Actual source code: itfunc.c
petsc-master 2020-08-25
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: PETSC_STATIC_INLINE PetscErrorCode ObjectView(PetscObject obj, PetscViewer viewer, PetscViewerFormat format)
10: {
13: PetscViewerPushFormat(viewer, format);
14: PetscObjectView(obj, viewer);
15: PetscViewerPopFormat(viewer);
16: return(0);
17: }
19: /*@
20: KSPComputeExtremeSingularValues - Computes the extreme singular values
21: for the preconditioned operator. Called after or during KSPSolve().
23: Not Collective
25: Input Parameter:
26: . ksp - iterative context obtained from KSPCreate()
28: Output Parameters:
29: . emin, emax - extreme singular values
31: Options Database Keys:
32: . -ksp_view_singularvalues - compute extreme singular values and print when KSPSolve completes.
34: Notes:
35: One must call KSPSetComputeSingularValues() before calling KSPSetUp()
36: (or use the option -ksp_view_eigenvalues) in order for this routine to work correctly.
38: Many users may just want to use the monitoring routine
39: KSPMonitorSingularValue() (which can be set with option -ksp_monitor_singular_value)
40: to print the extreme singular values at each iteration of the linear solve.
42: Estimates of the smallest singular value may be very inaccurate, especially if the Krylov method has not converged.
43: The largest singular value is usually accurate to within a few percent if the method has converged, but is still not
44: intended for eigenanalysis.
46: Disable restarts if using KSPGMRES, otherwise this estimate will only be using those iterations after the last
47: restart. See KSPGMRESSetRestart() for more details.
49: Level: advanced
51: .seealso: KSPSetComputeSingularValues(), KSPMonitorSingularValue(), KSPComputeEigenvalues(), KSP
52: @*/
53: PetscErrorCode KSPComputeExtremeSingularValues(KSP ksp,PetscReal *emax,PetscReal *emin)
54: {
61: if (!ksp->calc_sings) SETERRQ(PetscObjectComm((PetscObject)ksp),4,"Singular values not requested before KSPSetUp()");
63: if (ksp->ops->computeextremesingularvalues) {
64: (*ksp->ops->computeextremesingularvalues)(ksp,emax,emin);
65: } else {
66: *emin = -1.0;
67: *emax = -1.0;
68: }
69: return(0);
70: }
72: /*@
73: KSPComputeEigenvalues - Computes the extreme eigenvalues for the
74: preconditioned operator. Called after or during KSPSolve().
76: Not Collective
78: Input Parameters:
79: + ksp - iterative context obtained from KSPCreate()
80: - n - size of arrays r and c. The number of eigenvalues computed (neig) will, in
81: general, be less than this.
83: Output Parameters:
84: + r - real part of computed eigenvalues, provided by user with a dimension of at least n
85: . c - complex part of computed eigenvalues, provided by user with a dimension of at least n
86: - neig - actual number of eigenvalues computed (will be less than or equal to n)
88: Options Database Keys:
89: . -ksp_view_eigenvalues - Prints eigenvalues to stdout
91: Notes:
92: The number of eigenvalues estimated depends on the size of the Krylov space
93: generated during the KSPSolve() ; for example, with
94: CG it corresponds to the number of CG iterations, for GMRES it is the number
95: of GMRES iterations SINCE the last restart. Any extra space in r[] and c[]
96: will be ignored.
98: KSPComputeEigenvalues() does not usually provide accurate estimates; it is
99: intended only for assistance in understanding the convergence of iterative
100: methods, not for eigenanalysis. For accurate computation of eigenvalues we recommend using
101: the excellent package SLEPc.
103: One must call KSPSetComputeEigenvalues() before calling KSPSetUp()
104: in order for this routine to work correctly.
106: Many users may just want to use the monitoring routine
107: KSPMonitorSingularValue() (which can be set with option -ksp_monitor_singular_value)
108: to print the singular values at each iteration of the linear solve.
110: Level: advanced
112: .seealso: KSPSetComputeSingularValues(), KSPMonitorSingularValue(), KSPComputeExtremeSingularValues(), KSP
113: @*/
114: PetscErrorCode KSPComputeEigenvalues(KSP ksp,PetscInt n,PetscReal r[],PetscReal c[],PetscInt *neig)
115: {
122: if (n<0) SETERRQ(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Requested < 0 Eigenvalues");
124: if (!ksp->calc_sings) SETERRQ(PetscObjectComm((PetscObject)ksp),4,"Eigenvalues not requested before KSPSetUp()");
126: if (n && ksp->ops->computeeigenvalues) {
127: (*ksp->ops->computeeigenvalues)(ksp,n,r,c,neig);
128: } else {
129: *neig = 0;
130: }
131: return(0);
132: }
134: /*@
135: KSPComputeRitz - Computes the Ritz or harmonic Ritz pairs associated to the
136: smallest or largest in modulus, for the preconditioned operator.
137: Called after KSPSolve().
139: Not Collective
141: Input Parameters:
142: + ksp - iterative context obtained from KSPCreate()
143: . ritz - PETSC_TRUE or PETSC_FALSE for ritz pairs or harmonic Ritz pairs, respectively
144: . small - PETSC_TRUE or PETSC_FALSE for smallest or largest (harmonic) Ritz values, respectively
145: - nrit - number of (harmonic) Ritz pairs to compute
147: Output Parameters:
148: + nrit - actual number of computed (harmonic) Ritz pairs
149: . S - multidimensional vector with Ritz vectors
150: . tetar - real part of the Ritz values
151: - tetai - imaginary part of the Ritz values
153: Notes:
154: -For GMRES, the (harmonic) Ritz pairs are computed from the Hessenberg matrix obtained during
155: the last complete cycle, or obtained at the end of the solution if the method is stopped before
156: a restart. Then, the number of actual (harmonic) Ritz pairs computed is less or equal to the restart
157: parameter for GMRES if a complete cycle has been performed or less or equal to the number of GMRES
158: iterations.
159: -Moreover, for real matrices, the (harmonic) Ritz pairs are possibly complex-valued. In such a case,
160: the routine selects the complex (harmonic) Ritz value and its conjugate, and two successive columns of S
161: are equal to the real and the imaginary parts of the associated vectors.
162: -the (harmonic) Ritz pairs are given in order of increasing (harmonic) Ritz values in modulus
163: -this is currently not implemented when PETSc is built with complex numbers
165: One must call KSPSetComputeRitz() before calling KSPSetUp()
166: in order for this routine to work correctly.
168: Level: advanced
170: .seealso: KSPSetComputeRitz(), KSP
171: @*/
172: PetscErrorCode KSPComputeRitz(KSP ksp,PetscBool ritz,PetscBool small,PetscInt *nrit,Vec S[],PetscReal tetar[],PetscReal tetai[])
173: {
178: if (!ksp->calc_ritz) SETERRQ(PetscObjectComm((PetscObject)ksp),4,"Ritz pairs not requested before KSPSetUp()");
179: if (ksp->ops->computeritz) {(*ksp->ops->computeritz)(ksp,ritz,small,nrit,S,tetar,tetai);}
180: return(0);
181: }
182: /*@
183: KSPSetUpOnBlocks - Sets up the preconditioner for each block in
184: the block Jacobi, block Gauss-Seidel, and overlapping Schwarz
185: methods.
187: Collective on ksp
189: Input Parameter:
190: . ksp - the KSP context
192: Notes:
193: KSPSetUpOnBlocks() is a routine that the user can optinally call for
194: more precise profiling (via -log_view) of the setup phase for these
195: block preconditioners. If the user does not call KSPSetUpOnBlocks(),
196: it will automatically be called from within KSPSolve().
198: Calling KSPSetUpOnBlocks() is the same as calling PCSetUpOnBlocks()
199: on the PC context within the KSP context.
201: Level: advanced
203: .seealso: PCSetUpOnBlocks(), KSPSetUp(), PCSetUp(), KSP
204: @*/
205: PetscErrorCode KSPSetUpOnBlocks(KSP ksp)
206: {
207: PC pc;
209: PCFailedReason pcreason;
213: KSPGetPC(ksp,&pc);
214: PCSetUpOnBlocks(pc);
215: PCGetFailedReasonRank(pc,&pcreason);
216: /* TODO: this code was wrong and is still wrong, there is no way to propogate the failure to all processes; their is no code to handle a ksp->reason on only some ranks */
217: if (pcreason) {
218: ksp->reason = KSP_DIVERGED_PC_FAILED;
219: }
220: return(0);
221: }
223: /*@
224: KSPSetReusePreconditioner - reuse the current preconditioner, do not construct a new one even if the operator changes
226: Collective on ksp
228: Input Parameters:
229: + ksp - iterative context obtained from KSPCreate()
230: - flag - PETSC_TRUE to reuse the current preconditioner
232: Level: intermediate
234: .seealso: KSPCreate(), KSPSolve(), KSPDestroy(), PCSetReusePreconditioner(), KSP
235: @*/
236: PetscErrorCode KSPSetReusePreconditioner(KSP ksp,PetscBool flag)
237: {
238: PC pc;
243: KSPGetPC(ksp,&pc);
244: PCSetReusePreconditioner(pc,flag);
245: return(0);
246: }
248: /*@
249: KSPGetReusePreconditioner - Determines if the KSP reuses the current preconditioner even if the operator in the preconditioner has changed.
251: Collective on ksp
253: Input Parameters:
254: . ksp - iterative context obtained from KSPCreate()
256: Output Parameters:
257: . flag - the boolean flag
259: Level: intermediate
261: .seealso: KSPCreate(), KSPSolve(), KSPDestroy(), KSPSetReusePreconditioner(), KSP
262: @*/
263: PetscErrorCode KSPGetReusePreconditioner(KSP ksp,PetscBool *flag)
264: {
270: *flag = PETSC_FALSE;
271: if (ksp->pc) {
272: PCGetReusePreconditioner(ksp->pc,flag);
273: }
274: return(0);
275: }
277: /*@
278: KSPSetSkipPCSetFromOptions - prevents KSPSetFromOptions() from call PCSetFromOptions(). This is used if the same PC is shared by more than one KSP so its options are not resetable for each KSP
280: Collective on ksp
282: Input Parameters:
283: + ksp - iterative context obtained from KSPCreate()
284: - flag - PETSC_TRUE to skip calling the PCSetFromOptions()
286: Level: intermediate
288: .seealso: KSPCreate(), KSPSolve(), KSPDestroy(), PCSetReusePreconditioner(), KSP
289: @*/
290: PetscErrorCode KSPSetSkipPCSetFromOptions(KSP ksp,PetscBool flag)
291: {
294: ksp->skippcsetfromoptions = flag;
295: return(0);
296: }
298: /*@
299: KSPSetUp - Sets up the internal data structures for the
300: later use of an iterative solver.
302: Collective on ksp
304: Input Parameter:
305: . ksp - iterative context obtained from KSPCreate()
307: Level: developer
309: .seealso: KSPCreate(), KSPSolve(), KSPDestroy(), KSP
310: @*/
311: PetscErrorCode KSPSetUp(KSP ksp)
312: {
314: Mat A,B;
315: Mat mat,pmat;
316: MatNullSpace nullsp;
317: PCFailedReason pcreason;
322: /* reset the convergence flag from the previous solves */
323: ksp->reason = KSP_CONVERGED_ITERATING;
325: if (!((PetscObject)ksp)->type_name) {
326: KSPSetType(ksp,KSPGMRES);
327: }
328: KSPSetUpNorms_Private(ksp,PETSC_TRUE,&ksp->normtype,&ksp->pc_side);
330: if (ksp->dmActive && !ksp->setupstage) {
331: /* first time in so build matrix and vector data structures using DM */
332: if (!ksp->vec_rhs) {DMCreateGlobalVector(ksp->dm,&ksp->vec_rhs);}
333: if (!ksp->vec_sol) {DMCreateGlobalVector(ksp->dm,&ksp->vec_sol);}
334: DMCreateMatrix(ksp->dm,&A);
335: KSPSetOperators(ksp,A,A);
336: PetscObjectDereference((PetscObject)A);
337: }
339: if (ksp->dmActive) {
340: DMKSP kdm;
341: DMGetDMKSP(ksp->dm,&kdm);
343: if (kdm->ops->computeinitialguess && ksp->setupstage != KSP_SETUP_NEWRHS) {
344: /* only computes initial guess the first time through */
345: (*kdm->ops->computeinitialguess)(ksp,ksp->vec_sol,kdm->initialguessctx);
346: KSPSetInitialGuessNonzero(ksp,PETSC_TRUE);
347: }
348: if (kdm->ops->computerhs) {
349: (*kdm->ops->computerhs)(ksp,ksp->vec_rhs,kdm->rhsctx);
350: }
352: if (ksp->setupstage != KSP_SETUP_NEWRHS) {
353: if (kdm->ops->computeoperators) {
354: KSPGetOperators(ksp,&A,&B);
355: (*kdm->ops->computeoperators)(ksp,A,B,kdm->operatorsctx);
356: } else SETERRQ(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_WRONGSTATE,"You called KSPSetDM() but did not use DMKSPSetComputeOperators() or KSPSetDMActive(ksp,PETSC_FALSE);");
357: }
358: }
360: if (ksp->setupstage == KSP_SETUP_NEWRHS) return(0);
361: PetscLogEventBegin(KSP_SetUp,ksp,ksp->vec_rhs,ksp->vec_sol,0);
363: switch (ksp->setupstage) {
364: case KSP_SETUP_NEW:
365: (*ksp->ops->setup)(ksp);
366: break;
367: case KSP_SETUP_NEWMATRIX: { /* This should be replaced with a more general mechanism */
368: if (ksp->setupnewmatrix) {
369: (*ksp->ops->setup)(ksp);
370: }
371: } break;
372: default: break;
373: }
375: if (!ksp->pc) {KSPGetPC(ksp,&ksp->pc);}
376: PCGetOperators(ksp->pc,&mat,&pmat);
377: /* scale the matrix if requested */
378: if (ksp->dscale) {
379: PetscScalar *xx;
380: PetscInt i,n;
381: PetscBool zeroflag = PETSC_FALSE;
382: if (!ksp->pc) {KSPGetPC(ksp,&ksp->pc);}
383: if (!ksp->diagonal) { /* allocate vector to hold diagonal */
384: MatCreateVecs(pmat,&ksp->diagonal,NULL);
385: }
386: MatGetDiagonal(pmat,ksp->diagonal);
387: VecGetLocalSize(ksp->diagonal,&n);
388: VecGetArray(ksp->diagonal,&xx);
389: for (i=0; i<n; i++) {
390: if (xx[i] != 0.0) xx[i] = 1.0/PetscSqrtReal(PetscAbsScalar(xx[i]));
391: else {
392: xx[i] = 1.0;
393: zeroflag = PETSC_TRUE;
394: }
395: }
396: VecRestoreArray(ksp->diagonal,&xx);
397: if (zeroflag) {
398: PetscInfo(ksp,"Zero detected in diagonal of matrix, using 1 at those locations\n");
399: }
400: MatDiagonalScale(pmat,ksp->diagonal,ksp->diagonal);
401: if (mat != pmat) {MatDiagonalScale(mat,ksp->diagonal,ksp->diagonal);}
402: ksp->dscalefix2 = PETSC_FALSE;
403: }
404: PetscLogEventEnd(KSP_SetUp,ksp,ksp->vec_rhs,ksp->vec_sol,0);
405: PCSetErrorIfFailure(ksp->pc,ksp->errorifnotconverged);
406: PCSetUp(ksp->pc);
407: PCGetFailedReasonRank(ksp->pc,&pcreason);
408: /* TODO: this code was wrong and is still wrong, there is no way to propogate the failure to all processes; their is no code to handle a ksp->reason on only some ranks */
409: if (pcreason) {
410: ksp->reason = KSP_DIVERGED_PC_FAILED;
411: }
413: MatGetNullSpace(mat,&nullsp);
414: if (nullsp) {
415: PetscBool test = PETSC_FALSE;
416: PetscOptionsGetBool(((PetscObject)ksp)->options,((PetscObject)ksp)->prefix,"-ksp_test_null_space",&test,NULL);
417: if (test) {
418: MatNullSpaceTest(nullsp,mat,NULL);
419: }
420: }
421: ksp->setupstage = KSP_SETUP_NEWRHS;
422: return(0);
423: }
425: /*@
426: KSPConvergedReasonView - Displays the reason a KSP solve converged or diverged to a viewer
428: Collective on ksp
430: Parameter:
431: + ksp - iterative context obtained from KSPCreate()
432: - viewer - the viewer to display the reason
435: Options Database Keys:
436: + -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
437: - -ksp_converged_reason ::failed - only print reason and number of iterations when diverged
439: Level: beginner
441: .seealso: KSPCreate(), KSPSetUp(), KSPDestroy(), KSPSetTolerances(), KSPConvergedDefault(),
442: KSPSolveTranspose(), KSPGetIterationNumber(), KSP, KSPGetConvergedReason()
443: @*/
444: PetscErrorCode KSPConvergedReasonView(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
445: {
447: PetscBool isAscii;
450: if (!viewer) viewer = PETSC_VIEWER_STDOUT_(PetscObjectComm((PetscObject)ksp));
451: if (format != PETSC_VIEWER_DEFAULT) {PetscViewerPushFormat(viewer,format);}
452: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isAscii);
453: if (isAscii) {
454: PetscViewerASCIIAddTab(viewer,((PetscObject)ksp)->tablevel);
455: if (ksp->reason > 0 && format != PETSC_VIEWER_FAILED) {
456: if (((PetscObject) ksp)->prefix) {
457: PetscViewerASCIIPrintf(viewer,"Linear %s solve converged due to %s iterations %D\n",((PetscObject) ksp)->prefix,KSPConvergedReasons[ksp->reason],ksp->its);
458: } else {
459: PetscViewerASCIIPrintf(viewer,"Linear solve converged due to %s iterations %D\n",KSPConvergedReasons[ksp->reason],ksp->its);
460: }
461: } else if (ksp->reason <= 0) {
462: if (((PetscObject) ksp)->prefix) {
463: PetscViewerASCIIPrintf(viewer,"Linear %s solve did not converge due to %s iterations %D\n",((PetscObject) ksp)->prefix,KSPConvergedReasons[ksp->reason],ksp->its);
464: } else {
465: PetscViewerASCIIPrintf(viewer,"Linear solve did not converge due to %s iterations %D\n",KSPConvergedReasons[ksp->reason],ksp->its);
466: }
467: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
468: PCFailedReason reason;
469: PCGetFailedReason(ksp->pc,&reason);
470: PetscViewerASCIIPrintf(viewer," PC failed due to %s \n",PCFailedReasons[reason]);
471: }
472: }
473: PetscViewerASCIISubtractTab(viewer,((PetscObject)ksp)->tablevel);
474: }
475: if (format != PETSC_VIEWER_DEFAULT) {PetscViewerPopFormat(viewer);}
476: return(0);
477: }
479: /*@C
480: KSPConvergedReasonViewFromOptions - Processes command line options to determine if/how a KSPReason is to be viewed.
482: Collective on ksp
484: Input Parameters:
485: . ksp - the KSP object
487: Level: intermediate
489: @*/
490: PetscErrorCode KSPConvergedReasonViewFromOptions(KSP ksp)
491: {
492: PetscViewer viewer;
493: PetscBool flg;
494: PetscViewerFormat format;
495: PetscErrorCode ierr;
498: PetscOptionsGetViewer(PetscObjectComm((PetscObject)ksp),((PetscObject)ksp)->options,((PetscObject)ksp)->prefix,"-ksp_converged_reason",&viewer,&format,&flg);
499: if (flg) {
500: KSPConvergedReasonView(ksp, viewer, format);
501: PetscViewerDestroy(&viewer);
502: }
503: return(0);
504: }
506: #include <petscdraw.h>
508: static PetscErrorCode KSPViewEigenvalues_Internal(KSP ksp, PetscBool isExplicit, PetscViewer viewer, PetscViewerFormat format)
509: {
510: PetscReal *r, *c;
511: PetscInt n, i, neig;
512: PetscBool isascii, isdraw;
513: PetscMPIInt rank;
517: MPI_Comm_rank(PetscObjectComm((PetscObject) ksp), &rank);
518: PetscObjectTypeCompare((PetscObject) viewer, PETSCVIEWERASCII, &isascii);
519: PetscObjectTypeCompare((PetscObject) viewer, PETSCVIEWERDRAW, &isdraw);
520: if (isExplicit) {
521: VecGetSize(ksp->vec_sol,&n);
522: PetscMalloc2(n, &r, n, &c);
523: KSPComputeEigenvaluesExplicitly(ksp, n, r, c);
524: neig = n;
525: } else {
526: PetscInt nits;
528: KSPGetIterationNumber(ksp, &nits);
529: n = nits+2;
530: if (!nits) {PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any eigenvalues\n");return(0);}
531: PetscMalloc2(n, &r, n, &c);
532: KSPComputeEigenvalues(ksp, n, r, c, &neig);
533: }
534: if (isascii) {
535: PetscViewerASCIIPrintf(viewer, "%s computed eigenvalues\n", isExplicit ? "Explicitly" : "Iteratively");
536: for (i = 0; i < neig; ++i) {
537: if (c[i] >= 0.0) {PetscViewerASCIIPrintf(viewer, "%g + %gi\n", (double) r[i], (double) c[i]);}
538: else {PetscViewerASCIIPrintf(viewer, "%g - %gi\n", (double) r[i], -(double) c[i]);}
539: }
540: } else if (isdraw && !rank) {
541: PetscDraw draw;
542: PetscDrawSP drawsp;
544: if (format == PETSC_VIEWER_DRAW_CONTOUR) {
545: KSPPlotEigenContours_Private(ksp,neig,r,c);
546: } else {
547: if (!ksp->eigviewer) {PetscViewerDrawOpen(PETSC_COMM_SELF,NULL,isExplicit ? "Explicitly Computed Eigenvalues" : "Iteratively Computed Eigenvalues",PETSC_DECIDE,PETSC_DECIDE,400,400,&ksp->eigviewer);}
548: PetscViewerDrawGetDraw(ksp->eigviewer,0,&draw);
549: PetscDrawSPCreate(draw,1,&drawsp);
550: PetscDrawSPReset(drawsp);
551: for (i = 0; i < neig; ++i) {PetscDrawSPAddPoint(drawsp,r+i,c+i);}
552: PetscDrawSPDraw(drawsp,PETSC_TRUE);
553: PetscDrawSPSave(drawsp);
554: PetscDrawSPDestroy(&drawsp);
555: }
556: }
557: PetscFree2(r, c);
558: return(0);
559: }
561: static PetscErrorCode KSPViewSingularvalues_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
562: {
563: PetscReal smax, smin;
564: PetscInt nits;
565: PetscBool isascii;
569: PetscObjectTypeCompare((PetscObject) viewer, PETSCVIEWERASCII, &isascii);
570: KSPGetIterationNumber(ksp, &nits);
571: if (!nits) {PetscViewerASCIIPrintf(viewer, "Zero iterations in solver, cannot approximate any singular values\n");return(0);}
572: KSPComputeExtremeSingularValues(ksp, &smax, &smin);
573: if (isascii) {PetscViewerASCIIPrintf(viewer, "Iteratively computed extreme singular values: max %g min %g max/min %g\n",(double)smax,(double)smin,(double)(smax/smin));}
574: return(0);
575: }
577: static PetscErrorCode KSPViewFinalResidual_Internal(KSP ksp, PetscViewer viewer, PetscViewerFormat format)
578: {
579: PetscBool isascii;
583: PetscObjectTypeCompare((PetscObject) viewer, PETSCVIEWERASCII, &isascii);
584: if (ksp->dscale && !ksp->dscalefix) SETERRQ(PetscObjectComm((PetscObject) ksp), PETSC_ERR_ARG_WRONGSTATE, "Cannot compute final scale with -ksp_diagonal_scale except also with -ksp_diagonal_scale_fix");
585: if (isascii) {
586: Mat A;
587: Vec t;
588: PetscReal norm;
590: PCGetOperators(ksp->pc, &A, NULL);
591: VecDuplicate(ksp->vec_rhs, &t);
592: KSP_MatMult(ksp, A, ksp->vec_sol, t);
593: VecAYPX(t, -1.0, ksp->vec_rhs);
594: VecNorm(t, NORM_2, &norm);
595: VecDestroy(&t);
596: PetscViewerASCIIPrintf(viewer, "KSP final norm of residual %g\n", (double) norm);
597: }
598: return(0);
599: }
601: static PetscErrorCode KSPSolve_Private(KSP ksp,Vec b,Vec x)
602: {
604: PetscBool flg = PETSC_FALSE,inXisinB=PETSC_FALSE,guess_zero;
605: Mat mat,pmat;
606: MPI_Comm comm;
607: MatNullSpace nullsp;
608: Vec btmp,vec_rhs=NULL;
611: comm = PetscObjectComm((PetscObject)ksp);
612: if (x && x == b) {
613: if (!ksp->guess_zero) SETERRQ(comm,PETSC_ERR_ARG_INCOMP,"Cannot use x == b with nonzero initial guess");
614: VecDuplicate(b,&x);
615: inXisinB = PETSC_TRUE;
616: }
617: if (b) {
618: PetscObjectReference((PetscObject)b);
619: VecDestroy(&ksp->vec_rhs);
620: ksp->vec_rhs = b;
621: }
622: if (x) {
623: PetscObjectReference((PetscObject)x);
624: VecDestroy(&ksp->vec_sol);
625: ksp->vec_sol = x;
626: }
628: if (ksp->viewPre) {ObjectView((PetscObject) ksp, ksp->viewerPre, ksp->formatPre);}
630: if (ksp->presolve) {(*ksp->presolve)(ksp,ksp->vec_rhs,ksp->vec_sol,ksp->prectx);}
632: /* reset the residual history list if requested */
633: if (ksp->res_hist_reset) ksp->res_hist_len = 0;
635: PetscLogEventBegin(KSP_Solve,ksp,ksp->vec_rhs,ksp->vec_sol,0);
637: if (ksp->guess) {
638: PetscObjectState ostate,state;
640: KSPGuessSetUp(ksp->guess);
641: PetscObjectStateGet((PetscObject)ksp->vec_sol,&ostate);
642: KSPGuessFormGuess(ksp->guess,ksp->vec_rhs,ksp->vec_sol);
643: PetscObjectStateGet((PetscObject)ksp->vec_sol,&state);
644: if (state != ostate) {
645: ksp->guess_zero = PETSC_FALSE;
646: } else {
647: PetscInfo(ksp,"Using zero initial guess since the KSPGuess object did not change the vector\n");
648: ksp->guess_zero = PETSC_TRUE;
649: }
650: }
652: /* KSPSetUp() scales the matrix if needed */
653: KSPSetUp(ksp);
654: KSPSetUpOnBlocks(ksp);
656: VecSetErrorIfLocked(ksp->vec_sol,3);
658: PCGetOperators(ksp->pc,&mat,&pmat);
659: /* diagonal scale RHS if called for */
660: if (ksp->dscale) {
661: VecPointwiseMult(ksp->vec_rhs,ksp->vec_rhs,ksp->diagonal);
662: /* second time in, but matrix was scaled back to original */
663: if (ksp->dscalefix && ksp->dscalefix2) {
664: Mat mat,pmat;
666: PCGetOperators(ksp->pc,&mat,&pmat);
667: MatDiagonalScale(pmat,ksp->diagonal,ksp->diagonal);
668: if (mat != pmat) {MatDiagonalScale(mat,ksp->diagonal,ksp->diagonal);}
669: }
671: /* scale initial guess */
672: if (!ksp->guess_zero) {
673: if (!ksp->truediagonal) {
674: VecDuplicate(ksp->diagonal,&ksp->truediagonal);
675: VecCopy(ksp->diagonal,ksp->truediagonal);
676: VecReciprocal(ksp->truediagonal);
677: }
678: VecPointwiseMult(ksp->vec_sol,ksp->vec_sol,ksp->truediagonal);
679: }
680: }
681: PCPreSolve(ksp->pc,ksp);
683: if (ksp->guess_zero) { VecSet(ksp->vec_sol,0.0);}
684: if (ksp->guess_knoll) { /* The Knoll trick is independent on the KSPGuess specified */
685: PCApply(ksp->pc,ksp->vec_rhs,ksp->vec_sol);
686: KSP_RemoveNullSpace(ksp,ksp->vec_sol);
687: ksp->guess_zero = PETSC_FALSE;
688: }
690: /* can we mark the initial guess as zero for this solve? */
691: guess_zero = ksp->guess_zero;
692: if (!ksp->guess_zero) {
693: PetscReal norm;
695: VecNormAvailable(ksp->vec_sol,NORM_2,&flg,&norm);
696: if (flg && !norm) ksp->guess_zero = PETSC_TRUE;
697: }
698: if (ksp->transpose_solve) {
699: MatGetNullSpace(pmat,&nullsp);
700: } else {
701: MatGetTransposeNullSpace(pmat,&nullsp);
702: }
703: if (nullsp) {
704: VecDuplicate(ksp->vec_rhs,&btmp);
705: VecCopy(ksp->vec_rhs,btmp);
706: MatNullSpaceRemove(nullsp,btmp);
707: vec_rhs = ksp->vec_rhs;
708: ksp->vec_rhs = btmp;
709: }
710: VecLockReadPush(ksp->vec_rhs);
711: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
712: VecSetInf(ksp->vec_sol);
713: }
714: (*ksp->ops->solve)(ksp);
716: VecLockReadPop(ksp->vec_rhs);
717: if (nullsp) {
718: ksp->vec_rhs = vec_rhs;
719: VecDestroy(&btmp);
720: }
722: ksp->guess_zero = guess_zero;
724: if (!ksp->reason) SETERRQ(comm,PETSC_ERR_PLIB,"Internal error, solver returned without setting converged reason");
725: ksp->totalits += ksp->its;
727: if (ksp->viewReason) {KSPConvergedReasonView(ksp, ksp->viewerReason, ksp->formatReason);}
728: PCPostSolve(ksp->pc,ksp);
730: /* diagonal scale solution if called for */
731: if (ksp->dscale) {
732: VecPointwiseMult(ksp->vec_sol,ksp->vec_sol,ksp->diagonal);
733: /* unscale right hand side and matrix */
734: if (ksp->dscalefix) {
735: Mat mat,pmat;
737: VecReciprocal(ksp->diagonal);
738: VecPointwiseMult(ksp->vec_rhs,ksp->vec_rhs,ksp->diagonal);
739: PCGetOperators(ksp->pc,&mat,&pmat);
740: MatDiagonalScale(pmat,ksp->diagonal,ksp->diagonal);
741: if (mat != pmat) {MatDiagonalScale(mat,ksp->diagonal,ksp->diagonal);}
742: VecReciprocal(ksp->diagonal);
743: ksp->dscalefix2 = PETSC_TRUE;
744: }
745: }
746: PetscLogEventEnd(KSP_Solve,ksp,ksp->vec_rhs,ksp->vec_sol,0);
747: if (ksp->guess) {
748: KSPGuessUpdate(ksp->guess,ksp->vec_rhs,ksp->vec_sol);
749: }
750: if (ksp->postsolve) {
751: (*ksp->postsolve)(ksp,ksp->vec_rhs,ksp->vec_sol,ksp->postctx);
752: }
754: PCGetOperators(ksp->pc,&mat,&pmat);
755: if (ksp->viewEV) {KSPViewEigenvalues_Internal(ksp, PETSC_FALSE, ksp->viewerEV, ksp->formatEV);}
756: if (ksp->viewEVExp) {KSPViewEigenvalues_Internal(ksp, PETSC_TRUE, ksp->viewerEVExp, ksp->formatEVExp);}
757: if (ksp->viewSV) {KSPViewSingularvalues_Internal(ksp, ksp->viewerSV, ksp->formatSV);}
758: if (ksp->viewFinalRes) {KSPViewFinalResidual_Internal(ksp, ksp->viewerFinalRes, ksp->formatFinalRes);}
759: if (ksp->viewMat) {ObjectView((PetscObject) mat, ksp->viewerMat, ksp->formatMat);}
760: if (ksp->viewPMat) {ObjectView((PetscObject) pmat, ksp->viewerPMat, ksp->formatPMat);}
761: if (ksp->viewRhs) {ObjectView((PetscObject) ksp->vec_rhs, ksp->viewerRhs, ksp->formatRhs);}
762: if (ksp->viewSol) {ObjectView((PetscObject) ksp->vec_sol, ksp->viewerSol, ksp->formatSol);}
763: if (ksp->view) {ObjectView((PetscObject) ksp, ksp->viewer, ksp->format);}
764: if (ksp->viewDScale) {ObjectView((PetscObject) ksp->diagonal, ksp->viewerDScale, ksp->formatDScale);}
765: if (ksp->viewMatExp) {
766: Mat A, B;
768: PCGetOperators(ksp->pc, &A, NULL);
769: if (ksp->transpose_solve) {
770: Mat AT;
772: MatCreateTranspose(A, &AT);
773: MatComputeOperator(AT, MATAIJ, &B);
774: MatDestroy(&AT);
775: } else {
776: MatComputeOperator(A, MATAIJ, &B);
777: }
778: ObjectView((PetscObject) B, ksp->viewerMatExp, ksp->formatMatExp);
779: MatDestroy(&B);
780: }
781: if (ksp->viewPOpExp) {
782: Mat B;
784: KSPComputeOperator(ksp, MATAIJ, &B);
785: ObjectView((PetscObject) B, ksp->viewerPOpExp, ksp->formatPOpExp);
786: MatDestroy(&B);
787: }
789: if (inXisinB) {
790: VecCopy(x,b);
791: VecDestroy(&x);
792: }
793: PetscObjectSAWsBlock((PetscObject)ksp);
794: if (ksp->errorifnotconverged && ksp->reason < 0 && ksp->reason != KSP_DIVERGED_ITS) {
795: if (ksp->reason == KSP_DIVERGED_PC_FAILED) {
796: PCFailedReason reason;
797: PCGetFailedReason(ksp->pc,&reason);
798: SETERRQ2(comm,PETSC_ERR_NOT_CONVERGED,"KSPSolve has not converged, reason %s PC failed due to %s",KSPConvergedReasons[ksp->reason],PCFailedReasons[reason]);
799: } else SETERRQ1(comm,PETSC_ERR_NOT_CONVERGED,"KSPSolve has not converged, reason %s",KSPConvergedReasons[ksp->reason]);
800: }
801: return(0);
802: }
804: /*@
805: KSPSolve - Solves linear system.
807: Collective on ksp
809: Parameters:
810: + ksp - iterative context obtained from KSPCreate()
811: . b - the right hand side vector
812: - x - the solution (this may be the same vector as b, then b will be overwritten with answer)
814: Options Database Keys:
815: + -ksp_view_eigenvalues - compute preconditioned operators eigenvalues
816: . -ksp_view_eigenvalues_explicitly - compute the eigenvalues by forming the dense operator and using LAPACK
817: . -ksp_view_mat binary - save matrix to the default binary viewer
818: . -ksp_view_pmat binary - save matrix used to build preconditioner to the default binary viewer
819: . -ksp_view_rhs binary - save right hand side vector to the default binary viewer
820: . -ksp_view_solution binary - save computed solution vector to the default binary viewer
821: (can be read later with src/ksp/tutorials/ex10.c for testing solvers)
822: . -ksp_view_mat_explicit - for matrix-free operators, computes the matrix entries and views them
823: . -ksp_view_preconditioned_operator_explicit - computes the product of the preconditioner and matrix as an explicit matrix and views it
824: . -ksp_converged_reason - print reason for converged or diverged, also prints number of iterations
825: . -ksp_view_final_residual - print 2-norm of true linear system residual at the end of the solution process
826: - -ksp_view - print the ksp data structure at the end of the system solution
828: Notes:
830: If one uses KSPSetDM() then x or b need not be passed. Use KSPGetSolution() to access the solution in this case.
832: The operator is specified with KSPSetOperators().
834: Call KSPGetConvergedReason() to determine if the solver converged or failed and
835: why. The number of iterations can be obtained from KSPGetIterationNumber().
837: If you provide a matrix that has a MatSetNullSpace() and MatSetTransposeNullSpace() this will use that information to solve singular systems
838: in the least squares sense with a norm minimizing solution.
839: $
840: $ 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()
841: $
842: $ 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
843: $ 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
844: $ direction thus the solution which is a linear combination of the search directions has no component in the nullspace(A).
845: $
846: $ We recommend always using GMRES for such singular systems.
847: $ If nullspace(A) = nullspace(A') (note symmetric matrices always satisfy this property) then both left and right preconditioning will work
848: $ If nullspace(A) != nullspace(A') then left preconditioning will work but right preconditioning may not work (or it may).
850: Developer Note: The reason we cannot always solve nullspace(A) != nullspace(A') systems with right preconditioning is because we need to remove at each iteration
851: 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
852: such as diagonal scaling we cannot apply the inverse of the preconditioner to a vector and thus cannot compute the nullspace(AB).
855: If using a direct method (e.g., via the KSP solver
856: KSPPREONLY and a preconditioner such as PCLU/PCILU),
857: then its=1. See KSPSetTolerances() and KSPConvergedDefault()
858: for more details.
860: Understanding Convergence:
861: The routines KSPMonitorSet(), KSPComputeEigenvalues(), and
862: KSPComputeEigenvaluesExplicitly() provide information on additional
863: options to monitor convergence and print eigenvalue information.
865: Level: beginner
867: .seealso: KSPCreate(), KSPSetUp(), KSPDestroy(), KSPSetTolerances(), KSPConvergedDefault(),
868: KSPSolveTranspose(), KSPGetIterationNumber(), MatNullSpaceCreate(), MatSetNullSpace(), MatSetTransposeNullSpace(), KSP,
869: KSPConvergedReasonView()
870: @*/
871: PetscErrorCode KSPSolve(KSP ksp,Vec b,Vec x)
872: {
879: ksp->transpose_solve = PETSC_FALSE;
880: KSPSolve_Private(ksp,b,x);
881: return(0);
882: }
884: /*@
885: KSPSolveTranspose - Solves the transpose of a linear system.
887: Collective on ksp
889: Input Parameters:
890: + ksp - iterative context obtained from KSPCreate()
891: . b - right hand side vector
892: - x - solution vector
894: Notes:
895: For complex numbers this solve the non-Hermitian transpose system.
897: Developer Notes:
898: We need to implement a KSPSolveHermitianTranspose()
900: Level: developer
902: .seealso: KSPCreate(), KSPSetUp(), KSPDestroy(), KSPSetTolerances(), KSPConvergedDefault(),
903: KSPSolve(), KSP
904: @*/
905: PetscErrorCode KSPSolveTranspose(KSP ksp,Vec b,Vec x)
906: {
913: ksp->transpose_solve = PETSC_TRUE;
914: KSPSolve_Private(ksp,b,x);
915: return(0);
916: }
918: static PetscErrorCode KSPViewFinalMatResidual_Internal(KSP ksp, Mat B, Mat X, PetscViewer viewer, PetscViewerFormat format, PetscInt shift)
919: {
920: Mat A, R;
921: PetscReal *norms;
922: PetscInt i, N;
923: PetscBool flg;
927: PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &flg);
928: if (flg) {
929: PCGetOperators(ksp->pc, &A, NULL);
930: MatMatMult(A, X, MAT_INITIAL_MATRIX, PETSC_DEFAULT, &R);
931: MatAYPX(R, -1.0, B, SAME_NONZERO_PATTERN);
932: MatGetSize(R, NULL, &N);
933: PetscMalloc1(N, &norms);
934: MatGetColumnNorms(R, NORM_2, norms);
935: MatDestroy(&R);
936: for (i = 0; i < N; ++i) {
937: PetscViewerASCIIPrintf(viewer, "%s #%D %g\n", i == 0 ? "KSP final norm of residual" : " ", shift + i, (double)norms[i]);
938: }
939: PetscFree(norms);
940: }
941: return(0);
942: }
944: /*@
945: KSPMatSolve - Solves a linear system with multiple right-hand sides stored as a MATDENSE. Unlike KSPSolve(), B and X must be different matrices.
947: Input Parameters:
948: + ksp - iterative context
949: - B - block of right-hand sides
951: Output Parameter:
952: . X - block of solutions
954: Notes:
955: This is a stripped-down version of KSPSolve(), which only handles -ksp_view, -ksp_converged_reason, and -ksp_view_final_residual.
957: Level: intermediate
959: .seealso: KSPSolve(), MatMatSolve(), MATDENSE, KSPHPDDM, PCBJACOBI, PCASM
960: @*/
961: PetscErrorCode KSPMatSolve(KSP ksp, Mat B, Mat X)
962: {
963: Mat A, vB, vX;
964: Vec cb, cx;
965: PetscInt m1, M1, m2, M2, n1, N1, n2, N2, Bbn = PETSC_DECIDE;
966: PetscBool match;
975: if (!B->assembled) SETERRQ(PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
976: MatCheckPreallocated(X, 3);
977: if (!X->assembled) {
978: MatSetOption(X, MAT_NO_OFF_PROC_ENTRIES, PETSC_TRUE);
979: MatAssemblyBegin(X, MAT_FINAL_ASSEMBLY);
980: MatAssemblyEnd(X, MAT_FINAL_ASSEMBLY);
981: }
982: if (B == X) SETERRQ(PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_IDN, "B and X must be different matrices");
983: KSPGetOperators(ksp, &A, NULL);
984: MatGetLocalSize(A, &m1, NULL);
985: MatGetLocalSize(B, &m2, &n2);
986: MatGetSize(A, &M1, NULL);
987: MatGetSize(B, &M2, &N2);
988: if (m1 != m2 || M1 != M2) SETERRQ4(PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Cannot use a block of right-hand sides with (m2,M2) = (%D,%D) for a linear system with (m1,M1) = (%D,%D)", m2, M2, m1, M1);
989: MatGetLocalSize(X, &m1, &n1);
990: MatGetSize(X, &M1, &N1);
991: if (m1 != m2 || M1 != M2 || n1 != n2 || N1 != N2) SETERRQ8(PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible block of right-hand sides (m2,M2)x(n2,N2) = (%D,%D)x(%D,%D) and solutions (m1,M1)x(n1,N1) = (%D,%D)x(%D,%D)", m2, M2, n2, N2, m1, M1, n1, N1);
992: PetscObjectBaseTypeCompareAny((PetscObject)B, &match, MATSEQDENSE, MATMPIDENSE, "");
993: if (!match) SETERRQ(PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of right-hand sides not stored in a dense Mat");
994: PetscObjectBaseTypeCompareAny((PetscObject)X, &match, MATSEQDENSE, MATMPIDENSE, "");
995: if (!match) SETERRQ(PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Provided block of solutions not stored in a dense Mat");
996: KSPSetUp(ksp);
997: KSPSetUpOnBlocks(ksp);
998: if (ksp->ops->matsolve) {
999: if (ksp->guess_zero) {
1000: MatZeroEntries(X);
1001: }
1002: PetscLogEventBegin(KSP_MatSolve, ksp, B, X, 0);
1003: KSPGetMatSolveBlockSize(ksp, &Bbn);
1004: /* by default, do a single solve with all columns */
1005: if (Bbn == PETSC_DECIDE) Bbn = N2;
1006: else if (Bbn < 1) SETERRQ1(PetscObjectComm((PetscObject)ksp), PETSC_ERR_ARG_OUTOFRANGE, "KSPMatSolve() block size %D must be positive", Bbn);
1007: PetscInfo2(ksp, "KSP type %s solving using blocks of width at most %D\n", ((PetscObject)ksp)->type_name, Bbn);
1008: /* if -ksp_matsolve_block_size is greater than the actual number of columns, do a single solve with all columns */
1009: if (Bbn >= N2) {
1010: (*ksp->ops->matsolve)(ksp, B, X);
1011: if (ksp->viewFinalRes) {
1012: KSPViewFinalMatResidual_Internal(ksp, B, X, ksp->viewerFinalRes, ksp->formatFinalRes, 0);
1013: }
1014: if (ksp->viewReason) {
1015: KSPConvergedReasonView(ksp, ksp->viewerReason,PETSC_VIEWER_DEFAULT);
1016: }
1017: } else {
1018: for (n2 = 0; n2 < N2; n2 += Bbn) {
1019: MatDenseGetSubMatrix(B, n2, PetscMin(n2+Bbn, N2), &vB);
1020: MatDenseGetSubMatrix(X, n2, PetscMin(n2+Bbn, N2), &vX);
1021: (*ksp->ops->matsolve)(ksp, vB, vX);
1022: if (ksp->viewFinalRes) {
1023: KSPViewFinalMatResidual_Internal(ksp, vB, vX, ksp->viewerFinalRes, ksp->formatFinalRes, n2);
1024: }
1025: if (ksp->viewReason) {
1026: KSPConvergedReasonView(ksp, ksp->viewerReason,PETSC_VIEWER_DEFAULT);
1027: }
1028: MatDenseRestoreSubMatrix(B, &vB);
1029: MatDenseRestoreSubMatrix(X, &vX);
1030: }
1031: }
1032: if (ksp->view) {
1033: KSPView(ksp, ksp->viewer);
1034: }
1035: PetscLogEventEnd(KSP_MatSolve, ksp, B, X, 0);
1036: } else {
1037: PetscInfo1(ksp, "KSP type %s solving column by column\n", ((PetscObject)ksp)->type_name);
1038: for (n2 = 0; n2 < N2; ++n2) {
1039: MatDenseGetColumnVecRead(B, n2, &cb);
1040: MatDenseGetColumnVecWrite(X, n2, &cx);
1041: KSPSolve(ksp, cb, cx);
1042: MatDenseRestoreColumnVecWrite(X, n2, &cx);
1043: MatDenseRestoreColumnVecRead(B, n2, &cb);
1044: }
1045: }
1046: return(0);
1047: }
1049: /*@
1050: KSPSetMatSolveBlockSize - Sets the maximum number of columns treated simultaneously in KSPMatSolve().
1052: Logically collective
1054: Input Parameters:
1055: + ksp - iterative context
1056: - bs - block size
1058: Level: advanced
1060: .seealso: KSPMatSolve(), KSPGetMatSolveBlockSize(), -mat_mumps_icntl_27, -matmatmult_Bbn
1061: @*/
1062: PetscErrorCode KSPSetMatSolveBlockSize(KSP ksp, PetscInt bs)
1063: {
1069: PetscTryMethod(ksp, "KSPSetMatSolveBlockSize_C", (KSP, PetscInt), (ksp, bs));
1070: return(0);
1071: }
1073: /*@
1074: KSPGetMatSolveBlockSize - Gets the maximum number of columns treated simultaneously in KSPMatSolve().
1076: Input Parameter:
1077: . ksp - iterative context
1079: Output Parameter:
1080: . bs - block size
1082: Level: advanced
1084: .seealso: KSPMatSolve(), KSPSetMatSolveBlockSize(), -mat_mumps_icntl_27, -matmatmult_Bbn
1085: @*/
1086: PetscErrorCode KSPGetMatSolveBlockSize(KSP ksp, PetscInt *bs)
1087: {
1092: *bs = PETSC_DECIDE;
1093: PetscTryMethod(ksp, "KSPGetMatSolveBlockSize_C", (KSP, PetscInt*), (ksp, bs));
1094: return(0);
1095: }
1097: /*@
1098: KSPResetViewers - Resets all the viewers set from the options database during KSPSetFromOptions()
1100: Collective on ksp
1102: Input Parameter:
1103: . ksp - iterative context obtained from KSPCreate()
1105: Level: beginner
1107: .seealso: KSPCreate(), KSPSetUp(), KSPSolve(), KSPSetFromOptions(), KSP
1108: @*/
1109: PetscErrorCode KSPResetViewers(KSP ksp)
1110: {
1115: if (!ksp) return(0);
1116: PetscViewerDestroy(&ksp->viewer);
1117: PetscViewerDestroy(&ksp->viewerPre);
1118: PetscViewerDestroy(&ksp->viewerReason);
1119: PetscViewerDestroy(&ksp->viewerMat);
1120: PetscViewerDestroy(&ksp->viewerPMat);
1121: PetscViewerDestroy(&ksp->viewerRhs);
1122: PetscViewerDestroy(&ksp->viewerSol);
1123: PetscViewerDestroy(&ksp->viewerMatExp);
1124: PetscViewerDestroy(&ksp->viewerEV);
1125: PetscViewerDestroy(&ksp->viewerSV);
1126: PetscViewerDestroy(&ksp->viewerEVExp);
1127: PetscViewerDestroy(&ksp->viewerFinalRes);
1128: PetscViewerDestroy(&ksp->viewerPOpExp);
1129: PetscViewerDestroy(&ksp->viewerDScale);
1130: ksp->view = PETSC_FALSE;
1131: ksp->viewPre = PETSC_FALSE;
1132: ksp->viewReason = PETSC_FALSE;
1133: ksp->viewMat = PETSC_FALSE;
1134: ksp->viewPMat = PETSC_FALSE;
1135: ksp->viewRhs = PETSC_FALSE;
1136: ksp->viewSol = PETSC_FALSE;
1137: ksp->viewMatExp = PETSC_FALSE;
1138: ksp->viewEV = PETSC_FALSE;
1139: ksp->viewSV = PETSC_FALSE;
1140: ksp->viewEVExp = PETSC_FALSE;
1141: ksp->viewFinalRes = PETSC_FALSE;
1142: ksp->viewPOpExp = PETSC_FALSE;
1143: ksp->viewDScale = PETSC_FALSE;
1144: return(0);
1145: }
1147: /*@
1148: KSPReset - Resets a KSP context to the kspsetupcalled = 0 state and removes any allocated Vecs and Mats
1150: Collective on ksp
1152: Input Parameter:
1153: . ksp - iterative context obtained from KSPCreate()
1155: Level: beginner
1157: .seealso: KSPCreate(), KSPSetUp(), KSPSolve(), KSP
1158: @*/
1159: PetscErrorCode KSPReset(KSP ksp)
1160: {
1165: if (!ksp) return(0);
1166: if (ksp->ops->reset) {
1167: (*ksp->ops->reset)(ksp);
1168: }
1169: if (ksp->pc) {PCReset(ksp->pc);}
1170: if (ksp->guess) {
1171: KSPGuess guess = ksp->guess;
1172: if (guess->ops->reset) { (*guess->ops->reset)(guess); }
1173: }
1174: VecDestroyVecs(ksp->nwork,&ksp->work);
1175: VecDestroy(&ksp->vec_rhs);
1176: VecDestroy(&ksp->vec_sol);
1177: VecDestroy(&ksp->diagonal);
1178: VecDestroy(&ksp->truediagonal);
1180: KSPResetViewers(ksp);
1182: ksp->setupstage = KSP_SETUP_NEW;
1183: return(0);
1184: }
1186: /*@
1187: KSPDestroy - Destroys KSP context.
1189: Collective on ksp
1191: Input Parameter:
1192: . ksp - iterative context obtained from KSPCreate()
1194: Level: beginner
1196: .seealso: KSPCreate(), KSPSetUp(), KSPSolve(), KSP
1197: @*/
1198: PetscErrorCode KSPDestroy(KSP *ksp)
1199: {
1201: PC pc;
1204: if (!*ksp) return(0);
1206: if (--((PetscObject)(*ksp))->refct > 0) {*ksp = NULL; return(0);}
1208: PetscObjectSAWsViewOff((PetscObject)*ksp);
1210: /*
1211: Avoid a cascading call to PCReset(ksp->pc) from the following call:
1212: PCReset() shouldn't be called from KSPDestroy() as it is unprotected by pc's
1213: refcount (and may be shared, e.g., by other ksps).
1214: */
1215: pc = (*ksp)->pc;
1216: (*ksp)->pc = NULL;
1217: KSPReset((*ksp));
1218: (*ksp)->pc = pc;
1219: if ((*ksp)->ops->destroy) {(*(*ksp)->ops->destroy)(*ksp);}
1221: KSPGuessDestroy(&(*ksp)->guess);
1222: DMDestroy(&(*ksp)->dm);
1223: PCDestroy(&(*ksp)->pc);
1224: PetscFree((*ksp)->res_hist_alloc);
1225: if ((*ksp)->convergeddestroy) {
1226: (*(*ksp)->convergeddestroy)((*ksp)->cnvP);
1227: }
1228: KSPMonitorCancel((*ksp));
1229: PetscViewerDestroy(&(*ksp)->eigviewer);
1230: PetscHeaderDestroy(ksp);
1231: return(0);
1232: }
1234: /*@
1235: KSPSetPCSide - Sets the preconditioning side.
1237: Logically Collective on ksp
1239: Input Parameter:
1240: . ksp - iterative context obtained from KSPCreate()
1242: Output Parameter:
1243: . side - the preconditioning side, where side is one of
1244: .vb
1245: PC_LEFT - left preconditioning (default)
1246: PC_RIGHT - right preconditioning
1247: PC_SYMMETRIC - symmetric preconditioning
1248: .ve
1250: Options Database Keys:
1251: . -ksp_pc_side <right,left,symmetric>
1253: Notes:
1254: Left preconditioning is used by default for most Krylov methods except KSPFGMRES which only supports right preconditioning.
1256: For methods changing the side of the preconditioner changes the norm type that is used, see KSPSetNormType().
1258: Symmetric preconditioning is currently available only for the KSPQCG method. Note, however, that
1259: symmetric preconditioning can be emulated by using either right or left
1260: preconditioning and a pre or post processing step.
1262: Setting the PC side often affects the default norm type. See KSPSetNormType() for details.
1264: Level: intermediate
1266: .seealso: KSPGetPCSide(), KSPSetNormType(), KSPGetNormType(), KSP
1267: @*/
1268: PetscErrorCode KSPSetPCSide(KSP ksp,PCSide side)
1269: {
1273: ksp->pc_side = ksp->pc_side_set = side;
1274: return(0);
1275: }
1277: /*@
1278: KSPGetPCSide - Gets the preconditioning side.
1280: Not Collective
1282: Input Parameter:
1283: . ksp - iterative context obtained from KSPCreate()
1285: Output Parameter:
1286: . side - the preconditioning side, where side is one of
1287: .vb
1288: PC_LEFT - left preconditioning (default)
1289: PC_RIGHT - right preconditioning
1290: PC_SYMMETRIC - symmetric preconditioning
1291: .ve
1293: Level: intermediate
1295: .seealso: KSPSetPCSide(), KSP
1296: @*/
1297: PetscErrorCode KSPGetPCSide(KSP ksp,PCSide *side)
1298: {
1304: KSPSetUpNorms_Private(ksp,PETSC_TRUE,&ksp->normtype,&ksp->pc_side);
1305: *side = ksp->pc_side;
1306: return(0);
1307: }
1309: /*@
1310: KSPGetTolerances - Gets the relative, absolute, divergence, and maximum
1311: iteration tolerances used by the default KSP convergence tests.
1313: Not Collective
1315: Input Parameter:
1316: . ksp - the Krylov subspace context
1318: Output Parameters:
1319: + rtol - the relative convergence tolerance
1320: . abstol - the absolute convergence tolerance
1321: . dtol - the divergence tolerance
1322: - maxits - maximum number of iterations
1324: Notes:
1325: The user can specify NULL for any parameter that is not needed.
1327: Level: intermediate
1329: maximum, iterations
1331: .seealso: KSPSetTolerances(), KSP
1332: @*/
1333: PetscErrorCode KSPGetTolerances(KSP ksp,PetscReal *rtol,PetscReal *abstol,PetscReal *dtol,PetscInt *maxits)
1334: {
1337: if (abstol) *abstol = ksp->abstol;
1338: if (rtol) *rtol = ksp->rtol;
1339: if (dtol) *dtol = ksp->divtol;
1340: if (maxits) *maxits = ksp->max_it;
1341: return(0);
1342: }
1344: /*@
1345: KSPSetTolerances - Sets the relative, absolute, divergence, and maximum
1346: iteration tolerances used by the default KSP convergence testers.
1348: Logically Collective on ksp
1350: Input Parameters:
1351: + ksp - the Krylov subspace context
1352: . rtol - the relative convergence tolerance, relative decrease in the (possibly preconditioned) residual norm
1353: . abstol - the absolute convergence tolerance absolute size of the (possibly preconditioned) residual norm
1354: . dtol - the divergence tolerance, amount (possibly preconditioned) residual norm can increase before KSPConvergedDefault() concludes that the method is diverging
1355: - maxits - maximum number of iterations to use
1357: Options Database Keys:
1358: + -ksp_atol <abstol> - Sets abstol
1359: . -ksp_rtol <rtol> - Sets rtol
1360: . -ksp_divtol <dtol> - Sets dtol
1361: - -ksp_max_it <maxits> - Sets maxits
1363: Notes:
1364: Use PETSC_DEFAULT to retain the default value of any of the tolerances.
1366: See KSPConvergedDefault() for details how these parameters are used in the default convergence test. See also KSPSetConvergenceTest()
1367: for setting user-defined stopping criteria.
1369: Level: intermediate
1371: convergence, maximum, iterations
1373: .seealso: KSPGetTolerances(), KSPConvergedDefault(), KSPSetConvergenceTest(), KSP
1374: @*/
1375: PetscErrorCode KSPSetTolerances(KSP ksp,PetscReal rtol,PetscReal abstol,PetscReal dtol,PetscInt maxits)
1376: {
1384: if (rtol != PETSC_DEFAULT) {
1385: if (rtol < 0.0 || 1.0 <= rtol) SETERRQ1(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Relative tolerance %g must be non-negative and less than 1.0",(double)rtol);
1386: ksp->rtol = rtol;
1387: }
1388: if (abstol != PETSC_DEFAULT) {
1389: if (abstol < 0.0) SETERRQ1(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Absolute tolerance %g must be non-negative",(double)abstol);
1390: ksp->abstol = abstol;
1391: }
1392: if (dtol != PETSC_DEFAULT) {
1393: if (dtol < 0.0) SETERRQ1(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Divergence tolerance %g must be larger than 1.0",(double)dtol);
1394: ksp->divtol = dtol;
1395: }
1396: if (maxits != PETSC_DEFAULT) {
1397: if (maxits < 0) SETERRQ1(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Maximum number of iterations %D must be non-negative",maxits);
1398: ksp->max_it = maxits;
1399: }
1400: return(0);
1401: }
1403: /*@
1404: KSPSetInitialGuessNonzero - Tells the iterative solver that the
1405: initial guess is nonzero; otherwise KSP assumes the initial guess
1406: is to be zero (and thus zeros it out before solving).
1408: Logically Collective on ksp
1410: Input Parameters:
1411: + ksp - iterative context obtained from KSPCreate()
1412: - flg - PETSC_TRUE indicates the guess is non-zero, PETSC_FALSE indicates the guess is zero
1414: Options database keys:
1415: . -ksp_initial_guess_nonzero : use nonzero initial guess; this takes an optional truth value (0/1/no/yes/true/false)
1417: Level: beginner
1419: Notes:
1420: If this is not called the X vector is zeroed in the call to KSPSolve().
1422: .seealso: KSPGetInitialGuessNonzero(), KSPSetGuessType(), KSPGuessType, KSP
1423: @*/
1424: PetscErrorCode KSPSetInitialGuessNonzero(KSP ksp,PetscBool flg)
1425: {
1429: ksp->guess_zero = (PetscBool) !(int)flg;
1430: return(0);
1431: }
1433: /*@
1434: KSPGetInitialGuessNonzero - Determines whether the KSP solver is using
1435: a zero initial guess.
1437: Not Collective
1439: Input Parameter:
1440: . ksp - iterative context obtained from KSPCreate()
1442: Output Parameter:
1443: . flag - PETSC_TRUE if guess is nonzero, else PETSC_FALSE
1445: Level: intermediate
1447: .seealso: KSPSetInitialGuessNonzero(), KSP
1448: @*/
1449: PetscErrorCode KSPGetInitialGuessNonzero(KSP ksp,PetscBool *flag)
1450: {
1454: if (ksp->guess_zero) *flag = PETSC_FALSE;
1455: else *flag = PETSC_TRUE;
1456: return(0);
1457: }
1459: /*@
1460: KSPSetErrorIfNotConverged - Causes KSPSolve() to generate an error if the solver has not converged.
1462: Logically Collective on ksp
1464: Input Parameters:
1465: + ksp - iterative context obtained from KSPCreate()
1466: - flg - PETSC_TRUE indicates you want the error generated
1468: Options database keys:
1469: . -ksp_error_if_not_converged : this takes an optional truth value (0/1/no/yes/true/false)
1471: Level: intermediate
1473: Notes:
1474: Normally PETSc continues if a linear solver fails to converge, you can call KSPGetConvergedReason() after a KSPSolve()
1475: to determine if it has converged.
1478: .seealso: KSPGetErrorIfNotConverged(), KSP
1479: @*/
1480: PetscErrorCode KSPSetErrorIfNotConverged(KSP ksp,PetscBool flg)
1481: {
1485: ksp->errorifnotconverged = flg;
1486: return(0);
1487: }
1489: /*@
1490: KSPGetErrorIfNotConverged - Will KSPSolve() generate an error if the solver does not converge?
1492: Not Collective
1494: Input Parameter:
1495: . ksp - iterative context obtained from KSPCreate()
1497: Output Parameter:
1498: . flag - PETSC_TRUE if it will generate an error, else PETSC_FALSE
1500: Level: intermediate
1502: .seealso: KSPSetErrorIfNotConverged(), KSP
1503: @*/
1504: PetscErrorCode KSPGetErrorIfNotConverged(KSP ksp,PetscBool *flag)
1505: {
1509: *flag = ksp->errorifnotconverged;
1510: return(0);
1511: }
1513: /*@
1514: KSPSetInitialGuessKnoll - Tells the iterative solver to use PCApply(pc,b,..) to compute the initial guess (The Knoll trick)
1516: Logically Collective on ksp
1518: Input Parameters:
1519: + ksp - iterative context obtained from KSPCreate()
1520: - flg - PETSC_TRUE or PETSC_FALSE
1522: Level: advanced
1524: Developer Note: the Knoll trick is not currently implemented using the KSPGuess class
1526: .seealso: KSPGetInitialGuessKnoll(), KSPSetInitialGuessNonzero(), KSPGetInitialGuessNonzero(), KSP
1527: @*/
1528: PetscErrorCode KSPSetInitialGuessKnoll(KSP ksp,PetscBool flg)
1529: {
1533: ksp->guess_knoll = flg;
1534: return(0);
1535: }
1537: /*@
1538: KSPGetInitialGuessKnoll - Determines whether the KSP solver is using the Knoll trick (using PCApply(pc,b,...) to compute
1539: the initial guess
1541: Not Collective
1543: Input Parameter:
1544: . ksp - iterative context obtained from KSPCreate()
1546: Output Parameter:
1547: . flag - PETSC_TRUE if using Knoll trick, else PETSC_FALSE
1549: Level: advanced
1551: .seealso: KSPSetInitialGuessKnoll(), KSPSetInitialGuessNonzero(), KSPGetInitialGuessNonzero(), KSP
1552: @*/
1553: PetscErrorCode KSPGetInitialGuessKnoll(KSP ksp,PetscBool *flag)
1554: {
1558: *flag = ksp->guess_knoll;
1559: return(0);
1560: }
1562: /*@
1563: KSPGetComputeSingularValues - Gets the flag indicating whether the extreme singular
1564: values will be calculated via a Lanczos or Arnoldi process as the linear
1565: system is solved.
1567: Not Collective
1569: Input Parameter:
1570: . ksp - iterative context obtained from KSPCreate()
1572: Output Parameter:
1573: . flg - PETSC_TRUE or PETSC_FALSE
1575: Options Database Key:
1576: . -ksp_monitor_singular_value - Activates KSPSetComputeSingularValues()
1578: Notes:
1579: Currently this option is not valid for all iterative methods.
1581: Many users may just want to use the monitoring routine
1582: KSPMonitorSingularValue() (which can be set with option -ksp_monitor_singular_value)
1583: to print the singular values at each iteration of the linear solve.
1585: Level: advanced
1587: .seealso: KSPComputeExtremeSingularValues(), KSPMonitorSingularValue(), KSP
1588: @*/
1589: PetscErrorCode KSPGetComputeSingularValues(KSP ksp,PetscBool *flg)
1590: {
1594: *flg = ksp->calc_sings;
1595: return(0);
1596: }
1598: /*@
1599: KSPSetComputeSingularValues - Sets a flag so that the extreme singular
1600: values will be calculated via a Lanczos or Arnoldi process as the linear
1601: system is solved.
1603: Logically Collective on ksp
1605: Input Parameters:
1606: + ksp - iterative context obtained from KSPCreate()
1607: - flg - PETSC_TRUE or PETSC_FALSE
1609: Options Database Key:
1610: . -ksp_monitor_singular_value - Activates KSPSetComputeSingularValues()
1612: Notes:
1613: Currently this option is not valid for all iterative methods.
1615: Many users may just want to use the monitoring routine
1616: KSPMonitorSingularValue() (which can be set with option -ksp_monitor_singular_value)
1617: to print the singular values at each iteration of the linear solve.
1619: Level: advanced
1621: .seealso: KSPComputeExtremeSingularValues(), KSPMonitorSingularValue(), KSP
1622: @*/
1623: PetscErrorCode KSPSetComputeSingularValues(KSP ksp,PetscBool flg)
1624: {
1628: ksp->calc_sings = flg;
1629: return(0);
1630: }
1632: /*@
1633: KSPGetComputeEigenvalues - Gets the flag indicating that the extreme eigenvalues
1634: values will be calculated via a Lanczos or Arnoldi process as the linear
1635: system is solved.
1637: Not Collective
1639: Input Parameter:
1640: . ksp - iterative context obtained from KSPCreate()
1642: Output Parameter:
1643: . flg - PETSC_TRUE or PETSC_FALSE
1645: Notes:
1646: Currently this option is not valid for all iterative methods.
1648: Level: advanced
1650: .seealso: KSPComputeEigenvalues(), KSPComputeEigenvaluesExplicitly(), KSP
1651: @*/
1652: PetscErrorCode KSPGetComputeEigenvalues(KSP ksp,PetscBool *flg)
1653: {
1657: *flg = ksp->calc_sings;
1658: return(0);
1659: }
1661: /*@
1662: KSPSetComputeEigenvalues - Sets a flag so that the extreme eigenvalues
1663: values will be calculated via a Lanczos or Arnoldi process as the linear
1664: system is solved.
1666: Logically Collective on ksp
1668: Input Parameters:
1669: + ksp - iterative context obtained from KSPCreate()
1670: - flg - PETSC_TRUE or PETSC_FALSE
1672: Notes:
1673: Currently this option is not valid for all iterative methods.
1675: Level: advanced
1677: .seealso: KSPComputeEigenvalues(), KSPComputeEigenvaluesExplicitly(), KSP
1678: @*/
1679: PetscErrorCode KSPSetComputeEigenvalues(KSP ksp,PetscBool flg)
1680: {
1684: ksp->calc_sings = flg;
1685: return(0);
1686: }
1688: /*@
1689: KSPSetComputeRitz - Sets a flag so that the Ritz or harmonic Ritz pairs
1690: will be calculated via a Lanczos or Arnoldi process as the linear
1691: system is solved.
1693: Logically Collective on ksp
1695: Input Parameters:
1696: + ksp - iterative context obtained from KSPCreate()
1697: - flg - PETSC_TRUE or PETSC_FALSE
1699: Notes:
1700: Currently this option is only valid for the GMRES method.
1702: Level: advanced
1704: .seealso: KSPComputeRitz(), KSP
1705: @*/
1706: PetscErrorCode KSPSetComputeRitz(KSP ksp, PetscBool flg)
1707: {
1711: ksp->calc_ritz = flg;
1712: return(0);
1713: }
1715: /*@
1716: KSPGetRhs - Gets the right-hand-side vector for the linear system to
1717: be solved.
1719: Not Collective
1721: Input Parameter:
1722: . ksp - iterative context obtained from KSPCreate()
1724: Output Parameter:
1725: . r - right-hand-side vector
1727: Level: developer
1729: .seealso: KSPGetSolution(), KSPSolve(), KSP
1730: @*/
1731: PetscErrorCode KSPGetRhs(KSP ksp,Vec *r)
1732: {
1736: *r = ksp->vec_rhs;
1737: return(0);
1738: }
1740: /*@
1741: KSPGetSolution - Gets the location of the solution for the
1742: linear system to be solved. Note that this may not be where the solution
1743: is stored during the iterative process; see KSPBuildSolution().
1745: Not Collective
1747: Input Parameters:
1748: . ksp - iterative context obtained from KSPCreate()
1750: Output Parameters:
1751: . v - solution vector
1753: Level: developer
1755: .seealso: KSPGetRhs(), KSPBuildSolution(), KSPSolve(), KSP
1756: @*/
1757: PetscErrorCode KSPGetSolution(KSP ksp,Vec *v)
1758: {
1762: *v = ksp->vec_sol;
1763: return(0);
1764: }
1766: /*@
1767: KSPSetPC - Sets the preconditioner to be used to calculate the
1768: Section 1.5 Writing Application Codes with PETSc of the preconditioner on a vector.
1770: Collective on ksp
1772: Input Parameters:
1773: + ksp - iterative context obtained from KSPCreate()
1774: - pc - the preconditioner object (can be NULL)
1776: Notes:
1777: Use KSPGetPC() to retrieve the preconditioner context.
1779: Level: developer
1781: .seealso: KSPGetPC(), KSP
1782: @*/
1783: PetscErrorCode KSPSetPC(KSP ksp,PC pc)
1784: {
1789: if (pc) {
1792: }
1793: PetscObjectReference((PetscObject)pc);
1794: PCDestroy(&ksp->pc);
1795: ksp->pc = pc;
1796: PetscLogObjectParent((PetscObject)ksp,(PetscObject)ksp->pc);
1797: return(0);
1798: }
1800: /*@
1801: KSPGetPC - Returns a pointer to the preconditioner context
1802: set with KSPSetPC().
1804: Not Collective
1806: Input Parameters:
1807: . ksp - iterative context obtained from KSPCreate()
1809: Output Parameter:
1810: . pc - preconditioner context
1812: Level: developer
1814: .seealso: KSPSetPC(), KSP
1815: @*/
1816: PetscErrorCode KSPGetPC(KSP ksp,PC *pc)
1817: {
1823: if (!ksp->pc) {
1824: PCCreate(PetscObjectComm((PetscObject)ksp),&ksp->pc);
1825: PetscObjectIncrementTabLevel((PetscObject)ksp->pc,(PetscObject)ksp,0);
1826: PetscLogObjectParent((PetscObject)ksp,(PetscObject)ksp->pc);
1827: PetscObjectSetOptions((PetscObject)ksp->pc,((PetscObject)ksp)->options);
1828: }
1829: *pc = ksp->pc;
1830: return(0);
1831: }
1833: /*@
1834: KSPMonitor - runs the user provided monitor routines, if they exist
1836: Collective on ksp
1838: Input Parameters:
1839: + ksp - iterative context obtained from KSPCreate()
1840: . it - iteration number
1841: - rnorm - relative norm of the residual
1843: Notes:
1844: This routine is called by the KSP implementations.
1845: It does not typically need to be called by the user.
1847: Level: developer
1849: .seealso: KSPMonitorSet()
1850: @*/
1851: PetscErrorCode KSPMonitor(KSP ksp,PetscInt it,PetscReal rnorm)
1852: {
1853: PetscInt i, n = ksp->numbermonitors;
1857: for (i=0; i<n; i++) {
1858: (*ksp->monitor[i])(ksp,it,rnorm,ksp->monitorcontext[i]);
1859: }
1860: return(0);
1861: }
1863: /*@C
1864: KSPMonitorSet - Sets an ADDITIONAL function to be called at every iteration to monitor
1865: the residual/error etc.
1867: Logically Collective on ksp
1869: Input Parameters:
1870: + ksp - iterative context obtained from KSPCreate()
1871: . monitor - pointer to function (if this is NULL, it turns off monitoring
1872: . mctx - [optional] context for private data for the
1873: monitor routine (use NULL if no context is desired)
1874: - monitordestroy - [optional] routine that frees monitor context
1875: (may be NULL)
1877: Calling Sequence of monitor:
1878: $ monitor (KSP ksp, PetscInt it, PetscReal rnorm, void *mctx)
1880: + ksp - iterative context obtained from KSPCreate()
1881: . it - iteration number
1882: . rnorm - (estimated) 2-norm of (preconditioned) residual
1883: - mctx - optional monitoring context, as set by KSPMonitorSet()
1885: Options Database Keys:
1886: + -ksp_monitor - sets KSPMonitorDefault()
1887: . -ksp_monitor_true_residual - sets KSPMonitorTrueResidualNorm()
1888: . -ksp_monitor_max - sets KSPMonitorTrueResidualMaxNorm()
1889: . -ksp_monitor_lg_residualnorm - sets line graph monitor,
1890: uses KSPMonitorLGResidualNormCreate()
1891: . -ksp_monitor_lg_true_residualnorm - sets line graph monitor,
1892: uses KSPMonitorLGResidualNormCreate()
1893: . -ksp_monitor_singular_value - sets KSPMonitorSingularValue()
1894: - -ksp_monitor_cancel - cancels all monitors that have
1895: been hardwired into a code by
1896: calls to KSPMonitorSet(), but
1897: does not cancel those set via
1898: the options database.
1900: Notes:
1901: The default is to do nothing. To print the residual, or preconditioned
1902: residual if KSPSetNormType(ksp,KSP_NORM_PRECONDITIONED) was called, use
1903: KSPMonitorDefault() as the monitoring routine, with a ASCII viewer as the
1904: context.
1906: Several different monitoring routines may be set by calling
1907: KSPMonitorSet() multiple times; all will be called in the
1908: order in which they were set.
1910: Fortran Notes:
1911: Only a single monitor function can be set for each KSP object
1913: Level: beginner
1915: .seealso: KSPMonitorDefault(), KSPMonitorLGResidualNormCreate(), KSPMonitorCancel(), KSP
1916: @*/
1917: PetscErrorCode KSPMonitorSet(KSP ksp,PetscErrorCode (*monitor)(KSP,PetscInt,PetscReal,void*),void *mctx,PetscErrorCode (*monitordestroy)(void**))
1918: {
1919: PetscInt i;
1921: PetscBool identical;
1925: for (i=0; i<ksp->numbermonitors;i++) {
1926: PetscMonitorCompare((PetscErrorCode (*)(void))monitor,mctx,monitordestroy,(PetscErrorCode (*)(void))ksp->monitor[i],ksp->monitorcontext[i],ksp->monitordestroy[i],&identical);
1927: if (identical) return(0);
1928: }
1929: if (ksp->numbermonitors >= MAXKSPMONITORS) SETERRQ(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_OUTOFRANGE,"Too many KSP monitors set");
1930: ksp->monitor[ksp->numbermonitors] = monitor;
1931: ksp->monitordestroy[ksp->numbermonitors] = monitordestroy;
1932: ksp->monitorcontext[ksp->numbermonitors++] = (void*)mctx;
1933: return(0);
1934: }
1936: /*@
1937: KSPMonitorCancel - Clears all monitors for a KSP object.
1939: Logically Collective on ksp
1941: Input Parameters:
1942: . ksp - iterative context obtained from KSPCreate()
1944: Options Database Key:
1945: . -ksp_monitor_cancel - Cancels all monitors that have
1946: been hardwired into a code by calls to KSPMonitorSet(),
1947: but does not cancel those set via the options database.
1949: Level: intermediate
1951: .seealso: KSPMonitorDefault(), KSPMonitorLGResidualNormCreate(), KSPMonitorSet(), KSP
1952: @*/
1953: PetscErrorCode KSPMonitorCancel(KSP ksp)
1954: {
1956: PetscInt i;
1960: for (i=0; i<ksp->numbermonitors; i++) {
1961: if (ksp->monitordestroy[i]) {
1962: (*ksp->monitordestroy[i])(&ksp->monitorcontext[i]);
1963: }
1964: }
1965: ksp->numbermonitors = 0;
1966: return(0);
1967: }
1969: /*@C
1970: KSPGetMonitorContext - Gets the monitoring context, as set by
1971: KSPMonitorSet() for the FIRST monitor only.
1973: Not Collective
1975: Input Parameter:
1976: . ksp - iterative context obtained from KSPCreate()
1978: Output Parameter:
1979: . ctx - monitoring context
1981: Level: intermediate
1983: .seealso: KSPMonitorDefault(), KSPMonitorLGResidualNormCreate(), KSP
1984: @*/
1985: PetscErrorCode KSPGetMonitorContext(KSP ksp,void **ctx)
1986: {
1989: *ctx = (ksp->monitorcontext[0]);
1990: return(0);
1991: }
1993: /*@
1994: KSPSetResidualHistory - Sets the array used to hold the residual history.
1995: If set, this array will contain the residual norms computed at each
1996: iteration of the solver.
1998: Not Collective
2000: Input Parameters:
2001: + ksp - iterative context obtained from KSPCreate()
2002: . a - array to hold history
2003: . na - size of a
2004: - reset - PETSC_TRUE indicates the history counter is reset to zero
2005: for each new linear solve
2007: Level: advanced
2009: Notes:
2010: The array is NOT freed by PETSc so the user needs to keep track of
2011: it and destroy once the KSP object is destroyed.
2013: If 'a' is NULL then space is allocated for the history. If 'na' PETSC_DECIDE or PETSC_DEFAULT then a
2014: default array of length 10000 is allocated.
2016: .seealso: KSPGetResidualHistory(), KSP
2018: @*/
2019: PetscErrorCode KSPSetResidualHistory(KSP ksp,PetscReal a[],PetscInt na,PetscBool reset)
2020: {
2026: PetscFree(ksp->res_hist_alloc);
2027: if (na != PETSC_DECIDE && na != PETSC_DEFAULT && a) {
2028: ksp->res_hist = a;
2029: ksp->res_hist_max = na;
2030: } else {
2031: if (na != PETSC_DECIDE && na != PETSC_DEFAULT) ksp->res_hist_max = na;
2032: else ksp->res_hist_max = 10000; /* like default ksp->max_it */
2033: PetscCalloc1(ksp->res_hist_max,&ksp->res_hist_alloc);
2035: ksp->res_hist = ksp->res_hist_alloc;
2036: }
2037: ksp->res_hist_len = 0;
2038: ksp->res_hist_reset = reset;
2039: return(0);
2040: }
2042: /*@C
2043: KSPGetResidualHistory - Gets the array used to hold the residual history
2044: and the number of residuals it contains.
2046: Not Collective
2048: Input Parameter:
2049: . ksp - iterative context obtained from KSPCreate()
2051: Output Parameters:
2052: + a - pointer to array to hold history (or NULL)
2053: - na - number of used entries in a (or NULL)
2055: Level: advanced
2057: Notes:
2058: Can only be called after a KSPSetResidualHistory() otherwise a and na are set to zero
2060: The Fortran version of this routine has a calling sequence
2061: $ call KSPGetResidualHistory(KSP ksp, integer na, integer ierr)
2062: note that you have passed a Fortran array into KSPSetResidualHistory() and you need
2063: to access the residual values from this Fortran array you provided. Only the na (number of
2064: residual norms currently held) is set.
2066: .seealso: KSPGetResidualHistory(), KSP
2068: @*/
2069: PetscErrorCode KSPGetResidualHistory(KSP ksp,PetscReal *a[],PetscInt *na)
2070: {
2073: if (a) *a = ksp->res_hist;
2074: if (na) *na = ksp->res_hist_len;
2075: return(0);
2076: }
2078: /*@C
2079: KSPSetConvergenceTest - Sets the function to be used to determine
2080: convergence.
2082: Logically Collective on ksp
2084: Input Parameters:
2085: + ksp - iterative context obtained from KSPCreate()
2086: . converge - pointer to the function
2087: . cctx - context for private data for the convergence routine (may be null)
2088: - destroy - a routine for destroying the context (may be null)
2090: Calling sequence of converge:
2091: $ converge (KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason,void *mctx)
2093: + ksp - iterative context obtained from KSPCreate()
2094: . it - iteration number
2095: . rnorm - (estimated) 2-norm of (preconditioned) residual
2096: . reason - the reason why it has converged or diverged
2097: - cctx - optional convergence context, as set by KSPSetConvergenceTest()
2100: Notes:
2101: Must be called after the KSP type has been set so put this after
2102: a call to KSPSetType(), or KSPSetFromOptions().
2104: The default convergence test, KSPConvergedDefault(), aborts if the
2105: residual grows to more than 10000 times the initial residual.
2107: The default is a combination of relative and absolute tolerances.
2108: The residual value that is tested may be an approximation; routines
2109: that need exact values should compute them.
2111: In the default PETSc convergence test, the precise values of reason
2112: are macros such as KSP_CONVERGED_RTOL, which are defined in petscksp.h.
2114: Level: advanced
2116: .seealso: KSPConvergedDefault(), KSPGetConvergenceContext(), KSPSetTolerances(), KSP, KSPGetConvergenceTest(), KSPGetAndClearConvergenceTest()
2117: @*/
2118: PetscErrorCode KSPSetConvergenceTest(KSP ksp,PetscErrorCode (*converge)(KSP,PetscInt,PetscReal,KSPConvergedReason*,void*),void *cctx,PetscErrorCode (*destroy)(void*))
2119: {
2124: if (ksp->convergeddestroy) {
2125: (*ksp->convergeddestroy)(ksp->cnvP);
2126: }
2127: ksp->converged = converge;
2128: ksp->convergeddestroy = destroy;
2129: ksp->cnvP = (void*)cctx;
2130: return(0);
2131: }
2133: /*@C
2134: KSPGetConvergenceTest - Gets the function to be used to determine
2135: convergence.
2137: Logically Collective on ksp
2139: Input Parameter:
2140: . ksp - iterative context obtained from KSPCreate()
2142: Output Parameter:
2143: + converge - pointer to convergence test function
2144: . cctx - context for private data for the convergence routine (may be null)
2145: - destroy - a routine for destroying the context (may be null)
2147: Calling sequence of converge:
2148: $ converge (KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason,void *mctx)
2150: + ksp - iterative context obtained from KSPCreate()
2151: . it - iteration number
2152: . rnorm - (estimated) 2-norm of (preconditioned) residual
2153: . reason - the reason why it has converged or diverged
2154: - cctx - optional convergence context, as set by KSPSetConvergenceTest()
2156: Level: advanced
2158: .seealso: KSPConvergedDefault(), KSPGetConvergenceContext(), KSPSetTolerances(), KSP, KSPSetConvergenceTest(), KSPGetAndClearConvergenceTest()
2159: @*/
2160: PetscErrorCode KSPGetConvergenceTest(KSP ksp,PetscErrorCode (**converge)(KSP,PetscInt,PetscReal,KSPConvergedReason*,void*),void **cctx,PetscErrorCode (**destroy)(void*))
2161: {
2164: if (converge) *converge = ksp->converged;
2165: if (destroy) *destroy = ksp->convergeddestroy;
2166: if (cctx) *cctx = ksp->cnvP;
2167: return(0);
2168: }
2170: /*@C
2171: KSPGetAndClearConvergenceTest - Gets the function to be used to determine convergence. Removes the current test without calling destroy on the test context
2173: Logically Collective on ksp
2175: Input Parameter:
2176: . ksp - iterative context obtained from KSPCreate()
2178: Output Parameter:
2179: + converge - pointer to convergence test function
2180: . cctx - context for private data for the convergence routine
2181: - destroy - a routine for destroying the context
2183: Calling sequence of converge:
2184: $ converge (KSP ksp, PetscInt it, PetscReal rnorm, KSPConvergedReason *reason,void *mctx)
2186: + ksp - iterative context obtained from KSPCreate()
2187: . it - iteration number
2188: . rnorm - (estimated) 2-norm of (preconditioned) residual
2189: . reason - the reason why it has converged or diverged
2190: - cctx - optional convergence context, as set by KSPSetConvergenceTest()
2192: Level: advanced
2194: Notes: This is intended to be used to allow transferring the convergence test (and its context) to another testing object (for example another KSP) and then calling
2195: KSPSetConvergenceTest() on this original KSP. If you just called KSPGetConvergenceTest() followed by KSPSetConvergenceTest() the original context information
2196: would be destroyed and hence the transferred context would be invalid and trigger a crash on use
2198: .seealso: KSPConvergedDefault(), KSPGetConvergenceContext(), KSPSetTolerances(), KSP, KSPSetConvergenceTest(), KSPGetConvergenceTest()
2199: @*/
2200: PetscErrorCode KSPGetAndClearConvergenceTest(KSP ksp,PetscErrorCode (**converge)(KSP,PetscInt,PetscReal,KSPConvergedReason*,void*),void **cctx,PetscErrorCode (**destroy)(void*))
2201: {
2204: *converge = ksp->converged;
2205: *destroy = ksp->convergeddestroy;
2206: *cctx = ksp->cnvP;
2207: ksp->converged = NULL;
2208: ksp->cnvP = NULL;
2209: ksp->convergeddestroy = NULL;
2210: return(0);
2211: }
2213: /*@C
2214: KSPGetConvergenceContext - Gets the convergence context set with
2215: KSPSetConvergenceTest().
2217: Not Collective
2219: Input Parameter:
2220: . ksp - iterative context obtained from KSPCreate()
2222: Output Parameter:
2223: . ctx - monitoring context
2225: Level: advanced
2227: .seealso: KSPConvergedDefault(), KSPSetConvergenceTest(), KSP
2228: @*/
2229: PetscErrorCode KSPGetConvergenceContext(KSP ksp,void **ctx)
2230: {
2233: *ctx = ksp->cnvP;
2234: return(0);
2235: }
2237: /*@C
2238: KSPBuildSolution - Builds the approximate solution in a vector provided.
2239: This routine is NOT commonly needed (see KSPSolve()).
2241: Collective on ksp
2243: Input Parameter:
2244: . ctx - iterative context obtained from KSPCreate()
2246: Output Parameter:
2247: Provide exactly one of
2248: + v - location to stash solution.
2249: - V - the solution is returned in this location. This vector is created
2250: internally. This vector should NOT be destroyed by the user with
2251: VecDestroy().
2253: Notes:
2254: This routine can be used in one of two ways
2255: .vb
2256: KSPBuildSolution(ksp,NULL,&V);
2257: or
2258: KSPBuildSolution(ksp,v,NULL); or KSPBuildSolution(ksp,v,&v);
2259: .ve
2260: In the first case an internal vector is allocated to store the solution
2261: (the user cannot destroy this vector). In the second case the solution
2262: is generated in the vector that the user provides. Note that for certain
2263: methods, such as KSPCG, the second case requires a copy of the solution,
2264: while in the first case the call is essentially free since it simply
2265: returns the vector where the solution already is stored. For some methods
2266: like GMRES this is a reasonably expensive operation and should only be
2267: used in truly needed.
2269: Level: advanced
2271: .seealso: KSPGetSolution(), KSPBuildResidual(), KSP
2272: @*/
2273: PetscErrorCode KSPBuildSolution(KSP ksp,Vec v,Vec *V)
2274: {
2279: if (!V && !v) SETERRQ(PetscObjectComm((PetscObject)ksp),PETSC_ERR_ARG_WRONG,"Must provide either v or V");
2280: if (!V) V = &v;
2281: (*ksp->ops->buildsolution)(ksp,v,V);
2282: return(0);
2283: }
2285: /*@C
2286: KSPBuildResidual - Builds the residual in a vector provided.
2288: Collective on ksp
2290: Input Parameter:
2291: . ksp - iterative context obtained from KSPCreate()
2293: Output Parameters:
2294: + v - optional location to stash residual. If v is not provided,
2295: then a location is generated.
2296: . t - work vector. If not provided then one is generated.
2297: - V - the residual
2299: Notes:
2300: Regardless of whether or not v is provided, the residual is
2301: returned in V.
2303: Level: advanced
2305: .seealso: KSPBuildSolution()
2306: @*/
2307: PetscErrorCode KSPBuildResidual(KSP ksp,Vec t,Vec v,Vec *V)
2308: {
2310: PetscBool flag = PETSC_FALSE;
2311: Vec w = v,tt = t;
2315: if (!w) {
2316: VecDuplicate(ksp->vec_rhs,&w);
2317: PetscLogObjectParent((PetscObject)ksp,(PetscObject)w);
2318: }
2319: if (!tt) {
2320: VecDuplicate(ksp->vec_sol,&tt); flag = PETSC_TRUE;
2321: PetscLogObjectParent((PetscObject)ksp,(PetscObject)tt);
2322: }
2323: (*ksp->ops->buildresidual)(ksp,tt,w,V);
2324: if (flag) {VecDestroy(&tt);}
2325: return(0);
2326: }
2328: /*@
2329: KSPSetDiagonalScale - Tells KSP to symmetrically diagonally scale the system
2330: before solving. This actually CHANGES the matrix (and right hand side).
2332: Logically Collective on ksp
2334: Input Parameter:
2335: + ksp - the KSP context
2336: - scale - PETSC_TRUE or PETSC_FALSE
2338: Options Database Key:
2339: + -ksp_diagonal_scale -
2340: - -ksp_diagonal_scale_fix - scale the matrix back AFTER the solve
2343: Notes:
2344: Scales the matrix by D^(-1/2) A D^(-1/2) [D^(1/2) x ] = D^(-1/2) b
2345: where D_{ii} is 1/abs(A_{ii}) unless A_{ii} is zero and then it is 1.
2347: BE CAREFUL with this routine: it actually scales the matrix and right
2348: hand side that define the system. After the system is solved the matrix
2349: and right hand side remain scaled unless you use KSPSetDiagonalScaleFix()
2351: This should NOT be used within the SNES solves if you are using a line
2352: search.
2354: If you use this with the PCType Eisenstat preconditioner than you can
2355: use the PCEisenstatSetNoDiagonalScaling() option, or -pc_eisenstat_no_diagonal_scaling
2356: to save some unneeded, redundant flops.
2358: Level: intermediate
2360: .seealso: KSPGetDiagonalScale(), KSPSetDiagonalScaleFix(), KSP
2361: @*/
2362: PetscErrorCode KSPSetDiagonalScale(KSP ksp,PetscBool scale)
2363: {
2367: ksp->dscale = scale;
2368: return(0);
2369: }
2371: /*@
2372: KSPGetDiagonalScale - Checks if KSP solver scales the matrix and
2373: right hand side
2375: Not Collective
2377: Input Parameter:
2378: . ksp - the KSP context
2380: Output Parameter:
2381: . scale - PETSC_TRUE or PETSC_FALSE
2383: Notes:
2384: BE CAREFUL with this routine: it actually scales the matrix and right
2385: hand side that define the system. After the system is solved the matrix
2386: and right hand side remain scaled unless you use KSPSetDiagonalScaleFix()
2388: Level: intermediate
2390: .seealso: KSPSetDiagonalScale(), KSPSetDiagonalScaleFix(), KSP
2391: @*/
2392: PetscErrorCode KSPGetDiagonalScale(KSP ksp,PetscBool *scale)
2393: {
2397: *scale = ksp->dscale;
2398: return(0);
2399: }
2401: /*@
2402: KSPSetDiagonalScaleFix - Tells KSP to diagonally scale the system
2403: back after solving.
2405: Logically Collective on ksp
2407: Input Parameter:
2408: + ksp - the KSP context
2409: - fix - PETSC_TRUE to scale back after the system solve, PETSC_FALSE to not
2410: rescale (default)
2412: Notes:
2413: Must be called after KSPSetDiagonalScale()
2415: Using this will slow things down, because it rescales the matrix before and
2416: after each linear solve. This is intended mainly for testing to allow one
2417: to easily get back the original system to make sure the solution computed is
2418: accurate enough.
2420: Level: intermediate
2422: .seealso: KSPGetDiagonalScale(), KSPSetDiagonalScale(), KSPGetDiagonalScaleFix(), KSP
2423: @*/
2424: PetscErrorCode KSPSetDiagonalScaleFix(KSP ksp,PetscBool fix)
2425: {
2429: ksp->dscalefix = fix;
2430: return(0);
2431: }
2433: /*@
2434: KSPGetDiagonalScaleFix - Determines if KSP diagonally scales the system
2435: back after solving.
2437: Not Collective
2439: Input Parameter:
2440: . ksp - the KSP context
2442: Output Parameter:
2443: . fix - PETSC_TRUE to scale back after the system solve, PETSC_FALSE to not
2444: rescale (default)
2446: Notes:
2447: Must be called after KSPSetDiagonalScale()
2449: If PETSC_TRUE will slow things down, because it rescales the matrix before and
2450: after each linear solve. This is intended mainly for testing to allow one
2451: to easily get back the original system to make sure the solution computed is
2452: accurate enough.
2454: Level: intermediate
2456: .seealso: KSPGetDiagonalScale(), KSPSetDiagonalScale(), KSPSetDiagonalScaleFix(), KSP
2457: @*/
2458: PetscErrorCode KSPGetDiagonalScaleFix(KSP ksp,PetscBool *fix)
2459: {
2463: *fix = ksp->dscalefix;
2464: return(0);
2465: }
2467: /*@C
2468: KSPSetComputeOperators - set routine to compute the linear operators
2470: Logically Collective
2472: Input Arguments:
2473: + ksp - the KSP context
2474: . func - function to compute the operators
2475: - ctx - optional context
2477: Calling sequence of func:
2478: $ func(KSP ksp,Mat A,Mat B,void *ctx)
2480: + ksp - the KSP context
2481: . A - the linear operator
2482: . B - preconditioning matrix
2483: - ctx - optional user-provided context
2485: Notes:
2486: The user provided func() will be called automatically at the very next call to KSPSolve(). It will not be called at future KSPSolve() calls
2487: unless either KSPSetComputeOperators() or KSPSetOperators() is called before that KSPSolve() is called.
2489: To reuse the same preconditioner for the next KSPSolve() and not compute a new one based on the most recently computed matrix call KSPSetReusePreconditioner()
2491: Level: beginner
2493: .seealso: KSPSetOperators(), KSPSetComputeRHS(), DMKSPSetComputeOperators(), KSPSetComputeInitialGuess()
2494: @*/
2495: PetscErrorCode KSPSetComputeOperators(KSP ksp,PetscErrorCode (*func)(KSP,Mat,Mat,void*),void *ctx)
2496: {
2498: DM dm;
2502: KSPGetDM(ksp,&dm);
2503: DMKSPSetComputeOperators(dm,func,ctx);
2504: if (ksp->setupstage == KSP_SETUP_NEWRHS) ksp->setupstage = KSP_SETUP_NEWMATRIX;
2505: return(0);
2506: }
2508: /*@C
2509: KSPSetComputeRHS - set routine to compute the right hand side of the linear system
2511: Logically Collective
2513: Input Arguments:
2514: + ksp - the KSP context
2515: . func - function to compute the right hand side
2516: - ctx - optional context
2518: Calling sequence of func:
2519: $ func(KSP ksp,Vec b,void *ctx)
2521: + ksp - the KSP context
2522: . b - right hand side of linear system
2523: - ctx - optional user-provided context
2525: Notes:
2526: The routine you provide will be called EACH you call KSPSolve() to prepare the new right hand side for that solve
2528: Level: beginner
2530: .seealso: KSPSolve(), DMKSPSetComputeRHS(), KSPSetComputeOperators()
2531: @*/
2532: PetscErrorCode KSPSetComputeRHS(KSP ksp,PetscErrorCode (*func)(KSP,Vec,void*),void *ctx)
2533: {
2535: DM dm;
2539: KSPGetDM(ksp,&dm);
2540: DMKSPSetComputeRHS(dm,func,ctx);
2541: return(0);
2542: }
2544: /*@C
2545: KSPSetComputeInitialGuess - set routine to compute the initial guess of the linear system
2547: Logically Collective
2549: Input Arguments:
2550: + ksp - the KSP context
2551: . func - function to compute the initial guess
2552: - ctx - optional context
2554: Calling sequence of func:
2555: $ func(KSP ksp,Vec x,void *ctx)
2557: + ksp - the KSP context
2558: . x - solution vector
2559: - ctx - optional user-provided context
2561: Notes: This should only be used in conjunction with KSPSetComputeRHS(), KSPSetComputeOperators(), otherwise
2562: call KSPSetInitialGuessNonzero() and set the initial guess values in the solution vector passed to KSPSolve().
2564: Level: beginner
2566: .seealso: KSPSolve(), KSPSetComputeRHS(), KSPSetComputeOperators(), DMKSPSetComputeInitialGuess()
2567: @*/
2568: PetscErrorCode KSPSetComputeInitialGuess(KSP ksp,PetscErrorCode (*func)(KSP,Vec,void*),void *ctx)
2569: {
2571: DM dm;
2575: KSPGetDM(ksp,&dm);
2576: DMKSPSetComputeInitialGuess(dm,func,ctx);
2577: return(0);
2578: }