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 context 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_eigenvalues) 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 context obtained from `KSPCreate()`
79: - n - size of arrays `r` and `c`. The number of eigenvalues computed `neig` will, in
80: general, be less than this.
82: Output Parameters:
83: + r - real part of computed eigenvalues, provided by user with a dimension of at least `n`
84: . c - complex part of computed eigenvalues, provided by user with a dimension of at least `n`
85: - neig - actual number of eigenvalues computed (will be less than or equal to `n`)
87: Options Database Key:
88: . -ksp_view_eigenvalues - Prints eigenvalues to stdout
90: Level: advanced
92: Notes:
93: The number of eigenvalues estimated depends on the size of the Krylov space
94: generated during the `KSPSolve()` ; for example, with
95: `KSPCG` it corresponds to the number of CG iterations, for `KSPGMRES` it is the number
96: of GMRES iterations SINCE the last restart. Any extra space in `r` and `c`
97: will be ignored.
99: `KSPComputeEigenvalues()` does not usually provide accurate estimates; it is
100: intended only for assistance in understanding the convergence of iterative
101: methods, not for eigenanalysis. For accurate computation of eigenvalues we recommend using
102: the excellent package SLEPc.
104: One must call `KSPSetComputeEigenvalues()` before calling `KSPSetUp()`
105: in order for this routine to work correctly.
107: Many users may just want to use the monitoring routine
108: `KSPMonitorSingularValue()` (which can be set with option -ksp_monitor_singular_value)
109: to print the singular values at each iteration of the linear solve.
111: `KSPComputeRitz()` provides estimates for both the eigenvalues and their corresponding eigenvectors.
113: .seealso: [](ch_ksp), `KSPSetComputeEigenvalues()`, `KSPSetComputeSingularValues()`, `KSPMonitorSingularValue()`, `KSPComputeExtremeSingularValues()`, `KSP`, `KSPComputeRitz()`
114: @*/
115: PetscErrorCode KSPComputeEigenvalues(KSP ksp, PetscInt n, PetscReal r[], PetscReal c[], PetscInt *neig)
116: {
117: PetscFunctionBegin;
119: if (n) PetscAssertPointer(r, 3);
120: if (n) PetscAssertPointer(c, 4);
121: PetscCheck(n >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Requested < 0 Eigenvalues");
122: PetscAssertPointer(neig, 5);
123: PetscCheck(ksp->calc_sings, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Eigenvalues not requested before KSPSetUp()");
125: if (n && ksp->ops->computeeigenvalues) PetscUseTypeMethod(ksp, computeeigenvalues, n, r, c, neig);
126: else *neig = 0;
127: PetscFunctionReturn(PETSC_SUCCESS);
128: }
130: /*@
131: KSPComputeRitz - Computes the Ritz or harmonic Ritz pairs associated with the
132: smallest or largest in modulus, for the preconditioned operator.
134: Not Collective
136: Input Parameters:
137: + ksp - iterative context obtained from `KSPCreate()`
138: . ritz - `PETSC_TRUE` or `PETSC_FALSE` for Ritz pairs or harmonic Ritz pairs, respectively
139: - small - `PETSC_TRUE` or `PETSC_FALSE` for smallest or largest (harmonic) Ritz values, respectively
141: Output Parameters:
142: + nrit - On input number of (harmonic) Ritz pairs to compute; on output, actual number of computed (harmonic) Ritz pairs
143: . S - an array of the Ritz vectors, pass in an array of vectors of size `nrit`
144: . tetar - real part of the Ritz values, pass in an array of size `nrit`
145: - tetai - imaginary part of the Ritz values, pass in an array of size `nrit`
147: Level: advanced
149: Notes:
150: This only works with a `KSPType` of `KSPGMRES`.
152: One must call `KSPSetComputeRitz()` before calling `KSPSetUp()` in order for this routine to work correctly.
154: This routine must be called after `KSPSolve()`.
156: In `KSPGMRES`, the (harmonic) Ritz pairs are computed from the Hessenberg matrix obtained during
157: the last complete cycle of the GMRES solve, or during the partial cycle if the solve ended before
158: a restart (that is a complete GMRES cycle was never achieved).
160: The number of actual (harmonic) Ritz pairs computed is less than or equal to the restart
161: parameter for GMRES if a complete cycle has been performed or less or equal to the number of GMRES
162: iterations.
164: `KSPComputeEigenvalues()` provides estimates for only the eigenvalues (Ritz values).
166: For real matrices, the (harmonic) Ritz pairs can be complex-valued. In such a case,
167: the routine selects the complex (harmonic) Ritz value and its conjugate, and two successive entries of the
168: vectors `S` are equal to the real and the imaginary parts of the associated vectors.
169: When PETSc has been built with complex scalars, the real and imaginary parts of the Ritz
170: values are still returned in `tetar` and `tetai`, as is done in `KSPComputeEigenvalues()`, but
171: the Ritz vectors S are complex.
173: The (harmonic) Ritz pairs are given in order of increasing (harmonic) Ritz values in modulus.
175: The Ritz pairs do not necessarily accurately reflect the eigenvalues and eigenvectors of the operator, consider the
176: excellent package `SLEPc` if accurate values are required.
178: .seealso: [](ch_ksp), `KSPSetComputeRitz()`, `KSP`, `KSPGMRES`, `KSPComputeEigenvalues()`, `KSPSetComputeSingularValues()`, `KSPMonitorSingularValue()`
179: @*/
180: PetscErrorCode KSPComputeRitz(KSP ksp, PetscBool ritz, PetscBool small, PetscInt *nrit, Vec S[], PetscReal tetar[], PetscReal tetai[])
181: {
182: PetscFunctionBegin;
184: PetscCheck(ksp->calc_ritz, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "Ritz pairs not requested before KSPSetUp()");
185: PetscTryTypeMethod(ksp, computeritz, ritz, small, nrit, S, tetar, tetai);
186: PetscFunctionReturn(PETSC_SUCCESS);
187: }
188: /*@
189: KSPSetUpOnBlocks - Sets up the preconditioner for each block in
190: the block Jacobi, block Gauss-Seidel, and overlapping Schwarz
191: methods.
193: Collective
195: Input Parameter:
196: . ksp - the `KSP` context
198: Level: advanced
200: Notes:
201: `KSPSetUpOnBlocks()` is a routine that the user can optionally call for
202: more precise profiling (via -log_view) of the setup phase for these
203: block preconditioners. If the user does not call `KSPSetUpOnBlocks()`,
204: it will automatically be called from within `KSPSolve()`.
206: Calling `KSPSetUpOnBlocks()` is the same as calling `PCSetUpOnBlocks()`
207: on the PC context within the `KSP` context.
209: .seealso: [](ch_ksp), `PCSetUpOnBlocks()`, `KSPSetUp()`, `PCSetUp()`, `KSP`
210: @*/
211: PetscErrorCode KSPSetUpOnBlocks(KSP ksp)
212: {
213: PC pc;
214: PCFailedReason pcreason;
216: PetscFunctionBegin;
218: level++;
219: PetscCall(KSPGetPC(ksp, &pc));
220: PetscCall(PCSetUpOnBlocks(pc));
221: PetscCall(PCGetFailedReasonRank(pc, &pcreason));
222: level--;
223: /*
224: This is tricky since only a subset of MPI ranks may set this; each KSPSolve_*() is responsible for checking
225: this flag and initializing an appropriate vector with VecSetInf() so that the first norm computation can
226: produce a result at KSPCheckNorm() thus communicating the known problem to all MPI ranks so they may
227: terminate the Krylov solve. For many KSP implementations this is handled within KSPInitialResidual()
228: */
229: if (pcreason) ksp->reason = KSP_DIVERGED_PC_FAILED;
230: PetscFunctionReturn(PETSC_SUCCESS);
231: }
233: /*@
234: KSPSetReusePreconditioner - reuse the current preconditioner, do not construct a new one even if the operator changes
236: Collective
238: Input Parameters:
239: + ksp - iterative context obtained from `KSPCreate()`
240: - flag - `PETSC_TRUE` to reuse the current preconditioner
242: Level: intermediate
244: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `PCSetReusePreconditioner()`, `KSP`
245: @*/
246: PetscErrorCode KSPSetReusePreconditioner(KSP ksp, PetscBool flag)
247: {
248: PC pc;
250: PetscFunctionBegin;
252: PetscCall(KSPGetPC(ksp, &pc));
253: PetscCall(PCSetReusePreconditioner(pc, flag));
254: PetscFunctionReturn(PETSC_SUCCESS);
255: }
257: /*@
258: KSPGetReusePreconditioner - Determines if the `KSP` reuses the current preconditioner even if the operator in the preconditioner has changed.
260: Collective
262: Input Parameter:
263: . ksp - iterative context obtained from `KSPCreate()`
265: Output Parameter:
266: . flag - the boolean flag
268: Level: intermediate
270: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `KSPSetReusePreconditioner()`, `KSP`
271: @*/
272: PetscErrorCode KSPGetReusePreconditioner(KSP ksp, PetscBool *flag)
273: {
274: PetscFunctionBegin;
276: PetscAssertPointer(flag, 2);
277: *flag = PETSC_FALSE;
278: if (ksp->pc) PetscCall(PCGetReusePreconditioner(ksp->pc, flag));
279: PetscFunctionReturn(PETSC_SUCCESS);
280: }
282: /*@
283: KSPSetSkipPCSetFromOptions - prevents `KSPSetFromOptions()` from calling `PCSetFromOptions()`. This is used if the same `PC` is shared by more than one `KSP` so its options are not resettable for each `KSP`
285: Collective
287: Input Parameters:
288: + ksp - iterative context obtained from `KSPCreate()`
289: - flag - `PETSC_TRUE` to skip calling the `PCSetFromOptions()`
291: Level: intermediate
293: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `PCSetReusePreconditioner()`, `KSP`
294: @*/
295: PetscErrorCode KSPSetSkipPCSetFromOptions(KSP ksp, PetscBool flag)
296: {
297: PetscFunctionBegin;
299: ksp->skippcsetfromoptions = flag;
300: PetscFunctionReturn(PETSC_SUCCESS);
301: }
303: /*@
304: KSPSetUp - Sets up the internal data structures for the
305: later use of an iterative solver.
307: Collective
309: Input Parameter:
310: . ksp - iterative context obtained from `KSPCreate()`
312: Level: developer
314: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSolve()`, `KSPDestroy()`, `KSP`
315: @*/
316: PetscErrorCode KSPSetUp(KSP ksp)
317: {
318: Mat A, B;
319: Mat mat, pmat;
320: MatNullSpace nullsp;
321: PCFailedReason pcreason;
322: PC pc;
323: PetscBool pcmpi;
325: PetscFunctionBegin;
327: PetscCall(KSPGetPC(ksp, &pc));
328: PetscCall(PetscObjectTypeCompare((PetscObject)pc, PCMPI, &pcmpi));
329: if (pcmpi) {
330: PetscBool ksppreonly;
331: PetscCall(PetscObjectTypeCompare((PetscObject)ksp, KSPPREONLY, &ksppreonly));
332: if (!ksppreonly) PetscCall(KSPSetType(ksp, KSPPREONLY));
333: }
334: level++;
336: /* reset the convergence flag from the previous solves */
337: ksp->reason = KSP_CONVERGED_ITERATING;
339: if (!((PetscObject)ksp)->type_name) PetscCall(KSPSetType(ksp, KSPGMRES));
340: PetscCall(KSPSetUpNorms_Private(ksp, PETSC_TRUE, &ksp->normtype, &ksp->pc_side));
342: if (ksp->dmActive && !ksp->setupstage) {
343: /* first time in so build matrix and vector data structures using DM */
344: if (!ksp->vec_rhs) PetscCall(DMCreateGlobalVector(ksp->dm, &ksp->vec_rhs));
345: if (!ksp->vec_sol) PetscCall(DMCreateGlobalVector(ksp->dm, &ksp->vec_sol));
346: PetscCall(DMCreateMatrix(ksp->dm, &A));
347: PetscCall(KSPSetOperators(ksp, A, A));
348: PetscCall(PetscObjectDereference((PetscObject)A));
349: }
351: if (ksp->dmActive) {
352: DMKSP kdm;
353: PetscCall(DMGetDMKSP(ksp->dm, &kdm));
355: if (kdm->ops->computeinitialguess && ksp->setupstage != KSP_SETUP_NEWRHS) {
356: /* only computes initial guess the first time through */
357: PetscCallBack("KSP callback initial guess", (*kdm->ops->computeinitialguess)(ksp, ksp->vec_sol, kdm->initialguessctx));
358: PetscCall(KSPSetInitialGuessNonzero(ksp, PETSC_TRUE));
359: }
360: if (kdm->ops->computerhs) PetscCallBack("KSP callback rhs", (*kdm->ops->computerhs)(ksp, ksp->vec_rhs, kdm->rhsctx));
362: if (ksp->setupstage != KSP_SETUP_NEWRHS) {
363: PetscCheck(kdm->ops->computeoperators, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONGSTATE, "You called KSPSetDM() but did not use DMKSPSetComputeOperators() or KSPSetDMActive(ksp,PETSC_FALSE);");
364: PetscCall(KSPGetOperators(ksp, &A, &B));
365: PetscCallBack("KSP callback operators", (*kdm->ops->computeoperators)(ksp, A, B, kdm->operatorsctx));
366: }
367: }
369: if (ksp->setupstage == KSP_SETUP_NEWRHS) {
370: level--;
371: PetscFunctionReturn(PETSC_SUCCESS);
372: }
373: PetscCall(PetscLogEventBegin(KSP_SetUp, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
375: switch (ksp->setupstage) {
376: case KSP_SETUP_NEW:
377: PetscUseTypeMethod(ksp, setup);
378: break;
379: case KSP_SETUP_NEWMATRIX: /* This should be replaced with a more general mechanism */
380: if (ksp->setupnewmatrix) PetscUseTypeMethod(ksp, setup);
381: break;
382: default:
383: break;
384: }
386: if (!ksp->pc) PetscCall(KSPGetPC(ksp, &ksp->pc));
387: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
388: /* scale the matrix if requested */
389: if (ksp->dscale) {
390: PetscScalar *xx;
391: PetscInt i, n;
392: PetscBool zeroflag = PETSC_FALSE;
394: if (!ksp->diagonal) { /* allocate vector to hold diagonal */
395: PetscCall(MatCreateVecs(pmat, &ksp->diagonal, NULL));
396: }
397: PetscCall(MatGetDiagonal(pmat, ksp->diagonal));
398: PetscCall(VecGetLocalSize(ksp->diagonal, &n));
399: PetscCall(VecGetArray(ksp->diagonal, &xx));
400: for (i = 0; i < n; i++) {
401: if (xx[i] != 0.0) xx[i] = 1.0 / PetscSqrtReal(PetscAbsScalar(xx[i]));
402: else {
403: xx[i] = 1.0;
404: zeroflag = PETSC_TRUE;
405: }
406: }
407: PetscCall(VecRestoreArray(ksp->diagonal, &xx));
408: if (zeroflag) PetscCall(PetscInfo(ksp, "Zero detected in diagonal of matrix, using 1 at those locations\n"));
409: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
410: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
411: ksp->dscalefix2 = PETSC_FALSE;
412: }
413: PetscCall(PetscLogEventEnd(KSP_SetUp, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
414: PetscCall(PCSetErrorIfFailure(ksp->pc, ksp->errorifnotconverged));
415: PetscCall(PCSetUp(ksp->pc));
416: PetscCall(PCGetFailedReasonRank(ksp->pc, &pcreason));
417: /* 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 */
418: if (pcreason) ksp->reason = KSP_DIVERGED_PC_FAILED;
420: PetscCall(MatGetNullSpace(mat, &nullsp));
421: if (nullsp) {
422: PetscBool test = PETSC_FALSE;
423: PetscCall(PetscOptionsGetBool(((PetscObject)ksp)->options, ((PetscObject)ksp)->prefix, "-ksp_test_null_space", &test, NULL));
424: if (test) PetscCall(MatNullSpaceTest(nullsp, mat, NULL));
425: }
426: ksp->setupstage = KSP_SETUP_NEWRHS;
427: level--;
428: PetscFunctionReturn(PETSC_SUCCESS);
429: }
431: /*@C
432: KSPConvergedReasonView - Displays the reason a `KSP` solve converged or diverged to a viewer
434: Collective
436: Input Parameters:
437: + ksp - iterative context obtained from `KSPCreate()`
438: - viewer - the viewer to display the reason
440: Options Database Keys:
441: + -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
442: - -ksp_converged_reason ::failed - only print reason and number of iterations when diverged
444: Level: beginner
446: Note:
447: To change the format of the output call `PetscViewerPushFormat`(`viewer`,`format`) before this call. Use `PETSC_VIEWER_DEFAULT` for the default,
448: use `PETSC_VIEWER_FAILED` to only display a reason if it fails.
450: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
451: `KSPSolveTranspose()`, `KSPGetIterationNumber()`, `KSP`, `KSPGetConvergedReason()`, `PetscViewerPushFormat()`, `PetscViewerPopFormat()`
452: @*/
453: PetscErrorCode KSPConvergedReasonView(KSP ksp, PetscViewer viewer)
454: {
455: PetscBool isAscii;
456: PetscViewerFormat format;
458: PetscFunctionBegin;
459: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
460: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isAscii));
461: if (isAscii) {
462: PetscCall(PetscViewerGetFormat(viewer, &format));
463: PetscCall(PetscViewerASCIIAddTab(viewer, ((PetscObject)ksp)->tablevel));
464: if (ksp->reason > 0 && format != PETSC_VIEWER_FAILED) {
465: if (((PetscObject)ksp)->prefix) {
466: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve converged due to %s iterations %" PetscInt_FMT "\n", ((PetscObject)ksp)->prefix, KSPConvergedReasons[ksp->reason], ksp->its));
467: } else {
468: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve converged due to %s iterations %" PetscInt_FMT "\n", KSPConvergedReasons[ksp->reason], ksp->its));
469: }
470: } else if (ksp->reason <= 0) {
471: if (((PetscObject)ksp)->prefix) {
472: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve did not converge due to %s iterations %" PetscInt_FMT "\n", ((PetscObject)ksp)->prefix, KSPConvergedReasons[ksp->reason], ksp->its));
473: } else {
474: PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve did not converge due to %s iterations %" PetscInt_FMT "\n", KSPConvergedReasons[ksp->reason], ksp->its));
475: }
476: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
477: PCFailedReason reason;
478: PetscCall(PCGetFailedReason(ksp->pc, &reason));
479: PetscCall(PetscViewerASCIIPrintf(viewer, " PC failed due to %s \n", PCFailedReasons[reason]));
480: }
481: }
482: PetscCall(PetscViewerASCIISubtractTab(viewer, ((PetscObject)ksp)->tablevel));
483: }
484: PetscFunctionReturn(PETSC_SUCCESS);
485: }
487: /*@C
488: KSPConvergedReasonViewSet - Sets an ADDITIONAL function that is to be used at the
489: end of the linear solver to display the convergence reason of the linear solver.
491: Logically Collective
493: Input Parameters:
494: + ksp - the `KSP` context
495: . f - the ksp converged reason view function
496: . vctx - [optional] user-defined context for private data for the
497: ksp converged reason view routine (use `NULL` if no context is desired)
498: - reasonviewdestroy - [optional] routine that frees reasonview context
499: (may be `NULL`)
501: Options Database Keys:
502: + -ksp_converged_reason - sets a default `KSPConvergedReasonView()`
503: - -ksp_converged_reason_view_cancel - cancels all converged reason viewers that have
504: been hardwired into a code by
505: calls to `KSPConvergedReasonViewSet()`, but
506: does not cancel those set via
507: the options database.
509: Level: intermediate
511: Note:
512: Several different converged reason view routines may be set by calling
513: `KSPConvergedReasonViewSet()` multiple times; all will be called in the
514: order in which they were set.
516: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPConvergedReasonViewCancel()`
517: @*/
518: PetscErrorCode KSPConvergedReasonViewSet(KSP ksp, PetscErrorCode (*f)(KSP, void *), void *vctx, PetscErrorCode (*reasonviewdestroy)(void **))
519: {
520: PetscInt i;
521: PetscBool identical;
523: PetscFunctionBegin;
525: for (i = 0; i < ksp->numberreasonviews; i++) {
526: PetscCall(PetscMonitorCompare((PetscErrorCode(*)(void))f, vctx, reasonviewdestroy, (PetscErrorCode(*)(void))ksp->reasonview[i], ksp->reasonviewcontext[i], ksp->reasonviewdestroy[i], &identical));
527: if (identical) PetscFunctionReturn(PETSC_SUCCESS);
528: }
529: PetscCheck(ksp->numberreasonviews < MAXKSPREASONVIEWS, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Too many KSP reasonview set");
530: ksp->reasonview[ksp->numberreasonviews] = f;
531: ksp->reasonviewdestroy[ksp->numberreasonviews] = reasonviewdestroy;
532: ksp->reasonviewcontext[ksp->numberreasonviews++] = (void *)vctx;
533: PetscFunctionReturn(PETSC_SUCCESS);
534: }
536: /*@
537: KSPConvergedReasonViewCancel - Clears all the reasonview functions for a `KSP` object set with `KSPConvergedReasonViewSet()`.
539: Collective
541: Input Parameter:
542: . ksp - iterative context obtained from `KSPCreate()`
544: Level: intermediate
546: .seealso: [](ch_ksp), `KSPCreate()`, `KSPDestroy()`, `KSPReset()`, `KSPConvergedReasonViewSet()`
547: @*/
548: PetscErrorCode KSPConvergedReasonViewCancel(KSP ksp)
549: {
550: PetscInt i;
552: PetscFunctionBegin;
554: for (i = 0; i < ksp->numberreasonviews; i++) {
555: if (ksp->reasonviewdestroy[i]) PetscCall((*ksp->reasonviewdestroy[i])(&ksp->reasonviewcontext[i]));
556: }
557: ksp->numberreasonviews = 0;
558: PetscFunctionReturn(PETSC_SUCCESS);
559: }
561: /*@
562: KSPConvergedReasonViewFromOptions - Processes command line options to determine if/how a `KSPReason` is to be viewed.
564: Collective
566: Input Parameter:
567: . ksp - the `KSP` object
569: Level: intermediate
571: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPConvergedReasonViewSet()`
572: @*/
573: PetscErrorCode KSPConvergedReasonViewFromOptions(KSP ksp)
574: {
575: PetscViewer viewer;
576: PetscBool flg;
577: PetscViewerFormat format;
578: PetscInt i;
580: PetscFunctionBegin;
582: /* Call all user-provided reason review routines */
583: for (i = 0; i < ksp->numberreasonviews; i++) PetscCall((*ksp->reasonview[i])(ksp, ksp->reasonviewcontext[i]));
585: /* Call the default PETSc routine */
586: PetscCall(PetscOptionsGetViewer(PetscObjectComm((PetscObject)ksp), ((PetscObject)ksp)->options, ((PetscObject)ksp)->prefix, "-ksp_converged_reason", &viewer, &format, &flg));
587: if (flg) {
588: PetscCall(PetscViewerPushFormat(viewer, format));
589: PetscCall(KSPConvergedReasonView(ksp, viewer));
590: PetscCall(PetscViewerPopFormat(viewer));
591: PetscCall(PetscViewerDestroy(&viewer));
592: }
593: PetscFunctionReturn(PETSC_SUCCESS);
594: }
596: /*@C
597: KSPConvergedRateView - Displays the convergence rate <https://en.wikipedia.org/wiki/Coefficient_of_determination> of `KSPSolve()` to a viewer
599: Collective
601: Input Parameters:
602: + ksp - iterative context obtained from `KSPCreate()`
603: - viewer - the viewer to display the reason
605: Options Database Key:
606: . -ksp_converged_rate - print reason for convergence or divergence and the convergence rate (or 0.0 for divergence)
608: Level: intermediate
610: Notes:
611: To change the format of the output, call `PetscViewerPushFormat`(`viewer`,`format`) before this call.
613: 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,
614: the slope is $\log c$. The coefficient of determination is given by $1 - \frac{\sum_i (y_i - f(x_i))^2}{\sum_i (y_i - \bar y)}$,
616: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPGetConvergedRate()`, `KSPSetTolerances()`, `KSPConvergedDefault()`
617: @*/
618: PetscErrorCode KSPConvergedRateView(KSP ksp, PetscViewer viewer)
619: {
620: PetscViewerFormat format;
621: PetscBool isAscii;
622: PetscReal rrate, rRsq, erate = 0.0, eRsq = 0.0;
623: PetscInt its;
624: const char *prefix, *reason = KSPConvergedReasons[ksp->reason];
626: PetscFunctionBegin;
627: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
628: PetscCall(KSPGetIterationNumber(ksp, &its));
629: PetscCall(KSPComputeConvergenceRate(ksp, &rrate, &rRsq, &erate, &eRsq));
630: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
631: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isAscii));
632: if (isAscii) {
633: PetscCall(PetscViewerGetFormat(viewer, &format));
634: PetscCall(PetscViewerASCIIAddTab(viewer, ((PetscObject)ksp)->tablevel));
635: if (ksp->reason > 0) {
636: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve converged due to %s iterations %" PetscInt_FMT, prefix, reason, its));
637: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve converged due to %s iterations %" PetscInt_FMT, reason, its));
638: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
639: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
640: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
641: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
642: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
643: } else if (ksp->reason <= 0) {
644: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve did not converge due to %s iterations %" PetscInt_FMT, prefix, reason, its));
645: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve did not converge due to %s iterations %" PetscInt_FMT, reason, its));
646: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
647: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
648: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
649: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
650: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
651: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
652: PCFailedReason reason;
653: PetscCall(PCGetFailedReason(ksp->pc, &reason));
654: PetscCall(PetscViewerASCIIPrintf(viewer, " PC failed due to %s \n", PCFailedReasons[reason]));
655: }
656: }
657: PetscCall(PetscViewerASCIISubtractTab(viewer, ((PetscObject)ksp)->tablevel));
658: }
659: PetscFunctionReturn(PETSC_SUCCESS);
660: }
662: #include <petscdraw.h>
664: static PetscErrorCode KSPViewEigenvalues_Internal(KSP ksp, PetscBool isExplicit, PetscViewer viewer, PetscViewerFormat format)
665: {
666: PetscReal *r, *c;
667: PetscInt n, i, neig;
668: PetscBool isascii, isdraw;
669: PetscMPIInt rank;
671: PetscFunctionBegin;
672: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)ksp), &rank));
673: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
674: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
675: if (isExplicit) {
676: PetscCall(VecGetSize(ksp->vec_sol, &n));
677: PetscCall(PetscMalloc2(n, &r, n, &c));
678: PetscCall(KSPComputeEigenvaluesExplicitly(ksp, n, r, c));
679: neig = n;
680: } else {
681: PetscInt nits;
683: PetscCall(KSPGetIterationNumber(ksp, &nits));
684: n = nits + 2;
685: if (!nits) {
686: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any eigenvalues\n"));
687: PetscFunctionReturn(PETSC_SUCCESS);
688: }
689: PetscCall(PetscMalloc2(n, &r, n, &c));
690: PetscCall(KSPComputeEigenvalues(ksp, n, r, c, &neig));
691: }
692: if (isascii) {
693: PetscCall(PetscViewerASCIIPrintf(viewer, "%s computed eigenvalues\n", isExplicit ? "Explicitly" : "Iteratively"));
694: for (i = 0; i < neig; ++i) {
695: if (c[i] >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, "%g + %gi\n", (double)r[i], (double)c[i]));
696: else PetscCall(PetscViewerASCIIPrintf(viewer, "%g - %gi\n", (double)r[i], -(double)c[i]));
697: }
698: } else if (isdraw && rank == 0) {
699: PetscDraw draw;
700: PetscDrawSP drawsp;
702: if (format == PETSC_VIEWER_DRAW_CONTOUR) {
703: PetscCall(KSPPlotEigenContours_Private(ksp, neig, r, c));
704: } else {
705: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
706: PetscCall(PetscDrawSPCreate(draw, 1, &drawsp));
707: PetscCall(PetscDrawSPReset(drawsp));
708: for (i = 0; i < neig; ++i) PetscCall(PetscDrawSPAddPoint(drawsp, r + i, c + i));
709: PetscCall(PetscDrawSPDraw(drawsp, PETSC_TRUE));
710: PetscCall(PetscDrawSPSave(drawsp));
711: PetscCall(PetscDrawSPDestroy(&drawsp));
712: }
713: }
714: PetscCall(PetscFree2(r, c));
715: PetscFunctionReturn(PETSC_SUCCESS);
716: }
718: static PetscErrorCode KSPViewSingularvalues_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
719: {
720: PetscReal smax, smin;
721: PetscInt nits;
722: PetscBool isascii;
724: PetscFunctionBegin;
725: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
726: PetscCall(KSPGetIterationNumber(ksp, &nits));
727: if (!nits) {
728: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any singular values\n"));
729: PetscFunctionReturn(PETSC_SUCCESS);
730: }
731: PetscCall(KSPComputeExtremeSingularValues(ksp, &smax, &smin));
732: if (isascii) PetscCall(PetscViewerASCIIPrintf(viewer, "Iteratively computed extreme singular values: max %g min %g max/min %g\n", (double)smax, (double)smin, (double)(smax / smin)));
733: PetscFunctionReturn(PETSC_SUCCESS);
734: }
736: static PetscErrorCode KSPViewFinalResidual_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
737: {
738: PetscBool isascii;
740: PetscFunctionBegin;
741: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
742: 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");
743: if (isascii) {
744: Mat A;
745: Vec t;
746: PetscReal norm;
748: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
749: PetscCall(VecDuplicate(ksp->vec_rhs, &t));
750: PetscCall(KSP_MatMult(ksp, A, ksp->vec_sol, t));
751: PetscCall(VecAYPX(t, -1.0, ksp->vec_rhs));
752: PetscCall(VecViewFromOptions(t, (PetscObject)ksp, "-ksp_view_final_residual_vec"));
753: PetscCall(VecNorm(t, NORM_2, &norm));
754: PetscCall(VecDestroy(&t));
755: PetscCall(PetscViewerASCIIPrintf(viewer, "KSP final norm of residual %g\n", (double)norm));
756: }
757: PetscFunctionReturn(PETSC_SUCCESS);
758: }
760: static PetscErrorCode KSPMonitorPauseFinal_Internal(KSP ksp)
761: {
762: PetscInt i;
764: PetscFunctionBegin;
765: if (!ksp->pauseFinal) PetscFunctionReturn(PETSC_SUCCESS);
766: for (i = 0; i < ksp->numbermonitors; ++i) {
767: PetscViewerAndFormat *vf = (PetscViewerAndFormat *)ksp->monitorcontext[i];
768: PetscDraw draw;
769: PetscReal lpause;
771: if (!vf) continue;
772: if (vf->lg) {
773: if (!PetscCheckPointer(vf->lg, PETSC_OBJECT)) continue;
774: if (((PetscObject)vf->lg)->classid != PETSC_DRAWLG_CLASSID) continue;
775: PetscCall(PetscDrawLGGetDraw(vf->lg, &draw));
776: PetscCall(PetscDrawGetPause(draw, &lpause));
777: PetscCall(PetscDrawSetPause(draw, -1.0));
778: PetscCall(PetscDrawPause(draw));
779: PetscCall(PetscDrawSetPause(draw, lpause));
780: } else {
781: PetscBool isdraw;
783: if (!PetscCheckPointer(vf->viewer, PETSC_OBJECT)) continue;
784: if (((PetscObject)vf->viewer)->classid != PETSC_VIEWER_CLASSID) continue;
785: PetscCall(PetscObjectTypeCompare((PetscObject)vf->viewer, PETSCVIEWERDRAW, &isdraw));
786: 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: }
794: PetscFunctionReturn(PETSC_SUCCESS);
795: }
797: static PetscErrorCode KSPSolve_Private(KSP ksp, Vec b, Vec x)
798: {
799: PetscBool flg = PETSC_FALSE, inXisinB = PETSC_FALSE, guess_zero;
800: Mat mat, pmat;
801: MPI_Comm comm;
802: MatNullSpace nullsp;
803: Vec btmp, vec_rhs = NULL;
805: PetscFunctionBegin;
806: level++;
807: comm = PetscObjectComm((PetscObject)ksp);
808: if (x && x == b) {
809: PetscCheck(ksp->guess_zero, comm, PETSC_ERR_ARG_INCOMP, "Cannot use x == b with nonzero initial guess");
810: PetscCall(VecDuplicate(b, &x));
811: inXisinB = PETSC_TRUE;
812: }
813: if (b) {
814: PetscCall(PetscObjectReference((PetscObject)b));
815: PetscCall(VecDestroy(&ksp->vec_rhs));
816: ksp->vec_rhs = b;
817: }
818: if (x) {
819: PetscCall(PetscObjectReference((PetscObject)x));
820: PetscCall(VecDestroy(&ksp->vec_sol));
821: ksp->vec_sol = x;
822: }
824: if (ksp->viewPre) PetscCall(ObjectView((PetscObject)ksp, ksp->viewerPre, ksp->formatPre));
826: if (ksp->presolve) PetscCall((*ksp->presolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->prectx));
828: /* reset the residual history list if requested */
829: if (ksp->res_hist_reset) ksp->res_hist_len = 0;
830: if (ksp->err_hist_reset) ksp->err_hist_len = 0;
832: /* KSPSetUp() scales the matrix if needed */
833: PetscCall(KSPSetUp(ksp));
834: PetscCall(KSPSetUpOnBlocks(ksp));
836: if (ksp->guess) {
837: PetscObjectState ostate, state;
839: PetscCall(KSPGuessSetUp(ksp->guess));
840: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &ostate));
841: PetscCall(KSPGuessFormGuess(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
842: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &state));
843: if (state != ostate) {
844: ksp->guess_zero = PETSC_FALSE;
845: } else {
846: PetscCall(PetscInfo(ksp, "Using zero initial guess since the KSPGuess object did not change the vector\n"));
847: ksp->guess_zero = PETSC_TRUE;
848: }
849: }
851: PetscCall(VecSetErrorIfLocked(ksp->vec_sol, 3));
853: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
854: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
855: /* diagonal scale RHS if called for */
856: if (ksp->dscale) {
857: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
858: /* second time in, but matrix was scaled back to original */
859: if (ksp->dscalefix && ksp->dscalefix2) {
860: Mat mat, pmat;
862: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
863: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
864: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
865: }
867: /* scale initial guess */
868: if (!ksp->guess_zero) {
869: if (!ksp->truediagonal) {
870: PetscCall(VecDuplicate(ksp->diagonal, &ksp->truediagonal));
871: PetscCall(VecCopy(ksp->diagonal, ksp->truediagonal));
872: PetscCall(VecReciprocal(ksp->truediagonal));
873: }
874: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->truediagonal));
875: }
876: }
877: PetscCall(PCPreSolve(ksp->pc, ksp));
879: if (ksp->guess_zero && !ksp->guess_not_read) PetscCall(VecSet(ksp->vec_sol, 0.0));
880: if (ksp->guess_knoll) { /* The Knoll trick is independent on the KSPGuess specified */
881: PetscCall(PCApply(ksp->pc, ksp->vec_rhs, ksp->vec_sol));
882: PetscCall(KSP_RemoveNullSpace(ksp, ksp->vec_sol));
883: ksp->guess_zero = PETSC_FALSE;
884: }
886: /* can we mark the initial guess as zero for this solve? */
887: guess_zero = ksp->guess_zero;
888: if (!ksp->guess_zero) {
889: PetscReal norm;
891: PetscCall(VecNormAvailable(ksp->vec_sol, NORM_2, &flg, &norm));
892: if (flg && !norm) ksp->guess_zero = PETSC_TRUE;
893: }
894: if (ksp->transpose_solve) {
895: PetscCall(MatGetNullSpace(pmat, &nullsp));
896: } else {
897: PetscCall(MatGetTransposeNullSpace(pmat, &nullsp));
898: }
899: if (nullsp) {
900: PetscCall(VecDuplicate(ksp->vec_rhs, &btmp));
901: PetscCall(VecCopy(ksp->vec_rhs, btmp));
902: PetscCall(MatNullSpaceRemove(nullsp, btmp));
903: vec_rhs = ksp->vec_rhs;
904: ksp->vec_rhs = btmp;
905: }
906: PetscCall(VecLockReadPush(ksp->vec_rhs));
907: PetscUseTypeMethod(ksp, solve);
908: PetscCall(KSPMonitorPauseFinal_Internal(ksp));
910: PetscCall(VecLockReadPop(ksp->vec_rhs));
911: if (nullsp) {
912: ksp->vec_rhs = vec_rhs;
913: PetscCall(VecDestroy(&btmp));
914: }
916: ksp->guess_zero = guess_zero;
918: PetscCheck(ksp->reason, comm, PETSC_ERR_PLIB, "Internal error, solver returned without setting converged reason");
919: ksp->totalits += ksp->its;
921: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
923: if (ksp->viewRate) {
924: PetscCall(PetscViewerPushFormat(ksp->viewerRate, ksp->formatRate));
925: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
926: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
927: }
928: PetscCall(PCPostSolve(ksp->pc, ksp));
930: /* diagonal scale solution if called for */
931: if (ksp->dscale) {
932: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->diagonal));
933: /* unscale right hand side and matrix */
934: if (ksp->dscalefix) {
935: Mat mat, pmat;
937: PetscCall(VecReciprocal(ksp->diagonal));
938: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
939: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
940: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
941: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
942: PetscCall(VecReciprocal(ksp->diagonal));
943: ksp->dscalefix2 = PETSC_TRUE;
944: }
945: }
946: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
947: if (ksp->guess) PetscCall(KSPGuessUpdate(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
948: if (ksp->postsolve) PetscCall((*ksp->postsolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->postctx));
950: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
951: if (ksp->viewEV) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_FALSE, ksp->viewerEV, ksp->formatEV));
952: if (ksp->viewEVExp) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_TRUE, ksp->viewerEVExp, ksp->formatEVExp));
953: if (ksp->viewSV) PetscCall(KSPViewSingularvalues_Internal(ksp, ksp->viewerSV, ksp->formatSV));
954: if (ksp->viewFinalRes) PetscCall(KSPViewFinalResidual_Internal(ksp, ksp->viewerFinalRes, ksp->formatFinalRes));
955: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)mat, ksp->viewerMat, ksp->formatMat));
956: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)pmat, ksp->viewerPMat, ksp->formatPMat));
957: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)ksp->vec_rhs, ksp->viewerRhs, ksp->formatRhs));
958: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)ksp->vec_sol, ksp->viewerSol, ksp->formatSol));
959: if (ksp->view) PetscCall(ObjectView((PetscObject)ksp, ksp->viewer, ksp->format));
960: if (ksp->viewDScale) PetscCall(ObjectView((PetscObject)ksp->diagonal, ksp->viewerDScale, ksp->formatDScale));
961: if (ksp->viewMatExp) {
962: Mat A, B;
964: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
965: if (ksp->transpose_solve) {
966: Mat AT;
968: PetscCall(MatCreateTranspose(A, &AT));
969: PetscCall(MatComputeOperator(AT, MATAIJ, &B));
970: PetscCall(MatDestroy(&AT));
971: } else {
972: PetscCall(MatComputeOperator(A, MATAIJ, &B));
973: }
974: PetscCall(ObjectView((PetscObject)B, ksp->viewerMatExp, ksp->formatMatExp));
975: PetscCall(MatDestroy(&B));
976: }
977: if (ksp->viewPOpExp) {
978: Mat B;
980: PetscCall(KSPComputeOperator(ksp, MATAIJ, &B));
981: PetscCall(ObjectView((PetscObject)B, ksp->viewerPOpExp, ksp->formatPOpExp));
982: PetscCall(MatDestroy(&B));
983: }
985: if (inXisinB) {
986: PetscCall(VecCopy(x, b));
987: PetscCall(VecDestroy(&x));
988: }
989: PetscCall(PetscObjectSAWsBlock((PetscObject)ksp));
990: if (ksp->errorifnotconverged && ksp->reason < 0 && ((level == 1) || (ksp->reason != KSP_DIVERGED_ITS))) {
991: PCFailedReason reason;
993: 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]);
994: PetscCall(PCGetFailedReason(ksp->pc, &reason));
995: 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]);
996: }
997: level--;
998: PetscFunctionReturn(PETSC_SUCCESS);
999: }
1001: /*@
1002: KSPSolve - Solves linear system.
1004: Collective
1006: Input Parameters:
1007: + ksp - iterative context obtained from `KSPCreate()`
1008: . b - the right hand side vector
1009: - x - the solution (this may be the same vector as `b`, then `b` will be overwritten with answer)
1011: Options Database Keys:
1012: + -ksp_view_eigenvalues - compute preconditioned operators eigenvalues
1013: . -ksp_view_eigenvalues_explicit - compute the eigenvalues by forming the dense operator and using LAPACK
1014: . -ksp_view_mat binary - save matrix to the default binary viewer
1015: . -ksp_view_pmat binary - save matrix used to build preconditioner to the default binary viewer
1016: . -ksp_view_rhs binary - save right hand side vector to the default binary viewer
1017: . -ksp_view_solution binary - save computed solution vector to the default binary viewer
1018: (can be read later with src/ksp/tutorials/ex10.c for testing solvers)
1019: . -ksp_view_mat_explicit - for matrix-free operators, computes the matrix entries and views them
1020: . -ksp_view_preconditioned_operator_explicit - computes the product of the preconditioner and matrix as an explicit matrix and views it
1021: . -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
1022: . -ksp_view_final_residual - print 2-norm of true linear system residual at the end of the solution process
1023: . -ksp_error_if_not_converged - stop the program as soon as an error is detected in a `KSPSolve()`
1024: . -ksp_view_pre - print the ksp data structure before the system solution
1025: - -ksp_view - print the ksp data structure at the end of the system solution
1027: Level: beginner
1029: Notes:
1030: If one uses `KSPSetDM()` then `x` or `b` need not be passed. Use `KSPGetSolution()` to access the solution in this case.
1032: The operator is specified with `KSPSetOperators()`.
1034: `KSPSolve()` will normally return without generating an error regardless of whether the linear system was solved or if constructing the preconditioner failed.
1035: Call `KSPGetConvergedReason()` to determine if the solver converged or failed and why. The option -ksp_error_if_not_converged or function `KSPSetErrorIfNotConverged()`
1036: 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
1037: it may be fine that inner solvers in the preconditioner do not converge during the solution process.
1039: The number of iterations can be obtained from `KSPGetIterationNumber()`.
1041: If you provide a matrix that has a `MatSetNullSpace()` and `MatSetTransposeNullSpace()` this will use that information to solve singular systems
1042: in the least squares sense with a norm minimizing solution.
1044: 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()`
1046: `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
1047: 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
1048: direction thus the solution which is a linear combination of the search directions has no component in the nullspace(A).
1050: We recommend always using `KSPGMRES` for such singular systems.
1051: If nullspace(A) = nullspace(A') (note symmetric matrices always satisfy this property) then both left and right preconditioning will work
1052: If nullspace(A) != nullspace(A') then left preconditioning will work but right preconditioning may not work (or it may).
1054: Developer Notes:
1055: The reason we cannot always solve nullspace(A) != nullspace(A') systems with right preconditioning is because we need to remove at each iteration
1056: the nullspace(AB) from the search direction. While we know the nullspace(A) the nullspace(AB) equals B^-1 times the nullspace(A) but except for trivial preconditioners
1057: such as diagonal scaling we cannot apply the inverse of the preconditioner to a vector and thus cannot compute the nullspace(AB).
1059: If using a direct method (e.g., via the `KSP` solver
1060: `KSPPREONLY` and a preconditioner such as `PCLU` or `PCILU`,
1061: then its=1. See `KSPSetTolerances()` and `KSPConvergedDefault()`
1062: for more details.
1064: Understanding Convergence\:
1065: The routines `KSPMonitorSet()`, `KSPComputeEigenvalues()`, and
1066: `KSPComputeEigenvaluesExplicitly()` provide information on additional
1067: options to monitor convergence and print eigenvalue information.
1069: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1070: `KSPSolveTranspose()`, `KSPGetIterationNumber()`, `MatNullSpaceCreate()`, `MatSetNullSpace()`, `MatSetTransposeNullSpace()`, `KSP`,
1071: `KSPConvergedReasonView()`, `KSPCheckSolve()`, `KSPSetErrorIfNotConverged()`
1072: @*/
1073: PetscErrorCode KSPSolve(KSP ksp, Vec b, Vec x)
1074: {
1075: PetscFunctionBegin;
1079: ksp->transpose_solve = PETSC_FALSE;
1080: PetscCall(KSPSolve_Private(ksp, b, x));
1081: PetscFunctionReturn(PETSC_SUCCESS);
1082: }
1084: /*@
1085: KSPSolveTranspose - Solves a linear system with the transpose of the matrix, $ A^T x = b$.
1087: Collective
1089: Input Parameters:
1090: + ksp - iterative context obtained from `KSPCreate()`
1091: . b - right hand side vector
1092: - x - solution vector
1094: Level: developer
1096: Note:
1097: For complex numbers this solve the non-Hermitian transpose system.
1099: Developer Note:
1100: We need to implement a `KSPSolveHermitianTranspose()`
1102: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1103: `KSPSolve()`, `KSP`
1104: @*/
1105: PetscErrorCode KSPSolveTranspose(KSP ksp, Vec b, Vec x)
1106: {
1107: PetscFunctionBegin;
1111: if (ksp->transpose.use_explicittranspose) {
1112: Mat J, Jpre;
1113: PetscCall(KSPGetOperators(ksp, &J, &Jpre));
1114: if (!ksp->transpose.reuse_transpose) {
1115: PetscCall(MatTranspose(J, MAT_INITIAL_MATRIX, &ksp->transpose.AT));
1116: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_INITIAL_MATRIX, &ksp->transpose.BT));
1117: ksp->transpose.reuse_transpose = PETSC_TRUE;
1118: } else {
1119: PetscCall(MatTranspose(J, MAT_REUSE_MATRIX, &ksp->transpose.AT));
1120: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_REUSE_MATRIX, &ksp->transpose.BT));
1121: }
1122: if (J == Jpre && ksp->transpose.BT != ksp->transpose.AT) {
1123: PetscCall(PetscObjectReference((PetscObject)ksp->transpose.AT));
1124: ksp->transpose.BT = ksp->transpose.AT;
1125: }
1126: PetscCall(KSPSetOperators(ksp, ksp->transpose.AT, ksp->transpose.BT));
1127: } else {
1128: ksp->transpose_solve = PETSC_TRUE;
1129: }
1130: PetscCall(KSPSolve_Private(ksp, b, x));
1131: PetscFunctionReturn(PETSC_SUCCESS);
1132: }
1134: static PetscErrorCode KSPViewFinalMatResidual_Internal(KSP ksp, Mat B, Mat X, PetscViewer viewer, PetscViewerFormat format, PetscInt shift)
1135: {
1136: Mat A, R;
1137: PetscReal *norms;
1138: PetscInt i, N;
1139: PetscBool flg;
1141: PetscFunctionBegin;
1142: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &flg));
1143: if (flg) {
1144: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
1145: if (!ksp->transpose_solve) PetscCall(MatMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DEFAULT, &R));
1146: else PetscCall(MatTransposeMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DEFAULT, &R));
1147: PetscCall(MatAYPX(R, -1.0, B, SAME_NONZERO_PATTERN));
1148: PetscCall(MatGetSize(R, NULL, &N));
1149: PetscCall(PetscMalloc1(N, &norms));
1150: PetscCall(MatGetColumnNorms(R, NORM_2, norms));
1151: PetscCall(MatDestroy(&R));
1152: 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]));
1153: PetscCall(PetscFree(norms));
1154: }
1155: PetscFunctionReturn(PETSC_SUCCESS);
1156: }
1158: static PetscErrorCode KSPMatSolve_Private(KSP ksp, Mat B, Mat X)
1159: {
1160: Mat A, P, vB, vX;
1161: Vec cb, cx;
1162: PetscInt n1, N1, n2, N2, Bbn = PETSC_DECIDE;
1163: PetscBool match;
1165: PetscFunctionBegin;
1169: PetscCheckSameComm(ksp, 1, B, 2);
1170: PetscCheckSameComm(ksp, 1, X, 3);
1171: PetscCheckSameType(B, 2, X, 3);
1172: PetscCheck(B->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
1173: MatCheckPreallocated(X, 3);
1174: if (!X->assembled) {
1175: PetscCall(MatSetOption(X, MAT_NO_OFF_PROC_ENTRIES, PETSC_TRUE));
1176: PetscCall(MatAssemblyBegin(X, MAT_FINAL_ASSEMBLY));
1177: PetscCall(MatAssemblyEnd(X, MAT_FINAL_ASSEMBLY));
1178: }
1179: PetscCheck(B != X, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_IDN, "B and X must be different matrices");
1180: PetscCheck(!ksp->transpose_solve || !ksp->transpose.use_explicittranspose, PetscObjectComm((PetscObject)ksp), PETSC_ERR_SUP, "KSPMatSolveTranspose() does not support -ksp_use_explicittranspose");
1181: PetscCall(KSPGetOperators(ksp, &A, &P));
1182: PetscCall(MatGetLocalSize(B, NULL, &n2));
1183: PetscCall(MatGetLocalSize(X, NULL, &n1));
1184: PetscCall(MatGetSize(B, NULL, &N2));
1185: PetscCall(MatGetSize(X, NULL, &N1));
1186: 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);
1187: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)B, &match, MATSEQDENSE, MATMPIDENSE, ""));
1188: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of right-hand sides not stored in a dense Mat");
1189: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)X, &match, MATSEQDENSE, MATMPIDENSE, ""));
1190: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of solutions not stored in a dense Mat");
1191: PetscCall(KSPSetUp(ksp));
1192: PetscCall(KSPSetUpOnBlocks(ksp));
1193: if (ksp->ops->matsolve) {
1194: level++;
1195: if (ksp->guess_zero) PetscCall(MatZeroEntries(X));
1196: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1197: PetscCall(KSPGetMatSolveBatchSize(ksp, &Bbn));
1198: /* by default, do a single solve with all columns */
1199: if (Bbn == PETSC_DECIDE) Bbn = N2;
1200: else PetscCheck(Bbn >= 1, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "KSPMatSolve() batch size %" PetscInt_FMT " must be positive", Bbn);
1201: PetscCall(PetscInfo(ksp, "KSP type %s solving using batches of width at most %" PetscInt_FMT "\n", ((PetscObject)ksp)->type_name, Bbn));
1202: /* if -ksp_matsolve_batch_size is greater than the actual number of columns, do a single solve with all columns */
1203: if (Bbn >= N2) {
1204: PetscUseTypeMethod(ksp, matsolve, B, X);
1205: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, B, X, ksp->viewerFinalRes, ksp->formatFinalRes, 0));
1207: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1209: if (ksp->viewRate) {
1210: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1211: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1212: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1213: }
1214: } else {
1215: for (n2 = 0; n2 < N2; n2 += Bbn) {
1216: PetscCall(MatDenseGetSubMatrix(B, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vB));
1217: PetscCall(MatDenseGetSubMatrix(X, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vX));
1218: PetscUseTypeMethod(ksp, matsolve, vB, vX);
1219: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, vB, vX, ksp->viewerFinalRes, ksp->formatFinalRes, n2));
1221: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1223: if (ksp->viewRate) {
1224: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1225: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1226: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1227: }
1228: PetscCall(MatDenseRestoreSubMatrix(B, &vB));
1229: PetscCall(MatDenseRestoreSubMatrix(X, &vX));
1230: }
1231: }
1232: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)A, ksp->viewerMat, ksp->formatMat));
1233: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)P, ksp->viewerPMat, ksp->formatPMat));
1234: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)B, ksp->viewerRhs, ksp->formatRhs));
1235: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)X, ksp->viewerSol, ksp->formatSol));
1236: if (ksp->view) PetscCall(KSPView(ksp, ksp->viewer));
1237: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1238: if (ksp->errorifnotconverged && ksp->reason < 0 && (level == 1 || ksp->reason != KSP_DIVERGED_ITS)) {
1239: PCFailedReason reason;
1241: 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]);
1242: PetscCall(PCGetFailedReason(ksp->pc, &reason));
1243: 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]);
1244: }
1245: level--;
1246: } else {
1247: PetscCall(PetscInfo(ksp, "KSP type %s solving column by column\n", ((PetscObject)ksp)->type_name));
1248: for (n2 = 0; n2 < N2; ++n2) {
1249: PetscCall(MatDenseGetColumnVecRead(B, n2, &cb));
1250: PetscCall(MatDenseGetColumnVecWrite(X, n2, &cx));
1251: PetscCall(KSPSolve_Private(ksp, cb, cx));
1252: PetscCall(MatDenseRestoreColumnVecWrite(X, n2, &cx));
1253: PetscCall(MatDenseRestoreColumnVecRead(B, n2, &cb));
1254: }
1255: }
1256: PetscFunctionReturn(PETSC_SUCCESS);
1257: }
1259: /*@
1260: KSPMatSolve - Solves a linear system with multiple right-hand sides stored as a `MATDENSE`. Unlike `KSPSolve()`, `B` and `X` must be different matrices.
1262: Input Parameters:
1263: + ksp - iterative context
1264: - B - block of right-hand sides
1266: Output Parameter:
1267: . X - block of solutions
1269: Level: intermediate
1271: Note:
1272: This is a stripped-down version of `KSPSolve()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1274: .seealso: [](ch_ksp), `KSPSolve()`, `MatMatSolve()`, `KSPMatSolveTranspose()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`
1275: @*/
1276: PetscErrorCode KSPMatSolve(KSP ksp, Mat B, Mat X)
1277: {
1278: PetscFunctionBegin;
1279: ksp->transpose_solve = PETSC_FALSE;
1280: PetscCall(KSPMatSolve_Private(ksp, B, X));
1281: PetscFunctionReturn(PETSC_SUCCESS);
1282: }
1284: /*@
1285: KSPMatSolveTranspose - Solves a linear system with the transposed matrix with multiple right-hand sides stored as a `MATDENSE`. Unlike `KSPSolveTranspose()`,
1286: `B` and `X` must be different matrices and the transposed matrix cannot be assembled explicitly for the user.
1288: Input Parameters:
1289: + ksp - iterative context
1290: - B - block of right-hand sides
1292: Output Parameter:
1293: . X - block of solutions
1295: Level: intermediate
1297: Note:
1298: This is a stripped-down version of `KSPSolveTranspose()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1300: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `MatMatTransposeSolve()`, `KSPMatSolve()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`
1301: @*/
1302: PetscErrorCode KSPMatSolveTranspose(KSP ksp, Mat B, Mat X)
1303: {
1304: PetscFunctionBegin;
1305: ksp->transpose_solve = PETSC_TRUE;
1306: PetscCall(KSPMatSolve_Private(ksp, B, X));
1307: PetscFunctionReturn(PETSC_SUCCESS);
1308: }
1310: /*@
1311: KSPSetMatSolveBatchSize - Sets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1313: Logically Collective
1315: Input Parameters:
1316: + ksp - iterative context
1317: - bs - batch size
1319: Level: advanced
1321: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPGetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matmatmult_Bbn`
1322: @*/
1323: PetscErrorCode KSPSetMatSolveBatchSize(KSP ksp, PetscInt bs)
1324: {
1325: PetscFunctionBegin;
1328: ksp->nmax = bs;
1329: PetscFunctionReturn(PETSC_SUCCESS);
1330: }
1332: /*@
1333: KSPGetMatSolveBatchSize - Gets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1335: Input Parameter:
1336: . ksp - iterative context
1338: Output Parameter:
1339: . bs - batch size
1341: Level: advanced
1343: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPSetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matmatmult_Bbn`
1344: @*/
1345: PetscErrorCode KSPGetMatSolveBatchSize(KSP ksp, PetscInt *bs)
1346: {
1347: PetscFunctionBegin;
1349: PetscAssertPointer(bs, 2);
1350: *bs = ksp->nmax;
1351: PetscFunctionReturn(PETSC_SUCCESS);
1352: }
1354: /*@
1355: KSPResetViewers - Resets all the viewers set from the options database during `KSPSetFromOptions()`
1357: Collective
1359: Input Parameter:
1360: . ksp - iterative context obtained from `KSPCreate()`
1362: Level: beginner
1364: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSPSetFromOptions()`, `KSP`
1365: @*/
1366: PetscErrorCode KSPResetViewers(KSP ksp)
1367: {
1368: PetscFunctionBegin;
1370: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1371: PetscCall(PetscViewerDestroy(&ksp->viewer));
1372: PetscCall(PetscViewerDestroy(&ksp->viewerPre));
1373: PetscCall(PetscViewerDestroy(&ksp->viewerRate));
1374: PetscCall(PetscViewerDestroy(&ksp->viewerMat));
1375: PetscCall(PetscViewerDestroy(&ksp->viewerPMat));
1376: PetscCall(PetscViewerDestroy(&ksp->viewerRhs));
1377: PetscCall(PetscViewerDestroy(&ksp->viewerSol));
1378: PetscCall(PetscViewerDestroy(&ksp->viewerMatExp));
1379: PetscCall(PetscViewerDestroy(&ksp->viewerEV));
1380: PetscCall(PetscViewerDestroy(&ksp->viewerSV));
1381: PetscCall(PetscViewerDestroy(&ksp->viewerEVExp));
1382: PetscCall(PetscViewerDestroy(&ksp->viewerFinalRes));
1383: PetscCall(PetscViewerDestroy(&ksp->viewerPOpExp));
1384: PetscCall(PetscViewerDestroy(&ksp->viewerDScale));
1385: ksp->view = PETSC_FALSE;
1386: ksp->viewPre = PETSC_FALSE;
1387: ksp->viewMat = PETSC_FALSE;
1388: ksp->viewPMat = PETSC_FALSE;
1389: ksp->viewRhs = PETSC_FALSE;
1390: ksp->viewSol = PETSC_FALSE;
1391: ksp->viewMatExp = PETSC_FALSE;
1392: ksp->viewEV = PETSC_FALSE;
1393: ksp->viewSV = PETSC_FALSE;
1394: ksp->viewEVExp = PETSC_FALSE;
1395: ksp->viewFinalRes = PETSC_FALSE;
1396: ksp->viewPOpExp = PETSC_FALSE;
1397: ksp->viewDScale = PETSC_FALSE;
1398: PetscFunctionReturn(PETSC_SUCCESS);
1399: }
1401: /*@
1402: KSPReset - Resets a `KSP` context to the kspsetupcalled = 0 state and removes any allocated Vecs and Mats
1404: Collective
1406: Input Parameter:
1407: . ksp - iterative context obtained from `KSPCreate()`
1409: Level: beginner
1411: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1412: @*/
1413: PetscErrorCode KSPReset(KSP ksp)
1414: {
1415: PetscFunctionBegin;
1417: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1418: PetscTryTypeMethod(ksp, reset);
1419: if (ksp->pc) PetscCall(PCReset(ksp->pc));
1420: if (ksp->guess) {
1421: KSPGuess guess = ksp->guess;
1422: PetscTryTypeMethod(guess, reset);
1423: }
1424: PetscCall(VecDestroyVecs(ksp->nwork, &ksp->work));
1425: PetscCall(VecDestroy(&ksp->vec_rhs));
1426: PetscCall(VecDestroy(&ksp->vec_sol));
1427: PetscCall(VecDestroy(&ksp->diagonal));
1428: PetscCall(VecDestroy(&ksp->truediagonal));
1430: PetscCall(KSPResetViewers(ksp));
1432: ksp->setupstage = KSP_SETUP_NEW;
1433: ksp->nmax = PETSC_DECIDE;
1434: PetscFunctionReturn(PETSC_SUCCESS);
1435: }
1437: /*@C
1438: KSPDestroy - Destroys a `KSP` context.
1440: Collective
1442: Input Parameter:
1443: . ksp - iterative context obtained from `KSPCreate()`
1445: Level: beginner
1447: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1448: @*/
1449: PetscErrorCode KSPDestroy(KSP *ksp)
1450: {
1451: PC pc;
1453: PetscFunctionBegin;
1454: if (!*ksp) PetscFunctionReturn(PETSC_SUCCESS);
1456: if (--((PetscObject)(*ksp))->refct > 0) {
1457: *ksp = NULL;
1458: PetscFunctionReturn(PETSC_SUCCESS);
1459: }
1461: PetscCall(PetscObjectSAWsViewOff((PetscObject)*ksp));
1463: /*
1464: Avoid a cascading call to PCReset(ksp->pc) from the following call:
1465: PCReset() shouldn't be called from KSPDestroy() as it is unprotected by pc's
1466: refcount (and may be shared, e.g., by other ksps).
1467: */
1468: pc = (*ksp)->pc;
1469: (*ksp)->pc = NULL;
1470: PetscCall(KSPReset((*ksp)));
1471: (*ksp)->pc = pc;
1472: PetscTryTypeMethod((*ksp), destroy);
1474: if ((*ksp)->transpose.use_explicittranspose) {
1475: PetscCall(MatDestroy(&(*ksp)->transpose.AT));
1476: PetscCall(MatDestroy(&(*ksp)->transpose.BT));
1477: (*ksp)->transpose.reuse_transpose = PETSC_FALSE;
1478: }
1480: PetscCall(KSPGuessDestroy(&(*ksp)->guess));
1481: PetscCall(DMDestroy(&(*ksp)->dm));
1482: PetscCall(PCDestroy(&(*ksp)->pc));
1483: PetscCall(PetscFree((*ksp)->res_hist_alloc));
1484: PetscCall(PetscFree((*ksp)->err_hist_alloc));
1485: if ((*ksp)->convergeddestroy) PetscCall((*(*ksp)->convergeddestroy)((*ksp)->cnvP));
1486: PetscCall(KSPMonitorCancel((*ksp)));
1487: PetscCall(KSPConvergedReasonViewCancel((*ksp)));
1488: PetscCall(PetscHeaderDestroy(ksp));
1489: PetscFunctionReturn(PETSC_SUCCESS);
1490: }
1492: /*@
1493: KSPSetPCSide - Sets the preconditioning side.
1495: Logically Collective
1497: Input Parameter:
1498: . ksp - iterative context obtained from `KSPCreate()`
1500: Output Parameter:
1501: . side - the preconditioning side, where side is one of
1502: .vb
1503: PC_LEFT - left preconditioning (default)
1504: PC_RIGHT - right preconditioning
1505: PC_SYMMETRIC - symmetric preconditioning
1506: .ve
1508: Options Database Key:
1509: . -ksp_pc_side <right,left,symmetric> - `KSP` preconditioner side
1511: Level: intermediate
1513: Notes:
1514: Left preconditioning is used by default for most Krylov methods except `KSPFGMRES` which only supports right preconditioning.
1516: For methods changing the side of the preconditioner changes the norm type that is used, see `KSPSetNormType()`.
1518: Symmetric preconditioning is currently available only for the `KSPQCG` method. However, note that
1519: symmetric preconditioning can be emulated by using either right or left
1520: preconditioning and a pre or post processing step.
1522: Setting the `PCSide` often affects the default norm type. See `KSPSetNormType()` for details.
1524: .seealso: [](ch_ksp), `KSPGetPCSide()`, `KSPSetNormType()`, `KSPGetNormType()`, `KSP`
1525: @*/
1526: PetscErrorCode KSPSetPCSide(KSP ksp, PCSide side)
1527: {
1528: PetscFunctionBegin;
1531: ksp->pc_side = ksp->pc_side_set = side;
1532: PetscFunctionReturn(PETSC_SUCCESS);
1533: }
1535: /*@
1536: KSPGetPCSide - Gets the preconditioning side.
1538: Not Collective
1540: Input Parameter:
1541: . ksp - iterative context obtained from `KSPCreate()`
1543: Output Parameter:
1544: . side - the preconditioning side, where side is one of
1545: .vb
1546: PC_LEFT - left preconditioning (default)
1547: PC_RIGHT - right preconditioning
1548: PC_SYMMETRIC - symmetric preconditioning
1549: .ve
1551: Level: intermediate
1553: .seealso: [](ch_ksp), `KSPSetPCSide()`, `KSP`
1554: @*/
1555: PetscErrorCode KSPGetPCSide(KSP ksp, PCSide *side)
1556: {
1557: PetscFunctionBegin;
1559: PetscAssertPointer(side, 2);
1560: PetscCall(KSPSetUpNorms_Private(ksp, PETSC_TRUE, &ksp->normtype, &ksp->pc_side));
1561: *side = ksp->pc_side;
1562: PetscFunctionReturn(PETSC_SUCCESS);
1563: }
1565: /*@
1566: KSPGetTolerances - Gets the relative, absolute, divergence, and maximum
1567: iteration tolerances used by the default `KSP` convergence tests.
1569: Not Collective
1571: Input Parameter:
1572: . ksp - the Krylov subspace context
1574: Output Parameters:
1575: + rtol - the relative convergence tolerance
1576: . abstol - the absolute convergence tolerance
1577: . dtol - the divergence tolerance
1578: - maxits - maximum number of iterations
1580: Level: intermediate
1582: Note:
1583: The user can specify `NULL` for any parameter that is not needed.
1585: .seealso: [](ch_ksp), `KSPSetTolerances()`, `KSP`, `KSPSetMinimumIterations()`, `KSPGetMinimumIterations()`
1586: @*/
1587: PetscErrorCode KSPGetTolerances(KSP ksp, PetscReal *rtol, PetscReal *abstol, PetscReal *dtol, PetscInt *maxits)
1588: {
1589: PetscFunctionBegin;
1591: if (abstol) *abstol = ksp->abstol;
1592: if (rtol) *rtol = ksp->rtol;
1593: if (dtol) *dtol = ksp->divtol;
1594: if (maxits) *maxits = ksp->max_it;
1595: PetscFunctionReturn(PETSC_SUCCESS);
1596: }
1598: /*@
1599: KSPSetTolerances - Sets the relative, absolute, divergence, and maximum
1600: iteration tolerances used by the default `KSP` convergence testers.
1602: Logically Collective
1604: Input Parameters:
1605: + ksp - the Krylov subspace context
1606: . rtol - the relative convergence tolerance, relative decrease in the (possibly preconditioned) residual norm
1607: . abstol - the absolute convergence tolerance absolute size of the (possibly preconditioned) residual norm
1608: . dtol - the divergence tolerance, amount (possibly preconditioned) residual norm can increase before `KSPConvergedDefault()` concludes that the method is diverging
1609: - maxits - maximum number of iterations to use
1611: Options Database Keys:
1612: + -ksp_atol <abstol> - Sets `abstol`
1613: . -ksp_rtol <rtol> - Sets `rtol`
1614: . -ksp_divtol <dtol> - Sets `dtol`
1615: - -ksp_max_it <maxits> - Sets `maxits`
1617: Level: intermediate
1619: Notes:
1620: Use `PETSC_DEFAULT` to retain the default value of any of the tolerances.
1622: See `KSPConvergedDefault()` for details how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1623: for setting user-defined stopping criteria.
1625: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetMinimumIterations()`
1626: @*/
1627: PetscErrorCode KSPSetTolerances(KSP ksp, PetscReal rtol, PetscReal abstol, PetscReal dtol, PetscInt maxits)
1628: {
1629: PetscFunctionBegin;
1636: if (rtol != (PetscReal)PETSC_DEFAULT) {
1637: 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);
1638: ksp->rtol = rtol;
1639: }
1640: if (abstol != (PetscReal)PETSC_DEFAULT) {
1641: PetscCheck(abstol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Absolute tolerance %g must be non-negative", (double)abstol);
1642: ksp->abstol = abstol;
1643: }
1644: if (dtol != (PetscReal)PETSC_DEFAULT) {
1645: PetscCheck(dtol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Divergence tolerance %g must be larger than 1.0", (double)dtol);
1646: ksp->divtol = dtol;
1647: }
1648: if (maxits != PETSC_DEFAULT) {
1649: PetscCheck(maxits >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Maximum number of iterations %" PetscInt_FMT " must be non-negative", maxits);
1650: ksp->max_it = maxits;
1651: }
1652: PetscFunctionReturn(PETSC_SUCCESS);
1653: }
1655: /*@
1656: KSPSetMinimumIterations - Sets the minimum number of iterations to use, regardless of the tolerances
1658: Logically Collective
1660: Input Parameters:
1661: + ksp - the Krylov subspace context
1662: - minit - minimum number of iterations to use
1664: Options Database Key:
1665: . -ksp_min_it <minits> - Sets `minit`
1667: Level: intermediate
1669: Notes:
1670: Use `KSPSetTolerances()` to set a variety of other tolerances
1672: See `KSPConvergedDefault()` for details on how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1673: for setting user-defined stopping criteria.
1675: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPGetMinimumIterations()`
1676: @*/
1677: PetscErrorCode KSPSetMinimumIterations(KSP ksp, PetscInt minit)
1678: {
1679: PetscFunctionBegin;
1683: PetscCheck(minit >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Minimum number of iterations %" PetscInt_FMT " must be non-negative", minit);
1684: ksp->min_it = minit;
1685: PetscFunctionReturn(PETSC_SUCCESS);
1686: }
1688: /*@
1689: KSPGetMinimumIterations - Gets the minimum number of iterations to use, regardless of the tolerances, that was set with `KSPSetMinimumIterations()` or `-ksp_min_it`
1691: Not Collective
1693: Input Parameter:
1694: . ksp - the Krylov subspace context
1696: Output Parameter:
1697: . minit - minimum number of iterations to use
1699: Level: intermediate
1701: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPSetMinimumIterations()`
1702: @*/
1703: PetscErrorCode KSPGetMinimumIterations(KSP ksp, PetscInt *minit)
1704: {
1705: PetscFunctionBegin;
1707: PetscAssertPointer(minit, 2);
1709: *minit = ksp->min_it;
1710: PetscFunctionReturn(PETSC_SUCCESS);
1711: }
1713: /*@
1714: KSPSetInitialGuessNonzero - Tells the iterative solver that the
1715: initial guess is nonzero; otherwise `KSP` assumes the initial guess
1716: is to be zero (and thus zeros it out before solving).
1718: Logically Collective
1720: Input Parameters:
1721: + ksp - iterative context obtained from `KSPCreate()`
1722: - flg - ``PETSC_TRUE`` indicates the guess is non-zero, `PETSC_FALSE` indicates the guess is zero
1724: Options Database Key:
1725: . -ksp_initial_guess_nonzero <true,false> - use nonzero initial guess
1727: Level: beginner
1729: Note:
1730: If this is not called the X vector is zeroed in the call to `KSPSolve()`.
1732: .seealso: [](ch_ksp), `KSPGetInitialGuessNonzero()`, `KSPSetGuessType()`, `KSPGuessType`, `KSP`
1733: @*/
1734: PetscErrorCode KSPSetInitialGuessNonzero(KSP ksp, PetscBool flg)
1735: {
1736: PetscFunctionBegin;
1739: ksp->guess_zero = (PetscBool) !(int)flg;
1740: PetscFunctionReturn(PETSC_SUCCESS);
1741: }
1743: /*@
1744: KSPGetInitialGuessNonzero - Determines whether the `KSP` solver is using
1745: a zero initial guess.
1747: Not Collective
1749: Input Parameter:
1750: . ksp - iterative context obtained from `KSPCreate()`
1752: Output Parameter:
1753: . flag - `PETSC_TRUE` if guess is nonzero, else `PETSC_FALSE`
1755: Level: intermediate
1757: .seealso: [](ch_ksp), `KSPSetInitialGuessNonzero()`, `KSP`
1758: @*/
1759: PetscErrorCode KSPGetInitialGuessNonzero(KSP ksp, PetscBool *flag)
1760: {
1761: PetscFunctionBegin;
1763: PetscAssertPointer(flag, 2);
1764: if (ksp->guess_zero) *flag = PETSC_FALSE;
1765: else *flag = PETSC_TRUE;
1766: PetscFunctionReturn(PETSC_SUCCESS);
1767: }
1769: /*@
1770: KSPSetErrorIfNotConverged - Causes `KSPSolve()` to generate an error if the solver has not converged as soon as the error is detected.
1772: Logically Collective
1774: Input Parameters:
1775: + ksp - iterative context obtained from `KSPCreate()`
1776: - flg - `PETSC_TRUE` indicates you want the error generated
1778: Options Database Key:
1779: . -ksp_error_if_not_converged <true,false> - generate an error and stop the program
1781: Level: intermediate
1783: Notes:
1784: Normally PETSc continues if a linear solver fails to converge, you can call `KSPGetConvergedReason()` after a `KSPSolve()`
1785: to determine if it has converged.
1787: A `KSP_DIVERGED_ITS` will not generate an error in a `KSPSolve()` inside a nested linear solver
1789: .seealso: [](ch_ksp), `KSPGetErrorIfNotConverged()`, `KSP`
1790: @*/
1791: PetscErrorCode KSPSetErrorIfNotConverged(KSP ksp, PetscBool flg)
1792: {
1793: PetscFunctionBegin;
1796: ksp->errorifnotconverged = flg;
1797: PetscFunctionReturn(PETSC_SUCCESS);
1798: }
1800: /*@
1801: KSPGetErrorIfNotConverged - Will `KSPSolve()` generate an error if the solver does not converge?
1803: Not Collective
1805: Input Parameter:
1806: . ksp - iterative context obtained from KSPCreate()
1808: Output Parameter:
1809: . flag - `PETSC_TRUE` if it will generate an error, else `PETSC_FALSE`
1811: Level: intermediate
1813: .seealso: [](ch_ksp), `KSPSetErrorIfNotConverged()`, `KSP`
1814: @*/
1815: PetscErrorCode KSPGetErrorIfNotConverged(KSP ksp, PetscBool *flag)
1816: {
1817: PetscFunctionBegin;
1819: PetscAssertPointer(flag, 2);
1820: *flag = ksp->errorifnotconverged;
1821: PetscFunctionReturn(PETSC_SUCCESS);
1822: }
1824: /*@
1825: KSPSetInitialGuessKnoll - Tells the iterative solver to use `PCApply()` to compute the initial guess (The Knoll trick)
1827: Logically Collective
1829: Input Parameters:
1830: + ksp - iterative context obtained from `KSPCreate()`
1831: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1833: Level: advanced
1835: Developer Note:
1836: The Knoll trick is not currently implemented using the `KSPGuess` class
1838: .seealso: [](ch_ksp), `KSPGetInitialGuessKnoll()`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1839: @*/
1840: PetscErrorCode KSPSetInitialGuessKnoll(KSP ksp, PetscBool flg)
1841: {
1842: PetscFunctionBegin;
1845: ksp->guess_knoll = flg;
1846: PetscFunctionReturn(PETSC_SUCCESS);
1847: }
1849: /*@
1850: KSPGetInitialGuessKnoll - Determines whether the `KSP` solver is using the Knoll trick (using PCApply(pc,b,...) to compute
1851: the initial guess
1853: Not Collective
1855: Input Parameter:
1856: . ksp - iterative context obtained from `KSPCreate()`
1858: Output Parameter:
1859: . flag - `PETSC_TRUE` if using Knoll trick, else `PETSC_FALSE`
1861: Level: advanced
1863: .seealso: [](ch_ksp), `KSPSetInitialGuessKnoll()`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1864: @*/
1865: PetscErrorCode KSPGetInitialGuessKnoll(KSP ksp, PetscBool *flag)
1866: {
1867: PetscFunctionBegin;
1869: PetscAssertPointer(flag, 2);
1870: *flag = ksp->guess_knoll;
1871: PetscFunctionReturn(PETSC_SUCCESS);
1872: }
1874: /*@
1875: KSPGetComputeSingularValues - Gets the flag indicating whether the extreme singular
1876: values will be calculated via a Lanczos or Arnoldi process as the linear
1877: system is solved.
1879: Not Collective
1881: Input Parameter:
1882: . ksp - iterative context obtained from `KSPCreate()`
1884: Output Parameter:
1885: . flg - `PETSC_TRUE` or `PETSC_FALSE`
1887: Options Database Key:
1888: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
1890: Level: advanced
1892: Notes:
1893: Currently this option is not valid for all iterative methods.
1895: Many users may just want to use the monitoring routine
1896: `KSPMonitorSingularValue()` (which can be set with option -ksp_monitor_singular_value)
1897: to print the singular values at each iteration of the linear solve.
1899: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`
1900: @*/
1901: PetscErrorCode KSPGetComputeSingularValues(KSP ksp, PetscBool *flg)
1902: {
1903: PetscFunctionBegin;
1905: PetscAssertPointer(flg, 2);
1906: *flg = ksp->calc_sings;
1907: PetscFunctionReturn(PETSC_SUCCESS);
1908: }
1910: /*@
1911: KSPSetComputeSingularValues - Sets a flag so that the extreme singular
1912: values will be calculated via a Lanczos or Arnoldi process as the linear
1913: system is solved.
1915: Logically Collective
1917: Input Parameters:
1918: + ksp - iterative context obtained from `KSPCreate()`
1919: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1921: Options Database Key:
1922: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
1924: Level: advanced
1926: Notes:
1927: Currently this option is not valid for all iterative methods.
1929: Many users may just want to use the monitoring routine
1930: `KSPMonitorSingularValue()` (which can be set with option -ksp_monitor_singular_value)
1931: to print the singular values at each iteration of the linear solve.
1933: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`, `KSPSetComputeRitz()`
1934: @*/
1935: PetscErrorCode KSPSetComputeSingularValues(KSP ksp, PetscBool flg)
1936: {
1937: PetscFunctionBegin;
1940: ksp->calc_sings = flg;
1941: PetscFunctionReturn(PETSC_SUCCESS);
1942: }
1944: /*@
1945: KSPGetComputeEigenvalues - Gets the flag indicating that the extreme eigenvalues
1946: values will be calculated via a Lanczos or Arnoldi process as the linear
1947: system is solved.
1949: Not Collective
1951: Input Parameter:
1952: . ksp - iterative context obtained from `KSPCreate()`
1954: Output Parameter:
1955: . flg - `PETSC_TRUE` or `PETSC_FALSE`
1957: Level: advanced
1959: Note:
1960: Currently this option is not valid for all iterative methods.
1962: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`, `KSPSetComputeRitz()`
1963: @*/
1964: PetscErrorCode KSPGetComputeEigenvalues(KSP ksp, PetscBool *flg)
1965: {
1966: PetscFunctionBegin;
1968: PetscAssertPointer(flg, 2);
1969: *flg = ksp->calc_sings;
1970: PetscFunctionReturn(PETSC_SUCCESS);
1971: }
1973: /*@
1974: KSPSetComputeEigenvalues - Sets a flag so that the extreme eigenvalues
1975: values will be calculated via a Lanczos or Arnoldi process as the linear
1976: system is solved.
1978: Logically Collective
1980: Input Parameters:
1981: + ksp - iterative context obtained from `KSPCreate()`
1982: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1984: Level: advanced
1986: Note:
1987: Currently this option is not valid for all iterative methods.
1989: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`, `KSPSetComputeRitz()`
1990: @*/
1991: PetscErrorCode KSPSetComputeEigenvalues(KSP ksp, PetscBool flg)
1992: {
1993: PetscFunctionBegin;
1996: ksp->calc_sings = flg;
1997: PetscFunctionReturn(PETSC_SUCCESS);
1998: }
2000: /*@
2001: KSPSetComputeRitz - Sets a flag so that the Ritz or harmonic Ritz pairs
2002: will be calculated via a Lanczos or Arnoldi process as the linear
2003: system is solved.
2005: Logically Collective
2007: Input Parameters:
2008: + ksp - iterative context obtained from `KSPCreate()`
2009: - flg - `PETSC_TRUE` or `PETSC_FALSE`
2011: Level: advanced
2013: Note:
2014: Currently this option is only valid for the `KSPGMRES` method.
2016: .seealso: [](ch_ksp), `KSPComputeRitz()`, `KSP`
2017: @*/
2018: PetscErrorCode KSPSetComputeRitz(KSP ksp, PetscBool flg)
2019: {
2020: PetscFunctionBegin;
2023: ksp->calc_ritz = flg;
2024: PetscFunctionReturn(PETSC_SUCCESS);
2025: }
2027: /*@
2028: KSPGetRhs - Gets the right-hand-side vector for the linear system to
2029: be solved.
2031: Not Collective
2033: Input Parameter:
2034: . ksp - iterative context obtained from `KSPCreate()`
2036: Output Parameter:
2037: . r - right-hand-side vector
2039: Level: developer
2041: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPSolve()`, `KSP`
2042: @*/
2043: PetscErrorCode KSPGetRhs(KSP ksp, Vec *r)
2044: {
2045: PetscFunctionBegin;
2047: PetscAssertPointer(r, 2);
2048: *r = ksp->vec_rhs;
2049: PetscFunctionReturn(PETSC_SUCCESS);
2050: }
2052: /*@
2053: KSPGetSolution - Gets the location of the solution for the
2054: linear system to be solved. Note that this may not be where the solution
2055: is stored during the iterative process; see `KSPBuildSolution()`.
2057: Not Collective
2059: Input Parameter:
2060: . ksp - iterative context obtained from `KSPCreate()`
2062: Output Parameter:
2063: . v - solution vector
2065: Level: developer
2067: .seealso: [](ch_ksp), `KSPGetRhs()`, `KSPBuildSolution()`, `KSPSolve()`, `KSP`
2068: @*/
2069: PetscErrorCode KSPGetSolution(KSP ksp, Vec *v)
2070: {
2071: PetscFunctionBegin;
2073: PetscAssertPointer(v, 2);
2074: *v = ksp->vec_sol;
2075: PetscFunctionReturn(PETSC_SUCCESS);
2076: }
2078: /*@
2079: KSPSetPC - Sets the preconditioner to be used to calculate the
2080: application of the preconditioner on a vector.
2082: Collective
2084: Input Parameters:
2085: + ksp - iterative context obtained from `KSPCreate()`
2086: - pc - the preconditioner object (can be `NULL`)
2088: Level: developer
2090: Note:
2091: Use `KSPGetPC()` to retrieve the preconditioner context.
2093: .seealso: [](ch_ksp), `KSPGetPC()`, `KSP`
2094: @*/
2095: PetscErrorCode KSPSetPC(KSP ksp, PC pc)
2096: {
2097: PetscFunctionBegin;
2099: if (pc) {
2101: PetscCheckSameComm(ksp, 1, pc, 2);
2102: }
2103: if (ksp->pc != pc && ksp->setupstage) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2104: PetscCall(PetscObjectReference((PetscObject)pc));
2105: PetscCall(PCDestroy(&ksp->pc));
2106: ksp->pc = pc;
2107: PetscFunctionReturn(PETSC_SUCCESS);
2108: }
2110: PETSC_INTERN PetscErrorCode PCCreate_MPI(PC);
2112: // PetscClangLinter pragma disable: -fdoc-internal-linkage
2113: /*@C
2114: KSPCheckPCMPI - Checks if `-mpi_linear_solver_server` is active and the `PC` should be changed to `PCMPI`
2116: Collective
2118: Input Parameter:
2119: . ksp - iterative context obtained from `KSPCreate()`
2121: Level: developer
2123: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`, `PCMPIServerBegin()`, `PCMPIServerEnd()`
2124: @*/
2125: PETSC_INTERN PetscErrorCode KSPCheckPCMPI(KSP ksp)
2126: {
2127: PetscBool isPCMPI;
2129: PetscFunctionBegin;
2131: PetscCall(PetscObjectTypeCompare((PetscObject)ksp->pc, PCMPI, &isPCMPI));
2132: if (PCMPIServerActive && ksp->nestlevel == 0 && !isPCMPI) {
2133: const char *prefix;
2134: char *found = NULL;
2136: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
2137: if (prefix) PetscCall(PetscStrstr(prefix, "mpi_linear_solver_server_", &found));
2138: if (!found) PetscCall(KSPAppendOptionsPrefix(ksp, "mpi_linear_solver_server_"));
2139: PetscCall(PCSetType(ksp->pc, PCMPI));
2140: }
2141: PetscFunctionReturn(PETSC_SUCCESS);
2142: }
2144: /*@
2145: KSPGetPC - Returns a pointer to the preconditioner context
2146: set with `KSPSetPC()`.
2148: Not Collective
2150: Input Parameter:
2151: . ksp - iterative context obtained from `KSPCreate()`
2153: Output Parameter:
2154: . pc - preconditioner context
2156: Level: developer
2158: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`, `PC`
2159: @*/
2160: PetscErrorCode KSPGetPC(KSP ksp, PC *pc)
2161: {
2162: PetscFunctionBegin;
2164: PetscAssertPointer(pc, 2);
2165: if (!ksp->pc) {
2166: PetscCall(PCCreate(PetscObjectComm((PetscObject)ksp), &ksp->pc));
2167: PetscCall(PetscObjectIncrementTabLevel((PetscObject)ksp->pc, (PetscObject)ksp, 0));
2168: PetscCall(PetscObjectSetOptions((PetscObject)ksp->pc, ((PetscObject)ksp)->options));
2169: PetscCall(PCSetKSPNestLevel(ksp->pc, ksp->nestlevel));
2170: }
2171: PetscCall(KSPCheckPCMPI(ksp));
2172: *pc = ksp->pc;
2173: PetscFunctionReturn(PETSC_SUCCESS);
2174: }
2176: /*@
2177: KSPMonitor - runs the user provided monitor routines, if they exist
2179: Collective
2181: Input Parameters:
2182: + ksp - iterative context obtained from `KSPCreate()`
2183: . it - iteration number
2184: - rnorm - relative norm of the residual
2186: Level: developer
2188: Note:
2189: This routine is called by the `KSP` implementations.
2190: It does not typically need to be called by the user.
2192: .seealso: [](ch_ksp), `KSPMonitorSet()`
2193: @*/
2194: PetscErrorCode KSPMonitor(KSP ksp, PetscInt it, PetscReal rnorm)
2195: {
2196: PetscInt i, n = ksp->numbermonitors;
2198: PetscFunctionBegin;
2199: for (i = 0; i < n; i++) PetscCall((*ksp->monitor[i])(ksp, it, rnorm, ksp->monitorcontext[i]));
2200: PetscFunctionReturn(PETSC_SUCCESS);
2201: }
2203: /*@C
2204: KSPMonitorSet - Sets an ADDITIONAL function to be called at every iteration to monitor
2205: the residual/error etc.
2207: Logically Collective
2209: Input Parameters:
2210: + ksp - iterative context obtained from `KSPCreate()`
2211: . monitor - pointer to function (if this is `NULL`, it turns off monitoring
2212: . ctx - [optional] context for private data for the monitor routine (use `NULL` if no context is needed)
2213: - monitordestroy - [optional] routine that frees monitor context (may be `NULL`)
2215: Calling sequence of `monitor`:
2216: + ksp - iterative context obtained from `KSPCreate()`
2217: . it - iteration number
2218: . rnorm - (estimated) 2-norm of (preconditioned) residual
2219: - ctx - optional monitoring context, as set by `KSPMonitorSet()`
2221: Calling sequence of `monitordestroy`:
2222: . ctx - optional monitoring context, as set by `KSPMonitorSet()`
2224: Options Database Keys:
2225: + -ksp_monitor - sets `KSPMonitorResidual()`
2226: . -ksp_monitor draw - sets `KSPMonitorResidualDraw()` and plots residual
2227: . -ksp_monitor draw::draw_lg - sets `KSPMonitorResidualDrawLG()` and plots residual
2228: . -ksp_monitor_pause_final - Pauses any graphics when the solve finishes (only works for internal monitors)
2229: . -ksp_monitor_true_residual - sets `KSPMonitorTrueResidual()`
2230: . -ksp_monitor_true_residual draw::draw_lg - sets `KSPMonitorTrueResidualDrawLG()` and plots residual
2231: . -ksp_monitor_max - sets `KSPMonitorTrueResidualMax()`
2232: . -ksp_monitor_singular_value - sets `KSPMonitorSingularValue()`
2233: - -ksp_monitor_cancel - cancels all monitors that have
2234: been hardwired into a code by
2235: calls to `KSPMonitorSet()`, but
2236: does not cancel those set via
2237: the options database.
2239: Level: beginner
2241: Notes:
2242: The default is to do nothing. To print the residual, or preconditioned
2243: residual if `KSPSetNormType`(ksp,`KSP_NORM_PRECONDITIONED`) was called, use
2244: `KSPMonitorResidual()` as the monitoring routine, with a `PETSCVIEWERASCII` as the
2245: context.
2247: Several different monitoring routines may be set by calling
2248: `KSPMonitorSet()` multiple times; all will be called in the
2249: order in which they were set.
2251: Fortran Note:
2252: Only a single monitor function can be set for each `KSP` object
2254: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorCancel()`, `KSP`
2255: @*/
2256: PetscErrorCode KSPMonitorSet(KSP ksp, PetscErrorCode (*monitor)(KSP ksp, PetscInt it, PetscReal rnorm, void *ctx), void *ctx, PetscErrorCode (*monitordestroy)(void **ctx))
2257: {
2258: PetscInt i;
2259: PetscBool identical;
2261: PetscFunctionBegin;
2263: for (i = 0; i < ksp->numbermonitors; i++) {
2264: PetscCall(PetscMonitorCompare((PetscErrorCode(*)(void))monitor, ctx, monitordestroy, (PetscErrorCode(*)(void))ksp->monitor[i], ksp->monitorcontext[i], ksp->monitordestroy[i], &identical));
2265: if (identical) PetscFunctionReturn(PETSC_SUCCESS);
2266: }
2267: PetscCheck(ksp->numbermonitors < MAXKSPMONITORS, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Too many KSP monitors set");
2268: ksp->monitor[ksp->numbermonitors] = monitor;
2269: ksp->monitordestroy[ksp->numbermonitors] = monitordestroy;
2270: ksp->monitorcontext[ksp->numbermonitors++] = (void *)ctx;
2271: PetscFunctionReturn(PETSC_SUCCESS);
2272: }
2274: /*@
2275: KSPMonitorCancel - Clears all monitors for a `KSP` object.
2277: Logically Collective
2279: Input Parameter:
2280: . ksp - iterative context obtained from `KSPCreate()`
2282: Options Database Key:
2283: . -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.
2285: Level: intermediate
2287: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorSet()`, `KSP`
2288: @*/
2289: PetscErrorCode KSPMonitorCancel(KSP ksp)
2290: {
2291: PetscInt i;
2293: PetscFunctionBegin;
2295: for (i = 0; i < ksp->numbermonitors; i++) {
2296: if (ksp->monitordestroy[i]) PetscCall((*ksp->monitordestroy[i])(&ksp->monitorcontext[i]));
2297: }
2298: ksp->numbermonitors = 0;
2299: PetscFunctionReturn(PETSC_SUCCESS);
2300: }
2302: /*@C
2303: KSPGetMonitorContext - Gets the monitoring context, as set by `KSPMonitorSet()` for the FIRST monitor only.
2305: Not Collective
2307: Input Parameter:
2308: . ksp - iterative context obtained from `KSPCreate()`
2310: Output Parameter:
2311: . ctx - monitoring context
2313: Level: intermediate
2315: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSP`
2316: @*/
2317: PetscErrorCode KSPGetMonitorContext(KSP ksp, void *ctx)
2318: {
2319: PetscFunctionBegin;
2321: *(void **)ctx = ksp->monitorcontext[0];
2322: PetscFunctionReturn(PETSC_SUCCESS);
2323: }
2325: /*@
2326: KSPSetResidualHistory - Sets the array used to hold the residual history.
2327: If set, this array will contain the residual norms computed at each
2328: iteration of the solver.
2330: Not Collective
2332: Input Parameters:
2333: + ksp - iterative context obtained from `KSPCreate()`
2334: . a - array to hold history
2335: . na - size of `a`
2336: - reset - `PETSC_TRUE` indicates the history counter is reset to zero
2337: for each new linear solve
2339: Level: advanced
2341: Notes:
2342: 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.
2343: If 'a' is `NULL` then space is allocated for the history. If 'na' `PETSC_DECIDE` or `PETSC_DEFAULT` then a
2344: default array of length 10000 is allocated.
2346: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2348: .seealso: [](ch_ksp), `KSPGetResidualHistory()`, `KSP`
2349: @*/
2350: PetscErrorCode KSPSetResidualHistory(KSP ksp, PetscReal a[], PetscInt na, PetscBool reset)
2351: {
2352: PetscFunctionBegin;
2355: PetscCall(PetscFree(ksp->res_hist_alloc));
2356: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2357: ksp->res_hist = a;
2358: ksp->res_hist_max = (size_t)na;
2359: } else {
2360: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->res_hist_max = (size_t)na;
2361: else ksp->res_hist_max = 10000; /* like default ksp->max_it */
2362: PetscCall(PetscCalloc1(ksp->res_hist_max, &ksp->res_hist_alloc));
2364: ksp->res_hist = ksp->res_hist_alloc;
2365: }
2366: ksp->res_hist_len = 0;
2367: ksp->res_hist_reset = reset;
2368: PetscFunctionReturn(PETSC_SUCCESS);
2369: }
2371: /*@C
2372: KSPGetResidualHistory - Gets the array used to hold the residual history and the number of residuals it contains.
2374: Not Collective
2376: Input Parameter:
2377: . ksp - iterative context obtained from `KSPCreate()`
2379: Output Parameters:
2380: + a - pointer to array to hold history (or `NULL`)
2381: - na - number of used entries in a (or `NULL`)
2383: Level: advanced
2385: Note:
2386: This array is borrowed and should not be freed by the caller.
2388: Can only be called after a `KSPSetResidualHistory()` otherwise `a` and `na` are set to `NULL` and zero
2390: Fortran Note:
2391: The Fortran version of this routine has a calling sequence
2392: $ call KSPGetResidualHistory(KSP ksp, integer na, integer ierr)
2393: note that you have passed a Fortran array into `KSPSetResidualHistory()` and you need
2394: to access the residual values from this Fortran array you provided. Only the `na` (number of
2395: residual norms currently held) is set.
2397: .seealso: [](ch_ksp), `KSPSetResidualHistory()`, `KSP`
2398: @*/
2399: PetscErrorCode KSPGetResidualHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2400: {
2401: PetscFunctionBegin;
2403: if (a) *a = ksp->res_hist;
2404: if (na) *na = (PetscInt)ksp->res_hist_len;
2405: PetscFunctionReturn(PETSC_SUCCESS);
2406: }
2408: /*@
2409: 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.
2411: Not Collective
2413: Input Parameters:
2414: + ksp - iterative context obtained from `KSPCreate()`
2415: . a - array to hold history
2416: . na - size of `a`
2417: - reset - `PETSC_TRUE` indicates the history counter is reset to zero for each new linear solve
2419: Level: advanced
2421: Notes:
2422: 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.
2423: If 'a' is `NULL` then space is allocated for the history. If 'na' is `PETSC_DECIDE` or `PETSC_DEFAULT` then a default array of length 10000 is allocated.
2425: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2427: .seealso: [](ch_ksp), `KSPGetErrorHistory()`, `KSPSetResidualHistory()`, `KSP`
2428: @*/
2429: PetscErrorCode KSPSetErrorHistory(KSP ksp, PetscReal a[], PetscInt na, PetscBool reset)
2430: {
2431: PetscFunctionBegin;
2434: PetscCall(PetscFree(ksp->err_hist_alloc));
2435: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2436: ksp->err_hist = a;
2437: ksp->err_hist_max = (size_t)na;
2438: } else {
2439: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->err_hist_max = (size_t)na;
2440: else ksp->err_hist_max = 10000; /* like default ksp->max_it */
2441: PetscCall(PetscCalloc1(ksp->err_hist_max, &ksp->err_hist_alloc));
2443: ksp->err_hist = ksp->err_hist_alloc;
2444: }
2445: ksp->err_hist_len = 0;
2446: ksp->err_hist_reset = reset;
2447: PetscFunctionReturn(PETSC_SUCCESS);
2448: }
2450: /*@C
2451: KSPGetErrorHistory - Gets the array used to hold the error history and the number of residuals it contains.
2453: Not Collective
2455: Input Parameter:
2456: . ksp - iterative context obtained from `KSPCreate()`
2458: Output Parameters:
2459: + a - pointer to array to hold history (or `NULL`)
2460: - na - number of used entries in a (or `NULL`)
2462: Level: advanced
2464: Note:
2465: This array is borrowed and should not be freed by the caller.
2466: Can only be called after a `KSPSetErrorHistory()` otherwise `a` and `na` are set to `NULL` and zero
2468: Fortran Note:
2469: The Fortran version of this routine has a calling sequence
2470: $ call KSPGetErrorHistory(KSP ksp, integer na, integer ierr)
2471: note that you have passed a Fortran array into `KSPSetErrorHistory()` and you need
2472: to access the residual values from this Fortran array you provided. Only the `na` (number of
2473: residual norms currently held) is set.
2475: .seealso: [](ch_ksp), `KSPSetErrorHistory()`, `KSPGetResidualHistory()`, `KSP`
2476: @*/
2477: PetscErrorCode KSPGetErrorHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2478: {
2479: PetscFunctionBegin;
2481: if (a) *a = ksp->err_hist;
2482: if (na) *na = (PetscInt)ksp->err_hist_len;
2483: PetscFunctionReturn(PETSC_SUCCESS);
2484: }
2486: /*@
2487: KSPComputeConvergenceRate - Compute the convergence rate for the iteration <https:/en.wikipedia.org/wiki/Coefficient_of_determination>
2489: Not Collective
2491: Input Parameter:
2492: . ksp - The `KSP`
2494: Output Parameters:
2495: + cr - The residual contraction rate
2496: . rRsq - The coefficient of determination, $R^2$, indicating the linearity of the data
2497: . ce - The error contraction rate
2498: - eRsq - The coefficient of determination, $R^2$, indicating the linearity of the data
2500: Level: advanced
2502: Note:
2503: 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,
2504: the slope is $\log c$. The coefficient of determination is given by $1 - \frac{\sum_i (y_i - f(x_i))^2}{\sum_i (y_i - \bar y)}$,
2506: .seealso: [](ch_ksp), `KSP`, `KSPConvergedRateView()`
2507: @*/
2508: PetscErrorCode KSPComputeConvergenceRate(KSP ksp, PetscReal *cr, PetscReal *rRsq, PetscReal *ce, PetscReal *eRsq)
2509: {
2510: PetscReal const *hist;
2511: PetscReal *x, *y, slope, intercept, mean = 0.0, var = 0.0, res = 0.0;
2512: PetscInt n, k;
2514: PetscFunctionBegin;
2515: if (cr || rRsq) {
2516: PetscCall(KSPGetResidualHistory(ksp, &hist, &n));
2517: if (!n) {
2518: if (cr) *cr = 0.0;
2519: if (rRsq) *rRsq = -1.0;
2520: } else {
2521: PetscCall(PetscMalloc2(n, &x, n, &y));
2522: for (k = 0; k < n; ++k) {
2523: x[k] = k;
2524: y[k] = PetscLogReal(hist[k]);
2525: mean += y[k];
2526: }
2527: mean /= n;
2528: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2529: for (k = 0; k < n; ++k) {
2530: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2531: var += PetscSqr(y[k] - mean);
2532: }
2533: PetscCall(PetscFree2(x, y));
2534: if (cr) *cr = PetscExpReal(slope);
2535: if (rRsq) *rRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2536: }
2537: }
2538: if (ce || eRsq) {
2539: PetscCall(KSPGetErrorHistory(ksp, &hist, &n));
2540: if (!n) {
2541: if (ce) *ce = 0.0;
2542: if (eRsq) *eRsq = -1.0;
2543: } else {
2544: PetscCall(PetscMalloc2(n, &x, n, &y));
2545: for (k = 0; k < n; ++k) {
2546: x[k] = k;
2547: y[k] = PetscLogReal(hist[k]);
2548: mean += y[k];
2549: }
2550: mean /= n;
2551: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2552: for (k = 0; k < n; ++k) {
2553: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2554: var += PetscSqr(y[k] - mean);
2555: }
2556: PetscCall(PetscFree2(x, y));
2557: if (ce) *ce = PetscExpReal(slope);
2558: if (eRsq) *eRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2559: }
2560: }
2561: PetscFunctionReturn(PETSC_SUCCESS);
2562: }
2564: /*@C
2565: KSPSetConvergenceTest - Sets the function to be used to determine convergence.
2567: Logically Collective
2569: Input Parameters:
2570: + ksp - iterative context obtained from `KSPCreate()`
2571: . converge - pointer to the function
2572: . ctx - context for private data for the convergence routine (may be `NULL`)
2573: - destroy - a routine for destroying the context (may be `NULL`)
2575: Calling sequence of `converge`:
2576: + ksp - iterative context obtained from `KSPCreate()`
2577: . it - iteration number
2578: . rnorm - (estimated) 2-norm of (preconditioned) residual
2579: . reason - the reason why it has converged or diverged
2580: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2582: Calling sequence of `destroy`:
2583: . ctx - the context
2585: Level: advanced
2587: Notes:
2588: Must be called after the `KSP` type has been set so put this after
2589: a call to `KSPSetType()`, or `KSPSetFromOptions()`.
2591: The default convergence test, `KSPConvergedDefault()`, aborts if the
2592: residual grows to more than 10000 times the initial residual.
2594: The default is a combination of relative and absolute tolerances.
2595: The residual value that is tested may be an approximation; routines
2596: that need exact values should compute them.
2598: In the default PETSc convergence test, the precise values of reason
2599: are macros such as `KSP_CONVERGED_RTOL`, which are defined in petscksp.h.
2601: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPGetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2602: @*/
2603: PetscErrorCode KSPSetConvergenceTest(KSP ksp, PetscErrorCode (*converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void *ctx, PetscErrorCode (*destroy)(void *ctx))
2604: {
2605: PetscFunctionBegin;
2607: if (ksp->convergeddestroy) PetscCall((*ksp->convergeddestroy)(ksp->cnvP));
2608: ksp->converged = converge;
2609: ksp->convergeddestroy = destroy;
2610: ksp->cnvP = (void *)ctx;
2611: PetscFunctionReturn(PETSC_SUCCESS);
2612: }
2614: /*@C
2615: KSPGetConvergenceTest - Gets the function to be used to determine convergence.
2617: Logically Collective
2619: Input Parameter:
2620: . ksp - iterative context obtained from `KSPCreate()`
2622: Output Parameters:
2623: + converge - pointer to convergence test function
2624: . ctx - context for private data for the convergence routine (may be `NULL`)
2625: - destroy - a routine for destroying the context (may be `NULL`)
2627: Calling sequence of `converge`:
2628: + ksp - iterative context obtained from `KSPCreate()`
2629: . it - iteration number
2630: . rnorm - (estimated) 2-norm of (preconditioned) residual
2631: . reason - the reason why it has converged or diverged
2632: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2634: Calling sequence of `destroy`:
2635: . ctx - the convergence test context
2637: Level: advanced
2639: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2640: @*/
2641: PetscErrorCode KSPGetConvergenceTest(KSP ksp, PetscErrorCode (**converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void **ctx, PetscErrorCode (**destroy)(void *ctx))
2642: {
2643: PetscFunctionBegin;
2645: if (converge) *converge = ksp->converged;
2646: if (destroy) *destroy = ksp->convergeddestroy;
2647: if (ctx) *ctx = ksp->cnvP;
2648: PetscFunctionReturn(PETSC_SUCCESS);
2649: }
2651: /*@C
2652: KSPGetAndClearConvergenceTest - Gets the function to be used to determine convergence. Removes the current test without calling destroy on the test context
2654: Logically Collective
2656: Input Parameter:
2657: . ksp - iterative context obtained from `KSPCreate()`
2659: Output Parameters:
2660: + converge - pointer to convergence test function
2661: . ctx - context for private data for the convergence routine
2662: - destroy - a routine for destroying the context
2664: Calling sequence of `converge`:
2665: + ksp - iterative context obtained from `KSPCreate()`
2666: . it - iteration number
2667: . rnorm - (estimated) 2-norm of (preconditioned) residual
2668: . reason - the reason why it has converged or diverged
2669: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2671: Calling sequence of `destroy`:
2672: . ctx - the convergence test context
2674: Level: advanced
2676: Note:
2677: This is intended to be used to allow transferring the convergence test (and its context) to another testing object (for example another `KSP`)
2678: and then calling `KSPSetConvergenceTest()` on this original `KSP`. If you just called `KSPGetConvergenceTest()` followed
2679: by `KSPSetConvergenceTest()` the original context information
2680: would be destroyed and hence the transferred context would be invalid and trigger a crash on use
2682: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2683: @*/
2684: PetscErrorCode KSPGetAndClearConvergenceTest(KSP ksp, PetscErrorCode (**converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void **ctx, PetscErrorCode (**destroy)(void *ctx))
2685: {
2686: PetscFunctionBegin;
2688: *converge = ksp->converged;
2689: *destroy = ksp->convergeddestroy;
2690: *ctx = ksp->cnvP;
2691: ksp->converged = NULL;
2692: ksp->cnvP = NULL;
2693: ksp->convergeddestroy = NULL;
2694: PetscFunctionReturn(PETSC_SUCCESS);
2695: }
2697: /*@C
2698: KSPGetConvergenceContext - Gets the convergence context set with `KSPSetConvergenceTest()`.
2700: Not Collective
2702: Input Parameter:
2703: . ksp - iterative context obtained from `KSPCreate()`
2705: Output Parameter:
2706: . ctx - monitoring context
2708: Level: advanced
2710: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2711: @*/
2712: PetscErrorCode KSPGetConvergenceContext(KSP ksp, void *ctx)
2713: {
2714: PetscFunctionBegin;
2716: *(void **)ctx = ksp->cnvP;
2717: PetscFunctionReturn(PETSC_SUCCESS);
2718: }
2720: /*@C
2721: KSPBuildSolution - Builds the approximate solution in a vector provided.
2723: Collective
2725: Input Parameter:
2726: . ksp - iterative context obtained from `KSPCreate()`
2728: Output Parameter:
2729: Provide exactly one of
2730: + v - location to stash solution.
2731: - V - the solution is returned in this location. This vector is created
2732: internally. This vector should NOT be destroyed by the user with
2733: `VecDestroy()`.
2735: Level: developer
2737: Notes:
2738: This routine can be used in one of two ways
2739: .vb
2740: KSPBuildSolution(ksp,NULL,&V);
2741: or
2742: KSPBuildSolution(ksp,v,NULL); or KSPBuildSolution(ksp,v,&v);
2743: .ve
2744: In the first case an internal vector is allocated to store the solution
2745: (the user cannot destroy this vector). In the second case the solution
2746: is generated in the vector that the user provides. Note that for certain
2747: methods, such as `KSPCG`, the second case requires a copy of the solution,
2748: while in the first case the call is essentially free since it simply
2749: returns the vector where the solution already is stored. For some methods
2750: like `KSPGMRES` this is a reasonably expensive operation and should only be
2751: used in truly needed.
2753: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPBuildResidual()`, `KSP`
2754: @*/
2755: PetscErrorCode KSPBuildSolution(KSP ksp, Vec v, Vec *V)
2756: {
2757: PetscFunctionBegin;
2759: PetscCheck(V || v, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONG, "Must provide either v or V");
2760: if (!V) V = &v;
2761: PetscUseTypeMethod(ksp, buildsolution, v, V);
2762: PetscFunctionReturn(PETSC_SUCCESS);
2763: }
2765: /*@C
2766: KSPBuildResidual - Builds the residual in a vector provided.
2768: Collective
2770: Input Parameter:
2771: . ksp - iterative context obtained from `KSPCreate()`
2773: Output Parameters:
2774: + v - optional location to stash residual. If `v` is not provided,
2775: then a location is generated.
2776: . t - work vector. If not provided then one is generated.
2777: - V - the residual
2779: Level: advanced
2781: Note:
2782: Regardless of whether or not `v` is provided, the residual is
2783: returned in `V`.
2785: .seealso: [](ch_ksp), `KSP`, `KSPBuildSolution()`
2786: @*/
2787: PetscErrorCode KSPBuildResidual(KSP ksp, Vec t, Vec v, Vec *V)
2788: {
2789: PetscBool flag = PETSC_FALSE;
2790: Vec w = v, tt = t;
2792: PetscFunctionBegin;
2794: if (!w) PetscCall(VecDuplicate(ksp->vec_rhs, &w));
2795: if (!tt) {
2796: PetscCall(VecDuplicate(ksp->vec_sol, &tt));
2797: flag = PETSC_TRUE;
2798: }
2799: PetscUseTypeMethod(ksp, buildresidual, tt, w, V);
2800: if (flag) PetscCall(VecDestroy(&tt));
2801: PetscFunctionReturn(PETSC_SUCCESS);
2802: }
2804: /*@
2805: KSPSetDiagonalScale - Tells `KSP` to symmetrically diagonally scale the system
2806: before solving. This actually CHANGES the matrix (and right hand side).
2808: Logically Collective
2810: Input Parameters:
2811: + ksp - the `KSP` context
2812: - scale - `PETSC_TRUE` or `PETSC_FALSE`
2814: Options Database Keys:
2815: + -ksp_diagonal_scale - perform a diagonal scaling before the solve
2816: - -ksp_diagonal_scale_fix - scale the matrix back AFTER the solve
2818: Level: advanced
2820: Notes:
2821: Scales the matrix by D^(-1/2) A D^(-1/2) [D^(1/2) x ] = D^(-1/2) b
2822: where D_{ii} is 1/abs(A_{ii}) unless A_{ii} is zero and then it is 1.
2824: BE CAREFUL with this routine: it actually scales the matrix and right
2825: hand side that define the system. After the system is solved the matrix
2826: and right hand side remain scaled unless you use `KSPSetDiagonalScaleFix()`
2828: This should NOT be used within the `SNES` solves if you are using a line
2829: search.
2831: If you use this with the `PCType` `PCEISENSTAT` preconditioner than you can
2832: use the `PCEisenstatSetNoDiagonalScaling()` option, or -pc_eisenstat_no_diagonal_scaling
2833: to save some unneeded, redundant flops.
2835: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2836: @*/
2837: PetscErrorCode KSPSetDiagonalScale(KSP ksp, PetscBool scale)
2838: {
2839: PetscFunctionBegin;
2842: ksp->dscale = scale;
2843: PetscFunctionReturn(PETSC_SUCCESS);
2844: }
2846: /*@
2847: KSPGetDiagonalScale - Checks if `KSP` solver scales the matrix and right hand side, that is if `KSPSetDiagonalScale()` has been called
2849: Not Collective
2851: Input Parameter:
2852: . ksp - the `KSP` context
2854: Output Parameter:
2855: . scale - `PETSC_TRUE` or `PETSC_FALSE`
2857: Level: intermediate
2859: .seealso: [](ch_ksp), `KSP`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`
2860: @*/
2861: PetscErrorCode KSPGetDiagonalScale(KSP ksp, PetscBool *scale)
2862: {
2863: PetscFunctionBegin;
2865: PetscAssertPointer(scale, 2);
2866: *scale = ksp->dscale;
2867: PetscFunctionReturn(PETSC_SUCCESS);
2868: }
2870: /*@
2871: KSPSetDiagonalScaleFix - Tells `KSP` to diagonally scale the system back after solving.
2873: Logically Collective
2875: Input Parameters:
2876: + ksp - the `KSP` context
2877: - fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2878: rescale (default)
2880: Level: intermediate
2882: Notes:
2883: Must be called after `KSPSetDiagonalScale()`
2885: Using this will slow things down, because it rescales the matrix before and
2886: after each linear solve. This is intended mainly for testing to allow one
2887: to easily get back the original system to make sure the solution computed is
2888: accurate enough.
2890: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPGetDiagonalScaleFix()`, `KSP`
2891: @*/
2892: PetscErrorCode KSPSetDiagonalScaleFix(KSP ksp, PetscBool fix)
2893: {
2894: PetscFunctionBegin;
2897: ksp->dscalefix = fix;
2898: PetscFunctionReturn(PETSC_SUCCESS);
2899: }
2901: /*@
2902: KSPGetDiagonalScaleFix - Determines if `KSP` diagonally scales the system back after solving. That is `KSPSetDiagonalScaleFix()` has been called
2904: Not Collective
2906: Input Parameter:
2907: . ksp - the `KSP` context
2909: Output Parameter:
2910: . fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2911: rescale (default)
2913: Level: intermediate
2915: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2916: @*/
2917: PetscErrorCode KSPGetDiagonalScaleFix(KSP ksp, PetscBool *fix)
2918: {
2919: PetscFunctionBegin;
2921: PetscAssertPointer(fix, 2);
2922: *fix = ksp->dscalefix;
2923: PetscFunctionReturn(PETSC_SUCCESS);
2924: }
2926: /*@C
2927: KSPSetComputeOperators - set routine to compute the linear operators
2929: Logically Collective
2931: Input Parameters:
2932: + ksp - the `KSP` context
2933: . func - function to compute the operators
2934: - ctx - optional context
2936: Calling sequence of `func`:
2937: + ksp - the `KSP` context
2938: . A - the linear operator
2939: . B - the matrix from which the preconditioner is built, often `A`
2940: - ctx - optional user-provided context
2942: Level: beginner
2944: Notes:
2945: `func()` will be called automatically at the very next call to `KSPSolve()`. It will NOT be called at future `KSPSolve()` calls
2946: unless either `KSPSetComputeOperators()` or `KSPSetOperators()` is called before that `KSPSolve()` is called. This allows the same system to be solved several times
2947: with different right hand side functions but is a confusing API since one might expect it to be called for each `KSPSolve()`
2949: To reuse the same preconditioner for the next `KSPSolve()` and not compute a new one based on the most recently computed matrix call `KSPSetReusePreconditioner()`
2951: Developer Note:
2952: Perhaps this routine and `KSPSetComputeRHS()` could be combined into a new API that makes clear when new matrices are computing without requiring call this
2953: routine to indicate when the new matrix should be computed.
2955: .seealso: [](ch_ksp), `KSP`, `KSPSetOperators()`, `KSPSetComputeRHS()`, `DMKSPSetComputeOperators()`, `KSPSetComputeInitialGuess()`
2956: @*/
2957: PetscErrorCode KSPSetComputeOperators(KSP ksp, PetscErrorCode (*func)(KSP ksp, Mat A, Mat B, void *ctx), void *ctx)
2958: {
2959: DM dm;
2961: PetscFunctionBegin;
2963: PetscCall(KSPGetDM(ksp, &dm));
2964: PetscCall(DMKSPSetComputeOperators(dm, func, ctx));
2965: if (ksp->setupstage == KSP_SETUP_NEWRHS) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2966: PetscFunctionReturn(PETSC_SUCCESS);
2967: }
2969: /*@C
2970: KSPSetComputeRHS - set routine to compute the right hand side of the linear system
2972: Logically Collective
2974: Input Parameters:
2975: + ksp - the `KSP` context
2976: . func - function to compute the right hand side
2977: - ctx - optional context
2979: Calling sequence of `func`:
2980: + ksp - the `KSP` context
2981: . b - right hand side of linear system
2982: - ctx - optional user-provided context
2984: Level: beginner
2986: Note:
2987: The routine you provide will be called EACH you call `KSPSolve()` to prepare the new right hand side for that solve
2989: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `DMKSPSetComputeRHS()`, `KSPSetComputeOperators()`, `KSPSetOperators()`
2990: @*/
2991: PetscErrorCode KSPSetComputeRHS(KSP ksp, PetscErrorCode (*func)(KSP ksp, Vec b, void *ctx), void *ctx)
2992: {
2993: DM dm;
2995: PetscFunctionBegin;
2997: PetscCall(KSPGetDM(ksp, &dm));
2998: PetscCall(DMKSPSetComputeRHS(dm, func, ctx));
2999: PetscFunctionReturn(PETSC_SUCCESS);
3000: }
3002: /*@C
3003: KSPSetComputeInitialGuess - set routine to compute the initial guess of the linear system
3005: Logically Collective
3007: Input Parameters:
3008: + ksp - the `KSP` context
3009: . func - function to compute the initial guess
3010: - ctx - optional context
3012: Calling sequence of `func`:
3013: + ksp - the `KSP` context
3014: . x - solution vector
3015: - ctx - optional user-provided context
3017: Level: beginner
3019: Note:
3020: This should only be used in conjunction with `KSPSetComputeRHS()` and `KSPSetComputeOperators()`, otherwise
3021: call `KSPSetInitialGuessNonzero()` and set the initial guess values in the solution vector passed to `KSPSolve()` before calling the solver
3023: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `KSPSetComputeRHS()`, `KSPSetComputeOperators()`, `DMKSPSetComputeInitialGuess()`, `KSPSetInitialGuessNonzero()`
3024: @*/
3025: PetscErrorCode KSPSetComputeInitialGuess(KSP ksp, PetscErrorCode (*func)(KSP ksp, Vec x, void *ctx), void *ctx)
3026: {
3027: DM dm;
3029: PetscFunctionBegin;
3031: PetscCall(KSPGetDM(ksp, &dm));
3032: PetscCall(DMKSPSetComputeInitialGuess(dm, func, ctx));
3033: PetscFunctionReturn(PETSC_SUCCESS);
3034: }
3036: /*@
3037: KSPSetUseExplicitTranspose - Determines the explicit transpose of the operator is formed in `KSPSolveTranspose()`. In some configurations (like GPUs) it may
3038: be explicitly formed since the solve is much more efficient.
3040: Logically Collective
3042: Input Parameter:
3043: . ksp - the `KSP` context
3045: Output Parameter:
3046: . flg - `PETSC_TRUE` to transpose the system in `KSPSolveTranspose()`, `PETSC_FALSE` to not transpose (default)
3048: Level: advanced
3050: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `KSP`
3051: @*/
3052: PetscErrorCode KSPSetUseExplicitTranspose(KSP ksp, PetscBool flg)
3053: {
3054: PetscFunctionBegin;
3057: ksp->transpose.use_explicittranspose = flg;
3058: PetscFunctionReturn(PETSC_SUCCESS);
3059: }