Actual source code: matrix.c
1: /*
2: This is where the abstract matrix operations are defined
3: */
5: #include <petsc/private/matimpl.h>
6: #include <petsc/private/isimpl.h>
7: #include <petsc/private/vecimpl.h>
9: /* Logging support */
10: PetscClassId MAT_CLASSID;
11: PetscClassId MAT_COLORING_CLASSID;
12: PetscClassId MAT_FDCOLORING_CLASSID;
13: PetscClassId MAT_TRANSPOSECOLORING_CLASSID;
15: PetscLogEvent MAT_Mult, MAT_Mults, MAT_MultConstrained, MAT_MultAdd, MAT_MultTranspose;
16: PetscLogEvent MAT_MultTransposeConstrained, MAT_MultTransposeAdd, MAT_Solve, MAT_Solves, MAT_SolveAdd, MAT_SolveTranspose, MAT_MatSolve,MAT_MatTrSolve;
17: PetscLogEvent MAT_SolveTransposeAdd, MAT_SOR, MAT_ForwardSolve, MAT_BackwardSolve, MAT_LUFactor, MAT_LUFactorSymbolic;
18: PetscLogEvent MAT_LUFactorNumeric, MAT_CholeskyFactor, MAT_CholeskyFactorSymbolic, MAT_CholeskyFactorNumeric, MAT_ILUFactor;
19: PetscLogEvent MAT_ILUFactorSymbolic, MAT_ICCFactorSymbolic, MAT_Copy, MAT_Convert, MAT_Scale, MAT_AssemblyBegin;
20: PetscLogEvent MAT_QRFactorNumeric, MAT_QRFactorSymbolic, MAT_QRFactor;
21: PetscLogEvent MAT_AssemblyEnd, MAT_SetValues, MAT_GetValues, MAT_GetRow, MAT_GetRowIJ, MAT_CreateSubMats, MAT_GetOrdering, MAT_RedundantMat, MAT_GetSeqNonzeroStructure;
22: PetscLogEvent MAT_IncreaseOverlap, MAT_Partitioning, MAT_PartitioningND, MAT_Coarsen, MAT_ZeroEntries, MAT_Load, MAT_View, MAT_AXPY, MAT_FDColoringCreate;
23: PetscLogEvent MAT_FDColoringSetUp, MAT_FDColoringApply,MAT_Transpose,MAT_FDColoringFunction, MAT_CreateSubMat;
24: PetscLogEvent MAT_TransposeColoringCreate;
25: PetscLogEvent MAT_MatMult, MAT_MatMultSymbolic, MAT_MatMultNumeric;
26: PetscLogEvent MAT_PtAP, MAT_PtAPSymbolic, MAT_PtAPNumeric,MAT_RARt, MAT_RARtSymbolic, MAT_RARtNumeric;
27: PetscLogEvent MAT_MatTransposeMult, MAT_MatTransposeMultSymbolic, MAT_MatTransposeMultNumeric;
28: PetscLogEvent MAT_TransposeMatMult, MAT_TransposeMatMultSymbolic, MAT_TransposeMatMultNumeric;
29: PetscLogEvent MAT_MatMatMult, MAT_MatMatMultSymbolic, MAT_MatMatMultNumeric;
30: PetscLogEvent MAT_MultHermitianTranspose,MAT_MultHermitianTransposeAdd;
31: PetscLogEvent MAT_Getsymtranspose, MAT_Getsymtransreduced, MAT_GetBrowsOfAcols;
32: PetscLogEvent MAT_GetBrowsOfAocols, MAT_Getlocalmat, MAT_Getlocalmatcondensed, MAT_Seqstompi, MAT_Seqstompinum, MAT_Seqstompisym;
33: PetscLogEvent MAT_Applypapt, MAT_Applypapt_numeric, MAT_Applypapt_symbolic, MAT_GetSequentialNonzeroStructure;
34: PetscLogEvent MAT_GetMultiProcBlock;
35: PetscLogEvent MAT_CUSPARSECopyToGPU, MAT_CUSPARSECopyFromGPU, MAT_CUSPARSEGenerateTranspose, MAT_CUSPARSESolveAnalysis;
36: PetscLogEvent MAT_PreallCOO, MAT_SetVCOO;
37: PetscLogEvent MAT_SetValuesBatch;
38: PetscLogEvent MAT_ViennaCLCopyToGPU;
39: PetscLogEvent MAT_DenseCopyToGPU, MAT_DenseCopyFromGPU;
40: PetscLogEvent MAT_Merge,MAT_Residual,MAT_SetRandom;
41: PetscLogEvent MAT_FactorFactS,MAT_FactorInvS;
42: PetscLogEvent MATCOLORING_Apply,MATCOLORING_Comm,MATCOLORING_Local,MATCOLORING_ISCreate,MATCOLORING_SetUp,MATCOLORING_Weights;
43: PetscLogEvent MAT_H2Opus_Build,MAT_H2Opus_Compress,MAT_H2Opus_Orthog;
45: const char *const MatFactorTypes[] = {"NONE","LU","CHOLESKY","ILU","ICC","ILUDT","QR","MatFactorType","MAT_FACTOR_",NULL};
47: /*@
48: MatSetRandom - Sets all components of a matrix to random numbers. For sparse matrices that have been preallocated but not been assembled it randomly selects appropriate locations,
49: for sparse matrices that already have locations it fills the locations with random numbers
51: Logically Collective on Mat
53: Input Parameters:
54: + x - the matrix
55: - rctx - the random number context, formed by PetscRandomCreate(), or NULL and
56: it will create one internally.
58: Output Parameter:
59: . x - the matrix
61: Example of Usage:
62: .vb
63: PetscRandomCreate(PETSC_COMM_WORLD,&rctx);
64: MatSetRandom(x,rctx);
65: PetscRandomDestroy(rctx);
66: .ve
68: Level: intermediate
70: .seealso: MatZeroEntries(), MatSetValues(), PetscRandomCreate(), PetscRandomDestroy()
71: @*/
72: PetscErrorCode MatSetRandom(Mat x,PetscRandom rctx)
73: {
75: PetscRandom randObj = NULL;
82: if (!x->ops->setrandom) SETERRQ1(PetscObjectComm((PetscObject)x),PETSC_ERR_SUP,"Mat type %s",((PetscObject)x)->type_name);
84: if (!rctx) {
85: MPI_Comm comm;
86: PetscObjectGetComm((PetscObject)x,&comm);
87: PetscRandomCreate(comm,&randObj);
88: PetscRandomSetFromOptions(randObj);
89: rctx = randObj;
90: }
92: PetscLogEventBegin(MAT_SetRandom,x,rctx,0,0);
93: (*x->ops->setrandom)(x,rctx);
94: PetscLogEventEnd(MAT_SetRandom,x,rctx,0,0);
96: MatAssemblyBegin(x, MAT_FINAL_ASSEMBLY);
97: MatAssemblyEnd(x, MAT_FINAL_ASSEMBLY);
98: PetscRandomDestroy(&randObj);
99: return(0);
100: }
102: /*@
103: MatFactorGetErrorZeroPivot - returns the pivot value that was determined to be zero and the row it occurred in
105: Logically Collective on Mat
107: Input Parameter:
108: . mat - the factored matrix
110: Output Parameters:
111: + pivot - the pivot value computed
112: - row - the row that the zero pivot occurred. Note that this row must be interpreted carefully due to row reorderings and which processes
113: the share the matrix
115: Level: advanced
117: Notes:
118: This routine does not work for factorizations done with external packages.
120: This routine should only be called if MatGetFactorError() returns a value of MAT_FACTOR_NUMERIC_ZEROPIVOT
122: This can be called on non-factored matrices that come from, for example, matrices used in SOR.
124: .seealso: MatZeroEntries(), MatFactor(), MatGetFactor(), MatLUFactorSymbolic(), MatCholeskyFactorSymbolic(), MatFactorClearError(), MatFactorGetErrorZeroPivot()
125: @*/
126: PetscErrorCode MatFactorGetErrorZeroPivot(Mat mat,PetscReal *pivot,PetscInt *row)
127: {
130: *pivot = mat->factorerror_zeropivot_value;
131: *row = mat->factorerror_zeropivot_row;
132: return(0);
133: }
135: /*@
136: MatFactorGetError - gets the error code from a factorization
138: Logically Collective on Mat
140: Input Parameters:
141: . mat - the factored matrix
143: Output Parameter:
144: . err - the error code
146: Level: advanced
148: Notes:
149: This can be called on non-factored matrices that come from, for example, matrices used in SOR.
151: .seealso: MatZeroEntries(), MatFactor(), MatGetFactor(), MatLUFactorSymbolic(), MatCholeskyFactorSymbolic(), MatFactorClearError(), MatFactorGetErrorZeroPivot()
152: @*/
153: PetscErrorCode MatFactorGetError(Mat mat,MatFactorError *err)
154: {
157: *err = mat->factorerrortype;
158: return(0);
159: }
161: /*@
162: MatFactorClearError - clears the error code in a factorization
164: Logically Collective on Mat
166: Input Parameter:
167: . mat - the factored matrix
169: Level: developer
171: Notes:
172: This can be called on non-factored matrices that come from, for example, matrices used in SOR.
174: .seealso: MatZeroEntries(), MatFactor(), MatGetFactor(), MatLUFactorSymbolic(), MatCholeskyFactorSymbolic(), MatFactorGetError(), MatFactorGetErrorZeroPivot()
175: @*/
176: PetscErrorCode MatFactorClearError(Mat mat)
177: {
180: mat->factorerrortype = MAT_FACTOR_NOERROR;
181: mat->factorerror_zeropivot_value = 0.0;
182: mat->factorerror_zeropivot_row = 0;
183: return(0);
184: }
186: PETSC_INTERN PetscErrorCode MatFindNonzeroRowsOrCols_Basic(Mat mat,PetscBool cols,PetscReal tol,IS *nonzero)
187: {
188: PetscErrorCode ierr;
189: Vec r,l;
190: const PetscScalar *al;
191: PetscInt i,nz,gnz,N,n;
194: MatCreateVecs(mat,&r,&l);
195: if (!cols) { /* nonzero rows */
196: MatGetSize(mat,&N,NULL);
197: MatGetLocalSize(mat,&n,NULL);
198: VecSet(l,0.0);
199: VecSetRandom(r,NULL);
200: MatMult(mat,r,l);
201: VecGetArrayRead(l,&al);
202: } else { /* nonzero columns */
203: MatGetSize(mat,NULL,&N);
204: MatGetLocalSize(mat,NULL,&n);
205: VecSet(r,0.0);
206: VecSetRandom(l,NULL);
207: MatMultTranspose(mat,l,r);
208: VecGetArrayRead(r,&al);
209: }
210: if (tol <= 0.0) { for (i=0,nz=0;i<n;i++) if (al[i] != 0.0) nz++; }
211: else { for (i=0,nz=0;i<n;i++) if (PetscAbsScalar(al[i]) > tol) nz++; }
212: MPIU_Allreduce(&nz,&gnz,1,MPIU_INT,MPI_SUM,PetscObjectComm((PetscObject)mat));
213: if (gnz != N) {
214: PetscInt *nzr;
215: PetscMalloc1(nz,&nzr);
216: if (nz) {
217: if (tol < 0) { for (i=0,nz=0;i<n;i++) if (al[i] != 0.0) nzr[nz++] = i; }
218: else { for (i=0,nz=0;i<n;i++) if (PetscAbsScalar(al[i]) > tol) nzr[nz++] = i; }
219: }
220: ISCreateGeneral(PetscObjectComm((PetscObject)mat),nz,nzr,PETSC_OWN_POINTER,nonzero);
221: } else *nonzero = NULL;
222: if (!cols) { /* nonzero rows */
223: VecRestoreArrayRead(l,&al);
224: } else {
225: VecRestoreArrayRead(r,&al);
226: }
227: VecDestroy(&l);
228: VecDestroy(&r);
229: return(0);
230: }
232: /*@
233: MatFindNonzeroRows - Locate all rows that are not completely zero in the matrix
235: Input Parameter:
236: . A - the matrix
238: Output Parameter:
239: . keptrows - the rows that are not completely zero
241: Notes:
242: keptrows is set to NULL if all rows are nonzero.
244: Level: intermediate
246: @*/
247: PetscErrorCode MatFindNonzeroRows(Mat mat,IS *keptrows)
248: {
255: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
256: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
257: if (!mat->ops->findnonzerorows) {
258: MatFindNonzeroRowsOrCols_Basic(mat,PETSC_FALSE,0.0,keptrows);
259: } else {
260: (*mat->ops->findnonzerorows)(mat,keptrows);
261: }
262: return(0);
263: }
265: /*@
266: MatFindZeroRows - Locate all rows that are completely zero in the matrix
268: Input Parameter:
269: . A - the matrix
271: Output Parameter:
272: . zerorows - the rows that are completely zero
274: Notes:
275: zerorows is set to NULL if no rows are zero.
277: Level: intermediate
279: @*/
280: PetscErrorCode MatFindZeroRows(Mat mat,IS *zerorows)
281: {
283: IS keptrows;
284: PetscInt m, n;
290: MatFindNonzeroRows(mat, &keptrows);
291: /* MatFindNonzeroRows sets keptrows to NULL if there are no zero rows.
292: In keeping with this convention, we set zerorows to NULL if there are no zero
293: rows. */
294: if (keptrows == NULL) {
295: *zerorows = NULL;
296: } else {
297: MatGetOwnershipRange(mat,&m,&n);
298: ISComplement(keptrows,m,n,zerorows);
299: ISDestroy(&keptrows);
300: }
301: return(0);
302: }
304: /*@
305: MatGetDiagonalBlock - Returns the part of the matrix associated with the on-process coupling
307: Not Collective
309: Input Parameters:
310: . A - the matrix
312: Output Parameters:
313: . a - the diagonal part (which is a SEQUENTIAL matrix)
315: Notes:
316: see the manual page for MatCreateAIJ() for more information on the "diagonal part" of the matrix.
317: Use caution, as the reference count on the returned matrix is not incremented and it is used as
318: part of the containing MPI Mat's normal operation.
320: Level: advanced
322: @*/
323: PetscErrorCode MatGetDiagonalBlock(Mat A,Mat *a)
324: {
331: if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
332: if (!A->ops->getdiagonalblock) {
333: PetscMPIInt size;
334: MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);
335: if (size == 1) {
336: *a = A;
337: return(0);
338: } else SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Not coded for matrix type %s",((PetscObject)A)->type_name);
339: }
340: (*A->ops->getdiagonalblock)(A,a);
341: return(0);
342: }
344: /*@
345: MatGetTrace - Gets the trace of a matrix. The sum of the diagonal entries.
347: Collective on Mat
349: Input Parameters:
350: . mat - the matrix
352: Output Parameter:
353: . trace - the sum of the diagonal entries
355: Level: advanced
357: @*/
358: PetscErrorCode MatGetTrace(Mat mat,PetscScalar *trace)
359: {
361: Vec diag;
364: MatCreateVecs(mat,&diag,NULL);
365: MatGetDiagonal(mat,diag);
366: VecSum(diag,trace);
367: VecDestroy(&diag);
368: return(0);
369: }
371: /*@
372: MatRealPart - Zeros out the imaginary part of the matrix
374: Logically Collective on Mat
376: Input Parameters:
377: . mat - the matrix
379: Level: advanced
381: .seealso: MatImaginaryPart()
382: @*/
383: PetscErrorCode MatRealPart(Mat mat)
384: {
390: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
391: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
392: if (!mat->ops->realpart) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
393: MatCheckPreallocated(mat,1);
394: (*mat->ops->realpart)(mat);
395: return(0);
396: }
398: /*@C
399: MatGetGhosts - Get the global index of all ghost nodes defined by the sparse matrix
401: Collective on Mat
403: Input Parameter:
404: . mat - the matrix
406: Output Parameters:
407: + nghosts - number of ghosts (note for BAIJ matrices there is one ghost for each block)
408: - ghosts - the global indices of the ghost points
410: Notes:
411: the nghosts and ghosts are suitable to pass into VecCreateGhost()
413: Level: advanced
415: @*/
416: PetscErrorCode MatGetGhosts(Mat mat,PetscInt *nghosts,const PetscInt *ghosts[])
417: {
423: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
424: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
425: if (!mat->ops->getghosts) {
426: if (nghosts) *nghosts = 0;
427: if (ghosts) *ghosts = NULL;
428: } else {
429: (*mat->ops->getghosts)(mat,nghosts,ghosts);
430: }
431: return(0);
432: }
434: /*@
435: MatImaginaryPart - Moves the imaginary part of the matrix to the real part and zeros the imaginary part
437: Logically Collective on Mat
439: Input Parameters:
440: . mat - the matrix
442: Level: advanced
444: .seealso: MatRealPart()
445: @*/
446: PetscErrorCode MatImaginaryPart(Mat mat)
447: {
453: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
454: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
455: if (!mat->ops->imaginarypart) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
456: MatCheckPreallocated(mat,1);
457: (*mat->ops->imaginarypart)(mat);
458: return(0);
459: }
461: /*@
462: MatMissingDiagonal - Determine if sparse matrix is missing a diagonal entry (or block entry for BAIJ matrices)
464: Not Collective
466: Input Parameter:
467: . mat - the matrix
469: Output Parameters:
470: + missing - is any diagonal missing
471: - dd - first diagonal entry that is missing (optional) on this process
473: Level: advanced
475: .seealso: MatRealPart()
476: @*/
477: PetscErrorCode MatMissingDiagonal(Mat mat,PetscBool *missing,PetscInt *dd)
478: {
485: if (!mat->assembled) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix %s",((PetscObject)mat)->type_name);
486: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
487: if (!mat->ops->missingdiagonal) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
488: (*mat->ops->missingdiagonal)(mat,missing,dd);
489: return(0);
490: }
492: /*@C
493: MatGetRow - Gets a row of a matrix. You MUST call MatRestoreRow()
494: for each row that you get to ensure that your application does
495: not bleed memory.
497: Not Collective
499: Input Parameters:
500: + mat - the matrix
501: - row - the row to get
503: Output Parameters:
504: + ncols - if not NULL, the number of nonzeros in the row
505: . cols - if not NULL, the column numbers
506: - vals - if not NULL, the values
508: Notes:
509: This routine is provided for people who need to have direct access
510: to the structure of a matrix. We hope that we provide enough
511: high-level matrix routines that few users will need it.
513: MatGetRow() always returns 0-based column indices, regardless of
514: whether the internal representation is 0-based (default) or 1-based.
516: For better efficiency, set cols and/or vals to NULL if you do
517: not wish to extract these quantities.
519: The user can only examine the values extracted with MatGetRow();
520: the values cannot be altered. To change the matrix entries, one
521: must use MatSetValues().
523: You can only have one call to MatGetRow() outstanding for a particular
524: matrix at a time, per processor. MatGetRow() can only obtain rows
525: associated with the given processor, it cannot get rows from the
526: other processors; for that we suggest using MatCreateSubMatrices(), then
527: MatGetRow() on the submatrix. The row index passed to MatGetRow()
528: is in the global number of rows.
530: Fortran Notes:
531: The calling sequence from Fortran is
532: .vb
533: MatGetRow(matrix,row,ncols,cols,values,ierr)
534: Mat matrix (input)
535: integer row (input)
536: integer ncols (output)
537: integer cols(maxcols) (output)
538: double precision (or double complex) values(maxcols) output
539: .ve
540: where maxcols >= maximum nonzeros in any row of the matrix.
542: Caution:
543: Do not try to change the contents of the output arrays (cols and vals).
544: In some cases, this may corrupt the matrix.
546: Level: advanced
548: .seealso: MatRestoreRow(), MatSetValues(), MatGetValues(), MatCreateSubMatrices(), MatGetDiagonal()
549: @*/
550: PetscErrorCode MatGetRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[])
551: {
553: PetscInt incols;
558: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
559: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
560: if (!mat->ops->getrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
561: MatCheckPreallocated(mat,1);
562: if (row < mat->rmap->rstart || row >= mat->rmap->rend) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Only for local rows, %D not in [%D,%D)",row,mat->rmap->rstart,mat->rmap->rend);
563: PetscLogEventBegin(MAT_GetRow,mat,0,0,0);
564: (*mat->ops->getrow)(mat,row,&incols,(PetscInt**)cols,(PetscScalar**)vals);
565: if (ncols) *ncols = incols;
566: PetscLogEventEnd(MAT_GetRow,mat,0,0,0);
567: return(0);
568: }
570: /*@
571: MatConjugate - replaces the matrix values with their complex conjugates
573: Logically Collective on Mat
575: Input Parameters:
576: . mat - the matrix
578: Level: advanced
580: .seealso: VecConjugate()
581: @*/
582: PetscErrorCode MatConjugate(Mat mat)
583: {
584: #if defined(PETSC_USE_COMPLEX)
589: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
590: if (!mat->ops->conjugate) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not provided for matrix type %s, send email to petsc-maint@mcs.anl.gov",((PetscObject)mat)->type_name);
591: (*mat->ops->conjugate)(mat);
592: #else
594: #endif
595: return(0);
596: }
598: /*@C
599: MatRestoreRow - Frees any temporary space allocated by MatGetRow().
601: Not Collective
603: Input Parameters:
604: + mat - the matrix
605: . row - the row to get
606: . ncols, cols - the number of nonzeros and their columns
607: - vals - if nonzero the column values
609: Notes:
610: This routine should be called after you have finished examining the entries.
612: This routine zeros out ncols, cols, and vals. This is to prevent accidental
613: us of the array after it has been restored. If you pass NULL, it will
614: not zero the pointers. Use of cols or vals after MatRestoreRow is invalid.
616: Fortran Notes:
617: The calling sequence from Fortran is
618: .vb
619: MatRestoreRow(matrix,row,ncols,cols,values,ierr)
620: Mat matrix (input)
621: integer row (input)
622: integer ncols (output)
623: integer cols(maxcols) (output)
624: double precision (or double complex) values(maxcols) output
625: .ve
626: Where maxcols >= maximum nonzeros in any row of the matrix.
628: In Fortran MatRestoreRow() MUST be called after MatGetRow()
629: before another call to MatGetRow() can be made.
631: Level: advanced
633: .seealso: MatGetRow()
634: @*/
635: PetscErrorCode MatRestoreRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[])
636: {
642: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
643: if (!mat->ops->restorerow) return(0);
644: (*mat->ops->restorerow)(mat,row,ncols,(PetscInt **)cols,(PetscScalar **)vals);
645: if (ncols) *ncols = 0;
646: if (cols) *cols = NULL;
647: if (vals) *vals = NULL;
648: return(0);
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/MatRestoreRow() to disable the flag.
655: Not Collective
657: Input Parameters:
658: . mat - the matrix
660: Notes:
661: The flag is to ensure that users are aware of MatGetRow() only provides the upper triangular part of the row for the matrices in MATSBAIJ format.
663: Level: advanced
665: .seealso: MatRestoreRowUpperTriangular()
666: @*/
667: PetscErrorCode MatGetRowUpperTriangular(Mat mat)
668: {
674: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
675: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
676: MatCheckPreallocated(mat,1);
677: if (!mat->ops->getrowuppertriangular) return(0);
678: (*mat->ops->getrowuppertriangular)(mat);
679: return(0);
680: }
682: /*@
683: MatRestoreRowUpperTriangular - Disable calls to MatGetRow() for matrix in MATSBAIJ format.
685: Not Collective
687: Input Parameters:
688: . mat - the matrix
690: Notes:
691: This routine should be called after you have finished MatGetRow/MatRestoreRow().
693: Level: advanced
695: .seealso: MatGetRowUpperTriangular()
696: @*/
697: PetscErrorCode MatRestoreRowUpperTriangular(Mat mat)
698: {
704: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
705: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
706: MatCheckPreallocated(mat,1);
707: if (!mat->ops->restorerowuppertriangular) return(0);
708: (*mat->ops->restorerowuppertriangular)(mat);
709: return(0);
710: }
712: /*@C
713: MatSetOptionsPrefix - Sets the prefix used for searching for all
714: Mat options in the database.
716: Logically Collective on Mat
718: Input Parameters:
719: + A - the Mat context
720: - prefix - the prefix to prepend to all option names
722: Notes:
723: A hyphen (-) must NOT be given at the beginning of the prefix name.
724: The first character of all runtime options is AUTOMATICALLY the hyphen.
726: Level: advanced
728: .seealso: MatSetFromOptions()
729: @*/
730: PetscErrorCode MatSetOptionsPrefix(Mat A,const char prefix[])
731: {
736: PetscObjectSetOptionsPrefix((PetscObject)A,prefix);
737: return(0);
738: }
740: /*@C
741: MatAppendOptionsPrefix - Appends to the prefix used for searching for all
742: Mat options in the database.
744: Logically Collective on Mat
746: Input Parameters:
747: + A - the Mat context
748: - prefix - the prefix to prepend to all option names
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: Level: advanced
756: .seealso: MatGetOptionsPrefix()
757: @*/
758: PetscErrorCode MatAppendOptionsPrefix(Mat A,const char prefix[])
759: {
764: PetscObjectAppendOptionsPrefix((PetscObject)A,prefix);
765: return(0);
766: }
768: /*@C
769: MatGetOptionsPrefix - Gets the prefix used for searching for all
770: Mat options in the database.
772: Not Collective
774: Input Parameter:
775: . A - the Mat context
777: Output Parameter:
778: . prefix - pointer to the prefix string used
780: Notes:
781: On the fortran side, the user should pass in a string 'prefix' of
782: sufficient length to hold the prefix.
784: Level: advanced
786: .seealso: MatAppendOptionsPrefix()
787: @*/
788: PetscErrorCode MatGetOptionsPrefix(Mat A,const char *prefix[])
789: {
794: PetscObjectGetOptionsPrefix((PetscObject)A,prefix);
795: return(0);
796: }
798: /*@
799: MatResetPreallocation - Reset mat to use the original nonzero pattern provided by users.
801: Collective on Mat
803: Input Parameters:
804: . A - the Mat context
806: Notes:
807: The allocated memory will be shrunk after calling MatAssembly with MAT_FINAL_ASSEMBLY. Users can reset the preallocation to access the original memory.
808: Currently support MPIAIJ and SEQAIJ.
810: Level: beginner
812: .seealso: MatSeqAIJSetPreallocation(), MatMPIAIJSetPreallocation(), MatXAIJSetPreallocation()
813: @*/
814: PetscErrorCode MatResetPreallocation(Mat A)
815: {
821: PetscUseMethod(A,"MatResetPreallocation_C",(Mat),(A));
822: return(0);
823: }
825: /*@
826: MatSetUp - Sets up the internal matrix data structures for later use.
828: Collective on Mat
830: Input Parameters:
831: . A - the Mat context
833: Notes:
834: If the user has not set preallocation for this matrix then a default preallocation that is likely to be inefficient is used.
836: If a suitable preallocation routine is used, this function does not need to be called.
838: See the Performance chapter of the PETSc users manual for how to preallocate matrices
840: Level: beginner
842: .seealso: MatCreate(), MatDestroy()
843: @*/
844: PetscErrorCode MatSetUp(Mat A)
845: {
846: PetscMPIInt size;
851: if (!((PetscObject)A)->type_name) {
852: MPI_Comm_size(PetscObjectComm((PetscObject)A), &size);
853: if (size == 1) {
854: MatSetType(A, MATSEQAIJ);
855: } else {
856: MatSetType(A, MATMPIAIJ);
857: }
858: }
859: if (!A->preallocated && A->ops->setup) {
860: PetscInfo(A,"Warning not preallocating matrix storage\n");
861: (*A->ops->setup)(A);
862: }
863: PetscLayoutSetUp(A->rmap);
864: PetscLayoutSetUp(A->cmap);
865: A->preallocated = PETSC_TRUE;
866: return(0);
867: }
869: #if defined(PETSC_HAVE_SAWS)
870: #include <petscviewersaws.h>
871: #endif
873: /*@C
874: MatViewFromOptions - View from Options
876: Collective on Mat
878: Input Parameters:
879: + A - the Mat context
880: . obj - Optional object
881: - name - command line option
883: Level: intermediate
884: .seealso: Mat, MatView, PetscObjectViewFromOptions(), MatCreate()
885: @*/
886: PetscErrorCode MatViewFromOptions(Mat A,PetscObject obj,const char name[])
887: {
892: PetscObjectViewFromOptions((PetscObject)A,obj,name);
893: return(0);
894: }
896: /*@C
897: MatView - Visualizes a matrix object.
899: Collective on Mat
901: Input Parameters:
902: + mat - the matrix
903: - viewer - visualization context
905: Notes:
906: The available visualization contexts include
907: + PETSC_VIEWER_STDOUT_SELF - for sequential matrices
908: . PETSC_VIEWER_STDOUT_WORLD - for parallel matrices created on PETSC_COMM_WORLD
909: . PETSC_VIEWER_STDOUT_(comm) - for matrices created on MPI communicator comm
910: - PETSC_VIEWER_DRAW_WORLD - graphical display of nonzero structure
912: The user can open alternative visualization contexts with
913: + PetscViewerASCIIOpen() - Outputs matrix to a specified file
914: . PetscViewerBinaryOpen() - Outputs matrix in binary to a
915: specified file; corresponding input uses MatLoad()
916: . PetscViewerDrawOpen() - Outputs nonzero matrix structure to
917: an X window display
918: - PetscViewerSocketOpen() - Outputs matrix to Socket viewer.
919: Currently only the sequential dense and AIJ
920: matrix types support the Socket viewer.
922: The user can call PetscViewerPushFormat() to specify the output
923: format of ASCII printed objects (when using PETSC_VIEWER_STDOUT_SELF,
924: PETSC_VIEWER_STDOUT_WORLD and PetscViewerASCIIOpen). Available formats include
925: + PETSC_VIEWER_DEFAULT - default, prints matrix contents
926: . PETSC_VIEWER_ASCII_MATLAB - prints matrix contents in Matlab format
927: . PETSC_VIEWER_ASCII_DENSE - prints entire matrix including zeros
928: . PETSC_VIEWER_ASCII_COMMON - prints matrix contents, using a sparse
929: format common among all matrix types
930: . PETSC_VIEWER_ASCII_IMPL - prints matrix contents, using an implementation-specific
931: format (which is in many cases the same as the default)
932: . PETSC_VIEWER_ASCII_INFO - prints basic information about the matrix
933: size and structure (not the matrix entries)
934: - PETSC_VIEWER_ASCII_INFO_DETAIL - prints more detailed information about
935: the matrix structure
937: Options Database Keys:
938: + -mat_view ::ascii_info - Prints info on matrix at conclusion of MatAssemblyEnd()
939: . -mat_view ::ascii_info_detail - Prints more detailed info
940: . -mat_view - Prints matrix in ASCII format
941: . -mat_view ::ascii_matlab - Prints matrix in Matlab format
942: . -mat_view draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX().
943: . -display <name> - Sets display name (default is host)
944: . -draw_pause <sec> - Sets number of seconds to pause after display
945: . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (see Users-Manual: ch_matlab for details)
946: . -viewer_socket_machine <machine> -
947: . -viewer_socket_port <port> -
948: . -mat_view binary - save matrix to file in binary format
949: - -viewer_binary_filename <name> -
950: Level: beginner
952: Notes:
953: The ASCII viewers are only recommended for small matrices on at most a moderate number of processes,
954: the program will seemingly hang and take hours for larger matrices, for larger matrices one should use the binary format.
956: In the debugger you can do "call MatView(mat,0)" to display the matrix. (The same holds for any PETSc object viewer).
958: See the manual page for MatLoad() for the exact format of the binary file when the binary
959: viewer is used.
961: See share/petsc/matlab/PetscBinaryRead.m for a Matlab code that can read in the binary file when the binary
962: viewer is used and lib/petsc/bin/PetscBinaryIO.py for loading them into Python.
964: One can use '-mat_view draw -draw_pause -1' to pause the graphical display of matrix nonzero structure,
965: and then use the following mouse functions.
966: + left mouse: zoom in
967: . middle mouse: zoom out
968: - right mouse: continue with the simulation
970: .seealso: PetscViewerPushFormat(), PetscViewerASCIIOpen(), PetscViewerDrawOpen(),
971: PetscViewerSocketOpen(), PetscViewerBinaryOpen(), MatLoad()
972: @*/
973: PetscErrorCode MatView(Mat mat,PetscViewer viewer)
974: {
975: PetscErrorCode ierr;
976: PetscInt rows,cols,rbs,cbs;
977: PetscBool isascii,isstring,issaws;
978: PetscViewerFormat format;
979: PetscMPIInt size;
984: if (!viewer) {PetscViewerASCIIGetStdout(PetscObjectComm((PetscObject)mat),&viewer);}
987: MatCheckPreallocated(mat,1);
989: PetscViewerGetFormat(viewer,&format);
990: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
991: if (size == 1 && format == PETSC_VIEWER_LOAD_BALANCE) return(0);
993: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSTRING,&isstring);
994: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isascii);
995: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSAWS,&issaws);
996: if ((!isascii || (format != PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL)) && mat->factortype) {
997: SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"No viewers for factored matrix except ASCII info or info_detail");
998: }
1000: PetscLogEventBegin(MAT_View,mat,viewer,0,0);
1001: if (isascii) {
1002: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ORDER,"Must call MatAssemblyBegin/End() before viewing matrix");
1003: PetscObjectPrintClassNamePrefixType((PetscObject)mat,viewer);
1004: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
1005: MatNullSpace nullsp,transnullsp;
1007: PetscViewerASCIIPushTab(viewer);
1008: MatGetSize(mat,&rows,&cols);
1009: MatGetBlockSizes(mat,&rbs,&cbs);
1010: if (rbs != 1 || cbs != 1) {
1011: if (rbs != cbs) {PetscViewerASCIIPrintf(viewer,"rows=%D, cols=%D, rbs=%D, cbs=%D\n",rows,cols,rbs,cbs);}
1012: else {PetscViewerASCIIPrintf(viewer,"rows=%D, cols=%D, bs=%D\n",rows,cols,rbs);}
1013: } else {
1014: PetscViewerASCIIPrintf(viewer,"rows=%D, cols=%D\n",rows,cols);
1015: }
1016: if (mat->factortype) {
1017: MatSolverType solver;
1018: MatFactorGetSolverType(mat,&solver);
1019: PetscViewerASCIIPrintf(viewer,"package used to perform factorization: %s\n",solver);
1020: }
1021: if (mat->ops->getinfo) {
1022: MatInfo info;
1023: MatGetInfo(mat,MAT_GLOBAL_SUM,&info);
1024: PetscViewerASCIIPrintf(viewer,"total: nonzeros=%.f, allocated nonzeros=%.f\n",info.nz_used,info.nz_allocated);
1025: if (!mat->factortype) {
1026: PetscViewerASCIIPrintf(viewer,"total number of mallocs used during MatSetValues calls=%D\n",(PetscInt)info.mallocs);
1027: }
1028: }
1029: MatGetNullSpace(mat,&nullsp);
1030: MatGetTransposeNullSpace(mat,&transnullsp);
1031: if (nullsp) {PetscViewerASCIIPrintf(viewer," has attached null space\n");}
1032: if (transnullsp && transnullsp != nullsp) {PetscViewerASCIIPrintf(viewer," has attached transposed null space\n");}
1033: MatGetNearNullSpace(mat,&nullsp);
1034: if (nullsp) {PetscViewerASCIIPrintf(viewer," has attached near null space\n");}
1035: PetscViewerASCIIPushTab(viewer);
1036: MatProductView(mat,viewer);
1037: PetscViewerASCIIPopTab(viewer);
1038: }
1039: } else if (issaws) {
1040: #if defined(PETSC_HAVE_SAWS)
1041: PetscMPIInt rank;
1043: PetscObjectName((PetscObject)mat);
1044: MPI_Comm_rank(PETSC_COMM_WORLD,&rank);
1045: if (!((PetscObject)mat)->amsmem && rank == 0) {
1046: PetscObjectViewSAWs((PetscObject)mat,viewer);
1047: }
1048: #endif
1049: } else if (isstring) {
1050: const char *type;
1051: MatGetType(mat,&type);
1052: PetscViewerStringSPrintf(viewer," MatType: %-7.7s",type);
1053: if (mat->ops->view) {(*mat->ops->view)(mat,viewer);}
1054: }
1055: if ((format == PETSC_VIEWER_NATIVE || format == PETSC_VIEWER_LOAD_BALANCE) && mat->ops->viewnative) {
1056: PetscViewerASCIIPushTab(viewer);
1057: (*mat->ops->viewnative)(mat,viewer);
1058: PetscViewerASCIIPopTab(viewer);
1059: } else if (mat->ops->view) {
1060: PetscViewerASCIIPushTab(viewer);
1061: (*mat->ops->view)(mat,viewer);
1062: PetscViewerASCIIPopTab(viewer);
1063: }
1064: if (isascii) {
1065: PetscViewerGetFormat(viewer,&format);
1066: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
1067: PetscViewerASCIIPopTab(viewer);
1068: }
1069: }
1070: PetscLogEventEnd(MAT_View,mat,viewer,0,0);
1071: return(0);
1072: }
1074: #if defined(PETSC_USE_DEBUG)
1075: #include <../src/sys/totalview/tv_data_display.h>
1076: PETSC_UNUSED static int TV_display_type(const struct _p_Mat *mat)
1077: {
1078: TV_add_row("Local rows", "int", &mat->rmap->n);
1079: TV_add_row("Local columns", "int", &mat->cmap->n);
1080: TV_add_row("Global rows", "int", &mat->rmap->N);
1081: TV_add_row("Global columns", "int", &mat->cmap->N);
1082: TV_add_row("Typename", TV_ascii_string_type, ((PetscObject)mat)->type_name);
1083: return TV_format_OK;
1084: }
1085: #endif
1087: /*@C
1088: MatLoad - Loads a matrix that has been stored in binary/HDF5 format
1089: with MatView(). The matrix format is determined from the options database.
1090: Generates a parallel MPI matrix if the communicator has more than one
1091: processor. The default matrix type is AIJ.
1093: Collective on PetscViewer
1095: Input Parameters:
1096: + mat - the newly loaded matrix, this needs to have been created with MatCreate()
1097: or some related function before a call to MatLoad()
1098: - viewer - binary/HDF5 file viewer
1100: Options Database Keys:
1101: Used with block matrix formats (MATSEQBAIJ, ...) to specify
1102: block size
1103: . -matload_block_size <bs>
1105: Level: beginner
1107: Notes:
1108: If the Mat type has not yet been given then MATAIJ is used, call MatSetFromOptions() on the
1109: Mat before calling this routine if you wish to set it from the options database.
1111: MatLoad() automatically loads into the options database any options
1112: given in the file filename.info where filename is the name of the file
1113: that was passed to the PetscViewerBinaryOpen(). The options in the info
1114: file will be ignored if you use the -viewer_binary_skip_info option.
1116: If the type or size of mat is not set before a call to MatLoad, PETSc
1117: sets the default matrix type AIJ and sets the local and global sizes.
1118: If type and/or size is already set, then the same are used.
1120: In parallel, each processor can load a subset of rows (or the
1121: entire matrix). This routine is especially useful when a large
1122: matrix is stored on disk and only part of it is desired on each
1123: processor. For example, a parallel solver may access only some of
1124: the rows from each processor. The algorithm used here reads
1125: relatively small blocks of data rather than reading the entire
1126: matrix and then subsetting it.
1128: Viewer's PetscViewerType must be either PETSCVIEWERBINARY or PETSCVIEWERHDF5.
1129: Such viewer can be created using PetscViewerBinaryOpen()/PetscViewerHDF5Open(),
1130: or the sequence like
1131: $ PetscViewer v;
1132: $ PetscViewerCreate(PETSC_COMM_WORLD,&v);
1133: $ PetscViewerSetType(v,PETSCVIEWERBINARY);
1134: $ PetscViewerSetFromOptions(v);
1135: $ PetscViewerFileSetMode(v,FILE_MODE_READ);
1136: $ PetscViewerFileSetName(v,"datafile");
1137: The optional PetscViewerSetFromOptions() call allows to override PetscViewerSetType() using option
1138: $ -viewer_type {binary,hdf5}
1140: See the example src/ksp/ksp/tutorials/ex27.c with the first approach,
1141: and src/mat/tutorials/ex10.c with the second approach.
1143: Notes about the PETSc binary format:
1144: In case of PETSCVIEWERBINARY, a native PETSc binary format is used. Each of the blocks
1145: is read onto rank 0 and then shipped to its destination rank, one after another.
1146: Multiple objects, both matrices and vectors, can be stored within the same file.
1147: Their PetscObject name is ignored; they are loaded in the order of their storage.
1149: Most users should not need to know the details of the binary storage
1150: format, since MatLoad() and MatView() completely hide these details.
1151: But for anyone who's interested, the standard binary matrix storage
1152: format is
1154: $ PetscInt MAT_FILE_CLASSID
1155: $ PetscInt number of rows
1156: $ PetscInt number of columns
1157: $ PetscInt total number of nonzeros
1158: $ PetscInt *number nonzeros in each row
1159: $ PetscInt *column indices of all nonzeros (starting index is zero)
1160: $ PetscScalar *values of all nonzeros
1162: PETSc automatically does the byte swapping for
1163: machines that store the bytes reversed, e.g. DEC alpha, freebsd,
1164: linux, Windows and the paragon; thus if you write your own binary
1165: read/write routines you have to swap the bytes; see PetscBinaryRead()
1166: and PetscBinaryWrite() to see how this may be done.
1168: Notes about the HDF5 (MATLAB MAT-File Version 7.3) format:
1169: In case of PETSCVIEWERHDF5, a parallel HDF5 reader is used.
1170: Each processor's chunk is loaded independently by its owning rank.
1171: Multiple objects, both matrices and vectors, can be stored within the same file.
1172: They are looked up by their PetscObject name.
1174: As the MATLAB MAT-File Version 7.3 format is also a HDF5 flavor, we decided to use
1175: by default the same structure and naming of the AIJ arrays and column count
1176: within the HDF5 file. This means that a MAT file saved with -v7.3 flag, e.g.
1177: $ save example.mat A b -v7.3
1178: can be directly read by this routine (see Reference 1 for details).
1179: Note that depending on your MATLAB version, this format might be a default,
1180: otherwise you can set it as default in Preferences.
1182: Unless -nocompression flag is used to save the file in MATLAB,
1183: PETSc must be configured with ZLIB package.
1185: See also examples src/mat/tutorials/ex10.c and src/ksp/ksp/tutorials/ex27.c
1187: Current HDF5 (MAT-File) limitations:
1188: This reader currently supports only real MATSEQAIJ, MATMPIAIJ, MATSEQDENSE and MATMPIDENSE matrices.
1190: Corresponding MatView() is not yet implemented.
1192: The loaded matrix is actually a transpose of the original one in MATLAB,
1193: unless you push PETSC_VIEWER_HDF5_MAT format (see examples above).
1194: With this format, matrix is automatically transposed by PETSc,
1195: unless the matrix is marked as SPD or symmetric
1196: (see MatSetOption(), MAT_SPD, MAT_SYMMETRIC).
1198: References:
1199: 1. MATLAB(R) Documentation, manual page of save(), https://www.mathworks.com/help/matlab/ref/save.html#btox10b-1-version
1201: .seealso: PetscViewerBinaryOpen(), PetscViewerSetType(), MatView(), VecLoad()
1203: @*/
1204: PetscErrorCode MatLoad(Mat mat,PetscViewer viewer)
1205: {
1207: PetscBool flg;
1213: if (!((PetscObject)mat)->type_name) {
1214: MatSetType(mat,MATAIJ);
1215: }
1217: flg = PETSC_FALSE;
1218: PetscOptionsGetBool(((PetscObject)mat)->options,((PetscObject)mat)->prefix,"-matload_symmetric",&flg,NULL);
1219: if (flg) {
1220: MatSetOption(mat,MAT_SYMMETRIC,PETSC_TRUE);
1221: MatSetOption(mat,MAT_SYMMETRY_ETERNAL,PETSC_TRUE);
1222: }
1223: flg = PETSC_FALSE;
1224: PetscOptionsGetBool(((PetscObject)mat)->options,((PetscObject)mat)->prefix,"-matload_spd",&flg,NULL);
1225: if (flg) {
1226: MatSetOption(mat,MAT_SPD,PETSC_TRUE);
1227: }
1229: if (!mat->ops->load) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatLoad is not supported for type %s",((PetscObject)mat)->type_name);
1230: PetscLogEventBegin(MAT_Load,mat,viewer,0,0);
1231: (*mat->ops->load)(mat,viewer);
1232: PetscLogEventEnd(MAT_Load,mat,viewer,0,0);
1233: return(0);
1234: }
1236: static PetscErrorCode MatDestroy_Redundant(Mat_Redundant **redundant)
1237: {
1239: Mat_Redundant *redund = *redundant;
1240: PetscInt i;
1243: if (redund) {
1244: if (redund->matseq) { /* via MatCreateSubMatrices() */
1245: ISDestroy(&redund->isrow);
1246: ISDestroy(&redund->iscol);
1247: MatDestroySubMatrices(1,&redund->matseq);
1248: } else {
1249: PetscFree2(redund->send_rank,redund->recv_rank);
1250: PetscFree(redund->sbuf_j);
1251: PetscFree(redund->sbuf_a);
1252: for (i=0; i<redund->nrecvs; i++) {
1253: PetscFree(redund->rbuf_j[i]);
1254: PetscFree(redund->rbuf_a[i]);
1255: }
1256: PetscFree4(redund->sbuf_nz,redund->rbuf_nz,redund->rbuf_j,redund->rbuf_a);
1257: }
1259: if (redund->subcomm) {
1260: PetscCommDestroy(&redund->subcomm);
1261: }
1262: PetscFree(redund);
1263: }
1264: return(0);
1265: }
1267: /*@C
1268: MatDestroy - Frees space taken by a matrix.
1270: Collective on Mat
1272: Input Parameter:
1273: . A - the matrix
1275: Level: beginner
1277: @*/
1278: PetscErrorCode MatDestroy(Mat *A)
1279: {
1283: if (!*A) return(0);
1285: if (--((PetscObject)(*A))->refct > 0) {*A = NULL; return(0);}
1287: /* if memory was published with SAWs then destroy it */
1288: PetscObjectSAWsViewOff((PetscObject)*A);
1289: if ((*A)->ops->destroy) {
1290: (*(*A)->ops->destroy)(*A);
1291: }
1293: PetscFree((*A)->defaultvectype);
1294: PetscFree((*A)->bsizes);
1295: PetscFree((*A)->solvertype);
1296: for (PetscInt i=0; i<MAT_FACTOR_NUM_TYPES; i++) {
1297: PetscFree((*A)->preferredordering[i]);
1298: }
1299: MatDestroy_Redundant(&(*A)->redundant);
1300: MatProductClear(*A);
1301: MatNullSpaceDestroy(&(*A)->nullsp);
1302: MatNullSpaceDestroy(&(*A)->transnullsp);
1303: MatNullSpaceDestroy(&(*A)->nearnullsp);
1304: MatDestroy(&(*A)->schur);
1305: PetscLayoutDestroy(&(*A)->rmap);
1306: PetscLayoutDestroy(&(*A)->cmap);
1307: PetscHeaderDestroy(A);
1308: return(0);
1309: }
1311: /*@C
1312: MatSetValues - Inserts or adds a block of values into a matrix.
1313: These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
1314: MUST be called after all calls to MatSetValues() have been completed.
1316: Not Collective
1318: Input Parameters:
1319: + mat - the matrix
1320: . v - a logically two-dimensional array of values
1321: . m, idxm - the number of rows and their global indices
1322: . n, idxn - the number of columns and their global indices
1323: - addv - either ADD_VALUES or INSERT_VALUES, where
1324: ADD_VALUES adds values to any existing entries, and
1325: INSERT_VALUES replaces existing entries with new values
1327: Notes:
1328: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or
1329: MatSetUp() before using this routine
1331: By default the values, v, are row-oriented. See MatSetOption() for other options.
1333: Calls to MatSetValues() with the INSERT_VALUES and ADD_VALUES
1334: options cannot be mixed without intervening calls to the assembly
1335: routines.
1337: MatSetValues() uses 0-based row and column numbers in Fortran
1338: as well as in C.
1340: Negative indices may be passed in idxm and idxn, these rows and columns are
1341: simply ignored. This allows easily inserting element stiffness matrices
1342: with homogeneous Dirchlet boundary conditions that you don't want represented
1343: in the matrix.
1345: Efficiency Alert:
1346: The routine MatSetValuesBlocked() may offer much better efficiency
1347: for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ).
1349: Level: beginner
1351: Developer Notes:
1352: This is labeled with C so does not automatically generate Fortran stubs and interfaces
1353: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
1355: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1356: InsertMode, INSERT_VALUES, ADD_VALUES
1357: @*/
1358: PetscErrorCode MatSetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv)
1359: {
1365: if (!m || !n) return(0); /* no values to insert */
1368: MatCheckPreallocated(mat,1);
1370: if (mat->insertmode == NOT_SET_VALUES) {
1371: mat->insertmode = addv;
1372: } else if (PetscUnlikely(mat->insertmode != addv)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1373: if (PetscDefined(USE_DEBUG)) {
1374: PetscInt i,j;
1376: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1377: if (!mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1379: for (i=0; i<m; i++) {
1380: for (j=0; j<n; j++) {
1381: if (mat->erroriffailure && PetscIsInfOrNanScalar(v[i*n+j]))
1382: #if defined(PETSC_USE_COMPLEX)
1383: SETERRQ4(PETSC_COMM_SELF,PETSC_ERR_FP,"Inserting %g+ig at matrix entry (%D,%D)",(double)PetscRealPart(v[i*n+j]),(double)PetscImaginaryPart(v[i*n+j]),idxm[i],idxn[j]);
1384: #else
1385: SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_FP,"Inserting %g at matrix entry (%D,%D)",(double)v[i*n+j],idxm[i],idxn[j]);
1386: #endif
1387: }
1388: }
1389: for (i=0; i<m; i++) if (idxm[i] >= mat->rmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Cannot insert in row %D, maximum is %D",idxm[i],mat->rmap->N-1);
1390: for (i=0; i<n; i++) if (idxn[i] >= mat->cmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Cannot insert in column %D, maximum is %D",idxn[i],mat->cmap->N-1);
1391: }
1393: if (mat->assembled) {
1394: mat->was_assembled = PETSC_TRUE;
1395: mat->assembled = PETSC_FALSE;
1396: }
1397: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
1398: (*mat->ops->setvalues)(mat,m,idxm,n,idxn,v,addv);
1399: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
1400: return(0);
1401: }
1403: /*@
1404: MatSetValuesRowLocal - Inserts a row (block row for BAIJ matrices) of nonzero
1405: values into a matrix
1407: Not Collective
1409: Input Parameters:
1410: + mat - the matrix
1411: . row - the (block) row to set
1412: - v - a logically two-dimensional array of values
1414: Notes:
1415: By the values, v, are column-oriented (for the block version) and sorted
1417: All the nonzeros in the row must be provided
1419: The matrix must have previously had its column indices set
1421: The row must belong to this process
1423: Level: intermediate
1425: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1426: InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues(), MatSetValuesRow(), MatSetLocalToGlobalMapping()
1427: @*/
1428: PetscErrorCode MatSetValuesRowLocal(Mat mat,PetscInt row,const PetscScalar v[])
1429: {
1431: PetscInt globalrow;
1437: ISLocalToGlobalMappingApply(mat->rmap->mapping,1,&row,&globalrow);
1438: MatSetValuesRow(mat,globalrow,v);
1439: return(0);
1440: }
1442: /*@
1443: MatSetValuesRow - Inserts a row (block row for BAIJ matrices) of nonzero
1444: values into a matrix
1446: Not Collective
1448: Input Parameters:
1449: + mat - the matrix
1450: . row - the (block) row to set
1451: - 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
1453: Notes:
1454: The values, v, are column-oriented for the block version.
1456: All the nonzeros in the row must be provided
1458: THE MATRIX MUST HAVE PREVIOUSLY HAD ITS COLUMN INDICES SET. IT IS RARE THAT THIS ROUTINE IS USED, usually MatSetValues() is used.
1460: The row must belong to this process
1462: Level: advanced
1464: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1465: InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues()
1466: @*/
1467: PetscErrorCode MatSetValuesRow(Mat mat,PetscInt row,const PetscScalar v[])
1468: {
1474: MatCheckPreallocated(mat,1);
1476: if (PetscUnlikely(mat->insertmode == ADD_VALUES)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add and insert values");
1477: if (PetscUnlikely(mat->factortype)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1478: mat->insertmode = INSERT_VALUES;
1480: if (mat->assembled) {
1481: mat->was_assembled = PETSC_TRUE;
1482: mat->assembled = PETSC_FALSE;
1483: }
1484: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
1485: if (!mat->ops->setvaluesrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1486: (*mat->ops->setvaluesrow)(mat,row,v);
1487: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
1488: return(0);
1489: }
1491: /*@
1492: MatSetValuesStencil - Inserts or adds a block of values into a matrix.
1493: Using structured grid indexing
1495: Not Collective
1497: Input Parameters:
1498: + mat - the matrix
1499: . m - number of rows being entered
1500: . idxm - grid coordinates (and component number when dof > 1) for matrix rows being entered
1501: . n - number of columns being entered
1502: . idxn - grid coordinates (and component number when dof > 1) for matrix columns being entered
1503: . v - a logically two-dimensional array of values
1504: - addv - either ADD_VALUES or INSERT_VALUES, where
1505: ADD_VALUES adds values to any existing entries, and
1506: INSERT_VALUES replaces existing entries with new values
1508: Notes:
1509: By default the values, v, are row-oriented. See MatSetOption() for other options.
1511: Calls to MatSetValuesStencil() with the INSERT_VALUES and ADD_VALUES
1512: options cannot be mixed without intervening calls to the assembly
1513: routines.
1515: The grid coordinates are across the entire grid, not just the local portion
1517: MatSetValuesStencil() uses 0-based row and column numbers in Fortran
1518: as well as in C.
1520: For setting/accessing vector values via array coordinates you can use the DMDAVecGetArray() routine
1522: In order to use this routine you must either obtain the matrix with DMCreateMatrix()
1523: or call MatSetLocalToGlobalMapping() and MatSetStencil() first.
1525: The columns and rows in the stencil passed in MUST be contained within the
1526: ghost region of the given process as set with DMDACreateXXX() or MatSetStencil(). For example,
1527: if you create a DMDA with an overlap of one grid level and on a particular process its first
1528: local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1529: first i index you can use in your column and row indices in MatSetStencil() is 5.
1531: In Fortran idxm and idxn should be declared as
1532: $ MatStencil idxm(4,m),idxn(4,n)
1533: and the values inserted using
1534: $ idxm(MatStencil_i,1) = i
1535: $ idxm(MatStencil_j,1) = j
1536: $ idxm(MatStencil_k,1) = k
1537: $ idxm(MatStencil_c,1) = c
1538: etc
1540: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
1541: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
1542: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
1543: DM_BOUNDARY_PERIODIC boundary type.
1545: 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
1546: a single value per point) you can skip filling those indices.
1548: Inspired by the structured grid interface to the HYPRE package
1549: (https://computation.llnl.gov/projects/hypre-scalable-linear-solvers-multigrid-methods)
1551: Efficiency Alert:
1552: The routine MatSetValuesBlockedStencil() may offer much better efficiency
1553: for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ).
1555: Level: beginner
1557: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1558: MatSetValues(), MatSetValuesBlockedStencil(), MatSetStencil(), DMCreateMatrix(), DMDAVecGetArray(), MatStencil
1559: @*/
1560: PetscErrorCode MatSetValuesStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv)
1561: {
1563: PetscInt buf[8192],*bufm=NULL,*bufn=NULL,*jdxm,*jdxn;
1564: PetscInt j,i,dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp;
1565: PetscInt *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1568: if (!m || !n) return(0); /* no values to insert */
1574: if ((m+n) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
1575: jdxm = buf; jdxn = buf+m;
1576: } else {
1577: PetscMalloc2(m,&bufm,n,&bufn);
1578: jdxm = bufm; jdxn = bufn;
1579: }
1580: for (i=0; i<m; i++) {
1581: for (j=0; j<3-sdim; j++) dxm++;
1582: tmp = *dxm++ - starts[0];
1583: for (j=0; j<dim-1; j++) {
1584: if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = -1;
1585: else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
1586: }
1587: if (mat->stencil.noc) dxm++;
1588: jdxm[i] = tmp;
1589: }
1590: for (i=0; i<n; i++) {
1591: for (j=0; j<3-sdim; j++) dxn++;
1592: tmp = *dxn++ - starts[0];
1593: for (j=0; j<dim-1; j++) {
1594: if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = -1;
1595: else tmp = tmp*dims[j] + *(dxn-1) - starts[j+1];
1596: }
1597: if (mat->stencil.noc) dxn++;
1598: jdxn[i] = tmp;
1599: }
1600: MatSetValuesLocal(mat,m,jdxm,n,jdxn,v,addv);
1601: PetscFree2(bufm,bufn);
1602: return(0);
1603: }
1605: /*@
1606: MatSetValuesBlockedStencil - Inserts or adds a block of values into a matrix.
1607: Using structured grid indexing
1609: Not Collective
1611: Input Parameters:
1612: + mat - the matrix
1613: . m - number of rows being entered
1614: . idxm - grid coordinates for matrix rows being entered
1615: . n - number of columns being entered
1616: . idxn - grid coordinates for matrix columns being entered
1617: . v - a logically two-dimensional array of values
1618: - addv - either ADD_VALUES or INSERT_VALUES, where
1619: ADD_VALUES adds values to any existing entries, and
1620: INSERT_VALUES replaces existing entries with new values
1622: Notes:
1623: By default the values, v, are row-oriented and unsorted.
1624: See MatSetOption() for other options.
1626: Calls to MatSetValuesBlockedStencil() with the INSERT_VALUES and ADD_VALUES
1627: options cannot be mixed without intervening calls to the assembly
1628: routines.
1630: The grid coordinates are across the entire grid, not just the local portion
1632: MatSetValuesBlockedStencil() uses 0-based row and column numbers in Fortran
1633: as well as in C.
1635: For setting/accessing vector values via array coordinates you can use the DMDAVecGetArray() routine
1637: In order to use this routine you must either obtain the matrix with DMCreateMatrix()
1638: or call MatSetBlockSize(), MatSetLocalToGlobalMapping() and MatSetStencil() first.
1640: The columns and rows in the stencil passed in MUST be contained within the
1641: ghost region of the given process as set with DMDACreateXXX() or MatSetStencil(). For example,
1642: if you create a DMDA with an overlap of one grid level and on a particular process its first
1643: local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1644: first i index you can use in your column and row indices in MatSetStencil() is 5.
1646: In Fortran idxm and idxn should be declared as
1647: $ MatStencil idxm(4,m),idxn(4,n)
1648: and the values inserted using
1649: $ idxm(MatStencil_i,1) = i
1650: $ idxm(MatStencil_j,1) = j
1651: $ idxm(MatStencil_k,1) = k
1652: etc
1654: Negative indices may be passed in idxm and idxn, these rows and columns are
1655: simply ignored. This allows easily inserting element stiffness matrices
1656: with homogeneous Dirchlet boundary conditions that you don't want represented
1657: in the matrix.
1659: Inspired by the structured grid interface to the HYPRE package
1660: (https://computation.llnl.gov/projects/hypre-scalable-linear-solvers-multigrid-methods)
1662: Level: beginner
1664: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1665: MatSetValues(), MatSetValuesStencil(), MatSetStencil(), DMCreateMatrix(), DMDAVecGetArray(), MatStencil,
1666: MatSetBlockSize(), MatSetLocalToGlobalMapping()
1667: @*/
1668: PetscErrorCode MatSetValuesBlockedStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv)
1669: {
1671: PetscInt buf[8192],*bufm=NULL,*bufn=NULL,*jdxm,*jdxn;
1672: PetscInt j,i,dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp;
1673: PetscInt *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1676: if (!m || !n) return(0); /* no values to insert */
1683: if ((m+n) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
1684: jdxm = buf; jdxn = buf+m;
1685: } else {
1686: PetscMalloc2(m,&bufm,n,&bufn);
1687: jdxm = bufm; jdxn = bufn;
1688: }
1689: for (i=0; i<m; i++) {
1690: for (j=0; j<3-sdim; j++) dxm++;
1691: tmp = *dxm++ - starts[0];
1692: for (j=0; j<sdim-1; j++) {
1693: if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = -1;
1694: else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
1695: }
1696: dxm++;
1697: jdxm[i] = tmp;
1698: }
1699: for (i=0; i<n; i++) {
1700: for (j=0; j<3-sdim; j++) dxn++;
1701: tmp = *dxn++ - starts[0];
1702: for (j=0; j<sdim-1; j++) {
1703: if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = -1;
1704: else tmp = tmp*dims[j] + *(dxn-1) - starts[j+1];
1705: }
1706: dxn++;
1707: jdxn[i] = tmp;
1708: }
1709: MatSetValuesBlockedLocal(mat,m,jdxm,n,jdxn,v,addv);
1710: PetscFree2(bufm,bufn);
1711: return(0);
1712: }
1714: /*@
1715: MatSetStencil - Sets the grid information for setting values into a matrix via
1716: MatSetValuesStencil()
1718: Not Collective
1720: Input Parameters:
1721: + mat - the matrix
1722: . dim - dimension of the grid 1, 2, or 3
1723: . dims - number of grid points in x, y, and z direction, including ghost points on your processor
1724: . starts - starting point of ghost nodes on your processor in x, y, and z direction
1725: - dof - number of degrees of freedom per node
1727: Inspired by the structured grid interface to the HYPRE package
1728: (www.llnl.gov/CASC/hyper)
1730: For matrices generated with DMCreateMatrix() this routine is automatically called and so not needed by the
1731: user.
1733: Level: beginner
1735: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1736: MatSetValues(), MatSetValuesBlockedStencil(), MatSetValuesStencil()
1737: @*/
1738: PetscErrorCode MatSetStencil(Mat mat,PetscInt dim,const PetscInt dims[],const PetscInt starts[],PetscInt dof)
1739: {
1740: PetscInt i;
1747: mat->stencil.dim = dim + (dof > 1);
1748: for (i=0; i<dim; i++) {
1749: mat->stencil.dims[i] = dims[dim-i-1]; /* copy the values in backwards */
1750: mat->stencil.starts[i] = starts[dim-i-1];
1751: }
1752: mat->stencil.dims[dim] = dof;
1753: mat->stencil.starts[dim] = 0;
1754: mat->stencil.noc = (PetscBool)(dof == 1);
1755: return(0);
1756: }
1758: /*@C
1759: MatSetValuesBlocked - Inserts or adds a block of values into a matrix.
1761: Not Collective
1763: Input Parameters:
1764: + mat - the matrix
1765: . v - a logically two-dimensional array of values
1766: . m, idxm - the number of block rows and their global block indices
1767: . n, idxn - the number of block columns and their global block indices
1768: - addv - either ADD_VALUES or INSERT_VALUES, where
1769: ADD_VALUES adds values to any existing entries, and
1770: INSERT_VALUES replaces existing entries with new values
1772: Notes:
1773: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call
1774: MatXXXXSetPreallocation() or MatSetUp() before using this routine.
1776: The m and n count the NUMBER of blocks in the row direction and column direction,
1777: NOT the total number of rows/columns; for example, if the block size is 2 and
1778: you are passing in values for rows 2,3,4,5 then m would be 2 (not 4).
1779: The values in idxm would be 1 2; that is the first index for each block divided by
1780: the block size.
1782: Note that you must call MatSetBlockSize() when constructing this matrix (before
1783: preallocating it).
1785: By default the values, v, are row-oriented, so the layout of
1786: v is the same as for MatSetValues(). See MatSetOption() for other options.
1788: Calls to MatSetValuesBlocked() with the INSERT_VALUES and ADD_VALUES
1789: options cannot be mixed without intervening calls to the assembly
1790: routines.
1792: MatSetValuesBlocked() uses 0-based row and column numbers in Fortran
1793: as well as in C.
1795: Negative indices may be passed in idxm and idxn, these rows and columns are
1796: simply ignored. This allows easily inserting element stiffness matrices
1797: with homogeneous Dirchlet boundary conditions that you don't want represented
1798: in the matrix.
1800: Each time an entry is set within a sparse matrix via MatSetValues(),
1801: internal searching must be done to determine where to place the
1802: data in the matrix storage space. By instead inserting blocks of
1803: entries via MatSetValuesBlocked(), the overhead of matrix assembly is
1804: reduced.
1806: Example:
1807: $ Suppose m=n=2 and block size(bs) = 2 The array is
1808: $
1809: $ 1 2 | 3 4
1810: $ 5 6 | 7 8
1811: $ - - - | - - -
1812: $ 9 10 | 11 12
1813: $ 13 14 | 15 16
1814: $
1815: $ v[] should be passed in like
1816: $ v[] = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]
1817: $
1818: $ If you are not using row oriented storage of v (that is you called MatSetOption(mat,MAT_ROW_ORIENTED,PETSC_FALSE)) then
1819: $ v[] = [1,5,9,13,2,6,10,14,3,7,11,15,4,8,12,16]
1821: Level: intermediate
1823: .seealso: MatSetBlockSize(), MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal()
1824: @*/
1825: PetscErrorCode MatSetValuesBlocked(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv)
1826: {
1832: if (!m || !n) return(0); /* no values to insert */
1836: MatCheckPreallocated(mat,1);
1837: if (mat->insertmode == NOT_SET_VALUES) {
1838: mat->insertmode = addv;
1839: } else if (PetscUnlikely(mat->insertmode != addv)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1840: if (PetscDefined(USE_DEBUG)) {
1841: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1842: if (!mat->ops->setvaluesblocked && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1843: }
1844: if (PetscDefined(USE_DEBUG)) {
1845: PetscInt rbs,cbs,M,N,i;
1846: MatGetBlockSizes(mat,&rbs,&cbs);
1847: MatGetSize(mat,&M,&N);
1848: for (i=0; i<m; i++) {
1849: if (idxm[i]*rbs >= M) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row block index %D (index %D) greater than row length %D",i,idxm[i],M);
1850: }
1851: for (i=0; i<n; i++) {
1852: if (idxn[i]*cbs >= N) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column block index %D (index %D) great than column length %D",i,idxn[i],N);
1853: }
1854: }
1855: if (mat->assembled) {
1856: mat->was_assembled = PETSC_TRUE;
1857: mat->assembled = PETSC_FALSE;
1858: }
1859: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
1860: if (mat->ops->setvaluesblocked) {
1861: (*mat->ops->setvaluesblocked)(mat,m,idxm,n,idxn,v,addv);
1862: } else {
1863: PetscInt buf[8192],*bufr=NULL,*bufc=NULL,*iidxm,*iidxn;
1864: PetscInt i,j,bs,cbs;
1865: MatGetBlockSizes(mat,&bs,&cbs);
1866: if (m*bs+n*cbs <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
1867: iidxm = buf; iidxn = buf + m*bs;
1868: } else {
1869: PetscMalloc2(m*bs,&bufr,n*cbs,&bufc);
1870: iidxm = bufr; iidxn = bufc;
1871: }
1872: for (i=0; i<m; i++) {
1873: for (j=0; j<bs; j++) {
1874: iidxm[i*bs+j] = bs*idxm[i] + j;
1875: }
1876: }
1877: for (i=0; i<n; i++) {
1878: for (j=0; j<cbs; j++) {
1879: iidxn[i*cbs+j] = cbs*idxn[i] + j;
1880: }
1881: }
1882: MatSetValues(mat,m*bs,iidxm,n*cbs,iidxn,v,addv);
1883: PetscFree2(bufr,bufc);
1884: }
1885: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
1886: return(0);
1887: }
1889: /*@C
1890: MatGetValues - Gets a block of values from a matrix.
1892: Not Collective; can only return values that are owned by the give process
1894: Input Parameters:
1895: + mat - the matrix
1896: . v - a logically two-dimensional array for storing the values
1897: . m, idxm - the number of rows and their global indices
1898: - n, idxn - the number of columns and their global indices
1900: Notes:
1901: The user must allocate space (m*n PetscScalars) for the values, v.
1902: The values, v, are then returned in a row-oriented format,
1903: analogous to that used by default in MatSetValues().
1905: MatGetValues() uses 0-based row and column numbers in
1906: Fortran as well as in C.
1908: MatGetValues() requires that the matrix has been assembled
1909: with MatAssemblyBegin()/MatAssemblyEnd(). Thus, calls to
1910: MatSetValues() and MatGetValues() CANNOT be made in succession
1911: without intermediate matrix assembly.
1913: Negative row or column indices will be ignored and those locations in v[] will be
1914: left unchanged.
1916: For the standard row-based matrix formats, idxm[] can only contain rows owned by the requesting MPI rank.
1917: That is, rows with global index greater than or equal to restart and less than rend where restart and rend are obtainable
1918: from MatGetOwnershipRange(mat,&rstart,&rend).
1920: Level: advanced
1922: .seealso: MatGetRow(), MatCreateSubMatrices(), MatSetValues(), MatGetOwnershipRange(), MatGetValuesLocal()
1923: @*/
1924: PetscErrorCode MatGetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[])
1925: {
1931: if (!m || !n) return(0);
1935: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
1936: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1937: if (!mat->ops->getvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1938: MatCheckPreallocated(mat,1);
1940: PetscLogEventBegin(MAT_GetValues,mat,0,0,0);
1941: (*mat->ops->getvalues)(mat,m,idxm,n,idxn,v);
1942: PetscLogEventEnd(MAT_GetValues,mat,0,0,0);
1943: return(0);
1944: }
1946: /*@C
1947: MatGetValuesLocal - retrieves values from certain locations in a matrix using the local numbering of the indices
1948: defined previously by MatSetLocalToGlobalMapping()
1950: Not Collective
1952: Input Parameters:
1953: + mat - the matrix
1954: . nrow, irow - number of rows and their local indices
1955: - ncol, icol - number of columns and their local indices
1957: Output Parameter:
1958: . y - a logically two-dimensional array of values
1960: Notes:
1961: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatSetLocalToGlobalMapping() before using this routine.
1963: 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,
1964: are greater than or equal to restart and less than rend where restart and rend are obtainable from MatGetOwnershipRange(mat,&rstart,&rend). One can
1965: determine if the resulting global row associated with the local row r is owned by the requesting MPI rank by applying the ISLocalToGlobalMapping set
1966: with MatSetLocalToGlobalMapping().
1968: Developer Notes:
1969: This is labelled with C so does not automatically generate Fortran stubs and interfaces
1970: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
1972: Level: advanced
1974: .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetLocalToGlobalMapping(),
1975: MatSetValuesLocal(), MatGetValues()
1976: @*/
1977: PetscErrorCode MatGetValuesLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],PetscScalar y[])
1978: {
1984: MatCheckPreallocated(mat,1);
1985: if (!nrow || !ncol) return(0); /* no values to retrieve */
1988: if (PetscDefined(USE_DEBUG)) {
1989: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1990: if (!mat->ops->getvalueslocal && !mat->ops->getvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1991: }
1992: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
1993: PetscLogEventBegin(MAT_GetValues,mat,0,0,0);
1994: if (mat->ops->getvalueslocal) {
1995: (*mat->ops->getvalueslocal)(mat,nrow,irow,ncol,icol,y);
1996: } else {
1997: PetscInt buf[8192],*bufr=NULL,*bufc=NULL,*irowm,*icolm;
1998: if ((nrow+ncol) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
1999: irowm = buf; icolm = buf+nrow;
2000: } else {
2001: PetscMalloc2(nrow,&bufr,ncol,&bufc);
2002: irowm = bufr; icolm = bufc;
2003: }
2004: if (!mat->rmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"MatGetValuesLocal() cannot proceed without local-to-global row mapping (See MatSetLocalToGlobalMapping()).");
2005: if (!mat->cmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"MatGetValuesLocal() cannot proceed without local-to-global column mapping (See MatSetLocalToGlobalMapping()).");
2006: ISLocalToGlobalMappingApply(mat->rmap->mapping,nrow,irow,irowm);
2007: ISLocalToGlobalMappingApply(mat->cmap->mapping,ncol,icol,icolm);
2008: MatGetValues(mat,nrow,irowm,ncol,icolm,y);
2009: PetscFree2(bufr,bufc);
2010: }
2011: PetscLogEventEnd(MAT_GetValues,mat,0,0,0);
2012: return(0);
2013: }
2015: /*@
2016: MatSetValuesBatch - Adds (ADD_VALUES) many blocks of values into a matrix at once. The blocks must all be square and
2017: the same size. Currently, this can only be called once and creates the given matrix.
2019: Not Collective
2021: Input Parameters:
2022: + mat - the matrix
2023: . nb - the number of blocks
2024: . bs - the number of rows (and columns) in each block
2025: . rows - a concatenation of the rows for each block
2026: - v - a concatenation of logically two-dimensional arrays of values
2028: Notes:
2029: In the future, we will extend this routine to handle rectangular blocks, and to allow multiple calls for a given matrix.
2031: Level: advanced
2033: .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
2034: InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues()
2035: @*/
2036: PetscErrorCode MatSetValuesBatch(Mat mat, PetscInt nb, PetscInt bs, PetscInt rows[], const PetscScalar v[])
2037: {
2045: if (PetscUnlikelyDebug(mat->factortype)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2047: PetscLogEventBegin(MAT_SetValuesBatch,mat,0,0,0);
2048: if (mat->ops->setvaluesbatch) {
2049: (*mat->ops->setvaluesbatch)(mat,nb,bs,rows,v);
2050: } else {
2051: PetscInt b;
2052: for (b = 0; b < nb; ++b) {
2053: MatSetValues(mat, bs, &rows[b*bs], bs, &rows[b*bs], &v[b*bs*bs], ADD_VALUES);
2054: }
2055: }
2056: PetscLogEventEnd(MAT_SetValuesBatch,mat,0,0,0);
2057: return(0);
2058: }
2060: /*@
2061: MatSetLocalToGlobalMapping - Sets a local-to-global numbering for use by
2062: the routine MatSetValuesLocal() to allow users to insert matrix entries
2063: using a local (per-processor) numbering.
2065: Not Collective
2067: Input Parameters:
2068: + x - the matrix
2069: . rmapping - row mapping created with ISLocalToGlobalMappingCreate() or ISLocalToGlobalMappingCreateIS()
2070: - cmapping - column mapping
2072: Level: intermediate
2074: .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesLocal(), MatGetValuesLocal()
2075: @*/
2076: PetscErrorCode MatSetLocalToGlobalMapping(Mat x,ISLocalToGlobalMapping rmapping,ISLocalToGlobalMapping cmapping)
2077: {
2086: if (x->ops->setlocaltoglobalmapping) {
2087: (*x->ops->setlocaltoglobalmapping)(x,rmapping,cmapping);
2088: } else {
2089: PetscLayoutSetISLocalToGlobalMapping(x->rmap,rmapping);
2090: PetscLayoutSetISLocalToGlobalMapping(x->cmap,cmapping);
2091: }
2092: return(0);
2093: }
2095: /*@
2096: MatGetLocalToGlobalMapping - Gets the local-to-global numbering set by MatSetLocalToGlobalMapping()
2098: Not Collective
2100: Input Parameter:
2101: . A - the matrix
2103: Output Parameters:
2104: + rmapping - row mapping
2105: - cmapping - column mapping
2107: Level: advanced
2109: .seealso: MatSetValuesLocal()
2110: @*/
2111: PetscErrorCode MatGetLocalToGlobalMapping(Mat A,ISLocalToGlobalMapping *rmapping,ISLocalToGlobalMapping *cmapping)
2112: {
2118: if (rmapping) *rmapping = A->rmap->mapping;
2119: if (cmapping) *cmapping = A->cmap->mapping;
2120: return(0);
2121: }
2123: /*@
2124: MatSetLayouts - Sets the PetscLayout objects for rows and columns of a matrix
2126: Logically Collective on A
2128: Input Parameters:
2129: + A - the matrix
2130: . rmap - row layout
2131: - cmap - column layout
2133: Level: advanced
2135: .seealso: MatCreateVecs(), MatGetLocalToGlobalMapping(), MatGetLayouts()
2136: @*/
2137: PetscErrorCode MatSetLayouts(Mat A,PetscLayout rmap,PetscLayout cmap)
2138: {
2144: PetscLayoutReference(rmap,&A->rmap);
2145: PetscLayoutReference(cmap,&A->cmap);
2146: return(0);
2147: }
2149: /*@
2150: MatGetLayouts - Gets the PetscLayout objects for rows and columns
2152: Not Collective
2154: Input Parameter:
2155: . A - the matrix
2157: Output Parameters:
2158: + rmap - row layout
2159: - cmap - column layout
2161: Level: advanced
2163: .seealso: MatCreateVecs(), MatGetLocalToGlobalMapping(), MatSetLayouts()
2164: @*/
2165: PetscErrorCode MatGetLayouts(Mat A,PetscLayout *rmap,PetscLayout *cmap)
2166: {
2172: if (rmap) *rmap = A->rmap;
2173: if (cmap) *cmap = A->cmap;
2174: return(0);
2175: }
2177: /*@C
2178: MatSetValuesLocal - Inserts or adds values into certain locations of a matrix,
2179: using a local numbering of the nodes.
2181: Not Collective
2183: Input Parameters:
2184: + mat - the matrix
2185: . nrow, irow - number of rows and their local indices
2186: . ncol, icol - number of columns and their local indices
2187: . y - a logically two-dimensional array of values
2188: - addv - either INSERT_VALUES or ADD_VALUES, where
2189: ADD_VALUES adds values to any existing entries, and
2190: INSERT_VALUES replaces existing entries with new values
2192: Notes:
2193: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or
2194: MatSetUp() before using this routine
2196: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatSetLocalToGlobalMapping() before using this routine
2198: Calls to MatSetValuesLocal() with the INSERT_VALUES and ADD_VALUES
2199: options cannot be mixed without intervening calls to the assembly
2200: routines.
2202: These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
2203: MUST be called after all calls to MatSetValuesLocal() have been completed.
2205: Level: intermediate
2207: Developer Notes:
2208: This is labeled with C so does not automatically generate Fortran stubs and interfaces
2209: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
2211: .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetLocalToGlobalMapping(),
2212: MatSetValueLocal(), MatGetValuesLocal()
2213: @*/
2214: PetscErrorCode MatSetValuesLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv)
2215: {
2221: MatCheckPreallocated(mat,1);
2222: if (!nrow || !ncol) return(0); /* no values to insert */
2225: if (mat->insertmode == NOT_SET_VALUES) {
2226: mat->insertmode = addv;
2227: }
2228: else if (PetscUnlikely(mat->insertmode != addv)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
2229: if (PetscDefined(USE_DEBUG)) {
2230: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2231: if (!mat->ops->setvalueslocal && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2232: }
2234: if (mat->assembled) {
2235: mat->was_assembled = PETSC_TRUE;
2236: mat->assembled = PETSC_FALSE;
2237: }
2238: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
2239: if (mat->ops->setvalueslocal) {
2240: (*mat->ops->setvalueslocal)(mat,nrow,irow,ncol,icol,y,addv);
2241: } else {
2242: PetscInt buf[8192],*bufr=NULL,*bufc=NULL,*irowm,*icolm;
2243: if ((nrow+ncol) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
2244: irowm = buf; icolm = buf+nrow;
2245: } else {
2246: PetscMalloc2(nrow,&bufr,ncol,&bufc);
2247: irowm = bufr; icolm = bufc;
2248: }
2249: if (!mat->rmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"MatSetValuesLocal() cannot proceed without local-to-global row mapping (See MatSetLocalToGlobalMapping()).");
2250: if (!mat->cmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"MatSetValuesLocal() cannot proceed without local-to-global column mapping (See MatSetLocalToGlobalMapping()).");
2251: ISLocalToGlobalMappingApply(mat->rmap->mapping,nrow,irow,irowm);
2252: ISLocalToGlobalMappingApply(mat->cmap->mapping,ncol,icol,icolm);
2253: MatSetValues(mat,nrow,irowm,ncol,icolm,y,addv);
2254: PetscFree2(bufr,bufc);
2255: }
2256: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
2257: return(0);
2258: }
2260: /*@C
2261: MatSetValuesBlockedLocal - Inserts or adds values into certain locations of a matrix,
2262: using a local ordering of the nodes a block at a time.
2264: Not Collective
2266: Input Parameters:
2267: + x - the matrix
2268: . nrow, irow - number of rows and their local indices
2269: . ncol, icol - number of columns and their local indices
2270: . y - a logically two-dimensional array of values
2271: - addv - either INSERT_VALUES or ADD_VALUES, where
2272: ADD_VALUES adds values to any existing entries, and
2273: INSERT_VALUES replaces existing entries with new values
2275: Notes:
2276: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or
2277: MatSetUp() before using this routine
2279: If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatSetBlockSize() and MatSetLocalToGlobalMapping()
2280: before using this routineBefore calling MatSetValuesLocal(), the user must first set the
2282: Calls to MatSetValuesBlockedLocal() with the INSERT_VALUES and ADD_VALUES
2283: options cannot be mixed without intervening calls to the assembly
2284: routines.
2286: These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
2287: MUST be called after all calls to MatSetValuesBlockedLocal() have been completed.
2289: Level: intermediate
2291: Developer Notes:
2292: This is labeled with C so does not automatically generate Fortran stubs and interfaces
2293: because it requires multiple Fortran interfaces depending on which arguments are scalar or arrays.
2295: .seealso: MatSetBlockSize(), MatSetLocalToGlobalMapping(), MatAssemblyBegin(), MatAssemblyEnd(),
2296: MatSetValuesLocal(), MatSetValuesBlocked()
2297: @*/
2298: PetscErrorCode MatSetValuesBlockedLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv)
2299: {
2305: MatCheckPreallocated(mat,1);
2306: if (!nrow || !ncol) return(0); /* no values to insert */
2310: if (mat->insertmode == NOT_SET_VALUES) {
2311: mat->insertmode = addv;
2312: } else if (PetscUnlikely(mat->insertmode != addv)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
2313: if (PetscDefined(USE_DEBUG)) {
2314: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2315: if (!mat->ops->setvaluesblockedlocal && !mat->ops->setvaluesblocked && !mat->ops->setvalueslocal && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2316: }
2318: if (mat->assembled) {
2319: mat->was_assembled = PETSC_TRUE;
2320: mat->assembled = PETSC_FALSE;
2321: }
2322: if (PetscUnlikelyDebug(mat->rmap->mapping)) { /* Condition on the mapping existing, because MatSetValuesBlockedLocal_IS does not require it to be set. */
2323: PetscInt irbs, rbs;
2324: MatGetBlockSizes(mat, &rbs, NULL);
2325: ISLocalToGlobalMappingGetBlockSize(mat->rmap->mapping,&irbs);
2326: if (rbs != irbs) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Different row block sizes! mat %D, row l2g map %D",rbs,irbs);
2327: }
2328: if (PetscUnlikelyDebug(mat->cmap->mapping)) {
2329: PetscInt icbs, cbs;
2330: MatGetBlockSizes(mat,NULL,&cbs);
2331: ISLocalToGlobalMappingGetBlockSize(mat->cmap->mapping,&icbs);
2332: if (cbs != icbs) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Different col block sizes! mat %D, col l2g map %D",cbs,icbs);
2333: }
2334: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
2335: if (mat->ops->setvaluesblockedlocal) {
2336: (*mat->ops->setvaluesblockedlocal)(mat,nrow,irow,ncol,icol,y,addv);
2337: } else {
2338: PetscInt buf[8192],*bufr=NULL,*bufc=NULL,*irowm,*icolm;
2339: if ((nrow+ncol) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) {
2340: irowm = buf; icolm = buf + nrow;
2341: } else {
2342: PetscMalloc2(nrow,&bufr,ncol,&bufc);
2343: irowm = bufr; icolm = bufc;
2344: }
2345: ISLocalToGlobalMappingApplyBlock(mat->rmap->mapping,nrow,irow,irowm);
2346: ISLocalToGlobalMappingApplyBlock(mat->cmap->mapping,ncol,icol,icolm);
2347: MatSetValuesBlocked(mat,nrow,irowm,ncol,icolm,y,addv);
2348: PetscFree2(bufr,bufc);
2349: }
2350: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
2351: return(0);
2352: }
2354: /*@
2355: MatMultDiagonalBlock - Computes the matrix-vector product, y = Dx. Where D is defined by the inode or block structure of the diagonal
2357: Collective on Mat
2359: Input Parameters:
2360: + mat - the matrix
2361: - x - the vector to be multiplied
2363: Output Parameters:
2364: . y - the result
2366: Notes:
2367: The vectors x and y cannot be the same. I.e., one cannot
2368: call MatMult(A,y,y).
2370: Level: developer
2372: .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2373: @*/
2374: PetscErrorCode MatMultDiagonalBlock(Mat mat,Vec x,Vec y)
2375: {
2384: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2385: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2386: if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2387: MatCheckPreallocated(mat,1);
2389: if (!mat->ops->multdiagonalblock) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s does not have a multiply defined",((PetscObject)mat)->type_name);
2390: (*mat->ops->multdiagonalblock)(mat,x,y);
2391: PetscObjectStateIncrease((PetscObject)y);
2392: return(0);
2393: }
2395: /* --------------------------------------------------------*/
2396: /*@
2397: MatMult - Computes the matrix-vector product, y = Ax.
2399: Neighbor-wise Collective on Mat
2401: Input Parameters:
2402: + mat - the matrix
2403: - x - the vector to be multiplied
2405: Output Parameters:
2406: . y - the result
2408: Notes:
2409: The vectors x and y cannot be the same. I.e., one cannot
2410: call MatMult(A,y,y).
2412: Level: beginner
2414: .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2415: @*/
2416: PetscErrorCode MatMult(Mat mat,Vec x,Vec y)
2417: {
2425: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2426: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2427: if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2428: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2429: if (mat->rmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
2430: if (mat->cmap->n != x->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: local dim %D %D",mat->cmap->n,x->map->n);
2431: if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n);
2432: VecSetErrorIfLocked(y,3);
2433: if (mat->erroriffailure) {VecValidValues(x,2,PETSC_TRUE);}
2434: MatCheckPreallocated(mat,1);
2436: VecLockReadPush(x);
2437: if (!mat->ops->mult) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s does not have a multiply defined",((PetscObject)mat)->type_name);
2438: PetscLogEventBegin(MAT_Mult,mat,x,y,0);
2439: (*mat->ops->mult)(mat,x,y);
2440: PetscLogEventEnd(MAT_Mult,mat,x,y,0);
2441: if (mat->erroriffailure) {VecValidValues(y,3,PETSC_FALSE);}
2442: VecLockReadPop(x);
2443: return(0);
2444: }
2446: /*@
2447: MatMultTranspose - Computes matrix transpose times a vector y = A^T * x.
2449: Neighbor-wise Collective on Mat
2451: Input Parameters:
2452: + mat - the matrix
2453: - x - the vector to be multiplied
2455: Output Parameters:
2456: . y - the result
2458: Notes:
2459: The vectors x and y cannot be the same. I.e., one cannot
2460: call MatMultTranspose(A,y,y).
2462: For complex numbers this does NOT compute the Hermitian (complex conjugate) transpose multiple,
2463: use MatMultHermitianTranspose()
2465: Level: beginner
2467: .seealso: MatMult(), MatMultAdd(), MatMultTransposeAdd(), MatMultHermitianTranspose(), MatTranspose()
2468: @*/
2469: PetscErrorCode MatMultTranspose(Mat mat,Vec x,Vec y)
2470: {
2471: PetscErrorCode (*op)(Mat,Vec,Vec)=NULL,ierr;
2479: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2480: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2481: if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2482: if (mat->cmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
2483: if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
2484: if (mat->cmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->cmap->n,y->map->n);
2485: if (mat->rmap->n != x->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: local dim %D %D",mat->rmap->n,x->map->n);
2486: if (mat->erroriffailure) {VecValidValues(x,2,PETSC_TRUE);}
2487: MatCheckPreallocated(mat,1);
2489: if (!mat->ops->multtranspose) {
2490: if (mat->symmetric && mat->ops->mult) op = mat->ops->mult;
2491: if (!op) SETERRQ1(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);
2492: } else op = mat->ops->multtranspose;
2493: PetscLogEventBegin(MAT_MultTranspose,mat,x,y,0);
2494: VecLockReadPush(x);
2495: (*op)(mat,x,y);
2496: VecLockReadPop(x);
2497: PetscLogEventEnd(MAT_MultTranspose,mat,x,y,0);
2498: PetscObjectStateIncrease((PetscObject)y);
2499: if (mat->erroriffailure) {VecValidValues(y,3,PETSC_FALSE);}
2500: return(0);
2501: }
2503: /*@
2504: MatMultHermitianTranspose - Computes matrix Hermitian transpose times a vector.
2506: Neighbor-wise Collective on Mat
2508: Input Parameters:
2509: + mat - the matrix
2510: - x - the vector to be multilplied
2512: Output Parameters:
2513: . y - the result
2515: Notes:
2516: The vectors x and y cannot be the same. I.e., one cannot
2517: call MatMultHermitianTranspose(A,y,y).
2519: Also called the conjugate transpose, complex conjugate transpose, or adjoint.
2521: For real numbers MatMultTranspose() and MatMultHermitianTranspose() are identical.
2523: Level: beginner
2525: .seealso: MatMult(), MatMultAdd(), MatMultHermitianTransposeAdd(), MatMultTranspose()
2526: @*/
2527: PetscErrorCode MatMultHermitianTranspose(Mat mat,Vec x,Vec y)
2528: {
2537: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2538: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2539: if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2540: if (mat->cmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
2541: if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
2542: if (mat->cmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->cmap->n,y->map->n);
2543: if (mat->rmap->n != x->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: local dim %D %D",mat->rmap->n,x->map->n);
2544: MatCheckPreallocated(mat,1);
2546: PetscLogEventBegin(MAT_MultHermitianTranspose,mat,x,y,0);
2547: #if defined(PETSC_USE_COMPLEX)
2548: if (mat->ops->multhermitiantranspose || (mat->hermitian && mat->ops->mult)) {
2549: VecLockReadPush(x);
2550: if (mat->ops->multhermitiantranspose) {
2551: (*mat->ops->multhermitiantranspose)(mat,x,y);
2552: } else {
2553: (*mat->ops->mult)(mat,x,y);
2554: }
2555: VecLockReadPop(x);
2556: } else {
2557: Vec w;
2558: VecDuplicate(x,&w);
2559: VecCopy(x,w);
2560: VecConjugate(w);
2561: MatMultTranspose(mat,w,y);
2562: VecDestroy(&w);
2563: VecConjugate(y);
2564: }
2565: PetscObjectStateIncrease((PetscObject)y);
2566: #else
2567: MatMultTranspose(mat,x,y);
2568: #endif
2569: PetscLogEventEnd(MAT_MultHermitianTranspose,mat,x,y,0);
2570: return(0);
2571: }
2573: /*@
2574: MatMultAdd - Computes v3 = v2 + A * v1.
2576: Neighbor-wise Collective on Mat
2578: Input Parameters:
2579: + mat - the matrix
2580: - v1, v2 - the vectors
2582: Output Parameters:
2583: . v3 - the result
2585: Notes:
2586: The vectors v1 and v3 cannot be the same. I.e., one cannot
2587: call MatMultAdd(A,v1,v2,v1).
2589: Level: beginner
2591: .seealso: MatMultTranspose(), MatMult(), MatMultTransposeAdd()
2592: @*/
2593: PetscErrorCode MatMultAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2594: {
2604: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2605: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2606: if (mat->cmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->cmap->N,v1->map->N);
2607: /* if (mat->rmap->N != v2->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->rmap->N,v2->map->N);
2608: if (mat->rmap->N != v3->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->rmap->N,v3->map->N); */
2609: if (mat->rmap->n != v3->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: local dim %D %D",mat->rmap->n,v3->map->n);
2610: if (mat->rmap->n != v2->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: local dim %D %D",mat->rmap->n,v2->map->n);
2611: if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2612: MatCheckPreallocated(mat,1);
2614: if (!mat->ops->multadd) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"No MatMultAdd() for matrix type %s",((PetscObject)mat)->type_name);
2615: PetscLogEventBegin(MAT_MultAdd,mat,v1,v2,v3);
2616: VecLockReadPush(v1);
2617: (*mat->ops->multadd)(mat,v1,v2,v3);
2618: VecLockReadPop(v1);
2619: PetscLogEventEnd(MAT_MultAdd,mat,v1,v2,v3);
2620: PetscObjectStateIncrease((PetscObject)v3);
2621: return(0);
2622: }
2624: /*@
2625: MatMultTransposeAdd - Computes v3 = v2 + A' * v1.
2627: Neighbor-wise Collective on Mat
2629: Input Parameters:
2630: + mat - the matrix
2631: - v1, v2 - the vectors
2633: Output Parameters:
2634: . v3 - the result
2636: Notes:
2637: The vectors v1 and v3 cannot be the same. I.e., one cannot
2638: call MatMultTransposeAdd(A,v1,v2,v1).
2640: Level: beginner
2642: .seealso: MatMultTranspose(), MatMultAdd(), MatMult()
2643: @*/
2644: PetscErrorCode MatMultTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2645: {
2655: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2656: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2657: if (!mat->ops->multtransposeadd) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2658: if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2659: if (mat->rmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N);
2660: if (mat->cmap->N != v2->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N);
2661: if (mat->cmap->N != v3->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N);
2662: MatCheckPreallocated(mat,1);
2664: PetscLogEventBegin(MAT_MultTransposeAdd,mat,v1,v2,v3);
2665: VecLockReadPush(v1);
2666: (*mat->ops->multtransposeadd)(mat,v1,v2,v3);
2667: VecLockReadPop(v1);
2668: PetscLogEventEnd(MAT_MultTransposeAdd,mat,v1,v2,v3);
2669: PetscObjectStateIncrease((PetscObject)v3);
2670: return(0);
2671: }
2673: /*@
2674: MatMultHermitianTransposeAdd - Computes v3 = v2 + A^H * v1.
2676: Neighbor-wise Collective on Mat
2678: Input Parameters:
2679: + mat - the matrix
2680: - v1, v2 - the vectors
2682: Output Parameters:
2683: . v3 - the result
2685: Notes:
2686: The vectors v1 and v3 cannot be the same. I.e., one cannot
2687: call MatMultHermitianTransposeAdd(A,v1,v2,v1).
2689: Level: beginner
2691: .seealso: MatMultHermitianTranspose(), MatMultTranspose(), MatMultAdd(), MatMult()
2692: @*/
2693: PetscErrorCode MatMultHermitianTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2694: {
2704: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2705: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2706: if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2707: if (mat->rmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N);
2708: if (mat->cmap->N != v2->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N);
2709: if (mat->cmap->N != v3->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N);
2710: MatCheckPreallocated(mat,1);
2712: PetscLogEventBegin(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);
2713: VecLockReadPush(v1);
2714: if (mat->ops->multhermitiantransposeadd) {
2715: (*mat->ops->multhermitiantransposeadd)(mat,v1,v2,v3);
2716: } else {
2717: Vec w,z;
2718: VecDuplicate(v1,&w);
2719: VecCopy(v1,w);
2720: VecConjugate(w);
2721: VecDuplicate(v3,&z);
2722: MatMultTranspose(mat,w,z);
2723: VecDestroy(&w);
2724: VecConjugate(z);
2725: if (v2 != v3) {
2726: VecWAXPY(v3,1.0,v2,z);
2727: } else {
2728: VecAXPY(v3,1.0,z);
2729: }
2730: VecDestroy(&z);
2731: }
2732: VecLockReadPop(v1);
2733: PetscLogEventEnd(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);
2734: PetscObjectStateIncrease((PetscObject)v3);
2735: return(0);
2736: }
2738: /*@
2739: MatMultConstrained - The inner multiplication routine for a
2740: constrained matrix P^T A P.
2742: Neighbor-wise Collective on Mat
2744: Input Parameters:
2745: + mat - the matrix
2746: - x - the vector to be multilplied
2748: Output Parameters:
2749: . y - the result
2751: Notes:
2752: The vectors x and y cannot be the same. I.e., one cannot
2753: call MatMult(A,y,y).
2755: Level: beginner
2757: .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2758: @*/
2759: PetscErrorCode MatMultConstrained(Mat mat,Vec x,Vec y)
2760: {
2767: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2768: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2769: if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2770: if (mat->cmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2771: if (mat->rmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
2772: if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n);
2774: PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);
2775: VecLockReadPush(x);
2776: (*mat->ops->multconstrained)(mat,x,y);
2777: VecLockReadPop(x);
2778: PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);
2779: PetscObjectStateIncrease((PetscObject)y);
2780: return(0);
2781: }
2783: /*@
2784: MatMultTransposeConstrained - The inner multiplication routine for a
2785: constrained matrix P^T A^T P.
2787: Neighbor-wise Collective on Mat
2789: Input Parameters:
2790: + mat - the matrix
2791: - x - the vector to be multilplied
2793: Output Parameters:
2794: . y - the result
2796: Notes:
2797: The vectors x and y cannot be the same. I.e., one cannot
2798: call MatMult(A,y,y).
2800: Level: beginner
2802: .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2803: @*/
2804: PetscErrorCode MatMultTransposeConstrained(Mat mat,Vec x,Vec y)
2805: {
2812: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2813: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2814: if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2815: if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2816: if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
2818: PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);
2819: (*mat->ops->multtransposeconstrained)(mat,x,y);
2820: PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);
2821: PetscObjectStateIncrease((PetscObject)y);
2822: return(0);
2823: }
2825: /*@C
2826: MatGetFactorType - gets the type of factorization it is
2828: Not Collective
2830: Input Parameters:
2831: . mat - the matrix
2833: Output Parameters:
2834: . t - the type, one of MAT_FACTOR_NONE, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ILU, MAT_FACTOR_ICC,MAT_FACTOR_ILUDT
2836: Level: intermediate
2838: .seealso: MatFactorType, MatGetFactor(), MatSetFactorType()
2839: @*/
2840: PetscErrorCode MatGetFactorType(Mat mat,MatFactorType *t)
2841: {
2846: *t = mat->factortype;
2847: return(0);
2848: }
2850: /*@C
2851: MatSetFactorType - sets the type of factorization it is
2853: Logically Collective on Mat
2855: Input Parameters:
2856: + mat - the matrix
2857: - t - the type, one of MAT_FACTOR_NONE, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ILU, MAT_FACTOR_ICC,MAT_FACTOR_ILUDT
2859: Level: intermediate
2861: .seealso: MatFactorType, MatGetFactor(), MatGetFactorType()
2862: @*/
2863: PetscErrorCode MatSetFactorType(Mat mat, MatFactorType t)
2864: {
2868: mat->factortype = t;
2869: return(0);
2870: }
2872: /* ------------------------------------------------------------*/
2873: /*@C
2874: MatGetInfo - Returns information about matrix storage (number of
2875: nonzeros, memory, etc.).
2877: Collective on Mat if MAT_GLOBAL_MAX or MAT_GLOBAL_SUM is used as the flag
2879: Input Parameter:
2880: . mat - the matrix
2882: Output Parameters:
2883: + flag - flag indicating the type of parameters to be returned
2884: (MAT_LOCAL - local matrix, MAT_GLOBAL_MAX - maximum over all processors,
2885: MAT_GLOBAL_SUM - sum over all processors)
2886: - info - matrix information context
2888: Notes:
2889: The MatInfo context contains a variety of matrix data, including
2890: number of nonzeros allocated and used, number of mallocs during
2891: matrix assembly, etc. Additional information for factored matrices
2892: is provided (such as the fill ratio, number of mallocs during
2893: factorization, etc.). Much of this info is printed to PETSC_STDOUT
2894: when using the runtime options
2895: $ -info -mat_view ::ascii_info
2897: Example for C/C++ Users:
2898: See the file ${PETSC_DIR}/include/petscmat.h for a complete list of
2899: data within the MatInfo context. For example,
2900: .vb
2901: MatInfo info;
2902: Mat A;
2903: double mal, nz_a, nz_u;
2905: MatGetInfo(A,MAT_LOCAL,&info);
2906: mal = info.mallocs;
2907: nz_a = info.nz_allocated;
2908: .ve
2910: Example for Fortran Users:
2911: Fortran users should declare info as a double precision
2912: array of dimension MAT_INFO_SIZE, and then extract the parameters
2913: of interest. See the file ${PETSC_DIR}/include/petsc/finclude/petscmat.h
2914: a complete list of parameter names.
2915: .vb
2916: double precision info(MAT_INFO_SIZE)
2917: double precision mal, nz_a
2918: Mat A
2919: integer ierr
2921: call MatGetInfo(A,MAT_LOCAL,info,ierr)
2922: mal = info(MAT_INFO_MALLOCS)
2923: nz_a = info(MAT_INFO_NZ_ALLOCATED)
2924: .ve
2926: Level: intermediate
2928: Developer Note: fortran interface is not autogenerated as the f90
2929: interface definition cannot be generated correctly [due to MatInfo]
2931: .seealso: MatStashGetInfo()
2933: @*/
2934: PetscErrorCode MatGetInfo(Mat mat,MatInfoType flag,MatInfo *info)
2935: {
2942: if (!mat->ops->getinfo) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2943: MatCheckPreallocated(mat,1);
2944: (*mat->ops->getinfo)(mat,flag,info);
2945: return(0);
2946: }
2948: /*
2949: This is used by external packages where it is not easy to get the info from the actual
2950: matrix factorization.
2951: */
2952: PetscErrorCode MatGetInfo_External(Mat A,MatInfoType flag,MatInfo *info)
2953: {
2957: PetscMemzero(info,sizeof(MatInfo));
2958: return(0);
2959: }
2961: /* ----------------------------------------------------------*/
2963: /*@C
2964: MatLUFactor - Performs in-place LU factorization of matrix.
2966: Collective on Mat
2968: Input Parameters:
2969: + mat - the matrix
2970: . row - row permutation
2971: . col - column permutation
2972: - info - options for factorization, includes
2973: $ fill - expected fill as ratio of original fill.
2974: $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
2975: $ Run with the option -info to determine an optimal value to use
2977: Notes:
2978: Most users should employ the simplified KSP interface for linear solvers
2979: instead of working directly with matrix algebra routines such as this.
2980: See, e.g., KSPCreate().
2982: This changes the state of the matrix to a factored matrix; it cannot be used
2983: for example with MatSetValues() unless one first calls MatSetUnfactored().
2985: Level: developer
2987: .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(),
2988: MatGetOrdering(), MatSetUnfactored(), MatFactorInfo, MatGetFactor()
2990: Developer Note: fortran interface is not autogenerated as the f90
2991: interface definition cannot be generated correctly [due to MatFactorInfo]
2993: @*/
2994: PetscErrorCode MatLUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info)
2995: {
2997: MatFactorInfo tinfo;
3005: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3006: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3007: if (!mat->ops->lufactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3008: MatCheckPreallocated(mat,1);
3009: if (!info) {
3010: MatFactorInfoInitialize(&tinfo);
3011: info = &tinfo;
3012: }
3014: PetscLogEventBegin(MAT_LUFactor,mat,row,col,0);
3015: (*mat->ops->lufactor)(mat,row,col,info);
3016: PetscLogEventEnd(MAT_LUFactor,mat,row,col,0);
3017: PetscObjectStateIncrease((PetscObject)mat);
3018: return(0);
3019: }
3021: /*@C
3022: MatILUFactor - Performs in-place ILU factorization of matrix.
3024: Collective on Mat
3026: Input Parameters:
3027: + mat - the matrix
3028: . row - row permutation
3029: . col - column permutation
3030: - info - structure containing
3031: $ levels - number of levels of fill.
3032: $ expected fill - as ratio of original fill.
3033: $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices
3034: missing diagonal entries)
3036: Notes:
3037: Probably really in-place only when level of fill is zero, otherwise allocates
3038: new space to store factored matrix and deletes previous memory.
3040: Most users should employ the simplified KSP interface for linear solvers
3041: instead of working directly with matrix algebra routines such as this.
3042: See, e.g., KSPCreate().
3044: Level: developer
3046: .seealso: MatILUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo
3048: Developer Note: fortran interface is not autogenerated as the f90
3049: interface definition cannot be generated correctly [due to MatFactorInfo]
3051: @*/
3052: PetscErrorCode MatILUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info)
3053: {
3062: if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"matrix must be square");
3063: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3064: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3065: if (!mat->ops->ilufactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3066: MatCheckPreallocated(mat,1);
3068: PetscLogEventBegin(MAT_ILUFactor,mat,row,col,0);
3069: (*mat->ops->ilufactor)(mat,row,col,info);
3070: PetscLogEventEnd(MAT_ILUFactor,mat,row,col,0);
3071: PetscObjectStateIncrease((PetscObject)mat);
3072: return(0);
3073: }
3075: /*@C
3076: MatLUFactorSymbolic - Performs symbolic LU factorization of matrix.
3077: Call this routine before calling MatLUFactorNumeric().
3079: Collective on Mat
3081: Input Parameters:
3082: + fact - the factor matrix obtained with MatGetFactor()
3083: . mat - the matrix
3084: . row, col - row and column permutations
3085: - info - options for factorization, includes
3086: $ fill - expected fill as ratio of original fill.
3087: $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3088: $ Run with the option -info to determine an optimal value to use
3090: Notes:
3091: See Users-Manual: ch_mat for additional information about choosing the fill factor for better efficiency.
3093: Most users should employ the simplified KSP interface for linear solvers
3094: instead of working directly with matrix algebra routines such as this.
3095: See, e.g., KSPCreate().
3097: Level: developer
3099: .seealso: MatLUFactor(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo, MatFactorInfoInitialize()
3101: Developer Note: fortran interface is not autogenerated as the f90
3102: interface definition cannot be generated correctly [due to MatFactorInfo]
3104: @*/
3105: PetscErrorCode MatLUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info)
3106: {
3108: MatFactorInfo tinfo;
3117: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3118: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3119: if (!(fact)->ops->lufactorsymbolic) {
3120: MatSolverType stype;
3121: MatFactorGetSolverType(fact,&stype);
3122: SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic LU using solver package %s",((PetscObject)mat)->type_name,stype);
3123: }
3124: MatCheckPreallocated(mat,2);
3125: if (!info) {
3126: MatFactorInfoInitialize(&tinfo);
3127: info = &tinfo;
3128: }
3130: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_LUFactorSymbolic,mat,row,col,0);}
3131: (fact->ops->lufactorsymbolic)(fact,mat,row,col,info);
3132: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_LUFactorSymbolic,mat,row,col,0);}
3133: PetscObjectStateIncrease((PetscObject)fact);
3134: return(0);
3135: }
3137: /*@C
3138: MatLUFactorNumeric - Performs numeric LU factorization of a matrix.
3139: Call this routine after first calling MatLUFactorSymbolic().
3141: Collective on Mat
3143: Input Parameters:
3144: + fact - the factor matrix obtained with MatGetFactor()
3145: . mat - the matrix
3146: - info - options for factorization
3148: Notes:
3149: See MatLUFactor() for in-place factorization. See
3150: MatCholeskyFactorNumeric() for the symmetric, positive definite case.
3152: Most users should employ the simplified KSP interface for linear solvers
3153: instead of working directly with matrix algebra routines such as this.
3154: See, e.g., KSPCreate().
3156: Level: developer
3158: .seealso: MatLUFactorSymbolic(), MatLUFactor(), MatCholeskyFactor()
3160: Developer Note: fortran interface is not autogenerated as the f90
3161: interface definition cannot be generated correctly [due to MatFactorInfo]
3163: @*/
3164: PetscErrorCode MatLUFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info)
3165: {
3166: MatFactorInfo tinfo;
3174: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3175: if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dimensions are different %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N);
3177: if (!(fact)->ops->lufactornumeric) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s numeric LU",((PetscObject)mat)->type_name);
3178: MatCheckPreallocated(mat,2);
3179: if (!info) {
3180: MatFactorInfoInitialize(&tinfo);
3181: info = &tinfo;
3182: }
3184: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_LUFactorNumeric,mat,fact,0,0);}
3185: else {PetscLogEventBegin(MAT_LUFactor,mat,fact,0,0);}
3186: (fact->ops->lufactornumeric)(fact,mat,info);
3187: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_LUFactorNumeric,mat,fact,0,0);}
3188: else {PetscLogEventEnd(MAT_LUFactor,mat,fact,0,0);}
3189: MatViewFromOptions(fact,NULL,"-mat_factor_view");
3190: PetscObjectStateIncrease((PetscObject)fact);
3191: return(0);
3192: }
3194: /*@C
3195: MatCholeskyFactor - Performs in-place Cholesky factorization of a
3196: symmetric matrix.
3198: Collective on Mat
3200: Input Parameters:
3201: + mat - the matrix
3202: . perm - row and column permutations
3203: - f - expected fill as ratio of original fill
3205: Notes:
3206: See MatLUFactor() for the nonsymmetric case. See also
3207: MatCholeskyFactorSymbolic(), and MatCholeskyFactorNumeric().
3209: Most users should employ the simplified KSP interface for linear solvers
3210: instead of working directly with matrix algebra routines such as this.
3211: See, e.g., KSPCreate().
3213: Level: developer
3215: .seealso: MatLUFactor(), MatCholeskyFactorSymbolic(), MatCholeskyFactorNumeric()
3216: MatGetOrdering()
3218: Developer Note: fortran interface is not autogenerated as the f90
3219: interface definition cannot be generated correctly [due to MatFactorInfo]
3221: @*/
3222: PetscErrorCode MatCholeskyFactor(Mat mat,IS perm,const MatFactorInfo *info)
3223: {
3225: MatFactorInfo tinfo;
3232: if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"Matrix must be square");
3233: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3234: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3235: if (!mat->ops->choleskyfactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"In-place factorization for Mat type %s is not supported, try out-of-place factorization. See MatCholeskyFactorSymbolic/Numeric",((PetscObject)mat)->type_name);
3236: MatCheckPreallocated(mat,1);
3237: if (!info) {
3238: MatFactorInfoInitialize(&tinfo);
3239: info = &tinfo;
3240: }
3242: PetscLogEventBegin(MAT_CholeskyFactor,mat,perm,0,0);
3243: (*mat->ops->choleskyfactor)(mat,perm,info);
3244: PetscLogEventEnd(MAT_CholeskyFactor,mat,perm,0,0);
3245: PetscObjectStateIncrease((PetscObject)mat);
3246: return(0);
3247: }
3249: /*@C
3250: MatCholeskyFactorSymbolic - Performs symbolic Cholesky factorization
3251: of a symmetric matrix.
3253: Collective on Mat
3255: Input Parameters:
3256: + fact - the factor matrix obtained with MatGetFactor()
3257: . mat - the matrix
3258: . perm - row and column permutations
3259: - info - options for factorization, includes
3260: $ fill - expected fill as ratio of original fill.
3261: $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3262: $ Run with the option -info to determine an optimal value to use
3264: Notes:
3265: See MatLUFactorSymbolic() for the nonsymmetric case. See also
3266: MatCholeskyFactor() and MatCholeskyFactorNumeric().
3268: Most users should employ the simplified KSP interface for linear solvers
3269: instead of working directly with matrix algebra routines such as this.
3270: See, e.g., KSPCreate().
3272: Level: developer
3274: .seealso: MatLUFactorSymbolic(), MatCholeskyFactor(), MatCholeskyFactorNumeric()
3275: MatGetOrdering()
3277: Developer Note: fortran interface is not autogenerated as the f90
3278: interface definition cannot be generated correctly [due to MatFactorInfo]
3280: @*/
3281: PetscErrorCode MatCholeskyFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info)
3282: {
3284: MatFactorInfo tinfo;
3292: if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"Matrix must be square");
3293: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3294: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3295: if (!(fact)->ops->choleskyfactorsymbolic) {
3296: MatSolverType stype;
3297: MatFactorGetSolverType(fact,&stype);
3298: SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s symbolic factor Cholesky using solver package %s",((PetscObject)mat)->type_name,stype);
3299: }
3300: MatCheckPreallocated(mat,2);
3301: if (!info) {
3302: MatFactorInfoInitialize(&tinfo);
3303: info = &tinfo;
3304: }
3306: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_CholeskyFactorSymbolic,mat,perm,0,0);}
3307: (fact->ops->choleskyfactorsymbolic)(fact,mat,perm,info);
3308: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_CholeskyFactorSymbolic,mat,perm,0,0);}
3309: PetscObjectStateIncrease((PetscObject)fact);
3310: return(0);
3311: }
3313: /*@C
3314: MatCholeskyFactorNumeric - Performs numeric Cholesky factorization
3315: of a symmetric matrix. Call this routine after first calling
3316: MatCholeskyFactorSymbolic().
3318: Collective on Mat
3320: Input Parameters:
3321: + fact - the factor matrix obtained with MatGetFactor()
3322: . mat - the initial matrix
3323: . info - options for factorization
3324: - fact - the symbolic factor of mat
3326: Notes:
3327: Most users should employ the simplified KSP interface for linear solvers
3328: instead of working directly with matrix algebra routines such as this.
3329: See, e.g., KSPCreate().
3331: Level: developer
3333: .seealso: MatCholeskyFactorSymbolic(), MatCholeskyFactor(), MatLUFactorNumeric()
3335: Developer Note: fortran interface is not autogenerated as the f90
3336: interface definition cannot be generated correctly [due to MatFactorInfo]
3338: @*/
3339: PetscErrorCode MatCholeskyFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info)
3340: {
3341: MatFactorInfo tinfo;
3349: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3350: if (!(fact)->ops->choleskyfactornumeric) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s numeric factor Cholesky",((PetscObject)mat)->type_name);
3351: if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dim %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N);
3352: MatCheckPreallocated(mat,2);
3353: if (!info) {
3354: MatFactorInfoInitialize(&tinfo);
3355: info = &tinfo;
3356: }
3358: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_CholeskyFactorNumeric,mat,fact,0,0);}
3359: else {PetscLogEventBegin(MAT_CholeskyFactor,mat,fact,0,0);}
3360: (fact->ops->choleskyfactornumeric)(fact,mat,info);
3361: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_CholeskyFactorNumeric,mat,fact,0,0);}
3362: else {PetscLogEventEnd(MAT_CholeskyFactor,mat,fact,0,0);}
3363: MatViewFromOptions(fact,NULL,"-mat_factor_view");
3364: PetscObjectStateIncrease((PetscObject)fact);
3365: return(0);
3366: }
3368: /*@
3369: MatQRFactor - Performs in-place QR factorization of matrix.
3371: Collective on Mat
3373: Input Parameters:
3374: + mat - the matrix
3375: . col - column permutation
3376: - info - options for factorization, includes
3377: $ fill - expected fill as ratio of original fill.
3378: $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3379: $ Run with the option -info to determine an optimal value to use
3381: Notes:
3382: Most users should employ the simplified KSP interface for linear solvers
3383: instead of working directly with matrix algebra routines such as this.
3384: See, e.g., KSPCreate().
3386: This changes the state of the matrix to a factored matrix; it cannot be used
3387: for example with MatSetValues() unless one first calls MatSetUnfactored().
3389: Level: developer
3391: .seealso: MatQRFactorSymbolic(), MatQRFactorNumeric(), MatLUFactor(),
3392: MatSetUnfactored(), MatFactorInfo, MatGetFactor()
3394: Developer Note: fortran interface is not autogenerated as the f90
3395: interface definition cannot be generated correctly [due to MatFactorInfo]
3397: @*/
3398: PetscErrorCode MatQRFactor(Mat mat, IS col, const MatFactorInfo *info)
3399: {
3407: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3408: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3409: MatCheckPreallocated(mat,1);
3410: PetscLogEventBegin(MAT_QRFactor,mat,col,0,0);
3411: PetscUseMethod(mat,"MatQRFactor_C", (Mat,IS,const MatFactorInfo*), (mat, col, info));
3412: PetscLogEventEnd(MAT_QRFactor,mat,col,0,0);
3413: PetscObjectStateIncrease((PetscObject)mat);
3414: return(0);
3415: }
3417: /*@
3418: MatQRFactorSymbolic - Performs symbolic QR factorization of matrix.
3419: Call this routine before calling MatQRFactorNumeric().
3421: Collective on Mat
3423: Input Parameters:
3424: + fact - the factor matrix obtained with MatGetFactor()
3425: . mat - the matrix
3426: . col - column permutation
3427: - info - options for factorization, includes
3428: $ fill - expected fill as ratio of original fill.
3429: $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
3430: $ Run with the option -info to determine an optimal value to use
3432: Most users should employ the simplified KSP interface for linear solvers
3433: instead of working directly with matrix algebra routines such as this.
3434: See, e.g., KSPCreate().
3436: Level: developer
3438: .seealso: MatQRFactor(), MatQRFactorNumeric(), MatLUFactor(), MatFactorInfo, MatFactorInfoInitialize()
3440: Developer Note: fortran interface is not autogenerated as the f90
3441: interface definition cannot be generated correctly [due to MatFactorInfo]
3443: @*/
3444: PetscErrorCode MatQRFactorSymbolic(Mat fact,Mat mat,IS col,const MatFactorInfo *info)
3445: {
3447: MatFactorInfo tinfo;
3455: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3456: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3457: MatCheckPreallocated(mat,2);
3458: if (!info) {
3459: MatFactorInfoInitialize(&tinfo);
3460: info = &tinfo;
3461: }
3463: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_QRFactorSymbolic,fact,mat,col,0);}
3464: PetscUseMethod(fact,"MatQRFactorSymbolic_C", (Mat,Mat,IS,const MatFactorInfo*), (fact, mat, col, info));
3465: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_QRFactorSymbolic,fact,mat,col,0);}
3466: PetscObjectStateIncrease((PetscObject)fact);
3467: return(0);
3468: }
3470: /*@
3471: MatQRFactorNumeric - Performs numeric QR factorization of a matrix.
3472: Call this routine after first calling MatQRFactorSymbolic().
3474: Collective on Mat
3476: Input Parameters:
3477: + fact - the factor matrix obtained with MatGetFactor()
3478: . mat - the matrix
3479: - info - options for factorization
3481: Notes:
3482: See MatQRFactor() for in-place factorization.
3484: Most users should employ the simplified KSP interface for linear solvers
3485: instead of working directly with matrix algebra routines such as this.
3486: See, e.g., KSPCreate().
3488: Level: developer
3490: .seealso: MatQRFactorSymbolic(), MatLUFactor()
3492: Developer Note: fortran interface is not autogenerated as the f90
3493: interface definition cannot be generated correctly [due to MatFactorInfo]
3495: @*/
3496: PetscErrorCode MatQRFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info)
3497: {
3498: MatFactorInfo tinfo;
3506: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3507: if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dimensions are different %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N);
3509: MatCheckPreallocated(mat,2);
3510: if (!info) {
3511: MatFactorInfoInitialize(&tinfo);
3512: info = &tinfo;
3513: }
3515: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_QRFactorNumeric,mat,fact,0,0);}
3516: else {PetscLogEventBegin(MAT_QRFactor,mat,fact,0,0);}
3517: PetscUseMethod(fact,"MatQRFactorNumeric_C", (Mat,Mat,const MatFactorInfo*), (fact, mat, info));
3518: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_QRFactorNumeric,mat,fact,0,0);}
3519: else {PetscLogEventEnd(MAT_QRFactor,mat,fact,0,0);}
3520: MatViewFromOptions(fact,NULL,"-mat_factor_view");
3521: PetscObjectStateIncrease((PetscObject)fact);
3522: return(0);
3523: }
3525: /* ----------------------------------------------------------------*/
3526: /*@
3527: MatSolve - Solves A x = b, given a factored matrix.
3529: Neighbor-wise Collective on Mat
3531: Input Parameters:
3532: + mat - the factored matrix
3533: - b - the right-hand-side vector
3535: Output Parameter:
3536: . x - the result vector
3538: Notes:
3539: The vectors b and x cannot be the same. I.e., one cannot
3540: call MatSolve(A,x,x).
3542: Notes:
3543: Most users should employ the simplified KSP interface for linear solvers
3544: instead of working directly with matrix algebra routines such as this.
3545: See, e.g., KSPCreate().
3547: Level: developer
3549: .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd()
3550: @*/
3551: PetscErrorCode MatSolve(Mat mat,Vec b,Vec x)
3552: {
3562: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3563: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3564: if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3565: if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3566: if (!mat->rmap->N && !mat->cmap->N) return(0);
3567: MatCheckPreallocated(mat,1);
3569: PetscLogEventBegin(MAT_Solve,mat,b,x,0);
3570: if (mat->factorerrortype) {
3571: PetscInfo1(mat,"MatFactorError %D\n",mat->factorerrortype);
3572: VecSetInf(x);
3573: } else {
3574: if (!mat->ops->solve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3575: (*mat->ops->solve)(mat,b,x);
3576: }
3577: PetscLogEventEnd(MAT_Solve,mat,b,x,0);
3578: PetscObjectStateIncrease((PetscObject)x);
3579: return(0);
3580: }
3582: static PetscErrorCode MatMatSolve_Basic(Mat A,Mat B,Mat X,PetscBool trans)
3583: {
3585: Vec b,x;
3586: PetscInt m,N,i;
3587: PetscScalar *bb,*xx;
3588: PetscErrorCode (*f)(Mat,Vec,Vec);
3591: if (A->factorerrortype) {
3592: PetscInfo1(A,"MatFactorError %D\n",A->factorerrortype);
3593: MatSetInf(X);
3594: return(0);
3595: }
3596: f = trans ? A->ops->solvetranspose : A->ops->solve;
3597: if (!f) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)A)->type_name);
3599: MatDenseGetArrayRead(B,(const PetscScalar**)&bb);
3600: MatDenseGetArray(X,&xx);
3601: MatGetLocalSize(B,&m,NULL); /* number local rows */
3602: MatGetSize(B,NULL,&N); /* total columns in dense matrix */
3603: MatCreateVecs(A,&x,&b);
3604: for (i=0; i<N; i++) {
3605: VecPlaceArray(b,bb + i*m);
3606: VecPlaceArray(x,xx + i*m);
3607: (*f)(A,b,x);
3608: VecResetArray(x);
3609: VecResetArray(b);
3610: }
3611: VecDestroy(&b);
3612: VecDestroy(&x);
3613: MatDenseRestoreArrayRead(B,(const PetscScalar**)&bb);
3614: MatDenseRestoreArray(X,&xx);
3615: return(0);
3616: }
3618: /*@
3619: MatMatSolve - Solves A X = B, given a factored matrix.
3621: Neighbor-wise Collective on Mat
3623: Input Parameters:
3624: + A - the factored matrix
3625: - B - the right-hand-side matrix MATDENSE (or sparse -- when using MUMPS)
3627: Output Parameter:
3628: . X - the result matrix (dense matrix)
3630: Notes:
3631: If B is a MATDENSE matrix then one can call MatMatSolve(A,B,B) except with MKL_CPARDISO;
3632: otherwise, B and X cannot be the same.
3634: Notes:
3635: Most users should usually employ the simplified KSP interface for linear solvers
3636: instead of working directly with matrix algebra routines such as this.
3637: See, e.g., KSPCreate(). However KSP can only solve for one vector (column of X)
3638: at a time.
3640: Level: developer
3642: .seealso: MatMatSolveTranspose(), MatLUFactor(), MatCholeskyFactor()
3643: @*/
3644: PetscErrorCode MatMatSolve(Mat A,Mat B,Mat X)
3645: {
3655: if (A->cmap->N != X->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat X: global dim %D %D",A->cmap->N,X->rmap->N);
3656: if (A->rmap->N != B->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D",A->rmap->N,B->rmap->N);
3657: if (X->cmap->N != B->cmap->N) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Solution matrix must have same number of columns as rhs matrix");
3658: if (!A->rmap->N && !A->cmap->N) return(0);
3659: if (!A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3660: MatCheckPreallocated(A,1);
3662: PetscLogEventBegin(MAT_MatSolve,A,B,X,0);
3663: if (!A->ops->matsolve) {
3664: PetscInfo1(A,"Mat type %s using basic MatMatSolve\n",((PetscObject)A)->type_name);
3665: MatMatSolve_Basic(A,B,X,PETSC_FALSE);
3666: } else {
3667: (*A->ops->matsolve)(A,B,X);
3668: }
3669: PetscLogEventEnd(MAT_MatSolve,A,B,X,0);
3670: PetscObjectStateIncrease((PetscObject)X);
3671: return(0);
3672: }
3674: /*@
3675: MatMatSolveTranspose - Solves A^T X = B, given a factored matrix.
3677: Neighbor-wise Collective on Mat
3679: Input Parameters:
3680: + A - the factored matrix
3681: - B - the right-hand-side matrix (dense matrix)
3683: Output Parameter:
3684: . X - the result matrix (dense matrix)
3686: Notes:
3687: The matrices B and X cannot be the same. I.e., one cannot
3688: call MatMatSolveTranspose(A,X,X).
3690: Notes:
3691: Most users should usually employ the simplified KSP interface for linear solvers
3692: instead of working directly with matrix algebra routines such as this.
3693: See, e.g., KSPCreate(). However KSP can only solve for one vector (column of X)
3694: at a time.
3696: When using SuperLU_Dist or MUMPS as a parallel solver, PETSc will use their functionality to solve multiple right hand sides simultaneously.
3698: Level: developer
3700: .seealso: MatMatSolve(), MatLUFactor(), MatCholeskyFactor()
3701: @*/
3702: PetscErrorCode MatMatSolveTranspose(Mat A,Mat B,Mat X)
3703: {
3713: if (X == B) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_IDN,"X and B must be different matrices");
3714: if (A->cmap->N != X->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat X: global dim %D %D",A->cmap->N,X->rmap->N);
3715: if (A->rmap->N != B->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D",A->rmap->N,B->rmap->N);
3716: if (A->rmap->n != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: local dim %D %D",A->rmap->n,B->rmap->n);
3717: if (X->cmap->N < B->cmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Solution matrix must have same number of columns as rhs matrix");
3718: if (!A->rmap->N && !A->cmap->N) return(0);
3719: if (!A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3720: MatCheckPreallocated(A,1);
3722: PetscLogEventBegin(MAT_MatSolve,A,B,X,0);
3723: if (!A->ops->matsolvetranspose) {
3724: PetscInfo1(A,"Mat type %s using basic MatMatSolveTranspose\n",((PetscObject)A)->type_name);
3725: MatMatSolve_Basic(A,B,X,PETSC_TRUE);
3726: } else {
3727: (*A->ops->matsolvetranspose)(A,B,X);
3728: }
3729: PetscLogEventEnd(MAT_MatSolve,A,B,X,0);
3730: PetscObjectStateIncrease((PetscObject)X);
3731: return(0);
3732: }
3734: /*@
3735: MatMatTransposeSolve - Solves A X = B^T, given a factored matrix.
3737: Neighbor-wise Collective on Mat
3739: Input Parameters:
3740: + A - the factored matrix
3741: - Bt - the transpose of right-hand-side matrix
3743: Output Parameter:
3744: . X - the result matrix (dense matrix)
3746: Notes:
3747: Most users should usually employ the simplified KSP interface for linear solvers
3748: instead of working directly with matrix algebra routines such as this.
3749: See, e.g., KSPCreate(). However KSP can only solve for one vector (column of X)
3750: at a time.
3752: 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 format on the host processor and call MatMatTransposeSolve() to implement MUMPS' MatMatSolve().
3754: Level: developer
3756: .seealso: MatMatSolve(), MatMatSolveTranspose(), MatLUFactor(), MatCholeskyFactor()
3757: @*/
3758: PetscErrorCode MatMatTransposeSolve(Mat A,Mat Bt,Mat X)
3759: {
3770: if (X == Bt) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_IDN,"X and B must be different matrices");
3771: if (A->cmap->N != X->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat X: global dim %D %D",A->cmap->N,X->rmap->N);
3772: if (A->rmap->N != Bt->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat Bt: global dim %D %D",A->rmap->N,Bt->cmap->N);
3773: if (X->cmap->N < Bt->rmap->N) SETERRQ(PetscObjectComm((PetscObject)X),PETSC_ERR_ARG_SIZ,"Solution matrix must have same number of columns as row number of the rhs matrix");
3774: if (!A->rmap->N && !A->cmap->N) return(0);
3775: if (!A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3776: MatCheckPreallocated(A,1);
3778: if (!A->ops->mattransposesolve) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)A)->type_name);
3779: PetscLogEventBegin(MAT_MatTrSolve,A,Bt,X,0);
3780: (*A->ops->mattransposesolve)(A,Bt,X);
3781: PetscLogEventEnd(MAT_MatTrSolve,A,Bt,X,0);
3782: PetscObjectStateIncrease((PetscObject)X);
3783: return(0);
3784: }
3786: /*@
3787: MatForwardSolve - Solves L x = b, given a factored matrix, A = LU, or
3788: U^T*D^(1/2) x = b, given a factored symmetric matrix, A = U^T*D*U,
3790: Neighbor-wise Collective on Mat
3792: Input Parameters:
3793: + mat - the factored matrix
3794: - b - the right-hand-side vector
3796: Output Parameter:
3797: . x - the result vector
3799: Notes:
3800: MatSolve() should be used for most applications, as it performs
3801: a forward solve followed by a backward solve.
3803: The vectors b and x cannot be the same, i.e., one cannot
3804: call MatForwardSolve(A,x,x).
3806: For matrix in seqsbaij format with block size larger than 1,
3807: the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3808: MatForwardSolve() solves U^T*D y = b, and
3809: MatBackwardSolve() solves U x = y.
3810: Thus they do not provide a symmetric preconditioner.
3812: Most users should employ the simplified KSP interface for linear solvers
3813: instead of working directly with matrix algebra routines such as this.
3814: See, e.g., KSPCreate().
3816: Level: developer
3818: .seealso: MatSolve(), MatBackwardSolve()
3819: @*/
3820: PetscErrorCode MatForwardSolve(Mat mat,Vec b,Vec x)
3821: {
3831: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3832: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3833: if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3834: if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3835: if (!mat->rmap->N && !mat->cmap->N) return(0);
3836: MatCheckPreallocated(mat,1);
3838: if (!mat->ops->forwardsolve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3839: PetscLogEventBegin(MAT_ForwardSolve,mat,b,x,0);
3840: (*mat->ops->forwardsolve)(mat,b,x);
3841: PetscLogEventEnd(MAT_ForwardSolve,mat,b,x,0);
3842: PetscObjectStateIncrease((PetscObject)x);
3843: return(0);
3844: }
3846: /*@
3847: MatBackwardSolve - Solves U x = b, given a factored matrix, A = LU.
3848: D^(1/2) U x = b, given a factored symmetric matrix, A = U^T*D*U,
3850: Neighbor-wise Collective on Mat
3852: Input Parameters:
3853: + mat - the factored matrix
3854: - b - the right-hand-side vector
3856: Output Parameter:
3857: . x - the result vector
3859: Notes:
3860: MatSolve() should be used for most applications, as it performs
3861: a forward solve followed by a backward solve.
3863: The vectors b and x cannot be the same. I.e., one cannot
3864: call MatBackwardSolve(A,x,x).
3866: For matrix in seqsbaij format with block size larger than 1,
3867: the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3868: MatForwardSolve() solves U^T*D y = b, and
3869: MatBackwardSolve() solves U x = y.
3870: Thus they do not provide a symmetric preconditioner.
3872: Most users should employ the simplified KSP interface for linear solvers
3873: instead of working directly with matrix algebra routines such as this.
3874: See, e.g., KSPCreate().
3876: Level: developer
3878: .seealso: MatSolve(), MatForwardSolve()
3879: @*/
3880: PetscErrorCode MatBackwardSolve(Mat mat,Vec b,Vec x)
3881: {
3891: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3892: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3893: if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3894: if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3895: if (!mat->rmap->N && !mat->cmap->N) return(0);
3896: MatCheckPreallocated(mat,1);
3898: if (!mat->ops->backwardsolve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3899: PetscLogEventBegin(MAT_BackwardSolve,mat,b,x,0);
3900: (*mat->ops->backwardsolve)(mat,b,x);
3901: PetscLogEventEnd(MAT_BackwardSolve,mat,b,x,0);
3902: PetscObjectStateIncrease((PetscObject)x);
3903: return(0);
3904: }
3906: /*@
3907: MatSolveAdd - Computes x = y + inv(A)*b, given a factored matrix.
3909: Neighbor-wise Collective on Mat
3911: Input Parameters:
3912: + mat - the factored matrix
3913: . b - the right-hand-side vector
3914: - y - the vector to be added to
3916: Output Parameter:
3917: . x - the result vector
3919: Notes:
3920: The vectors b and x cannot be the same. I.e., one cannot
3921: call MatSolveAdd(A,x,y,x).
3923: Most users should employ the simplified KSP interface for linear solvers
3924: instead of working directly with matrix algebra routines such as this.
3925: See, e.g., KSPCreate().
3927: Level: developer
3929: .seealso: MatSolve(), MatSolveTranspose(), MatSolveTransposeAdd()
3930: @*/
3931: PetscErrorCode MatSolveAdd(Mat mat,Vec b,Vec y,Vec x)
3932: {
3933: PetscScalar one = 1.0;
3934: Vec tmp;
3946: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3947: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3948: if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3949: if (mat->rmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
3950: if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3951: if (x->map->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n);
3952: if (!mat->rmap->N && !mat->cmap->N) return(0);
3953: MatCheckPreallocated(mat,1);
3955: PetscLogEventBegin(MAT_SolveAdd,mat,b,x,y);
3956: if (mat->factorerrortype) {
3957: PetscInfo1(mat,"MatFactorError %D\n",mat->factorerrortype);
3958: VecSetInf(x);
3959: } else if (mat->ops->solveadd) {
3960: (*mat->ops->solveadd)(mat,b,y,x);
3961: } else {
3962: /* do the solve then the add manually */
3963: if (x != y) {
3964: MatSolve(mat,b,x);
3965: VecAXPY(x,one,y);
3966: } else {
3967: VecDuplicate(x,&tmp);
3968: PetscLogObjectParent((PetscObject)mat,(PetscObject)tmp);
3969: VecCopy(x,tmp);
3970: MatSolve(mat,b,x);
3971: VecAXPY(x,one,tmp);
3972: VecDestroy(&tmp);
3973: }
3974: }
3975: PetscLogEventEnd(MAT_SolveAdd,mat,b,x,y);
3976: PetscObjectStateIncrease((PetscObject)x);
3977: return(0);
3978: }
3980: /*@
3981: MatSolveTranspose - Solves A' x = b, given a factored matrix.
3983: Neighbor-wise Collective on Mat
3985: Input Parameters:
3986: + mat - the factored matrix
3987: - b - the right-hand-side vector
3989: Output Parameter:
3990: . x - the result vector
3992: Notes:
3993: The vectors b and x cannot be the same. I.e., one cannot
3994: call MatSolveTranspose(A,x,x).
3996: Most users should employ the simplified KSP interface for linear solvers
3997: instead of working directly with matrix algebra routines such as this.
3998: See, e.g., KSPCreate().
4000: Level: developer
4002: .seealso: MatSolve(), MatSolveAdd(), MatSolveTransposeAdd()
4003: @*/
4004: PetscErrorCode MatSolveTranspose(Mat mat,Vec b,Vec x)
4005: {
4015: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
4016: if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
4017: if (mat->cmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N);
4018: if (!mat->rmap->N && !mat->cmap->N) return(0);
4019: MatCheckPreallocated(mat,1);
4020: PetscLogEventBegin(MAT_SolveTranspose,mat,b,x,0);
4021: if (mat->factorerrortype) {
4022: PetscInfo1(mat,"MatFactorError %D\n",mat->factorerrortype);
4023: VecSetInf(x);
4024: } else {
4025: if (!mat->ops->solvetranspose) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s",((PetscObject)mat)->type_name);
4026: (*mat->ops->solvetranspose)(mat,b,x);
4027: }
4028: PetscLogEventEnd(MAT_SolveTranspose,mat,b,x,0);
4029: PetscObjectStateIncrease((PetscObject)x);
4030: return(0);
4031: }
4033: /*@
4034: MatSolveTransposeAdd - Computes x = y + inv(Transpose(A)) b, given a
4035: factored matrix.
4037: Neighbor-wise Collective on Mat
4039: Input Parameters:
4040: + mat - the factored matrix
4041: . b - the right-hand-side vector
4042: - y - the vector to be added to
4044: Output Parameter:
4045: . x - the result vector
4047: Notes:
4048: The vectors b and x cannot be the same. I.e., one cannot
4049: call MatSolveTransposeAdd(A,x,y,x).
4051: Most users should employ the simplified KSP interface for linear solvers
4052: instead of working directly with matrix algebra routines such as this.
4053: See, e.g., KSPCreate().
4055: Level: developer
4057: .seealso: MatSolve(), MatSolveAdd(), MatSolveTranspose()
4058: @*/
4059: PetscErrorCode MatSolveTransposeAdd(Mat mat,Vec b,Vec y,Vec x)
4060: {
4061: PetscScalar one = 1.0;
4063: Vec tmp;
4074: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
4075: if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
4076: if (mat->cmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N);
4077: if (mat->cmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
4078: if (x->map->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n);
4079: if (!mat->rmap->N && !mat->cmap->N) return(0);
4080: MatCheckPreallocated(mat,1);
4082: PetscLogEventBegin(MAT_SolveTransposeAdd,mat,b,x,y);
4083: if (mat->factorerrortype) {
4084: PetscInfo1(mat,"MatFactorError %D\n",mat->factorerrortype);
4085: VecSetInf(x);
4086: } else if (mat->ops->solvetransposeadd) {
4087: (*mat->ops->solvetransposeadd)(mat,b,y,x);
4088: } else {
4089: /* do the solve then the add manually */
4090: if (x != y) {
4091: MatSolveTranspose(mat,b,x);
4092: VecAXPY(x,one,y);
4093: } else {
4094: VecDuplicate(x,&tmp);
4095: PetscLogObjectParent((PetscObject)mat,(PetscObject)tmp);
4096: VecCopy(x,tmp);
4097: MatSolveTranspose(mat,b,x);
4098: VecAXPY(x,one,tmp);
4099: VecDestroy(&tmp);
4100: }
4101: }
4102: PetscLogEventEnd(MAT_SolveTransposeAdd,mat,b,x,y);
4103: PetscObjectStateIncrease((PetscObject)x);
4104: return(0);
4105: }
4106: /* ----------------------------------------------------------------*/
4108: /*@
4109: MatSOR - Computes relaxation (SOR, Gauss-Seidel) sweeps.
4111: Neighbor-wise Collective on Mat
4113: Input Parameters:
4114: + mat - the matrix
4115: . b - the right hand side
4116: . omega - the relaxation factor
4117: . flag - flag indicating the type of SOR (see below)
4118: . shift - diagonal shift
4119: . its - the number of iterations
4120: - lits - the number of local iterations
4122: Output Parameter:
4123: . x - the solution (can contain an initial guess, use option SOR_ZERO_INITIAL_GUESS to indicate no guess)
4125: SOR Flags:
4126: + SOR_FORWARD_SWEEP - forward SOR
4127: . SOR_BACKWARD_SWEEP - backward SOR
4128: . SOR_SYMMETRIC_SWEEP - SSOR (symmetric SOR)
4129: . SOR_LOCAL_FORWARD_SWEEP - local forward SOR
4130: . SOR_LOCAL_BACKWARD_SWEEP - local forward SOR
4131: . SOR_LOCAL_SYMMETRIC_SWEEP - local SSOR
4132: . SOR_APPLY_UPPER, SOR_APPLY_LOWER - applies
4133: upper/lower triangular part of matrix to
4134: vector (with omega)
4135: - SOR_ZERO_INITIAL_GUESS - zero initial guess
4137: Notes:
4138: SOR_LOCAL_FORWARD_SWEEP, SOR_LOCAL_BACKWARD_SWEEP, and
4139: SOR_LOCAL_SYMMETRIC_SWEEP perform separate independent smoothings
4140: on each processor.
4142: Application programmers will not generally use MatSOR() directly,
4143: but instead will employ the KSP/PC interface.
4145: Notes:
4146: for BAIJ, SBAIJ, and AIJ matrices with Inodes this does a block SOR smoothing, otherwise it does a pointwise smoothing
4148: Notes for Advanced Users:
4149: The flags are implemented as bitwise inclusive or operations.
4150: For example, use (SOR_ZERO_INITIAL_GUESS | SOR_SYMMETRIC_SWEEP)
4151: to specify a zero initial guess for SSOR.
4153: Most users should employ the simplified KSP interface for linear solvers
4154: instead of working directly with matrix algebra routines such as this.
4155: See, e.g., KSPCreate().
4157: Vectors x and b CANNOT be the same
4159: Developer Note: We should add block SOR support for AIJ matrices with block size set to great than one and no inodes
4161: Level: developer
4163: @*/
4164: PetscErrorCode MatSOR(Mat mat,Vec b,PetscReal omega,MatSORType flag,PetscReal shift,PetscInt its,PetscInt lits,Vec x)
4165: {
4175: if (!mat->ops->sor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4176: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4177: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4178: if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
4179: if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
4180: if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
4181: if (its <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D positive",its);
4182: if (lits <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires local its %D positive",lits);
4183: if (b == x) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_IDN,"b and x vector cannot be the same");
4185: MatCheckPreallocated(mat,1);
4186: PetscLogEventBegin(MAT_SOR,mat,b,x,0);
4187: ierr =(*mat->ops->sor)(mat,b,omega,flag,shift,its,lits,x);
4188: PetscLogEventEnd(MAT_SOR,mat,b,x,0);
4189: PetscObjectStateIncrease((PetscObject)x);
4190: return(0);
4191: }
4193: /*
4194: Default matrix copy routine.
4195: */
4196: PetscErrorCode MatCopy_Basic(Mat A,Mat B,MatStructure str)
4197: {
4198: PetscErrorCode ierr;
4199: PetscInt i,rstart = 0,rend = 0,nz;
4200: const PetscInt *cwork;
4201: const PetscScalar *vwork;
4204: if (B->assembled) {
4205: MatZeroEntries(B);
4206: }
4207: if (str == SAME_NONZERO_PATTERN) {
4208: MatGetOwnershipRange(A,&rstart,&rend);
4209: for (i=rstart; i<rend; i++) {
4210: MatGetRow(A,i,&nz,&cwork,&vwork);
4211: MatSetValues(B,1,&i,nz,cwork,vwork,INSERT_VALUES);
4212: MatRestoreRow(A,i,&nz,&cwork,&vwork);
4213: }
4214: } else {
4215: MatAYPX(B,0.0,A,str);
4216: }
4217: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
4218: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
4219: return(0);
4220: }
4222: /*@
4223: MatCopy - Copies a matrix to another matrix.
4225: Collective on Mat
4227: Input Parameters:
4228: + A - the matrix
4229: - str - SAME_NONZERO_PATTERN or DIFFERENT_NONZERO_PATTERN
4231: Output Parameter:
4232: . B - where the copy is put
4234: Notes:
4235: If you use SAME_NONZERO_PATTERN then the two matrices must have the same nonzero pattern or the routine will crash.
4237: MatCopy() copies the matrix entries of a matrix to another existing
4238: matrix (after first zeroing the second matrix). A related routine is
4239: MatConvert(), which first creates a new matrix and then copies the data.
4241: Level: intermediate
4243: .seealso: MatConvert(), MatDuplicate()
4245: @*/
4246: PetscErrorCode MatCopy(Mat A,Mat B,MatStructure str)
4247: {
4249: PetscInt i;
4257: MatCheckPreallocated(B,2);
4258: if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4259: if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4260: if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim (%D,%D) (%D,%D)",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N);
4261: MatCheckPreallocated(A,1);
4262: if (A == B) return(0);
4264: PetscLogEventBegin(MAT_Copy,A,B,0,0);
4265: if (A->ops->copy) {
4266: (*A->ops->copy)(A,B,str);
4267: } else { /* generic conversion */
4268: MatCopy_Basic(A,B,str);
4269: }
4271: B->stencil.dim = A->stencil.dim;
4272: B->stencil.noc = A->stencil.noc;
4273: for (i=0; i<=A->stencil.dim; i++) {
4274: B->stencil.dims[i] = A->stencil.dims[i];
4275: B->stencil.starts[i] = A->stencil.starts[i];
4276: }
4278: PetscLogEventEnd(MAT_Copy,A,B,0,0);
4279: PetscObjectStateIncrease((PetscObject)B);
4280: return(0);
4281: }
4283: /*@C
4284: MatConvert - Converts a matrix to another matrix, either of the same
4285: or different type.
4287: Collective on Mat
4289: Input Parameters:
4290: + mat - the matrix
4291: . newtype - new matrix type. Use MATSAME to create a new matrix of the
4292: same type as the original matrix.
4293: - reuse - denotes if the destination matrix is to be created or reused.
4294: 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
4295: 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).
4297: Output Parameter:
4298: . M - pointer to place new matrix
4300: Notes:
4301: MatConvert() first creates a new matrix and then copies the data from
4302: the first matrix. A related routine is MatCopy(), which copies the matrix
4303: entries of one matrix to another already existing matrix context.
4305: Cannot be used to convert a sequential matrix to parallel or parallel to sequential,
4306: the MPI communicator of the generated matrix is always the same as the communicator
4307: of the input matrix.
4309: Level: intermediate
4311: .seealso: MatCopy(), MatDuplicate()
4312: @*/
4313: PetscErrorCode MatConvert(Mat mat, MatType newtype,MatReuse reuse,Mat *M)
4314: {
4316: PetscBool sametype,issame,flg,issymmetric,ishermitian;
4317: char convname[256],mtype[256];
4318: Mat B;
4324: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4325: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4326: MatCheckPreallocated(mat,1);
4328: PetscOptionsGetString(((PetscObject)mat)->options,((PetscObject)mat)->prefix,"-matconvert_type",mtype,sizeof(mtype),&flg);
4329: if (flg) newtype = mtype;
4331: PetscObjectTypeCompare((PetscObject)mat,newtype,&sametype);
4332: PetscStrcmp(newtype,"same",&issame);
4333: if ((reuse == MAT_INPLACE_MATRIX) && (mat != *M)) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"MAT_INPLACE_MATRIX requires same input and output matrix");
4334: if ((reuse == MAT_REUSE_MATRIX) && (mat == *M)) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"MAT_REUSE_MATRIX means reuse matrix in final argument, perhaps you mean MAT_INPLACE_MATRIX");
4336: if ((reuse == MAT_INPLACE_MATRIX) && (issame || sametype)) {
4337: PetscInfo3(mat,"Early return for inplace %s %d %d\n",((PetscObject)mat)->type_name,sametype,issame);
4338: return(0);
4339: }
4341: /* Cache Mat options because some converter use MatHeaderReplace */
4342: issymmetric = mat->symmetric;
4343: ishermitian = mat->hermitian;
4345: if ((sametype || issame) && (reuse==MAT_INITIAL_MATRIX) && mat->ops->duplicate) {
4346: PetscInfo3(mat,"Calling duplicate for initial matrix %s %d %d\n",((PetscObject)mat)->type_name,sametype,issame);
4347: (*mat->ops->duplicate)(mat,MAT_COPY_VALUES,M);
4348: } else {
4349: PetscErrorCode (*conv)(Mat, MatType,MatReuse,Mat*)=NULL;
4350: const char *prefix[3] = {"seq","mpi",""};
4351: PetscInt i;
4352: /*
4353: Order of precedence:
4354: 0) See if newtype is a superclass of the current matrix.
4355: 1) See if a specialized converter is known to the current matrix.
4356: 2) See if a specialized converter is known to the desired matrix class.
4357: 3) See if a good general converter is registered for the desired class
4358: (as of 6/27/03 only MATMPIADJ falls into this category).
4359: 4) See if a good general converter is known for the current matrix.
4360: 5) Use a really basic converter.
4361: */
4363: /* 0) See if newtype is a superclass of the current matrix.
4364: i.e mat is mpiaij and newtype is aij */
4365: for (i=0; i<2; i++) {
4366: PetscStrncpy(convname,prefix[i],sizeof(convname));
4367: PetscStrlcat(convname,newtype,sizeof(convname));
4368: PetscStrcmp(convname,((PetscObject)mat)->type_name,&flg);
4369: PetscInfo3(mat,"Check superclass %s %s -> %d\n",convname,((PetscObject)mat)->type_name,flg);
4370: if (flg) {
4371: if (reuse == MAT_INPLACE_MATRIX) {
4372: PetscInfo(mat,"Early return\n");
4373: return(0);
4374: } else if (reuse == MAT_INITIAL_MATRIX && mat->ops->duplicate) {
4375: PetscInfo(mat,"Calling MatDuplicate\n");
4376: (*mat->ops->duplicate)(mat,MAT_COPY_VALUES,M);
4377: return(0);
4378: } else if (reuse == MAT_REUSE_MATRIX && mat->ops->copy) {
4379: PetscInfo(mat,"Calling MatCopy\n");
4380: MatCopy(mat,*M,SAME_NONZERO_PATTERN);
4381: return(0);
4382: }
4383: }
4384: }
4385: /* 1) See if a specialized converter is known to the current matrix and the desired class */
4386: for (i=0; i<3; i++) {
4387: PetscStrncpy(convname,"MatConvert_",sizeof(convname));
4388: PetscStrlcat(convname,((PetscObject)mat)->type_name,sizeof(convname));
4389: PetscStrlcat(convname,"_",sizeof(convname));
4390: PetscStrlcat(convname,prefix[i],sizeof(convname));
4391: PetscStrlcat(convname,issame ? ((PetscObject)mat)->type_name : newtype,sizeof(convname));
4392: PetscStrlcat(convname,"_C",sizeof(convname));
4393: PetscObjectQueryFunction((PetscObject)mat,convname,&conv);
4394: PetscInfo3(mat,"Check specialized (1) %s (%s) -> %d\n",convname,((PetscObject)mat)->type_name,!!conv);
4395: if (conv) goto foundconv;
4396: }
4398: /* 2) See if a specialized converter is known to the desired matrix class. */
4399: MatCreate(PetscObjectComm((PetscObject)mat),&B);
4400: MatSetSizes(B,mat->rmap->n,mat->cmap->n,mat->rmap->N,mat->cmap->N);
4401: MatSetType(B,newtype);
4402: for (i=0; i<3; i++) {
4403: PetscStrncpy(convname,"MatConvert_",sizeof(convname));
4404: PetscStrlcat(convname,((PetscObject)mat)->type_name,sizeof(convname));
4405: PetscStrlcat(convname,"_",sizeof(convname));
4406: PetscStrlcat(convname,prefix[i],sizeof(convname));
4407: PetscStrlcat(convname,newtype,sizeof(convname));
4408: PetscStrlcat(convname,"_C",sizeof(convname));
4409: PetscObjectQueryFunction((PetscObject)B,convname,&conv);
4410: PetscInfo3(mat,"Check specialized (2) %s (%s) -> %d\n",convname,((PetscObject)B)->type_name,!!conv);
4411: if (conv) {
4412: MatDestroy(&B);
4413: goto foundconv;
4414: }
4415: }
4417: /* 3) See if a good general converter is registered for the desired class */
4418: conv = B->ops->convertfrom;
4419: PetscInfo2(mat,"Check convertfrom (%s) -> %d\n",((PetscObject)B)->type_name,!!conv);
4420: MatDestroy(&B);
4421: if (conv) goto foundconv;
4423: /* 4) See if a good general converter is known for the current matrix */
4424: if (mat->ops->convert) conv = mat->ops->convert;
4426: PetscInfo2(mat,"Check general convert (%s) -> %d\n",((PetscObject)mat)->type_name,!!conv);
4427: if (conv) goto foundconv;
4429: /* 5) Use a really basic converter. */
4430: PetscInfo(mat,"Using MatConvert_Basic\n");
4431: conv = MatConvert_Basic;
4433: foundconv:
4434: PetscLogEventBegin(MAT_Convert,mat,0,0,0);
4435: (*conv)(mat,newtype,reuse,M);
4436: if (mat->rmap->mapping && mat->cmap->mapping && !(*M)->rmap->mapping && !(*M)->cmap->mapping) {
4437: /* the block sizes must be same if the mappings are copied over */
4438: (*M)->rmap->bs = mat->rmap->bs;
4439: (*M)->cmap->bs = mat->cmap->bs;
4440: PetscObjectReference((PetscObject)mat->rmap->mapping);
4441: PetscObjectReference((PetscObject)mat->cmap->mapping);
4442: (*M)->rmap->mapping = mat->rmap->mapping;
4443: (*M)->cmap->mapping = mat->cmap->mapping;
4444: }
4445: (*M)->stencil.dim = mat->stencil.dim;
4446: (*M)->stencil.noc = mat->stencil.noc;
4447: for (i=0; i<=mat->stencil.dim; i++) {
4448: (*M)->stencil.dims[i] = mat->stencil.dims[i];
4449: (*M)->stencil.starts[i] = mat->stencil.starts[i];
4450: }
4451: PetscLogEventEnd(MAT_Convert,mat,0,0,0);
4452: }
4453: PetscObjectStateIncrease((PetscObject)*M);
4455: /* Copy Mat options */
4456: if (issymmetric) {
4457: MatSetOption(*M,MAT_SYMMETRIC,PETSC_TRUE);
4458: }
4459: if (ishermitian) {
4460: MatSetOption(*M,MAT_HERMITIAN,PETSC_TRUE);
4461: }
4462: return(0);
4463: }
4465: /*@C
4466: MatFactorGetSolverType - Returns name of the package providing the factorization routines
4468: Not Collective
4470: Input Parameter:
4471: . mat - the matrix, must be a factored matrix
4473: Output Parameter:
4474: . type - the string name of the package (do not free this string)
4476: Notes:
4477: In Fortran you pass in a empty string and the package name will be copied into it.
4478: (Make sure the string is long enough)
4480: Level: intermediate
4482: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor()
4483: @*/
4484: PetscErrorCode MatFactorGetSolverType(Mat mat, MatSolverType *type)
4485: {
4486: PetscErrorCode ierr, (*conv)(Mat,MatSolverType*);
4491: if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix");
4492: PetscObjectQueryFunction((PetscObject)mat,"MatFactorGetSolverType_C",&conv);
4493: if (!conv) {
4494: *type = MATSOLVERPETSC;
4495: } else {
4496: (*conv)(mat,type);
4497: }
4498: return(0);
4499: }
4501: typedef struct _MatSolverTypeForSpecifcType* MatSolverTypeForSpecifcType;
4502: struct _MatSolverTypeForSpecifcType {
4503: MatType mtype;
4504: /* no entry for MAT_FACTOR_NONE */
4505: PetscErrorCode (*createfactor[MAT_FACTOR_NUM_TYPES-1])(Mat,MatFactorType,Mat*);
4506: MatSolverTypeForSpecifcType next;
4507: };
4509: typedef struct _MatSolverTypeHolder* MatSolverTypeHolder;
4510: struct _MatSolverTypeHolder {
4511: char *name;
4512: MatSolverTypeForSpecifcType handlers;
4513: MatSolverTypeHolder next;
4514: };
4516: static MatSolverTypeHolder MatSolverTypeHolders = NULL;
4518: /*@C
4519: MatSolverTypeRegister - Registers a MatSolverType that works for a particular matrix type
4521: Input Parameters:
4522: + package - name of the package, for example petsc or superlu
4523: . mtype - the matrix type that works with this package
4524: . ftype - the type of factorization supported by the package
4525: - createfactor - routine that will create the factored matrix ready to be used
4527: Level: intermediate
4529: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor()
4530: @*/
4531: PetscErrorCode MatSolverTypeRegister(MatSolverType package,MatType mtype,MatFactorType ftype,PetscErrorCode (*createfactor)(Mat,MatFactorType,Mat*))
4532: {
4533: PetscErrorCode ierr;
4534: MatSolverTypeHolder next = MatSolverTypeHolders,prev = NULL;
4535: PetscBool flg;
4536: MatSolverTypeForSpecifcType inext,iprev = NULL;
4539: MatInitializePackage();
4540: if (!next) {
4541: PetscNew(&MatSolverTypeHolders);
4542: PetscStrallocpy(package,&MatSolverTypeHolders->name);
4543: PetscNew(&MatSolverTypeHolders->handlers);
4544: PetscStrallocpy(mtype,(char **)&MatSolverTypeHolders->handlers->mtype);
4545: MatSolverTypeHolders->handlers->createfactor[(int)ftype-1] = createfactor;
4546: return(0);
4547: }
4548: while (next) {
4549: PetscStrcasecmp(package,next->name,&flg);
4550: if (flg) {
4551: if (!next->handlers) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"MatSolverTypeHolder is missing handlers");
4552: inext = next->handlers;
4553: while (inext) {
4554: PetscStrcasecmp(mtype,inext->mtype,&flg);
4555: if (flg) {
4556: inext->createfactor[(int)ftype-1] = createfactor;
4557: return(0);
4558: }
4559: iprev = inext;
4560: inext = inext->next;
4561: }
4562: PetscNew(&iprev->next);
4563: PetscStrallocpy(mtype,(char **)&iprev->next->mtype);
4564: iprev->next->createfactor[(int)ftype-1] = createfactor;
4565: return(0);
4566: }
4567: prev = next;
4568: next = next->next;
4569: }
4570: PetscNew(&prev->next);
4571: PetscStrallocpy(package,&prev->next->name);
4572: PetscNew(&prev->next->handlers);
4573: PetscStrallocpy(mtype,(char **)&prev->next->handlers->mtype);
4574: prev->next->handlers->createfactor[(int)ftype-1] = createfactor;
4575: return(0);
4576: }
4578: /*@C
4579: MatSolveTypeGet - Gets the function that creates the factor matrix if it exist
4581: Input Parameters:
4582: + type - name of the package, for example petsc or superlu
4583: . ftype - the type of factorization supported by the type
4584: - mtype - the matrix type that works with this type
4586: Output Parameters:
4587: + foundtype - PETSC_TRUE if the type was registered
4588: . foundmtype - PETSC_TRUE if the type supports the requested mtype
4589: - createfactor - routine that will create the factored matrix ready to be used or NULL if not found
4591: Level: intermediate
4593: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatSolverTypeRegister(), MatGetFactor()
4594: @*/
4595: PetscErrorCode MatSolverTypeGet(MatSolverType type,MatType mtype,MatFactorType ftype,PetscBool *foundtype,PetscBool *foundmtype,PetscErrorCode (**createfactor)(Mat,MatFactorType,Mat*))
4596: {
4597: PetscErrorCode ierr;
4598: MatSolverTypeHolder next = MatSolverTypeHolders;
4599: PetscBool flg;
4600: MatSolverTypeForSpecifcType inext;
4603: if (foundtype) *foundtype = PETSC_FALSE;
4604: if (foundmtype) *foundmtype = PETSC_FALSE;
4605: if (createfactor) *createfactor = NULL;
4607: if (type) {
4608: while (next) {
4609: PetscStrcasecmp(type,next->name,&flg);
4610: if (flg) {
4611: if (foundtype) *foundtype = PETSC_TRUE;
4612: inext = next->handlers;
4613: while (inext) {
4614: PetscStrbeginswith(mtype,inext->mtype,&flg);
4615: if (flg) {
4616: if (foundmtype) *foundmtype = PETSC_TRUE;
4617: if (createfactor) *createfactor = inext->createfactor[(int)ftype-1];
4618: return(0);
4619: }
4620: inext = inext->next;
4621: }
4622: }
4623: next = next->next;
4624: }
4625: } else {
4626: while (next) {
4627: inext = next->handlers;
4628: while (inext) {
4629: PetscStrcmp(mtype,inext->mtype,&flg);
4630: if (flg && inext->createfactor[(int)ftype-1]) {
4631: if (foundtype) *foundtype = PETSC_TRUE;
4632: if (foundmtype) *foundmtype = PETSC_TRUE;
4633: if (createfactor) *createfactor = inext->createfactor[(int)ftype-1];
4634: return(0);
4635: }
4636: inext = inext->next;
4637: }
4638: next = next->next;
4639: }
4640: /* try with base classes inext->mtype */
4641: next = MatSolverTypeHolders;
4642: while (next) {
4643: inext = next->handlers;
4644: while (inext) {
4645: PetscStrbeginswith(mtype,inext->mtype,&flg);
4646: if (flg && inext->createfactor[(int)ftype-1]) {
4647: if (foundtype) *foundtype = PETSC_TRUE;
4648: if (foundmtype) *foundmtype = PETSC_TRUE;
4649: if (createfactor) *createfactor = inext->createfactor[(int)ftype-1];
4650: return(0);
4651: }
4652: inext = inext->next;
4653: }
4654: next = next->next;
4655: }
4656: }
4657: return(0);
4658: }
4660: PetscErrorCode MatSolverTypeDestroy(void)
4661: {
4662: PetscErrorCode ierr;
4663: MatSolverTypeHolder next = MatSolverTypeHolders,prev;
4664: MatSolverTypeForSpecifcType inext,iprev;
4667: while (next) {
4668: PetscFree(next->name);
4669: inext = next->handlers;
4670: while (inext) {
4671: PetscFree(inext->mtype);
4672: iprev = inext;
4673: inext = inext->next;
4674: PetscFree(iprev);
4675: }
4676: prev = next;
4677: next = next->next;
4678: PetscFree(prev);
4679: }
4680: MatSolverTypeHolders = NULL;
4681: return(0);
4682: }
4684: /*@C
4685: MatFactorGetCanUseOrdering - Indicates if the factorization can use the ordering provided in MatLUFactorSymbolic(), MatCholeskyFactorSymbolic()
4687: Logically Collective on Mat
4689: Input Parameters:
4690: . mat - the matrix
4692: Output Parameters:
4693: . flg - PETSC_TRUE if uses the ordering
4695: Notes:
4696: Most internal PETSc factorizations use the ordering passed to the factorization routine but external
4697: packages do not, thus we want to skip generating the ordering when it is not needed or used.
4699: Level: developer
4701: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor(), MatLUFactorSymbolic(), MatCholeskyFactorSymbolic()
4702: @*/
4703: PetscErrorCode MatFactorGetCanUseOrdering(Mat mat, PetscBool *flg)
4704: {
4706: *flg = mat->canuseordering;
4707: return(0);
4708: }
4710: /*@C
4711: MatFactorGetPreferredOrdering - The preferred ordering for a particular matrix factor object
4713: Logically Collective on Mat
4715: Input Parameters:
4716: . mat - the matrix
4718: Output Parameters:
4719: . otype - the preferred type
4721: Level: developer
4723: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor(), MatLUFactorSymbolic(), MatCholeskyFactorSymbolic()
4724: @*/
4725: PetscErrorCode MatFactorGetPreferredOrdering(Mat mat, MatFactorType ftype, MatOrderingType *otype)
4726: {
4728: *otype = mat->preferredordering[ftype];
4729: if (!*otype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"MatFactor did not have a preferred ordering");
4730: return(0);
4731: }
4733: /*@C
4734: MatGetFactor - Returns a matrix suitable to calls to MatXXFactorSymbolic()
4736: Collective on Mat
4738: Input Parameters:
4739: + mat - the matrix
4740: . type - name of solver type, for example, superlu, petsc (to use PETSc's default)
4741: - ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU,
4743: Output Parameters:
4744: . f - the factor matrix used with MatXXFactorSymbolic() calls
4746: Notes:
4747: Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
4748: such as pastix, superlu, mumps etc.
4750: PETSc must have been ./configure to use the external solver, using the option --download-package
4752: Developer Notes:
4753: This should actually be called MatCreateFactor() since it creates a new factor object
4755: Level: intermediate
4757: .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatFactorGetCanUseOrdering(), MatSolverTypeRegister()
4758: @*/
4759: PetscErrorCode MatGetFactor(Mat mat, MatSolverType type,MatFactorType ftype,Mat *f)
4760: {
4761: PetscErrorCode ierr,(*conv)(Mat,MatFactorType,Mat*);
4762: PetscBool foundtype,foundmtype;
4768: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4769: MatCheckPreallocated(mat,1);
4771: MatSolverTypeGet(type,((PetscObject)mat)->type_name,ftype,&foundtype,&foundmtype,&conv);
4772: if (!foundtype) {
4773: if (type) {
4774: SETERRQ4(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],((PetscObject)mat)->type_name,type);
4775: } else {
4776: SETERRQ2(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);
4777: }
4778: }
4779: if (!foundmtype) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_MISSING_FACTOR,"MatSolverType %s does not support matrix type %s",type,((PetscObject)mat)->type_name);
4780: if (!conv) SETERRQ3(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);
4782: (*conv)(mat,ftype,f);
4783: return(0);
4784: }
4786: /*@C
4787: MatGetFactorAvailable - Returns a a flag if matrix supports particular type and factor type
4789: Not Collective
4791: Input Parameters:
4792: + mat - the matrix
4793: . type - name of solver type, for example, superlu, petsc (to use PETSc's default)
4794: - ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU,
4796: Output Parameter:
4797: . flg - PETSC_TRUE if the factorization is available
4799: Notes:
4800: Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
4801: such as pastix, superlu, mumps etc.
4803: PETSc must have been ./configure to use the external solver, using the option --download-package
4805: Developer Notes:
4806: This should actually be called MatCreateFactorAvailable() since MatGetFactor() creates a new factor object
4808: Level: intermediate
4810: .seealso: MatCopy(), MatDuplicate(), MatGetFactor(), MatSolverTypeRegister()
4811: @*/
4812: PetscErrorCode MatGetFactorAvailable(Mat mat, MatSolverType type,MatFactorType ftype,PetscBool *flg)
4813: {
4814: PetscErrorCode ierr, (*gconv)(Mat,MatFactorType,Mat*);
4820: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4821: MatCheckPreallocated(mat,1);
4823: *flg = PETSC_FALSE;
4824: MatSolverTypeGet(type,((PetscObject)mat)->type_name,ftype,NULL,NULL,&gconv);
4825: if (gconv) {
4826: *flg = PETSC_TRUE;
4827: }
4828: return(0);
4829: }
4831: #include <petscdmtypes.h>
4833: /*@
4834: MatDuplicate - Duplicates a matrix including the non-zero structure.
4836: Collective on Mat
4838: Input Parameters:
4839: + mat - the matrix
4840: - op - One of MAT_DO_NOT_COPY_VALUES, MAT_COPY_VALUES, or MAT_SHARE_NONZERO_PATTERN.
4841: See the manual page for MatDuplicateOption for an explanation of these options.
4843: Output Parameter:
4844: . M - pointer to place new matrix
4846: Level: intermediate
4848: Notes:
4849: You cannot change the nonzero pattern for the parent or child matrix if you use MAT_SHARE_NONZERO_PATTERN.
4850: 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 is duplicated and the internal data structures created for the reuse of previous matrix operations are not duplicated. User should not use MatDuplicate() to create new matrix M if M is intended to be reused as the product of matrix operation.
4852: .seealso: MatCopy(), MatConvert(), MatDuplicateOption
4853: @*/
4854: PetscErrorCode MatDuplicate(Mat mat,MatDuplicateOption op,Mat *M)
4855: {
4857: Mat B;
4858: PetscInt i;
4859: DM dm;
4860: void (*viewf)(void);
4866: if (op == MAT_COPY_VALUES && !mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"MAT_COPY_VALUES not allowed for unassembled matrix");
4867: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4868: MatCheckPreallocated(mat,1);
4870: *M = NULL;
4871: if (!mat->ops->duplicate) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not written for matrix type %s\n",((PetscObject)mat)->type_name);
4872: PetscLogEventBegin(MAT_Convert,mat,0,0,0);
4873: (*mat->ops->duplicate)(mat,op,M);
4874: B = *M;
4876: MatGetOperation(mat,MATOP_VIEW,&viewf);
4877: if (viewf) {
4878: MatSetOperation(B,MATOP_VIEW,viewf);
4879: }
4881: B->stencil.dim = mat->stencil.dim;
4882: B->stencil.noc = mat->stencil.noc;
4883: for (i=0; i<=mat->stencil.dim; i++) {
4884: B->stencil.dims[i] = mat->stencil.dims[i];
4885: B->stencil.starts[i] = mat->stencil.starts[i];
4886: }
4888: B->nooffproczerorows = mat->nooffproczerorows;
4889: B->nooffprocentries = mat->nooffprocentries;
4891: PetscObjectQuery((PetscObject) mat, "__PETSc_dm", (PetscObject*) &dm);
4892: if (dm) {
4893: PetscObjectCompose((PetscObject) B, "__PETSc_dm", (PetscObject) dm);
4894: }
4895: PetscLogEventEnd(MAT_Convert,mat,0,0,0);
4896: PetscObjectStateIncrease((PetscObject)B);
4897: return(0);
4898: }
4900: /*@
4901: MatGetDiagonal - Gets the diagonal of a matrix.
4903: Logically Collective on Mat
4905: Input Parameters:
4906: + mat - the matrix
4907: - v - the vector for storing the diagonal
4909: Output Parameter:
4910: . v - the diagonal of the matrix
4912: Level: intermediate
4914: Note:
4915: Currently only correct in parallel for square matrices.
4917: .seealso: MatGetRow(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMaxAbs()
4918: @*/
4919: PetscErrorCode MatGetDiagonal(Mat mat,Vec v)
4920: {
4927: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4928: if (!mat->ops->getdiagonal) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4929: MatCheckPreallocated(mat,1);
4931: (*mat->ops->getdiagonal)(mat,v);
4932: PetscObjectStateIncrease((PetscObject)v);
4933: return(0);
4934: }
4936: /*@C
4937: MatGetRowMin - Gets the minimum value (of the real part) of each
4938: row of the matrix
4940: Logically Collective on Mat
4942: Input Parameter:
4943: . mat - the matrix
4945: Output Parameters:
4946: + v - the vector for storing the maximums
4947: - idx - the indices of the column found for each row (optional)
4949: Level: intermediate
4951: Notes:
4952: The result of this call are the same as if one converted the matrix to dense format
4953: and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
4955: This code is only implemented for a couple of matrix formats.
4957: .seealso: MatGetDiagonal(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMaxAbs(),
4958: MatGetRowMax()
4959: @*/
4960: PetscErrorCode MatGetRowMin(Mat mat,Vec v,PetscInt idx[])
4961: {
4968: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4970: if (!mat->cmap->N) {
4971: VecSet(v,PETSC_MAX_REAL);
4972: if (idx) {
4973: PetscInt i,m = mat->rmap->n;
4974: for (i=0; i<m; i++) idx[i] = -1;
4975: }
4976: } else {
4977: if (!mat->ops->getrowmin) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4978: MatCheckPreallocated(mat,1);
4979: }
4980: (*mat->ops->getrowmin)(mat,v,idx);
4981: PetscObjectStateIncrease((PetscObject)v);
4982: return(0);
4983: }
4985: /*@C
4986: MatGetRowMinAbs - Gets the minimum value (in absolute value) of each
4987: row of the matrix
4989: Logically Collective on Mat
4991: Input Parameter:
4992: . mat - the matrix
4994: Output Parameters:
4995: + v - the vector for storing the minimums
4996: - idx - the indices of the column found for each row (or NULL if not needed)
4998: Level: intermediate
5000: Notes:
5001: if a row is completely empty or has only 0.0 values then the idx[] value for that
5002: row is 0 (the first column).
5004: This code is only implemented for a couple of matrix formats.
5006: .seealso: MatGetDiagonal(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMax(), MatGetRowMaxAbs(), MatGetRowMin()
5007: @*/
5008: PetscErrorCode MatGetRowMinAbs(Mat mat,Vec v,PetscInt idx[])
5009: {
5016: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5017: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5019: if (!mat->cmap->N) {
5020: VecSet(v,0.0);
5021: if (idx) {
5022: PetscInt i,m = mat->rmap->n;
5023: for (i=0; i<m; i++) idx[i] = -1;
5024: }
5025: } else {
5026: if (!mat->ops->getrowminabs) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5027: MatCheckPreallocated(mat,1);
5028: if (idx) {PetscArrayzero(idx,mat->rmap->n);}
5029: (*mat->ops->getrowminabs)(mat,v,idx);
5030: }
5031: PetscObjectStateIncrease((PetscObject)v);
5032: return(0);
5033: }
5035: /*@C
5036: MatGetRowMax - Gets the maximum value (of the real part) of each
5037: row of the matrix
5039: Logically Collective on Mat
5041: Input Parameter:
5042: . mat - the matrix
5044: Output Parameters:
5045: + v - the vector for storing the maximums
5046: - idx - the indices of the column found for each row (optional)
5048: Level: intermediate
5050: Notes:
5051: The result of this call are the same as if one converted the matrix to dense format
5052: and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
5054: This code is only implemented for a couple of matrix formats.
5056: .seealso: MatGetDiagonal(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMaxAbs(), MatGetRowMin()
5057: @*/
5058: PetscErrorCode MatGetRowMax(Mat mat,Vec v,PetscInt idx[])
5059: {
5066: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5068: if (!mat->cmap->N) {
5069: VecSet(v,PETSC_MIN_REAL);
5070: if (idx) {
5071: PetscInt i,m = mat->rmap->n;
5072: for (i=0; i<m; i++) idx[i] = -1;
5073: }
5074: } else {
5075: if (!mat->ops->getrowmax) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5076: MatCheckPreallocated(mat,1);
5077: (*mat->ops->getrowmax)(mat,v,idx);
5078: }
5079: PetscObjectStateIncrease((PetscObject)v);
5080: return(0);
5081: }
5083: /*@C
5084: MatGetRowMaxAbs - Gets the maximum value (in absolute value) of each
5085: row of the matrix
5087: Logically Collective on Mat
5089: Input Parameter:
5090: . mat - the matrix
5092: Output Parameters:
5093: + v - the vector for storing the maximums
5094: - idx - the indices of the column found for each row (or NULL if not needed)
5096: Level: intermediate
5098: Notes:
5099: if a row is completely empty or has only 0.0 values then the idx[] value for that
5100: row is 0 (the first column).
5102: This code is only implemented for a couple of matrix formats.
5104: .seealso: MatGetDiagonal(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMax(), MatGetRowMin()
5105: @*/
5106: PetscErrorCode MatGetRowMaxAbs(Mat mat,Vec v,PetscInt idx[])
5107: {
5114: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5116: if (!mat->cmap->N) {
5117: VecSet(v,0.0);
5118: if (idx) {
5119: PetscInt i,m = mat->rmap->n;
5120: for (i=0; i<m; i++) idx[i] = -1;
5121: }
5122: } else {
5123: if (!mat->ops->getrowmaxabs) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5124: MatCheckPreallocated(mat,1);
5125: if (idx) {PetscArrayzero(idx,mat->rmap->n);}
5126: (*mat->ops->getrowmaxabs)(mat,v,idx);
5127: }
5128: PetscObjectStateIncrease((PetscObject)v);
5129: return(0);
5130: }
5132: /*@
5133: MatGetRowSum - Gets the sum of each row of the matrix
5135: Logically or Neighborhood Collective on Mat
5137: Input Parameters:
5138: . mat - the matrix
5140: Output Parameter:
5141: . v - the vector for storing the sum of rows
5143: Level: intermediate
5145: Notes:
5146: This code is slow since it is not currently specialized for different formats
5148: .seealso: MatGetDiagonal(), MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRowMax(), MatGetRowMin()
5149: @*/
5150: PetscErrorCode MatGetRowSum(Mat mat, Vec v)
5151: {
5152: Vec ones;
5159: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5160: MatCheckPreallocated(mat,1);
5161: MatCreateVecs(mat,&ones,NULL);
5162: VecSet(ones,1.);
5163: MatMult(mat,ones,v);
5164: VecDestroy(&ones);
5165: return(0);
5166: }
5168: /*@
5169: MatTranspose - Computes an in-place or out-of-place transpose of a matrix.
5171: Collective on Mat
5173: Input Parameters:
5174: + mat - the matrix to transpose
5175: - reuse - either MAT_INITIAL_MATRIX, MAT_REUSE_MATRIX, or MAT_INPLACE_MATRIX
5177: Output Parameter:
5178: . B - the transpose
5180: Notes:
5181: If you use MAT_INPLACE_MATRIX then you must pass in &mat for B
5183: MAT_REUSE_MATRIX causes the B matrix from a previous call to this function with MAT_INITIAL_MATRIX to be used
5185: Consider using MatCreateTranspose() instead if you only need a matrix that behaves like the transpose, but don't need the storage to be changed.
5187: Level: intermediate
5189: .seealso: MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse
5190: @*/
5191: PetscErrorCode MatTranspose(Mat mat,MatReuse reuse,Mat *B)
5192: {
5198: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5199: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5200: if (!mat->ops->transpose) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5201: if (reuse == MAT_INPLACE_MATRIX && mat != *B) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"MAT_INPLACE_MATRIX requires last matrix to match first");
5202: if (reuse == MAT_REUSE_MATRIX && mat == *B) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Perhaps you mean MAT_INPLACE_MATRIX");
5203: MatCheckPreallocated(mat,1);
5205: PetscLogEventBegin(MAT_Transpose,mat,0,0,0);
5206: (*mat->ops->transpose)(mat,reuse,B);
5207: PetscLogEventEnd(MAT_Transpose,mat,0,0,0);
5208: if (B) {PetscObjectStateIncrease((PetscObject)*B);}
5209: return(0);
5210: }
5212: /*@
5213: MatIsTranspose - Test whether a matrix is another one's transpose,
5214: or its own, in which case it tests symmetry.
5216: Collective on Mat
5218: Input Parameters:
5219: + A - the matrix to test
5220: - B - the matrix to test against, this can equal the first parameter
5222: Output Parameters:
5223: . flg - the result
5225: Notes:
5226: Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm
5227: has a running time of the order of the number of nonzeros; the parallel
5228: test involves parallel copies of the block-offdiagonal parts of the matrix.
5230: Level: intermediate
5232: .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian()
5233: @*/
5234: PetscErrorCode MatIsTranspose(Mat A,Mat B,PetscReal tol,PetscBool *flg)
5235: {
5236: PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscBool*),(*g)(Mat,Mat,PetscReal,PetscBool*);
5242: PetscObjectQueryFunction((PetscObject)A,"MatIsTranspose_C",&f);
5243: PetscObjectQueryFunction((PetscObject)B,"MatIsTranspose_C",&g);
5244: *flg = PETSC_FALSE;
5245: if (f && g) {
5246: if (f == g) {
5247: (*f)(A,B,tol,flg);
5248: } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for symmetry test");
5249: } else {
5250: MatType mattype;
5251: if (!f) {
5252: MatGetType(A,&mattype);
5253: } else {
5254: MatGetType(B,&mattype);
5255: }
5256: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type %s does not support checking for transpose",mattype);
5257: }
5258: return(0);
5259: }
5261: /*@
5262: MatHermitianTranspose - Computes an in-place or out-of-place transpose of a matrix in complex conjugate.
5264: Collective on Mat
5266: Input Parameters:
5267: + mat - the matrix to transpose and complex conjugate
5268: - reuse - either MAT_INITIAL_MATRIX, MAT_REUSE_MATRIX, or MAT_INPLACE_MATRIX
5270: Output Parameter:
5271: . B - the Hermitian
5273: Level: intermediate
5275: .seealso: MatTranspose(), MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse
5276: @*/
5277: PetscErrorCode MatHermitianTranspose(Mat mat,MatReuse reuse,Mat *B)
5278: {
5282: MatTranspose(mat,reuse,B);
5283: #if defined(PETSC_USE_COMPLEX)
5284: MatConjugate(*B);
5285: #endif
5286: return(0);
5287: }
5289: /*@
5290: MatIsHermitianTranspose - Test whether a matrix is another one's Hermitian transpose,
5292: Collective on Mat
5294: Input Parameters:
5295: + A - the matrix to test
5296: - B - the matrix to test against, this can equal the first parameter
5298: Output Parameters:
5299: . flg - the result
5301: Notes:
5302: Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm
5303: has a running time of the order of the number of nonzeros; the parallel
5304: test involves parallel copies of the block-offdiagonal parts of the matrix.
5306: Level: intermediate
5308: .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian(), MatIsTranspose()
5309: @*/
5310: PetscErrorCode MatIsHermitianTranspose(Mat A,Mat B,PetscReal tol,PetscBool *flg)
5311: {
5312: PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscBool*),(*g)(Mat,Mat,PetscReal,PetscBool*);
5318: PetscObjectQueryFunction((PetscObject)A,"MatIsHermitianTranspose_C",&f);
5319: PetscObjectQueryFunction((PetscObject)B,"MatIsHermitianTranspose_C",&g);
5320: if (f && g) {
5321: if (f==g) {
5322: (*f)(A,B,tol,flg);
5323: } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for Hermitian test");
5324: }
5325: return(0);
5326: }
5328: /*@
5329: MatPermute - Creates a new matrix with rows and columns permuted from the
5330: original.
5332: Collective on Mat
5334: Input Parameters:
5335: + mat - the matrix to permute
5336: . row - row permutation, each processor supplies only the permutation for its rows
5337: - col - column permutation, each processor supplies only the permutation for its columns
5339: Output Parameters:
5340: . B - the permuted matrix
5342: Level: advanced
5344: Note:
5345: The index sets map from row/col of permuted matrix to row/col of original matrix.
5346: The index sets should be on the same communicator as Mat and have the same local sizes.
5348: Developer Note:
5349: If you want to implement MatPermute for a matrix type, and your approach doesn't
5350: exploit the fact that row and col are permutations, consider implementing the
5351: more general MatCreateSubMatrix() instead.
5353: .seealso: MatGetOrdering(), ISAllGather()
5355: @*/
5356: PetscErrorCode MatPermute(Mat mat,IS row,IS col,Mat *B)
5357: {
5368: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5369: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5370: if (!mat->ops->permute && !mat->ops->createsubmatrix) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatPermute not available for Mat type %s",((PetscObject)mat)->type_name);
5371: MatCheckPreallocated(mat,1);
5373: if (mat->ops->permute) {
5374: (*mat->ops->permute)(mat,row,col,B);
5375: PetscObjectStateIncrease((PetscObject)*B);
5376: } else {
5377: MatCreateSubMatrix(mat, row, col, MAT_INITIAL_MATRIX, B);
5378: }
5379: return(0);
5380: }
5382: /*@
5383: MatEqual - Compares two matrices.
5385: Collective on Mat
5387: Input Parameters:
5388: + A - the first matrix
5389: - B - the second matrix
5391: Output Parameter:
5392: . flg - PETSC_TRUE if the matrices are equal; PETSC_FALSE otherwise.
5394: Level: intermediate
5396: @*/
5397: PetscErrorCode MatEqual(Mat A,Mat B,PetscBool *flg)
5398: {
5408: MatCheckPreallocated(B,2);
5409: if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5410: if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5411: if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D %D %D",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N);
5412: if (!A->ops->equal) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)A)->type_name);
5413: if (!B->ops->equal) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)B)->type_name);
5414: if (A->ops->equal != B->ops->equal) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"A is type: %s\nB is type: %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
5415: MatCheckPreallocated(A,1);
5417: (*A->ops->equal)(A,B,flg);
5418: return(0);
5419: }
5421: /*@
5422: MatDiagonalScale - Scales a matrix on the left and right by diagonal
5423: matrices that are stored as vectors. Either of the two scaling
5424: matrices can be NULL.
5426: Collective on Mat
5428: Input Parameters:
5429: + mat - the matrix to be scaled
5430: . l - the left scaling vector (or NULL)
5431: - r - the right scaling vector (or NULL)
5433: Notes:
5434: MatDiagonalScale() computes A = LAR, where
5435: L = a diagonal matrix (stored as a vector), R = a diagonal matrix (stored as a vector)
5436: The L scales the rows of the matrix, the R scales the columns of the matrix.
5438: Level: intermediate
5440: .seealso: MatScale(), MatShift(), MatDiagonalSet()
5441: @*/
5442: PetscErrorCode MatDiagonalScale(Mat mat,Vec l,Vec r)
5443: {
5451: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5452: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5453: MatCheckPreallocated(mat,1);
5454: if (!l && !r) return(0);
5456: if (!mat->ops->diagonalscale) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5457: PetscLogEventBegin(MAT_Scale,mat,0,0,0);
5458: (*mat->ops->diagonalscale)(mat,l,r);
5459: PetscLogEventEnd(MAT_Scale,mat,0,0,0);
5460: PetscObjectStateIncrease((PetscObject)mat);
5461: return(0);
5462: }
5464: /*@
5465: MatScale - Scales all elements of a matrix by a given number.
5467: Logically Collective on Mat
5469: Input Parameters:
5470: + mat - the matrix to be scaled
5471: - a - the scaling value
5473: Output Parameter:
5474: . mat - the scaled matrix
5476: Level: intermediate
5478: .seealso: MatDiagonalScale()
5479: @*/
5480: PetscErrorCode MatScale(Mat mat,PetscScalar a)
5481: {
5487: if (a != (PetscScalar)1.0 && !mat->ops->scale) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5488: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5489: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5491: MatCheckPreallocated(mat,1);
5493: PetscLogEventBegin(MAT_Scale,mat,0,0,0);
5494: if (a != (PetscScalar)1.0) {
5495: (*mat->ops->scale)(mat,a);
5496: PetscObjectStateIncrease((PetscObject)mat);
5497: }
5498: PetscLogEventEnd(MAT_Scale,mat,0,0,0);
5499: return(0);
5500: }
5502: /*@
5503: MatNorm - Calculates various norms of a matrix.
5505: Collective on Mat
5507: Input Parameters:
5508: + mat - the matrix
5509: - type - the type of norm, NORM_1, NORM_FROBENIUS, NORM_INFINITY
5511: Output Parameter:
5512: . nrm - the resulting norm
5514: Level: intermediate
5516: @*/
5517: PetscErrorCode MatNorm(Mat mat,NormType type,PetscReal *nrm)
5518: {
5526: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5527: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5528: if (!mat->ops->norm) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5529: MatCheckPreallocated(mat,1);
5531: (*mat->ops->norm)(mat,type,nrm);
5532: return(0);
5533: }
5535: /*
5536: This variable is used to prevent counting of MatAssemblyBegin() that
5537: are called from within a MatAssemblyEnd().
5538: */
5539: static PetscInt MatAssemblyEnd_InUse = 0;
5540: /*@
5541: MatAssemblyBegin - Begins assembling the matrix. This routine should
5542: be called after completing all calls to MatSetValues().
5544: Collective on Mat
5546: Input Parameters:
5547: + mat - the matrix
5548: - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY
5550: Notes:
5551: MatSetValues() generally caches the values. The matrix is ready to
5552: use only after MatAssemblyBegin() and MatAssemblyEnd() have been called.
5553: Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES
5554: in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before
5555: using the matrix.
5557: ALL processes that share a matrix MUST call MatAssemblyBegin() and MatAssemblyEnd() the SAME NUMBER of times, and each time with the
5558: 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
5559: a global collective operation requring all processes that share the matrix.
5561: Space for preallocated nonzeros that is not filled by a call to MatSetValues() or a related routine are compressed
5562: out by assembly. If you intend to use that extra space on a subsequent assembly, be sure to insert explicit zeros
5563: before MAT_FINAL_ASSEMBLY so the space is not compressed out.
5565: Level: beginner
5567: .seealso: MatAssemblyEnd(), MatSetValues(), MatAssembled()
5568: @*/
5569: PetscErrorCode MatAssemblyBegin(Mat mat,MatAssemblyType type)
5570: {
5576: MatCheckPreallocated(mat,1);
5577: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix.\nDid you forget to call MatSetUnfactored()?");
5578: if (mat->assembled) {
5579: mat->was_assembled = PETSC_TRUE;
5580: mat->assembled = PETSC_FALSE;
5581: }
5583: if (!MatAssemblyEnd_InUse) {
5584: PetscLogEventBegin(MAT_AssemblyBegin,mat,0,0,0);
5585: if (mat->ops->assemblybegin) {(*mat->ops->assemblybegin)(mat,type);}
5586: PetscLogEventEnd(MAT_AssemblyBegin,mat,0,0,0);
5587: } else if (mat->ops->assemblybegin) {
5588: (*mat->ops->assemblybegin)(mat,type);
5589: }
5590: return(0);
5591: }
5593: /*@
5594: MatAssembled - Indicates if a matrix has been assembled and is ready for
5595: use; for example, in matrix-vector product.
5597: Not Collective
5599: Input Parameter:
5600: . mat - the matrix
5602: Output Parameter:
5603: . assembled - PETSC_TRUE or PETSC_FALSE
5605: Level: advanced
5607: .seealso: MatAssemblyEnd(), MatSetValues(), MatAssemblyBegin()
5608: @*/
5609: PetscErrorCode MatAssembled(Mat mat,PetscBool *assembled)
5610: {
5614: *assembled = mat->assembled;
5615: return(0);
5616: }
5618: /*@
5619: MatAssemblyEnd - Completes assembling the matrix. This routine should
5620: be called after MatAssemblyBegin().
5622: Collective on Mat
5624: Input Parameters:
5625: + mat - the matrix
5626: - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY
5628: Options Database Keys:
5629: + -mat_view ::ascii_info - Prints info on matrix at conclusion of MatEndAssembly()
5630: . -mat_view ::ascii_info_detail - Prints more detailed info
5631: . -mat_view - Prints matrix in ASCII format
5632: . -mat_view ::ascii_matlab - Prints matrix in Matlab format
5633: . -mat_view draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX().
5634: . -display <name> - Sets display name (default is host)
5635: . -draw_pause <sec> - Sets number of seconds to pause after display
5636: . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (See Users-Manual: ch_matlab)
5637: . -viewer_socket_machine <machine> - Machine to use for socket
5638: . -viewer_socket_port <port> - Port number to use for socket
5639: - -mat_view binary:filename[:append] - Save matrix to file in binary format
5641: Notes:
5642: MatSetValues() generally caches the values. The matrix is ready to
5643: use only after MatAssemblyBegin() and MatAssemblyEnd() have been called.
5644: Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES
5645: in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before
5646: using the matrix.
5648: Space for preallocated nonzeros that is not filled by a call to MatSetValues() or a related routine are compressed
5649: out by assembly. If you intend to use that extra space on a subsequent assembly, be sure to insert explicit zeros
5650: before MAT_FINAL_ASSEMBLY so the space is not compressed out.
5652: Level: beginner
5654: .seealso: MatAssemblyBegin(), MatSetValues(), PetscDrawOpenX(), PetscDrawCreate(), MatView(), MatAssembled(), PetscViewerSocketOpen()
5655: @*/
5656: PetscErrorCode MatAssemblyEnd(Mat mat,MatAssemblyType type)
5657: {
5658: PetscErrorCode ierr;
5659: static PetscInt inassm = 0;
5660: PetscBool flg = PETSC_FALSE;
5666: inassm++;
5667: MatAssemblyEnd_InUse++;
5668: if (MatAssemblyEnd_InUse == 1) { /* Do the logging only the first time through */
5669: PetscLogEventBegin(MAT_AssemblyEnd,mat,0,0,0);
5670: if (mat->ops->assemblyend) {
5671: (*mat->ops->assemblyend)(mat,type);
5672: }
5673: PetscLogEventEnd(MAT_AssemblyEnd,mat,0,0,0);
5674: } else if (mat->ops->assemblyend) {
5675: (*mat->ops->assemblyend)(mat,type);
5676: }
5678: /* Flush assembly is not a true assembly */
5679: if (type != MAT_FLUSH_ASSEMBLY) {
5680: mat->num_ass++;
5681: mat->assembled = PETSC_TRUE;
5682: mat->ass_nonzerostate = mat->nonzerostate;
5683: }
5685: mat->insertmode = NOT_SET_VALUES;
5686: MatAssemblyEnd_InUse--;
5687: PetscObjectStateIncrease((PetscObject)mat);
5688: if (!mat->symmetric_eternal) {
5689: mat->symmetric_set = PETSC_FALSE;
5690: mat->hermitian_set = PETSC_FALSE;
5691: mat->structurally_symmetric_set = PETSC_FALSE;
5692: }
5693: if (inassm == 1 && type != MAT_FLUSH_ASSEMBLY) {
5694: MatViewFromOptions(mat,NULL,"-mat_view");
5696: if (mat->checksymmetryonassembly) {
5697: MatIsSymmetric(mat,mat->checksymmetrytol,&flg);
5698: if (flg) {
5699: PetscPrintf(PetscObjectComm((PetscObject)mat),"Matrix is symmetric (tolerance %g)\n",(double)mat->checksymmetrytol);
5700: } else {
5701: PetscPrintf(PetscObjectComm((PetscObject)mat),"Matrix is not symmetric (tolerance %g)\n",(double)mat->checksymmetrytol);
5702: }
5703: }
5704: if (mat->nullsp && mat->checknullspaceonassembly) {
5705: MatNullSpaceTest(mat->nullsp,mat,NULL);
5706: }
5707: }
5708: inassm--;
5709: return(0);
5710: }
5712: /*@
5713: MatSetOption - Sets a parameter option for a matrix. Some options
5714: may be specific to certain storage formats. Some options
5715: determine how values will be inserted (or added). Sorted,
5716: row-oriented input will generally assemble the fastest. The default
5717: is row-oriented.
5719: Logically Collective on Mat for certain operations, such as MAT_SPD, not collective for MAT_ROW_ORIENTED, see MatOption
5721: Input Parameters:
5722: + mat - the matrix
5723: . option - the option, one of those listed below (and possibly others),
5724: - flg - turn the option on (PETSC_TRUE) or off (PETSC_FALSE)
5726: Options Describing Matrix Structure:
5727: + MAT_SPD - symmetric positive definite
5728: . MAT_SYMMETRIC - symmetric in terms of both structure and value
5729: . MAT_HERMITIAN - transpose is the complex conjugation
5730: . MAT_STRUCTURALLY_SYMMETRIC - symmetric nonzero structure
5731: - MAT_SYMMETRY_ETERNAL - if you would like the symmetry/Hermitian flag
5732: you set to be kept with all future use of the matrix
5733: including after MatAssemblyBegin/End() which could
5734: potentially change the symmetry structure, i.e. you
5735: KNOW the matrix will ALWAYS have the property you set.
5736: Note that setting this flag alone implies nothing about whether the matrix is symmetric/Hermitian;
5737: the relevant flags must be set independently.
5739: Options For Use with MatSetValues():
5740: Insert a logically dense subblock, which can be
5741: . MAT_ROW_ORIENTED - row-oriented (default)
5743: Note these options reflect the data you pass in with MatSetValues(); it has
5744: nothing to do with how the data is stored internally in the matrix
5745: data structure.
5747: When (re)assembling a matrix, we can restrict the input for
5748: efficiency/debugging purposes. These options include:
5749: + MAT_NEW_NONZERO_LOCATIONS - additional insertions will be allowed if they generate a new nonzero (slow)
5750: . MAT_FORCE_DIAGONAL_ENTRIES - forces diagonal entries to be allocated
5751: . MAT_IGNORE_OFF_PROC_ENTRIES - drops off-processor entries
5752: . MAT_NEW_NONZERO_LOCATION_ERR - generates an error for new matrix entry
5753: . MAT_USE_HASH_TABLE - uses a hash table to speed up matrix assembly
5754: . MAT_NO_OFF_PROC_ENTRIES - you know each process will only set values for its own rows, will generate an error if
5755: any process sets values for another process. This avoids all reductions in the MatAssembly routines and thus improves
5756: performance for very large process counts.
5757: - MAT_SUBSET_OFF_PROC_ENTRIES - you know that the first assembly after setting this flag will set a superset
5758: of the off-process entries required for all subsequent assemblies. This avoids a rendezvous step in the MatAssembly
5759: functions, instead sending only neighbor messages.
5761: Notes:
5762: Except for MAT_UNUSED_NONZERO_LOCATION_ERR and MAT_ROW_ORIENTED all processes that share the matrix must pass the same value in flg!
5764: Some options are relevant only for particular matrix types and
5765: are thus ignored by others. Other options are not supported by
5766: certain matrix types and will generate an error message if set.
5768: If using a Fortran 77 module to compute a matrix, one may need to
5769: use the column-oriented option (or convert to the row-oriented
5770: format).
5772: MAT_NEW_NONZERO_LOCATIONS set to PETSC_FALSE indicates that any add or insertion
5773: that would generate a new entry in the nonzero structure is instead
5774: ignored. Thus, if memory has not alredy been allocated for this particular
5775: data, then the insertion is ignored. For dense matrices, in which
5776: the entire array is allocated, no entries are ever ignored.
5777: Set after the first MatAssemblyEnd(). If this option is set then the MatAssemblyBegin/End() processes has one less global reduction
5779: MAT_NEW_NONZERO_LOCATION_ERR set to PETSC_TRUE indicates that any add or insertion
5780: that would generate a new entry in the nonzero structure instead produces
5781: an error. (Currently supported for AIJ and BAIJ formats only.) If this option is set then the MatAssemblyBegin/End() processes has one less global reduction
5783: MAT_NEW_NONZERO_ALLOCATION_ERR set to PETSC_TRUE indicates that any add or insertion
5784: that would generate a new entry that has not been preallocated will
5785: instead produce an error. (Currently supported for AIJ and BAIJ formats
5786: only.) This is a useful flag when debugging matrix memory preallocation.
5787: If this option is set then the MatAssemblyBegin/End() processes has one less global reduction
5789: MAT_IGNORE_OFF_PROC_ENTRIES set to PETSC_TRUE indicates entries destined for
5790: other processors should be dropped, rather than stashed.
5791: This is useful if you know that the "owning" processor is also
5792: always generating the correct matrix entries, so that PETSc need
5793: not transfer duplicate entries generated on another processor.
5795: MAT_USE_HASH_TABLE indicates that a hash table be used to improve the
5796: searches during matrix assembly. When this flag is set, the hash table
5797: is created during the first Matrix Assembly. This hash table is
5798: used the next time through, during MatSetVaules()/MatSetVaulesBlocked()
5799: to improve the searching of indices. MAT_NEW_NONZERO_LOCATIONS flag
5800: should be used with MAT_USE_HASH_TABLE flag. This option is currently
5801: supported by MATMPIBAIJ format only.
5803: MAT_KEEP_NONZERO_PATTERN indicates when MatZeroRows() is called the zeroed entries
5804: are kept in the nonzero structure
5806: MAT_IGNORE_ZERO_ENTRIES - for AIJ/IS matrices this will stop zero values from creating
5807: a zero location in the matrix
5809: MAT_USE_INODES - indicates using inode version of the code - works with AIJ matrix types
5811: MAT_NO_OFF_PROC_ZERO_ROWS - you know each process will only zero its own rows. This avoids all reductions in the
5812: zero row routines and thus improves performance for very large process counts.
5814: MAT_IGNORE_LOWER_TRIANGULAR - For SBAIJ matrices will ignore any insertions you make in the lower triangular
5815: part of the matrix (since they should match the upper triangular part).
5817: MAT_SORTED_FULL - each process provides exactly its local rows; all column indices for a given row are passed in a
5818: single call to MatSetValues(), preallocation is perfect, row oriented, INSERT_VALUES is used. Common
5819: with finite difference schemes with non-periodic boundary conditions.
5821: Level: intermediate
5823: .seealso: MatOption, Mat
5825: @*/
5826: PetscErrorCode MatSetOption(Mat mat,MatOption op,PetscBool flg)
5827: {
5832: if (op > 0) {
5835: }
5837: if (((int) op) <= MAT_OPTION_MIN || ((int) op) >= MAT_OPTION_MAX) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Options %d is out of range",(int)op);
5839: switch (op) {
5840: case MAT_FORCE_DIAGONAL_ENTRIES:
5841: mat->force_diagonals = flg;
5842: return(0);
5843: case MAT_NO_OFF_PROC_ENTRIES:
5844: mat->nooffprocentries = flg;
5845: return(0);
5846: case MAT_SUBSET_OFF_PROC_ENTRIES:
5847: mat->assembly_subset = flg;
5848: if (!mat->assembly_subset) { /* See the same logic in VecAssembly wrt VEC_SUBSET_OFF_PROC_ENTRIES */
5849: #if !defined(PETSC_HAVE_MPIUNI)
5850: MatStashScatterDestroy_BTS(&mat->stash);
5851: #endif
5852: mat->stash.first_assembly_done = PETSC_FALSE;
5853: }
5854: return(0);
5855: case MAT_NO_OFF_PROC_ZERO_ROWS:
5856: mat->nooffproczerorows = flg;
5857: return(0);
5858: case MAT_SPD:
5859: mat->spd_set = PETSC_TRUE;
5860: mat->spd = flg;
5861: if (flg) {
5862: mat->symmetric = PETSC_TRUE;
5863: mat->structurally_symmetric = PETSC_TRUE;
5864: mat->symmetric_set = PETSC_TRUE;
5865: mat->structurally_symmetric_set = PETSC_TRUE;
5866: }
5867: break;
5868: case MAT_SYMMETRIC:
5869: mat->symmetric = flg;
5870: if (flg) mat->structurally_symmetric = PETSC_TRUE;
5871: mat->symmetric_set = PETSC_TRUE;
5872: mat->structurally_symmetric_set = flg;
5873: #if !defined(PETSC_USE_COMPLEX)
5874: mat->hermitian = flg;
5875: mat->hermitian_set = PETSC_TRUE;
5876: #endif
5877: break;
5878: case MAT_HERMITIAN:
5879: mat->hermitian = flg;
5880: if (flg) mat->structurally_symmetric = PETSC_TRUE;
5881: mat->hermitian_set = PETSC_TRUE;
5882: mat->structurally_symmetric_set = flg;
5883: #if !defined(PETSC_USE_COMPLEX)
5884: mat->symmetric = flg;
5885: mat->symmetric_set = PETSC_TRUE;
5886: #endif
5887: break;
5888: case MAT_STRUCTURALLY_SYMMETRIC:
5889: mat->structurally_symmetric = flg;
5890: mat->structurally_symmetric_set = PETSC_TRUE;
5891: break;
5892: case MAT_SYMMETRY_ETERNAL:
5893: mat->symmetric_eternal = flg;
5894: break;
5895: case MAT_STRUCTURE_ONLY:
5896: mat->structure_only = flg;
5897: break;
5898: case MAT_SORTED_FULL:
5899: mat->sortedfull = flg;
5900: break;
5901: default:
5902: break;
5903: }
5904: if (mat->ops->setoption) {
5905: (*mat->ops->setoption)(mat,op,flg);
5906: }
5907: return(0);
5908: }
5910: /*@
5911: MatGetOption - Gets a parameter option that has been set for a matrix.
5913: Logically Collective on Mat for certain operations, such as MAT_SPD, not collective for MAT_ROW_ORIENTED, see MatOption
5915: Input Parameters:
5916: + mat - the matrix
5917: - option - the option, this only responds to certain options, check the code for which ones
5919: Output Parameter:
5920: . flg - turn the option on (PETSC_TRUE) or off (PETSC_FALSE)
5922: Notes:
5923: Can only be called after MatSetSizes() and MatSetType() have been set.
5925: Level: intermediate
5927: .seealso: MatOption, MatSetOption()
5929: @*/
5930: PetscErrorCode MatGetOption(Mat mat,MatOption op,PetscBool *flg)
5931: {
5936: if (((int) op) <= MAT_OPTION_MIN || ((int) op) >= MAT_OPTION_MAX) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Options %d is out of range",(int)op);
5937: if (!((PetscObject)mat)->type_name) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_TYPENOTSET,"Cannot get options until type and size have been set, see MatSetType() and MatSetSizes()");
5939: switch (op) {
5940: case MAT_NO_OFF_PROC_ENTRIES:
5941: *flg = mat->nooffprocentries;
5942: break;
5943: case MAT_NO_OFF_PROC_ZERO_ROWS:
5944: *flg = mat->nooffproczerorows;
5945: break;
5946: case MAT_SYMMETRIC:
5947: *flg = mat->symmetric;
5948: break;
5949: case MAT_HERMITIAN:
5950: *flg = mat->hermitian;
5951: break;
5952: case MAT_STRUCTURALLY_SYMMETRIC:
5953: *flg = mat->structurally_symmetric;
5954: break;
5955: case MAT_SYMMETRY_ETERNAL:
5956: *flg = mat->symmetric_eternal;
5957: break;
5958: case MAT_SPD:
5959: *flg = mat->spd;
5960: break;
5961: default:
5962: break;
5963: }
5964: return(0);
5965: }
5967: /*@
5968: MatZeroEntries - Zeros all entries of a matrix. For sparse matrices
5969: this routine retains the old nonzero structure.
5971: Logically Collective on Mat
5973: Input Parameters:
5974: . mat - the matrix
5976: Level: intermediate
5978: Notes:
5979: 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.
5980: See the Performance chapter of the users manual for information on preallocating matrices.
5982: .seealso: MatZeroRows()
5983: @*/
5984: PetscErrorCode MatZeroEntries(Mat mat)
5985: {
5991: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5992: if (mat->insertmode != NOT_SET_VALUES) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for matrices where you have set values but not yet assembled");
5993: if (!mat->ops->zeroentries) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5994: MatCheckPreallocated(mat,1);
5996: PetscLogEventBegin(MAT_ZeroEntries,mat,0,0,0);
5997: (*mat->ops->zeroentries)(mat);
5998: PetscLogEventEnd(MAT_ZeroEntries,mat,0,0,0);
5999: PetscObjectStateIncrease((PetscObject)mat);
6000: return(0);
6001: }
6003: /*@
6004: MatZeroRowsColumns - Zeros all entries (except possibly the main diagonal)
6005: of a set of rows and columns of a matrix.
6007: Collective on Mat
6009: Input Parameters:
6010: + mat - the matrix
6011: . numRows - the number of rows to remove
6012: . rows - the global row indices
6013: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6014: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6015: - b - optional vector of right hand side, that will be adjusted by provided solution
6017: Notes:
6018: This does not change the nonzero structure of the matrix, it merely zeros those entries in the matrix.
6020: The user can set a value in the diagonal entry (or for the AIJ and
6021: row formats can optionally remove the main diagonal entry from the
6022: nonzero structure as well, by passing 0.0 as the final argument).
6024: For the parallel case, all processes that share the matrix (i.e.,
6025: those in the communicator used for matrix creation) MUST call this
6026: routine, regardless of whether any rows being zeroed are owned by
6027: them.
6029: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6030: list only rows local to itself).
6032: The option MAT_NO_OFF_PROC_ZERO_ROWS does not apply to this routine.
6034: Level: intermediate
6036: .seealso: MatZeroRowsIS(), MatZeroRows(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6037: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6038: @*/
6039: PetscErrorCode MatZeroRowsColumns(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
6040: {
6047: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6048: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6049: if (!mat->ops->zerorowscolumns) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6050: MatCheckPreallocated(mat,1);
6052: (*mat->ops->zerorowscolumns)(mat,numRows,rows,diag,x,b);
6053: MatViewFromOptions(mat,NULL,"-mat_view");
6054: PetscObjectStateIncrease((PetscObject)mat);
6055: return(0);
6056: }
6058: /*@
6059: MatZeroRowsColumnsIS - Zeros all entries (except possibly the main diagonal)
6060: of a set of rows and columns of a matrix.
6062: Collective on Mat
6064: Input Parameters:
6065: + mat - the matrix
6066: . is - the rows to zero
6067: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6068: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6069: - b - optional vector of right hand side, that will be adjusted by provided solution
6071: Notes:
6072: This does not change the nonzero structure of the matrix, it merely zeros those entries in the matrix.
6074: The user can set a value in the diagonal entry (or for the AIJ and
6075: row formats can optionally remove the main diagonal entry from the
6076: nonzero structure as well, by passing 0.0 as the final argument).
6078: For the parallel case, all processes that share the matrix (i.e.,
6079: those in the communicator used for matrix creation) MUST call this
6080: routine, regardless of whether any rows being zeroed are owned by
6081: them.
6083: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6084: list only rows local to itself).
6086: The option MAT_NO_OFF_PROC_ZERO_ROWS does not apply to this routine.
6088: Level: intermediate
6090: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6091: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRows(), MatZeroRowsColumnsStencil()
6092: @*/
6093: PetscErrorCode MatZeroRowsColumnsIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b)
6094: {
6096: PetscInt numRows;
6097: const PetscInt *rows;
6104: ISGetLocalSize(is,&numRows);
6105: ISGetIndices(is,&rows);
6106: MatZeroRowsColumns(mat,numRows,rows,diag,x,b);
6107: ISRestoreIndices(is,&rows);
6108: return(0);
6109: }
6111: /*@
6112: MatZeroRows - Zeros all entries (except possibly the main diagonal)
6113: of a set of rows of a matrix.
6115: Collective on Mat
6117: Input Parameters:
6118: + mat - the matrix
6119: . numRows - the number of rows to remove
6120: . rows - the global row indices
6121: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6122: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6123: - b - optional vector of right hand side, that will be adjusted by provided solution
6125: Notes:
6126: For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
6127: but does not release memory. For the dense and block diagonal
6128: formats this does not alter the nonzero structure.
6130: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6131: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6132: merely zeroed.
6134: The user can set a value in the diagonal entry (or for the AIJ and
6135: row formats can optionally remove the main diagonal entry from the
6136: nonzero structure as well, by passing 0.0 as the final argument).
6138: For the parallel case, all processes that share the matrix (i.e.,
6139: those in the communicator used for matrix creation) MUST call this
6140: routine, regardless of whether any rows being zeroed are owned by
6141: them.
6143: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6144: list only rows local to itself).
6146: You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it
6147: owns that are to be zeroed. This saves a global synchronization in the implementation.
6149: Level: intermediate
6151: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6152: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6153: @*/
6154: PetscErrorCode MatZeroRows(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
6155: {
6162: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6163: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6164: if (!mat->ops->zerorows) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6165: MatCheckPreallocated(mat,1);
6167: (*mat->ops->zerorows)(mat,numRows,rows,diag,x,b);
6168: MatViewFromOptions(mat,NULL,"-mat_view");
6169: PetscObjectStateIncrease((PetscObject)mat);
6170: return(0);
6171: }
6173: /*@
6174: MatZeroRowsIS - Zeros all entries (except possibly the main diagonal)
6175: of a set of rows of a matrix.
6177: Collective on Mat
6179: Input Parameters:
6180: + mat - the matrix
6181: . is - index set of rows to remove (if NULL then no row is removed)
6182: . diag - value put in all diagonals of eliminated rows
6183: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6184: - b - optional vector of right hand side, that will be adjusted by provided solution
6186: Notes:
6187: For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
6188: but does not release memory. For the dense and block diagonal
6189: formats this does not alter the nonzero structure.
6191: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6192: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6193: merely zeroed.
6195: The user can set a value in the diagonal entry (or for the AIJ and
6196: row formats can optionally remove the main diagonal entry from the
6197: nonzero structure as well, by passing 0.0 as the final argument).
6199: For the parallel case, all processes that share the matrix (i.e.,
6200: those in the communicator used for matrix creation) MUST call this
6201: routine, regardless of whether any rows being zeroed are owned by
6202: them.
6204: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6205: list only rows local to itself).
6207: You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it
6208: owns that are to be zeroed. This saves a global synchronization in the implementation.
6210: Level: intermediate
6212: .seealso: MatZeroRows(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6213: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6214: @*/
6215: PetscErrorCode MatZeroRowsIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b)
6216: {
6217: PetscInt numRows = 0;
6218: const PetscInt *rows = NULL;
6224: if (is) {
6226: ISGetLocalSize(is,&numRows);
6227: ISGetIndices(is,&rows);
6228: }
6229: MatZeroRows(mat,numRows,rows,diag,x,b);
6230: if (is) {
6231: ISRestoreIndices(is,&rows);
6232: }
6233: return(0);
6234: }
6236: /*@
6237: MatZeroRowsStencil - Zeros all entries (except possibly the main diagonal)
6238: of a set of rows of a matrix. These rows must be local to the process.
6240: Collective on Mat
6242: Input Parameters:
6243: + mat - the matrix
6244: . numRows - the number of rows to remove
6245: . rows - the grid coordinates (and component number when dof > 1) for matrix rows
6246: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
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: Notes:
6251: For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
6252: but does not release memory. For the dense and block diagonal
6253: formats this does not alter the nonzero structure.
6255: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6256: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6257: merely zeroed.
6259: The user can set a value in the diagonal entry (or for the AIJ and
6260: row formats can optionally remove the main diagonal entry from the
6261: nonzero structure as well, by passing 0.0 as the final argument).
6263: For the parallel case, all processes that share the matrix (i.e.,
6264: those in the communicator used for matrix creation) MUST call this
6265: routine, regardless of whether any rows being zeroed are owned by
6266: them.
6268: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6269: list only rows local to itself).
6271: The grid coordinates are across the entire grid, not just the local portion
6273: In Fortran idxm and idxn should be declared as
6274: $ MatStencil idxm(4,m)
6275: and the values inserted using
6276: $ idxm(MatStencil_i,1) = i
6277: $ idxm(MatStencil_j,1) = j
6278: $ idxm(MatStencil_k,1) = k
6279: $ idxm(MatStencil_c,1) = c
6280: etc
6282: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
6283: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
6284: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
6285: DM_BOUNDARY_PERIODIC boundary type.
6287: 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
6288: a single value per point) you can skip filling those indices.
6290: Level: intermediate
6292: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsl(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6293: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6294: @*/
6295: PetscErrorCode MatZeroRowsStencil(Mat mat,PetscInt numRows,const MatStencil rows[],PetscScalar diag,Vec x,Vec b)
6296: {
6297: PetscInt dim = mat->stencil.dim;
6298: PetscInt sdim = dim - (1 - (PetscInt) mat->stencil.noc);
6299: PetscInt *dims = mat->stencil.dims+1;
6300: PetscInt *starts = mat->stencil.starts;
6301: PetscInt *dxm = (PetscInt*) rows;
6302: PetscInt *jdxm, i, j, tmp, numNewRows = 0;
6310: PetscMalloc1(numRows, &jdxm);
6311: for (i = 0; i < numRows; ++i) {
6312: /* Skip unused dimensions (they are ordered k, j, i, c) */
6313: for (j = 0; j < 3-sdim; ++j) dxm++;
6314: /* Local index in X dir */
6315: tmp = *dxm++ - starts[0];
6316: /* Loop over remaining dimensions */
6317: for (j = 0; j < dim-1; ++j) {
6318: /* If nonlocal, set index to be negative */
6319: if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
6320: /* Update local index */
6321: else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
6322: }
6323: /* Skip component slot if necessary */
6324: if (mat->stencil.noc) dxm++;
6325: /* Local row number */
6326: if (tmp >= 0) {
6327: jdxm[numNewRows++] = tmp;
6328: }
6329: }
6330: MatZeroRowsLocal(mat,numNewRows,jdxm,diag,x,b);
6331: PetscFree(jdxm);
6332: return(0);
6333: }
6335: /*@
6336: MatZeroRowsColumnsStencil - Zeros all row and column entries (except possibly the main diagonal)
6337: of a set of rows and columns of a matrix.
6339: Collective on Mat
6341: Input Parameters:
6342: + mat - the matrix
6343: . numRows - the number of rows/columns to remove
6344: . rows - the grid coordinates (and component number when dof > 1) for matrix rows
6345: . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
6346: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6347: - b - optional vector of right hand side, that will be adjusted by provided solution
6349: Notes:
6350: For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
6351: but does not release memory. For the dense and block diagonal
6352: formats this does not alter the nonzero structure.
6354: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6355: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6356: merely zeroed.
6358: The user can set a value in the diagonal entry (or for the AIJ and
6359: row formats can optionally remove the main diagonal entry from the
6360: nonzero structure as well, by passing 0.0 as the final argument).
6362: For the parallel case, all processes that share the matrix (i.e.,
6363: those in the communicator used for matrix creation) MUST call this
6364: routine, regardless of whether any rows being zeroed are owned by
6365: them.
6367: Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
6368: list only rows local to itself, but the row/column numbers are given in local numbering).
6370: The grid coordinates are across the entire grid, not just the local portion
6372: In Fortran idxm and idxn should be declared as
6373: $ MatStencil idxm(4,m)
6374: and the values inserted using
6375: $ idxm(MatStencil_i,1) = i
6376: $ idxm(MatStencil_j,1) = j
6377: $ idxm(MatStencil_k,1) = k
6378: $ idxm(MatStencil_c,1) = c
6379: etc
6381: For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
6382: obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
6383: etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the
6384: DM_BOUNDARY_PERIODIC boundary type.
6386: 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
6387: a single value per point) you can skip filling those indices.
6389: Level: intermediate
6391: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6392: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRows()
6393: @*/
6394: PetscErrorCode MatZeroRowsColumnsStencil(Mat mat,PetscInt numRows,const MatStencil rows[],PetscScalar diag,Vec x,Vec b)
6395: {
6396: PetscInt dim = mat->stencil.dim;
6397: PetscInt sdim = dim - (1 - (PetscInt) mat->stencil.noc);
6398: PetscInt *dims = mat->stencil.dims+1;
6399: PetscInt *starts = mat->stencil.starts;
6400: PetscInt *dxm = (PetscInt*) rows;
6401: PetscInt *jdxm, i, j, tmp, numNewRows = 0;
6409: PetscMalloc1(numRows, &jdxm);
6410: for (i = 0; i < numRows; ++i) {
6411: /* Skip unused dimensions (they are ordered k, j, i, c) */
6412: for (j = 0; j < 3-sdim; ++j) dxm++;
6413: /* Local index in X dir */
6414: tmp = *dxm++ - starts[0];
6415: /* Loop over remaining dimensions */
6416: for (j = 0; j < dim-1; ++j) {
6417: /* If nonlocal, set index to be negative */
6418: if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
6419: /* Update local index */
6420: else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
6421: }
6422: /* Skip component slot if necessary */
6423: if (mat->stencil.noc) dxm++;
6424: /* Local row number */
6425: if (tmp >= 0) {
6426: jdxm[numNewRows++] = tmp;
6427: }
6428: }
6429: MatZeroRowsColumnsLocal(mat,numNewRows,jdxm,diag,x,b);
6430: PetscFree(jdxm);
6431: return(0);
6432: }
6434: /*@C
6435: MatZeroRowsLocal - Zeros all entries (except possibly the main diagonal)
6436: of a set of rows of a matrix; using local numbering of rows.
6438: Collective on Mat
6440: Input Parameters:
6441: + mat - the matrix
6442: . numRows - the number of rows to remove
6443: . rows - the local row indices
6444: . diag - value put in all diagonals of eliminated rows
6445: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6446: - b - optional vector of right hand side, that will be adjusted by provided solution
6448: Notes:
6449: Before calling MatZeroRowsLocal(), the user must first set the
6450: local-to-global mapping by calling MatSetLocalToGlobalMapping().
6452: For the AIJ matrix formats this removes the old nonzero structure,
6453: but does not release memory. For the dense and block diagonal
6454: formats this does not alter the nonzero structure.
6456: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6457: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6458: merely zeroed.
6460: The user can set a value in the diagonal entry (or for the AIJ and
6461: row formats can optionally remove the main diagonal entry from the
6462: nonzero structure as well, by passing 0.0 as the final argument).
6464: You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it
6465: owns that are to be zeroed. This saves a global synchronization in the implementation.
6467: Level: intermediate
6469: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRows(), MatSetOption(),
6470: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6471: @*/
6472: PetscErrorCode MatZeroRowsLocal(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
6473: {
6480: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6481: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6482: MatCheckPreallocated(mat,1);
6484: if (mat->ops->zerorowslocal) {
6485: (*mat->ops->zerorowslocal)(mat,numRows,rows,diag,x,b);
6486: } else {
6487: IS is, newis;
6488: const PetscInt *newRows;
6490: if (!mat->rmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Need to provide local to global mapping to matrix first");
6491: ISCreateGeneral(PETSC_COMM_SELF,numRows,rows,PETSC_COPY_VALUES,&is);
6492: ISLocalToGlobalMappingApplyIS(mat->rmap->mapping,is,&newis);
6493: ISGetIndices(newis,&newRows);
6494: (*mat->ops->zerorows)(mat,numRows,newRows,diag,x,b);
6495: ISRestoreIndices(newis,&newRows);
6496: ISDestroy(&newis);
6497: ISDestroy(&is);
6498: }
6499: PetscObjectStateIncrease((PetscObject)mat);
6500: return(0);
6501: }
6503: /*@
6504: MatZeroRowsLocalIS - Zeros all entries (except possibly the main diagonal)
6505: of a set of rows of a matrix; using local numbering of rows.
6507: Collective on Mat
6509: Input Parameters:
6510: + mat - the matrix
6511: . is - index set of rows to remove
6512: . diag - value put in all diagonals of eliminated rows
6513: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6514: - b - optional vector of right hand side, that will be adjusted by provided solution
6516: Notes:
6517: Before calling MatZeroRowsLocalIS(), the user must first set the
6518: local-to-global mapping by calling MatSetLocalToGlobalMapping().
6520: For the AIJ matrix formats this removes the old nonzero structure,
6521: but does not release memory. For the dense and block diagonal
6522: formats this does not alter the nonzero structure.
6524: If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
6525: of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
6526: merely zeroed.
6528: The user can set a value in the diagonal entry (or for the AIJ and
6529: row formats can optionally remove the main diagonal entry from the
6530: nonzero structure as well, by passing 0.0 as the final argument).
6532: You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it
6533: owns that are to be zeroed. This saves a global synchronization in the implementation.
6535: Level: intermediate
6537: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRows(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6538: MatZeroRowsColumnsLocal(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6539: @*/
6540: PetscErrorCode MatZeroRowsLocalIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b)
6541: {
6543: PetscInt numRows;
6544: const PetscInt *rows;
6550: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6551: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6552: MatCheckPreallocated(mat,1);
6554: ISGetLocalSize(is,&numRows);
6555: ISGetIndices(is,&rows);
6556: MatZeroRowsLocal(mat,numRows,rows,diag,x,b);
6557: ISRestoreIndices(is,&rows);
6558: return(0);
6559: }
6561: /*@
6562: MatZeroRowsColumnsLocal - Zeros all entries (except possibly the main diagonal)
6563: of a set of rows and columns of a matrix; using local numbering of rows.
6565: Collective on Mat
6567: Input Parameters:
6568: + mat - the matrix
6569: . numRows - the number of rows to remove
6570: . rows - the global row indices
6571: . diag - value put in all diagonals of eliminated rows
6572: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6573: - b - optional vector of right hand side, that will be adjusted by provided solution
6575: Notes:
6576: Before calling MatZeroRowsColumnsLocal(), the user must first set the
6577: local-to-global mapping by calling MatSetLocalToGlobalMapping().
6579: The user can set a value in the diagonal entry (or for the AIJ and
6580: row formats can optionally remove the main diagonal entry from the
6581: nonzero structure as well, by passing 0.0 as the final argument).
6583: Level: intermediate
6585: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6586: MatZeroRows(), MatZeroRowsColumnsLocalIS(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6587: @*/
6588: PetscErrorCode MatZeroRowsColumnsLocal(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
6589: {
6591: IS is, newis;
6592: const PetscInt *newRows;
6598: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6599: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6600: MatCheckPreallocated(mat,1);
6602: if (!mat->cmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Need to provide local to global mapping to matrix first");
6603: ISCreateGeneral(PETSC_COMM_SELF,numRows,rows,PETSC_COPY_VALUES,&is);
6604: ISLocalToGlobalMappingApplyIS(mat->cmap->mapping,is,&newis);
6605: ISGetIndices(newis,&newRows);
6606: (*mat->ops->zerorowscolumns)(mat,numRows,newRows,diag,x,b);
6607: ISRestoreIndices(newis,&newRows);
6608: ISDestroy(&newis);
6609: ISDestroy(&is);
6610: PetscObjectStateIncrease((PetscObject)mat);
6611: return(0);
6612: }
6614: /*@
6615: MatZeroRowsColumnsLocalIS - Zeros all entries (except possibly the main diagonal)
6616: of a set of rows and columns of a matrix; using local numbering of rows.
6618: Collective on Mat
6620: Input Parameters:
6621: + mat - the matrix
6622: . is - index set of rows to remove
6623: . diag - value put in all diagonals of eliminated rows
6624: . x - optional vector of solutions for zeroed rows (other entries in vector are not used)
6625: - b - optional vector of right hand side, that will be adjusted by provided solution
6627: Notes:
6628: Before calling MatZeroRowsColumnsLocalIS(), the user must first set the
6629: local-to-global mapping by calling MatSetLocalToGlobalMapping().
6631: The user can set a value in the diagonal entry (or for the AIJ and
6632: row formats can optionally remove the main diagonal entry from the
6633: nonzero structure as well, by passing 0.0 as the final argument).
6635: Level: intermediate
6637: .seealso: MatZeroRowsIS(), MatZeroRowsColumns(), MatZeroRowsLocalIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(),
6638: MatZeroRowsColumnsLocal(), MatZeroRows(), MatZeroRowsColumnsIS(), MatZeroRowsColumnsStencil()
6639: @*/
6640: PetscErrorCode MatZeroRowsColumnsLocalIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b)
6641: {
6643: PetscInt numRows;
6644: const PetscInt *rows;
6650: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6651: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6652: MatCheckPreallocated(mat,1);
6654: ISGetLocalSize(is,&numRows);
6655: ISGetIndices(is,&rows);
6656: MatZeroRowsColumnsLocal(mat,numRows,rows,diag,x,b);
6657: ISRestoreIndices(is,&rows);
6658: return(0);
6659: }
6661: /*@C
6662: MatGetSize - Returns the numbers of rows and columns in a matrix.
6664: Not Collective
6666: Input Parameter:
6667: . mat - the matrix
6669: Output Parameters:
6670: + m - the number of global rows
6671: - n - the number of global columns
6673: Note: both output parameters can be NULL on input.
6675: Level: beginner
6677: .seealso: MatGetLocalSize()
6678: @*/
6679: PetscErrorCode MatGetSize(Mat mat,PetscInt *m,PetscInt *n)
6680: {
6683: if (m) *m = mat->rmap->N;
6684: if (n) *n = mat->cmap->N;
6685: return(0);
6686: }
6688: /*@C
6689: MatGetLocalSize - Returns the number of local rows and local columns
6690: of a matrix, that is the local size of the left and right vectors as returned by MatCreateVecs().
6692: Not Collective
6694: Input Parameter:
6695: . mat - the matrix
6697: Output Parameters:
6698: + m - the number of local rows
6699: - n - the number of local columns
6701: Note: both output parameters can be NULL on input.
6703: Level: beginner
6705: .seealso: MatGetSize()
6706: @*/
6707: PetscErrorCode MatGetLocalSize(Mat mat,PetscInt *m,PetscInt *n)
6708: {
6713: if (m) *m = mat->rmap->n;
6714: if (n) *n = mat->cmap->n;
6715: return(0);
6716: }
6718: /*@C
6719: MatGetOwnershipRangeColumn - Returns the range of matrix columns associated with rows of a vector one multiplies by that owned by
6720: this processor. (The columns of the "diagonal block")
6722: Not Collective, unless matrix has not been allocated, then collective on Mat
6724: Input Parameter:
6725: . mat - the matrix
6727: Output Parameters:
6728: + m - the global index of the first local column
6729: - n - one more than the global index of the last local column
6731: Notes:
6732: both output parameters can be NULL on input.
6734: Level: developer
6736: .seealso: MatGetOwnershipRange(), MatGetOwnershipRanges(), MatGetOwnershipRangesColumn()
6738: @*/
6739: PetscErrorCode MatGetOwnershipRangeColumn(Mat mat,PetscInt *m,PetscInt *n)
6740: {
6746: MatCheckPreallocated(mat,1);
6747: if (m) *m = mat->cmap->rstart;
6748: if (n) *n = mat->cmap->rend;
6749: return(0);
6750: }
6752: /*@C
6753: MatGetOwnershipRange - Returns the range of matrix rows owned by
6754: this processor, assuming that the matrix is laid out with the first
6755: n1 rows on the first processor, the next n2 rows on the second, etc.
6756: For certain parallel layouts this range may not be well defined.
6758: Not Collective
6760: Input Parameter:
6761: . mat - the matrix
6763: Output Parameters:
6764: + m - the global index of the first local row
6765: - n - one more than the global index of the last local row
6767: Note: Both output parameters can be NULL on input.
6768: $ This function requires that the matrix be preallocated. If you have not preallocated, consider using
6769: $ PetscSplitOwnership(MPI_Comm comm, PetscInt *n, PetscInt *N)
6770: $ and then MPI_Scan() to calculate prefix sums of the local sizes.
6772: Level: beginner
6774: .seealso: MatGetOwnershipRanges(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn(), PetscSplitOwnership(), PetscSplitOwnershipBlock()
6776: @*/
6777: PetscErrorCode MatGetOwnershipRange(Mat mat,PetscInt *m,PetscInt *n)
6778: {
6784: MatCheckPreallocated(mat,1);
6785: if (m) *m = mat->rmap->rstart;
6786: if (n) *n = mat->rmap->rend;
6787: return(0);
6788: }
6790: /*@C
6791: MatGetOwnershipRanges - Returns the range of matrix rows owned by
6792: each process
6794: Not Collective, unless matrix has not been allocated, then collective on Mat
6796: Input Parameters:
6797: . mat - the matrix
6799: Output Parameters:
6800: . ranges - start of each processors portion plus one more than the total length at the end
6802: Level: beginner
6804: .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn()
6806: @*/
6807: PetscErrorCode MatGetOwnershipRanges(Mat mat,const PetscInt **ranges)
6808: {
6814: MatCheckPreallocated(mat,1);
6815: PetscLayoutGetRanges(mat->rmap,ranges);
6816: return(0);
6817: }
6819: /*@C
6820: MatGetOwnershipRangesColumn - Returns the range of matrix columns associated with rows of a vector one multiplies by that owned by
6821: this processor. (The columns of the "diagonal blocks" for each process)
6823: Not Collective, unless matrix has not been allocated, then collective on Mat
6825: Input Parameters:
6826: . mat - the matrix
6828: Output Parameters:
6829: . ranges - start of each processors portion plus one more then the total length at the end
6831: Level: beginner
6833: .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRanges()
6835: @*/
6836: PetscErrorCode MatGetOwnershipRangesColumn(Mat mat,const PetscInt **ranges)
6837: {
6843: MatCheckPreallocated(mat,1);
6844: PetscLayoutGetRanges(mat->cmap,ranges);
6845: return(0);
6846: }
6848: /*@C
6849: MatGetOwnershipIS - Get row and column ownership as index sets
6851: Not Collective
6853: Input Parameter:
6854: . A - matrix of type Elemental or ScaLAPACK
6856: Output Parameters:
6857: + rows - rows in which this process owns elements
6858: - cols - columns in which this process owns elements
6860: Level: intermediate
6862: .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatSetValues(), MATELEMENTAL
6863: @*/
6864: PetscErrorCode MatGetOwnershipIS(Mat A,IS *rows,IS *cols)
6865: {
6866: PetscErrorCode ierr,(*f)(Mat,IS*,IS*);
6869: MatCheckPreallocated(A,1);
6870: PetscObjectQueryFunction((PetscObject)A,"MatGetOwnershipIS_C",&f);
6871: if (f) {
6872: (*f)(A,rows,cols);
6873: } else { /* Create a standard row-based partition, each process is responsible for ALL columns in their row block */
6874: if (rows) {ISCreateStride(PETSC_COMM_SELF,A->rmap->n,A->rmap->rstart,1,rows);}
6875: if (cols) {ISCreateStride(PETSC_COMM_SELF,A->cmap->N,0,1,cols);}
6876: }
6877: return(0);
6878: }
6880: /*@C
6881: MatILUFactorSymbolic - Performs symbolic ILU factorization of a matrix.
6882: Uses levels of fill only, not drop tolerance. Use MatLUFactorNumeric()
6883: to complete the factorization.
6885: Collective on Mat
6887: Input Parameters:
6888: + mat - the matrix
6889: . row - row permutation
6890: . column - column permutation
6891: - info - structure containing
6892: $ levels - number of levels of fill.
6893: $ expected fill - as ratio of original fill.
6894: $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices
6895: missing diagonal entries)
6897: Output Parameters:
6898: . fact - new matrix that has been symbolically factored
6900: Notes:
6901: See Users-Manual: ch_mat for additional information about choosing the fill factor for better efficiency.
6903: Most users should employ the simplified KSP interface for linear solvers
6904: instead of working directly with matrix algebra routines such as this.
6905: See, e.g., KSPCreate().
6907: Level: developer
6909: .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor()
6910: MatGetOrdering(), MatFactorInfo
6912: Note: this uses the definition of level of fill as in Y. Saad, 2003
6914: Developer Note: fortran interface is not autogenerated as the f90
6915: interface definition cannot be generated correctly [due to MatFactorInfo]
6917: References:
6918: Y. Saad, Iterative methods for sparse linear systems Philadelphia: Society for Industrial and Applied Mathematics, 2003
6919: @*/
6920: PetscErrorCode MatILUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info)
6921: {
6931: if (info->levels < 0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Levels of fill negative %D",(PetscInt)info->levels);
6932: if (info->fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %g",(double)info->fill);
6933: if (!fact->ops->ilufactorsymbolic) {
6934: MatSolverType stype;
6935: MatFactorGetSolverType(fact,&stype);
6936: SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic ILU using solver type %s",((PetscObject)mat)->type_name,stype);
6937: }
6938: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6939: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6940: MatCheckPreallocated(mat,2);
6942: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_ILUFactorSymbolic,mat,row,col,0);}
6943: (fact->ops->ilufactorsymbolic)(fact,mat,row,col,info);
6944: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_ILUFactorSymbolic,mat,row,col,0);}
6945: return(0);
6946: }
6948: /*@C
6949: MatICCFactorSymbolic - Performs symbolic incomplete
6950: Cholesky factorization for a symmetric matrix. Use
6951: MatCholeskyFactorNumeric() to complete the factorization.
6953: Collective on Mat
6955: Input Parameters:
6956: + mat - the matrix
6957: . perm - row and column permutation
6958: - info - structure containing
6959: $ levels - number of levels of fill.
6960: $ expected fill - as ratio of original fill.
6962: Output Parameter:
6963: . fact - the factored matrix
6965: Notes:
6966: Most users should employ the KSP interface for linear solvers
6967: instead of working directly with matrix algebra routines such as this.
6968: See, e.g., KSPCreate().
6970: Level: developer
6972: .seealso: MatCholeskyFactorNumeric(), MatCholeskyFactor(), MatFactorInfo
6974: Note: this uses the definition of level of fill as in Y. Saad, 2003
6976: Developer Note: fortran interface is not autogenerated as the f90
6977: interface definition cannot be generated correctly [due to MatFactorInfo]
6979: References:
6980: Y. Saad, Iterative methods for sparse linear systems Philadelphia: Society for Industrial and Applied Mathematics, 2003
6981: @*/
6982: PetscErrorCode MatICCFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info)
6983: {
6992: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6993: if (info->levels < 0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Levels negative %D",(PetscInt) info->levels);
6994: if (info->fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %g",(double)info->fill);
6995: if (!(fact)->ops->iccfactorsymbolic) {
6996: MatSolverType stype;
6997: MatFactorGetSolverType(fact,&stype);
6998: SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic ICC using solver type %s",((PetscObject)mat)->type_name,stype);
6999: }
7000: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7001: MatCheckPreallocated(mat,2);
7003: if (!fact->trivialsymbolic) {PetscLogEventBegin(MAT_ICCFactorSymbolic,mat,perm,0,0);}
7004: (fact->ops->iccfactorsymbolic)(fact,mat,perm,info);
7005: if (!fact->trivialsymbolic) {PetscLogEventEnd(MAT_ICCFactorSymbolic,mat,perm,0,0);}
7006: return(0);
7007: }
7009: /*@C
7010: MatCreateSubMatrices - Extracts several submatrices from a matrix. If submat
7011: points to an array of valid matrices, they may be reused to store the new
7012: submatrices.
7014: Collective on Mat
7016: Input Parameters:
7017: + mat - the matrix
7018: . n - the number of submatrixes to be extracted (on this processor, may be zero)
7019: . irow, icol - index sets of rows and columns to extract
7020: - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
7022: Output Parameter:
7023: . submat - the array of submatrices
7025: Notes:
7026: MatCreateSubMatrices() can extract ONLY sequential submatrices
7027: (from both sequential and parallel matrices). Use MatCreateSubMatrix()
7028: to extract a parallel submatrix.
7030: Some matrix types place restrictions on the row and column
7031: indices, such as that they be sorted or that they be equal to each other.
7033: The index sets may not have duplicate entries.
7035: When extracting submatrices from a parallel matrix, each processor can
7036: form a different submatrix by setting the rows and columns of its
7037: individual index sets according to the local submatrix desired.
7039: When finished using the submatrices, the user should destroy
7040: them with MatDestroySubMatrices().
7042: MAT_REUSE_MATRIX can only be used when the nonzero structure of the
7043: original matrix has not changed from that last call to MatCreateSubMatrices().
7045: This routine creates the matrices in submat; you should NOT create them before
7046: calling it. It also allocates the array of matrix pointers submat.
7048: For BAIJ matrices the index sets must respect the block structure, that is if they
7049: request one row/column in a block, they must request all rows/columns that are in
7050: that block. For example, if the block size is 2 you cannot request just row 0 and
7051: column 0.
7053: Fortran Note:
7054: The Fortran interface is slightly different from that given below; it
7055: requires one to pass in as submat a Mat (integer) array of size at least n+1.
7057: Level: advanced
7059: .seealso: MatDestroySubMatrices(), MatCreateSubMatrix(), MatGetRow(), MatGetDiagonal(), MatReuse
7060: @*/
7061: PetscErrorCode MatCreateSubMatrices(Mat mat,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *submat[])
7062: {
7064: PetscInt i;
7065: PetscBool eq;
7070: if (n) {
7075: }
7077: if (n && scall == MAT_REUSE_MATRIX) {
7080: }
7081: if (!mat->ops->createsubmatrices) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7082: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7083: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7084: MatCheckPreallocated(mat,1);
7086: PetscLogEventBegin(MAT_CreateSubMats,mat,0,0,0);
7087: (*mat->ops->createsubmatrices)(mat,n,irow,icol,scall,submat);
7088: PetscLogEventEnd(MAT_CreateSubMats,mat,0,0,0);
7089: for (i=0; i<n; i++) {
7090: (*submat)[i]->factortype = MAT_FACTOR_NONE; /* in case in place factorization was previously done on submatrix */
7091: ISEqualUnsorted(irow[i],icol[i],&eq);
7092: if (eq) {
7093: MatPropagateSymmetryOptions(mat,(*submat)[i]);
7094: }
7095: }
7096: return(0);
7097: }
7099: /*@C
7100: MatCreateSubMatricesMPI - Extracts MPI submatrices across a sub communicator of mat (by pairs of IS that may live on subcomms).
7102: Collective on Mat
7104: Input Parameters:
7105: + mat - the matrix
7106: . n - the number of submatrixes to be extracted
7107: . irow, icol - index sets of rows and columns to extract
7108: - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
7110: Output Parameter:
7111: . submat - the array of submatrices
7113: Level: advanced
7115: .seealso: MatCreateSubMatrices(), MatCreateSubMatrix(), MatGetRow(), MatGetDiagonal(), MatReuse
7116: @*/
7117: PetscErrorCode MatCreateSubMatricesMPI(Mat mat,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *submat[])
7118: {
7120: PetscInt i;
7121: PetscBool eq;
7126: if (n) {
7131: }
7133: if (n && scall == MAT_REUSE_MATRIX) {
7136: }
7137: if (!mat->ops->createsubmatricesmpi) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7138: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7139: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7140: MatCheckPreallocated(mat,1);
7142: PetscLogEventBegin(MAT_CreateSubMats,mat,0,0,0);
7143: (*mat->ops->createsubmatricesmpi)(mat,n,irow,icol,scall,submat);
7144: PetscLogEventEnd(MAT_CreateSubMats,mat,0,0,0);
7145: for (i=0; i<n; i++) {
7146: ISEqualUnsorted(irow[i],icol[i],&eq);
7147: if (eq) {
7148: MatPropagateSymmetryOptions(mat,(*submat)[i]);
7149: }
7150: }
7151: return(0);
7152: }
7154: /*@C
7155: MatDestroyMatrices - Destroys an array of matrices.
7157: Collective on Mat
7159: Input Parameters:
7160: + n - the number of local matrices
7161: - mat - the matrices (note that this is a pointer to the array of matrices)
7163: Level: advanced
7165: Notes:
7166: Frees not only the matrices, but also the array that contains the matrices
7167: In Fortran will not free the array.
7169: .seealso: MatCreateSubMatrices() MatDestroySubMatrices()
7170: @*/
7171: PetscErrorCode MatDestroyMatrices(PetscInt n,Mat *mat[])
7172: {
7174: PetscInt i;
7177: if (!*mat) return(0);
7178: if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Trying to destroy negative number of matrices %D",n);
7181: for (i=0; i<n; i++) {
7182: MatDestroy(&(*mat)[i]);
7183: }
7185: /* memory is allocated even if n = 0 */
7186: PetscFree(*mat);
7187: return(0);
7188: }
7190: /*@C
7191: MatDestroySubMatrices - Destroys a set of matrices obtained with MatCreateSubMatrices().
7193: Collective on Mat
7195: Input Parameters:
7196: + n - the number of local matrices
7197: - mat - the matrices (note that this is a pointer to the array of matrices, just to match the calling
7198: sequence of MatCreateSubMatrices())
7200: Level: advanced
7202: Notes:
7203: Frees not only the matrices, but also the array that contains the matrices
7204: In Fortran will not free the array.
7206: .seealso: MatCreateSubMatrices()
7207: @*/
7208: PetscErrorCode MatDestroySubMatrices(PetscInt n,Mat *mat[])
7209: {
7211: Mat mat0;
7214: if (!*mat) return(0);
7215: /* mat[] is an array of length n+1, see MatCreateSubMatrices_xxx() */
7216: if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Trying to destroy negative number of matrices %D",n);
7219: mat0 = (*mat)[0];
7220: if (mat0 && mat0->ops->destroysubmatrices) {
7221: (mat0->ops->destroysubmatrices)(n,mat);
7222: } else {
7223: MatDestroyMatrices(n,mat);
7224: }
7225: return(0);
7226: }
7228: /*@C
7229: MatGetSeqNonzeroStructure - Extracts the sequential nonzero structure from a matrix.
7231: Collective on Mat
7233: Input Parameters:
7234: . mat - the matrix
7236: Output Parameter:
7237: . matstruct - the sequential matrix with the nonzero structure of mat
7239: Level: intermediate
7241: .seealso: MatDestroySeqNonzeroStructure(), MatCreateSubMatrices(), MatDestroyMatrices()
7242: @*/
7243: PetscErrorCode MatGetSeqNonzeroStructure(Mat mat,Mat *matstruct)
7244: {
7252: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7253: MatCheckPreallocated(mat,1);
7255: if (!mat->ops->getseqnonzerostructure) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not for matrix type %s\n",((PetscObject)mat)->type_name);
7256: PetscLogEventBegin(MAT_GetSeqNonzeroStructure,mat,0,0,0);
7257: (*mat->ops->getseqnonzerostructure)(mat,matstruct);
7258: PetscLogEventEnd(MAT_GetSeqNonzeroStructure,mat,0,0,0);
7259: return(0);
7260: }
7262: /*@C
7263: MatDestroySeqNonzeroStructure - Destroys matrix obtained with MatGetSeqNonzeroStructure().
7265: Collective on Mat
7267: Input Parameters:
7268: . mat - the matrix (note that this is a pointer to the array of matrices, just to match the calling
7269: sequence of MatGetSequentialNonzeroStructure())
7271: Level: advanced
7273: Notes:
7274: Frees not only the matrices, but also the array that contains the matrices
7276: .seealso: MatGetSeqNonzeroStructure()
7277: @*/
7278: PetscErrorCode MatDestroySeqNonzeroStructure(Mat *mat)
7279: {
7284: MatDestroy(mat);
7285: return(0);
7286: }
7288: /*@
7289: MatIncreaseOverlap - Given a set of submatrices indicated by index sets,
7290: replaces the index sets by larger ones that represent submatrices with
7291: additional overlap.
7293: Collective on Mat
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:
7302: . -mat_increase_overlap_scalable - use a scalable algorithm to compute the overlap (supported by MPIAIJ matrix)
7304: Level: developer
7306: .seealso: MatCreateSubMatrices()
7307: @*/
7308: PetscErrorCode MatIncreaseOverlap(Mat mat,PetscInt n,IS is[],PetscInt ov)
7309: {
7315: if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Must have one or more domains, you have %D",n);
7316: if (n) {
7319: }
7320: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7321: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7322: MatCheckPreallocated(mat,1);
7324: if (!ov) return(0);
7325: if (!mat->ops->increaseoverlap) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7326: PetscLogEventBegin(MAT_IncreaseOverlap,mat,0,0,0);
7327: (*mat->ops->increaseoverlap)(mat,n,is,ov);
7328: PetscLogEventEnd(MAT_IncreaseOverlap,mat,0,0,0);
7329: return(0);
7330: }
7332: PetscErrorCode MatIncreaseOverlapSplit_Single(Mat,IS*,PetscInt);
7334: /*@
7335: MatIncreaseOverlapSplit - Given a set of submatrices indicated by index sets across
7336: a sub communicator, replaces the index sets by larger ones that represent submatrices with
7337: additional overlap.
7339: Collective on Mat
7341: Input Parameters:
7342: + mat - the matrix
7343: . n - the number of index sets
7344: . is - the array of index sets (these index sets will changed during the call)
7345: - ov - the additional overlap requested
7347: Options Database:
7348: . -mat_increase_overlap_scalable - use a scalable algorithm to compute the overlap (supported by MPIAIJ matrix)
7350: Level: developer
7352: .seealso: MatCreateSubMatrices()
7353: @*/
7354: PetscErrorCode MatIncreaseOverlapSplit(Mat mat,PetscInt n,IS is[],PetscInt ov)
7355: {
7356: PetscInt i;
7362: if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Must have one or more domains, you have %D",n);
7363: if (n) {
7366: }
7367: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7368: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7369: MatCheckPreallocated(mat,1);
7370: if (!ov) return(0);
7371: PetscLogEventBegin(MAT_IncreaseOverlap,mat,0,0,0);
7372: for (i=0; i<n; i++) {
7373: MatIncreaseOverlapSplit_Single(mat,&is[i],ov);
7374: }
7375: PetscLogEventEnd(MAT_IncreaseOverlap,mat,0,0,0);
7376: return(0);
7377: }
7379: /*@
7380: MatGetBlockSize - Returns the matrix block size.
7382: Not Collective
7384: Input Parameter:
7385: . mat - the matrix
7387: Output Parameter:
7388: . bs - block size
7390: Notes:
7391: Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ. These formats ALWAYS have square block storage in the matrix.
7393: If the block size has not been set yet this routine returns 1.
7395: Level: intermediate
7397: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSizes()
7398: @*/
7399: PetscErrorCode MatGetBlockSize(Mat mat,PetscInt *bs)
7400: {
7404: *bs = PetscAbs(mat->rmap->bs);
7405: return(0);
7406: }
7408: /*@
7409: MatGetBlockSizes - Returns the matrix block row and column sizes.
7411: Not Collective
7413: Input Parameter:
7414: . mat - the matrix
7416: Output Parameters:
7417: + rbs - row block size
7418: - cbs - column block size
7420: Notes:
7421: Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ. These formats ALWAYS have square block storage in the matrix.
7422: If you pass a different block size for the columns than the rows, the row block size determines the square block storage.
7424: If a block size has not been set yet this routine returns 1.
7426: Level: intermediate
7428: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSize(), MatSetBlockSizes()
7429: @*/
7430: PetscErrorCode MatGetBlockSizes(Mat mat,PetscInt *rbs, PetscInt *cbs)
7431: {
7436: if (rbs) *rbs = PetscAbs(mat->rmap->bs);
7437: if (cbs) *cbs = PetscAbs(mat->cmap->bs);
7438: return(0);
7439: }
7441: /*@
7442: MatSetBlockSize - Sets the matrix block size.
7444: Logically Collective on Mat
7446: Input Parameters:
7447: + mat - the matrix
7448: - bs - block size
7450: Notes:
7451: Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ. These formats ALWAYS have square block storage in the matrix.
7452: This must be called before MatSetUp() or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later.
7454: For MATMPIAIJ and MATSEQAIJ matrix formats, this function can be called at a later stage, provided that the specified block size
7455: is compatible with the matrix local sizes.
7457: Level: intermediate
7459: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSizes(), MatGetBlockSizes()
7460: @*/
7461: PetscErrorCode MatSetBlockSize(Mat mat,PetscInt bs)
7462: {
7468: MatSetBlockSizes(mat,bs,bs);
7469: return(0);
7470: }
7472: /*@
7473: MatSetVariableBlockSizes - Sets a diagonal blocks of the matrix that need not be of the same size
7475: Logically Collective on Mat
7477: Input Parameters:
7478: + mat - the matrix
7479: . nblocks - the number of blocks on this process
7480: - bsizes - the block sizes
7482: Notes:
7483: Currently used by PCVPBJACOBI for SeqAIJ matrices
7485: Level: intermediate
7487: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSizes(), MatGetBlockSizes(), MatGetVariableBlockSizes()
7488: @*/
7489: PetscErrorCode MatSetVariableBlockSizes(Mat mat,PetscInt nblocks,PetscInt *bsizes)
7490: {
7492: PetscInt i,ncnt = 0, nlocal;
7496: if (nblocks < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Number of local blocks must be great than or equal to zero");
7497: MatGetLocalSize(mat,&nlocal,NULL);
7498: for (i=0; i<nblocks; i++) ncnt += bsizes[i];
7499: if (ncnt != nlocal) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Sum of local block sizes %D does not equal local size of matrix %D",ncnt,nlocal);
7500: PetscFree(mat->bsizes);
7501: mat->nblocks = nblocks;
7502: PetscMalloc1(nblocks,&mat->bsizes);
7503: PetscArraycpy(mat->bsizes,bsizes,nblocks);
7504: return(0);
7505: }
7507: /*@C
7508: MatGetVariableBlockSizes - Gets a diagonal blocks of the matrix that need not be of the same size
7510: Logically Collective on Mat
7512: Input Parameter:
7513: . mat - the matrix
7515: Output Parameters:
7516: + nblocks - the number of blocks on this process
7517: - bsizes - the block sizes
7519: Notes: Currently not supported from Fortran
7521: Level: intermediate
7523: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSizes(), MatGetBlockSizes(), MatSetVariableBlockSizes()
7524: @*/
7525: PetscErrorCode MatGetVariableBlockSizes(Mat mat,PetscInt *nblocks,const PetscInt **bsizes)
7526: {
7529: *nblocks = mat->nblocks;
7530: *bsizes = mat->bsizes;
7531: return(0);
7532: }
7534: /*@
7535: MatSetBlockSizes - Sets the matrix block row and column sizes.
7537: Logically Collective on Mat
7539: Input Parameters:
7540: + mat - the matrix
7541: . rbs - row block size
7542: - cbs - column block size
7544: Notes:
7545: Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ. These formats ALWAYS have square block storage in the matrix.
7546: If you pass a different block size for the columns than the rows, the row block size determines the square block storage.
7547: This must be called before MatSetUp() or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later.
7549: For MATMPIAIJ and MATSEQAIJ matrix formats, this function can be called at a later stage, provided that the specified block sizes
7550: are compatible with the matrix local sizes.
7552: The row and column block size determine the blocksize of the "row" and "column" vectors returned by MatCreateVecs().
7554: Level: intermediate
7556: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSize(), MatGetBlockSizes()
7557: @*/
7558: PetscErrorCode MatSetBlockSizes(Mat mat,PetscInt rbs,PetscInt cbs)
7559: {
7566: if (mat->ops->setblocksizes) {
7567: (*mat->ops->setblocksizes)(mat,rbs,cbs);
7568: }
7569: if (mat->rmap->refcnt) {
7570: ISLocalToGlobalMapping l2g = NULL;
7571: PetscLayout nmap = NULL;
7573: PetscLayoutDuplicate(mat->rmap,&nmap);
7574: if (mat->rmap->mapping) {
7575: ISLocalToGlobalMappingDuplicate(mat->rmap->mapping,&l2g);
7576: }
7577: PetscLayoutDestroy(&mat->rmap);
7578: mat->rmap = nmap;
7579: mat->rmap->mapping = l2g;
7580: }
7581: if (mat->cmap->refcnt) {
7582: ISLocalToGlobalMapping l2g = NULL;
7583: PetscLayout nmap = NULL;
7585: PetscLayoutDuplicate(mat->cmap,&nmap);
7586: if (mat->cmap->mapping) {
7587: ISLocalToGlobalMappingDuplicate(mat->cmap->mapping,&l2g);
7588: }
7589: PetscLayoutDestroy(&mat->cmap);
7590: mat->cmap = nmap;
7591: mat->cmap->mapping = l2g;
7592: }
7593: PetscLayoutSetBlockSize(mat->rmap,rbs);
7594: PetscLayoutSetBlockSize(mat->cmap,cbs);
7595: return(0);
7596: }
7598: /*@
7599: MatSetBlockSizesFromMats - Sets the matrix block row and column sizes to match a pair of matrices
7601: Logically Collective on Mat
7603: Input Parameters:
7604: + mat - the matrix
7605: . fromRow - matrix from which to copy row block size
7606: - fromCol - matrix from which to copy column block size (can be same as fromRow)
7608: Level: developer
7610: .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize(), MatSetBlockSizes()
7611: @*/
7612: PetscErrorCode MatSetBlockSizesFromMats(Mat mat,Mat fromRow,Mat fromCol)
7613: {
7620: if (fromRow->rmap->bs > 0) {PetscLayoutSetBlockSize(mat->rmap,fromRow->rmap->bs);}
7621: if (fromCol->cmap->bs > 0) {PetscLayoutSetBlockSize(mat->cmap,fromCol->cmap->bs);}
7622: return(0);
7623: }
7625: /*@
7626: MatResidual - Default routine to calculate the residual.
7628: Collective on Mat
7630: Input Parameters:
7631: + mat - the matrix
7632: . b - the right-hand-side
7633: - x - the approximate solution
7635: Output Parameter:
7636: . r - location to store the residual
7638: Level: developer
7640: .seealso: PCMGSetResidual()
7641: @*/
7642: PetscErrorCode MatResidual(Mat mat,Vec b,Vec x,Vec r)
7643: {
7652: MatCheckPreallocated(mat,1);
7653: PetscLogEventBegin(MAT_Residual,mat,0,0,0);
7654: if (!mat->ops->residual) {
7655: MatMult(mat,x,r);
7656: VecAYPX(r,-1.0,b);
7657: } else {
7658: (*mat->ops->residual)(mat,b,x,r);
7659: }
7660: PetscLogEventEnd(MAT_Residual,mat,0,0,0);
7661: return(0);
7662: }
7664: /*@C
7665: MatGetRowIJ - Returns the compressed row storage i and j indices for sequential matrices.
7667: Collective on Mat
7669: Input Parameters:
7670: + mat - the matrix
7671: . shift - 0 or 1 indicating we want the indices starting at 0 or 1
7672: . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be symmetrized
7673: - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
7674: inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
7675: always used.
7677: Output Parameters:
7678: + n - number of rows in the (possibly compressed) matrix
7679: . ia - the row pointers; that is ia[0] = 0, ia[row] = ia[row-1] + number of elements in that row of the matrix
7680: . ja - the column indices
7681: - done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers
7682: are responsible for handling the case when done == PETSC_FALSE and ia and ja are not set
7684: Level: developer
7686: Notes:
7687: You CANNOT change any of the ia[] or ja[] values.
7689: Use MatRestoreRowIJ() when you are finished accessing the ia[] and ja[] values.
7691: Fortran Notes:
7692: In Fortran use
7693: $
7694: $ PetscInt ia(1), ja(1)
7695: $ PetscOffset iia, jja
7696: $ call MatGetRowIJ(mat,shift,symmetric,inodecompressed,n,ia,iia,ja,jja,done,ierr)
7697: $ ! Access the ith and jth entries via ia(iia + i) and ja(jja + j)
7699: or
7700: $
7701: $ PetscInt, pointer :: ia(:),ja(:)
7702: $ call MatGetRowIJF90(mat,shift,symmetric,inodecompressed,n,ia,ja,done,ierr)
7703: $ ! Access the ith and jth entries via ia(i) and ja(j)
7705: .seealso: MatGetColumnIJ(), MatRestoreRowIJ(), MatSeqAIJGetArray()
7706: @*/
7707: PetscErrorCode MatGetRowIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
7708: {
7718: MatCheckPreallocated(mat,1);
7719: if (!mat->ops->getrowij) *done = PETSC_FALSE;
7720: else {
7721: *done = PETSC_TRUE;
7722: PetscLogEventBegin(MAT_GetRowIJ,mat,0,0,0);
7723: (*mat->ops->getrowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);
7724: PetscLogEventEnd(MAT_GetRowIJ,mat,0,0,0);
7725: }
7726: return(0);
7727: }
7729: /*@C
7730: MatGetColumnIJ - Returns the compressed column storage i and j indices for sequential matrices.
7732: Collective on Mat
7734: Input Parameters:
7735: + mat - the matrix
7736: . shift - 1 or zero indicating we want the indices starting at 0 or 1
7737: . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
7738: symmetrized
7739: . inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
7740: inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
7741: always used.
7742: . n - number of columns in the (possibly compressed) matrix
7743: . ia - the column pointers; that is ia[0] = 0, ia[col] = i[col-1] + number of elements in that col of the matrix
7744: - ja - the row indices
7746: Output Parameters:
7747: . done - PETSC_TRUE or PETSC_FALSE, indicating whether the values have been returned
7749: Level: developer
7751: .seealso: MatGetRowIJ(), MatRestoreColumnIJ()
7752: @*/
7753: PetscErrorCode MatGetColumnIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
7754: {
7764: MatCheckPreallocated(mat,1);
7765: if (!mat->ops->getcolumnij) *done = PETSC_FALSE;
7766: else {
7767: *done = PETSC_TRUE;
7768: (*mat->ops->getcolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);
7769: }
7770: return(0);
7771: }
7773: /*@C
7774: MatRestoreRowIJ - Call after you are completed with the ia,ja indices obtained with
7775: MatGetRowIJ().
7777: Collective on Mat
7779: Input Parameters:
7780: + mat - the matrix
7781: . shift - 1 or zero indicating we want the indices starting at 0 or 1
7782: . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
7783: symmetrized
7784: . inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
7785: inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
7786: always used.
7787: . n - size of (possibly compressed) matrix
7788: . ia - the row pointers
7789: - ja - the column indices
7791: Output Parameters:
7792: . done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned
7794: Note:
7795: This routine zeros out n, ia, and ja. This is to prevent accidental
7796: us of the array after it has been restored. If you pass NULL, it will
7797: not zero the pointers. Use of ia or ja after MatRestoreRowIJ() is invalid.
7799: Level: developer
7801: .seealso: MatGetRowIJ(), MatRestoreColumnIJ()
7802: @*/
7803: PetscErrorCode MatRestoreRowIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
7804: {
7813: MatCheckPreallocated(mat,1);
7815: if (!mat->ops->restorerowij) *done = PETSC_FALSE;
7816: else {
7817: *done = PETSC_TRUE;
7818: (*mat->ops->restorerowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);
7819: if (n) *n = 0;
7820: if (ia) *ia = NULL;
7821: if (ja) *ja = NULL;
7822: }
7823: return(0);
7824: }
7826: /*@C
7827: MatRestoreColumnIJ - Call after you are completed with the ia,ja indices obtained with
7828: MatGetColumnIJ().
7830: Collective on Mat
7832: Input Parameters:
7833: + mat - the matrix
7834: . shift - 1 or zero indicating we want the indices starting at 0 or 1
7835: . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
7836: symmetrized
7837: - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
7838: inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
7839: always used.
7841: Output Parameters:
7842: + n - size of (possibly compressed) matrix
7843: . ia - the column pointers
7844: . ja - the row indices
7845: - done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned
7847: Level: developer
7849: .seealso: MatGetColumnIJ(), MatRestoreRowIJ()
7850: @*/
7851: PetscErrorCode MatRestoreColumnIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
7852: {
7861: MatCheckPreallocated(mat,1);
7863: if (!mat->ops->restorecolumnij) *done = PETSC_FALSE;
7864: else {
7865: *done = PETSC_TRUE;
7866: (*mat->ops->restorecolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);
7867: if (n) *n = 0;
7868: if (ia) *ia = NULL;
7869: if (ja) *ja = NULL;
7870: }
7871: return(0);
7872: }
7874: /*@C
7875: MatColoringPatch -Used inside matrix coloring routines that
7876: use MatGetRowIJ() and/or MatGetColumnIJ().
7878: Collective on Mat
7880: Input Parameters:
7881: + mat - the matrix
7882: . ncolors - max color value
7883: . n - number of entries in colorarray
7884: - colorarray - array indicating color for each column
7886: Output Parameters:
7887: . iscoloring - coloring generated using colorarray information
7889: Level: developer
7891: .seealso: MatGetRowIJ(), MatGetColumnIJ()
7893: @*/
7894: PetscErrorCode MatColoringPatch(Mat mat,PetscInt ncolors,PetscInt n,ISColoringValue colorarray[],ISColoring *iscoloring)
7895: {
7903: MatCheckPreallocated(mat,1);
7905: if (!mat->ops->coloringpatch) {
7906: ISColoringCreate(PetscObjectComm((PetscObject)mat),ncolors,n,colorarray,PETSC_OWN_POINTER,iscoloring);
7907: } else {
7908: (*mat->ops->coloringpatch)(mat,ncolors,n,colorarray,iscoloring);
7909: }
7910: return(0);
7911: }
7913: /*@
7914: MatSetUnfactored - Resets a factored matrix to be treated as unfactored.
7916: Logically Collective on Mat
7918: Input Parameter:
7919: . mat - the factored matrix to be reset
7921: Notes:
7922: This routine should be used only with factored matrices formed by in-place
7923: factorization via ILU(0) (or by in-place LU factorization for the MATSEQDENSE
7924: format). This option can save memory, for example, when solving nonlinear
7925: systems with a matrix-free Newton-Krylov method and a matrix-based, in-place
7926: ILU(0) preconditioner.
7928: Note that one can specify in-place ILU(0) factorization by calling
7929: .vb
7930: PCType(pc,PCILU);
7931: PCFactorSeUseInPlace(pc);
7932: .ve
7933: or by using the options -pc_type ilu -pc_factor_in_place
7935: In-place factorization ILU(0) can also be used as a local
7936: solver for the blocks within the block Jacobi or additive Schwarz
7937: methods (runtime option: -sub_pc_factor_in_place). See Users-Manual: ch_pc
7938: for details on setting local solver options.
7940: Most users should employ the simplified KSP interface for linear solvers
7941: instead of working directly with matrix algebra routines such as this.
7942: See, e.g., KSPCreate().
7944: Level: developer
7946: .seealso: PCFactorSetUseInPlace(), PCFactorGetUseInPlace()
7948: @*/
7949: PetscErrorCode MatSetUnfactored(Mat mat)
7950: {
7956: MatCheckPreallocated(mat,1);
7957: mat->factortype = MAT_FACTOR_NONE;
7958: if (!mat->ops->setunfactored) return(0);
7959: (*mat->ops->setunfactored)(mat);
7960: return(0);
7961: }
7963: /*MC
7964: MatDenseGetArrayF90 - Accesses a matrix array from Fortran90.
7966: Synopsis:
7967: MatDenseGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:,:)},integer ierr)
7969: Not collective
7971: Input Parameter:
7972: . x - matrix
7974: Output Parameters:
7975: + xx_v - the Fortran90 pointer to the array
7976: - ierr - error code
7978: Example of Usage:
7979: .vb
7980: PetscScalar, pointer xx_v(:,:)
7981: ....
7982: call MatDenseGetArrayF90(x,xx_v,ierr)
7983: a = xx_v(3)
7984: call MatDenseRestoreArrayF90(x,xx_v,ierr)
7985: .ve
7987: Level: advanced
7989: .seealso: MatDenseRestoreArrayF90(), MatDenseGetArray(), MatDenseRestoreArray(), MatSeqAIJGetArrayF90()
7991: M*/
7993: /*MC
7994: MatDenseRestoreArrayF90 - Restores a matrix array that has been
7995: accessed with MatDenseGetArrayF90().
7997: Synopsis:
7998: MatDenseRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:,:)},integer ierr)
8000: Not collective
8002: Input Parameters:
8003: + x - matrix
8004: - xx_v - the Fortran90 pointer to the array
8006: Output Parameter:
8007: . ierr - error code
8009: Example of Usage:
8010: .vb
8011: PetscScalar, pointer xx_v(:,:)
8012: ....
8013: call MatDenseGetArrayF90(x,xx_v,ierr)
8014: a = xx_v(3)
8015: call MatDenseRestoreArrayF90(x,xx_v,ierr)
8016: .ve
8018: Level: advanced
8020: .seealso: MatDenseGetArrayF90(), MatDenseGetArray(), MatDenseRestoreArray(), MatSeqAIJRestoreArrayF90()
8022: M*/
8024: /*MC
8025: MatSeqAIJGetArrayF90 - Accesses a matrix array from Fortran90.
8027: Synopsis:
8028: MatSeqAIJGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
8030: Not collective
8032: Input Parameter:
8033: . x - matrix
8035: Output Parameters:
8036: + xx_v - the Fortran90 pointer to the array
8037: - ierr - error code
8039: Example of Usage:
8040: .vb
8041: PetscScalar, pointer xx_v(:)
8042: ....
8043: call MatSeqAIJGetArrayF90(x,xx_v,ierr)
8044: a = xx_v(3)
8045: call MatSeqAIJRestoreArrayF90(x,xx_v,ierr)
8046: .ve
8048: Level: advanced
8050: .seealso: MatSeqAIJRestoreArrayF90(), MatSeqAIJGetArray(), MatSeqAIJRestoreArray(), MatDenseGetArrayF90()
8052: M*/
8054: /*MC
8055: MatSeqAIJRestoreArrayF90 - Restores a matrix array that has been
8056: accessed with MatSeqAIJGetArrayF90().
8058: Synopsis:
8059: MatSeqAIJRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
8061: Not collective
8063: Input Parameters:
8064: + x - matrix
8065: - xx_v - the Fortran90 pointer to the array
8067: Output Parameter:
8068: . ierr - error code
8070: Example of Usage:
8071: .vb
8072: PetscScalar, pointer xx_v(:)
8073: ....
8074: call MatSeqAIJGetArrayF90(x,xx_v,ierr)
8075: a = xx_v(3)
8076: call MatSeqAIJRestoreArrayF90(x,xx_v,ierr)
8077: .ve
8079: Level: advanced
8081: .seealso: MatSeqAIJGetArrayF90(), MatSeqAIJGetArray(), MatSeqAIJRestoreArray(), MatDenseRestoreArrayF90()
8083: M*/
8085: /*@
8086: MatCreateSubMatrix - Gets a single submatrix on the same number of processors
8087: as the original matrix.
8089: Collective on Mat
8091: Input Parameters:
8092: + mat - the original matrix
8093: . isrow - parallel IS containing the rows this processor should obtain
8094: . 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.
8095: - cll - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
8097: Output Parameter:
8098: . newmat - the new submatrix, of the same type as the old
8100: Level: advanced
8102: Notes:
8103: The submatrix will be able to be multiplied with vectors using the same layout as iscol.
8105: Some matrix types place restrictions on the row and column indices, such
8106: as that they be sorted or that they be equal to each other.
8108: The index sets may not have duplicate entries.
8110: The first time this is called you should use a cll of MAT_INITIAL_MATRIX,
8111: the MatCreateSubMatrix() routine will create the newmat for you. Any additional calls
8112: to this routine with a mat of the same nonzero structure and with a call of MAT_REUSE_MATRIX
8113: will reuse the matrix generated the first time. You should call MatDestroy() on newmat when
8114: you are finished using it.
8116: The communicator of the newly obtained matrix is ALWAYS the same as the communicator of
8117: the input matrix.
8119: If iscol is NULL then all columns are obtained (not supported in Fortran).
8121: Example usage:
8122: Consider the following 8x8 matrix with 34 non-zero values, that is
8123: assembled across 3 processors. Let's assume that proc0 owns 3 rows,
8124: proc1 owns 3 rows, proc2 owns 2 rows. This division can be shown
8125: as follows:
8127: .vb
8128: 1 2 0 | 0 3 0 | 0 4
8129: Proc0 0 5 6 | 7 0 0 | 8 0
8130: 9 0 10 | 11 0 0 | 12 0
8131: -------------------------------------
8132: 13 0 14 | 15 16 17 | 0 0
8133: Proc1 0 18 0 | 19 20 21 | 0 0
8134: 0 0 0 | 22 23 0 | 24 0
8135: -------------------------------------
8136: Proc2 25 26 27 | 0 0 28 | 29 0
8137: 30 0 0 | 31 32 33 | 0 34
8138: .ve
8140: Suppose isrow = [0 1 | 4 | 6 7] and iscol = [1 2 | 3 4 5 | 6]. The resulting submatrix is
8142: .vb
8143: 2 0 | 0 3 0 | 0
8144: Proc0 5 6 | 7 0 0 | 8
8145: -------------------------------
8146: Proc1 18 0 | 19 20 21 | 0
8147: -------------------------------
8148: Proc2 26 27 | 0 0 28 | 29
8149: 0 0 | 31 32 33 | 0
8150: .ve
8152: .seealso: MatCreateSubMatrices(), MatCreateSubMatricesMPI(), MatCreateSubMatrixVirtual(), MatSubMatrixVirtualUpdate()
8153: @*/
8154: PetscErrorCode MatCreateSubMatrix(Mat mat,IS isrow,IS iscol,MatReuse cll,Mat *newmat)
8155: {
8157: PetscMPIInt size;
8158: Mat *local;
8159: IS iscoltmp;
8160: PetscBool flg;
8169: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
8170: if (cll == MAT_IGNORE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Cannot use MAT_IGNORE_MATRIX");
8172: MatCheckPreallocated(mat,1);
8173: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
8175: if (!iscol || isrow == iscol) {
8176: PetscBool stride;
8177: PetscMPIInt grabentirematrix = 0,grab;
8178: PetscObjectTypeCompare((PetscObject)isrow,ISSTRIDE,&stride);
8179: if (stride) {
8180: PetscInt first,step,n,rstart,rend;
8181: ISStrideGetInfo(isrow,&first,&step);
8182: if (step == 1) {
8183: MatGetOwnershipRange(mat,&rstart,&rend);
8184: if (rstart == first) {
8185: ISGetLocalSize(isrow,&n);
8186: if (n == rend-rstart) {
8187: grabentirematrix = 1;
8188: }
8189: }
8190: }
8191: }
8192: MPIU_Allreduce(&grabentirematrix,&grab,1,MPI_INT,MPI_MIN,PetscObjectComm((PetscObject)mat));
8193: if (grab) {
8194: PetscInfo(mat,"Getting entire matrix as submatrix\n");
8195: if (cll == MAT_INITIAL_MATRIX) {
8196: *newmat = mat;
8197: PetscObjectReference((PetscObject)mat);
8198: }
8199: return(0);
8200: }
8201: }
8203: if (!iscol) {
8204: ISCreateStride(PetscObjectComm((PetscObject)mat),mat->cmap->n,mat->cmap->rstart,1,&iscoltmp);
8205: } else {
8206: iscoltmp = iscol;
8207: }
8209: /* if original matrix is on just one processor then use submatrix generated */
8210: if (mat->ops->createsubmatrices && !mat->ops->createsubmatrix && size == 1 && cll == MAT_REUSE_MATRIX) {
8211: MatCreateSubMatrices(mat,1,&isrow,&iscoltmp,MAT_REUSE_MATRIX,&newmat);
8212: goto setproperties;
8213: } else if (mat->ops->createsubmatrices && !mat->ops->createsubmatrix && size == 1) {
8214: MatCreateSubMatrices(mat,1,&isrow,&iscoltmp,MAT_INITIAL_MATRIX,&local);
8215: *newmat = *local;
8216: PetscFree(local);
8217: goto setproperties;
8218: } else if (!mat->ops->createsubmatrix) {
8219: /* Create a new matrix type that implements the operation using the full matrix */
8220: PetscLogEventBegin(MAT_CreateSubMat,mat,0,0,0);
8221: switch (cll) {
8222: case MAT_INITIAL_MATRIX:
8223: MatCreateSubMatrixVirtual(mat,isrow,iscoltmp,newmat);
8224: break;
8225: case MAT_REUSE_MATRIX:
8226: MatSubMatrixVirtualUpdate(*newmat,mat,isrow,iscoltmp);
8227: break;
8228: default: SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Invalid MatReuse, must be either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX");
8229: }
8230: PetscLogEventEnd(MAT_CreateSubMat,mat,0,0,0);
8231: goto setproperties;
8232: }
8234: if (!mat->ops->createsubmatrix) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
8235: PetscLogEventBegin(MAT_CreateSubMat,mat,0,0,0);
8236: (*mat->ops->createsubmatrix)(mat,isrow,iscoltmp,cll,newmat);
8237: PetscLogEventEnd(MAT_CreateSubMat,mat,0,0,0);
8239: setproperties:
8240: ISEqualUnsorted(isrow,iscoltmp,&flg);
8241: if (flg) {
8242: MatPropagateSymmetryOptions(mat,*newmat);
8243: }
8244: if (!iscol) {ISDestroy(&iscoltmp);}
8245: if (*newmat && cll == MAT_INITIAL_MATRIX) {PetscObjectStateIncrease((PetscObject)*newmat);}
8246: return(0);
8247: }
8249: /*@
8250: MatPropagateSymmetryOptions - Propagates symmetry options set on a matrix to another matrix
8252: Not Collective
8254: Input Parameters:
8255: + A - the matrix we wish to propagate options from
8256: - B - the matrix we wish to propagate options to
8258: Level: beginner
8260: Notes: Propagates the options associated to MAT_SYMMETRY_ETERNAL, MAT_STRUCTURALLY_SYMMETRIC, MAT_HERMITIAN, MAT_SPD and MAT_SYMMETRIC
8262: .seealso: MatSetOption()
8263: @*/
8264: PetscErrorCode MatPropagateSymmetryOptions(Mat A, Mat B)
8265: {
8271: if (A->symmetric_eternal) { /* symmetric_eternal does not have a corresponding *set flag */
8272: MatSetOption(B,MAT_SYMMETRY_ETERNAL,A->symmetric_eternal);
8273: }
8274: if (A->structurally_symmetric_set) {
8275: MatSetOption(B,MAT_STRUCTURALLY_SYMMETRIC,A->structurally_symmetric);
8276: }
8277: if (A->hermitian_set) {
8278: MatSetOption(B,MAT_HERMITIAN,A->hermitian);
8279: }
8280: if (A->spd_set) {
8281: MatSetOption(B,MAT_SPD,A->spd);
8282: }
8283: if (A->symmetric_set) {
8284: MatSetOption(B,MAT_SYMMETRIC,A->symmetric);
8285: }
8286: return(0);
8287: }
8289: /*@
8290: MatStashSetInitialSize - sets the sizes of the matrix stash, that is
8291: used during the assembly process to store values that belong to
8292: other processors.
8294: Not Collective
8296: Input Parameters:
8297: + mat - the matrix
8298: . size - the initial size of the stash.
8299: - bsize - the initial size of the block-stash(if used).
8301: Options Database Keys:
8302: + -matstash_initial_size <size> or <size0,size1,...sizep-1>
8303: - -matstash_block_initial_size <bsize> or <bsize0,bsize1,...bsizep-1>
8305: Level: intermediate
8307: Notes:
8308: The block-stash is used for values set with MatSetValuesBlocked() while
8309: the stash is used for values set with MatSetValues()
8311: Run with the option -info and look for output of the form
8312: MatAssemblyBegin_MPIXXX:Stash has MM entries, uses nn mallocs.
8313: to determine the appropriate value, MM, to use for size and
8314: MatAssemblyBegin_MPIXXX:Block-Stash has BMM entries, uses nn mallocs.
8315: to determine the value, BMM to use for bsize
8317: .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashGetInfo()
8319: @*/
8320: PetscErrorCode MatStashSetInitialSize(Mat mat,PetscInt size, PetscInt bsize)
8321: {
8327: MatStashSetInitialSize_Private(&mat->stash,size);
8328: MatStashSetInitialSize_Private(&mat->bstash,bsize);
8329: return(0);
8330: }
8332: /*@
8333: MatInterpolateAdd - w = y + A*x or A'*x depending on the shape of
8334: the matrix
8336: Neighbor-wise Collective on Mat
8338: Input Parameters:
8339: + mat - the matrix
8340: . x,y - the vectors
8341: - w - where the result is stored
8343: Level: intermediate
8345: Notes:
8346: w may be the same vector as y.
8348: This allows one to use either the restriction or interpolation (its transpose)
8349: matrix to do the interpolation
8351: .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict()
8353: @*/
8354: PetscErrorCode MatInterpolateAdd(Mat A,Vec x,Vec y,Vec w)
8355: {
8357: PetscInt M,N,Ny;
8364: MatGetSize(A,&M,&N);
8365: VecGetSize(y,&Ny);
8366: if (M == Ny) {
8367: MatMultAdd(A,x,y,w);
8368: } else {
8369: MatMultTransposeAdd(A,x,y,w);
8370: }
8371: return(0);
8372: }
8374: /*@
8375: MatInterpolate - y = A*x or A'*x depending on the shape of
8376: the matrix
8378: Neighbor-wise Collective on Mat
8380: Input Parameters:
8381: + mat - the matrix
8382: - x,y - the vectors
8384: Level: intermediate
8386: Notes:
8387: This allows one to use either the restriction or interpolation (its transpose)
8388: matrix to do the interpolation
8390: .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict()
8392: @*/
8393: PetscErrorCode MatInterpolate(Mat A,Vec x,Vec y)
8394: {
8396: PetscInt M,N,Ny;
8402: MatGetSize(A,&M,&N);
8403: VecGetSize(y,&Ny);
8404: if (M == Ny) {
8405: MatMult(A,x,y);
8406: } else {
8407: MatMultTranspose(A,x,y);
8408: }
8409: return(0);
8410: }
8412: /*@
8413: MatRestrict - y = A*x or A'*x
8415: Neighbor-wise Collective on Mat
8417: Input Parameters:
8418: + mat - the matrix
8419: - x,y - the vectors
8421: Level: intermediate
8423: Notes:
8424: This allows one to use either the restriction or interpolation (its transpose)
8425: matrix to do the restriction
8427: .seealso: MatMultAdd(), MatMultTransposeAdd(), MatInterpolate()
8429: @*/
8430: PetscErrorCode MatRestrict(Mat A,Vec x,Vec y)
8431: {
8433: PetscInt M,N,Ny;
8439: MatGetSize(A,&M,&N);
8440: VecGetSize(y,&Ny);
8441: if (M == Ny) {
8442: MatMult(A,x,y);
8443: } else {
8444: MatMultTranspose(A,x,y);
8445: }
8446: return(0);
8447: }
8449: /*@
8450: MatMatInterpolateAdd - Y = W + A*X or W + A'*X
8452: Neighbor-wise Collective on Mat
8454: Input Parameters:
8455: + mat - the matrix
8456: - w, x - the input dense matrices
8458: Output Parameters:
8459: . y - the output dense matrix
8461: Level: intermediate
8463: Notes:
8464: This allows one to use either the restriction or interpolation (its transpose)
8465: matrix to do the interpolation. y matrix can be reused if already created with the proper sizes,
8466: otherwise it will be recreated. y must be initialized to NULL if not supplied.
8468: .seealso: MatInterpolateAdd(), MatMatInterpolate(), MatMatRestrict()
8470: @*/
8471: PetscErrorCode MatMatInterpolateAdd(Mat A,Mat x,Mat w,Mat *y)
8472: {
8474: PetscInt M,N,Mx,Nx,Mo,My = 0,Ny = 0;
8475: PetscBool trans = PETSC_TRUE;
8476: MatReuse reuse = MAT_INITIAL_MATRIX;
8484: MatGetSize(A,&M,&N);
8485: MatGetSize(x,&Mx,&Nx);
8486: if (N == Mx) trans = PETSC_FALSE;
8487: else if (M != Mx) SETERRQ4(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Size mismatch: A %Dx%D, X %Dx%D",M,N,Mx,Nx);
8488: Mo = trans ? N : M;
8489: if (*y) {
8490: MatGetSize(*y,&My,&Ny);
8491: if (Mo == My && Nx == Ny) { reuse = MAT_REUSE_MATRIX; }
8492: else {
8493: if (w && *y == w) SETERRQ6(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Cannot reuse y and w, size mismatch: A %Dx%D, X %Dx%D, Y %Dx%D",M,N,Mx,Nx,My,Ny);
8494: MatDestroy(y);
8495: }
8496: }
8498: if (w && *y == w) { /* this is to minimize changes in PCMG */
8499: PetscBool flg;
8501: PetscObjectQuery((PetscObject)*y,"__MatMatIntAdd_w",(PetscObject*)&w);
8502: if (w) {
8503: PetscInt My,Ny,Mw,Nw;
8505: PetscObjectTypeCompare((PetscObject)*y,((PetscObject)w)->type_name,&flg);
8506: MatGetSize(*y,&My,&Ny);
8507: MatGetSize(w,&Mw,&Nw);
8508: if (!flg || My != Mw || Ny != Nw) w = NULL;
8509: }
8510: if (!w) {
8511: MatDuplicate(*y,MAT_COPY_VALUES,&w);
8512: PetscObjectCompose((PetscObject)*y,"__MatMatIntAdd_w",(PetscObject)w);
8513: PetscLogObjectParent((PetscObject)*y,(PetscObject)w);
8514: PetscObjectDereference((PetscObject)w);
8515: } else {
8516: MatCopy(*y,w,UNKNOWN_NONZERO_PATTERN);
8517: }
8518: }
8519: if (!trans) {
8520: MatMatMult(A,x,reuse,PETSC_DEFAULT,y);
8521: } else {
8522: MatTransposeMatMult(A,x,reuse,PETSC_DEFAULT,y);
8523: }
8524: if (w) {
8525: MatAXPY(*y,1.0,w,UNKNOWN_NONZERO_PATTERN);
8526: }
8527: return(0);
8528: }
8530: /*@
8531: MatMatInterpolate - Y = A*X or A'*X
8533: Neighbor-wise Collective on Mat
8535: Input Parameters:
8536: + mat - the matrix
8537: - x - the input dense matrix
8539: Output Parameters:
8540: . y - the output dense matrix
8542: Level: intermediate
8544: Notes:
8545: This allows one to use either the restriction or interpolation (its transpose)
8546: matrix to do the interpolation. y matrix can be reused if already created with the proper sizes,
8547: otherwise it will be recreated. y must be initialized to NULL if not supplied.
8549: .seealso: MatInterpolate(), MatRestrict(), MatMatRestrict()
8551: @*/
8552: PetscErrorCode MatMatInterpolate(Mat A,Mat x,Mat *y)
8553: {
8557: MatMatInterpolateAdd(A,x,NULL,y);
8558: return(0);
8559: }
8561: /*@
8562: MatMatRestrict - Y = A*X or A'*X
8564: Neighbor-wise Collective on Mat
8566: Input Parameters:
8567: + mat - the matrix
8568: - x - the input dense matrix
8570: Output Parameters:
8571: . y - the output dense matrix
8573: Level: intermediate
8575: Notes:
8576: This allows one to use either the restriction or interpolation (its transpose)
8577: matrix to do the restriction. y matrix can be reused if already created with the proper sizes,
8578: otherwise it will be recreated. y must be initialized to NULL if not supplied.
8580: .seealso: MatRestrict(), MatInterpolate(), MatMatInterpolate()
8581: @*/
8582: PetscErrorCode MatMatRestrict(Mat A,Mat x,Mat *y)
8583: {
8587: MatMatInterpolateAdd(A,x,NULL,y);
8588: return(0);
8589: }
8591: /*@
8592: MatGetNullSpace - retrieves the null space of a matrix.
8594: Logically Collective on Mat
8596: Input Parameters:
8597: + mat - the matrix
8598: - nullsp - the null space object
8600: Level: developer
8602: .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace(), MatSetNullSpace()
8603: @*/
8604: PetscErrorCode MatGetNullSpace(Mat mat, MatNullSpace *nullsp)
8605: {
8609: *nullsp = (mat->symmetric_set && mat->symmetric && !mat->nullsp) ? mat->transnullsp : mat->nullsp;
8610: return(0);
8611: }
8613: /*@
8614: MatSetNullSpace - attaches a null space to a matrix.
8616: Logically Collective on Mat
8618: Input Parameters:
8619: + mat - the matrix
8620: - nullsp - the null space object
8622: Level: advanced
8624: Notes:
8625: This null space is used by the linear solvers. Overwrites any previous null space that may have been attached
8627: For inconsistent singular systems (linear systems where the right hand side is not in the range of the operator) you also likely should
8628: call MatSetTransposeNullSpace(). This allows the linear system to be solved in a least squares sense.
8630: You can remove the null space by calling this routine with an nullsp of NULL
8632: The fundamental theorem of linear algebra (Gilbert Strang, Introduction to Applied Mathematics, page 72) states that
8633: 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).
8634: 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
8635: 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
8636: 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).
8638: Krylov solvers can produce the minimal norm solution to the least squares problem by utilizing MatNullSpaceRemove().
8640: If the matrix is known to be symmetric because it is an SBAIJ matrix or one as called MatSetOption(mat,MAT_SYMMETRIC or MAT_SYMMETRIC_ETERNAL,PETSC_TRUE); this
8641: routine also automatically calls MatSetTransposeNullSpace().
8643: .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace(), MatGetNullSpace(), MatSetTransposeNullSpace(), MatGetTransposeNullSpace(), MatNullSpaceRemove()
8644: @*/
8645: PetscErrorCode MatSetNullSpace(Mat mat,MatNullSpace nullsp)
8646: {
8652: if (nullsp) {PetscObjectReference((PetscObject)nullsp);}
8653: MatNullSpaceDestroy(&mat->nullsp);
8654: mat->nullsp = nullsp;
8655: if (mat->symmetric_set && mat->symmetric) {
8656: MatSetTransposeNullSpace(mat,nullsp);
8657: }
8658: return(0);
8659: }
8661: /*@
8662: MatGetTransposeNullSpace - retrieves the null space of the transpose of a matrix.
8664: Logically Collective on Mat
8666: Input Parameters:
8667: + mat - the matrix
8668: - nullsp - the null space object
8670: Level: developer
8672: .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace(), MatSetTransposeNullSpace(), MatSetNullSpace(), MatGetNullSpace()
8673: @*/
8674: PetscErrorCode MatGetTransposeNullSpace(Mat mat, MatNullSpace *nullsp)
8675: {
8680: *nullsp = (mat->symmetric_set && mat->symmetric && !mat->transnullsp) ? mat->nullsp : mat->transnullsp;
8681: return(0);
8682: }
8684: /*@
8685: MatSetTransposeNullSpace - attaches a null space to a matrix.
8687: Logically Collective on Mat
8689: Input Parameters:
8690: + mat - the matrix
8691: - nullsp - the null space object
8693: Level: advanced
8695: Notes:
8696: For inconsistent singular systems (linear systems where the right hand side is not in the range of the operator) this allows the linear system to be solved in a least squares sense.
8697: You must also call MatSetNullSpace()
8699: The fundamental theorem of linear algebra (Gilbert Strang, Introduction to Applied Mathematics, page 72) states that
8700: 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).
8701: 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
8702: 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
8703: 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).
8705: Krylov solvers can produce the minimal norm solution to the least squares problem by utilizing MatNullSpaceRemove().
8707: .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace(), MatGetNullSpace(), MatSetNullSpace(), MatGetTransposeNullSpace(), MatNullSpaceRemove()
8708: @*/
8709: PetscErrorCode MatSetTransposeNullSpace(Mat mat,MatNullSpace nullsp)
8710: {
8716: if (nullsp) {PetscObjectReference((PetscObject)nullsp);}
8717: MatNullSpaceDestroy(&mat->transnullsp);
8718: mat->transnullsp = nullsp;
8719: return(0);
8720: }
8722: /*@
8723: MatSetNearNullSpace - attaches a null space to a matrix, which is often the null space (rigid body modes) of the operator without boundary conditions
8724: This null space will be used to provide near null space vectors to a multigrid preconditioner built from this matrix.
8726: Logically Collective on Mat
8728: Input Parameters:
8729: + mat - the matrix
8730: - nullsp - the null space object
8732: Level: advanced
8734: Notes:
8735: Overwrites any previous near null space that may have been attached
8737: You can remove the null space by calling this routine with an nullsp of NULL
8739: .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNullSpace(), MatNullSpaceCreateRigidBody(), MatGetNearNullSpace()
8740: @*/
8741: PetscErrorCode MatSetNearNullSpace(Mat mat,MatNullSpace nullsp)
8742: {
8749: MatCheckPreallocated(mat,1);
8750: if (nullsp) {PetscObjectReference((PetscObject)nullsp);}
8751: MatNullSpaceDestroy(&mat->nearnullsp);
8752: mat->nearnullsp = nullsp;
8753: return(0);
8754: }
8756: /*@
8757: MatGetNearNullSpace - Get null space attached with MatSetNearNullSpace()
8759: Not Collective
8761: Input Parameter:
8762: . mat - the matrix
8764: Output Parameter:
8765: . nullsp - the null space object, NULL if not set
8767: Level: developer
8769: .seealso: MatSetNearNullSpace(), MatGetNullSpace(), MatNullSpaceCreate()
8770: @*/
8771: PetscErrorCode MatGetNearNullSpace(Mat mat,MatNullSpace *nullsp)
8772: {
8777: MatCheckPreallocated(mat,1);
8778: *nullsp = mat->nearnullsp;
8779: return(0);
8780: }
8782: /*@C
8783: MatICCFactor - Performs in-place incomplete Cholesky factorization of matrix.
8785: Collective on Mat
8787: Input Parameters:
8788: + mat - the matrix
8789: . row - row/column permutation
8790: . fill - expected fill factor >= 1.0
8791: - level - level of fill, for ICC(k)
8793: Notes:
8794: Probably really in-place only when level of fill is zero, otherwise allocates
8795: new space to store factored matrix and deletes previous memory.
8797: Most users should employ the simplified KSP interface for linear solvers
8798: instead of working directly with matrix algebra routines such as this.
8799: See, e.g., KSPCreate().
8801: Level: developer
8803: .seealso: MatICCFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor()
8805: Developer Note: fortran interface is not autogenerated as the f90
8806: interface definition cannot be generated correctly [due to MatFactorInfo]
8808: @*/
8809: PetscErrorCode MatICCFactor(Mat mat,IS row,const MatFactorInfo *info)
8810: {
8818: if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"matrix must be square");
8819: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
8820: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
8821: if (!mat->ops->iccfactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
8822: MatCheckPreallocated(mat,1);
8823: (*mat->ops->iccfactor)(mat,row,info);
8824: PetscObjectStateIncrease((PetscObject)mat);
8825: return(0);
8826: }
8828: /*@
8829: MatDiagonalScaleLocal - Scales columns of a matrix given the scaling values including the
8830: ghosted ones.
8832: Not Collective
8834: Input Parameters:
8835: + mat - the matrix
8836: - diag = the diagonal values, including ghost ones
8838: Level: developer
8840: Notes:
8841: Works only for MPIAIJ and MPIBAIJ matrices
8843: .seealso: MatDiagonalScale()
8844: @*/
8845: PetscErrorCode MatDiagonalScaleLocal(Mat mat,Vec diag)
8846: {
8848: PetscMPIInt size;
8855: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled");
8856: PetscLogEventBegin(MAT_Scale,mat,0,0,0);
8857: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
8858: if (size == 1) {
8859: PetscInt n,m;
8860: VecGetSize(diag,&n);
8861: MatGetSize(mat,NULL,&m);
8862: if (m == n) {
8863: MatDiagonalScale(mat,NULL,diag);
8864: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only supported for sequential matrices when no ghost points/periodic conditions");
8865: } else {
8866: PetscUseMethod(mat,"MatDiagonalScaleLocal_C",(Mat,Vec),(mat,diag));
8867: }
8868: PetscLogEventEnd(MAT_Scale,mat,0,0,0);
8869: PetscObjectStateIncrease((PetscObject)mat);
8870: return(0);
8871: }
8873: /*@
8874: MatGetInertia - Gets the inertia from a factored matrix
8876: Collective on Mat
8878: Input Parameter:
8879: . mat - the matrix
8881: Output Parameters:
8882: + nneg - number of negative eigenvalues
8883: . nzero - number of zero eigenvalues
8884: - npos - number of positive eigenvalues
8886: Level: advanced
8888: Notes:
8889: Matrix must have been factored by MatCholeskyFactor()
8891: @*/
8892: PetscErrorCode MatGetInertia(Mat mat,PetscInt *nneg,PetscInt *nzero,PetscInt *npos)
8893: {
8899: if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
8900: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Numeric factor mat is not assembled");
8901: if (!mat->ops->getinertia) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
8902: (*mat->ops->getinertia)(mat,nneg,nzero,npos);
8903: return(0);
8904: }
8906: /* ----------------------------------------------------------------*/
8907: /*@C
8908: MatSolves - Solves A x = b, given a factored matrix, for a collection of vectors
8910: Neighbor-wise Collective on Mats
8912: Input Parameters:
8913: + mat - the factored matrix
8914: - b - the right-hand-side vectors
8916: Output Parameter:
8917: . x - the result vectors
8919: Notes:
8920: The vectors b and x cannot be the same. I.e., one cannot
8921: call MatSolves(A,x,x).
8923: Notes:
8924: Most users should employ the simplified KSP interface for linear solvers
8925: instead of working directly with matrix algebra routines such as this.
8926: See, e.g., KSPCreate().
8928: Level: developer
8930: .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd(), MatSolve()
8931: @*/
8932: PetscErrorCode MatSolves(Mat mat,Vecs b,Vecs x)
8933: {
8939: if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors");
8940: if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
8941: if (!mat->rmap->N && !mat->cmap->N) return(0);
8943: if (!mat->ops->solves) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
8944: MatCheckPreallocated(mat,1);
8945: PetscLogEventBegin(MAT_Solves,mat,0,0,0);
8946: (*mat->ops->solves)(mat,b,x);
8947: PetscLogEventEnd(MAT_Solves,mat,0,0,0);
8948: return(0);
8949: }
8951: /*@
8952: MatIsSymmetric - Test whether a matrix is symmetric
8954: Collective on Mat
8956: Input Parameters:
8957: + A - the matrix to test
8958: - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact transpose)
8960: Output Parameters:
8961: . flg - the result
8963: Notes:
8964: For real numbers MatIsSymmetric() and MatIsHermitian() return identical results
8966: Level: intermediate
8968: .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetricKnown()
8969: @*/
8970: PetscErrorCode MatIsSymmetric(Mat A,PetscReal tol,PetscBool *flg)
8971: {
8978: if (!A->symmetric_set) {
8979: if (!A->ops->issymmetric) {
8980: MatType mattype;
8981: MatGetType(A,&mattype);
8982: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type %s does not support checking for symmetric",mattype);
8983: }
8984: (*A->ops->issymmetric)(A,tol,flg);
8985: if (!tol) {
8986: MatSetOption(A,MAT_SYMMETRIC,*flg);
8987: }
8988: } else if (A->symmetric) {
8989: *flg = PETSC_TRUE;
8990: } else if (!tol) {
8991: *flg = PETSC_FALSE;
8992: } else {
8993: if (!A->ops->issymmetric) {
8994: MatType mattype;
8995: MatGetType(A,&mattype);
8996: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type %s does not support checking for symmetric",mattype);
8997: }
8998: (*A->ops->issymmetric)(A,tol,flg);
8999: }
9000: return(0);
9001: }
9003: /*@
9004: MatIsHermitian - Test whether a matrix is Hermitian
9006: Collective on Mat
9008: Input Parameters:
9009: + A - the matrix to test
9010: - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact Hermitian)
9012: Output Parameters:
9013: . flg - the result
9015: Level: intermediate
9017: .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(),
9018: MatIsSymmetricKnown(), MatIsSymmetric()
9019: @*/
9020: PetscErrorCode MatIsHermitian(Mat A,PetscReal tol,PetscBool *flg)
9021: {
9028: if (!A->hermitian_set) {
9029: if (!A->ops->ishermitian) {
9030: MatType mattype;
9031: MatGetType(A,&mattype);
9032: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type %s does not support checking for hermitian",mattype);
9033: }
9034: (*A->ops->ishermitian)(A,tol,flg);
9035: if (!tol) {
9036: MatSetOption(A,MAT_HERMITIAN,*flg);
9037: }
9038: } else if (A->hermitian) {
9039: *flg = PETSC_TRUE;
9040: } else if (!tol) {
9041: *flg = PETSC_FALSE;
9042: } else {
9043: if (!A->ops->ishermitian) {
9044: MatType mattype;
9045: MatGetType(A,&mattype);
9046: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type %s does not support checking for hermitian",mattype);
9047: }
9048: (*A->ops->ishermitian)(A,tol,flg);
9049: }
9050: return(0);
9051: }
9053: /*@
9054: MatIsSymmetricKnown - Checks the flag on the matrix to see if it is symmetric.
9056: Not Collective
9058: Input Parameter:
9059: . A - the matrix to check
9061: Output Parameters:
9062: + set - if the symmetric flag is set (this tells you if the next flag is valid)
9063: - flg - the result
9065: Level: advanced
9067: Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsSymmetric()
9068: if you want it explicitly checked
9070: .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric()
9071: @*/
9072: PetscErrorCode MatIsSymmetricKnown(Mat A,PetscBool *set,PetscBool *flg)
9073: {
9078: if (A->symmetric_set) {
9079: *set = PETSC_TRUE;
9080: *flg = A->symmetric;
9081: } else {
9082: *set = PETSC_FALSE;
9083: }
9084: return(0);
9085: }
9087: /*@
9088: MatIsHermitianKnown - Checks the flag on the matrix to see if it is hermitian.
9090: Not Collective
9092: Input Parameter:
9093: . A - the matrix to check
9095: Output Parameters:
9096: + set - if the hermitian flag is set (this tells you if the next flag is valid)
9097: - flg - the result
9099: Level: advanced
9101: Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsHermitian()
9102: if you want it explicitly checked
9104: .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric()
9105: @*/
9106: PetscErrorCode MatIsHermitianKnown(Mat A,PetscBool *set,PetscBool *flg)
9107: {
9112: if (A->hermitian_set) {
9113: *set = PETSC_TRUE;
9114: *flg = A->hermitian;
9115: } else {
9116: *set = PETSC_FALSE;
9117: }
9118: return(0);
9119: }
9121: /*@
9122: MatIsStructurallySymmetric - Test whether a matrix is structurally symmetric
9124: Collective on Mat
9126: Input Parameter:
9127: . A - the matrix to test
9129: Output Parameters:
9130: . flg - the result
9132: Level: intermediate
9134: .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsSymmetric(), MatSetOption()
9135: @*/
9136: PetscErrorCode MatIsStructurallySymmetric(Mat A,PetscBool *flg)
9137: {
9143: if (!A->structurally_symmetric_set) {
9144: if (!A->ops->isstructurallysymmetric) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Matrix of type %s does not support checking for structural symmetric",((PetscObject)A)->type_name);
9145: (*A->ops->isstructurallysymmetric)(A,flg);
9146: MatSetOption(A,MAT_STRUCTURALLY_SYMMETRIC,*flg);
9147: } else *flg = A->structurally_symmetric;
9148: return(0);
9149: }
9151: /*@
9152: MatStashGetInfo - Gets how many values are currently in the matrix stash, i.e. need
9153: to be communicated to other processors during the MatAssemblyBegin/End() process
9155: Not collective
9157: Input Parameter:
9158: . vec - the vector
9160: Output Parameters:
9161: + nstash - the size of the stash
9162: . reallocs - the number of additional mallocs incurred.
9163: . bnstash - the size of the block stash
9164: - breallocs - the number of additional mallocs incurred.in the block stash
9166: Level: advanced
9168: .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashSetInitialSize()
9170: @*/
9171: PetscErrorCode MatStashGetInfo(Mat mat,PetscInt *nstash,PetscInt *reallocs,PetscInt *bnstash,PetscInt *breallocs)
9172: {
9176: MatStashGetInfo_Private(&mat->stash,nstash,reallocs);
9177: MatStashGetInfo_Private(&mat->bstash,bnstash,breallocs);
9178: return(0);
9179: }
9181: /*@C
9182: MatCreateVecs - Get vector(s) compatible with the matrix, i.e. with the same
9183: parallel layout
9185: Collective on Mat
9187: Input Parameter:
9188: . mat - the matrix
9190: Output Parameters:
9191: + right - (optional) vector that the matrix can be multiplied against
9192: - left - (optional) vector that the matrix vector product can be stored in
9194: Notes:
9195: 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().
9197: Notes:
9198: These are new vectors which are not owned by the Mat, they should be destroyed in VecDestroy() when no longer needed
9200: Level: advanced
9202: .seealso: MatCreate(), VecDestroy()
9203: @*/
9204: PetscErrorCode MatCreateVecs(Mat mat,Vec *right,Vec *left)
9205: {
9211: if (mat->ops->getvecs) {
9212: (*mat->ops->getvecs)(mat,right,left);
9213: } else {
9214: PetscInt rbs,cbs;
9215: MatGetBlockSizes(mat,&rbs,&cbs);
9216: if (right) {
9217: if (mat->cmap->n < 0) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"PetscLayout for columns not yet setup");
9218: VecCreate(PetscObjectComm((PetscObject)mat),right);
9219: VecSetSizes(*right,mat->cmap->n,PETSC_DETERMINE);
9220: VecSetBlockSize(*right,cbs);
9221: VecSetType(*right,mat->defaultvectype);
9222: PetscLayoutReference(mat->cmap,&(*right)->map);
9223: }
9224: if (left) {
9225: if (mat->rmap->n < 0) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"PetscLayout for rows not yet setup");
9226: VecCreate(PetscObjectComm((PetscObject)mat),left);
9227: VecSetSizes(*left,mat->rmap->n,PETSC_DETERMINE);
9228: VecSetBlockSize(*left,rbs);
9229: VecSetType(*left,mat->defaultvectype);
9230: PetscLayoutReference(mat->rmap,&(*left)->map);
9231: }
9232: }
9233: return(0);
9234: }
9236: /*@C
9237: MatFactorInfoInitialize - Initializes a MatFactorInfo data structure
9238: with default values.
9240: Not Collective
9242: Input Parameters:
9243: . info - the MatFactorInfo data structure
9245: Notes:
9246: The solvers are generally used through the KSP and PC objects, for example
9247: PCLU, PCILU, PCCHOLESKY, PCICC
9249: Level: developer
9251: .seealso: MatFactorInfo
9253: Developer Note: fortran interface is not autogenerated as the f90
9254: interface definition cannot be generated correctly [due to MatFactorInfo]
9256: @*/
9258: PetscErrorCode MatFactorInfoInitialize(MatFactorInfo *info)
9259: {
9263: PetscMemzero(info,sizeof(MatFactorInfo));
9264: return(0);
9265: }
9267: /*@
9268: MatFactorSetSchurIS - Set indices corresponding to the Schur complement you wish to have computed
9270: Collective on Mat
9272: Input Parameters:
9273: + mat - the factored matrix
9274: - is - the index set defining the Schur indices (0-based)
9276: Notes:
9277: Call MatFactorSolveSchurComplement() or MatFactorSolveSchurComplementTranspose() after this call to solve a Schur complement system.
9279: You can call MatFactorGetSchurComplement() or MatFactorCreateSchurComplement() after this call.
9281: Level: developer
9283: .seealso: MatGetFactor(), MatFactorGetSchurComplement(), MatFactorRestoreSchurComplement(), MatFactorCreateSchurComplement(), MatFactorSolveSchurComplement(),
9284: MatFactorSolveSchurComplementTranspose(), MatFactorSolveSchurComplement()
9286: @*/
9287: PetscErrorCode MatFactorSetSchurIS(Mat mat,IS is)
9288: {
9289: PetscErrorCode ierr,(*f)(Mat,IS);
9297: if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix");
9298: PetscObjectQueryFunction((PetscObject)mat,"MatFactorSetSchurIS_C",&f);
9299: if (!f) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"The selected MatSolverType does not support Schur complement computation. You should use MATSOLVERMUMPS or MATSOLVERMKL_PARDISO");
9300: MatDestroy(&mat->schur);
9301: (*f)(mat,is);
9302: if (!mat->schur) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_PLIB,"Schur complement has not been created");
9303: return(0);
9304: }
9306: /*@
9307: MatFactorCreateSchurComplement - Create a Schur complement matrix object using Schur data computed during the factorization step
9309: Logically Collective on Mat
9311: Input Parameters:
9312: + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface
9313: . S - location where to return the Schur complement, can be NULL
9314: - status - the status of the Schur complement matrix, can be NULL
9316: Notes:
9317: You must call MatFactorSetSchurIS() before calling this routine.
9319: The routine provides a copy of the Schur matrix stored within the solver data structures.
9320: The caller must destroy the object when it is no longer needed.
9321: If MatFactorInvertSchurComplement() has been called, the routine gets back the inverse.
9323: 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)
9325: Developer Notes:
9326: The reason this routine exists is because the representation of the Schur complement within the factor matrix may be different than a standard PETSc
9327: matrix representation and we normally do not want to use the time or memory to make a copy as a regular PETSc matrix.
9329: See MatCreateSchurComplement() or MatGetSchurComplement() for ways to create virtual or approximate Schur complements.
9331: Level: advanced
9333: References:
9335: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorGetSchurComplement(), MatFactorSchurStatus
9336: @*/
9337: PetscErrorCode MatFactorCreateSchurComplement(Mat F,Mat* S,MatFactorSchurStatus* status)
9338: {
9345: if (S) {
9346: PetscErrorCode (*f)(Mat,Mat*);
9348: PetscObjectQueryFunction((PetscObject)F,"MatFactorCreateSchurComplement_C",&f);
9349: if (f) {
9350: (*f)(F,S);
9351: } else {
9352: MatDuplicate(F->schur,MAT_COPY_VALUES,S);
9353: }
9354: }
9355: if (status) *status = F->schur_status;
9356: return(0);
9357: }
9359: /*@
9360: MatFactorGetSchurComplement - Gets access to a Schur complement matrix using the current Schur data within a factored matrix
9362: Logically Collective on Mat
9364: Input Parameters:
9365: + F - the factored matrix obtained by calling MatGetFactor()
9366: . *S - location where to return the Schur complement, can be NULL
9367: - status - the status of the Schur complement matrix, can be NULL
9369: Notes:
9370: You must call MatFactorSetSchurIS() before calling this routine.
9372: Schur complement mode is currently implemented for sequential matrices.
9373: The routine returns a the Schur Complement stored within the data strutures of the solver.
9374: If MatFactorInvertSchurComplement() has previously been called, the returned matrix is actually the inverse of the Schur complement.
9375: The returned matrix should not be destroyed; the caller should call MatFactorRestoreSchurComplement() when the object is no longer needed.
9377: Use MatFactorCreateSchurComplement() to create a copy of the Schur complement matrix that is within a factored matrix
9379: See MatCreateSchurComplement() or MatGetSchurComplement() for ways to create virtual or approximate Schur complements.
9381: Level: advanced
9383: References:
9385: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorRestoreSchurComplement(), MatFactorCreateSchurComplement(), MatFactorSchurStatus
9386: @*/
9387: PetscErrorCode MatFactorGetSchurComplement(Mat F,Mat* S,MatFactorSchurStatus* status)
9388: {
9393: if (S) *S = F->schur;
9394: if (status) *status = F->schur_status;
9395: return(0);
9396: }
9398: /*@
9399: MatFactorRestoreSchurComplement - Restore the Schur complement matrix object obtained from a call to MatFactorGetSchurComplement
9401: Logically Collective on Mat
9403: Input Parameters:
9404: + F - the factored matrix obtained by calling MatGetFactor()
9405: . *S - location where the Schur complement is stored
9406: - status - the status of the Schur complement matrix (see MatFactorSchurStatus)
9408: Notes:
9410: Level: advanced
9412: References:
9414: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorRestoreSchurComplement(), MatFactorCreateSchurComplement(), MatFactorSchurStatus
9415: @*/
9416: PetscErrorCode MatFactorRestoreSchurComplement(Mat F,Mat* S,MatFactorSchurStatus status)
9417: {
9422: if (S) {
9424: *S = NULL;
9425: }
9426: F->schur_status = status;
9427: MatFactorUpdateSchurStatus_Private(F);
9428: return(0);
9429: }
9431: /*@
9432: MatFactorSolveSchurComplementTranspose - Solve the transpose of the Schur complement system computed during the factorization step
9434: Logically Collective on Mat
9436: Input Parameters:
9437: + F - the factored matrix obtained by calling MatGetFactor()
9438: . rhs - location where the right hand side of the Schur complement system is stored
9439: - sol - location where the solution of the Schur complement system has to be returned
9441: Notes:
9442: The sizes of the vectors should match the size of the Schur complement
9444: Must be called after MatFactorSetSchurIS()
9446: Level: advanced
9448: References:
9450: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorSolveSchurComplement()
9451: @*/
9452: PetscErrorCode MatFactorSolveSchurComplementTranspose(Mat F, Vec rhs, Vec sol)
9453: {
9465: MatFactorFactorizeSchurComplement(F);
9466: switch (F->schur_status) {
9467: case MAT_FACTOR_SCHUR_FACTORED:
9468: MatSolveTranspose(F->schur,rhs,sol);
9469: break;
9470: case MAT_FACTOR_SCHUR_INVERTED:
9471: MatMultTranspose(F->schur,rhs,sol);
9472: break;
9473: default:
9474: SETERRQ1(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Unhandled MatFactorSchurStatus %D",F->schur_status);
9475: }
9476: return(0);
9477: }
9479: /*@
9480: MatFactorSolveSchurComplement - Solve the Schur complement system computed during the factorization step
9482: Logically Collective on Mat
9484: Input Parameters:
9485: + F - the factored matrix obtained by calling MatGetFactor()
9486: . rhs - location where the right hand side of the Schur complement system is stored
9487: - sol - location where the solution of the Schur complement system has to be returned
9489: Notes:
9490: The sizes of the vectors should match the size of the Schur complement
9492: Must be called after MatFactorSetSchurIS()
9494: Level: advanced
9496: References:
9498: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorSolveSchurComplementTranspose()
9499: @*/
9500: PetscErrorCode MatFactorSolveSchurComplement(Mat F, Vec rhs, Vec sol)
9501: {
9513: MatFactorFactorizeSchurComplement(F);
9514: switch (F->schur_status) {
9515: case MAT_FACTOR_SCHUR_FACTORED:
9516: MatSolve(F->schur,rhs,sol);
9517: break;
9518: case MAT_FACTOR_SCHUR_INVERTED:
9519: MatMult(F->schur,rhs,sol);
9520: break;
9521: default:
9522: SETERRQ1(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Unhandled MatFactorSchurStatus %D",F->schur_status);
9523: }
9524: return(0);
9525: }
9527: /*@
9528: MatFactorInvertSchurComplement - Invert the Schur complement matrix computed during the factorization step
9530: Logically Collective on Mat
9532: Input Parameters:
9533: . F - the factored matrix obtained by calling MatGetFactor()
9535: Notes:
9536: Must be called after MatFactorSetSchurIS().
9538: Call MatFactorGetSchurComplement() or MatFactorCreateSchurComplement() AFTER this call to actually compute the inverse and get access to it.
9540: Level: advanced
9542: References:
9544: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorGetSchurComplement(), MatFactorCreateSchurComplement()
9545: @*/
9546: PetscErrorCode MatFactorInvertSchurComplement(Mat F)
9547: {
9553: if (F->schur_status == MAT_FACTOR_SCHUR_INVERTED) return(0);
9554: MatFactorFactorizeSchurComplement(F);
9555: MatFactorInvertSchurComplement_Private(F);
9556: F->schur_status = MAT_FACTOR_SCHUR_INVERTED;
9557: return(0);
9558: }
9560: /*@
9561: MatFactorFactorizeSchurComplement - Factorize the Schur complement matrix computed during the factorization step
9563: Logically Collective on Mat
9565: Input Parameters:
9566: . F - the factored matrix obtained by calling MatGetFactor()
9568: Notes:
9569: Must be called after MatFactorSetSchurIS().
9571: Level: advanced
9573: References:
9575: .seealso: MatGetFactor(), MatFactorSetSchurIS(), MatFactorInvertSchurComplement()
9576: @*/
9577: PetscErrorCode MatFactorFactorizeSchurComplement(Mat F)
9578: {
9584: if (F->schur_status == MAT_FACTOR_SCHUR_INVERTED || F->schur_status == MAT_FACTOR_SCHUR_FACTORED) return(0);
9585: MatFactorFactorizeSchurComplement_Private(F);
9586: F->schur_status = MAT_FACTOR_SCHUR_FACTORED;
9587: return(0);
9588: }
9590: /*@
9591: MatPtAP - Creates the matrix product C = P^T * A * P
9593: Neighbor-wise Collective on Mat
9595: Input Parameters:
9596: + A - the matrix
9597: . P - the projection matrix
9598: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9599: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(P)), use PETSC_DEFAULT if you do not have a good estimate
9600: if the result is a dense matrix this is irrelevant
9602: Output Parameters:
9603: . C - the product matrix
9605: Notes:
9606: C will be created and must be destroyed by the user with MatDestroy().
9608: For matrix types without special implementation the function fallbacks to MatMatMult() followed by MatTransposeMatMult().
9610: Level: intermediate
9612: .seealso: MatMatMult(), MatRARt()
9613: @*/
9614: PetscErrorCode MatPtAP(Mat A,Mat P,MatReuse scall,PetscReal fill,Mat *C)
9615: {
9619: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*C,5);
9620: if (scall == MAT_INPLACE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Inplace product not supported");
9622: if (scall == MAT_INITIAL_MATRIX) {
9623: MatProductCreate(A,P,NULL,C);
9624: MatProductSetType(*C,MATPRODUCT_PtAP);
9625: MatProductSetAlgorithm(*C,"default");
9626: MatProductSetFill(*C,fill);
9628: (*C)->product->api_user = PETSC_TRUE;
9629: MatProductSetFromOptions(*C);
9630: if (!(*C)->ops->productsymbolic) SETERRQ3(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);
9631: MatProductSymbolic(*C);
9632: } else { /* scall == MAT_REUSE_MATRIX */
9633: MatProductReplaceMats(A,P,NULL,*C);
9634: }
9636: MatProductNumeric(*C);
9637: if (A->symmetric_set && A->symmetric) {
9638: MatSetOption(*C,MAT_SYMMETRIC,PETSC_TRUE);
9639: }
9640: return(0);
9641: }
9643: /*@
9644: MatRARt - Creates the matrix product C = R * A * R^T
9646: Neighbor-wise Collective on Mat
9648: Input Parameters:
9649: + A - the matrix
9650: . R - the projection matrix
9651: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9652: - fill - expected fill as ratio of nnz(C)/nnz(A), use PETSC_DEFAULT if you do not have a good estimate
9653: if the result is a dense matrix this is irrelevant
9655: Output Parameters:
9656: . C - the product matrix
9658: Notes:
9659: C will be created and must be destroyed by the user with MatDestroy().
9661: This routine is currently only implemented for pairs of AIJ matrices and classes
9662: which inherit from AIJ. Due to PETSc sparse matrix block row distribution among processes,
9663: parallel MatRARt is implemented via explicit transpose of R, which could be very expensive.
9664: We recommend using MatPtAP().
9666: Level: intermediate
9668: .seealso: MatMatMult(), MatPtAP()
9669: @*/
9670: PetscErrorCode MatRARt(Mat A,Mat R,MatReuse scall,PetscReal fill,Mat *C)
9671: {
9675: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*C,5);
9676: if (scall == MAT_INPLACE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Inplace product not supported");
9678: if (scall == MAT_INITIAL_MATRIX) {
9679: MatProductCreate(A,R,NULL,C);
9680: MatProductSetType(*C,MATPRODUCT_RARt);
9681: MatProductSetAlgorithm(*C,"default");
9682: MatProductSetFill(*C,fill);
9684: (*C)->product->api_user = PETSC_TRUE;
9685: MatProductSetFromOptions(*C);
9686: if (!(*C)->ops->productsymbolic) SETERRQ3(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);
9687: MatProductSymbolic(*C);
9688: } else { /* scall == MAT_REUSE_MATRIX */
9689: MatProductReplaceMats(A,R,NULL,*C);
9690: }
9692: MatProductNumeric(*C);
9693: if (A->symmetric_set && A->symmetric) {
9694: MatSetOption(*C,MAT_SYMMETRIC,PETSC_TRUE);
9695: }
9696: return(0);
9697: }
9699: static PetscErrorCode MatProduct_Private(Mat A,Mat B,MatReuse scall,PetscReal fill,MatProductType ptype, Mat *C)
9700: {
9704: if (scall == MAT_INPLACE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Inplace product not supported");
9706: if (scall == MAT_INITIAL_MATRIX) {
9707: PetscInfo1(A,"Calling MatProduct API with MAT_INITIAL_MATRIX and product type %s\n",MatProductTypes[ptype]);
9708: MatProductCreate(A,B,NULL,C);
9709: MatProductSetType(*C,ptype);
9710: MatProductSetAlgorithm(*C,MATPRODUCTALGORITHM_DEFAULT);
9711: MatProductSetFill(*C,fill);
9713: (*C)->product->api_user = PETSC_TRUE;
9714: MatProductSetFromOptions(*C);
9715: MatProductSymbolic(*C);
9716: } else { /* scall == MAT_REUSE_MATRIX */
9717: Mat_Product *product = (*C)->product;
9719: PetscInfo2(A,"Calling MatProduct API with MAT_REUSE_MATRIX %s product present and product type %s\n",product ? "with" : "without",MatProductTypes[ptype]);
9720: if (!product) {
9721: /* user provide the dense matrix *C without calling MatProductCreate() */
9722: PetscBool isdense;
9724: PetscObjectBaseTypeCompareAny((PetscObject)(*C),&isdense,MATSEQDENSE,MATMPIDENSE,"");
9725: if (isdense) {
9726: /* user wants to reuse an assembled dense matrix */
9727: /* Create product -- see MatCreateProduct() */
9728: MatProductCreate_Private(A,B,NULL,*C);
9729: product = (*C)->product;
9730: product->fill = fill;
9731: product->api_user = PETSC_TRUE;
9732: product->clear = PETSC_TRUE;
9734: MatProductSetType(*C,ptype);
9735: MatProductSetFromOptions(*C);
9736: if (!(*C)->ops->productsymbolic) SETERRQ3(PetscObjectComm((PetscObject)(*C)),PETSC_ERR_SUP,"MatProduct %s not supported for %s and %s",MatProductTypes[ptype],((PetscObject)A)->type_name,((PetscObject)B)->type_name);
9737: MatProductSymbolic(*C);
9738: } else SETERRQ(PetscObjectComm((PetscObject)(*C)),PETSC_ERR_SUP,"Call MatProductCreate() first");
9739: } else { /* user may change input matrices A or B when REUSE */
9740: MatProductReplaceMats(A,B,NULL,*C);
9741: }
9742: }
9743: MatProductNumeric(*C);
9744: return(0);
9745: }
9747: /*@
9748: MatMatMult - Performs Matrix-Matrix Multiplication C=A*B.
9750: Neighbor-wise Collective on Mat
9752: Input Parameters:
9753: + A - the left matrix
9754: . B - the right matrix
9755: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9756: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if you do not have a good estimate
9757: if the result is a dense matrix this is irrelevant
9759: Output Parameters:
9760: . C - the product matrix
9762: Notes:
9763: Unless scall is MAT_REUSE_MATRIX C will be created.
9765: 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
9766: call to this function with MAT_INITIAL_MATRIX.
9768: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value actually needed.
9770: If you have many matrices with the same non-zero structure to multiply, you should use MatProductCreate()/MatProductSymbolic()/MatProductReplaceMats(), and call MatProductNumeric() repeatedly.
9772: 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, rather than first having MatMatMult() create it for you. You can NEVER do this if the matrix C is sparse.
9774: Example of Usage:
9775: .vb
9776: MatProductCreate(A,B,NULL,&C);
9777: MatProductSetType(C,MATPRODUCT_AB);
9778: MatProductSymbolic(C);
9779: MatProductNumeric(C); // compute C=A * B
9780: MatProductReplaceMats(A1,B1,NULL,C); // compute C=A1 * B1
9781: MatProductNumeric(C);
9782: MatProductReplaceMats(A2,NULL,NULL,C); // compute C=A2 * B1
9783: MatProductNumeric(C);
9784: .ve
9786: Level: intermediate
9788: .seealso: MatTransposeMatMult(), MatMatTransposeMult(), MatPtAP(), MatProductCreate(), MatProductSymbolic(), MatProductReplaceMats(), MatProductNumeric()
9789: @*/
9790: PetscErrorCode MatMatMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
9791: {
9795: MatProduct_Private(A,B,scall,fill,MATPRODUCT_AB,C);
9796: return(0);
9797: }
9799: /*@
9800: MatMatTransposeMult - Performs Matrix-Matrix Multiplication C=A*B^T.
9802: Neighbor-wise Collective on Mat
9804: Input Parameters:
9805: + A - the left matrix
9806: . B - the right matrix
9807: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9808: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if not known
9810: Output Parameters:
9811: . C - the product matrix
9813: Notes:
9814: C will be created if MAT_INITIAL_MATRIX and must be destroyed by the user with MatDestroy().
9816: MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call
9818: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
9819: actually needed.
9821: This routine is currently only implemented for pairs of SeqAIJ matrices, for the SeqDense class,
9822: and for pairs of MPIDense matrices.
9824: Options Database Keys:
9825: . -matmattransmult_mpidense_mpidense_via {allgatherv,cyclic} - Choose between algorthims for MPIDense matrices: the
9826: first redundantly copies the transposed B matrix on each process and requiers O(log P) communication complexity;
9827: the second never stores more than one portion of the B matrix at a time by requires O(P) communication complexity.
9829: Level: intermediate
9831: .seealso: MatMatMult(), MatTransposeMatMult() MatPtAP()
9832: @*/
9833: PetscErrorCode MatMatTransposeMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
9834: {
9838: MatProduct_Private(A,B,scall,fill,MATPRODUCT_ABt,C);
9839: return(0);
9840: }
9842: /*@
9843: MatTransposeMatMult - Performs Matrix-Matrix Multiplication C=A^T*B.
9845: Neighbor-wise Collective on Mat
9847: Input Parameters:
9848: + A - the left matrix
9849: . B - the right matrix
9850: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9851: - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if not known
9853: Output Parameters:
9854: . C - the product matrix
9856: Notes:
9857: C will be created if MAT_INITIAL_MATRIX and must be destroyed by the user with MatDestroy().
9859: MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call.
9861: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
9862: actually needed.
9864: This routine is currently implemented for pairs of AIJ matrices and pairs of SeqDense matrices and classes
9865: which inherit from SeqAIJ. C will be of same type as the input matrices.
9867: Level: intermediate
9869: .seealso: MatMatMult(), MatMatTransposeMult(), MatPtAP()
9870: @*/
9871: PetscErrorCode MatTransposeMatMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
9872: {
9876: MatProduct_Private(A,B,scall,fill,MATPRODUCT_AtB,C);
9877: return(0);
9878: }
9880: /*@
9881: MatMatMatMult - Performs Matrix-Matrix-Matrix Multiplication D=A*B*C.
9883: Neighbor-wise Collective on Mat
9885: Input Parameters:
9886: + A - the left matrix
9887: . B - the middle matrix
9888: . C - the right matrix
9889: . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9890: - 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
9891: if the result is a dense matrix this is irrelevant
9893: Output Parameters:
9894: . D - the product matrix
9896: Notes:
9897: Unless scall is MAT_REUSE_MATRIX D will be created.
9899: MAT_REUSE_MATRIX can only be used if the matrices A, B and C have the same nonzero pattern as in the previous call
9901: To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
9902: actually needed.
9904: If you have many matrices with the same non-zero structure to multiply, you
9905: should use MAT_REUSE_MATRIX in all calls but the first or
9907: Level: intermediate
9909: .seealso: MatMatMult, MatPtAP()
9910: @*/
9911: PetscErrorCode MatMatMatMult(Mat A,Mat B,Mat C,MatReuse scall,PetscReal fill,Mat *D)
9912: {
9916: if (scall == MAT_REUSE_MATRIX) MatCheckProduct(*D,6);
9917: if (scall == MAT_INPLACE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Inplace product not supported");
9919: if (scall == MAT_INITIAL_MATRIX) {
9920: MatProductCreate(A,B,C,D);
9921: MatProductSetType(*D,MATPRODUCT_ABC);
9922: MatProductSetAlgorithm(*D,"default");
9923: MatProductSetFill(*D,fill);
9925: (*D)->product->api_user = PETSC_TRUE;
9926: MatProductSetFromOptions(*D);
9927: if (!(*D)->ops->productsymbolic) SETERRQ4(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,((PetscObject)C)->type_name);
9928: MatProductSymbolic(*D);
9929: } else { /* user may change input matrices when REUSE */
9930: MatProductReplaceMats(A,B,C,*D);
9931: }
9932: MatProductNumeric(*D);
9933: return(0);
9934: }
9936: /*@
9937: MatCreateRedundantMatrix - Create redundant matrices and put them into processors of subcommunicators.
9939: Collective on Mat
9941: Input Parameters:
9942: + mat - the matrix
9943: . nsubcomm - the number of subcommunicators (= number of redundant parallel or sequential matrices)
9944: . subcomm - MPI communicator split from the communicator where mat resides in (or MPI_COMM_NULL if nsubcomm is used)
9945: - reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
9947: Output Parameter:
9948: . matredundant - redundant matrix
9950: Notes:
9951: MAT_REUSE_MATRIX can only be used when the nonzero structure of the
9952: original matrix has not changed from that last call to MatCreateRedundantMatrix().
9954: This routine creates the duplicated matrices in subcommunicators; you should NOT create them before
9955: calling it.
9957: Level: advanced
9959: .seealso: MatDestroy()
9960: @*/
9961: PetscErrorCode MatCreateRedundantMatrix(Mat mat,PetscInt nsubcomm,MPI_Comm subcomm,MatReuse reuse,Mat *matredundant)
9962: {
9964: MPI_Comm comm;
9965: PetscMPIInt size;
9966: PetscInt mloc_sub,nloc_sub,rstart,rend,M=mat->rmap->N,N=mat->cmap->N,bs=mat->rmap->bs;
9967: Mat_Redundant *redund=NULL;
9968: PetscSubcomm psubcomm=NULL;
9969: MPI_Comm subcomm_in=subcomm;
9970: Mat *matseq;
9971: IS isrow,iscol;
9972: PetscBool newsubcomm=PETSC_FALSE;
9976: if (nsubcomm && reuse == MAT_REUSE_MATRIX) {
9979: }
9981: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
9982: if (size == 1 || nsubcomm == 1) {
9983: if (reuse == MAT_INITIAL_MATRIX) {
9984: MatDuplicate(mat,MAT_COPY_VALUES,matredundant);
9985: } else {
9986: if (*matredundant == mat) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
9987: MatCopy(mat,*matredundant,SAME_NONZERO_PATTERN);
9988: }
9989: return(0);
9990: }
9992: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
9993: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
9994: MatCheckPreallocated(mat,1);
9996: PetscLogEventBegin(MAT_RedundantMat,mat,0,0,0);
9997: if (subcomm_in == MPI_COMM_NULL && reuse == MAT_INITIAL_MATRIX) { /* get subcomm if user does not provide subcomm */
9998: /* create psubcomm, then get subcomm */
9999: PetscObjectGetComm((PetscObject)mat,&comm);
10000: MPI_Comm_size(comm,&size);
10001: if (nsubcomm < 1 || nsubcomm > size) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"nsubcomm must between 1 and %D",size);
10003: PetscSubcommCreate(comm,&psubcomm);
10004: PetscSubcommSetNumber(psubcomm,nsubcomm);
10005: PetscSubcommSetType(psubcomm,PETSC_SUBCOMM_CONTIGUOUS);
10006: PetscSubcommSetFromOptions(psubcomm);
10007: PetscCommDuplicate(PetscSubcommChild(psubcomm),&subcomm,NULL);
10008: newsubcomm = PETSC_TRUE;
10009: PetscSubcommDestroy(&psubcomm);
10010: }
10012: /* get isrow, iscol and a local sequential matrix matseq[0] */
10013: if (reuse == MAT_INITIAL_MATRIX) {
10014: mloc_sub = PETSC_DECIDE;
10015: nloc_sub = PETSC_DECIDE;
10016: if (bs < 1) {
10017: PetscSplitOwnership(subcomm,&mloc_sub,&M);
10018: PetscSplitOwnership(subcomm,&nloc_sub,&N);
10019: } else {
10020: PetscSplitOwnershipBlock(subcomm,bs,&mloc_sub,&M);
10021: PetscSplitOwnershipBlock(subcomm,bs,&nloc_sub,&N);
10022: }
10023: MPI_Scan(&mloc_sub,&rend,1,MPIU_INT,MPI_SUM,subcomm);
10024: rstart = rend - mloc_sub;
10025: ISCreateStride(PETSC_COMM_SELF,mloc_sub,rstart,1,&isrow);
10026: ISCreateStride(PETSC_COMM_SELF,N,0,1,&iscol);
10027: } else { /* reuse == MAT_REUSE_MATRIX */
10028: if (*matredundant == mat) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10029: /* retrieve subcomm */
10030: PetscObjectGetComm((PetscObject)(*matredundant),&subcomm);
10031: redund = (*matredundant)->redundant;
10032: isrow = redund->isrow;
10033: iscol = redund->iscol;
10034: matseq = redund->matseq;
10035: }
10036: MatCreateSubMatrices(mat,1,&isrow,&iscol,reuse,&matseq);
10038: /* get matredundant over subcomm */
10039: if (reuse == MAT_INITIAL_MATRIX) {
10040: MatCreateMPIMatConcatenateSeqMat(subcomm,matseq[0],nloc_sub,reuse,matredundant);
10042: /* create a supporting struct and attach it to C for reuse */
10043: PetscNewLog(*matredundant,&redund);
10044: (*matredundant)->redundant = redund;
10045: redund->isrow = isrow;
10046: redund->iscol = iscol;
10047: redund->matseq = matseq;
10048: if (newsubcomm) {
10049: redund->subcomm = subcomm;
10050: } else {
10051: redund->subcomm = MPI_COMM_NULL;
10052: }
10053: } else {
10054: MatCreateMPIMatConcatenateSeqMat(subcomm,matseq[0],PETSC_DECIDE,reuse,matredundant);
10055: }
10056: PetscLogEventEnd(MAT_RedundantMat,mat,0,0,0);
10057: return(0);
10058: }
10060: /*@C
10061: MatGetMultiProcBlock - Create multiple [bjacobi] 'parallel submatrices' from
10062: a given 'mat' object. Each submatrix can span multiple procs.
10064: Collective on Mat
10066: Input Parameters:
10067: + mat - the matrix
10068: . subcomm - the subcommunicator obtained by com_split(comm)
10069: - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
10071: Output Parameter:
10072: . subMat - 'parallel submatrices each spans a given subcomm
10074: Notes:
10075: The submatrix partition across processors is dictated by 'subComm' a
10076: communicator obtained by com_split(comm). The comm_split
10077: is not restriced to be grouped with consecutive original ranks.
10079: Due the comm_split() usage, the parallel layout of the submatrices
10080: map directly to the layout of the original matrix [wrt the local
10081: row,col partitioning]. So the original 'DiagonalMat' naturally maps
10082: into the 'DiagonalMat' of the subMat, hence it is used directly from
10083: the subMat. However the offDiagMat looses some columns - and this is
10084: reconstructed with MatSetValues()
10086: Level: advanced
10088: .seealso: MatCreateSubMatrices()
10089: @*/
10090: PetscErrorCode MatGetMultiProcBlock(Mat mat, MPI_Comm subComm, MatReuse scall,Mat *subMat)
10091: {
10093: PetscMPIInt commsize,subCommSize;
10096: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&commsize);
10097: MPI_Comm_size(subComm,&subCommSize);
10098: if (subCommSize > commsize) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"CommSize %D < SubCommZize %D",commsize,subCommSize);
10100: if (scall == MAT_REUSE_MATRIX && *subMat == mat) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10101: PetscLogEventBegin(MAT_GetMultiProcBlock,mat,0,0,0);
10102: (*mat->ops->getmultiprocblock)(mat,subComm,scall,subMat);
10103: PetscLogEventEnd(MAT_GetMultiProcBlock,mat,0,0,0);
10104: return(0);
10105: }
10107: /*@
10108: MatGetLocalSubMatrix - Gets a reference to a submatrix specified in local numbering
10110: Not Collective
10112: Input Parameters:
10113: + mat - matrix to extract local submatrix from
10114: . isrow - local row indices for submatrix
10115: - iscol - local column indices for submatrix
10117: Output Parameter:
10118: . submat - the submatrix
10120: Level: intermediate
10122: Notes:
10123: The submat should be returned with MatRestoreLocalSubMatrix().
10125: Depending on the format of mat, the returned submat may not implement MatMult(). Its communicator may be
10126: the same as mat, it may be PETSC_COMM_SELF, or some other subcomm of mat's.
10128: The submat always implements MatSetValuesLocal(). If isrow and iscol have the same block size, then
10129: MatSetValuesBlockedLocal() will also be implemented.
10131: The mat must have had a ISLocalToGlobalMapping provided to it with MatSetLocalToGlobalMapping(). Note that
10132: matrices obtained with DMCreateMatrix() generally already have the local to global mapping provided.
10134: .seealso: MatRestoreLocalSubMatrix(), MatCreateLocalRef(), MatSetLocalToGlobalMapping()
10135: @*/
10136: PetscErrorCode MatGetLocalSubMatrix(Mat mat,IS isrow,IS iscol,Mat *submat)
10137: {
10146: if (!mat->rmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Matrix must have local to global mapping provided before this call");
10148: if (mat->ops->getlocalsubmatrix) {
10149: (*mat->ops->getlocalsubmatrix)(mat,isrow,iscol,submat);
10150: } else {
10151: MatCreateLocalRef(mat,isrow,iscol,submat);
10152: }
10153: return(0);
10154: }
10156: /*@
10157: MatRestoreLocalSubMatrix - Restores a reference to a submatrix specified in local numbering
10159: Not Collective
10161: Input Parameters:
10162: + mat - matrix to extract local submatrix from
10163: . isrow - local row indices for submatrix
10164: . iscol - local column indices for submatrix
10165: - submat - the submatrix
10167: Level: intermediate
10169: .seealso: MatGetLocalSubMatrix()
10170: @*/
10171: PetscErrorCode MatRestoreLocalSubMatrix(Mat mat,IS isrow,IS iscol,Mat *submat)
10172: {
10181: if (*submat) {
10183: }
10185: if (mat->ops->restorelocalsubmatrix) {
10186: (*mat->ops->restorelocalsubmatrix)(mat,isrow,iscol,submat);
10187: } else {
10188: MatDestroy(submat);
10189: }
10190: *submat = NULL;
10191: return(0);
10192: }
10194: /* --------------------------------------------------------*/
10195: /*@
10196: MatFindZeroDiagonals - Finds all the rows of a matrix that have zero or no diagonal entry in the matrix
10198: Collective on Mat
10200: Input Parameter:
10201: . mat - the matrix
10203: Output Parameter:
10204: . is - if any rows have zero diagonals this contains the list of them
10206: Level: developer
10208: .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
10209: @*/
10210: PetscErrorCode MatFindZeroDiagonals(Mat mat,IS *is)
10211: {
10217: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
10218: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
10220: if (!mat->ops->findzerodiagonals) {
10221: Vec diag;
10222: const PetscScalar *a;
10223: PetscInt *rows;
10224: PetscInt rStart, rEnd, r, nrow = 0;
10226: MatCreateVecs(mat, &diag, NULL);
10227: MatGetDiagonal(mat, diag);
10228: MatGetOwnershipRange(mat, &rStart, &rEnd);
10229: VecGetArrayRead(diag, &a);
10230: for (r = 0; r < rEnd-rStart; ++r) if (a[r] == 0.0) ++nrow;
10231: PetscMalloc1(nrow, &rows);
10232: nrow = 0;
10233: for (r = 0; r < rEnd-rStart; ++r) if (a[r] == 0.0) rows[nrow++] = r+rStart;
10234: VecRestoreArrayRead(diag, &a);
10235: VecDestroy(&diag);
10236: ISCreateGeneral(PetscObjectComm((PetscObject) mat), nrow, rows, PETSC_OWN_POINTER, is);
10237: } else {
10238: (*mat->ops->findzerodiagonals)(mat, is);
10239: }
10240: return(0);
10241: }
10243: /*@
10244: MatFindOffBlockDiagonalEntries - Finds all the rows of a matrix that have entries outside of the main diagonal block (defined by the matrix block size)
10246: Collective on Mat
10248: Input Parameter:
10249: . mat - the matrix
10251: Output Parameter:
10252: . is - contains the list of rows with off block diagonal entries
10254: Level: developer
10256: .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
10257: @*/
10258: PetscErrorCode MatFindOffBlockDiagonalEntries(Mat mat,IS *is)
10259: {
10265: if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
10266: if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
10268: if (!mat->ops->findoffblockdiagonalentries) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s does not have a find off block diagonal entries defined",((PetscObject)mat)->type_name);
10269: (*mat->ops->findoffblockdiagonalentries)(mat,is);
10270: return(0);
10271: }
10273: /*@C
10274: MatInvertBlockDiagonal - Inverts the block diagonal entries.
10276: Collective on Mat
10278: Input Parameters:
10279: . mat - the matrix
10281: Output Parameters:
10282: . values - the block inverses in column major order (FORTRAN-like)
10284: Note:
10285: This routine is not available from Fortran.
10287: Level: advanced
10289: .seealso: MatInvertBockDiagonalMat
10290: @*/
10291: PetscErrorCode MatInvertBlockDiagonal(Mat mat,const PetscScalar **values)
10292: {
10297: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
10298: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
10299: if (!mat->ops->invertblockdiagonal) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for type %s",((PetscObject)mat)->type_name);
10300: (*mat->ops->invertblockdiagonal)(mat,values);
10301: return(0);
10302: }
10304: /*@C
10305: MatInvertVariableBlockDiagonal - Inverts the block diagonal entries.
10307: Collective on Mat
10309: Input Parameters:
10310: + mat - the matrix
10311: . nblocks - the number of blocks
10312: - bsizes - the size of each block
10314: Output Parameters:
10315: . values - the block inverses in column major order (FORTRAN-like)
10317: Note:
10318: This routine is not available from Fortran.
10320: Level: advanced
10322: .seealso: MatInvertBockDiagonal()
10323: @*/
10324: PetscErrorCode MatInvertVariableBlockDiagonal(Mat mat,PetscInt nblocks,const PetscInt *bsizes,PetscScalar *values)
10325: {
10330: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
10331: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
10332: if (!mat->ops->invertvariableblockdiagonal) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for type",((PetscObject)mat)->type_name);
10333: (*mat->ops->invertvariableblockdiagonal)(mat,nblocks,bsizes,values);
10334: return(0);
10335: }
10337: /*@
10338: MatInvertBlockDiagonalMat - set matrix C to be the inverted block diagonal of matrix A
10340: Collective on Mat
10342: Input Parameters:
10343: . A - the matrix
10345: Output Parameters:
10346: . C - matrix with inverted block diagonal of A. This matrix should be created and may have its type set.
10348: Notes: the blocksize of the matrix is used to determine the blocks on the diagonal of C
10350: Level: advanced
10352: .seealso: MatInvertBockDiagonal()
10353: @*/
10354: PetscErrorCode MatInvertBlockDiagonalMat(Mat A,Mat C)
10355: {
10356: PetscErrorCode ierr;
10357: const PetscScalar *vals;
10358: PetscInt *dnnz;
10359: PetscInt M,N,m,n,rstart,rend,bs,i,j;
10362: MatInvertBlockDiagonal(A,&vals);
10363: MatGetBlockSize(A,&bs);
10364: MatGetSize(A,&M,&N);
10365: MatGetLocalSize(A,&m,&n);
10366: MatSetSizes(C,m,n,M,N);
10367: MatSetBlockSize(C,bs);
10368: PetscMalloc1(m/bs,&dnnz);
10369: for (j = 0; j < m/bs; j++) dnnz[j] = 1;
10370: MatXAIJSetPreallocation(C,bs,dnnz,NULL,NULL,NULL);
10371: PetscFree(dnnz);
10372: MatGetOwnershipRange(C,&rstart,&rend);
10373: MatSetOption(C,MAT_ROW_ORIENTED,PETSC_FALSE);
10374: for (i = rstart/bs; i < rend/bs; i++) {
10375: MatSetValuesBlocked(C,1,&i,1,&i,&vals[(i-rstart/bs)*bs*bs],INSERT_VALUES);
10376: }
10377: MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);
10378: MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);
10379: MatSetOption(C,MAT_ROW_ORIENTED,PETSC_TRUE);
10380: return(0);
10381: }
10383: /*@C
10384: MatTransposeColoringDestroy - Destroys a coloring context for matrix product C=A*B^T that was created
10385: via MatTransposeColoringCreate().
10387: Collective on MatTransposeColoring
10389: Input Parameter:
10390: . c - coloring context
10392: Level: intermediate
10394: .seealso: MatTransposeColoringCreate()
10395: @*/
10396: PetscErrorCode MatTransposeColoringDestroy(MatTransposeColoring *c)
10397: {
10398: PetscErrorCode ierr;
10399: MatTransposeColoring matcolor=*c;
10402: if (!matcolor) return(0);
10403: if (--((PetscObject)matcolor)->refct > 0) {matcolor = NULL; return(0);}
10405: PetscFree3(matcolor->ncolumns,matcolor->nrows,matcolor->colorforrow);
10406: PetscFree(matcolor->rows);
10407: PetscFree(matcolor->den2sp);
10408: PetscFree(matcolor->colorforcol);
10409: PetscFree(matcolor->columns);
10410: if (matcolor->brows>0) {
10411: PetscFree(matcolor->lstart);
10412: }
10413: PetscHeaderDestroy(c);
10414: return(0);
10415: }
10417: /*@C
10418: MatTransColoringApplySpToDen - Given a symbolic matrix product C=A*B^T for which
10419: a MatTransposeColoring context has been created, computes a dense B^T by Apply
10420: MatTransposeColoring to sparse B.
10422: Collective on MatTransposeColoring
10424: Input Parameters:
10425: + B - sparse matrix B
10426: . Btdense - symbolic dense matrix B^T
10427: - coloring - coloring context created with MatTransposeColoringCreate()
10429: Output Parameter:
10430: . Btdense - dense matrix B^T
10432: Level: advanced
10434: Notes:
10435: These are used internally for some implementations of MatRARt()
10437: .seealso: MatTransposeColoringCreate(), MatTransposeColoringDestroy(), MatTransColoringApplyDenToSp()
10439: @*/
10440: PetscErrorCode MatTransColoringApplySpToDen(MatTransposeColoring coloring,Mat B,Mat Btdense)
10441: {
10449: if (!B->ops->transcoloringapplysptoden) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for this matrix type %s",((PetscObject)B)->type_name);
10450: (B->ops->transcoloringapplysptoden)(coloring,B,Btdense);
10451: return(0);
10452: }
10454: /*@C
10455: MatTransColoringApplyDenToSp - Given a symbolic matrix product Csp=A*B^T for which
10456: a MatTransposeColoring context has been created and a dense matrix Cden=A*Btdense
10457: in which Btdens is obtained from MatTransColoringApplySpToDen(), recover sparse matrix
10458: Csp from Cden.
10460: Collective on MatTransposeColoring
10462: Input Parameters:
10463: + coloring - coloring context created with MatTransposeColoringCreate()
10464: - Cden - matrix product of a sparse matrix and a dense matrix Btdense
10466: Output Parameter:
10467: . Csp - sparse matrix
10469: Level: advanced
10471: Notes:
10472: These are used internally for some implementations of MatRARt()
10474: .seealso: MatTransposeColoringCreate(), MatTransposeColoringDestroy(), MatTransColoringApplySpToDen()
10476: @*/
10477: PetscErrorCode MatTransColoringApplyDenToSp(MatTransposeColoring matcoloring,Mat Cden,Mat Csp)
10478: {
10486: if (!Csp->ops->transcoloringapplydentosp) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for this matrix type %s",((PetscObject)Csp)->type_name);
10487: (Csp->ops->transcoloringapplydentosp)(matcoloring,Cden,Csp);
10488: MatAssemblyBegin(Csp,MAT_FINAL_ASSEMBLY);
10489: MatAssemblyEnd(Csp,MAT_FINAL_ASSEMBLY);
10490: return(0);
10491: }
10493: /*@C
10494: MatTransposeColoringCreate - Creates a matrix coloring context for matrix product C=A*B^T.
10496: Collective on Mat
10498: Input Parameters:
10499: + mat - the matrix product C
10500: - iscoloring - the coloring of the matrix; usually obtained with MatColoringCreate() or DMCreateColoring()
10502: Output Parameter:
10503: . color - the new coloring context
10505: Level: intermediate
10507: .seealso: MatTransposeColoringDestroy(), MatTransColoringApplySpToDen(),
10508: MatTransColoringApplyDenToSp()
10509: @*/
10510: PetscErrorCode MatTransposeColoringCreate(Mat mat,ISColoring iscoloring,MatTransposeColoring *color)
10511: {
10512: MatTransposeColoring c;
10513: MPI_Comm comm;
10514: PetscErrorCode ierr;
10517: PetscLogEventBegin(MAT_TransposeColoringCreate,mat,0,0,0);
10518: PetscObjectGetComm((PetscObject)mat,&comm);
10519: PetscHeaderCreate(c,MAT_TRANSPOSECOLORING_CLASSID,"MatTransposeColoring","Matrix product C=A*B^T via coloring","Mat",comm,MatTransposeColoringDestroy,NULL);
10521: c->ctype = iscoloring->ctype;
10522: if (mat->ops->transposecoloringcreate) {
10523: (*mat->ops->transposecoloringcreate)(mat,iscoloring,c);
10524: } else SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Code not yet written for matrix type %s",((PetscObject)mat)->type_name);
10526: *color = c;
10527: PetscLogEventEnd(MAT_TransposeColoringCreate,mat,0,0,0);
10528: return(0);
10529: }
10531: /*@
10532: MatGetNonzeroState - Returns a 64 bit integer representing the current state of nonzeros in the matrix. If the
10533: matrix has had no new nonzero locations added to the matrix since the previous call then the value will be the
10534: same, otherwise it will be larger
10536: Not Collective
10538: Input Parameter:
10539: . A - the matrix
10541: Output Parameter:
10542: . state - the current state
10544: Notes:
10545: You can only compare states from two different calls to the SAME matrix, you cannot compare calls between
10546: different matrices
10548: Level: intermediate
10550: @*/
10551: PetscErrorCode MatGetNonzeroState(Mat mat,PetscObjectState *state)
10552: {
10555: *state = mat->nonzerostate;
10556: return(0);
10557: }
10559: /*@
10560: MatCreateMPIMatConcatenateSeqMat - Creates a single large PETSc matrix by concatenating sequential
10561: matrices from each processor
10563: Collective
10565: Input Parameters:
10566: + comm - the communicators the parallel matrix will live on
10567: . seqmat - the input sequential matrices
10568: . n - number of local columns (or PETSC_DECIDE)
10569: - reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
10571: Output Parameter:
10572: . mpimat - the parallel matrix generated
10574: Level: advanced
10576: Notes:
10577: The number of columns of the matrix in EACH processor MUST be the same.
10579: @*/
10580: PetscErrorCode MatCreateMPIMatConcatenateSeqMat(MPI_Comm comm,Mat seqmat,PetscInt n,MatReuse reuse,Mat *mpimat)
10581: {
10585: if (!seqmat->ops->creatempimatconcatenateseqmat) SETERRQ1(PetscObjectComm((PetscObject)seqmat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)seqmat)->type_name);
10586: if (reuse == MAT_REUSE_MATRIX && seqmat == *mpimat) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"MAT_REUSE_MATRIX means reuse the matrix passed in as the final argument, not the original matrix");
10588: PetscLogEventBegin(MAT_Merge,seqmat,0,0,0);
10589: (*seqmat->ops->creatempimatconcatenateseqmat)(comm,seqmat,n,reuse,mpimat);
10590: PetscLogEventEnd(MAT_Merge,seqmat,0,0,0);
10591: return(0);
10592: }
10594: /*@
10595: MatSubdomainsCreateCoalesce - Creates index subdomains by coalescing adjacent
10596: ranks' ownership ranges.
10598: Collective on A
10600: Input Parameters:
10601: + A - the matrix to create subdomains from
10602: - N - requested number of subdomains
10604: Output Parameters:
10605: + n - number of subdomains resulting on this rank
10606: - iss - IS list with indices of subdomains on this rank
10608: Level: advanced
10610: Notes:
10611: number of subdomains must be smaller than the communicator size
10612: @*/
10613: PetscErrorCode MatSubdomainsCreateCoalesce(Mat A,PetscInt N,PetscInt *n,IS *iss[])
10614: {
10615: MPI_Comm comm,subcomm;
10616: PetscMPIInt size,rank,color;
10617: PetscInt rstart,rend,k;
10618: PetscErrorCode ierr;
10621: PetscObjectGetComm((PetscObject)A,&comm);
10622: MPI_Comm_size(comm,&size);
10623: MPI_Comm_rank(comm,&rank);
10624: if (N < 1 || N >= (PetscInt)size) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"number of subdomains must be > 0 and < %D, got N = %D",size,N);
10625: *n = 1;
10626: k = ((PetscInt)size)/N + ((PetscInt)size%N>0); /* There are up to k ranks to a color */
10627: color = rank/k;
10628: MPI_Comm_split(comm,color,rank,&subcomm);
10629: PetscMalloc1(1,iss);
10630: MatGetOwnershipRange(A,&rstart,&rend);
10631: ISCreateStride(subcomm,rend-rstart,rstart,1,iss[0]);
10632: MPI_Comm_free(&subcomm);
10633: return(0);
10634: }
10636: /*@
10637: MatGalerkin - Constructs the coarse grid problem via Galerkin projection.
10639: If the interpolation and restriction operators are the same, uses MatPtAP.
10640: If they are not the same, use MatMatMatMult.
10642: Once the coarse grid problem is constructed, correct for interpolation operators
10643: that are not of full rank, which can legitimately happen in the case of non-nested
10644: geometric multigrid.
10646: Input Parameters:
10647: + restrct - restriction operator
10648: . dA - fine grid matrix
10649: . interpolate - interpolation operator
10650: . reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
10651: - fill - expected fill, use PETSC_DEFAULT if you do not have a good estimate
10653: Output Parameters:
10654: . A - the Galerkin coarse matrix
10656: Options Database Key:
10657: . -pc_mg_galerkin <both,pmat,mat,none>
10659: Level: developer
10661: .seealso: MatPtAP(), MatMatMatMult()
10662: @*/
10663: PetscErrorCode MatGalerkin(Mat restrct, Mat dA, Mat interpolate, MatReuse reuse, PetscReal fill, Mat *A)
10664: {
10666: IS zerorows;
10667: Vec diag;
10670: if (reuse == MAT_INPLACE_MATRIX) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Inplace product not supported");
10671: /* Construct the coarse grid matrix */
10672: if (interpolate == restrct) {
10673: MatPtAP(dA,interpolate,reuse,fill,A);
10674: } else {
10675: MatMatMatMult(restrct,dA,interpolate,reuse,fill,A);
10676: }
10678: /* If the interpolation matrix is not of full rank, A will have zero rows.
10679: This can legitimately happen in the case of non-nested geometric multigrid.
10680: In that event, we set the rows of the matrix to the rows of the identity,
10681: ignoring the equations (as the RHS will also be zero). */
10683: MatFindZeroRows(*A, &zerorows);
10685: if (zerorows != NULL) { /* if there are any zero rows */
10686: MatCreateVecs(*A, &diag, NULL);
10687: MatGetDiagonal(*A, diag);
10688: VecISSet(diag, zerorows, 1.0);
10689: MatDiagonalSet(*A, diag, INSERT_VALUES);
10690: VecDestroy(&diag);
10691: ISDestroy(&zerorows);
10692: }
10693: return(0);
10694: }
10696: /*@C
10697: MatSetOperation - Allows user to set a matrix operation for any matrix type
10699: Logically Collective on Mat
10701: Input Parameters:
10702: + mat - the matrix
10703: . op - the name of the operation
10704: - f - the function that provides the operation
10706: Level: developer
10708: Usage:
10709: $ extern PetscErrorCode usermult(Mat,Vec,Vec);
10710: $ MatCreateXXX(comm,...&A);
10711: $ MatSetOperation(A,MATOP_MULT,(void(*)(void))usermult);
10713: Notes:
10714: See the file include/petscmat.h for a complete list of matrix
10715: operations, which all have the form MATOP_<OPERATION>, where
10716: <OPERATION> is the name (in all capital letters) of the
10717: user interface routine (e.g., MatMult() -> MATOP_MULT).
10719: All user-provided functions (except for MATOP_DESTROY) should have the same calling
10720: sequence as the usual matrix interface routines, since they
10721: are intended to be accessed via the usual matrix interface
10722: routines, e.g.,
10723: $ MatMult(Mat,Vec,Vec) -> usermult(Mat,Vec,Vec)
10725: In particular each function MUST return an error code of 0 on success and
10726: nonzero on failure.
10728: This routine is distinct from MatShellSetOperation() in that it can be called on any matrix type.
10730: .seealso: MatGetOperation(), MatCreateShell(), MatShellSetContext(), MatShellSetOperation()
10731: @*/
10732: PetscErrorCode MatSetOperation(Mat mat,MatOperation op,void (*f)(void))
10733: {
10736: if (op == MATOP_VIEW && !mat->ops->viewnative && f != (void (*)(void))(mat->ops->view)) {
10737: mat->ops->viewnative = mat->ops->view;
10738: }
10739: (((void(**)(void))mat->ops)[op]) = f;
10740: return(0);
10741: }
10743: /*@C
10744: MatGetOperation - Gets a matrix operation for any matrix type.
10746: Not Collective
10748: Input Parameters:
10749: + mat - the matrix
10750: - op - the name of the operation
10752: Output Parameter:
10753: . f - the function that provides the operation
10755: Level: developer
10757: Usage:
10758: $ PetscErrorCode (*usermult)(Mat,Vec,Vec);
10759: $ MatGetOperation(A,MATOP_MULT,(void(**)(void))&usermult);
10761: Notes:
10762: See the file include/petscmat.h for a complete list of matrix
10763: operations, which all have the form MATOP_<OPERATION>, where
10764: <OPERATION> is the name (in all capital letters) of the
10765: user interface routine (e.g., MatMult() -> MATOP_MULT).
10767: This routine is distinct from MatShellGetOperation() in that it can be called on any matrix type.
10769: .seealso: MatSetOperation(), MatCreateShell(), MatShellGetContext(), MatShellGetOperation()
10770: @*/
10771: PetscErrorCode MatGetOperation(Mat mat,MatOperation op,void(**f)(void))
10772: {
10775: *f = (((void (**)(void))mat->ops)[op]);
10776: return(0);
10777: }
10779: /*@
10780: MatHasOperation - Determines whether the given matrix supports the particular
10781: operation.
10783: Not Collective
10785: Input Parameters:
10786: + mat - the matrix
10787: - op - the operation, for example, MATOP_GET_DIAGONAL
10789: Output Parameter:
10790: . has - either PETSC_TRUE or PETSC_FALSE
10792: Level: advanced
10794: Notes:
10795: See the file include/petscmat.h for a complete list of matrix
10796: operations, which all have the form MATOP_<OPERATION>, where
10797: <OPERATION> is the name (in all capital letters) of the
10798: user-level routine. E.g., MatNorm() -> MATOP_NORM.
10800: .seealso: MatCreateShell()
10801: @*/
10802: PetscErrorCode MatHasOperation(Mat mat,MatOperation op,PetscBool *has)
10803: {
10808: /* symbolic product can be set before matrix type */
10811: if (mat->ops->hasoperation) {
10812: (*mat->ops->hasoperation)(mat,op,has);
10813: } else {
10814: if (((void**)mat->ops)[op]) *has = PETSC_TRUE;
10815: else {
10816: *has = PETSC_FALSE;
10817: if (op == MATOP_CREATE_SUBMATRIX) {
10818: PetscMPIInt size;
10820: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
10821: if (size == 1) {
10822: MatHasOperation(mat,MATOP_CREATE_SUBMATRICES,has);
10823: }
10824: }
10825: }
10826: }
10827: return(0);
10828: }
10830: /*@
10831: MatHasCongruentLayouts - Determines whether the rows and columns layouts
10832: of the matrix are congruent
10834: Collective on mat
10836: Input Parameters:
10837: . mat - the matrix
10839: Output Parameter:
10840: . cong - either PETSC_TRUE or PETSC_FALSE
10842: Level: beginner
10844: Notes:
10846: .seealso: MatCreate(), MatSetSizes()
10847: @*/
10848: PetscErrorCode MatHasCongruentLayouts(Mat mat,PetscBool *cong)
10849: {
10856: if (!mat->rmap || !mat->cmap) {
10857: *cong = mat->rmap == mat->cmap ? PETSC_TRUE : PETSC_FALSE;
10858: return(0);
10859: }
10860: if (mat->congruentlayouts == PETSC_DECIDE) { /* first time we compare rows and cols layouts */
10861: PetscLayoutCompare(mat->rmap,mat->cmap,cong);
10862: if (*cong) mat->congruentlayouts = 1;
10863: else mat->congruentlayouts = 0;
10864: } else *cong = mat->congruentlayouts ? PETSC_TRUE : PETSC_FALSE;
10865: return(0);
10866: }
10868: PetscErrorCode MatSetInf(Mat A)
10869: {
10873: if (!A->ops->setinf) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"No support for this operation for this matrix type");
10874: (*A->ops->setinf)(A);
10875: return(0);
10876: }