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`
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 Keys:
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: CG it corresponds to the number of CG iterations, for GMRES 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 GMRES, 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: Notes:
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: Notes:
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.
539: Collective
541: Input Parameter:
542: . ksp - iterative context obtained from `KSPCreate()`
544: Level: intermediate
546: .seealso: [](ch_ksp), `KSPCreate()`, `KSPDestroy()`, `KSPReset()`
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()`
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 reason a `KSP` solve converged or diverged 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: References:
617: . * - `//en.wikipedia.org/wiki/Coefficient_of_determination`
619: .seealso: [](ch_ksp), `KSPConvergedReasonView()`, `KSPGetConvergedRate()`, `KSPSetTolerances()`, `KSPConvergedDefault()`
620: @*/
621: PetscErrorCode KSPConvergedRateView(KSP ksp, PetscViewer viewer)
622: {
623: PetscViewerFormat format;
624: PetscBool isAscii;
625: PetscReal rrate, rRsq, erate = 0.0, eRsq = 0.0;
626: PetscInt its;
627: const char *prefix, *reason = KSPConvergedReasons[ksp->reason];
629: PetscFunctionBegin;
630: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
631: PetscCall(KSPGetIterationNumber(ksp, &its));
632: PetscCall(KSPComputeConvergenceRate(ksp, &rrate, &rRsq, &erate, &eRsq));
633: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
634: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isAscii));
635: if (isAscii) {
636: PetscCall(PetscViewerGetFormat(viewer, &format));
637: PetscCall(PetscViewerASCIIAddTab(viewer, ((PetscObject)ksp)->tablevel));
638: if (ksp->reason > 0) {
639: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve converged due to %s iterations %" PetscInt_FMT, prefix, reason, its));
640: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve converged due to %s iterations %" PetscInt_FMT, reason, its));
641: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
642: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
643: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
644: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
645: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
646: } else if (ksp->reason <= 0) {
647: if (prefix) PetscCall(PetscViewerASCIIPrintf(viewer, "Linear %s solve did not converge due to %s iterations %" PetscInt_FMT, prefix, reason, its));
648: else PetscCall(PetscViewerASCIIPrintf(viewer, "Linear solve did not converge due to %s iterations %" PetscInt_FMT, reason, its));
649: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
650: if (rRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " res rate %g R^2 %g", (double)rrate, (double)rRsq));
651: if (eRsq >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " error rate %g R^2 %g", (double)erate, (double)eRsq));
652: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
653: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
654: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
655: PCFailedReason reason;
656: PetscCall(PCGetFailedReason(ksp->pc, &reason));
657: PetscCall(PetscViewerASCIIPrintf(viewer, " PC failed due to %s \n", PCFailedReasons[reason]));
658: }
659: }
660: PetscCall(PetscViewerASCIISubtractTab(viewer, ((PetscObject)ksp)->tablevel));
661: }
662: PetscFunctionReturn(PETSC_SUCCESS);
663: }
665: #include <petscdraw.h>
667: static PetscErrorCode KSPViewEigenvalues_Internal(KSP ksp, PetscBool isExplicit, PetscViewer viewer, PetscViewerFormat format)
668: {
669: PetscReal *r, *c;
670: PetscInt n, i, neig;
671: PetscBool isascii, isdraw;
672: PetscMPIInt rank;
674: PetscFunctionBegin;
675: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)ksp), &rank));
676: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
677: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
678: if (isExplicit) {
679: PetscCall(VecGetSize(ksp->vec_sol, &n));
680: PetscCall(PetscMalloc2(n, &r, n, &c));
681: PetscCall(KSPComputeEigenvaluesExplicitly(ksp, n, r, c));
682: neig = n;
683: } else {
684: PetscInt nits;
686: PetscCall(KSPGetIterationNumber(ksp, &nits));
687: n = nits + 2;
688: if (!nits) {
689: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any eigenvalues\n"));
690: PetscFunctionReturn(PETSC_SUCCESS);
691: }
692: PetscCall(PetscMalloc2(n, &r, n, &c));
693: PetscCall(KSPComputeEigenvalues(ksp, n, r, c, &neig));
694: }
695: if (isascii) {
696: PetscCall(PetscViewerASCIIPrintf(viewer, "%s computed eigenvalues\n", isExplicit ? "Explicitly" : "Iteratively"));
697: for (i = 0; i < neig; ++i) {
698: if (c[i] >= 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, "%g + %gi\n", (double)r[i], (double)c[i]));
699: else PetscCall(PetscViewerASCIIPrintf(viewer, "%g - %gi\n", (double)r[i], -(double)c[i]));
700: }
701: } else if (isdraw && rank == 0) {
702: PetscDraw draw;
703: PetscDrawSP drawsp;
705: if (format == PETSC_VIEWER_DRAW_CONTOUR) {
706: PetscCall(KSPPlotEigenContours_Private(ksp, neig, r, c));
707: } else {
708: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
709: PetscCall(PetscDrawSPCreate(draw, 1, &drawsp));
710: PetscCall(PetscDrawSPReset(drawsp));
711: for (i = 0; i < neig; ++i) PetscCall(PetscDrawSPAddPoint(drawsp, r + i, c + i));
712: PetscCall(PetscDrawSPDraw(drawsp, PETSC_TRUE));
713: PetscCall(PetscDrawSPSave(drawsp));
714: PetscCall(PetscDrawSPDestroy(&drawsp));
715: }
716: }
717: PetscCall(PetscFree2(r, c));
718: PetscFunctionReturn(PETSC_SUCCESS);
719: }
721: static PetscErrorCode KSPViewSingularvalues_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
722: {
723: PetscReal smax, smin;
724: PetscInt nits;
725: PetscBool isascii;
727: PetscFunctionBegin;
728: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
729: PetscCall(KSPGetIterationNumber(ksp, &nits));
730: if (!nits) {
731: PetscCall(PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any singular values\n"));
732: PetscFunctionReturn(PETSC_SUCCESS);
733: }
734: PetscCall(KSPComputeExtremeSingularValues(ksp, &smax, &smin));
735: 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)));
736: PetscFunctionReturn(PETSC_SUCCESS);
737: }
739: static PetscErrorCode KSPViewFinalResidual_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
740: {
741: PetscBool isascii;
743: PetscFunctionBegin;
744: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii));
745: 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");
746: if (isascii) {
747: Mat A;
748: Vec t;
749: PetscReal norm;
751: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
752: PetscCall(VecDuplicate(ksp->vec_rhs, &t));
753: PetscCall(KSP_MatMult(ksp, A, ksp->vec_sol, t));
754: PetscCall(VecAYPX(t, -1.0, ksp->vec_rhs));
755: PetscCall(VecViewFromOptions(t, (PetscObject)ksp, "-ksp_view_final_residual_vec"));
756: PetscCall(VecNorm(t, NORM_2, &norm));
757: PetscCall(VecDestroy(&t));
758: PetscCall(PetscViewerASCIIPrintf(viewer, "KSP final norm of residual %g\n", (double)norm));
759: }
760: PetscFunctionReturn(PETSC_SUCCESS);
761: }
763: static PetscErrorCode KSPMonitorPauseFinal_Internal(KSP ksp)
764: {
765: PetscInt i;
767: PetscFunctionBegin;
768: if (!ksp->pauseFinal) PetscFunctionReturn(PETSC_SUCCESS);
769: for (i = 0; i < ksp->numbermonitors; ++i) {
770: PetscViewerAndFormat *vf = (PetscViewerAndFormat *)ksp->monitorcontext[i];
771: PetscDraw draw;
772: PetscReal lpause;
774: if (!vf) continue;
775: if (vf->lg) {
776: if (!PetscCheckPointer(vf->lg, PETSC_OBJECT)) continue;
777: if (((PetscObject)vf->lg)->classid != PETSC_DRAWLG_CLASSID) continue;
778: PetscCall(PetscDrawLGGetDraw(vf->lg, &draw));
779: PetscCall(PetscDrawGetPause(draw, &lpause));
780: PetscCall(PetscDrawSetPause(draw, -1.0));
781: PetscCall(PetscDrawPause(draw));
782: PetscCall(PetscDrawSetPause(draw, lpause));
783: } else {
784: PetscBool isdraw;
786: if (!PetscCheckPointer(vf->viewer, PETSC_OBJECT)) continue;
787: if (((PetscObject)vf->viewer)->classid != PETSC_VIEWER_CLASSID) continue;
788: PetscCall(PetscObjectTypeCompare((PetscObject)vf->viewer, PETSCVIEWERDRAW, &isdraw));
789: if (!isdraw) continue;
790: PetscCall(PetscViewerDrawGetDraw(vf->viewer, 0, &draw));
791: PetscCall(PetscDrawGetPause(draw, &lpause));
792: PetscCall(PetscDrawSetPause(draw, -1.0));
793: PetscCall(PetscDrawPause(draw));
794: PetscCall(PetscDrawSetPause(draw, lpause));
795: }
796: }
797: PetscFunctionReturn(PETSC_SUCCESS);
798: }
800: static PetscErrorCode KSPSolve_Private(KSP ksp, Vec b, Vec x)
801: {
802: PetscBool flg = PETSC_FALSE, inXisinB = PETSC_FALSE, guess_zero;
803: Mat mat, pmat;
804: MPI_Comm comm;
805: MatNullSpace nullsp;
806: Vec btmp, vec_rhs = NULL;
808: PetscFunctionBegin;
809: level++;
810: comm = PetscObjectComm((PetscObject)ksp);
811: if (x && x == b) {
812: PetscCheck(ksp->guess_zero, comm, PETSC_ERR_ARG_INCOMP, "Cannot use x == b with nonzero initial guess");
813: PetscCall(VecDuplicate(b, &x));
814: inXisinB = PETSC_TRUE;
815: }
816: if (b) {
817: PetscCall(PetscObjectReference((PetscObject)b));
818: PetscCall(VecDestroy(&ksp->vec_rhs));
819: ksp->vec_rhs = b;
820: }
821: if (x) {
822: PetscCall(PetscObjectReference((PetscObject)x));
823: PetscCall(VecDestroy(&ksp->vec_sol));
824: ksp->vec_sol = x;
825: }
827: if (ksp->viewPre) PetscCall(ObjectView((PetscObject)ksp, ksp->viewerPre, ksp->formatPre));
829: if (ksp->presolve) PetscCall((*ksp->presolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->prectx));
831: /* reset the residual history list if requested */
832: if (ksp->res_hist_reset) ksp->res_hist_len = 0;
833: if (ksp->err_hist_reset) ksp->err_hist_len = 0;
835: /* KSPSetUp() scales the matrix if needed */
836: PetscCall(KSPSetUp(ksp));
837: PetscCall(KSPSetUpOnBlocks(ksp));
839: if (ksp->guess) {
840: PetscObjectState ostate, state;
842: PetscCall(KSPGuessSetUp(ksp->guess));
843: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &ostate));
844: PetscCall(KSPGuessFormGuess(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
845: PetscCall(PetscObjectStateGet((PetscObject)ksp->vec_sol, &state));
846: if (state != ostate) {
847: ksp->guess_zero = PETSC_FALSE;
848: } else {
849: PetscCall(PetscInfo(ksp, "Using zero initial guess since the KSPGuess object did not change the vector\n"));
850: ksp->guess_zero = PETSC_TRUE;
851: }
852: }
854: PetscCall(VecSetErrorIfLocked(ksp->vec_sol, 3));
856: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
857: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
858: /* diagonal scale RHS if called for */
859: if (ksp->dscale) {
860: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
861: /* second time in, but matrix was scaled back to original */
862: if (ksp->dscalefix && ksp->dscalefix2) {
863: Mat mat, pmat;
865: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
866: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
867: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
868: }
870: /* scale initial guess */
871: if (!ksp->guess_zero) {
872: if (!ksp->truediagonal) {
873: PetscCall(VecDuplicate(ksp->diagonal, &ksp->truediagonal));
874: PetscCall(VecCopy(ksp->diagonal, ksp->truediagonal));
875: PetscCall(VecReciprocal(ksp->truediagonal));
876: }
877: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->truediagonal));
878: }
879: }
880: PetscCall(PCPreSolve(ksp->pc, ksp));
882: if (ksp->guess_zero && !ksp->guess_not_read) PetscCall(VecSet(ksp->vec_sol, 0.0));
883: if (ksp->guess_knoll) { /* The Knoll trick is independent on the KSPGuess specified */
884: PetscCall(PCApply(ksp->pc, ksp->vec_rhs, ksp->vec_sol));
885: PetscCall(KSP_RemoveNullSpace(ksp, ksp->vec_sol));
886: ksp->guess_zero = PETSC_FALSE;
887: }
889: /* can we mark the initial guess as zero for this solve? */
890: guess_zero = ksp->guess_zero;
891: if (!ksp->guess_zero) {
892: PetscReal norm;
894: PetscCall(VecNormAvailable(ksp->vec_sol, NORM_2, &flg, &norm));
895: if (flg && !norm) ksp->guess_zero = PETSC_TRUE;
896: }
897: if (ksp->transpose_solve) {
898: PetscCall(MatGetNullSpace(pmat, &nullsp));
899: } else {
900: PetscCall(MatGetTransposeNullSpace(pmat, &nullsp));
901: }
902: if (nullsp) {
903: PetscCall(VecDuplicate(ksp->vec_rhs, &btmp));
904: PetscCall(VecCopy(ksp->vec_rhs, btmp));
905: PetscCall(MatNullSpaceRemove(nullsp, btmp));
906: vec_rhs = ksp->vec_rhs;
907: ksp->vec_rhs = btmp;
908: }
909: PetscCall(VecLockReadPush(ksp->vec_rhs));
910: PetscUseTypeMethod(ksp, solve);
911: PetscCall(KSPMonitorPauseFinal_Internal(ksp));
913: PetscCall(VecLockReadPop(ksp->vec_rhs));
914: if (nullsp) {
915: ksp->vec_rhs = vec_rhs;
916: PetscCall(VecDestroy(&btmp));
917: }
919: ksp->guess_zero = guess_zero;
921: PetscCheck(ksp->reason, comm, PETSC_ERR_PLIB, "Internal error, solver returned without setting converged reason");
922: ksp->totalits += ksp->its;
924: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
926: if (ksp->viewRate) {
927: PetscCall(PetscViewerPushFormat(ksp->viewerRate, ksp->formatRate));
928: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
929: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
930: }
931: PetscCall(PCPostSolve(ksp->pc, ksp));
933: /* diagonal scale solution if called for */
934: if (ksp->dscale) {
935: PetscCall(VecPointwiseMult(ksp->vec_sol, ksp->vec_sol, ksp->diagonal));
936: /* unscale right hand side and matrix */
937: if (ksp->dscalefix) {
938: Mat mat, pmat;
940: PetscCall(VecReciprocal(ksp->diagonal));
941: PetscCall(VecPointwiseMult(ksp->vec_rhs, ksp->vec_rhs, ksp->diagonal));
942: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
943: PetscCall(MatDiagonalScale(pmat, ksp->diagonal, ksp->diagonal));
944: if (mat != pmat) PetscCall(MatDiagonalScale(mat, ksp->diagonal, ksp->diagonal));
945: PetscCall(VecReciprocal(ksp->diagonal));
946: ksp->dscalefix2 = PETSC_TRUE;
947: }
948: }
949: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_Solve : KSP_SolveTranspose, ksp, ksp->vec_rhs, ksp->vec_sol, 0));
950: if (ksp->guess) PetscCall(KSPGuessUpdate(ksp->guess, ksp->vec_rhs, ksp->vec_sol));
951: if (ksp->postsolve) PetscCall((*ksp->postsolve)(ksp, ksp->vec_rhs, ksp->vec_sol, ksp->postctx));
953: PetscCall(PCGetOperators(ksp->pc, &mat, &pmat));
954: if (ksp->viewEV) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_FALSE, ksp->viewerEV, ksp->formatEV));
955: if (ksp->viewEVExp) PetscCall(KSPViewEigenvalues_Internal(ksp, PETSC_TRUE, ksp->viewerEVExp, ksp->formatEVExp));
956: if (ksp->viewSV) PetscCall(KSPViewSingularvalues_Internal(ksp, ksp->viewerSV, ksp->formatSV));
957: if (ksp->viewFinalRes) PetscCall(KSPViewFinalResidual_Internal(ksp, ksp->viewerFinalRes, ksp->formatFinalRes));
958: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)mat, ksp->viewerMat, ksp->formatMat));
959: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)pmat, ksp->viewerPMat, ksp->formatPMat));
960: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)ksp->vec_rhs, ksp->viewerRhs, ksp->formatRhs));
961: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)ksp->vec_sol, ksp->viewerSol, ksp->formatSol));
962: if (ksp->view) PetscCall(ObjectView((PetscObject)ksp, ksp->viewer, ksp->format));
963: if (ksp->viewDScale) PetscCall(ObjectView((PetscObject)ksp->diagonal, ksp->viewerDScale, ksp->formatDScale));
964: if (ksp->viewMatExp) {
965: Mat A, B;
967: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
968: if (ksp->transpose_solve) {
969: Mat AT;
971: PetscCall(MatCreateTranspose(A, &AT));
972: PetscCall(MatComputeOperator(AT, MATAIJ, &B));
973: PetscCall(MatDestroy(&AT));
974: } else {
975: PetscCall(MatComputeOperator(A, MATAIJ, &B));
976: }
977: PetscCall(ObjectView((PetscObject)B, ksp->viewerMatExp, ksp->formatMatExp));
978: PetscCall(MatDestroy(&B));
979: }
980: if (ksp->viewPOpExp) {
981: Mat B;
983: PetscCall(KSPComputeOperator(ksp, MATAIJ, &B));
984: PetscCall(ObjectView((PetscObject)B, ksp->viewerPOpExp, ksp->formatPOpExp));
985: PetscCall(MatDestroy(&B));
986: }
988: if (inXisinB) {
989: PetscCall(VecCopy(x, b));
990: PetscCall(VecDestroy(&x));
991: }
992: PetscCall(PetscObjectSAWsBlock((PetscObject)ksp));
993: if (ksp->errorifnotconverged && ksp->reason < 0 && ((level == 1) || (ksp->reason != KSP_DIVERGED_ITS))) {
994: PCFailedReason reason;
996: 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]);
997: PetscCall(PCGetFailedReason(ksp->pc, &reason));
998: 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]);
999: }
1000: level--;
1001: PetscFunctionReturn(PETSC_SUCCESS);
1002: }
1004: /*@
1005: KSPSolve - Solves linear system.
1007: Collective
1009: Input Parameters:
1010: + ksp - iterative context obtained from `KSPCreate()`
1011: . b - the right hand side vector
1012: - x - the solution (this may be the same vector as b, then b will be overwritten with answer)
1014: Options Database Keys:
1015: + -ksp_view_eigenvalues - compute preconditioned operators eigenvalues
1016: . -ksp_view_eigenvalues_explicit - compute the eigenvalues by forming the dense operator and using LAPACK
1017: . -ksp_view_mat binary - save matrix to the default binary viewer
1018: . -ksp_view_pmat binary - save matrix used to build preconditioner to the default binary viewer
1019: . -ksp_view_rhs binary - save right hand side vector to the default binary viewer
1020: . -ksp_view_solution binary - save computed solution vector to the default binary viewer
1021: (can be read later with src/ksp/tutorials/ex10.c for testing solvers)
1022: . -ksp_view_mat_explicit - for matrix-free operators, computes the matrix entries and views them
1023: . -ksp_view_preconditioned_operator_explicit - computes the product of the preconditioner and matrix as an explicit matrix and views it
1024: . -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
1025: . -ksp_view_final_residual - print 2-norm of true linear system residual at the end of the solution process
1026: . -ksp_error_if_not_converged - stop the program as soon as an error is detected in a `KSPSolve()`
1027: - -ksp_view - print the ksp data structure at the end of the system solution
1029: Level: beginner
1031: Notes:
1032: If one uses `KSPSetDM()` then x or b need not be passed. Use `KSPGetSolution()` to access the solution in this case.
1034: The operator is specified with `KSPSetOperators()`.
1036: `KSPSolve()` will normally return without generating an error regardless of whether the linear system was solved or if constructing the preconditioner failed.
1037: Call `KSPGetConvergedReason()` to determine if the solver converged or failed and why. The option -ksp_error_if_not_converged or function `KSPSetErrorIfNotConverged()`
1038: will cause `KSPSolve()` to error as soon as an error occurs in the linear solver. In inner KSPSolves() KSP_DIVERGED_ITS is not treated as an error because when using nested solvers
1039: it may be fine that inner solvers in the preconditioner do not converge during the solution process.
1041: The number of iterations can be obtained from `KSPGetIterationNumber()`.
1043: If you provide a matrix that has a `MatSetNullSpace()` and `MatSetTransposeNullSpace()` this will use that information to solve singular systems
1044: in the least squares sense with a norm minimizing solution.
1046: 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()`
1048: `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
1049: 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
1050: direction thus the solution which is a linear combination of the search directions has no component in the nullspace(A).
1052: We recommend always using `KSPGMRES` for such singular systems.
1053: If nullspace(A) = nullspace(A') (note symmetric matrices always satisfy this property) then both left and right preconditioning will work
1054: If nullspace(A) != nullspace(A') then left preconditioning will work but right preconditioning may not work (or it may).
1056: Developer Notes:
1057: The reason we cannot always solve nullspace(A) != nullspace(A') systems with right preconditioning is because we need to remove at each iteration
1058: 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
1059: such as diagonal scaling we cannot apply the inverse of the preconditioner to a vector and thus cannot compute the nullspace(AB).
1061: If using a direct method (e.g., via the `KSP` solver
1062: `KSPPREONLY` and a preconditioner such as `PCLU` or `PCILU`,
1063: then its=1. See `KSPSetTolerances()` and `KSPConvergedDefault()`
1064: for more details.
1066: Understanding Convergence\:
1067: The routines `KSPMonitorSet()`, `KSPComputeEigenvalues()`, and
1068: `KSPComputeEigenvaluesExplicitly()` provide information on additional
1069: options to monitor convergence and print eigenvalue information.
1071: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1072: `KSPSolveTranspose()`, `KSPGetIterationNumber()`, `MatNullSpaceCreate()`, `MatSetNullSpace()`, `MatSetTransposeNullSpace()`, `KSP`,
1073: `KSPConvergedReasonView()`, `KSPCheckSolve()`, `KSPSetErrorIfNotConverged()`
1074: @*/
1075: PetscErrorCode KSPSolve(KSP ksp, Vec b, Vec x)
1076: {
1077: PetscFunctionBegin;
1081: ksp->transpose_solve = PETSC_FALSE;
1082: PetscCall(KSPSolve_Private(ksp, b, x));
1083: PetscFunctionReturn(PETSC_SUCCESS);
1084: }
1086: /*@
1087: KSPSolveTranspose - Solves a linear system with the transposed matrix.
1089: Collective
1091: Input Parameters:
1092: + ksp - iterative context obtained from `KSPCreate()`
1093: . b - right hand side vector
1094: - x - solution vector
1096: Level: developer
1098: Note:
1099: For complex numbers this solve the non-Hermitian transpose system.
1101: Developer Notes:
1102: We need to implement a `KSPSolveHermitianTranspose()`
1104: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPDestroy()`, `KSPSetTolerances()`, `KSPConvergedDefault()`,
1105: `KSPSolve()`, `KSP`
1106: @*/
1107: PetscErrorCode KSPSolveTranspose(KSP ksp, Vec b, Vec x)
1108: {
1109: PetscFunctionBegin;
1113: if (ksp->transpose.use_explicittranspose) {
1114: Mat J, Jpre;
1115: PetscCall(KSPGetOperators(ksp, &J, &Jpre));
1116: if (!ksp->transpose.reuse_transpose) {
1117: PetscCall(MatTranspose(J, MAT_INITIAL_MATRIX, &ksp->transpose.AT));
1118: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_INITIAL_MATRIX, &ksp->transpose.BT));
1119: ksp->transpose.reuse_transpose = PETSC_TRUE;
1120: } else {
1121: PetscCall(MatTranspose(J, MAT_REUSE_MATRIX, &ksp->transpose.AT));
1122: if (J != Jpre) PetscCall(MatTranspose(Jpre, MAT_REUSE_MATRIX, &ksp->transpose.BT));
1123: }
1124: if (J == Jpre && ksp->transpose.BT != ksp->transpose.AT) {
1125: PetscCall(PetscObjectReference((PetscObject)ksp->transpose.AT));
1126: ksp->transpose.BT = ksp->transpose.AT;
1127: }
1128: PetscCall(KSPSetOperators(ksp, ksp->transpose.AT, ksp->transpose.BT));
1129: } else {
1130: ksp->transpose_solve = PETSC_TRUE;
1131: }
1132: PetscCall(KSPSolve_Private(ksp, b, x));
1133: PetscFunctionReturn(PETSC_SUCCESS);
1134: }
1136: static PetscErrorCode KSPViewFinalMatResidual_Internal(KSP ksp, Mat B, Mat X, PetscViewer viewer, PetscViewerFormat format, PetscInt shift)
1137: {
1138: Mat A, R;
1139: PetscReal *norms;
1140: PetscInt i, N;
1141: PetscBool flg;
1143: PetscFunctionBegin;
1144: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &flg));
1145: if (flg) {
1146: PetscCall(PCGetOperators(ksp->pc, &A, NULL));
1147: if (!ksp->transpose_solve) PetscCall(MatMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DEFAULT, &R));
1148: else PetscCall(MatTransposeMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DEFAULT, &R));
1149: PetscCall(MatAYPX(R, -1.0, B, SAME_NONZERO_PATTERN));
1150: PetscCall(MatGetSize(R, NULL, &N));
1151: PetscCall(PetscMalloc1(N, &norms));
1152: PetscCall(MatGetColumnNorms(R, NORM_2, norms));
1153: PetscCall(MatDestroy(&R));
1154: 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]));
1155: PetscCall(PetscFree(norms));
1156: }
1157: PetscFunctionReturn(PETSC_SUCCESS);
1158: }
1160: static PetscErrorCode KSPMatSolve_Private(KSP ksp, Mat B, Mat X)
1161: {
1162: Mat A, P, vB, vX;
1163: Vec cb, cx;
1164: PetscInt n1, N1, n2, N2, Bbn = PETSC_DECIDE;
1165: PetscBool match;
1167: PetscFunctionBegin;
1171: PetscCheckSameComm(ksp, 1, B, 2);
1172: PetscCheckSameComm(ksp, 1, X, 3);
1173: PetscCheckSameType(B, 2, X, 3);
1174: PetscCheck(B->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
1175: MatCheckPreallocated(X, 3);
1176: if (!X->assembled) {
1177: PetscCall(MatSetOption(X, MAT_NO_OFF_PROC_ENTRIES, PETSC_TRUE));
1178: PetscCall(MatAssemblyBegin(X, MAT_FINAL_ASSEMBLY));
1179: PetscCall(MatAssemblyEnd(X, MAT_FINAL_ASSEMBLY));
1180: }
1181: PetscCheck(B != X, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_IDN, "B and X must be different matrices");
1182: PetscCheck(!ksp->transpose_solve || !ksp->transpose.use_explicittranspose, PetscObjectComm((PetscObject)ksp), PETSC_ERR_SUP, "KSPMatSolveTranspose() does not support -ksp_use_explicittranspose");
1183: PetscCall(KSPGetOperators(ksp, &A, &P));
1184: PetscCall(MatGetLocalSize(B, NULL, &n2));
1185: PetscCall(MatGetLocalSize(X, NULL, &n1));
1186: PetscCall(MatGetSize(B, NULL, &N2));
1187: PetscCall(MatGetSize(X, NULL, &N1));
1188: 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);
1189: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)B, &match, MATSEQDENSE, MATMPIDENSE, ""));
1190: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of right-hand sides not stored in a dense Mat");
1191: PetscCall(PetscObjectBaseTypeCompareAny((PetscObject)X, &match, MATSEQDENSE, MATMPIDENSE, ""));
1192: PetscCheck(match, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of solutions not stored in a dense Mat");
1193: PetscCall(KSPSetUp(ksp));
1194: PetscCall(KSPSetUpOnBlocks(ksp));
1195: if (ksp->ops->matsolve) {
1196: level++;
1197: if (ksp->guess_zero) PetscCall(MatZeroEntries(X));
1198: PetscCall(PetscLogEventBegin(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1199: PetscCall(KSPGetMatSolveBatchSize(ksp, &Bbn));
1200: /* by default, do a single solve with all columns */
1201: if (Bbn == PETSC_DECIDE) Bbn = N2;
1202: else PetscCheck(Bbn >= 1, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "KSPMatSolve() batch size %" PetscInt_FMT " must be positive", Bbn);
1203: PetscCall(PetscInfo(ksp, "KSP type %s solving using batches of width at most %" PetscInt_FMT "\n", ((PetscObject)ksp)->type_name, Bbn));
1204: /* if -ksp_matsolve_batch_size is greater than the actual number of columns, do a single solve with all columns */
1205: if (Bbn >= N2) {
1206: PetscUseTypeMethod(ksp, matsolve, B, X);
1207: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, B, X, ksp->viewerFinalRes, ksp->formatFinalRes, 0));
1209: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1211: if (ksp->viewRate) {
1212: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1213: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1214: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1215: }
1216: } else {
1217: for (n2 = 0; n2 < N2; n2 += Bbn) {
1218: PetscCall(MatDenseGetSubMatrix(B, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vB));
1219: PetscCall(MatDenseGetSubMatrix(X, PETSC_DECIDE, PETSC_DECIDE, n2, PetscMin(n2 + Bbn, N2), &vX));
1220: PetscUseTypeMethod(ksp, matsolve, vB, vX);
1221: if (ksp->viewFinalRes) PetscCall(KSPViewFinalMatResidual_Internal(ksp, vB, vX, ksp->viewerFinalRes, ksp->formatFinalRes, n2));
1223: PetscCall(KSPConvergedReasonViewFromOptions(ksp));
1225: if (ksp->viewRate) {
1226: PetscCall(PetscViewerPushFormat(ksp->viewerRate, PETSC_VIEWER_DEFAULT));
1227: PetscCall(KSPConvergedRateView(ksp, ksp->viewerRate));
1228: PetscCall(PetscViewerPopFormat(ksp->viewerRate));
1229: }
1230: PetscCall(MatDenseRestoreSubMatrix(B, &vB));
1231: PetscCall(MatDenseRestoreSubMatrix(X, &vX));
1232: }
1233: }
1234: if (ksp->viewMat) PetscCall(ObjectView((PetscObject)A, ksp->viewerMat, ksp->formatMat));
1235: if (ksp->viewPMat) PetscCall(ObjectView((PetscObject)P, ksp->viewerPMat, ksp->formatPMat));
1236: if (ksp->viewRhs) PetscCall(ObjectView((PetscObject)B, ksp->viewerRhs, ksp->formatRhs));
1237: if (ksp->viewSol) PetscCall(ObjectView((PetscObject)X, ksp->viewerSol, ksp->formatSol));
1238: if (ksp->view) PetscCall(KSPView(ksp, ksp->viewer));
1239: PetscCall(PetscLogEventEnd(!ksp->transpose_solve ? KSP_MatSolve : KSP_MatSolveTranspose, ksp, B, X, 0));
1240: if (ksp->errorifnotconverged && ksp->reason < 0 && (level == 1 || ksp->reason != KSP_DIVERGED_ITS)) {
1241: PCFailedReason reason;
1243: 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]);
1244: PetscCall(PCGetFailedReason(ksp->pc, &reason));
1245: 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]);
1246: }
1247: level--;
1248: } else {
1249: PetscCall(PetscInfo(ksp, "KSP type %s solving column by column\n", ((PetscObject)ksp)->type_name));
1250: for (n2 = 0; n2 < N2; ++n2) {
1251: PetscCall(MatDenseGetColumnVecRead(B, n2, &cb));
1252: PetscCall(MatDenseGetColumnVecWrite(X, n2, &cx));
1253: PetscCall(KSPSolve_Private(ksp, cb, cx));
1254: PetscCall(MatDenseRestoreColumnVecWrite(X, n2, &cx));
1255: PetscCall(MatDenseRestoreColumnVecRead(B, n2, &cb));
1256: }
1257: }
1258: PetscFunctionReturn(PETSC_SUCCESS);
1259: }
1261: /*@
1262: KSPMatSolve - Solves a linear system with multiple right-hand sides stored as a `MATDENSE`. Unlike `KSPSolve()`, `B` and `X` must be different matrices.
1264: Input Parameters:
1265: + ksp - iterative context
1266: - B - block of right-hand sides
1268: Output Parameter:
1269: . X - block of solutions
1271: Level: intermediate
1273: Note:
1274: This is a stripped-down version of `KSPSolve()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1276: .seealso: [](ch_ksp), `KSPSolve()`, `MatMatSolve()`, `KSPMatSolveTranspose()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`
1277: @*/
1278: PetscErrorCode KSPMatSolve(KSP ksp, Mat B, Mat X)
1279: {
1280: PetscFunctionBegin;
1281: ksp->transpose_solve = PETSC_FALSE;
1282: PetscCall(KSPMatSolve_Private(ksp, B, X));
1283: PetscFunctionReturn(PETSC_SUCCESS);
1284: }
1286: /*@
1287: KSPMatSolveTranspose - Solves a linear system with the transposed matrix with multiple right-hand sides stored as a `MATDENSE`. Unlike `KSPSolveTranspose()`, `B` and `X` must be different matrices and the transposed matrix cannot be assembled explicitly for the user.
1289: Input Parameters:
1290: + ksp - iterative context
1291: - B - block of right-hand sides
1293: Output Parameter:
1294: . X - block of solutions
1296: Level: intermediate
1298: Notes:
1299: This is a stripped-down version of `KSPSolveTranspose()`, which only handles `-ksp_view`, `-ksp_converged_reason`, `-ksp_converged_rate`, and `-ksp_view_final_residual`.
1301: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `MatMatTransposeSolve()`, `KSPMatSolve()`, `MATDENSE`, `KSPHPDDM`, `PCBJACOBI`, `PCASM`
1302: @*/
1303: PetscErrorCode KSPMatSolveTranspose(KSP ksp, Mat B, Mat X)
1304: {
1305: PetscFunctionBegin;
1306: ksp->transpose_solve = PETSC_TRUE;
1307: PetscCall(KSPMatSolve_Private(ksp, B, X));
1308: PetscFunctionReturn(PETSC_SUCCESS);
1309: }
1311: /*@
1312: KSPSetMatSolveBatchSize - Sets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1314: Logically Collective
1316: Input Parameters:
1317: + ksp - iterative context
1318: - bs - batch size
1320: Level: advanced
1322: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPGetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matmatmult_Bbn`
1323: @*/
1324: PetscErrorCode KSPSetMatSolveBatchSize(KSP ksp, PetscInt bs)
1325: {
1326: PetscFunctionBegin;
1329: ksp->nmax = bs;
1330: PetscFunctionReturn(PETSC_SUCCESS);
1331: }
1333: /*@
1334: KSPGetMatSolveBatchSize - Gets the maximum number of columns treated simultaneously in `KSPMatSolve()`.
1336: Input Parameter:
1337: . ksp - iterative context
1339: Output Parameter:
1340: . bs - batch size
1342: Level: advanced
1344: .seealso: [](ch_ksp), `KSPMatSolve()`, `KSPSetMatSolveBatchSize()`, `-mat_mumps_icntl_27`, `-matmatmult_Bbn`
1345: @*/
1346: PetscErrorCode KSPGetMatSolveBatchSize(KSP ksp, PetscInt *bs)
1347: {
1348: PetscFunctionBegin;
1350: PetscAssertPointer(bs, 2);
1351: *bs = ksp->nmax;
1352: PetscFunctionReturn(PETSC_SUCCESS);
1353: }
1355: /*@
1356: KSPResetViewers - Resets all the viewers set from the options database during `KSPSetFromOptions()`
1358: Collective
1360: Input Parameter:
1361: . ksp - iterative context obtained from `KSPCreate()`
1363: Level: beginner
1365: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSPSetFromOptions()`, `KSP`
1366: @*/
1367: PetscErrorCode KSPResetViewers(KSP ksp)
1368: {
1369: PetscFunctionBegin;
1371: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1372: PetscCall(PetscViewerDestroy(&ksp->viewer));
1373: PetscCall(PetscViewerDestroy(&ksp->viewerPre));
1374: PetscCall(PetscViewerDestroy(&ksp->viewerRate));
1375: PetscCall(PetscViewerDestroy(&ksp->viewerMat));
1376: PetscCall(PetscViewerDestroy(&ksp->viewerPMat));
1377: PetscCall(PetscViewerDestroy(&ksp->viewerRhs));
1378: PetscCall(PetscViewerDestroy(&ksp->viewerSol));
1379: PetscCall(PetscViewerDestroy(&ksp->viewerMatExp));
1380: PetscCall(PetscViewerDestroy(&ksp->viewerEV));
1381: PetscCall(PetscViewerDestroy(&ksp->viewerSV));
1382: PetscCall(PetscViewerDestroy(&ksp->viewerEVExp));
1383: PetscCall(PetscViewerDestroy(&ksp->viewerFinalRes));
1384: PetscCall(PetscViewerDestroy(&ksp->viewerPOpExp));
1385: PetscCall(PetscViewerDestroy(&ksp->viewerDScale));
1386: ksp->view = PETSC_FALSE;
1387: ksp->viewPre = PETSC_FALSE;
1388: ksp->viewMat = PETSC_FALSE;
1389: ksp->viewPMat = PETSC_FALSE;
1390: ksp->viewRhs = PETSC_FALSE;
1391: ksp->viewSol = PETSC_FALSE;
1392: ksp->viewMatExp = PETSC_FALSE;
1393: ksp->viewEV = PETSC_FALSE;
1394: ksp->viewSV = PETSC_FALSE;
1395: ksp->viewEVExp = PETSC_FALSE;
1396: ksp->viewFinalRes = PETSC_FALSE;
1397: ksp->viewPOpExp = PETSC_FALSE;
1398: ksp->viewDScale = PETSC_FALSE;
1399: PetscFunctionReturn(PETSC_SUCCESS);
1400: }
1402: /*@
1403: KSPReset - Resets a `KSP` context to the kspsetupcalled = 0 state and removes any allocated Vecs and Mats
1405: Collective
1407: Input Parameter:
1408: . ksp - iterative context obtained from `KSPCreate()`
1410: Level: beginner
1412: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1413: @*/
1414: PetscErrorCode KSPReset(KSP ksp)
1415: {
1416: PetscFunctionBegin;
1418: if (!ksp) PetscFunctionReturn(PETSC_SUCCESS);
1419: PetscTryTypeMethod(ksp, reset);
1420: if (ksp->pc) PetscCall(PCReset(ksp->pc));
1421: if (ksp->guess) {
1422: KSPGuess guess = ksp->guess;
1423: PetscTryTypeMethod(guess, reset);
1424: }
1425: PetscCall(VecDestroyVecs(ksp->nwork, &ksp->work));
1426: PetscCall(VecDestroy(&ksp->vec_rhs));
1427: PetscCall(VecDestroy(&ksp->vec_sol));
1428: PetscCall(VecDestroy(&ksp->diagonal));
1429: PetscCall(VecDestroy(&ksp->truediagonal));
1431: PetscCall(KSPResetViewers(ksp));
1433: ksp->setupstage = KSP_SETUP_NEW;
1434: ksp->nmax = PETSC_DECIDE;
1435: PetscFunctionReturn(PETSC_SUCCESS);
1436: }
1438: /*@C
1439: KSPDestroy - Destroys `KSP` context.
1441: Collective
1443: Input Parameter:
1444: . ksp - iterative context obtained from `KSPCreate()`
1446: Level: beginner
1448: .seealso: [](ch_ksp), `KSPCreate()`, `KSPSetUp()`, `KSPSolve()`, `KSP`
1449: @*/
1450: PetscErrorCode KSPDestroy(KSP *ksp)
1451: {
1452: PC pc;
1454: PetscFunctionBegin;
1455: if (!*ksp) PetscFunctionReturn(PETSC_SUCCESS);
1457: if (--((PetscObject)(*ksp))->refct > 0) {
1458: *ksp = NULL;
1459: PetscFunctionReturn(PETSC_SUCCESS);
1460: }
1462: PetscCall(PetscObjectSAWsViewOff((PetscObject)*ksp));
1464: /*
1465: Avoid a cascading call to PCReset(ksp->pc) from the following call:
1466: PCReset() shouldn't be called from KSPDestroy() as it is unprotected by pc's
1467: refcount (and may be shared, e.g., by other ksps).
1468: */
1469: pc = (*ksp)->pc;
1470: (*ksp)->pc = NULL;
1471: PetscCall(KSPReset((*ksp)));
1472: (*ksp)->pc = pc;
1473: PetscTryTypeMethod((*ksp), destroy);
1475: if ((*ksp)->transpose.use_explicittranspose) {
1476: PetscCall(MatDestroy(&(*ksp)->transpose.AT));
1477: PetscCall(MatDestroy(&(*ksp)->transpose.BT));
1478: (*ksp)->transpose.reuse_transpose = PETSC_FALSE;
1479: }
1481: PetscCall(KSPGuessDestroy(&(*ksp)->guess));
1482: PetscCall(DMDestroy(&(*ksp)->dm));
1483: PetscCall(PCDestroy(&(*ksp)->pc));
1484: PetscCall(PetscFree((*ksp)->res_hist_alloc));
1485: PetscCall(PetscFree((*ksp)->err_hist_alloc));
1486: if ((*ksp)->convergeddestroy) PetscCall((*(*ksp)->convergeddestroy)((*ksp)->cnvP));
1487: PetscCall(KSPMonitorCancel((*ksp)));
1488: PetscCall(KSPConvergedReasonViewCancel((*ksp)));
1489: PetscCall(PetscHeaderDestroy(ksp));
1490: PetscFunctionReturn(PETSC_SUCCESS);
1491: }
1493: /*@
1494: KSPSetPCSide - Sets the preconditioning side.
1496: Logically Collective
1498: Input Parameter:
1499: . ksp - iterative context obtained from `KSPCreate()`
1501: Output Parameter:
1502: . side - the preconditioning side, where side is one of
1503: .vb
1504: PC_LEFT - left preconditioning (default)
1505: PC_RIGHT - right preconditioning
1506: PC_SYMMETRIC - symmetric preconditioning
1507: .ve
1509: Options Database Key:
1510: . -ksp_pc_side <right,left,symmetric> - `KSP` preconditioner side
1512: Level: intermediate
1514: Notes:
1515: Left preconditioning is used by default for most Krylov methods except `KSPFGMRES` which only supports right preconditioning.
1517: For methods changing the side of the preconditioner changes the norm type that is used, see `KSPSetNormType()`.
1519: Symmetric preconditioning is currently available only for the `KSPQCG` method. However, note that
1520: symmetric preconditioning can be emulated by using either right or left
1521: preconditioning and a pre or post processing step.
1523: Setting the `PCSide` often affects the default norm type. See `KSPSetNormType()` for details.
1525: .seealso: [](ch_ksp), `KSPGetPCSide()`, `KSPSetNormType()`, `KSPGetNormType()`, `KSP`
1526: @*/
1527: PetscErrorCode KSPSetPCSide(KSP ksp, PCSide side)
1528: {
1529: PetscFunctionBegin;
1532: ksp->pc_side = ksp->pc_side_set = side;
1533: PetscFunctionReturn(PETSC_SUCCESS);
1534: }
1536: /*@
1537: KSPGetPCSide - Gets the preconditioning side.
1539: Not Collective
1541: Input Parameter:
1542: . ksp - iterative context obtained from `KSPCreate()`
1544: Output Parameter:
1545: . side - the preconditioning side, where side is one of
1546: .vb
1547: PC_LEFT - left preconditioning (default)
1548: PC_RIGHT - right preconditioning
1549: PC_SYMMETRIC - symmetric preconditioning
1550: .ve
1552: Level: intermediate
1554: .seealso: [](ch_ksp), `KSPSetPCSide()`, `KSP`
1555: @*/
1556: PetscErrorCode KSPGetPCSide(KSP ksp, PCSide *side)
1557: {
1558: PetscFunctionBegin;
1560: PetscAssertPointer(side, 2);
1561: PetscCall(KSPSetUpNorms_Private(ksp, PETSC_TRUE, &ksp->normtype, &ksp->pc_side));
1562: *side = ksp->pc_side;
1563: PetscFunctionReturn(PETSC_SUCCESS);
1564: }
1566: /*@
1567: KSPGetTolerances - Gets the relative, absolute, divergence, and maximum
1568: iteration tolerances used by the default `KSP` convergence tests.
1570: Not Collective
1572: Input Parameter:
1573: . ksp - the Krylov subspace context
1575: Output Parameters:
1576: + rtol - the relative convergence tolerance
1577: . abstol - the absolute convergence tolerance
1578: . dtol - the divergence tolerance
1579: - maxits - maximum number of iterations
1581: Level: intermediate
1583: Notes:
1584: The user can specify `NULL` for any parameter that is not needed.
1586: .seealso: [](ch_ksp), `KSPSetTolerances()`, `KSP`, `KSPSetMinimumIterations()`, `KSPGetMinimumIterations()`
1587: @*/
1588: PetscErrorCode KSPGetTolerances(KSP ksp, PetscReal *rtol, PetscReal *abstol, PetscReal *dtol, PetscInt *maxits)
1589: {
1590: PetscFunctionBegin;
1592: if (abstol) *abstol = ksp->abstol;
1593: if (rtol) *rtol = ksp->rtol;
1594: if (dtol) *dtol = ksp->divtol;
1595: if (maxits) *maxits = ksp->max_it;
1596: PetscFunctionReturn(PETSC_SUCCESS);
1597: }
1599: /*@
1600: KSPSetTolerances - Sets the relative, absolute, divergence, and maximum
1601: iteration tolerances used by the default `KSP` convergence testers.
1603: Logically Collective
1605: Input Parameters:
1606: + ksp - the Krylov subspace context
1607: . rtol - the relative convergence tolerance, relative decrease in the (possibly preconditioned) residual norm
1608: . abstol - the absolute convergence tolerance absolute size of the (possibly preconditioned) residual norm
1609: . dtol - the divergence tolerance, amount (possibly preconditioned) residual norm can increase before `KSPConvergedDefault()` concludes that the method is diverging
1610: - maxits - maximum number of iterations to use
1612: Options Database Keys:
1613: + -ksp_atol <abstol> - Sets `abstol`
1614: . -ksp_rtol <rtol> - Sets `rtol`
1615: . -ksp_divtol <dtol> - Sets `dtol`
1616: - -ksp_max_it <maxits> - Sets `maxits`
1618: Level: intermediate
1620: Notes:
1621: Use `PETSC_DEFAULT` to retain the default value of any of the tolerances.
1623: See `KSPConvergedDefault()` for details how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1624: for setting user-defined stopping criteria.
1626: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetMinimumIterations()`
1627: @*/
1628: PetscErrorCode KSPSetTolerances(KSP ksp, PetscReal rtol, PetscReal abstol, PetscReal dtol, PetscInt maxits)
1629: {
1630: PetscFunctionBegin;
1637: if (rtol != (PetscReal)PETSC_DEFAULT) {
1638: 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);
1639: ksp->rtol = rtol;
1640: }
1641: if (abstol != (PetscReal)PETSC_DEFAULT) {
1642: PetscCheck(abstol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Absolute tolerance %g must be non-negative", (double)abstol);
1643: ksp->abstol = abstol;
1644: }
1645: if (dtol != (PetscReal)PETSC_DEFAULT) {
1646: PetscCheck(dtol >= 0.0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Divergence tolerance %g must be larger than 1.0", (double)dtol);
1647: ksp->divtol = dtol;
1648: }
1649: if (maxits != PETSC_DEFAULT) {
1650: PetscCheck(maxits >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Maximum number of iterations %" PetscInt_FMT " must be non-negative", maxits);
1651: ksp->max_it = maxits;
1652: }
1653: PetscFunctionReturn(PETSC_SUCCESS);
1654: }
1656: /*@
1657: KSPSetMinimumIterations - Sets the minimum number of iterations to use, regardless of the tolerances
1659: Logically Collective
1661: Input Parameters:
1662: + ksp - the Krylov subspace context
1663: - minit - minimum number of iterations to use
1665: Options Database Keys:
1666: . -ksp_min_it <minits> - Sets `minit`
1668: Level: intermediate
1670: Notes:
1671: Use `KSPSetTolerances()` to set a variety of other tolerances
1673: See `KSPConvergedDefault()` for details on how these parameters are used in the default convergence test. See also `KSPSetConvergenceTest()`
1674: for setting user-defined stopping criteria.
1676: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPGetMinimumIterations()`
1677: @*/
1678: PetscErrorCode KSPSetMinimumIterations(KSP ksp, PetscInt minit)
1679: {
1680: PetscFunctionBegin;
1684: PetscCheck(minit >= 0, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Minimum number of iterations %" PetscInt_FMT " must be non-negative", minit);
1685: ksp->min_it = minit;
1686: PetscFunctionReturn(PETSC_SUCCESS);
1687: }
1689: /*@
1690: KSPGetMinimumIterations - Gets the minimum number of iterations to use, regardless of the tolerances, that was set with `KSPSetMinimumIterations()` or `-ksp_min_it`
1692: Not Collective
1694: Input Parameter:
1695: . ksp - the Krylov subspace context
1697: Output Parameter:
1698: . minit - minimum number of iterations to use
1700: Level: intermediate
1702: .seealso: [](ch_ksp), `KSPGetTolerances()`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSP`, `KSPSetTolerances()`, `KSPSetMinimumIterations()`
1703: @*/
1704: PetscErrorCode KSPGetMinimumIterations(KSP ksp, PetscInt *minit)
1705: {
1706: PetscFunctionBegin;
1708: PetscAssertPointer(minit, 2);
1710: *minit = ksp->min_it;
1711: PetscFunctionReturn(PETSC_SUCCESS);
1712: }
1714: /*@
1715: KSPSetInitialGuessNonzero - Tells the iterative solver that the
1716: initial guess is nonzero; otherwise `KSP` assumes the initial guess
1717: is to be zero (and thus zeros it out before solving).
1719: Logically Collective
1721: Input Parameters:
1722: + ksp - iterative context obtained from `KSPCreate()`
1723: - flg - ``PETSC_TRUE`` indicates the guess is non-zero, `PETSC_FALSE` indicates the guess is zero
1725: Options Database Key:
1726: . -ksp_initial_guess_nonzero <true,false> - use nonzero initial guess
1728: Level: beginner
1730: Notes:
1731: If this is not called the X vector is zeroed in the call to `KSPSolve()`.
1733: .seealso: [](ch_ksp), `KSPGetInitialGuessNonzero()`, `KSPSetGuessType()`, `KSPGuessType`, `KSP`
1734: @*/
1735: PetscErrorCode KSPSetInitialGuessNonzero(KSP ksp, PetscBool flg)
1736: {
1737: PetscFunctionBegin;
1740: ksp->guess_zero = (PetscBool) !(int)flg;
1741: PetscFunctionReturn(PETSC_SUCCESS);
1742: }
1744: /*@
1745: KSPGetInitialGuessNonzero - Determines whether the `KSP` solver is using
1746: a zero initial guess.
1748: Not Collective
1750: Input Parameter:
1751: . ksp - iterative context obtained from `KSPCreate()`
1753: Output Parameter:
1754: . flag - `PETSC_TRUE` if guess is nonzero, else `PETSC_FALSE`
1756: Level: intermediate
1758: .seealso: [](ch_ksp), `KSPSetInitialGuessNonzero()`, `KSP`
1759: @*/
1760: PetscErrorCode KSPGetInitialGuessNonzero(KSP ksp, PetscBool *flag)
1761: {
1762: PetscFunctionBegin;
1764: PetscAssertPointer(flag, 2);
1765: if (ksp->guess_zero) *flag = PETSC_FALSE;
1766: else *flag = PETSC_TRUE;
1767: PetscFunctionReturn(PETSC_SUCCESS);
1768: }
1770: /*@
1771: KSPSetErrorIfNotConverged - Causes `KSPSolve()` to generate an error if the solver has not converged as soon as the error is detected.
1773: Logically Collective
1775: Input Parameters:
1776: + ksp - iterative context obtained from `KSPCreate()`
1777: - flg - `PETSC_TRUE` indicates you want the error generated
1779: Options Database Key:
1780: . -ksp_error_if_not_converged <true,false> - generate an error and stop the program
1782: Level: intermediate
1784: Notes:
1785: Normally PETSc continues if a linear solver fails to converge, you can call `KSPGetConvergedReason()` after a `KSPSolve()`
1786: to determine if it has converged.
1788: A `KSP_DIVERGED_ITS` will not generate an error in a `KSPSolve()` inside a nested linear solver
1790: .seealso: [](ch_ksp), `KSPGetErrorIfNotConverged()`, `KSP`
1791: @*/
1792: PetscErrorCode KSPSetErrorIfNotConverged(KSP ksp, PetscBool flg)
1793: {
1794: PetscFunctionBegin;
1797: ksp->errorifnotconverged = flg;
1798: PetscFunctionReturn(PETSC_SUCCESS);
1799: }
1801: /*@
1802: KSPGetErrorIfNotConverged - Will `KSPSolve()` generate an error if the solver does not converge?
1804: Not Collective
1806: Input Parameter:
1807: . ksp - iterative context obtained from KSPCreate()
1809: Output Parameter:
1810: . flag - `PETSC_TRUE` if it will generate an error, else `PETSC_FALSE`
1812: Level: intermediate
1814: .seealso: [](ch_ksp), `KSPSetErrorIfNotConverged()`, `KSP`
1815: @*/
1816: PetscErrorCode KSPGetErrorIfNotConverged(KSP ksp, PetscBool *flag)
1817: {
1818: PetscFunctionBegin;
1820: PetscAssertPointer(flag, 2);
1821: *flag = ksp->errorifnotconverged;
1822: PetscFunctionReturn(PETSC_SUCCESS);
1823: }
1825: /*@
1826: KSPSetInitialGuessKnoll - Tells the iterative solver to use `PCApply()` to compute the initial guess (The Knoll trick)
1828: Logically Collective
1830: Input Parameters:
1831: + ksp - iterative context obtained from `KSPCreate()`
1832: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1834: Level: advanced
1836: Developer Notes:
1837: The Knoll trick is not currently implemented using the `KSPGuess` class
1839: .seealso: [](ch_ksp), `KSPGetInitialGuessKnoll()`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1840: @*/
1841: PetscErrorCode KSPSetInitialGuessKnoll(KSP ksp, PetscBool flg)
1842: {
1843: PetscFunctionBegin;
1846: ksp->guess_knoll = flg;
1847: PetscFunctionReturn(PETSC_SUCCESS);
1848: }
1850: /*@
1851: KSPGetInitialGuessKnoll - Determines whether the `KSP` solver is using the Knoll trick (using PCApply(pc,b,...) to compute
1852: the initial guess
1854: Not Collective
1856: Input Parameter:
1857: . ksp - iterative context obtained from `KSPCreate()`
1859: Output Parameter:
1860: . flag - `PETSC_TRUE` if using Knoll trick, else `PETSC_FALSE`
1862: Level: advanced
1864: .seealso: [](ch_ksp), `KSPSetInitialGuessKnoll()`, `KSPSetInitialGuessNonzero()`, `KSPGetInitialGuessNonzero()`, `KSP`
1865: @*/
1866: PetscErrorCode KSPGetInitialGuessKnoll(KSP ksp, PetscBool *flag)
1867: {
1868: PetscFunctionBegin;
1870: PetscAssertPointer(flag, 2);
1871: *flag = ksp->guess_knoll;
1872: PetscFunctionReturn(PETSC_SUCCESS);
1873: }
1875: /*@
1876: KSPGetComputeSingularValues - Gets the flag indicating whether the extreme singular
1877: values will be calculated via a Lanczos or Arnoldi process as the linear
1878: system is solved.
1880: Not Collective
1882: Input Parameter:
1883: . ksp - iterative context obtained from `KSPCreate()`
1885: Output Parameter:
1886: . flg - `PETSC_TRUE` or `PETSC_FALSE`
1888: Options Database Key:
1889: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
1891: Level: advanced
1893: Notes:
1894: Currently this option is not valid for all iterative methods.
1896: Many users may just want to use the monitoring routine
1897: `KSPMonitorSingularValue()` (which can be set with option -ksp_monitor_singular_value)
1898: to print the singular values at each iteration of the linear solve.
1900: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`
1901: @*/
1902: PetscErrorCode KSPGetComputeSingularValues(KSP ksp, PetscBool *flg)
1903: {
1904: PetscFunctionBegin;
1906: PetscAssertPointer(flg, 2);
1907: *flg = ksp->calc_sings;
1908: PetscFunctionReturn(PETSC_SUCCESS);
1909: }
1911: /*@
1912: KSPSetComputeSingularValues - Sets a flag so that the extreme singular
1913: values will be calculated via a Lanczos or Arnoldi process as the linear
1914: system is solved.
1916: Logically Collective
1918: Input Parameters:
1919: + ksp - iterative context obtained from `KSPCreate()`
1920: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1922: Options Database Key:
1923: . -ksp_monitor_singular_value - Activates `KSPSetComputeSingularValues()`
1925: Level: advanced
1927: Notes:
1928: Currently this option is not valid for all iterative methods.
1930: Many users may just want to use the monitoring routine
1931: `KSPMonitorSingularValue()` (which can be set with option -ksp_monitor_singular_value)
1932: to print the singular values at each iteration of the linear solve.
1934: .seealso: [](ch_ksp), `KSPComputeExtremeSingularValues()`, `KSPMonitorSingularValue()`, `KSP`
1935: @*/
1936: PetscErrorCode KSPSetComputeSingularValues(KSP ksp, PetscBool flg)
1937: {
1938: PetscFunctionBegin;
1941: ksp->calc_sings = flg;
1942: PetscFunctionReturn(PETSC_SUCCESS);
1943: }
1945: /*@
1946: KSPGetComputeEigenvalues - Gets the flag indicating that the extreme eigenvalues
1947: values will be calculated via a Lanczos or Arnoldi process as the linear
1948: system is solved.
1950: Not Collective
1952: Input Parameter:
1953: . ksp - iterative context obtained from `KSPCreate()`
1955: Output Parameter:
1956: . flg - `PETSC_TRUE` or `PETSC_FALSE`
1958: Level: advanced
1960: Note:
1961: Currently this option is not valid for all iterative methods.
1963: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`
1964: @*/
1965: PetscErrorCode KSPGetComputeEigenvalues(KSP ksp, PetscBool *flg)
1966: {
1967: PetscFunctionBegin;
1969: PetscAssertPointer(flg, 2);
1970: *flg = ksp->calc_sings;
1971: PetscFunctionReturn(PETSC_SUCCESS);
1972: }
1974: /*@
1975: KSPSetComputeEigenvalues - Sets a flag so that the extreme eigenvalues
1976: values will be calculated via a Lanczos or Arnoldi process as the linear
1977: system is solved.
1979: Logically Collective
1981: Input Parameters:
1982: + ksp - iterative context obtained from `KSPCreate()`
1983: - flg - `PETSC_TRUE` or `PETSC_FALSE`
1985: Level: advanced
1987: Note:
1988: Currently this option is not valid for all iterative methods.
1990: .seealso: [](ch_ksp), `KSPComputeEigenvalues()`, `KSPComputeEigenvaluesExplicitly()`, `KSP`
1991: @*/
1992: PetscErrorCode KSPSetComputeEigenvalues(KSP ksp, PetscBool flg)
1993: {
1994: PetscFunctionBegin;
1997: ksp->calc_sings = flg;
1998: PetscFunctionReturn(PETSC_SUCCESS);
1999: }
2001: /*@
2002: KSPSetComputeRitz - Sets a flag so that the Ritz or harmonic Ritz pairs
2003: will be calculated via a Lanczos or Arnoldi process as the linear
2004: system is solved.
2006: Logically Collective
2008: Input Parameters:
2009: + ksp - iterative context obtained from `KSPCreate()`
2010: - flg - `PETSC_TRUE` or `PETSC_FALSE`
2012: Level: advanced
2014: Note:
2015: Currently this option is only valid for the GMRES method.
2017: .seealso: [](ch_ksp), `KSPComputeRitz()`, `KSP`
2018: @*/
2019: PetscErrorCode KSPSetComputeRitz(KSP ksp, PetscBool flg)
2020: {
2021: PetscFunctionBegin;
2024: ksp->calc_ritz = flg;
2025: PetscFunctionReturn(PETSC_SUCCESS);
2026: }
2028: /*@
2029: KSPGetRhs - Gets the right-hand-side vector for the linear system to
2030: be solved.
2032: Not Collective
2034: Input Parameter:
2035: . ksp - iterative context obtained from `KSPCreate()`
2037: Output Parameter:
2038: . r - right-hand-side vector
2040: Level: developer
2042: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPSolve()`, `KSP`
2043: @*/
2044: PetscErrorCode KSPGetRhs(KSP ksp, Vec *r)
2045: {
2046: PetscFunctionBegin;
2048: PetscAssertPointer(r, 2);
2049: *r = ksp->vec_rhs;
2050: PetscFunctionReturn(PETSC_SUCCESS);
2051: }
2053: /*@
2054: KSPGetSolution - Gets the location of the solution for the
2055: linear system to be solved. Note that this may not be where the solution
2056: is stored during the iterative process; see `KSPBuildSolution()`.
2058: Not Collective
2060: Input Parameter:
2061: . ksp - iterative context obtained from `KSPCreate()`
2063: Output Parameter:
2064: . v - solution vector
2066: Level: developer
2068: .seealso: [](ch_ksp), `KSPGetRhs()`, `KSPBuildSolution()`, `KSPSolve()`, `KSP`
2069: @*/
2070: PetscErrorCode KSPGetSolution(KSP ksp, Vec *v)
2071: {
2072: PetscFunctionBegin;
2074: PetscAssertPointer(v, 2);
2075: *v = ksp->vec_sol;
2076: PetscFunctionReturn(PETSC_SUCCESS);
2077: }
2079: /*@
2080: KSPSetPC - Sets the preconditioner to be used to calculate the
2081: application of the preconditioner on a vector.
2083: Collective
2085: Input Parameters:
2086: + ksp - iterative context obtained from `KSPCreate()`
2087: - pc - the preconditioner object (can be `NULL`)
2089: Level: developer
2091: Note:
2092: Use `KSPGetPC()` to retrieve the preconditioner context.
2094: .seealso: [](ch_ksp), `KSPGetPC()`, `KSP`
2095: @*/
2096: PetscErrorCode KSPSetPC(KSP ksp, PC pc)
2097: {
2098: PetscFunctionBegin;
2100: if (pc) {
2102: PetscCheckSameComm(ksp, 1, pc, 2);
2103: }
2104: if (ksp->pc != pc && ksp->setupstage) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2105: PetscCall(PetscObjectReference((PetscObject)pc));
2106: PetscCall(PCDestroy(&ksp->pc));
2107: ksp->pc = pc;
2108: PetscFunctionReturn(PETSC_SUCCESS);
2109: }
2111: PETSC_INTERN PetscErrorCode PCCreate_MPI(PC);
2113: // PetscClangLinter pragma disable: -fdoc-internal-linkage
2114: /*@C
2115: KSPCheckPCMPI - Checks if `-mpi_linear_solver_server` is active and the `PC` should be changed to `PCMPI`
2117: Collective
2119: Input Parameter:
2120: . ksp - iterative context obtained from `KSPCreate()`
2122: Level: developer
2124: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`, `PCMPIServerBegin()`, `PCMPIServerEnd()`
2125: @*/
2126: PETSC_INTERN PetscErrorCode KSPCheckPCMPI(KSP ksp)
2127: {
2128: PetscBool isPCMPI;
2130: PetscFunctionBegin;
2132: PetscCall(PetscObjectTypeCompare((PetscObject)ksp->pc, PCMPI, &isPCMPI));
2133: if (PCMPIServerActive && ksp->nestlevel == 0 && !isPCMPI) {
2134: const char *prefix;
2135: char *found = NULL;
2137: PetscCall(KSPGetOptionsPrefix(ksp, &prefix));
2138: if (prefix) PetscCall(PetscStrstr(prefix, "mpi_linear_solver_server_", &found));
2139: if (!found) PetscCall(KSPAppendOptionsPrefix(ksp, "mpi_linear_solver_server_"));
2140: PetscCall(PCSetType(ksp->pc, PCMPI));
2141: }
2142: PetscFunctionReturn(PETSC_SUCCESS);
2143: }
2145: /*@
2146: KSPGetPC - Returns a pointer to the preconditioner context
2147: set with `KSPSetPC()`.
2149: Not Collective
2151: Input Parameter:
2152: . ksp - iterative context obtained from `KSPCreate()`
2154: Output Parameter:
2155: . pc - preconditioner context
2157: Level: developer
2159: .seealso: [](ch_ksp), `KSPSetPC()`, `KSP`
2160: @*/
2161: PetscErrorCode KSPGetPC(KSP ksp, PC *pc)
2162: {
2163: PetscFunctionBegin;
2165: PetscAssertPointer(pc, 2);
2166: if (!ksp->pc) {
2167: PetscCall(PCCreate(PetscObjectComm((PetscObject)ksp), &ksp->pc));
2168: PetscCall(PetscObjectIncrementTabLevel((PetscObject)ksp->pc, (PetscObject)ksp, 0));
2169: PetscCall(PetscObjectSetOptions((PetscObject)ksp->pc, ((PetscObject)ksp)->options));
2170: PetscCall(PCSetKSPNestLevel(ksp->pc, ksp->nestlevel));
2171: }
2172: PetscCall(KSPCheckPCMPI(ksp));
2173: *pc = ksp->pc;
2174: PetscFunctionReturn(PETSC_SUCCESS);
2175: }
2177: /*@
2178: KSPMonitor - runs the user provided monitor routines, if they exist
2180: Collective
2182: Input Parameters:
2183: + ksp - iterative context obtained from `KSPCreate()`
2184: . it - iteration number
2185: - rnorm - relative norm of the residual
2187: Level: developer
2189: Notes:
2190: This routine is called by the `KSP` implementations.
2191: It does not typically need to be called by the user.
2193: .seealso: [](ch_ksp), `KSPMonitorSet()`
2194: @*/
2195: PetscErrorCode KSPMonitor(KSP ksp, PetscInt it, PetscReal rnorm)
2196: {
2197: PetscInt i, n = ksp->numbermonitors;
2199: PetscFunctionBegin;
2200: for (i = 0; i < n; i++) PetscCall((*ksp->monitor[i])(ksp, it, rnorm, ksp->monitorcontext[i]));
2201: PetscFunctionReturn(PETSC_SUCCESS);
2202: }
2204: /*@C
2205: KSPMonitorSet - Sets an ADDITIONAL function to be called at every iteration to monitor
2206: the residual/error etc.
2208: Logically Collective
2210: Input Parameters:
2211: + ksp - iterative context obtained from `KSPCreate()`
2212: . monitor - pointer to function (if this is `NULL`, it turns off monitoring
2213: . ctx - [optional] context for private data for the monitor routine (use `NULL` if no context is needed)
2214: - monitordestroy - [optional] routine that frees monitor context (may be `NULL`)
2216: Calling sequence of `monitor`:
2217: + ksp - iterative context obtained from `KSPCreate()`
2218: . it - iteration number
2219: . rnorm - (estimated) 2-norm of (preconditioned) residual
2220: - ctx - optional monitoring context, as set by `KSPMonitorSet()`
2222: Calling sequence of `monitordestroy`:
2223: . ctx - optional monitoring context, as set by `KSPMonitorSet()`
2225: Options Database Keys:
2226: + -ksp_monitor - sets `KSPMonitorResidual()`
2227: . -ksp_monitor draw - sets `KSPMonitorResidualDraw()` and plots residual
2228: . -ksp_monitor draw::draw_lg - sets `KSPMonitorResidualDrawLG()` and plots residual
2229: . -ksp_monitor_pause_final - Pauses any graphics when the solve finishes (only works for internal monitors)
2230: . -ksp_monitor_true_residual - sets `KSPMonitorTrueResidual()`
2231: . -ksp_monitor_true_residual draw::draw_lg - sets `KSPMonitorTrueResidualDrawLG()` and plots residual
2232: . -ksp_monitor_max - sets `KSPMonitorTrueResidualMax()`
2233: . -ksp_monitor_singular_value - sets `KSPMonitorSingularValue()`
2234: - -ksp_monitor_cancel - cancels all monitors that have
2235: been hardwired into a code by
2236: calls to `KSPMonitorSet()`, but
2237: does not cancel those set via
2238: the options database.
2240: Level: beginner
2242: Notes:
2243: The default is to do nothing. To print the residual, or preconditioned
2244: residual if `KSPSetNormType`(ksp,`KSP_NORM_PRECONDITIONED`) was called, use
2245: `KSPMonitorResidual()` as the monitoring routine, with a `PETSCVIEWERASCII` as the
2246: context.
2248: Several different monitoring routines may be set by calling
2249: `KSPMonitorSet()` multiple times; all will be called in the
2250: order in which they were set.
2252: Fortran Notes:
2253: Only a single monitor function can be set for each `KSP` object
2255: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorCancel()`, `KSP`
2256: @*/
2257: PetscErrorCode KSPMonitorSet(KSP ksp, PetscErrorCode (*monitor)(KSP ksp, PetscInt it, PetscReal rnorm, void *ctx), void *ctx, PetscErrorCode (*monitordestroy)(void **ctx))
2258: {
2259: PetscInt i;
2260: PetscBool identical;
2262: PetscFunctionBegin;
2264: for (i = 0; i < ksp->numbermonitors; i++) {
2265: PetscCall(PetscMonitorCompare((PetscErrorCode(*)(void))monitor, ctx, monitordestroy, (PetscErrorCode(*)(void))ksp->monitor[i], ksp->monitorcontext[i], ksp->monitordestroy[i], &identical));
2266: if (identical) PetscFunctionReturn(PETSC_SUCCESS);
2267: }
2268: PetscCheck(ksp->numbermonitors < MAXKSPMONITORS, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "Too many KSP monitors set");
2269: ksp->monitor[ksp->numbermonitors] = monitor;
2270: ksp->monitordestroy[ksp->numbermonitors] = monitordestroy;
2271: ksp->monitorcontext[ksp->numbermonitors++] = (void *)ctx;
2272: PetscFunctionReturn(PETSC_SUCCESS);
2273: }
2275: /*@
2276: KSPMonitorCancel - Clears all monitors for a `KSP` object.
2278: Logically Collective
2280: Input Parameter:
2281: . ksp - iterative context obtained from `KSPCreate()`
2283: Options Database Key:
2284: . -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.
2286: Level: intermediate
2288: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSPMonitorSet()`, `KSP`
2289: @*/
2290: PetscErrorCode KSPMonitorCancel(KSP ksp)
2291: {
2292: PetscInt i;
2294: PetscFunctionBegin;
2296: for (i = 0; i < ksp->numbermonitors; i++) {
2297: if (ksp->monitordestroy[i]) PetscCall((*ksp->monitordestroy[i])(&ksp->monitorcontext[i]));
2298: }
2299: ksp->numbermonitors = 0;
2300: PetscFunctionReturn(PETSC_SUCCESS);
2301: }
2303: /*@C
2304: KSPGetMonitorContext - Gets the monitoring context, as set by `KSPMonitorSet()` for the FIRST monitor only.
2306: Not Collective
2308: Input Parameter:
2309: . ksp - iterative context obtained from `KSPCreate()`
2311: Output Parameter:
2312: . ctx - monitoring context
2314: Level: intermediate
2316: .seealso: [](ch_ksp), `KSPMonitorResidual()`, `KSP`
2317: @*/
2318: PetscErrorCode KSPGetMonitorContext(KSP ksp, void *ctx)
2319: {
2320: PetscFunctionBegin;
2322: *(void **)ctx = ksp->monitorcontext[0];
2323: PetscFunctionReturn(PETSC_SUCCESS);
2324: }
2326: /*@
2327: KSPSetResidualHistory - Sets the array used to hold the residual history.
2328: If set, this array will contain the residual norms computed at each
2329: iteration of the solver.
2331: Not Collective
2333: Input Parameters:
2334: + ksp - iterative context obtained from `KSPCreate()`
2335: . a - array to hold history
2336: . na - size of a
2337: - reset - `PETSC_TRUE` indicates the history counter is reset to zero
2338: for each new linear solve
2340: Level: advanced
2342: Notes:
2343: If provided, he array is NOT freed by PETSc so the user needs to keep track of it and destroy once the `KSP` object is destroyed.
2344: If 'a' is `NULL` then space is allocated for the history. If 'na' `PETSC_DECIDE` or `PETSC_DEFAULT` then a
2345: default array of length 10000 is allocated.
2347: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2349: .seealso: [](ch_ksp), `KSPGetResidualHistory()`, `KSP`
2350: @*/
2351: PetscErrorCode KSPSetResidualHistory(KSP ksp, PetscReal a[], PetscInt na, PetscBool reset)
2352: {
2353: PetscFunctionBegin;
2356: PetscCall(PetscFree(ksp->res_hist_alloc));
2357: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2358: ksp->res_hist = a;
2359: ksp->res_hist_max = (size_t)na;
2360: } else {
2361: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->res_hist_max = (size_t)na;
2362: else ksp->res_hist_max = 10000; /* like default ksp->max_it */
2363: PetscCall(PetscCalloc1(ksp->res_hist_max, &ksp->res_hist_alloc));
2365: ksp->res_hist = ksp->res_hist_alloc;
2366: }
2367: ksp->res_hist_len = 0;
2368: ksp->res_hist_reset = reset;
2369: PetscFunctionReturn(PETSC_SUCCESS);
2370: }
2372: /*@C
2373: KSPGetResidualHistory - Gets the array used to hold the residual history and the number of residuals it contains.
2375: Not Collective
2377: Input Parameter:
2378: . ksp - iterative context obtained from `KSPCreate()`
2380: Output Parameters:
2381: + a - pointer to array to hold history (or `NULL`)
2382: - na - number of used entries in a (or `NULL`)
2384: Level: advanced
2386: Note:
2387: This array is borrowed and should not be freed by the caller.
2389: Can only be called after a `KSPSetResidualHistory()` otherwise `a` and `na` are set to `NULL` and zero
2391: Fortran Notes:
2392: The Fortran version of this routine has a calling sequence
2393: $ call KSPGetResidualHistory(KSP ksp, integer na, integer ierr)
2394: note that you have passed a Fortran array into `KSPSetResidualHistory()` and you need
2395: to access the residual values from this Fortran array you provided. Only the `na` (number of
2396: residual norms currently held) is set.
2398: .seealso: [](ch_ksp), `KSPSetResidualHistory()`, `KSP`
2399: @*/
2400: PetscErrorCode KSPGetResidualHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2401: {
2402: PetscFunctionBegin;
2404: if (a) *a = ksp->res_hist;
2405: if (na) *na = (PetscInt)ksp->res_hist_len;
2406: PetscFunctionReturn(PETSC_SUCCESS);
2407: }
2409: /*@
2410: 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.
2412: Not Collective
2414: Input Parameters:
2415: + ksp - iterative context obtained from `KSPCreate()`
2416: . a - array to hold history
2417: . na - size of `a`
2418: - reset - `PETSC_TRUE` indicates the history counter is reset to zero for each new linear solve
2420: Level: advanced
2422: Notes:
2423: If provided, the array is NOT freed by PETSc so the user needs to keep track of it and destroy once the `KSP` object is destroyed.
2424: 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.
2426: If the array is not long enough then once the iterations is longer than the array length `KSPSolve()` stops recording the history
2428: .seealso: [](ch_ksp), `KSPGetErrorHistory()`, `KSPSetResidualHistory()`, `KSP`
2429: @*/
2430: PetscErrorCode KSPSetErrorHistory(KSP ksp, PetscReal a[], PetscInt na, PetscBool reset)
2431: {
2432: PetscFunctionBegin;
2435: PetscCall(PetscFree(ksp->err_hist_alloc));
2436: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2437: ksp->err_hist = a;
2438: ksp->err_hist_max = (size_t)na;
2439: } else {
2440: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->err_hist_max = (size_t)na;
2441: else ksp->err_hist_max = 10000; /* like default ksp->max_it */
2442: PetscCall(PetscCalloc1(ksp->err_hist_max, &ksp->err_hist_alloc));
2444: ksp->err_hist = ksp->err_hist_alloc;
2445: }
2446: ksp->err_hist_len = 0;
2447: ksp->err_hist_reset = reset;
2448: PetscFunctionReturn(PETSC_SUCCESS);
2449: }
2451: /*@C
2452: KSPGetErrorHistory - Gets the array used to hold the error history and the number of residuals it contains.
2454: Not Collective
2456: Input Parameter:
2457: . ksp - iterative context obtained from `KSPCreate()`
2459: Output Parameters:
2460: + a - pointer to array to hold history (or `NULL`)
2461: - na - number of used entries in a (or `NULL`)
2463: Level: advanced
2465: Notes:
2466: This array is borrowed and should not be freed by the caller.
2467: Can only be called after a `KSPSetErrorHistory()` otherwise `a` and `na` are set to `NULL` and zero
2469: Fortran Notes:
2470: The Fortran version of this routine has a calling sequence
2471: $ call KSPGetErrorHistory(KSP ksp, integer na, integer ierr)
2472: note that you have passed a Fortran array into `KSPSetErrorHistory()` and you need
2473: to access the residual values from this Fortran array you provided. Only the `na` (number of
2474: residual norms currently held) is set.
2476: .seealso: [](ch_ksp), `KSPSetErrorHistory()`, `KSPGetResidualHistory()`, `KSP`
2477: @*/
2478: PetscErrorCode KSPGetErrorHistory(KSP ksp, const PetscReal *a[], PetscInt *na)
2479: {
2480: PetscFunctionBegin;
2482: if (a) *a = ksp->err_hist;
2483: if (na) *na = (PetscInt)ksp->err_hist_len;
2484: PetscFunctionReturn(PETSC_SUCCESS);
2485: }
2487: /*@
2488: KSPComputeConvergenceRate - Compute the convergence rate for the iteration
2490: Not collective
2492: Input Parameter:
2493: . ksp - The `KSP`
2495: Output Parameters:
2496: + cr - The residual contraction rate
2497: . rRsq - The coefficient of determination, R^2, indicating the linearity of the data
2498: . ce - The error contraction rate
2499: - eRsq - The coefficient of determination, R^2, indicating the linearity of the data
2501: Level: advanced
2503: Note:
2504: 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,
2505: the slope is $\log c$. The coefficient of determination is given by $1 - \frac{\sum_i (y_i - f(x_i))^2}{\sum_i (y_i - \bar y)}$,
2507: References:
2508: . * - `//en.wikipedia.org/wiki/Coefficient_of_determination`
2510: .seealso: [](ch_ksp), `KSP`, `KSPConvergedRateView()`
2511: @*/
2512: PetscErrorCode KSPComputeConvergenceRate(KSP ksp, PetscReal *cr, PetscReal *rRsq, PetscReal *ce, PetscReal *eRsq)
2513: {
2514: PetscReal const *hist;
2515: PetscReal *x, *y, slope, intercept, mean = 0.0, var = 0.0, res = 0.0;
2516: PetscInt n, k;
2518: PetscFunctionBegin;
2519: if (cr || rRsq) {
2520: PetscCall(KSPGetResidualHistory(ksp, &hist, &n));
2521: if (!n) {
2522: if (cr) *cr = 0.0;
2523: if (rRsq) *rRsq = -1.0;
2524: } else {
2525: PetscCall(PetscMalloc2(n, &x, n, &y));
2526: for (k = 0; k < n; ++k) {
2527: x[k] = k;
2528: y[k] = PetscLogReal(hist[k]);
2529: mean += y[k];
2530: }
2531: mean /= n;
2532: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2533: for (k = 0; k < n; ++k) {
2534: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2535: var += PetscSqr(y[k] - mean);
2536: }
2537: PetscCall(PetscFree2(x, y));
2538: if (cr) *cr = PetscExpReal(slope);
2539: if (rRsq) *rRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2540: }
2541: }
2542: if (ce || eRsq) {
2543: PetscCall(KSPGetErrorHistory(ksp, &hist, &n));
2544: if (!n) {
2545: if (ce) *ce = 0.0;
2546: if (eRsq) *eRsq = -1.0;
2547: } else {
2548: PetscCall(PetscMalloc2(n, &x, n, &y));
2549: for (k = 0; k < n; ++k) {
2550: x[k] = k;
2551: y[k] = PetscLogReal(hist[k]);
2552: mean += y[k];
2553: }
2554: mean /= n;
2555: PetscCall(PetscLinearRegression(n, x, y, &slope, &intercept));
2556: for (k = 0; k < n; ++k) {
2557: res += PetscSqr(y[k] - (slope * x[k] + intercept));
2558: var += PetscSqr(y[k] - mean);
2559: }
2560: PetscCall(PetscFree2(x, y));
2561: if (ce) *ce = PetscExpReal(slope);
2562: if (eRsq) *eRsq = var < PETSC_MACHINE_EPSILON ? 0.0 : 1.0 - (res / var);
2563: }
2564: }
2565: PetscFunctionReturn(PETSC_SUCCESS);
2566: }
2568: /*@C
2569: KSPSetConvergenceTest - Sets the function to be used to determine convergence.
2571: Logically Collective
2573: Input Parameters:
2574: + ksp - iterative context obtained from `KSPCreate()`
2575: . converge - pointer to the function
2576: . ctx - context for private data for the convergence routine (may be null)
2577: - destroy - a routine for destroying the context (may be null)
2579: Calling sequence of `converge`:
2580: + ksp - iterative context obtained from `KSPCreate()`
2581: . it - iteration number
2582: . rnorm - (estimated) 2-norm of (preconditioned) residual
2583: . reason - the reason why it has converged or diverged
2584: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2586: Calling sequence of `destroy`:
2587: . ctx - the context
2589: Level: advanced
2591: Notes:
2592: Must be called after the `KSP` type has been set so put this after
2593: a call to `KSPSetType()`, or `KSPSetFromOptions()`.
2595: The default convergence test, `KSPConvergedDefault()`, aborts if the
2596: residual grows to more than 10000 times the initial residual.
2598: The default is a combination of relative and absolute tolerances.
2599: The residual value that is tested may be an approximation; routines
2600: that need exact values should compute them.
2602: In the default PETSc convergence test, the precise values of reason
2603: are macros such as `KSP_CONVERGED_RTOL`, which are defined in petscksp.h.
2605: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPGetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2606: @*/
2607: PetscErrorCode KSPSetConvergenceTest(KSP ksp, PetscErrorCode (*converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void *ctx, PetscErrorCode (*destroy)(void *ctx))
2608: {
2609: PetscFunctionBegin;
2611: if (ksp->convergeddestroy) PetscCall((*ksp->convergeddestroy)(ksp->cnvP));
2612: ksp->converged = converge;
2613: ksp->convergeddestroy = destroy;
2614: ksp->cnvP = (void *)ctx;
2615: PetscFunctionReturn(PETSC_SUCCESS);
2616: }
2618: /*@C
2619: KSPGetConvergenceTest - Gets the function to be used to determine convergence.
2621: Logically Collective
2623: Input Parameter:
2624: . ksp - iterative context obtained from `KSPCreate()`
2626: Output Parameters:
2627: + converge - pointer to convergence test function
2628: . ctx - context for private data for the convergence routine (may be null)
2629: - destroy - a routine for destroying the context (may be null)
2631: Calling sequence of `converge`:
2632: + ksp - iterative context obtained from `KSPCreate()`
2633: . it - iteration number
2634: . rnorm - (estimated) 2-norm of (preconditioned) residual
2635: . reason - the reason why it has converged or diverged
2636: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2638: Calling sequence of `destroy`:
2639: . ctx - the convergence test context
2641: Level: advanced
2643: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetAndClearConvergenceTest()`
2644: @*/
2645: PetscErrorCode KSPGetConvergenceTest(KSP ksp, PetscErrorCode (**converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void **ctx, PetscErrorCode (**destroy)(void *ctx))
2646: {
2647: PetscFunctionBegin;
2649: if (converge) *converge = ksp->converged;
2650: if (destroy) *destroy = ksp->convergeddestroy;
2651: if (ctx) *ctx = ksp->cnvP;
2652: PetscFunctionReturn(PETSC_SUCCESS);
2653: }
2655: /*@C
2656: KSPGetAndClearConvergenceTest - Gets the function to be used to determine convergence. Removes the current test without calling destroy on the test context
2658: Logically Collective
2660: Input Parameter:
2661: . ksp - iterative context obtained from `KSPCreate()`
2663: Output Parameters:
2664: + converge - pointer to convergence test function
2665: . ctx - context for private data for the convergence routine
2666: - destroy - a routine for destroying the context
2668: Calling sequence of `converge`:
2669: + ksp - iterative context obtained from `KSPCreate()`
2670: . it - iteration number
2671: . rnorm - (estimated) 2-norm of (preconditioned) residual
2672: . reason - the reason why it has converged or diverged
2673: - ctx - optional convergence context, as set by `KSPSetConvergenceTest()`
2675: Calling sequence of `destroy`:
2676: . ctx - the convergence test context
2678: Level: advanced
2680: Note:
2681: This is intended to be used to allow transferring the convergence test (and its context) to another testing object (for example another `KSP`)
2682: and then calling `KSPSetConvergenceTest()` on this original `KSP`. If you just called `KSPGetConvergenceTest()` followed
2683: by `KSPSetConvergenceTest()` the original context information
2684: would be destroyed and hence the transferred context would be invalid and trigger a crash on use
2686: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPGetConvergenceContext()`, `KSPSetTolerances()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2687: @*/
2688: PetscErrorCode KSPGetAndClearConvergenceTest(KSP ksp, PetscErrorCode (**converge)(KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason, void *ctx), void **ctx, PetscErrorCode (**destroy)(void *ctx))
2689: {
2690: PetscFunctionBegin;
2692: *converge = ksp->converged;
2693: *destroy = ksp->convergeddestroy;
2694: *ctx = ksp->cnvP;
2695: ksp->converged = NULL;
2696: ksp->cnvP = NULL;
2697: ksp->convergeddestroy = NULL;
2698: PetscFunctionReturn(PETSC_SUCCESS);
2699: }
2701: /*@C
2702: KSPGetConvergenceContext - Gets the convergence context set with `KSPSetConvergenceTest()`.
2704: Not Collective
2706: Input Parameter:
2707: . ksp - iterative context obtained from `KSPCreate()`
2709: Output Parameter:
2710: . ctx - monitoring context
2712: Level: advanced
2714: .seealso: [](ch_ksp), `KSP`, `KSPConvergedDefault()`, `KSPSetConvergenceTest()`, `KSPGetConvergenceTest()`
2715: @*/
2716: PetscErrorCode KSPGetConvergenceContext(KSP ksp, void *ctx)
2717: {
2718: PetscFunctionBegin;
2720: *(void **)ctx = ksp->cnvP;
2721: PetscFunctionReturn(PETSC_SUCCESS);
2722: }
2724: /*@C
2725: KSPBuildSolution - Builds the approximate solution in a vector provided.
2727: Collective
2729: Input Parameter:
2730: . ksp - iterative context obtained from `KSPCreate()`
2732: Output Parameter:
2733: Provide exactly one of
2734: + v - location to stash solution.
2735: - V - the solution is returned in this location. This vector is created
2736: internally. This vector should NOT be destroyed by the user with
2737: `VecDestroy()`.
2739: Level: developer
2741: Notes:
2742: This routine can be used in one of two ways
2743: .vb
2744: KSPBuildSolution(ksp,NULL,&V);
2745: or
2746: KSPBuildSolution(ksp,v,NULL); or KSPBuildSolution(ksp,v,&v);
2747: .ve
2748: In the first case an internal vector is allocated to store the solution
2749: (the user cannot destroy this vector). In the second case the solution
2750: is generated in the vector that the user provides. Note that for certain
2751: methods, such as `KSPCG`, the second case requires a copy of the solution,
2752: while in the first case the call is essentially free since it simply
2753: returns the vector where the solution already is stored. For some methods
2754: like `KSPGMRES` this is a reasonably expensive operation and should only be
2755: used in truly needed.
2757: .seealso: [](ch_ksp), `KSPGetSolution()`, `KSPBuildResidual()`, `KSP`
2758: @*/
2759: PetscErrorCode KSPBuildSolution(KSP ksp, Vec v, Vec *V)
2760: {
2761: PetscFunctionBegin;
2763: PetscCheck(V || v, PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_WRONG, "Must provide either v or V");
2764: if (!V) V = &v;
2765: PetscUseTypeMethod(ksp, buildsolution, v, V);
2766: PetscFunctionReturn(PETSC_SUCCESS);
2767: }
2769: /*@C
2770: KSPBuildResidual - Builds the residual in a vector provided.
2772: Collective
2774: Input Parameter:
2775: . ksp - iterative context obtained from `KSPCreate()`
2777: Output Parameters:
2778: + v - optional location to stash residual. If `v` is not provided,
2779: then a location is generated.
2780: . t - work vector. If not provided then one is generated.
2781: - V - the residual
2783: Level: advanced
2785: Note:
2786: Regardless of whether or not `v` is provided, the residual is
2787: returned in `V`.
2789: .seealso: [](ch_ksp), `KSP`, `KSPBuildSolution()`
2790: @*/
2791: PetscErrorCode KSPBuildResidual(KSP ksp, Vec t, Vec v, Vec *V)
2792: {
2793: PetscBool flag = PETSC_FALSE;
2794: Vec w = v, tt = t;
2796: PetscFunctionBegin;
2798: if (!w) PetscCall(VecDuplicate(ksp->vec_rhs, &w));
2799: if (!tt) {
2800: PetscCall(VecDuplicate(ksp->vec_sol, &tt));
2801: flag = PETSC_TRUE;
2802: }
2803: PetscUseTypeMethod(ksp, buildresidual, tt, w, V);
2804: if (flag) PetscCall(VecDestroy(&tt));
2805: PetscFunctionReturn(PETSC_SUCCESS);
2806: }
2808: /*@
2809: KSPSetDiagonalScale - Tells `KSP` to symmetrically diagonally scale the system
2810: before solving. This actually CHANGES the matrix (and right hand side).
2812: Logically Collective
2814: Input Parameters:
2815: + ksp - the `KSP` context
2816: - scale - `PETSC_TRUE` or `PETSC_FALSE`
2818: Options Database Keys:
2819: + -ksp_diagonal_scale - perform a diagonal scaling before the solve
2820: - -ksp_diagonal_scale_fix - scale the matrix back AFTER the solve
2822: Level: advanced
2824: Notes:
2825: Scales the matrix by D^(-1/2) A D^(-1/2) [D^(1/2) x ] = D^(-1/2) b
2826: where D_{ii} is 1/abs(A_{ii}) unless A_{ii} is zero and then it is 1.
2828: BE CAREFUL with this routine: it actually scales the matrix and right
2829: hand side that define the system. After the system is solved the matrix
2830: and right hand side remain scaled unless you use `KSPSetDiagonalScaleFix()`
2832: This should NOT be used within the `SNES` solves if you are using a line
2833: search.
2835: If you use this with the `PCType` `PCEISENSTAT` preconditioner than you can
2836: use the `PCEisenstatSetNoDiagonalScaling()` option, or -pc_eisenstat_no_diagonal_scaling
2837: to save some unneeded, redundant flops.
2839: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2840: @*/
2841: PetscErrorCode KSPSetDiagonalScale(KSP ksp, PetscBool scale)
2842: {
2843: PetscFunctionBegin;
2846: ksp->dscale = scale;
2847: PetscFunctionReturn(PETSC_SUCCESS);
2848: }
2850: /*@
2851: KSPGetDiagonalScale - Checks if `KSP` solver scales the matrix and right hand side, that is if `KSPSetDiagonalScale()` has been called
2853: Not Collective
2855: Input Parameter:
2856: . ksp - the `KSP` context
2858: Output Parameter:
2859: . scale - `PETSC_TRUE` or `PETSC_FALSE`
2861: Level: intermediate
2863: .seealso: [](ch_ksp), `KSP`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`
2864: @*/
2865: PetscErrorCode KSPGetDiagonalScale(KSP ksp, PetscBool *scale)
2866: {
2867: PetscFunctionBegin;
2869: PetscAssertPointer(scale, 2);
2870: *scale = ksp->dscale;
2871: PetscFunctionReturn(PETSC_SUCCESS);
2872: }
2874: /*@
2875: KSPSetDiagonalScaleFix - Tells `KSP` to diagonally scale the system back after solving.
2877: Logically Collective
2879: Input Parameters:
2880: + ksp - the `KSP` context
2881: - fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2882: rescale (default)
2884: Level: intermediate
2886: Notes:
2887: Must be called after `KSPSetDiagonalScale()`
2889: Using this will slow things down, because it rescales the matrix before and
2890: after each linear solve. This is intended mainly for testing to allow one
2891: to easily get back the original system to make sure the solution computed is
2892: accurate enough.
2894: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPGetDiagonalScaleFix()`, `KSP`
2895: @*/
2896: PetscErrorCode KSPSetDiagonalScaleFix(KSP ksp, PetscBool fix)
2897: {
2898: PetscFunctionBegin;
2901: ksp->dscalefix = fix;
2902: PetscFunctionReturn(PETSC_SUCCESS);
2903: }
2905: /*@
2906: KSPGetDiagonalScaleFix - Determines if `KSP` diagonally scales the system back after solving. That is `KSPSetDiagonalScaleFix()` has been called
2908: Not Collective
2910: Input Parameter:
2911: . ksp - the `KSP` context
2913: Output Parameter:
2914: . fix - `PETSC_TRUE` to scale back after the system solve, `PETSC_FALSE` to not
2915: rescale (default)
2917: Level: intermediate
2919: .seealso: [](ch_ksp), `KSPGetDiagonalScale()`, `KSPSetDiagonalScale()`, `KSPSetDiagonalScaleFix()`, `KSP`
2920: @*/
2921: PetscErrorCode KSPGetDiagonalScaleFix(KSP ksp, PetscBool *fix)
2922: {
2923: PetscFunctionBegin;
2925: PetscAssertPointer(fix, 2);
2926: *fix = ksp->dscalefix;
2927: PetscFunctionReturn(PETSC_SUCCESS);
2928: }
2930: /*@C
2931: KSPSetComputeOperators - set routine to compute the linear operators
2933: Logically Collective
2935: Input Parameters:
2936: + ksp - the `KSP` context
2937: . func - function to compute the operators
2938: - ctx - optional context
2940: Calling sequence of `func`:
2941: + ksp - the `KSP` context
2942: . A - the linear operator
2943: . B - the matrix from which the preconditioner is built, often `A`
2944: - ctx - optional user-provided context
2946: Level: beginner
2948: Notes:
2949: The user provided func() will be called automatically at the very next call to `KSPSolve()`. It will NOT be called at future `KSPSolve()` calls
2950: unless either `KSPSetComputeOperators()` or `KSPSetOperators()` is called before that `KSPSolve()` is called. This allows the same system to be solved several times
2951: with different right hand side functions but is a confusing API since one might expect it to be called for each `KSPSolve()`
2953: To reuse the same preconditioner for the next `KSPSolve()` and not compute a new one based on the most recently computed matrix call `KSPSetReusePreconditioner()`
2955: Developer Notes:
2956: Perhaps this routine and `KSPSetComputeRHS()` could be combined into a new API that makes clear when new matrices are computing without requiring call this
2957: routine to indicate when the new matrix should be computed.
2959: .seealso: [](ch_ksp), `KSP`, `KSPSetOperators()`, `KSPSetComputeRHS()`, `DMKSPSetComputeOperators()`, `KSPSetComputeInitialGuess()`
2960: @*/
2961: PetscErrorCode KSPSetComputeOperators(KSP ksp, PetscErrorCode (*func)(KSP ksp, Mat A, Mat B, void *ctx), void *ctx)
2962: {
2963: DM dm;
2965: PetscFunctionBegin;
2967: PetscCall(KSPGetDM(ksp, &dm));
2968: PetscCall(DMKSPSetComputeOperators(dm, func, ctx));
2969: if (ksp->setupstage == KSP_SETUP_NEWRHS) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2970: PetscFunctionReturn(PETSC_SUCCESS);
2971: }
2973: /*@C
2974: KSPSetComputeRHS - set routine to compute the right hand side of the linear system
2976: Logically Collective
2978: Input Parameters:
2979: + ksp - the `KSP` context
2980: . func - function to compute the right hand side
2981: - ctx - optional context
2983: Calling sequence of `func`:
2984: + ksp - the `KSP` context
2985: . b - right hand side of linear system
2986: - ctx - optional user-provided context
2988: Level: beginner
2990: Notes:
2991: The routine you provide will be called EACH you call `KSPSolve()` to prepare the new right hand side for that solve
2993: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `DMKSPSetComputeRHS()`, `KSPSetComputeOperators()`, `KSPSetOperators()`
2994: @*/
2995: PetscErrorCode KSPSetComputeRHS(KSP ksp, PetscErrorCode (*func)(KSP ksp, Vec b, void *ctx), void *ctx)
2996: {
2997: DM dm;
2999: PetscFunctionBegin;
3001: PetscCall(KSPGetDM(ksp, &dm));
3002: PetscCall(DMKSPSetComputeRHS(dm, func, ctx));
3003: PetscFunctionReturn(PETSC_SUCCESS);
3004: }
3006: /*@C
3007: KSPSetComputeInitialGuess - set routine to compute the initial guess of the linear system
3009: Logically Collective
3011: Input Parameters:
3012: + ksp - the `KSP` context
3013: . func - function to compute the initial guess
3014: - ctx - optional context
3016: Calling sequence of `func`:
3017: + ksp - the `KSP` context
3018: . x - solution vector
3019: - ctx - optional user-provided context
3021: Level: beginner
3023: Notes:
3024: This should only be used in conjunction with `KSPSetComputeRHS()` and `KSPSetComputeOperators()`, otherwise
3025: call `KSPSetInitialGuessNonzero()` and set the initial guess values in the solution vector passed to `KSPSolve()` before calling the solver
3027: .seealso: [](ch_ksp), `KSP`, `KSPSolve()`, `KSPSetComputeRHS()`, `KSPSetComputeOperators()`, `DMKSPSetComputeInitialGuess()`, `KSPSetInitialGuessNonzero()`
3028: @*/
3029: PetscErrorCode KSPSetComputeInitialGuess(KSP ksp, PetscErrorCode (*func)(KSP ksp, Vec x, void *ctx), void *ctx)
3030: {
3031: DM dm;
3033: PetscFunctionBegin;
3035: PetscCall(KSPGetDM(ksp, &dm));
3036: PetscCall(DMKSPSetComputeInitialGuess(dm, func, ctx));
3037: PetscFunctionReturn(PETSC_SUCCESS);
3038: }
3040: /*@
3041: KSPSetUseExplicitTranspose - Determines the explicit transpose of the operator is formed in `KSPSolveTranspose()`. In some configurations (like GPUs) it may
3042: be explicitly formed when possible since the solve is much more efficient.
3044: Logically Collective
3046: Input Parameter:
3047: . ksp - the `KSP` context
3049: Output Parameter:
3050: . flg - `PETSC_TRUE` to transpose the system in `KSPSolveTranspose()`, `PETSC_FALSE` to not transpose (default)
3052: Level: advanced
3054: .seealso: [](ch_ksp), `KSPSolveTranspose()`, `KSP`
3055: @*/
3056: PetscErrorCode KSPSetUseExplicitTranspose(KSP ksp, PetscBool flg)
3057: {
3058: PetscFunctionBegin;
3061: ksp->transpose.use_explicittranspose = flg;
3062: PetscFunctionReturn(PETSC_SUCCESS);
3063: }