Actual source code: precon.c
1: /*
2: The PC (preconditioner) interface routines, callable by users.
3: */
4: #include <petsc/private/pcimpl.h>
5: #include <petscdm.h>
7: /* Logging support */
8: PetscClassId PC_CLASSID;
9: PetscLogEvent PC_SetUp, PC_SetUpOnBlocks, PC_Apply, PC_MatApply, PC_ApplyCoarse, PC_ApplyMultiple, PC_ApplySymmetricLeft;
10: PetscLogEvent PC_ApplySymmetricRight, PC_ModifySubMatrices, PC_ApplyOnBlocks, PC_ApplyTransposeOnBlocks;
11: PetscInt PetscMGLevelId;
12: PetscLogStage PCMPIStage;
14: PETSC_INTERN PetscErrorCode PCGetDefaultType_Private(PC pc, const char *type[])
15: {
16: PetscMPIInt size;
17: PetscBool hasopblock, hasopsolve, flg1, flg2, set, flg3, isnormal;
19: PetscFunctionBegin;
20: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)pc), &size));
21: if (pc->pmat) {
22: PetscCall(MatHasOperation(pc->pmat, MATOP_GET_DIAGONAL_BLOCK, &hasopblock));
23: PetscCall(MatHasOperation(pc->pmat, MATOP_SOLVE, &hasopsolve));
24: if (size == 1) {
25: PetscCall(MatGetFactorAvailable(pc->pmat, "petsc", MAT_FACTOR_ICC, &flg1));
26: PetscCall(MatGetFactorAvailable(pc->pmat, "petsc", MAT_FACTOR_ILU, &flg2));
27: PetscCall(MatIsSymmetricKnown(pc->pmat, &set, &flg3));
28: PetscCall(PetscObjectTypeCompareAny((PetscObject)pc->pmat, &isnormal, MATNORMAL, MATNORMALHERMITIAN, NULL));
29: if (flg1 && (!flg2 || (set && flg3))) {
30: *type = PCICC;
31: } else if (flg2) {
32: *type = PCILU;
33: } else if (isnormal) {
34: *type = PCNONE;
35: } else if (hasopblock) { /* likely is a parallel matrix run on one processor */
36: *type = PCBJACOBI;
37: } else if (hasopsolve) {
38: *type = PCMAT;
39: } else {
40: *type = PCNONE;
41: }
42: } else {
43: if (hasopblock) {
44: *type = PCBJACOBI;
45: } else if (hasopsolve) {
46: *type = PCMAT;
47: } else {
48: *type = PCNONE;
49: }
50: }
51: } else *type = NULL;
52: PetscFunctionReturn(PETSC_SUCCESS);
53: }
55: /*@
56: PCReset - Resets a `PC` context to the pcsetupcalled = 0 state and removes any allocated `Vec`s and `Mat`s
58: Collective
60: Input Parameter:
61: . pc - the preconditioner context
63: Level: developer
65: Note:
66: This allows a `PC` to be reused for a different sized linear system but using the same options that have been previously set in `pc`
68: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCSetUp()`
69: @*/
70: PetscErrorCode PCReset(PC pc)
71: {
72: PetscFunctionBegin;
74: PetscTryTypeMethod(pc, reset);
75: PetscCall(VecDestroy(&pc->diagonalscaleright));
76: PetscCall(VecDestroy(&pc->diagonalscaleleft));
77: PetscCall(MatDestroy(&pc->pmat));
78: PetscCall(MatDestroy(&pc->mat));
80: pc->setupcalled = 0;
81: PetscFunctionReturn(PETSC_SUCCESS);
82: }
84: /*@
85: PCDestroy - Destroys `PC` context that was created with `PCCreate()`.
87: Collective
89: Input Parameter:
90: . pc - the preconditioner context
92: Level: developer
94: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCSetUp()`
95: @*/
96: PetscErrorCode PCDestroy(PC *pc)
97: {
98: PetscFunctionBegin;
99: if (!*pc) PetscFunctionReturn(PETSC_SUCCESS);
101: if (--((PetscObject)*pc)->refct > 0) {
102: *pc = NULL;
103: PetscFunctionReturn(PETSC_SUCCESS);
104: }
106: PetscCall(PCReset(*pc));
108: /* if memory was published with SAWs then destroy it */
109: PetscCall(PetscObjectSAWsViewOff((PetscObject)*pc));
110: PetscTryTypeMethod(*pc, destroy);
111: PetscCall(DMDestroy(&(*pc)->dm));
112: PetscCall(PetscHeaderDestroy(pc));
113: PetscFunctionReturn(PETSC_SUCCESS);
114: }
116: /*@
117: PCGetDiagonalScale - Indicates if the preconditioner applies an additional left and right
118: scaling as needed by certain time-stepping codes.
120: Logically Collective
122: Input Parameter:
123: . pc - the preconditioner context
125: Output Parameter:
126: . flag - `PETSC_TRUE` if it applies the scaling
128: Level: developer
130: Note:
131: If this returns `PETSC_TRUE` then the system solved via the Krylov method is, for left and right preconditioning,
133: $$
134: \begin{align*}
135: D M A D^{-1} y = D M b \\
136: D A M D^{-1} z = D b.
137: \end{align*}
138: $$
140: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCSetUp()`, `PCDiagonalScaleLeft()`, `PCDiagonalScaleRight()`, `PCSetDiagonalScale()`
141: @*/
142: PetscErrorCode PCGetDiagonalScale(PC pc, PetscBool *flag)
143: {
144: PetscFunctionBegin;
146: PetscAssertPointer(flag, 2);
147: *flag = pc->diagonalscale;
148: PetscFunctionReturn(PETSC_SUCCESS);
149: }
151: /*@
152: PCSetDiagonalScale - Indicates the left scaling to use to apply an additional left and right
153: scaling as needed by certain time-stepping codes.
155: Logically Collective
157: Input Parameters:
158: + pc - the preconditioner context
159: - s - scaling vector
161: Level: intermediate
163: Notes:
164: The system solved via the Krylov method is, for left and right preconditioning,
165: $$
166: \begin{align*}
167: D M A D^{-1} y = D M b \\
168: D A M D^{-1} z = D b.
169: \end{align*}
170: $$
172: `PCDiagonalScaleLeft()` scales a vector by $D$. `PCDiagonalScaleRight()` scales a vector by $D^{-1}$.
174: .seealso: [](ch_ksp), `PCCreate()`, `PCSetUp()`, `PCDiagonalScaleLeft()`, `PCDiagonalScaleRight()`, `PCGetDiagonalScale()`
175: @*/
176: PetscErrorCode PCSetDiagonalScale(PC pc, Vec s)
177: {
178: PetscFunctionBegin;
181: pc->diagonalscale = PETSC_TRUE;
183: PetscCall(PetscObjectReference((PetscObject)s));
184: PetscCall(VecDestroy(&pc->diagonalscaleleft));
186: pc->diagonalscaleleft = s;
188: PetscCall(VecDuplicate(s, &pc->diagonalscaleright));
189: PetscCall(VecCopy(s, pc->diagonalscaleright));
190: PetscCall(VecReciprocal(pc->diagonalscaleright));
191: PetscFunctionReturn(PETSC_SUCCESS);
192: }
194: /*@
195: PCDiagonalScaleLeft - Scales a vector by the left scaling as needed by certain time-stepping codes.
197: Logically Collective
199: Input Parameters:
200: + pc - the preconditioner context
201: . in - input vector
202: - out - scaled vector (maybe the same as in)
204: Level: intermediate
206: Notes:
207: The system solved via the Krylov method is, for left and right preconditioning,
209: $$
210: \begin{align*}
211: D M A D^{-1} y = D M b \\
212: D A M D^{-1} z = D b.
213: \end{align*}
214: $$
216: `PCDiagonalScaleLeft()` scales a vector by $D$. `PCDiagonalScaleRight()` scales a vector by $D^{-1}$.
218: If diagonal scaling is turned off and `in` is not `out` then `in` is copied to `out`
220: .seealso: [](ch_ksp), `PCCreate()`, `PCSetUp()`, `PCSetDiagonalScale()`, `PCDiagonalScaleRight()`, `MatDiagonalScale()`
221: @*/
222: PetscErrorCode PCDiagonalScaleLeft(PC pc, Vec in, Vec out)
223: {
224: PetscFunctionBegin;
228: if (pc->diagonalscale) {
229: PetscCall(VecPointwiseMult(out, pc->diagonalscaleleft, in));
230: } else if (in != out) {
231: PetscCall(VecCopy(in, out));
232: }
233: PetscFunctionReturn(PETSC_SUCCESS);
234: }
236: /*@
237: PCDiagonalScaleRight - Scales a vector by the right scaling as needed by certain time-stepping codes.
239: Logically Collective
241: Input Parameters:
242: + pc - the preconditioner context
243: . in - input vector
244: - out - scaled vector (maybe the same as in)
246: Level: intermediate
248: Notes:
249: The system solved via the Krylov method is, for left and right preconditioning,
251: $$
252: \begin{align*}
253: D M A D^{-1} y = D M b \\
254: D A M D^{-1} z = D b.
255: \end{align*}
256: $$
258: `PCDiagonalScaleLeft()` scales a vector by $D$. `PCDiagonalScaleRight()` scales a vector by $D^{-1}$.
260: If diagonal scaling is turned off and `in` is not `out` then `in` is copied to `out`
262: .seealso: [](ch_ksp), `PCCreate()`, `PCSetUp()`, `PCDiagonalScaleLeft()`, `PCSetDiagonalScale()`, `MatDiagonalScale()`
263: @*/
264: PetscErrorCode PCDiagonalScaleRight(PC pc, Vec in, Vec out)
265: {
266: PetscFunctionBegin;
270: if (pc->diagonalscale) {
271: PetscCall(VecPointwiseMult(out, pc->diagonalscaleright, in));
272: } else if (in != out) {
273: PetscCall(VecCopy(in, out));
274: }
275: PetscFunctionReturn(PETSC_SUCCESS);
276: }
278: /*@
279: PCSetUseAmat - Sets a flag to indicate that when the preconditioner needs to apply (part of) the
280: operator during the preconditioning process it applies the Amat provided to `TSSetRHSJacobian()`,
281: `TSSetIJacobian()`, `SNESSetJacobian()`, `KSPSetOperators()` or `PCSetOperators()` not the Pmat.
283: Logically Collective
285: Input Parameters:
286: + pc - the preconditioner context
287: - flg - `PETSC_TRUE` to use the Amat, `PETSC_FALSE` to use the Pmat (default is false)
289: Options Database Key:
290: . -pc_use_amat <true,false> - use the amat argument to `KSPSetOperators()` or `PCSetOperators()` to apply the operator
292: Level: intermediate
294: Note:
295: For the common case in which the linear system matrix and the matrix used to construct the
296: preconditioner are identical, this routine has no affect.
298: .seealso: [](ch_ksp), `PC`, `PCGetUseAmat()`, `PCBJACOBI`, `PCMG`, `PCFIELDSPLIT`, `PCCOMPOSITE`,
299: `KSPSetOperators()`, `PCSetOperators()`
300: @*/
301: PetscErrorCode PCSetUseAmat(PC pc, PetscBool flg)
302: {
303: PetscFunctionBegin;
305: pc->useAmat = flg;
306: PetscFunctionReturn(PETSC_SUCCESS);
307: }
309: /*@
310: PCSetErrorIfFailure - Causes `PC` to generate an error if a floating point exception, for example a zero pivot, is detected.
312: Logically Collective
314: Input Parameters:
315: + pc - iterative context obtained from `PCCreate()`
316: - flg - `PETSC_TRUE` indicates you want the error generated
318: Level: advanced
320: Notes:
321: Normally PETSc continues if a linear solver fails due to a failed setup of a preconditioner, you can call `KSPGetConvergedReason()` after a `KSPSolve()`
322: to determine if it has converged or failed. Or use -ksp_error_if_not_converged to cause the program to terminate as soon as lack of convergence is
323: detected.
325: This is propagated into `KSP`s used by this `PC`, which then propagate it into `PC`s used by those `KSP`s
327: .seealso: [](ch_ksp), `PC`, `KSPSetErrorIfNotConverged()`, `PCGetInitialGuessNonzero()`, `PCSetInitialGuessKnoll()`, `PCGetInitialGuessKnoll()`
328: @*/
329: PetscErrorCode PCSetErrorIfFailure(PC pc, PetscBool flg)
330: {
331: PetscFunctionBegin;
334: pc->erroriffailure = flg;
335: PetscFunctionReturn(PETSC_SUCCESS);
336: }
338: /*@
339: PCGetUseAmat - Gets a flag to indicate that when the preconditioner needs to apply (part of) the
340: operator during the preconditioning process it applies the Amat provided to `TSSetRHSJacobian()`,
341: `TSSetIJacobian()`, `SNESSetJacobian()`, `KSPSetOperators()` or `PCSetOperators()` not the Pmat.
343: Logically Collective
345: Input Parameter:
346: . pc - the preconditioner context
348: Output Parameter:
349: . flg - `PETSC_TRUE` to use the Amat, `PETSC_FALSE` to use the Pmat (default is false)
351: Level: intermediate
353: Note:
354: For the common case in which the linear system matrix and the matrix used to construct the
355: preconditioner are identical, this routine is does nothing.
357: .seealso: [](ch_ksp), `PC`, `PCSetUseAmat()`, `PCBJACOBI`, `PCMG`, `PCFIELDSPLIT`, `PCCOMPOSITE`
358: @*/
359: PetscErrorCode PCGetUseAmat(PC pc, PetscBool *flg)
360: {
361: PetscFunctionBegin;
363: *flg = pc->useAmat;
364: PetscFunctionReturn(PETSC_SUCCESS);
365: }
367: /*@
368: PCSetKSPNestLevel - sets the amount of nesting the `KSP` that contains this `PC` has
370: Collective
372: Input Parameters:
373: + pc - the `PC`
374: - level - the nest level
376: Level: developer
378: .seealso: [](ch_ksp), `KSPSetUp()`, `KSPSolve()`, `KSPDestroy()`, `KSP`, `KSPGMRES`, `KSPType`, `KSPGetNestLevel()`, `PCGetKSPNestLevel()`, `KSPSetNestLevel()`
379: @*/
380: PetscErrorCode PCSetKSPNestLevel(PC pc, PetscInt level)
381: {
382: PetscFunctionBegin;
385: pc->kspnestlevel = level;
386: PetscFunctionReturn(PETSC_SUCCESS);
387: }
389: /*@
390: PCGetKSPNestLevel - gets the amount of nesting the `KSP` that contains this `PC` has
392: Not Collective
394: Input Parameter:
395: . pc - the `PC`
397: Output Parameter:
398: . level - the nest level
400: Level: developer
402: .seealso: [](ch_ksp), `KSPSetUp()`, `KSPSolve()`, `KSPDestroy()`, `KSP`, `KSPGMRES`, `KSPType`, `KSPSetNestLevel()`, `PCSetKSPNestLevel()`, `KSPGetNestLevel()`
403: @*/
404: PetscErrorCode PCGetKSPNestLevel(PC pc, PetscInt *level)
405: {
406: PetscFunctionBegin;
408: PetscAssertPointer(level, 2);
409: *level = pc->kspnestlevel;
410: PetscFunctionReturn(PETSC_SUCCESS);
411: }
413: /*@
414: PCCreate - Creates a preconditioner context, `PC`
416: Collective
418: Input Parameter:
419: . comm - MPI communicator
421: Output Parameter:
422: . newpc - location to put the preconditioner context
424: Level: developer
426: Note:
427: The default preconditioner for sparse matrices is `PCILU` or `PCICC` with 0 fill on one process and block Jacobi (`PCBJACOBI`) with `PCILU` or `PCICC`
428: in parallel. For dense matrices it is always `PCNONE`.
430: .seealso: [](ch_ksp), `PC`, `PCSetUp()`, `PCApply()`, `PCDestroy()`
431: @*/
432: PetscErrorCode PCCreate(MPI_Comm comm, PC *newpc)
433: {
434: PC pc;
436: PetscFunctionBegin;
437: PetscAssertPointer(newpc, 2);
438: PetscCall(PCInitializePackage());
440: PetscCall(PetscHeaderCreate(pc, PC_CLASSID, "PC", "Preconditioner", "PC", comm, PCDestroy, PCView));
441: pc->mat = NULL;
442: pc->pmat = NULL;
443: pc->setupcalled = 0;
444: pc->setfromoptionscalled = 0;
445: pc->data = NULL;
446: pc->diagonalscale = PETSC_FALSE;
447: pc->diagonalscaleleft = NULL;
448: pc->diagonalscaleright = NULL;
450: pc->modifysubmatrices = NULL;
451: pc->modifysubmatricesP = NULL;
453: *newpc = pc;
454: PetscFunctionReturn(PETSC_SUCCESS);
455: }
457: /*@
458: PCApply - Applies the preconditioner to a vector.
460: Collective
462: Input Parameters:
463: + pc - the preconditioner context
464: - x - input vector
466: Output Parameter:
467: . y - output vector
469: Level: developer
471: .seealso: [](ch_ksp), `PC`, `PCApplyTranspose()`, `PCApplyBAorAB()`
472: @*/
473: PetscErrorCode PCApply(PC pc, Vec x, Vec y)
474: {
475: PetscInt m, n, mv, nv;
477: PetscFunctionBegin;
481: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
482: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
483: /* use pmat to check vector sizes since for KSPLSQR the pmat may be of a different size than mat */
484: PetscCall(MatGetLocalSize(pc->pmat, &m, &n));
485: PetscCall(VecGetLocalSize(x, &mv));
486: PetscCall(VecGetLocalSize(y, &nv));
487: /* check pmat * y = x is feasible */
488: PetscCheck(mv == m, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Preconditioner number of local rows %" PetscInt_FMT " does not equal input vector size %" PetscInt_FMT, m, mv);
489: PetscCheck(nv == n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Preconditioner number of local columns %" PetscInt_FMT " does not equal output vector size %" PetscInt_FMT, n, nv);
490: PetscCall(VecSetErrorIfLocked(y, 3));
492: PetscCall(PCSetUp(pc));
493: PetscCall(VecLockReadPush(x));
494: PetscCall(PetscLogEventBegin(PC_Apply, pc, x, y, 0));
495: PetscUseTypeMethod(pc, apply, x, y);
496: PetscCall(PetscLogEventEnd(PC_Apply, pc, x, y, 0));
497: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
498: PetscCall(VecLockReadPop(x));
499: PetscFunctionReturn(PETSC_SUCCESS);
500: }
502: /*@
503: PCMatApply - Applies the preconditioner to multiple vectors stored as a `MATDENSE`. Like `PCApply()`, `Y` and `X` must be different matrices.
505: Collective
507: Input Parameters:
508: + pc - the preconditioner context
509: - X - block of input vectors
511: Output Parameter:
512: . Y - block of output vectors
514: Level: developer
516: .seealso: [](ch_ksp), `PC`, `PCApply()`, `KSPMatSolve()`
517: @*/
518: PetscErrorCode PCMatApply(PC pc, Mat X, Mat Y)
519: {
520: Mat A;
521: Vec cy, cx;
522: PetscInt m1, M1, m2, M2, n1, N1, n2, N2, m3, M3, n3, N3;
523: PetscBool match;
525: PetscFunctionBegin;
529: PetscCheckSameComm(pc, 1, X, 2);
530: PetscCheckSameComm(pc, 1, Y, 3);
531: PetscCheck(Y != X, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "Y and X must be different matrices");
532: PetscCall(PCGetOperators(pc, NULL, &A));
533: PetscCall(MatGetLocalSize(A, &m3, &n3));
534: PetscCall(MatGetLocalSize(X, &m2, &n2));
535: PetscCall(MatGetLocalSize(Y, &m1, &n1));
536: PetscCall(MatGetSize(A, &M3, &N3));
537: PetscCall(MatGetSize(X, &M2, &N2));
538: PetscCall(MatGetSize(Y, &M1, &N1));
539: PetscCheck(n1 == n2 && N1 == N2, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible number of columns between block of input vectors (n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ") and block of output vectors (n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ")", n2, N2, n1, N1);
540: PetscCheck(m2 == m3 && M2 == M3, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible layout between block of input vectors (m,M) = (%" PetscInt_FMT ",%" PetscInt_FMT ") and Pmat (m,M)x(n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ")x(%" PetscInt_FMT ",%" PetscInt_FMT ")", m2, M2, m3, M3, n3, N3);
541: PetscCheck(m1 == n3 && M1 == N3, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible layout between block of output vectors (m,M) = (%" PetscInt_FMT ",%" PetscInt_FMT ") and Pmat (m,M)x(n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ")x(%" PetscInt_FMT ",%" PetscInt_FMT ")", m1, M1, m3, M3, n3, N3);
542: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)Y, &match, MATSEQDENSE, MATMPIDENSE, ""));
543: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of output vectors not stored in a dense Mat");
544: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)X, &match, MATSEQDENSE, MATMPIDENSE, ""));
545: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of input vectors not stored in a dense Mat");
546: PetscCall(PCSetUp(pc));
547: if (pc->ops->matapply) {
548: PetscCall(PetscLogEventBegin(PC_MatApply, pc, X, Y, 0));
549: PetscUseTypeMethod(pc, matapply, X, Y);
550: PetscCall(PetscLogEventEnd(PC_MatApply, pc, X, Y, 0));
551: } else {
552: PetscCall(PetscInfo(pc, "PC type %s applying column by column\n", ((PetscObject)pc)->type_name));
553: for (n1 = 0; n1 < N1; ++n1) {
554: PetscCall(MatDenseGetColumnVecRead(X, n1, &cx));
555: PetscCall(MatDenseGetColumnVecWrite(Y, n1, &cy));
556: PetscCall(PCApply(pc, cx, cy));
557: PetscCall(MatDenseRestoreColumnVecWrite(Y, n1, &cy));
558: PetscCall(MatDenseRestoreColumnVecRead(X, n1, &cx));
559: }
560: }
561: PetscFunctionReturn(PETSC_SUCCESS);
562: }
564: /*@
565: PCApplySymmetricLeft - Applies the left part of a symmetric preconditioner to a vector.
567: Collective
569: Input Parameters:
570: + pc - the preconditioner context
571: - x - input vector
573: Output Parameter:
574: . y - output vector
576: Level: developer
578: Note:
579: Currently, this routine is implemented only for `PCICC` and `PCJACOBI` preconditioners.
581: .seealso: [](ch_ksp), `PC`, `PCApply()`, `PCApplySymmetricRight()`
582: @*/
583: PetscErrorCode PCApplySymmetricLeft(PC pc, Vec x, Vec y)
584: {
585: PetscFunctionBegin;
589: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
590: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
591: PetscCall(PCSetUp(pc));
592: PetscCall(VecLockReadPush(x));
593: PetscCall(PetscLogEventBegin(PC_ApplySymmetricLeft, pc, x, y, 0));
594: PetscUseTypeMethod(pc, applysymmetricleft, x, y);
595: PetscCall(PetscLogEventEnd(PC_ApplySymmetricLeft, pc, x, y, 0));
596: PetscCall(VecLockReadPop(x));
597: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
598: PetscFunctionReturn(PETSC_SUCCESS);
599: }
601: /*@
602: PCApplySymmetricRight - Applies the right part of a symmetric preconditioner to a vector.
604: Collective
606: Input Parameters:
607: + pc - the preconditioner context
608: - x - input vector
610: Output Parameter:
611: . y - output vector
613: Level: developer
615: Note:
616: Currently, this routine is implemented only for `PCICC` and `PCJACOBI` preconditioners.
618: .seealso: [](ch_ksp), `PC`, `PCApply()`, `PCApplySymmetricLeft()`
619: @*/
620: PetscErrorCode PCApplySymmetricRight(PC pc, Vec x, Vec y)
621: {
622: PetscFunctionBegin;
626: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
627: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
628: PetscCall(PCSetUp(pc));
629: PetscCall(VecLockReadPush(x));
630: PetscCall(PetscLogEventBegin(PC_ApplySymmetricRight, pc, x, y, 0));
631: PetscUseTypeMethod(pc, applysymmetricright, x, y);
632: PetscCall(PetscLogEventEnd(PC_ApplySymmetricRight, pc, x, y, 0));
633: PetscCall(VecLockReadPop(x));
634: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
635: PetscFunctionReturn(PETSC_SUCCESS);
636: }
638: /*@
639: PCApplyTranspose - Applies the transpose of preconditioner to a vector.
641: Collective
643: Input Parameters:
644: + pc - the preconditioner context
645: - x - input vector
647: Output Parameter:
648: . y - output vector
650: Level: developer
652: Note:
653: For complex numbers this applies the non-Hermitian transpose.
655: Developer Note:
656: We need to implement a `PCApplyHermitianTranspose()`
658: .seealso: [](ch_ksp), `PC`, `PCApply()`, `PCApplyBAorAB()`, `PCApplyBAorABTranspose()`, `PCApplyTransposeExists()`
659: @*/
660: PetscErrorCode PCApplyTranspose(PC pc, Vec x, Vec y)
661: {
662: PetscFunctionBegin;
666: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
667: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 2, PETSC_TRUE));
668: PetscCall(PCSetUp(pc));
669: PetscCall(VecLockReadPush(x));
670: PetscCall(PetscLogEventBegin(PC_Apply, pc, x, y, 0));
671: PetscUseTypeMethod(pc, applytranspose, x, y);
672: PetscCall(PetscLogEventEnd(PC_Apply, pc, x, y, 0));
673: PetscCall(VecLockReadPop(x));
674: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 3, PETSC_FALSE));
675: PetscFunctionReturn(PETSC_SUCCESS);
676: }
678: /*@
679: PCApplyTransposeExists - Test whether the preconditioner has a transpose apply operation
681: Collective
683: Input Parameter:
684: . pc - the preconditioner context
686: Output Parameter:
687: . flg - `PETSC_TRUE` if a transpose operation is defined
689: Level: developer
691: .seealso: [](ch_ksp), `PC`, `PCApplyTranspose()`
692: @*/
693: PetscErrorCode PCApplyTransposeExists(PC pc, PetscBool *flg)
694: {
695: PetscFunctionBegin;
697: PetscAssertPointer(flg, 2);
698: if (pc->ops->applytranspose) *flg = PETSC_TRUE;
699: else *flg = PETSC_FALSE;
700: PetscFunctionReturn(PETSC_SUCCESS);
701: }
703: /*@
704: PCApplyBAorAB - Applies the preconditioner and operator to a vector. $y = B*A*x $ or $ y = A*B*x$.
706: Collective
708: Input Parameters:
709: + pc - the preconditioner context
710: . side - indicates the preconditioner side, one of `PC_LEFT`, `PC_RIGHT`, or `PC_SYMMETRIC`
711: . x - input vector
712: - work - work vector
714: Output Parameter:
715: . y - output vector
717: Level: developer
719: Note:
720: If the `PC` has had `PCSetDiagonalScale()` set then $ D M A D^{-1} $ for left preconditioning or $ D A M D^{-1} $ is actually applied.
721: The specific `KSPSolve()` method must also be written to handle the post-solve "correction" for the diagonal scaling.
723: .seealso: [](ch_ksp), `PC`, `PCApply()`, `PCApplyTranspose()`, `PCApplyBAorABTranspose()`
724: @*/
725: PetscErrorCode PCApplyBAorAB(PC pc, PCSide side, Vec x, Vec y, Vec work)
726: {
727: PetscFunctionBegin;
733: PetscCheckSameComm(pc, 1, x, 3);
734: PetscCheckSameComm(pc, 1, y, 4);
735: PetscCheckSameComm(pc, 1, work, 5);
736: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
737: PetscCheck(side == PC_LEFT || side == PC_SYMMETRIC || side == PC_RIGHT, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_OUTOFRANGE, "Side must be right, left, or symmetric");
738: PetscCheck(!pc->diagonalscale || side != PC_SYMMETRIC, PetscObjectComm((PetscObject)pc), PETSC_ERR_SUP, "Cannot include diagonal scaling with symmetric preconditioner application");
739: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 3, PETSC_TRUE));
741: PetscCall(PCSetUp(pc));
742: if (pc->diagonalscale) {
743: if (pc->ops->applyBA) {
744: Vec work2; /* this is expensive, but to fix requires a second work vector argument to PCApplyBAorAB() */
745: PetscCall(VecDuplicate(x, &work2));
746: PetscCall(PCDiagonalScaleRight(pc, x, work2));
747: PetscUseTypeMethod(pc, applyBA, side, work2, y, work);
748: PetscCall(PCDiagonalScaleLeft(pc, y, y));
749: PetscCall(VecDestroy(&work2));
750: } else if (side == PC_RIGHT) {
751: PetscCall(PCDiagonalScaleRight(pc, x, y));
752: PetscCall(PCApply(pc, y, work));
753: PetscCall(MatMult(pc->mat, work, y));
754: PetscCall(PCDiagonalScaleLeft(pc, y, y));
755: } else if (side == PC_LEFT) {
756: PetscCall(PCDiagonalScaleRight(pc, x, y));
757: PetscCall(MatMult(pc->mat, y, work));
758: PetscCall(PCApply(pc, work, y));
759: PetscCall(PCDiagonalScaleLeft(pc, y, y));
760: } else PetscCheck(side != PC_SYMMETRIC, PetscObjectComm((PetscObject)pc), PETSC_ERR_SUP, "Cannot provide diagonal scaling with symmetric application of preconditioner");
761: } else {
762: if (pc->ops->applyBA) {
763: PetscUseTypeMethod(pc, applyBA, side, x, y, work);
764: } else if (side == PC_RIGHT) {
765: PetscCall(PCApply(pc, x, work));
766: PetscCall(MatMult(pc->mat, work, y));
767: } else if (side == PC_LEFT) {
768: PetscCall(MatMult(pc->mat, x, work));
769: PetscCall(PCApply(pc, work, y));
770: } else if (side == PC_SYMMETRIC) {
771: /* There's an extra copy here; maybe should provide 2 work vectors instead? */
772: PetscCall(PCApplySymmetricRight(pc, x, work));
773: PetscCall(MatMult(pc->mat, work, y));
774: PetscCall(VecCopy(y, work));
775: PetscCall(PCApplySymmetricLeft(pc, work, y));
776: }
777: }
778: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 4, PETSC_FALSE));
779: PetscFunctionReturn(PETSC_SUCCESS);
780: }
782: /*@
783: PCApplyBAorABTranspose - Applies the transpose of the preconditioner
784: and operator to a vector. That is, applies $B^T * A^T$ with left preconditioning,
785: NOT $(B*A)^T = A^T*B^T$.
787: Collective
789: Input Parameters:
790: + pc - the preconditioner context
791: . side - indicates the preconditioner side, one of `PC_LEFT`, `PC_RIGHT`, or `PC_SYMMETRIC`
792: . x - input vector
793: - work - work vector
795: Output Parameter:
796: . y - output vector
798: Level: developer
800: Note:
801: This routine is used internally so that the same Krylov code can be used to solve $A x = b$ and $A^T x = b$, with a preconditioner
802: defined by $B^T$. This is why this has the funny form that it computes $B^T * A^T$
804: .seealso: [](ch_ksp), `PC`, `PCApply()`, `PCApplyTranspose()`, `PCApplyBAorAB()`
805: @*/
806: PetscErrorCode PCApplyBAorABTranspose(PC pc, PCSide side, Vec x, Vec y, Vec work)
807: {
808: PetscFunctionBegin;
813: PetscCheck(x != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "x and y must be different vectors");
814: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(x, 3, PETSC_TRUE));
815: if (pc->ops->applyBAtranspose) {
816: PetscUseTypeMethod(pc, applyBAtranspose, side, x, y, work);
817: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 4, PETSC_FALSE));
818: PetscFunctionReturn(PETSC_SUCCESS);
819: }
820: PetscCheck(side == PC_LEFT || side == PC_RIGHT, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_OUTOFRANGE, "Side must be right or left");
822: PetscCall(PCSetUp(pc));
823: if (side == PC_RIGHT) {
824: PetscCall(PCApplyTranspose(pc, x, work));
825: PetscCall(MatMultTranspose(pc->mat, work, y));
826: } else if (side == PC_LEFT) {
827: PetscCall(MatMultTranspose(pc->mat, x, work));
828: PetscCall(PCApplyTranspose(pc, work, y));
829: }
830: /* add support for PC_SYMMETRIC */
831: if (pc->erroriffailure) PetscCall(VecValidValues_Internal(y, 4, PETSC_FALSE));
832: PetscFunctionReturn(PETSC_SUCCESS);
833: }
835: /*@
836: PCApplyRichardsonExists - Determines whether a particular preconditioner has a
837: built-in fast application of Richardson's method.
839: Not Collective
841: Input Parameter:
842: . pc - the preconditioner
844: Output Parameter:
845: . exists - `PETSC_TRUE` or `PETSC_FALSE`
847: Level: developer
849: .seealso: [](ch_ksp), `PC`, `KSPRICHARDSON`, `PCApplyRichardson()`
850: @*/
851: PetscErrorCode PCApplyRichardsonExists(PC pc, PetscBool *exists)
852: {
853: PetscFunctionBegin;
855: PetscAssertPointer(exists, 2);
856: if (pc->ops->applyrichardson) *exists = PETSC_TRUE;
857: else *exists = PETSC_FALSE;
858: PetscFunctionReturn(PETSC_SUCCESS);
859: }
861: /*@
862: PCApplyRichardson - Applies several steps of Richardson iteration with
863: the particular preconditioner. This routine is usually used by the
864: Krylov solvers and not the application code directly.
866: Collective
868: Input Parameters:
869: + pc - the preconditioner context
870: . b - the right-hand side
871: . w - one work vector
872: . rtol - relative decrease in residual norm convergence criteria
873: . abstol - absolute residual norm convergence criteria
874: . dtol - divergence residual norm increase criteria
875: . its - the number of iterations to apply.
876: - guesszero - if the input x contains nonzero initial guess
878: Output Parameters:
879: + outits - number of iterations actually used (for SOR this always equals its)
880: . reason - the reason the apply terminated
881: - y - the solution (also contains initial guess if guesszero is `PETSC_FALSE`
883: Level: developer
885: Notes:
886: Most preconditioners do not support this function. Use the command
887: `PCApplyRichardsonExists()` to determine if one does.
889: Except for the `PCMG` this routine ignores the convergence tolerances
890: and always runs for the number of iterations
892: .seealso: [](ch_ksp), `PC`, `PCApplyRichardsonExists()`
893: @*/
894: PetscErrorCode PCApplyRichardson(PC pc, Vec b, Vec y, Vec w, PetscReal rtol, PetscReal abstol, PetscReal dtol, PetscInt its, PetscBool guesszero, PetscInt *outits, PCRichardsonConvergedReason *reason)
895: {
896: PetscFunctionBegin;
901: PetscCheck(b != y, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_IDN, "b and y must be different vectors");
902: PetscCall(PCSetUp(pc));
903: PetscUseTypeMethod(pc, applyrichardson, b, y, w, rtol, abstol, dtol, its, guesszero, outits, reason);
904: PetscFunctionReturn(PETSC_SUCCESS);
905: }
907: /*@
908: PCSetFailedReason - Sets the reason a `PCSetUp()` failed or `PC_NOERROR` if it did not fail
910: Logically Collective
912: Input Parameters:
913: + pc - the preconditioner context
914: - reason - the reason it failedx
916: Level: advanced
918: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCApply()`, `PCDestroy()`, `PCFailedReason`
919: @*/
920: PetscErrorCode PCSetFailedReason(PC pc, PCFailedReason reason)
921: {
922: PetscFunctionBegin;
924: pc->failedreason = reason;
925: PetscFunctionReturn(PETSC_SUCCESS);
926: }
928: /*@
929: PCGetFailedReason - Gets the reason a `PCSetUp()` failed or `PC_NOERROR` if it did not fail
931: Logically Collective
933: Input Parameter:
934: . pc - the preconditioner context
936: Output Parameter:
937: . reason - the reason it failed
939: Level: advanced
941: Note:
942: This is the maximum over reason over all ranks in the PC communicator. It is only valid after
943: a call `KSPCheckDot()` or `KSPCheckNorm()` inside a `KSPSolve()` or `PCReduceFailedReason()`.
944: It is not valid immediately after a `PCSetUp()` or `PCApply()`, then use `PCGetFailedReasonRank()`
946: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCApply()`, `PCDestroy()`, `PCGetFailedReasonRank()`, `PCSetFailedReason()`
947: @*/
948: PetscErrorCode PCGetFailedReason(PC pc, PCFailedReason *reason)
949: {
950: PetscFunctionBegin;
952: if (pc->setupcalled < 0) *reason = (PCFailedReason)pc->setupcalled;
953: else *reason = pc->failedreason;
954: PetscFunctionReturn(PETSC_SUCCESS);
955: }
957: /*@
958: PCGetFailedReasonRank - Gets the reason a `PCSetUp()` failed or `PC_NOERROR` if it did not fail on this MPI rank
960: Not Collective
962: Input Parameter:
963: . pc - the preconditioner context
965: Output Parameter:
966: . reason - the reason it failed
968: Level: advanced
970: Note:
971: Different processes may have different reasons or no reason, see `PCGetFailedReason()`
973: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCApply()`, `PCDestroy()`, `PCGetFailedReason()`, `PCSetFailedReason()`, `PCReduceFailedReason()`
974: @*/
975: PetscErrorCode PCGetFailedReasonRank(PC pc, PCFailedReason *reason)
976: {
977: PetscFunctionBegin;
979: if (pc->setupcalled < 0) *reason = (PCFailedReason)pc->setupcalled;
980: else *reason = pc->failedreason;
981: PetscFunctionReturn(PETSC_SUCCESS);
982: }
984: /*@
985: PCReduceFailedReason - Reduce the failed reason among the MPI processes that share the `PC`
987: Collective
989: Input Parameter:
990: . pc - the preconditioner context
992: Level: advanced
994: Note:
995: Different MPI processes may have different reasons or no reason, see `PCGetFailedReason()`. This routine
996: makes them have a common value (failure if any MPI process had a failure).
998: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCApply()`, `PCDestroy()`, `PCGetFailedReason()`, `PCSetFailedReason()`
999: @*/
1000: PetscErrorCode PCReduceFailedReason(PC pc)
1001: {
1002: PetscInt buf;
1004: PetscFunctionBegin;
1006: buf = (PetscInt)pc->failedreason;
1007: PetscCall(MPIU_Allreduce(MPI_IN_PLACE, &buf, 1, MPIU_INT, MPI_MAX, PetscObjectComm((PetscObject)pc)));
1008: pc->failedreason = (PCFailedReason)buf;
1009: PetscFunctionReturn(PETSC_SUCCESS);
1010: }
1012: /* Next line needed to deactivate KSP_Solve logging */
1013: #include <petsc/private/kspimpl.h>
1015: /*
1016: a setupcall of 0 indicates never setup,
1017: 1 indicates has been previously setup
1018: -1 indicates a PCSetUp() was attempted and failed
1019: */
1020: /*@
1021: PCSetUp - Prepares for the use of a preconditioner.
1023: Collective
1025: Input Parameter:
1026: . pc - the preconditioner context
1028: Level: developer
1030: .seealso: [](ch_ksp), `PC`, `PCCreate()`, `PCApply()`, `PCDestroy()`
1031: @*/
1032: PetscErrorCode PCSetUp(PC pc)
1033: {
1034: const char *def;
1035: PetscObjectState matstate, matnonzerostate;
1037: PetscFunctionBegin;
1039: PetscCheck(pc->mat, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_WRONGSTATE, "Matrix must be set first");
1041: if (pc->setupcalled && pc->reusepreconditioner) {
1042: PetscCall(PetscInfo(pc, "Leaving PC with identical preconditioner since reuse preconditioner is set\n"));
1043: PetscFunctionReturn(PETSC_SUCCESS);
1044: }
1046: PetscCall(PetscObjectStateGet((PetscObject)pc->pmat, &matstate));
1047: PetscCall(MatGetNonzeroState(pc->pmat, &matnonzerostate));
1048: if (!pc->setupcalled) {
1049: //PetscCall(PetscInfo(pc, "Setting up PC for first time\n"));
1050: pc->flag = DIFFERENT_NONZERO_PATTERN;
1051: } else if (matstate == pc->matstate) PetscFunctionReturn(PETSC_SUCCESS);
1052: else {
1053: if (matnonzerostate != pc->matnonzerostate) {
1054: PetscCall(PetscInfo(pc, "Setting up PC with different nonzero pattern\n"));
1055: pc->flag = DIFFERENT_NONZERO_PATTERN;
1056: } else {
1057: //PetscCall(PetscInfo(pc, "Setting up PC with same nonzero pattern\n"));
1058: pc->flag = SAME_NONZERO_PATTERN;
1059: }
1060: }
1061: pc->matstate = matstate;
1062: pc->matnonzerostate = matnonzerostate;
1064: if (!((PetscObject)pc)->type_name) {
1065: PetscCall(PCGetDefaultType_Private(pc, &def));
1066: PetscCall(PCSetType(pc, def));
1067: }
1069: PetscCall(MatSetErrorIfFailure(pc->pmat, pc->erroriffailure));
1070: PetscCall(MatSetErrorIfFailure(pc->mat, pc->erroriffailure));
1071: PetscCall(PetscLogEventBegin(PC_SetUp, pc, 0, 0, 0));
1072: if (pc->ops->setup) {
1073: /* do not log solves and applications of preconditioners while constructing preconditioners; perhaps they should be logged separately from the regular solves */
1074: PetscCall(KSPInitializePackage());
1075: PetscCall(PetscLogEventDeactivatePush(KSP_Solve));
1076: PetscCall(PetscLogEventDeactivatePush(PC_Apply));
1077: PetscUseTypeMethod(pc, setup);
1078: PetscCall(PetscLogEventDeactivatePop(KSP_Solve));
1079: PetscCall(PetscLogEventDeactivatePop(PC_Apply));
1080: }
1081: PetscCall(PetscLogEventEnd(PC_SetUp, pc, 0, 0, 0));
1082: if (!pc->setupcalled) pc->setupcalled = 1;
1083: PetscFunctionReturn(PETSC_SUCCESS);
1084: }
1086: /*@
1087: PCSetUpOnBlocks - Sets up the preconditioner for each block in
1088: the block Jacobi, block Gauss-Seidel, and overlapping Schwarz
1089: methods.
1091: Collective
1093: Input Parameter:
1094: . pc - the preconditioner context
1096: Level: developer
1098: Note:
1099: For nested preconditioners such as `PCBJACOBI` `PCSetUp()` is not called on each sub-`KSP` when `PCSetUp()` is
1100: called on the outer `PC`, this routine ensures it is called.
1102: .seealso: [](ch_ksp), `PC`, `PCSetUp()`, `PCCreate()`, `PCApply()`, `PCDestroy()`
1103: @*/
1104: PetscErrorCode PCSetUpOnBlocks(PC pc)
1105: {
1106: PetscFunctionBegin;
1108: if (!pc->ops->setuponblocks) PetscFunctionReturn(PETSC_SUCCESS);
1109: PetscCall(PetscLogEventBegin(PC_SetUpOnBlocks, pc, 0, 0, 0));
1110: PetscUseTypeMethod(pc, setuponblocks);
1111: PetscCall(PetscLogEventEnd(PC_SetUpOnBlocks, pc, 0, 0, 0));
1112: PetscFunctionReturn(PETSC_SUCCESS);
1113: }
1115: /*@C
1116: PCSetModifySubMatrices - Sets a user-defined routine for modifying the
1117: submatrices that arise within certain subdomain-based preconditioners such as `PCASM`
1119: Logically Collective
1121: Input Parameters:
1122: + pc - the preconditioner context
1123: . func - routine for modifying the submatrices
1124: - ctx - optional user-defined context (may be `NULL`)
1126: Calling sequence of `func`:
1127: + pc - the preconditioner context
1128: . nsub - number of index sets
1129: . row - an array of index sets that contain the global row numbers
1130: that comprise each local submatrix
1131: . col - an array of index sets that contain the global column numbers
1132: that comprise each local submatrix
1133: . submat - array of local submatrices
1134: - ctx - optional user-defined context for private data for the
1135: user-defined func routine (may be `NULL`)
1137: Level: advanced
1139: Notes:
1140: The basic submatrices are extracted from the preconditioner matrix as
1141: usual; the user can then alter these (for example, to set different boundary
1142: conditions for each submatrix) before they are used for the local solves.
1144: `PCSetModifySubMatrices()` MUST be called before `KSPSetUp()` and
1145: `KSPSolve()`.
1147: A routine set by `PCSetModifySubMatrices()` is currently called within
1148: the block Jacobi (`PCBJACOBI`) and additive Schwarz (`PCASM`)
1149: preconditioners. All other preconditioners ignore this routine.
1151: .seealso: [](ch_ksp), `PC`, `PCBJACOBI`, `PCASM`, `PCModifySubMatrices()`
1152: @*/
1153: PetscErrorCode PCSetModifySubMatrices(PC pc, PetscErrorCode (*func)(PC pc, PetscInt nsub, const IS row[], const IS col[], Mat submat[], void *ctx), void *ctx)
1154: {
1155: PetscFunctionBegin;
1157: pc->modifysubmatrices = func;
1158: pc->modifysubmatricesP = ctx;
1159: PetscFunctionReturn(PETSC_SUCCESS);
1160: }
1162: /*@C
1163: PCModifySubMatrices - Calls an optional user-defined routine within
1164: certain preconditioners if one has been set with `PCSetModifySubMatrices()`.
1166: Collective
1168: Input Parameters:
1169: + pc - the preconditioner context
1170: . nsub - the number of local submatrices
1171: . row - an array of index sets that contain the global row numbers
1172: that comprise each local submatrix
1173: . col - an array of index sets that contain the global column numbers
1174: that comprise each local submatrix
1175: . submat - array of local submatrices
1176: - ctx - optional user-defined context for private data for the
1177: user-defined routine (may be `NULL`)
1179: Output Parameter:
1180: . submat - array of local submatrices (the entries of which may
1181: have been modified)
1183: Level: developer
1185: Note:
1186: The user should NOT generally call this routine, as it will
1187: automatically be called within certain preconditioners.
1189: .seealso: [](ch_ksp), `PC`, `PCSetModifySubMatrices()`
1190: @*/
1191: PetscErrorCode PCModifySubMatrices(PC pc, PetscInt nsub, const IS row[], const IS col[], Mat submat[], void *ctx)
1192: {
1193: PetscFunctionBegin;
1195: if (!pc->modifysubmatrices) PetscFunctionReturn(PETSC_SUCCESS);
1196: PetscCall(PetscLogEventBegin(PC_ModifySubMatrices, pc, 0, 0, 0));
1197: PetscCall((*pc->modifysubmatrices)(pc, nsub, row, col, submat, ctx));
1198: PetscCall(PetscLogEventEnd(PC_ModifySubMatrices, pc, 0, 0, 0));
1199: PetscFunctionReturn(PETSC_SUCCESS);
1200: }
1202: /*@
1203: PCSetOperators - Sets the matrix associated with the linear system and
1204: a (possibly) different one associated with the preconditioner.
1206: Logically Collective
1208: Input Parameters:
1209: + pc - the preconditioner context
1210: . Amat - the matrix that defines the linear system
1211: - Pmat - the matrix to be used in constructing the preconditioner, usually the same as Amat.
1213: Level: intermediate
1215: Notes:
1216: Passing a `NULL` for `Amat` or `Pmat` removes the matrix that is currently used.
1218: If you wish to replace either `Amat` or `Pmat` but leave the other one untouched then
1219: first call `KSPGetOperators()` to get the one you wish to keep, call `PetscObjectReference()`
1220: on it and then pass it back in in your call to `KSPSetOperators()`.
1222: More Notes about Repeated Solution of Linear Systems:
1223: PETSc does NOT reset the matrix entries of either `Amat` or `Pmat`
1224: to zero after a linear solve; the user is completely responsible for
1225: matrix assembly. See the routine `MatZeroEntries()` if desiring to
1226: zero all elements of a matrix.
1228: .seealso: [](ch_ksp), `PC`, `PCGetOperators()`, `MatZeroEntries()`
1229: @*/
1230: PetscErrorCode PCSetOperators(PC pc, Mat Amat, Mat Pmat)
1231: {
1232: PetscInt m1, n1, m2, n2;
1234: PetscFunctionBegin;
1238: if (Amat) PetscCheckSameComm(pc, 1, Amat, 2);
1239: if (Pmat) PetscCheckSameComm(pc, 1, Pmat, 3);
1240: if (pc->setupcalled && pc->mat && pc->pmat && Amat && Pmat) {
1241: PetscCall(MatGetLocalSize(Amat, &m1, &n1));
1242: PetscCall(MatGetLocalSize(pc->mat, &m2, &n2));
1243: PetscCheck(m1 == m2 && n1 == n2, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Cannot change local size of Amat after use old sizes %" PetscInt_FMT " %" PetscInt_FMT " new sizes %" PetscInt_FMT " %" PetscInt_FMT, m2, n2, m1, n1);
1244: PetscCall(MatGetLocalSize(Pmat, &m1, &n1));
1245: PetscCall(MatGetLocalSize(pc->pmat, &m2, &n2));
1246: PetscCheck(m1 == m2 && n1 == n2, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Cannot change local size of Pmat after use old sizes %" PetscInt_FMT " %" PetscInt_FMT " new sizes %" PetscInt_FMT " %" PetscInt_FMT, m2, n2, m1, n1);
1247: }
1249: if (Pmat != pc->pmat) {
1250: /* changing the operator that defines the preconditioner thus reneed to clear current states so new preconditioner is built */
1251: pc->matnonzerostate = -1;
1252: pc->matstate = -1;
1253: }
1255: /* reference first in case the matrices are the same */
1256: if (Amat) PetscCall(PetscObjectReference((PetscObject)Amat));
1257: PetscCall(MatDestroy(&pc->mat));
1258: if (Pmat) PetscCall(PetscObjectReference((PetscObject)Pmat));
1259: PetscCall(MatDestroy(&pc->pmat));
1260: pc->mat = Amat;
1261: pc->pmat = Pmat;
1262: PetscFunctionReturn(PETSC_SUCCESS);
1263: }
1265: /*@
1266: PCSetReusePreconditioner - reuse the current preconditioner even if the operator in the preconditioner has changed.
1268: Logically Collective
1270: Input Parameters:
1271: + pc - the preconditioner context
1272: - flag - `PETSC_TRUE` do not compute a new preconditioner, `PETSC_FALSE` do compute a new preconditioner
1274: Level: intermediate
1276: Note:
1277: Normally if a matrix inside a `PC` changes the `PC` automatically updates itself using information from the changed matrix. This option
1278: prevents this.
1280: .seealso: [](ch_ksp), `PC`, `PCGetOperators()`, `MatZeroEntries()`, `PCGetReusePreconditioner()`, `KSPSetReusePreconditioner()`
1281: @*/
1282: PetscErrorCode PCSetReusePreconditioner(PC pc, PetscBool flag)
1283: {
1284: PetscFunctionBegin;
1287: pc->reusepreconditioner = flag;
1288: PetscFunctionReturn(PETSC_SUCCESS);
1289: }
1291: /*@
1292: PCGetReusePreconditioner - Determines if the `PC` reuses the current preconditioner even if the operator in the preconditioner has changed.
1294: Not Collective
1296: Input Parameter:
1297: . pc - the preconditioner context
1299: Output Parameter:
1300: . flag - `PETSC_TRUE` do not compute a new preconditioner, `PETSC_FALSE` do compute a new preconditioner
1302: Level: intermediate
1304: .seealso: [](ch_ksp), `PC`, `PCGetOperators()`, `MatZeroEntries()`, `PCSetReusePreconditioner()`
1305: @*/
1306: PetscErrorCode PCGetReusePreconditioner(PC pc, PetscBool *flag)
1307: {
1308: PetscFunctionBegin;
1310: PetscAssertPointer(flag, 2);
1311: *flag = pc->reusepreconditioner;
1312: PetscFunctionReturn(PETSC_SUCCESS);
1313: }
1315: /*@
1316: PCGetOperators - Gets the matrix associated with the linear system and
1317: possibly a different one associated with the preconditioner.
1319: Not Collective, though parallel `Mat`s are returned if `pc` is parallel
1321: Input Parameter:
1322: . pc - the preconditioner context
1324: Output Parameters:
1325: + Amat - the matrix defining the linear system
1326: - Pmat - the matrix from which the preconditioner is constructed, usually the same as Amat.
1328: Level: intermediate
1330: Note:
1331: Does not increase the reference count of the matrices, so you should not destroy them
1333: Alternative usage: If the operators have NOT been set with `KSPSetOperators()`/`PCSetOperators()` then the operators
1334: are created in `PC` and returned to the user. In this case, if both operators
1335: mat and pmat are requested, two DIFFERENT operators will be returned. If
1336: only one is requested both operators in the PC will be the same (i.e. as
1337: if one had called `KSPSetOperators()`/`PCSetOperators()` with the same argument for both Mats).
1338: The user must set the sizes of the returned matrices and their type etc just
1339: as if the user created them with `MatCreate()`. For example,
1341: .vb
1342: KSP/PCGetOperators(ksp/pc,&Amat,NULL); is equivalent to
1343: set size, type, etc of Amat
1345: MatCreate(comm,&mat);
1346: KSP/PCSetOperators(ksp/pc,Amat,Amat);
1347: PetscObjectDereference((PetscObject)mat);
1348: set size, type, etc of Amat
1349: .ve
1351: and
1353: .vb
1354: KSP/PCGetOperators(ksp/pc,&Amat,&Pmat); is equivalent to
1355: set size, type, etc of Amat and Pmat
1357: MatCreate(comm,&Amat);
1358: MatCreate(comm,&Pmat);
1359: KSP/PCSetOperators(ksp/pc,Amat,Pmat);
1360: PetscObjectDereference((PetscObject)Amat);
1361: PetscObjectDereference((PetscObject)Pmat);
1362: set size, type, etc of Amat and Pmat
1363: .ve
1365: The rationale for this support is so that when creating a `TS`, `SNES`, or `KSP` the hierarchy
1366: of underlying objects (i.e. `SNES`, `KSP`, `PC`, `Mat`) and their lifespans can be completely
1367: managed by the top most level object (i.e. the `TS`, `SNES`, or `KSP`). Another way to look
1368: at this is when you create a `SNES` you do not NEED to create a `KSP` and attach it to
1369: the `SNES` object (the `SNES` object manages it for you). Similarly when you create a KSP
1370: you do not need to attach a `PC` to it (the `KSP` object manages the `PC` object for you).
1371: Thus, why should YOU have to create the `Mat` and attach it to the `SNES`/`KSP`/`PC`, when
1372: it can be created for you?
1374: .seealso: [](ch_ksp), `PC`, `PCSetOperators()`, `KSPGetOperators()`, `KSPSetOperators()`, `PCGetOperatorsSet()`
1375: @*/
1376: PetscErrorCode PCGetOperators(PC pc, Mat *Amat, Mat *Pmat)
1377: {
1378: PetscFunctionBegin;
1380: if (Amat) {
1381: if (!pc->mat) {
1382: if (pc->pmat && !Pmat) { /* Pmat has been set, but user did not request it, so use for Amat */
1383: pc->mat = pc->pmat;
1384: PetscCall(PetscObjectReference((PetscObject)pc->mat));
1385: } else { /* both Amat and Pmat are empty */
1386: PetscCall(MatCreate(PetscObjectComm((PetscObject)pc), &pc->mat));
1387: if (!Pmat) { /* user did NOT request Pmat, so make same as Amat */
1388: pc->pmat = pc->mat;
1389: PetscCall(PetscObjectReference((PetscObject)pc->pmat));
1390: }
1391: }
1392: }
1393: *Amat = pc->mat;
1394: }
1395: if (Pmat) {
1396: if (!pc->pmat) {
1397: if (pc->mat && !Amat) { /* Amat has been set but was not requested, so use for pmat */
1398: pc->pmat = pc->mat;
1399: PetscCall(PetscObjectReference((PetscObject)pc->pmat));
1400: } else {
1401: PetscCall(MatCreate(PetscObjectComm((PetscObject)pc), &pc->pmat));
1402: if (!Amat) { /* user did NOT request Amat, so make same as Pmat */
1403: pc->mat = pc->pmat;
1404: PetscCall(PetscObjectReference((PetscObject)pc->mat));
1405: }
1406: }
1407: }
1408: *Pmat = pc->pmat;
1409: }
1410: PetscFunctionReturn(PETSC_SUCCESS);
1411: }
1413: /*@
1414: PCGetOperatorsSet - Determines if the matrix associated with the linear system and
1415: possibly a different one associated with the preconditioner have been set in the `PC`.
1417: Not Collective, though the results on all processes should be the same
1419: Input Parameter:
1420: . pc - the preconditioner context
1422: Output Parameters:
1423: + mat - the matrix associated with the linear system was set
1424: - pmat - matrix associated with the preconditioner was set, usually the same
1426: Level: intermediate
1428: .seealso: [](ch_ksp), `PC`, `PCSetOperators()`, `KSPGetOperators()`, `KSPSetOperators()`, `PCGetOperators()`
1429: @*/
1430: PetscErrorCode PCGetOperatorsSet(PC pc, PetscBool *mat, PetscBool *pmat)
1431: {
1432: PetscFunctionBegin;
1434: if (mat) *mat = (pc->mat) ? PETSC_TRUE : PETSC_FALSE;
1435: if (pmat) *pmat = (pc->pmat) ? PETSC_TRUE : PETSC_FALSE;
1436: PetscFunctionReturn(PETSC_SUCCESS);
1437: }
1439: /*@
1440: PCFactorGetMatrix - Gets the factored matrix from the
1441: preconditioner context. This routine is valid only for the `PCLU`,
1442: `PCILU`, `PCCHOLESKY`, and `PCICC` methods.
1444: Not Collective though `mat` is parallel if `pc` is parallel
1446: Input Parameter:
1447: . pc - the preconditioner context
1449: Output Parameters:
1450: . mat - the factored matrix
1452: Level: advanced
1454: Note:
1455: Does not increase the reference count for `mat` so DO NOT destroy it
1457: .seealso: [](ch_ksp), `PC`, `PCLU`, `PCILU`, `PCCHOLESKY`, `PCICC`
1458: @*/
1459: PetscErrorCode PCFactorGetMatrix(PC pc, Mat *mat)
1460: {
1461: PetscFunctionBegin;
1463: PetscAssertPointer(mat, 2);
1464: PetscCall(PCFactorSetUpMatSolverType(pc));
1465: PetscUseTypeMethod(pc, getfactoredmatrix, mat);
1466: PetscFunctionReturn(PETSC_SUCCESS);
1467: }
1469: /*@
1470: PCSetOptionsPrefix - Sets the prefix used for searching for all
1471: `PC` options in the database.
1473: Logically Collective
1475: Input Parameters:
1476: + pc - the preconditioner context
1477: - prefix - the prefix string to prepend to all `PC` option requests
1479: Note:
1480: A hyphen (-) must NOT be given at the beginning of the prefix name.
1481: The first character of all runtime options is AUTOMATICALLY the
1482: hyphen.
1484: Level: advanced
1486: .seealso: [](ch_ksp), `PC`, `PCSetFromOptions`, `PCAppendOptionsPrefix()`, `PCGetOptionsPrefix()`
1487: @*/
1488: PetscErrorCode PCSetOptionsPrefix(PC pc, const char prefix[])
1489: {
1490: PetscFunctionBegin;
1492: PetscCall(PetscObjectSetOptionsPrefix((PetscObject)pc, prefix));
1493: PetscFunctionReturn(PETSC_SUCCESS);
1494: }
1496: /*@
1497: PCAppendOptionsPrefix - Appends to the prefix used for searching for all
1498: `PC` options in the database.
1500: Logically Collective
1502: Input Parameters:
1503: + pc - the preconditioner context
1504: - prefix - the prefix string to prepend to all `PC` option requests
1506: Note:
1507: A hyphen (-) must NOT be given at the beginning of the prefix name.
1508: The first character of all runtime options is AUTOMATICALLY the
1509: hyphen.
1511: Level: advanced
1513: .seealso: [](ch_ksp), `PC`, `PCSetFromOptions`, `PCSetOptionsPrefix()`, `PCGetOptionsPrefix()`
1514: @*/
1515: PetscErrorCode PCAppendOptionsPrefix(PC pc, const char prefix[])
1516: {
1517: PetscFunctionBegin;
1519: PetscCall(PetscObjectAppendOptionsPrefix((PetscObject)pc, prefix));
1520: PetscFunctionReturn(PETSC_SUCCESS);
1521: }
1523: /*@
1524: PCGetOptionsPrefix - Gets the prefix used for searching for all
1525: PC options in the database.
1527: Not Collective
1529: Input Parameter:
1530: . pc - the preconditioner context
1532: Output Parameter:
1533: . prefix - pointer to the prefix string used, is returned
1535: Level: advanced
1537: Fortran Note:
1538: The user should pass in a string `prefix` of
1539: sufficient length to hold the prefix.
1541: .seealso: [](ch_ksp), `PC`, `PCSetFromOptions`, `PCSetOptionsPrefix()`, `PCAppendOptionsPrefix()`
1542: @*/
1543: PetscErrorCode PCGetOptionsPrefix(PC pc, const char *prefix[])
1544: {
1545: PetscFunctionBegin;
1547: PetscAssertPointer(prefix, 2);
1548: PetscCall(PetscObjectGetOptionsPrefix((PetscObject)pc, prefix));
1549: PetscFunctionReturn(PETSC_SUCCESS);
1550: }
1552: /*
1553: Indicates the right-hand side will be changed by KSPSolve(), this occurs for a few
1554: preconditioners including BDDC and Eisentat that transform the equations before applying
1555: the Krylov methods
1556: */
1557: PETSC_INTERN PetscErrorCode PCPreSolveChangeRHS(PC pc, PetscBool *change)
1558: {
1559: PetscFunctionBegin;
1561: PetscAssertPointer(change, 2);
1562: *change = PETSC_FALSE;
1563: PetscTryMethod(pc, "PCPreSolveChangeRHS_C", (PC, PetscBool *), (pc, change));
1564: PetscFunctionReturn(PETSC_SUCCESS);
1565: }
1567: /*@
1568: PCPreSolve - Optional pre-solve phase, intended for any
1569: preconditioner-specific actions that must be performed before
1570: the iterative solve itself.
1572: Collective
1574: Input Parameters:
1575: + pc - the preconditioner context
1576: - ksp - the Krylov subspace context
1578: Level: developer
1580: Example Usage:
1581: .vb
1582: PCPreSolve(pc,ksp);
1583: KSPSolve(ksp,b,x);
1584: PCPostSolve(pc,ksp);
1585: .ve
1587: Notes:
1588: The pre-solve phase is distinct from the `PCSetUp()` phase.
1590: `KSPSolve()` calls this directly, so is rarely called by the user.
1592: .seealso: [](ch_ksp), `PC`, `PCPostSolve()`
1593: @*/
1594: PetscErrorCode PCPreSolve(PC pc, KSP ksp)
1595: {
1596: Vec x, rhs;
1598: PetscFunctionBegin;
1601: pc->presolvedone++;
1602: PetscCheck(pc->presolvedone <= 2, PetscObjectComm((PetscObject)pc), PETSC_ERR_SUP, "Cannot embed PCPreSolve() more than twice");
1603: PetscCall(KSPGetSolution(ksp, &x));
1604: PetscCall(KSPGetRhs(ksp, &rhs));
1606: if (pc->ops->presolve) PetscUseTypeMethod(pc, presolve, ksp, rhs, x);
1607: else if (pc->presolve) PetscCall(pc->presolve(pc, ksp));
1608: PetscFunctionReturn(PETSC_SUCCESS);
1609: }
1611: /*@C
1612: PCSetPreSolve - Sets function used by `PCPreSolve()` which is intended for any
1613: preconditioner-specific actions that must be performed before
1614: the iterative solve itself.
1616: Logically Collective
1618: Input Parameters:
1619: + pc - the preconditioner object
1620: - presolve - the function to call before the solve
1622: Calling sequence of `presolve`:
1623: + pc - the `PC` context
1624: - ksp - the `KSP` context
1626: Level: developer
1628: .seealso: [](ch_ksp), `PC`, `PCSetUp()`, `PCPreSolve()`
1629: @*/
1630: PetscErrorCode PCSetPreSolve(PC pc, PetscErrorCode (*presolve)(PC pc, KSP ksp))
1631: {
1632: PetscFunctionBegin;
1634: pc->presolve = presolve;
1635: PetscFunctionReturn(PETSC_SUCCESS);
1636: }
1638: /*@
1639: PCPostSolve - Optional post-solve phase, intended for any
1640: preconditioner-specific actions that must be performed after
1641: the iterative solve itself.
1643: Collective
1645: Input Parameters:
1646: + pc - the preconditioner context
1647: - ksp - the Krylov subspace context
1649: Example Usage:
1650: .vb
1651: PCPreSolve(pc,ksp);
1652: KSPSolve(ksp,b,x);
1653: PCPostSolve(pc,ksp);
1654: .ve
1656: Level: developer
1658: Note:
1659: `KSPSolve()` calls this routine directly, so it is rarely called by the user.
1661: .seealso: [](ch_ksp), `PC`, `PCSetPostSolve()`, `PCSetPresolve()`, `PCPreSolve()`, `KSPSolve()`
1662: @*/
1663: PetscErrorCode PCPostSolve(PC pc, KSP ksp)
1664: {
1665: Vec x, rhs;
1667: PetscFunctionBegin;
1670: pc->presolvedone--;
1671: PetscCall(KSPGetSolution(ksp, &x));
1672: PetscCall(KSPGetRhs(ksp, &rhs));
1673: PetscTryTypeMethod(pc, postsolve, ksp, rhs, x);
1674: PetscFunctionReturn(PETSC_SUCCESS);
1675: }
1677: /*@
1678: PCLoad - Loads a `PC` that has been stored in binary with `PCView()`.
1680: Collective
1682: Input Parameters:
1683: + newdm - the newly loaded `PC`, this needs to have been created with `PCCreate()` or
1684: some related function before a call to `PCLoad()`.
1685: - viewer - binary file viewer, obtained from `PetscViewerBinaryOpen()`
1687: Level: intermediate
1689: Note:
1690: The type is determined by the data in the file, any `PCType` set into the `PC` before this call is ignored.
1692: .seealso: [](ch_ksp), `PC`, `PetscViewerBinaryOpen()`, `PCView()`, `MatLoad()`, `VecLoad()`
1693: @*/
1694: PetscErrorCode PCLoad(PC newdm, PetscViewer viewer)
1695: {
1696: PetscBool isbinary;
1697: PetscInt classid;
1698: char type[256];
1700: PetscFunctionBegin;
1703: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
1704: PetscCheck(isbinary, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Invalid viewer; open viewer with PetscViewerBinaryOpen()");
1706: PetscCall(PetscViewerBinaryRead(viewer, &classid, 1, NULL, PETSC_INT));
1707: PetscCheck(classid == PC_FILE_CLASSID, PetscObjectComm((PetscObject)newdm), PETSC_ERR_ARG_WRONG, "Not PC next in file");
1708: PetscCall(PetscViewerBinaryRead(viewer, type, 256, NULL, PETSC_CHAR));
1709: PetscCall(PCSetType(newdm, type));
1710: PetscTryTypeMethod(newdm, load, viewer);
1711: PetscFunctionReturn(PETSC_SUCCESS);
1712: }
1714: #include <petscdraw.h>
1715: #if defined(PETSC_HAVE_SAWS)
1716: #include <petscviewersaws.h>
1717: #endif
1719: /*@
1720: PCViewFromOptions - View from the `PC` based on options in the options database
1722: Collective
1724: Input Parameters:
1725: + A - the `PC` context
1726: . obj - Optional object that provides the options prefix
1727: - name - command line option
1729: Level: intermediate
1731: .seealso: [](ch_ksp), `PC`, `PCView`, `PetscObjectViewFromOptions()`, `PCCreate()`
1732: @*/
1733: PetscErrorCode PCViewFromOptions(PC A, PetscObject obj, const char name[])
1734: {
1735: PetscFunctionBegin;
1737: PetscCall(PetscObjectViewFromOptions((PetscObject)A, obj, name));
1738: PetscFunctionReturn(PETSC_SUCCESS);
1739: }
1741: /*@
1742: PCView - Prints information about the `PC`
1744: Collective
1746: Input Parameters:
1747: + pc - the `PC` context
1748: - viewer - optional visualization context
1750: Level: developer
1752: Notes:
1753: The available visualization contexts include
1754: + `PETSC_VIEWER_STDOUT_SELF` - standard output (default)
1755: - `PETSC_VIEWER_STDOUT_WORLD` - synchronized standard
1756: output where only the first processor opens
1757: the file. All other processors send their
1758: data to the first processor to print.
1760: The user can open an alternative visualization contexts with
1761: `PetscViewerASCIIOpen()` (output to a specified file).
1763: .seealso: [](ch_ksp), `PC`, `PetscViewer`, `KSPView()`, `PetscViewerASCIIOpen()`
1764: @*/
1765: PetscErrorCode PCView(PC pc, PetscViewer viewer)
1766: {
1767: PCType cstr;
1768: PetscViewerFormat format;
1769: PetscBool iascii, isstring, isbinary, isdraw, pop = PETSC_FALSE;
1770: #if defined(PETSC_HAVE_SAWS)
1771: PetscBool issaws;
1772: #endif
1774: PetscFunctionBegin;
1776: if (!viewer) PetscCall(PetscViewerASCIIGetStdout(PetscObjectComm((PetscObject)pc), &viewer));
1778: PetscCheckSameComm(pc, 1, viewer, 2);
1780: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
1781: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSTRING, &isstring));
1782: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
1783: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
1784: #if defined(PETSC_HAVE_SAWS)
1785: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSAWS, &issaws));
1786: #endif
1788: if (iascii) {
1789: PetscCall(PetscObjectPrintClassNamePrefixType((PetscObject)pc, viewer));
1790: if (!pc->setupcalled) PetscCall(PetscViewerASCIIPrintf(viewer, " PC has not been set up so information may be incomplete\n"));
1791: PetscCall(PetscViewerASCIIPushTab(viewer));
1792: PetscTryTypeMethod(pc, view, viewer);
1793: PetscCall(PetscViewerASCIIPopTab(viewer));
1794: if (pc->mat) {
1795: PetscCall(PetscViewerGetFormat(viewer, &format));
1796: if (format != PETSC_VIEWER_ASCII_INFO_DETAIL) {
1797: PetscCall(PetscViewerPushFormat(viewer, PETSC_VIEWER_ASCII_INFO));
1798: pop = PETSC_TRUE;
1799: }
1800: if (pc->pmat == pc->mat) {
1801: PetscCall(PetscViewerASCIIPrintf(viewer, " linear system matrix = precond matrix:\n"));
1802: PetscCall(PetscViewerASCIIPushTab(viewer));
1803: PetscCall(MatView(pc->mat, viewer));
1804: PetscCall(PetscViewerASCIIPopTab(viewer));
1805: } else {
1806: if (pc->pmat) {
1807: PetscCall(PetscViewerASCIIPrintf(viewer, " linear system matrix followed by preconditioner matrix:\n"));
1808: } else {
1809: PetscCall(PetscViewerASCIIPrintf(viewer, " linear system matrix:\n"));
1810: }
1811: PetscCall(PetscViewerASCIIPushTab(viewer));
1812: PetscCall(MatView(pc->mat, viewer));
1813: if (pc->pmat) PetscCall(MatView(pc->pmat, viewer));
1814: PetscCall(PetscViewerASCIIPopTab(viewer));
1815: }
1816: if (pop) PetscCall(PetscViewerPopFormat(viewer));
1817: }
1818: } else if (isstring) {
1819: PetscCall(PCGetType(pc, &cstr));
1820: PetscCall(PetscViewerStringSPrintf(viewer, " PCType: %-7.7s", cstr));
1821: PetscTryTypeMethod(pc, view, viewer);
1822: if (pc->mat) PetscCall(MatView(pc->mat, viewer));
1823: if (pc->pmat && pc->pmat != pc->mat) PetscCall(MatView(pc->pmat, viewer));
1824: } else if (isbinary) {
1825: PetscInt classid = PC_FILE_CLASSID;
1826: MPI_Comm comm;
1827: PetscMPIInt rank;
1828: char type[256];
1830: PetscCall(PetscObjectGetComm((PetscObject)pc, &comm));
1831: PetscCallMPI(MPI_Comm_rank(comm, &rank));
1832: if (rank == 0) {
1833: PetscCall(PetscViewerBinaryWrite(viewer, &classid, 1, PETSC_INT));
1834: PetscCall(PetscStrncpy(type, ((PetscObject)pc)->type_name, 256));
1835: PetscCall(PetscViewerBinaryWrite(viewer, type, 256, PETSC_CHAR));
1836: }
1837: PetscTryTypeMethod(pc, view, viewer);
1838: } else if (isdraw) {
1839: PetscDraw draw;
1840: char str[25];
1841: PetscReal x, y, bottom, h;
1842: PetscInt n;
1844: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
1845: PetscCall(PetscDrawGetCurrentPoint(draw, &x, &y));
1846: if (pc->mat) {
1847: PetscCall(MatGetSize(pc->mat, &n, NULL));
1848: PetscCall(PetscSNPrintf(str, 25, "PC: %s (%" PetscInt_FMT ")", ((PetscObject)pc)->type_name, n));
1849: } else {
1850: PetscCall(PetscSNPrintf(str, 25, "PC: %s", ((PetscObject)pc)->type_name));
1851: }
1852: PetscCall(PetscDrawStringBoxed(draw, x, y, PETSC_DRAW_RED, PETSC_DRAW_BLACK, str, NULL, &h));
1853: bottom = y - h;
1854: PetscCall(PetscDrawPushCurrentPoint(draw, x, bottom));
1855: PetscTryTypeMethod(pc, view, viewer);
1856: PetscCall(PetscDrawPopCurrentPoint(draw));
1857: #if defined(PETSC_HAVE_SAWS)
1858: } else if (issaws) {
1859: PetscMPIInt rank;
1861: PetscCall(PetscObjectName((PetscObject)pc));
1862: PetscCallMPI(MPI_Comm_rank(PETSC_COMM_WORLD, &rank));
1863: if (!((PetscObject)pc)->amsmem && rank == 0) PetscCall(PetscObjectViewSAWs((PetscObject)pc, viewer));
1864: if (pc->mat) PetscCall(MatView(pc->mat, viewer));
1865: if (pc->pmat && pc->pmat != pc->mat) PetscCall(MatView(pc->pmat, viewer));
1866: #endif
1867: }
1868: PetscFunctionReturn(PETSC_SUCCESS);
1869: }
1871: /*@C
1872: PCRegister - Adds a method (`PCType`) to the preconditioner package.
1874: Not collective. No Fortran Support
1876: Input Parameters:
1877: + sname - name of a new user-defined solver
1878: - function - routine to create method context
1880: Example Usage:
1881: .vb
1882: PCRegister("my_solver", MySolverCreate);
1883: .ve
1885: Then, your solver can be chosen with the procedural interface via
1886: $ PCSetType(pc, "my_solver")
1887: or at runtime via the option
1888: $ -pc_type my_solver
1890: Level: advanced
1892: Note:
1893: `PCRegister()` may be called multiple times to add several user-defined preconditioners.
1895: .seealso: [](ch_ksp), `PC`, `PCType`, `PCRegisterAll()`
1896: @*/
1897: PetscErrorCode PCRegister(const char sname[], PetscErrorCode (*function)(PC))
1898: {
1899: PetscFunctionBegin;
1900: PetscCall(PCInitializePackage());
1901: PetscCall(PetscFunctionListAdd(&PCList, sname, function));
1902: PetscFunctionReturn(PETSC_SUCCESS);
1903: }
1905: static PetscErrorCode MatMult_PC(Mat A, Vec X, Vec Y)
1906: {
1907: PC pc;
1909: PetscFunctionBegin;
1910: PetscCall(MatShellGetContext(A, &pc));
1911: PetscCall(PCApply(pc, X, Y));
1912: PetscFunctionReturn(PETSC_SUCCESS);
1913: }
1915: /*@
1916: PCComputeOperator - Computes the explicit preconditioned operator.
1918: Collective
1920: Input Parameters:
1921: + pc - the preconditioner object
1922: - mattype - the `MatType` to be used for the operator
1924: Output Parameter:
1925: . mat - the explicit preconditioned operator
1927: Level: advanced
1929: Note:
1930: This computation is done by applying the operators to columns of the identity matrix.
1931: This routine is costly in general, and is recommended for use only with relatively small systems.
1932: Currently, this routine uses a dense matrix format when `mattype` == `NULL`
1934: .seealso: [](ch_ksp), `PC`, `KSPComputeOperator()`, `MatType`
1935: @*/
1936: PetscErrorCode PCComputeOperator(PC pc, MatType mattype, Mat *mat)
1937: {
1938: PetscInt N, M, m, n;
1939: Mat A, Apc;
1941: PetscFunctionBegin;
1943: PetscAssertPointer(mat, 3);
1944: PetscCall(PCGetOperators(pc, &A, NULL));
1945: PetscCall(MatGetLocalSize(A, &m, &n));
1946: PetscCall(MatGetSize(A, &M, &N));
1947: PetscCall(MatCreateShell(PetscObjectComm((PetscObject)pc), m, n, M, N, pc, &Apc));
1948: PetscCall(MatShellSetOperation(Apc, MATOP_MULT, (void (*)(void))MatMult_PC));
1949: PetscCall(MatComputeOperator(Apc, mattype, mat));
1950: PetscCall(MatDestroy(&Apc));
1951: PetscFunctionReturn(PETSC_SUCCESS);
1952: }
1954: /*@
1955: PCSetCoordinates - sets the coordinates of all the nodes on the local process
1957: Collective
1959: Input Parameters:
1960: + pc - the solver context
1961: . dim - the dimension of the coordinates 1, 2, or 3
1962: . nloc - the blocked size of the coordinates array
1963: - coords - the coordinates array
1965: Level: intermediate
1967: Note:
1968: `coords` is an array of the dim coordinates for the nodes on
1969: the local processor, of size `dim`*`nloc`.
1970: If there are 108 equation on a processor
1971: for a displacement finite element discretization of elasticity (so
1972: that there are nloc = 36 = 108/3 nodes) then the array must have 108
1973: double precision values (ie, 3 * 36). These x y z coordinates
1974: should be ordered for nodes 0 to N-1 like so: [ 0.x, 0.y, 0.z, 1.x,
1975: ... , N-1.z ].
1977: .seealso: [](ch_ksp), `PC`, `MatSetNearNullSpace()`
1978: @*/
1979: PetscErrorCode PCSetCoordinates(PC pc, PetscInt dim, PetscInt nloc, PetscReal coords[])
1980: {
1981: PetscFunctionBegin;
1984: PetscTryMethod(pc, "PCSetCoordinates_C", (PC, PetscInt, PetscInt, PetscReal[]), (pc, dim, nloc, coords));
1985: PetscFunctionReturn(PETSC_SUCCESS);
1986: }
1988: /*@
1989: PCGetInterpolations - Gets interpolation matrices for all levels (except level 0)
1991: Logically Collective
1993: Input Parameter:
1994: . pc - the precondition context
1996: Output Parameters:
1997: + num_levels - the number of levels
1998: - interpolations - the interpolation matrices (size of `num_levels`-1)
2000: Level: advanced
2002: Developer Note:
2003: Why is this here instead of in `PCMG` etc?
2005: .seealso: [](ch_ksp), `PC`, `PCMG`, `PCMGGetRestriction()`, `PCMGSetInterpolation()`, `PCMGGetInterpolation()`, `PCGetCoarseOperators()`
2006: @*/
2007: PetscErrorCode PCGetInterpolations(PC pc, PetscInt *num_levels, Mat *interpolations[])
2008: {
2009: PetscFunctionBegin;
2011: PetscAssertPointer(num_levels, 2);
2012: PetscAssertPointer(interpolations, 3);
2013: PetscUseMethod(pc, "PCGetInterpolations_C", (PC, PetscInt *, Mat *[]), (pc, num_levels, interpolations));
2014: PetscFunctionReturn(PETSC_SUCCESS);
2015: }
2017: /*@
2018: PCGetCoarseOperators - Gets coarse operator matrices for all levels (except the finest level)
2020: Logically Collective
2022: Input Parameter:
2023: . pc - the precondition context
2025: Output Parameters:
2026: + num_levels - the number of levels
2027: - coarseOperators - the coarse operator matrices (size of `num_levels`-1)
2029: Level: advanced
2031: Developer Note:
2032: Why is this here instead of in `PCMG` etc?
2034: .seealso: [](ch_ksp), `PC`, `PCMG`, `PCMGGetRestriction()`, `PCMGSetInterpolation()`, `PCMGGetRScale()`, `PCMGGetInterpolation()`, `PCGetInterpolations()`
2035: @*/
2036: PetscErrorCode PCGetCoarseOperators(PC pc, PetscInt *num_levels, Mat *coarseOperators[])
2037: {
2038: PetscFunctionBegin;
2040: PetscAssertPointer(num_levels, 2);
2041: PetscAssertPointer(coarseOperators, 3);
2042: PetscUseMethod(pc, "PCGetCoarseOperators_C", (PC, PetscInt *, Mat *[]), (pc, num_levels, coarseOperators));
2043: PetscFunctionReturn(PETSC_SUCCESS);
2044: }