Actual source code: itfunc.c
1: /*
2: Interface KSP routines that the user calls.
3: */
5: #include <petsc/private/kspimpl.h>
6: #include <petsc/private/matimpl.h>
7: #include <petscdm.h>
9: /* number of nested levels of KSPSetUp/Solve(). This is used to determine if KSP_DIVERGED_ITS should be fatal. */
10: static PetscInt level = 0;
12: static inline PetscErrorCode ObjectView(PetscObject obj, PetscViewer viewer, PetscViewerFormat format)
13: {
14: PetscCall(PetscViewerPushFormat(viewer, format));
15: PetscCall(PetscObjectView(obj, viewer));
16: PetscCall(PetscViewerPopFormat(viewer));
17: return PETSC_SUCCESS;
18: }
20: /*@
21: KSPComputeExtremeSingularValues - Computes the extreme singular values
22: for the preconditioned operator. Called after or during `KSPSolve()`.
24: Not Collective
26: Input Parameter:
27: . ksp - iterative solver obtained from `KSPCreate()`
29: Output Parameters:
30: + emax - maximum estimated singular value
31: - emin - minimum estimated singular value
33: Options Database Key:
34: . -ksp_view_singularvalues - compute extreme singular values and print when `KSPSolve()` completes.
36: Level: advanced
38: Notes:
39: One must call `KSPSetComputeSingularValues()` before calling `KSPSetUp()`
40: (or use the option `-ksp_view_singularvalues`) in order for this routine to work correctly.
42: Many users may just want to use the monitoring routine
43: `KSPMonitorSingularValue()` (which can be set with option `-ksp_monitor_singular_value`)
44: to print the extreme singular values at each iteration of the linear solve.
46: Estimates of the smallest singular value may be very inaccurate, especially if the Krylov method has not converged.
47: The largest singular value is usually accurate to within a few percent if the method has converged, but is still not
48: intended for eigenanalysis. Consider the excellent package SLEPc if accurate values are required.
50: Disable restarts if using `KSPGMRES`, otherwise this estimate will only be using those iterations after the last
51: restart. See `KSPGMRESSetRestart()` for more details.
53: .seealso: [](ch_ksp), `KSPSetComputeSingularValues()`, `KSPMonitorSingularValue()`, `KSPComputeEigenvalues()`, `KSP`, `KSPComputeRitz()`
54: @*/
55: PetscErrorCode KSPComputeExtremeSingularValues(KSP ksp, PetscReal *emax, PetscReal *emin)
56: {
57: PetscFunctionBegin;
59: PetscAssertPointer(emax, 2);
60: PetscAssertPointer(emin, 3);
61: PetscCheck(ksp->calc_sings, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Singular values not requested before KSPSetUp()");
63: if (ksp->ops->computeextremesingularvalues) PetscUseTypeMethod(ksp, computeextremesingularvalues, emax, emin);
64: else {
65: *emin = -1.0;
66: *emax = -1.0;
67: }
68: PetscFunctionReturn(PETSC_SUCCESS);
69: }
71: /*@
72: KSPComputeEigenvalues - Computes the extreme eigenvalues for the
73: preconditioned operator. Called after or during `KSPSolve()`.
75: Not Collective
77: Input Parameters:
78: + ksp - iterative solver obtained from `KSPCreate()`
79: - n - size of arrays `r` and `c`. The number of eigenvalues computed `neig` will, in general, be less than this.
81: Output Parameters:
82: + r - real part of computed eigenvalues, provided by user with a dimension of at least `n`
83: . c - complex part of computed eigenvalues, provided by user with a dimension of at least `n`
84: - neig - actual number of eigenvalues computed (will be less than or equal to `n`)
86: Options Database Key:
87: . -ksp_view_eigenvalues - Prints eigenvalues to stdout
89: Level: advanced
91: Notes:
92: The number of eigenvalues estimated depends on the size of the Krylov space
93: generated during the `KSPSolve()` ; for example, with
94: `KSPCG` it corresponds to the number of CG iterations, for `KSPGMRES` it is the number
95: of GMRES iterations SINCE the last restart. Any extra space in `r` and `c`
96: will be ignored.
98: `KSPComputeEigenvalues()` does not usually provide accurate estimates; it is
99: intended only for assistance in understanding the convergence of iterative
100: methods, not for eigenanalysis. For accurate computation of eigenvalues we recommend using
101: the excellent package SLEPc.
103: One must call `KSPSetComputeEigenvalues()` before calling `KSPSetUp()`
104: in order for this routine to work correctly.
106: Many users may just want to use the monitoring routine
107: `KSPMonitorSingularValue()` (which can be set with option `-ksp_monitor_singular_value`)
108: to print the singular values at each iteration of the linear solve.
110: `KSPComputeRitz()` provides estimates for both the eigenvalues and their corresponding eigenvectors.
112: .seealso: [](ch_ksp), `KSPSetComputeEigenvalues()`, `KSPSetComputeSingularValues()`, `KSPMonitorSingularValue()`, `KSPComputeExtremeSingularValues()`, `KSP`, `KSPComputeRitz()`
113: @*/
114: PetscErrorCode KSPComputeEigenvalues(KSP ksp, PetscInt n, PetscReal r[], PetscReal c[], PetscInt *neig)
115: {
116: PetscFunctionBegin;
118: if (n) PetscAssertPointer(r, 3);
119: if (n) PetscAssertPointer(c, 4);
120: PetscCheck(n >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Requested < 0 Eigenvalues");
121: PetscAssertPointer(neig, 5);
122: PetscCheck(ksp->calc_sings, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Eigenvalues not requested before KSPSetUp()");
124: if (n && ksp->ops->computeeigenvalues) PetscUseTypeMethod(ksp, computeeigenvalues, n, r, c, neig);
125: else *neig = 0;
126: PetscFunctionReturn(PETSC_SUCCESS);
127: }
129: /*@
130: KSPComputeRitz - Computes the Ritz or harmonic Ritz pairs associated with the
131: smallest or largest in modulus, for the preconditioned operator.
133: Not Collective
135: Input Parameters:
136: + ksp - iterative solver obtained from `KSPCreate()`
137: . ritz - `PETSC_TRUE` or `PETSC_FALSE` for Ritz pairs or harmonic Ritz pairs, respectively
138: - small - `PETSC_TRUE` or `PETSC_FALSE` for smallest or largest (harmonic) Ritz values, respectively
140: Output Parameters:
141: + nrit - On input number of (harmonic) Ritz pairs to compute; on output, actual number of computed (harmonic) Ritz pairs
142: . S - an array of the Ritz vectors, pass in an array of vectors of size `nrit`
143: . tetar - real part of the Ritz values, pass in an array of size `nrit`
144: - tetai - imaginary part of the Ritz values, pass in an array of size `nrit`
146: Level: advanced
148: Notes:
149: This only works with a `KSPType` of `KSPGMRES`.
151: One must call `KSPSetComputeRitz()` before calling `KSPSetUp()` in order for this routine to work correctly.
153: This routine must be called after `KSPSolve()`.
155: In `KSPGMRES`, the (harmonic) Ritz pairs are computed from the Hessenberg matrix obtained during
156: the last complete cycle of the GMRES solve, or during the partial cycle if the solve ended before
157: a restart (that is a complete GMRES cycle was never achieved).
159: The number of actual (harmonic) Ritz pairs computed is less than or equal to the restart
160: parameter for GMRES if a complete cycle has been performed or less or equal to the number of GMRES
161: iterations.
163: `KSPComputeEigenvalues()` provides estimates for only the eigenvalues (Ritz values).
165: For real matrices, the (harmonic) Ritz pairs can be complex-valued. In such a case,
166: the routine selects the complex (harmonic) Ritz value and its conjugate, and two successive entries of the
167: vectors `S` are equal to the real and the imaginary parts of the associated vectors.
168: When PETSc has been built with complex scalars, the real and imaginary parts of the Ritz
169: values are still returned in `tetar` and `tetai`, as is done in `KSPComputeEigenvalues()`, but
170: the Ritz vectors S are complex.
172: The (harmonic) Ritz pairs are given in order of increasing (harmonic) Ritz values in modulus.
174: The Ritz pairs do not necessarily accurately reflect the eigenvalues and eigenvectors of the operator, consider the
175: excellent package SLEPc if accurate values are required.
177: .seealso: [](ch_ksp), `KSPSetComputeRitz()`, `KSP`, `KSPGMRES`, `KSPComputeEigenvalues()`, `KSPSetComputeSingularValues()`, `KSPMonitorSingularValue()`
178: @*/
179: PetscErrorCode KSPComputeRitz(KSP ksp, PetscBool ritz, PetscBool small, PetscInt *nrit, Vec S[], PetscReal tetar[], PetscReal tetai[])
180: {
181: PetscFunctionBegin;
183: PetscCheck(ksp->calc_ritz, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Ritz pairs not requested before KSPSetUp()");
184: PetscTryTypeMethod(ksp, computeritz, ritz, small, nrit, S, tetar, tetai);
185: PetscFunctionReturn(PETSC_SUCCESS);
186: }
188: /*@
189: KSPSetUpOnBlocks - Sets up the preconditioner for each block in
190: the block Jacobi `PCJACOBI`, overlapping Schwarz `PCASM`, and fieldsplit `PCFIELDSPLIT` preconditioners
192: Collective
194: Input Parameter:
195: . ksp - the `KSP` context
197: Level: advanced
199: Notes:
200: `KSPSetUpOnBlocks()` is a routine that the user can optionally call for
201: more precise profiling (via `-log_view`) of the setup phase for these
202: block preconditioners. If the user does not call `KSPSetUpOnBlocks()`,
203: it will automatically be called from within `KSPSolve()`.
205: Calling `KSPSetUpOnBlocks()` is the same as calling `PCSetUpOnBlocks()`
206: on the `PC` context within the `KSP` context.
208: .seealso: [](ch_ksp), `PCSetUpOnBlocks()`, `KSPSetUp()`, `PCSetUp()`, `KSP`
209: @*/
210: PetscErrorCode KSPSetUpOnBlocks(KSP ksp)
211: {
212: PC pc;
213: PCFailedReason pcreason;
215: PetscFunctionBegin;
217: level++;
218: PetscCall(KSPGetPC(ksp, &pc));
219: PetscCall(PCSetUpOnBlocks(pc));
220: PetscCall(PCGetFailedReason(pc, &pcreason));
221: level--;
222: /*
223: This is tricky since only a subset of MPI ranks may set this; each KSPSolve_*() is responsible for checking
224: this flag and initializing an appropriate vector with VecFlag() so that the first norm computation can
225: produce a result at KSPCheckNorm() thus communicating the known problem to all MPI ranks so they may
226: terminate the Krylov solve. For many KSP implementations this is handled within KSPInitialResidual()
227: */
228: if (pcreason) ksp->reason = KSP_DIVERGED_PC_FAILED;
229: PetscFunctionReturn(PETSC_SUCCESS);
230: }
232: /*@
233: KSPSetReusePreconditioner - reuse the current preconditioner for future `KSPSolve()`, do not construct a new preconditioner even if the `Mat` operator
234: in the `KSP` has different values
236: Collective
238: Input Parameters:
239: + ksp - iterative solver obtained from `KSPCreate()`
240: - flag - `PETSC_TRUE` to reuse the current preconditioner, or `PETSC_FALSE` to construct a new preconditioner
242: Options Database Key:
243: . -ksp_reuse_preconditioner <true,false> - reuse the previously computed preconditioner
245: Level: intermediate
247: Notes:
248: When using `SNES` one can use `SNESSetLagPreconditioner()` to determine when preconditioners are reused.
250: Reusing the preconditioner reduces the time needed to form new preconditioners but may (significantly) increase the number
251: of iterations needed for future solves depending on how much the matrix entries have changed.
253: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `KSP`, `KSPGetReusePreconditioner()`,
254: `SNESSetLagPreconditioner()`, `SNES`
255: @*/
256: PetscErrorCode KSPSetReusePreconditioner(KSP ksp, PetscBool flag)
257: {
258: PC pc;
260: PetscFunctionBegin;
262: PetscCall(KSPGetPC(ksp, &pc));
263: PetscCall(PCSetReusePreconditioner(pc, flag));
264: PetscFunctionReturn(PETSC_SUCCESS);
265: }
267: /*@
268: KSPGetReusePreconditioner - Determines if the `KSP` reuses the current preconditioner even if the `Mat` operator in the `KSP` has changed.
270: Collective
272: Input Parameter:
273: . ksp - iterative solver obtained from `KSPCreate()`
275: Output Parameter:
276: . flag - the boolean flag indicating if the current preconditioner should be reused
278: Level: intermediate
280: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `KSPSetReusePreconditioner()`, `KSP`
281: @*/
282: PetscErrorCode KSPGetReusePreconditioner(KSP ksp, PetscBool *flag)
283: {
284: PetscFunctionBegin;
286: PetscAssertPointer(flag, 2);
287: *flag = PETSC_FALSE;
288: if (ksp->pc) PetscCall(PCGetReusePreconditioner(ksp->pc, flag));
289: PetscFunctionReturn(PETSC_SUCCESS);
290: }
292: /*@
293: KSPSetSkipPCSetFromOptions - prevents `KSPSetFromOptions()` from calling `PCSetFromOptions()`.
294: This is used if the same `PC` is shared by more than one `KSP` so its options are not reset for each `KSP`
296: Collective
298: Input Parameters:
299: + ksp - iterative solver obtained from `KSPCreate()`
300: - flag - `PETSC_TRUE` to skip calling the `PCSetFromOptions()`
302: Level: developer
304: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `PCSetReusePreconditioner()`, `KSP`
305: @*/
306: PetscErrorCode KSPSetSkipPCSetFromOptions(KSP ksp, PetscBool flag)
307: {
308: PetscFunctionBegin;
310: ksp->skippcsetfromoptions = flag;
311: PetscFunctionReturn(PETSC_SUCCESS);
312: }
314: /*@
315: KSPSetUp - Sets up the internal data structures for the
316: later use `KSPSolve()` the `KSP` linear iterative solver.
318: Collective
320: Input Parameter:
321: . ksp - iterative solver, `KSP`, obtained from `KSPCreate()`
323: Level: developer
325: Note:
326: This is called automatically by `KSPSolve()` so usually does not need to be called directly.
328: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `KSP`, `KSPSetUpOnBlocks()`
329: @*/
330: PetscErrorCode KSPSetUp(KSP ksp)
331: {
332: Mat A, B;
333: Mat mat, pmat;
334: MatNullSpace nullsp;
335: PCFailedReason pcreason;
336: PC pc;
337: PetscBool pcmpi;
339: PetscFunctionBegin;
341: PetscCall(KSPGetPC(ksp, &pc));
342: PetscCall(PetscObjectTypeCompare((PetscObject)pc, PCMPI, &pcmpi));
343: if (pcmpi) {
344: PetscBool ksppreonly;
345: PetscCall(PetscObjectTypeCompare((PetscObject)ksp, KSPPREONLY, &ksppreonly));
346: if (!ksppreonly) PetscCall(KSPSetType(ksp, KSPPREONLY));
347: }
348: level++;
350: /* reset the convergence flag from the previous solves */
351: ksp->reason = KSP_CONVERGED_ITERATING;
353: if (!((PetscObject)ksp)->type_name) PetscCall(KSPSetType(ksp, KSPGMRES));
354: PetscCall(KSPSetUpNorms_Private(ksp, PETSC_TRUE, &ksp->normtype, &ksp->pc_side));
356: if (ksp->dmActive && !ksp->setupstage) {
357: /* first time in so build matrix and vector data structures using DM */
358: if (!ksp->vec_rhs) PetscCall(DMCreateGlobalVector(ksp->dm, &ksp->vec_rhs));
359: if (!ksp->vec_sol) PetscCall(DMCreateGlobalVector(ksp->dm, &ksp->vec_sol));
360: PetscCall(DMCreateMatrix(ksp->dm, &A));
361: PetscCall(KSPSetOperators(ksp, A, A));
362: PetscCall(PetscObjectDereference((PetscObject)A));
363: }
365: if (ksp->dmActive) {
366: DMKSP kdm;
367: PetscCall(DMGetDMKSP(ksp->dm, &kdm));
369: if (kdm->ops->computeinitialguess && ksp->setupstage != KSP_SETUP_NEWRHS) {
370: /* only computes initial guess the first time through */
371: PetscCallBack("KSP callback initial guess", (*kdm->ops->computeinitialguess)(ksp, ksp->vec_sol, kdm->initialguessctx));
372: PetscCall(KSPSetInitialGuessNonzero(ksp, PETSC_TRUE));
373: }
374: if (kdm->ops->computerhs) PetscCallBack("KSP callback rhs", (*kdm->ops->computerhs)(ksp, ksp->vec_rhs, kdm->rhsctx));
376: if (ksp->setupstage != KSP_SETUP_NEWRHS) {
377: PetscCheck(kdm->ops->computeoperators, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "You called KSPSetDM() but did not use DMKSPSetComputeOperators() or KSPSetDMActive(ksp,PETSC_FALSE);");
378: PetscCall(KSPGetOperators(ksp, &A, &B));
379: PetscCallBack("KSP callback operators", (*kdm->ops->computeoperators)(ksp, A, B, kdm->operatorsctx));
380: }
381: }
383: if (ksp->setupstage == KSP_SETUP_NEWRHS) {
384: level--;
385: PetscFunctionReturn(PETSC_SUCCESS);
386: }
387: PetscCall(PetscLogEventBegin(KSP_SetUp, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
389: switch (ksp->setupstage) {
390: case KSP_SETUP_NEW:
391: PetscUseTypeMethod(ksp, setup);
392: break;
393: case KSP_SETUP_NEWMATRIX: /* This should be replaced with a more general mechanism */
394: if (ksp->setupnewmatrix) PetscUseTypeMethod(ksp, setup);
395: break;
396: default:
397: break;
398: }
400: if (!ksp->pc) PetscCall(KSPGetPC(ksp, &ksp->pc));
401: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
402: /* scale the matrix if requested */
403: if (ksp->dscale) {
404: PetscScalar *xx;
405: PetscInt i, n;
406: PetscBool zeroflag = PETSC_FALSE;
408: if (!ksp->diagonal) { /* allocate vector to hold diagonal */
409: PetscCall(MatCreateVecs(pmat, &ksp->diagonal, NULL));
410: }
411: PetscCall(MatGetDiagonal(pmat, ksp->diagonal));
412: PetscCall(VecGetLocalSize(ksp->diagonal, &n));
413: PetscCall(VecGetArray(ksp->diagonal, &xx));
414: for (i = 0; i < n; i++) {
415: if (xx[i] != 0.0) xx[i] = 1.0 / PetscSqrtReal(PetscAbsScalar(xx[i]));
416: else {
417: xx[i] = 1.0;
418: zeroflag = PETSC_TRUE;
419: }
420: }
421: PetscCall(VecRestoreArray(ksp->diagonal, &xx));
422: if (zeroflag) PetscCall(PetscInfo(ksp, "Zero detected in diagonal of matrix, using 1 at those locations\n"));
423: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
424: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
425: ksp->dscalefix2 = PETSC_FALSE;
426: }
427: PetscCall(PetscLogEventEnd(KSP_SetUp, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
428: PetscCall(PCSetErrorIfFailure(ksp->pc, ksp->errorifnotconverged));
429: PetscCall(PCSetUp(ksp->pc));
430: PetscCall(PCGetFailedReason(ksp->pc, &pcreason));
431: /* TODO: this code was wrong and is still wrong, there is no way to propagate the failure to all processes; their is no code to handle a ksp->reason on only some ranks */
432: if (pcreason) ksp->reason = KSP_DIVERGED_PC_FAILED;
434: PetscCall(MatGetNullSpace(mat, &nullsp));
435: if (nullsp) {
436: PetscBool test = PETSC_FALSE;
437: PetscCall(PetscOptionsGetBool(((PetscObject)ksp)->options, ((PetscObject)ksp)->prefix, "-ksp_test_null_space", &test, NULL));
438: if (test) PetscCall(MatNullSpaceTest(nullsp, mat, NULL));
439: }
440: ksp->setupstage = KSP_SETUP_NEWRHS;
441: level--;
442: PetscFunctionReturn(PETSC_SUCCESS);
443: }
445: /*@
446: KSPConvergedReasonView - Displays the reason a `KSP` solve converged or diverged, `KSPConvergedReason` to a `PetscViewer`
448: Collective
450: Input Parameters:
451: + ksp - iterative solver obtained from `KSPCreate()`
452: - viewer - the `PetscViewer` on which to display the reason
454: Options Database Keys:
455: + -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
456: - -ksp_converged_reason ::failed - only print reason and number of iterations when diverged
458: Level: beginner
460: Note:
461: Use `KSPConvergedReasonViewFromOptions()` to display the reason based on values in the PETSc options database.
463: To change the format of the output call `PetscViewerPushFormat`(`viewer`,`format`) before this call. Use `PETSC_VIEWER_DEFAULT` for the default,
464: use `PETSC_VIEWER_FAILED` to only display a reason if it fails.
466: .seealso: [](ch_ksp), `KSPConvergedReasonViewFromOptions()`, `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
467: `KSPSolveTranspose()`, `KSPGetIterationNumber()`, `KSP`, `KSPGetConvergedReason()`, `PetscViewerPushFormat()`, `PetscViewerPopFormat()`
468: @*/
469: PetscErrorCode KSPConvergedReasonView(KSP ksp, PetscViewer viewer)
470: {
471: PetscBool isAscii;
472: PetscViewerFormat format;
474: PetscFunctionBegin;
475: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
476: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isAscii));
477: if (isAscii) {
478: PetscCall(PetscViewerGetFormat(viewer, &format));
479: PetscCall(PetscViewerASCIIAddTab(viewer, ((PetscObject)ksp)->tablevel + 1));
480: if (ksp->reason > 0 && format != PETSC_VIEWER_FAILED) {
481: if (((PetscObject)ksp)->prefix) {
482: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve converged due to %s iterations %" PetscInt_FMT "\n", ((PetscObject)ksp)->prefix, KSPConvergedReasons[ksp->reason], ksp->its));
483: } else {
484: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve converged due to %s iterations %" PetscInt_FMT "\n", KSPConvergedReasons[ksp->reason], ksp->its));
485: }
486: } else if (ksp->reason <= 0) {
487: if (((PetscObject)ksp)->prefix) {
488: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve did not converge due to %s iterations %" PetscInt_FMT "\n", ((PetscObject)ksp)->prefix, KSPConvergedReasons[ksp->reason], ksp->its));
489: } else {
490: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve did not converge due to %s iterations %" PetscInt_FMT "\n", KSPConvergedReasons[ksp->reason], ksp->its));
491: }
492: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
493: PCFailedReason reason;
494: PetscCall(PCGetFailedReason(ksp->pc, &reason));
495: PetscCall(PetscViewerASCIIPrintf(viewer, " PC failed due to %s\n", PCFailedReasons[reason]));
496: }
497: }
498: PetscCall(PetscViewerASCIISubtractTab(viewer, ((PetscObject)ksp)->tablevel + 1));
499: }
500: PetscFunctionReturn(PETSC_SUCCESS);
501: }
503: /*@C
504: KSPConvergedReasonViewSet - Sets an ADDITIONAL function that is to be used at the
505: end of the linear solver to display the convergence reason of the linear solver.
507: Logically Collective
509: Input Parameters:
510: + ksp - the `KSP` context
511: . f - the `ksp` converged reason view function, see `KSPConvergedReasonViewFn`
512: . ctx - [optional] context for private data for the `KSPConvergedReason` view routine (use `NULL` if context is not needed)
513: - reasonviewdestroy - [optional] routine that frees `ctx` (may be `NULL`), see `PetscCtxDestroyFn` for the calling sequence
515: Options Database Keys:
516: + -ksp_converged_reason - sets a default `KSPConvergedReasonView()`
517: - -ksp_converged_reason_view_cancel - cancels all converged reason viewers that have been hardwired into a code by
518: calls to `KSPConvergedReasonViewSet()`, but does not cancel those set via the options database.
520: Level: intermediate
522: Note:
523: Several different converged reason view routines may be set by calling
524: `KSPConvergedReasonViewSet()` multiple times; all will be called in the
525: order in which they were set.
527: Developer Note:
528: Should be named KSPConvergedReasonViewAdd().
530: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPConvergedReasonViewFn`, `KSPConvergedReasonViewCancel()`, `PetscCtxDestroyFn`
531: @*/
532: PetscErrorCode KSPConvergedReasonViewSet(KSP ksp, KSPConvergedReasonViewFn *f, PetscCtx ctx, PetscCtxDestroyFn *reasonviewdestroy)
533: {
534: PetscFunctionBegin;
536: for (PetscInt i = 0; i < ksp->numberreasonviews; i++) {
537: PetscBool identical;
539: PetscCall(PetscMonitorCompare((PetscErrorCode (*)(void))(PetscVoidFn *)f, ctx, reasonviewdestroy, (PetscErrorCode (*)(void))(PetscVoidFn *)ksp->reasonview[i], ksp->reasonviewcontext[i], ksp->reasonviewdestroy[i], &identical));
540: if (identical) PetscFunctionReturn(PETSC_SUCCESS);
541: }
542: PetscCheck(ksp->numberreasonviews < MAXKSPREASONVIEWS, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Too many KSP reasonview set");
543: ksp->reasonview[ksp->numberreasonviews] = f;
544: ksp->reasonviewdestroy[ksp->numberreasonviews] = reasonviewdestroy;
545: ksp->reasonviewcontext[ksp->numberreasonviews++] = ctx;
546: PetscFunctionReturn(PETSC_SUCCESS);
547: }
549: /*@
550: KSPConvergedReasonViewCancel - Clears all the `KSPConvergedReason` view functions for a `KSP` object set with `KSPConvergedReasonViewSet()`
551: as well as the default viewer.
553: Collective
555: Input Parameter:
556: . ksp - iterative solver obtained from `KSPCreate()`
558: Level: intermediate
560: .seealso: [](ch_ksp), `KSPCreate()`, `KSPDestroy()`, `KSPReset()`, `KSPConvergedReasonViewSet()`
561: @*/
562: PetscErrorCode KSPConvergedReasonViewCancel(KSP ksp)
563: {
564: PetscInt i;
566: PetscFunctionBegin;
568: for (i = 0; i < ksp->numberreasonviews; i++) {
569: if (ksp->reasonviewdestroy[i]) PetscCall((*ksp->reasonviewdestroy[i])(&ksp->reasonviewcontext[i]));
570: }
571: ksp->numberreasonviews = 0;
572: PetscCall(PetscViewerDestroy(&ksp->convergedreasonviewer));
573: PetscFunctionReturn(PETSC_SUCCESS);
574: }
576: /*@
577: KSPConvergedReasonViewFromOptions - Processes command line options to determine if/how a `KSPConvergedReason` is to be viewed.
579: Collective
581: Input Parameter:
582: . ksp - the `KSP` object
584: Level: intermediate
586: Note:
587: This is called automatically at the conclusion of `KSPSolve()` so is rarely called directly by user code.
589: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPConvergedReasonViewSet()`
590: @*/
591: PetscErrorCode KSPConvergedReasonViewFromOptions(KSP ksp)
592: {
593: PetscFunctionBegin;
594: /* Call all user-provided reason review routines */
595: for (PetscInt i = 0; i < ksp->numberreasonviews; i++) PetscCall((*ksp->reasonview[i])(ksp, ksp->reasonviewcontext[i]));
597: /* Call the default PETSc routine */
598: if (ksp->convergedreasonviewer) {
599: PetscCall(PetscViewerPushFormat(ksp->convergedreasonviewer, ksp->convergedreasonformat));
600: PetscCall(KSPConvergedReasonView(ksp, ksp->convergedreasonviewer));
601: PetscCall(PetscViewerPopFormat(ksp->convergedreasonviewer));
602: }
603: PetscFunctionReturn(PETSC_SUCCESS);
604: }
606: /*@
607: KSPConvergedRateView - Displays the convergence rate <https://en.wikipedia.org/wiki/Coefficient_of_determination> of `KSPSolve()` to a viewer
609: Collective
611: Input Parameters:
612: + ksp - iterative solver obtained from `KSPCreate()`
613: - viewer - the `PetscViewer` to display the reason
615: Options Database Key:
616: . -ksp_converged_rate - print reason for convergence or divergence and the convergence rate (or 0.0 for divergence)
618: Level: intermediate
620: Notes:
621: To change the format of the output, call `PetscViewerPushFormat`(`viewer`,`format`) before this call.
623: Suppose that the residual is reduced linearly, $r_k = c^k r_0$, which means $\log r_k = \log r_0 + k \log c$. After linear regression,
624: the slope is $\log c$. The coefficient of determination is given by $1 - \frac{\sum_i (y_i - f(x_i))^2}{\sum_i (y_i - \bar y)}$,
626: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPGetConvergedRate()`, `KSPSetTolerances()`, `KSPConvergedDefault()`
627: @*/
628: PetscErrorCode KSPConvergedRateView(KSP ksp, PetscViewer viewer)
629: {
630: PetscViewerFormat format;
631: PetscBool isAscii;
632: PetscReal rrate, rRsq, erate = 0.0, eRsq = 0.0;
633: PetscInt its;
634: const char *prefix, *reason = KSPConvergedReasons[ksp->reason];
636: PetscFunctionBegin;
637: PetscCall(KSPGetIterationNumber(ksp, &its));
638: PetscCall(KSPComputeConvergenceRate(ksp, &rrate, &rRsq, &erate, &eRsq));
639: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
640: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isAscii));
641: if (isAscii) {
642: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
643: PetscCall(PetscViewerGetFormat(viewer, &format));
644: PetscCall(PetscViewerASCIIAddTab(viewer, ((PetscObject)ksp)->tablevel));
645: if (ksp->reason > 0) {
646: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve converged due to %s iterations %" PetscInt_FMT, prefix, reason, its));
647: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve converged due to %s iterations %" PetscInt_FMT, reason, its));
648: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
649: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
650: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
651: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
652: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
653: } else if (ksp->reason <= 0) {
654: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve did not converge due to %s iterations %" PetscInt_FMT, prefix, reason, its));
655: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve did not converge due to %s iterations %" PetscInt_FMT, reason, its));
656: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
657: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
658: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
659: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
660: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
661: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
662: PCFailedReason reason;
663: PetscCall(PCGetFailedReason(ksp->pc, &reason));
664: PetscCall(PetscViewerASCIIPrintf(viewer, " PC failed due to %s\n", PCFailedReasons[reason]));
665: }
666: }
667: PetscCall(PetscViewerASCIISubtractTab(viewer, ((PetscObject)ksp)->tablevel));
668: }
669: PetscFunctionReturn(PETSC_SUCCESS);
670: }
672: #include <petscdraw.h>
674: static PetscErrorCode KSPViewEigenvalues_Internal(KSP ksp, PetscBool isExplicit, PetscViewer viewer, PetscViewerFormat format)
675: {
676: PetscReal *r, *c;
677: PetscInt n, i, neig;
678: PetscBool isascii, isdraw;
679: PetscMPIInt rank;
681: PetscFunctionBegin;
682: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)ksp), &rank));
683: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
684: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
685: if (isExplicit) {
686: PetscCall(VecGetSize(ksp->vec_sol, &n));
687: PetscCall(PetscMalloc2(n, &r, n, &c));
688: PetscCall(KSPComputeEigenvaluesExplicitly(ksp, n, r, c));
689: neig = n;
690: } else {
691: PetscInt nits;
693: PetscCall(KSPGetIterationNumber(ksp, &nits));
694: n = nits + 2;
695: if (!nits) {
696: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any eigenvalues\n"));
697: PetscFunctionReturn(PETSC_SUCCESS);
698: }
699: PetscCall(PetscMalloc2(n, &r, n, &c));
700: PetscCall(KSPComputeEigenvalues(ksp, n, r, c, &neig));
701: }
702: if (isascii) {
703: PetscCall(PetscViewerASCIIPrintf(viewer, "%s computed eigenvalues\n", isExplicit ? "Explicitly" : "Iteratively"));
704: for (i = 0; i < neig; ++i) {
705: if (c[i] >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, "%g + %gi\n", (double)r[i], (double)c[i]));
706: else PetscCall(PetscViewerASCIIPrintf(viewer, "%g - %gi\n", (double)r[i], -(double)c[i]));
707: }
708: } else if (isdraw && rank == 0) {
709: PetscDraw draw;
710: PetscDrawSP drawsp;
712: if (format == PETSC_VIEWER_DRAW_CONTOUR) {
713: PetscCall(KSPPlotEigenContours_Private(ksp, neig, r, c));
714: } else {
715: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
716: PetscCall(PetscDrawSPCreate(draw, 1, &drawsp));
717: PetscCall(PetscDrawSPReset(drawsp));
718: for (i = 0; i < neig; ++i) PetscCall(PetscDrawSPAddPoint(drawsp, r + i, c + i));
719: PetscCall(PetscDrawSPDraw(drawsp, PETSC_TRUE));
720: PetscCall(PetscDrawSPSave(drawsp));
721: PetscCall(PetscDrawSPDestroy(&drawsp));
722: }
723: }
724: PetscCall(PetscFree2(r, c));
725: PetscFunctionReturn(PETSC_SUCCESS);
726: }
728: static PetscErrorCode KSPViewSingularvalues_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
729: {
730: PetscReal smax, smin;
731: PetscInt nits;
732: PetscBool isascii;
734: PetscFunctionBegin;
735: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
736: PetscCall(KSPGetIterationNumber(ksp, &nits));
737: if (!nits) {
738: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any singular values\n"));
739: PetscFunctionReturn(PETSC_SUCCESS);
740: }
741: PetscCall(KSPComputeExtremeSingularValues(ksp, &smax, &smin));
742: if (isascii) PetscCall(PetscViewerASCIIPrintf(viewer, "Iteratively computed extreme %svalues: max %g min %g max/min %g\n", smin < 0 ? "eigen" : "singular ", (double)smax, (double)smin, (double)(smax / smin)));
743: PetscFunctionReturn(PETSC_SUCCESS);
744: }
746: static PetscErrorCode KSPViewFinalResidual_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
747: {
748: PetscBool isascii;
750: PetscFunctionBegin;
751: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
752: PetscCheck(!ksp->dscale || ksp->dscalefix, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Cannot compute final scale with -ksp_diagonal_scale except also with -ksp_diagonal_scale_fix");
753: if (isascii) {
754: Mat A;
755: Vec t;
756: PetscReal norm;
758: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
759: PetscCall(VecDuplicate(ksp->vec_rhs, &t));
760: PetscCall(KSP_MatMult(ksp, A, ksp->vec_sol, t));
761: PetscCall(VecAYPX(t, -1.0, ksp->vec_rhs));
762: PetscCall(PetscOptionsPushCreateViewerOff(PETSC_FALSE));
763: PetscCall(VecViewFromOptions(t, (PetscObject)ksp, "-ksp_view_final_residual_vec"));
764: PetscCall(PetscOptionsPopCreateViewerOff());
765: PetscCall(VecNorm(t, NORM_2, &norm));
766: PetscCall(VecDestroy(&t));
767: PetscCall(PetscViewerASCIIPrintf(viewer, "KSP final norm of residual %g\n", (double)norm));
768: }
769: PetscFunctionReturn(PETSC_SUCCESS);
770: }
772: PETSC_SINGLE_LIBRARY_INTERN PetscErrorCode PetscMonitorPauseFinal_Internal(PetscInt n, PetscCtx ctx[])
773: {
774: PetscFunctionBegin;
775: for (PetscInt i = 0; i < n; ++i) {
776: PetscViewerAndFormat *vf = (PetscViewerAndFormat *)ctx[i];
777: PetscDraw draw;
778: PetscReal lpause;
779: PetscBool isdraw;
781: if (!vf) continue;
782: if (!PetscCheckPointer(vf->viewer, PETSC_OBJECT)) continue;
783: if (((PetscObject)vf->viewer)->classid != PETSC_VIEWER_CLASSID) continue;
784: PetscCall(PetscObjectTypeCompare((PetscObject)vf->viewer, PETSCVIEWERDRAW, &isdraw));
785: if (!isdraw) continue;
787: PetscCall(PetscViewerDrawGetDraw(vf->viewer, 0, &draw));
788: PetscCall(PetscDrawGetPause(draw, &lpause));
789: PetscCall(PetscDrawSetPause(draw, -1.0));
790: PetscCall(PetscDrawPause(draw));
791: PetscCall(PetscDrawSetPause(draw, lpause));
792: }
793: PetscFunctionReturn(PETSC_SUCCESS);
794: }
796: static PetscErrorCode KSPMonitorPauseFinal_Internal(KSP ksp)
797: {
798: PetscFunctionBegin;
799: if (!ksp->pauseFinal) PetscFunctionReturn(PETSC_SUCCESS);
800: PetscCall(PetscMonitorPauseFinal_Internal(ksp->numbermonitors, ksp->monitorcontext));
801: PetscFunctionReturn(PETSC_SUCCESS);
802: }
804: static PetscErrorCode KSPSolve_Private(KSP ksp, Vec b, Vec x)
805: {
806: PetscBool flg = PETSC_FALSE, inXisinB = PETSC_FALSE, guess_zero;
807: Mat mat, pmat;
808: MPI_Comm comm;
809: MatNullSpace nullsp;
810: Vec btmp, vec_rhs = NULL;
812: PetscFunctionBegin;
813: level++;
814: comm = PetscObjectComm((PetscObject)ksp);
815: if (x && x == b) {
816: PetscCheck(ksp->guess_zero, comm, PETSC_ERR_ARG_INCOMP, "Cannot use x == b with nonzero initial guess");
817: PetscCall(VecDuplicate(b, &x));
818: inXisinB = PETSC_TRUE;
819: }
820: if (b) {
821: PetscCall(PetscObjectReference((PetscObject)b));
822: PetscCall(VecDestroy(&ksp->vec_rhs));
823: ksp->vec_rhs = b;
824: }
825: if (x) {
826: PetscCall(PetscObjectReference((PetscObject)x));
827: PetscCall(VecDestroy(&ksp->vec_sol));
828: ksp->vec_sol = x;
829: }
831: if (ksp->viewPre) PetscCall(ObjectView((PetscObject)ksp, ksp->viewerPre, ksp->formatPre));
833: if (ksp->presolve) PetscCall((*ksp->presolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->prectx));
835: /* reset the residual history list if requested */
836: if (ksp->res_hist_reset) ksp->res_hist_len = 0;
837: if (ksp->err_hist_reset) ksp->err_hist_len = 0;
839: /* KSPSetUp() scales the matrix if needed */
840: PetscCall(KSPSetUp(ksp));
841: PetscCall(KSPSetUpOnBlocks(ksp));
843: if (ksp->guess) {
844: PetscObjectState ostate, state;
846: PetscCall(KSPGuessSetUp(ksp->guess));
847: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &ostate));
848: PetscCall(KSPGuessFormGuess(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
849: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &state));
850: if (state != ostate) {
851: ksp->guess_zero = PETSC_FALSE;
852: } else {
853: PetscCall(PetscInfo(ksp, "Using zero initial guess since the KSPGuess object did not change the vector\n"));
854: ksp->guess_zero = PETSC_TRUE;
855: }
856: }
858: PetscCall(VecSetErrorIfLocked(ksp->vec_sol, 3));
860: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
861: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
862: /* diagonal scale RHS if called for */
863: if (ksp->dscale) {
864: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
865: /* second time in, but matrix was scaled back to original */
866: if (ksp->dscalefix && ksp->dscalefix2) {
867: Mat mat, pmat;
869: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
870: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
871: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
872: }
874: /* scale initial guess */
875: if (!ksp->guess_zero) {
876: if (!ksp->truediagonal) {
877: PetscCall(VecDuplicate(ksp->diagonal, &ksp->truediagonal));
878: PetscCall(VecCopy(ksp->diagonal, ksp->truediagonal));
879: PetscCall(VecReciprocal(ksp->truediagonal));
880: }
881: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->truediagonal));
882: }
883: }
884: PetscCall(PCPreSolve(ksp->pc, ksp));
886: if (ksp->guess_zero && !ksp->guess_not_read) PetscCall(VecSet(ksp->vec_sol, 0.0));
887: if (ksp->guess_knoll) { /* The Knoll trick is independent on the KSPGuess specified */
888: PetscCall(PCApply(ksp->pc, ksp->vec_rhs, ksp->vec_sol));
889: PetscCall(KSP_RemoveNullSpace(ksp, ksp->vec_sol));
890: ksp->guess_zero = PETSC_FALSE;
891: }
893: /* can we mark the initial guess as zero for this solve? */
894: guess_zero = ksp->guess_zero;
895: if (!ksp->guess_zero) {
896: PetscReal norm;
898: PetscCall(VecNormAvailable(ksp->vec_sol, NORM_2, &flg, &norm));
899: if (flg && !norm) ksp->guess_zero = PETSC_TRUE;
900: }
901: if (ksp->transpose_solve) {
902: PetscCall(MatGetNullSpace(mat, &nullsp));
903: } else {
904: PetscCall(MatGetTransposeNullSpace(mat, &nullsp));
905: }
906: if (nullsp) {
907: PetscCall(VecDuplicate(ksp->vec_rhs, &btmp));
908: PetscCall(VecCopy(ksp->vec_rhs, btmp));
909: PetscCall(MatNullSpaceRemove(nullsp, btmp));
910: vec_rhs = ksp->vec_rhs;
911: ksp->vec_rhs = btmp;
912: }
913: PetscCall(VecLockReadPush(ksp->vec_rhs));
914: PetscUseTypeMethod(ksp, solve);
915: PetscCall(KSPMonitorPauseFinal_Internal(ksp));
917: PetscCall(VecLockReadPop(ksp->vec_rhs));
918: if (nullsp) {
919: ksp->vec_rhs = vec_rhs;
920: PetscCall(VecDestroy(&btmp));
921: }
923: ksp->guess_zero = guess_zero;
925: PetscCheck(ksp->reason, comm, PETSC_ERR_PLIB, "Internal error, solver returned without setting converged reason");
926: ksp->totalits += ksp->its;
928: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
930: if (ksp->viewRate) {
931: PetscCall(PetscViewerPushFormat(ksp->viewerRate, ksp->formatRate));
932: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
933: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
934: }
935: PetscCall(PCPostSolve(ksp->pc, ksp));
937: /* diagonal scale solution if called for */
938: if (ksp->dscale) {
939: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->diagonal));
940: /* unscale right-hand side and matrix */
941: if (ksp->dscalefix) {
942: Mat mat, pmat;
944: PetscCall(VecReciprocal(ksp->diagonal));
945: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
946: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
947: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
948: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
949: PetscCall(VecReciprocal(ksp->diagonal));
950: ksp->dscalefix2 = PETSC_TRUE;
951: }
952: }
953: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
954: if (ksp->guess) PetscCall(KSPGuessUpdate(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
955: if (ksp->postsolve) PetscCall((*ksp->postsolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->postctx));
957: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
958: if (ksp->viewEV) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_FALSE, ksp->viewerEV, ksp->formatEV));
959: if (ksp->viewEVExp) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_TRUE, ksp->viewerEVExp, ksp->formatEVExp));
960: if (ksp->viewSV) PetscCall(KSPViewSingularvalues_Internal(ksp, ksp->viewerSV, ksp->formatSV));
961: if (ksp->viewFinalRes) PetscCall(KSPViewFinalResidual_Internal(ksp, ksp->viewerFinalRes, ksp->formatFinalRes));
962: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)mat, ksp->viewerMat, ksp->formatMat));
963: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)pmat, ksp->viewerPMat, ksp->formatPMat));
964: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)ksp->vec_rhs, ksp->viewerRhs, ksp->formatRhs));
965: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)ksp->vec_sol, ksp->viewerSol, ksp->formatSol));
966: if (ksp->view) PetscCall(ObjectView((PetscObject)ksp, ksp->viewer, ksp->format));
967: if (ksp->viewDScale) PetscCall(ObjectView((PetscObject)ksp->diagonal, ksp->viewerDScale, ksp->formatDScale));
968: if (ksp->viewMatExp) {
969: Mat A, B;
971: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
972: if (ksp->transpose_solve) {
973: Mat AT;
975: PetscCall(MatCreateTranspose(A, &AT));
976: PetscCall(MatComputeOperator(AT, MATAIJ, &B));
977: PetscCall(MatDestroy(&AT));
978: } else {
979: PetscCall(MatComputeOperator(A, MATAIJ, &B));
980: }
981: PetscCall(ObjectView((PetscObject)B, ksp->viewerMatExp, ksp->formatMatExp));
982: PetscCall(MatDestroy(&B));
983: }
984: if (ksp->viewPOpExp) {
985: Mat B;
987: PetscCall(KSPComputeOperator(ksp, MATAIJ, &B));
988: PetscCall(ObjectView((PetscObject)B, ksp->viewerPOpExp, ksp->formatPOpExp));
989: PetscCall(MatDestroy(&B));
990: }
992: if (inXisinB) {
993: PetscCall(VecCopy(x, b));
994: PetscCall(VecDestroy(&x));
995: }
996: PetscCall(PetscObjectSAWsBlock((PetscObject)ksp));
997: if (ksp->errorifnotconverged && ksp->reason < 0 && ((level == 1) || (ksp->reason != KSP_DIVERGED_ITS))) {
998: PCFailedReason reason;
1000: PetscCheck(ksp->reason == KSP_DIVERGED_PC_FAILED, comm, PETSC_ERR_NOT_CONVERGED, "KSPSolve%s() has not converged, reason %s", !ksp->transpose_solve ? "" : "Transpose", KSPConvergedReasons[ksp->reason]);
1001: PetscCall(PCGetFailedReason(ksp->pc, &reason));
1002: SETERRQ(comm, PETSC_ERR_NOT_CONVERGED, "KSPSolve%s() has not converged, reason %s PC failed due to %s", !ksp->transpose_solve ? "" : "Transpose", KSPConvergedReasons[ksp->reason], PCFailedReasons[reason]);
1003: }
1004: level--;
1005: PetscFunctionReturn(PETSC_SUCCESS);
1006: }
1008: /*@
1009: KSPSolve - Solves a linear system associated with `KSP` object
1011: Collective
1013: Input Parameters:
1014: + ksp - iterative solver obtained from `KSPCreate()`
1015: . b - the right-hand side vector
1016: - x - the solution (this may be the same vector as `b`, then `b` will be overwritten with the answer)
1018: Options Database Keys:
1019: + -ksp_view_eigenvalues - compute preconditioned operators eigenvalues
1020: . -ksp_view_eigenvalues_explicit - compute the eigenvalues by forming the dense operator and using LAPACK
1021: . -ksp_view_mat binary - save matrix to the default binary viewer
1022: . -ksp_view_pmat binary - save matrix used to build preconditioner to the default binary viewer
1023: . -ksp_view_rhs binary - save right-hand side vector to the default binary viewer
1024: . -ksp_view_solution binary - save computed solution vector to the default binary viewer
1025: (can be read later with src/ksp/tutorials/ex10.c for testing solvers)
1026: . -ksp_view_mat_explicit - for matrix-free operators, computes the matrix entries and views them
1027: . -ksp_view_preconditioned_operator_explicit - computes the product of the preconditioner and matrix as an explicit matrix and views it
1028: . -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
1029: . -ksp_view_final_residual - print 2-norm of true linear system residual at the end of the solution process
1030: . -ksp_view_final_residual_vec - print true linear system residual vector at the end of the solution process;
1031: `-ksp_view_final_residual` must to be called first to enable this option
1032: . -ksp_error_if_not_converged - stop the program as soon as an error is detected in a `KSPSolve()`
1033: . -ksp_view_pre - print the ksp data structure before the system solution
1034: - -ksp_view - print the ksp data structure at the end of the system solution
1036: Level: beginner
1038: Notes:
1039: See `KSPSetFromOptions()` for additional options database keys that affect `KSPSolve()`
1041: If one uses `KSPSetDM()` then `x` or `b` need not be passed. Use `KSPGetSolution()` to access the solution in this case.
1043: The operator is specified with `KSPSetOperators()`.
1045: `KSPSolve()` will normally return without generating an error regardless of whether the linear system was solved or if constructing the preconditioner failed.
1046: Call `KSPGetConvergedReason()` to determine if the solver converged or failed and why. The option -ksp_error_if_not_converged or function `KSPSetErrorIfNotConverged()`
1047: will cause `KSPSolve()` to error as soon as an error occurs in the linear solver. In inner `KSPSolve()` `KSP_DIVERGED_ITS` is not treated as an error because when using nested solvers
1048: it may be fine that inner solvers in the preconditioner do not converge during the solution process.
1050: The number of iterations can be obtained from `KSPGetIterationNumber()`.
1052: If you provide a matrix that has a `MatSetNullSpace()` and `MatSetTransposeNullSpace()` this will use that information to solve singular systems
1053: in the least squares sense with a norm minimizing solution.
1055: $A x = b $ where $b = b_p + b_t$ where $b_t$ is not in the range of $A$ (and hence by the fundamental theorem of linear algebra is in the nullspace(A'), see `MatSetNullSpace()`).
1057: `KSP` first removes $b_t$ producing the linear system $A x = b_p$ (which has multiple solutions) and solves this to find the $\|x\|$ minimizing solution (and hence
1058: it finds the solution $x$ orthogonal to the nullspace(A). The algorithm is simply in each iteration of the Krylov method we remove the nullspace(A) from the search
1059: direction thus the solution which is a linear combination of the search directions has no component in the nullspace(A).
1061: We recommend always using `KSPGMRES` for such singular systems.
1062: If $ nullspace(A) = nullspace(A^T)$ (note symmetric matrices always satisfy this property) then both left and right preconditioning will work
1063: If $nullspace(A) \neq nullspace(A^T)$ then left preconditioning will work but right preconditioning may not work (or it may).
1065: Developer Notes:
1066: The reason we cannot always solve $nullspace(A) \neq nullspace(A^T)$ systems with right preconditioning is because we need to remove at each iteration
1067: $ nullspace(AB) $ from the search direction. While we know the $nullspace(A)$, $nullspace(AB)$ equals $B^{-1}$ times $nullspace(A)$ but except for trivial preconditioners
1068: such as diagonal scaling we cannot apply the inverse of the preconditioner to a vector and thus cannot compute $nullspace(AB)$.
1070: If using a direct method (e.g., via the `KSP` solver
1071: `KSPPREONLY` and a preconditioner such as `PCLU` or `PCCHOLESKY` then usually one iteration of the `KSP` method will be needed for convergence.
1073: To solve a linear system with the transpose of the matrix use `KSPSolveTranspose()`.
1075: Understanding Convergence\:
1076: The manual pages `KSPMonitorSet()`, `KSPComputeEigenvalues()`, and
1077: `KSPComputeEigenvaluesExplicitly()` provide information on additional
1078: options to monitor convergence and print eigenvalue information.
1080: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1081: `KSPSolveTranspose()`, `KSPGetIterationNumber()`, `MatNullSpaceCreate()`, `MatSetNullSpace()`, `MatSetTransposeNullSpace()`, `KSP`,
1082: `KSPConvergedReasonView()`, `KSPCheckSolve()`, `KSPSetErrorIfNotConverged()`
1083: @*/
1084: PetscErrorCode KSPSolve(KSP ksp, Vec b, Vec x)
1085: {
1086: PetscBool isPCMPI;
1088: PetscFunctionBegin;
1092: ksp->transpose_solve = PETSC_FALSE;
1093: PetscCall(KSPSolve_Private(ksp, b, x));
1094: PetscCall(PetscObjectTypeCompare((PetscObject)ksp->pc, PCMPI, &isPCMPI));
1095: if (PCMPIServerActive && isPCMPI) {
1096: KSP subksp;
1098: PetscCall(PCMPIGetKSP(ksp->pc, &subksp));
1099: ksp->its = subksp->its;
1100: ksp->reason = subksp->reason;
1101: }
1102: PetscFunctionReturn(PETSC_SUCCESS);
1103: }
1105: static PetscErrorCode KSPUseExplicitTranspose_Private(KSP ksp)
1106: {
1107: Mat J, Jpre;
1109: PetscFunctionBegin;
1110: PetscCall(KSPGetOperators(ksp, &J, &Jpre));
1111: if (!ksp->transpose.reuse_transpose) {
1112: PetscCall(MatTranspose(J, MAT_INITIAL_MATRIX, &ksp->transpose.AT));
1113: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_INITIAL_MATRIX, &ksp->transpose.BT));
1114: ksp->transpose.reuse_transpose = PETSC_TRUE;
1115: } else {
1116: PetscCall(MatTranspose(J, MAT_REUSE_MATRIX, &ksp->transpose.AT));
1117: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_REUSE_MATRIX, &ksp->transpose.BT));
1118: }
1119: if (J == Jpre && ksp->transpose.BT != ksp->transpose.AT) {
1120: PetscCall(PetscObjectReference((PetscObject)ksp->transpose.AT));
1121: ksp->transpose.BT = ksp->transpose.AT;
1122: }
1123: PetscCall(KSPSetOperators(ksp, ksp->transpose.AT, ksp->transpose.BT));
1124: PetscFunctionReturn(PETSC_SUCCESS);
1125: }
1127: /*@
1128: KSPSolveTranspose - Solves a linear system with the transpose of the matrix associated with the `KSP` object, $A^T x = b$.
1130: Collective
1132: Input Parameters:
1133: + ksp - iterative solver obtained from `KSPCreate()`
1134: . b - right-hand side vector
1135: - x - solution vector
1137: Level: developer
1139: Note:
1140: For complex numbers, this solve the non-Hermitian transpose system.
1142: Developer Note:
1143: We need to implement a `KSPSolveHermitianTranspose()`
1145: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1146: `KSPSolve()`, `KSP`, `KSPSetOperators()`
1147: @*/
1148: PetscErrorCode KSPSolveTranspose(KSP ksp, Vec b, Vec x)
1149: {
1150: PetscFunctionBegin;
1154: if (ksp->transpose.use_explicittranspose) {
1155: Mat J, Jpre;
1156: PetscCall(KSPGetOperators(ksp, &J, &Jpre));
1157: if (!ksp->transpose.reuse_transpose) {
1158: PetscCall(MatTranspose(J, MAT_INITIAL_MATRIX, &ksp->transpose.AT));
1159: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_INITIAL_MATRIX, &ksp->transpose.BT));
1160: ksp->transpose.reuse_transpose = PETSC_TRUE;
1161: } else {
1162: PetscCall(MatTranspose(J, MAT_REUSE_MATRIX, &ksp->transpose.AT));
1163: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_REUSE_MATRIX, &ksp->transpose.BT));
1164: }
1165: if (J == Jpre && ksp->transpose.BT != ksp->transpose.AT) {
1166: PetscCall(PetscObjectReference((PetscObject)ksp->transpose.AT));
1167: ksp->transpose.BT = ksp->transpose.AT;
1168: }
1169: PetscCall(KSPSetOperators(ksp, ksp->transpose.AT, ksp->transpose.BT));
1170: } else {
1171: ksp->transpose_solve = PETSC_TRUE;
1172: }
1173: PetscCall(KSPSolve_Private(ksp, b, x));
1174: PetscFunctionReturn(PETSC_SUCCESS);
1175: }
1177: static PetscErrorCode KSPViewFinalMatResidual_Internal(KSP ksp, Mat B, Mat X, PetscViewer viewer, PetscViewerFormat format, PetscInt shift)
1178: {
1179: Mat A, R;
1180: PetscReal *norms;
1181: PetscInt i, N;
1182: PetscBool flg;
1184: PetscFunctionBegin;
1185: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &flg));
1186: if (flg) {
1187: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
1188: if (!ksp->transpose_solve) PetscCall(MatMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DETERMINE, &R));
1189: else PetscCall(MatTransposeMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DETERMINE, &R));
1190: PetscCall(MatAYPX(R, -1.0, B, SAME_NONZERO_PATTERN));
1191: PetscCall(MatGetSize(R, NULL, &N));
1192: PetscCall(PetscMalloc1(N, &norms));
1193: PetscCall(MatGetColumnNorms(R, NORM_2, norms));
1194: PetscCall(MatDestroy(&R));
1195: for (i = 0; i < N; ++i) PetscCall(PetscViewerASCIIPrintf(viewer, "%s #%" PetscInt_FMT " %g\n", i == 0 ? "KSP final norm of residual" : " ", shift + i, (double)norms[i]));
1196: PetscCall(PetscFree(norms));
1197: }
1198: PetscFunctionReturn(PETSC_SUCCESS);
1199: }
1201: static PetscErrorCode KSPMatSolve_Private(KSP ksp, Mat B, Mat X)
1202: {
1203: Mat A, P, vB, vX;
1204: Vec cb, cx;
1205: PetscInt n1, N1, n2, N2, Bbn = PETSC_DECIDE;
1206: PetscBool match;
1208: PetscFunctionBegin;
1212: PetscCheckSameComm(ksp, 1, B, 2);
1213: PetscCheckSameComm(ksp, 1, X, 3);
1214: PetscCheckSameType(B, 2, X, 3);
1215: PetscCheck(B->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
1216: MatCheckPreallocated(X, 3);
1217: if (!X->assembled) {
1218: PetscCall(MatSetOption(X, MAT_NO_OFF_PROC_ENTRIES, PETSC_TRUE));
1219: PetscCall(MatAssemblyBegin(X, MAT_FINAL_ASSEMBLY));
1220: PetscCall(MatAssemblyEnd(X, MAT_FINAL_ASSEMBLY));
1221: }
1222: PetscCheck(B != X, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_IDN, "B and X must be different matrices");
1223: PetscCall(KSPGetOperators(ksp, &A, &P));
1224: PetscCall(MatGetLocalSize(B, NULL, &n2));
1225: PetscCall(MatGetLocalSize(X, NULL, &n1));
1226: PetscCall(MatGetSize(B, NULL, &N2));
1227: PetscCall(MatGetSize(X, NULL, &N1));
1228: PetscCheck(n1 == n2 && N1 == N2, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible number of columns between block of right-hand sides (n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ") and block of solutions (n,N) = (%" PetscInt_FMT ",%" PetscInt_FMT ")", n2, N2, n1, N1);
1229: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)B, &match, MATSEQDENSE, MATMPIDENSE, ""));
1230: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of right-hand sides not stored in a dense Mat");
1231: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)X, &match, MATSEQDENSE, MATMPIDENSE, ""));
1232: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of solutions not stored in a dense Mat");
1233: PetscCall(KSPSetUp(ksp));
1234: PetscCall(KSPSetUpOnBlocks(ksp));
1235: if (ksp->ops->matsolve) {
1236: level++;
1237: if (ksp->guess_zero) PetscCall(MatZeroEntries(X));
1238: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1239: PetscCall(KSPGetMatSolveBatchSize(ksp, &Bbn));
1240: /* by default, do a single solve with all columns */
1241: if (Bbn == PETSC_DECIDE) Bbn = N2;
1242: else PetscCheck(Bbn >= 1, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "KSPMatSolve() batch size %" PetscInt_FMT " must be positive", Bbn);
1243: PetscCall(PetscInfo(ksp, "KSP type %s%s solving using batches of width at most %" PetscInt_FMT "\n", ((PetscObject)ksp)->type_name, ksp->transpose_solve ? " transpose" : "", Bbn));
1244: /* if -ksp_matsolve_batch_size is greater than the actual number of columns, do a single solve with all columns */
1245: if (Bbn >= N2) {
1246: PetscUseTypeMethod(ksp, matsolve, B, X);
1247: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, B, X, ksp->viewerFinalRes, ksp->formatFinalRes, 0));
1249: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1251: if (ksp->viewRate) {
1252: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1253: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1254: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1255: }
1256: } else {
1257: for (n2 = 0; n2 < N2; n2 += Bbn) {
1258: PetscCall(MatDenseGetSubMatrix(B, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vB));
1259: PetscCall(MatDenseGetSubMatrix(X, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vX));
1260: PetscUseTypeMethod(ksp, matsolve, vB, vX);
1261: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, vB, vX, ksp->viewerFinalRes, ksp->formatFinalRes, n2));
1263: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1265: if (ksp->viewRate) {
1266: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1267: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1268: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1269: }
1270: PetscCall(MatDenseRestoreSubMatrix(B, &vB));
1271: PetscCall(MatDenseRestoreSubMatrix(X, &vX));
1272: }
1273: }
1274: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)A, ksp->viewerMat, ksp->formatMat));
1275: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)P, ksp->viewerPMat, ksp->formatPMat));
1276: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)B, ksp->viewerRhs, ksp->formatRhs));
1277: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)X, ksp->viewerSol, ksp->formatSol));
1278: if (ksp->view) PetscCall(KSPView(ksp, ksp->viewer));
1279: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1280: if (ksp->errorifnotconverged && ksp->reason < 0 && (level == 1 || ksp->reason != KSP_DIVERGED_ITS)) {
1281: PCFailedReason reason;
1283: PetscCheck(ksp->reason == KSP_DIVERGED_PC_FAILED, PetscObjectComm((PetscObject)ksp), PETSC_ERR_NOT_CONVERGED, "KSPMatSolve%s() has not converged, reason %s", !ksp->transpose_solve ? "" : "Transpose", KSPConvergedReasons[ksp->reason]);
1284: PetscCall(PCGetFailedReason(ksp->pc, &reason));
1285: SETERRQ(PetscObjectComm((PetscObject)ksp), PETSC_ERR_NOT_CONVERGED, "KSPMatSolve%s() has not converged, reason %s PC failed due to %s", !ksp->transpose_solve ? "" : "Transpose", KSPConvergedReasons[ksp->reason], PCFailedReasons[reason]);
1286: }
1287: level--;
1288: } else {
1289: PetscCall(PetscInfo(ksp, "KSP type %s solving column by column\n", ((PetscObject)ksp)->type_name));
1290: for (n2 = 0; n2 < N2; ++n2) {
1291: PetscCall(MatDenseGetColumnVecRead(B, n2, &cb));
1292: PetscCall(MatDenseGetColumnVecWrite(X, n2, &cx));
1293: PetscCall(KSPSolve_Private(ksp, cb, cx));
1294: PetscCall(MatDenseRestoreColumnVecWrite(X, n2, &cx));
1295: PetscCall(MatDenseRestoreColumnVecRead(B, n2, &cb));
1296: }
1297: }
1298: PetscFunctionReturn(PETSC_SUCCESS);
1299: }
1301: /*@
1302: KSPMatSolve - Solves a linear system with multiple right-hand sides stored as a `MATDENSE`.
1304: Input Parameters:
1305: + ksp - iterative solver
1306: - B - block of right-hand sides
1308: Output Parameter:
1309: . X - block of solutions
1311: Level: intermediate
1313: Notes:
1314: This is a stripped-down version of `KSPSolve()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1316: Unlike with `KSPSolve()`, `B` and `X` must be different matrices.
1318: .seealso: [](ch_ksp), `KSPSolve()`, `MatMatSolve()`, `KSPMatSolveTranspose()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`, `KSPSetMatSolveBatchSize()`
1319: @*/
1320: PetscErrorCode KSPMatSolve(KSP ksp, Mat B, Mat X)
1321: {
1322: PetscFunctionBegin;
1323: ksp->transpose_solve = PETSC_FALSE;
1324: PetscCall(KSPMatSolve_Private(ksp, B, X));
1325: PetscFunctionReturn(PETSC_SUCCESS);
1326: }
1328: /*@
1329: KSPMatSolveTranspose - Solves a linear system with the transposed matrix with multiple right-hand sides stored as a `MATDENSE`.
1331: Input Parameters:
1332: + ksp - iterative solver
1333: - B - block of right-hand sides
1335: Output Parameter:
1336: . X - block of solutions
1338: Level: intermediate
1340: Notes:
1341: This is a stripped-down version of `KSPSolveTranspose()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1343: Unlike `KSPSolveTranspose()`,
1344: `B` and `X` must be different matrices and the transposed matrix cannot be assembled explicitly for the user.
1346: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `MatMatTransposeSolve()`, `KSPMatSolve()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`
1347: @*/
1348: PetscErrorCode KSPMatSolveTranspose(KSP ksp, Mat B, Mat X)
1349: {
1350: PetscFunctionBegin;
1351: if (ksp->transpose.use_explicittranspose) PetscCall(KSPUseExplicitTranspose_Private(ksp));
1352: else ksp->transpose_solve = PETSC_TRUE;
1353: PetscCall(KSPMatSolve_Private(ksp, B, X));
1354: PetscFunctionReturn(PETSC_SUCCESS);
1355: }
1357: /*@
1358: KSPSetMatSolveBatchSize - Sets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1360: Logically Collective
1362: Input Parameters:
1363: + ksp - the `KSP` iterative solver
1364: - bs - batch size
1366: Level: advanced
1368: Note:
1369: Using a larger block size can improve the efficiency of the solver.
1371: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPGetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matproduct_batch_size`
1372: @*/
1373: PetscErrorCode KSPSetMatSolveBatchSize(KSP ksp, PetscInt bs)
1374: {
1375: PetscFunctionBegin;
1378: ksp->nmax = bs;
1379: PetscFunctionReturn(PETSC_SUCCESS);
1380: }
1382: /*@
1383: KSPGetMatSolveBatchSize - Gets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1385: Input Parameter:
1386: . ksp - iterative solver context
1388: Output Parameter:
1389: . bs - batch size
1391: Level: advanced
1393: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPSetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matproduct_batch_size`
1394: @*/
1395: PetscErrorCode KSPGetMatSolveBatchSize(KSP ksp, PetscInt *bs)
1396: {
1397: PetscFunctionBegin;
1399: PetscAssertPointer(bs, 2);
1400: *bs = ksp->nmax;
1401: PetscFunctionReturn(PETSC_SUCCESS);
1402: }
1404: /*@
1405: KSPResetViewers - Resets all the viewers set from the options database during `KSPSetFromOptions()`
1407: Collective
1409: Input Parameter:
1410: . ksp - the `KSP` iterative solver context obtained from `KSPCreate()`
1412: Level: beginner
1414: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSPSetFromOptions()`, `KSP`
1415: @*/
1416: PetscErrorCode KSPResetViewers(KSP ksp)
1417: {
1418: PetscFunctionBegin;
1420: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1421: PetscCall(PetscViewerDestroy(&ksp->viewer));
1422: PetscCall(PetscViewerDestroy(&ksp->viewerPre));
1423: PetscCall(PetscViewerDestroy(&ksp->viewerRate));
1424: PetscCall(PetscViewerDestroy(&ksp->viewerMat));
1425: PetscCall(PetscViewerDestroy(&ksp->viewerPMat));
1426: PetscCall(PetscViewerDestroy(&ksp->viewerRhs));
1427: PetscCall(PetscViewerDestroy(&ksp->viewerSol));
1428: PetscCall(PetscViewerDestroy(&ksp->viewerMatExp));
1429: PetscCall(PetscViewerDestroy(&ksp->viewerEV));
1430: PetscCall(PetscViewerDestroy(&ksp->viewerSV));
1431: PetscCall(PetscViewerDestroy(&ksp->viewerEVExp));
1432: PetscCall(PetscViewerDestroy(&ksp->viewerFinalRes));
1433: PetscCall(PetscViewerDestroy(&ksp->viewerPOpExp));
1434: PetscCall(PetscViewerDestroy(&ksp->viewerDScale));
1435: ksp->view = PETSC_FALSE;
1436: ksp->viewPre = PETSC_FALSE;
1437: ksp->viewMat = PETSC_FALSE;
1438: ksp->viewPMat = PETSC_FALSE;
1439: ksp->viewRhs = PETSC_FALSE;
1440: ksp->viewSol = PETSC_FALSE;
1441: ksp->viewMatExp = PETSC_FALSE;
1442: ksp->viewEV = PETSC_FALSE;
1443: ksp->viewSV = PETSC_FALSE;
1444: ksp->viewEVExp = PETSC_FALSE;
1445: ksp->viewFinalRes = PETSC_FALSE;
1446: ksp->viewPOpExp = PETSC_FALSE;
1447: ksp->viewDScale = PETSC_FALSE;
1448: PetscFunctionReturn(PETSC_SUCCESS);
1449: }
1451: /*@
1452: KSPReset - Removes any allocated `Vec` and `Mat` from the `KSP` data structures.
1454: Collective
1456: Input Parameter:
1457: . ksp - iterative solver obtained from `KSPCreate()`
1459: Level: intermediate
1461: Notes:
1462: Any options set in the `KSP`, including those set with `KSPSetFromOptions()` remain.
1464: Call `KSPReset()` only before you call `KSPSetOperators()` with a different sized matrix than the previous matrix used with the `KSP`.
1466: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1467: @*/
1468: PetscErrorCode KSPReset(KSP ksp)
1469: {
1470: PetscFunctionBegin;
1472: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1473: PetscTryTypeMethod(ksp, reset);
1474: if (ksp->pc) PetscCall(PCReset(ksp->pc));
1475: if (ksp->guess) {
1476: KSPGuess guess = ksp->guess;
1477: PetscTryTypeMethod(guess, reset);
1478: }
1479: PetscCall(VecDestroyVecs(ksp->nwork, &ksp->work));
1480: PetscCall(VecDestroy(&ksp->vec_rhs));
1481: PetscCall(VecDestroy(&ksp->vec_sol));
1482: PetscCall(VecDestroy(&ksp->diagonal));
1483: PetscCall(VecDestroy(&ksp->truediagonal));
1485: ksp->setupstage = KSP_SETUP_NEW;
1486: ksp->nmax = PETSC_DECIDE;
1487: PetscFunctionReturn(PETSC_SUCCESS);
1488: }
1490: /*@
1491: KSPDestroy - Destroys a `KSP` context.
1493: Collective
1495: Input Parameter:
1496: . ksp - iterative solver obtained from `KSPCreate()`
1498: Level: beginner
1500: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1501: @*/
1502: PetscErrorCode KSPDestroy(KSP *ksp)
1503: {
1504: PC pc;
1506: PetscFunctionBegin;
1507: if (!*ksp) PetscFunctionReturn(PETSC_SUCCESS);
1509: if (--((PetscObject)*ksp)->refct > 0) {
1510: *ksp = NULL;
1511: PetscFunctionReturn(PETSC_SUCCESS);
1512: }
1514: PetscCall(PetscObjectSAWsViewOff((PetscObject)*ksp));
1516: /*
1517: Avoid a cascading call to PCReset(ksp->pc) from the following call:
1518: PCReset() shouldn't be called from KSPDestroy() as it is unprotected by pc's
1519: refcount (and may be shared, e.g., by other ksps).
1520: */
1521: pc = (*ksp)->pc;
1522: (*ksp)->pc = NULL;
1523: PetscCall(KSPReset(*ksp));
1524: PetscCall(KSPResetViewers(*ksp));
1525: (*ksp)->pc = pc;
1526: PetscTryTypeMethod(*ksp, destroy);
1528: if ((*ksp)->transpose.use_explicittranspose) {
1529: PetscCall(MatDestroy(&(*ksp)->transpose.AT));
1530: PetscCall(MatDestroy(&(*ksp)->transpose.BT));
1531: (*ksp)->transpose.reuse_transpose = PETSC_FALSE;
1532: }
1534: PetscCall(KSPGuessDestroy(&(*ksp)->guess));
1535: PetscCall(DMDestroy(&(*ksp)->dm));
1536: PetscCall(PCDestroy(&(*ksp)->pc));
1537: PetscCall(PetscFree((*ksp)->res_hist_alloc));
1538: PetscCall(PetscFree((*ksp)->err_hist_alloc));
1539: if ((*ksp)->convergeddestroy) PetscCall((*(*ksp)->convergeddestroy)(&(*ksp)->cnvP));
1540: PetscCall(KSPMonitorCancel(*ksp));
1541: PetscCall(KSPConvergedReasonViewCancel(*ksp));
1542: PetscCall(PetscHeaderDestroy(ksp));
1543: PetscFunctionReturn(PETSC_SUCCESS);
1544: }
1546: /*@
1547: KSPSetPCSide - Sets the preconditioning side.
1549: Logically Collective
1551: Input Parameter:
1552: . ksp - iterative solver obtained from `KSPCreate()`
1554: Output Parameter:
1555: . side - the preconditioning side, where side is one of
1556: .vb
1557: PC_LEFT - left preconditioning (default)
1558: PC_RIGHT - right preconditioning
1559: PC_SYMMETRIC - symmetric preconditioning
1560: .ve
1562: Options Database Key:
1563: . -ksp_pc_side <right,left,symmetric> - `KSP` preconditioner side
1565: Level: intermediate
1567: Notes:
1568: Left preconditioning is used by default for most Krylov methods except `KSPFGMRES` which only supports right preconditioning.
1570: For methods changing the side of the preconditioner changes the norm type that is used, see `KSPSetNormType()`.
1572: Symmetric preconditioning is currently available only for the `KSPQCG` method. However, note that
1573: symmetric preconditioning can be emulated by using either right or left
1574: preconditioning, modifying the application of the matrix (with a custom `Mat` argument to `KSPSetOperators()`,
1575: and using a pre 'KSPSetPreSolve()` or post processing `KSPSetPostSolve()` step).
1577: Setting the `PCSide` often affects the default norm type. See `KSPSetNormType()` for details.
1579: .seealso: [](ch_ksp), `KSPGetPCSide()`, `KSPSetNormType()`, `KSPGetNormType()`, `KSP`, `KSPSetPreSolve()`, `KSPSetPostSolve()`
1580: @*/
1581: PetscErrorCode KSPSetPCSide(KSP ksp, PCSide side)
1582: {
1583: PetscFunctionBegin;
1586: ksp->pc_side = ksp->pc_side_set = side;
1587: PetscFunctionReturn(PETSC_SUCCESS);
1588: }
1590: /*@
1591: KSPGetPCSide - Gets the preconditioning side.
1593: Not Collective
1595: Input Parameter:
1596: . ksp - iterative solver obtained from `KSPCreate()`
1598: Output Parameter:
1599: . side - the preconditioning side, where side is one of
1600: .vb
1601: PC_LEFT - left preconditioning (default)
1602: PC_RIGHT - right preconditioning
1603: PC_SYMMETRIC - symmetric preconditioning
1604: .ve
1606: Level: intermediate
1608: .seealso: [](ch_ksp), `KSPSetPCSide()`, `KSP`
1609: @*/
1610: PetscErrorCode KSPGetPCSide(KSP ksp, PCSide *side)
1611: {
1612: PetscFunctionBegin;
1614: PetscAssertPointer(side, 2);
1615: PetscCall(KSPSetUpNorms_Private(ksp, PETSC_TRUE, &ksp->normtype, &ksp->pc_side));
1616: *side = ksp->pc_side;
1617: PetscFunctionReturn(PETSC_SUCCESS);
1618: }
1620: /*@
1621: KSPGetTolerances - Gets the relative, absolute, divergence, and maximum
1622: iteration tolerances used by the default `KSP` convergence tests.
1624: Not Collective
1626: Input Parameter:
1627: . ksp - the Krylov subspace context
1629: Output Parameters:
1630: + rtol - the relative convergence tolerance
1631: . abstol - the absolute convergence tolerance
1632: . dtol - the divergence tolerance
1633: - maxits - maximum number of iterations
1635: Level: intermediate
1637: Note:
1638: The user can specify `NULL` for any parameter that is not needed.
1640: .seealso: [](ch_ksp), `KSPSetTolerances()`, `KSP`, `KSPSetMinimumIterations()`, `KSPGetMinimumIterations()`
1641: @*/
1642: PetscErrorCode KSPGetTolerances(KSP ksp, PeOp PetscReal *rtol, PeOp PetscReal *abstol, PeOp PetscReal *dtol, PeOp PetscInt *maxits)
1643: {
1644: PetscFunctionBegin;
1646: if (abstol) *abstol = ksp->abstol;
1647: if (rtol) *rtol = ksp->rtol;
1648: if (dtol) *dtol = ksp->divtol;
1649: if (maxits) *maxits = ksp->max_it;
1650: PetscFunctionReturn(PETSC_SUCCESS);
1651: }
1653: /*@
1654: KSPSetTolerances - Sets the relative, absolute, divergence, and maximum
1655: iteration tolerances used by the default `KSP` convergence testers.
1657: Logically Collective
1659: Input Parameters:
1660: + ksp - the Krylov subspace context
1661: . rtol - the relative convergence tolerance, relative decrease in the (possibly preconditioned) residual norm
1662: . abstol - the absolute convergence tolerance absolute size of the (possibly preconditioned) residual norm
1663: . dtol - the divergence tolerance, amount (possibly preconditioned) residual norm can increase before `KSPConvergedDefault()` concludes that the method is diverging
1664: - maxits - maximum number of iterations to use
1666: Options Database Keys:
1667: + -ksp_atol <abstol> - Sets `abstol`
1668: . -ksp_rtol <rtol> - Sets `rtol`
1669: . -ksp_divtol <dtol> - Sets `dtol`
1670: - -ksp_max_it <maxits> - Sets `maxits`
1672: Level: intermediate
1674: Notes:
1675: The tolerances are with respect to a norm of the residual of the equation $ \| b - A x^n \|$, they do not directly use the error of the equation.
1676: The norm used depends on the `KSPNormType` that has been set with `KSPSetNormType()`, the default depends on the `KSPType` used.
1678: All parameters must be non-negative.
1680: Use `PETSC_CURRENT` to retain the current value of any of the parameters. The deprecated `PETSC_DEFAULT` also retains the current value (though the name is confusing).
1682: Use `PETSC_DETERMINE` to use the default value for the given `KSP`. The default value is the value when the object's type is set.
1684: For `dtol` and `maxits` use `PETSC_UNLIMITED` to indicate there is no upper bound on these values
1686: See `KSPConvergedDefault()` for details how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1687: for setting user-defined stopping criteria.
1689: Fortran Note:
1690: Use `PETSC_CURRENT_INTEGER`, `PETSC_CURRENT_REAL`, `PETSC_DETERMINE_INTEGER`, or `PETSC_DETERMINE_REAL`
1692: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetMinimumIterations()`
1693: @*/
1694: PetscErrorCode KSPSetTolerances(KSP ksp, PetscReal rtol, PetscReal abstol, PetscReal dtol, PetscInt maxits)
1695: {
1696: PetscFunctionBegin;
1703: if (rtol == (PetscReal)PETSC_DETERMINE) {
1704: ksp->rtol = ksp->default_rtol;
1705: } else if (rtol != (PetscReal)PETSC_CURRENT) {
1706: PetscCheck(rtol >= 0.0 && rtol < 1.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Relative tolerance %g must be non-negative and less than 1.0", (double)rtol);
1707: ksp->rtol = rtol;
1708: }
1709: if (abstol == (PetscReal)PETSC_DETERMINE) {
1710: ksp->abstol = ksp->default_abstol;
1711: } else if (abstol != (PetscReal)PETSC_CURRENT) {
1712: PetscCheck(abstol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Absolute tolerance %g must be non-negative", (double)abstol);
1713: ksp->abstol = abstol;
1714: }
1715: if (dtol == (PetscReal)PETSC_DETERMINE) {
1716: ksp->divtol = ksp->default_divtol;
1717: } else if (dtol == (PetscReal)PETSC_UNLIMITED) {
1718: ksp->divtol = PETSC_MAX_REAL;
1719: } else if (dtol != (PetscReal)PETSC_CURRENT) {
1720: PetscCheck(dtol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Divergence tolerance %g must be larger than 1.0", (double)dtol);
1721: ksp->divtol = dtol;
1722: }
1723: if (maxits == PETSC_DETERMINE) {
1724: ksp->max_it = ksp->default_max_it;
1725: } else if (maxits == PETSC_UNLIMITED) {
1726: ksp->max_it = PETSC_INT_MAX;
1727: } else if (maxits != PETSC_CURRENT) {
1728: PetscCheck(maxits >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Maximum number of iterations %" PetscInt_FMT " must be non-negative", maxits);
1729: ksp->max_it = maxits;
1730: }
1731: PetscFunctionReturn(PETSC_SUCCESS);
1732: }
1734: /*@
1735: KSPSetMinimumIterations - Sets the minimum number of iterations to use, regardless of the tolerances
1737: Logically Collective
1739: Input Parameters:
1740: + ksp - the Krylov subspace context
1741: - minit - minimum number of iterations to use
1743: Options Database Key:
1744: . -ksp_min_it <minits> - Sets `minit`
1746: Level: intermediate
1748: Notes:
1749: Use `KSPSetTolerances()` to set a variety of other tolerances
1751: See `KSPConvergedDefault()` for details on how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1752: for setting user-defined stopping criteria.
1754: If the initial residual norm is small enough solvers may return immediately without computing any improvement to the solution. Using this routine
1755: prevents that which usually ensures the solution is changed (often minimally) from the previous solution. This option may be used with ODE integrators
1756: to ensure the integrator does not fall into a false steady-state solution of the ODE.
1758: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPGetMinimumIterations()`
1759: @*/
1760: PetscErrorCode KSPSetMinimumIterations(KSP ksp, PetscInt minit)
1761: {
1762: PetscFunctionBegin;
1766: PetscCheck(minit >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Minimum number of iterations %" PetscInt_FMT " must be non-negative", minit);
1767: ksp->min_it = minit;
1768: PetscFunctionReturn(PETSC_SUCCESS);
1769: }
1771: /*@
1772: KSPGetMinimumIterations - Gets the minimum number of iterations to use, regardless of the tolerances, that was set with `KSPSetMinimumIterations()` or `-ksp_min_it`
1774: Not Collective
1776: Input Parameter:
1777: . ksp - the Krylov subspace context
1779: Output Parameter:
1780: . minit - minimum number of iterations to use
1782: Level: intermediate
1784: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPSetMinimumIterations()`
1785: @*/
1786: PetscErrorCode KSPGetMinimumIterations(KSP ksp, PetscInt *minit)
1787: {
1788: PetscFunctionBegin;
1790: PetscAssertPointer(minit, 2);
1792: *minit = ksp->min_it;
1793: PetscFunctionReturn(PETSC_SUCCESS);
1794: }
1796: /*@
1797: KSPSetInitialGuessNonzero - Tells the iterative solver that the
1798: initial guess is nonzero; otherwise `KSP` assumes the initial guess
1799: is to be zero (and thus zeros it out before solving).
1801: Logically Collective
1803: Input Parameters:
1804: + ksp - iterative solver obtained from `KSPCreate()`
1805: - flg - ``PETSC_TRUE`` indicates the guess is non-zero, `PETSC_FALSE` indicates the guess is zero
1807: Options Database Key:
1808: . -ksp_initial_guess_nonzero <true,false> - use nonzero initial guess
1810: Level: beginner
1812: .seealso: [](ch_ksp), `KSPGetInitialGuessNonzero()`, `KSPGuessSetType()`, `KSPGuessType`, `KSP`
1813: @*/
1814: PetscErrorCode KSPSetInitialGuessNonzero(KSP ksp, PetscBool flg)
1815: {
1816: PetscFunctionBegin;
1819: ksp->guess_zero = (PetscBool)!flg;
1820: PetscFunctionReturn(PETSC_SUCCESS);
1821: }
1823: /*@
1824: KSPGetInitialGuessNonzero - Determines whether the `KSP` solver is using
1825: a zero initial guess.
1827: Not Collective
1829: Input Parameter:
1830: . ksp - iterative solver obtained from `KSPCreate()`
1832: Output Parameter:
1833: . flag - `PETSC_TRUE` if guess is nonzero, else `PETSC_FALSE`
1835: Level: intermediate
1837: .seealso: [](ch_ksp), `KSPSetInitialGuessNonzero()`, `KSP`
1838: @*/
1839: PetscErrorCode KSPGetInitialGuessNonzero(KSP ksp, PetscBool *flag)
1840: {
1841: PetscFunctionBegin;
1843: PetscAssertPointer(flag, 2);
1844: if (ksp->guess_zero) *flag = PETSC_FALSE;
1845: else *flag = PETSC_TRUE;
1846: PetscFunctionReturn(PETSC_SUCCESS);
1847: }
1849: /*@
1850: KSPSetErrorIfNotConverged - Causes `KSPSolve()` to generate an error if the solver has not converged as soon as the error is detected.
1852: Logically Collective
1854: Input Parameters:
1855: + ksp - iterative solver obtained from `KSPCreate()`
1856: - flg - `PETSC_TRUE` indicates you want the error generated
1858: Options Database Key:
1859: . -ksp_error_if_not_converged <true,false> - generate an error and stop the program
1861: Level: intermediate
1863: Notes:
1864: Normally PETSc continues if a linear solver fails to converge, you can call `KSPGetConvergedReason()` after a `KSPSolve()`
1865: to determine if it has converged. This functionality is mostly helpful while running in a debugger (`-start_in_debugger`) to determine exactly where
1866: the failure occurs and why.
1868: A `KSP_DIVERGED_ITS` will not generate an error in a `KSPSolve()` inside a nested linear solver
1870: .seealso: [](ch_ksp), `KSPGetErrorIfNotConverged()`, `KSP`
1871: @*/
1872: PetscErrorCode KSPSetErrorIfNotConverged(KSP ksp, PetscBool flg)
1873: {
1874: PetscFunctionBegin;
1877: ksp->errorifnotconverged = flg;
1878: PetscFunctionReturn(PETSC_SUCCESS);
1879: }
1881: /*@
1882: KSPGetErrorIfNotConverged - Will `KSPSolve()` generate an error if the solver does not converge?
1884: Not Collective
1886: Input Parameter:
1887: . ksp - iterative solver obtained from KSPCreate()
1889: Output Parameter:
1890: . flag - `PETSC_TRUE` if it will generate an error, else `PETSC_FALSE`
1892: Level: intermediate
1894: .seealso: [](ch_ksp), `KSPSetErrorIfNotConverged()`, `KSP`
1895: @*/
1896: PetscErrorCode KSPGetErrorIfNotConverged(KSP ksp, PetscBool *flag)
1897: {
1898: PetscFunctionBegin;
1900: PetscAssertPointer(flag, 2);
1901: *flag = ksp->errorifnotconverged;
1902: PetscFunctionReturn(PETSC_SUCCESS);
1903: }
1905: /*@
1906: KSPSetInitialGuessKnoll - Tells the iterative solver to use `PCApply()` on the right hand side vector to compute the initial guess (The Knoll trick)
1908: Logically Collective
1910: Input Parameters:
1911: + ksp - iterative solver obtained from `KSPCreate()`
1912: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1914: Level: advanced
1916: Developer Note:
1917: The Knoll trick is not currently implemented using the `KSPGuess` class which provides a variety of ways of computing
1918: an initial guess based on previous solves.
1920: .seealso: [](ch_ksp), `KSPGetInitialGuessKnoll()`, `KSPGuess`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1921: @*/
1922: PetscErrorCode KSPSetInitialGuessKnoll(KSP ksp, PetscBool flg)
1923: {
1924: PetscFunctionBegin;
1927: ksp->guess_knoll = flg;
1928: PetscFunctionReturn(PETSC_SUCCESS);
1929: }
1931: /*@
1932: KSPGetInitialGuessKnoll - Determines whether the `KSP` solver is using the Knoll trick (using PCApply(pc,b,...) to compute
1933: the initial guess
1935: Not Collective
1937: Input Parameter:
1938: . ksp - iterative solver obtained from `KSPCreate()`
1940: Output Parameter:
1941: . flag - `PETSC_TRUE` if using Knoll trick, else `PETSC_FALSE`
1943: Level: advanced
1945: .seealso: [](ch_ksp), `KSPSetInitialGuessKnoll()`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1946: @*/
1947: PetscErrorCode KSPGetInitialGuessKnoll(KSP ksp, PetscBool *flag)
1948: {
1949: PetscFunctionBegin;
1951: PetscAssertPointer(flag, 2);
1952: *flag = ksp->guess_knoll;
1953: PetscFunctionReturn(PETSC_SUCCESS);
1954: }
1956: /*@
1957: KSPGetComputeSingularValues - Gets the flag indicating whether the extreme singular
1958: values will be calculated via a Lanczos or Arnoldi process as the linear
1959: system is solved.
1961: Not Collective
1963: Input Parameter:
1964: . ksp - iterative solver obtained from `KSPCreate()`
1966: Output Parameter:
1967: . flg - `PETSC_TRUE` or `PETSC_FALSE`
1969: Options Database Key:
1970: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
1972: Level: advanced
1974: Notes:
1975: This option is not valid for `KSPType`.
1977: Many users may just want to use the monitoring routine
1978: `KSPMonitorSingularValue()` (which can be set with option `-ksp_monitor_singular_value`)
1979: to print the singular values at each iteration of the linear solve.
1981: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`
1982: @*/
1983: PetscErrorCode KSPGetComputeSingularValues(KSP ksp, PetscBool *flg)
1984: {
1985: PetscFunctionBegin;
1987: PetscAssertPointer(flg, 2);
1988: *flg = ksp->calc_sings;
1989: PetscFunctionReturn(PETSC_SUCCESS);
1990: }
1992: /*@
1993: KSPSetComputeSingularValues - Sets a flag so that the extreme singular
1994: values will be calculated via a Lanczos or Arnoldi process as the linear
1995: system is solved.
1997: Logically Collective
1999: Input Parameters:
2000: + ksp - iterative solver obtained from `KSPCreate()`
2001: - flg - `PETSC_TRUE` or `PETSC_FALSE`
2003: Options Database Key:
2004: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
2006: Level: advanced
2008: Notes:
2009: This option is not valid for all iterative methods.
2011: Many users may just want to use the monitoring routine
2012: `KSPMonitorSingularValue()` (which can be set with option `-ksp_monitor_singular_value`)
2013: to print the singular values at each iteration of the linear solve.
2015: Consider using the excellent package SLEPc for accurate efficient computations of singular or eigenvalues.
2017: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`, `KSPSetComputeRitz()`
2018: @*/
2019: PetscErrorCode KSPSetComputeSingularValues(KSP ksp, PetscBool flg)
2020: {
2021: PetscFunctionBegin;
2024: ksp->calc_sings = flg;
2025: PetscFunctionReturn(PETSC_SUCCESS);
2026: }
2028: /*@
2029: KSPGetComputeEigenvalues - Gets the flag indicating that the extreme eigenvalues
2030: values will be calculated via a Lanczos or Arnoldi process as the linear
2031: system is solved.
2033: Not Collective
2035: Input Parameter:
2036: . ksp - iterative solver obtained from `KSPCreate()`
2038: Output Parameter:
2039: . flg - `PETSC_TRUE` or `PETSC_FALSE`
2041: Level: advanced
2043: Note:
2044: Currently this option is not valid for all iterative methods.
2046: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`, `KSPSetComputeRitz()`
2047: @*/
2048: PetscErrorCode KSPGetComputeEigenvalues(KSP ksp, PetscBool *flg)
2049: {
2050: PetscFunctionBegin;
2052: PetscAssertPointer(flg, 2);
2053: *flg = ksp->calc_sings;
2054: PetscFunctionReturn(PETSC_SUCCESS);
2055: }
2057: /*@
2058: KSPSetComputeEigenvalues - Sets a flag so that the extreme eigenvalues
2059: values will be calculated via a Lanczos or Arnoldi process as the linear
2060: system is solved.
2062: Logically Collective
2064: Input Parameters:
2065: + ksp - iterative solver obtained from `KSPCreate()`
2066: - flg - `PETSC_TRUE` or `PETSC_FALSE`
2068: Level: advanced
2070: Note:
2071: Currently this option is not valid for all iterative methods.
2073: Consider using the excellent package SLEPc for accurate efficient computations of singular or eigenvalues.
2075: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`, `KSPSetComputeRitz()`
2076: @*/
2077: PetscErrorCode KSPSetComputeEigenvalues(KSP ksp, PetscBool flg)
2078: {
2079: PetscFunctionBegin;
2082: ksp->calc_sings = flg;
2083: PetscFunctionReturn(PETSC_SUCCESS);
2084: }
2086: /*@
2087: KSPSetComputeRitz - Sets a flag so that the Ritz or harmonic Ritz pairs
2088: will be calculated via a Lanczos or Arnoldi process as the linear
2089: system is solved.
2091: Logically Collective
2093: Input Parameters:
2094: + ksp - iterative solver obtained from `KSPCreate()`
2095: - flg - `PETSC_TRUE` or `PETSC_FALSE`
2097: Level: advanced
2099: Note:
2100: Currently this option is only valid for the `KSPGMRES` method.
2102: .seealso: [](ch_ksp), `KSPComputeRitz()`, `KSP`, `KSPComputeEigenvalues()`, `KSPComputeExtremeSingularValues()`
2103: @*/
2104: PetscErrorCode KSPSetComputeRitz(KSP ksp, PetscBool flg)
2105: {
2106: PetscFunctionBegin;
2109: ksp->calc_ritz = flg;
2110: PetscFunctionReturn(PETSC_SUCCESS);
2111: }
2113: /*@
2114: KSPGetRhs - Gets the right-hand-side vector for the linear system to
2115: be solved.
2117: Not Collective
2119: Input Parameter:
2120: . ksp - iterative solver obtained from `KSPCreate()`
2122: Output Parameter:
2123: . r - right-hand-side vector
2125: Level: developer
2127: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPSolve()`, `KSP`
2128: @*/
2129: PetscErrorCode KSPGetRhs(KSP ksp, Vec *r)
2130: {
2131: PetscFunctionBegin;
2133: PetscAssertPointer(r, 2);
2134: *r = ksp->vec_rhs;
2135: PetscFunctionReturn(PETSC_SUCCESS);
2136: }
2138: /*@
2139: KSPGetSolution - Gets the location of the solution for the
2140: linear system to be solved.
2142: Not Collective
2144: Input Parameter:
2145: . ksp - iterative solver obtained from `KSPCreate()`
2147: Output Parameter:
2148: . v - solution vector
2150: Level: developer
2152: Note:
2153: If this is called during a `KSPSolve()` the vector's values may not represent the solution
2154: to the linear system.
2156: .seealso: [](ch_ksp), `KSPGetRhs()`, `KSPBuildSolution()`, `KSPSolve()`, `KSP`
2157: @*/
2158: PetscErrorCode KSPGetSolution(KSP ksp, Vec *v)
2159: {
2160: PetscFunctionBegin;
2162: PetscAssertPointer(v, 2);
2163: *v = ksp->vec_sol;
2164: PetscFunctionReturn(PETSC_SUCCESS);
2165: }
2167: /*@
2168: KSPSetPC - Sets the preconditioner to be used to calculate the
2169: application of the preconditioner on a vector into a `KSP`.
2171: Collective
2173: Input Parameters:
2174: + ksp - the `KSP` iterative solver obtained from `KSPCreate()`
2175: - pc - the preconditioner object (if `NULL` it returns the `PC` currently held by the `KSP`)
2177: Level: developer
2179: Note:
2180: This routine is almost never used since `KSP` creates its own `PC` when needed.
2181: Use `KSPGetPC()` to retrieve the preconditioner context instead of creating a new one.
2183: .seealso: [](ch_ksp), `KSPGetPC()`, `KSP`
2184: @*/
2185: PetscErrorCode KSPSetPC(KSP ksp, PC pc)
2186: {
2187: PetscFunctionBegin;
2189: if (pc) {
2191: PetscCheckSameComm(ksp, 1, pc, 2);
2192: }
2193: if (ksp->pc != pc && ksp->setupstage) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2194: PetscCall(PetscObjectReference((PetscObject)pc));
2195: PetscCall(PCDestroy(&ksp->pc));
2196: ksp->pc = pc;
2197: PetscFunctionReturn(PETSC_SUCCESS);
2198: }
2200: PETSC_INTERN PetscErrorCode PCCreate_MPI(PC);
2202: // PetscClangLinter pragma disable: -fdoc-internal-linkage
2203: /*@C
2204: KSPCheckPCMPI - Checks if `-mpi_linear_solver_server` is active and the `PC` should be changed to `PCMPI`
2206: Collective, No Fortran Support
2208: Input Parameter:
2209: . ksp - iterative solver obtained from `KSPCreate()`
2211: Level: developer
2213: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`, `PCMPIServerBegin()`, `PCMPIServerEnd()`
2214: @*/
2215: PETSC_INTERN PetscErrorCode KSPCheckPCMPI(KSP ksp)
2216: {
2217: PetscBool isPCMPI;
2219: PetscFunctionBegin;
2221: PetscCall(PetscObjectTypeCompare((PetscObject)ksp->pc, PCMPI, &isPCMPI));
2222: if (PCMPIServerActive && ksp->nestlevel == 0 && !isPCMPI) {
2223: const char *prefix;
2224: char *found = NULL;
2226: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
2227: if (prefix) PetscCall(PetscStrstr(prefix, "mpi_linear_solver_server_", &found));
2228: if (!found) PetscCall(KSPAppendOptionsPrefix(ksp, "mpi_linear_solver_server_"));
2229: PetscCall(PetscInfo(NULL, "In MPI Linear Solver Server and detected (root) PC that must be changed to PCMPI\n"));
2230: PetscCall(PCSetType(ksp->pc, PCMPI));
2231: }
2232: PetscFunctionReturn(PETSC_SUCCESS);
2233: }
2235: /*@
2236: KSPGetPC - Returns a pointer to the preconditioner context with the `KSP`
2238: Not Collective
2240: Input Parameter:
2241: . ksp - iterative solver obtained from `KSPCreate()`
2243: Output Parameter:
2244: . pc - preconditioner context
2246: Level: beginner
2248: Note:
2249: The `PC` is created if it does not already exist.
2251: Developer Note:
2252: Calls `KSPCheckPCMPI()` to check if the `KSP` is effected by `-mpi_linear_solver_server`
2254: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`, `PC`
2255: @*/
2256: PetscErrorCode KSPGetPC(KSP ksp, PC *pc)
2257: {
2258: PetscFunctionBegin;
2260: PetscAssertPointer(pc, 2);
2261: if (!ksp->pc) {
2262: PetscCall(PCCreate(PetscObjectComm((PetscObject)ksp), &ksp->pc));
2263: PetscCall(PetscObjectIncrementTabLevel((PetscObject)ksp->pc, (PetscObject)ksp, 0));
2264: PetscCall(PetscObjectSetOptions((PetscObject)ksp->pc, ((PetscObject)ksp)->options));
2265: PetscCall(PCSetKSPNestLevel(ksp->pc, ksp->nestlevel));
2266: }
2267: PetscCall(KSPCheckPCMPI(ksp));
2268: *pc = ksp->pc;
2269: PetscFunctionReturn(PETSC_SUCCESS);
2270: }
2272: /*@
2273: KSPMonitor - runs the user provided monitor routines, if they exist
2275: Collective
2277: Input Parameters:
2278: + ksp - iterative solver obtained from `KSPCreate()`
2279: . it - iteration number
2280: - rnorm - relative norm of the residual
2282: Level: developer
2284: Notes:
2285: This routine is called by the `KSP` implementations.
2286: It does not typically need to be called by the user.
2288: For Krylov methods that do not keep a running value of the current solution (such as `KSPGMRES`) this
2289: cannot be called after the `KSPConvergedReason` has been set but before the final solution has been computed.
2291: .seealso: [](ch_ksp), `KSPMonitorSet()`
2292: @*/
2293: PetscErrorCode KSPMonitor(KSP ksp, PetscInt it, PetscReal rnorm)
2294: {
2295: PetscInt i, n = ksp->numbermonitors;
2297: PetscFunctionBegin;
2298: for (i = 0; i < n; i++) PetscCall((*ksp->monitor[i])(ksp, it, rnorm, ksp->monitorcontext[i]));
2299: PetscFunctionReturn(PETSC_SUCCESS);
2300: }
2302: /*@C
2303: KSPMonitorSet - Sets an ADDITIONAL function to be called at every iteration to monitor, i.e. display in some way, perhaps by printing in the terminal,
2304: the residual norm computed in a `KSPSolve()`
2306: Logically Collective
2308: Input Parameters:
2309: + ksp - iterative solver obtained from `KSPCreate()`
2310: . monitor - pointer to function (if this is `NULL`, it turns off monitoring, see `KSPMonitorFn`
2311: . ctx - [optional] context for private data for the monitor routine (use `NULL` if no context is needed)
2312: - monitordestroy - [optional] routine that frees monitor context (may be `NULL`), see `PetscCtxDestroyFn` for the calling sequence
2314: Options Database Keys:
2315: + -ksp_monitor - sets `KSPMonitorResidual()`
2316: . -ksp_monitor hdf5:filename - sets `KSPMonitorResidualView()` and saves residual
2317: . -ksp_monitor draw - sets `KSPMonitorResidualView()` and plots residual
2318: . -ksp_monitor draw::draw_lg - sets `KSPMonitorResidualDrawLG()` and plots residual
2319: . -ksp_monitor_pause_final - Pauses any graphics when the solve finishes (only works for internal monitors)
2320: . -ksp_monitor_true_residual - sets `KSPMonitorTrueResidual()`
2321: . -ksp_monitor_true_residual draw::draw_lg - sets `KSPMonitorTrueResidualDrawLG()` and plots residual
2322: . -ksp_monitor_max - sets `KSPMonitorTrueResidualMax()`
2323: . -ksp_monitor_singular_value - sets `KSPMonitorSingularValue()`
2324: - -ksp_monitor_cancel - cancels all monitors that have been hardwired into a code by calls to `KSPMonitorSet()`, but
2325: does not cancel those set via the options database.
2327: Level: beginner
2329: Notes:
2330: The options database option `-ksp_monitor` and related options are the easiest way to turn on `KSP` iteration monitoring
2332: `KSPMonitorRegister()` provides a way to associate an options database key with `KSP` monitor function.
2334: The default is to do no monitoring. To print the residual, or preconditioned
2335: residual if `KSPSetNormType`(ksp,`KSP_NORM_PRECONDITIONED`) was called, use
2336: `KSPMonitorResidual()` as the monitoring routine, with a `PETSCVIEWERASCII` as the
2337: context.
2339: Several different monitoring routines may be set by calling
2340: `KSPMonitorSet()` multiple times; they will be called in the
2341: order in which they were set.
2343: Fortran Note:
2344: Only a single monitor function can be set for each `KSP` object
2346: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorRegister()`, `KSPMonitorCancel()`, `KSP`, `PetscCtxDestroyFn`
2347: @*/
2348: PetscErrorCode KSPMonitorSet(KSP ksp, KSPMonitorFn *monitor, PetscCtx ctx, PetscCtxDestroyFn *monitordestroy)
2349: {
2350: PetscFunctionBegin;
2352: for (PetscInt i = 0; i < ksp->numbermonitors; i++) {
2353: PetscBool identical;
2355: PetscCall(PetscMonitorCompare((PetscErrorCode (*)(void))(PetscVoidFn *)monitor, ctx, monitordestroy, (PetscErrorCode (*)(void))(PetscVoidFn *)ksp->monitor[i], ksp->monitorcontext[i], ksp->monitordestroy[i], &identical));
2356: if (identical) PetscFunctionReturn(PETSC_SUCCESS);
2357: }
2358: PetscCheck(ksp->numbermonitors < MAXKSPMONITORS, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Too many KSP monitors set");
2359: ksp->monitor[ksp->numbermonitors] = monitor;
2360: ksp->monitordestroy[ksp->numbermonitors] = monitordestroy;
2361: ksp->monitorcontext[ksp->numbermonitors++] = ctx;
2362: PetscFunctionReturn(PETSC_SUCCESS);
2363: }
2365: /*@
2366: KSPMonitorCancel - Clears all monitors for a `KSP` object.
2368: Logically Collective
2370: Input Parameter:
2371: . ksp - iterative solver obtained from `KSPCreate()`
2373: Options Database Key:
2374: . -ksp_monitor_cancel - Cancels all monitors that have been hardwired into a code by calls to `KSPMonitorSet()`, but does not cancel those set via the options database.
2376: Level: intermediate
2378: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorSet()`, `KSP`
2379: @*/
2380: PetscErrorCode KSPMonitorCancel(KSP ksp)
2381: {
2382: PetscInt i;
2384: PetscFunctionBegin;
2386: for (i = 0; i < ksp->numbermonitors; i++) {
2387: if (ksp->monitordestroy[i]) PetscCall((*ksp->monitordestroy[i])(&ksp->monitorcontext[i]));
2388: }
2389: ksp->numbermonitors = 0;
2390: PetscFunctionReturn(PETSC_SUCCESS);
2391: }
2393: /*@C
2394: KSPGetMonitorContext - Gets the monitoring context, as set by `KSPMonitorSet()` for the FIRST monitor only.
2396: Not Collective
2398: Input Parameter:
2399: . ksp - iterative solver obtained from `KSPCreate()`
2401: Output Parameter:
2402: . ctx - monitoring context
2404: Level: intermediate
2406: Fortran Notes:
2407: This only works when the context is a Fortran derived type or a `PetscObject`. Declare `ctx` with
2408: .vb
2409: type(tUsertype), pointer :: ctx
2410: .ve
2412: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSP`
2413: @*/
2414: PetscErrorCode KSPGetMonitorContext(KSP ksp, PetscCtxRt ctx)
2415: {
2416: PetscFunctionBegin;
2418: *(void **)ctx = ksp->monitorcontext[0];
2419: PetscFunctionReturn(PETSC_SUCCESS);
2420: }
2422: /*@
2423: KSPSetResidualHistory - Sets the array used to hold the residual history.
2424: If set, this array will contain the residual norms computed at each
2425: iteration of the solver.
2427: Not Collective
2429: Input Parameters:
2430: + ksp - iterative solver obtained from `KSPCreate()`
2431: . a - array to hold history
2432: . na - size of `a`
2433: - reset - `PETSC_TRUE` indicates the history counter is reset to zero
2434: for each new linear solve
2436: Level: advanced
2438: Notes:
2439: If provided, `a` is NOT freed by PETSc so the user needs to keep track of it and destroy once the `KSP` object is destroyed.
2440: If 'a' is `NULL` then space is allocated for the history. If 'na' `PETSC_DECIDE` or (deprecated) `PETSC_DEFAULT` then a
2441: default array of length 10,000 is allocated.
2443: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2445: .seealso: [](ch_ksp), `KSPGetResidualHistory()`, `KSP`
2446: @*/
2447: PetscErrorCode KSPSetResidualHistory(KSP ksp, PetscReal a[], PetscCount na, PetscBool reset)
2448: {
2449: PetscFunctionBegin;
2452: PetscCall(PetscFree(ksp->res_hist_alloc));
2453: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2454: ksp->res_hist = a;
2455: ksp->res_hist_max = na;
2456: } else {
2457: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->res_hist_max = (size_t)na;
2458: else ksp->res_hist_max = 10000; /* like default ksp->max_it */
2459: PetscCall(PetscCalloc1(ksp->res_hist_max, &ksp->res_hist_alloc));
2461: ksp->res_hist = ksp->res_hist_alloc;
2462: }
2463: ksp->res_hist_len = 0;
2464: ksp->res_hist_reset = reset;
2465: PetscFunctionReturn(PETSC_SUCCESS);
2466: }
2468: /*@C
2469: KSPGetResidualHistory - Gets the array used to hold the residual history and the number of residuals it contains.
2471: Not Collective
2473: Input Parameter:
2474: . ksp - iterative solver obtained from `KSPCreate()`
2476: Output Parameters:
2477: + a - pointer to array to hold history (or `NULL`)
2478: - na - number of used entries in a (or `NULL`). Note this has different meanings depending on the `reset` argument to `KSPSetResidualHistory()`
2480: Level: advanced
2482: Note:
2483: This array is borrowed and should not be freed by the caller.
2485: Can only be called after a `KSPSetResidualHistory()` otherwise `a` and `na` are set to `NULL` and zero
2487: When `reset` was `PETSC_TRUE` since a residual is computed before the first iteration, the value of `na` is generally one more than the value
2488: returned with `KSPGetIterationNumber()`.
2490: Some Krylov methods may not compute the final residual norm when convergence is declared because the maximum number of iterations allowed has been reached.
2491: In this situation, when `reset` was `PETSC_TRUE`, `na` will then equal the number of iterations reported with `KSPGetIterationNumber()`
2493: Some Krylov methods (such as `KSPSTCG`), under certain circumstances, do not compute the final residual norm. In this situation, when `reset` was `PETSC_TRUE`,
2494: `na` will then equal the number of iterations reported with `KSPGetIterationNumber()`
2496: `KSPBCGSL` does not record the residual norms for the "subiterations" hence the results from `KSPGetResidualHistory()` and `KSPGetIterationNumber()` will be different
2498: Fortran Note:
2499: Call `KSPRestoreResidualHistory()` when access to the history is no longer needed.
2501: .seealso: [](ch_ksp), `KSPSetResidualHistory()`, `KSP`, `KSPGetIterationNumber()`, `KSPSTCG`, `KSPBCGSL`
2502: @*/
2503: PetscErrorCode KSPGetResidualHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2504: {
2505: PetscFunctionBegin;
2507: if (a) *a = ksp->res_hist;
2508: if (na) PetscCall(PetscIntCast(ksp->res_hist_len, na));
2509: PetscFunctionReturn(PETSC_SUCCESS);
2510: }
2512: /*@
2513: KSPSetErrorHistory - Sets the array used to hold the error history. If set, this array will contain the error norms computed at each iteration of the solver.
2515: Not Collective
2517: Input Parameters:
2518: + ksp - iterative solver obtained from `KSPCreate()`
2519: . a - array to hold history
2520: . na - size of `a`
2521: - reset - `PETSC_TRUE` indicates the history counter is reset to zero for each new linear solve
2523: Level: advanced
2525: Notes:
2526: If provided, `a` is NOT freed by PETSc so the user needs to keep track of it and destroy once the `KSP` object is destroyed.
2527: If 'a' is `NULL` then space is allocated for the history. If 'na' is `PETSC_DECIDE` or (deprecated) `PETSC_DEFAULT` then a default array of length 1,0000 is allocated.
2529: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2531: .seealso: [](ch_ksp), `KSPGetErrorHistory()`, `KSPSetResidualHistory()`, `KSP`
2532: @*/
2533: PetscErrorCode KSPSetErrorHistory(KSP ksp, PetscReal a[], PetscCount na, PetscBool reset)
2534: {
2535: PetscFunctionBegin;
2538: PetscCall(PetscFree(ksp->err_hist_alloc));
2539: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2540: ksp->err_hist = a;
2541: ksp->err_hist_max = na;
2542: } else {
2543: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->err_hist_max = (size_t)na;
2544: else ksp->err_hist_max = 10000; /* like default ksp->max_it */
2545: PetscCall(PetscCalloc1(ksp->err_hist_max, &ksp->err_hist_alloc));
2546: ksp->err_hist = ksp->err_hist_alloc;
2547: }
2548: ksp->err_hist_len = 0;
2549: ksp->err_hist_reset = reset;
2550: PetscFunctionReturn(PETSC_SUCCESS);
2551: }
2553: /*@C
2554: KSPGetErrorHistory - Gets the array used to hold the error history and the number of residuals it contains.
2556: Not Collective
2558: Input Parameter:
2559: . ksp - iterative solver obtained from `KSPCreate()`
2561: Output Parameters:
2562: + a - pointer to array to hold history (or `NULL`)
2563: - na - number of used entries in a (or `NULL`)
2565: Level: advanced
2567: Note:
2568: This array is borrowed and should not be freed by the caller.
2569: Can only be called after a `KSPSetErrorHistory()` otherwise `a` and `na` are set to `NULL` and zero
2571: Fortran Note:
2572: .vb
2573: PetscReal, pointer :: a(:)
2574: .ve
2576: .seealso: [](ch_ksp), `KSPSetErrorHistory()`, `KSPGetResidualHistory()`, `KSP`
2577: @*/
2578: PetscErrorCode KSPGetErrorHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2579: {
2580: PetscFunctionBegin;
2582: if (a) *a = ksp->err_hist;
2583: if (na) PetscCall(PetscIntCast(ksp->err_hist_len, na));
2584: PetscFunctionReturn(PETSC_SUCCESS);
2585: }
2587: /*@
2588: KSPComputeConvergenceRate - Compute the convergence rate for the iteration <https:/en.wikipedia.org/wiki/Coefficient_of_determination>
2590: Not Collective
2592: Input Parameter:
2593: . ksp - The `KSP`
2595: Output Parameters:
2596: + cr - The residual contraction rate
2597: . rRsq - The coefficient of determination, $R^2$, indicating the linearity of the data
2598: . ce - The error contraction rate
2599: - eRsq - The coefficient of determination, $R^2$, indicating the linearity of the data
2601: Level: advanced
2603: Note:
2604: Suppose that the residual is reduced linearly, $r_k = c^k r_0$, which means $log r_k = log r_0 + k log c$. After linear regression,
2605: the slope is $\log c$. The coefficient of determination is given by $1 - \frac{\sum_i (y_i - f(x_i))^2}{\sum_i (y_i - \bar y)}$,
2607: .seealso: [](ch_ksp), `KSP`, `KSPConvergedRateView()`
2608: @*/
2609: PetscErrorCode KSPComputeConvergenceRate(KSP ksp, PetscReal *cr, PetscReal *rRsq, PetscReal *ce, PetscReal *eRsq)
2610: {
2611: PetscReal const *hist;
2612: PetscReal *x, *y, slope, intercept, mean = 0.0, var = 0.0, res = 0.0;
2613: PetscInt n, k;
2615: PetscFunctionBegin;
2616: if (cr || rRsq) {
2617: PetscCall(KSPGetResidualHistory(ksp, &hist, &n));
2618: if (!n) {
2619: if (cr) *cr = 0.0;
2620: if (rRsq) *rRsq = -1.0;
2621: } else {
2622: PetscCall(PetscMalloc2(n, &x, n, &y));
2623: for (k = 0; k < n; ++k) {
2624: x[k] = k;
2625: y[k] = PetscLogReal(hist[k]);
2626: mean += y[k];
2627: }
2628: mean /= n;
2629: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2630: for (k = 0; k < n; ++k) {
2631: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2632: var += PetscSqr(y[k] - mean);
2633: }
2634: PetscCall(PetscFree2(x, y));
2635: if (cr) *cr = PetscExpReal(slope);
2636: if (rRsq) *rRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2637: }
2638: }
2639: if (ce || eRsq) {
2640: PetscCall(KSPGetErrorHistory(ksp, &hist, &n));
2641: if (!n) {
2642: if (ce) *ce = 0.0;
2643: if (eRsq) *eRsq = -1.0;
2644: } else {
2645: PetscCall(PetscMalloc2(n, &x, n, &y));
2646: for (k = 0; k < n; ++k) {
2647: x[k] = k;
2648: y[k] = PetscLogReal(hist[k]);
2649: mean += y[k];
2650: }
2651: mean /= n;
2652: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2653: for (k = 0; k < n; ++k) {
2654: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2655: var += PetscSqr(y[k] - mean);
2656: }
2657: PetscCall(PetscFree2(x, y));
2658: if (ce) *ce = PetscExpReal(slope);
2659: if (eRsq) *eRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2660: }
2661: }
2662: PetscFunctionReturn(PETSC_SUCCESS);
2663: }
2665: /*@C
2666: KSPSetConvergenceTest - Sets the function to be used to determine convergence of `KSPSolve()`
2668: Logically Collective
2670: Input Parameters:
2671: + ksp - iterative solver obtained from `KSPCreate()`
2672: . converge - pointer to the function, see `KSPConvergenceTestFn`
2673: . ctx - context for private data for the convergence routine (may be `NULL`)
2674: - destroy - a routine for destroying the context (may be `NULL`)
2676: Level: advanced
2678: Notes:
2679: Must be called after the `KSP` type has been set so put this after
2680: a call to `KSPSetType()`, or `KSPSetFromOptions()`.
2682: The default convergence test, `KSPConvergedDefault()`, aborts if the
2683: residual grows to more than 10000 times the initial residual.
2685: The default is a combination of relative and absolute tolerances.
2686: The residual value that is tested may be an approximation; routines
2687: that need exact values should compute them.
2689: In the default PETSc convergence test, the precise values of reason
2690: are macros such as `KSP_CONVERGED_RTOL`, which are defined in petscksp.h.
2692: .seealso: [](ch_ksp), `KSP`, `KSPConvergenceTestFn`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPGetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2693: @*/
2694: PetscErrorCode KSPSetConvergenceTest(KSP ksp, KSPConvergenceTestFn *converge, PetscCtx ctx, PetscCtxDestroyFn *destroy)
2695: {
2696: PetscFunctionBegin;
2698: if (ksp->convergeddestroy) PetscCall((*ksp->convergeddestroy)(&ksp->cnvP));
2699: ksp->converged = converge;
2700: ksp->convergeddestroy = destroy;
2701: ksp->cnvP = ctx;
2702: PetscFunctionReturn(PETSC_SUCCESS);
2703: }
2705: /*@C
2706: KSPGetConvergenceTest - Gets the function to be used to determine convergence.
2708: Logically Collective
2710: Input Parameter:
2711: . ksp - iterative solver obtained from `KSPCreate()`
2713: Output Parameters:
2714: + converge - pointer to convergence test function, see `KSPConvergenceTestFn`
2715: . ctx - context for private data for the convergence routine (may be `NULL`)
2716: - destroy - a routine for destroying the context (may be `NULL`)
2718: Level: advanced
2720: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2721: @*/
2722: PetscErrorCode KSPGetConvergenceTest(KSP ksp, KSPConvergenceTestFn **converge, PetscCtxRt ctx, PetscCtxDestroyFn **destroy)
2723: {
2724: PetscFunctionBegin;
2726: if (converge) *converge = ksp->converged;
2727: if (destroy) *destroy = ksp->convergeddestroy;
2728: if (ctx) *(void **)ctx = ksp->cnvP;
2729: PetscFunctionReturn(PETSC_SUCCESS);
2730: }
2732: /*@C
2733: KSPGetAndClearConvergenceTest - Gets the function to be used to determine convergence. Removes the current test without calling destroy on the test context
2735: Logically Collective
2737: Input Parameter:
2738: . ksp - iterative solver obtained from `KSPCreate()`
2740: Output Parameters:
2741: + converge - pointer to convergence test function, see `KSPConvergenceTestFn`
2742: . ctx - context for private data for the convergence routine
2743: - destroy - a routine for destroying the context
2745: Level: advanced
2747: Note:
2748: This is intended to be used to allow transferring the convergence test (and its context) to another testing object (for example another `KSP`)
2749: and then calling `KSPSetConvergenceTest()` on this original `KSP`. If you just called `KSPGetConvergenceTest()` followed
2750: by `KSPSetConvergenceTest()` the original context information
2751: would be destroyed and hence the transferred context would be invalid and trigger a crash on use
2753: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2754: @*/
2755: PetscErrorCode KSPGetAndClearConvergenceTest(KSP ksp, KSPConvergenceTestFn **converge, PetscCtxRt ctx, PetscCtxDestroyFn **destroy)
2756: {
2757: PetscFunctionBegin;
2759: *converge = ksp->converged;
2760: *destroy = ksp->convergeddestroy;
2761: *(void **)ctx = ksp->cnvP;
2762: ksp->converged = NULL;
2763: ksp->cnvP = NULL;
2764: ksp->convergeddestroy = NULL;
2765: PetscFunctionReturn(PETSC_SUCCESS);
2766: }
2768: /*@C
2769: KSPGetConvergenceContext - Gets the convergence context set with `KSPSetConvergenceTest()`.
2771: Not Collective
2773: Input Parameter:
2774: . ksp - iterative solver obtained from `KSPCreate()`
2776: Output Parameter:
2777: . ctx - monitoring context
2779: Level: advanced
2781: Fortran Note:
2782: This only works when the context is a Fortran derived type or a `PetscObject`. Declare `ctx` with
2783: .vb
2784: type(tUsertype), pointer :: ctx
2785: .ve
2787: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2788: @*/
2789: PetscErrorCode KSPGetConvergenceContext(KSP ksp, PetscCtxRt ctx)
2790: {
2791: PetscFunctionBegin;
2793: *(void **)ctx = ksp->cnvP;
2794: PetscFunctionReturn(PETSC_SUCCESS);
2795: }
2797: /*@
2798: KSPBuildSolution - Builds the approximate solution in a vector provided.
2800: Collective
2802: Input Parameter:
2803: . ksp - iterative solver obtained from `KSPCreate()`
2805: Output Parameter:
2806: Provide exactly one of
2807: + v - location to stash solution, optional, otherwise pass `NULL`
2808: - V - the solution is returned in this location. This vector is created internally. This vector should NOT be destroyed by the user with `VecDestroy()`.
2810: Level: developer
2812: Notes:
2813: This routine can be used in one of two ways
2814: .vb
2815: KSPBuildSolution(ksp,NULL,&V);
2816: or
2817: KSPBuildSolution(ksp,v,NULL); or KSPBuildSolution(ksp,v,&v);
2818: .ve
2819: In the first case an internal vector is allocated to store the solution
2820: (the user cannot destroy this vector). In the second case the solution
2821: is generated in the vector that the user provides. Note that for certain
2822: methods, such as `KSPCG`, the second case requires a copy of the solution,
2823: while in the first case the call is essentially free since it simply
2824: returns the vector where the solution already is stored. For some methods
2825: like `KSPGMRES` during the solve this is a reasonably expensive operation and should only be
2826: used if truly needed.
2828: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPBuildResidual()`, `KSP`
2829: @*/
2830: PetscErrorCode KSPBuildSolution(KSP ksp, Vec v, Vec *V)
2831: {
2832: PetscFunctionBegin;
2834: PetscCheck(V || v, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONG, "Must provide either v or V");
2835: if (!V) V = &v;
2836: if (ksp->reason != KSP_CONVERGED_ITERATING) {
2837: if (!v) PetscCall(KSPGetSolution(ksp, V));
2838: else PetscCall(VecCopy(ksp->vec_sol, v));
2839: } else {
2840: PetscUseTypeMethod(ksp, buildsolution, v, V);
2841: }
2842: PetscFunctionReturn(PETSC_SUCCESS);
2843: }
2845: /*@
2846: KSPBuildResidual - Builds the residual in a vector provided.
2848: Collective
2850: Input Parameter:
2851: . ksp - iterative solver obtained from `KSPCreate()`
2853: Output Parameters:
2854: + t - work vector. If not provided then one is generated.
2855: . v - optional location to stash residual. If `v` is not provided, then a location is generated.
2856: - V - the residual
2858: Level: advanced
2860: Note:
2861: Regardless of whether or not `v` is provided, the residual is
2862: returned in `V`.
2864: .seealso: [](ch_ksp), `KSP`, `KSPBuildSolution()`
2865: @*/
2866: PetscErrorCode KSPBuildResidual(KSP ksp, Vec t, Vec v, Vec *V)
2867: {
2868: PetscBool flag = PETSC_FALSE;
2869: Vec w = v, tt = t;
2871: PetscFunctionBegin;
2873: if (!w) PetscCall(VecDuplicate(ksp->vec_rhs, &w));
2874: if (!tt) {
2875: PetscCall(VecDuplicate(ksp->vec_sol, &tt));
2876: flag = PETSC_TRUE;
2877: }
2878: PetscUseTypeMethod(ksp, buildresidual, tt, w, V);
2879: if (flag) PetscCall(VecDestroy(&tt));
2880: PetscFunctionReturn(PETSC_SUCCESS);
2881: }
2883: /*@
2884: KSPSetDiagonalScale - Tells `KSP` to symmetrically diagonally scale the system
2885: before solving. This actually CHANGES the matrix (and right-hand side).
2887: Logically Collective
2889: Input Parameters:
2890: + ksp - the `KSP` context
2891: - scale - `PETSC_TRUE` or `PETSC_FALSE`
2893: Options Database Keys:
2894: + -ksp_diagonal_scale - perform a diagonal scaling before the solve
2895: - -ksp_diagonal_scale_fix - scale the matrix back AFTER the solve
2897: Level: advanced
2899: Notes:
2900: Scales the matrix by $D^{-1/2} A D^{-1/2} [D^{1/2} x ] = D^{-1/2} b $
2901: where $D_{ii}$ is $1/abs(A_{ii}) $ unless $A_{ii}$ is zero and then it is 1.
2903: BE CAREFUL with this routine: it actually scales the matrix and right
2904: hand side that define the system. After the system is solved the matrix
2905: and right-hand side remain scaled unless you use `KSPSetDiagonalScaleFix()`
2907: This should NOT be used within the `SNES` solves if you are using a line
2908: search.
2910: If you use this with the `PCType` `PCEISENSTAT` preconditioner than you can
2911: use the `PCEisenstatSetNoDiagonalScaling()` option, or `-pc_eisenstat_no_diagonal_scaling`
2912: to save some unneeded, redundant flops.
2914: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2915: @*/
2916: PetscErrorCode KSPSetDiagonalScale(KSP ksp, PetscBool scale)
2917: {
2918: PetscFunctionBegin;
2921: ksp->dscale = scale;
2922: PetscFunctionReturn(PETSC_SUCCESS);
2923: }
2925: /*@
2926: KSPGetDiagonalScale - Checks if `KSP` solver scales the matrix and right-hand side, that is if `KSPSetDiagonalScale()` has been called
2928: Not Collective
2930: Input Parameter:
2931: . ksp - the `KSP` context
2933: Output Parameter:
2934: . scale - `PETSC_TRUE` or `PETSC_FALSE`
2936: Level: intermediate
2938: .seealso: [](ch_ksp), `KSP`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`
2939: @*/
2940: PetscErrorCode KSPGetDiagonalScale(KSP ksp, PetscBool *scale)
2941: {
2942: PetscFunctionBegin;
2944: PetscAssertPointer(scale, 2);
2945: *scale = ksp->dscale;
2946: PetscFunctionReturn(PETSC_SUCCESS);
2947: }
2949: /*@
2950: KSPSetDiagonalScaleFix - Tells `KSP` to diagonally scale the system back after solving.
2952: Logically Collective
2954: Input Parameters:
2955: + ksp - the `KSP` context
2956: - fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2957: rescale (default)
2959: Level: intermediate
2961: Notes:
2962: Must be called after `KSPSetDiagonalScale()`
2964: Using this will slow things down, because it rescales the matrix before and
2965: after each linear solve. This is intended mainly for testing to allow one
2966: to easily get back the original system to make sure the solution computed is
2967: accurate enough.
2969: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPGetDiagonalScaleFix()`, `KSP`
2970: @*/
2971: PetscErrorCode KSPSetDiagonalScaleFix(KSP ksp, PetscBool fix)
2972: {
2973: PetscFunctionBegin;
2976: ksp->dscalefix = fix;
2977: PetscFunctionReturn(PETSC_SUCCESS);
2978: }
2980: /*@
2981: KSPGetDiagonalScaleFix - Determines if `KSP` diagonally scales the system back after solving. That is `KSPSetDiagonalScaleFix()` has been called
2983: Not Collective
2985: Input Parameter:
2986: . ksp - the `KSP` context
2988: Output Parameter:
2989: . fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2990: rescale (default)
2992: Level: intermediate
2994: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2995: @*/
2996: PetscErrorCode KSPGetDiagonalScaleFix(KSP ksp, PetscBool *fix)
2997: {
2998: PetscFunctionBegin;
3000: PetscAssertPointer(fix, 2);
3001: *fix = ksp->dscalefix;
3002: PetscFunctionReturn(PETSC_SUCCESS);
3003: }
3005: /*@C
3006: KSPSetComputeOperators - set routine to compute the linear operators
3008: Logically Collective
3010: Input Parameters:
3011: + ksp - the `KSP` context
3012: . func - function to compute the operators, see `KSPComputeOperatorsFn` for the calling sequence
3013: - ctx - optional context
3015: Level: beginner
3017: Notes:
3018: `func()` will be called automatically at the very next call to `KSPSolve()`. It will NOT be called at future `KSPSolve()` calls
3019: unless either `KSPSetComputeOperators()` or `KSPSetOperators()` is called before that `KSPSolve()` is called. This allows the same system to be solved several times
3020: with different right-hand side functions but is a confusing API since one might expect it to be called for each `KSPSolve()`
3022: To reuse the same preconditioner for the next `KSPSolve()` and not compute a new one based on the most recently computed matrix call `KSPSetReusePreconditioner()`
3024: Developer Note:
3025: Perhaps this routine and `KSPSetComputeRHS()` could be combined into a new API that makes clear when new matrices are computing without requiring call this
3026: routine to indicate when the new matrix should be computed.
3028: .seealso: [](ch_ksp), `KSP`, `KSPSetOperators()`, `KSPSetComputeRHS()`, `DMKSPSetComputeOperators()`, `KSPSetComputeInitialGuess()`, `KSPComputeOperatorsFn`
3029: @*/
3030: PetscErrorCode KSPSetComputeOperators(KSP ksp, KSPComputeOperatorsFn *func, PetscCtx ctx)
3031: {
3032: DM dm;
3034: PetscFunctionBegin;
3036: PetscCall(KSPGetDM(ksp, &dm));
3037: PetscCall(DMKSPSetComputeOperators(dm, func, ctx));
3038: if (ksp->setupstage == KSP_SETUP_NEWRHS) ksp->setupstage = KSP_SETUP_NEWMATRIX;
3039: PetscFunctionReturn(PETSC_SUCCESS);
3040: }
3042: /*@C
3043: KSPSetComputeRHS - set routine to compute the right-hand side of the linear system
3045: Logically Collective
3047: Input Parameters:
3048: + ksp - the `KSP` context
3049: . func - function to compute the right-hand side, see `KSPComputeRHSFn` for the calling sequence
3050: - ctx - optional context
3052: Level: beginner
3054: Note:
3055: The routine you provide will be called EACH you call `KSPSolve()` to prepare the new right-hand side for that solve
3057: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `DMKSPSetComputeRHS()`, `KSPSetComputeOperators()`, `KSPSetOperators()`, `KSPComputeRHSFn`
3058: @*/
3059: PetscErrorCode KSPSetComputeRHS(KSP ksp, KSPComputeRHSFn *func, PetscCtx ctx)
3060: {
3061: DM dm;
3063: PetscFunctionBegin;
3065: PetscCall(KSPGetDM(ksp, &dm));
3066: PetscCall(DMKSPSetComputeRHS(dm, func, ctx));
3067: PetscFunctionReturn(PETSC_SUCCESS);
3068: }
3070: /*@C
3071: KSPSetComputeInitialGuess - set routine to compute the initial guess of the linear system
3073: Logically Collective
3075: Input Parameters:
3076: + ksp - the `KSP` context
3077: . func - function to compute the initial guess, see `KSPComputeInitialGuessFn` for calling sequence
3078: - ctx - optional context
3080: Level: beginner
3082: Note:
3083: This should only be used in conjunction with `KSPSetComputeRHS()` and `KSPSetComputeOperators()`, otherwise
3084: call `KSPSetInitialGuessNonzero()` and set the initial guess values in the solution vector passed to `KSPSolve()` before calling the solver
3086: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `KSPSetComputeRHS()`, `KSPSetComputeOperators()`, `DMKSPSetComputeInitialGuess()`, `KSPSetInitialGuessNonzero()`,
3087: `KSPComputeInitialGuessFn`
3088: @*/
3089: PetscErrorCode KSPSetComputeInitialGuess(KSP ksp, KSPComputeInitialGuessFn *func, PetscCtx ctx)
3090: {
3091: DM dm;
3093: PetscFunctionBegin;
3095: PetscCall(KSPGetDM(ksp, &dm));
3096: PetscCall(DMKSPSetComputeInitialGuess(dm, func, ctx));
3097: PetscFunctionReturn(PETSC_SUCCESS);
3098: }
3100: /*@
3101: KSPSetUseExplicitTranspose - Determines the explicit transpose of the operator is formed in `KSPSolveTranspose()`. In some configurations (like GPUs) it may
3102: be explicitly formed since the solve is much more efficient.
3104: Logically Collective
3106: Input Parameter:
3107: . ksp - the `KSP` context
3109: Output Parameter:
3110: . flg - `PETSC_TRUE` to transpose the system in `KSPSolveTranspose()`, `PETSC_FALSE` to not transpose (default)
3112: Level: advanced
3114: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `KSP`
3115: @*/
3116: PetscErrorCode KSPSetUseExplicitTranspose(KSP ksp, PetscBool flg)
3117: {
3118: PetscFunctionBegin;
3121: ksp->transpose.use_explicittranspose = flg;
3122: PetscFunctionReturn(PETSC_SUCCESS);
3123: }