Actual source code: matrix.c
1: /*
2: This is where the abstract matrix operations are defined
3: Portions of this code are under:
4: Copyright (c) 2022 Advanced Micro Devices, Inc. All rights reserved.
5: */
7: #include <petsc/private/matimpl.h>
8: #include <petsc/private/isimpl.h>
9: #include <petsc/private/vecimpl.h>
11: /* Logging support */
12: PetscClassId MAT_CLASSID;
13: PetscClassId MAT_COLORING_CLASSID;
14: PetscClassId MAT_FDCOLORING_CLASSID;
15: PetscClassId MAT_TRANSPOSECOLORING_CLASSID;
17: PetscLogEvent MAT_Mult, MAT_MultAdd, MAT_MultTranspose;
18: PetscLogEvent MAT_MultTransposeAdd, MAT_Solve, MAT_Solves, MAT_SolveAdd, MAT_SolveTranspose, MAT_MatSolve, MAT_MatTrSolve;
19: PetscLogEvent MAT_SolveTransposeAdd, MAT_SOR, MAT_ForwardSolve, MAT_BackwardSolve, MAT_LUFactor, MAT_LUFactorSymbolic;
20: PetscLogEvent MAT_LUFactorNumeric, MAT_CholeskyFactor, MAT_CholeskyFactorSymbolic, MAT_CholeskyFactorNumeric, MAT_ILUFactor;
21: PetscLogEvent MAT_ILUFactorSymbolic, MAT_ICCFactorSymbolic, MAT_Copy, MAT_Convert, MAT_Scale, MAT_AssemblyBegin;
22: PetscLogEvent MAT_QRFactorNumeric, MAT_QRFactorSymbolic, MAT_QRFactor;
23: PetscLogEvent MAT_AssemblyEnd, MAT_SetValues, MAT_GetValues, MAT_GetRow, MAT_GetRowIJ, MAT_CreateSubMats, MAT_GetOrdering, MAT_RedundantMat, MAT_GetSeqNonzeroStructure;
24: PetscLogEvent MAT_IncreaseOverlap, MAT_Partitioning, MAT_PartitioningND, MAT_Coarsen, MAT_ZeroEntries, MAT_Load, MAT_View, MAT_AXPY, MAT_FDColoringCreate;
25: PetscLogEvent MAT_FDColoringSetUp, MAT_FDColoringApply, MAT_Transpose, MAT_FDColoringFunction, MAT_CreateSubMat;
26: PetscLogEvent MAT_TransposeColoringCreate;
27: PetscLogEvent MAT_MatMult, MAT_MatMultSymbolic, MAT_MatMultNumeric;
28: PetscLogEvent MAT_PtAP, MAT_PtAPSymbolic, MAT_PtAPNumeric, MAT_RARt, MAT_RARtSymbolic, MAT_RARtNumeric;
29: PetscLogEvent MAT_MatTransposeMult, MAT_MatTransposeMultSymbolic, MAT_MatTransposeMultNumeric;
30: PetscLogEvent MAT_TransposeMatMult, MAT_TransposeMatMultSymbolic, MAT_TransposeMatMultNumeric;
31: PetscLogEvent MAT_MatMatMult, MAT_MatMatMultSymbolic, MAT_MatMatMultNumeric;
32: PetscLogEvent MAT_MultHermitianTranspose, MAT_MultHermitianTransposeAdd;
33: PetscLogEvent MAT_Getsymtransreduced, MAT_GetBrowsOfAcols;
34: PetscLogEvent MAT_GetBrowsOfAocols, MAT_Getlocalmat, MAT_Getlocalmatcondensed, MAT_Seqstompi, MAT_Seqstompinum, MAT_Seqstompisym;
35: PetscLogEvent MAT_GetMultiProcBlock;
36: PetscLogEvent MAT_CUSPARSECopyToGPU, MAT_CUSPARSECopyFromGPU, MAT_CUSPARSEGenerateTranspose, MAT_CUSPARSESolveAnalysis;
37: PetscLogEvent MAT_HIPSPARSECopyToGPU, MAT_HIPSPARSECopyFromGPU, MAT_HIPSPARSEGenerateTranspose, MAT_HIPSPARSESolveAnalysis;
38: PetscLogEvent MAT_PreallCOO, MAT_SetVCOO;
39: PetscLogEvent MAT_SetValuesBatch;
40: PetscLogEvent MAT_ViennaCLCopyToGPU;
41: PetscLogEvent MAT_DenseCopyToGPU, MAT_DenseCopyFromGPU;
42: PetscLogEvent MAT_Merge, MAT_Residual, MAT_SetRandom;
43: PetscLogEvent MAT_FactorFactS, MAT_FactorInvS;
44: PetscLogEvent MATCOLORING_Apply, MATCOLORING_Comm, MATCOLORING_Local, MATCOLORING_ISCreate, MATCOLORING_SetUp, MATCOLORING_Weights;
45: PetscLogEvent MAT_H2Opus_Build, MAT_H2Opus_Compress, MAT_H2Opus_Orthog, MAT_H2Opus_LR;
47: const char *const MatFactorTypes[] = {"NONE", "LU", "CHOLESKY", "ILU", "ICC", "ILUDT", "QR", "MatFactorType", "MAT_FACTOR_", NULL};
49: /*@
50: MatSetRandom - Sets all components of a matrix to random numbers.
52: Logically Collective
54: Input Parameters:
55: + x - the matrix
56: - rctx - the `PetscRandom` object, formed by `PetscRandomCreate()`, or `NULL` and
57: it will create one internally.
59: Example:
60: .vb
61: PetscRandomCreate(PETSC_COMM_WORLD,&rctx);
62: MatSetRandom(x,rctx);
63: PetscRandomDestroy(rctx);
64: .ve
66: Level: intermediate
68: Notes:
69: For sparse matrices that have been preallocated but not been assembled it randomly selects appropriate locations,
71: for sparse matrices that already have locations it fills the locations with random numbers.
73: It generates an error if used on sparse matrices that have not been preallocated.
75: .seealso: [](ch_matrices), `Mat`, `PetscRandom`, `PetscRandomCreate()`, `MatZeroEntries()`, `MatSetValues()`, `PetscRandomCreate()`, `PetscRandomDestroy()`
76: @*/
77: PetscErrorCode MatSetRandom(Mat x, PetscRandom rctx)
78: {
79: PetscRandom randObj = NULL;
81: PetscFunctionBegin;
85: MatCheckPreallocated(x, 1);
87: if (!rctx) {
88: MPI_Comm comm;
89: PetscCall(PetscObjectGetComm((PetscObject)x, &comm));
90: PetscCall(PetscRandomCreate(comm, &randObj));
91: PetscCall(PetscRandomSetType(randObj, x->defaultrandtype));
92: PetscCall(PetscRandomSetFromOptions(randObj));
93: rctx = randObj;
94: }
95: PetscCall(PetscLogEventBegin(MAT_SetRandom, x, rctx, 0, 0));
96: PetscUseTypeMethod(x, setrandom, rctx);
97: PetscCall(PetscLogEventEnd(MAT_SetRandom, x, rctx, 0, 0));
99: PetscCall(MatAssemblyBegin(x, MAT_FINAL_ASSEMBLY));
100: PetscCall(MatAssemblyEnd(x, MAT_FINAL_ASSEMBLY));
101: PetscCall(PetscRandomDestroy(&randObj));
102: PetscFunctionReturn(PETSC_SUCCESS);
103: }
105: /*@
106: MatFactorGetErrorZeroPivot - returns the pivot value that was determined to be zero and the row it occurred in
108: Logically Collective
110: Input Parameter:
111: . mat - the factored matrix
113: Output Parameters:
114: + pivot - the pivot value computed
115: - row - the row that the zero pivot occurred. This row value must be interpreted carefully due to row reorderings and which processes
116: the share the matrix
118: Level: advanced
120: Notes:
121: This routine does not work for factorizations done with external packages.
123: This routine should only be called if `MatGetFactorError()` returns a value of `MAT_FACTOR_NUMERIC_ZEROPIVOT`
125: This can also be called on non-factored matrices that come from, for example, matrices used in SOR.
127: .seealso: [](ch_matrices), `Mat`, `MatZeroEntries()`, `MatFactor()`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`, `MatFactorClearError()`, `MatFactorGetErrorZeroPivot()`,
128: `MAT_FACTOR_NUMERIC_ZEROPIVOT`
129: @*/
130: PetscErrorCode MatFactorGetErrorZeroPivot(Mat mat, PetscReal *pivot, PetscInt *row)
131: {
132: PetscFunctionBegin;
136: *pivot = mat->factorerror_zeropivot_value;
137: *row = mat->factorerror_zeropivot_row;
138: PetscFunctionReturn(PETSC_SUCCESS);
139: }
141: /*@
142: MatFactorGetError - gets the error code from a factorization
144: Logically Collective
146: Input Parameter:
147: . mat - the factored matrix
149: Output Parameter:
150: . err - the error code
152: Level: advanced
154: Note:
155: This can also be called on non-factored matrices that come from, for example, matrices used in SOR.
157: .seealso: [](ch_matrices), `Mat`, `MatZeroEntries()`, `MatFactor()`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`,
158: `MatFactorClearError()`, `MatFactorGetErrorZeroPivot()`, `MatFactorError`
159: @*/
160: PetscErrorCode MatFactorGetError(Mat mat, MatFactorError *err)
161: {
162: PetscFunctionBegin;
165: *err = mat->factorerrortype;
166: PetscFunctionReturn(PETSC_SUCCESS);
167: }
169: /*@
170: MatFactorClearError - clears the error code in a factorization
172: Logically Collective
174: Input Parameter:
175: . mat - the factored matrix
177: Level: developer
179: Note:
180: This can also be called on non-factored matrices that come from, for example, matrices used in SOR.
182: .seealso: [](ch_matrices), `Mat`, `MatZeroEntries()`, `MatFactor()`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`, `MatFactorGetError()`, `MatFactorGetErrorZeroPivot()`,
183: `MatGetErrorCode()`, `MatFactorError`
184: @*/
185: PetscErrorCode MatFactorClearError(Mat mat)
186: {
187: PetscFunctionBegin;
189: mat->factorerrortype = MAT_FACTOR_NOERROR;
190: mat->factorerror_zeropivot_value = 0.0;
191: mat->factorerror_zeropivot_row = 0;
192: PetscFunctionReturn(PETSC_SUCCESS);
193: }
195: PETSC_INTERN PetscErrorCode MatFindNonzeroRowsOrCols_Basic(Mat mat, PetscBool cols, PetscReal tol, IS *nonzero)
196: {
197: Vec r, l;
198: const PetscScalar *al;
199: PetscInt i, nz, gnz, N, n;
201: PetscFunctionBegin;
202: PetscCall(MatCreateVecs(mat, &r, &l));
203: if (!cols) { /* nonzero rows */
204: PetscCall(MatGetSize(mat, &N, NULL));
205: PetscCall(MatGetLocalSize(mat, &n, NULL));
206: PetscCall(VecSet(l, 0.0));
207: PetscCall(VecSetRandom(r, NULL));
208: PetscCall(MatMult(mat, r, l));
209: PetscCall(VecGetArrayRead(l, &al));
210: } else { /* nonzero columns */
211: PetscCall(MatGetSize(mat, NULL, &N));
212: PetscCall(MatGetLocalSize(mat, NULL, &n));
213: PetscCall(VecSet(r, 0.0));
214: PetscCall(VecSetRandom(l, NULL));
215: PetscCall(MatMultTranspose(mat, l, r));
216: PetscCall(VecGetArrayRead(r, &al));
217: }
218: if (tol <= 0.0) {
219: for (i = 0, nz = 0; i < n; i++)
220: if (al[i] != 0.0) nz++;
221: } else {
222: for (i = 0, nz = 0; i < n; i++)
223: if (PetscAbsScalar(al[i]) > tol) nz++;
224: }
225: PetscCall(MPIU_Allreduce(&nz, &gnz, 1, MPIU_INT, MPI_SUM, PetscObjectComm((PetscObject)mat)));
226: if (gnz != N) {
227: PetscInt *nzr;
228: PetscCall(PetscMalloc1(nz, &nzr));
229: if (nz) {
230: if (tol < 0) {
231: for (i = 0, nz = 0; i < n; i++)
232: if (al[i] != 0.0) nzr[nz++] = i;
233: } else {
234: for (i = 0, nz = 0; i < n; i++)
235: if (PetscAbsScalar(al[i]) > tol) nzr[nz++] = i;
236: }
237: }
238: PetscCall(ISCreateGeneral(PetscObjectComm((PetscObject)mat), nz, nzr, PETSC_OWN_POINTER, nonzero));
239: } else *nonzero = NULL;
240: if (!cols) { /* nonzero rows */
241: PetscCall(VecRestoreArrayRead(l, &al));
242: } else {
243: PetscCall(VecRestoreArrayRead(r, &al));
244: }
245: PetscCall(VecDestroy(&l));
246: PetscCall(VecDestroy(&r));
247: PetscFunctionReturn(PETSC_SUCCESS);
248: }
250: /*@
251: MatFindNonzeroRows - Locate all rows that are not completely zero in the matrix
253: Input Parameter:
254: . A - the matrix
256: Output Parameter:
257: . keptrows - the rows that are not completely zero
259: Level: intermediate
261: Note:
262: `keptrows` is set to `NULL` if all rows are nonzero.
264: .seealso: [](ch_matrices), `Mat`, `MatFindZeroRows()`
265: @*/
266: PetscErrorCode MatFindNonzeroRows(Mat mat, IS *keptrows)
267: {
268: PetscFunctionBegin;
272: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
273: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
274: if (mat->ops->findnonzerorows) PetscUseTypeMethod(mat, findnonzerorows, keptrows);
275: else PetscCall(MatFindNonzeroRowsOrCols_Basic(mat, PETSC_FALSE, 0.0, keptrows));
276: PetscFunctionReturn(PETSC_SUCCESS);
277: }
279: /*@
280: MatFindZeroRows - Locate all rows that are completely zero in the matrix
282: Input Parameter:
283: . A - the matrix
285: Output Parameter:
286: . zerorows - the rows that are completely zero
288: Level: intermediate
290: Note:
291: `zerorows` is set to `NULL` if no rows are zero.
293: .seealso: [](ch_matrices), `Mat`, `MatFindNonzeroRows()`
294: @*/
295: PetscErrorCode MatFindZeroRows(Mat mat, IS *zerorows)
296: {
297: IS keptrows;
298: PetscInt m, n;
300: PetscFunctionBegin;
304: PetscCall(MatFindNonzeroRows(mat, &keptrows));
305: /* MatFindNonzeroRows sets keptrows to NULL if there are no zero rows.
306: In keeping with this convention, we set zerorows to NULL if there are no zero
307: rows. */
308: if (keptrows == NULL) {
309: *zerorows = NULL;
310: } else {
311: PetscCall(MatGetOwnershipRange(mat, &m, &n));
312: PetscCall(ISComplement(keptrows, m, n, zerorows));
313: PetscCall(ISDestroy(&keptrows));
314: }
315: PetscFunctionReturn(PETSC_SUCCESS);
316: }
318: /*@
319: MatGetDiagonalBlock - Returns the part of the matrix associated with the on-process coupling
321: Not Collective
323: Input Parameter:
324: . A - the matrix
326: Output Parameter:
327: . a - the diagonal part (which is a SEQUENTIAL matrix)
329: Level: advanced
331: Notes:
332: See `MatCreateAIJ()` for more information on the "diagonal part" of the matrix.
334: Use caution, as the reference count on the returned matrix is not incremented and it is used as part of `A`'s normal operation.
336: .seealso: [](ch_matrices), `Mat`, `MatCreateAIJ()`, `MATAIJ`, `MATBAIJ`, `MATSBAIJ`
337: @*/
338: PetscErrorCode MatGetDiagonalBlock(Mat A, Mat *a)
339: {
340: PetscFunctionBegin;
344: PetscCheck(!A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
345: if (A->ops->getdiagonalblock) PetscUseTypeMethod(A, getdiagonalblock, a);
346: else {
347: PetscMPIInt size;
349: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)A), &size));
350: PetscCheck(size == 1, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Not for parallel matrix type %s", ((PetscObject)A)->type_name);
351: *a = A;
352: }
353: PetscFunctionReturn(PETSC_SUCCESS);
354: }
356: /*@
357: MatGetTrace - Gets the trace of a matrix. The sum of the diagonal entries.
359: Collective
361: Input Parameter:
362: . mat - the matrix
364: Output Parameter:
365: . trace - the sum of the diagonal entries
367: Level: advanced
369: .seealso: [](ch_matrices), `Mat`
370: @*/
371: PetscErrorCode MatGetTrace(Mat mat, PetscScalar *trace)
372: {
373: Vec diag;
375: PetscFunctionBegin;
378: PetscCall(MatCreateVecs(mat, &diag, NULL));
379: PetscCall(MatGetDiagonal(mat, diag));
380: PetscCall(VecSum(diag, trace));
381: PetscCall(VecDestroy(&diag));
382: PetscFunctionReturn(PETSC_SUCCESS);
383: }
385: /*@
386: MatRealPart - Zeros out the imaginary part of the matrix
388: Logically Collective
390: Input Parameter:
391: . mat - the matrix
393: Level: advanced
395: .seealso: [](ch_matrices), `Mat`, `MatImaginaryPart()`
396: @*/
397: PetscErrorCode MatRealPart(Mat mat)
398: {
399: PetscFunctionBegin;
402: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
403: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
404: MatCheckPreallocated(mat, 1);
405: PetscUseTypeMethod(mat, realpart);
406: PetscFunctionReturn(PETSC_SUCCESS);
407: }
409: /*@C
410: MatGetGhosts - Get the global indices of all ghost nodes defined by the sparse matrix
412: Collective
414: Input Parameter:
415: . mat - the matrix
417: Output Parameters:
418: + nghosts - number of ghosts (for `MATBAIJ` and `MATSBAIJ` matrices there is one ghost for each block)
419: - ghosts - the global indices of the ghost points
421: Level: advanced
423: Note:
424: `nghosts` and `ghosts` are suitable to pass into `VecCreateGhost()`
426: .seealso: [](ch_matrices), `Mat`, `VecCreateGhost()`
427: @*/
428: PetscErrorCode MatGetGhosts(Mat mat, PetscInt *nghosts, const PetscInt *ghosts[])
429: {
430: PetscFunctionBegin;
433: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
434: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
435: if (mat->ops->getghosts) PetscUseTypeMethod(mat, getghosts, nghosts, ghosts);
436: else {
437: if (nghosts) *nghosts = 0;
438: if (ghosts) *ghosts = NULL;
439: }
440: PetscFunctionReturn(PETSC_SUCCESS);
441: }
443: /*@
444: MatImaginaryPart - Moves the imaginary part of the matrix to the real part and zeros the imaginary part
446: Logically Collective
448: Input Parameter:
449: . mat - the matrix
451: Level: advanced
453: .seealso: [](ch_matrices), `Mat`, `MatRealPart()`
454: @*/
455: PetscErrorCode MatImaginaryPart(Mat mat)
456: {
457: PetscFunctionBegin;
460: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
461: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
462: MatCheckPreallocated(mat, 1);
463: PetscUseTypeMethod(mat, imaginarypart);
464: PetscFunctionReturn(PETSC_SUCCESS);
465: }
467: /*@
468: MatMissingDiagonal - Determine if sparse matrix is missing a diagonal entry (or block entry for `MATBAIJ` and `MATSBAIJ` matrices)
470: Not Collective
472: Input Parameter:
473: . mat - the matrix
475: Output Parameters:
476: + missing - is any diagonal missing
477: - dd - first diagonal entry that is missing (optional) on this process
479: Level: advanced
481: .seealso: [](ch_matrices), `Mat`
482: @*/
483: PetscErrorCode MatMissingDiagonal(Mat mat, PetscBool *missing, PetscInt *dd)
484: {
485: PetscFunctionBegin;
489: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix %s", ((PetscObject)mat)->type_name);
490: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
491: PetscUseTypeMethod(mat, missingdiagonal, missing, dd);
492: PetscFunctionReturn(PETSC_SUCCESS);
493: }
495: /*@C
496: MatGetRow - Gets a row of a matrix. You MUST call `MatRestoreRow()`
497: for each row that you get to ensure that your application does
498: not bleed memory.
500: Not Collective
502: Input Parameters:
503: + mat - the matrix
504: - row - the row to get
506: Output Parameters:
507: + ncols - if not `NULL`, the number of nonzeros in the row
508: . cols - if not `NULL`, the column numbers
509: - vals - if not `NULL`, the values
511: Level: advanced
513: Notes:
514: This routine is provided for people who need to have direct access
515: to the structure of a matrix. We hope that we provide enough
516: high-level matrix routines that few users will need it.
518: `MatGetRow()` always returns 0-based column indices, regardless of
519: whether the internal representation is 0-based (default) or 1-based.
521: For better efficiency, set cols and/or vals to `NULL` if you do
522: not wish to extract these quantities.
524: The user can only examine the values extracted with `MatGetRow()`;
525: the values cannot be altered. To change the matrix entries, one
526: must use `MatSetValues()`.
528: You can only have one call to `MatGetRow()` outstanding for a particular
529: matrix at a time, per processor. `MatGetRow()` can only obtain rows
530: associated with the given processor, it cannot get rows from the
531: other processors; for that we suggest using `MatCreateSubMatrices()`, then
532: MatGetRow() on the submatrix. The row index passed to `MatGetRow()`
533: is in the global number of rows.
535: Use `MatGetRowIJ()` and `MatRestoreRowIJ()` to access all the local indices of the sparse matrix.
537: Use `MatSeqAIJGetArray()` and similar functions to access the numerical values for certain matrix types directly.
539: Fortran Note:
540: The calling sequence is
541: .vb
542: MatGetRow(matrix,row,ncols,cols,values,ierr)
543: Mat matrix (input)
544: integer row (input)
545: integer ncols (output)
546: integer cols(maxcols) (output)
547: double precision (or double complex) values(maxcols) output
548: .ve
549: where maxcols >= maximum nonzeros in any row of the matrix.
551: Caution:
552: Do not try to change the contents of the output arrays (`cols` and `vals`).
553: In some cases, this may corrupt the matrix.
555: .seealso: [](ch_matrices), `Mat`, `MatRestoreRow()`, `MatSetValues()`, `MatGetValues()`, `MatCreateSubMatrices()`, `MatGetDiagonal()`, `MatGetRowIJ()`, `MatRestoreRowIJ()`
556: @*/
557: PetscErrorCode MatGetRow(Mat mat, PetscInt row, PetscInt *ncols, const PetscInt *cols[], const PetscScalar *vals[])
558: {
559: PetscInt incols;
561: PetscFunctionBegin;
564: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
565: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
566: MatCheckPreallocated(mat, 1);
567: PetscCheck(row >= mat->rmap->rstart && row < mat->rmap->rend, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Only for local rows, %" PetscInt_FMT " not in [%" PetscInt_FMT ",%" PetscInt_FMT ")", row, mat->rmap->rstart, mat->rmap->rend);
568: PetscCall(PetscLogEventBegin(MAT_GetRow, mat, 0, 0, 0));
569: PetscUseTypeMethod(mat, getrow, row, &incols, (PetscInt **)cols, (PetscScalar **)vals);
570: if (ncols) *ncols = incols;
571: PetscCall(PetscLogEventEnd(MAT_GetRow, mat, 0, 0, 0));
572: PetscFunctionReturn(PETSC_SUCCESS);
573: }
575: /*@
576: MatConjugate - replaces the matrix values with their complex conjugates
578: Logically Collective
580: Input Parameter:
581: . mat - the matrix
583: Level: advanced
585: .seealso: [](ch_matrices), `Mat`, `MatRealPart()`, `MatImaginaryPart()`, `VecConjugate()`, `MatTranspose()`
586: @*/
587: PetscErrorCode MatConjugate(Mat mat)
588: {
589: PetscFunctionBegin;
591: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
592: if (PetscDefined(USE_COMPLEX) && mat->hermitian != PETSC_BOOL3_TRUE) {
593: PetscUseTypeMethod(mat, conjugate);
594: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
595: }
596: PetscFunctionReturn(PETSC_SUCCESS);
597: }
599: /*@C
600: MatRestoreRow - Frees any temporary space allocated by `MatGetRow()`.
602: Not Collective
604: Input Parameters:
605: + mat - the matrix
606: . row - the row to get
607: . ncols - the number of nonzeros
608: . cols - the columns of the nonzeros
609: - vals - if nonzero the column values
611: Level: advanced
613: Notes:
614: This routine should be called after you have finished examining the entries.
616: This routine zeros out `ncols`, `cols`, and `vals`. This is to prevent accidental
617: us of the array after it has been restored. If you pass `NULL`, it will
618: not zero the pointers. Use of `cols` or `vals` after `MatRestoreRow()` is invalid.
620: Fortran Notes:
621: The calling sequence is
622: .vb
623: MatRestoreRow(matrix,row,ncols,cols,values,ierr)
624: Mat matrix (input)
625: integer row (input)
626: integer ncols (output)
627: integer cols(maxcols) (output)
628: double precision (or double complex) values(maxcols) output
629: .ve
630: Where maxcols >= maximum nonzeros in any row of the matrix.
632: In Fortran `MatRestoreRow()` MUST be called after `MatGetRow()`
633: before another call to `MatGetRow()` can be made.
635: .seealso: [](ch_matrices), `Mat`, `MatGetRow()`
636: @*/
637: PetscErrorCode MatRestoreRow(Mat mat, PetscInt row, PetscInt *ncols, const PetscInt *cols[], const PetscScalar *vals[])
638: {
639: PetscFunctionBegin;
642: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
643: if (!mat->ops->restorerow) PetscFunctionReturn(PETSC_SUCCESS);
644: PetscUseTypeMethod(mat, restorerow, row, ncols, (PetscInt **)cols, (PetscScalar **)vals);
645: if (ncols) *ncols = 0;
646: if (cols) *cols = NULL;
647: if (vals) *vals = NULL;
648: PetscFunctionReturn(PETSC_SUCCESS);
649: }
651: /*@
652: MatGetRowUpperTriangular - Sets a flag to enable calls to `MatGetRow()` for matrix in `MATSBAIJ` format.
653: You should call `MatRestoreRowUpperTriangular()` after calling` MatGetRow()` and `MatRestoreRow()` to disable the flag.
655: Not Collective
657: Input Parameter:
658: . mat - the matrix
660: Level: advanced
662: Note:
663: The flag is to ensure that users are aware that `MatGetRow()` only provides the upper triangular part of the row for the matrices in `MATSBAIJ` format.
665: .seealso: [](ch_matrices), `Mat`, `MATSBAIJ`, `MatRestoreRowUpperTriangular()`
666: @*/
667: PetscErrorCode MatGetRowUpperTriangular(Mat mat)
668: {
669: PetscFunctionBegin;
672: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
673: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
674: MatCheckPreallocated(mat, 1);
675: if (!mat->ops->getrowuppertriangular) PetscFunctionReturn(PETSC_SUCCESS);
676: PetscUseTypeMethod(mat, getrowuppertriangular);
677: PetscFunctionReturn(PETSC_SUCCESS);
678: }
680: /*@
681: MatRestoreRowUpperTriangular - Disable calls to `MatGetRow()` for matrix in `MATSBAIJ` format.
683: Not Collective
685: Input Parameter:
686: . mat - the matrix
688: Level: advanced
690: Note:
691: This routine should be called after you have finished calls to `MatGetRow()` and `MatRestoreRow()`.
693: .seealso: [](ch_matrices), `Mat`, `MATSBAIJ`, `MatGetRowUpperTriangular()`
694: @*/
695: PetscErrorCode MatRestoreRowUpperTriangular(Mat mat)
696: {
697: PetscFunctionBegin;
700: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
701: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
702: MatCheckPreallocated(mat, 1);
703: if (!mat->ops->restorerowuppertriangular) PetscFunctionReturn(PETSC_SUCCESS);
704: PetscUseTypeMethod(mat, restorerowuppertriangular);
705: PetscFunctionReturn(PETSC_SUCCESS);
706: }
708: /*@C
709: MatSetOptionsPrefix - Sets the prefix used for searching for all
710: `Mat` options in the database.
712: Logically Collective
714: Input Parameters:
715: + A - the matrix
716: - prefix - the prefix to prepend to all option names
718: Level: advanced
720: Notes:
721: A hyphen (-) must NOT be given at the beginning of the prefix name.
722: The first character of all runtime options is AUTOMATICALLY the hyphen.
724: This is NOT used for options for the factorization of the matrix. Normally the
725: prefix is automatically passed in from the PC calling the factorization. To set
726: it directly use `MatSetOptionsPrefixFactor()`
728: .seealso: [](ch_matrices), `Mat`, `MatSetFromOptions()`, `MatSetOptionsPrefixFactor()`
729: @*/
730: PetscErrorCode MatSetOptionsPrefix(Mat A, const char prefix[])
731: {
732: PetscFunctionBegin;
734: PetscCall(PetscObjectSetOptionsPrefix((PetscObject)A, prefix));
735: PetscFunctionReturn(PETSC_SUCCESS);
736: }
738: /*@C
739: MatSetOptionsPrefixFactor - Sets the prefix used for searching for all matrix factor options in the database for
740: for matrices created with `MatGetFactor()`
742: Logically Collective
744: Input Parameters:
745: + A - the matrix
746: - prefix - the prefix to prepend to all option names for the factored matrix
748: Level: developer
750: Notes:
751: A hyphen (-) must NOT be given at the beginning of the prefix name.
752: The first character of all runtime options is AUTOMATICALLY the hyphen.
754: Normally the prefix is automatically passed in from the `PC` calling the factorization. To set
755: it directly when not using `KSP`/`PC` use `MatSetOptionsPrefixFactor()`
757: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatSetFromOptions()`, `MatSetOptionsPrefix()`, `MatAppendOptionsPrefixFactor()`
758: @*/
759: PetscErrorCode MatSetOptionsPrefixFactor(Mat A, const char prefix[])
760: {
761: PetscFunctionBegin;
763: if (prefix) {
765: PetscCheck(prefix[0] != '-', PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONG, "Options prefix should not begin with a hyphen");
766: if (prefix != A->factorprefix) {
767: PetscCall(PetscFree(A->factorprefix));
768: PetscCall(PetscStrallocpy(prefix, &A->factorprefix));
769: }
770: } else PetscCall(PetscFree(A->factorprefix));
771: PetscFunctionReturn(PETSC_SUCCESS);
772: }
774: /*@C
775: MatAppendOptionsPrefixFactor - Appends to the prefix used for searching for all matrix factor options in the database for
776: for matrices created with `MatGetFactor()`
778: Logically Collective
780: Input Parameters:
781: + A - the matrix
782: - prefix - the prefix to prepend to all option names for the factored matrix
784: Level: developer
786: Notes:
787: A hyphen (-) must NOT be given at the beginning of the prefix name.
788: The first character of all runtime options is AUTOMATICALLY the hyphen.
790: Normally the prefix is automatically passed in from the `PC` calling the factorization. To set
791: it directly when not using `KSP`/`PC` use `MatAppendOptionsPrefixFactor()`
793: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, PetscOptionsCreate()`, `PetscOptionsDestroy()`, `PetscObjectSetOptionsPrefix()`, `PetscObjectPrependOptionsPrefix()`,
794: `PetscObjectGetOptionsPrefix()`, `TSAppendOptionsPrefix()`, `SNESAppendOptionsPrefix()`, `KSPAppendOptionsPrefix()`, `MatSetOptionsPrefixFactor()`,
795: `MatSetOptionsPrefix()`
796: @*/
797: PetscErrorCode MatAppendOptionsPrefixFactor(Mat A, const char prefix[])
798: {
799: size_t len1, len2, new_len;
801: PetscFunctionBegin;
803: if (!prefix) PetscFunctionReturn(PETSC_SUCCESS);
804: if (!A->factorprefix) {
805: PetscCall(MatSetOptionsPrefixFactor(A, prefix));
806: PetscFunctionReturn(PETSC_SUCCESS);
807: }
808: PetscCheck(prefix[0] != '-', PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONG, "Options prefix should not begin with a hyphen");
810: PetscCall(PetscStrlen(A->factorprefix, &len1));
811: PetscCall(PetscStrlen(prefix, &len2));
812: new_len = len1 + len2 + 1;
813: PetscCall(PetscRealloc(new_len * sizeof(*(A->factorprefix)), &A->factorprefix));
814: PetscCall(PetscStrncpy(A->factorprefix + len1, prefix, len2 + 1));
815: PetscFunctionReturn(PETSC_SUCCESS);
816: }
818: /*@C
819: MatAppendOptionsPrefix - Appends to the prefix used for searching for all
820: matrix options in the database.
822: Logically Collective
824: Input Parameters:
825: + A - the matrix
826: - prefix - the prefix to prepend to all option names
828: Level: advanced
830: Note:
831: A hyphen (-) must NOT be given at the beginning of the prefix name.
832: The first character of all runtime options is AUTOMATICALLY the hyphen.
834: .seealso: [](ch_matrices), `Mat`, `MatGetOptionsPrefix()`, `MatAppendOptionsPrefixFactor()`, `MatSetOptionsPrefix()`
835: @*/
836: PetscErrorCode MatAppendOptionsPrefix(Mat A, const char prefix[])
837: {
838: PetscFunctionBegin;
840: PetscCall(PetscObjectAppendOptionsPrefix((PetscObject)A, prefix));
841: PetscFunctionReturn(PETSC_SUCCESS);
842: }
844: /*@C
845: MatGetOptionsPrefix - Gets the prefix used for searching for all
846: matrix options in the database.
848: Not Collective
850: Input Parameter:
851: . A - the matrix
853: Output Parameter:
854: . prefix - pointer to the prefix string used
856: Level: advanced
858: Fortran Note:
859: The user should pass in a string `prefix` of
860: sufficient length to hold the prefix.
862: .seealso: [](ch_matrices), `Mat`, `MatAppendOptionsPrefix()`, `MatSetOptionsPrefix()`, `MatAppendOptionsPrefixFactor()`, `MatSetOptionsPrefixFactor()`
863: @*/
864: PetscErrorCode MatGetOptionsPrefix(Mat A, const char *prefix[])
865: {
866: PetscFunctionBegin;
869: PetscCall(PetscObjectGetOptionsPrefix((PetscObject)A, prefix));
870: PetscFunctionReturn(PETSC_SUCCESS);
871: }
873: /*@
874: MatResetPreallocation - Reset matrix to use the original nonzero pattern provided by users.
876: Collective
878: Input Parameter:
879: . A - the matrix
881: Level: beginner
883: Notes:
884: The allocated memory will be shrunk after calling `MatAssemblyBegin()` and `MatAssemblyEnd()` with `MAT_FINAL_ASSEMBLY`.
886: Users can reset the preallocation to access the original memory.
888: Currently only supported for `MATAIJ` matrices.
890: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJSetPreallocation()`, `MatMPIAIJSetPreallocation()`, `MatXAIJSetPreallocation()`
891: @*/
892: PetscErrorCode MatResetPreallocation(Mat A)
893: {
894: PetscFunctionBegin;
897: PetscCheck(A->insertmode == NOT_SET_VALUES, PETSC_COMM_SELF, PETSC_ERR_SUP, "Cannot reset preallocation after setting some values but not yet calling MatAssemblyBegin()/MatAsssemblyEnd()");
898: if (A->num_ass == 0) PetscFunctionReturn(PETSC_SUCCESS);
899: PetscUseMethod(A, "MatResetPreallocation_C", (Mat), (A));
900: PetscFunctionReturn(PETSC_SUCCESS);
901: }
903: /*@
904: MatSetUp - Sets up the internal matrix data structures for later use.
906: Collective
908: Input Parameter:
909: . A - the matrix
911: Level: intermediate
913: Notes:
914: If the user has not set preallocation for this matrix then an efficient algorithm will be used for the first round of
915: setting values in the matrix.
917: If a suitable preallocation routine is used, this function does not need to be called.
919: This routine is called internally by other matrix functions when needed so rarely needs to be called by users
921: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `MatCreate()`, `MatDestroy()`, `MatXAIJSetPreallocation()`
922: @*/
923: PetscErrorCode MatSetUp(Mat A)
924: {
925: PetscFunctionBegin;
927: if (!((PetscObject)A)->type_name) {
928: PetscMPIInt size;
930: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)A), &size));
931: PetscCall(MatSetType(A, size == 1 ? MATSEQAIJ : MATMPIAIJ));
932: }
933: if (!A->preallocated) PetscTryTypeMethod(A, setup);
934: PetscCall(PetscLayoutSetUp(A->rmap));
935: PetscCall(PetscLayoutSetUp(A->cmap));
936: A->preallocated = PETSC_TRUE;
937: PetscFunctionReturn(PETSC_SUCCESS);
938: }
940: #if defined(PETSC_HAVE_SAWS)
941: #include <petscviewersaws.h>
942: #endif
944: /*
945: If threadsafety is on extraneous matrices may be printed
947: This flag cannot be stored in the matrix because the original matrix in MatView() may assemble a new matrix which is passed into MatViewFromOptions()
948: */
949: #if !defined(PETSC_HAVE_THREADSAFETY)
950: static PetscInt insidematview = 0;
951: #endif
953: /*@C
954: MatViewFromOptions - View properties of the matrix based on options set in the options database
956: Collective
958: Input Parameters:
959: + A - the matrix
960: . obj - optional additional object that provides the options prefix to use
961: - name - command line option
963: Options Database Key:
964: . -mat_view [viewertype]:... - the viewer and its options
966: Level: intermediate
968: Notes:
969: .vb
970: If no value is provided ascii:stdout is used
971: ascii[:[filename][:[format][:append]]] defaults to stdout - format can be one of ascii_info, ascii_info_detail, or ascii_matlab,
972: for example ascii::ascii_info prints just the information about the object not all details
973: unless :append is given filename opens in write mode, overwriting what was already there
974: binary[:[filename][:[format][:append]]] defaults to the file binaryoutput
975: draw[:drawtype[:filename]] for example, draw:tikz, draw:tikz:figure.tex or draw:x
976: socket[:port] defaults to the standard output port
977: saws[:communicatorname] publishes object to the Scientific Application Webserver (SAWs)
978: .ve
980: .seealso: [](ch_matrices), `Mat`, `MatView()`, `PetscObjectViewFromOptions()`, `MatCreate()`
981: @*/
982: PetscErrorCode MatViewFromOptions(Mat A, PetscObject obj, const char name[])
983: {
984: PetscFunctionBegin;
986: #if !defined(PETSC_HAVE_THREADSAFETY)
987: if (insidematview) PetscFunctionReturn(PETSC_SUCCESS);
988: #endif
989: PetscCall(PetscObjectViewFromOptions((PetscObject)A, obj, name));
990: PetscFunctionReturn(PETSC_SUCCESS);
991: }
993: /*@C
994: MatView - display information about a matrix in a variety ways
996: Collective
998: Input Parameters:
999: + mat - the matrix
1000: - viewer - visualization context
1002: Options Database Keys:
1003: + -mat_view ::ascii_info - Prints info on matrix at conclusion of `MatAssemblyEnd()`
1004: . -mat_view ::ascii_info_detail - Prints more detailed info
1005: . -mat_view - Prints matrix in ASCII format
1006: . -mat_view ::ascii_matlab - Prints matrix in Matlab format
1007: . -mat_view draw - PetscDraws nonzero structure of matrix, using `MatView()` and `PetscDrawOpenX()`.
1008: . -display <name> - Sets display name (default is host)
1009: . -draw_pause <sec> - Sets number of seconds to pause after display
1010: . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (see Users-Manual: ch_matlab for details)
1011: . -viewer_socket_machine <machine> -
1012: . -viewer_socket_port <port> -
1013: . -mat_view binary - save matrix to file in binary format
1014: - -viewer_binary_filename <name> -
1016: Level: beginner
1018: Notes:
1019: The available visualization contexts include
1020: + `PETSC_VIEWER_STDOUT_SELF` - for sequential matrices
1021: . `PETSC_VIEWER_STDOUT_WORLD` - for parallel matrices created on `PETSC_COMM_WORLD`
1022: . `PETSC_VIEWER_STDOUT_`(comm) - for matrices created on MPI communicator comm
1023: - `PETSC_VIEWER_DRAW_WORLD` - graphical display of nonzero structure
1025: The user can open alternative visualization contexts with
1026: + `PetscViewerASCIIOpen()` - Outputs matrix to a specified file
1027: . `PetscViewerBinaryOpen()` - Outputs matrix in binary to a
1028: specified file; corresponding input uses MatLoad()
1029: . `PetscViewerDrawOpen()` - Outputs nonzero matrix structure to
1030: an X window display
1031: - `PetscViewerSocketOpen()` - Outputs matrix to Socket viewer.
1032: Currently only the sequential dense and AIJ
1033: matrix types support the Socket viewer.
1035: The user can call `PetscViewerPushFormat()` to specify the output
1036: format of ASCII printed objects (when using `PETSC_VIEWER_STDOUT_SELF`,
1037: `PETSC_VIEWER_STDOUT_WORLD` and `PetscViewerASCIIOpen()`). Available formats include
1038: + `PETSC_VIEWER_DEFAULT` - default, prints matrix contents
1039: . `PETSC_VIEWER_ASCII_MATLAB` - prints matrix contents in Matlab format
1040: . `PETSC_VIEWER_ASCII_DENSE` - prints entire matrix including zeros
1041: . `PETSC_VIEWER_ASCII_COMMON` - prints matrix contents, using a sparse
1042: format common among all matrix types
1043: . `PETSC_VIEWER_ASCII_IMPL` - prints matrix contents, using an implementation-specific
1044: format (which is in many cases the same as the default)
1045: . `PETSC_VIEWER_ASCII_INFO` - prints basic information about the matrix
1046: size and structure (not the matrix entries)
1047: - `PETSC_VIEWER_ASCII_INFO_DETAIL` - prints more detailed information about
1048: the matrix structure
1050: The ASCII viewers are only recommended for small matrices on at most a moderate number of processes,
1051: the program will seemingly hang and take hours for larger matrices, for larger matrices one should use the binary format.
1053: In the debugger you can do "call MatView(mat,0)" to display the matrix. (The same holds for any PETSc object viewer).
1055: See the manual page for `MatLoad()` for the exact format of the binary file when the binary
1056: viewer is used.
1058: See share/petsc/matlab/PetscBinaryRead.m for a Matlab code that can read in the binary file when the binary
1059: viewer is used and lib/petsc/bin/PetscBinaryIO.py for loading them into Python.
1061: One can use '-mat_view draw -draw_pause -1' to pause the graphical display of matrix nonzero structure,
1062: and then use the following mouse functions.
1063: .vb
1064: left mouse: zoom in
1065: middle mouse: zoom out
1066: right mouse: continue with the simulation
1067: .ve
1069: .seealso: [](ch_matrices), `Mat`, `PetscViewerPushFormat()`, `PetscViewerASCIIOpen()`, `PetscViewerDrawOpen()`, `PetscViewer`,
1070: `PetscViewerSocketOpen()`, `PetscViewerBinaryOpen()`, `MatLoad()`, `MatViewFromOptions()`
1071: @*/
1072: PetscErrorCode MatView(Mat mat, PetscViewer viewer)
1073: {
1074: PetscInt rows, cols, rbs, cbs;
1075: PetscBool isascii, isstring, issaws;
1076: PetscViewerFormat format;
1077: PetscMPIInt size;
1079: PetscFunctionBegin;
1082: if (!viewer) PetscCall(PetscViewerASCIIGetStdout(PetscObjectComm((PetscObject)mat), &viewer));
1084: PetscCheckSameComm(mat, 1, viewer, 2);
1086: PetscCall(PetscViewerGetFormat(viewer, &format));
1087: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size));
1088: if (size == 1 && format == PETSC_VIEWER_LOAD_BALANCE) PetscFunctionReturn(PETSC_SUCCESS);
1090: #if !defined(PETSC_HAVE_THREADSAFETY)
1091: insidematview++;
1092: #endif
1093: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSTRING, &isstring));
1094: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
1095: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSAWS, &issaws));
1096: PetscCheck((isascii && (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL)) || !mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "No viewers for factored matrix except ASCII, info, or info_detail");
1098: PetscCall(PetscLogEventBegin(MAT_View, mat, viewer, 0, 0));
1099: if (isascii) {
1100: if (!mat->preallocated) {
1101: PetscCall(PetscViewerASCIIPrintf(viewer, "Matrix has not been preallocated yet\n"));
1102: #if !defined(PETSC_HAVE_THREADSAFETY)
1103: insidematview--;
1104: #endif
1105: PetscCall(PetscLogEventEnd(MAT_View, mat, viewer, 0, 0));
1106: PetscFunctionReturn(PETSC_SUCCESS);
1107: }
1108: if (!mat->assembled) {
1109: PetscCall(PetscViewerASCIIPrintf(viewer, "Matrix has not been assembled yet\n"));
1110: #if !defined(PETSC_HAVE_THREADSAFETY)
1111: insidematview--;
1112: #endif
1113: PetscCall(PetscLogEventEnd(MAT_View, mat, viewer, 0, 0));
1114: PetscFunctionReturn(PETSC_SUCCESS);
1115: }
1116: PetscCall(PetscObjectPrintClassNamePrefixType((PetscObject)mat, viewer));
1117: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
1118: MatNullSpace nullsp, transnullsp;
1120: PetscCall(PetscViewerASCIIPushTab(viewer));
1121: PetscCall(MatGetSize(mat, &rows, &cols));
1122: PetscCall(MatGetBlockSizes(mat, &rbs, &cbs));
1123: if (rbs != 1 || cbs != 1) {
1124: if (rbs != cbs) PetscCall(PetscViewerASCIIPrintf(viewer, "rows=%" PetscInt_FMT ", cols=%" PetscInt_FMT ", rbs=%" PetscInt_FMT ", cbs=%" PetscInt_FMT "\n", rows, cols, rbs, cbs));
1125: else PetscCall(PetscViewerASCIIPrintf(viewer, "rows=%" PetscInt_FMT ", cols=%" PetscInt_FMT ", bs=%" PetscInt_FMT "\n", rows, cols, rbs));
1126: } else PetscCall(PetscViewerASCIIPrintf(viewer, "rows=%" PetscInt_FMT ", cols=%" PetscInt_FMT "\n", rows, cols));
1127: if (mat->factortype) {
1128: MatSolverType solver;
1129: PetscCall(MatFactorGetSolverType(mat, &solver));
1130: PetscCall(PetscViewerASCIIPrintf(viewer, "package used to perform factorization: %s\n", solver));
1131: }
1132: if (mat->ops->getinfo) {
1133: MatInfo info;
1134: PetscCall(MatGetInfo(mat, MAT_GLOBAL_SUM, &info));
1135: PetscCall(PetscViewerASCIIPrintf(viewer, "total: nonzeros=%.f, allocated nonzeros=%.f\n", info.nz_used, info.nz_allocated));
1136: if (!mat->factortype) PetscCall(PetscViewerASCIIPrintf(viewer, "total number of mallocs used during MatSetValues calls=%" PetscInt_FMT "\n", (PetscInt)info.mallocs));
1137: }
1138: PetscCall(MatGetNullSpace(mat, &nullsp));
1139: PetscCall(MatGetTransposeNullSpace(mat, &transnullsp));
1140: if (nullsp) PetscCall(PetscViewerASCIIPrintf(viewer, " has attached null space\n"));
1141: if (transnullsp && transnullsp != nullsp) PetscCall(PetscViewerASCIIPrintf(viewer, " has attached transposed null space\n"));
1142: PetscCall(MatGetNearNullSpace(mat, &nullsp));
1143: if (nullsp) PetscCall(PetscViewerASCIIPrintf(viewer, " has attached near null space\n"));
1144: PetscCall(PetscViewerASCIIPushTab(viewer));
1145: PetscCall(MatProductView(mat, viewer));
1146: PetscCall(PetscViewerASCIIPopTab(viewer));
1147: }
1148: } else if (issaws) {
1149: #if defined(PETSC_HAVE_SAWS)
1150: PetscMPIInt rank;
1152: PetscCall(PetscObjectName((PetscObject)mat));
1153: PetscCallMPI(MPI_Comm_rank(PETSC_COMM_WORLD, &rank));
1154: if (!((PetscObject)mat)->amsmem && rank == 0) PetscCall(PetscObjectViewSAWs((PetscObject)mat, viewer));
1155: #endif
1156: } else if (isstring) {
1157: const char *type;
1158: PetscCall(MatGetType(mat, &type));
1159: PetscCall(PetscViewerStringSPrintf(viewer, " MatType: %-7.7s", type));
1160: PetscTryTypeMethod(mat, view, viewer);
1161: }
1162: if ((format == PETSC_VIEWER_NATIVE || format == PETSC_VIEWER_LOAD_BALANCE) && mat->ops->viewnative) {
1163: PetscCall(PetscViewerASCIIPushTab(viewer));
1164: PetscUseTypeMethod(mat, viewnative, viewer);
1165: PetscCall(PetscViewerASCIIPopTab(viewer));
1166: } else if (mat->ops->view) {
1167: PetscCall(PetscViewerASCIIPushTab(viewer));
1168: PetscUseTypeMethod(mat, view, viewer);
1169: PetscCall(PetscViewerASCIIPopTab(viewer));
1170: }
1171: if (isascii) {
1172: PetscCall(PetscViewerGetFormat(viewer, &format));
1173: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) PetscCall(PetscViewerASCIIPopTab(viewer));
1174: }
1175: PetscCall(PetscLogEventEnd(MAT_View, mat, viewer, 0, 0));
1176: #if !defined(PETSC_HAVE_THREADSAFETY)
1177: insidematview--;
1178: #endif
1179: PetscFunctionReturn(PETSC_SUCCESS);
1180: }
1182: #if defined(PETSC_USE_DEBUG)
1183: #include <../src/sys/totalview/tv_data_display.h>
1184: PETSC_UNUSED static int TV_display_type(const struct _p_Mat *mat)
1185: {
1186: TV_add_row("Local rows", "int", &mat->rmap->n);
1187: TV_add_row("Local columns", "int", &mat->cmap->n);
1188: TV_add_row("Global rows", "int", &mat->rmap->N);
1189: TV_add_row("Global columns", "int", &mat->cmap->N);
1190: TV_add_row("Typename", TV_ascii_string_type, ((PetscObject)mat)->type_name);
1191: return TV_format_OK;
1192: }
1193: #endif
1195: /*@C
1196: MatLoad - Loads a matrix that has been stored in binary/HDF5 format
1197: with `MatView()`. The matrix format is determined from the options database.
1198: Generates a parallel MPI matrix if the communicator has more than one
1199: processor. The default matrix type is `MATAIJ`.
1201: Collective
1203: Input Parameters:
1204: + mat - the newly loaded matrix, this needs to have been created with `MatCreate()`
1205: or some related function before a call to `MatLoad()`
1206: - viewer - `PETSCVIEWERBINARY`/`PETSCVIEWERHDF5` file viewer
1208: Options Database Keys:
1209: Used with block matrix formats (`MATSEQBAIJ`, ...) to specify
1210: block size
1211: . -matload_block_size <bs> - set block size
1213: Level: beginner
1215: Notes:
1216: If the `Mat` type has not yet been given then `MATAIJ` is used, call `MatSetFromOptions()` on the
1217: `Mat` before calling this routine if you wish to set it from the options database.
1219: `MatLoad()` automatically loads into the options database any options
1220: given in the file filename.info where filename is the name of the file
1221: that was passed to the `PetscViewerBinaryOpen()`. The options in the info
1222: file will be ignored if you use the -viewer_binary_skip_info option.
1224: If the type or size of mat is not set before a call to `MatLoad()`, PETSc
1225: sets the default matrix type AIJ and sets the local and global sizes.
1226: If type and/or size is already set, then the same are used.
1228: In parallel, each processor can load a subset of rows (or the
1229: entire matrix). This routine is especially useful when a large
1230: matrix is stored on disk and only part of it is desired on each
1231: processor. For example, a parallel solver may access only some of
1232: the rows from each processor. The algorithm used here reads
1233: relatively small blocks of data rather than reading the entire
1234: matrix and then subsetting it.
1236: Viewer's `PetscViewerType` must be either `PETSCVIEWERBINARY` or `PETSCVIEWERHDF5`.
1237: Such viewer can be created using `PetscViewerBinaryOpen()` or `PetscViewerHDF5Open()`,
1238: or the sequence like
1239: .vb
1240: `PetscViewer` v;
1241: `PetscViewerCreate`(`PETSC_COMM_WORLD`,&v);
1242: `PetscViewerSetType`(v,`PETSCVIEWERBINARY`);
1243: `PetscViewerSetFromOptions`(v);
1244: `PetscViewerFileSetMode`(v,`FILE_MODE_READ`);
1245: `PetscViewerFileSetName`(v,"datafile");
1246: .ve
1247: The optional `PetscViewerSetFromOptions()` call allows overriding `PetscViewerSetType()` using the option
1248: $ -viewer_type {binary, hdf5}
1250: See the example src/ksp/ksp/tutorials/ex27.c with the first approach,
1251: and src/mat/tutorials/ex10.c with the second approach.
1253: In case of `PETSCVIEWERBINARY`, a native PETSc binary format is used. Each of the blocks
1254: is read onto rank 0 and then shipped to its destination rank, one after another.
1255: Multiple objects, both matrices and vectors, can be stored within the same file.
1256: Their PetscObject name is ignored; they are loaded in the order of their storage.
1258: Most users should not need to know the details of the binary storage
1259: format, since `MatLoad()` and `MatView()` completely hide these details.
1260: But for anyone who's interested, the standard binary matrix storage
1261: format is
1263: .vb
1264: PetscInt MAT_FILE_CLASSID
1265: PetscInt number of rows
1266: PetscInt number of columns
1267: PetscInt total number of nonzeros
1268: PetscInt *number nonzeros in each row
1269: PetscInt *column indices of all nonzeros (starting index is zero)
1270: PetscScalar *values of all nonzeros
1271: .ve
1273: PETSc automatically does the byte swapping for
1274: machines that store the bytes reversed. Thus if you write your own binary
1275: read/write routines you have to swap the bytes; see `PetscBinaryRead()`
1276: and `PetscBinaryWrite()` to see how this may be done.
1278: In case of `PETSCVIEWERHDF5`, a parallel HDF5 reader is used.
1279: Each processor's chunk is loaded independently by its owning rank.
1280: Multiple objects, both matrices and vectors, can be stored within the same file.
1281: They are looked up by their PetscObject name.
1283: As the MATLAB MAT-File Version 7.3 format is also a HDF5 flavor, we decided to use
1284: by default the same structure and naming of the AIJ arrays and column count
1285: within the HDF5 file. This means that a MAT file saved with -v7.3 flag, e.g.
1286: $ save example.mat A b -v7.3
1287: can be directly read by this routine (see Reference 1 for details).
1289: Depending on your MATLAB version, this format might be a default,
1290: otherwise you can set it as default in Preferences.
1292: Unless -nocompression flag is used to save the file in MATLAB,
1293: PETSc must be configured with ZLIB package.
1295: See also examples src/mat/tutorials/ex10.c and src/ksp/ksp/tutorials/ex27.c
1297: This reader currently supports only real `MATSEQAIJ`, `MATMPIAIJ`, `MATSEQDENSE` and `MATMPIDENSE` matrices for `PETSCVIEWERHDF5`
1299: Corresponding `MatView()` is not yet implemented.
1301: The loaded matrix is actually a transpose of the original one in MATLAB,
1302: unless you push `PETSC_VIEWER_HDF5_MAT` format (see examples above).
1303: With this format, matrix is automatically transposed by PETSc,
1304: unless the matrix is marked as SPD or symmetric
1305: (see `MatSetOption()`, `MAT_SPD`, `MAT_SYMMETRIC`).
1307: References:
1308: . * - MATLAB(R) Documentation, manual page of save(), https://www.mathworks.com/help/matlab/ref/save.html#btox10b-1-version
1310: .seealso: [](ch_matrices), `Mat`, `PetscViewerBinaryOpen()`, `PetscViewerSetType()`, `MatView()`, `VecLoad()`
1311: @*/
1312: PetscErrorCode MatLoad(Mat mat, PetscViewer viewer)
1313: {
1314: PetscBool flg;
1316: PetscFunctionBegin;
1320: if (!((PetscObject)mat)->type_name) PetscCall(MatSetType(mat, MATAIJ));
1322: flg = PETSC_FALSE;
1323: PetscCall(PetscOptionsGetBool(((PetscObject)mat)->options, ((PetscObject)mat)->prefix, "-matload_symmetric", &flg, NULL));
1324: if (flg) {
1325: PetscCall(MatSetOption(mat, MAT_SYMMETRIC, PETSC_TRUE));
1326: PetscCall(MatSetOption(mat, MAT_SYMMETRY_ETERNAL, PETSC_TRUE));
1327: }
1328: flg = PETSC_FALSE;
1329: PetscCall(PetscOptionsGetBool(((PetscObject)mat)->options, ((PetscObject)mat)->prefix, "-matload_spd", &flg, NULL));
1330: if (flg) PetscCall(MatSetOption(mat, MAT_SPD, PETSC_TRUE));
1332: PetscCall(PetscLogEventBegin(MAT_Load, mat, viewer, 0, 0));
1333: PetscUseTypeMethod(mat, load, viewer);
1334: PetscCall(PetscLogEventEnd(MAT_Load, mat, viewer, 0, 0));
1335: PetscFunctionReturn(PETSC_SUCCESS);
1336: }
1338: static PetscErrorCode MatDestroy_Redundant(Mat_Redundant **redundant)
1339: {
1340: Mat_Redundant *redund = *redundant;
1342: PetscFunctionBegin;
1343: if (redund) {
1344: if (redund->matseq) { /* via MatCreateSubMatrices() */
1345: PetscCall(ISDestroy(&redund->isrow));
1346: PetscCall(ISDestroy(&redund->iscol));
1347: PetscCall(MatDestroySubMatrices(1, &redund->matseq));
1348: } else {
1349: PetscCall(PetscFree2(redund->send_rank, redund->recv_rank));
1350: PetscCall(PetscFree(redund->sbuf_j));
1351: PetscCall(PetscFree(redund->sbuf_a));
1352: for (PetscInt i = 0; i < redund->nrecvs; i++) {
1353: PetscCall(PetscFree(redund->rbuf_j[i]));
1354: PetscCall(PetscFree(redund->rbuf_a[i]));
1355: }
1356: PetscCall(PetscFree4(redund->sbuf_nz, redund->rbuf_nz, redund->rbuf_j, redund->rbuf_a));
1357: }
1359: if (redund->subcomm) PetscCall(PetscCommDestroy(&redund->subcomm));
1360: PetscCall(PetscFree(redund));
1361: }
1362: PetscFunctionReturn(PETSC_SUCCESS);
1363: }
1365: /*@C
1366: MatDestroy - Frees space taken by a matrix.
1368: Collective
1370: Input Parameter:
1371: . A - the matrix
1373: Level: beginner
1375: Developer Note:
1376: Some special arrays of matrices are not destroyed in this routine but instead by the routines called by
1377: `MatDestroySubMatrices()`. Thus one must be sure that any changes here must also be made in those routines.
1378: `MatHeaderMerge()` and `MatHeaderReplace()` also manipulate the data in the `Mat` object and likely need changes
1379: if changes are needed here.
1381: .seealso: [](ch_matrices), `Mat`, `MatCreate()`
1382: @*/
1383: PetscErrorCode MatDestroy(Mat *A)
1384: {
1385: PetscFunctionBegin;
1386: if (!*A) PetscFunctionReturn(PETSC_SUCCESS);
1388: if (--((PetscObject)(*A))->refct > 0) {
1389: *A = NULL;
1390: PetscFunctionReturn(PETSC_SUCCESS);
1391: }
1393: /* if memory was published with SAWs then destroy it */
1394: PetscCall(PetscObjectSAWsViewOff((PetscObject)*A));
1395: PetscTryTypeMethod((*A), destroy);
1397: PetscCall(PetscFree((*A)->factorprefix));
1398: PetscCall(PetscFree((*A)->defaultvectype));
1399: PetscCall(PetscFree((*A)->defaultrandtype));
1400: PetscCall(PetscFree((*A)->bsizes));
1401: PetscCall(PetscFree((*A)->solvertype));
1402: for (PetscInt i = 0; i < MAT_FACTOR_NUM_TYPES; i++) PetscCall(PetscFree((*A)->preferredordering[i]));
1403: if ((*A)->redundant && (*A)->redundant->matseq[0] == *A) (*A)->redundant->matseq[0] = NULL;
1404: PetscCall(MatDestroy_Redundant(&(*A)->redundant));
1405: PetscCall(MatProductClear(*A));
1406: PetscCall(MatNullSpaceDestroy(&(*A)->nullsp));
1407: PetscCall(MatNullSpaceDestroy(&(*A)->transnullsp));
1408: PetscCall(MatNullSpaceDestroy(&(*A)->nearnullsp));
1409: PetscCall(MatDestroy(&(*A)->schur));
1410: PetscCall(PetscLayoutDestroy(&(*A)->rmap));
1411: PetscCall(PetscLayoutDestroy(&(*A)->cmap));
1412: PetscCall(PetscHeaderDestroy(A));
1413: PetscFunctionReturn(PETSC_SUCCESS);
1414: }
1416: /*@C
1417: MatSetValues - Inserts or adds a block of values into a matrix.
1418: These values may be cached, so `MatAssemblyBegin()` and `MatAssemblyEnd()`
1419: MUST be called after all calls to `MatSetValues()` have been completed.
1421: Not Collective
1423: Input Parameters:
1424: + mat - the matrix
1425: . v - a logically two-dimensional array of values
1426: . m - the number of rows
1427: . idxm - the global indices of the rows
1428: . n - the number of columns
1429: . idxn - the global indices of the columns
1430: - addv - either `ADD_VALUES` to add values to any existing entries, or `INSERT_VALUES` to replace existing entries with new values
1432: Level: beginner
1434: Notes:
1435: By default the values, `v`, are stored row-oriented. See `MatSetOption()` for other options.
1437: Calls to `MatSetValues()` with the `INSERT_VALUES` and `ADD_VALUES`
1438: options cannot be mixed without intervening calls to the assembly
1439: routines.
1441: `MatSetValues()` uses 0-based row and column numbers in Fortran
1442: as well as in C.
1444: Negative indices may be passed in `idxm` and `idxn`, these rows and columns are
1445: simply ignored. This allows easily inserting element stiffness matrices
1446: with homogeneous Dirchlet boundary conditions that you don't want represented
1447: in the matrix.
1449: Efficiency Alert:
1450: The routine `MatSetValuesBlocked()` may offer much better efficiency
1451: for users of block sparse formats (`MATSEQBAIJ` and `MATMPIBAIJ`).
1453: Developer Note:
1454: This is labeled with C so does not automatically generate Fortran stubs and interfaces
1455: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
1457: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`,
1458: `InsertMode`, `INSERT_VALUES`, `ADD_VALUES`
1459: @*/
1460: PetscErrorCode MatSetValues(Mat mat, PetscInt m, const PetscInt idxm[], PetscInt n, const PetscInt idxn[], const PetscScalar v[], InsertMode addv)
1461: {
1462: PetscFunctionBeginHot;
1465: if (!m || !n) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
1468: MatCheckPreallocated(mat, 1);
1470: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
1471: else PetscCheck(mat->insertmode == addv, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add values and insert values");
1473: if (PetscDefined(USE_DEBUG)) {
1474: PetscInt i, j;
1476: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
1477: for (i = 0; i < m; i++) {
1478: for (j = 0; j < n; j++) {
1479: if (mat->erroriffailure && PetscIsInfOrNanScalar(v[i * n + j]))
1480: #if defined(PETSC_USE_COMPLEX)
1481: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_FP, "Inserting %g+i%g at matrix entry (%" PetscInt_FMT ",%" PetscInt_FMT ")", (double)PetscRealPart(v[i * n + j]), (double)PetscImaginaryPart(v[i * n + j]), idxm[i], idxn[j]);
1482: #else
1483: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_FP, "Inserting %g at matrix entry (%" PetscInt_FMT ",%" PetscInt_FMT ")", (double)v[i * n + j], idxm[i], idxn[j]);
1484: #endif
1485: }
1486: }
1487: for (i = 0; i < m; i++) PetscCheck(idxm[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Cannot insert in row %" PetscInt_FMT ", maximum is %" PetscInt_FMT, idxm[i], mat->rmap->N - 1);
1488: for (i = 0; i < n; i++) PetscCheck(idxn[i] < mat->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Cannot insert in column %" PetscInt_FMT ", maximum is %" PetscInt_FMT, idxn[i], mat->cmap->N - 1);
1489: }
1491: if (mat->assembled) {
1492: mat->was_assembled = PETSC_TRUE;
1493: mat->assembled = PETSC_FALSE;
1494: }
1495: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
1496: PetscUseTypeMethod(mat, setvalues, m, idxm, n, idxn, v, addv);
1497: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
1498: PetscFunctionReturn(PETSC_SUCCESS);
1499: }
1501: /*@C
1502: MatSetValuesIS - Inserts or adds a block of values into a matrix using an `IS` to indicate the rows and columns
1503: These values may be cached, so `MatAssemblyBegin()` and `MatAssemblyEnd()`
1504: MUST be called after all calls to `MatSetValues()` have been completed.
1506: Not Collective
1508: Input Parameters:
1509: + mat - the matrix
1510: . v - a logically two-dimensional array of values
1511: . ism - the rows to provide
1512: . isn - the columns to provide
1513: - addv - either `ADD_VALUES` to add values to any existing entries, or `INSERT_VALUES` to replace existing entries with new values
1515: Level: beginner
1517: Notes:
1518: By default the values, `v`, are stored row-oriented. See `MatSetOption()` for other options.
1520: Calls to `MatSetValues()` with the `INSERT_VALUES` and `ADD_VALUES`
1521: options cannot be mixed without intervening calls to the assembly
1522: routines.
1524: `MatSetValues()` uses 0-based row and column numbers in Fortran
1525: as well as in C.
1527: Negative indices may be passed in `ism` and `isn`, these rows and columns are
1528: simply ignored. This allows easily inserting element stiffness matrices
1529: with homogeneous Dirchlet boundary conditions that you don't want represented
1530: in the matrix.
1532: Efficiency Alert:
1533: The routine `MatSetValuesBlocked()` may offer much better efficiency
1534: for users of block sparse formats (`MATSEQBAIJ` and `MATMPIBAIJ`).
1536: This is currently not optimized for any particular `ISType`
1538: Developer Notes:
1539: This is labeled with C so does not automatically generate Fortran stubs and interfaces
1540: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
1542: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MatSetValues()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`,
1543: `InsertMode`, `INSERT_VALUES`, `ADD_VALUES`, `MatSetValues()`
1544: @*/
1545: PetscErrorCode MatSetValuesIS(Mat mat, IS ism, IS isn, const PetscScalar v[], InsertMode addv)
1546: {
1547: PetscInt m, n;
1548: const PetscInt *rows, *cols;
1550: PetscFunctionBeginHot;
1552: PetscCall(ISGetIndices(ism, &rows));
1553: PetscCall(ISGetIndices(isn, &cols));
1554: PetscCall(ISGetLocalSize(ism, &m));
1555: PetscCall(ISGetLocalSize(isn, &n));
1556: PetscCall(MatSetValues(mat, m, rows, n, cols, v, addv));
1557: PetscCall(ISRestoreIndices(ism, &rows));
1558: PetscCall(ISRestoreIndices(isn, &cols));
1559: PetscFunctionReturn(PETSC_SUCCESS);
1560: }
1562: /*@
1563: MatSetValuesRowLocal - Inserts a row (block row for `MATBAIJ` matrices) of nonzero
1564: values into a matrix
1566: Not Collective
1568: Input Parameters:
1569: + mat - the matrix
1570: . row - the (block) row to set
1571: - v - a logically two-dimensional array of values
1573: Level: intermediate
1575: Notes:
1576: The values, `v`, are column-oriented (for the block version) and sorted
1578: All the nonzeros in the row must be provided
1580: The matrix must have previously had its column indices set, likely by having been assembled.
1582: The row must belong to this process
1584: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`,
1585: `InsertMode`, `INSERT_VALUES`, `ADD_VALUES`, `MatSetValues()`, `MatSetValuesRow()`, `MatSetLocalToGlobalMapping()`
1586: @*/
1587: PetscErrorCode MatSetValuesRowLocal(Mat mat, PetscInt row, const PetscScalar v[])
1588: {
1589: PetscInt globalrow;
1591: PetscFunctionBegin;
1595: PetscCall(ISLocalToGlobalMappingApply(mat->rmap->mapping, 1, &row, &globalrow));
1596: PetscCall(MatSetValuesRow(mat, globalrow, v));
1597: PetscFunctionReturn(PETSC_SUCCESS);
1598: }
1600: /*@
1601: MatSetValuesRow - Inserts a row (block row for `MATBAIJ` matrices) of nonzero
1602: values into a matrix
1604: Not Collective
1606: Input Parameters:
1607: + mat - the matrix
1608: . row - the (block) row to set
1609: - v - a logically two-dimensional (column major) array of values for block matrices with blocksize larger than one, otherwise a one dimensional array of values
1611: Level: advanced
1613: Notes:
1614: The values, `v`, are column-oriented for the block version.
1616: All the nonzeros in the row must be provided
1618: THE MATRIX MUST HAVE PREVIOUSLY HAD ITS COLUMN INDICES SET. IT IS RARE THAT THIS ROUTINE IS USED, usually `MatSetValues()` is used.
1620: The row must belong to this process
1622: .seealso: [](ch_matrices), `Mat`, `MatSetValues()`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`,
1623: `InsertMode`, `INSERT_VALUES`, `ADD_VALUES`, `MatSetValues()`
1624: @*/
1625: PetscErrorCode MatSetValuesRow(Mat mat, PetscInt row, const PetscScalar v[])
1626: {
1627: PetscFunctionBeginHot;
1630: MatCheckPreallocated(mat, 1);
1632: PetscCheck(mat->insertmode != ADD_VALUES, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add and insert values");
1633: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
1634: mat->insertmode = INSERT_VALUES;
1636: if (mat->assembled) {
1637: mat->was_assembled = PETSC_TRUE;
1638: mat->assembled = PETSC_FALSE;
1639: }
1640: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
1641: PetscUseTypeMethod(mat, setvaluesrow, row, v);
1642: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
1643: PetscFunctionReturn(PETSC_SUCCESS);
1644: }
1646: /*@
1647: MatSetValuesStencil - Inserts or adds a block of values into a matrix.
1648: Using structured grid indexing
1650: Not Collective
1652: Input Parameters:
1653: + mat - the matrix
1654: . m - number of rows being entered
1655: . idxm - grid coordinates (and component number when dof > 1) for matrix rows being entered
1656: . n - number of columns being entered
1657: . idxn - grid coordinates (and component number when dof > 1) for matrix columns being entered
1658: . v - a logically two-dimensional array of values
1659: - addv - either `ADD_VALUES` to add to existing entries at that location or `INSERT_VALUES` to replace existing entries with new values
1661: Level: beginner
1663: Notes:
1664: By default the values, `v`, are row-oriented. See `MatSetOption()` for other options.
1666: Calls to `MatSetValuesStencil()` with the `INSERT_VALUES` and `ADD_VALUES`
1667: options cannot be mixed without intervening calls to the assembly
1668: routines.
1670: The grid coordinates are across the entire grid, not just the local portion
1672: `MatSetValuesStencil()` uses 0-based row and column numbers in Fortran
1673: as well as in C.
1675: For setting/accessing vector values via array coordinates you can use the `DMDAVecGetArray()` routine
1677: In order to use this routine you must either obtain the matrix with `DMCreateMatrix()`
1678: or call `MatSetLocalToGlobalMapping()` and `MatSetStencil()` first.
1680: The columns and rows in the stencil passed in MUST be contained within the
1681: ghost region of the given process as set with DMDACreateXXX() or `MatSetStencil()`. For example,
1682: if you create a `DMDA` with an overlap of one grid level and on a particular process its first
1683: local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1684: first i index you can use in your column and row indices in `MatSetStencil()` is 5.
1686: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
1687: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
1688: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
1689: `DM_BOUNDARY_PERIODIC` boundary type.
1691: For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have
1692: a single value per point) you can skip filling those indices.
1694: Inspired by the structured grid interface to the HYPRE package
1695: (https://computation.llnl.gov/projects/hypre-scalable-linear-solvers-multigrid-methods)
1697: Efficiency Alert:
1698: The routine `MatSetValuesBlockedStencil()` may offer much better efficiency
1699: for users of block sparse formats (`MATSEQBAIJ` and `MATMPIBAIJ`).
1701: Fortran Note:
1702: `idxm` and `idxn` should be declared as
1703: $ MatStencil idxm(4,m),idxn(4,n)
1704: and the values inserted using
1705: .vb
1706: idxm(MatStencil_i,1) = i
1707: idxm(MatStencil_j,1) = j
1708: idxm(MatStencil_k,1) = k
1709: idxm(MatStencil_c,1) = c
1710: etc
1711: .ve
1713: .seealso: [](ch_matrices), `Mat`, `DMDA`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`
1714: `MatSetValues()`, `MatSetValuesBlockedStencil()`, `MatSetStencil()`, `DMCreateMatrix()`, `DMDAVecGetArray()`, `MatStencil`
1715: @*/
1716: PetscErrorCode MatSetValuesStencil(Mat mat, PetscInt m, const MatStencil idxm[], PetscInt n, const MatStencil idxn[], const PetscScalar v[], InsertMode addv)
1717: {
1718: PetscInt buf[8192], *bufm = NULL, *bufn = NULL, *jdxm, *jdxn;
1719: PetscInt j, i, dim = mat->stencil.dim, *dims = mat->stencil.dims + 1, tmp;
1720: PetscInt *starts = mat->stencil.starts, *dxm = (PetscInt *)idxm, *dxn = (PetscInt *)idxn, sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1722: PetscFunctionBegin;
1723: if (!m || !n) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
1729: if ((m + n) <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
1730: jdxm = buf;
1731: jdxn = buf + m;
1732: } else {
1733: PetscCall(PetscMalloc2(m, &bufm, n, &bufn));
1734: jdxm = bufm;
1735: jdxn = bufn;
1736: }
1737: for (i = 0; i < m; i++) {
1738: for (j = 0; j < 3 - sdim; j++) dxm++;
1739: tmp = *dxm++ - starts[0];
1740: for (j = 0; j < dim - 1; j++) {
1741: if ((*dxm++ - starts[j + 1]) < 0 || tmp < 0) tmp = -1;
1742: else tmp = tmp * dims[j] + *(dxm - 1) - starts[j + 1];
1743: }
1744: if (mat->stencil.noc) dxm++;
1745: jdxm[i] = tmp;
1746: }
1747: for (i = 0; i < n; i++) {
1748: for (j = 0; j < 3 - sdim; j++) dxn++;
1749: tmp = *dxn++ - starts[0];
1750: for (j = 0; j < dim - 1; j++) {
1751: if ((*dxn++ - starts[j + 1]) < 0 || tmp < 0) tmp = -1;
1752: else tmp = tmp * dims[j] + *(dxn - 1) - starts[j + 1];
1753: }
1754: if (mat->stencil.noc) dxn++;
1755: jdxn[i] = tmp;
1756: }
1757: PetscCall(MatSetValuesLocal(mat, m, jdxm, n, jdxn, v, addv));
1758: PetscCall(PetscFree2(bufm, bufn));
1759: PetscFunctionReturn(PETSC_SUCCESS);
1760: }
1762: /*@
1763: MatSetValuesBlockedStencil - Inserts or adds a block of values into a matrix.
1764: Using structured grid indexing
1766: Not Collective
1768: Input Parameters:
1769: + mat - the matrix
1770: . m - number of rows being entered
1771: . idxm - grid coordinates for matrix rows being entered
1772: . n - number of columns being entered
1773: . idxn - grid coordinates for matrix columns being entered
1774: . v - a logically two-dimensional array of values
1775: - addv - either `ADD_VALUES` to add to existing entries or `INSERT_VALUES` to replace existing entries with new values
1777: Level: beginner
1779: Notes:
1780: By default the values, `v`, are row-oriented and unsorted.
1781: See `MatSetOption()` for other options.
1783: Calls to `MatSetValuesBlockedStencil()` with the `INSERT_VALUES` and `ADD_VALUES`
1784: options cannot be mixed without intervening calls to the assembly
1785: routines.
1787: The grid coordinates are across the entire grid, not just the local portion
1789: `MatSetValuesBlockedStencil()` uses 0-based row and column numbers in Fortran
1790: as well as in C.
1792: For setting/accessing vector values via array coordinates you can use the `DMDAVecGetArray()` routine
1794: In order to use this routine you must either obtain the matrix with `DMCreateMatrix()`
1795: or call `MatSetBlockSize()`, `MatSetLocalToGlobalMapping()` and `MatSetStencil()` first.
1797: The columns and rows in the stencil passed in MUST be contained within the
1798: ghost region of the given process as set with DMDACreateXXX() or `MatSetStencil()`. For example,
1799: if you create a `DMDA` with an overlap of one grid level and on a particular process its first
1800: local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1801: first i index you can use in your column and row indices in `MatSetStencil()` is 5.
1803: Negative indices may be passed in idxm and idxn, these rows and columns are
1804: simply ignored. This allows easily inserting element stiffness matrices
1805: with homogeneous Dirchlet boundary conditions that you don't want represented
1806: in the matrix.
1808: Inspired by the structured grid interface to the HYPRE package
1809: (https://computation.llnl.gov/projects/hypre-scalable-linear-solvers-multigrid-methods)
1811: Fortran Note:
1812: `idxm` and `idxn` should be declared as
1813: $ MatStencil idxm(4,m),idxn(4,n)
1814: and the values inserted using
1815: .vb
1816: idxm(MatStencil_i,1) = i
1817: idxm(MatStencil_j,1) = j
1818: idxm(MatStencil_k,1) = k
1819: etc
1820: .ve
1822: .seealso: [](ch_matrices), `Mat`, `DMDA`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`
1823: `MatSetValues()`, `MatSetValuesStencil()`, `MatSetStencil()`, `DMCreateMatrix()`, `DMDAVecGetArray()`, `MatStencil`,
1824: `MatSetBlockSize()`, `MatSetLocalToGlobalMapping()`
1825: @*/
1826: PetscErrorCode MatSetValuesBlockedStencil(Mat mat, PetscInt m, const MatStencil idxm[], PetscInt n, const MatStencil idxn[], const PetscScalar v[], InsertMode addv)
1827: {
1828: PetscInt buf[8192], *bufm = NULL, *bufn = NULL, *jdxm, *jdxn;
1829: PetscInt j, i, dim = mat->stencil.dim, *dims = mat->stencil.dims + 1, tmp;
1830: PetscInt *starts = mat->stencil.starts, *dxm = (PetscInt *)idxm, *dxn = (PetscInt *)idxn, sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1832: PetscFunctionBegin;
1833: if (!m || !n) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
1840: if ((m + n) <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
1841: jdxm = buf;
1842: jdxn = buf + m;
1843: } else {
1844: PetscCall(PetscMalloc2(m, &bufm, n, &bufn));
1845: jdxm = bufm;
1846: jdxn = bufn;
1847: }
1848: for (i = 0; i < m; i++) {
1849: for (j = 0; j < 3 - sdim; j++) dxm++;
1850: tmp = *dxm++ - starts[0];
1851: for (j = 0; j < sdim - 1; j++) {
1852: if ((*dxm++ - starts[j + 1]) < 0 || tmp < 0) tmp = -1;
1853: else tmp = tmp * dims[j] + *(dxm - 1) - starts[j + 1];
1854: }
1855: dxm++;
1856: jdxm[i] = tmp;
1857: }
1858: for (i = 0; i < n; i++) {
1859: for (j = 0; j < 3 - sdim; j++) dxn++;
1860: tmp = *dxn++ - starts[0];
1861: for (j = 0; j < sdim - 1; j++) {
1862: if ((*dxn++ - starts[j + 1]) < 0 || tmp < 0) tmp = -1;
1863: else tmp = tmp * dims[j] + *(dxn - 1) - starts[j + 1];
1864: }
1865: dxn++;
1866: jdxn[i] = tmp;
1867: }
1868: PetscCall(MatSetValuesBlockedLocal(mat, m, jdxm, n, jdxn, v, addv));
1869: PetscCall(PetscFree2(bufm, bufn));
1870: PetscFunctionReturn(PETSC_SUCCESS);
1871: }
1873: /*@
1874: MatSetStencil - Sets the grid information for setting values into a matrix via
1875: `MatSetValuesStencil()`
1877: Not Collective
1879: Input Parameters:
1880: + mat - the matrix
1881: . dim - dimension of the grid 1, 2, or 3
1882: . dims - number of grid points in x, y, and z direction, including ghost points on your processor
1883: . starts - starting point of ghost nodes on your processor in x, y, and z direction
1884: - dof - number of degrees of freedom per node
1886: Level: beginner
1888: Notes:
1889: Inspired by the structured grid interface to the HYPRE package
1890: (www.llnl.gov/CASC/hyper)
1892: For matrices generated with `DMCreateMatrix()` this routine is automatically called and so not needed by the
1893: user.
1895: .seealso: [](ch_matrices), `Mat`, `MatStencil`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`
1896: `MatSetValues()`, `MatSetValuesBlockedStencil()`, `MatSetValuesStencil()`
1897: @*/
1898: PetscErrorCode MatSetStencil(Mat mat, PetscInt dim, const PetscInt dims[], const PetscInt starts[], PetscInt dof)
1899: {
1900: PetscFunctionBegin;
1905: mat->stencil.dim = dim + (dof > 1);
1906: for (PetscInt i = 0; i < dim; i++) {
1907: mat->stencil.dims[i] = dims[dim - i - 1]; /* copy the values in backwards */
1908: mat->stencil.starts[i] = starts[dim - i - 1];
1909: }
1910: mat->stencil.dims[dim] = dof;
1911: mat->stencil.starts[dim] = 0;
1912: mat->stencil.noc = (PetscBool)(dof == 1);
1913: PetscFunctionReturn(PETSC_SUCCESS);
1914: }
1916: /*@C
1917: MatSetValuesBlocked - Inserts or adds a block of values into a matrix.
1919: Not Collective
1921: Input Parameters:
1922: + mat - the matrix
1923: . v - a logically two-dimensional array of values
1924: . m - the number of block rows
1925: . idxm - the global block indices
1926: . n - the number of block columns
1927: . idxn - the global block indices
1928: - addv - either `ADD_VALUES` to add values to any existing entries, or `INSERT_VALUES` replaces existing entries with new values
1930: Level: intermediate
1932: Notes:
1933: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call
1934: MatXXXXSetPreallocation() or `MatSetUp()` before using this routine.
1936: The `m` and `n` count the NUMBER of blocks in the row direction and column direction,
1937: NOT the total number of rows/columns; for example, if the block size is 2 and
1938: you are passing in values for rows 2,3,4,5 then m would be 2 (not 4).
1939: The values in idxm would be 1 2; that is the first index for each block divided by
1940: the block size.
1942: You must call `MatSetBlockSize()` when constructing this matrix (before
1943: preallocating it).
1945: By default the values, `v`, are row-oriented, so the layout of
1946: `v` is the same as for `MatSetValues()`. See `MatSetOption()` for other options.
1948: Calls to `MatSetValuesBlocked()` with the `INSERT_VALUES` and `ADD_VALUES`
1949: options cannot be mixed without intervening calls to the assembly
1950: routines.
1952: `MatSetValuesBlocked()` uses 0-based row and column numbers in Fortran
1953: as well as in C.
1955: Negative indices may be passed in `idxm` and `idxn`, these rows and columns are
1956: simply ignored. This allows easily inserting element stiffness matrices
1957: with homogeneous Dirchlet boundary conditions that you don't want represented
1958: in the matrix.
1960: Each time an entry is set within a sparse matrix via `MatSetValues()`,
1961: internal searching must be done to determine where to place the
1962: data in the matrix storage space. By instead inserting blocks of
1963: entries via `MatSetValuesBlocked()`, the overhead of matrix assembly is
1964: reduced.
1966: Example:
1967: .vb
1968: Suppose m=n=2 and block size(bs) = 2 The array is
1970: 1 2 | 3 4
1971: 5 6 | 7 8
1972: - - - | - - -
1973: 9 10 | 11 12
1974: 13 14 | 15 16
1976: v[] should be passed in like
1977: v[] = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]
1979: If you are not using row oriented storage of v (that is you called MatSetOption(mat,MAT_ROW_ORIENTED,PETSC_FALSE)) then
1980: v[] = [1,5,9,13,2,6,10,14,3,7,11,15,4,8,12,16]
1981: .ve
1983: .seealso: [](ch_matrices), `Mat`, `MatSetBlockSize()`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValues()`, `MatSetValuesBlockedLocal()`
1984: @*/
1985: PetscErrorCode MatSetValuesBlocked(Mat mat, PetscInt m, const PetscInt idxm[], PetscInt n, const PetscInt idxn[], const PetscScalar v[], InsertMode addv)
1986: {
1987: PetscFunctionBeginHot;
1990: if (!m || !n) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
1993: MatCheckPreallocated(mat, 1);
1994: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
1995: else PetscCheck(mat->insertmode == addv, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add values and insert values");
1996: if (PetscDefined(USE_DEBUG)) {
1997: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
1998: PetscCheck(mat->ops->setvaluesblocked || mat->ops->setvalues, PETSC_COMM_SELF, PETSC_ERR_SUP, "Mat type %s", ((PetscObject)mat)->type_name);
1999: }
2000: if (PetscDefined(USE_DEBUG)) {
2001: PetscInt rbs, cbs, M, N, i;
2002: PetscCall(MatGetBlockSizes(mat, &rbs, &cbs));
2003: PetscCall(MatGetSize(mat, &M, &N));
2004: for (i = 0; i < m; i++) PetscCheck(idxm[i] * rbs < M, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row block index %" PetscInt_FMT " (index %" PetscInt_FMT ") greater than row length %" PetscInt_FMT, i, idxm[i], M);
2005: for (i = 0; i < n; i++) PetscCheck(idxn[i] * cbs < N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column block index %" PetscInt_FMT " (index %" PetscInt_FMT ") great than column length %" PetscInt_FMT, i, idxn[i], N);
2006: }
2007: if (mat->assembled) {
2008: mat->was_assembled = PETSC_TRUE;
2009: mat->assembled = PETSC_FALSE;
2010: }
2011: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
2012: if (mat->ops->setvaluesblocked) {
2013: PetscUseTypeMethod(mat, setvaluesblocked, m, idxm, n, idxn, v, addv);
2014: } else {
2015: PetscInt buf[8192], *bufr = NULL, *bufc = NULL, *iidxm, *iidxn;
2016: PetscInt i, j, bs, cbs;
2018: PetscCall(MatGetBlockSizes(mat, &bs, &cbs));
2019: if (m * bs + n * cbs <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
2020: iidxm = buf;
2021: iidxn = buf + m * bs;
2022: } else {
2023: PetscCall(PetscMalloc2(m * bs, &bufr, n * cbs, &bufc));
2024: iidxm = bufr;
2025: iidxn = bufc;
2026: }
2027: for (i = 0; i < m; i++) {
2028: for (j = 0; j < bs; j++) iidxm[i * bs + j] = bs * idxm[i] + j;
2029: }
2030: if (m != n || bs != cbs || idxm != idxn) {
2031: for (i = 0; i < n; i++) {
2032: for (j = 0; j < cbs; j++) iidxn[i * cbs + j] = cbs * idxn[i] + j;
2033: }
2034: } else iidxn = iidxm;
2035: PetscCall(MatSetValues(mat, m * bs, iidxm, n * cbs, iidxn, v, addv));
2036: PetscCall(PetscFree2(bufr, bufc));
2037: }
2038: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
2039: PetscFunctionReturn(PETSC_SUCCESS);
2040: }
2042: /*@C
2043: MatGetValues - Gets a block of local values from a matrix.
2045: Not Collective; can only return values that are owned by the give process
2047: Input Parameters:
2048: + mat - the matrix
2049: . v - a logically two-dimensional array for storing the values
2050: . m - the number of rows
2051: . idxm - the global indices of the rows
2052: . n - the number of columns
2053: - idxn - the global indices of the columns
2055: Level: advanced
2057: Notes:
2058: The user must allocate space (m*n `PetscScalar`s) for the values, `v`.
2059: The values, `v`, are then returned in a row-oriented format,
2060: analogous to that used by default in `MatSetValues()`.
2062: `MatGetValues()` uses 0-based row and column numbers in
2063: Fortran as well as in C.
2065: `MatGetValues()` requires that the matrix has been assembled
2066: with `MatAssemblyBegin()`/`MatAssemblyEnd()`. Thus, calls to
2067: `MatSetValues()` and `MatGetValues()` CANNOT be made in succession
2068: without intermediate matrix assembly.
2070: Negative row or column indices will be ignored and those locations in `v` will be
2071: left unchanged.
2073: For the standard row-based matrix formats, `idxm` can only contain rows owned by the requesting MPI rank.
2074: That is, rows with global index greater than or equal to rstart and less than rend where rstart and rend are obtainable
2075: from `MatGetOwnershipRange`(mat,&rstart,&rend).
2077: .seealso: [](ch_matrices), `Mat`, `MatGetRow()`, `MatCreateSubMatrices()`, `MatSetValues()`, `MatGetOwnershipRange()`, `MatGetValuesLocal()`, `MatGetValue()`
2078: @*/
2079: PetscErrorCode MatGetValues(Mat mat, PetscInt m, const PetscInt idxm[], PetscInt n, const PetscInt idxn[], PetscScalar v[])
2080: {
2081: PetscFunctionBegin;
2084: if (!m || !n) PetscFunctionReturn(PETSC_SUCCESS);
2088: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2089: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2090: MatCheckPreallocated(mat, 1);
2092: PetscCall(PetscLogEventBegin(MAT_GetValues, mat, 0, 0, 0));
2093: PetscUseTypeMethod(mat, getvalues, m, idxm, n, idxn, v);
2094: PetscCall(PetscLogEventEnd(MAT_GetValues, mat, 0, 0, 0));
2095: PetscFunctionReturn(PETSC_SUCCESS);
2096: }
2098: /*@C
2099: MatGetValuesLocal - retrieves values from certain locations in a matrix using the local numbering of the indices
2100: defined previously by `MatSetLocalToGlobalMapping()`
2102: Not Collective
2104: Input Parameters:
2105: + mat - the matrix
2106: . nrow - number of rows
2107: . irow - the row local indices
2108: . ncol - number of columns
2109: - icol - the column local indices
2111: Output Parameter:
2112: . y - a logically two-dimensional array of values
2114: Level: advanced
2116: Notes:
2117: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call `MatSetLocalToGlobalMapping()` before using this routine.
2119: This routine can only return values that are owned by the requesting MPI rank. That is, for standard matrix formats, rows that, in the global numbering,
2120: are greater than or equal to rstart and less than rend where rstart and rend are obtainable from `MatGetOwnershipRange`(mat,&rstart,&rend). One can
2121: determine if the resulting global row associated with the local row r is owned by the requesting MPI rank by applying the `ISLocalToGlobalMapping` set
2122: with `MatSetLocalToGlobalMapping()`.
2124: Developer Note:
2125: This is labelled with C so does not automatically generate Fortran stubs and interfaces
2126: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
2128: .seealso: [](ch_matrices), `Mat`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValues()`, `MatSetLocalToGlobalMapping()`,
2129: `MatSetValuesLocal()`, `MatGetValues()`
2130: @*/
2131: PetscErrorCode MatGetValuesLocal(Mat mat, PetscInt nrow, const PetscInt irow[], PetscInt ncol, const PetscInt icol[], PetscScalar y[])
2132: {
2133: PetscFunctionBeginHot;
2136: MatCheckPreallocated(mat, 1);
2137: if (!nrow || !ncol) PetscFunctionReturn(PETSC_SUCCESS); /* no values to retrieve */
2140: if (PetscDefined(USE_DEBUG)) {
2141: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2142: PetscCheck(mat->ops->getvalueslocal || mat->ops->getvalues, PETSC_COMM_SELF, PETSC_ERR_SUP, "Mat type %s", ((PetscObject)mat)->type_name);
2143: }
2144: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2145: PetscCall(PetscLogEventBegin(MAT_GetValues, mat, 0, 0, 0));
2146: if (mat->ops->getvalueslocal) PetscUseTypeMethod(mat, getvalueslocal, nrow, irow, ncol, icol, y);
2147: else {
2148: PetscInt buf[8192], *bufr = NULL, *bufc = NULL, *irowm, *icolm;
2149: if ((nrow + ncol) <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
2150: irowm = buf;
2151: icolm = buf + nrow;
2152: } else {
2153: PetscCall(PetscMalloc2(nrow, &bufr, ncol, &bufc));
2154: irowm = bufr;
2155: icolm = bufc;
2156: }
2157: PetscCheck(mat->rmap->mapping, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "MatGetValuesLocal() cannot proceed without local-to-global row mapping (See MatSetLocalToGlobalMapping()).");
2158: PetscCheck(mat->cmap->mapping, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "MatGetValuesLocal() cannot proceed without local-to-global column mapping (See MatSetLocalToGlobalMapping()).");
2159: PetscCall(ISLocalToGlobalMappingApply(mat->rmap->mapping, nrow, irow, irowm));
2160: PetscCall(ISLocalToGlobalMappingApply(mat->cmap->mapping, ncol, icol, icolm));
2161: PetscCall(MatGetValues(mat, nrow, irowm, ncol, icolm, y));
2162: PetscCall(PetscFree2(bufr, bufc));
2163: }
2164: PetscCall(PetscLogEventEnd(MAT_GetValues, mat, 0, 0, 0));
2165: PetscFunctionReturn(PETSC_SUCCESS);
2166: }
2168: /*@
2169: MatSetValuesBatch - Adds (`ADD_VALUES`) many blocks of values into a matrix at once. The blocks must all be square and
2170: the same size. Currently, this can only be called once and creates the given matrix.
2172: Not Collective
2174: Input Parameters:
2175: + mat - the matrix
2176: . nb - the number of blocks
2177: . bs - the number of rows (and columns) in each block
2178: . rows - a concatenation of the rows for each block
2179: - v - a concatenation of logically two-dimensional arrays of values
2181: Level: advanced
2183: Note:
2184: `MatSetPreallocationCOO()` and `MatSetValuesCOO()` may be a better way to provide the values
2186: In the future, we will extend this routine to handle rectangular blocks, and to allow multiple calls for a given matrix.
2188: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValuesBlocked()`, `MatSetValuesLocal()`,
2189: `InsertMode`, `INSERT_VALUES`, `ADD_VALUES`, `MatSetValues()`, `MatSetPreallocationCOO()`, `MatSetValuesCOO()`
2190: @*/
2191: PetscErrorCode MatSetValuesBatch(Mat mat, PetscInt nb, PetscInt bs, PetscInt rows[], const PetscScalar v[])
2192: {
2193: PetscFunctionBegin;
2198: PetscAssert(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2200: PetscCall(PetscLogEventBegin(MAT_SetValuesBatch, mat, 0, 0, 0));
2201: if (mat->ops->setvaluesbatch) PetscUseTypeMethod(mat, setvaluesbatch, nb, bs, rows, v);
2202: else {
2203: for (PetscInt b = 0; b < nb; ++b) PetscCall(MatSetValues(mat, bs, &rows[b * bs], bs, &rows[b * bs], &v[b * bs * bs], ADD_VALUES));
2204: }
2205: PetscCall(PetscLogEventEnd(MAT_SetValuesBatch, mat, 0, 0, 0));
2206: PetscFunctionReturn(PETSC_SUCCESS);
2207: }
2209: /*@
2210: MatSetLocalToGlobalMapping - Sets a local-to-global numbering for use by
2211: the routine `MatSetValuesLocal()` to allow users to insert matrix entries
2212: using a local (per-processor) numbering.
2214: Not Collective
2216: Input Parameters:
2217: + x - the matrix
2218: . rmapping - row mapping created with `ISLocalToGlobalMappingCreate()` or `ISLocalToGlobalMappingCreateIS()`
2219: - cmapping - column mapping
2221: Level: intermediate
2223: Note:
2224: If the matrix is obtained with `DMCreateMatrix()` then this may already have been called on the matrix
2226: .seealso: [](ch_matrices), `Mat`, `DM`, `DMCreateMatrix()`, `MatGetLocalToGlobalMapping()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValues()`, `MatSetValuesLocal()`, `MatGetValuesLocal()`
2227: @*/
2228: PetscErrorCode MatSetLocalToGlobalMapping(Mat x, ISLocalToGlobalMapping rmapping, ISLocalToGlobalMapping cmapping)
2229: {
2230: PetscFunctionBegin;
2235: if (x->ops->setlocaltoglobalmapping) PetscUseTypeMethod(x, setlocaltoglobalmapping, rmapping, cmapping);
2236: else {
2237: PetscCall(PetscLayoutSetISLocalToGlobalMapping(x->rmap, rmapping));
2238: PetscCall(PetscLayoutSetISLocalToGlobalMapping(x->cmap, cmapping));
2239: }
2240: PetscFunctionReturn(PETSC_SUCCESS);
2241: }
2243: /*@
2244: MatGetLocalToGlobalMapping - Gets the local-to-global numbering set by `MatSetLocalToGlobalMapping()`
2246: Not Collective
2248: Input Parameter:
2249: . A - the matrix
2251: Output Parameters:
2252: + rmapping - row mapping
2253: - cmapping - column mapping
2255: Level: advanced
2257: .seealso: [](ch_matrices), `Mat`, `MatSetLocalToGlobalMapping()`, `MatSetValuesLocal()`
2258: @*/
2259: PetscErrorCode MatGetLocalToGlobalMapping(Mat A, ISLocalToGlobalMapping *rmapping, ISLocalToGlobalMapping *cmapping)
2260: {
2261: PetscFunctionBegin;
2264: if (rmapping) {
2266: *rmapping = A->rmap->mapping;
2267: }
2268: if (cmapping) {
2270: *cmapping = A->cmap->mapping;
2271: }
2272: PetscFunctionReturn(PETSC_SUCCESS);
2273: }
2275: /*@
2276: MatSetLayouts - Sets the `PetscLayout` objects for rows and columns of a matrix
2278: Logically Collective
2280: Input Parameters:
2281: + A - the matrix
2282: . rmap - row layout
2283: - cmap - column layout
2285: Level: advanced
2287: Note:
2288: The `PetscLayout` objects are usually created automatically for the matrix so this routine rarely needs to be called.
2290: .seealso: [](ch_matrices), `Mat`, `PetscLayout`, `MatCreateVecs()`, `MatGetLocalToGlobalMapping()`, `MatGetLayouts()`
2291: @*/
2292: PetscErrorCode MatSetLayouts(Mat A, PetscLayout rmap, PetscLayout cmap)
2293: {
2294: PetscFunctionBegin;
2296: PetscCall(PetscLayoutReference(rmap, &A->rmap));
2297: PetscCall(PetscLayoutReference(cmap, &A->cmap));
2298: PetscFunctionReturn(PETSC_SUCCESS);
2299: }
2301: /*@
2302: MatGetLayouts - Gets the `PetscLayout` objects for rows and columns
2304: Not Collective
2306: Input Parameter:
2307: . A - the matrix
2309: Output Parameters:
2310: + rmap - row layout
2311: - cmap - column layout
2313: Level: advanced
2315: .seealso: [](ch_matrices), `Mat`, [Matrix Layouts](sec_matlayout), `PetscLayout`, `MatCreateVecs()`, `MatGetLocalToGlobalMapping()`, `MatSetLayouts()`
2316: @*/
2317: PetscErrorCode MatGetLayouts(Mat A, PetscLayout *rmap, PetscLayout *cmap)
2318: {
2319: PetscFunctionBegin;
2322: if (rmap) {
2324: *rmap = A->rmap;
2325: }
2326: if (cmap) {
2328: *cmap = A->cmap;
2329: }
2330: PetscFunctionReturn(PETSC_SUCCESS);
2331: }
2333: /*@C
2334: MatSetValuesLocal - Inserts or adds values into certain locations of a matrix,
2335: using a local numbering of the nodes.
2337: Not Collective
2339: Input Parameters:
2340: + mat - the matrix
2341: . nrow - number of rows
2342: . irow - the row local indices
2343: . ncol - number of columns
2344: . icol - the column local indices
2345: . y - a logically two-dimensional array of values
2346: - addv - either `INSERT_VALUES` to add values to any existing entries, or `INSERT_VALUES` to replace existing entries with new values
2348: Level: intermediate
2350: Notes:
2351: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call MatXXXXSetPreallocation() or
2352: `MatSetUp()` before using this routine
2354: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call `MatSetLocalToGlobalMapping()` before using this routine
2356: Calls to `MatSetValuesLocal()` with the `INSERT_VALUES` and `ADD_VALUES`
2357: options cannot be mixed without intervening calls to the assembly
2358: routines.
2360: These values may be cached, so `MatAssemblyBegin()` and `MatAssemblyEnd()`
2361: MUST be called after all calls to `MatSetValuesLocal()` have been completed.
2363: Developer Note:
2364: This is labeled with C so does not automatically generate Fortran stubs and interfaces
2365: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
2367: .seealso: [](ch_matrices), `Mat`, `MatAssemblyBegin()`, `MatAssemblyEnd()`, `MatSetValues()`, `MatSetLocalToGlobalMapping()`,
2368: `MatGetValuesLocal()`
2369: @*/
2370: PetscErrorCode MatSetValuesLocal(Mat mat, PetscInt nrow, const PetscInt irow[], PetscInt ncol, const PetscInt icol[], const PetscScalar y[], InsertMode addv)
2371: {
2372: PetscFunctionBeginHot;
2375: MatCheckPreallocated(mat, 1);
2376: if (!nrow || !ncol) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
2379: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
2380: else PetscCheck(mat->insertmode == addv, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add values and insert values");
2381: if (PetscDefined(USE_DEBUG)) {
2382: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2383: PetscCheck(mat->ops->setvalueslocal || mat->ops->setvalues, PETSC_COMM_SELF, PETSC_ERR_SUP, "Mat type %s", ((PetscObject)mat)->type_name);
2384: }
2386: if (mat->assembled) {
2387: mat->was_assembled = PETSC_TRUE;
2388: mat->assembled = PETSC_FALSE;
2389: }
2390: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
2391: if (mat->ops->setvalueslocal) PetscUseTypeMethod(mat, setvalueslocal, nrow, irow, ncol, icol, y, addv);
2392: else {
2393: PetscInt buf[8192], *bufr = NULL, *bufc = NULL;
2394: const PetscInt *irowm, *icolm;
2396: if ((!mat->rmap->mapping && !mat->cmap->mapping) || (nrow + ncol) <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
2397: bufr = buf;
2398: bufc = buf + nrow;
2399: irowm = bufr;
2400: icolm = bufc;
2401: } else {
2402: PetscCall(PetscMalloc2(nrow, &bufr, ncol, &bufc));
2403: irowm = bufr;
2404: icolm = bufc;
2405: }
2406: if (mat->rmap->mapping) PetscCall(ISLocalToGlobalMappingApply(mat->rmap->mapping, nrow, irow, bufr));
2407: else irowm = irow;
2408: if (mat->cmap->mapping) {
2409: if (mat->cmap->mapping != mat->rmap->mapping || ncol != nrow || icol != irow) {
2410: PetscCall(ISLocalToGlobalMappingApply(mat->cmap->mapping, ncol, icol, bufc));
2411: } else icolm = irowm;
2412: } else icolm = icol;
2413: PetscCall(MatSetValues(mat, nrow, irowm, ncol, icolm, y, addv));
2414: if (bufr != buf) PetscCall(PetscFree2(bufr, bufc));
2415: }
2416: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
2417: PetscFunctionReturn(PETSC_SUCCESS);
2418: }
2420: /*@C
2421: MatSetValuesBlockedLocal - Inserts or adds values into certain locations of a matrix,
2422: using a local ordering of the nodes a block at a time.
2424: Not Collective
2426: Input Parameters:
2427: + x - the matrix
2428: . nrow - number of rows
2429: . irow - the row local indices
2430: . ncol - number of columns
2431: . icol - the column local indices
2432: . y - a logically two-dimensional array of values
2433: - addv - either `ADD_VALUES` to add values to any existing entries, or `INSERT_VALUES` to replace existing entries with new values
2435: Level: intermediate
2437: Notes:
2438: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call MatXXXXSetPreallocation() or
2439: `MatSetUp()` before using this routine
2441: If you create the matrix yourself (that is not with a call to `DMCreateMatrix()`) then you MUST call `MatSetBlockSize()` and `MatSetLocalToGlobalMapping()`
2442: before using this routineBefore calling `MatSetValuesLocal()`, the user must first set the
2444: Calls to `MatSetValuesBlockedLocal()` with the `INSERT_VALUES` and `ADD_VALUES`
2445: options cannot be mixed without intervening calls to the assembly
2446: routines.
2448: These values may be cached, so `MatAssemblyBegin()` and `MatAssemblyEnd()`
2449: MUST be called after all calls to `MatSetValuesBlockedLocal()` have been completed.
2451: Developer Note:
2452: This is labeled with C so does not automatically generate Fortran stubs and interfaces
2453: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
2455: .seealso: [](ch_matrices), `Mat`, `MatSetBlockSize()`, `MatSetLocalToGlobalMapping()`, `MatAssemblyBegin()`, `MatAssemblyEnd()`,
2456: `MatSetValuesLocal()`, `MatSetValuesBlocked()`
2457: @*/
2458: PetscErrorCode MatSetValuesBlockedLocal(Mat mat, PetscInt nrow, const PetscInt irow[], PetscInt ncol, const PetscInt icol[], const PetscScalar y[], InsertMode addv)
2459: {
2460: PetscFunctionBeginHot;
2463: MatCheckPreallocated(mat, 1);
2464: if (!nrow || !ncol) PetscFunctionReturn(PETSC_SUCCESS); /* no values to insert */
2467: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
2468: else PetscCheck(mat->insertmode == addv, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add values and insert values");
2469: if (PetscDefined(USE_DEBUG)) {
2470: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2471: PetscCheck(mat->ops->setvaluesblockedlocal || mat->ops->setvaluesblocked || mat->ops->setvalueslocal || mat->ops->setvalues, PETSC_COMM_SELF, PETSC_ERR_SUP, "Mat type %s", ((PetscObject)mat)->type_name);
2472: }
2474: if (mat->assembled) {
2475: mat->was_assembled = PETSC_TRUE;
2476: mat->assembled = PETSC_FALSE;
2477: }
2478: if (PetscUnlikelyDebug(mat->rmap->mapping)) { /* Condition on the mapping existing, because MatSetValuesBlockedLocal_IS does not require it to be set. */
2479: PetscInt irbs, rbs;
2480: PetscCall(MatGetBlockSizes(mat, &rbs, NULL));
2481: PetscCall(ISLocalToGlobalMappingGetBlockSize(mat->rmap->mapping, &irbs));
2482: PetscCheck(rbs == irbs, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Different row block sizes! mat %" PetscInt_FMT ", row l2g map %" PetscInt_FMT, rbs, irbs);
2483: }
2484: if (PetscUnlikelyDebug(mat->cmap->mapping)) {
2485: PetscInt icbs, cbs;
2486: PetscCall(MatGetBlockSizes(mat, NULL, &cbs));
2487: PetscCall(ISLocalToGlobalMappingGetBlockSize(mat->cmap->mapping, &icbs));
2488: PetscCheck(cbs == icbs, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Different col block sizes! mat %" PetscInt_FMT ", col l2g map %" PetscInt_FMT, cbs, icbs);
2489: }
2490: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
2491: if (mat->ops->setvaluesblockedlocal) PetscUseTypeMethod(mat, setvaluesblockedlocal, nrow, irow, ncol, icol, y, addv);
2492: else {
2493: PetscInt buf[8192], *bufr = NULL, *bufc = NULL;
2494: const PetscInt *irowm, *icolm;
2496: if ((!mat->rmap->mapping && !mat->cmap->mapping) || (nrow + ncol) <= (PetscInt)(sizeof(buf) / sizeof(PetscInt))) {
2497: bufr = buf;
2498: bufc = buf + nrow;
2499: irowm = bufr;
2500: icolm = bufc;
2501: } else {
2502: PetscCall(PetscMalloc2(nrow, &bufr, ncol, &bufc));
2503: irowm = bufr;
2504: icolm = bufc;
2505: }
2506: if (mat->rmap->mapping) PetscCall(ISLocalToGlobalMappingApplyBlock(mat->rmap->mapping, nrow, irow, bufr));
2507: else irowm = irow;
2508: if (mat->cmap->mapping) {
2509: if (mat->cmap->mapping != mat->rmap->mapping || ncol != nrow || icol != irow) {
2510: PetscCall(ISLocalToGlobalMappingApplyBlock(mat->cmap->mapping, ncol, icol, bufc));
2511: } else icolm = irowm;
2512: } else icolm = icol;
2513: PetscCall(MatSetValuesBlocked(mat, nrow, irowm, ncol, icolm, y, addv));
2514: if (bufr != buf) PetscCall(PetscFree2(bufr, bufc));
2515: }
2516: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
2517: PetscFunctionReturn(PETSC_SUCCESS);
2518: }
2520: /*@
2521: MatMultDiagonalBlock - Computes the matrix-vector product, y = Dx. Where D is defined by the inode or block structure of the diagonal
2523: Collective
2525: Input Parameters:
2526: + mat - the matrix
2527: - x - the vector to be multiplied
2529: Output Parameter:
2530: . y - the result
2532: Level: developer
2534: Note:
2535: The vectors `x` and `y` cannot be the same. I.e., one cannot
2536: call `MatMultDiagonalBlock`(A,y,y).
2538: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `MatMultTranspose()`, `MatMultAdd()`, `MatMultTransposeAdd()`
2539: @*/
2540: PetscErrorCode MatMultDiagonalBlock(Mat mat, Vec x, Vec y)
2541: {
2542: PetscFunctionBegin;
2548: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2549: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2550: PetscCheck(x != y, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "x and y must be different vectors");
2551: MatCheckPreallocated(mat, 1);
2553: PetscUseTypeMethod(mat, multdiagonalblock, x, y);
2554: PetscCall(PetscObjectStateIncrease((PetscObject)y));
2555: PetscFunctionReturn(PETSC_SUCCESS);
2556: }
2558: /*@
2559: MatMult - Computes the matrix-vector product, y = Ax.
2561: Neighbor-wise Collective
2563: Input Parameters:
2564: + mat - the matrix
2565: - x - the vector to be multiplied
2567: Output Parameter:
2568: . y - the result
2570: Level: beginner
2572: Note:
2573: The vectors `x` and `y` cannot be the same. I.e., one cannot
2574: call `MatMult`(A,y,y).
2576: .seealso: [](ch_matrices), `Mat`, `MatMultTranspose()`, `MatMultAdd()`, `MatMultTransposeAdd()`
2577: @*/
2578: PetscErrorCode MatMult(Mat mat, Vec x, Vec y)
2579: {
2580: PetscFunctionBegin;
2584: VecCheckAssembled(x);
2586: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2587: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2588: PetscCheck(x != y, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "x and y must be different vectors");
2589: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
2590: PetscCheck(mat->rmap->N == y->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, y->map->N);
2591: PetscCheck(mat->cmap->n == x->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->n, x->map->n);
2592: PetscCheck(mat->rmap->n == y->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, y->map->n);
2593: PetscCall(VecSetErrorIfLocked(y, 3));
2594: if (mat->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
2595: MatCheckPreallocated(mat, 1);
2597: PetscCall(VecLockReadPush(x));
2598: PetscCall(PetscLogEventBegin(MAT_Mult, mat, x, y, 0));
2599: PetscUseTypeMethod(mat, mult, x, y);
2600: PetscCall(PetscLogEventEnd(MAT_Mult, mat, x, y, 0));
2601: if (mat->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
2602: PetscCall(VecLockReadPop(x));
2603: PetscFunctionReturn(PETSC_SUCCESS);
2604: }
2606: /*@
2607: MatMultTranspose - Computes matrix transpose times a vector y = A^T * x.
2609: Neighbor-wise Collective
2611: Input Parameters:
2612: + mat - the matrix
2613: - x - the vector to be multiplied
2615: Output Parameter:
2616: . y - the result
2618: Level: beginner
2620: Notes:
2621: The vectors `x` and `y` cannot be the same. I.e., one cannot
2622: call `MatMultTranspose`(A,y,y).
2624: For complex numbers this does NOT compute the Hermitian (complex conjugate) transpose multiple,
2625: use `MatMultHermitianTranspose()`
2627: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `MatMultAdd()`, `MatMultTransposeAdd()`, `MatMultHermitianTranspose()`, `MatTranspose()`
2628: @*/
2629: PetscErrorCode MatMultTranspose(Mat mat, Vec x, Vec y)
2630: {
2631: PetscErrorCode (*op)(Mat, Vec, Vec) = NULL;
2633: PetscFunctionBegin;
2637: VecCheckAssembled(x);
2640: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2641: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2642: PetscCheck(x != y, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "x and y must be different vectors");
2643: PetscCheck(mat->cmap->N == y->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, y->map->N);
2644: PetscCheck(mat->rmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, x->map->N);
2645: PetscCheck(mat->cmap->n == y->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->n, y->map->n);
2646: PetscCheck(mat->rmap->n == x->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, x->map->n);
2647: if (mat->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
2648: MatCheckPreallocated(mat, 1);
2650: if (!mat->ops->multtranspose) {
2651: if (mat->symmetric == PETSC_BOOL3_TRUE && mat->ops->mult) op = mat->ops->mult;
2652: PetscCheck(op, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Matrix type %s does not have a multiply transpose defined or is symmetric and does not have a multiply defined", ((PetscObject)mat)->type_name);
2653: } else op = mat->ops->multtranspose;
2654: PetscCall(PetscLogEventBegin(MAT_MultTranspose, mat, x, y, 0));
2655: PetscCall(VecLockReadPush(x));
2656: PetscCall((*op)(mat, x, y));
2657: PetscCall(VecLockReadPop(x));
2658: PetscCall(PetscLogEventEnd(MAT_MultTranspose, mat, x, y, 0));
2659: PetscCall(PetscObjectStateIncrease((PetscObject)y));
2660: if (mat->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
2661: PetscFunctionReturn(PETSC_SUCCESS);
2662: }
2664: /*@
2665: MatMultHermitianTranspose - Computes matrix Hermitian transpose times a vector.
2667: Neighbor-wise Collective
2669: Input Parameters:
2670: + mat - the matrix
2671: - x - the vector to be multiplied
2673: Output Parameter:
2674: . y - the result
2676: Level: beginner
2678: Notes:
2679: The vectors `x` and `y` cannot be the same. I.e., one cannot
2680: call `MatMultHermitianTranspose`(A,y,y).
2682: Also called the conjugate transpose, complex conjugate transpose, or adjoint.
2684: For real numbers `MatMultTranspose()` and `MatMultHermitianTranspose()` are identical.
2686: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `MatMultAdd()`, `MatMultHermitianTransposeAdd()`, `MatMultTranspose()`
2687: @*/
2688: PetscErrorCode MatMultHermitianTranspose(Mat mat, Vec x, Vec y)
2689: {
2690: PetscFunctionBegin;
2696: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2697: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2698: PetscCheck(x != y, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "x and y must be different vectors");
2699: PetscCheck(mat->cmap->N == y->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, y->map->N);
2700: PetscCheck(mat->rmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, x->map->N);
2701: PetscCheck(mat->cmap->n == y->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->n, y->map->n);
2702: PetscCheck(mat->rmap->n == x->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, x->map->n);
2703: MatCheckPreallocated(mat, 1);
2705: PetscCall(PetscLogEventBegin(MAT_MultHermitianTranspose, mat, x, y, 0));
2706: #if defined(PETSC_USE_COMPLEX)
2707: if (mat->ops->multhermitiantranspose || (mat->hermitian == PETSC_BOOL3_TRUE && mat->ops->mult)) {
2708: PetscCall(VecLockReadPush(x));
2709: if (mat->ops->multhermitiantranspose) PetscUseTypeMethod(mat, multhermitiantranspose, x, y);
2710: else PetscUseTypeMethod(mat, mult, x, y);
2711: PetscCall(VecLockReadPop(x));
2712: } else {
2713: Vec w;
2714: PetscCall(VecDuplicate(x, &w));
2715: PetscCall(VecCopy(x, w));
2716: PetscCall(VecConjugate(w));
2717: PetscCall(MatMultTranspose(mat, w, y));
2718: PetscCall(VecDestroy(&w));
2719: PetscCall(VecConjugate(y));
2720: }
2721: PetscCall(PetscObjectStateIncrease((PetscObject)y));
2722: #else
2723: PetscCall(MatMultTranspose(mat, x, y));
2724: #endif
2725: PetscCall(PetscLogEventEnd(MAT_MultHermitianTranspose, mat, x, y, 0));
2726: PetscFunctionReturn(PETSC_SUCCESS);
2727: }
2729: /*@
2730: MatMultAdd - Computes v3 = v2 + A * v1.
2732: Neighbor-wise Collective
2734: Input Parameters:
2735: + mat - the matrix
2736: . v1 - the vector to be multiplied by `mat`
2737: - v2 - the vector to be added to the result
2739: Output Parameter:
2740: . v3 - the result
2742: Level: beginner
2744: Note:
2745: The vectors `v1` and `v3` cannot be the same. I.e., one cannot
2746: call `MatMultAdd`(A,v1,v2,v1).
2748: .seealso: [](ch_matrices), `Mat`, `MatMultTranspose()`, `MatMult()`, `MatMultTransposeAdd()`
2749: @*/
2750: PetscErrorCode MatMultAdd(Mat mat, Vec v1, Vec v2, Vec v3)
2751: {
2752: PetscFunctionBegin;
2759: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2760: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2761: PetscCheck(mat->cmap->N == v1->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v1: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, v1->map->N);
2762: /* PetscCheck(mat->rmap->N == v2->map->N,PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %" PetscInt_FMT " %" PetscInt_FMT,mat->rmap->N,v2->map->N);
2763: PetscCheck(mat->rmap->N == v3->map->N,PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %" PetscInt_FMT " %" PetscInt_FMT,mat->rmap->N,v3->map->N); */
2764: PetscCheck(mat->rmap->n == v3->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec v3: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, v3->map->n);
2765: PetscCheck(mat->rmap->n == v2->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec v2: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, v2->map->n);
2766: PetscCheck(v1 != v3, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "v1 and v3 must be different vectors");
2767: MatCheckPreallocated(mat, 1);
2769: PetscCall(PetscLogEventBegin(MAT_MultAdd, mat, v1, v2, v3));
2770: PetscCall(VecLockReadPush(v1));
2771: PetscUseTypeMethod(mat, multadd, v1, v2, v3);
2772: PetscCall(VecLockReadPop(v1));
2773: PetscCall(PetscLogEventEnd(MAT_MultAdd, mat, v1, v2, v3));
2774: PetscCall(PetscObjectStateIncrease((PetscObject)v3));
2775: PetscFunctionReturn(PETSC_SUCCESS);
2776: }
2778: /*@
2779: MatMultTransposeAdd - Computes v3 = v2 + A' * v1.
2781: Neighbor-wise Collective
2783: Input Parameters:
2784: + mat - the matrix
2785: . v1 - the vector to be multiplied by the transpose of the matrix
2786: - v2 - the vector to be added to the result
2788: Output Parameter:
2789: . v3 - the result
2791: Level: beginner
2793: Note:
2794: The vectors `v1` and `v3` cannot be the same. I.e., one cannot
2795: call `MatMultTransposeAdd`(A,v1,v2,v1).
2797: .seealso: [](ch_matrices), `Mat`, `MatMultTranspose()`, `MatMultAdd()`, `MatMult()`
2798: @*/
2799: PetscErrorCode MatMultTransposeAdd(Mat mat, Vec v1, Vec v2, Vec v3)
2800: {
2801: PetscErrorCode (*op)(Mat, Vec, Vec, Vec) = (!mat->ops->multtransposeadd && mat->symmetric) ? mat->ops->multadd : mat->ops->multtransposeadd;
2803: PetscFunctionBegin;
2810: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2811: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2812: PetscCheck(mat->rmap->N == v1->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v1: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, v1->map->N);
2813: PetscCheck(mat->cmap->N == v2->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v2: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, v2->map->N);
2814: PetscCheck(mat->cmap->N == v3->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v3: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, v3->map->N);
2815: PetscCheck(v1 != v3, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "v1 and v3 must be different vectors");
2816: PetscCheck(op, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Mat type %s", ((PetscObject)mat)->type_name);
2817: MatCheckPreallocated(mat, 1);
2819: PetscCall(PetscLogEventBegin(MAT_MultTransposeAdd, mat, v1, v2, v3));
2820: PetscCall(VecLockReadPush(v1));
2821: PetscCall((*op)(mat, v1, v2, v3));
2822: PetscCall(VecLockReadPop(v1));
2823: PetscCall(PetscLogEventEnd(MAT_MultTransposeAdd, mat, v1, v2, v3));
2824: PetscCall(PetscObjectStateIncrease((PetscObject)v3));
2825: PetscFunctionReturn(PETSC_SUCCESS);
2826: }
2828: /*@
2829: MatMultHermitianTransposeAdd - Computes v3 = v2 + A^H * v1.
2831: Neighbor-wise Collective
2833: Input Parameters:
2834: + mat - the matrix
2835: . v1 - the vector to be multiplied by the Hermitian transpose
2836: - v2 - the vector to be added to the result
2838: Output Parameter:
2839: . v3 - the result
2841: Level: beginner
2843: Note:
2844: The vectors `v1` and `v3` cannot be the same. I.e., one cannot
2845: call `MatMultHermitianTransposeAdd`(A,v1,v2,v1).
2847: .seealso: [](ch_matrices), `Mat`, `MatMultHermitianTranspose()`, `MatMultTranspose()`, `MatMultAdd()`, `MatMult()`
2848: @*/
2849: PetscErrorCode MatMultHermitianTransposeAdd(Mat mat, Vec v1, Vec v2, Vec v3)
2850: {
2851: PetscFunctionBegin;
2858: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
2859: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
2860: PetscCheck(v1 != v3, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "v1 and v3 must be different vectors");
2861: PetscCheck(mat->rmap->N == v1->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v1: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, v1->map->N);
2862: PetscCheck(mat->cmap->N == v2->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v2: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, v2->map->N);
2863: PetscCheck(mat->cmap->N == v3->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec v3: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, v3->map->N);
2864: MatCheckPreallocated(mat, 1);
2866: PetscCall(PetscLogEventBegin(MAT_MultHermitianTransposeAdd, mat, v1, v2, v3));
2867: PetscCall(VecLockReadPush(v1));
2868: if (mat->ops->multhermitiantransposeadd) PetscUseTypeMethod(mat, multhermitiantransposeadd, v1, v2, v3);
2869: else {
2870: Vec w, z;
2871: PetscCall(VecDuplicate(v1, &w));
2872: PetscCall(VecCopy(v1, w));
2873: PetscCall(VecConjugate(w));
2874: PetscCall(VecDuplicate(v3, &z));
2875: PetscCall(MatMultTranspose(mat, w, z));
2876: PetscCall(VecDestroy(&w));
2877: PetscCall(VecConjugate(z));
2878: if (v2 != v3) {
2879: PetscCall(VecWAXPY(v3, 1.0, v2, z));
2880: } else {
2881: PetscCall(VecAXPY(v3, 1.0, z));
2882: }
2883: PetscCall(VecDestroy(&z));
2884: }
2885: PetscCall(VecLockReadPop(v1));
2886: PetscCall(PetscLogEventEnd(MAT_MultHermitianTransposeAdd, mat, v1, v2, v3));
2887: PetscCall(PetscObjectStateIncrease((PetscObject)v3));
2888: PetscFunctionReturn(PETSC_SUCCESS);
2889: }
2891: /*@C
2892: MatGetFactorType - gets the type of factorization it is
2894: Not Collective
2896: Input Parameter:
2897: . mat - the matrix
2899: Output Parameter:
2900: . t - the type, one of `MAT_FACTOR_NONE`, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ILU`, `MAT_FACTOR_ICC,MAT_FACTOR_ILUDT`, `MAT_FACTOR_QR`
2902: Level: intermediate
2904: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorType`, `MatGetFactor()`, `MatSetFactorType()`, `MAT_FACTOR_NONE`, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ILU`,
2905: `MAT_FACTOR_ICC,MAT_FACTOR_ILUDT`, `MAT_FACTOR_QR`
2906: @*/
2907: PetscErrorCode MatGetFactorType(Mat mat, MatFactorType *t)
2908: {
2909: PetscFunctionBegin;
2913: *t = mat->factortype;
2914: PetscFunctionReturn(PETSC_SUCCESS);
2915: }
2917: /*@C
2918: MatSetFactorType - sets the type of factorization it is
2920: Logically Collective
2922: Input Parameters:
2923: + mat - the matrix
2924: - t - the type, one of `MAT_FACTOR_NONE`, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ILU`, `MAT_FACTOR_ICC,MAT_FACTOR_ILUDT`, `MAT_FACTOR_QR`
2926: Level: intermediate
2928: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorType`, `MatGetFactor()`, `MatGetFactorType()`, `MAT_FACTOR_NONE`, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ILU`,
2929: `MAT_FACTOR_ICC,MAT_FACTOR_ILUDT`, `MAT_FACTOR_QR`
2930: @*/
2931: PetscErrorCode MatSetFactorType(Mat mat, MatFactorType t)
2932: {
2933: PetscFunctionBegin;
2936: mat->factortype = t;
2937: PetscFunctionReturn(PETSC_SUCCESS);
2938: }
2940: /*@C
2941: MatGetInfo - Returns information about matrix storage (number of
2942: nonzeros, memory, etc.).
2944: Collective if `MAT_GLOBAL_MAX` or `MAT_GLOBAL_SUM` is used as the flag
2946: Input Parameters:
2947: + mat - the matrix
2948: - flag - flag indicating the type of parameters to be returned (`MAT_LOCAL` - local matrix, `MAT_GLOBAL_MAX` - maximum over all processors, `MAT_GLOBAL_SUM` - sum over all processors)
2950: Output Parameter:
2951: . info - matrix information context
2953: Options Database Key:
2954: . -mat_view ::ascii_info - print matrix info to `PETSC_STDOUT`
2956: Notes:
2957: The `MatInfo` context contains a variety of matrix data, including
2958: number of nonzeros allocated and used, number of mallocs during
2959: matrix assembly, etc. Additional information for factored matrices
2960: is provided (such as the fill ratio, number of mallocs during
2961: factorization, etc.).
2963: Example:
2964: See the file ${PETSC_DIR}/include/petscmat.h for a complete list of
2965: data within the MatInfo context. For example,
2966: .vb
2967: MatInfo info;
2968: Mat A;
2969: double mal, nz_a, nz_u;
2971: MatGetInfo(A, MAT_LOCAL, &info);
2972: mal = info.mallocs;
2973: nz_a = info.nz_allocated;
2974: .ve
2976: Fortran users should declare info as a double precision
2977: array of dimension `MAT_INFO_SIZE`, and then extract the parameters
2978: of interest. See the file ${PETSC_DIR}/include/petsc/finclude/petscmat.h
2979: a complete list of parameter names.
2980: .vb
2981: double precision info(MAT_INFO_SIZE)
2982: double precision mal, nz_a
2983: Mat A
2984: integer ierr
2986: call MatGetInfo(A, MAT_LOCAL, info, ierr)
2987: mal = info(MAT_INFO_MALLOCS)
2988: nz_a = info(MAT_INFO_NZ_ALLOCATED)
2989: .ve
2991: Level: intermediate
2993: Developer Note:
2994: The Fortran interface is not autogenerated as the
2995: interface definition cannot be generated correctly [due to `MatInfo` argument]
2997: .seealso: [](ch_matrices), `Mat`, `MatInfo`, `MatStashGetInfo()`
2998: @*/
2999: PetscErrorCode MatGetInfo(Mat mat, MatInfoType flag, MatInfo *info)
3000: {
3001: PetscFunctionBegin;
3005: MatCheckPreallocated(mat, 1);
3006: PetscUseTypeMethod(mat, getinfo, flag, info);
3007: PetscFunctionReturn(PETSC_SUCCESS);
3008: }
3010: /*
3011: This is used by external packages where it is not easy to get the info from the actual
3012: matrix factorization.
3013: */
3014: PetscErrorCode MatGetInfo_External(Mat A, MatInfoType flag, MatInfo *info)
3015: {
3016: PetscFunctionBegin;
3017: PetscCall(PetscMemzero(info, sizeof(MatInfo)));
3018: PetscFunctionReturn(PETSC_SUCCESS);
3019: }
3021: /*@C
3022: MatLUFactor - Performs in-place LU factorization of matrix.
3024: Collective
3026: Input Parameters:
3027: + mat - the matrix
3028: . row - row permutation
3029: . col - column permutation
3030: - info - options for factorization, includes
3031: .vb
3032: fill - expected fill as ratio of original fill.
3033: dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3034: Run with the option -info to determine an optimal value to use
3035: .ve
3036: Level: developer
3038: Notes:
3039: Most users should employ the `KSP` interface for linear solvers
3040: instead of working directly with matrix algebra routines such as this.
3041: See, e.g., `KSPCreate()`.
3043: This changes the state of the matrix to a factored matrix; it cannot be used
3044: for example with `MatSetValues()` unless one first calls `MatSetUnfactored()`.
3046: This is really in-place only for dense matrices, the preferred approach is to use `MatGetFactor()`, `MatLUFactorSymbolic()`, and `MatLUFactorNumeric()`
3047: when not using `KSP`.
3049: Developer Note:
3050: The Fortran interface is not autogenerated as the
3051: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3053: .seealso: [](ch_matrices), [Matrix Factorization](sec_matfactor), `Mat`, `MatFactorType`, `MatLUFactorSymbolic()`, `MatLUFactorNumeric()`, `MatCholeskyFactor()`,
3054: `MatGetOrdering()`, `MatSetUnfactored()`, `MatFactorInfo`, `MatGetFactor()`
3055: @*/
3056: PetscErrorCode MatLUFactor(Mat mat, IS row, IS col, const MatFactorInfo *info)
3057: {
3058: MatFactorInfo tinfo;
3060: PetscFunctionBegin;
3066: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3067: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3068: MatCheckPreallocated(mat, 1);
3069: if (!info) {
3070: PetscCall(MatFactorInfoInitialize(&tinfo));
3071: info = &tinfo;
3072: }
3074: PetscCall(PetscLogEventBegin(MAT_LUFactor, mat, row, col, 0));
3075: PetscUseTypeMethod(mat, lufactor, row, col, info);
3076: PetscCall(PetscLogEventEnd(MAT_LUFactor, mat, row, col, 0));
3077: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
3078: PetscFunctionReturn(PETSC_SUCCESS);
3079: }
3081: /*@C
3082: MatILUFactor - Performs in-place ILU factorization of matrix.
3084: Collective
3086: Input Parameters:
3087: + mat - the matrix
3088: . row - row permutation
3089: . col - column permutation
3090: - info - structure containing
3091: .vb
3092: levels - number of levels of fill.
3093: expected fill - as ratio of original fill.
3094: 1 or 0 - indicating force fill on diagonal (improves robustness for matrices
3095: missing diagonal entries)
3096: .ve
3098: Level: developer
3100: Notes:
3101: Most users should employ the `KSP` interface for linear solvers
3102: instead of working directly with matrix algebra routines such as this.
3103: See, e.g., `KSPCreate()`.
3105: Probably really in-place only when level of fill is zero, otherwise allocates
3106: new space to store factored matrix and deletes previous memory. The preferred approach is to use `MatGetFactor()`, `MatILUFactorSymbolic()`, and `MatILUFactorNumeric()`
3107: when not using `KSP`.
3109: Developer Note:
3110: The Fortran interface is not autogenerated as the
3111: interface definition cannot be generated correctly [due to MatFactorInfo]
3113: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatILUFactorSymbolic()`, `MatLUFactorNumeric()`, `MatCholeskyFactor()`, `MatFactorInfo`
3114: @*/
3115: PetscErrorCode MatILUFactor(Mat mat, IS row, IS col, const MatFactorInfo *info)
3116: {
3117: PetscFunctionBegin;
3123: PetscCheck(mat->rmap->N == mat->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONG, "matrix must be square");
3124: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3125: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3126: MatCheckPreallocated(mat, 1);
3128: PetscCall(PetscLogEventBegin(MAT_ILUFactor, mat, row, col, 0));
3129: PetscUseTypeMethod(mat, ilufactor, row, col, info);
3130: PetscCall(PetscLogEventEnd(MAT_ILUFactor, mat, row, col, 0));
3131: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
3132: PetscFunctionReturn(PETSC_SUCCESS);
3133: }
3135: /*@C
3136: MatLUFactorSymbolic - Performs symbolic LU factorization of matrix.
3137: Call this routine before calling `MatLUFactorNumeric()` and after `MatGetFactor()`.
3139: Collective
3141: Input Parameters:
3142: + fact - the factor matrix obtained with `MatGetFactor()`
3143: . mat - the matrix
3144: . row - the row permutation
3145: . col - the column permutation
3146: - info - options for factorization, includes
3147: .vb
3148: fill - expected fill as ratio of original fill. Run with the option -info to determine an optimal value to use
3149: dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3150: .ve
3152: Level: developer
3154: Notes:
3155: See [Matrix Factorization](sec_matfactor) for additional information about factorizations
3157: Most users should employ the simplified `KSP` interface for linear solvers
3158: instead of working directly with matrix algebra routines such as this.
3159: See, e.g., `KSPCreate()`.
3161: Developer Note:
3162: The Fortran interface is not autogenerated as the
3163: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3165: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatLUFactor()`, `MatLUFactorNumeric()`, `MatCholeskyFactor()`, `MatFactorInfo`, `MatFactorInfoInitialize()`
3166: @*/
3167: PetscErrorCode MatLUFactorSymbolic(Mat fact, Mat mat, IS row, IS col, const MatFactorInfo *info)
3168: {
3169: MatFactorInfo tinfo;
3171: PetscFunctionBegin;
3179: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3180: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3181: MatCheckPreallocated(mat, 2);
3182: if (!info) {
3183: PetscCall(MatFactorInfoInitialize(&tinfo));
3184: info = &tinfo;
3185: }
3187: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_LUFactorSymbolic, mat, row, col, 0));
3188: PetscUseTypeMethod(fact, lufactorsymbolic, mat, row, col, info);
3189: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_LUFactorSymbolic, mat, row, col, 0));
3190: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3191: PetscFunctionReturn(PETSC_SUCCESS);
3192: }
3194: /*@C
3195: MatLUFactorNumeric - Performs numeric LU factorization of a matrix.
3196: Call this routine after first calling `MatLUFactorSymbolic()` and `MatGetFactor()`.
3198: Collective
3200: Input Parameters:
3201: + fact - the factor matrix obtained with `MatGetFactor()`
3202: . mat - the matrix
3203: - info - options for factorization
3205: Level: developer
3207: Notes:
3208: See `MatLUFactor()` for in-place factorization. See
3209: `MatCholeskyFactorNumeric()` for the symmetric, positive definite case.
3211: Most users should employ the `KSP` interface for linear solvers
3212: instead of working directly with matrix algebra routines such as this.
3213: See, e.g., `KSPCreate()`.
3215: Developer Note:
3216: The Fortran interface is not autogenerated as the
3217: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3219: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatFactorInfo`, `MatLUFactorSymbolic()`, `MatLUFactor()`, `MatCholeskyFactor()`
3220: @*/
3221: PetscErrorCode MatLUFactorNumeric(Mat fact, Mat mat, const MatFactorInfo *info)
3222: {
3223: MatFactorInfo tinfo;
3225: PetscFunctionBegin;
3231: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3232: PetscCheck(mat->rmap->N == (fact)->rmap->N && mat->cmap->N == (fact)->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Mat fact: global dimensions are different %" PetscInt_FMT " should = %" PetscInt_FMT " %" PetscInt_FMT " should = %" PetscInt_FMT,
3233: mat->rmap->N, (fact)->rmap->N, mat->cmap->N, (fact)->cmap->N);
3235: MatCheckPreallocated(mat, 2);
3236: if (!info) {
3237: PetscCall(MatFactorInfoInitialize(&tinfo));
3238: info = &tinfo;
3239: }
3241: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_LUFactorNumeric, mat, fact, 0, 0));
3242: else PetscCall(PetscLogEventBegin(MAT_LUFactor, mat, fact, 0, 0));
3243: PetscUseTypeMethod(fact, lufactornumeric, mat, info);
3244: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_LUFactorNumeric, mat, fact, 0, 0));
3245: else PetscCall(PetscLogEventEnd(MAT_LUFactor, mat, fact, 0, 0));
3246: PetscCall(MatViewFromOptions(fact, NULL, "-mat_factor_view"));
3247: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3248: PetscFunctionReturn(PETSC_SUCCESS);
3249: }
3251: /*@C
3252: MatCholeskyFactor - Performs in-place Cholesky factorization of a
3253: symmetric matrix.
3255: Collective
3257: Input Parameters:
3258: + mat - the matrix
3259: . perm - row and column permutations
3260: - f - expected fill as ratio of original fill
3262: Level: developer
3264: Notes:
3265: See `MatLUFactor()` for the nonsymmetric case. See also `MatGetFactor()`,
3266: `MatCholeskyFactorSymbolic()`, and `MatCholeskyFactorNumeric()`.
3268: Most users should employ the `KSP` interface for linear solvers
3269: instead of working directly with matrix algebra routines such as this.
3270: See, e.g., `KSPCreate()`.
3272: Developer Note:
3273: The Fortran interface is not autogenerated as the
3274: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3276: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatFactorInfo`, `MatLUFactor()`, `MatCholeskyFactorSymbolic()`, `MatCholeskyFactorNumeric()`
3277: `MatGetOrdering()`
3278: @*/
3279: PetscErrorCode MatCholeskyFactor(Mat mat, IS perm, const MatFactorInfo *info)
3280: {
3281: MatFactorInfo tinfo;
3283: PetscFunctionBegin;
3288: PetscCheck(mat->rmap->N == mat->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONG, "Matrix must be square");
3289: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3290: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3291: MatCheckPreallocated(mat, 1);
3292: if (!info) {
3293: PetscCall(MatFactorInfoInitialize(&tinfo));
3294: info = &tinfo;
3295: }
3297: PetscCall(PetscLogEventBegin(MAT_CholeskyFactor, mat, perm, 0, 0));
3298: PetscUseTypeMethod(mat, choleskyfactor, perm, info);
3299: PetscCall(PetscLogEventEnd(MAT_CholeskyFactor, mat, perm, 0, 0));
3300: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
3301: PetscFunctionReturn(PETSC_SUCCESS);
3302: }
3304: /*@C
3305: MatCholeskyFactorSymbolic - Performs symbolic Cholesky factorization
3306: of a symmetric matrix.
3308: Collective
3310: Input Parameters:
3311: + fact - the factor matrix obtained with `MatGetFactor()`
3312: . mat - the matrix
3313: . perm - row and column permutations
3314: - info - options for factorization, includes
3315: .vb
3316: fill - expected fill as ratio of original fill.
3317: dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3318: Run with the option -info to determine an optimal value to use
3319: .ve
3321: Level: developer
3323: Notes:
3324: See `MatLUFactorSymbolic()` for the nonsymmetric case. See also
3325: `MatCholeskyFactor()` and `MatCholeskyFactorNumeric()`.
3327: Most users should employ the `KSP` interface for linear solvers
3328: instead of working directly with matrix algebra routines such as this.
3329: See, e.g., `KSPCreate()`.
3331: Developer Note:
3332: The Fortran interface is not autogenerated as the
3333: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3335: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorInfo`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactor()`, `MatCholeskyFactorNumeric()`
3336: `MatGetOrdering()`
3337: @*/
3338: PetscErrorCode MatCholeskyFactorSymbolic(Mat fact, Mat mat, IS perm, const MatFactorInfo *info)
3339: {
3340: MatFactorInfo tinfo;
3342: PetscFunctionBegin;
3349: PetscCheck(mat->rmap->N == mat->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONG, "Matrix must be square");
3350: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3351: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3352: MatCheckPreallocated(mat, 2);
3353: if (!info) {
3354: PetscCall(MatFactorInfoInitialize(&tinfo));
3355: info = &tinfo;
3356: }
3358: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_CholeskyFactorSymbolic, mat, perm, 0, 0));
3359: PetscUseTypeMethod(fact, choleskyfactorsymbolic, mat, perm, info);
3360: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_CholeskyFactorSymbolic, mat, perm, 0, 0));
3361: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3362: PetscFunctionReturn(PETSC_SUCCESS);
3363: }
3365: /*@C
3366: MatCholeskyFactorNumeric - Performs numeric Cholesky factorization
3367: of a symmetric matrix. Call this routine after first calling `MatGetFactor()` and
3368: `MatCholeskyFactorSymbolic()`.
3370: Collective
3372: Input Parameters:
3373: + fact - the factor matrix obtained with `MatGetFactor()`, where the factored values are stored
3374: . mat - the initial matrix that is to be factored
3375: - info - options for factorization
3377: Level: developer
3379: Note:
3380: Most users should employ the `KSP` interface for linear solvers
3381: instead of working directly with matrix algebra routines such as this.
3382: See, e.g., `KSPCreate()`.
3384: Developer Note:
3385: The Fortran interface is not autogenerated as the
3386: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3388: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorInfo`, `MatGetFactor()`, `MatCholeskyFactorSymbolic()`, `MatCholeskyFactor()`, `MatLUFactorNumeric()`
3389: @*/
3390: PetscErrorCode MatCholeskyFactorNumeric(Mat fact, Mat mat, const MatFactorInfo *info)
3391: {
3392: MatFactorInfo tinfo;
3394: PetscFunctionBegin;
3400: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3401: PetscCheck(mat->rmap->N == (fact)->rmap->N && mat->cmap->N == (fact)->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Mat fact: global dim %" PetscInt_FMT " should = %" PetscInt_FMT " %" PetscInt_FMT " should = %" PetscInt_FMT,
3402: mat->rmap->N, (fact)->rmap->N, mat->cmap->N, (fact)->cmap->N);
3403: MatCheckPreallocated(mat, 2);
3404: if (!info) {
3405: PetscCall(MatFactorInfoInitialize(&tinfo));
3406: info = &tinfo;
3407: }
3409: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_CholeskyFactorNumeric, mat, fact, 0, 0));
3410: else PetscCall(PetscLogEventBegin(MAT_CholeskyFactor, mat, fact, 0, 0));
3411: PetscUseTypeMethod(fact, choleskyfactornumeric, mat, info);
3412: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_CholeskyFactorNumeric, mat, fact, 0, 0));
3413: else PetscCall(PetscLogEventEnd(MAT_CholeskyFactor, mat, fact, 0, 0));
3414: PetscCall(MatViewFromOptions(fact, NULL, "-mat_factor_view"));
3415: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3416: PetscFunctionReturn(PETSC_SUCCESS);
3417: }
3419: /*@
3420: MatQRFactor - Performs in-place QR factorization of matrix.
3422: Collective
3424: Input Parameters:
3425: + mat - the matrix
3426: . col - column permutation
3427: - info - options for factorization, includes
3428: .vb
3429: fill - expected fill as ratio of original fill.
3430: dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3431: Run with the option -info to determine an optimal value to use
3432: .ve
3434: Level: developer
3436: Notes:
3437: Most users should employ the `KSP` interface for linear solvers
3438: instead of working directly with matrix algebra routines such as this.
3439: See, e.g., `KSPCreate()`.
3441: This changes the state of the matrix to a factored matrix; it cannot be used
3442: for example with `MatSetValues()` unless one first calls `MatSetUnfactored()`.
3444: Developer Note:
3445: The Fortran interface is not autogenerated as the
3446: interface definition cannot be generated correctly [due to MatFactorInfo]
3448: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorInfo`, `MatGetFactor()`, `MatQRFactorSymbolic()`, `MatQRFactorNumeric()`, `MatLUFactor()`,
3449: `MatSetUnfactored()`, `MatFactorInfo`, `MatGetFactor()`
3450: @*/
3451: PetscErrorCode MatQRFactor(Mat mat, IS col, const MatFactorInfo *info)
3452: {
3453: PetscFunctionBegin;
3458: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3459: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3460: MatCheckPreallocated(mat, 1);
3461: PetscCall(PetscLogEventBegin(MAT_QRFactor, mat, col, 0, 0));
3462: PetscUseMethod(mat, "MatQRFactor_C", (Mat, IS, const MatFactorInfo *), (mat, col, info));
3463: PetscCall(PetscLogEventEnd(MAT_QRFactor, mat, col, 0, 0));
3464: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
3465: PetscFunctionReturn(PETSC_SUCCESS);
3466: }
3468: /*@
3469: MatQRFactorSymbolic - Performs symbolic QR factorization of matrix.
3470: Call this routine after `MatGetFactor()` but before calling `MatQRFactorNumeric()`.
3472: Collective
3474: Input Parameters:
3475: + fact - the factor matrix obtained with `MatGetFactor()`
3476: . mat - the matrix
3477: . col - column permutation
3478: - info - options for factorization, includes
3479: .vb
3480: fill - expected fill as ratio of original fill.
3481: dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3482: Run with the option -info to determine an optimal value to use
3483: .ve
3485: Level: developer
3487: Note:
3488: Most users should employ the `KSP` interface for linear solvers
3489: instead of working directly with matrix algebra routines such as this.
3490: See, e.g., `KSPCreate()`.
3492: Developer Note:
3493: The Fortran interface is not autogenerated as the
3494: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3496: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatFactorInfo`, `MatQRFactor()`, `MatQRFactorNumeric()`, `MatLUFactor()`, `MatFactorInfo`, `MatFactorInfoInitialize()`
3497: @*/
3498: PetscErrorCode MatQRFactorSymbolic(Mat fact, Mat mat, IS col, const MatFactorInfo *info)
3499: {
3500: MatFactorInfo tinfo;
3502: PetscFunctionBegin;
3509: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3510: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3511: MatCheckPreallocated(mat, 2);
3512: if (!info) {
3513: PetscCall(MatFactorInfoInitialize(&tinfo));
3514: info = &tinfo;
3515: }
3517: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_QRFactorSymbolic, fact, mat, col, 0));
3518: PetscUseMethod(fact, "MatQRFactorSymbolic_C", (Mat, Mat, IS, const MatFactorInfo *), (fact, mat, col, info));
3519: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_QRFactorSymbolic, fact, mat, col, 0));
3520: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3521: PetscFunctionReturn(PETSC_SUCCESS);
3522: }
3524: /*@
3525: MatQRFactorNumeric - Performs numeric QR factorization of a matrix.
3526: Call this routine after first calling `MatGetFactor()`, and `MatQRFactorSymbolic()`.
3528: Collective
3530: Input Parameters:
3531: + fact - the factor matrix obtained with `MatGetFactor()`
3532: . mat - the matrix
3533: - info - options for factorization
3535: Level: developer
3537: Notes:
3538: See `MatQRFactor()` for in-place factorization.
3540: Most users should employ the `KSP` interface for linear solvers
3541: instead of working directly with matrix algebra routines such as this.
3542: See, e.g., `KSPCreate()`.
3544: Developer Note:
3545: The Fortran interface is not autogenerated as the
3546: interface definition cannot be generated correctly [due to `MatFactorInfo`]
3548: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorInfo`, `MatGetFactor()`, `MatQRFactor()`, `MatQRFactorSymbolic()`, `MatLUFactor()`
3549: @*/
3550: PetscErrorCode MatQRFactorNumeric(Mat fact, Mat mat, const MatFactorInfo *info)
3551: {
3552: MatFactorInfo tinfo;
3554: PetscFunctionBegin;
3559: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3560: PetscCheck(mat->rmap->N == fact->rmap->N && mat->cmap->N == fact->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Mat fact: global dimensions are different %" PetscInt_FMT " should = %" PetscInt_FMT " %" PetscInt_FMT " should = %" PetscInt_FMT,
3561: mat->rmap->N, (fact)->rmap->N, mat->cmap->N, (fact)->cmap->N);
3563: MatCheckPreallocated(mat, 2);
3564: if (!info) {
3565: PetscCall(MatFactorInfoInitialize(&tinfo));
3566: info = &tinfo;
3567: }
3569: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_QRFactorNumeric, mat, fact, 0, 0));
3570: else PetscCall(PetscLogEventBegin(MAT_QRFactor, mat, fact, 0, 0));
3571: PetscUseMethod(fact, "MatQRFactorNumeric_C", (Mat, Mat, const MatFactorInfo *), (fact, mat, info));
3572: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_QRFactorNumeric, mat, fact, 0, 0));
3573: else PetscCall(PetscLogEventEnd(MAT_QRFactor, mat, fact, 0, 0));
3574: PetscCall(MatViewFromOptions(fact, NULL, "-mat_factor_view"));
3575: PetscCall(PetscObjectStateIncrease((PetscObject)fact));
3576: PetscFunctionReturn(PETSC_SUCCESS);
3577: }
3579: /*@
3580: MatSolve - Solves A x = b, given a factored matrix.
3582: Neighbor-wise Collective
3584: Input Parameters:
3585: + mat - the factored matrix
3586: - b - the right-hand-side vector
3588: Output Parameter:
3589: . x - the result vector
3591: Level: developer
3593: Notes:
3594: The vectors `b` and `x` cannot be the same. I.e., one cannot
3595: call `MatSolve`(A,x,x).
3597: Most users should employ the `KSP` interface for linear solvers
3598: instead of working directly with matrix algebra routines such as this.
3599: See, e.g., `KSPCreate()`.
3601: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatLUFactor()`, `MatSolveAdd()`, `MatSolveTranspose()`, `MatSolveTransposeAdd()`
3602: @*/
3603: PetscErrorCode MatSolve(Mat mat, Vec b, Vec x)
3604: {
3605: PetscFunctionBegin;
3610: PetscCheckSameComm(mat, 1, b, 2);
3611: PetscCheckSameComm(mat, 1, x, 3);
3612: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
3613: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
3614: PetscCheck(mat->rmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, b->map->N);
3615: PetscCheck(mat->rmap->n == b->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, b->map->n);
3616: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3617: MatCheckPreallocated(mat, 1);
3619: PetscCall(PetscLogEventBegin(MAT_Solve, mat, b, x, 0));
3620: if (mat->factorerrortype) {
3621: PetscCall(PetscInfo(mat, "MatFactorError %d\n", mat->factorerrortype));
3622: PetscCall(VecSetInf(x));
3623: } else PetscUseTypeMethod(mat, solve, b, x);
3624: PetscCall(PetscLogEventEnd(MAT_Solve, mat, b, x, 0));
3625: PetscCall(PetscObjectStateIncrease((PetscObject)x));
3626: PetscFunctionReturn(PETSC_SUCCESS);
3627: }
3629: static PetscErrorCode MatMatSolve_Basic(Mat A, Mat B, Mat X, PetscBool trans)
3630: {
3631: Vec b, x;
3632: PetscInt N, i;
3633: PetscErrorCode (*f)(Mat, Vec, Vec);
3634: PetscBool Abound, Bneedconv = PETSC_FALSE, Xneedconv = PETSC_FALSE;
3636: PetscFunctionBegin;
3637: if (A->factorerrortype) {
3638: PetscCall(PetscInfo(A, "MatFactorError %d\n", A->factorerrortype));
3639: PetscCall(MatSetInf(X));
3640: PetscFunctionReturn(PETSC_SUCCESS);
3641: }
3642: f = (!trans || (!A->ops->solvetranspose && A->symmetric)) ? A->ops->solve : A->ops->solvetranspose;
3643: PetscCheck(f, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Mat type %s", ((PetscObject)A)->type_name);
3644: PetscCall(MatBoundToCPU(A, &Abound));
3645: if (!Abound) {
3646: PetscCall(PetscObjectTypeCompareAny((PetscObject)B, &Bneedconv, MATSEQDENSE, MATMPIDENSE, ""));
3647: PetscCall(PetscObjectTypeCompareAny((PetscObject)X, &Xneedconv, MATSEQDENSE, MATMPIDENSE, ""));
3648: }
3649: #if PetscDefined(HAVE_CUDA)
3650: if (Bneedconv) PetscCall(MatConvert(B, MATDENSECUDA, MAT_INPLACE_MATRIX, &B));
3651: if (Xneedconv) PetscCall(MatConvert(X, MATDENSECUDA, MAT_INPLACE_MATRIX, &X));
3652: #elif PetscDefined(HAVE_HIP)
3653: if (Bneedconv) PetscCall(MatConvert(B, MATDENSEHIP, MAT_INPLACE_MATRIX, &B));
3654: if (Xneedconv) PetscCall(MatConvert(X, MATDENSEHIP, MAT_INPLACE_MATRIX, &X));
3655: #endif
3656: PetscCall(MatGetSize(B, NULL, &N));
3657: for (i = 0; i < N; i++) {
3658: PetscCall(MatDenseGetColumnVecRead(B, i, &b));
3659: PetscCall(MatDenseGetColumnVecWrite(X, i, &x));
3660: PetscCall((*f)(A, b, x));
3661: PetscCall(MatDenseRestoreColumnVecWrite(X, i, &x));
3662: PetscCall(MatDenseRestoreColumnVecRead(B, i, &b));
3663: }
3664: if (Bneedconv) PetscCall(MatConvert(B, MATDENSE, MAT_INPLACE_MATRIX, &B));
3665: if (Xneedconv) PetscCall(MatConvert(X, MATDENSE, MAT_INPLACE_MATRIX, &X));
3666: PetscFunctionReturn(PETSC_SUCCESS);
3667: }
3669: /*@
3670: MatMatSolve - Solves A X = B, given a factored matrix.
3672: Neighbor-wise Collective
3674: Input Parameters:
3675: + A - the factored matrix
3676: - B - the right-hand-side matrix `MATDENSE` (or sparse `MATAIJ`-- when using MUMPS)
3678: Output Parameter:
3679: . X - the result matrix (dense matrix)
3681: Level: developer
3683: Note:
3684: If `B` is a `MATDENSE` matrix then one can call `MatMatSolve`(A,B,B) except with `MATSOLVERMKL_CPARDISO`;
3685: otherwise, `B` and `X` cannot be the same.
3687: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatSolve()`, `MatMatSolveTranspose()`, `MatLUFactor()`, `MatCholeskyFactor()`
3688: @*/
3689: PetscErrorCode MatMatSolve(Mat A, Mat B, Mat X)
3690: {
3691: PetscFunctionBegin;
3696: PetscCheckSameComm(A, 1, B, 2);
3697: PetscCheckSameComm(A, 1, X, 3);
3698: PetscCheck(A->cmap->N == X->rmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat X: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->cmap->N, X->rmap->N);
3699: PetscCheck(A->rmap->N == B->rmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat B: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->rmap->N, B->rmap->N);
3700: PetscCheck(X->cmap->N == B->cmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Solution matrix must have same number of columns as rhs matrix");
3701: if (!A->rmap->N && !A->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3702: PetscCheck(A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Unfactored matrix");
3703: MatCheckPreallocated(A, 1);
3705: PetscCall(PetscLogEventBegin(MAT_MatSolve, A, B, X, 0));
3706: if (!A->ops->matsolve) {
3707: PetscCall(PetscInfo(A, "Mat type %s using basic MatMatSolve\n", ((PetscObject)A)->type_name));
3708: PetscCall(MatMatSolve_Basic(A, B, X, PETSC_FALSE));
3709: } else PetscUseTypeMethod(A, matsolve, B, X);
3710: PetscCall(PetscLogEventEnd(MAT_MatSolve, A, B, X, 0));
3711: PetscCall(PetscObjectStateIncrease((PetscObject)X));
3712: PetscFunctionReturn(PETSC_SUCCESS);
3713: }
3715: /*@
3716: MatMatSolveTranspose - Solves A^T X = B, given a factored matrix.
3718: Neighbor-wise Collective
3720: Input Parameters:
3721: + A - the factored matrix
3722: - B - the right-hand-side matrix (`MATDENSE` matrix)
3724: Output Parameter:
3725: . X - the result matrix (dense matrix)
3727: Level: developer
3729: Note:
3730: The matrices `B` and `X` cannot be the same. I.e., one cannot
3731: call `MatMatSolveTranspose`(A,X,X).
3733: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatSolveTranspose()`, `MatMatSolve()`, `MatLUFactor()`, `MatCholeskyFactor()`
3734: @*/
3735: PetscErrorCode MatMatSolveTranspose(Mat A, Mat B, Mat X)
3736: {
3737: PetscFunctionBegin;
3742: PetscCheckSameComm(A, 1, B, 2);
3743: PetscCheckSameComm(A, 1, X, 3);
3744: PetscCheck(X != B, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_IDN, "X and B must be different matrices");
3745: PetscCheck(A->cmap->N == X->rmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat X: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->cmap->N, X->rmap->N);
3746: PetscCheck(A->rmap->N == B->rmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat B: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->rmap->N, B->rmap->N);
3747: PetscCheck(A->rmap->n == B->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat A,Mat B: local dim %" PetscInt_FMT " %" PetscInt_FMT, A->rmap->n, B->rmap->n);
3748: PetscCheck(X->cmap->N >= B->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Solution matrix must have same number of columns as rhs matrix");
3749: if (!A->rmap->N && !A->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3750: PetscCheck(A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Unfactored matrix");
3751: MatCheckPreallocated(A, 1);
3753: PetscCall(PetscLogEventBegin(MAT_MatSolve, A, B, X, 0));
3754: if (!A->ops->matsolvetranspose) {
3755: PetscCall(PetscInfo(A, "Mat type %s using basic MatMatSolveTranspose\n", ((PetscObject)A)->type_name));
3756: PetscCall(MatMatSolve_Basic(A, B, X, PETSC_TRUE));
3757: } else PetscUseTypeMethod(A, matsolvetranspose, B, X);
3758: PetscCall(PetscLogEventEnd(MAT_MatSolve, A, B, X, 0));
3759: PetscCall(PetscObjectStateIncrease((PetscObject)X));
3760: PetscFunctionReturn(PETSC_SUCCESS);
3761: }
3763: /*@
3764: MatMatTransposeSolve - Solves A X = B^T, given a factored matrix.
3766: Neighbor-wise Collective
3768: Input Parameters:
3769: + A - the factored matrix
3770: - Bt - the transpose of right-hand-side matrix as a `MATDENSE`
3772: Output Parameter:
3773: . X - the result matrix (dense matrix)
3775: Level: developer
3777: Note:
3778: For MUMPS, it only supports centralized sparse compressed column format on the host processor for right hand side matrix. User must create B^T in sparse compressed row
3779: format on the host processor and call `MatMatTransposeSolve()` to implement MUMPS' `MatMatSolve()`.
3781: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatMatSolve()`, `MatMatSolveTranspose()`, `MatLUFactor()`, `MatCholeskyFactor()`
3782: @*/
3783: PetscErrorCode MatMatTransposeSolve(Mat A, Mat Bt, Mat X)
3784: {
3785: PetscFunctionBegin;
3790: PetscCheckSameComm(A, 1, Bt, 2);
3791: PetscCheckSameComm(A, 1, X, 3);
3793: PetscCheck(X != Bt, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_IDN, "X and B must be different matrices");
3794: PetscCheck(A->cmap->N == X->rmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat X: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->cmap->N, X->rmap->N);
3795: PetscCheck(A->rmap->N == Bt->cmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat Bt: global dim %" PetscInt_FMT " %" PetscInt_FMT, A->rmap->N, Bt->cmap->N);
3796: PetscCheck(X->cmap->N >= Bt->rmap->N, PetscObjectComm((PetscObject)X), PETSC_ERR_ARG_SIZ, "Solution matrix must have same number of columns as row number of the rhs matrix");
3797: if (!A->rmap->N && !A->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3798: PetscCheck(A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Unfactored matrix");
3799: MatCheckPreallocated(A, 1);
3801: PetscCall(PetscLogEventBegin(MAT_MatTrSolve, A, Bt, X, 0));
3802: PetscUseTypeMethod(A, mattransposesolve, Bt, X);
3803: PetscCall(PetscLogEventEnd(MAT_MatTrSolve, A, Bt, X, 0));
3804: PetscCall(PetscObjectStateIncrease((PetscObject)X));
3805: PetscFunctionReturn(PETSC_SUCCESS);
3806: }
3808: /*@
3809: MatForwardSolve - Solves L x = b, given a factored matrix, A = LU, or
3810: U^T*D^(1/2) x = b, given a factored symmetric matrix, A = U^T*D*U,
3812: Neighbor-wise Collective
3814: Input Parameters:
3815: + mat - the factored matrix
3816: - b - the right-hand-side vector
3818: Output Parameter:
3819: . x - the result vector
3821: Level: developer
3823: Notes:
3824: `MatSolve()` should be used for most applications, as it performs
3825: a forward solve followed by a backward solve.
3827: The vectors `b` and `x` cannot be the same, i.e., one cannot
3828: call `MatForwardSolve`(A,x,x).
3830: For matrix in `MATSEQBAIJ` format with block size larger than 1,
3831: the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3832: `MatForwardSolve()` solves U^T*D y = b, and
3833: `MatBackwardSolve()` solves U x = y.
3834: Thus they do not provide a symmetric preconditioner.
3836: .seealso: [](ch_matrices), `Mat`, `MatBackwardSolve()`, `MatGetFactor()`, `MatSolve()`, `MatBackwardSolve()`
3837: @*/
3838: PetscErrorCode MatForwardSolve(Mat mat, Vec b, Vec x)
3839: {
3840: PetscFunctionBegin;
3845: PetscCheckSameComm(mat, 1, b, 2);
3846: PetscCheckSameComm(mat, 1, x, 3);
3847: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
3848: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
3849: PetscCheck(mat->rmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, b->map->N);
3850: PetscCheck(mat->rmap->n == b->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, b->map->n);
3851: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3852: MatCheckPreallocated(mat, 1);
3854: PetscCall(PetscLogEventBegin(MAT_ForwardSolve, mat, b, x, 0));
3855: PetscUseTypeMethod(mat, forwardsolve, b, x);
3856: PetscCall(PetscLogEventEnd(MAT_ForwardSolve, mat, b, x, 0));
3857: PetscCall(PetscObjectStateIncrease((PetscObject)x));
3858: PetscFunctionReturn(PETSC_SUCCESS);
3859: }
3861: /*@
3862: MatBackwardSolve - Solves U x = b, given a factored matrix, A = LU.
3863: D^(1/2) U x = b, given a factored symmetric matrix, A = U^T*D*U,
3865: Neighbor-wise Collective
3867: Input Parameters:
3868: + mat - the factored matrix
3869: - b - the right-hand-side vector
3871: Output Parameter:
3872: . x - the result vector
3874: Level: developer
3876: Notes:
3877: `MatSolve()` should be used for most applications, as it performs
3878: a forward solve followed by a backward solve.
3880: The vectors `b` and `x` cannot be the same. I.e., one cannot
3881: call `MatBackwardSolve`(A,x,x).
3883: For matrix in `MATSEQBAIJ` format with block size larger than 1,
3884: the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3885: `MatForwardSolve()` solves U^T*D y = b, and
3886: `MatBackwardSolve()` solves U x = y.
3887: Thus they do not provide a symmetric preconditioner.
3889: .seealso: [](ch_matrices), `Mat`, `MatForwardSolve()`, `MatGetFactor()`, `MatSolve()`, `MatForwardSolve()`
3890: @*/
3891: PetscErrorCode MatBackwardSolve(Mat mat, Vec b, Vec x)
3892: {
3893: PetscFunctionBegin;
3898: PetscCheckSameComm(mat, 1, b, 2);
3899: PetscCheckSameComm(mat, 1, x, 3);
3900: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
3901: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
3902: PetscCheck(mat->rmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, b->map->N);
3903: PetscCheck(mat->rmap->n == b->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, b->map->n);
3904: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3905: MatCheckPreallocated(mat, 1);
3907: PetscCall(PetscLogEventBegin(MAT_BackwardSolve, mat, b, x, 0));
3908: PetscUseTypeMethod(mat, backwardsolve, b, x);
3909: PetscCall(PetscLogEventEnd(MAT_BackwardSolve, mat, b, x, 0));
3910: PetscCall(PetscObjectStateIncrease((PetscObject)x));
3911: PetscFunctionReturn(PETSC_SUCCESS);
3912: }
3914: /*@
3915: MatSolveAdd - Computes x = y + inv(A)*b, given a factored matrix.
3917: Neighbor-wise Collective
3919: Input Parameters:
3920: + mat - the factored matrix
3921: . b - the right-hand-side vector
3922: - y - the vector to be added to
3924: Output Parameter:
3925: . x - the result vector
3927: Level: developer
3929: Note:
3930: The vectors `b` and `x` cannot be the same. I.e., one cannot
3931: call `MatSolveAdd`(A,x,y,x).
3933: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatSolve()`, `MatGetFactor()`, `MatSolveTranspose()`, `MatSolveTransposeAdd()`
3934: @*/
3935: PetscErrorCode MatSolveAdd(Mat mat, Vec b, Vec y, Vec x)
3936: {
3937: PetscScalar one = 1.0;
3938: Vec tmp;
3940: PetscFunctionBegin;
3946: PetscCheckSameComm(mat, 1, b, 2);
3947: PetscCheckSameComm(mat, 1, y, 3);
3948: PetscCheckSameComm(mat, 1, x, 4);
3949: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
3950: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
3951: PetscCheck(mat->rmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, b->map->N);
3952: PetscCheck(mat->rmap->N == y->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, y->map->N);
3953: PetscCheck(mat->rmap->n == b->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, b->map->n);
3954: PetscCheck(x->map->n == y->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Vec x,Vec y: local dim %" PetscInt_FMT " %" PetscInt_FMT, x->map->n, y->map->n);
3955: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
3956: MatCheckPreallocated(mat, 1);
3958: PetscCall(PetscLogEventBegin(MAT_SolveAdd, mat, b, x, y));
3959: if (mat->factorerrortype) {
3960: PetscCall(PetscInfo(mat, "MatFactorError %d\n", mat->factorerrortype));
3961: PetscCall(VecSetInf(x));
3962: } else if (mat->ops->solveadd) {
3963: PetscUseTypeMethod(mat, solveadd, b, y, x);
3964: } else {
3965: /* do the solve then the add manually */
3966: if (x != y) {
3967: PetscCall(MatSolve(mat, b, x));
3968: PetscCall(VecAXPY(x, one, y));
3969: } else {
3970: PetscCall(VecDuplicate(x, &tmp));
3971: PetscCall(VecCopy(x, tmp));
3972: PetscCall(MatSolve(mat, b, x));
3973: PetscCall(VecAXPY(x, one, tmp));
3974: PetscCall(VecDestroy(&tmp));
3975: }
3976: }
3977: PetscCall(PetscLogEventEnd(MAT_SolveAdd, mat, b, x, y));
3978: PetscCall(PetscObjectStateIncrease((PetscObject)x));
3979: PetscFunctionReturn(PETSC_SUCCESS);
3980: }
3982: /*@
3983: MatSolveTranspose - Solves A' x = b, given a factored matrix.
3985: Neighbor-wise Collective
3987: Input Parameters:
3988: + mat - the factored matrix
3989: - b - the right-hand-side vector
3991: Output Parameter:
3992: . x - the result vector
3994: Level: developer
3996: Notes:
3997: The vectors `b` and `x` cannot be the same. I.e., one cannot
3998: call `MatSolveTranspose`(A,x,x).
4000: Most users should employ the `KSP` interface for linear solvers
4001: instead of working directly with matrix algebra routines such as this.
4002: See, e.g., `KSPCreate()`.
4004: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `KSP`, `MatSolve()`, `MatSolveAdd()`, `MatSolveTransposeAdd()`
4005: @*/
4006: PetscErrorCode MatSolveTranspose(Mat mat, Vec b, Vec x)
4007: {
4008: PetscErrorCode (*f)(Mat, Vec, Vec) = (!mat->ops->solvetranspose && mat->symmetric) ? mat->ops->solve : mat->ops->solvetranspose;
4010: PetscFunctionBegin;
4015: PetscCheckSameComm(mat, 1, b, 2);
4016: PetscCheckSameComm(mat, 1, x, 3);
4017: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
4018: PetscCheck(mat->rmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, x->map->N);
4019: PetscCheck(mat->cmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, b->map->N);
4020: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
4021: MatCheckPreallocated(mat, 1);
4022: PetscCall(PetscLogEventBegin(MAT_SolveTranspose, mat, b, x, 0));
4023: if (mat->factorerrortype) {
4024: PetscCall(PetscInfo(mat, "MatFactorError %d\n", mat->factorerrortype));
4025: PetscCall(VecSetInf(x));
4026: } else {
4027: PetscCheck(f, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Matrix type %s", ((PetscObject)mat)->type_name);
4028: PetscCall((*f)(mat, b, x));
4029: }
4030: PetscCall(PetscLogEventEnd(MAT_SolveTranspose, mat, b, x, 0));
4031: PetscCall(PetscObjectStateIncrease((PetscObject)x));
4032: PetscFunctionReturn(PETSC_SUCCESS);
4033: }
4035: /*@
4036: MatSolveTransposeAdd - Computes x = y + inv(Transpose(A)) b, given a
4037: factored matrix.
4039: Neighbor-wise Collective
4041: Input Parameters:
4042: + mat - the factored matrix
4043: . b - the right-hand-side vector
4044: - y - the vector to be added to
4046: Output Parameter:
4047: . x - the result vector
4049: Level: developer
4051: Note:
4052: The vectors `b` and `x` cannot be the same. I.e., one cannot
4053: call `MatSolveTransposeAdd`(A,x,y,x).
4055: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatSolve()`, `MatSolveAdd()`, `MatSolveTranspose()`
4056: @*/
4057: PetscErrorCode MatSolveTransposeAdd(Mat mat, Vec b, Vec y, Vec x)
4058: {
4059: PetscScalar one = 1.0;
4060: Vec tmp;
4061: PetscErrorCode (*f)(Mat, Vec, Vec, Vec) = (!mat->ops->solvetransposeadd && mat->symmetric) ? mat->ops->solveadd : mat->ops->solvetransposeadd;
4063: PetscFunctionBegin;
4069: PetscCheckSameComm(mat, 1, b, 2);
4070: PetscCheckSameComm(mat, 1, y, 3);
4071: PetscCheckSameComm(mat, 1, x, 4);
4072: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
4073: PetscCheck(mat->rmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, x->map->N);
4074: PetscCheck(mat->cmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, b->map->N);
4075: PetscCheck(mat->cmap->N == y->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec y: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, y->map->N);
4076: PetscCheck(x->map->n == y->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Vec x,Vec y: local dim %" PetscInt_FMT " %" PetscInt_FMT, x->map->n, y->map->n);
4077: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
4078: MatCheckPreallocated(mat, 1);
4080: PetscCall(PetscLogEventBegin(MAT_SolveTransposeAdd, mat, b, x, y));
4081: if (mat->factorerrortype) {
4082: PetscCall(PetscInfo(mat, "MatFactorError %d\n", mat->factorerrortype));
4083: PetscCall(VecSetInf(x));
4084: } else if (f) {
4085: PetscCall((*f)(mat, b, y, x));
4086: } else {
4087: /* do the solve then the add manually */
4088: if (x != y) {
4089: PetscCall(MatSolveTranspose(mat, b, x));
4090: PetscCall(VecAXPY(x, one, y));
4091: } else {
4092: PetscCall(VecDuplicate(x, &tmp));
4093: PetscCall(VecCopy(x, tmp));
4094: PetscCall(MatSolveTranspose(mat, b, x));
4095: PetscCall(VecAXPY(x, one, tmp));
4096: PetscCall(VecDestroy(&tmp));
4097: }
4098: }
4099: PetscCall(PetscLogEventEnd(MAT_SolveTransposeAdd, mat, b, x, y));
4100: PetscCall(PetscObjectStateIncrease((PetscObject)x));
4101: PetscFunctionReturn(PETSC_SUCCESS);
4102: }
4104: /*@
4105: MatSOR - Computes relaxation (SOR, Gauss-Seidel) sweeps.
4107: Neighbor-wise Collective
4109: Input Parameters:
4110: + mat - the matrix
4111: . b - the right hand side
4112: . omega - the relaxation factor
4113: . flag - flag indicating the type of SOR (see below)
4114: . shift - diagonal shift
4115: . its - the number of iterations
4116: - lits - the number of local iterations
4118: Output Parameter:
4119: . x - the solution (can contain an initial guess, use option `SOR_ZERO_INITIAL_GUESS` to indicate no guess)
4121: SOR Flags:
4122: + `SOR_FORWARD_SWEEP` - forward SOR
4123: . `SOR_BACKWARD_SWEEP` - backward SOR
4124: . `SOR_SYMMETRIC_SWEEP` - SSOR (symmetric SOR)
4125: . `SOR_LOCAL_FORWARD_SWEEP` - local forward SOR
4126: . `SOR_LOCAL_BACKWARD_SWEEP` - local forward SOR
4127: . `SOR_LOCAL_SYMMETRIC_SWEEP` - local SSOR
4128: . `SOR_EISENSTAT` - SOR with Eisenstat trick
4129: . `SOR_APPLY_UPPER`, `SOR_APPLY_LOWER` - applies
4130: upper/lower triangular part of matrix to
4131: vector (with omega)
4132: - `SOR_ZERO_INITIAL_GUESS` - zero initial guess
4134: Level: developer
4136: Notes:
4137: `SOR_LOCAL_FORWARD_SWEEP`, `SOR_LOCAL_BACKWARD_SWEEP`, and
4138: `SOR_LOCAL_SYMMETRIC_SWEEP` perform separate independent smoothings
4139: on each processor.
4141: Application programmers will not generally use `MatSOR()` directly,
4142: but instead will employ the `KSP`/`PC` interface.
4144: For `MATBAIJ`, `MATSBAIJ`, and `MATAIJ` matrices with Inodes this does a block SOR smoothing, otherwise it does a pointwise smoothing
4146: Most users should employ the `KSP` interface for linear solvers
4147: instead of working directly with matrix algebra routines such as this.
4148: See, e.g., `KSPCreate()`.
4150: Vectors `x` and `b` CANNOT be the same
4152: The flags are implemented as bitwise inclusive or operations.
4153: For example, use (`SOR_ZERO_INITIAL_GUESS` | `SOR_SYMMETRIC_SWEEP`)
4154: to specify a zero initial guess for SSOR.
4156: Developer Note:
4157: We should add block SOR support for `MATAIJ` matrices with block size set to great than one and no inodes
4159: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `KSP`, `PC`, `MatGetFactor()`
4160: @*/
4161: PetscErrorCode MatSOR(Mat mat, Vec b, PetscReal omega, MatSORType flag, PetscReal shift, PetscInt its, PetscInt lits, Vec x)
4162: {
4163: PetscFunctionBegin;
4168: PetscCheckSameComm(mat, 1, b, 2);
4169: PetscCheckSameComm(mat, 1, x, 8);
4170: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4171: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4172: PetscCheck(mat->cmap->N == x->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec x: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->cmap->N, x->map->N);
4173: PetscCheck(mat->rmap->N == b->map->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: global dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->N, b->map->N);
4174: PetscCheck(mat->rmap->n == b->map->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Mat mat,Vec b: local dim %" PetscInt_FMT " %" PetscInt_FMT, mat->rmap->n, b->map->n);
4175: PetscCheck(its > 0, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Relaxation requires global its %" PetscInt_FMT " positive", its);
4176: PetscCheck(lits > 0, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Relaxation requires local its %" PetscInt_FMT " positive", lits);
4177: PetscCheck(b != x, PETSC_COMM_SELF, PETSC_ERR_ARG_IDN, "b and x vector cannot be the same");
4179: MatCheckPreallocated(mat, 1);
4180: PetscCall(PetscLogEventBegin(MAT_SOR, mat, b, x, 0));
4181: PetscUseTypeMethod(mat, sor, b, omega, flag, shift, its, lits, x);
4182: PetscCall(PetscLogEventEnd(MAT_SOR, mat, b, x, 0));
4183: PetscCall(PetscObjectStateIncrease((PetscObject)x));
4184: PetscFunctionReturn(PETSC_SUCCESS);
4185: }
4187: /*
4188: Default matrix copy routine.
4189: */
4190: PetscErrorCode MatCopy_Basic(Mat A, Mat B, MatStructure str)
4191: {
4192: PetscInt i, rstart = 0, rend = 0, nz;
4193: const PetscInt *cwork;
4194: const PetscScalar *vwork;
4196: PetscFunctionBegin;
4197: if (B->assembled) PetscCall(MatZeroEntries(B));
4198: if (str == SAME_NONZERO_PATTERN) {
4199: PetscCall(MatGetOwnershipRange(A, &rstart, &rend));
4200: for (i = rstart; i < rend; i++) {
4201: PetscCall(MatGetRow(A, i, &nz, &cwork, &vwork));
4202: PetscCall(MatSetValues(B, 1, &i, nz, cwork, vwork, INSERT_VALUES));
4203: PetscCall(MatRestoreRow(A, i, &nz, &cwork, &vwork));
4204: }
4205: } else {
4206: PetscCall(MatAYPX(B, 0.0, A, str));
4207: }
4208: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
4209: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
4210: PetscFunctionReturn(PETSC_SUCCESS);
4211: }
4213: /*@
4214: MatCopy - Copies a matrix to another matrix.
4216: Collective
4218: Input Parameters:
4219: + A - the matrix
4220: - str - `SAME_NONZERO_PATTERN` or `DIFFERENT_NONZERO_PATTERN`
4222: Output Parameter:
4223: . B - where the copy is put
4225: Level: intermediate
4227: Notes:
4228: If you use `SAME_NONZERO_PATTERN` then the two matrices must have the same nonzero pattern or the routine will crash.
4230: `MatCopy()` copies the matrix entries of a matrix to another existing
4231: matrix (after first zeroing the second matrix). A related routine is
4232: `MatConvert()`, which first creates a new matrix and then copies the data.
4234: .seealso: [](ch_matrices), `Mat`, `MatConvert()`, `MatDuplicate()`
4235: @*/
4236: PetscErrorCode MatCopy(Mat A, Mat B, MatStructure str)
4237: {
4238: PetscInt i;
4240: PetscFunctionBegin;
4245: PetscCheckSameComm(A, 1, B, 2);
4246: MatCheckPreallocated(B, 2);
4247: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4248: PetscCheck(!A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4249: PetscCheck(A->rmap->N == B->rmap->N && A->cmap->N == B->cmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat B: global dim (%" PetscInt_FMT ",%" PetscInt_FMT ") (%" PetscInt_FMT ",%" PetscInt_FMT ")", A->rmap->N, B->rmap->N,
4250: A->cmap->N, B->cmap->N);
4251: MatCheckPreallocated(A, 1);
4252: if (A == B) PetscFunctionReturn(PETSC_SUCCESS);
4254: PetscCall(PetscLogEventBegin(MAT_Copy, A, B, 0, 0));
4255: if (A->ops->copy) PetscUseTypeMethod(A, copy, B, str);
4256: else PetscCall(MatCopy_Basic(A, B, str));
4258: B->stencil.dim = A->stencil.dim;
4259: B->stencil.noc = A->stencil.noc;
4260: for (i = 0; i <= A->stencil.dim; i++) {
4261: B->stencil.dims[i] = A->stencil.dims[i];
4262: B->stencil.starts[i] = A->stencil.starts[i];
4263: }
4265: PetscCall(PetscLogEventEnd(MAT_Copy, A, B, 0, 0));
4266: PetscCall(PetscObjectStateIncrease((PetscObject)B));
4267: PetscFunctionReturn(PETSC_SUCCESS);
4268: }
4270: /*@C
4271: MatConvert - Converts a matrix to another matrix, either of the same
4272: or different type.
4274: Collective
4276: Input Parameters:
4277: + mat - the matrix
4278: . newtype - new matrix type. Use `MATSAME` to create a new matrix of the
4279: same type as the original matrix.
4280: - reuse - denotes if the destination matrix is to be created or reused.
4281: Use `MAT_INPLACE_MATRIX` for inplace conversion (that is when you want the input mat to be changed to contain the matrix in the new format), otherwise use
4282: `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX` (can only be used after the first call was made with `MAT_INITIAL_MATRIX`, causes the matrix space in M to be reused).
4284: Output Parameter:
4285: . M - pointer to place new matrix
4287: Level: intermediate
4289: Notes:
4290: `MatConvert()` first creates a new matrix and then copies the data from
4291: the first matrix. A related routine is `MatCopy()`, which copies the matrix
4292: entries of one matrix to another already existing matrix context.
4294: Cannot be used to convert a sequential matrix to parallel or parallel to sequential,
4295: the MPI communicator of the generated matrix is always the same as the communicator
4296: of the input matrix.
4298: .seealso: [](ch_matrices), `Mat`, `MatCopy()`, `MatDuplicate()`, `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, `MAT_INPLACE_MATRIX`
4299: @*/
4300: PetscErrorCode MatConvert(Mat mat, MatType newtype, MatReuse reuse, Mat *M)
4301: {
4302: PetscBool sametype, issame, flg;
4303: PetscBool3 issymmetric, ishermitian;
4304: char convname[256], mtype[256];
4305: Mat B;
4307: PetscFunctionBegin;
4311: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4312: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4313: MatCheckPreallocated(mat, 1);
4315: PetscCall(PetscOptionsGetString(((PetscObject)mat)->options, ((PetscObject)mat)->prefix, "-matconvert_type", mtype, sizeof(mtype), &flg));
4316: if (flg) newtype = mtype;
4318: PetscCall(PetscObjectTypeCompare((PetscObject)mat, newtype, &sametype));
4319: PetscCall(PetscStrcmp(newtype, "same", &issame));
4320: PetscCheck(!(reuse == MAT_INPLACE_MATRIX) || !(mat != *M), PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "MAT_INPLACE_MATRIX requires same input and output matrix");
4321: PetscCheck(!(reuse == MAT_REUSE_MATRIX) || !(mat == *M), PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "MAT_REUSE_MATRIX means reuse matrix in final argument, perhaps you mean MAT_INPLACE_MATRIX");
4323: if ((reuse == MAT_INPLACE_MATRIX) && (issame || sametype)) {
4324: PetscCall(PetscInfo(mat, "Early return for inplace %s %d %d\n", ((PetscObject)mat)->type_name, sametype, issame));
4325: PetscFunctionReturn(PETSC_SUCCESS);
4326: }
4328: /* Cache Mat options because some converters use MatHeaderReplace */
4329: issymmetric = mat->symmetric;
4330: ishermitian = mat->hermitian;
4332: if ((sametype || issame) && (reuse == MAT_INITIAL_MATRIX) && mat->ops->duplicate) {
4333: PetscCall(PetscInfo(mat, "Calling duplicate for initial matrix %s %d %d\n", ((PetscObject)mat)->type_name, sametype, issame));
4334: PetscUseTypeMethod(mat, duplicate, MAT_COPY_VALUES, M);
4335: } else {
4336: PetscErrorCode (*conv)(Mat, MatType, MatReuse, Mat *) = NULL;
4337: const char *prefix[3] = {"seq", "mpi", ""};
4338: PetscInt i;
4339: /*
4340: Order of precedence:
4341: 0) See if newtype is a superclass of the current matrix.
4342: 1) See if a specialized converter is known to the current matrix.
4343: 2) See if a specialized converter is known to the desired matrix class.
4344: 3) See if a good general converter is registered for the desired class
4345: (as of 6/27/03 only MATMPIADJ falls into this category).
4346: 4) See if a good general converter is known for the current matrix.
4347: 5) Use a really basic converter.
4348: */
4350: /* 0) See if newtype is a superclass of the current matrix.
4351: i.e mat is mpiaij and newtype is aij */
4352: for (i = 0; i < 2; i++) {
4353: PetscCall(PetscStrncpy(convname, prefix[i], sizeof(convname)));
4354: PetscCall(PetscStrlcat(convname, newtype, sizeof(convname)));
4355: PetscCall(PetscStrcmp(convname, ((PetscObject)mat)->type_name, &flg));
4356: PetscCall(PetscInfo(mat, "Check superclass %s %s -> %d\n", convname, ((PetscObject)mat)->type_name, flg));
4357: if (flg) {
4358: if (reuse == MAT_INPLACE_MATRIX) {
4359: PetscCall(PetscInfo(mat, "Early return\n"));
4360: PetscFunctionReturn(PETSC_SUCCESS);
4361: } else if (reuse == MAT_INITIAL_MATRIX && mat->ops->duplicate) {
4362: PetscCall(PetscInfo(mat, "Calling MatDuplicate\n"));
4363: PetscUseTypeMethod(mat, duplicate, MAT_COPY_VALUES, M);
4364: PetscFunctionReturn(PETSC_SUCCESS);
4365: } else if (reuse == MAT_REUSE_MATRIX && mat->ops->copy) {
4366: PetscCall(PetscInfo(mat, "Calling MatCopy\n"));
4367: PetscCall(MatCopy(mat, *M, SAME_NONZERO_PATTERN));
4368: PetscFunctionReturn(PETSC_SUCCESS);
4369: }
4370: }
4371: }
4372: /* 1) See if a specialized converter is known to the current matrix and the desired class */
4373: for (i = 0; i < 3; i++) {
4374: PetscCall(PetscStrncpy(convname, "MatConvert_", sizeof(convname)));
4375: PetscCall(PetscStrlcat(convname, ((PetscObject)mat)->type_name, sizeof(convname)));
4376: PetscCall(PetscStrlcat(convname, "_", sizeof(convname)));
4377: PetscCall(PetscStrlcat(convname, prefix[i], sizeof(convname)));
4378: PetscCall(PetscStrlcat(convname, issame ? ((PetscObject)mat)->type_name : newtype, sizeof(convname)));
4379: PetscCall(PetscStrlcat(convname, "_C", sizeof(convname)));
4380: PetscCall(PetscObjectQueryFunction((PetscObject)mat, convname, &conv));
4381: PetscCall(PetscInfo(mat, "Check specialized (1) %s (%s) -> %d\n", convname, ((PetscObject)mat)->type_name, !!conv));
4382: if (conv) goto foundconv;
4383: }
4385: /* 2) See if a specialized converter is known to the desired matrix class. */
4386: PetscCall(MatCreate(PetscObjectComm((PetscObject)mat), &B));
4387: PetscCall(MatSetSizes(B, mat->rmap->n, mat->cmap->n, mat->rmap->N, mat->cmap->N));
4388: PetscCall(MatSetType(B, newtype));
4389: for (i = 0; i < 3; i++) {
4390: PetscCall(PetscStrncpy(convname, "MatConvert_", sizeof(convname)));
4391: PetscCall(PetscStrlcat(convname, ((PetscObject)mat)->type_name, sizeof(convname)));
4392: PetscCall(PetscStrlcat(convname, "_", sizeof(convname)));
4393: PetscCall(PetscStrlcat(convname, prefix[i], sizeof(convname)));
4394: PetscCall(PetscStrlcat(convname, newtype, sizeof(convname)));
4395: PetscCall(PetscStrlcat(convname, "_C", sizeof(convname)));
4396: PetscCall(PetscObjectQueryFunction((PetscObject)B, convname, &conv));
4397: PetscCall(PetscInfo(mat, "Check specialized (2) %s (%s) -> %d\n", convname, ((PetscObject)B)->type_name, !!conv));
4398: if (conv) {
4399: PetscCall(MatDestroy(&B));
4400: goto foundconv;
4401: }
4402: }
4404: /* 3) See if a good general converter is registered for the desired class */
4405: conv = B->ops->convertfrom;
4406: PetscCall(PetscInfo(mat, "Check convertfrom (%s) -> %d\n", ((PetscObject)B)->type_name, !!conv));
4407: PetscCall(MatDestroy(&B));
4408: if (conv) goto foundconv;
4410: /* 4) See if a good general converter is known for the current matrix */
4411: if (mat->ops->convert) conv = mat->ops->convert;
4412: PetscCall(PetscInfo(mat, "Check general convert (%s) -> %d\n", ((PetscObject)mat)->type_name, !!conv));
4413: if (conv) goto foundconv;
4415: /* 5) Use a really basic converter. */
4416: PetscCall(PetscInfo(mat, "Using MatConvert_Basic\n"));
4417: conv = MatConvert_Basic;
4419: foundconv:
4420: PetscCall(PetscLogEventBegin(MAT_Convert, mat, 0, 0, 0));
4421: PetscCall((*conv)(mat, newtype, reuse, M));
4422: if (mat->rmap->mapping && mat->cmap->mapping && !(*M)->rmap->mapping && !(*M)->cmap->mapping) {
4423: /* the block sizes must be same if the mappings are copied over */
4424: (*M)->rmap->bs = mat->rmap->bs;
4425: (*M)->cmap->bs = mat->cmap->bs;
4426: PetscCall(PetscObjectReference((PetscObject)mat->rmap->mapping));
4427: PetscCall(PetscObjectReference((PetscObject)mat->cmap->mapping));
4428: (*M)->rmap->mapping = mat->rmap->mapping;
4429: (*M)->cmap->mapping = mat->cmap->mapping;
4430: }
4431: (*M)->stencil.dim = mat->stencil.dim;
4432: (*M)->stencil.noc = mat->stencil.noc;
4433: for (i = 0; i <= mat->stencil.dim; i++) {
4434: (*M)->stencil.dims[i] = mat->stencil.dims[i];
4435: (*M)->stencil.starts[i] = mat->stencil.starts[i];
4436: }
4437: PetscCall(PetscLogEventEnd(MAT_Convert, mat, 0, 0, 0));
4438: }
4439: PetscCall(PetscObjectStateIncrease((PetscObject)*M));
4441: /* Copy Mat options */
4442: if (issymmetric == PETSC_BOOL3_TRUE) PetscCall(MatSetOption(*M, MAT_SYMMETRIC, PETSC_TRUE));
4443: else if (issymmetric == PETSC_BOOL3_FALSE) PetscCall(MatSetOption(*M, MAT_SYMMETRIC, PETSC_FALSE));
4444: if (ishermitian == PETSC_BOOL3_TRUE) PetscCall(MatSetOption(*M, MAT_HERMITIAN, PETSC_TRUE));
4445: else if (ishermitian == PETSC_BOOL3_FALSE) PetscCall(MatSetOption(*M, MAT_HERMITIAN, PETSC_FALSE));
4446: PetscFunctionReturn(PETSC_SUCCESS);
4447: }
4449: /*@C
4450: MatFactorGetSolverType - Returns name of the package providing the factorization routines
4452: Not Collective
4454: Input Parameter:
4455: . mat - the matrix, must be a factored matrix
4457: Output Parameter:
4458: . type - the string name of the package (do not free this string)
4460: Level: intermediate
4462: Fortran Note:
4463: Pass in an empty string and the package name will be copied into it. Make sure the string is long enough.
4465: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatSolverType`, `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatGetFactor()`
4466: @*/
4467: PetscErrorCode MatFactorGetSolverType(Mat mat, MatSolverType *type)
4468: {
4469: PetscErrorCode (*conv)(Mat, MatSolverType *);
4471: PetscFunctionBegin;
4475: PetscCheck(mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Only for factored matrix");
4476: PetscCall(PetscObjectQueryFunction((PetscObject)mat, "MatFactorGetSolverType_C", &conv));
4477: if (conv) PetscCall((*conv)(mat, type));
4478: else *type = MATSOLVERPETSC;
4479: PetscFunctionReturn(PETSC_SUCCESS);
4480: }
4482: typedef struct _MatSolverTypeForSpecifcType *MatSolverTypeForSpecifcType;
4483: struct _MatSolverTypeForSpecifcType {
4484: MatType mtype;
4485: /* no entry for MAT_FACTOR_NONE */
4486: PetscErrorCode (*createfactor[MAT_FACTOR_NUM_TYPES - 1])(Mat, MatFactorType, Mat *);
4487: MatSolverTypeForSpecifcType next;
4488: };
4490: typedef struct _MatSolverTypeHolder *MatSolverTypeHolder;
4491: struct _MatSolverTypeHolder {
4492: char *name;
4493: MatSolverTypeForSpecifcType handlers;
4494: MatSolverTypeHolder next;
4495: };
4497: static MatSolverTypeHolder MatSolverTypeHolders = NULL;
4499: /*@C
4500: MatSolverTypeRegister - Registers a `MatSolverType` that works for a particular matrix type
4502: Input Parameters:
4503: + package - name of the package, for example petsc or superlu
4504: . mtype - the matrix type that works with this package
4505: . ftype - the type of factorization supported by the package
4506: - createfactor - routine that will create the factored matrix ready to be used
4508: Level: developer
4510: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorGetSolverType()`, `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatGetFactor()`
4511: @*/
4512: PetscErrorCode MatSolverTypeRegister(MatSolverType package, MatType mtype, MatFactorType ftype, PetscErrorCode (*createfactor)(Mat, MatFactorType, Mat *))
4513: {
4514: MatSolverTypeHolder next = MatSolverTypeHolders, prev = NULL;
4515: PetscBool flg;
4516: MatSolverTypeForSpecifcType inext, iprev = NULL;
4518: PetscFunctionBegin;
4519: PetscCall(MatInitializePackage());
4520: if (!next) {
4521: PetscCall(PetscNew(&MatSolverTypeHolders));
4522: PetscCall(PetscStrallocpy(package, &MatSolverTypeHolders->name));
4523: PetscCall(PetscNew(&MatSolverTypeHolders->handlers));
4524: PetscCall(PetscStrallocpy(mtype, (char **)&MatSolverTypeHolders->handlers->mtype));
4525: MatSolverTypeHolders->handlers->createfactor[(int)ftype - 1] = createfactor;
4526: PetscFunctionReturn(PETSC_SUCCESS);
4527: }
4528: while (next) {
4529: PetscCall(PetscStrcasecmp(package, next->name, &flg));
4530: if (flg) {
4531: PetscCheck(next->handlers, PETSC_COMM_SELF, PETSC_ERR_PLIB, "MatSolverTypeHolder is missing handlers");
4532: inext = next->handlers;
4533: while (inext) {
4534: PetscCall(PetscStrcasecmp(mtype, inext->mtype, &flg));
4535: if (flg) {
4536: inext->createfactor[(int)ftype - 1] = createfactor;
4537: PetscFunctionReturn(PETSC_SUCCESS);
4538: }
4539: iprev = inext;
4540: inext = inext->next;
4541: }
4542: PetscCall(PetscNew(&iprev->next));
4543: PetscCall(PetscStrallocpy(mtype, (char **)&iprev->next->mtype));
4544: iprev->next->createfactor[(int)ftype - 1] = createfactor;
4545: PetscFunctionReturn(PETSC_SUCCESS);
4546: }
4547: prev = next;
4548: next = next->next;
4549: }
4550: PetscCall(PetscNew(&prev->next));
4551: PetscCall(PetscStrallocpy(package, &prev->next->name));
4552: PetscCall(PetscNew(&prev->next->handlers));
4553: PetscCall(PetscStrallocpy(mtype, (char **)&prev->next->handlers->mtype));
4554: prev->next->handlers->createfactor[(int)ftype - 1] = createfactor;
4555: PetscFunctionReturn(PETSC_SUCCESS);
4556: }
4558: /*@C
4559: MatSolverTypeGet - Gets the function that creates the factor matrix if it exist
4561: Input Parameters:
4562: + type - name of the package, for example petsc or superlu
4563: . ftype - the type of factorization supported by the type
4564: - mtype - the matrix type that works with this type
4566: Output Parameters:
4567: + foundtype - `PETSC_TRUE` if the type was registered
4568: . foundmtype - `PETSC_TRUE` if the type supports the requested mtype
4569: - createfactor - routine that will create the factored matrix ready to be used or `NULL` if not found
4571: Level: developer
4573: .seealso: [](ch_matrices), `Mat`, `MatFactorType`, `MatType`, `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatSolverTypeRegister()`, `MatGetFactor()`
4574: @*/
4575: PetscErrorCode MatSolverTypeGet(MatSolverType type, MatType mtype, MatFactorType ftype, PetscBool *foundtype, PetscBool *foundmtype, PetscErrorCode (**createfactor)(Mat, MatFactorType, Mat *))
4576: {
4577: MatSolverTypeHolder next = MatSolverTypeHolders;
4578: PetscBool flg;
4579: MatSolverTypeForSpecifcType inext;
4581: PetscFunctionBegin;
4582: if (foundtype) *foundtype = PETSC_FALSE;
4583: if (foundmtype) *foundmtype = PETSC_FALSE;
4584: if (createfactor) *createfactor = NULL;
4586: if (type) {
4587: while (next) {
4588: PetscCall(PetscStrcasecmp(type, next->name, &flg));
4589: if (flg) {
4590: if (foundtype) *foundtype = PETSC_TRUE;
4591: inext = next->handlers;
4592: while (inext) {
4593: PetscCall(PetscStrbeginswith(mtype, inext->mtype, &flg));
4594: if (flg) {
4595: if (foundmtype) *foundmtype = PETSC_TRUE;
4596: if (createfactor) *createfactor = inext->createfactor[(int)ftype - 1];
4597: PetscFunctionReturn(PETSC_SUCCESS);
4598: }
4599: inext = inext->next;
4600: }
4601: }
4602: next = next->next;
4603: }
4604: } else {
4605: while (next) {
4606: inext = next->handlers;
4607: while (inext) {
4608: PetscCall(PetscStrcmp(mtype, inext->mtype, &flg));
4609: if (flg && inext->createfactor[(int)ftype - 1]) {
4610: if (foundtype) *foundtype = PETSC_TRUE;
4611: if (foundmtype) *foundmtype = PETSC_TRUE;
4612: if (createfactor) *createfactor = inext->createfactor[(int)ftype - 1];
4613: PetscFunctionReturn(PETSC_SUCCESS);
4614: }
4615: inext = inext->next;
4616: }
4617: next = next->next;
4618: }
4619: /* try with base classes inext->mtype */
4620: next = MatSolverTypeHolders;
4621: while (next) {
4622: inext = next->handlers;
4623: while (inext) {
4624: PetscCall(PetscStrbeginswith(mtype, inext->mtype, &flg));
4625: if (flg && inext->createfactor[(int)ftype - 1]) {
4626: if (foundtype) *foundtype = PETSC_TRUE;
4627: if (foundmtype) *foundmtype = PETSC_TRUE;
4628: if (createfactor) *createfactor = inext->createfactor[(int)ftype - 1];
4629: PetscFunctionReturn(PETSC_SUCCESS);
4630: }
4631: inext = inext->next;
4632: }
4633: next = next->next;
4634: }
4635: }
4636: PetscFunctionReturn(PETSC_SUCCESS);
4637: }
4639: PetscErrorCode MatSolverTypeDestroy(void)
4640: {
4641: MatSolverTypeHolder next = MatSolverTypeHolders, prev;
4642: MatSolverTypeForSpecifcType inext, iprev;
4644: PetscFunctionBegin;
4645: while (next) {
4646: PetscCall(PetscFree(next->name));
4647: inext = next->handlers;
4648: while (inext) {
4649: PetscCall(PetscFree(inext->mtype));
4650: iprev = inext;
4651: inext = inext->next;
4652: PetscCall(PetscFree(iprev));
4653: }
4654: prev = next;
4655: next = next->next;
4656: PetscCall(PetscFree(prev));
4657: }
4658: MatSolverTypeHolders = NULL;
4659: PetscFunctionReturn(PETSC_SUCCESS);
4660: }
4662: /*@C
4663: MatFactorGetCanUseOrdering - Indicates if the factorization can use the ordering provided in `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`
4665: Logically Collective
4667: Input Parameter:
4668: . mat - the matrix
4670: Output Parameter:
4671: . flg - `PETSC_TRUE` if uses the ordering
4673: Level: developer
4675: Note:
4676: Most internal PETSc factorizations use the ordering passed to the factorization routine but external
4677: packages do not, thus we want to skip generating the ordering when it is not needed or used.
4679: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`
4680: @*/
4681: PetscErrorCode MatFactorGetCanUseOrdering(Mat mat, PetscBool *flg)
4682: {
4683: PetscFunctionBegin;
4684: *flg = mat->canuseordering;
4685: PetscFunctionReturn(PETSC_SUCCESS);
4686: }
4688: /*@C
4689: MatFactorGetPreferredOrdering - The preferred ordering for a particular matrix factor object
4691: Logically Collective
4693: Input Parameters:
4694: + mat - the matrix obtained with `MatGetFactor()`
4695: - ftype - the factorization type to be used
4697: Output Parameter:
4698: . otype - the preferred ordering type
4700: Level: developer
4702: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatFactorType`, `MatOrderingType`, `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatCholeskyFactorSymbolic()`
4703: @*/
4704: PetscErrorCode MatFactorGetPreferredOrdering(Mat mat, MatFactorType ftype, MatOrderingType *otype)
4705: {
4706: PetscFunctionBegin;
4707: *otype = mat->preferredordering[ftype];
4708: PetscCheck(*otype, PETSC_COMM_SELF, PETSC_ERR_PLIB, "MatFactor did not have a preferred ordering");
4709: PetscFunctionReturn(PETSC_SUCCESS);
4710: }
4712: /*@C
4713: MatGetFactor - Returns a matrix suitable to calls to MatXXFactorSymbolic()
4715: Collective
4717: Input Parameters:
4718: + mat - the matrix
4719: . type - name of solver type, for example, superlu, petsc (to use PETSc's default)
4720: - ftype - factor type, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ICC`, `MAT_FACTOR_ILU`, `MAT_FACTOR_QR`
4722: Output Parameter:
4723: . f - the factor matrix used with MatXXFactorSymbolic() calls
4725: Options Database Key:
4726: . -mat_factor_bind_factorization <host, device> - Where to do matrix factorization? Default is device (might consume more device memory.
4727: One can choose host to save device memory). Currently only supported with `MATSEQAIJCUSPARSE` matrices.
4729: Level: intermediate
4731: Notes:
4732: Users usually access the factorization solvers via `KSP`
4734: Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
4735: such as pastix, superlu, mumps etc.
4737: PETSc must have been ./configure to use the external solver, using the option --download-package
4739: Some of the packages have options for controlling the factorization, these are in the form -prefix_mat_packagename_packageoption
4740: where prefix is normally obtained from the calling `KSP`/`PC`. If `MatGetFactor()` is called directly one can set
4741: call `MatSetOptionsPrefixFactor()` on the originating matrix or `MatSetOptionsPrefix()` on the resulting factor matrix.
4743: Developer Note:
4744: This should actually be called `MatCreateFactor()` since it creates a new factor object
4746: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `KSP`, `MatSolverType`, `MatFactorType`, `MatCopy()`, `MatDuplicate()`, `MatGetFactorAvailable()`, `MatFactorGetCanUseOrdering()`, `MatSolverTypeRegister()`,
4747: `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ICC`, `MAT_FACTOR_ILU`, `MAT_FACTOR_QR`
4748: @*/
4749: PetscErrorCode MatGetFactor(Mat mat, MatSolverType type, MatFactorType ftype, Mat *f)
4750: {
4751: PetscBool foundtype, foundmtype;
4752: PetscErrorCode (*conv)(Mat, MatFactorType, Mat *);
4754: PetscFunctionBegin;
4758: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4759: MatCheckPreallocated(mat, 1);
4761: PetscCall(MatSolverTypeGet(type, ((PetscObject)mat)->type_name, ftype, &foundtype, &foundmtype, &conv));
4762: if (!foundtype) {
4763: if (type) {
4764: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_MISSING_FACTOR, "Could not locate solver type %s for factorization type %s and matrix type %s. Perhaps you must ./configure with --download-%s", type, MatFactorTypes[ftype],
4765: ((PetscObject)mat)->type_name, type);
4766: } else {
4767: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_MISSING_FACTOR, "Could not locate a solver type for factorization type %s and matrix type %s.", MatFactorTypes[ftype], ((PetscObject)mat)->type_name);
4768: }
4769: }
4770: PetscCheck(foundmtype, PetscObjectComm((PetscObject)mat), PETSC_ERR_MISSING_FACTOR, "MatSolverType %s does not support matrix type %s", type, ((PetscObject)mat)->type_name);
4771: PetscCheck(conv, PetscObjectComm((PetscObject)mat), PETSC_ERR_MISSING_FACTOR, "MatSolverType %s does not support factorization type %s for matrix type %s", type, MatFactorTypes[ftype], ((PetscObject)mat)->type_name);
4773: PetscCall((*conv)(mat, ftype, f));
4774: if (mat->factorprefix) PetscCall(MatSetOptionsPrefix(*f, mat->factorprefix));
4775: PetscFunctionReturn(PETSC_SUCCESS);
4776: }
4778: /*@C
4779: MatGetFactorAvailable - Returns a a flag if matrix supports particular type and factor type
4781: Not Collective
4783: Input Parameters:
4784: + mat - the matrix
4785: . type - name of solver type, for example, superlu, petsc (to use PETSc's default)
4786: - ftype - factor type, `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ICC`, `MAT_FACTOR_ILU`, `MAT_FACTOR_QR`
4788: Output Parameter:
4789: . flg - PETSC_TRUE if the factorization is available
4791: Level: intermediate
4793: Notes:
4794: Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
4795: such as pastix, superlu, mumps etc.
4797: PETSc must have been ./configure to use the external solver, using the option --download-package
4799: Developer Note:
4800: This should actually be called MatCreateFactorAvailable() since MatGetFactor() creates a new factor object
4802: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatSolverType`, `MatFactorType`, `MatGetFactor()`, `MatCopy()`, `MatDuplicate()`, `MatGetFactor()`, `MatSolverTypeRegister()`,
4803: `MAT_FACTOR_LU`, `MAT_FACTOR_CHOLESKY`, `MAT_FACTOR_ICC`, `MAT_FACTOR_ILU`, `MAT_FACTOR_QR`
4804: @*/
4805: PetscErrorCode MatGetFactorAvailable(Mat mat, MatSolverType type, MatFactorType ftype, PetscBool *flg)
4806: {
4807: PetscErrorCode (*gconv)(Mat, MatFactorType, Mat *);
4809: PetscFunctionBegin;
4814: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4815: MatCheckPreallocated(mat, 1);
4817: PetscCall(MatSolverTypeGet(type, ((PetscObject)mat)->type_name, ftype, NULL, NULL, &gconv));
4818: *flg = gconv ? PETSC_TRUE : PETSC_FALSE;
4819: PetscFunctionReturn(PETSC_SUCCESS);
4820: }
4822: /*@
4823: MatDuplicate - Duplicates a matrix including the non-zero structure.
4825: Collective
4827: Input Parameters:
4828: + mat - the matrix
4829: - op - One of `MAT_DO_NOT_COPY_VALUES`, `MAT_COPY_VALUES`, or `MAT_SHARE_NONZERO_PATTERN`.
4830: See the manual page for `MatDuplicateOption()` for an explanation of these options.
4832: Output Parameter:
4833: . M - pointer to place new matrix
4835: Level: intermediate
4837: Notes:
4838: You cannot change the nonzero pattern for the parent or child matrix if you use `MAT_SHARE_NONZERO_PATTERN`.
4840: May be called with an unassembled input `Mat` if `MAT_DO_NOT_COPY_VALUES` is used, in which case the output `Mat` is unassembled as well.
4842: When original mat is a product of matrix operation, e.g., an output of `MatMatMult()` or `MatCreateSubMatrix()`, only the simple matrix data structure of mat
4843: is duplicated and the internal data structures created for the reuse of previous matrix operations are not duplicated.
4844: User should not use `MatDuplicate()` to create new matrix M if M is intended to be reused as the product of matrix operation.
4846: .seealso: [](ch_matrices), `Mat`, `MatCopy()`, `MatConvert()`, `MatDuplicateOption`
4847: @*/
4848: PetscErrorCode MatDuplicate(Mat mat, MatDuplicateOption op, Mat *M)
4849: {
4850: Mat B;
4851: VecType vtype;
4852: PetscInt i;
4853: PetscObject dm;
4854: void (*viewf)(void);
4856: PetscFunctionBegin;
4860: PetscCheck(op != MAT_COPY_VALUES || mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "MAT_COPY_VALUES not allowed for unassembled matrix");
4861: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4862: MatCheckPreallocated(mat, 1);
4864: *M = NULL;
4865: PetscCall(PetscLogEventBegin(MAT_Convert, mat, 0, 0, 0));
4866: PetscUseTypeMethod(mat, duplicate, op, M);
4867: PetscCall(PetscLogEventEnd(MAT_Convert, mat, 0, 0, 0));
4868: B = *M;
4870: PetscCall(MatGetOperation(mat, MATOP_VIEW, &viewf));
4871: if (viewf) PetscCall(MatSetOperation(B, MATOP_VIEW, viewf));
4872: PetscCall(MatGetVecType(mat, &vtype));
4873: PetscCall(MatSetVecType(B, vtype));
4875: B->stencil.dim = mat->stencil.dim;
4876: B->stencil.noc = mat->stencil.noc;
4877: for (i = 0; i <= mat->stencil.dim; i++) {
4878: B->stencil.dims[i] = mat->stencil.dims[i];
4879: B->stencil.starts[i] = mat->stencil.starts[i];
4880: }
4882: B->nooffproczerorows = mat->nooffproczerorows;
4883: B->nooffprocentries = mat->nooffprocentries;
4885: PetscCall(PetscObjectQuery((PetscObject)mat, "__PETSc_dm", &dm));
4886: if (dm) PetscCall(PetscObjectCompose((PetscObject)B, "__PETSc_dm", dm));
4887: PetscCall(PetscObjectStateIncrease((PetscObject)B));
4888: PetscFunctionReturn(PETSC_SUCCESS);
4889: }
4891: /*@
4892: MatGetDiagonal - Gets the diagonal of a matrix as a `Vec`
4894: Logically Collective
4896: Input Parameter:
4897: . mat - the matrix
4899: Output Parameter:
4900: . v - the diagonal of the matrix
4902: Level: intermediate
4904: Note:
4905: Currently only correct in parallel for square matrices.
4907: .seealso: [](ch_matrices), `Mat`, `Vec`, `MatGetRow()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMaxAbs()`
4908: @*/
4909: PetscErrorCode MatGetDiagonal(Mat mat, Vec v)
4910: {
4911: PetscFunctionBegin;
4915: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4916: MatCheckPreallocated(mat, 1);
4918: PetscUseTypeMethod(mat, getdiagonal, v);
4919: PetscCall(PetscObjectStateIncrease((PetscObject)v));
4920: PetscFunctionReturn(PETSC_SUCCESS);
4921: }
4923: /*@C
4924: MatGetRowMin - Gets the minimum value (of the real part) of each
4925: row of the matrix
4927: Logically Collective
4929: Input Parameter:
4930: . mat - the matrix
4932: Output Parameters:
4933: + v - the vector for storing the maximums
4934: - idx - the indices of the column found for each row (optional)
4936: Level: intermediate
4938: Note:
4939: The result of this call are the same as if one converted the matrix to dense format
4940: and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
4942: This code is only implemented for a couple of matrix formats.
4944: .seealso: [](ch_matrices), `Mat`, `MatGetDiagonal()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMaxAbs()`, `MatGetRowMinAbs()`,
4945: `MatGetRowMax()`
4946: @*/
4947: PetscErrorCode MatGetRowMin(Mat mat, Vec v, PetscInt idx[])
4948: {
4949: PetscFunctionBegin;
4953: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4955: if (!mat->cmap->N) {
4956: PetscCall(VecSet(v, PETSC_MAX_REAL));
4957: if (idx) {
4958: PetscInt i, m = mat->rmap->n;
4959: for (i = 0; i < m; i++) idx[i] = -1;
4960: }
4961: } else {
4962: MatCheckPreallocated(mat, 1);
4963: }
4964: PetscUseTypeMethod(mat, getrowmin, v, idx);
4965: PetscCall(PetscObjectStateIncrease((PetscObject)v));
4966: PetscFunctionReturn(PETSC_SUCCESS);
4967: }
4969: /*@C
4970: MatGetRowMinAbs - Gets the minimum value (in absolute value) of each
4971: row of the matrix
4973: Logically Collective
4975: Input Parameter:
4976: . mat - the matrix
4978: Output Parameters:
4979: + v - the vector for storing the minimums
4980: - idx - the indices of the column found for each row (or `NULL` if not needed)
4982: Level: intermediate
4984: Notes:
4985: if a row is completely empty or has only 0.0 values then the idx[] value for that
4986: row is 0 (the first column).
4988: This code is only implemented for a couple of matrix formats.
4990: .seealso: [](ch_matrices), `Mat`, `MatGetDiagonal()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMax()`, `MatGetRowMaxAbs()`, `MatGetRowMin()`
4991: @*/
4992: PetscErrorCode MatGetRowMinAbs(Mat mat, Vec v, PetscInt idx[])
4993: {
4994: PetscFunctionBegin;
4998: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4999: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5001: if (!mat->cmap->N) {
5002: PetscCall(VecSet(v, 0.0));
5003: if (idx) {
5004: PetscInt i, m = mat->rmap->n;
5005: for (i = 0; i < m; i++) idx[i] = -1;
5006: }
5007: } else {
5008: MatCheckPreallocated(mat, 1);
5009: if (idx) PetscCall(PetscArrayzero(idx, mat->rmap->n));
5010: PetscUseTypeMethod(mat, getrowminabs, v, idx);
5011: }
5012: PetscCall(PetscObjectStateIncrease((PetscObject)v));
5013: PetscFunctionReturn(PETSC_SUCCESS);
5014: }
5016: /*@C
5017: MatGetRowMax - Gets the maximum value (of the real part) of each
5018: row of the matrix
5020: Logically Collective
5022: Input Parameter:
5023: . mat - the matrix
5025: Output Parameters:
5026: + v - the vector for storing the maximums
5027: - idx - the indices of the column found for each row (optional)
5029: Level: intermediate
5031: Notes:
5032: The result of this call are the same as if one converted the matrix to dense format
5033: and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
5035: This code is only implemented for a couple of matrix formats.
5037: .seealso: [](ch_matrices), `Mat`, `MatGetDiagonal()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMaxAbs()`, `MatGetRowMin()`, `MatGetRowMinAbs()`
5038: @*/
5039: PetscErrorCode MatGetRowMax(Mat mat, Vec v, PetscInt idx[])
5040: {
5041: PetscFunctionBegin;
5045: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5047: if (!mat->cmap->N) {
5048: PetscCall(VecSet(v, PETSC_MIN_REAL));
5049: if (idx) {
5050: PetscInt i, m = mat->rmap->n;
5051: for (i = 0; i < m; i++) idx[i] = -1;
5052: }
5053: } else {
5054: MatCheckPreallocated(mat, 1);
5055: PetscUseTypeMethod(mat, getrowmax, v, idx);
5056: }
5057: PetscCall(PetscObjectStateIncrease((PetscObject)v));
5058: PetscFunctionReturn(PETSC_SUCCESS);
5059: }
5061: /*@C
5062: MatGetRowMaxAbs - Gets the maximum value (in absolute value) of each
5063: row of the matrix
5065: Logically Collective
5067: Input Parameter:
5068: . mat - the matrix
5070: Output Parameters:
5071: + v - the vector for storing the maximums
5072: - idx - the indices of the column found for each row (or `NULL` if not needed)
5074: Level: intermediate
5076: Notes:
5077: if a row is completely empty or has only 0.0 values then the idx[] value for that
5078: row is 0 (the first column).
5080: This code is only implemented for a couple of matrix formats.
5082: .seealso: [](ch_matrices), `Mat`, `MatGetDiagonal()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMax()`, `MatGetRowMin()`, `MatGetRowMinAbs()`
5083: @*/
5084: PetscErrorCode MatGetRowMaxAbs(Mat mat, Vec v, PetscInt idx[])
5085: {
5086: PetscFunctionBegin;
5090: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5092: if (!mat->cmap->N) {
5093: PetscCall(VecSet(v, 0.0));
5094: if (idx) {
5095: PetscInt i, m = mat->rmap->n;
5096: for (i = 0; i < m; i++) idx[i] = -1;
5097: }
5098: } else {
5099: MatCheckPreallocated(mat, 1);
5100: if (idx) PetscCall(PetscArrayzero(idx, mat->rmap->n));
5101: PetscUseTypeMethod(mat, getrowmaxabs, v, idx);
5102: }
5103: PetscCall(PetscObjectStateIncrease((PetscObject)v));
5104: PetscFunctionReturn(PETSC_SUCCESS);
5105: }
5107: /*@
5108: MatGetRowSum - Gets the sum of each row of the matrix
5110: Logically or Neighborhood Collective
5112: Input Parameter:
5113: . mat - the matrix
5115: Output Parameter:
5116: . v - the vector for storing the sum of rows
5118: Level: intermediate
5120: Notes:
5121: This code is slow since it is not currently specialized for different formats
5123: .seealso: [](ch_matrices), `Mat`, `MatGetDiagonal()`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRowMax()`, `MatGetRowMin()`, `MatGetRowMaxAbs()`, `MatGetRowMinAbs()`
5124: @*/
5125: PetscErrorCode MatGetRowSum(Mat mat, Vec v)
5126: {
5127: Vec ones;
5129: PetscFunctionBegin;
5133: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5134: MatCheckPreallocated(mat, 1);
5135: PetscCall(MatCreateVecs(mat, &ones, NULL));
5136: PetscCall(VecSet(ones, 1.));
5137: PetscCall(MatMult(mat, ones, v));
5138: PetscCall(VecDestroy(&ones));
5139: PetscFunctionReturn(PETSC_SUCCESS);
5140: }
5142: /*@
5143: MatTransposeSetPrecursor - Set the matrix from which the second matrix will receive numerical transpose data with a call to `MatTranspose`(A,`MAT_REUSE_MATRIX`,&B)
5144: when B was not obtained with `MatTranspose`(A,`MAT_INITIAL_MATRIX`,&B)
5146: Collective
5148: Input Parameter:
5149: . mat - the matrix to provide the transpose
5151: Output Parameter:
5152: . mat - the matrix to contain the transpose; it MUST have the nonzero structure of the transpose of A or the code will crash or generate incorrect results
5154: Level: advanced
5156: Note:
5157: Normally the use of `MatTranspose`(A, `MAT_REUSE_MATRIX`, &B) requires that `B` was obtained with a call to `MatTranspose`(A, `MAT_INITIAL_MATRIX`, &B). This
5158: routine allows bypassing that call.
5160: .seealso: [](ch_matrices), `Mat`, `MatTransposeSymbolic()`, `MatTranspose()`, `MatMultTranspose()`, `MatMultTransposeAdd()`, `MatIsTranspose()`, `MatReuse`, `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, `MAT_INPLACE_MATRIX`
5161: @*/
5162: PetscErrorCode MatTransposeSetPrecursor(Mat mat, Mat B)
5163: {
5164: PetscContainer rB = NULL;
5165: MatParentState *rb = NULL;
5167: PetscFunctionBegin;
5168: PetscCall(PetscNew(&rb));
5169: rb->id = ((PetscObject)mat)->id;
5170: rb->state = 0;
5171: PetscCall(MatGetNonzeroState(mat, &rb->nonzerostate));
5172: PetscCall(PetscContainerCreate(PetscObjectComm((PetscObject)B), &rB));
5173: PetscCall(PetscContainerSetPointer(rB, rb));
5174: PetscCall(PetscContainerSetUserDestroy(rB, PetscContainerUserDestroyDefault));
5175: PetscCall(PetscObjectCompose((PetscObject)B, "MatTransposeParent", (PetscObject)rB));
5176: PetscCall(PetscObjectDereference((PetscObject)rB));
5177: PetscFunctionReturn(PETSC_SUCCESS);
5178: }
5180: /*@
5181: MatTranspose - Computes an in-place or out-of-place transpose of a matrix.
5183: Collective
5185: Input Parameters:
5186: + mat - the matrix to transpose
5187: - reuse - either `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, or `MAT_INPLACE_MATRIX`
5189: Output Parameter:
5190: . B - the transpose
5192: Level: intermediate
5194: Notes:
5195: If you use `MAT_INPLACE_MATRIX` then you must pass in &mat for B
5197: `MAT_REUSE_MATRIX` uses the B matrix obtained from a previous call to this function with `MAT_INITIAL_MATRIX`. If you already have a matrix to contain the
5198: transpose, call `MatTransposeSetPrecursor`(mat,B) before calling this routine.
5200: If the nonzero structure of mat changed from the previous call to this function with the same matrices an error will be generated for some matrix types.
5202: Consider using `MatCreateTranspose()` instead if you only need a matrix that behaves like the transpose, but don't need the storage to be changed.
5204: If mat is unchanged from the last call this function returns immediately without recomputing the result
5206: If you only need the symbolic transpose, and not the numerical values, use `MatTransposeSymbolic()`
5208: .seealso: [](ch_matrices), `Mat`, `MatTransposeSetPrecursor()`, `MatMultTranspose()`, `MatMultTransposeAdd()`, `MatIsTranspose()`, `MatReuse`, `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, `MAT_INPLACE_MATRIX`,
5209: `MatTransposeSymbolic()`, `MatCreateTranspose()`
5210: @*/
5211: PetscErrorCode MatTranspose(Mat mat, MatReuse reuse, Mat *B)
5212: {
5213: PetscContainer rB = NULL;
5214: MatParentState *rb = NULL;
5216: PetscFunctionBegin;
5219: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5220: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5221: PetscCheck(reuse != MAT_INPLACE_MATRIX || mat == *B, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "MAT_INPLACE_MATRIX requires last matrix to match first");
5222: PetscCheck(reuse != MAT_REUSE_MATRIX || mat != *B, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Perhaps you mean MAT_INPLACE_MATRIX");
5223: MatCheckPreallocated(mat, 1);
5224: if (reuse == MAT_REUSE_MATRIX) {
5225: PetscCall(PetscObjectQuery((PetscObject)*B, "MatTransposeParent", (PetscObject *)&rB));
5226: PetscCheck(rB, PetscObjectComm((PetscObject)*B), PETSC_ERR_ARG_WRONG, "Reuse matrix used was not generated from call to MatTranspose(). Suggest MatTransposeSetPrecursor().");
5227: PetscCall(PetscContainerGetPointer(rB, (void **)&rb));
5228: PetscCheck(rb->id == ((PetscObject)mat)->id, PetscObjectComm((PetscObject)*B), PETSC_ERR_ARG_WRONG, "Reuse matrix used was not generated from input matrix");
5229: if (rb->state == ((PetscObject)mat)->state) PetscFunctionReturn(PETSC_SUCCESS);
5230: }
5232: PetscCall(PetscLogEventBegin(MAT_Transpose, mat, 0, 0, 0));
5233: if (reuse != MAT_INPLACE_MATRIX || mat->symmetric != PETSC_BOOL3_TRUE) {
5234: PetscUseTypeMethod(mat, transpose, reuse, B);
5235: PetscCall(PetscObjectStateIncrease((PetscObject)*B));
5236: }
5237: PetscCall(PetscLogEventEnd(MAT_Transpose, mat, 0, 0, 0));
5239: if (reuse == MAT_INITIAL_MATRIX) PetscCall(MatTransposeSetPrecursor(mat, *B));
5240: if (reuse != MAT_INPLACE_MATRIX) {
5241: PetscCall(PetscObjectQuery((PetscObject)*B, "MatTransposeParent", (PetscObject *)&rB));
5242: PetscCall(PetscContainerGetPointer(rB, (void **)&rb));
5243: rb->state = ((PetscObject)mat)->state;
5244: rb->nonzerostate = mat->nonzerostate;
5245: }
5246: PetscFunctionReturn(PETSC_SUCCESS);
5247: }
5249: /*@
5250: MatTransposeSymbolic - Computes the symbolic part of the transpose of a matrix.
5252: Collective
5254: Input Parameter:
5255: . A - the matrix to transpose
5257: Output Parameter:
5258: . B - the transpose. This is a complete matrix but the numerical portion is invalid. One can call `MatTranspose`(A,`MAT_REUSE_MATRIX`,&B) to compute the
5259: numerical portion.
5261: Level: intermediate
5263: Note:
5264: This is not supported for many matrix types, use `MatTranspose()` in those cases
5266: .seealso: [](ch_matrices), `Mat`, `MatTransposeSetPrecursor()`, `MatTranspose()`, `MatMultTranspose()`, `MatMultTransposeAdd()`, `MatIsTranspose()`, `MatReuse`, `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, `MAT_INPLACE_MATRIX`
5267: @*/
5268: PetscErrorCode MatTransposeSymbolic(Mat A, Mat *B)
5269: {
5270: PetscFunctionBegin;
5273: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5274: PetscCheck(!A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5275: PetscCall(PetscLogEventBegin(MAT_Transpose, A, 0, 0, 0));
5276: PetscUseTypeMethod(A, transposesymbolic, B);
5277: PetscCall(PetscLogEventEnd(MAT_Transpose, A, 0, 0, 0));
5279: PetscCall(MatTransposeSetPrecursor(A, *B));
5280: PetscFunctionReturn(PETSC_SUCCESS);
5281: }
5283: PetscErrorCode MatTransposeCheckNonzeroState_Private(Mat A, Mat B)
5284: {
5285: PetscContainer rB;
5286: MatParentState *rb;
5288: PetscFunctionBegin;
5291: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5292: PetscCheck(!A->factortype, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5293: PetscCall(PetscObjectQuery((PetscObject)B, "MatTransposeParent", (PetscObject *)&rB));
5294: PetscCheck(rB, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_WRONG, "Reuse matrix used was not generated from call to MatTranspose()");
5295: PetscCall(PetscContainerGetPointer(rB, (void **)&rb));
5296: PetscCheck(rb->id == ((PetscObject)A)->id, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_WRONG, "Reuse matrix used was not generated from input matrix");
5297: PetscCheck(rb->nonzerostate == A->nonzerostate, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_WRONGSTATE, "Reuse matrix has changed nonzero structure");
5298: PetscFunctionReturn(PETSC_SUCCESS);
5299: }
5301: /*@
5302: MatIsTranspose - Test whether a matrix is another one's transpose,
5303: or its own, in which case it tests symmetry.
5305: Collective
5307: Input Parameters:
5308: + A - the matrix to test
5309: . B - the matrix to test against, this can equal the first parameter
5310: - tol - tolerance, differences between entries smaller than this are counted as zero
5312: Output Parameter:
5313: . flg - the result
5315: Level: intermediate
5317: Notes:
5318: Only available for `MATAIJ` matrices.
5320: The sequential algorithm has a running time of the order of the number of nonzeros; the parallel
5321: test involves parallel copies of the block-offdiagonal parts of the matrix.
5323: .seealso: [](ch_matrices), `Mat`, `MatTranspose()`, `MatIsSymmetric()`, `MatIsHermitian()`
5324: @*/
5325: PetscErrorCode MatIsTranspose(Mat A, Mat B, PetscReal tol, PetscBool *flg)
5326: {
5327: PetscErrorCode (*f)(Mat, Mat, PetscReal, PetscBool *), (*g)(Mat, Mat, PetscReal, PetscBool *);
5329: PetscFunctionBegin;
5333: PetscCall(PetscObjectQueryFunction((PetscObject)A, "MatIsTranspose_C", &f));
5334: PetscCall(PetscObjectQueryFunction((PetscObject)B, "MatIsTranspose_C", &g));
5335: *flg = PETSC_FALSE;
5336: if (f && g) {
5337: PetscCheck(f == g, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_NOTSAMETYPE, "Matrices do not have the same comparator for symmetry test");
5338: PetscCall((*f)(A, B, tol, flg));
5339: } else {
5340: MatType mattype;
5342: PetscCall(MatGetType(f ? B : A, &mattype));
5343: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Matrix of type %s does not support checking for transpose", mattype);
5344: }
5345: PetscFunctionReturn(PETSC_SUCCESS);
5346: }
5348: /*@
5349: MatHermitianTranspose - Computes an in-place or out-of-place Hermitian transpose of a matrix in complex conjugate.
5351: Collective
5353: Input Parameters:
5354: + mat - the matrix to transpose and complex conjugate
5355: - reuse - either `MAT_INITIAL_MATRIX`, `MAT_REUSE_MATRIX`, or `MAT_INPLACE_MATRIX`
5357: Output Parameter:
5358: . B - the Hermitian transpose
5360: Level: intermediate
5362: .seealso: [](ch_matrices), `Mat`, `MatTranspose()`, `MatMultTranspose()`, `MatMultTransposeAdd()`, `MatIsTranspose()`, `MatReuse`
5363: @*/
5364: PetscErrorCode MatHermitianTranspose(Mat mat, MatReuse reuse, Mat *B)
5365: {
5366: PetscFunctionBegin;
5367: PetscCall(MatTranspose(mat, reuse, B));
5368: #if defined(PETSC_USE_COMPLEX)
5369: PetscCall(MatConjugate(*B));
5370: #endif
5371: PetscFunctionReturn(PETSC_SUCCESS);
5372: }
5374: /*@
5375: MatIsHermitianTranspose - Test whether a matrix is another one's Hermitian transpose,
5377: Collective
5379: Input Parameters:
5380: + A - the matrix to test
5381: . B - the matrix to test against, this can equal the first parameter
5382: - tol - tolerance, differences between entries smaller than this are counted as zero
5384: Output Parameter:
5385: . flg - the result
5387: Level: intermediate
5389: Notes:
5390: Only available for `MATAIJ` matrices.
5392: The sequential algorithm
5393: has a running time of the order of the number of nonzeros; the parallel
5394: test involves parallel copies of the block-offdiagonal parts of the matrix.
5396: .seealso: [](ch_matrices), `Mat`, `MatTranspose()`, `MatIsSymmetric()`, `MatIsHermitian()`, `MatIsTranspose()`
5397: @*/
5398: PetscErrorCode MatIsHermitianTranspose(Mat A, Mat B, PetscReal tol, PetscBool *flg)
5399: {
5400: PetscErrorCode (*f)(Mat, Mat, PetscReal, PetscBool *), (*g)(Mat, Mat, PetscReal, PetscBool *);
5402: PetscFunctionBegin;
5406: PetscCall(PetscObjectQueryFunction((PetscObject)A, "MatIsHermitianTranspose_C", &f));
5407: PetscCall(PetscObjectQueryFunction((PetscObject)B, "MatIsHermitianTranspose_C", &g));
5408: if (f && g) {
5409: PetscCheck(f != g, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_NOTSAMETYPE, "Matrices do not have the same comparator for Hermitian test");
5410: PetscCall((*f)(A, B, tol, flg));
5411: }
5412: PetscFunctionReturn(PETSC_SUCCESS);
5413: }
5415: /*@
5416: MatPermute - Creates a new matrix with rows and columns permuted from the
5417: original.
5419: Collective
5421: Input Parameters:
5422: + mat - the matrix to permute
5423: . row - row permutation, each processor supplies only the permutation for its rows
5424: - col - column permutation, each processor supplies only the permutation for its columns
5426: Output Parameter:
5427: . B - the permuted matrix
5429: Level: advanced
5431: Note:
5432: The index sets map from row/col of permuted matrix to row/col of original matrix.
5433: The index sets should be on the same communicator as mat and have the same local sizes.
5435: Developer Note:
5436: If you want to implement `MatPermute()` for a matrix type, and your approach doesn't
5437: exploit the fact that row and col are permutations, consider implementing the
5438: more general `MatCreateSubMatrix()` instead.
5440: .seealso: [](ch_matrices), `Mat`, `MatGetOrdering()`, `ISAllGather()`, `MatCreateSubMatrix()`
5441: @*/
5442: PetscErrorCode MatPermute(Mat mat, IS row, IS col, Mat *B)
5443: {
5444: PetscFunctionBegin;
5450: PetscCheckSameComm(mat, 1, row, 2);
5451: if (row != col) PetscCheckSameComm(row, 2, col, 3);
5452: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5453: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5454: PetscCheck(mat->ops->permute || mat->ops->createsubmatrix, PETSC_COMM_SELF, PETSC_ERR_SUP, "MatPermute not available for Mat type %s", ((PetscObject)mat)->type_name);
5455: MatCheckPreallocated(mat, 1);
5457: if (mat->ops->permute) {
5458: PetscUseTypeMethod(mat, permute, row, col, B);
5459: PetscCall(PetscObjectStateIncrease((PetscObject)*B));
5460: } else {
5461: PetscCall(MatCreateSubMatrix(mat, row, col, MAT_INITIAL_MATRIX, B));
5462: }
5463: PetscFunctionReturn(PETSC_SUCCESS);
5464: }
5466: /*@
5467: MatEqual - Compares two matrices.
5469: Collective
5471: Input Parameters:
5472: + A - the first matrix
5473: - B - the second matrix
5475: Output Parameter:
5476: . flg - `PETSC_TRUE` if the matrices are equal; `PETSC_FALSE` otherwise.
5478: Level: intermediate
5480: .seealso: [](ch_matrices), `Mat`
5481: @*/
5482: PetscErrorCode MatEqual(Mat A, Mat B, PetscBool *flg)
5483: {
5484: PetscFunctionBegin;
5490: PetscCheckSameComm(A, 1, B, 2);
5491: MatCheckPreallocated(A, 1);
5492: MatCheckPreallocated(B, 2);
5493: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5494: PetscCheck(B->assembled, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5495: PetscCheck(A->rmap->N == B->rmap->N && A->cmap->N == B->cmap->N, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_SIZ, "Mat A,Mat B: global dim %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT, A->rmap->N, B->rmap->N, A->cmap->N,
5496: B->cmap->N);
5497: if (A->ops->equal && A->ops->equal == B->ops->equal) {
5498: PetscUseTypeMethod(A, equal, B, flg);
5499: } else {
5500: PetscCall(MatMultEqual(A, B, 10, flg));
5501: }
5502: PetscFunctionReturn(PETSC_SUCCESS);
5503: }
5505: /*@
5506: MatDiagonalScale - Scales a matrix on the left and right by diagonal
5507: matrices that are stored as vectors. Either of the two scaling
5508: matrices can be `NULL`.
5510: Collective
5512: Input Parameters:
5513: + mat - the matrix to be scaled
5514: . l - the left scaling vector (or `NULL`)
5515: - r - the right scaling vector (or `NULL`)
5517: Level: intermediate
5519: Note:
5520: `MatDiagonalScale()` computes A = LAR, where
5521: L = a diagonal matrix (stored as a vector), R = a diagonal matrix (stored as a vector)
5522: The L scales the rows of the matrix, the R scales the columns of the matrix.
5524: .seealso: [](ch_matrices), `Mat`, `MatScale()`, `MatShift()`, `MatDiagonalSet()`
5525: @*/
5526: PetscErrorCode MatDiagonalScale(Mat mat, Vec l, Vec r)
5527: {
5528: PetscFunctionBegin;
5531: if (l) {
5533: PetscCheckSameComm(mat, 1, l, 2);
5534: }
5535: if (r) {
5537: PetscCheckSameComm(mat, 1, r, 3);
5538: }
5539: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5540: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5541: MatCheckPreallocated(mat, 1);
5542: if (!l && !r) PetscFunctionReturn(PETSC_SUCCESS);
5544: PetscCall(PetscLogEventBegin(MAT_Scale, mat, 0, 0, 0));
5545: PetscUseTypeMethod(mat, diagonalscale, l, r);
5546: PetscCall(PetscLogEventEnd(MAT_Scale, mat, 0, 0, 0));
5547: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
5548: if (l != r) mat->symmetric = PETSC_BOOL3_FALSE;
5549: PetscFunctionReturn(PETSC_SUCCESS);
5550: }
5552: /*@
5553: MatScale - Scales all elements of a matrix by a given number.
5555: Logically Collective
5557: Input Parameters:
5558: + mat - the matrix to be scaled
5559: - a - the scaling value
5561: Level: intermediate
5563: .seealso: [](ch_matrices), `Mat`, `MatDiagonalScale()`
5564: @*/
5565: PetscErrorCode MatScale(Mat mat, PetscScalar a)
5566: {
5567: PetscFunctionBegin;
5570: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5571: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5573: MatCheckPreallocated(mat, 1);
5575: PetscCall(PetscLogEventBegin(MAT_Scale, mat, 0, 0, 0));
5576: if (a != (PetscScalar)1.0) {
5577: PetscUseTypeMethod(mat, scale, a);
5578: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
5579: }
5580: PetscCall(PetscLogEventEnd(MAT_Scale, mat, 0, 0, 0));
5581: PetscFunctionReturn(PETSC_SUCCESS);
5582: }
5584: /*@
5585: MatNorm - Calculates various norms of a matrix.
5587: Collective
5589: Input Parameters:
5590: + mat - the matrix
5591: - type - the type of norm, `NORM_1`, `NORM_FROBENIUS`, `NORM_INFINITY`
5593: Output Parameter:
5594: . nrm - the resulting norm
5596: Level: intermediate
5598: .seealso: [](ch_matrices), `Mat`
5599: @*/
5600: PetscErrorCode MatNorm(Mat mat, NormType type, PetscReal *nrm)
5601: {
5602: PetscFunctionBegin;
5607: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
5608: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
5609: MatCheckPreallocated(mat, 1);
5611: PetscUseTypeMethod(mat, norm, type, nrm);
5612: PetscFunctionReturn(PETSC_SUCCESS);
5613: }
5615: /*
5616: This variable is used to prevent counting of MatAssemblyBegin() that
5617: are called from within a MatAssemblyEnd().
5618: */
5619: static PetscInt MatAssemblyEnd_InUse = 0;
5620: /*@
5621: MatAssemblyBegin - Begins assembling the matrix. This routine should
5622: be called after completing all calls to `MatSetValues()`.
5624: Collective
5626: Input Parameters:
5627: + mat - the matrix
5628: - type - type of assembly, either `MAT_FLUSH_ASSEMBLY` or `MAT_FINAL_ASSEMBLY`
5630: Level: beginner
5632: Notes:
5633: `MatSetValues()` generally caches the values that belong to other MPI ranks. The matrix is ready to
5634: use only after `MatAssemblyBegin()` and `MatAssemblyEnd()` have been called.
5636: Use `MAT_FLUSH_ASSEMBLY` when switching between `ADD_VALUES` and `INSERT_VALUES`
5637: in `MatSetValues()`; use `MAT_FINAL_ASSEMBLY` for the final assembly before
5638: using the matrix.
5640: ALL processes that share a matrix MUST call `MatAssemblyBegin()` and `MatAssemblyEnd()` the SAME NUMBER of times, and each time with the
5641: same flag of `MAT_FLUSH_ASSEMBLY` or `MAT_FINAL_ASSEMBLY` for all processes. Thus you CANNOT locally change from `ADD_VALUES` to `INSERT_VALUES`, that is
5642: a global collective operation requiring all processes that share the matrix.
5644: Space for preallocated nonzeros that is not filled by a call to `MatSetValues()` or a related routine are compressed
5645: out by assembly. If you intend to use that extra space on a subsequent assembly, be sure to insert explicit zeros
5646: before `MAT_FINAL_ASSEMBLY` so the space is not compressed out.
5648: .seealso: [](ch_matrices), `Mat`, `MatAssemblyEnd()`, `MatSetValues()`, `MatAssembled()`
5649: @*/
5650: PetscErrorCode MatAssemblyBegin(Mat mat, MatAssemblyType type)
5651: {
5652: PetscFunctionBegin;
5655: MatCheckPreallocated(mat, 1);
5656: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix.\nDid you forget to call MatSetUnfactored()?");
5657: if (mat->assembled) {
5658: mat->was_assembled = PETSC_TRUE;
5659: mat->assembled = PETSC_FALSE;
5660: }
5662: if (!MatAssemblyEnd_InUse) {
5663: PetscCall(PetscLogEventBegin(MAT_AssemblyBegin, mat, 0, 0, 0));
5664: PetscTryTypeMethod(mat, assemblybegin, type);
5665: PetscCall(PetscLogEventEnd(MAT_AssemblyBegin, mat, 0, 0, 0));
5666: } else PetscTryTypeMethod(mat, assemblybegin, type);
5667: PetscFunctionReturn(PETSC_SUCCESS);
5668: }
5670: /*@
5671: MatAssembled - Indicates if a matrix has been assembled and is ready for
5672: use; for example, in matrix-vector product.
5674: Not Collective
5676: Input Parameter:
5677: . mat - the matrix
5679: Output Parameter:
5680: . assembled - `PETSC_TRUE` or `PETSC_FALSE`
5682: Level: advanced
5684: .seealso: [](ch_matrices), `Mat`, `MatAssemblyEnd()`, `MatSetValues()`, `MatAssemblyBegin()`
5685: @*/
5686: PetscErrorCode MatAssembled(Mat mat, PetscBool *assembled)
5687: {
5688: PetscFunctionBegin;
5691: *assembled = mat->assembled;
5692: PetscFunctionReturn(PETSC_SUCCESS);
5693: }
5695: /*@
5696: MatAssemblyEnd - Completes assembling the matrix. This routine should
5697: be called after `MatAssemblyBegin()`.
5699: Collective
5701: Input Parameters:
5702: + mat - the matrix
5703: - type - type of assembly, either `MAT_FLUSH_ASSEMBLY` or `MAT_FINAL_ASSEMBLY`
5705: Options Database Keys:
5706: + -mat_view ::ascii_info - Prints info on matrix at conclusion of `MatAssemblyEnd()`
5707: . -mat_view ::ascii_info_detail - Prints more detailed info
5708: . -mat_view - Prints matrix in ASCII format
5709: . -mat_view ::ascii_matlab - Prints matrix in Matlab format
5710: . -mat_view draw - draws nonzero structure of matrix, using `MatView()` and `PetscDrawOpenX()`.
5711: . -display <name> - Sets display name (default is host)
5712: . -draw_pause <sec> - Sets number of seconds to pause after display
5713: . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (See [Using MATLAB with PETSc](ch_matlab))
5714: . -viewer_socket_machine <machine> - Machine to use for socket
5715: . -viewer_socket_port <port> - Port number to use for socket
5716: - -mat_view binary:filename[:append] - Save matrix to file in binary format
5718: Level: beginner
5720: .seealso: [](ch_matrices), `Mat`, `MatAssemblyBegin()`, `MatSetValues()`, `PetscDrawOpenX()`, `PetscDrawCreate()`, `MatView()`, `MatAssembled()`, `PetscViewerSocketOpen()`
5721: @*/
5722: PetscErrorCode MatAssemblyEnd(Mat mat, MatAssemblyType type)
5723: {
5724: static PetscInt inassm = 0;
5725: PetscBool flg = PETSC_FALSE;
5727: PetscFunctionBegin;
5731: inassm++;
5732: MatAssemblyEnd_InUse++;
5733: if (MatAssemblyEnd_InUse == 1) { /* Do the logging only the first time through */
5734: PetscCall(PetscLogEventBegin(MAT_AssemblyEnd, mat, 0, 0, 0));
5735: PetscTryTypeMethod(mat, assemblyend, type);
5736: PetscCall(PetscLogEventEnd(MAT_AssemblyEnd, mat, 0, 0, 0));
5737: } else PetscTryTypeMethod(mat, assemblyend, type);
5739: /* Flush assembly is not a true assembly */
5740: if (type != MAT_FLUSH_ASSEMBLY) {
5741: if (mat->num_ass) {
5742: if (!mat->symmetry_eternal) {
5743: mat->symmetric = PETSC_BOOL3_UNKNOWN;
5744: mat->hermitian = PETSC_BOOL3_UNKNOWN;
5745: }
5746: if (!mat->structural_symmetry_eternal && mat->ass_nonzerostate != mat->nonzerostate) mat->structurally_symmetric = PETSC_BOOL3_UNKNOWN;
5747: if (!mat->spd_eternal) mat->spd = PETSC_BOOL3_UNKNOWN;
5748: }
5749: mat->num_ass++;
5750: mat->assembled = PETSC_TRUE;
5751: mat->ass_nonzerostate = mat->nonzerostate;
5752: }
5754: mat->insertmode = NOT_SET_VALUES;
5755: MatAssemblyEnd_InUse--;
5756: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
5757: if (inassm == 1 && type != MAT_FLUSH_ASSEMBLY) {
5758: PetscCall(MatViewFromOptions(mat, NULL, "-mat_view"));
5760: if (mat->checksymmetryonassembly) {
5761: PetscCall(MatIsSymmetric(mat, mat->checksymmetrytol, &flg));
5762: if (flg) {
5763: PetscCall(PetscPrintf(PetscObjectComm((PetscObject)mat), "Matrix is symmetric (tolerance %g)\n", (double)mat->checksymmetrytol));
5764: } else {
5765: PetscCall(PetscPrintf(PetscObjectComm((PetscObject)mat), "Matrix is not symmetric (tolerance %g)\n", (double)mat->checksymmetrytol));
5766: }
5767: }
5768: if (mat->nullsp && mat->checknullspaceonassembly) PetscCall(MatNullSpaceTest(mat->nullsp, mat, NULL));
5769: }
5770: inassm--;
5771: PetscFunctionReturn(PETSC_SUCCESS);
5772: }
5774: /*@
5775: MatSetOption - Sets a parameter option for a matrix. Some options
5776: may be specific to certain storage formats. Some options
5777: determine how values will be inserted (or added). Sorted,
5778: row-oriented input will generally assemble the fastest. The default
5779: is row-oriented.
5781: Logically Collective for certain operations, such as `MAT_SPD`, not collective for `MAT_ROW_ORIENTED`, see `MatOption`
5783: Input Parameters:
5784: + mat - the matrix
5785: . option - the option, one of those listed below (and possibly others),
5786: - flg - turn the option on (`PETSC_TRUE`) or off (`PETSC_FALSE`)
5788: Options Describing Matrix Structure:
5789: + `MAT_SPD` - symmetric positive definite
5790: . `MAT_SYMMETRIC` - symmetric in terms of both structure and value
5791: . `MAT_HERMITIAN` - transpose is the complex conjugation
5792: . `MAT_STRUCTURALLY_SYMMETRIC` - symmetric nonzero structure
5793: . `MAT_SYMMETRY_ETERNAL` - indicates the symmetry (or Hermitian structure) or its absence will persist through any changes to the matrix
5794: . `MAT_STRUCTURAL_SYMMETRY_ETERNAL` - indicates the structural symmetry or its absence will persist through any changes to the matrix
5795: - `MAT_SPD_ETERNAL` - indicates the value of `MAT_SPD` (true or false) will persist through any changes to the matrix
5797: These are not really options of the matrix, they are knowledge about the structure of the matrix that users may provide so that they
5798: do not need to be computed (usually at a high cost)
5800: Options For Use with `MatSetValues()`:
5801: Insert a logically dense subblock, which can be
5802: . `MAT_ROW_ORIENTED` - row-oriented (default)
5804: These options reflect the data you pass in with `MatSetValues()`; it has
5805: nothing to do with how the data is stored internally in the matrix
5806: data structure.
5808: When (re)assembling a matrix, we can restrict the input for
5809: efficiency/debugging purposes. These options include
5810: + `MAT_NEW_NONZERO_LOCATIONS` - additional insertions will be allowed if they generate a new nonzero (slow)
5811: . `MAT_FORCE_DIAGONAL_ENTRIES` - forces diagonal entries to be allocated
5812: . `MAT_IGNORE_OFF_PROC_ENTRIES` - drops off-processor entries
5813: . `MAT_NEW_NONZERO_LOCATION_ERR` - generates an error for new matrix entry
5814: . `MAT_USE_HASH_TABLE` - uses a hash table to speed up matrix assembly
5815: . `MAT_NO_OFF_PROC_ENTRIES` - you know each process will only set values for its own rows, will generate an error if
5816: any process sets values for another process. This avoids all reductions in the MatAssembly routines and thus improves
5817: performance for very large process counts.
5818: - `MAT_SUBSET_OFF_PROC_ENTRIES` - you know that the first assembly after setting this flag will set a superset
5819: of the off-process entries required for all subsequent assemblies. This avoids a rendezvous step in the MatAssembly
5820: functions, instead sending only neighbor messages.
5822: Level: intermediate
5824: Notes:
5825: Except for `MAT_UNUSED_NONZERO_LOCATION_ERR` and `MAT_ROW_ORIENTED` all processes that share the matrix must pass the same value in flg!
5827: Some options are relevant only for particular matrix types and
5828: are thus ignored by others. Other options are not supported by
5829: certain matrix types and will generate an error message if set.
5831: If using Fortran to compute a matrix, one may need to
5832: use the column-oriented option (or convert to the row-oriented
5833: format).
5835: `MAT_NEW_NONZERO_LOCATIONS` set to `PETSC_FALSE` indicates that any add or insertion
5836: that would generate a new entry in the nonzero structure is instead
5837: ignored. Thus, if memory has not already been allocated for this particular
5838: data, then the insertion is ignored. For dense matrices, in which
5839: the entire array is allocated, no entries are ever ignored.
5840: Set after the first `MatAssemblyEnd()`. If this option is set then the MatAssemblyBegin/End() processes has one less global reduction
5842: `MAT_NEW_NONZERO_LOCATION_ERR` set to PETSC_TRUE indicates that any add or insertion
5843: that would generate a new entry in the nonzero structure instead produces
5844: an error. (Currently supported for `MATAIJ` and `MATBAIJ` formats only.) If this option is set then the `MatAssemblyBegin()`/`MatAssemblyEnd()` processes has one less global reduction
5846: `MAT_NEW_NONZERO_ALLOCATION_ERR` set to `PETSC_TRUE` indicates that any add or insertion
5847: that would generate a new entry that has not been preallocated will
5848: instead produce an error. (Currently supported for `MATAIJ` and `MATBAIJ` formats
5849: only.) This is a useful flag when debugging matrix memory preallocation.
5850: If this option is set then the `MatAssemblyBegin()`/`MatAssemblyEnd()` processes has one less global reduction
5852: `MAT_IGNORE_OFF_PROC_ENTRIES` set to `PETSC_TRUE` indicates entries destined for
5853: other processors should be dropped, rather than stashed.
5854: This is useful if you know that the "owning" processor is also
5855: always generating the correct matrix entries, so that PETSc need
5856: not transfer duplicate entries generated on another processor.
5858: `MAT_USE_HASH_TABLE` indicates that a hash table be used to improve the
5859: searches during matrix assembly. When this flag is set, the hash table
5860: is created during the first matrix assembly. This hash table is
5861: used the next time through, during `MatSetValues()`/`MatSetValuesBlocked()`
5862: to improve the searching of indices. `MAT_NEW_NONZERO_LOCATIONS` flag
5863: should be used with `MAT_USE_HASH_TABLE` flag. This option is currently
5864: supported by `MATMPIBAIJ` format only.
5866: `MAT_KEEP_NONZERO_PATTERN` indicates when `MatZeroRows()` is called the zeroed entries
5867: are kept in the nonzero structure
5869: `MAT_IGNORE_ZERO_ENTRIES` - for `MATAIJ` and `MATIS` matrices this will stop zero values from creating
5870: a zero location in the matrix
5872: `MAT_USE_INODES` - indicates using inode version of the code - works with `MATAIJ` matrix types
5874: `MAT_NO_OFF_PROC_ZERO_ROWS` - you know each process will only zero its own rows. This avoids all reductions in the
5875: zero row routines and thus improves performance for very large process counts.
5877: `MAT_IGNORE_LOWER_TRIANGULAR` - For `MATSBAIJ` matrices will ignore any insertions you make in the lower triangular
5878: part of the matrix (since they should match the upper triangular part).
5880: `MAT_SORTED_FULL` - each process provides exactly its local rows; all column indices for a given row are passed in a
5881: single call to `MatSetValues()`, preallocation is perfect, row oriented, `INSERT_VALUES` is used. Common
5882: with finite difference schemes with non-periodic boundary conditions.
5884: Developer Note:
5885: `MAT_SYMMETRY_ETERNAL`, `MAT_STRUCTURAL_SYMMETRY_ETERNAL`, and `MAT_SPD_ETERNAL` are used by `MatAssemblyEnd()` and in other
5886: places where otherwise the value of `MAT_SYMMETRIC`, `MAT_STRUCTURAL_SYMMETRIC` or `MAT_SPD` would need to be changed back
5887: to `PETSC_BOOL3_UNKNOWN` because the matrix values had changed so the code cannot be certain that the related property had
5888: not changed.
5890: .seealso: [](ch_matrices), `MatOption`, `Mat`, `MatGetOption()`
5891: @*/
5892: PetscErrorCode MatSetOption(Mat mat, MatOption op, PetscBool flg)
5893: {
5894: PetscFunctionBegin;
5896: if (op > 0) {
5899: }
5901: PetscCheck(((int)op) > MAT_OPTION_MIN && ((int)op) < MAT_OPTION_MAX, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Options %d is out of range", (int)op);
5903: switch (op) {
5904: case MAT_FORCE_DIAGONAL_ENTRIES:
5905: mat->force_diagonals = flg;
5906: PetscFunctionReturn(PETSC_SUCCESS);
5907: case MAT_NO_OFF_PROC_ENTRIES:
5908: mat->nooffprocentries = flg;
5909: PetscFunctionReturn(PETSC_SUCCESS);
5910: case MAT_SUBSET_OFF_PROC_ENTRIES:
5911: mat->assembly_subset = flg;
5912: if (!mat->assembly_subset) { /* See the same logic in VecAssembly wrt VEC_SUBSET_OFF_PROC_ENTRIES */
5913: #if !defined(PETSC_HAVE_MPIUNI)
5914: PetscCall(MatStashScatterDestroy_BTS(&mat->stash));
5915: #endif
5916: mat->stash.first_assembly_done = PETSC_FALSE;
5917: }
5918: PetscFunctionReturn(PETSC_SUCCESS);
5919: case MAT_NO_OFF_PROC_ZERO_ROWS:
5920: mat->nooffproczerorows = flg;
5921: PetscFunctionReturn(PETSC_SUCCESS);
5922: case MAT_SPD:
5923: if (flg) {
5924: mat->spd = PETSC_BOOL3_TRUE;
5925: mat->symmetric = PETSC_BOOL3_TRUE;
5926: mat->structurally_symmetric = PETSC_BOOL3_TRUE;
5927: } else {
5928: mat->spd = PETSC_BOOL3_FALSE;
5929: }
5930: break;
5931: case MAT_SYMMETRIC:
5932: mat->symmetric = flg ? PETSC_BOOL3_TRUE : PETSC_BOOL3_FALSE;
5933: if (flg) mat->structurally_symmetric = PETSC_BOOL3_TRUE;
5934: #if !defined(PETSC_USE_COMPLEX)
5935: mat->hermitian = flg ? PETSC_BOOL3_TRUE : PETSC_BOOL3_FALSE;
5936: #endif
5937: break;
5938: case MAT_HERMITIAN:
5939: mat->hermitian = flg ? PETSC_BOOL3_TRUE : PETSC_BOOL3_FALSE;
5940: if (flg) mat->structurally_symmetric = PETSC_BOOL3_TRUE;
5941: #if !defined(PETSC_USE_COMPLEX)
5942: mat->symmetric = flg ? PETSC_BOOL3_TRUE : PETSC_BOOL3_FALSE;
5943: #endif
5944: break;
5945: case MAT_STRUCTURALLY_SYMMETRIC:
5946: mat->structurally_symmetric = flg ? PETSC_BOOL3_TRUE : PETSC_BOOL3_FALSE;
5947: break;
5948: case MAT_SYMMETRY_ETERNAL:
5949: PetscCheck(mat->symmetric != PETSC_BOOL3_UNKNOWN, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Cannot set MAT_SYMMETRY_ETERNAL without first setting MAT_SYMMETRIC to true or false");
5950: mat->symmetry_eternal = flg;
5951: if (flg) mat->structural_symmetry_eternal = PETSC_TRUE;
5952: break;
5953: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
5954: PetscCheck(mat->structurally_symmetric != PETSC_BOOL3_UNKNOWN, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Cannot set MAT_STRUCTURAL_SYMMETRY_ETERNAL without first setting MAT_STRUCTURAL_SYMMETRIC to true or false");
5955: mat->structural_symmetry_eternal = flg;
5956: break;
5957: case MAT_SPD_ETERNAL:
5958: PetscCheck(mat->spd != PETSC_BOOL3_UNKNOWN, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Cannot set MAT_SPD_ETERNAL without first setting MAT_SPD to true or false");
5959: mat->spd_eternal = flg;
5960: if (flg) {
5961: mat->structural_symmetry_eternal = PETSC_TRUE;
5962: mat->symmetry_eternal = PETSC_TRUE;
5963: }
5964: break;
5965: case MAT_STRUCTURE_ONLY:
5966: mat->structure_only = flg;
5967: break;
5968: case MAT_SORTED_FULL:
5969: mat->sortedfull = flg;
5970: break;
5971: default:
5972: break;
5973: }
5974: PetscTryTypeMethod(mat, setoption, op, flg);
5975: PetscFunctionReturn(PETSC_SUCCESS);
5976: }
5978: /*@
5979: MatGetOption - Gets a parameter option that has been set for a matrix.
5981: Logically Collective
5983: Input Parameters:
5984: + mat - the matrix
5985: - option - the option, this only responds to certain options, check the code for which ones
5987: Output Parameter:
5988: . flg - turn the option on (`PETSC_TRUE`) or off (`PETSC_FALSE`)
5990: Level: intermediate
5992: Notes:
5993: Can only be called after `MatSetSizes()` and `MatSetType()` have been set.
5995: Certain option values may be unknown, for those use the routines `MatIsSymmetric()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, or
5996: `MatIsSymmetricKnown()`, `MatIsHermitianKnown()`, `MatIsStructurallySymmetricKnown()`
5998: .seealso: [](ch_matrices), `Mat`, `MatOption`, `MatSetOption()`, `MatIsSymmetric()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`,
5999: `MatIsSymmetricKnown()`, `MatIsHermitianKnown()`, `MatIsStructurallySymmetricKnown()`
6000: @*/
6001: PetscErrorCode MatGetOption(Mat mat, MatOption op, PetscBool *flg)
6002: {
6003: PetscFunctionBegin;
6007: PetscCheck(((int)op) > MAT_OPTION_MIN && ((int)op) < MAT_OPTION_MAX, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Options %d is out of range", (int)op);
6008: PetscCheck(((PetscObject)mat)->type_name, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_TYPENOTSET, "Cannot get options until type and size have been set, see MatSetType() and MatSetSizes()");
6010: switch (op) {
6011: case MAT_NO_OFF_PROC_ENTRIES:
6012: *flg = mat->nooffprocentries;
6013: break;
6014: case MAT_NO_OFF_PROC_ZERO_ROWS:
6015: *flg = mat->nooffproczerorows;
6016: break;
6017: case MAT_SYMMETRIC:
6018: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Use MatIsSymmetric() or MatIsSymmetricKnown()");
6019: break;
6020: case MAT_HERMITIAN:
6021: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Use MatIsHermitian() or MatIsHermitianKnown()");
6022: break;
6023: case MAT_STRUCTURALLY_SYMMETRIC:
6024: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Use MatIsStructurallySymmetric() or MatIsStructurallySymmetricKnown()");
6025: break;
6026: case MAT_SPD:
6027: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "Use MatIsSPDKnown()");
6028: break;
6029: case MAT_SYMMETRY_ETERNAL:
6030: *flg = mat->symmetry_eternal;
6031: break;
6032: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
6033: *flg = mat->symmetry_eternal;
6034: break;
6035: default:
6036: break;
6037: }
6038: PetscFunctionReturn(PETSC_SUCCESS);
6039: }
6041: /*@
6042: MatZeroEntries - Zeros all entries of a matrix. For sparse matrices
6043: this routine retains the old nonzero structure.
6045: Logically Collective
6047: Input Parameter:
6048: . mat - the matrix
6050: Level: intermediate
6052: Note:
6053: If the matrix was not preallocated then a default, likely poor preallocation will be set in the matrix, so this should be called after the preallocation phase.
6054: See the Performance chapter of the users manual for information on preallocating matrices.
6056: .seealso: [](ch_matrices), `Mat`, `MatZeroRows()`, `MatZeroRowsColumns()`
6057: @*/
6058: PetscErrorCode MatZeroEntries(Mat mat)
6059: {
6060: PetscFunctionBegin;
6063: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6064: PetscCheck(mat->insertmode == NOT_SET_VALUES, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for matrices where you have set values but not yet assembled");
6065: MatCheckPreallocated(mat, 1);
6067: PetscCall(PetscLogEventBegin(MAT_ZeroEntries, mat, 0, 0, 0));
6068: PetscUseTypeMethod(mat, zeroentries);
6069: PetscCall(PetscLogEventEnd(MAT_ZeroEntries, mat, 0, 0, 0));
6070: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
6071: PetscFunctionReturn(PETSC_SUCCESS);
6072: }
6074: /*@
6075: MatZeroRowsColumns - Zeros all entries (except possibly the main diagonal)
6076: of a set of rows and columns of a matrix.
6078: Collective
6080: Input Parameters:
6081: + mat - the matrix
6082: . numRows - the number of rows/columns to zero
6083: . rows - the global row indices
6084: . diag - value put in the diagonal of the eliminated rows
6085: . x - optional vector of the solution for zeroed rows (other entries in vector are not used), these must be set before this call
6086: - b - optional vector of the right hand side, that will be adjusted by provided solution entries
6088: Level: intermediate
6090: Notes:
6091: This routine, along with `MatZeroRows()`, is typically used to eliminate known Dirichlet boundary conditions from a linear system.
6093: For each zeroed row, the value of the corresponding `b` is set to diag times the value of the corresponding `x`.
6094: The other entries of `b` will be adjusted by the known values of `x` times the corresponding matrix entries in the columns that are being eliminated
6096: If the resulting linear system is to be solved with `KSP` then one can (but does not have to) call `KSPSetInitialGuessNonzero()` to allow the
6097: Krylov method to take advantage of the known solution on the zeroed rows.
6099: For the parallel case, all processes that share the matrix (i.e.,
6100: those in the communicator used for matrix creation) MUST call this
6101: routine, regardless of whether any rows being zeroed are owned by
6102: them.
6104: Unlike `MatZeroRows()` this does not change the nonzero structure of the matrix, it merely zeros those entries in the matrix.
6106: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6107: list only rows local to itself).
6109: The option `MAT_NO_OFF_PROC_ZERO_ROWS` does not apply to this routine.
6111: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRows()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6112: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6113: @*/
6114: PetscErrorCode MatZeroRowsColumns(Mat mat, PetscInt numRows, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
6115: {
6116: PetscFunctionBegin;
6120: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6121: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6122: MatCheckPreallocated(mat, 1);
6124: PetscUseTypeMethod(mat, zerorowscolumns, numRows, rows, diag, x, b);
6125: PetscCall(MatViewFromOptions(mat, NULL, "-mat_view"));
6126: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
6127: PetscFunctionReturn(PETSC_SUCCESS);
6128: }
6130: /*@
6131: MatZeroRowsColumnsIS - Zeros all entries (except possibly the main diagonal)
6132: of a set of rows and columns of a matrix.
6134: Collective
6136: Input Parameters:
6137: + mat - the matrix
6138: . is - the rows to zero
6139: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6140: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6141: - b - optional vector of right hand side, that will be adjusted by provided solution
6143: Level: intermediate
6145: Note:
6146: See `MatZeroRowsColumns()` for details on how this routine operates.
6148: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6149: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRows()`, `MatZeroRowsColumnsStencil()`
6150: @*/
6151: PetscErrorCode MatZeroRowsColumnsIS(Mat mat, IS is, PetscScalar diag, Vec x, Vec b)
6152: {
6153: PetscInt numRows;
6154: const PetscInt *rows;
6156: PetscFunctionBegin;
6161: PetscCall(ISGetLocalSize(is, &numRows));
6162: PetscCall(ISGetIndices(is, &rows));
6163: PetscCall(MatZeroRowsColumns(mat, numRows, rows, diag, x, b));
6164: PetscCall(ISRestoreIndices(is, &rows));
6165: PetscFunctionReturn(PETSC_SUCCESS);
6166: }
6168: /*@
6169: MatZeroRows - Zeros all entries (except possibly the main diagonal)
6170: of a set of rows of a matrix.
6172: Collective
6174: Input Parameters:
6175: + mat - the matrix
6176: . numRows - the number of rows to zero
6177: . rows - the global row indices
6178: . diag - value put in the diagonal of the zeroed rows
6179: . x - optional vector of solutions for zeroed rows (other entries in vector are not used), these must be set before this call
6180: - b - optional vector of right hand side, that will be adjusted by provided solution entries
6182: Level: intermediate
6184: Notes:
6185: This routine, along with `MatZeroRowsColumns()`, is typically used to eliminate known Dirichlet boundary conditions from a linear system.
6187: For each zeroed row, the value of the corresponding `b` is set to `diag` times the value of the corresponding `x`.
6189: If the resulting linear system is to be solved with `KSP` then one can (but does not have to) call `KSPSetInitialGuessNonzero()` to allow the
6190: Krylov method to take advantage of the known solution on the zeroed rows.
6192: May be followed by using a `PC` of type `PCREDISTRIBUTE` to solve the reduced problem (`PCDISTRIBUTE` completely eliminates the zeroed rows and their corresponding columns)
6193: from the matrix.
6195: Unlike `MatZeroRowsColumns()` for the `MATAIJ` and `MATBAIJ` matrix formats this removes the old nonzero structure, from the eliminated rows of the matrix
6196: but does not release memory. Because of this removal matrix-vector products with the adjusted matrix will be a bit faster. For the dense and block diagonal
6197: formats this does not alter the nonzero structure.
6199: If the option `MatSetOption`(mat,`MAT_KEEP_NONZERO_PATTERN`,`PETSC_TRUE`) the nonzero structure
6200: of the matrix is not changed the values are
6201: merely zeroed.
6203: The user can set a value in the diagonal entry (or for the `MATAIJ` format
6204: formats can optionally remove the main diagonal entry from the
6205: nonzero structure as well, by passing 0.0 as the final argument).
6207: For the parallel case, all processes that share the matrix (i.e.,
6208: those in the communicator used for matrix creation) MUST call this
6209: routine, regardless of whether any rows being zeroed are owned by
6210: them.
6212: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6213: list only rows local to itself).
6215: You can call `MatSetOption`(mat,`MAT_NO_OFF_PROC_ZERO_ROWS`,`PETSC_TRUE`) if each process indicates only rows it
6216: owns that are to be zeroed. This saves a global synchronization in the implementation.
6218: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6219: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`, `PCREDISTRIBUTE`
6220: @*/
6221: PetscErrorCode MatZeroRows(Mat mat, PetscInt numRows, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
6222: {
6223: PetscFunctionBegin;
6227: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6228: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6229: MatCheckPreallocated(mat, 1);
6231: PetscUseTypeMethod(mat, zerorows, numRows, rows, diag, x, b);
6232: PetscCall(MatViewFromOptions(mat, NULL, "-mat_view"));
6233: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
6234: PetscFunctionReturn(PETSC_SUCCESS);
6235: }
6237: /*@
6238: MatZeroRowsIS - Zeros all entries (except possibly the main diagonal)
6239: of a set of rows of a matrix.
6241: Collective
6243: Input Parameters:
6244: + mat - the matrix
6245: . is - index set of rows to remove (if `NULL` then no row is removed)
6246: . diag - value put in all diagonals of eliminated rows
6247: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6248: - b - optional vector of right hand side, that will be adjusted by provided solution
6250: Level: intermediate
6252: Note:
6253: See `MatZeroRows()` for details on how this routine operates.
6255: .seealso: [](ch_matrices), `Mat`, `MatZeroRows()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6256: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6257: @*/
6258: PetscErrorCode MatZeroRowsIS(Mat mat, IS is, PetscScalar diag, Vec x, Vec b)
6259: {
6260: PetscInt numRows = 0;
6261: const PetscInt *rows = NULL;
6263: PetscFunctionBegin;
6266: if (is) {
6268: PetscCall(ISGetLocalSize(is, &numRows));
6269: PetscCall(ISGetIndices(is, &rows));
6270: }
6271: PetscCall(MatZeroRows(mat, numRows, rows, diag, x, b));
6272: if (is) PetscCall(ISRestoreIndices(is, &rows));
6273: PetscFunctionReturn(PETSC_SUCCESS);
6274: }
6276: /*@
6277: MatZeroRowsStencil - Zeros all entries (except possibly the main diagonal)
6278: of a set of rows of a matrix. These rows must be local to the process.
6280: Collective
6282: Input Parameters:
6283: + mat - the matrix
6284: . numRows - the number of rows to remove
6285: . rows - the grid coordinates (and component number when dof > 1) for matrix rows
6286: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6287: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6288: - b - optional vector of right hand side, that will be adjusted by provided solution
6290: Level: intermediate
6292: Notes:
6293: See `MatZeroRows()` for details on how this routine operates.
6295: The grid coordinates are across the entire grid, not just the local portion
6297: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
6298: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
6299: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
6300: `DM_BOUNDARY_PERIODIC` boundary type.
6302: For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have
6303: a single value per point) you can skip filling those indices.
6305: Fortran Note:
6306: `idxm` and `idxn` should be declared as
6307: $ MatStencil idxm(4, m)
6308: and the values inserted using
6309: .vb
6310: idxm(MatStencil_i, 1) = i
6311: idxm(MatStencil_j, 1) = j
6312: idxm(MatStencil_k, 1) = k
6313: idxm(MatStencil_c, 1) = c
6314: etc
6315: .ve
6317: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsl()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6318: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6319: @*/
6320: PetscErrorCode MatZeroRowsStencil(Mat mat, PetscInt numRows, const MatStencil rows[], PetscScalar diag, Vec x, Vec b)
6321: {
6322: PetscInt dim = mat->stencil.dim;
6323: PetscInt sdim = dim - (1 - (PetscInt)mat->stencil.noc);
6324: PetscInt *dims = mat->stencil.dims + 1;
6325: PetscInt *starts = mat->stencil.starts;
6326: PetscInt *dxm = (PetscInt *)rows;
6327: PetscInt *jdxm, i, j, tmp, numNewRows = 0;
6329: PetscFunctionBegin;
6334: PetscCall(PetscMalloc1(numRows, &jdxm));
6335: for (i = 0; i < numRows; ++i) {
6336: /* Skip unused dimensions (they are ordered k, j, i, c) */
6337: for (j = 0; j < 3 - sdim; ++j) dxm++;
6338: /* Local index in X dir */
6339: tmp = *dxm++ - starts[0];
6340: /* Loop over remaining dimensions */
6341: for (j = 0; j < dim - 1; ++j) {
6342: /* If nonlocal, set index to be negative */
6343: if ((*dxm++ - starts[j + 1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
6344: /* Update local index */
6345: else tmp = tmp * dims[j] + *(dxm - 1) - starts[j + 1];
6346: }
6347: /* Skip component slot if necessary */
6348: if (mat->stencil.noc) dxm++;
6349: /* Local row number */
6350: if (tmp >= 0) jdxm[numNewRows++] = tmp;
6351: }
6352: PetscCall(MatZeroRowsLocal(mat, numNewRows, jdxm, diag, x, b));
6353: PetscCall(PetscFree(jdxm));
6354: PetscFunctionReturn(PETSC_SUCCESS);
6355: }
6357: /*@
6358: MatZeroRowsColumnsStencil - Zeros all row and column entries (except possibly the main diagonal)
6359: of a set of rows and columns of a matrix.
6361: Collective
6363: Input Parameters:
6364: + mat - the matrix
6365: . numRows - the number of rows/columns to remove
6366: . rows - the grid coordinates (and component number when dof > 1) for matrix rows
6367: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6368: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6369: - b - optional vector of right hand side, that will be adjusted by provided solution
6371: Level: intermediate
6373: Notes:
6374: See `MatZeroRowsColumns()` for details on how this routine operates.
6376: The grid coordinates are across the entire grid, not just the local portion
6378: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
6379: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
6380: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
6381: `DM_BOUNDARY_PERIODIC` boundary type.
6383: For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have
6384: a single value per point) you can skip filling those indices.
6386: Fortran Note:
6387: `idxm` and `idxn` should be declared as
6388: $ MatStencil idxm(4, m)
6389: and the values inserted using
6390: .vb
6391: idxm(MatStencil_i, 1) = i
6392: idxm(MatStencil_j, 1) = j
6393: idxm(MatStencil_k, 1) = k
6394: idxm(MatStencil_c, 1) = c
6395: etc
6396: .ve
6398: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6399: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRows()`
6400: @*/
6401: PetscErrorCode MatZeroRowsColumnsStencil(Mat mat, PetscInt numRows, const MatStencil rows[], PetscScalar diag, Vec x, Vec b)
6402: {
6403: PetscInt dim = mat->stencil.dim;
6404: PetscInt sdim = dim - (1 - (PetscInt)mat->stencil.noc);
6405: PetscInt *dims = mat->stencil.dims + 1;
6406: PetscInt *starts = mat->stencil.starts;
6407: PetscInt *dxm = (PetscInt *)rows;
6408: PetscInt *jdxm, i, j, tmp, numNewRows = 0;
6410: PetscFunctionBegin;
6415: PetscCall(PetscMalloc1(numRows, &jdxm));
6416: for (i = 0; i < numRows; ++i) {
6417: /* Skip unused dimensions (they are ordered k, j, i, c) */
6418: for (j = 0; j < 3 - sdim; ++j) dxm++;
6419: /* Local index in X dir */
6420: tmp = *dxm++ - starts[0];
6421: /* Loop over remaining dimensions */
6422: for (j = 0; j < dim - 1; ++j) {
6423: /* If nonlocal, set index to be negative */
6424: if ((*dxm++ - starts[j + 1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
6425: /* Update local index */
6426: else tmp = tmp * dims[j] + *(dxm - 1) - starts[j + 1];
6427: }
6428: /* Skip component slot if necessary */
6429: if (mat->stencil.noc) dxm++;
6430: /* Local row number */
6431: if (tmp >= 0) jdxm[numNewRows++] = tmp;
6432: }
6433: PetscCall(MatZeroRowsColumnsLocal(mat, numNewRows, jdxm, diag, x, b));
6434: PetscCall(PetscFree(jdxm));
6435: PetscFunctionReturn(PETSC_SUCCESS);
6436: }
6438: /*@C
6439: MatZeroRowsLocal - Zeros all entries (except possibly the main diagonal)
6440: of a set of rows of a matrix; using local numbering of rows.
6442: Collective
6444: Input Parameters:
6445: + mat - the matrix
6446: . numRows - the number of rows to remove
6447: . rows - the local row indices
6448: . diag - value put in all diagonals of eliminated rows
6449: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6450: - b - optional vector of right hand side, that will be adjusted by provided solution
6452: Level: intermediate
6454: Notes:
6455: Before calling `MatZeroRowsLocal()`, the user must first set the
6456: local-to-global mapping by calling MatSetLocalToGlobalMapping(), this is often already set for matrices obtained with `DMCreateMatrix()`.
6458: See `MatZeroRows()` for details on how this routine operates.
6460: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRows()`, `MatSetOption()`,
6461: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6462: @*/
6463: PetscErrorCode MatZeroRowsLocal(Mat mat, PetscInt numRows, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
6464: {
6465: PetscFunctionBegin;
6469: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6470: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6471: MatCheckPreallocated(mat, 1);
6473: if (mat->ops->zerorowslocal) {
6474: PetscUseTypeMethod(mat, zerorowslocal, numRows, rows, diag, x, b);
6475: } else {
6476: IS is, newis;
6477: const PetscInt *newRows;
6479: PetscCheck(mat->rmap->mapping, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Need to provide local to global mapping to matrix first");
6480: PetscCall(ISCreateGeneral(PETSC_COMM_SELF, numRows, rows, PETSC_COPY_VALUES, &is));
6481: PetscCall(ISLocalToGlobalMappingApplyIS(mat->rmap->mapping, is, &newis));
6482: PetscCall(ISGetIndices(newis, &newRows));
6483: PetscUseTypeMethod(mat, zerorows, numRows, newRows, diag, x, b);
6484: PetscCall(ISRestoreIndices(newis, &newRows));
6485: PetscCall(ISDestroy(&newis));
6486: PetscCall(ISDestroy(&is));
6487: }
6488: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
6489: PetscFunctionReturn(PETSC_SUCCESS);
6490: }
6492: /*@
6493: MatZeroRowsLocalIS - Zeros all entries (except possibly the main diagonal)
6494: of a set of rows of a matrix; using local numbering of rows.
6496: Collective
6498: Input Parameters:
6499: + mat - the matrix
6500: . is - index set of rows to remove
6501: . diag - value put in all diagonals of eliminated rows
6502: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6503: - b - optional vector of right hand side, that will be adjusted by provided solution
6505: Level: intermediate
6507: Notes:
6508: Before calling `MatZeroRowsLocalIS()`, the user must first set the
6509: local-to-global mapping by calling `MatSetLocalToGlobalMapping()`, this is often already set for matrices obtained with `DMCreateMatrix()`.
6511: See `MatZeroRows()` for details on how this routine operates.
6513: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRows()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6514: `MatZeroRowsColumnsLocal()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6515: @*/
6516: PetscErrorCode MatZeroRowsLocalIS(Mat mat, IS is, PetscScalar diag, Vec x, Vec b)
6517: {
6518: PetscInt numRows;
6519: const PetscInt *rows;
6521: PetscFunctionBegin;
6525: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6526: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6527: MatCheckPreallocated(mat, 1);
6529: PetscCall(ISGetLocalSize(is, &numRows));
6530: PetscCall(ISGetIndices(is, &rows));
6531: PetscCall(MatZeroRowsLocal(mat, numRows, rows, diag, x, b));
6532: PetscCall(ISRestoreIndices(is, &rows));
6533: PetscFunctionReturn(PETSC_SUCCESS);
6534: }
6536: /*@
6537: MatZeroRowsColumnsLocal - Zeros all entries (except possibly the main diagonal)
6538: of a set of rows and columns of a matrix; using local numbering of rows.
6540: Collective
6542: Input Parameters:
6543: + mat - the matrix
6544: . numRows - the number of rows to remove
6545: . rows - the global row indices
6546: . diag - value put in all diagonals of eliminated rows
6547: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6548: - b - optional vector of right hand side, that will be adjusted by provided solution
6550: Level: intermediate
6552: Notes:
6553: Before calling `MatZeroRowsColumnsLocal()`, the user must first set the
6554: local-to-global mapping by calling `MatSetLocalToGlobalMapping()`, this is often already set for matrices obtained with `DMCreateMatrix()`.
6556: See `MatZeroRowsColumns()` for details on how this routine operates.
6558: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6559: `MatZeroRows()`, `MatZeroRowsColumnsLocalIS()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6560: @*/
6561: PetscErrorCode MatZeroRowsColumnsLocal(Mat mat, PetscInt numRows, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
6562: {
6563: IS is, newis;
6564: const PetscInt *newRows;
6566: PetscFunctionBegin;
6570: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6571: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6572: MatCheckPreallocated(mat, 1);
6574: PetscCheck(mat->cmap->mapping, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Need to provide local to global mapping to matrix first");
6575: PetscCall(ISCreateGeneral(PETSC_COMM_SELF, numRows, rows, PETSC_COPY_VALUES, &is));
6576: PetscCall(ISLocalToGlobalMappingApplyIS(mat->cmap->mapping, is, &newis));
6577: PetscCall(ISGetIndices(newis, &newRows));
6578: PetscUseTypeMethod(mat, zerorowscolumns, numRows, newRows, diag, x, b);
6579: PetscCall(ISRestoreIndices(newis, &newRows));
6580: PetscCall(ISDestroy(&newis));
6581: PetscCall(ISDestroy(&is));
6582: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
6583: PetscFunctionReturn(PETSC_SUCCESS);
6584: }
6586: /*@
6587: MatZeroRowsColumnsLocalIS - Zeros all entries (except possibly the main diagonal)
6588: of a set of rows and columns of a matrix; using local numbering of rows.
6590: Collective
6592: Input Parameters:
6593: + mat - the matrix
6594: . is - index set of rows to remove
6595: . diag - value put in all diagonals of eliminated rows
6596: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6597: - b - optional vector of right hand side, that will be adjusted by provided solution
6599: Level: intermediate
6601: Notes:
6602: Before calling `MatZeroRowsColumnsLocalIS()`, the user must first set the
6603: local-to-global mapping by calling `MatSetLocalToGlobalMapping()`, this is often already set for matrices obtained with `DMCreateMatrix()`.
6605: See `MatZeroRowsColumns()` for details on how this routine operates.
6607: .seealso: [](ch_matrices), `Mat`, `MatZeroRowsIS()`, `MatZeroRowsColumns()`, `MatZeroRowsLocalIS()`, `MatZeroRowsStencil()`, `MatZeroEntries()`, `MatZeroRowsLocal()`, `MatSetOption()`,
6608: `MatZeroRowsColumnsLocal()`, `MatZeroRows()`, `MatZeroRowsColumnsIS()`, `MatZeroRowsColumnsStencil()`
6609: @*/
6610: PetscErrorCode MatZeroRowsColumnsLocalIS(Mat mat, IS is, PetscScalar diag, Vec x, Vec b)
6611: {
6612: PetscInt numRows;
6613: const PetscInt *rows;
6615: PetscFunctionBegin;
6619: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6620: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6621: MatCheckPreallocated(mat, 1);
6623: PetscCall(ISGetLocalSize(is, &numRows));
6624: PetscCall(ISGetIndices(is, &rows));
6625: PetscCall(MatZeroRowsColumnsLocal(mat, numRows, rows, diag, x, b));
6626: PetscCall(ISRestoreIndices(is, &rows));
6627: PetscFunctionReturn(PETSC_SUCCESS);
6628: }
6630: /*@C
6631: MatGetSize - Returns the numbers of rows and columns in a matrix.
6633: Not Collective
6635: Input Parameter:
6636: . mat - the matrix
6638: Output Parameters:
6639: + m - the number of global rows
6640: - n - the number of global columns
6642: Level: beginner
6644: Note:
6645: Both output parameters can be `NULL` on input.
6647: .seealso: [](ch_matrices), `Mat`, `MatSetSizes()`, `MatGetLocalSize()`
6648: @*/
6649: PetscErrorCode MatGetSize(Mat mat, PetscInt *m, PetscInt *n)
6650: {
6651: PetscFunctionBegin;
6653: if (m) *m = mat->rmap->N;
6654: if (n) *n = mat->cmap->N;
6655: PetscFunctionReturn(PETSC_SUCCESS);
6656: }
6658: /*@C
6659: MatGetLocalSize - For most matrix formats, excluding `MATELEMENTAL` and `MATSCALAPACK`, Returns the number of local rows and local columns
6660: of a matrix. For all matrices this is the local size of the left and right vectors as returned by `MatCreateVecs()`.
6662: Not Collective
6664: Input Parameter:
6665: . mat - the matrix
6667: Output Parameters:
6668: + m - the number of local rows, use `NULL` to not obtain this value
6669: - n - the number of local columns, use `NULL` to not obtain this value
6671: Level: beginner
6673: .seealso: [](ch_matrices), `Mat`, `MatSetSizes()`, `MatGetSize()`
6674: @*/
6675: PetscErrorCode MatGetLocalSize(Mat mat, PetscInt *m, PetscInt *n)
6676: {
6677: PetscFunctionBegin;
6681: if (m) *m = mat->rmap->n;
6682: if (n) *n = mat->cmap->n;
6683: PetscFunctionReturn(PETSC_SUCCESS);
6684: }
6686: /*@C
6687: MatGetOwnershipRangeColumn - Returns the range of matrix columns associated with rows of a vector one multiplies this matrix by that are owned by
6688: this processor. (The columns of the "diagonal block" for most sparse matrix formats). See [Matrix Layouts](sec_matlayout) for details on matrix layouts.
6690: Not Collective, unless matrix has not been allocated, then collective
6692: Input Parameter:
6693: . mat - the matrix
6695: Output Parameters:
6696: + m - the global index of the first local column, use `NULL` to not obtain this value
6697: - n - one more than the global index of the last local column, use `NULL` to not obtain this value
6699: Level: developer
6701: .seealso: [](ch_matrices), `Mat`, `MatGetOwnershipRange()`, `MatGetOwnershipRanges()`, `MatGetOwnershipRangesColumn()`, `PetscLayout`
6702: @*/
6703: PetscErrorCode MatGetOwnershipRangeColumn(Mat mat, PetscInt *m, PetscInt *n)
6704: {
6705: PetscFunctionBegin;
6710: MatCheckPreallocated(mat, 1);
6711: if (m) *m = mat->cmap->rstart;
6712: if (n) *n = mat->cmap->rend;
6713: PetscFunctionReturn(PETSC_SUCCESS);
6714: }
6716: /*@C
6717: MatGetOwnershipRange - For matrices that own values by row, excludes `MATELEMENTAL` and `MATSCALAPACK`, returns the range of matrix rows owned by
6718: this MPI rank. For all matrices it returns the range of matrix rows associated with rows of a vector that would contain the result of a matrix
6719: vector product with this matrix. See [Matrix Layouts](sec_matlayout) for details on matrix layouts
6721: Not Collective
6723: Input Parameter:
6724: . mat - the matrix
6726: Output Parameters:
6727: + m - the global index of the first local row, use `NULL` to not obtain this value
6728: - n - one more than the global index of the last local row, use `NULL` to not obtain this value
6730: Level: beginner
6732: Note:
6733: This function requires that the matrix be preallocated. If you have not preallocated, consider using
6734: `PetscSplitOwnership`(`MPI_Comm` comm, `PetscInt` *n, `PetscInt` *N)
6735: and then `MPI_Scan()` to calculate prefix sums of the local sizes.
6737: .seealso: [](ch_matrices), `Mat`, `MatGetOwnershipRanges()`, `MatGetOwnershipRangeColumn()`, `MatGetOwnershipRangesColumn()`, `PetscSplitOwnership()`, `PetscSplitOwnershipBlock()`,
6738: `PetscLayout`
6739: @*/
6740: PetscErrorCode MatGetOwnershipRange(Mat mat, PetscInt *m, PetscInt *n)
6741: {
6742: PetscFunctionBegin;
6747: MatCheckPreallocated(mat, 1);
6748: if (m) *m = mat->rmap->rstart;
6749: if (n) *n = mat->rmap->rend;
6750: PetscFunctionReturn(PETSC_SUCCESS);
6751: }
6753: /*@C
6754: MatGetOwnershipRanges - For matrices that own values by row, excludes `MATELEMENTAL` and `MATSCALAPACK`, returns the range of matrix rows owned by
6755: each process. For all matrices it returns the ranges of matrix rows associated with rows of a vector that would contain the result of a matrix
6756: vector product with this matrix. See [Matrix Layouts](sec_matlayout) for details on matrix layouts
6758: Not Collective, unless matrix has not been allocated
6760: Input Parameter:
6761: . mat - the matrix
6763: Output Parameter:
6764: . ranges - start of each processors portion plus one more than the total length at the end
6766: Level: beginner
6768: .seealso: [](ch_matrices), `Mat`, `MatGetOwnershipRange()`, `MatGetOwnershipRangeColumn()`, `MatGetOwnershipRangesColumn()`, `PetscLayout`
6769: @*/
6770: PetscErrorCode MatGetOwnershipRanges(Mat mat, const PetscInt **ranges)
6771: {
6772: PetscFunctionBegin;
6775: MatCheckPreallocated(mat, 1);
6776: PetscCall(PetscLayoutGetRanges(mat->rmap, ranges));
6777: PetscFunctionReturn(PETSC_SUCCESS);
6778: }
6780: /*@C
6781: MatGetOwnershipRangesColumn - Returns the ranges of matrix columns associated with rows of a vector one multiplies this vector by that are owned by
6782: each processor. (The columns of the "diagonal blocks", for most sparse matrix formats). See [Matrix Layouts](sec_matlayout) for details on matrix layouts.
6784: Not Collective, unless matrix has not been allocated
6786: Input Parameter:
6787: . mat - the matrix
6789: Output Parameter:
6790: . ranges - start of each processors portion plus one more then the total length at the end
6792: Level: beginner
6794: .seealso: [](ch_matrices), `Mat`, `MatGetOwnershipRange()`, `MatGetOwnershipRangeColumn()`, `MatGetOwnershipRanges()`
6795: @*/
6796: PetscErrorCode MatGetOwnershipRangesColumn(Mat mat, const PetscInt **ranges)
6797: {
6798: PetscFunctionBegin;
6801: MatCheckPreallocated(mat, 1);
6802: PetscCall(PetscLayoutGetRanges(mat->cmap, ranges));
6803: PetscFunctionReturn(PETSC_SUCCESS);
6804: }
6806: /*@C
6807: MatGetOwnershipIS - Get row and column ownership of a matrices' values as index sets. For most matrices, excluding `MATELEMENTAL` and `MATSCALAPACK`, this
6808: corresponds to values returned by `MatGetOwnershipRange()`, `MatGetOwnershipRangeColumn()`. For `MATELEMENTAL` and `MATSCALAPACK` the ownership
6809: is more complicated. See [Matrix Layouts](sec_matlayout) for details on matrix layouts.
6811: Not Collective
6813: Input Parameter:
6814: . A - matrix
6816: Output Parameters:
6817: + rows - rows in which this process owns elements, , use `NULL` to not obtain this value
6818: - cols - columns in which this process owns elements, use `NULL` to not obtain this value
6820: Level: intermediate
6822: .seealso: [](ch_matrices), `Mat`, `MatGetOwnershipRange()`, `MatGetOwnershipRangeColumn()`, `MatSetValues()`, ``MATELEMENTAL``, ``MATSCALAPACK``
6823: @*/
6824: PetscErrorCode MatGetOwnershipIS(Mat A, IS *rows, IS *cols)
6825: {
6826: PetscErrorCode (*f)(Mat, IS *, IS *);
6828: PetscFunctionBegin;
6829: MatCheckPreallocated(A, 1);
6830: PetscCall(PetscObjectQueryFunction((PetscObject)A, "MatGetOwnershipIS_C", &f));
6831: if (f) {
6832: PetscCall((*f)(A, rows, cols));
6833: } else { /* Create a standard row-based partition, each process is responsible for ALL columns in their row block */
6834: if (rows) PetscCall(ISCreateStride(PETSC_COMM_SELF, A->rmap->n, A->rmap->rstart, 1, rows));
6835: if (cols) PetscCall(ISCreateStride(PETSC_COMM_SELF, A->cmap->N, 0, 1, cols));
6836: }
6837: PetscFunctionReturn(PETSC_SUCCESS);
6838: }
6840: /*@C
6841: MatILUFactorSymbolic - Performs symbolic ILU factorization of a matrix obtained with `MatGetFactor()`
6842: Uses levels of fill only, not drop tolerance. Use `MatLUFactorNumeric()`
6843: to complete the factorization.
6845: Collective
6847: Input Parameters:
6848: + fact - the factorized matrix obtained with `MatGetFactor()`
6849: . mat - the matrix
6850: . row - row permutation
6851: . col - column permutation
6852: - info - structure containing
6853: .vb
6854: levels - number of levels of fill.
6855: expected fill - as ratio of original fill.
6856: 1 or 0 - indicating force fill on diagonal (improves robustness for matrices
6857: missing diagonal entries)
6858: .ve
6860: Level: developer
6862: Notes:
6863: See [Matrix Factorization](sec_matfactor) for additional information.
6865: Most users should employ the `KSP` interface for linear solvers
6866: instead of working directly with matrix algebra routines such as this.
6867: See, e.g., `KSPCreate()`.
6869: Uses the definition of level of fill as in Y. Saad, 2003
6871: Developer Note:
6872: The Fortran interface is not autogenerated as the
6873: interface definition cannot be generated correctly [due to `MatFactorInfo`]
6875: References:
6876: . * - Y. Saad, Iterative methods for sparse linear systems Philadelphia: Society for Industrial and Applied Mathematics, 2003
6878: .seealso: [](ch_matrices), `Mat`, [Matrix Factorization](sec_matfactor), `MatGetFactor()`, `MatLUFactorSymbolic()`, `MatLUFactorNumeric()`, `MatCholeskyFactor()`
6879: `MatGetOrdering()`, `MatFactorInfo`
6880: @*/
6881: PetscErrorCode MatILUFactorSymbolic(Mat fact, Mat mat, IS row, IS col, const MatFactorInfo *info)
6882: {
6883: PetscFunctionBegin;
6890: PetscCheck(info->levels >= 0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Levels of fill negative %" PetscInt_FMT, (PetscInt)info->levels);
6891: PetscCheck(info->fill >= 1.0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Expected fill less than 1.0 %g", (double)info->fill);
6892: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6893: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6894: MatCheckPreallocated(mat, 2);
6896: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_ILUFactorSymbolic, mat, row, col, 0));
6897: PetscUseTypeMethod(fact, ilufactorsymbolic, mat, row, col, info);
6898: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_ILUFactorSymbolic, mat, row, col, 0));
6899: PetscFunctionReturn(PETSC_SUCCESS);
6900: }
6902: /*@C
6903: MatICCFactorSymbolic - Performs symbolic incomplete
6904: Cholesky factorization for a symmetric matrix. Use
6905: `MatCholeskyFactorNumeric()` to complete the factorization.
6907: Collective
6909: Input Parameters:
6910: + fact - the factorized matrix obtained with `MatGetFactor()`
6911: . mat - the matrix to be factored
6912: . perm - row and column permutation
6913: - info - structure containing
6914: .vb
6915: levels - number of levels of fill.
6916: expected fill - as ratio of original fill.
6917: .ve
6919: Level: developer
6921: Notes:
6922: Most users should employ the `KSP` interface for linear solvers
6923: instead of working directly with matrix algebra routines such as this.
6924: See, e.g., `KSPCreate()`.
6926: This uses the definition of level of fill as in Y. Saad, 2003
6928: Developer Note:
6929: The Fortran interface is not autogenerated as the
6930: interface definition cannot be generated correctly [due to `MatFactorInfo`]
6932: References:
6933: . * - Y. Saad, Iterative methods for sparse linear systems Philadelphia: Society for Industrial and Applied Mathematics, 2003
6935: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatCholeskyFactorNumeric()`, `MatCholeskyFactor()`, `MatFactorInfo`
6936: @*/
6937: PetscErrorCode MatICCFactorSymbolic(Mat fact, Mat mat, IS perm, const MatFactorInfo *info)
6938: {
6939: PetscFunctionBegin;
6945: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
6946: PetscCheck(info->levels >= 0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Levels negative %" PetscInt_FMT, (PetscInt)info->levels);
6947: PetscCheck(info->fill >= 1.0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Expected fill less than 1.0 %g", (double)info->fill);
6948: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
6949: MatCheckPreallocated(mat, 2);
6951: if (!fact->trivialsymbolic) PetscCall(PetscLogEventBegin(MAT_ICCFactorSymbolic, mat, perm, 0, 0));
6952: PetscUseTypeMethod(fact, iccfactorsymbolic, mat, perm, info);
6953: if (!fact->trivialsymbolic) PetscCall(PetscLogEventEnd(MAT_ICCFactorSymbolic, mat, perm, 0, 0));
6954: PetscFunctionReturn(PETSC_SUCCESS);
6955: }
6957: /*@C
6958: MatCreateSubMatrices - Extracts several submatrices from a matrix. If submat
6959: points to an array of valid matrices, they may be reused to store the new
6960: submatrices.
6962: Collective
6964: Input Parameters:
6965: + mat - the matrix
6966: . n - the number of submatrixes to be extracted (on this processor, may be zero)
6967: . irow - index set of rows to extract
6968: . icol - index set of columns to extract
6969: - scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
6971: Output Parameter:
6972: . submat - the array of submatrices
6974: Level: advanced
6976: Notes:
6977: `MatCreateSubMatrices()` can extract ONLY sequential submatrices
6978: (from both sequential and parallel matrices). Use `MatCreateSubMatrix()`
6979: to extract a parallel submatrix.
6981: Some matrix types place restrictions on the row and column
6982: indices, such as that they be sorted or that they be equal to each other.
6984: The index sets may not have duplicate entries.
6986: When extracting submatrices from a parallel matrix, each processor can
6987: form a different submatrix by setting the rows and columns of its
6988: individual index sets according to the local submatrix desired.
6990: When finished using the submatrices, the user should destroy
6991: them with `MatDestroySubMatrices()`.
6993: `MAT_REUSE_MATRIX` can only be used when the nonzero structure of the
6994: original matrix has not changed from that last call to `MatCreateSubMatrices()`.
6996: This routine creates the matrices in submat; you should NOT create them before
6997: calling it. It also allocates the array of matrix pointers submat.
6999: For `MATBAIJ` matrices the index sets must respect the block structure, that is if they
7000: request one row/column in a block, they must request all rows/columns that are in
7001: that block. For example, if the block size is 2 you cannot request just row 0 and
7002: column 0.
7004: Fortran Note:
7005: The Fortran interface is slightly different from that given below; it
7006: requires one to pass in as `submat` a `Mat` (integer) array of size at least n+1.
7008: .seealso: [](ch_matrices), `Mat`, `MatDestroySubMatrices()`, `MatCreateSubMatrix()`, `MatGetRow()`, `MatGetDiagonal()`, `MatReuse`
7009: @*/
7010: PetscErrorCode MatCreateSubMatrices(Mat mat, PetscInt n, const IS irow[], const IS icol[], MatReuse scall, Mat *submat[])
7011: {
7012: PetscInt i;
7013: PetscBool eq;
7015: PetscFunctionBegin;
7018: if (n) {
7023: }
7025: if (n && scall == MAT_REUSE_MATRIX) {
7028: }
7029: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
7030: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
7031: MatCheckPreallocated(mat, 1);
7032: PetscCall(PetscLogEventBegin(MAT_CreateSubMats, mat, 0, 0, 0));
7033: PetscUseTypeMethod(mat, createsubmatrices, n, irow, icol, scall, submat);
7034: PetscCall(PetscLogEventEnd(MAT_CreateSubMats, mat, 0, 0, 0));
7035: for (i = 0; i < n; i++) {
7036: (*submat)[i]->factortype = MAT_FACTOR_NONE; /* in case in place factorization was previously done on submatrix */
7037: PetscCall(ISEqualUnsorted(irow[i], icol[i], &eq));
7038: if (eq) PetscCall(MatPropagateSymmetryOptions(mat, (*submat)[i]));
7039: #if defined(PETSC_HAVE_VIENNACL) || defined(PETSC_HAVE_CUDA) || defined(PETSC_HAVE_HIP)
7040: if (mat->boundtocpu && mat->bindingpropagates) {
7041: PetscCall(MatBindToCPU((*submat)[i], PETSC_TRUE));
7042: PetscCall(MatSetBindingPropagates((*submat)[i], PETSC_TRUE));
7043: }
7044: #endif
7045: }
7046: PetscFunctionReturn(PETSC_SUCCESS);
7047: }
7049: /*@C
7050: MatCreateSubMatricesMPI - Extracts MPI submatrices across a sub communicator of mat (by pairs of `IS` that may live on subcomms).
7052: Collective
7054: Input Parameters:
7055: + mat - the matrix
7056: . n - the number of submatrixes to be extracted
7057: . irow - index set of rows to extract
7058: . icol - index set of columns to extract
7059: - scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
7061: Output Parameter:
7062: . submat - the array of submatrices
7064: Level: advanced
7066: Note:
7067: This is used by `PCGASM`
7069: .seealso: [](ch_matrices), `Mat`, `PCGASM`, `MatCreateSubMatrices()`, `MatCreateSubMatrix()`, `MatGetRow()`, `MatGetDiagonal()`, `MatReuse`
7070: @*/
7071: PetscErrorCode MatCreateSubMatricesMPI(Mat mat, PetscInt n, const IS irow[], const IS icol[], MatReuse scall, Mat *submat[])
7072: {
7073: PetscInt i;
7074: PetscBool eq;
7076: PetscFunctionBegin;
7079: if (n) {
7084: }
7086: if (n && scall == MAT_REUSE_MATRIX) {
7089: }
7090: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
7091: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
7092: MatCheckPreallocated(mat, 1);
7094: PetscCall(PetscLogEventBegin(MAT_CreateSubMats, mat, 0, 0, 0));
7095: PetscUseTypeMethod(mat, createsubmatricesmpi, n, irow, icol, scall, submat);
7096: PetscCall(PetscLogEventEnd(MAT_CreateSubMats, mat, 0, 0, 0));
7097: for (i = 0; i < n; i++) {
7098: PetscCall(ISEqualUnsorted(irow[i], icol[i], &eq));
7099: if (eq) PetscCall(MatPropagateSymmetryOptions(mat, (*submat)[i]));
7100: }
7101: PetscFunctionReturn(PETSC_SUCCESS);
7102: }
7104: /*@C
7105: MatDestroyMatrices - Destroys an array of matrices.
7107: Collective
7109: Input Parameters:
7110: + n - the number of local matrices
7111: - mat - the matrices (this is a pointer to the array of matrices)
7113: Level: advanced
7115: Note:
7116: Frees not only the matrices, but also the array that contains the matrices
7118: Fortran Note:
7119: This does not free the array.
7121: .seealso: [](ch_matrices), `Mat`, `MatCreateSubMatrices()` `MatDestroySubMatrices()`
7122: @*/
7123: PetscErrorCode MatDestroyMatrices(PetscInt n, Mat *mat[])
7124: {
7125: PetscInt i;
7127: PetscFunctionBegin;
7128: if (!*mat) PetscFunctionReturn(PETSC_SUCCESS);
7129: PetscCheck(n >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Trying to destroy negative number of matrices %" PetscInt_FMT, n);
7132: for (i = 0; i < n; i++) PetscCall(MatDestroy(&(*mat)[i]));
7134: /* memory is allocated even if n = 0 */
7135: PetscCall(PetscFree(*mat));
7136: PetscFunctionReturn(PETSC_SUCCESS);
7137: }
7139: /*@C
7140: MatDestroySubMatrices - Destroys a set of matrices obtained with `MatCreateSubMatrices()`.
7142: Collective
7144: Input Parameters:
7145: + n - the number of local matrices
7146: - mat - the matrices (this is a pointer to the array of matrices, just to match the calling
7147: sequence of `MatCreateSubMatrices()`)
7149: Level: advanced
7151: Note:
7152: Frees not only the matrices, but also the array that contains the matrices
7154: Fortran Note:
7155: This does not free the array.
7157: .seealso: [](ch_matrices), `Mat`, `MatCreateSubMatrices()`, `MatDestroyMatrices()`
7158: @*/
7159: PetscErrorCode MatDestroySubMatrices(PetscInt n, Mat *mat[])
7160: {
7161: Mat mat0;
7163: PetscFunctionBegin;
7164: if (!*mat) PetscFunctionReturn(PETSC_SUCCESS);
7165: /* mat[] is an array of length n+1, see MatCreateSubMatrices_xxx() */
7166: PetscCheck(n >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Trying to destroy negative number of matrices %" PetscInt_FMT, n);
7169: mat0 = (*mat)[0];
7170: if (mat0 && mat0->ops->destroysubmatrices) {
7171: PetscCall((*mat0->ops->destroysubmatrices)(n, mat));
7172: } else {
7173: PetscCall(MatDestroyMatrices(n, mat));
7174: }
7175: PetscFunctionReturn(PETSC_SUCCESS);
7176: }
7178: /*@C
7179: MatGetSeqNonzeroStructure - Extracts the nonzero structure from a matrix and stores it, in its entirety, on each process
7181: Collective
7183: Input Parameter:
7184: . mat - the matrix
7186: Output Parameter:
7187: . matstruct - the sequential matrix with the nonzero structure of mat
7189: Level: developer
7191: .seealso: [](ch_matrices), `Mat`, `MatDestroySeqNonzeroStructure()`, `MatCreateSubMatrices()`, `MatDestroyMatrices()`
7192: @*/
7193: PetscErrorCode MatGetSeqNonzeroStructure(Mat mat, Mat *matstruct)
7194: {
7195: PetscFunctionBegin;
7200: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
7201: MatCheckPreallocated(mat, 1);
7203: PetscCall(PetscLogEventBegin(MAT_GetSeqNonzeroStructure, mat, 0, 0, 0));
7204: PetscUseTypeMethod(mat, getseqnonzerostructure, matstruct);
7205: PetscCall(PetscLogEventEnd(MAT_GetSeqNonzeroStructure, mat, 0, 0, 0));
7206: PetscFunctionReturn(PETSC_SUCCESS);
7207: }
7209: /*@C
7210: MatDestroySeqNonzeroStructure - Destroys matrix obtained with `MatGetSeqNonzeroStructure()`.
7212: Collective
7214: Input Parameter:
7215: . mat - the matrix (this is a pointer to the array of matrices, just to match the calling
7216: sequence of `MatGetSeqNonzeroStructure()`)
7218: Level: advanced
7220: Note:
7221: Frees not only the matrices, but also the array that contains the matrices
7223: .seealso: [](ch_matrices), `Mat`, `MatGetSeqNonzeroStructure()`
7224: @*/
7225: PetscErrorCode MatDestroySeqNonzeroStructure(Mat *mat)
7226: {
7227: PetscFunctionBegin;
7229: PetscCall(MatDestroy(mat));
7230: PetscFunctionReturn(PETSC_SUCCESS);
7231: }
7233: /*@
7234: MatIncreaseOverlap - Given a set of submatrices indicated by index sets,
7235: replaces the index sets by larger ones that represent submatrices with
7236: additional overlap.
7238: Collective
7240: Input Parameters:
7241: + mat - the matrix
7242: . n - the number of index sets
7243: . is - the array of index sets (these index sets will changed during the call)
7244: - ov - the additional overlap requested
7246: Options Database Key:
7247: . -mat_increase_overlap_scalable - use a scalable algorithm to compute the overlap (supported by MPIAIJ matrix)
7249: Level: developer
7251: Note:
7252: The computed overlap preserves the matrix block sizes when the blocks are square.
7253: That is: if a matrix nonzero for a given block would increase the overlap all columns associated with
7254: that block are included in the overlap regardless of whether each specific column would increase the overlap.
7256: .seealso: [](ch_matrices), `Mat`, `PCASM`, `MatSetBlockSize()`, `MatIncreaseOverlapSplit()`, `MatCreateSubMatrices()`
7257: @*/
7258: PetscErrorCode MatIncreaseOverlap(Mat mat, PetscInt n, IS is[], PetscInt ov)
7259: {
7260: PetscInt i, bs, cbs;
7262: PetscFunctionBegin;
7266: PetscCheck(n >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Must have one or more domains, you have %" PetscInt_FMT, n);
7267: if (n) {
7270: }
7271: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
7272: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
7273: MatCheckPreallocated(mat, 1);
7275: if (!ov || !n) PetscFunctionReturn(PETSC_SUCCESS);
7276: PetscCall(PetscLogEventBegin(MAT_IncreaseOverlap, mat, 0, 0, 0));
7277: PetscUseTypeMethod(mat, increaseoverlap, n, is, ov);
7278: PetscCall(PetscLogEventEnd(MAT_IncreaseOverlap, mat, 0, 0, 0));
7279: PetscCall(MatGetBlockSizes(mat, &bs, &cbs));
7280: if (bs == cbs) {
7281: for (i = 0; i < n; i++) PetscCall(ISSetBlockSize(is[i], bs));
7282: }
7283: PetscFunctionReturn(PETSC_SUCCESS);
7284: }
7286: PetscErrorCode MatIncreaseOverlapSplit_Single(Mat, IS *, PetscInt);
7288: /*@
7289: MatIncreaseOverlapSplit - Given a set of submatrices indicated by index sets across
7290: a sub communicator, replaces the index sets by larger ones that represent submatrices with
7291: additional overlap.
7293: Collective
7295: Input Parameters:
7296: + mat - the matrix
7297: . n - the number of index sets
7298: . is - the array of index sets (these index sets will changed during the call)
7299: - ov - the additional overlap requested
7301: ` Options Database Key:
7302: . -mat_increase_overlap_scalable - use a scalable algorithm to compute the overlap (supported by MPIAIJ matrix)
7304: Level: developer
7306: .seealso: [](ch_matrices), `Mat`, `MatCreateSubMatrices()`, `MatIncreaseOverlap()`
7307: @*/
7308: PetscErrorCode MatIncreaseOverlapSplit(Mat mat, PetscInt n, IS is[], PetscInt ov)
7309: {
7310: PetscInt i;
7312: PetscFunctionBegin;
7315: PetscCheck(n >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Must have one or more domains, you have %" PetscInt_FMT, n);
7316: if (n) {
7319: }
7320: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
7321: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
7322: MatCheckPreallocated(mat, 1);
7323: if (!ov) PetscFunctionReturn(PETSC_SUCCESS);
7324: PetscCall(PetscLogEventBegin(MAT_IncreaseOverlap, mat, 0, 0, 0));
7325: for (i = 0; i < n; i++) PetscCall(MatIncreaseOverlapSplit_Single(mat, &is[i], ov));
7326: PetscCall(PetscLogEventEnd(MAT_IncreaseOverlap, mat, 0, 0, 0));
7327: PetscFunctionReturn(PETSC_SUCCESS);
7328: }
7330: /*@
7331: MatGetBlockSize - Returns the matrix block size.
7333: Not Collective
7335: Input Parameter:
7336: . mat - the matrix
7338: Output Parameter:
7339: . bs - block size
7341: Level: intermediate
7343: Notes:
7344: Block row formats are `MATBAIJ` and `MATSBAIJ` ALWAYS have square block storage in the matrix.
7346: If the block size has not been set yet this routine returns 1.
7348: .seealso: [](ch_matrices), `Mat`, `MATBAIJ`, `MATSBAIJ`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSizes()`
7349: @*/
7350: PetscErrorCode MatGetBlockSize(Mat mat, PetscInt *bs)
7351: {
7352: PetscFunctionBegin;
7355: *bs = PetscAbs(mat->rmap->bs);
7356: PetscFunctionReturn(PETSC_SUCCESS);
7357: }
7359: /*@
7360: MatGetBlockSizes - Returns the matrix block row and column sizes.
7362: Not Collective
7364: Input Parameter:
7365: . mat - the matrix
7367: Output Parameters:
7368: + rbs - row block size
7369: - cbs - column block size
7371: Level: intermediate
7373: Notes:
7374: Block row formats are `MATBAIJ` and `MATSBAIJ` ALWAYS have square block storage in the matrix.
7375: If you pass a different block size for the columns than the rows, the row block size determines the square block storage.
7377: If a block size has not been set yet this routine returns 1.
7379: .seealso: [](ch_matrices), `Mat`, `MATBAIJ`, `MATSBAIJ`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSize()`, `MatSetBlockSizes()`
7380: @*/
7381: PetscErrorCode MatGetBlockSizes(Mat mat, PetscInt *rbs, PetscInt *cbs)
7382: {
7383: PetscFunctionBegin;
7387: if (rbs) *rbs = PetscAbs(mat->rmap->bs);
7388: if (cbs) *cbs = PetscAbs(mat->cmap->bs);
7389: PetscFunctionReturn(PETSC_SUCCESS);
7390: }
7392: /*@
7393: MatSetBlockSize - Sets the matrix block size.
7395: Logically Collective
7397: Input Parameters:
7398: + mat - the matrix
7399: - bs - block size
7401: Level: intermediate
7403: Notes:
7404: Block row formats are `MATBAIJ` and `MATSBAIJ` formats ALWAYS have square block storage in the matrix.
7405: This must be called before `MatSetUp()` or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later.
7407: For `MATAIJ` matrix format, this function can be called at a later stage, provided that the specified block size
7408: is compatible with the matrix local sizes.
7410: .seealso: [](ch_matrices), `Mat`, `MATBAIJ`, `MATSBAIJ`, `MATAIJ`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSizes()`, `MatGetBlockSizes()`
7411: @*/
7412: PetscErrorCode MatSetBlockSize(Mat mat, PetscInt bs)
7413: {
7414: PetscFunctionBegin;
7417: PetscCall(MatSetBlockSizes(mat, bs, bs));
7418: PetscFunctionReturn(PETSC_SUCCESS);
7419: }
7421: typedef struct {
7422: PetscInt n;
7423: IS *is;
7424: Mat *mat;
7425: PetscObjectState nonzerostate;
7426: Mat C;
7427: } EnvelopeData;
7429: static PetscErrorCode EnvelopeDataDestroy(EnvelopeData *edata)
7430: {
7431: for (PetscInt i = 0; i < edata->n; i++) PetscCall(ISDestroy(&edata->is[i]));
7432: PetscCall(PetscFree(edata->is));
7433: PetscCall(PetscFree(edata));
7434: return PETSC_SUCCESS;
7435: }
7437: /*
7438: MatComputeVariableBlockEnvelope - Given a matrix whose nonzeros are in blocks along the diagonal this computes and stores
7439: the sizes of these blocks in the matrix. An individual block may lie over several processes.
7441: Collective
7443: Input Parameter:
7444: . mat - the matrix
7446: Notes:
7447: There can be zeros within the blocks
7449: The blocks can overlap between processes, including laying on more than two processes
7451: .seealso: [](ch_matrices), `Mat`, `MatInvertVariableBlockEnvelope()`, `MatSetVariableBlockSizes()`
7452: */
7453: static PetscErrorCode MatComputeVariableBlockEnvelope(Mat mat)
7454: {
7455: PetscInt n, *sizes, *starts, i = 0, env = 0, tbs = 0, lblocks = 0, rstart, II, ln = 0, cnt = 0, cstart, cend;
7456: PetscInt *diag, *odiag, sc;
7457: VecScatter scatter;
7458: PetscScalar *seqv;
7459: const PetscScalar *parv;
7460: const PetscInt *ia, *ja;
7461: PetscBool set, flag, done;
7462: Mat AA = mat, A;
7463: MPI_Comm comm;
7464: PetscMPIInt rank, size, tag;
7465: MPI_Status status;
7466: PetscContainer container;
7467: EnvelopeData *edata;
7468: Vec seq, par;
7469: IS isglobal;
7471: PetscFunctionBegin;
7473: PetscCall(MatIsSymmetricKnown(mat, &set, &flag));
7474: if (!set || !flag) {
7475: /* TOO: only needs nonzero structure of transpose */
7476: PetscCall(MatTranspose(mat, MAT_INITIAL_MATRIX, &AA));
7477: PetscCall(MatAXPY(AA, 1.0, mat, DIFFERENT_NONZERO_PATTERN));
7478: }
7479: PetscCall(MatAIJGetLocalMat(AA, &A));
7480: PetscCall(MatGetRowIJ(A, 0, PETSC_FALSE, PETSC_FALSE, &n, &ia, &ja, &done));
7481: PetscCheck(done, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Unable to get IJ structure from matrix");
7483: PetscCall(MatGetLocalSize(mat, &n, NULL));
7484: PetscCall(PetscObjectGetNewTag((PetscObject)mat, &tag));
7485: PetscCall(PetscObjectGetComm((PetscObject)mat, &comm));
7486: PetscCallMPI(MPI_Comm_size(comm, &size));
7487: PetscCallMPI(MPI_Comm_rank(comm, &rank));
7489: PetscCall(PetscMalloc2(n, &sizes, n, &starts));
7491: if (rank > 0) {
7492: PetscCallMPI(MPI_Recv(&env, 1, MPIU_INT, rank - 1, tag, comm, &status));
7493: PetscCallMPI(MPI_Recv(&tbs, 1, MPIU_INT, rank - 1, tag, comm, &status));
7494: }
7495: PetscCall(MatGetOwnershipRange(mat, &rstart, NULL));
7496: for (i = 0; i < n; i++) {
7497: env = PetscMax(env, ja[ia[i + 1] - 1]);
7498: II = rstart + i;
7499: if (env == II) {
7500: starts[lblocks] = tbs;
7501: sizes[lblocks++] = 1 + II - tbs;
7502: tbs = 1 + II;
7503: }
7504: }
7505: if (rank < size - 1) {
7506: PetscCallMPI(MPI_Send(&env, 1, MPIU_INT, rank + 1, tag, comm));
7507: PetscCallMPI(MPI_Send(&tbs, 1, MPIU_INT, rank + 1, tag, comm));
7508: }
7510: PetscCall(MatRestoreRowIJ(A, 0, PETSC_FALSE, PETSC_FALSE, &n, &ia, &ja, &done));
7511: if (!set || !flag) PetscCall(MatDestroy(&AA));
7512: PetscCall(MatDestroy(&A));
7514: PetscCall(PetscNew(&edata));
7515: PetscCall(MatGetNonzeroState(mat, &edata->nonzerostate));
7516: edata->n = lblocks;
7517: /* create IS needed for extracting blocks from the original matrix */
7518: PetscCall(PetscMalloc1(lblocks, &edata->is));
7519: for (PetscInt i = 0; i < lblocks; i++) PetscCall(ISCreateStride(PETSC_COMM_SELF, sizes[i], starts[i], 1, &edata->is[i]));
7521: /* Create the resulting inverse matrix structure with preallocation information */
7522: PetscCall(MatCreate(PetscObjectComm((PetscObject)mat), &edata->C));
7523: PetscCall(MatSetSizes(edata->C, mat->rmap->n, mat->cmap->n, mat->rmap->N, mat->cmap->N));
7524: PetscCall(MatSetBlockSizesFromMats(edata->C, mat, mat));
7525: PetscCall(MatSetType(edata->C, MATAIJ));
7527: /* Communicate the start and end of each row, from each block to the correct rank */
7528: /* TODO: Use PetscSF instead of VecScatter */
7529: for (PetscInt i = 0; i < lblocks; i++) ln += sizes[i];
7530: PetscCall(VecCreateSeq(PETSC_COMM_SELF, 2 * ln, &seq));
7531: PetscCall(VecGetArrayWrite(seq, &seqv));
7532: for (PetscInt i = 0; i < lblocks; i++) {
7533: for (PetscInt j = 0; j < sizes[i]; j++) {
7534: seqv[cnt] = starts[i];
7535: seqv[cnt + 1] = starts[i] + sizes[i];
7536: cnt += 2;
7537: }
7538: }
7539: PetscCall(VecRestoreArrayWrite(seq, &seqv));
7540: PetscCallMPI(MPI_Scan(&cnt, &sc, 1, MPIU_INT, MPI_SUM, PetscObjectComm((PetscObject)mat)));
7541: sc -= cnt;
7542: PetscCall(VecCreateMPI(PetscObjectComm((PetscObject)mat), 2 * mat->rmap->n, 2 * mat->rmap->N, &par));
7543: PetscCall(ISCreateStride(PETSC_COMM_SELF, cnt, sc, 1, &isglobal));
7544: PetscCall(VecScatterCreate(seq, NULL, par, isglobal, &scatter));
7545: PetscCall(ISDestroy(&isglobal));
7546: PetscCall(VecScatterBegin(scatter, seq, par, INSERT_VALUES, SCATTER_FORWARD));
7547: PetscCall(VecScatterEnd(scatter, seq, par, INSERT_VALUES, SCATTER_FORWARD));
7548: PetscCall(VecScatterDestroy(&scatter));
7549: PetscCall(VecDestroy(&seq));
7550: PetscCall(MatGetOwnershipRangeColumn(mat, &cstart, &cend));
7551: PetscCall(PetscMalloc2(mat->rmap->n, &diag, mat->rmap->n, &odiag));
7552: PetscCall(VecGetArrayRead(par, &parv));
7553: cnt = 0;
7554: PetscCall(MatGetSize(mat, NULL, &n));
7555: for (PetscInt i = 0; i < mat->rmap->n; i++) {
7556: PetscInt start, end, d = 0, od = 0;
7558: start = (PetscInt)PetscRealPart(parv[cnt]);
7559: end = (PetscInt)PetscRealPart(parv[cnt + 1]);
7560: cnt += 2;
7562: if (start < cstart) {
7563: od += cstart - start + n - cend;
7564: d += cend - cstart;
7565: } else if (start < cend) {
7566: od += n - cend;
7567: d += cend - start;
7568: } else od += n - start;
7569: if (end <= cstart) {
7570: od -= cstart - end + n - cend;
7571: d -= cend - cstart;
7572: } else if (end < cend) {
7573: od -= n - cend;
7574: d -= cend - end;
7575: } else od -= n - end;
7577: odiag[i] = od;
7578: diag[i] = d;
7579: }
7580: PetscCall(VecRestoreArrayRead(par, &parv));
7581: PetscCall(VecDestroy(&par));
7582: PetscCall(MatXAIJSetPreallocation(edata->C, mat->rmap->bs, diag, odiag, NULL, NULL));
7583: PetscCall(PetscFree2(diag, odiag));
7584: PetscCall(PetscFree2(sizes, starts));
7586: PetscCall(PetscContainerCreate(PETSC_COMM_SELF, &container));
7587: PetscCall(PetscContainerSetPointer(container, edata));
7588: PetscCall(PetscContainerSetUserDestroy(container, (PetscErrorCode(*)(void *))EnvelopeDataDestroy));
7589: PetscCall(PetscObjectCompose((PetscObject)mat, "EnvelopeData", (PetscObject)container));
7590: PetscCall(PetscObjectDereference((PetscObject)container));
7591: PetscFunctionReturn(PETSC_SUCCESS);
7592: }
7594: /*@
7595: MatInvertVariableBlockEnvelope - set matrix C to be the inverted block diagonal of matrix A
7597: Collective
7599: Input Parameters:
7600: + A - the matrix
7601: - reuse - indicates if the `C` matrix was obtained from a previous call to this routine
7603: Output Parameter:
7604: . C - matrix with inverted block diagonal of `A`
7606: Level: advanced
7608: Note:
7609: For efficiency the matrix `A` should have all the nonzero entries clustered in smallish blocks along the diagonal.
7611: .seealso: [](ch_matrices), `Mat`, `MatInvertBlockDiagonal()`, `MatComputeBlockDiagonal()`
7612: @*/
7613: PetscErrorCode MatInvertVariableBlockEnvelope(Mat A, MatReuse reuse, Mat *C)
7614: {
7615: PetscContainer container;
7616: EnvelopeData *edata;
7617: PetscObjectState nonzerostate;
7619: PetscFunctionBegin;
7620: PetscCall(PetscObjectQuery((PetscObject)A, "EnvelopeData", (PetscObject *)&container));
7621: if (!container) {
7622: PetscCall(MatComputeVariableBlockEnvelope(A));
7623: PetscCall(PetscObjectQuery((PetscObject)A, "EnvelopeData", (PetscObject *)&container));
7624: }
7625: PetscCall(PetscContainerGetPointer(container, (void **)&edata));
7626: PetscCall(MatGetNonzeroState(A, &nonzerostate));
7627: PetscCheck(nonzerostate <= edata->nonzerostate, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Cannot handle changes to matrix nonzero structure");
7628: PetscCheck(reuse != MAT_REUSE_MATRIX || *C == edata->C, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "C matrix must be the same as previously output");
7630: PetscCall(MatCreateSubMatrices(A, edata->n, edata->is, edata->is, MAT_INITIAL_MATRIX, &edata->mat));
7631: *C = edata->C;
7633: for (PetscInt i = 0; i < edata->n; i++) {
7634: Mat D;
7635: PetscScalar *dvalues;
7637: PetscCall(MatConvert(edata->mat[i], MATSEQDENSE, MAT_INITIAL_MATRIX, &D));
7638: PetscCall(MatSetOption(*C, MAT_ROW_ORIENTED, PETSC_FALSE));
7639: PetscCall(MatSeqDenseInvert(D));
7640: PetscCall(MatDenseGetArray(D, &dvalues));
7641: PetscCall(MatSetValuesIS(*C, edata->is[i], edata->is[i], dvalues, INSERT_VALUES));
7642: PetscCall(MatDestroy(&D));
7643: }
7644: PetscCall(MatDestroySubMatrices(edata->n, &edata->mat));
7645: PetscCall(MatAssemblyBegin(*C, MAT_FINAL_ASSEMBLY));
7646: PetscCall(MatAssemblyEnd(*C, MAT_FINAL_ASSEMBLY));
7647: PetscFunctionReturn(PETSC_SUCCESS);
7648: }
7650: /*@
7651: MatSetVariableBlockSizes - Sets diagonal point-blocks of the matrix that need not be of the same size
7653: Logically Collective
7655: Input Parameters:
7656: + mat - the matrix
7657: . nblocks - the number of blocks on this process, each block can only exist on a single process
7658: - bsizes - the block sizes
7660: Level: intermediate
7662: Notes:
7663: Currently used by `PCVPBJACOBI` for `MATAIJ` matrices
7665: Each variable point-block set of degrees of freedom must live on a single MPI rank. That is a point block cannot straddle two MPI ranks.
7667: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSizes()`, `MatGetBlockSizes()`, `MatGetVariableBlockSizes()`,
7668: `MatComputeVariableBlockEnvelope()`, `PCVPBJACOBI`
7669: @*/
7670: PetscErrorCode MatSetVariableBlockSizes(Mat mat, PetscInt nblocks, PetscInt *bsizes)
7671: {
7672: PetscInt i, ncnt = 0, nlocal;
7674: PetscFunctionBegin;
7676: PetscCheck(nblocks >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Number of local blocks must be great than or equal to zero");
7677: PetscCall(MatGetLocalSize(mat, &nlocal, NULL));
7678: for (i = 0; i < nblocks; i++) ncnt += bsizes[i];
7679: PetscCheck(ncnt == nlocal, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Sum of local block sizes %" PetscInt_FMT " does not equal local size of matrix %" PetscInt_FMT, ncnt, nlocal);
7680: PetscCall(PetscFree(mat->bsizes));
7681: mat->nblocks = nblocks;
7682: PetscCall(PetscMalloc1(nblocks, &mat->bsizes));
7683: PetscCall(PetscArraycpy(mat->bsizes, bsizes, nblocks));
7684: PetscFunctionReturn(PETSC_SUCCESS);
7685: }
7687: /*@C
7688: MatGetVariableBlockSizes - Gets a diagonal blocks of the matrix that need not be of the same size
7690: Logically Collective; No Fortran Support
7692: Input Parameter:
7693: . mat - the matrix
7695: Output Parameters:
7696: + nblocks - the number of blocks on this process
7697: - bsizes - the block sizes
7699: Level: intermediate
7701: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSizes()`, `MatGetBlockSizes()`, `MatSetVariableBlockSizes()`, `MatComputeVariableBlockEnvelope()`
7702: @*/
7703: PetscErrorCode MatGetVariableBlockSizes(Mat mat, PetscInt *nblocks, const PetscInt **bsizes)
7704: {
7705: PetscFunctionBegin;
7707: *nblocks = mat->nblocks;
7708: *bsizes = mat->bsizes;
7709: PetscFunctionReturn(PETSC_SUCCESS);
7710: }
7712: /*@
7713: MatSetBlockSizes - Sets the matrix block row and column sizes.
7715: Logically Collective
7717: Input Parameters:
7718: + mat - the matrix
7719: . rbs - row block size
7720: - cbs - column block size
7722: Level: intermediate
7724: Notes:
7725: Block row formats are `MATBAIJ` and `MATSBAIJ`. These formats ALWAYS have square block storage in the matrix.
7726: If you pass a different block size for the columns than the rows, the row block size determines the square block storage.
7727: This must be called before `MatSetUp()` or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later.
7729: For `MATAIJ` matrix this function can be called at a later stage, provided that the specified block sizes
7730: are compatible with the matrix local sizes.
7732: The row and column block size determine the blocksize of the "row" and "column" vectors returned by `MatCreateVecs()`.
7734: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSize()`, `MatGetBlockSizes()`
7735: @*/
7736: PetscErrorCode MatSetBlockSizes(Mat mat, PetscInt rbs, PetscInt cbs)
7737: {
7738: PetscFunctionBegin;
7742: PetscTryTypeMethod(mat, setblocksizes, rbs, cbs);
7743: if (mat->rmap->refcnt) {
7744: ISLocalToGlobalMapping l2g = NULL;
7745: PetscLayout nmap = NULL;
7747: PetscCall(PetscLayoutDuplicate(mat->rmap, &nmap));
7748: if (mat->rmap->mapping) PetscCall(ISLocalToGlobalMappingDuplicate(mat->rmap->mapping, &l2g));
7749: PetscCall(PetscLayoutDestroy(&mat->rmap));
7750: mat->rmap = nmap;
7751: mat->rmap->mapping = l2g;
7752: }
7753: if (mat->cmap->refcnt) {
7754: ISLocalToGlobalMapping l2g = NULL;
7755: PetscLayout nmap = NULL;
7757: PetscCall(PetscLayoutDuplicate(mat->cmap, &nmap));
7758: if (mat->cmap->mapping) PetscCall(ISLocalToGlobalMappingDuplicate(mat->cmap->mapping, &l2g));
7759: PetscCall(PetscLayoutDestroy(&mat->cmap));
7760: mat->cmap = nmap;
7761: mat->cmap->mapping = l2g;
7762: }
7763: PetscCall(PetscLayoutSetBlockSize(mat->rmap, rbs));
7764: PetscCall(PetscLayoutSetBlockSize(mat->cmap, cbs));
7765: PetscFunctionReturn(PETSC_SUCCESS);
7766: }
7768: /*@
7769: MatSetBlockSizesFromMats - Sets the matrix block row and column sizes to match a pair of matrices
7771: Logically Collective
7773: Input Parameters:
7774: + mat - the matrix
7775: . fromRow - matrix from which to copy row block size
7776: - fromCol - matrix from which to copy column block size (can be same as fromRow)
7778: Level: developer
7780: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqBAIJ()`, `MatCreateBAIJ()`, `MatGetBlockSize()`, `MatSetBlockSizes()`
7781: @*/
7782: PetscErrorCode MatSetBlockSizesFromMats(Mat mat, Mat fromRow, Mat fromCol)
7783: {
7784: PetscFunctionBegin;
7788: if (fromRow->rmap->bs > 0) PetscCall(PetscLayoutSetBlockSize(mat->rmap, fromRow->rmap->bs));
7789: if (fromCol->cmap->bs > 0) PetscCall(PetscLayoutSetBlockSize(mat->cmap, fromCol->cmap->bs));
7790: PetscFunctionReturn(PETSC_SUCCESS);
7791: }
7793: /*@
7794: MatResidual - Default routine to calculate the residual r = b - Ax
7796: Collective
7798: Input Parameters:
7799: + mat - the matrix
7800: . b - the right-hand-side
7801: - x - the approximate solution
7803: Output Parameter:
7804: . r - location to store the residual
7806: Level: developer
7808: .seealso: [](ch_matrices), `Mat`, `MatMult()`, `MatMultAdd()`, `PCMGSetResidual()`
7809: @*/
7810: PetscErrorCode MatResidual(Mat mat, Vec b, Vec x, Vec r)
7811: {
7812: PetscFunctionBegin;
7818: MatCheckPreallocated(mat, 1);
7819: PetscCall(PetscLogEventBegin(MAT_Residual, mat, 0, 0, 0));
7820: if (!mat->ops->residual) {
7821: PetscCall(MatMult(mat, x, r));
7822: PetscCall(VecAYPX(r, -1.0, b));
7823: } else {
7824: PetscUseTypeMethod(mat, residual, b, x, r);
7825: }
7826: PetscCall(PetscLogEventEnd(MAT_Residual, mat, 0, 0, 0));
7827: PetscFunctionReturn(PETSC_SUCCESS);
7828: }
7830: /*MC
7831: MatGetRowIJF90 - Obtains the compressed row storage i and j indices for the local rows of a sparse matrix
7833: Synopsis:
7834: MatGetRowIJF90(Mat A, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt n, {PetscInt, pointer :: ia(:)}, {PetscInt, pointer :: ja(:)}, PetscBool done,integer ierr)
7836: Not Collective
7838: Input Parameters:
7839: + A - the matrix
7840: . shift - 0 or 1 indicating we want the indices starting at 0 or 1
7841: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be symmetrized
7842: - inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
7843: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
7844: always used.
7846: Output Parameters:
7847: + n - number of local rows in the (possibly compressed) matrix
7848: . ia - the row pointers; that is ia[0] = 0, ia[row] = ia[row-1] + number of elements in that row of the matrix
7849: . ja - the column indices
7850: - done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers
7851: are responsible for handling the case when done == `PETSC_FALSE` and ia and ja are not set
7853: Level: developer
7855: Note:
7856: Use `MatRestoreRowIJF90()` when you no longer need access to the data
7858: .seealso: [](ch_matrices), [](sec_fortranarrays), `Mat`, `MATMPIAIJ`, `MatGetRowIJ()`, `MatRestoreRowIJ()`, `MatRestoreRowIJF90()`
7859: M*/
7861: /*MC
7862: MatRestoreRowIJF90 - restores the compressed row storage i and j indices for the local rows of a sparse matrix obtained with `MatGetRowIJF90()`
7864: Synopsis:
7865: MatRestoreRowIJF90(Mat A, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt n, {PetscInt, pointer :: ia(:)}, {PetscInt, pointer :: ja(:)}, PetscBool done,integer ierr)
7867: Not Collective
7869: Input Parameters:
7870: + A - the matrix
7871: . shift - 0 or 1 indicating we want the indices starting at 0 or 1
7872: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be symmetrized
7873: inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
7874: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
7875: always used.
7876: . n - number of local rows in the (possibly compressed) matrix
7877: . ia - the row pointers; that is ia[0] = 0, ia[row] = ia[row-1] + number of elements in that row of the matrix
7878: . ja - the column indices
7879: - done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers
7880: are responsible for handling the case when done == `PETSC_FALSE` and ia and ja are not set
7882: Level: developer
7884: .seealso: [](ch_matrices), [](sec_fortranarrays), `Mat`, `MATMPIAIJ`, `MatGetRowIJ()`, `MatRestoreRowIJ()`, `MatGetRowIJF90()`
7885: M*/
7887: /*@C
7888: MatGetRowIJ - Returns the compressed row storage i and j indices for the local rows of a sparse matrix
7890: Collective
7892: Input Parameters:
7893: + mat - the matrix
7894: . shift - 0 or 1 indicating we want the indices starting at 0 or 1
7895: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be symmetrized
7896: - inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
7897: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
7898: always used.
7900: Output Parameters:
7901: + n - number of local rows in the (possibly compressed) matrix, use `NULL` if not needed
7902: . ia - the row pointers; that is ia[0] = 0, ia[row] = ia[row-1] + number of elements in that row of the matrix, use `NULL` if not needed
7903: . ja - the column indices, use `NULL` if not needed
7904: - done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers
7905: are responsible for handling the case when done == `PETSC_FALSE` and ia and ja are not set
7907: Level: developer
7909: Notes:
7910: You CANNOT change any of the ia[] or ja[] values.
7912: Use `MatRestoreRowIJ()` when you are finished accessing the ia[] and ja[] values.
7914: Fortran Notes:
7915: Use
7916: .vb
7917: PetscInt, pointer :: ia(:),ja(:)
7918: call MatGetRowIJF90(mat,shift,symmetric,inodecompressed,n,ia,ja,done,ierr)
7919: ! Access the ith and jth entries via ia(i) and ja(j)
7920: .ve
7921: `MatGetRowIJ()` Fortran binding is deprecated (since PETSc 3.19), use `MatGetRowIJF90()`
7923: .seealso: [](ch_matrices), `Mat`, `MATAIJ`, `MatGetRowIJF90()`, `MatGetColumnIJ()`, `MatRestoreRowIJ()`, `MatSeqAIJGetArray()`
7924: @*/
7925: PetscErrorCode MatGetRowIJ(Mat mat, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
7926: {
7927: PetscFunctionBegin;
7934: MatCheckPreallocated(mat, 1);
7935: if (!mat->ops->getrowij && done) *done = PETSC_FALSE;
7936: else {
7937: if (done) *done = PETSC_TRUE;
7938: PetscCall(PetscLogEventBegin(MAT_GetRowIJ, mat, 0, 0, 0));
7939: PetscUseTypeMethod(mat, getrowij, shift, symmetric, inodecompressed, n, ia, ja, done);
7940: PetscCall(PetscLogEventEnd(MAT_GetRowIJ, mat, 0, 0, 0));
7941: }
7942: PetscFunctionReturn(PETSC_SUCCESS);
7943: }
7945: /*@C
7946: MatGetColumnIJ - Returns the compressed column storage i and j indices for sequential matrices.
7948: Collective
7950: Input Parameters:
7951: + mat - the matrix
7952: . shift - 1 or zero indicating we want the indices starting at 0 or 1
7953: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be
7954: symmetrized
7955: . inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
7956: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
7957: always used.
7958: . n - number of columns in the (possibly compressed) matrix
7959: . ia - the column pointers; that is ia[0] = 0, ia[col] = i[col-1] + number of elements in that col of the matrix
7960: - ja - the row indices
7962: Output Parameter:
7963: . done - `PETSC_TRUE` or `PETSC_FALSE`, indicating whether the values have been returned
7965: Level: developer
7967: .seealso: [](ch_matrices), `Mat`, `MatGetRowIJ()`, `MatRestoreColumnIJ()`
7968: @*/
7969: PetscErrorCode MatGetColumnIJ(Mat mat, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
7970: {
7971: PetscFunctionBegin;
7978: MatCheckPreallocated(mat, 1);
7979: if (!mat->ops->getcolumnij) *done = PETSC_FALSE;
7980: else {
7981: *done = PETSC_TRUE;
7982: PetscUseTypeMethod(mat, getcolumnij, shift, symmetric, inodecompressed, n, ia, ja, done);
7983: }
7984: PetscFunctionReturn(PETSC_SUCCESS);
7985: }
7987: /*@C
7988: MatRestoreRowIJ - Call after you are completed with the ia,ja indices obtained with `MatGetRowIJ()`.
7990: Collective
7992: Input Parameters:
7993: + mat - the matrix
7994: . shift - 1 or zero indicating we want the indices starting at 0 or 1
7995: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be symmetrized
7996: . inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
7997: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
7998: always used.
7999: . n - size of (possibly compressed) matrix
8000: . ia - the row pointers
8001: - ja - the column indices
8003: Output Parameter:
8004: . done - `PETSC_TRUE` or `PETSC_FALSE` indicated that the values have been returned
8006: Level: developer
8008: Note:
8009: This routine zeros out `n`, `ia`, and `ja`. This is to prevent accidental
8010: us of the array after it has been restored. If you pass `NULL`, it will
8011: not zero the pointers. Use of ia or ja after `MatRestoreRowIJ()` is invalid.
8013: Fortran Note:
8014: `MatRestoreRowIJ()` Fortran binding is deprecated (since PETSc 3.19), use `MatRestoreRowIJF90()`
8016: .seealso: [](ch_matrices), `Mat`, `MatGetRowIJ()`, `MatRestoreRowIJF90()`, `MatRestoreColumnIJ()`
8017: @*/
8018: PetscErrorCode MatRestoreRowIJ(Mat mat, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
8019: {
8020: PetscFunctionBegin;
8026: MatCheckPreallocated(mat, 1);
8028: if (!mat->ops->restorerowij && done) *done = PETSC_FALSE;
8029: else {
8030: if (done) *done = PETSC_TRUE;
8031: PetscUseTypeMethod(mat, restorerowij, shift, symmetric, inodecompressed, n, ia, ja, done);
8032: if (n) *n = 0;
8033: if (ia) *ia = NULL;
8034: if (ja) *ja = NULL;
8035: }
8036: PetscFunctionReturn(PETSC_SUCCESS);
8037: }
8039: /*@C
8040: MatRestoreColumnIJ - Call after you are completed with the ia,ja indices obtained with `MatGetColumnIJ()`.
8042: Collective
8044: Input Parameters:
8045: + mat - the matrix
8046: . shift - 1 or zero indicating we want the indices starting at 0 or 1
8047: . symmetric - `PETSC_TRUE` or `PETSC_FALSE` indicating the matrix data structure should be symmetrized
8048: - inodecompressed - `PETSC_TRUE` or `PETSC_FALSE` indicating if the nonzero structure of the
8049: inodes or the nonzero elements is wanted. For `MATBAIJ` matrices the compressed version is
8050: always used.
8052: Output Parameters:
8053: + n - size of (possibly compressed) matrix
8054: . ia - the column pointers
8055: . ja - the row indices
8056: - done - `PETSC_TRUE` or `PETSC_FALSE` indicated that the values have been returned
8058: Level: developer
8060: .seealso: [](ch_matrices), `Mat`, `MatGetColumnIJ()`, `MatRestoreRowIJ()`
8061: @*/
8062: PetscErrorCode MatRestoreColumnIJ(Mat mat, PetscInt shift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
8063: {
8064: PetscFunctionBegin;
8070: MatCheckPreallocated(mat, 1);
8072: if (!mat->ops->restorecolumnij) *done = PETSC_FALSE;
8073: else {
8074: *done = PETSC_TRUE;
8075: PetscUseTypeMethod(mat, restorecolumnij, shift, symmetric, inodecompressed, n, ia, ja, done);
8076: if (n) *n = 0;
8077: if (ia) *ia = NULL;
8078: if (ja) *ja = NULL;
8079: }
8080: PetscFunctionReturn(PETSC_SUCCESS);
8081: }
8083: /*@C
8084: MatColoringPatch -Used inside matrix coloring routines that use `MatGetRowIJ()` and/or `MatGetColumnIJ()`.
8086: Collective
8088: Input Parameters:
8089: + mat - the matrix
8090: . ncolors - maximum color value
8091: . n - number of entries in colorarray
8092: - colorarray - array indicating color for each column
8094: Output Parameter:
8095: . iscoloring - coloring generated using colorarray information
8097: Level: developer
8099: .seealso: [](ch_matrices), `Mat`, `MatGetRowIJ()`, `MatGetColumnIJ()`
8100: @*/
8101: PetscErrorCode MatColoringPatch(Mat mat, PetscInt ncolors, PetscInt n, ISColoringValue colorarray[], ISColoring *iscoloring)
8102: {
8103: PetscFunctionBegin;
8108: MatCheckPreallocated(mat, 1);
8110: if (!mat->ops->coloringpatch) {
8111: PetscCall(ISColoringCreate(PetscObjectComm((PetscObject)mat), ncolors, n, colorarray, PETSC_OWN_POINTER, iscoloring));
8112: } else {
8113: PetscUseTypeMethod(mat, coloringpatch, ncolors, n, colorarray, iscoloring);
8114: }
8115: PetscFunctionReturn(PETSC_SUCCESS);
8116: }
8118: /*@
8119: MatSetUnfactored - Resets a factored matrix to be treated as unfactored.
8121: Logically Collective
8123: Input Parameter:
8124: . mat - the factored matrix to be reset
8126: Level: developer
8128: Notes:
8129: This routine should be used only with factored matrices formed by in-place
8130: factorization via ILU(0) (or by in-place LU factorization for the `MATSEQDENSE`
8131: format). This option can save memory, for example, when solving nonlinear
8132: systems with a matrix-free Newton-Krylov method and a matrix-based, in-place
8133: ILU(0) preconditioner.
8135: One can specify in-place ILU(0) factorization by calling
8136: .vb
8137: PCType(pc,PCILU);
8138: PCFactorSeUseInPlace(pc);
8139: .ve
8140: or by using the options -pc_type ilu -pc_factor_in_place
8142: In-place factorization ILU(0) can also be used as a local
8143: solver for the blocks within the block Jacobi or additive Schwarz
8144: methods (runtime option: -sub_pc_factor_in_place). See Users-Manual: ch_pc
8145: for details on setting local solver options.
8147: Most users should employ the `KSP` interface for linear solvers
8148: instead of working directly with matrix algebra routines such as this.
8149: See, e.g., `KSPCreate()`.
8151: .seealso: [](ch_matrices), `Mat`, `PCFactorSetUseInPlace()`, `PCFactorGetUseInPlace()`
8152: @*/
8153: PetscErrorCode MatSetUnfactored(Mat mat)
8154: {
8155: PetscFunctionBegin;
8158: MatCheckPreallocated(mat, 1);
8159: mat->factortype = MAT_FACTOR_NONE;
8160: if (!mat->ops->setunfactored) PetscFunctionReturn(PETSC_SUCCESS);
8161: PetscUseTypeMethod(mat, setunfactored);
8162: PetscFunctionReturn(PETSC_SUCCESS);
8163: }
8165: /*MC
8166: MatDenseGetArrayF90 - Accesses a matrix array from Fortran
8168: Synopsis:
8169: MatDenseGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:,:)},integer ierr)
8171: Not Collective
8173: Input Parameter:
8174: . x - matrix
8176: Output Parameters:
8177: + xx_v - the Fortran pointer to the array
8178: - ierr - error code
8180: Example of Usage:
8181: .vb
8182: PetscScalar, pointer xx_v(:,:)
8183: ....
8184: call MatDenseGetArrayF90(x,xx_v,ierr)
8185: a = xx_v(3)
8186: call MatDenseRestoreArrayF90(x,xx_v,ierr)
8187: .ve
8189: Level: advanced
8191: .seealso: [](ch_matrices), `Mat`, `MatDenseRestoreArrayF90()`, `MatDenseGetArray()`, `MatDenseRestoreArray()`, `MatSeqAIJGetArrayF90()`
8192: M*/
8194: /*MC
8195: MatDenseRestoreArrayF90 - Restores a matrix array that has been
8196: accessed with `MatDenseGetArrayF90()`.
8198: Synopsis:
8199: MatDenseRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:,:)},integer ierr)
8201: Not Collective
8203: Input Parameters:
8204: + x - matrix
8205: - xx_v - the Fortran90 pointer to the array
8207: Output Parameter:
8208: . ierr - error code
8210: Example of Usage:
8211: .vb
8212: PetscScalar, pointer xx_v(:,:)
8213: ....
8214: call MatDenseGetArrayF90(x,xx_v,ierr)
8215: a = xx_v(3)
8216: call MatDenseRestoreArrayF90(x,xx_v,ierr)
8217: .ve
8219: Level: advanced
8221: .seealso: [](ch_matrices), `Mat`, `MatDenseGetArrayF90()`, `MatDenseGetArray()`, `MatDenseRestoreArray()`, `MatSeqAIJRestoreArrayF90()`
8222: M*/
8224: /*MC
8225: MatSeqAIJGetArrayF90 - Accesses a matrix array from Fortran.
8227: Synopsis:
8228: MatSeqAIJGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
8230: Not Collective
8232: Input Parameter:
8233: . x - matrix
8235: Output Parameters:
8236: + xx_v - the Fortran pointer to the array
8237: - ierr - error code
8239: Example of Usage:
8240: .vb
8241: PetscScalar, pointer xx_v(:)
8242: ....
8243: call MatSeqAIJGetArrayF90(x,xx_v,ierr)
8244: a = xx_v(3)
8245: call MatSeqAIJRestoreArrayF90(x,xx_v,ierr)
8246: .ve
8248: Level: advanced
8250: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJRestoreArrayF90()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`, `MatDenseGetArrayF90()`
8251: M*/
8253: /*MC
8254: MatSeqAIJRestoreArrayF90 - Restores a matrix array that has been
8255: accessed with `MatSeqAIJGetArrayF90()`.
8257: Synopsis:
8258: MatSeqAIJRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
8260: Not Collective
8262: Input Parameters:
8263: + x - matrix
8264: - xx_v - the Fortran90 pointer to the array
8266: Output Parameter:
8267: . ierr - error code
8269: Example of Usage:
8270: .vb
8271: PetscScalar, pointer xx_v(:)
8272: ....
8273: call MatSeqAIJGetArrayF90(x,xx_v,ierr)
8274: a = xx_v(3)
8275: call MatSeqAIJRestoreArrayF90(x,xx_v,ierr)
8276: .ve
8278: Level: advanced
8280: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArrayF90()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`, `MatDenseRestoreArrayF90()`
8281: M*/
8283: /*@
8284: MatCreateSubMatrix - Gets a single submatrix on the same number of processors
8285: as the original matrix.
8287: Collective
8289: Input Parameters:
8290: + mat - the original matrix
8291: . isrow - parallel `IS` containing the rows this processor should obtain
8292: . iscol - parallel `IS` containing all columns you wish to keep. Each process should list the columns that will be in IT's "diagonal part" in the new matrix.
8293: - cll - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
8295: Output Parameter:
8296: . newmat - the new submatrix, of the same type as the original matrix
8298: Level: advanced
8300: Notes:
8301: The submatrix will be able to be multiplied with vectors using the same layout as `iscol`.
8303: Some matrix types place restrictions on the row and column indices, such
8304: as that they be sorted or that they be equal to each other. For `MATBAIJ` and `MATSBAIJ` matrices the indices must include all rows/columns of a block;
8305: for example, if the block size is 3 one cannot select the 0 and 2 rows without selecting the 1 row.
8307: The index sets may not have duplicate entries.
8309: The first time this is called you should use a cll of `MAT_INITIAL_MATRIX`,
8310: the `MatCreateSubMatrix()` routine will create the newmat for you. Any additional calls
8311: to this routine with a mat of the same nonzero structure and with a call of `MAT_REUSE_MATRIX`
8312: will reuse the matrix generated the first time. You should call `MatDestroy()` on `newmat` when
8313: you are finished using it.
8315: The communicator of the newly obtained matrix is ALWAYS the same as the communicator of
8316: the input matrix.
8318: If `iscol` is `NULL` then all columns are obtained (not supported in Fortran).
8320: Example usage:
8321: Consider the following 8x8 matrix with 34 non-zero values, that is
8322: assembled across 3 processors. Let's assume that proc0 owns 3 rows,
8323: proc1 owns 3 rows, proc2 owns 2 rows. This division can be shown
8324: as follows
8325: .vb
8326: 1 2 0 | 0 3 0 | 0 4
8327: Proc0 0 5 6 | 7 0 0 | 8 0
8328: 9 0 10 | 11 0 0 | 12 0
8329: -------------------------------------
8330: 13 0 14 | 15 16 17 | 0 0
8331: Proc1 0 18 0 | 19 20 21 | 0 0
8332: 0 0 0 | 22 23 0 | 24 0
8333: -------------------------------------
8334: Proc2 25 26 27 | 0 0 28 | 29 0
8335: 30 0 0 | 31 32 33 | 0 34
8336: .ve
8338: Suppose `isrow` = [0 1 | 4 | 6 7] and `iscol` = [1 2 | 3 4 5 | 6]. The resulting submatrix is
8340: .vb
8341: 2 0 | 0 3 0 | 0
8342: Proc0 5 6 | 7 0 0 | 8
8343: -------------------------------
8344: Proc1 18 0 | 19 20 21 | 0
8345: -------------------------------
8346: Proc2 26 27 | 0 0 28 | 29
8347: 0 0 | 31 32 33 | 0
8348: .ve
8350: .seealso: [](ch_matrices), `Mat`, `MatCreateSubMatrices()`, `MatCreateSubMatricesMPI()`, `MatCreateSubMatrixVirtual()`, `MatSubMatrixVirtualUpdate()`
8351: @*/
8352: PetscErrorCode MatCreateSubMatrix(Mat mat, IS isrow, IS iscol, MatReuse cll, Mat *newmat)
8353: {
8354: PetscMPIInt size;
8355: Mat *local;
8356: IS iscoltmp;
8357: PetscBool flg;
8359: PetscFunctionBegin;
8366: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
8367: PetscCheck(cll != MAT_IGNORE_MATRIX, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Cannot use MAT_IGNORE_MATRIX");
8369: MatCheckPreallocated(mat, 1);
8370: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size));
8372: if (!iscol || isrow == iscol) {
8373: PetscBool stride;
8374: PetscMPIInt grabentirematrix = 0, grab;
8375: PetscCall(PetscObjectTypeCompare((PetscObject)isrow, ISSTRIDE, &stride));
8376: if (stride) {
8377: PetscInt first, step, n, rstart, rend;
8378: PetscCall(ISStrideGetInfo(isrow, &first, &step));
8379: if (step == 1) {
8380: PetscCall(MatGetOwnershipRange(mat, &rstart, &rend));
8381: if (rstart == first) {
8382: PetscCall(ISGetLocalSize(isrow, &n));
8383: if (n == rend - rstart) grabentirematrix = 1;
8384: }
8385: }
8386: }
8387: PetscCall(MPIU_Allreduce(&grabentirematrix, &grab, 1, MPI_INT, MPI_MIN, PetscObjectComm((PetscObject)mat)));
8388: if (grab) {
8389: PetscCall(PetscInfo(mat, "Getting entire matrix as submatrix\n"));
8390: if (cll == MAT_INITIAL_MATRIX) {
8391: *newmat = mat;
8392: PetscCall(PetscObjectReference((PetscObject)mat));
8393: }
8394: PetscFunctionReturn(PETSC_SUCCESS);
8395: }
8396: }
8398: if (!iscol) {
8399: PetscCall(ISCreateStride(PetscObjectComm((PetscObject)mat), mat->cmap->n, mat->cmap->rstart, 1, &iscoltmp));
8400: } else {
8401: iscoltmp = iscol;
8402: }
8404: /* if original matrix is on just one processor then use submatrix generated */
8405: if (mat->ops->createsubmatrices && !mat->ops->createsubmatrix && size == 1 && cll == MAT_REUSE_MATRIX) {
8406: PetscCall(MatCreateSubMatrices(mat, 1, &isrow, &iscoltmp, MAT_REUSE_MATRIX, &newmat));
8407: goto setproperties;
8408: } else if (mat->ops->createsubmatrices && !mat->ops->createsubmatrix && size == 1) {
8409: PetscCall(MatCreateSubMatrices(mat, 1, &isrow, &iscoltmp, MAT_INITIAL_MATRIX, &local));
8410: *newmat = *local;
8411: PetscCall(PetscFree(local));
8412: goto setproperties;
8413: } else if (!mat->ops->createsubmatrix) {
8414: /* Create a new matrix type that implements the operation using the full matrix */
8415: PetscCall(PetscLogEventBegin(MAT_CreateSubMat, mat, 0, 0, 0));
8416: switch (cll) {
8417: case MAT_INITIAL_MATRIX:
8418: PetscCall(MatCreateSubMatrixVirtual(mat, isrow, iscoltmp, newmat));
8419: break;
8420: case MAT_REUSE_MATRIX:
8421: PetscCall(MatSubMatrixVirtualUpdate(*newmat, mat, isrow, iscoltmp));
8422: break;
8423: default:
8424: SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "Invalid MatReuse, must be either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX");
8425: }
8426: PetscCall(PetscLogEventEnd(MAT_CreateSubMat, mat, 0, 0, 0));
8427: goto setproperties;
8428: }
8430: PetscCall(PetscLogEventBegin(MAT_CreateSubMat, mat, 0, 0, 0));
8431: PetscUseTypeMethod(mat, createsubmatrix, isrow, iscoltmp, cll, newmat);
8432: PetscCall(PetscLogEventEnd(MAT_CreateSubMat, mat, 0, 0, 0));
8434: setproperties:
8435: PetscCall(ISEqualUnsorted(isrow, iscoltmp, &flg));
8436: if (flg) PetscCall(MatPropagateSymmetryOptions(mat, *newmat));
8437: if (!iscol) PetscCall(ISDestroy(&iscoltmp));
8438: if (*newmat && cll == MAT_INITIAL_MATRIX) PetscCall(PetscObjectStateIncrease((PetscObject)*newmat));
8439: PetscFunctionReturn(PETSC_SUCCESS);
8440: }
8442: /*@
8443: MatPropagateSymmetryOptions - Propagates symmetry options set on a matrix to another matrix
8445: Not Collective
8447: Input Parameters:
8448: + A - the matrix we wish to propagate options from
8449: - B - the matrix we wish to propagate options to
8451: Level: beginner
8453: Note:
8454: Propagates the options associated to `MAT_SYMMETRY_ETERNAL`, `MAT_STRUCTURALLY_SYMMETRIC`, `MAT_HERMITIAN`, `MAT_SPD`, `MAT_SYMMETRIC`, and `MAT_STRUCTURAL_SYMMETRY_ETERNAL`
8456: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MatIsSymmetricKnown()`, `MatIsSPDKnown()`, `MatIsHermitianKnown()`, MatIsStructurallySymmetricKnown()`
8457: @*/
8458: PetscErrorCode MatPropagateSymmetryOptions(Mat A, Mat B)
8459: {
8460: PetscFunctionBegin;
8463: B->symmetry_eternal = A->symmetry_eternal;
8464: B->structural_symmetry_eternal = A->structural_symmetry_eternal;
8465: B->symmetric = A->symmetric;
8466: B->structurally_symmetric = A->structurally_symmetric;
8467: B->spd = A->spd;
8468: B->hermitian = A->hermitian;
8469: PetscFunctionReturn(PETSC_SUCCESS);
8470: }
8472: /*@
8473: MatStashSetInitialSize - sets the sizes of the matrix stash, that is
8474: used during the assembly process to store values that belong to
8475: other processors.
8477: Not Collective
8479: Input Parameters:
8480: + mat - the matrix
8481: . size - the initial size of the stash.
8482: - bsize - the initial size of the block-stash(if used).
8484: Options Database Keys:
8485: + -matstash_initial_size <size> or <size0,size1,...sizep-1>
8486: - -matstash_block_initial_size <bsize> or <bsize0,bsize1,...bsizep-1>
8488: Level: intermediate
8490: Notes:
8491: The block-stash is used for values set with `MatSetValuesBlocked()` while
8492: the stash is used for values set with `MatSetValues()`
8494: Run with the option -info and look for output of the form
8495: MatAssemblyBegin_MPIXXX:Stash has MM entries, uses nn mallocs.
8496: to determine the appropriate value, MM, to use for size and
8497: MatAssemblyBegin_MPIXXX:Block-Stash has BMM entries, uses nn mallocs.
8498: to determine the value, BMM to use for bsize
8500: .seealso: [](ch_matrices), `MatAssemblyBegin()`, `MatAssemblyEnd()`, `Mat`, `MatStashGetInfo()`
8501: @*/
8502: PetscErrorCode MatStashSetInitialSize(Mat mat, PetscInt size, PetscInt bsize)
8503: {
8504: PetscFunctionBegin;
8507: PetscCall(MatStashSetInitialSize_Private(&mat->stash, size));
8508: PetscCall(MatStashSetInitialSize_Private(&mat->bstash, bsize));
8509: PetscFunctionReturn(PETSC_SUCCESS);
8510: }
8512: /*@
8513: MatInterpolateAdd - w = y + A*x or A'*x depending on the shape of
8514: the matrix
8516: Neighbor-wise Collective
8518: Input Parameters:
8519: + mat - the matrix
8520: . x - the vector to be multiplied by the interpolation operator
8521: - y - the vector to be added to the result
8523: Output Parameter:
8524: . w - the resulting vector
8526: Level: intermediate
8528: Notes:
8529: `w` may be the same vector as `y`.
8531: This allows one to use either the restriction or interpolation (its transpose)
8532: matrix to do the interpolation
8534: .seealso: [](ch_matrices), `Mat`, `MatMultAdd()`, `MatMultTransposeAdd()`, `MatRestrict()`, `PCMG`
8535: @*/
8536: PetscErrorCode MatInterpolateAdd(Mat A, Vec x, Vec y, Vec w)
8537: {
8538: PetscInt M, N, Ny;
8540: PetscFunctionBegin;
8545: PetscCall(MatGetSize(A, &M, &N));
8546: PetscCall(VecGetSize(y, &Ny));
8547: if (M == Ny) {
8548: PetscCall(MatMultAdd(A, x, y, w));
8549: } else {
8550: PetscCall(MatMultTransposeAdd(A, x, y, w));
8551: }
8552: PetscFunctionReturn(PETSC_SUCCESS);
8553: }
8555: /*@
8556: MatInterpolate - y = A*x or A'*x depending on the shape of
8557: the matrix
8559: Neighbor-wise Collective
8561: Input Parameters:
8562: + mat - the matrix
8563: - x - the vector to be interpolated
8565: Output Parameter:
8566: . y - the resulting vector
8568: Level: intermediate
8570: Note:
8571: This allows one to use either the restriction or interpolation (its transpose)
8572: matrix to do the interpolation
8574: .seealso: [](ch_matrices), `Mat`, `MatMultAdd()`, `MatMultTransposeAdd()`, `MatRestrict()`, `PCMG`
8575: @*/
8576: PetscErrorCode MatInterpolate(Mat A, Vec x, Vec y)
8577: {
8578: PetscInt M, N, Ny;
8580: PetscFunctionBegin;
8584: PetscCall(MatGetSize(A, &M, &N));
8585: PetscCall(VecGetSize(y, &Ny));
8586: if (M == Ny) {
8587: PetscCall(MatMult(A, x, y));
8588: } else {
8589: PetscCall(MatMultTranspose(A, x, y));
8590: }
8591: PetscFunctionReturn(PETSC_SUCCESS);
8592: }
8594: /*@
8595: MatRestrict - y = A*x or A'*x
8597: Neighbor-wise Collective
8599: Input Parameters:
8600: + mat - the matrix
8601: - x - the vector to be restricted
8603: Output Parameter:
8604: . y - the resulting vector
8606: Level: intermediate
8608: Note:
8609: This allows one to use either the restriction or interpolation (its transpose)
8610: matrix to do the restriction
8612: .seealso: [](ch_matrices), `Mat`, `MatMultAdd()`, `MatMultTransposeAdd()`, `MatInterpolate()`, `PCMG`
8613: @*/
8614: PetscErrorCode MatRestrict(Mat A, Vec x, Vec y)
8615: {
8616: PetscInt M, N, Ny;
8618: PetscFunctionBegin;
8622: PetscCall(MatGetSize(A, &M, &N));
8623: PetscCall(VecGetSize(y, &Ny));
8624: if (M == Ny) {
8625: PetscCall(MatMult(A, x, y));
8626: } else {
8627: PetscCall(MatMultTranspose(A, x, y));
8628: }
8629: PetscFunctionReturn(PETSC_SUCCESS);
8630: }
8632: /*@
8633: MatMatInterpolateAdd - Y = W + A*X or W + A'*X
8635: Neighbor-wise Collective
8637: Input Parameters:
8638: + mat - the matrix
8639: . x - the input dense matrix to be multiplied
8640: - w - the input dense matrix to be added to the result
8642: Output Parameter:
8643: . y - the output dense matrix
8645: Level: intermediate
8647: Note:
8648: This allows one to use either the restriction or interpolation (its transpose)
8649: matrix to do the interpolation. y matrix can be reused if already created with the proper sizes,
8650: otherwise it will be recreated. y must be initialized to `NULL` if not supplied.
8652: .seealso: [](ch_matrices), `Mat`, `MatInterpolateAdd()`, `MatMatInterpolate()`, `MatMatRestrict()`, `PCMG`
8653: @*/
8654: PetscErrorCode MatMatInterpolateAdd(Mat A, Mat x, Mat w, Mat *y)
8655: {
8656: PetscInt M, N, Mx, Nx, Mo, My = 0, Ny = 0;
8657: PetscBool trans = PETSC_TRUE;
8658: MatReuse reuse = MAT_INITIAL_MATRIX;
8660: PetscFunctionBegin;
8666: PetscCall(MatGetSize(A, &M, &N));
8667: PetscCall(MatGetSize(x, &Mx, &Nx));
8668: if (N == Mx) trans = PETSC_FALSE;
8669: else PetscCheck(M == Mx, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Size mismatch: A %" PetscInt_FMT "x%" PetscInt_FMT ", X %" PetscInt_FMT "x%" PetscInt_FMT, M, N, Mx, Nx);
8670: Mo = trans ? N : M;
8671: if (*y) {
8672: PetscCall(MatGetSize(*y, &My, &Ny));
8673: if (Mo == My && Nx == Ny) {
8674: reuse = MAT_REUSE_MATRIX;
8675: } else {
8676: PetscCheck(w || *y != w, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Cannot reuse y and w, size mismatch: A %" PetscInt_FMT "x%" PetscInt_FMT ", X %" PetscInt_FMT "x%" PetscInt_FMT ", Y %" PetscInt_FMT "x%" PetscInt_FMT, M, N, Mx, Nx, My, Ny);
8677: PetscCall(MatDestroy(y));
8678: }
8679: }
8681: if (w && *y == w) { /* this is to minimize changes in PCMG */
8682: PetscBool flg;
8684: PetscCall(PetscObjectQuery((PetscObject)*y, "__MatMatIntAdd_w", (PetscObject *)&w));
8685: if (w) {
8686: PetscInt My, Ny, Mw, Nw;
8688: PetscCall(PetscObjectTypeCompare((PetscObject)*y, ((PetscObject)w)->type_name, &flg));
8689: PetscCall(MatGetSize(*y, &My, &Ny));
8690: PetscCall(MatGetSize(w, &Mw, &Nw));
8691: if (!flg || My != Mw || Ny != Nw) w = NULL;
8692: }
8693: if (!w) {
8694: PetscCall(MatDuplicate(*y, MAT_COPY_VALUES, &w));
8695: PetscCall(PetscObjectCompose((PetscObject)*y, "__MatMatIntAdd_w", (PetscObject)w));
8696: PetscCall(PetscObjectDereference((PetscObject)w));
8697: } else {
8698: PetscCall(MatCopy(*y, w, UNKNOWN_NONZERO_PATTERN));
8699: }
8700: }
8701: if (!trans) {
8702: PetscCall(MatMatMult(A, x, reuse, PETSC_DEFAULT, y));
8703: } else {
8704: PetscCall(MatTransposeMatMult(A, x, reuse, PETSC_DEFAULT, y));
8705: }
8706: if (w) PetscCall(MatAXPY(*y, 1.0, w, UNKNOWN_NONZERO_PATTERN));
8707: PetscFunctionReturn(PETSC_SUCCESS);
8708: }
8710: /*@
8711: MatMatInterpolate - Y = A*X or A'*X
8713: Neighbor-wise Collective
8715: Input Parameters:
8716: + mat - the matrix
8717: - x - the input dense matrix
8719: Output Parameter:
8720: . y - the output dense matrix
8722: Level: intermediate
8724: Note:
8725: This allows one to use either the restriction or interpolation (its transpose)
8726: matrix to do the interpolation. y matrix can be reused if already created with the proper sizes,
8727: otherwise it will be recreated. y must be initialized to `NULL` if not supplied.
8729: .seealso: [](ch_matrices), `Mat`, `MatInterpolate()`, `MatRestrict()`, `MatMatRestrict()`, `PCMG`
8730: @*/
8731: PetscErrorCode MatMatInterpolate(Mat A, Mat x, Mat *y)
8732: {
8733: PetscFunctionBegin;
8734: PetscCall(MatMatInterpolateAdd(A, x, NULL, y));
8735: PetscFunctionReturn(PETSC_SUCCESS);
8736: }
8738: /*@
8739: MatMatRestrict - Y = A*X or A'*X
8741: Neighbor-wise Collective
8743: Input Parameters:
8744: + mat - the matrix
8745: - x - the input dense matrix
8747: Output Parameter:
8748: . y - the output dense matrix
8750: Level: intermediate
8752: Note:
8753: This allows one to use either the restriction or interpolation (its transpose)
8754: matrix to do the restriction. y matrix can be reused if already created with the proper sizes,
8755: otherwise it will be recreated. y must be initialized to `NULL` if not supplied.
8757: .seealso: [](ch_matrices), `Mat`, `MatRestrict()`, `MatInterpolate()`, `MatMatInterpolate()`, `PCMG`
8758: @*/
8759: PetscErrorCode MatMatRestrict(Mat A, Mat x, Mat *y)
8760: {
8761: PetscFunctionBegin;
8762: PetscCall(MatMatInterpolateAdd(A, x, NULL, y));
8763: PetscFunctionReturn(PETSC_SUCCESS);
8764: }
8766: /*@
8767: MatGetNullSpace - retrieves the null space of a matrix.
8769: Logically Collective
8771: Input Parameters:
8772: + mat - the matrix
8773: - nullsp - the null space object
8775: Level: developer
8777: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatNullSpaceCreate()`, `MatSetNearNullSpace()`, `MatSetNullSpace()`, `MatNullSpace`
8778: @*/
8779: PetscErrorCode MatGetNullSpace(Mat mat, MatNullSpace *nullsp)
8780: {
8781: PetscFunctionBegin;
8784: *nullsp = (mat->symmetric == PETSC_BOOL3_TRUE && !mat->nullsp) ? mat->transnullsp : mat->nullsp;
8785: PetscFunctionReturn(PETSC_SUCCESS);
8786: }
8788: /*@
8789: MatSetNullSpace - attaches a null space to a matrix.
8791: Logically Collective
8793: Input Parameters:
8794: + mat - the matrix
8795: - nullsp - the null space object
8797: Level: advanced
8799: Notes:
8800: This null space is used by the `KSP` linear solvers to solve singular systems.
8802: Overwrites any previous null space that may have been attached. You can remove the null space from the matrix object by calling this routine with an nullsp of `NULL`
8804: For inconsistent singular systems (linear systems where the right hand side is not in the range of the operator) the `KSP` residuals will not converge to
8805: to zero but the linear system will still be solved in a least squares sense.
8807: The fundamental theorem of linear algebra (Gilbert Strang, Introduction to Applied Mathematics, page 72) states that
8808: the domain of a matrix A (from R^n to R^m (m rows, n columns) R^n = the direct sum of the null space of A, n(A), + the range of A^T, R(A^T).
8809: Similarly R^m = direct sum n(A^T) + R(A). Hence the linear system A x = b has a solution only if b in R(A) (or correspondingly b is orthogonal to
8810: n(A^T)) and if x is a solution then x + alpha n(A) is a solution for any alpha. The minimum norm solution is orthogonal to n(A). For problems without a solution
8811: the solution that minimizes the norm of the residual (the least squares solution) can be obtained by solving A x = \hat{b} where \hat{b} is b orthogonalized to the n(A^T).
8812: This \hat{b} can be obtained by calling MatNullSpaceRemove() with the null space of the transpose of the matrix.
8814: If the matrix is known to be symmetric because it is an `MATSBAIJ` matrix or one as called
8815: `MatSetOption`(mat,`MAT_SYMMETRIC` or possibly `MAT_SYMMETRY_ETERNAL`,`PETSC_TRUE`); this
8816: routine also automatically calls `MatSetTransposeNullSpace()`.
8818: The user should call `MatNullSpaceDestroy()`.
8820: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatNullSpaceCreate()`, `MatSetNearNullSpace()`, `MatGetNullSpace()`, `MatSetTransposeNullSpace()`, `MatGetTransposeNullSpace()`, `MatNullSpaceRemove()`,
8821: `KSPSetPCSide()`
8822: @*/
8823: PetscErrorCode MatSetNullSpace(Mat mat, MatNullSpace nullsp)
8824: {
8825: PetscFunctionBegin;
8828: if (nullsp) PetscCall(PetscObjectReference((PetscObject)nullsp));
8829: PetscCall(MatNullSpaceDestroy(&mat->nullsp));
8830: mat->nullsp = nullsp;
8831: if (mat->symmetric == PETSC_BOOL3_TRUE) PetscCall(MatSetTransposeNullSpace(mat, nullsp));
8832: PetscFunctionReturn(PETSC_SUCCESS);
8833: }
8835: /*@
8836: MatGetTransposeNullSpace - retrieves the null space of the transpose of a matrix.
8838: Logically Collective
8840: Input Parameters:
8841: + mat - the matrix
8842: - nullsp - the null space object
8844: Level: developer
8846: .seealso: [](ch_matrices), `Mat`, `MatNullSpace`, `MatCreate()`, `MatNullSpaceCreate()`, `MatSetNearNullSpace()`, `MatSetTransposeNullSpace()`, `MatSetNullSpace()`, `MatGetNullSpace()`
8847: @*/
8848: PetscErrorCode MatGetTransposeNullSpace(Mat mat, MatNullSpace *nullsp)
8849: {
8850: PetscFunctionBegin;
8854: *nullsp = (mat->symmetric == PETSC_BOOL3_TRUE && !mat->transnullsp) ? mat->nullsp : mat->transnullsp;
8855: PetscFunctionReturn(PETSC_SUCCESS);
8856: }
8858: /*@
8859: MatSetTransposeNullSpace - attaches the null space of a transpose of a matrix to the matrix
8861: Logically Collective
8863: Input Parameters:
8864: + mat - the matrix
8865: - nullsp - the null space object
8867: Level: advanced
8869: Notes:
8870: This allows solving singular linear systems defined by the transpose of the matrix using `KSP` solvers with left preconditioning.
8872: See `MatSetNullSpace()`
8874: .seealso: [](ch_matrices), `Mat`, `MatNullSpace`, `MatCreate()`, `MatNullSpaceCreate()`, `MatSetNearNullSpace()`, `MatGetNullSpace()`, `MatSetNullSpace()`, `MatGetTransposeNullSpace()`, `MatNullSpaceRemove()`, `KSPSetPCSide()`
8875: @*/
8876: PetscErrorCode MatSetTransposeNullSpace(Mat mat, MatNullSpace nullsp)
8877: {
8878: PetscFunctionBegin;
8881: if (nullsp) PetscCall(PetscObjectReference((PetscObject)nullsp));
8882: PetscCall(MatNullSpaceDestroy(&mat->transnullsp));
8883: mat->transnullsp = nullsp;
8884: PetscFunctionReturn(PETSC_SUCCESS);
8885: }
8887: /*@
8888: MatSetNearNullSpace - attaches a null space to a matrix, which is often the null space (rigid body modes) of the operator without boundary conditions
8889: This null space will be used to provide near null space vectors to a multigrid preconditioner built from this matrix.
8891: Logically Collective
8893: Input Parameters:
8894: + mat - the matrix
8895: - nullsp - the null space object
8897: Level: advanced
8899: Notes:
8900: Overwrites any previous near null space that may have been attached
8902: You can remove the null space by calling this routine with an nullsp of `NULL`
8904: .seealso: [](ch_matrices), `Mat`, `MatNullSpace`, `MatCreate()`, `MatNullSpaceCreate()`, `MatSetNullSpace()`, `MatNullSpaceCreateRigidBody()`, `MatGetNearNullSpace()`
8905: @*/
8906: PetscErrorCode MatSetNearNullSpace(Mat mat, MatNullSpace nullsp)
8907: {
8908: PetscFunctionBegin;
8912: MatCheckPreallocated(mat, 1);
8913: if (nullsp) PetscCall(PetscObjectReference((PetscObject)nullsp));
8914: PetscCall(MatNullSpaceDestroy(&mat->nearnullsp));
8915: mat->nearnullsp = nullsp;
8916: PetscFunctionReturn(PETSC_SUCCESS);
8917: }
8919: /*@
8920: MatGetNearNullSpace - Get null space attached with `MatSetNearNullSpace()`
8922: Not Collective
8924: Input Parameter:
8925: . mat - the matrix
8927: Output Parameter:
8928: . nullsp - the null space object, `NULL` if not set
8930: Level: advanced
8932: .seealso: [](ch_matrices), `Mat`, `MatNullSpace`, `MatSetNearNullSpace()`, `MatGetNullSpace()`, `MatNullSpaceCreate()`
8933: @*/
8934: PetscErrorCode MatGetNearNullSpace(Mat mat, MatNullSpace *nullsp)
8935: {
8936: PetscFunctionBegin;
8940: MatCheckPreallocated(mat, 1);
8941: *nullsp = mat->nearnullsp;
8942: PetscFunctionReturn(PETSC_SUCCESS);
8943: }
8945: /*@C
8946: MatICCFactor - Performs in-place incomplete Cholesky factorization of matrix.
8948: Collective
8950: Input Parameters:
8951: + mat - the matrix
8952: . row - row/column permutation
8953: - info - information on desired factorization process
8955: Level: developer
8957: Notes:
8958: Probably really in-place only when level of fill is zero, otherwise allocates
8959: new space to store factored matrix and deletes previous memory.
8961: Most users should employ the `KSP` interface for linear solvers
8962: instead of working directly with matrix algebra routines such as this.
8963: See, e.g., `KSPCreate()`.
8965: Developer Note:
8966: The Fortran interface is not autogenerated as the
8967: interface definition cannot be generated correctly [due to `MatFactorInfo`]
8969: .seealso: [](ch_matrices), `Mat`, `MatFactorInfo`, `MatGetFactor()`, `MatICCFactorSymbolic()`, `MatLUFactorNumeric()`, `MatCholeskyFactor()`
8970: @*/
8971: PetscErrorCode MatICCFactor(Mat mat, IS row, const MatFactorInfo *info)
8972: {
8973: PetscFunctionBegin;
8978: PetscCheck(mat->rmap->N == mat->cmap->N, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONG, "matrix must be square");
8979: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
8980: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
8981: MatCheckPreallocated(mat, 1);
8982: PetscUseTypeMethod(mat, iccfactor, row, info);
8983: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
8984: PetscFunctionReturn(PETSC_SUCCESS);
8985: }
8987: /*@
8988: MatDiagonalScaleLocal - Scales columns of a matrix given the scaling values including the
8989: ghosted ones.
8991: Not Collective
8993: Input Parameters:
8994: + mat - the matrix
8995: - diag - the diagonal values, including ghost ones
8997: Level: developer
8999: Notes:
9000: Works only for `MATMPIAIJ` and `MATMPIBAIJ` matrices
9002: This allows one to avoid during communication to perform the scaling that must be done with `MatDiagonalScale()`
9004: .seealso: [](ch_matrices), `Mat`, `MatDiagonalScale()`
9005: @*/
9006: PetscErrorCode MatDiagonalScaleLocal(Mat mat, Vec diag)
9007: {
9008: PetscMPIInt size;
9010: PetscFunctionBegin;
9015: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Matrix must be already assembled");
9016: PetscCall(PetscLogEventBegin(MAT_Scale, mat, 0, 0, 0));
9017: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size));
9018: if (size == 1) {
9019: PetscInt n, m;
9020: PetscCall(VecGetSize(diag, &n));
9021: PetscCall(MatGetSize(mat, NULL, &m));
9022: PetscCheck(m == n, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only supported for sequential matrices when no ghost points/periodic conditions");
9023: PetscCall(MatDiagonalScale(mat, NULL, diag));
9024: } else {
9025: PetscUseMethod(mat, "MatDiagonalScaleLocal_C", (Mat, Vec), (mat, diag));
9026: }
9027: PetscCall(PetscLogEventEnd(MAT_Scale, mat, 0, 0, 0));
9028: PetscCall(PetscObjectStateIncrease((PetscObject)mat));
9029: PetscFunctionReturn(PETSC_SUCCESS);
9030: }
9032: /*@
9033: MatGetInertia - Gets the inertia from a factored matrix
9035: Collective
9037: Input Parameter:
9038: . mat - the matrix
9040: Output Parameters:
9041: + nneg - number of negative eigenvalues
9042: . nzero - number of zero eigenvalues
9043: - npos - number of positive eigenvalues
9045: Level: advanced
9047: Note:
9048: Matrix must have been factored by `MatCholeskyFactor()`
9050: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatCholeskyFactor()`
9051: @*/
9052: PetscErrorCode MatGetInertia(Mat mat, PetscInt *nneg, PetscInt *nzero, PetscInt *npos)
9053: {
9054: PetscFunctionBegin;
9057: PetscCheck(mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Unfactored matrix");
9058: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Numeric factor mat is not assembled");
9059: PetscUseTypeMethod(mat, getinertia, nneg, nzero, npos);
9060: PetscFunctionReturn(PETSC_SUCCESS);
9061: }
9063: /*@C
9064: MatSolves - Solves A x = b, given a factored matrix, for a collection of vectors
9066: Neighbor-wise Collective
9068: Input Parameters:
9069: + mat - the factored matrix obtained with `MatGetFactor()`
9070: - b - the right-hand-side vectors
9072: Output Parameter:
9073: . x - the result vectors
9075: Level: developer
9077: Note:
9078: The vectors `b` and `x` cannot be the same. I.e., one cannot
9079: call `MatSolves`(A,x,x).
9081: .seealso: [](ch_matrices), `Mat`, `Vecs`, `MatSolveAdd()`, `MatSolveTranspose()`, `MatSolveTransposeAdd()`, `MatSolve()`
9082: @*/
9083: PetscErrorCode MatSolves(Mat mat, Vecs b, Vecs x)
9084: {
9085: PetscFunctionBegin;
9088: PetscCheck(x != b, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_IDN, "x and b must be different vectors");
9089: PetscCheck(mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Unfactored matrix");
9090: if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(PETSC_SUCCESS);
9092: MatCheckPreallocated(mat, 1);
9093: PetscCall(PetscLogEventBegin(MAT_Solves, mat, 0, 0, 0));
9094: PetscUseTypeMethod(mat, solves, b, x);
9095: PetscCall(PetscLogEventEnd(MAT_Solves, mat, 0, 0, 0));
9096: PetscFunctionReturn(PETSC_SUCCESS);
9097: }
9099: /*@
9100: MatIsSymmetric - Test whether a matrix is symmetric
9102: Collective
9104: Input Parameters:
9105: + A - the matrix to test
9106: - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact transpose)
9108: Output Parameter:
9109: . flg - the result
9111: Level: intermediate
9113: Notes:
9114: For real numbers `MatIsSymmetric()` and `MatIsHermitian()` return identical results
9116: If the matrix does not yet know if it is symmetric or not this can be an expensive operation, also available `MatIsSymmetricKnown()`
9118: One can declare that a matrix is symmetric with `MatSetOption`(mat,`MAT_SYMMETRIC`,`PETSC_TRUE`) and if it is known to remain symmetric
9119: after changes to the matrices values one can call `MatSetOption`(mat,`MAT_SYMMETRY_ETERNAL`,`PETSC_TRUE`)
9121: .seealso: [](ch_matrices), `Mat`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, `MatSetOption()`, `MatIsSymmetricKnown()`,
9122: `MAT_SYMMETRIC`, `MAT_SYMMETRY_ETERNAL`, `MatSetOption()`
9123: @*/
9124: PetscErrorCode MatIsSymmetric(Mat A, PetscReal tol, PetscBool *flg)
9125: {
9126: PetscFunctionBegin;
9130: if (A->symmetric == PETSC_BOOL3_TRUE) *flg = PETSC_TRUE;
9131: else if (A->symmetric == PETSC_BOOL3_FALSE) *flg = PETSC_FALSE;
9132: else {
9133: PetscUseTypeMethod(A, issymmetric, tol, flg);
9134: if (!tol) PetscCall(MatSetOption(A, MAT_SYMMETRIC, *flg));
9135: }
9136: PetscFunctionReturn(PETSC_SUCCESS);
9137: }
9139: /*@
9140: MatIsHermitian - Test whether a matrix is Hermitian
9142: Collective
9144: Input Parameters:
9145: + A - the matrix to test
9146: - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact Hermitian)
9148: Output Parameter:
9149: . flg - the result
9151: Level: intermediate
9153: Notes:
9154: For real numbers `MatIsSymmetric()` and `MatIsHermitian()` return identical results
9156: If the matrix does not yet know if it is Hermitian or not this can be an expensive operation, also available `MatIsHermitianKnown()`
9158: One can declare that a matrix is Hermitian with `MatSetOption`(mat,`MAT_HERMITIAN`,`PETSC_TRUE`) and if it is known to remain Hermitian
9159: after changes to the matrices values one can call `MatSetOption`(mat,`MAT_SYMEMTRY_ETERNAL`,`PETSC_TRUE`)
9161: .seealso: [](ch_matrices), `Mat`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitianKnown()`, `MatIsStructurallySymmetric()`, `MatSetOption()`,
9162: `MatIsSymmetricKnown()`, `MatIsSymmetric()`, `MAT_HERMITIAN`, `MAT_SYMMETRY_ETERNAL`, `MatSetOption()`
9163: @*/
9164: PetscErrorCode MatIsHermitian(Mat A, PetscReal tol, PetscBool *flg)
9165: {
9166: PetscFunctionBegin;
9170: if (A->hermitian == PETSC_BOOL3_TRUE) *flg = PETSC_TRUE;
9171: else if (A->hermitian == PETSC_BOOL3_FALSE) *flg = PETSC_FALSE;
9172: else {
9173: PetscUseTypeMethod(A, ishermitian, tol, flg);
9174: if (!tol) PetscCall(MatSetOption(A, MAT_HERMITIAN, *flg));
9175: }
9176: PetscFunctionReturn(PETSC_SUCCESS);
9177: }
9179: /*@
9180: MatIsSymmetricKnown - Checks if a matrix knows if it is symmetric or not and its symmetric state
9182: Not Collective
9184: Input Parameter:
9185: . A - the matrix to check
9187: Output Parameters:
9188: + set - `PETSC_TRUE` if the matrix knows its symmetry state (this tells you if the next flag is valid)
9189: - flg - the result (only valid if set is `PETSC_TRUE`)
9191: Level: advanced
9193: Notes:
9194: Does not check the matrix values directly, so this may return unknown (set = `PETSC_FALSE`). Use `MatIsSymmetric()`
9195: if you want it explicitly checked
9197: One can declare that a matrix is symmetric with `MatSetOption`(mat,`MAT_SYMMETRIC`,`PETSC_TRUE`) and if it is known to remain symmetric
9198: after changes to the matrices values one can call `MatSetOption`(mat,`MAT_SYMMETRY_ETERNAL`,`PETSC_TRUE`)
9200: .seealso: [](ch_matrices), `Mat`, `MAT_SYMMETRY_ETERNAL`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, `MatSetOption()`, `MatIsSymmetric()`, `MatIsHermitianKnown()`
9201: @*/
9202: PetscErrorCode MatIsSymmetricKnown(Mat A, PetscBool *set, PetscBool *flg)
9203: {
9204: PetscFunctionBegin;
9208: if (A->symmetric != PETSC_BOOL3_UNKNOWN) {
9209: *set = PETSC_TRUE;
9210: *flg = PetscBool3ToBool(A->symmetric);
9211: } else {
9212: *set = PETSC_FALSE;
9213: }
9214: PetscFunctionReturn(PETSC_SUCCESS);
9215: }
9217: /*@
9218: MatIsSPDKnown - Checks if a matrix knows if it is symmetric positive definite or not and its symmetric positive definite state
9220: Not Collective
9222: Input Parameter:
9223: . A - the matrix to check
9225: Output Parameters:
9226: + set - `PETSC_TRUE` if the matrix knows its symmetric positive definite state (this tells you if the next flag is valid)
9227: - flg - the result (only valid if set is `PETSC_TRUE`)
9229: Level: advanced
9231: Notes:
9232: Does not check the matrix values directly, so this may return unknown (set = `PETSC_FALSE`).
9234: One can declare that a matrix is SPD with `MatSetOption`(mat,`MAT_SPD`,`PETSC_TRUE`) and if it is known to remain SPD
9235: after changes to the matrices values one can call `MatSetOption`(mat,`MAT_SPD_ETERNAL`,`PETSC_TRUE`)
9237: .seealso: [](ch_matrices), `Mat`, `MAT_SPD_ETERNAL`, `MAT_SPD`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, `MatSetOption()`, `MatIsSymmetric()`, `MatIsHermitianKnown()`
9238: @*/
9239: PetscErrorCode MatIsSPDKnown(Mat A, PetscBool *set, PetscBool *flg)
9240: {
9241: PetscFunctionBegin;
9245: if (A->spd != PETSC_BOOL3_UNKNOWN) {
9246: *set = PETSC_TRUE;
9247: *flg = PetscBool3ToBool(A->spd);
9248: } else {
9249: *set = PETSC_FALSE;
9250: }
9251: PetscFunctionReturn(PETSC_SUCCESS);
9252: }
9254: /*@
9255: MatIsHermitianKnown - Checks if a matrix knows if it is Hermitian or not and its Hermitian state
9257: Not Collective
9259: Input Parameter:
9260: . A - the matrix to check
9262: Output Parameters:
9263: + set - `PETSC_TRUE` if the matrix knows its Hermitian state (this tells you if the next flag is valid)
9264: - flg - the result (only valid if set is `PETSC_TRUE`)
9266: Level: advanced
9268: Notes:
9269: Does not check the matrix values directly, so this may return unknown (set = `PETSC_FALSE`). Use `MatIsHermitian()`
9270: if you want it explicitly checked
9272: One can declare that a matrix is Hermitian with `MatSetOption`(mat,`MAT_HERMITIAN`,`PETSC_TRUE`) and if it is known to remain Hermitian
9273: after changes to the matrices values one can call `MatSetOption`(mat,`MAT_SYMMETRY_ETERNAL`,`PETSC_TRUE`)
9275: .seealso: [](ch_matrices), `Mat`, `MAT_SYMMETRY_ETERNAL`, `MAT_HERMITIAN`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, `MatSetOption()`, `MatIsSymmetric()`
9276: @*/
9277: PetscErrorCode MatIsHermitianKnown(Mat A, PetscBool *set, PetscBool *flg)
9278: {
9279: PetscFunctionBegin;
9283: if (A->hermitian != PETSC_BOOL3_UNKNOWN) {
9284: *set = PETSC_TRUE;
9285: *flg = PetscBool3ToBool(A->hermitian);
9286: } else {
9287: *set = PETSC_FALSE;
9288: }
9289: PetscFunctionReturn(PETSC_SUCCESS);
9290: }
9292: /*@
9293: MatIsStructurallySymmetric - Test whether a matrix is structurally symmetric
9295: Collective
9297: Input Parameter:
9298: . A - the matrix to test
9300: Output Parameter:
9301: . flg - the result
9303: Level: intermediate
9305: Notes:
9306: If the matrix does yet know it is structurally symmetric this can be an expensive operation, also available `MatIsStructurallySymmetricKnown()`
9308: One can declare that a matrix is structurally symmetric with `MatSetOption`(mat,`MAT_STRUCTURALLY_SYMMETRIC`,`PETSC_TRUE`) and if it is known to remain structurally
9309: symmetric after changes to the matrices values one can call `MatSetOption`(mat,`MAT_STRUCTURAL_SYMMETRY_ETERNAL`,`PETSC_TRUE`)
9311: .seealso: [](ch_matrices), `Mat`, `MAT_STRUCTURALLY_SYMMETRIC`, `MAT_STRUCTURAL_SYMMETRY_ETERNAL`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsSymmetric()`, `MatSetOption()`, `MatIsStructurallySymmetricKnown()`
9312: @*/
9313: PetscErrorCode MatIsStructurallySymmetric(Mat A, PetscBool *flg)
9314: {
9315: PetscFunctionBegin;
9318: if (A->structurally_symmetric != PETSC_BOOL3_UNKNOWN) {
9319: *flg = PetscBool3ToBool(A->structurally_symmetric);
9320: } else {
9321: PetscUseTypeMethod(A, isstructurallysymmetric, flg);
9322: PetscCall(MatSetOption(A, MAT_STRUCTURALLY_SYMMETRIC, *flg));
9323: }
9324: PetscFunctionReturn(PETSC_SUCCESS);
9325: }
9327: /*@
9328: MatIsStructurallySymmetricKnown - Checks if a matrix knows if it is structurally symmetric or not and its structurally symmetric state
9330: Not Collective
9332: Input Parameter:
9333: . A - the matrix to check
9335: Output Parameters:
9336: + set - PETSC_TRUE if the matrix knows its structurally symmetric state (this tells you if the next flag is valid)
9337: - flg - the result (only valid if set is PETSC_TRUE)
9339: Level: advanced
9341: Notes:
9342: One can declare that a matrix is structurally symmetric with `MatSetOption`(mat,`MAT_STRUCTURALLY_SYMMETRIC`,`PETSC_TRUE`) and if it is known to remain structurally
9343: symmetric after changes to the matrices values one can call `MatSetOption`(mat,`MAT_STRUCTURAL_SYMMETRY_ETERNAL`,`PETSC_TRUE`)
9345: Use `MatIsStructurallySymmetric()` to explicitly check if a matrix is structurally symmetric (this is an expensive operation)
9347: .seealso: [](ch_matrices), `Mat`, `MAT_STRUCTURALLY_SYMMETRIC`, `MatTranspose()`, `MatIsTranspose()`, `MatIsHermitian()`, `MatIsStructurallySymmetric()`, `MatSetOption()`, `MatIsSymmetric()`, `MatIsHermitianKnown()`
9348: @*/
9349: PetscErrorCode MatIsStructurallySymmetricKnown(Mat A, PetscBool *set, PetscBool *flg)
9350: {
9351: PetscFunctionBegin;
9355: if (A->structurally_symmetric != PETSC_BOOL3_UNKNOWN) {
9356: *set = PETSC_TRUE;
9357: *flg = PetscBool3ToBool(A->structurally_symmetric);
9358: } else {
9359: *set = PETSC_FALSE;
9360: }
9361: PetscFunctionReturn(PETSC_SUCCESS);
9362: }
9364: /*@
9365: MatStashGetInfo - Gets how many values are currently in the matrix stash, i.e. need
9366: to be communicated to other processors during the `MatAssemblyBegin()`/`MatAssemblyEnd()` process
9368: Not Collective
9370: Input Parameter:
9371: . mat - the matrix
9373: Output Parameters:
9374: + nstash - the size of the stash
9375: . reallocs - the number of additional mallocs incurred.
9376: . bnstash - the size of the block stash
9377: - breallocs - the number of additional mallocs incurred.in the block stash
9379: Level: advanced
9381: .seealso: [](ch_matrices), `MatAssemblyBegin()`, `MatAssemblyEnd()`, `Mat`, `MatStashSetInitialSize()`
9382: @*/
9383: PetscErrorCode MatStashGetInfo(Mat mat, PetscInt *nstash, PetscInt *reallocs, PetscInt *bnstash, PetscInt *breallocs)
9384: {
9385: PetscFunctionBegin;
9386: PetscCall(MatStashGetInfo_Private(&mat->stash, nstash, reallocs));
9387: PetscCall(MatStashGetInfo_Private(&mat->bstash, bnstash, breallocs));
9388: PetscFunctionReturn(PETSC_SUCCESS);
9389: }
9391: /*@C
9392: MatCreateVecs - Get vector(s) compatible with the matrix, i.e. with the same
9393: parallel layout, `PetscLayout` for rows and columns
9395: Collective
9397: Input Parameter:
9398: . mat - the matrix
9400: Output Parameters:
9401: + right - (optional) vector that the matrix can be multiplied against
9402: - left - (optional) vector that the matrix vector product can be stored in
9404: Level: advanced
9406: Notes:
9407: The blocksize of the returned vectors is determined by the row and column block sizes set with `MatSetBlockSizes()` or the single blocksize (same for both) set by `MatSetBlockSize()`.
9409: These are new vectors which are not owned by the mat, they should be destroyed in `VecDestroy()` when no longer needed
9411: .seealso: [](ch_matrices), `Mat`, `Vec`, `VecCreate()`, `VecDestroy()`, `DMCreateGlobalVector()`
9412: @*/
9413: PetscErrorCode MatCreateVecs(Mat mat, Vec *right, Vec *left)
9414: {
9415: PetscFunctionBegin;
9418: if (mat->ops->getvecs) {
9419: PetscUseTypeMethod(mat, getvecs, right, left);
9420: } else {
9421: PetscInt rbs, cbs;
9422: PetscCall(MatGetBlockSizes(mat, &rbs, &cbs));
9423: if (right) {
9424: PetscCheck(mat->cmap->n >= 0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "PetscLayout for columns not yet setup");
9425: PetscCall(VecCreate(PetscObjectComm((PetscObject)mat), right));
9426: PetscCall(VecSetSizes(*right, mat->cmap->n, PETSC_DETERMINE));
9427: PetscCall(VecSetBlockSize(*right, cbs));
9428: PetscCall(VecSetType(*right, mat->defaultvectype));
9429: #if defined(PETSC_HAVE_VIENNACL) || defined(PETSC_HAVE_CUDA) || defined(PETSC_HAVE_HIP)
9430: if (mat->boundtocpu && mat->bindingpropagates) {
9431: PetscCall(VecSetBindingPropagates(*right, PETSC_TRUE));
9432: PetscCall(VecBindToCPU(*right, PETSC_TRUE));
9433: }
9434: #endif
9435: PetscCall(PetscLayoutReference(mat->cmap, &(*right)->map));
9436: }
9437: if (left) {
9438: PetscCheck(mat->rmap->n >= 0, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "PetscLayout for rows not yet setup");
9439: PetscCall(VecCreate(PetscObjectComm((PetscObject)mat), left));
9440: PetscCall(VecSetSizes(*left, mat->rmap->n, PETSC_DETERMINE));
9441: PetscCall(VecSetBlockSize(*left, rbs));
9442: PetscCall(VecSetType(*left, mat->defaultvectype));
9443: #if defined(PETSC_HAVE_VIENNACL) || defined(PETSC_HAVE_CUDA) || defined(PETSC_HAVE_HIP)
9444: if (mat->boundtocpu && mat->bindingpropagates) {
9445: PetscCall(VecSetBindingPropagates(*left, PETSC_TRUE));
9446: PetscCall(VecBindToCPU(*left, PETSC_TRUE));
9447: }
9448: #endif
9449: PetscCall(PetscLayoutReference(mat->rmap, &(*left)->map));
9450: }
9451: }
9452: PetscFunctionReturn(PETSC_SUCCESS);
9453: }
9455: /*@C
9456: MatFactorInfoInitialize - Initializes a `MatFactorInfo` data structure
9457: with default values.
9459: Not Collective
9461: Input Parameter:
9462: . info - the `MatFactorInfo` data structure
9464: Level: developer
9466: Notes:
9467: The solvers are generally used through the `KSP` and `PC` objects, for example
9468: `PCLU`, `PCILU`, `PCCHOLESKY`, `PCICC`
9470: Once the data structure is initialized one may change certain entries as desired for the particular factorization to be performed
9472: Developer Note:
9473: The Fortran interface is not autogenerated as the
9474: interface definition cannot be generated correctly [due to `MatFactorInfo`]
9476: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorInfo`
9477: @*/
9478: PetscErrorCode MatFactorInfoInitialize(MatFactorInfo *info)
9479: {
9480: PetscFunctionBegin;
9481: PetscCall(PetscMemzero(info, sizeof(MatFactorInfo)));
9482: PetscFunctionReturn(PETSC_SUCCESS);
9483: }
9485: /*@
9486: MatFactorSetSchurIS - Set indices corresponding to the Schur complement you wish to have computed
9488: Collective
9490: Input Parameters:
9491: + mat - the factored matrix
9492: - is - the index set defining the Schur indices (0-based)
9494: Level: advanced
9496: Notes:
9497: Call `MatFactorSolveSchurComplement()` or `MatFactorSolveSchurComplementTranspose()` after this call to solve a Schur complement system.
9499: You can call `MatFactorGetSchurComplement()` or `MatFactorCreateSchurComplement()` after this call.
9501: This functionality is only supported for `MATSOLVERMUMPS` and `MATSOLVERMKL_PARDISO`
9503: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorGetSchurComplement()`, `MatFactorRestoreSchurComplement()`, `MatFactorCreateSchurComplement()`, `MatFactorSolveSchurComplement()`,
9504: `MatFactorSolveSchurComplementTranspose()`, `MatFactorSolveSchurComplement()`, `MATSOLVERMUMPS`, `MATSOLVERMKL_PARDISO`
9505: @*/
9506: PetscErrorCode MatFactorSetSchurIS(Mat mat, IS is)
9507: {
9508: PetscErrorCode (*f)(Mat, IS);
9510: PetscFunctionBegin;
9515: PetscCheckSameComm(mat, 1, is, 2);
9516: PetscCheck(mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Only for factored matrix");
9517: PetscCall(PetscObjectQueryFunction((PetscObject)mat, "MatFactorSetSchurIS_C", &f));
9518: PetscCheck(f, PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "The selected MatSolverType does not support Schur complement computation. You should use MATSOLVERMUMPS or MATSOLVERMKL_PARDISO");
9519: PetscCall(MatDestroy(&mat->schur));
9520: PetscCall((*f)(mat, is));
9521: PetscCheck(mat->schur, PetscObjectComm((PetscObject)mat), PETSC_ERR_PLIB, "Schur complement has not been created");
9522: PetscFunctionReturn(PETSC_SUCCESS);
9523: }
9525: /*@
9526: MatFactorCreateSchurComplement - Create a Schur complement matrix object using Schur data computed during the factorization step
9528: Logically Collective
9530: Input Parameters:
9531: + F - the factored matrix obtained by calling `MatGetFactor()`
9532: . S - location where to return the Schur complement, can be `NULL`
9533: - status - the status of the Schur complement matrix, can be `NULL`
9535: Level: advanced
9537: Notes:
9538: You must call `MatFactorSetSchurIS()` before calling this routine.
9540: This functionality is only supported for `MATSOLVERMUMPS` and `MATSOLVERMKL_PARDISO`
9542: The routine provides a copy of the Schur matrix stored within the solver data structures.
9543: The caller must destroy the object when it is no longer needed.
9544: If `MatFactorInvertSchurComplement()` has been called, the routine gets back the inverse.
9546: Use `MatFactorGetSchurComplement()` to get access to the Schur complement matrix inside the factored matrix instead of making a copy of it (which this function does)
9548: See `MatCreateSchurComplement()` or `MatGetSchurComplement()` for ways to create virtual or approximate Schur complements.
9550: Developer Note:
9551: The reason this routine exists is because the representation of the Schur complement within the factor matrix may be different than a standard PETSc
9552: matrix representation and we normally do not want to use the time or memory to make a copy as a regular PETSc matrix.
9554: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorGetSchurComplement()`, `MatFactorSchurStatus`, `MATSOLVERMUMPS`, `MATSOLVERMKL_PARDISO`
9555: @*/
9556: PetscErrorCode MatFactorCreateSchurComplement(Mat F, Mat *S, MatFactorSchurStatus *status)
9557: {
9558: PetscFunctionBegin;
9562: if (S) {
9563: PetscErrorCode (*f)(Mat, Mat *);
9565: PetscCall(PetscObjectQueryFunction((PetscObject)F, "MatFactorCreateSchurComplement_C", &f));
9566: if (f) {
9567: PetscCall((*f)(F, S));
9568: } else {
9569: PetscCall(MatDuplicate(F->schur, MAT_COPY_VALUES, S));
9570: }
9571: }
9572: if (status) *status = F->schur_status;
9573: PetscFunctionReturn(PETSC_SUCCESS);
9574: }
9576: /*@
9577: MatFactorGetSchurComplement - Gets access to a Schur complement matrix using the current Schur data within a factored matrix
9579: Logically Collective
9581: Input Parameters:
9582: + F - the factored matrix obtained by calling `MatGetFactor()`
9583: . *S - location where to return the Schur complement, can be `NULL`
9584: - status - the status of the Schur complement matrix, can be `NULL`
9586: Level: advanced
9588: Notes:
9589: You must call `MatFactorSetSchurIS()` before calling this routine.
9591: Schur complement mode is currently implemented for sequential matrices with factor type of `MATSOLVERMUMPS`
9593: The routine returns a the Schur Complement stored within the data structures of the solver.
9595: If `MatFactorInvertSchurComplement()` has previously been called, the returned matrix is actually the inverse of the Schur complement.
9597: The returned matrix should not be destroyed; the caller should call `MatFactorRestoreSchurComplement()` when the object is no longer needed.
9599: Use `MatFactorCreateSchurComplement()` to create a copy of the Schur complement matrix that is within a factored matrix
9601: See `MatCreateSchurComplement()` or `MatGetSchurComplement()` for ways to create virtual or approximate Schur complements.
9603: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorRestoreSchurComplement()`, `MatFactorCreateSchurComplement()`, `MatFactorSchurStatus`
9604: @*/
9605: PetscErrorCode MatFactorGetSchurComplement(Mat F, Mat *S, MatFactorSchurStatus *status)
9606: {
9607: PetscFunctionBegin;
9611: if (S) *S = F->schur;
9612: if (status) *status = F->schur_status;
9613: PetscFunctionReturn(PETSC_SUCCESS);
9614: }
9616: static PetscErrorCode MatFactorUpdateSchurStatus_Private(Mat F)
9617: {
9618: Mat S = F->schur;
9620: PetscFunctionBegin;
9621: switch (F->schur_status) {
9622: case MAT_FACTOR_SCHUR_UNFACTORED: // fall-through
9623: case MAT_FACTOR_SCHUR_INVERTED:
9624: if (S) {
9625: S->ops->solve = NULL;
9626: S->ops->matsolve = NULL;
9627: S->ops->solvetranspose = NULL;
9628: S->ops->matsolvetranspose = NULL;
9629: S->ops->solveadd = NULL;
9630: S->ops->solvetransposeadd = NULL;
9631: S->factortype = MAT_FACTOR_NONE;
9632: PetscCall(PetscFree(S->solvertype));
9633: }
9634: case MAT_FACTOR_SCHUR_FACTORED: // fall-through
9635: break;
9636: default:
9637: SETERRQ(PetscObjectComm((PetscObject)F), PETSC_ERR_SUP, "Unhandled MatFactorSchurStatus %d", F->schur_status);
9638: }
9639: PetscFunctionReturn(PETSC_SUCCESS);
9640: }
9642: /*@
9643: MatFactorRestoreSchurComplement - Restore the Schur complement matrix object obtained from a call to `MatFactorGetSchurComplement()`
9645: Logically Collective
9647: Input Parameters:
9648: + F - the factored matrix obtained by calling `MatGetFactor()`
9649: . *S - location where the Schur complement is stored
9650: - status - the status of the Schur complement matrix (see `MatFactorSchurStatus`)
9652: Level: advanced
9654: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorRestoreSchurComplement()`, `MatFactorCreateSchurComplement()`, `MatFactorSchurStatus`
9655: @*/
9656: PetscErrorCode MatFactorRestoreSchurComplement(Mat F, Mat *S, MatFactorSchurStatus status)
9657: {
9658: PetscFunctionBegin;
9660: if (S) {
9662: *S = NULL;
9663: }
9664: F->schur_status = status;
9665: PetscCall(MatFactorUpdateSchurStatus_Private(F));
9666: PetscFunctionReturn(PETSC_SUCCESS);
9667: }
9669: /*@
9670: MatFactorSolveSchurComplementTranspose - Solve the transpose of the Schur complement system computed during the factorization step
9672: Logically Collective
9674: Input Parameters:
9675: + F - the factored matrix obtained by calling `MatGetFactor()`
9676: . rhs - location where the right hand side of the Schur complement system is stored
9677: - sol - location where the solution of the Schur complement system has to be returned
9679: Level: advanced
9681: Notes:
9682: The sizes of the vectors should match the size of the Schur complement
9684: Must be called after `MatFactorSetSchurIS()`
9686: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorSolveSchurComplement()`
9687: @*/
9688: PetscErrorCode MatFactorSolveSchurComplementTranspose(Mat F, Vec rhs, Vec sol)
9689: {
9690: PetscFunctionBegin;
9697: PetscCheckSameComm(F, 1, rhs, 2);
9698: PetscCheckSameComm(F, 1, sol, 3);
9699: PetscCall(MatFactorFactorizeSchurComplement(F));
9700: switch (F->schur_status) {
9701: case MAT_FACTOR_SCHUR_FACTORED:
9702: PetscCall(MatSolveTranspose(F->schur, rhs, sol));
9703: break;
9704: case MAT_FACTOR_SCHUR_INVERTED:
9705: PetscCall(MatMultTranspose(F->schur, rhs, sol));
9706: break;
9707: default:
9708: SETERRQ(PetscObjectComm((PetscObject)F), PETSC_ERR_SUP, "Unhandled MatFactorSchurStatus %d", F->schur_status);
9709: }
9710: PetscFunctionReturn(PETSC_SUCCESS);
9711: }
9713: /*@
9714: MatFactorSolveSchurComplement - Solve the Schur complement system computed during the factorization step
9716: Logically Collective
9718: Input Parameters:
9719: + F - the factored matrix obtained by calling `MatGetFactor()`
9720: . rhs - location where the right hand side of the Schur complement system is stored
9721: - sol - location where the solution of the Schur complement system has to be returned
9723: Level: advanced
9725: Notes:
9726: The sizes of the vectors should match the size of the Schur complement
9728: Must be called after `MatFactorSetSchurIS()`
9730: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorSolveSchurComplementTranspose()`
9731: @*/
9732: PetscErrorCode MatFactorSolveSchurComplement(Mat F, Vec rhs, Vec sol)
9733: {
9734: PetscFunctionBegin;
9741: PetscCheckSameComm(F, 1, rhs, 2);
9742: PetscCheckSameComm(F, 1, sol, 3);
9743: PetscCall(MatFactorFactorizeSchurComplement(F));
9744: switch (F->schur_status) {
9745: case MAT_FACTOR_SCHUR_FACTORED:
9746: PetscCall(MatSolve(F->schur, rhs, sol));
9747: break;
9748: case MAT_FACTOR_SCHUR_INVERTED:
9749: PetscCall(MatMult(F->schur, rhs, sol));
9750: break;
9751: default:
9752: SETERRQ(PetscObjectComm((PetscObject)F), PETSC_ERR_SUP, "Unhandled MatFactorSchurStatus %d", F->schur_status);
9753: }
9754: PetscFunctionReturn(PETSC_SUCCESS);
9755: }
9757: PETSC_EXTERN PetscErrorCode MatSeqDenseInvertFactors_Private(Mat);
9758: #if PetscDefined(HAVE_CUDA)
9759: PETSC_SINGLE_LIBRARY_INTERN PetscErrorCode MatSeqDenseCUDAInvertFactors_Internal(Mat);
9760: #endif
9762: /* Schur status updated in the interface */
9763: static PetscErrorCode MatFactorInvertSchurComplement_Private(Mat F)
9764: {
9765: Mat S = F->schur;
9767: PetscFunctionBegin;
9768: if (S) {
9769: PetscMPIInt size;
9770: PetscBool isdense, isdensecuda;
9772: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)S), &size));
9773: PetscCheck(size <= 1, PetscObjectComm((PetscObject)S), PETSC_ERR_SUP, "Not yet implemented");
9774: PetscCall(PetscObjectTypeCompare((PetscObject)S, MATSEQDENSE, &isdense));
9775: PetscCall(PetscObjectTypeCompare((PetscObject)S, MATSEQDENSECUDA, &isdensecuda));
9776: PetscCheck(isdense || isdensecuda, PetscObjectComm((PetscObject)S), PETSC_ERR_SUP, "Not implemented for type %s", ((PetscObject)S)->type_name);
9777: PetscCall(PetscLogEventBegin(MAT_FactorInvS, F, 0, 0, 0));
9778: if (isdense) {
9779: PetscCall(MatSeqDenseInvertFactors_Private(S));
9780: } else if (isdensecuda) {
9781: #if defined(PETSC_HAVE_CUDA)
9782: PetscCall(MatSeqDenseCUDAInvertFactors_Internal(S));
9783: #endif
9784: }
9785: // HIP??????????????
9786: PetscCall(PetscLogEventEnd(MAT_FactorInvS, F, 0, 0, 0));
9787: }
9788: PetscFunctionReturn(PETSC_SUCCESS);
9789: }
9791: /*@
9792: MatFactorInvertSchurComplement - Invert the Schur complement matrix computed during the factorization step
9794: Logically Collective
9796: Input Parameter:
9797: . F - the factored matrix obtained by calling `MatGetFactor()`
9799: Level: advanced
9801: Notes:
9802: Must be called after `MatFactorSetSchurIS()`.
9804: Call `MatFactorGetSchurComplement()` or `MatFactorCreateSchurComplement()` AFTER this call to actually compute the inverse and get access to it.
9806: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorGetSchurComplement()`, `MatFactorCreateSchurComplement()`
9807: @*/
9808: PetscErrorCode MatFactorInvertSchurComplement(Mat F)
9809: {
9810: PetscFunctionBegin;
9813: if (F->schur_status == MAT_FACTOR_SCHUR_INVERTED) PetscFunctionReturn(PETSC_SUCCESS);
9814: PetscCall(MatFactorFactorizeSchurComplement(F));
9815: PetscCall(MatFactorInvertSchurComplement_Private(F));
9816: F->schur_status = MAT_FACTOR_SCHUR_INVERTED;
9817: PetscFunctionReturn(PETSC_SUCCESS);
9818: }
9820: /*@
9821: MatFactorFactorizeSchurComplement - Factorize the Schur complement matrix computed during the factorization step
9823: Logically Collective
9825: Input Parameter:
9826: . F - the factored matrix obtained by calling `MatGetFactor()`
9828: Level: advanced
9830: Note:
9831: Must be called after `MatFactorSetSchurIS()`
9833: .seealso: [](ch_matrices), `Mat`, `MatGetFactor()`, `MatFactorSetSchurIS()`, `MatFactorInvertSchurComplement()`
9834: @*/
9835: PetscErrorCode MatFactorFactorizeSchurComplement(Mat F)
9836: {
9837: MatFactorInfo info;
9839: PetscFunctionBegin;
9842: if (F->schur_status == MAT_FACTOR_SCHUR_INVERTED || F->schur_status == MAT_FACTOR_SCHUR_FACTORED) PetscFunctionReturn(PETSC_SUCCESS);
9843: PetscCall(PetscLogEventBegin(MAT_FactorFactS, F, 0, 0, 0));
9844: PetscCall(PetscMemzero(&info, sizeof(MatFactorInfo)));
9845: if (F->factortype == MAT_FACTOR_CHOLESKY) { /* LDL^t regarded as Cholesky */
9846: PetscCall(MatCholeskyFactor(F->schur, NULL, &info));
9847: } else {
9848: PetscCall(MatLUFactor(F->schur, NULL, NULL, &info));
9849: }
9850: PetscCall(PetscLogEventEnd(MAT_FactorFactS, F, 0, 0, 0));
9851: F->schur_status = MAT_FACTOR_SCHUR_FACTORED;
9852: PetscFunctionReturn(PETSC_SUCCESS);
9853: }
9855: /*@
9856: MatPtAP - Creates the matrix product C = P^T * A * P
9858: Neighbor-wise Collective
9860: Input Parameters:
9861: + A - the matrix
9862: . P - the projection matrix
9863: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
9864: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(P)), use `PETSC_DEFAULT` if you do not have a good estimate
9865: if the result is a dense matrix this is irrelevant
9867: Output Parameter:
9868: . C - the product matrix
9870: Level: intermediate
9872: Notes:
9873: C will be created and must be destroyed by the user with `MatDestroy()`.
9875: An alternative approach to this function is to use `MatProductCreate()` and set the desired options before the computation is done
9877: Developer Note:
9878: For matrix types without special implementation the function fallbacks to `MatMatMult()` followed by `MatTransposeMatMult()`.
9880: .seealso: [](ch_matrices), `Mat`, `MatProductCreate()`, `MatMatMult()`, `MatRARt()`
9881: @*/
9882: PetscErrorCode MatPtAP(Mat A, Mat P, MatReuse scall, PetscReal fill, Mat *C)
9883: {
9884: PetscFunctionBegin;
9885: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*C, 5);
9886: PetscCheck(scall != MAT_INPLACE_MATRIX, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Inplace product not supported");
9888: if (scall == MAT_INITIAL_MATRIX) {
9889: PetscCall(MatProductCreate(A, P, NULL, C));
9890: PetscCall(MatProductSetType(*C, MATPRODUCT_PtAP));
9891: PetscCall(MatProductSetAlgorithm(*C, "default"));
9892: PetscCall(MatProductSetFill(*C, fill));
9894: (*C)->product->api_user = PETSC_TRUE;
9895: PetscCall(MatProductSetFromOptions(*C));
9896: PetscCheck((*C)->ops->productsymbolic, PetscObjectComm((PetscObject)(*C)), PETSC_ERR_SUP, "MatProduct %s not supported for A %s and P %s", MatProductTypes[MATPRODUCT_PtAP], ((PetscObject)A)->type_name, ((PetscObject)P)->type_name);
9897: PetscCall(MatProductSymbolic(*C));
9898: } else { /* scall == MAT_REUSE_MATRIX */
9899: PetscCall(MatProductReplaceMats(A, P, NULL, *C));
9900: }
9902: PetscCall(MatProductNumeric(*C));
9903: (*C)->symmetric = A->symmetric;
9904: (*C)->spd = A->spd;
9905: PetscFunctionReturn(PETSC_SUCCESS);
9906: }
9908: /*@
9909: MatRARt - Creates the matrix product C = R * A * R^T
9911: Neighbor-wise Collective
9913: Input Parameters:
9914: + A - the matrix
9915: . R - the projection matrix
9916: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
9917: - fill - expected fill as ratio of nnz(C)/nnz(A), use `PETSC_DEFAULT` if you do not have a good estimate
9918: if the result is a dense matrix this is irrelevant
9920: Output Parameter:
9921: . C - the product matrix
9923: Level: intermediate
9925: Notes:
9926: C will be created and must be destroyed by the user with `MatDestroy()`.
9928: An alternative approach to this function is to use `MatProductCreate()` and set the desired options before the computation is done
9930: This routine is currently only implemented for pairs of `MATAIJ` matrices and classes
9931: which inherit from `MATAIJ`. Due to PETSc sparse matrix block row distribution among processes,
9932: parallel MatRARt is implemented via explicit transpose of R, which could be very expensive.
9933: We recommend using MatPtAP().
9935: .seealso: [](ch_matrices), `Mat`, `MatProductCreate()`, `MatMatMult()`, `MatPtAP()`
9936: @*/
9937: PetscErrorCode MatRARt(Mat A, Mat R, MatReuse scall, PetscReal fill, Mat *C)
9938: {
9939: PetscFunctionBegin;
9940: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*C, 5);
9941: PetscCheck(scall != MAT_INPLACE_MATRIX, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Inplace product not supported");
9943: if (scall == MAT_INITIAL_MATRIX) {
9944: PetscCall(MatProductCreate(A, R, NULL, C));
9945: PetscCall(MatProductSetType(*C, MATPRODUCT_RARt));
9946: PetscCall(MatProductSetAlgorithm(*C, "default"));
9947: PetscCall(MatProductSetFill(*C, fill));
9949: (*C)->product->api_user = PETSC_TRUE;
9950: PetscCall(MatProductSetFromOptions(*C));
9951: PetscCheck((*C)->ops->productsymbolic, PetscObjectComm((PetscObject)(*C)), PETSC_ERR_SUP, "MatProduct %s not supported for A %s and R %s", MatProductTypes[MATPRODUCT_RARt], ((PetscObject)A)->type_name, ((PetscObject)R)->type_name);
9952: PetscCall(MatProductSymbolic(*C));
9953: } else { /* scall == MAT_REUSE_MATRIX */
9954: PetscCall(MatProductReplaceMats(A, R, NULL, *C));
9955: }
9957: PetscCall(MatProductNumeric(*C));
9958: if (A->symmetric == PETSC_BOOL3_TRUE) PetscCall(MatSetOption(*C, MAT_SYMMETRIC, PETSC_TRUE));
9959: PetscFunctionReturn(PETSC_SUCCESS);
9960: }
9962: static PetscErrorCode MatProduct_Private(Mat A, Mat B, MatReuse scall, PetscReal fill, MatProductType ptype, Mat *C)
9963: {
9964: PetscFunctionBegin;
9965: PetscCheck(scall != MAT_INPLACE_MATRIX, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Inplace product not supported");
9967: if (scall == MAT_INITIAL_MATRIX) {
9968: PetscCall(PetscInfo(A, "Calling MatProduct API with MAT_INITIAL_MATRIX and product type %s\n", MatProductTypes[ptype]));
9969: PetscCall(MatProductCreate(A, B, NULL, C));
9970: PetscCall(MatProductSetType(*C, ptype));
9971: PetscCall(MatProductSetAlgorithm(*C, MATPRODUCTALGORITHMDEFAULT));
9972: PetscCall(MatProductSetFill(*C, fill));
9974: (*C)->product->api_user = PETSC_TRUE;
9975: PetscCall(MatProductSetFromOptions(*C));
9976: PetscCall(MatProductSymbolic(*C));
9977: } else { /* scall == MAT_REUSE_MATRIX */
9978: Mat_Product *product = (*C)->product;
9979: PetscBool isdense;
9981: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)(*C), &isdense, MATSEQDENSE, MATMPIDENSE, ""));
9982: if (isdense && product && product->type != ptype) {
9983: PetscCall(MatProductClear(*C));
9984: product = NULL;
9985: }
9986: PetscCall(PetscInfo(A, "Calling MatProduct API with MAT_REUSE_MATRIX %s product present and product type %s\n", product ? "with" : "without", MatProductTypes[ptype]));
9987: if (!product) { /* user provide the dense matrix *C without calling MatProductCreate() or reusing it from previous calls */
9988: PetscCheck(isdense, PetscObjectComm((PetscObject)(*C)), PETSC_ERR_SUP, "Call MatProductCreate() first");
9989: PetscCall(MatProductCreate_Private(A, B, NULL, *C));
9990: product = (*C)->product;
9991: product->fill = fill;
9992: product->api_user = PETSC_TRUE;
9993: product->clear = PETSC_TRUE;
9995: PetscCall(MatProductSetType(*C, ptype));
9996: PetscCall(MatProductSetFromOptions(*C));
9997: PetscCheck((*C)->ops->productsymbolic, PetscObjectComm((PetscObject)(*C)), PETSC_ERR_SUP, "MatProduct %s not supported for %s and %s", MatProductTypes[ptype], ((PetscObject)A)->type_name, ((PetscObject)B)->type_name);
9998: PetscCall(MatProductSymbolic(*C));
9999: } else { /* user may change input matrices A or B when REUSE */
10000: PetscCall(MatProductReplaceMats(A, B, NULL, *C));
10001: }
10002: }
10003: PetscCall(MatProductNumeric(*C));
10004: PetscFunctionReturn(PETSC_SUCCESS);
10005: }
10007: /*@
10008: MatMatMult - Performs matrix-matrix multiplication C=A*B.
10010: Neighbor-wise Collective
10012: Input Parameters:
10013: + A - the left matrix
10014: . B - the right matrix
10015: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10016: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use `PETSC_DEFAULT` if you do not have a good estimate
10017: if the result is a dense matrix this is irrelevant
10019: Output Parameter:
10020: . C - the product matrix
10022: Notes:
10023: Unless scall is `MAT_REUSE_MATRIX` C will be created.
10025: `MAT_REUSE_MATRIX` can only be used if the matrices A and B have the same nonzero pattern as in the previous call and C was obtained from a previous
10026: call to this function with `MAT_INITIAL_MATRIX`.
10028: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value actually needed.
10030: In the special case where matrix B (and hence C) are dense you can create the correctly sized matrix C yourself and then call this routine with `MAT_REUSE_MATRIX`,
10031: rather than first having `MatMatMult()` create it for you. You can NEVER do this if the matrix C is sparse.
10033: Example of Usage:
10034: .vb
10035: MatProductCreate(A,B,NULL,&C);
10036: MatProductSetType(C,MATPRODUCT_AB);
10037: MatProductSymbolic(C);
10038: MatProductNumeric(C); // compute C=A * B
10039: MatProductReplaceMats(A1,B1,NULL,C); // compute C=A1 * B1
10040: MatProductNumeric(C);
10041: MatProductReplaceMats(A2,NULL,NULL,C); // compute C=A2 * B1
10042: MatProductNumeric(C);
10043: .ve
10045: Level: intermediate
10047: .seealso: [](ch_matrices), `Mat`, `MatProductType`, `MATPRODUCT_AB`, `MatTransposeMatMult()`, `MatMatTransposeMult()`, `MatPtAP()`, `MatProductCreate()`, `MatProductSymbolic()`, `MatProductReplaceMats()`, `MatProductNumeric()`
10048: @*/
10049: PetscErrorCode MatMatMult(Mat A, Mat B, MatReuse scall, PetscReal fill, Mat *C)
10050: {
10051: PetscFunctionBegin;
10052: PetscCall(MatProduct_Private(A, B, scall, fill, MATPRODUCT_AB, C));
10053: PetscFunctionReturn(PETSC_SUCCESS);
10054: }
10056: /*@
10057: MatMatTransposeMult - Performs matrix-matrix multiplication C=A*B^T.
10059: Neighbor-wise Collective
10061: Input Parameters:
10062: + A - the left matrix
10063: . B - the right matrix
10064: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10065: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use `PETSC_DEFAULT` if not known
10067: Output Parameter:
10068: . C - the product matrix
10070: Level: intermediate
10072: Notes:
10073: C will be created if `MAT_INITIAL_MATRIX` and must be destroyed by the user with `MatDestroy()`.
10075: `MAT_REUSE_MATRIX` can only be used if the matrices A and B have the same nonzero pattern as in the previous call
10077: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
10078: actually needed.
10080: This routine is currently only implemented for pairs of `MATSEQAIJ` matrices, for the `MATSEQDENSE` class,
10081: and for pairs of `MATMPIDENSE` matrices.
10083: This routine is shorthand for using `MatProductCreate()` with the `MatProductType` of `MATPRODUCT_ABt`
10085: Options Database Keys:
10086: . -matmattransmult_mpidense_mpidense_via {allgatherv,cyclic} - Choose between algorithms for `MATMPIDENSE` matrices: the
10087: first redundantly copies the transposed B matrix on each process and requires O(log P) communication complexity;
10088: the second never stores more than one portion of the B matrix at a time by requires O(P) communication complexity.
10090: .seealso: [](ch_matrices), `Mat`, `MatProductCreate()`, `MATPRODUCT_ABt`, `MatMatMult()`, `MatTransposeMatMult()` `MatPtAP()`, `MatProductCreate()`, `MatProductAlgorithm`, `MatProductType`, `MATPRODUCT_ABt`
10091: @*/
10092: PetscErrorCode MatMatTransposeMult(Mat A, Mat B, MatReuse scall, PetscReal fill, Mat *C)
10093: {
10094: PetscFunctionBegin;
10095: PetscCall(MatProduct_Private(A, B, scall, fill, MATPRODUCT_ABt, C));
10096: if (A == B) PetscCall(MatSetOption(*C, MAT_SYMMETRIC, PETSC_TRUE));
10097: PetscFunctionReturn(PETSC_SUCCESS);
10098: }
10100: /*@
10101: MatTransposeMatMult - Performs matrix-matrix multiplication C=A^T*B.
10103: Neighbor-wise Collective
10105: Input Parameters:
10106: + A - the left matrix
10107: . B - the right matrix
10108: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10109: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use `PETSC_DEFAULT` if not known
10111: Output Parameter:
10112: . C - the product matrix
10114: Level: intermediate
10116: Notes:
10117: C will be created if `MAT_INITIAL_MATRIX` and must be destroyed by the user with `MatDestroy()`.
10119: `MAT_REUSE_MATRIX` can only be used if the matrices A and B have the same nonzero pattern as in the previous call.
10121: This routine is shorthand for using `MatProductCreate()` with the `MatProductType` of `MATPRODUCT_AtB`
10123: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
10124: actually needed.
10126: This routine is currently implemented for pairs of `MATAIJ` matrices and pairs of `MATSEQDENSE` matrices and classes
10127: which inherit from `MATSEQAIJ`. C will be of the same type as the input matrices.
10129: .seealso: [](ch_matrices), `Mat`, `MatProductCreate()`, `MATPRODUCT_AtB`, `MatMatMult()`, `MatMatTransposeMult()`, `MatPtAP()`
10130: @*/
10131: PetscErrorCode MatTransposeMatMult(Mat A, Mat B, MatReuse scall, PetscReal fill, Mat *C)
10132: {
10133: PetscFunctionBegin;
10134: PetscCall(MatProduct_Private(A, B, scall, fill, MATPRODUCT_AtB, C));
10135: PetscFunctionReturn(PETSC_SUCCESS);
10136: }
10138: /*@
10139: MatMatMatMult - Performs matrix-matrix-matrix multiplication D=A*B*C.
10141: Neighbor-wise Collective
10143: Input Parameters:
10144: + A - the left matrix
10145: . B - the middle matrix
10146: . C - the right matrix
10147: . scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10148: - fill - expected fill as ratio of nnz(D)/(nnz(A) + nnz(B)+nnz(C)), use `PETSC_DEFAULT` if you do not have a good estimate
10149: if the result is a dense matrix this is irrelevant
10151: Output Parameter:
10152: . D - the product matrix
10154: Level: intermediate
10156: Notes:
10157: Unless scall is `MAT_REUSE_MATRIX` D will be created.
10159: `MAT_REUSE_MATRIX` can only be used if the matrices A, B and C have the same nonzero pattern as in the previous call
10161: This routine is shorthand for using `MatProductCreate()` with the `MatProductType` of `MATPRODUCT_ABC`
10163: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
10164: actually needed.
10166: If you have many matrices with the same non-zero structure to multiply, you
10167: should use `MAT_REUSE_MATRIX` in all calls but the first
10169: .seealso: [](ch_matrices), `Mat`, `MatProductCreate()`, `MATPRODUCT_ABC`, `MatMatMult`, `MatPtAP()`, `MatMatTransposeMult()`, `MatTransposeMatMult()`
10170: @*/
10171: PetscErrorCode MatMatMatMult(Mat A, Mat B, Mat C, MatReuse scall, PetscReal fill, Mat *D)
10172: {
10173: PetscFunctionBegin;
10174: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*D, 6);
10175: PetscCheck(scall != MAT_INPLACE_MATRIX, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Inplace product not supported");
10177: if (scall == MAT_INITIAL_MATRIX) {
10178: PetscCall(MatProductCreate(A, B, C, D));
10179: PetscCall(MatProductSetType(*D, MATPRODUCT_ABC));
10180: PetscCall(MatProductSetAlgorithm(*D, "default"));
10181: PetscCall(MatProductSetFill(*D, fill));
10183: (*D)->product->api_user = PETSC_TRUE;
10184: PetscCall(MatProductSetFromOptions(*D));
10185: PetscCheck((*D)->ops->productsymbolic, PetscObjectComm((PetscObject)(*D)), PETSC_ERR_SUP, "MatProduct %s not supported for A %s, B %s and C %s", MatProductTypes[MATPRODUCT_ABC], ((PetscObject)A)->type_name, ((PetscObject)B)->type_name,
10186: ((PetscObject)C)->type_name);
10187: PetscCall(MatProductSymbolic(*D));
10188: } else { /* user may change input matrices when REUSE */
10189: PetscCall(MatProductReplaceMats(A, B, C, *D));
10190: }
10191: PetscCall(MatProductNumeric(*D));
10192: PetscFunctionReturn(PETSC_SUCCESS);
10193: }
10195: /*@
10196: MatCreateRedundantMatrix - Create redundant matrices and put them into processors of subcommunicators.
10198: Collective
10200: Input Parameters:
10201: + mat - the matrix
10202: . nsubcomm - the number of subcommunicators (= number of redundant parallel or sequential matrices)
10203: . subcomm - MPI communicator split from the communicator where mat resides in (or `MPI_COMM_NULL` if nsubcomm is used)
10204: - reuse - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10206: Output Parameter:
10207: . matredundant - redundant matrix
10209: Level: advanced
10211: Notes:
10212: `MAT_REUSE_MATRIX` can only be used when the nonzero structure of the
10213: original matrix has not changed from that last call to MatCreateRedundantMatrix().
10215: This routine creates the duplicated matrices in the subcommunicators; you should NOT create them before
10216: calling it.
10218: `PetscSubcommCreate()` can be used to manage the creation of the subcomm but need not be.
10220: .seealso: [](ch_matrices), `Mat`, `MatDestroy()`, `PetscSubcommCreate()`, `PetscSubcomm`
10221: @*/
10222: PetscErrorCode MatCreateRedundantMatrix(Mat mat, PetscInt nsubcomm, MPI_Comm subcomm, MatReuse reuse, Mat *matredundant)
10223: {
10224: MPI_Comm comm;
10225: PetscMPIInt size;
10226: PetscInt mloc_sub, nloc_sub, rstart, rend, M = mat->rmap->N, N = mat->cmap->N, bs = mat->rmap->bs;
10227: Mat_Redundant *redund = NULL;
10228: PetscSubcomm psubcomm = NULL;
10229: MPI_Comm subcomm_in = subcomm;
10230: Mat *matseq;
10231: IS isrow, iscol;
10232: PetscBool newsubcomm = PETSC_FALSE;
10234: PetscFunctionBegin;
10236: if (nsubcomm && reuse == MAT_REUSE_MATRIX) {
10239: }
10241: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size));
10242: if (size == 1 || nsubcomm == 1) {
10243: if (reuse == MAT_INITIAL_MATRIX) {
10244: PetscCall(MatDuplicate(mat, MAT_COPY_VALUES, matredundant));
10245: } else {
10246: PetscCheck(*matredundant != mat, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10247: PetscCall(MatCopy(mat, *matredundant, SAME_NONZERO_PATTERN));
10248: }
10249: PetscFunctionReturn(PETSC_SUCCESS);
10250: }
10252: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
10253: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
10254: MatCheckPreallocated(mat, 1);
10256: PetscCall(PetscLogEventBegin(MAT_RedundantMat, mat, 0, 0, 0));
10257: if (subcomm_in == MPI_COMM_NULL && reuse == MAT_INITIAL_MATRIX) { /* get subcomm if user does not provide subcomm */
10258: /* create psubcomm, then get subcomm */
10259: PetscCall(PetscObjectGetComm((PetscObject)mat, &comm));
10260: PetscCallMPI(MPI_Comm_size(comm, &size));
10261: PetscCheck(nsubcomm >= 1 && nsubcomm <= size, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "nsubcomm must between 1 and %d", size);
10263: PetscCall(PetscSubcommCreate(comm, &psubcomm));
10264: PetscCall(PetscSubcommSetNumber(psubcomm, nsubcomm));
10265: PetscCall(PetscSubcommSetType(psubcomm, PETSC_SUBCOMM_CONTIGUOUS));
10266: PetscCall(PetscSubcommSetFromOptions(psubcomm));
10267: PetscCall(PetscCommDuplicate(PetscSubcommChild(psubcomm), &subcomm, NULL));
10268: newsubcomm = PETSC_TRUE;
10269: PetscCall(PetscSubcommDestroy(&psubcomm));
10270: }
10272: /* get isrow, iscol and a local sequential matrix matseq[0] */
10273: if (reuse == MAT_INITIAL_MATRIX) {
10274: mloc_sub = PETSC_DECIDE;
10275: nloc_sub = PETSC_DECIDE;
10276: if (bs < 1) {
10277: PetscCall(PetscSplitOwnership(subcomm, &mloc_sub, &M));
10278: PetscCall(PetscSplitOwnership(subcomm, &nloc_sub, &N));
10279: } else {
10280: PetscCall(PetscSplitOwnershipBlock(subcomm, bs, &mloc_sub, &M));
10281: PetscCall(PetscSplitOwnershipBlock(subcomm, bs, &nloc_sub, &N));
10282: }
10283: PetscCallMPI(MPI_Scan(&mloc_sub, &rend, 1, MPIU_INT, MPI_SUM, subcomm));
10284: rstart = rend - mloc_sub;
10285: PetscCall(ISCreateStride(PETSC_COMM_SELF, mloc_sub, rstart, 1, &isrow));
10286: PetscCall(ISCreateStride(PETSC_COMM_SELF, N, 0, 1, &iscol));
10287: } else { /* reuse == MAT_REUSE_MATRIX */
10288: PetscCheck(*matredundant != mat, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10289: /* retrieve subcomm */
10290: PetscCall(PetscObjectGetComm((PetscObject)(*matredundant), &subcomm));
10291: redund = (*matredundant)->redundant;
10292: isrow = redund->isrow;
10293: iscol = redund->iscol;
10294: matseq = redund->matseq;
10295: }
10296: PetscCall(MatCreateSubMatrices(mat, 1, &isrow, &iscol, reuse, &matseq));
10298: /* get matredundant over subcomm */
10299: if (reuse == MAT_INITIAL_MATRIX) {
10300: PetscCall(MatCreateMPIMatConcatenateSeqMat(subcomm, matseq[0], nloc_sub, reuse, matredundant));
10302: /* create a supporting struct and attach it to C for reuse */
10303: PetscCall(PetscNew(&redund));
10304: (*matredundant)->redundant = redund;
10305: redund->isrow = isrow;
10306: redund->iscol = iscol;
10307: redund->matseq = matseq;
10308: if (newsubcomm) {
10309: redund->subcomm = subcomm;
10310: } else {
10311: redund->subcomm = MPI_COMM_NULL;
10312: }
10313: } else {
10314: PetscCall(MatCreateMPIMatConcatenateSeqMat(subcomm, matseq[0], PETSC_DECIDE, reuse, matredundant));
10315: }
10316: #if defined(PETSC_HAVE_VIENNACL) || defined(PETSC_HAVE_CUDA) || defined(PETSC_HAVE_HIP)
10317: if (matseq[0]->boundtocpu && matseq[0]->bindingpropagates) {
10318: PetscCall(MatBindToCPU(*matredundant, PETSC_TRUE));
10319: PetscCall(MatSetBindingPropagates(*matredundant, PETSC_TRUE));
10320: }
10321: #endif
10322: PetscCall(PetscLogEventEnd(MAT_RedundantMat, mat, 0, 0, 0));
10323: PetscFunctionReturn(PETSC_SUCCESS);
10324: }
10326: /*@C
10327: MatGetMultiProcBlock - Create multiple 'parallel submatrices' from
10328: a given `Mat`. Each submatrix can span multiple procs.
10330: Collective
10332: Input Parameters:
10333: + mat - the matrix
10334: . subcomm - the sub communicator obtained as if by `MPI_Comm_split(PetscObjectComm((PetscObject)mat))`
10335: - scall - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10337: Output Parameter:
10338: . subMat - parallel sub-matrices each spanning a given `subcomm`
10340: Level: advanced
10342: Notes:
10343: The submatrix partition across processors is dictated by `subComm` a
10344: communicator obtained by `MPI_comm_split()` or via `PetscSubcommCreate()`. The `subComm`
10345: is not restricted to be grouped with consecutive original ranks.
10347: Due the `MPI_Comm_split()` usage, the parallel layout of the submatrices
10348: map directly to the layout of the original matrix [wrt the local
10349: row,col partitioning]. So the original 'DiagonalMat' naturally maps
10350: into the 'DiagonalMat' of the `subMat`, hence it is used directly from
10351: the `subMat`. However the offDiagMat looses some columns - and this is
10352: reconstructed with `MatSetValues()`
10354: This is used by `PCBJACOBI` when a single block spans multiple MPI ranks
10356: .seealso: [](ch_matrices), `Mat`, `MatCreateRedundantMatrix()`, `MatCreateSubMatrices()`, `PCBJACOBI`
10357: @*/
10358: PetscErrorCode MatGetMultiProcBlock(Mat mat, MPI_Comm subComm, MatReuse scall, Mat *subMat)
10359: {
10360: PetscMPIInt commsize, subCommSize;
10362: PetscFunctionBegin;
10363: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &commsize));
10364: PetscCallMPI(MPI_Comm_size(subComm, &subCommSize));
10365: PetscCheck(subCommSize <= commsize, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_OUTOFRANGE, "CommSize %d < SubCommZize %d", commsize, subCommSize);
10367: PetscCheck(scall != MAT_REUSE_MATRIX || *subMat != mat, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10368: PetscCall(PetscLogEventBegin(MAT_GetMultiProcBlock, mat, 0, 0, 0));
10369: PetscUseTypeMethod(mat, getmultiprocblock, subComm, scall, subMat);
10370: PetscCall(PetscLogEventEnd(MAT_GetMultiProcBlock, mat, 0, 0, 0));
10371: PetscFunctionReturn(PETSC_SUCCESS);
10372: }
10374: /*@
10375: MatGetLocalSubMatrix - Gets a reference to a submatrix specified in local numbering
10377: Not Collective
10379: Input Parameters:
10380: + mat - matrix to extract local submatrix from
10381: . isrow - local row indices for submatrix
10382: - iscol - local column indices for submatrix
10384: Output Parameter:
10385: . submat - the submatrix
10387: Level: intermediate
10389: Notes:
10390: `submat` should be disposed of with `MatRestoreLocalSubMatrix()`.
10392: Depending on the format of `mat`, the returned submat may not implement `MatMult()`. Its communicator may be
10393: the same as mat, it may be `PETSC_COMM_SELF`, or some other subcomm of `mat`'s.
10395: `submat` always implements `MatSetValuesLocal()`. If `isrow` and `iscol` have the same block size, then
10396: `MatSetValuesBlockedLocal()` will also be implemented.
10398: `mat` must have had a `ISLocalToGlobalMapping` provided to it with `MatSetLocalToGlobalMapping()`.
10399: Matrices obtained with `DMCreateMatrix()` generally already have the local to global mapping provided.
10401: .seealso: [](ch_matrices), `Mat`, `MatRestoreLocalSubMatrix()`, `MatCreateLocalRef()`, `MatSetLocalToGlobalMapping()`
10402: @*/
10403: PetscErrorCode MatGetLocalSubMatrix(Mat mat, IS isrow, IS iscol, Mat *submat)
10404: {
10405: PetscFunctionBegin;
10409: PetscCheckSameComm(isrow, 2, iscol, 3);
10411: PetscCheck(mat->rmap->mapping, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Matrix must have local to global mapping provided before this call");
10413: if (mat->ops->getlocalsubmatrix) {
10414: PetscUseTypeMethod(mat, getlocalsubmatrix, isrow, iscol, submat);
10415: } else {
10416: PetscCall(MatCreateLocalRef(mat, isrow, iscol, submat));
10417: }
10418: PetscFunctionReturn(PETSC_SUCCESS);
10419: }
10421: /*@
10422: MatRestoreLocalSubMatrix - Restores a reference to a submatrix specified in local numbering obtained with `MatGetLocalSubMatrix()`
10424: Not Collective
10426: Input Parameters:
10427: + mat - matrix to extract local submatrix from
10428: . isrow - local row indices for submatrix
10429: . iscol - local column indices for submatrix
10430: - submat - the submatrix
10432: Level: intermediate
10434: .seealso: [](ch_matrices), `Mat`, `MatGetLocalSubMatrix()`
10435: @*/
10436: PetscErrorCode MatRestoreLocalSubMatrix(Mat mat, IS isrow, IS iscol, Mat *submat)
10437: {
10438: PetscFunctionBegin;
10442: PetscCheckSameComm(isrow, 2, iscol, 3);
10446: if (mat->ops->restorelocalsubmatrix) {
10447: PetscUseTypeMethod(mat, restorelocalsubmatrix, isrow, iscol, submat);
10448: } else {
10449: PetscCall(MatDestroy(submat));
10450: }
10451: *submat = NULL;
10452: PetscFunctionReturn(PETSC_SUCCESS);
10453: }
10455: /*@
10456: MatFindZeroDiagonals - Finds all the rows of a matrix that have zero or no diagonal entry in the matrix
10458: Collective
10460: Input Parameter:
10461: . mat - the matrix
10463: Output Parameter:
10464: . is - if any rows have zero diagonals this contains the list of them
10466: Level: developer
10468: .seealso: [](ch_matrices), `Mat`, `MatMultTranspose()`, `MatMultAdd()`, `MatMultTransposeAdd()`
10469: @*/
10470: PetscErrorCode MatFindZeroDiagonals(Mat mat, IS *is)
10471: {
10472: PetscFunctionBegin;
10475: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
10476: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
10478: if (!mat->ops->findzerodiagonals) {
10479: Vec diag;
10480: const PetscScalar *a;
10481: PetscInt *rows;
10482: PetscInt rStart, rEnd, r, nrow = 0;
10484: PetscCall(MatCreateVecs(mat, &diag, NULL));
10485: PetscCall(MatGetDiagonal(mat, diag));
10486: PetscCall(MatGetOwnershipRange(mat, &rStart, &rEnd));
10487: PetscCall(VecGetArrayRead(diag, &a));
10488: for (r = 0; r < rEnd - rStart; ++r)
10489: if (a[r] == 0.0) ++nrow;
10490: PetscCall(PetscMalloc1(nrow, &rows));
10491: nrow = 0;
10492: for (r = 0; r < rEnd - rStart; ++r)
10493: if (a[r] == 0.0) rows[nrow++] = r + rStart;
10494: PetscCall(VecRestoreArrayRead(diag, &a));
10495: PetscCall(VecDestroy(&diag));
10496: PetscCall(ISCreateGeneral(PetscObjectComm((PetscObject)mat), nrow, rows, PETSC_OWN_POINTER, is));
10497: } else {
10498: PetscUseTypeMethod(mat, findzerodiagonals, is);
10499: }
10500: PetscFunctionReturn(PETSC_SUCCESS);
10501: }
10503: /*@
10504: MatFindOffBlockDiagonalEntries - Finds all the rows of a matrix that have entries outside of the main diagonal block (defined by the matrix block size)
10506: Collective
10508: Input Parameter:
10509: . mat - the matrix
10511: Output Parameter:
10512: . is - contains the list of rows with off block diagonal entries
10514: Level: developer
10516: .seealso: [](ch_matrices), `Mat`, `MatMultTranspose()`, `MatMultAdd()`, `MatMultTransposeAdd()`
10517: @*/
10518: PetscErrorCode MatFindOffBlockDiagonalEntries(Mat mat, IS *is)
10519: {
10520: PetscFunctionBegin;
10523: PetscCheck(mat->assembled, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
10524: PetscCheck(!mat->factortype, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
10526: PetscUseTypeMethod(mat, findoffblockdiagonalentries, is);
10527: PetscFunctionReturn(PETSC_SUCCESS);
10528: }
10530: /*@C
10531: MatInvertBlockDiagonal - Inverts the block diagonal entries.
10533: Collective; No Fortran Support
10535: Input Parameter:
10536: . mat - the matrix
10538: Output Parameter:
10539: . values - the block inverses in column major order (FORTRAN-like)
10541: Level: advanced
10543: Notes:
10544: The size of the blocks is determined by the block size of the matrix.
10546: The blocks never overlap between two MPI ranks, use `MatInvertVariableBlockEnvelope()` for that case
10548: The blocks all have the same size, use `MatInvertVariableBlockDiagonal()` for variable block size
10550: .seealso: [](ch_matrices), `Mat`, `MatInvertVariableBlockEnvelope()`, `MatInvertBlockDiagonalMat()`
10551: @*/
10552: PetscErrorCode MatInvertBlockDiagonal(Mat mat, const PetscScalar **values)
10553: {
10554: PetscFunctionBegin;
10556: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
10557: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
10558: PetscUseTypeMethod(mat, invertblockdiagonal, values);
10559: PetscFunctionReturn(PETSC_SUCCESS);
10560: }
10562: /*@C
10563: MatInvertVariableBlockDiagonal - Inverts the point block diagonal entries.
10565: Collective; No Fortran Support
10567: Input Parameters:
10568: + mat - the matrix
10569: . nblocks - the number of blocks on the process, set with `MatSetVariableBlockSizes()`
10570: - bsizes - the size of each block on the process, set with `MatSetVariableBlockSizes()`
10572: Output Parameter:
10573: . values - the block inverses in column major order (FORTRAN-like)
10575: Level: advanced
10577: Notes:
10578: Use `MatInvertBlockDiagonal()` if all blocks have the same size
10580: The blocks never overlap between two MPI ranks, use `MatInvertVariableBlockEnvelope()` for that case
10582: .seealso: [](ch_matrices), `Mat`, `MatInvertBlockDiagonal()`, `MatSetVariableBlockSizes()`, `MatInvertVariableBlockEnvelope()`
10583: @*/
10584: PetscErrorCode MatInvertVariableBlockDiagonal(Mat mat, PetscInt nblocks, const PetscInt *bsizes, PetscScalar *values)
10585: {
10586: PetscFunctionBegin;
10588: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
10589: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
10590: PetscUseTypeMethod(mat, invertvariableblockdiagonal, nblocks, bsizes, values);
10591: PetscFunctionReturn(PETSC_SUCCESS);
10592: }
10594: /*@
10595: MatInvertBlockDiagonalMat - set the values of matrix C to be the inverted block diagonal of matrix A
10597: Collective
10599: Input Parameters:
10600: + A - the matrix
10601: - C - matrix with inverted block diagonal of `A`. This matrix should be created and may have its type set.
10603: Level: advanced
10605: Note:
10606: The blocksize of the matrix is used to determine the blocks on the diagonal of `C`
10608: .seealso: [](ch_matrices), `Mat`, `MatInvertBlockDiagonal()`
10609: @*/
10610: PetscErrorCode MatInvertBlockDiagonalMat(Mat A, Mat C)
10611: {
10612: const PetscScalar *vals;
10613: PetscInt *dnnz;
10614: PetscInt m, rstart, rend, bs, i, j;
10616: PetscFunctionBegin;
10617: PetscCall(MatInvertBlockDiagonal(A, &vals));
10618: PetscCall(MatGetBlockSize(A, &bs));
10619: PetscCall(MatGetLocalSize(A, &m, NULL));
10620: PetscCall(MatSetLayouts(C, A->rmap, A->cmap));
10621: PetscCall(PetscMalloc1(m / bs, &dnnz));
10622: for (j = 0; j < m / bs; j++) dnnz[j] = 1;
10623: PetscCall(MatXAIJSetPreallocation(C, bs, dnnz, NULL, NULL, NULL));
10624: PetscCall(PetscFree(dnnz));
10625: PetscCall(MatGetOwnershipRange(C, &rstart, &rend));
10626: PetscCall(MatSetOption(C, MAT_ROW_ORIENTED, PETSC_FALSE));
10627: for (i = rstart / bs; i < rend / bs; i++) PetscCall(MatSetValuesBlocked(C, 1, &i, 1, &i, &vals[(i - rstart / bs) * bs * bs], INSERT_VALUES));
10628: PetscCall(MatAssemblyBegin(C, MAT_FINAL_ASSEMBLY));
10629: PetscCall(MatAssemblyEnd(C, MAT_FINAL_ASSEMBLY));
10630: PetscCall(MatSetOption(C, MAT_ROW_ORIENTED, PETSC_TRUE));
10631: PetscFunctionReturn(PETSC_SUCCESS);
10632: }
10634: /*@C
10635: MatTransposeColoringDestroy - Destroys a coloring context for matrix product C=A*B^T that was created
10636: via `MatTransposeColoringCreate()`.
10638: Collective
10640: Input Parameter:
10641: . c - coloring context
10643: Level: intermediate
10645: .seealso: [](ch_matrices), `Mat`, `MatTransposeColoringCreate()`
10646: @*/
10647: PetscErrorCode MatTransposeColoringDestroy(MatTransposeColoring *c)
10648: {
10649: MatTransposeColoring matcolor = *c;
10651: PetscFunctionBegin;
10652: if (!matcolor) PetscFunctionReturn(PETSC_SUCCESS);
10653: if (--((PetscObject)matcolor)->refct > 0) {
10654: matcolor = NULL;
10655: PetscFunctionReturn(PETSC_SUCCESS);
10656: }
10658: PetscCall(PetscFree3(matcolor->ncolumns, matcolor->nrows, matcolor->colorforrow));
10659: PetscCall(PetscFree(matcolor->rows));
10660: PetscCall(PetscFree(matcolor->den2sp));
10661: PetscCall(PetscFree(matcolor->colorforcol));
10662: PetscCall(PetscFree(matcolor->columns));
10663: if (matcolor->brows > 0) PetscCall(PetscFree(matcolor->lstart));
10664: PetscCall(PetscHeaderDestroy(c));
10665: PetscFunctionReturn(PETSC_SUCCESS);
10666: }
10668: /*@C
10669: MatTransColoringApplySpToDen - Given a symbolic matrix product C=A*B^T for which
10670: a `MatTransposeColoring` context has been created, computes a dense B^T by applying
10671: `MatTransposeColoring` to sparse B.
10673: Collective
10675: Input Parameters:
10676: + coloring - coloring context created with `MatTransposeColoringCreate()`
10677: - B - sparse matrix
10679: Output Parameter:
10680: . Btdense - dense matrix B^T
10682: Level: developer
10684: Note:
10685: These are used internally for some implementations of `MatRARt()`
10687: .seealso: [](ch_matrices), `Mat`, `MatTransposeColoringCreate()`, `MatTransposeColoringDestroy()`, `MatTransColoringApplyDenToSp()`
10688: @*/
10689: PetscErrorCode MatTransColoringApplySpToDen(MatTransposeColoring coloring, Mat B, Mat Btdense)
10690: {
10691: PetscFunctionBegin;
10696: PetscCall((*B->ops->transcoloringapplysptoden)(coloring, B, Btdense));
10697: PetscFunctionReturn(PETSC_SUCCESS);
10698: }
10700: /*@C
10701: MatTransColoringApplyDenToSp - Given a symbolic matrix product Csp=A*B^T for which
10702: a `MatTransposeColoring` context has been created and a dense matrix Cden=A*Btdense
10703: in which Btdens is obtained from `MatTransColoringApplySpToDen()`, recover sparse matrix
10704: `Csp` from `Cden`.
10706: Collective
10708: Input Parameters:
10709: + matcoloring - coloring context created with `MatTransposeColoringCreate()`
10710: - Cden - matrix product of a sparse matrix and a dense matrix Btdense
10712: Output Parameter:
10713: . Csp - sparse matrix
10715: Level: developer
10717: Note:
10718: These are used internally for some implementations of `MatRARt()`
10720: .seealso: [](ch_matrices), `Mat`, `MatTransposeColoringCreate()`, `MatTransposeColoringDestroy()`, `MatTransColoringApplySpToDen()`
10721: @*/
10722: PetscErrorCode MatTransColoringApplyDenToSp(MatTransposeColoring matcoloring, Mat Cden, Mat Csp)
10723: {
10724: PetscFunctionBegin;
10729: PetscCall((*Csp->ops->transcoloringapplydentosp)(matcoloring, Cden, Csp));
10730: PetscCall(MatAssemblyBegin(Csp, MAT_FINAL_ASSEMBLY));
10731: PetscCall(MatAssemblyEnd(Csp, MAT_FINAL_ASSEMBLY));
10732: PetscFunctionReturn(PETSC_SUCCESS);
10733: }
10735: /*@C
10736: MatTransposeColoringCreate - Creates a matrix coloring context for the matrix product C=A*B^T.
10738: Collective
10740: Input Parameters:
10741: + mat - the matrix product C
10742: - iscoloring - the coloring of the matrix; usually obtained with `MatColoringCreate()` or `DMCreateColoring()`
10744: Output Parameter:
10745: . color - the new coloring context
10747: Level: intermediate
10749: .seealso: [](ch_matrices), `Mat`, `MatTransposeColoringDestroy()`, `MatTransColoringApplySpToDen()`,
10750: `MatTransColoringApplyDenToSp()`
10751: @*/
10752: PetscErrorCode MatTransposeColoringCreate(Mat mat, ISColoring iscoloring, MatTransposeColoring *color)
10753: {
10754: MatTransposeColoring c;
10755: MPI_Comm comm;
10757: PetscFunctionBegin;
10758: PetscCall(PetscLogEventBegin(MAT_TransposeColoringCreate, mat, 0, 0, 0));
10759: PetscCall(PetscObjectGetComm((PetscObject)mat, &comm));
10760: PetscCall(PetscHeaderCreate(c, MAT_TRANSPOSECOLORING_CLASSID, "MatTransposeColoring", "Matrix product C=A*B^T via coloring", "Mat", comm, MatTransposeColoringDestroy, NULL));
10762: c->ctype = iscoloring->ctype;
10763: PetscUseTypeMethod(mat, transposecoloringcreate, iscoloring, c);
10765: *color = c;
10766: PetscCall(PetscLogEventEnd(MAT_TransposeColoringCreate, mat, 0, 0, 0));
10767: PetscFunctionReturn(PETSC_SUCCESS);
10768: }
10770: /*@
10771: MatGetNonzeroState - Returns a 64-bit integer representing the current state of nonzeros in the matrix. If the
10772: matrix has had no new nonzero locations added to (or removed from) the matrix since the previous call then the value will be the
10773: same, otherwise it will be larger
10775: Not Collective
10777: Input Parameter:
10778: . A - the matrix
10780: Output Parameter:
10781: . state - the current state
10783: Level: intermediate
10785: Notes:
10786: You can only compare states from two different calls to the SAME matrix, you cannot compare calls between
10787: different matrices
10789: Use `PetscObjectStateGet()` to check for changes to the numerical values in a matrix
10791: Use the result of `PetscObjectGetId()` to compare if a previously checked matrix is the same as the current matrix, do not compare object pointers.
10793: .seealso: [](ch_matrices), `Mat`, `PetscObjectStateGet()`, `PetscObjectGetId()`
10794: @*/
10795: PetscErrorCode MatGetNonzeroState(Mat mat, PetscObjectState *state)
10796: {
10797: PetscFunctionBegin;
10799: *state = mat->nonzerostate;
10800: PetscFunctionReturn(PETSC_SUCCESS);
10801: }
10803: /*@
10804: MatCreateMPIMatConcatenateSeqMat - Creates a single large PETSc matrix by concatenating sequential
10805: matrices from each processor
10807: Collective
10809: Input Parameters:
10810: + comm - the communicators the parallel matrix will live on
10811: . seqmat - the input sequential matrices
10812: . n - number of local columns (or `PETSC_DECIDE`)
10813: - reuse - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10815: Output Parameter:
10816: . mpimat - the parallel matrix generated
10818: Level: developer
10820: Note:
10821: The number of columns of the matrix in EACH processor MUST be the same.
10823: .seealso: [](ch_matrices), `Mat`
10824: @*/
10825: PetscErrorCode MatCreateMPIMatConcatenateSeqMat(MPI_Comm comm, Mat seqmat, PetscInt n, MatReuse reuse, Mat *mpimat)
10826: {
10827: PetscMPIInt size;
10829: PetscFunctionBegin;
10830: PetscCallMPI(MPI_Comm_size(comm, &size));
10831: if (size == 1) {
10832: if (reuse == MAT_INITIAL_MATRIX) {
10833: PetscCall(MatDuplicate(seqmat, MAT_COPY_VALUES, mpimat));
10834: } else {
10835: PetscCall(MatCopy(seqmat, *mpimat, SAME_NONZERO_PATTERN));
10836: }
10837: PetscFunctionReturn(PETSC_SUCCESS);
10838: }
10840: PetscCheck(reuse != MAT_REUSE_MATRIX || seqmat != *mpimat, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10842: PetscCall(PetscLogEventBegin(MAT_Merge, seqmat, 0, 0, 0));
10843: PetscCall((*seqmat->ops->creatempimatconcatenateseqmat)(comm, seqmat, n, reuse, mpimat));
10844: PetscCall(PetscLogEventEnd(MAT_Merge, seqmat, 0, 0, 0));
10845: PetscFunctionReturn(PETSC_SUCCESS);
10846: }
10848: /*@
10849: MatSubdomainsCreateCoalesce - Creates index subdomains by coalescing adjacent ranks' ownership ranges.
10851: Collective
10853: Input Parameters:
10854: + A - the matrix to create subdomains from
10855: - N - requested number of subdomains
10857: Output Parameters:
10858: + n - number of subdomains resulting on this rank
10859: - iss - `IS` list with indices of subdomains on this rank
10861: Level: advanced
10863: Note:
10864: The number of subdomains must be smaller than the communicator size
10866: .seealso: [](ch_matrices), `Mat`, `IS`
10867: @*/
10868: PetscErrorCode MatSubdomainsCreateCoalesce(Mat A, PetscInt N, PetscInt *n, IS *iss[])
10869: {
10870: MPI_Comm comm, subcomm;
10871: PetscMPIInt size, rank, color;
10872: PetscInt rstart, rend, k;
10874: PetscFunctionBegin;
10875: PetscCall(PetscObjectGetComm((PetscObject)A, &comm));
10876: PetscCallMPI(MPI_Comm_size(comm, &size));
10877: PetscCallMPI(MPI_Comm_rank(comm, &rank));
10878: PetscCheck(N >= 1 && N < (PetscInt)size, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "number of subdomains must be > 0 and < %d, got N = %" PetscInt_FMT, size, N);
10879: *n = 1;
10880: k = ((PetscInt)size) / N + ((PetscInt)size % N > 0); /* There are up to k ranks to a color */
10881: color = rank / k;
10882: PetscCallMPI(MPI_Comm_split(comm, color, rank, &subcomm));
10883: PetscCall(PetscMalloc1(1, iss));
10884: PetscCall(MatGetOwnershipRange(A, &rstart, &rend));
10885: PetscCall(ISCreateStride(subcomm, rend - rstart, rstart, 1, iss[0]));
10886: PetscCallMPI(MPI_Comm_free(&subcomm));
10887: PetscFunctionReturn(PETSC_SUCCESS);
10888: }
10890: /*@
10891: MatGalerkin - Constructs the coarse grid problem matrix via Galerkin projection.
10893: If the interpolation and restriction operators are the same, uses `MatPtAP()`.
10894: If they are not the same, uses `MatMatMatMult()`.
10896: Once the coarse grid problem is constructed, correct for interpolation operators
10897: that are not of full rank, which can legitimately happen in the case of non-nested
10898: geometric multigrid.
10900: Input Parameters:
10901: + restrct - restriction operator
10902: . dA - fine grid matrix
10903: . interpolate - interpolation operator
10904: . reuse - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
10905: - fill - expected fill, use `PETSC_DEFAULT` if you do not have a good estimate
10907: Output Parameter:
10908: . A - the Galerkin coarse matrix
10910: Options Database Key:
10911: . -pc_mg_galerkin <both,pmat,mat,none> - for what matrices the Galerkin process should be used
10913: Level: developer
10915: .seealso: [](ch_matrices), `Mat`, `MatPtAP()`, `MatMatMatMult()`
10916: @*/
10917: PetscErrorCode MatGalerkin(Mat restrct, Mat dA, Mat interpolate, MatReuse reuse, PetscReal fill, Mat *A)
10918: {
10919: IS zerorows;
10920: Vec diag;
10922: PetscFunctionBegin;
10923: PetscCheck(reuse != MAT_INPLACE_MATRIX, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Inplace product not supported");
10924: /* Construct the coarse grid matrix */
10925: if (interpolate == restrct) {
10926: PetscCall(MatPtAP(dA, interpolate, reuse, fill, A));
10927: } else {
10928: PetscCall(MatMatMatMult(restrct, dA, interpolate, reuse, fill, A));
10929: }
10931: /* If the interpolation matrix is not of full rank, A will have zero rows.
10932: This can legitimately happen in the case of non-nested geometric multigrid.
10933: In that event, we set the rows of the matrix to the rows of the identity,
10934: ignoring the equations (as the RHS will also be zero). */
10936: PetscCall(MatFindZeroRows(*A, &zerorows));
10938: if (zerorows != NULL) { /* if there are any zero rows */
10939: PetscCall(MatCreateVecs(*A, &diag, NULL));
10940: PetscCall(MatGetDiagonal(*A, diag));
10941: PetscCall(VecISSet(diag, zerorows, 1.0));
10942: PetscCall(MatDiagonalSet(*A, diag, INSERT_VALUES));
10943: PetscCall(VecDestroy(&diag));
10944: PetscCall(ISDestroy(&zerorows));
10945: }
10946: PetscFunctionReturn(PETSC_SUCCESS);
10947: }
10949: /*@C
10950: MatSetOperation - Allows user to set a matrix operation for any matrix type
10952: Logically Collective
10954: Input Parameters:
10955: + mat - the matrix
10956: . op - the name of the operation
10957: - f - the function that provides the operation
10959: Level: developer
10961: Usage:
10962: .vb
10963: extern PetscErrorCode usermult(Mat, Vec, Vec);
10965: PetscCall(MatCreateXXX(comm, ..., &A));
10966: PetscCall(MatSetOperation(A, MATOP_MULT, (PetscVoidFunction)usermult));
10967: .ve
10969: Notes:
10970: See the file `include/petscmat.h` for a complete list of matrix
10971: operations, which all have the form MATOP_<OPERATION>, where
10972: <OPERATION> is the name (in all capital letters) of the
10973: user interface routine (e.g., `MatMult()` -> `MATOP_MULT`).
10975: All user-provided functions (except for `MATOP_DESTROY`) should have the same calling
10976: sequence as the usual matrix interface routines, since they
10977: are intended to be accessed via the usual matrix interface
10978: routines, e.g.,
10979: .vb
10980: MatMult(Mat, Vec, Vec) -> usermult(Mat, Vec, Vec)
10981: .ve
10983: In particular each function MUST return `PETSC_SUCCESS` on success and
10984: nonzero on failure.
10986: This routine is distinct from `MatShellSetOperation()` in that it can be called on any matrix type.
10988: .seealso: [](ch_matrices), `Mat`, `MatGetOperation()`, `MatCreateShell()`, `MatShellSetContext()`, `MatShellSetOperation()`
10989: @*/
10990: PetscErrorCode MatSetOperation(Mat mat, MatOperation op, void (*f)(void))
10991: {
10992: PetscFunctionBegin;
10994: if (op == MATOP_VIEW && !mat->ops->viewnative && f != (void (*)(void))(mat->ops->view)) mat->ops->viewnative = mat->ops->view;
10995: (((void (**)(void))mat->ops)[op]) = f;
10996: PetscFunctionReturn(PETSC_SUCCESS);
10997: }
10999: /*@C
11000: MatGetOperation - Gets a matrix operation for any matrix type.
11002: Not Collective
11004: Input Parameters:
11005: + mat - the matrix
11006: - op - the name of the operation
11008: Output Parameter:
11009: . f - the function that provides the operation
11011: Level: developer
11013: Usage:
11014: .vb
11015: PetscErrorCode (*usermult)(Mat, Vec, Vec);
11016: MatGetOperation(A, MATOP_MULT, (void (**)(void))&usermult);
11017: .ve
11019: Notes:
11020: See the file include/petscmat.h for a complete list of matrix
11021: operations, which all have the form MATOP_<OPERATION>, where
11022: <OPERATION> is the name (in all capital letters) of the
11023: user interface routine (e.g., `MatMult()` -> `MATOP_MULT`).
11025: This routine is distinct from `MatShellGetOperation()` in that it can be called on any matrix type.
11027: .seealso: [](ch_matrices), `Mat`, `MatSetOperation()`, `MatCreateShell()`, `MatShellGetContext()`, `MatShellGetOperation()`
11028: @*/
11029: PetscErrorCode MatGetOperation(Mat mat, MatOperation op, void (**f)(void))
11030: {
11031: PetscFunctionBegin;
11033: *f = (((void (**)(void))mat->ops)[op]);
11034: PetscFunctionReturn(PETSC_SUCCESS);
11035: }
11037: /*@
11038: MatHasOperation - Determines whether the given matrix supports the particular operation.
11040: Not Collective
11042: Input Parameters:
11043: + mat - the matrix
11044: - op - the operation, for example, `MATOP_GET_DIAGONAL`
11046: Output Parameter:
11047: . has - either `PETSC_TRUE` or `PETSC_FALSE`
11049: Level: advanced
11051: Note:
11052: See `MatSetOperation()` for additional discussion on naming convention and usage of `op`.
11054: .seealso: [](ch_matrices), `Mat`, `MatCreateShell()`, `MatGetOperation()`, `MatSetOperation()`
11055: @*/
11056: PetscErrorCode MatHasOperation(Mat mat, MatOperation op, PetscBool *has)
11057: {
11058: PetscFunctionBegin;
11061: if (mat->ops->hasoperation) {
11062: PetscUseTypeMethod(mat, hasoperation, op, has);
11063: } else {
11064: if (((void **)mat->ops)[op]) *has = PETSC_TRUE;
11065: else {
11066: *has = PETSC_FALSE;
11067: if (op == MATOP_CREATE_SUBMATRIX) {
11068: PetscMPIInt size;
11070: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size));
11071: if (size == 1) PetscCall(MatHasOperation(mat, MATOP_CREATE_SUBMATRICES, has));
11072: }
11073: }
11074: }
11075: PetscFunctionReturn(PETSC_SUCCESS);
11076: }
11078: /*@
11079: MatHasCongruentLayouts - Determines whether the rows and columns layouts of the matrix are congruent
11081: Collective
11083: Input Parameter:
11084: . mat - the matrix
11086: Output Parameter:
11087: . cong - either `PETSC_TRUE` or `PETSC_FALSE`
11089: Level: beginner
11091: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatSetSizes()`, `PetscLayout`
11092: @*/
11093: PetscErrorCode MatHasCongruentLayouts(Mat mat, PetscBool *cong)
11094: {
11095: PetscFunctionBegin;
11099: if (!mat->rmap || !mat->cmap) {
11100: *cong = mat->rmap == mat->cmap ? PETSC_TRUE : PETSC_FALSE;
11101: PetscFunctionReturn(PETSC_SUCCESS);
11102: }
11103: if (mat->congruentlayouts == PETSC_DECIDE) { /* first time we compare rows and cols layouts */
11104: PetscCall(PetscLayoutSetUp(mat->rmap));
11105: PetscCall(PetscLayoutSetUp(mat->cmap));
11106: PetscCall(PetscLayoutCompare(mat->rmap, mat->cmap, cong));
11107: if (*cong) mat->congruentlayouts = 1;
11108: else mat->congruentlayouts = 0;
11109: } else *cong = mat->congruentlayouts ? PETSC_TRUE : PETSC_FALSE;
11110: PetscFunctionReturn(PETSC_SUCCESS);
11111: }
11113: PetscErrorCode MatSetInf(Mat A)
11114: {
11115: PetscFunctionBegin;
11116: PetscUseTypeMethod(A, setinf);
11117: PetscFunctionReturn(PETSC_SUCCESS);
11118: }
11120: /*@C
11121: MatCreateGraph - create a scalar matrix (that is a matrix with one vertex for each block vertex in the original matrix), for use in graph algorithms
11122: and possibly removes small values from the graph structure.
11124: Collective
11126: Input Parameters:
11127: + A - the matrix
11128: . sym - `PETSC_TRUE` indicates that the graph should be symmetrized
11129: . scale - `PETSC_TRUE` indicates that the graph edge weights should be symmetrically scaled with the diagonal entry
11130: - filter - filter value - < 0: does nothing; == 0: removes only 0.0 entries; otherwise: removes entries with abs(entries) <= value
11132: Output Parameter:
11133: . graph - the resulting graph
11135: Level: advanced
11137: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `PCGAMG`
11138: @*/
11139: PetscErrorCode MatCreateGraph(Mat A, PetscBool sym, PetscBool scale, PetscReal filter, Mat *graph)
11140: {
11141: PetscFunctionBegin;
11146: PetscUseTypeMethod(A, creategraph, sym, scale, filter, graph);
11147: PetscFunctionReturn(PETSC_SUCCESS);
11148: }
11150: /*@
11151: MatEliminateZeros - eliminate the nondiagonal zero entries in place from the nonzero structure of a sparse `Mat` in place,
11152: meaning the same memory is used for the matrix, and no new memory is allocated.
11154: Collective
11156: Input Parameter:
11157: . A - the matrix
11159: Level: intermediate
11161: Developer Note:
11162: The entries in the sparse matrix data structure are shifted to fill in the unneeded locations in the data. Thus the end
11163: of the arrays in the data structure are unneeded.
11165: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatCreateGraph()`, `MatChop()`
11166: @*/
11167: PetscErrorCode MatEliminateZeros(Mat A)
11168: {
11169: PetscFunctionBegin;
11171: PetscUseTypeMethod(A, eliminatezeros);
11172: PetscFunctionReturn(PETSC_SUCCESS);
11173: }