1: /*
2: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
3: SLEPc - Scalable Library for Eigenvalue Problem Computations
4: Copyright (c) 2002-2019, Universitat Politecnica de Valencia, Spain
6: This file is part of SLEPc.
7: SLEPc is distributed under a 2-clause BSD license (see LICENSE).
8: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
9: */
10: /*
11: SLEPc eigensolver: "krylovschur"
13: Method: Krylov-Schur
15: Algorithm:
17: Single-vector Krylov-Schur method for non-symmetric problems,
18: including harmonic extraction.
20: References:
22: [1] "Krylov-Schur Methods in SLEPc", SLEPc Technical Report STR-7,
23: available at http://slepc.upv.es.
25: [2] G.W. Stewart, "A Krylov-Schur Algorithm for Large Eigenproblems",
26: SIAM J. Matrix Anal. App. 23(3):601-614, 2001.
28: [3] "Practical Implementation of Harmonic Krylov-Schur", SLEPc Technical
29: Report STR-9, available at http://slepc.upv.es.
30: */
32: #include <slepc/private/epsimpl.h> /*I "slepceps.h" I*/
33: #include "krylovschur.h"
35: PetscErrorCode EPSGetArbitraryValues(EPS eps,PetscScalar *rr,PetscScalar *ri) 36: {
38: PetscInt i,newi,ld,n,l;
39: Vec xr=eps->work[0],xi=eps->work[1];
40: PetscScalar re,im,*Zr,*Zi,*X;
43: DSGetLeadingDimension(eps->ds,&ld);
44: DSGetDimensions(eps->ds,&n,NULL,&l,NULL,NULL);
45: for (i=l;i<n;i++) {
46: re = eps->eigr[i];
47: im = eps->eigi[i];
48: STBackTransform(eps->st,1,&re,&im);
49: newi = i;
50: DSVectors(eps->ds,DS_MAT_X,&newi,NULL);
51: DSGetArray(eps->ds,DS_MAT_X,&X);
52: Zr = X+i*ld;
53: if (newi==i+1) Zi = X+newi*ld;
54: else Zi = NULL;
55: EPSComputeRitzVector(eps,Zr,Zi,eps->V,xr,xi);
56: DSRestoreArray(eps->ds,DS_MAT_X,&X);
57: (*eps->arbitrary)(re,im,xr,xi,rr+i,ri+i,eps->arbitraryctx);
58: }
59: return(0);
60: }
62: static PetscErrorCode EstimateRange(Mat A,PetscReal *left,PetscReal *right) 63: {
65: PetscInt nconv;
66: PetscScalar eig0;
67: EPS eps;
70: *left = 0.0; *right = 0.0;
71: EPSCreate(PetscObjectComm((PetscObject)A),&eps);
72: EPSSetOptionsPrefix(eps,"eps_filter_");
73: EPSSetOperators(eps,A,NULL);
74: EPSSetProblemType(eps,EPS_HEP);
75: EPSSetTolerances(eps,1e-3,50);
76: EPSSetWhichEigenpairs(eps,EPS_SMALLEST_REAL);
77: EPSSolve(eps);
78: EPSGetConverged(eps,&nconv);
79: if (nconv>0) {
80: EPSGetEigenvalue(eps,0,&eig0,NULL);
81: } else eig0 = eps->eigr[0];
82: *left = PetscRealPart(eig0);
83: EPSSetWhichEigenpairs(eps,EPS_LARGEST_REAL);
84: EPSSolve(eps);
85: EPSGetConverged(eps,&nconv);
86: if (nconv>0) {
87: EPSGetEigenvalue(eps,0,&eig0,NULL);
88: } else eig0 = eps->eigr[0];
89: *right = PetscRealPart(eig0);
90: EPSDestroy(&eps);
91: return(0);
92: }
94: static PetscErrorCode EPSSetUp_KrylovSchur_Filter(EPS eps) 95: {
97: SlepcSC sc;
98: PetscReal rleft,rright;
99: Mat A;
102: if (eps->intb >= PETSC_MAX_REAL && eps->inta <= PETSC_MIN_REAL) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONG,"The defined computational interval should have at least one of their sides bounded");
103: if (!eps->ishermitian) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Polynomial filter only available for symmetric/Hermitian eigenproblems");
104: if (eps->isgeneralized) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Polynomial filters not available for generalized eigenproblems");
105: if (eps->arbitrary) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Arbitrary selection of eigenpairs cannot be used with polynomial filters");
106: if (eps->tol==PETSC_DEFAULT) eps->tol = SLEPC_DEFAULT_TOL*1e-2; /* use tighter tolerance */
107: STFilterSetInterval(eps->st,eps->inta,eps->intb);
108: STGetMatrix(eps->st,0,&A);
109: STFilterGetRange(eps->st,&rleft,&rright);
110: if (!rleft && !rright) {
111: EstimateRange(A,&rleft,&rright);
112: PetscInfo2(eps,"Setting eigenvalue range to [%g,%g]\n",(double)rleft,(double)rright);
113: STFilterSetRange(eps->st,rleft,rright);
114: }
115: if (!eps->ncv && eps->nev==1) eps->nev = 40; /* user did not provide nev estimation */
116: EPSSetDimensions_Default(eps,eps->nev,&eps->ncv,&eps->mpd);
117: if (eps->ncv>eps->nev+eps->mpd) SETERRQ(PetscObjectComm((PetscObject)eps),1,"The value of ncv must not be larger than nev+mpd");
118: if (!eps->max_it) eps->max_it = PetscMax(100,2*eps->n/eps->ncv);
120: DSGetSlepcSC(eps->ds,&sc);
121: sc->rg = NULL;
122: sc->comparison = SlepcCompareLargestReal;
123: sc->comparisonctx = NULL;
124: sc->map = NULL;
125: sc->mapobj = NULL;
126: return(0);
127: }
129: PetscErrorCode EPSSetUp_KrylovSchur(EPS eps)130: {
131: PetscErrorCode ierr;
132: PetscReal eta;
133: PetscBool isfilt=PETSC_FALSE;
134: BVOrthogType otype;
135: BVOrthogBlockType obtype;
136: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
137: enum { EPS_KS_DEFAULT,EPS_KS_SYMM,EPS_KS_SLICE,EPS_KS_FILTER,EPS_KS_INDEF,EPS_KS_TWOSIDED } variant;
140: /* spectrum slicing requires special treatment of default values */
141: if (eps->which==EPS_ALL) {
142: PetscObjectTypeCompare((PetscObject)eps->st,STFILTER,&isfilt);
143: if (isfilt) {
144: EPSSetUp_KrylovSchur_Filter(eps);
145: } else {
146: EPSSetUp_KrylovSchur_Slice(eps);
147: }
148: } else {
149: EPSSetDimensions_Default(eps,eps->nev,&eps->ncv,&eps->mpd);
150: if (eps->ncv>eps->nev+eps->mpd) SETERRQ(PetscObjectComm((PetscObject)eps),1,"The value of ncv must not be larger than nev+mpd");
151: if (!eps->max_it) eps->max_it = PetscMax(100,2*eps->n/eps->ncv);
152: if (!eps->which) { EPSSetWhichEigenpairs_Default(eps); }
153: }
154: if (!ctx->lock && eps->mpd<eps->ncv) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Should not use mpd parameter in non-locking variant");
156: if (eps->isgeneralized && eps->ishermitian && !eps->ispositive && eps->arbitrary) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Arbitrary selection of eigenpairs not implemented for indefinite problems");
157: if (eps->ishermitian && eps->ispositive && (eps->which==EPS_LARGEST_IMAGINARY || eps->which==EPS_SMALLEST_IMAGINARY)) SETERRQ(PetscObjectComm((PetscObject)eps),1,"Wrong value of eps->which");
159: if (!eps->extraction) {
160: EPSSetExtraction(eps,EPS_RITZ);
161: } else if (eps->extraction!=EPS_RITZ && eps->extraction!=EPS_HARMONIC) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Unsupported extraction type");
162: if (eps->extraction==EPS_HARMONIC && ctx->lock) { PetscInfo(eps,"Locking was requested but will be deactivated since is not supported with harmonic extraction\n"); }
164: if (!ctx->keep) ctx->keep = 0.5;
166: EPSAllocateSolution(eps,1);
167: EPS_SetInnerProduct(eps);
168: if (eps->arbitrary) {
169: EPSSetWorkVecs(eps,2);
170: } else if (eps->ishermitian && !eps->ispositive){
171: EPSSetWorkVecs(eps,1);
172: }
174: /* dispatch solve method */
175: if (eps->ishermitian) {
176: if (eps->which==EPS_ALL) {
177: if (eps->isgeneralized && !eps->ispositive) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Spectrum slicing not implemented for indefinite problems");
178: else variant = isfilt? EPS_KS_FILTER: EPS_KS_SLICE;
179: } else if (eps->isgeneralized && !eps->ispositive) {
180: variant = EPS_KS_INDEF;
181: } else {
182: switch (eps->extraction) {
183: case EPS_RITZ: variant = EPS_KS_SYMM; break;
184: case EPS_HARMONIC: variant = EPS_KS_DEFAULT; break;
185: default: SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Unsupported extraction type");
186: }
187: }
188: } else if (eps->twosided) {
189: variant = EPS_KS_TWOSIDED;
190: } else {
191: switch (eps->extraction) {
192: case EPS_RITZ: variant = EPS_KS_DEFAULT; break;
193: case EPS_HARMONIC: variant = EPS_KS_DEFAULT; break;
194: default: SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Unsupported extraction type");
195: }
196: }
197: switch (variant) {
198: case EPS_KS_DEFAULT:
199: eps->ops->solve = EPSSolve_KrylovSchur_Default;
200: eps->ops->computevectors = EPSComputeVectors_Schur;
201: DSSetType(eps->ds,DSNHEP);
202: DSAllocate(eps->ds,eps->ncv+1);
203: break;
204: case EPS_KS_SYMM:
205: case EPS_KS_FILTER:
206: eps->ops->solve = EPSSolve_KrylovSchur_Symm;
207: eps->ops->computevectors = EPSComputeVectors_Hermitian;
208: DSSetType(eps->ds,DSHEP);
209: DSSetCompact(eps->ds,PETSC_TRUE);
210: DSSetExtraRow(eps->ds,PETSC_TRUE);
211: DSAllocate(eps->ds,eps->ncv+1);
212: break;
213: case EPS_KS_SLICE:
214: if (eps->stopping!=EPSStoppingBasic) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_SUP,"Spectrum slicing does not support user-defined stopping test");
215: eps->ops->solve = EPSSolve_KrylovSchur_Slice;
216: eps->ops->computevectors = EPSComputeVectors_Slice;
217: break;
218: case EPS_KS_INDEF:
219: eps->ops->solve = EPSSolve_KrylovSchur_Indefinite;
220: eps->ops->computevectors = EPSComputeVectors_Indefinite;
221: DSSetType(eps->ds,DSGHIEP);
222: DSSetCompact(eps->ds,PETSC_TRUE);
223: DSAllocate(eps->ds,eps->ncv+1);
224: /* force reorthogonalization for pseudo-Lanczos */
225: BVGetOrthogonalization(eps->V,&otype,NULL,&eta,&obtype);
226: BVSetOrthogonalization(eps->V,otype,BV_ORTHOG_REFINE_ALWAYS,eta,obtype);
227: break;
228: case EPS_KS_TWOSIDED:
229: eps->ops->solve = EPSSolve_KrylovSchur_TwoSided;
230: eps->ops->computevectors = EPSComputeVectors_Schur;
231: DSSetType(eps->ds,DSNHEP);
232: DSAllocate(eps->ds,eps->ncv+1);
233: DSSetType(eps->dsts,DSNHEP);
234: DSAllocate(eps->dsts,eps->ncv+1);
235: break;
236: default: SETERRQ(PetscObjectComm((PetscObject)eps),1,"Unexpected error");
237: }
238: return(0);
239: }
241: PetscErrorCode EPSSolve_KrylovSchur_Default(EPS eps)242: {
243: PetscErrorCode ierr;
244: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
245: PetscInt i,j,*pj,k,l,nv,ld,nconv;
246: Mat U;
247: PetscScalar *S,*Q,*g;
248: PetscReal beta,gamma=1.0;
249: PetscBool breakdown,harmonic;
252: DSGetLeadingDimension(eps->ds,&ld);
253: harmonic = (eps->extraction==EPS_HARMONIC || eps->extraction==EPS_REFINED_HARMONIC)?PETSC_TRUE:PETSC_FALSE;
254: if (harmonic) { PetscMalloc1(ld,&g); }
255: if (eps->arbitrary) pj = &j;
256: else pj = NULL;
258: /* Get the starting Arnoldi vector */
259: EPSGetStartVector(eps,0,NULL);
260: l = 0;
262: /* Restart loop */
263: while (eps->reason == EPS_CONVERGED_ITERATING) {
264: eps->its++;
266: /* Compute an nv-step Arnoldi factorization */
267: nv = PetscMin(eps->nconv+eps->mpd,eps->ncv);
268: DSGetArray(eps->ds,DS_MAT_A,&S);
269: EPSBasicArnoldi(eps,PETSC_FALSE,S,ld,eps->nconv+l,&nv,&beta,&breakdown);
270: DSRestoreArray(eps->ds,DS_MAT_A,&S);
271: DSSetDimensions(eps->ds,nv,0,eps->nconv,eps->nconv+l);
272: if (l==0) {
273: DSSetState(eps->ds,DS_STATE_INTERMEDIATE);
274: } else {
275: DSSetState(eps->ds,DS_STATE_RAW);
276: }
277: BVSetActiveColumns(eps->V,eps->nconv,nv);
279: /* Compute translation of Krylov decomposition if harmonic extraction used */
280: if (harmonic) {
281: DSTranslateHarmonic(eps->ds,eps->target,beta,PETSC_FALSE,g,&gamma);
282: }
284: /* Solve projected problem */
285: DSSolve(eps->ds,eps->eigr,eps->eigi);
286: if (eps->arbitrary) {
287: EPSGetArbitraryValues(eps,eps->rr,eps->ri);
288: j=1;
289: }
290: DSSort(eps->ds,eps->eigr,eps->eigi,eps->rr,eps->ri,pj);
291: DSSynchronize(eps->ds,eps->eigr,eps->eigi);
293: /* Check convergence */
294: EPSKrylovConvergence(eps,PETSC_FALSE,eps->nconv,nv-eps->nconv,beta,0.0,gamma,&k);
295: (*eps->stopping)(eps,eps->its,eps->max_it,k,eps->nev,&eps->reason,eps->stoppingctx);
296: nconv = k;
298: /* Update l */
299: if (eps->reason != EPS_CONVERGED_ITERATING || breakdown || k==nv) l = 0;
300: else {
301: l = PetscMax(1,(PetscInt)((nv-k)*ctx->keep));
302: #if !defined(PETSC_USE_COMPLEX)
303: DSGetArray(eps->ds,DS_MAT_A,&S);
304: if (S[k+l+(k+l-1)*ld] != 0.0) {
305: if (k+l<nv-1) l = l+1;
306: else l = l-1;
307: }
308: DSRestoreArray(eps->ds,DS_MAT_A,&S);
309: #endif
310: }
311: if ((!ctx->lock || harmonic) && l>0) { l += k; k = 0; } /* non-locking variant: reset no. of converged pairs */
313: if (eps->reason == EPS_CONVERGED_ITERATING) {
314: if (breakdown || k==nv) {
315: /* Start a new Arnoldi factorization */
316: PetscInfo2(eps,"Breakdown in Krylov-Schur method (it=%D norm=%g)\n",eps->its,(double)beta);
317: if (k<eps->nev) {
318: EPSGetStartVector(eps,k,&breakdown);
319: if (breakdown) {
320: eps->reason = EPS_DIVERGED_BREAKDOWN;
321: PetscInfo(eps,"Unable to generate more start vectors\n");
322: }
323: }
324: } else {
325: /* Undo translation of Krylov decomposition */
326: if (harmonic) {
327: DSSetDimensions(eps->ds,nv,0,k,l);
328: DSTranslateHarmonic(eps->ds,0.0,beta,PETSC_TRUE,g,&gamma);
329: /* gamma u^ = u - U*g~ */
330: BVMultColumn(eps->V,-1.0,1.0,nv,g);
331: BVScaleColumn(eps->V,nv,1.0/gamma);
332: }
333: /* Prepare the Rayleigh quotient for restart */
334: DSGetArray(eps->ds,DS_MAT_A,&S);
335: DSGetArray(eps->ds,DS_MAT_Q,&Q);
336: for (i=k;i<k+l;i++) {
337: S[k+l+i*ld] = Q[nv-1+i*ld]*beta*gamma;
338: }
339: DSRestoreArray(eps->ds,DS_MAT_A,&S);
340: DSRestoreArray(eps->ds,DS_MAT_Q,&Q);
341: }
342: }
343: /* Update the corresponding vectors V(:,idx) = V*Q(:,idx) */
344: DSGetMat(eps->ds,DS_MAT_Q,&U);
345: BVMultInPlace(eps->V,U,eps->nconv,k+l);
346: MatDestroy(&U);
348: if (eps->reason == EPS_CONVERGED_ITERATING && !breakdown) {
349: BVCopyColumn(eps->V,nv,k+l);
350: }
351: eps->nconv = k;
352: EPSMonitor(eps,eps->its,nconv,eps->eigr,eps->eigi,eps->errest,nv);
353: }
355: if (harmonic) { PetscFree(g); }
356: /* truncate Schur decomposition and change the state to raw so that
357: DSVectors() computes eigenvectors from scratch */
358: DSSetDimensions(eps->ds,eps->nconv,0,0,0);
359: DSSetState(eps->ds,DS_STATE_RAW);
360: return(0);
361: }
363: static PetscErrorCode EPSKrylovSchurSetRestart_KrylovSchur(EPS eps,PetscReal keep)364: {
365: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
368: if (keep==PETSC_DEFAULT) ctx->keep = 0.5;
369: else {
370: if (keep<0.1 || keep>0.9) SETERRQ1(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"The keep argument %g must be in the range [0.1,0.9]",keep);
371: ctx->keep = keep;
372: }
373: return(0);
374: }
376: /*@
377: EPSKrylovSchurSetRestart - Sets the restart parameter for the Krylov-Schur
378: method, in particular the proportion of basis vectors that must be kept
379: after restart.
381: Logically Collective on EPS383: Input Parameters:
384: + eps - the eigenproblem solver context
385: - keep - the number of vectors to be kept at restart
387: Options Database Key:
388: . -eps_krylovschur_restart - Sets the restart parameter
390: Notes:
391: Allowed values are in the range [0.1,0.9]. The default is 0.5.
393: Level: advanced
395: .seealso: EPSKrylovSchurGetRestart()
396: @*/
397: PetscErrorCode EPSKrylovSchurSetRestart(EPS eps,PetscReal keep)398: {
404: PetscTryMethod(eps,"EPSKrylovSchurSetRestart_C",(EPS,PetscReal),(eps,keep));
405: return(0);
406: }
408: static PetscErrorCode EPSKrylovSchurGetRestart_KrylovSchur(EPS eps,PetscReal *keep)409: {
410: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
413: *keep = ctx->keep;
414: return(0);
415: }
417: /*@
418: EPSKrylovSchurGetRestart - Gets the restart parameter used in the
419: Krylov-Schur method.
421: Not Collective
423: Input Parameter:
424: . eps - the eigenproblem solver context
426: Output Parameter:
427: . keep - the restart parameter
429: Level: advanced
431: .seealso: EPSKrylovSchurSetRestart()
432: @*/
433: PetscErrorCode EPSKrylovSchurGetRestart(EPS eps,PetscReal *keep)434: {
440: PetscUseMethod(eps,"EPSKrylovSchurGetRestart_C",(EPS,PetscReal*),(eps,keep));
441: return(0);
442: }
444: static PetscErrorCode EPSKrylovSchurSetLocking_KrylovSchur(EPS eps,PetscBool lock)445: {
446: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
449: ctx->lock = lock;
450: return(0);
451: }
453: /*@
454: EPSKrylovSchurSetLocking - Choose between locking and non-locking variants of
455: the Krylov-Schur method.
457: Logically Collective on EPS459: Input Parameters:
460: + eps - the eigenproblem solver context
461: - lock - true if the locking variant must be selected
463: Options Database Key:
464: . -eps_krylovschur_locking - Sets the locking flag
466: Notes:
467: The default is to lock converged eigenpairs when the method restarts.
468: This behaviour can be changed so that all directions are kept in the
469: working subspace even if already converged to working accuracy (the
470: non-locking variant).
472: Level: advanced
474: .seealso: EPSKrylovSchurGetLocking()
475: @*/
476: PetscErrorCode EPSKrylovSchurSetLocking(EPS eps,PetscBool lock)477: {
483: PetscTryMethod(eps,"EPSKrylovSchurSetLocking_C",(EPS,PetscBool),(eps,lock));
484: return(0);
485: }
487: static PetscErrorCode EPSKrylovSchurGetLocking_KrylovSchur(EPS eps,PetscBool *lock)488: {
489: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
492: *lock = ctx->lock;
493: return(0);
494: }
496: /*@
497: EPSKrylovSchurGetLocking - Gets the locking flag used in the Krylov-Schur
498: method.
500: Not Collective
502: Input Parameter:
503: . eps - the eigenproblem solver context
505: Output Parameter:
506: . lock - the locking flag
508: Level: advanced
510: .seealso: EPSKrylovSchurSetLocking()
511: @*/
512: PetscErrorCode EPSKrylovSchurGetLocking(EPS eps,PetscBool *lock)513: {
519: PetscUseMethod(eps,"EPSKrylovSchurGetLocking_C",(EPS,PetscBool*),(eps,lock));
520: return(0);
521: }
523: static PetscErrorCode EPSKrylovSchurSetPartitions_KrylovSchur(EPS eps,PetscInt npart)524: {
525: PetscErrorCode ierr;
526: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
527: PetscMPIInt size;
530: if (ctx->npart!=npart) {
531: if (ctx->commset) { PetscSubcommDestroy(&ctx->subc); }
532: EPSDestroy(&ctx->eps);
533: }
534: if (npart == PETSC_DEFAULT || npart == PETSC_DECIDE) {
535: ctx->npart = 1;
536: } else {
537: MPI_Comm_size(PetscObjectComm((PetscObject)eps),&size);
538: if (npart<1 || npart>size) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"Illegal value of npart");
539: ctx->npart = npart;
540: }
541: eps->state = EPS_STATE_INITIAL;
542: return(0);
543: }
545: /*@
546: EPSKrylovSchurSetPartitions - Sets the number of partitions for the
547: case of doing spectrum slicing for a computational interval with the
548: communicator split in several sub-communicators.
550: Logically Collective on EPS552: Input Parameters:
553: + eps - the eigenproblem solver context
554: - npart - number of partitions
556: Options Database Key:
557: . -eps_krylovschur_partitions <npart> - Sets the number of partitions
559: Notes:
560: By default, npart=1 so all processes in the communicator participate in
561: the processing of the whole interval. If npart>1 then the interval is
562: divided into npart subintervals, each of them being processed by a
563: subset of processes.
565: The interval is split proportionally unless the separation points are
566: specified with EPSKrylovSchurSetSubintervals().
568: Level: advanced
570: .seealso: EPSKrylovSchurSetSubintervals(), EPSSetInterval()
571: @*/
572: PetscErrorCode EPSKrylovSchurSetPartitions(EPS eps,PetscInt npart)573: {
579: PetscTryMethod(eps,"EPSKrylovSchurSetPartitions_C",(EPS,PetscInt),(eps,npart));
580: return(0);
581: }
583: static PetscErrorCode EPSKrylovSchurGetPartitions_KrylovSchur(EPS eps,PetscInt *npart)584: {
585: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
588: *npart = ctx->npart;
589: return(0);
590: }
592: /*@
593: EPSKrylovSchurGetPartitions - Gets the number of partitions of the
594: communicator in case of spectrum slicing.
596: Not Collective
598: Input Parameter:
599: . eps - the eigenproblem solver context
601: Output Parameter:
602: . npart - number of partitions
604: Level: advanced
606: .seealso: EPSKrylovSchurSetPartitions()
607: @*/
608: PetscErrorCode EPSKrylovSchurGetPartitions(EPS eps,PetscInt *npart)609: {
615: PetscUseMethod(eps,"EPSKrylovSchurGetPartitions_C",(EPS,PetscInt*),(eps,npart));
616: return(0);
617: }
619: static PetscErrorCode EPSKrylovSchurSetDetectZeros_KrylovSchur(EPS eps,PetscBool detect)620: {
621: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
624: ctx->detect = detect;
625: eps->state = EPS_STATE_INITIAL;
626: return(0);
627: }
629: /*@
630: EPSKrylovSchurSetDetectZeros - Sets a flag to enforce detection of
631: zeros during the factorizations throughout the spectrum slicing computation.
633: Logically Collective on EPS635: Input Parameters:
636: + eps - the eigenproblem solver context
637: - detect - check for zeros
639: Options Database Key:
640: . -eps_krylovschur_detect_zeros - Check for zeros; this takes an optional
641: bool value (0/1/no/yes/true/false)
643: Notes:
644: A zero in the factorization indicates that a shift coincides with an eigenvalue.
646: This flag is turned off by default, and may be necessary in some cases,
647: especially when several partitions are being used. This feature currently
648: requires an external package for factorizations with support for zero
649: detection, e.g. MUMPS.
651: Level: advanced
653: .seealso: EPSKrylovSchurSetPartitions(), EPSSetInterval()
654: @*/
655: PetscErrorCode EPSKrylovSchurSetDetectZeros(EPS eps,PetscBool detect)656: {
662: PetscTryMethod(eps,"EPSKrylovSchurSetDetectZeros_C",(EPS,PetscBool),(eps,detect));
663: return(0);
664: }
666: static PetscErrorCode EPSKrylovSchurGetDetectZeros_KrylovSchur(EPS eps,PetscBool *detect)667: {
668: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
671: *detect = ctx->detect;
672: return(0);
673: }
675: /*@
676: EPSKrylovSchurGetDetectZeros - Gets the flag that enforces zero detection
677: in spectrum slicing.
679: Not Collective
681: Input Parameter:
682: . eps - the eigenproblem solver context
684: Output Parameter:
685: . detect - whether zeros detection is enforced during factorizations
687: Level: advanced
689: .seealso: EPSKrylovSchurSetDetectZeros()
690: @*/
691: PetscErrorCode EPSKrylovSchurGetDetectZeros(EPS eps,PetscBool *detect)692: {
698: PetscUseMethod(eps,"EPSKrylovSchurGetDetectZeros_C",(EPS,PetscBool*),(eps,detect));
699: return(0);
700: }
702: static PetscErrorCode EPSKrylovSchurSetDimensions_KrylovSchur(EPS eps,PetscInt nev,PetscInt ncv,PetscInt mpd)703: {
704: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
707: if (nev<1) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"Illegal value of nev. Must be > 0");
708: ctx->nev = nev;
709: if (ncv == PETSC_DECIDE || ncv == PETSC_DEFAULT) {
710: ctx->ncv = 0;
711: } else {
712: if (ncv<1) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"Illegal value of ncv. Must be > 0");
713: ctx->ncv = ncv;
714: }
715: if (mpd == PETSC_DECIDE || mpd == PETSC_DEFAULT) {
716: ctx->mpd = 0;
717: } else {
718: if (mpd<1) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"Illegal value of mpd. Must be > 0");
719: ctx->mpd = mpd;
720: }
721: eps->state = EPS_STATE_INITIAL;
722: return(0);
723: }
725: /*@
726: EPSKrylovSchurSetDimensions - Sets the dimensions used for each subsolve
727: step in case of doing spectrum slicing for a computational interval.
728: The meaning of the parameters is the same as in EPSSetDimensions().
730: Logically Collective on EPS732: Input Parameters:
733: + eps - the eigenproblem solver context
734: . nev - number of eigenvalues to compute
735: . ncv - the maximum dimension of the subspace to be used by the subsolve
736: - mpd - the maximum dimension allowed for the projected problem
738: Options Database Key:
739: + -eps_krylovschur_nev <nev> - Sets the number of eigenvalues
740: . -eps_krylovschur_ncv <ncv> - Sets the dimension of the subspace
741: - -eps_krylovschur_mpd <mpd> - Sets the maximum projected dimension
743: Level: advanced
745: .seealso: EPSKrylovSchurGetDimensions(), EPSSetDimensions(), EPSSetInterval()
746: @*/
747: PetscErrorCode EPSKrylovSchurSetDimensions(EPS eps,PetscInt nev,PetscInt ncv,PetscInt mpd)748: {
756: PetscTryMethod(eps,"EPSKrylovSchurSetDimensions_C",(EPS,PetscInt,PetscInt,PetscInt),(eps,nev,ncv,mpd));
757: return(0);
758: }
760: static PetscErrorCode EPSKrylovSchurGetDimensions_KrylovSchur(EPS eps,PetscInt *nev,PetscInt *ncv,PetscInt *mpd)761: {
762: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
765: if (nev) *nev = ctx->nev;
766: if (ncv) *ncv = ctx->ncv;
767: if (mpd) *mpd = ctx->mpd;
768: return(0);
769: }
771: /*@
772: EPSKrylovSchurGetDimensions - Gets the dimensions used for each subsolve
773: step in case of doing spectrum slicing for a computational interval.
775: Not Collective
777: Input Parameter:
778: . eps - the eigenproblem solver context
780: Output Parameters:
781: + nev - number of eigenvalues to compute
782: . ncv - the maximum dimension of the subspace to be used by the subsolve
783: - mpd - the maximum dimension allowed for the projected problem
785: Level: advanced
787: .seealso: EPSKrylovSchurSetDimensions()
788: @*/
789: PetscErrorCode EPSKrylovSchurGetDimensions(EPS eps,PetscInt *nev,PetscInt *ncv,PetscInt *mpd)790: {
795: PetscUseMethod(eps,"EPSKrylovSchurGetDimensions_C",(EPS,PetscInt*,PetscInt*,PetscInt*),(eps,nev,ncv,mpd));
796: return(0);
797: }
799: static PetscErrorCode EPSKrylovSchurSetSubintervals_KrylovSchur(EPS eps,PetscReal* subint)800: {
801: PetscErrorCode ierr;
802: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
803: PetscInt i;
806: if (subint[0]!=eps->inta || subint[ctx->npart]!=eps->intb) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONG,"First and last values must match the endpoints of EPSSetInterval()");
807: for (i=0;i<ctx->npart;i++) if (subint[i]>subint[i+1]) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONG,"Array must contain values in ascending order");
808: if (ctx->subintervals) { PetscFree(ctx->subintervals); }
809: PetscMalloc1(ctx->npart+1,&ctx->subintervals);
810: for (i=0;i<ctx->npart+1;i++) ctx->subintervals[i] = subint[i];
811: ctx->subintset = PETSC_TRUE;
812: eps->state = EPS_STATE_INITIAL;
813: return(0);
814: }
816: /*@C
817: EPSKrylovSchurSetSubintervals - Sets the points that delimit the
818: subintervals to be used in spectrum slicing with several partitions.
820: Logically Collective on EPS822: Input Parameters:
823: + eps - the eigenproblem solver context
824: - subint - array of real values specifying subintervals
826: Notes:
827: This function must be called after EPSKrylovSchurSetPartitions(). For npart
828: partitions, the argument subint must contain npart+1 real values sorted in
829: ascending order: subint_0, subint_1, ..., subint_npart, where the first
830: and last values must coincide with the interval endpoints set with
831: EPSSetInterval().
833: The subintervals are then defined by two consecutive points: [subint_0,subint_1],
834: [subint_1,subint_2], and so on.
836: Level: advanced
838: .seealso: EPSKrylovSchurSetPartitions(), EPSKrylovSchurGetSubintervals(), EPSSetInterval()
839: @*/
840: PetscErrorCode EPSKrylovSchurSetSubintervals(EPS eps,PetscReal *subint)841: {
846: PetscTryMethod(eps,"EPSKrylovSchurSetSubintervals_C",(EPS,PetscReal*),(eps,subint));
847: return(0);
848: }
850: static PetscErrorCode EPSKrylovSchurGetSubintervals_KrylovSchur(EPS eps,PetscReal **subint)851: {
852: PetscErrorCode ierr;
853: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
854: PetscInt i;
857: if (!ctx->subintset) {
858: if (!eps->state) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Must call EPSSetUp() first");
859: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
860: }
861: PetscMalloc1(ctx->npart+1,subint);
862: for (i=0;i<=ctx->npart;i++) (*subint)[i] = ctx->subintervals[i];
863: return(0);
864: }
866: /*@C
867: EPSKrylovSchurGetSubintervals - Returns the points that delimit the
868: subintervals used in spectrum slicing with several partitions.
870: Logically Collective on EPS872: Input Parameter:
873: . eps - the eigenproblem solver context
875: Output Parameter:
876: . subint - array of real values specifying subintervals
878: Notes:
879: If the user passed values with EPSKrylovSchurSetSubintervals(), then the
880: same values are returned. Otherwise, the values computed internally are
881: obtained.
883: This function is only available for spectrum slicing runs.
885: The returned array has length npart+1 (see EPSKrylovSchurGetPartitions())
886: and should be freed by the user.
888: Fortran Notes:
889: The calling sequence from Fortran is
890: .vb
891: EPSKrylovSchurGetSubintervals(eps,subint,ierr)
892: double precision subint(npart+1) output
893: .ve
895: Level: advanced
897: .seealso: EPSKrylovSchurSetSubintervals(), EPSKrylovSchurGetPartitions(), EPSSetInterval()
898: @*/
899: PetscErrorCode EPSKrylovSchurGetSubintervals(EPS eps,PetscReal **subint)900: {
906: PetscUseMethod(eps,"EPSKrylovSchurGetSubintervals_C",(EPS,PetscReal**),(eps,subint));
907: return(0);
908: }
910: static PetscErrorCode EPSKrylovSchurGetInertias_KrylovSchur(EPS eps,PetscInt *n,PetscReal **shifts,PetscInt **inertias)911: {
912: PetscErrorCode ierr;
913: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
914: PetscInt i,numsh;
915: EPS_SR sr = ctx->sr;
918: if (!eps->state) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Must call EPSSetUp() first");
919: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
920: switch (eps->state) {
921: case EPS_STATE_INITIAL:
922: break;
923: case EPS_STATE_SETUP:
924: numsh = ctx->npart+1;
925: if (n) *n = numsh;
926: if (shifts) {
927: PetscMalloc1(numsh,shifts);
928: (*shifts)[0] = eps->inta;
929: if (ctx->npart==1) (*shifts)[1] = eps->intb;
930: else for (i=1;i<numsh;i++) (*shifts)[i] = ctx->subintervals[i];
931: }
932: if (inertias) {
933: PetscMalloc1(numsh,inertias);
934: (*inertias)[0] = (sr->dir==1)?sr->inertia0:sr->inertia1;
935: if (ctx->npart==1) (*inertias)[1] = (sr->dir==1)?sr->inertia1:sr->inertia0;
936: else for (i=1;i<numsh;i++) (*inertias)[i] = (*inertias)[i-1]+ctx->nconv_loc[i-1];
937: }
938: break;
939: case EPS_STATE_SOLVED:
940: case EPS_STATE_EIGENVECTORS:
941: numsh = ctx->nshifts;
942: if (n) *n = numsh;
943: if (shifts) {
944: PetscMalloc1(numsh,shifts);
945: for (i=0;i<numsh;i++) (*shifts)[i] = ctx->shifts[i];
946: }
947: if (inertias) {
948: PetscMalloc1(numsh,inertias);
949: for (i=0;i<numsh;i++) (*inertias)[i] = ctx->inertias[i];
950: }
951: break;
952: }
953: return(0);
954: }
956: /*@C
957: EPSKrylovSchurGetInertias - Gets the values of the shifts and their
958: corresponding inertias in case of doing spectrum slicing for a
959: computational interval.
961: Not Collective
963: Input Parameter:
964: . eps - the eigenproblem solver context
966: Output Parameters:
967: + n - number of shifts, including the endpoints of the interval
968: . shifts - the values of the shifts used internally in the solver
969: - inertias - the values of the inertia in each shift
971: Notes:
972: If called after EPSSolve(), all shifts used internally by the solver are
973: returned (including both endpoints and any intermediate ones). If called
974: before EPSSolve() and after EPSSetUp() then only the information of the
975: endpoints of subintervals is available.
977: This function is only available for spectrum slicing runs.
979: The returned arrays should be freed by the user. Can pass NULL in any of
980: the two arrays if not required.
982: Fortran Notes:
983: The calling sequence from Fortran is
984: .vb
985: EPSKrylovSchurGetInertias(eps,n,shifts,inertias,ierr)
986: integer n
987: double precision shifts(*)
988: integer inertias(*)
989: .ve
990: The arrays should be at least of length n. The value of n can be determined
991: by an initial call
992: .vb
993: EPSKrylovSchurGetInertias(eps,n,PETSC_NULL_REAL,PETSC_NULL_INTEGER,ierr)
994: .ve
996: Level: advanced
998: .seealso: EPSSetInterval(), EPSKrylovSchurSetSubintervals()
999: @*/
1000: PetscErrorCode EPSKrylovSchurGetInertias(EPS eps,PetscInt *n,PetscReal **shifts,PetscInt **inertias)1001: {
1007: PetscUseMethod(eps,"EPSKrylovSchurGetInertias_C",(EPS,PetscInt*,PetscReal**,PetscInt**),(eps,n,shifts,inertias));
1008: return(0);
1009: }
1011: static PetscErrorCode EPSKrylovSchurGetSubcommInfo_KrylovSchur(EPS eps,PetscInt *k,PetscInt *n,Vec *v)1012: {
1013: PetscErrorCode ierr;
1014: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
1015: EPS_SR sr = ((EPS_KRYLOVSCHUR*)ctx->eps->data)->sr;
1018: if (!eps->state) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Must call EPSSetUp() first");
1019: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
1020: if (k) *k = (ctx->npart==1)? 0: ctx->subc->color;
1021: if (n) *n = sr->numEigs;
1022: if (v) {
1023: BVCreateVec(sr->V,v);
1024: }
1025: return(0);
1026: }
1028: /*@C
1029: EPSKrylovSchurGetSubcommInfo - Gets information related to the case of
1030: doing spectrum slicing for a computational interval with multiple
1031: communicators.
1033: Collective on the subcommunicator (if v is given)
1035: Input Parameter:
1036: . eps - the eigenproblem solver context
1038: Output Parameters:
1039: + k - index of the subinterval for the calling process
1040: . n - number of eigenvalues found in the k-th subinterval
1041: - v - a vector owned by processes in the subcommunicator with dimensions
1042: compatible for locally computed eigenvectors (or NULL)
1044: Notes:
1045: This function is only available for spectrum slicing runs.
1047: The returned Vec should be destroyed by the user.
1049: Level: advanced
1051: .seealso: EPSSetInterval(), EPSKrylovSchurSetPartitions(), EPSKrylovSchurGetSubcommPairs()
1052: @*/
1053: PetscErrorCode EPSKrylovSchurGetSubcommInfo(EPS eps,PetscInt *k,PetscInt *n,Vec *v)1054: {
1059: PetscUseMethod(eps,"EPSKrylovSchurGetSubcommInfo_C",(EPS,PetscInt*,PetscInt*,Vec*),(eps,k,n,v));
1060: return(0);
1061: }
1063: static PetscErrorCode EPSKrylovSchurGetSubcommPairs_KrylovSchur(EPS eps,PetscInt i,PetscScalar *eig,Vec v)1064: {
1065: PetscErrorCode ierr;
1066: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
1067: EPS_SR sr = ((EPS_KRYLOVSCHUR*)ctx->eps->data)->sr;
1070: EPSCheckSolved(eps,1);
1071: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
1072: if (i<0 || i>=sr->numEigs) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_OUTOFRANGE,"Argument 2 out of range");
1073: if (eig) *eig = sr->eigr[sr->perm[i]];
1074: if (v) { BVCopyVec(sr->V,sr->perm[i],v); }
1075: return(0);
1076: }
1078: /*@
1079: EPSKrylovSchurGetSubcommPairs - Gets the i-th eigenpair stored
1080: internally in the subcommunicator to which the calling process belongs.
1082: Collective on the subcommunicator (if v is given)
1084: Input Parameter:
1085: + eps - the eigenproblem solver context
1086: - i - index of the solution
1088: Output Parameters:
1089: + eig - the eigenvalue
1090: - v - the eigenvector
1092: Notes:
1093: It is allowed to pass NULL for v if the eigenvector is not required.
1094: Otherwise, the caller must provide a valid Vec objects, i.e.,
1095: it must be created by the calling program with EPSKrylovSchurGetSubcommInfo().
1097: The index i should be a value between 0 and n-1, where n is the number of
1098: vectors in the local subinterval, see EPSKrylovSchurGetSubcommInfo().
1100: Level: advanced
1102: .seealso: EPSSetInterval(), EPSKrylovSchurSetPartitions(), EPSKrylovSchurGetSubcommInfo(), EPSKrylovSchurGetSubcommMats()
1103: @*/
1104: PetscErrorCode EPSKrylovSchurGetSubcommPairs(EPS eps,PetscInt i,PetscScalar *eig,Vec v)1105: {
1111: PetscUseMethod(eps,"EPSKrylovSchurGetSubcommPairs_C",(EPS,PetscInt,PetscScalar*,Vec),(eps,i,eig,v));
1112: return(0);
1113: }
1115: static PetscErrorCode EPSKrylovSchurGetSubcommMats_KrylovSchur(EPS eps,Mat *A,Mat *B)1116: {
1117: PetscErrorCode ierr;
1118: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
1121: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
1122: if (!eps->state) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Must call EPSSetUp() first");
1123: EPSGetOperators(ctx->eps,A,B);
1124: return(0);
1125: }
1127: /*@C
1128: EPSKrylovSchurGetSubcommMats - Gets the eigenproblem matrices stored
1129: internally in the subcommunicator to which the calling process belongs.
1131: Collective on the subcommunicator
1133: Input Parameter:
1134: . eps - the eigenproblem solver context
1136: Output Parameters:
1137: + A - the matrix associated with the eigensystem
1138: - B - the second matrix in the case of generalized eigenproblems
1140: Notes:
1141: This is the analog of EPSGetOperators(), but returns the matrices distributed
1142: differently (in the subcommunicator rather than in the parent communicator).
1144: These matrices should not be modified by the user.
1146: Level: advanced
1148: .seealso: EPSSetInterval(), EPSKrylovSchurSetPartitions(), EPSKrylovSchurGetSubcommInfo()
1149: @*/
1150: PetscErrorCode EPSKrylovSchurGetSubcommMats(EPS eps,Mat *A,Mat *B)1151: {
1156: PetscTryMethod(eps,"EPSKrylovSchurGetSubcommMats_C",(EPS,Mat*,Mat*),(eps,A,B));
1157: return(0);
1158: }
1160: static PetscErrorCode EPSKrylovSchurUpdateSubcommMats_KrylovSchur(EPS eps,PetscScalar a,PetscScalar ap,Mat Au,PetscScalar b,PetscScalar bp, Mat Bu,MatStructure str,PetscBool globalup)1161: {
1162: PetscErrorCode ierr;
1163: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data,*subctx;
1164: Mat A,B=NULL,Ag,Bg=NULL;
1165: PetscBool reuse=PETSC_TRUE;
1168: if (!ctx->sr) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Only available in interval computations, see EPSSetInterval()");
1169: if (!eps->state) SETERRQ(PetscObjectComm((PetscObject)eps),PETSC_ERR_ARG_WRONGSTATE,"Must call EPSSetUp() first");
1170: EPSGetOperators(eps,&Ag,&Bg);
1171: EPSGetOperators(ctx->eps,&A,&B);
1173: MatScale(A,a);
1174: if (Au) {
1175: MatAXPY(A,ap,Au,str);
1176: }
1177: if (B) MatScale(B,b);
1178: if (Bu) {
1179: MatAXPY(B,bp,Bu,str);
1180: }
1181: EPSSetOperators(ctx->eps,A,B);
1183: /* Update stored matrix state */
1184: subctx = (EPS_KRYLOVSCHUR*)ctx->eps->data;
1185: PetscObjectStateGet((PetscObject)A,&subctx->Astate);
1186: if (B) { PetscObjectStateGet((PetscObject)B,&subctx->Bstate); }
1188: /* Update matrices in the parent communicator if requested by user */
1189: if (globalup) {
1190: if (ctx->npart>1) {
1191: if (!ctx->isrow) {
1192: MatGetOwnershipIS(Ag,&ctx->isrow,&ctx->iscol);
1193: reuse = PETSC_FALSE;
1194: }
1195: if (str==DIFFERENT_NONZERO_PATTERN) reuse = PETSC_FALSE;
1196: if (ctx->submata && !reuse) {
1197: MatDestroyMatrices(1,&ctx->submata);
1198: }
1199: MatCreateSubMatrices(A,1,&ctx->isrow,&ctx->iscol,(reuse)?MAT_REUSE_MATRIX:MAT_INITIAL_MATRIX,&ctx->submata);
1200: MatCreateMPIMatConcatenateSeqMat(((PetscObject)Ag)->comm,ctx->submata[0],PETSC_DECIDE,MAT_REUSE_MATRIX,&Ag);
1201: if (B) {
1202: if (ctx->submatb && !reuse) {
1203: MatDestroyMatrices(1,&ctx->submatb);
1204: }
1205: MatCreateSubMatrices(B,1,&ctx->isrow,&ctx->iscol,(reuse)?MAT_REUSE_MATRIX:MAT_INITIAL_MATRIX,&ctx->submatb);
1206: MatCreateMPIMatConcatenateSeqMat(((PetscObject)Bg)->comm,ctx->submatb[0],PETSC_DECIDE,MAT_REUSE_MATRIX,&Bg);
1207: }
1208: }
1209: PetscObjectStateGet((PetscObject)Ag,&ctx->Astate);
1210: if (Bg) { PetscObjectStateGet((PetscObject)Bg,&ctx->Bstate); }
1211: }
1212: EPSSetOperators(eps,Ag,Bg);
1213: return(0);
1214: }
1216: /*@
1217: EPSKrylovSchurUpdateSubcommMats - Update the eigenproblem matrices stored
1218: internally in the subcommunicator to which the calling process belongs.
1220: Collective on EPS1222: Input Parameters:
1223: + eps - the eigenproblem solver context
1224: . s - scalar that multiplies the existing A matrix
1225: . a - scalar used in the axpy operation on A
1226: . Au - matrix used in the axpy operation on A
1227: . t - scalar that multiplies the existing B matrix
1228: . b - scalar used in the axpy operation on B
1229: . Bu - matrix used in the axpy operation on B
1230: . str - structure flag
1231: - globalup - flag indicating if global matrices must be updated
1233: Notes:
1234: This function modifies the eigenproblem matrices at the subcommunicator level,
1235: and optionally updates the global matrices in the parent communicator. The updates
1236: are expressed as A <-- s*A + a*Au, B <-- t*B + b*Bu.
1238: It is possible to update one of the matrices, or both.
1240: The matrices Au and Bu must be equal in all subcommunicators.
1242: The str flag is passed to the MatAXPY() operations to perform the updates.
1244: If globalup is true, communication is carried out to reconstruct the updated
1245: matrices in the parent communicator. The user must be warned that if global
1246: matrices are not in sync with subcommunicator matrices, the errors computed
1247: by EPSComputeError() will be wrong even if the computed solution is correct
1248: (the synchronization may be done only once at the end).
1250: Level: advanced
1252: .seealso: EPSSetInterval(), EPSKrylovSchurSetPartitions(), EPSKrylovSchurGetSubcommMats()
1253: @*/
1254: PetscErrorCode EPSKrylovSchurUpdateSubcommMats(EPS eps,PetscScalar s,PetscScalar a,Mat Au,PetscScalar t,PetscScalar b,Mat Bu,MatStructure str,PetscBool globalup)1255: {
1268: PetscTryMethod(eps,"EPSKrylovSchurUpdateSubcommMats_C",(EPS,PetscScalar,PetscScalar,Mat,PetscScalar,PetscScalar,Mat,MatStructure,PetscBool),(eps,s,a,Au,t,b,Bu,str,globalup));
1269: return(0);
1270: }
1272: PetscErrorCode EPSSetFromOptions_KrylovSchur(PetscOptionItems *PetscOptionsObject,EPS eps)1273: {
1274: PetscErrorCode ierr;
1275: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
1276: PetscBool flg,lock,b,f1,f2,f3;
1277: PetscReal keep;
1278: PetscInt i,j,k;
1281: PetscOptionsHead(PetscOptionsObject,"EPS Krylov-Schur Options");
1283: PetscOptionsReal("-eps_krylovschur_restart","Proportion of vectors kept after restart","EPSKrylovSchurSetRestart",0.5,&keep,&flg);
1284: if (flg) { EPSKrylovSchurSetRestart(eps,keep); }
1286: PetscOptionsBool("-eps_krylovschur_locking","Choose between locking and non-locking variants","EPSKrylovSchurSetLocking",PETSC_TRUE,&lock,&flg);
1287: if (flg) { EPSKrylovSchurSetLocking(eps,lock); }
1289: i = ctx->npart;
1290: PetscOptionsInt("-eps_krylovschur_partitions","Number of partitions of the communicator for spectrum slicing","EPSKrylovSchurSetPartitions",ctx->npart,&i,&flg);
1291: if (flg) { EPSKrylovSchurSetPartitions(eps,i); }
1293: b = ctx->detect;
1294: PetscOptionsBool("-eps_krylovschur_detect_zeros","Check zeros during factorizations at subinterval boundaries","EPSKrylovSchurSetDetectZeros",ctx->detect,&b,&flg);
1295: if (flg) { EPSKrylovSchurSetDetectZeros(eps,b); }
1297: i = 1;
1298: j = k = PETSC_DECIDE;
1299: PetscOptionsInt("-eps_krylovschur_nev","Number of eigenvalues to compute in each subsolve (only for spectrum slicing)","EPSKrylovSchurSetDimensions",40,&i,&f1);
1300: PetscOptionsInt("-eps_krylovschur_ncv","Number of basis vectors in each subsolve (only for spectrum slicing)","EPSKrylovSchurSetDimensions",80,&j,&f2);
1301: PetscOptionsInt("-eps_krylovschur_mpd","Maximum dimension of projected problem in each subsolve (only for spectrum slicing)","EPSKrylovSchurSetDimensions",80,&k,&f3);
1302: if (f1 || f2 || f3) { EPSKrylovSchurSetDimensions(eps,i,j,k); }
1304: PetscOptionsTail();
1305: return(0);
1306: }
1308: PetscErrorCode EPSView_KrylovSchur(EPS eps,PetscViewer viewer)1309: {
1310: PetscErrorCode ierr;
1311: EPS_KRYLOVSCHUR *ctx = (EPS_KRYLOVSCHUR*)eps->data;
1312: PetscBool isascii,isfilt;
1315: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isascii);
1316: if (isascii) {
1317: PetscViewerASCIIPrintf(viewer," %d%% of basis vectors kept after restart\n",(int)(100*ctx->keep));
1318: PetscViewerASCIIPrintf(viewer," using the %slocking variant\n",ctx->lock?"":"non-");
1319: if (eps->which==EPS_ALL) {
1320: PetscObjectTypeCompare((PetscObject)eps->st,STFILTER,&isfilt);
1321: if (isfilt) {
1322: PetscViewerASCIIPrintf(viewer," using filtering to extract all eigenvalues in an interval\n");
1323: } else {
1324: PetscViewerASCIIPrintf(viewer," doing spectrum slicing with nev=%D, ncv=%D, mpd=%D\n",ctx->nev,ctx->ncv,ctx->mpd);
1325: if (ctx->npart>1) {
1326: PetscViewerASCIIPrintf(viewer," multi-communicator spectrum slicing with %D partitions\n",ctx->npart);
1327: if (ctx->detect) { PetscViewerASCIIPrintf(viewer," detecting zeros when factorizing at subinterval boundaries\n"); }
1328: }
1329: }
1330: }
1331: }
1332: return(0);
1333: }
1335: PetscErrorCode EPSDestroy_KrylovSchur(EPS eps)1336: {
1340: PetscFree(eps->data);
1341: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetRestart_C",NULL);
1342: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetRestart_C",NULL);
1343: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetLocking_C",NULL);
1344: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetLocking_C",NULL);
1345: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetPartitions_C",NULL);
1346: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetPartitions_C",NULL);
1347: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetDetectZeros_C",NULL);
1348: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetDetectZeros_C",NULL);
1349: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetDimensions_C",NULL);
1350: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetDimensions_C",NULL);
1351: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetSubintervals_C",NULL);
1352: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubintervals_C",NULL);
1353: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetInertias_C",NULL);
1354: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommInfo_C",NULL);
1355: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommPairs_C",NULL);
1356: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommMats_C",NULL);
1357: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurUpdateSubcommMats_C",NULL);
1358: return(0);
1359: }
1361: PetscErrorCode EPSReset_KrylovSchur(EPS eps)1362: {
1364: PetscBool isfilt;
1367: PetscObjectTypeCompare((PetscObject)eps->st,STFILTER,&isfilt);
1368: if (eps->which==EPS_ALL && !isfilt) {
1369: EPSReset_KrylovSchur_Slice(eps);
1370: }
1371: return(0);
1372: }
1374: PetscErrorCode EPSSetDefaultST_KrylovSchur(EPS eps)1375: {
1379: if (eps->which==EPS_ALL) {
1380: if (!((PetscObject)eps->st)->type_name) {
1381: STSetType(eps->st,STSINVERT);
1382: }
1383: }
1384: return(0);
1385: }
1387: SLEPC_EXTERN PetscErrorCode EPSCreate_KrylovSchur(EPS eps)1388: {
1389: EPS_KRYLOVSCHUR *ctx;
1390: PetscErrorCode ierr;
1393: PetscNewLog(eps,&ctx);
1394: eps->data = (void*)ctx;
1395: ctx->lock = PETSC_TRUE;
1396: ctx->nev = 1;
1397: ctx->npart = 1;
1398: ctx->detect = PETSC_FALSE;
1399: ctx->global = PETSC_TRUE;
1401: eps->useds = PETSC_TRUE;
1402: eps->hasts = PETSC_TRUE;
1404: /* solve and computevectors determined at setup */
1405: eps->ops->setup = EPSSetUp_KrylovSchur;
1406: eps->ops->setfromoptions = EPSSetFromOptions_KrylovSchur;
1407: eps->ops->destroy = EPSDestroy_KrylovSchur;
1408: eps->ops->reset = EPSReset_KrylovSchur;
1409: eps->ops->view = EPSView_KrylovSchur;
1410: eps->ops->backtransform = EPSBackTransform_Default;
1411: eps->ops->setdefaultst = EPSSetDefaultST_KrylovSchur;
1413: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetRestart_C",EPSKrylovSchurSetRestart_KrylovSchur);
1414: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetRestart_C",EPSKrylovSchurGetRestart_KrylovSchur);
1415: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetLocking_C",EPSKrylovSchurSetLocking_KrylovSchur);
1416: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetLocking_C",EPSKrylovSchurGetLocking_KrylovSchur);
1417: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetPartitions_C",EPSKrylovSchurSetPartitions_KrylovSchur);
1418: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetPartitions_C",EPSKrylovSchurGetPartitions_KrylovSchur);
1419: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetDetectZeros_C",EPSKrylovSchurSetDetectZeros_KrylovSchur);
1420: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetDetectZeros_C",EPSKrylovSchurGetDetectZeros_KrylovSchur);
1421: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetDimensions_C",EPSKrylovSchurSetDimensions_KrylovSchur);
1422: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetDimensions_C",EPSKrylovSchurGetDimensions_KrylovSchur);
1423: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurSetSubintervals_C",EPSKrylovSchurSetSubintervals_KrylovSchur);
1424: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubintervals_C",EPSKrylovSchurGetSubintervals_KrylovSchur);
1425: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetInertias_C",EPSKrylovSchurGetInertias_KrylovSchur);
1426: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommInfo_C",EPSKrylovSchurGetSubcommInfo_KrylovSchur);
1427: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommPairs_C",EPSKrylovSchurGetSubcommPairs_KrylovSchur);
1428: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurGetSubcommMats_C",EPSKrylovSchurGetSubcommMats_KrylovSchur);
1429: PetscObjectComposeFunction((PetscObject)eps,"EPSKrylovSchurUpdateSubcommMats_C",EPSKrylovSchurUpdateSubcommMats_KrylovSchur);
1430: return(0);
1431: }