Actual source code: pciss.c
slepc-3.17.0 2022-03-31
1: /*
2: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
3: SLEPc - Scalable Library for Eigenvalue Problem Computations
4: Copyright (c) 2002-, 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: "ciss"
13: Method: Contour Integral Spectral Slicing
15: Algorithm:
17: Contour integral based on Sakurai-Sugiura method to construct a
18: subspace, with various eigenpair extractions (Rayleigh-Ritz,
19: explicit moment).
21: Based on code contributed by Y. Maeda, T. Sakurai.
23: References:
25: [1] J. Asakura, T. Sakurai, H. Tadano, T. Ikegami, K. Kimura, "A
26: numerical method for polynomial eigenvalue problems using contour
27: integral", Japan J. Indust. Appl. Math. 27:73-90, 2010.
28: */
30: #include <slepc/private/pepimpl.h>
31: #include <slepc/private/slepccontour.h>
33: typedef struct {
34: /* parameters */
35: PetscInt N; /* number of integration points (32) */
36: PetscInt L; /* block size (16) */
37: PetscInt M; /* moment degree (N/4 = 4) */
38: PetscReal delta; /* threshold of singular value (1e-12) */
39: PetscInt L_max; /* maximum number of columns of the source matrix V */
40: PetscReal spurious_threshold; /* discard spurious eigenpairs */
41: PetscBool isreal; /* T(z) is real for real z */
42: PetscInt npart; /* number of partitions */
43: PetscInt refine_inner;
44: PetscInt refine_blocksize;
45: PEPCISSExtraction extraction;
46: /* private data */
47: SlepcContourData contour;
48: PetscReal *sigma; /* threshold for numerical rank */
49: PetscScalar *weight;
50: PetscScalar *omega;
51: PetscScalar *pp;
52: BV V;
53: BV S;
54: BV Y;
55: PetscBool useconj;
56: Mat J,*Psplit; /* auxiliary matrices */
57: BV pV;
58: PetscObjectId rgid;
59: PetscObjectState rgstate;
60: } PEP_CISS;
62: static PetscErrorCode PEPComputeFunction(PEP pep,PetscScalar lambda,Mat T,Mat P,PetscBool deriv)
63: {
64: PetscInt i;
65: PetscScalar *coeff;
66: Mat *A,*K;
67: MatStructure str,strp;
68: PEP_CISS *ctx = (PEP_CISS*)pep->data;
69: SlepcContourData contour = ctx->contour;
71: A = (contour->pA)?contour->pA:pep->A;
72: K = (contour->pP)?contour->pP:ctx->Psplit;
73: PetscMalloc1(pep->nmat,&coeff);
74: if (deriv) PEPEvaluateBasisDerivative(pep,lambda,0,coeff,NULL);
75: else PEPEvaluateBasis(pep,lambda,0,coeff,NULL);
76: STGetMatStructure(pep->st,&str);
77: MatZeroEntries(T);
78: if (!deriv && T != P) {
79: STGetSplitPreconditionerInfo(pep->st,NULL,&strp);
80: MatZeroEntries(P);
81: }
82: i = deriv?1:0;
83: for (;i<pep->nmat;i++) {
84: MatAXPY(T,coeff[i],A[i],str);
85: if (!deriv && T != P) MatAXPY(P,coeff[i],K[i],strp);
86: }
87: PetscFree(coeff);
88: PetscFunctionReturn(0);
89: }
91: /*
92: Set up KSP solvers for every integration point
93: */
94: static PetscErrorCode PEPCISSSetUp(PEP pep,Mat T,Mat P)
95: {
96: PEP_CISS *ctx = (PEP_CISS*)pep->data;
97: SlepcContourData contour;
98: PetscInt i,p_id;
99: Mat Amat,Pmat;
101: if (!ctx->contour || !ctx->contour->ksp) PEPCISSGetKSPs(pep,NULL,NULL);
102: contour = ctx->contour;
103: for (i=0;i<contour->npoints;i++) {
104: p_id = i*contour->subcomm->n + contour->subcomm->color;
105: MatDuplicate(T,MAT_DO_NOT_COPY_VALUES,&Amat);
106: if (T != P) MatDuplicate(P,MAT_DO_NOT_COPY_VALUES,&Pmat); else Pmat = Amat;
107: PEPComputeFunction(pep,ctx->omega[p_id],Amat,Pmat,PETSC_FALSE);
108: PEP_KSPSetOperators(contour->ksp[i],Amat,Pmat);
109: MatDestroy(&Amat);
110: if (T != P) MatDestroy(&Pmat);
111: }
112: PetscFunctionReturn(0);
113: }
115: /*
116: Y_i = F(z_i)^{-1}Fp(z_i)V for every integration point, Y=[Y_i] is in the context
117: */
118: static PetscErrorCode PEPCISSSolve(PEP pep,Mat dT,BV V,PetscInt L_start,PetscInt L_end)
119: {
120: PEP_CISS *ctx = (PEP_CISS*)pep->data;
121: SlepcContourData contour;
122: PetscInt i,p_id;
123: Mat MV,BMV=NULL,MC;
125: contour = ctx->contour;
126: BVSetActiveColumns(V,L_start,L_end);
127: BVGetMat(V,&MV);
128: for (i=0;i<contour->npoints;i++) {
129: p_id = i*contour->subcomm->n + contour->subcomm->color;
130: PEPComputeFunction(pep,ctx->omega[p_id],dT,NULL,PETSC_TRUE);
131: BVSetActiveColumns(ctx->Y,i*ctx->L+L_start,i*ctx->L+L_end);
132: BVGetMat(ctx->Y,&MC);
133: if (!i) {
134: MatProductCreate(dT,MV,NULL,&BMV);
135: MatProductSetType(BMV,MATPRODUCT_AB);
136: MatProductSetFromOptions(BMV);
137: MatProductSymbolic(BMV);
138: }
139: MatProductNumeric(BMV);
140: KSPMatSolve(contour->ksp[i],BMV,MC);
141: BVRestoreMat(ctx->Y,&MC);
142: }
143: MatDestroy(&BMV);
144: BVRestoreMat(V,&MV);
145: PetscFunctionReturn(0);
146: }
148: PetscErrorCode PEPSetUp_CISS(PEP pep)
149: {
150: PEP_CISS *ctx = (PEP_CISS*)pep->data;
151: SlepcContourData contour;
152: PetscInt i,nwork,nsplit;
153: PetscBool istrivial,isellipse,flg;
154: PetscObjectId id;
155: PetscObjectState state;
156: Vec v0;
158: if (pep->ncv==PETSC_DEFAULT) pep->ncv = ctx->L_max*ctx->M;
159: else {
160: ctx->L_max = pep->ncv/ctx->M;
161: if (!ctx->L_max) {
162: ctx->L_max = 1;
163: pep->ncv = ctx->L_max*ctx->M;
164: }
165: }
166: ctx->L = PetscMin(ctx->L,ctx->L_max);
167: if (pep->max_it==PETSC_DEFAULT) pep->max_it = 5;
168: if (pep->mpd==PETSC_DEFAULT) pep->mpd = pep->ncv;
169: if (!pep->which) pep->which = PEP_ALL;
171: PEPCheckUnsupported(pep,PEP_FEATURE_STOPPING);
172: PEPCheckIgnored(pep,PEP_FEATURE_SCALE);
174: /* check region */
175: RGIsTrivial(pep->rg,&istrivial);
177: RGGetComplement(pep->rg,&flg);
179: PetscObjectTypeCompare((PetscObject)pep->rg,RGELLIPSE,&isellipse);
182: /* if the region has changed, then reset contour data */
183: PetscObjectGetId((PetscObject)pep->rg,&id);
184: PetscObjectStateGet((PetscObject)pep->rg,&state);
185: if (ctx->rgid && (id != ctx->rgid || state != ctx->rgstate)) {
186: SlepcContourDataDestroy(&ctx->contour);
187: PetscInfo(pep,"Resetting the contour data structure due to a change of region\n");
188: ctx->rgid = id; ctx->rgstate = state;
189: }
191: /* create contour data structure */
192: if (!ctx->contour) {
193: RGCanUseConjugates(pep->rg,ctx->isreal,&ctx->useconj);
194: SlepcContourDataCreate(ctx->useconj?ctx->N/2:ctx->N,ctx->npart,(PetscObject)pep,&ctx->contour);
195: }
197: PEPAllocateSolution(pep,0);
198: if (ctx->weight) PetscFree4(ctx->weight,ctx->omega,ctx->pp,ctx->sigma);
199: PetscMalloc4(ctx->N,&ctx->weight,ctx->N,&ctx->omega,ctx->N,&ctx->pp,ctx->L_max*ctx->M,&ctx->sigma);
200: PetscLogObjectMemory((PetscObject)pep,3*ctx->N*sizeof(PetscScalar)+ctx->L_max*ctx->N*sizeof(PetscReal));
202: /* allocate basis vectors */
203: BVDestroy(&ctx->S);
204: BVDuplicateResize(pep->V,ctx->L*ctx->M,&ctx->S);
205: PetscLogObjectParent((PetscObject)pep,(PetscObject)ctx->S);
206: BVDestroy(&ctx->V);
207: BVDuplicateResize(pep->V,ctx->L,&ctx->V);
208: PetscLogObjectParent((PetscObject)pep,(PetscObject)ctx->V);
210: /* check if a user-defined split preconditioner has been set */
211: STGetSplitPreconditionerInfo(pep->st,&nsplit,NULL);
212: if (nsplit) {
213: PetscFree(ctx->Psplit);
214: PetscMalloc1(nsplit,&ctx->Psplit);
215: for (i=0;i<nsplit;i++) STGetSplitPreconditionerTerm(pep->st,i,&ctx->Psplit[i]);
216: }
218: contour = ctx->contour;
219: SlepcContourRedundantMat(contour,pep->nmat,pep->A,ctx->Psplit);
220: if (!ctx->J) {
221: MatDuplicate(contour->pA?contour->pA[0]:pep->A[0],MAT_DO_NOT_COPY_VALUES,&ctx->J);
222: PetscLogObjectParent((PetscObject)pep,(PetscObject)ctx->J);
223: }
224: if (contour->pA) {
225: BVGetColumn(ctx->V,0,&v0);
226: SlepcContourScatterCreate(contour,v0);
227: BVRestoreColumn(ctx->V,0,&v0);
228: BVDestroy(&ctx->pV);
229: BVCreate(PetscObjectComm((PetscObject)contour->xsub),&ctx->pV);
230: BVSetSizesFromVec(ctx->pV,contour->xsub,pep->n);
231: BVSetFromOptions(ctx->pV);
232: BVResize(ctx->pV,ctx->L,PETSC_FALSE);
233: PetscLogObjectParent((PetscObject)pep,(PetscObject)ctx->pV);
234: }
236: BVDestroy(&ctx->Y);
237: if (contour->pA) {
238: BVCreate(PetscObjectComm((PetscObject)contour->xsub),&ctx->Y);
239: BVSetSizesFromVec(ctx->Y,contour->xsub,pep->n);
240: BVSetFromOptions(ctx->Y);
241: BVResize(ctx->Y,contour->npoints*ctx->L,PETSC_FALSE);
242: } else BVDuplicateResize(pep->V,contour->npoints*ctx->L,&ctx->Y);
244: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) DSSetType(pep->ds,DSGNHEP);
245: else if (ctx->extraction == PEP_CISS_EXTRACTION_CAA) DSSetType(pep->ds,DSNHEP);
246: else {
247: DSSetType(pep->ds,DSPEP);
248: DSPEPSetDegree(pep->ds,pep->nmat-1);
249: DSPEPSetCoefficients(pep->ds,pep->pbc);
250: }
251: DSAllocate(pep->ds,pep->ncv);
252: nwork = 2;
253: PEPSetWorkVecs(pep,nwork);
254: PetscFunctionReturn(0);
255: }
257: PetscErrorCode PEPSolve_CISS(PEP pep)
258: {
259: PEP_CISS *ctx = (PEP_CISS*)pep->data;
260: SlepcContourData contour = ctx->contour;
261: Mat X,M,E,T,P;
262: PetscInt i,j,ld,L_add=0,nv=0,L_base=ctx->L,inner,*inside,nsplit;
263: PetscScalar *Mu,*H0,*H1,*rr,*temp,center;
264: PetscReal error,max_error,radius,rgscale,est_eig,eta;
265: PetscBool isellipse,*fl1;
266: Vec si;
267: SlepcSC sc;
268: PetscRandom rand;
270: DSSetFromOptions(pep->ds);
271: DSGetSlepcSC(pep->ds,&sc);
272: sc->comparison = SlepcCompareLargestMagnitude;
273: sc->comparisonctx = NULL;
274: sc->map = NULL;
275: sc->mapobj = NULL;
276: DSGetLeadingDimension(pep->ds,&ld);
277: RGComputeQuadrature(pep->rg,RG_QUADRULE_TRAPEZOIDAL,ctx->N,ctx->omega,ctx->pp,ctx->weight);
278: STGetSplitPreconditionerInfo(pep->st,&nsplit,NULL);
279: if (contour->pA) {
280: T = contour->pA[0];
281: P = nsplit? contour->pP[0]: T;
282: } else {
283: T = pep->A[0];
284: P = nsplit? ctx->Psplit[0]: T;
285: }
286: PEPCISSSetUp(pep,T,P);
287: BVSetActiveColumns(ctx->V,0,ctx->L);
288: BVSetRandomSign(ctx->V);
289: BVGetRandomContext(ctx->V,&rand);
290: if (contour->pA) BVScatter(ctx->V,ctx->pV,contour->scatterin,contour->xdup);
291: PEPCISSSolve(pep,ctx->J,(contour->pA)?ctx->pV:ctx->V,0,ctx->L);
292: PetscObjectTypeCompare((PetscObject)pep->rg,RGELLIPSE,&isellipse);
293: if (isellipse) {
294: BVTraceQuadrature(ctx->Y,ctx->V,ctx->L,ctx->L,ctx->weight,contour->scatterin,contour->subcomm,contour->npoints,ctx->useconj,&est_eig);
295: PetscInfo(pep,"Estimated eigenvalue count: %f\n",(double)est_eig);
296: eta = PetscPowReal(10.0,-PetscLog10Real(pep->tol)/ctx->N);
297: L_add = PetscMax(0,(PetscInt)PetscCeilReal((est_eig*eta)/ctx->M)-ctx->L);
298: if (L_add>ctx->L_max-ctx->L) {
299: PetscInfo(pep,"Number of eigenvalues inside the contour path may be too large\n");
300: L_add = ctx->L_max-ctx->L;
301: }
302: }
303: /* Updates L after estimate the number of eigenvalue */
304: if (L_add>0) {
305: PetscInfo(pep,"Changing L %" PetscInt_FMT " -> %" PetscInt_FMT " by Estimate #Eig\n",ctx->L,ctx->L+L_add);
306: BVCISSResizeBases(ctx->S,contour->pA?ctx->pV:ctx->V,ctx->Y,ctx->L,ctx->L+L_add,ctx->M,contour->npoints);
307: BVSetActiveColumns(ctx->V,ctx->L,ctx->L+L_add);
308: BVSetRandomSign(ctx->V);
309: if (contour->pA) BVScatter(ctx->V,ctx->pV,contour->scatterin,contour->xdup);
310: ctx->L += L_add;
311: PEPCISSSolve(pep,ctx->J,(contour->pA)?ctx->pV:ctx->V,ctx->L-L_add,ctx->L);
312: }
314: PetscMalloc2(ctx->L*ctx->L*ctx->M*2,&Mu,ctx->L*ctx->M*ctx->L*ctx->M,&H0);
315: for (i=0;i<ctx->refine_blocksize;i++) {
316: BVDotQuadrature(ctx->Y,(contour->pA)?ctx->pV:ctx->V,Mu,ctx->M,ctx->L,ctx->L,ctx->weight,ctx->pp,contour->subcomm,contour->npoints,ctx->useconj);
317: CISS_BlockHankel(Mu,0,ctx->L,ctx->M,H0);
318: PetscLogEventBegin(PEP_CISS_SVD,pep,0,0,0);
319: SlepcCISS_BH_SVD(H0,ctx->L*ctx->M,ctx->delta,ctx->sigma,&nv);
320: PetscLogEventEnd(PEP_CISS_SVD,pep,0,0,0);
321: if (ctx->sigma[0]<=ctx->delta || nv < ctx->L*ctx->M || ctx->L == ctx->L_max) break;
322: L_add = L_base;
323: if (ctx->L+L_add>ctx->L_max) L_add = ctx->L_max-ctx->L;
324: PetscInfo(pep,"Changing L %" PetscInt_FMT " -> %" PetscInt_FMT " by SVD(H0)\n",ctx->L,ctx->L+L_add);
325: BVCISSResizeBases(ctx->S,contour->pA?ctx->pV:ctx->V,ctx->Y,ctx->L,ctx->L+L_add,ctx->M,contour->npoints);
326: BVSetActiveColumns(ctx->V,ctx->L,ctx->L+L_add);
327: BVSetRandomSign(ctx->V);
328: if (contour->pA) BVScatter(ctx->V,ctx->pV,contour->scatterin,contour->xdup);
329: ctx->L += L_add;
330: PEPCISSSolve(pep,ctx->J,(contour->pA)?ctx->pV:ctx->V,ctx->L-L_add,ctx->L);
331: if (L_add) {
332: PetscFree2(Mu,H0);
333: PetscMalloc2(ctx->L*ctx->L*ctx->M*2,&Mu,ctx->L*ctx->M*ctx->L*ctx->M,&H0);
334: }
335: }
337: RGGetScale(pep->rg,&rgscale);
338: RGEllipseGetParameters(pep->rg,¢er,&radius,NULL);
340: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) PetscMalloc1(ctx->L*ctx->M*ctx->L*ctx->M,&H1);
342: while (pep->reason == PEP_CONVERGED_ITERATING) {
343: pep->its++;
344: for (inner=0;inner<=ctx->refine_inner;inner++) {
345: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) {
346: BVDotQuadrature(ctx->Y,(contour->pA)?ctx->pV:ctx->V,Mu,ctx->M,ctx->L,ctx->L,ctx->weight,ctx->pp,contour->subcomm,contour->npoints,ctx->useconj);
347: CISS_BlockHankel(Mu,0,ctx->L,ctx->M,H0);
348: PetscLogEventBegin(PEP_CISS_SVD,pep,0,0,0);
349: SlepcCISS_BH_SVD(H0,ctx->L*ctx->M,ctx->delta,ctx->sigma,&nv);
350: PetscLogEventEnd(PEP_CISS_SVD,pep,0,0,0);
351: } else {
352: BVSumQuadrature(ctx->S,ctx->Y,ctx->M,ctx->L,ctx->L,ctx->weight,ctx->pp,contour->scatterin,contour->subcomm,contour->npoints,ctx->useconj);
353: /* compute SVD of S */
354: BVSVDAndRank(ctx->S,ctx->M,ctx->L,ctx->delta,(ctx->extraction==PEP_CISS_EXTRACTION_CAA)?BV_SVD_METHOD_QR_CAA:BV_SVD_METHOD_QR,H0,ctx->sigma,&nv);
355: }
356: PetscInfo(pep,"Estimated rank: nv = %" PetscInt_FMT "\n",nv);
357: if (ctx->sigma[0]>ctx->delta && nv==ctx->L*ctx->M && inner!=ctx->refine_inner) {
358: BVSumQuadrature(ctx->S,ctx->Y,ctx->M,ctx->L,ctx->L,ctx->weight,ctx->pp,contour->scatterin,contour->subcomm,contour->npoints,ctx->useconj);
359: BVSetActiveColumns(ctx->S,0,ctx->L);
360: BVSetActiveColumns(ctx->V,0,ctx->L);
361: BVCopy(ctx->S,ctx->V);
362: if (contour->pA) BVScatter(ctx->V,ctx->pV,contour->scatterin,contour->xdup);
363: PEPCISSSolve(pep,ctx->J,(contour->pA)?ctx->pV:ctx->V,0,ctx->L);
364: } else break;
365: }
366: pep->nconv = 0;
367: if (nv == 0) { pep->reason = PEP_CONVERGED_TOL; break; }
368: else {
369: /* Extracting eigenpairs */
370: DSSetDimensions(pep->ds,nv,0,0);
371: DSSetState(pep->ds,DS_STATE_RAW);
372: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) {
373: CISS_BlockHankel(Mu,0,ctx->L,ctx->M,H0);
374: CISS_BlockHankel(Mu,1,ctx->L,ctx->M,H1);
375: DSGetArray(pep->ds,DS_MAT_A,&temp);
376: for (j=0;j<nv;j++)
377: for (i=0;i<nv;i++)
378: temp[i+j*ld] = H1[i+j*ctx->L*ctx->M];
379: DSRestoreArray(pep->ds,DS_MAT_A,&temp);
380: DSGetArray(pep->ds,DS_MAT_B,&temp);
381: for (j=0;j<nv;j++)
382: for (i=0;i<nv;i++)
383: temp[i+j*ld] = H0[i+j*ctx->L*ctx->M];
384: DSRestoreArray(pep->ds,DS_MAT_B,&temp);
385: } else if (ctx->extraction == PEP_CISS_EXTRACTION_CAA) {
386: BVSetActiveColumns(ctx->S,0,nv);
387: DSGetArray(pep->ds,DS_MAT_A,&temp);
388: for (i=0;i<nv;i++) PetscArraycpy(temp+i*ld,H0+i*nv,nv);
389: DSRestoreArray(pep->ds,DS_MAT_A,&temp);
390: } else {
391: BVSetActiveColumns(ctx->S,0,nv);
392: for (i=0;i<pep->nmat;i++) {
393: DSGetMat(pep->ds,DSMatExtra[i],&E);
394: BVMatProject(ctx->S,pep->A[i],ctx->S,E);
395: DSRestoreMat(pep->ds,DSMatExtra[i],&E);
396: }
397: nv = (pep->nmat-1)*nv;
398: }
399: DSSolve(pep->ds,pep->eigr,pep->eigi);
400: DSSynchronize(pep->ds,pep->eigr,pep->eigi);
401: if (ctx->extraction == PEP_CISS_EXTRACTION_CAA || ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) {
402: for (i=0;i<nv;i++) {
403: pep->eigr[i] = (pep->eigr[i]*radius+center)*rgscale;
404: }
405: }
406: PetscMalloc3(nv,&fl1,nv,&inside,nv,&rr);
407: DSVectors(pep->ds,DS_MAT_X,NULL,NULL);
408: DSGetMat(pep->ds,DS_MAT_X,&X);
409: SlepcCISS_isGhost(X,nv,ctx->sigma,ctx->spurious_threshold,fl1);
410: MatDestroy(&X);
411: RGCheckInside(pep->rg,nv,pep->eigr,pep->eigi,inside);
412: for (i=0;i<nv;i++) {
413: if (fl1[i] && inside[i]>=0) {
414: rr[i] = 1.0;
415: pep->nconv++;
416: } else rr[i] = 0.0;
417: }
418: DSSort(pep->ds,pep->eigr,pep->eigi,rr,NULL,&pep->nconv);
419: DSSynchronize(pep->ds,pep->eigr,pep->eigi);
420: if (ctx->extraction == PEP_CISS_EXTRACTION_CAA || ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) {
421: for (i=0;i<nv;i++) pep->eigr[i] = (pep->eigr[i]*radius+center)*rgscale;
422: }
423: PetscFree3(fl1,inside,rr);
424: BVSetActiveColumns(pep->V,0,nv);
425: DSVectors(pep->ds,DS_MAT_X,NULL,NULL);
426: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) {
427: BVSumQuadrature(ctx->S,ctx->Y,ctx->M,ctx->L,ctx->L,ctx->weight,ctx->pp,contour->scatterin,contour->subcomm,contour->npoints,ctx->useconj);
428: BVSetActiveColumns(ctx->S,0,nv);
429: BVCopy(ctx->S,pep->V);
430: DSGetMat(pep->ds,DS_MAT_X,&X);
431: BVMultInPlace(ctx->S,X,0,pep->nconv);
432: BVMultInPlace(pep->V,X,0,pep->nconv);
433: MatDestroy(&X);
434: } else {
435: DSGetMat(pep->ds,DS_MAT_X,&X);
436: BVMultInPlace(ctx->S,X,0,pep->nconv);
437: MatDestroy(&X);
438: BVSetActiveColumns(ctx->S,0,pep->nconv);
439: BVCopy(ctx->S,pep->V);
440: }
441: max_error = 0.0;
442: for (i=0;i<pep->nconv;i++) {
443: BVGetColumn(pep->V,i,&si);
444: VecNormalize(si,NULL);
445: PEPComputeResidualNorm_Private(pep,pep->eigr[i],0,si,NULL,pep->work,&error);
446: (*pep->converged)(pep,pep->eigr[i],0,error,&error,pep->convergedctx);
447: BVRestoreColumn(pep->V,i,&si);
448: max_error = PetscMax(max_error,error);
449: }
450: if (max_error <= pep->tol) pep->reason = PEP_CONVERGED_TOL;
451: else if (pep->its > pep->max_it) pep->reason = PEP_DIVERGED_ITS;
452: else {
453: if (pep->nconv > ctx->L) nv = pep->nconv;
454: else if (ctx->L > nv) nv = ctx->L;
455: nv = PetscMin(nv,ctx->L*ctx->M);
456: MatCreateSeqDense(PETSC_COMM_SELF,nv,ctx->L,NULL,&M);
457: MatSetRandom(M,rand);
458: BVSetActiveColumns(ctx->S,0,nv);
459: BVMultInPlace(ctx->S,M,0,ctx->L);
460: MatDestroy(&M);
461: BVSetActiveColumns(ctx->S,0,ctx->L);
462: BVSetActiveColumns(ctx->V,0,ctx->L);
463: BVCopy(ctx->S,ctx->V);
464: if (contour->pA) BVScatter(ctx->V,ctx->pV,contour->scatterin,contour->xdup);
465: PEPCISSSolve(pep,ctx->J,(contour->pA)?ctx->pV:ctx->V,0,ctx->L);
466: }
467: }
468: }
469: PetscFree2(Mu,H0);
470: if (ctx->extraction == PEP_CISS_EXTRACTION_HANKEL) PetscFree(H1);
471: PetscFunctionReturn(0);
472: }
474: static PetscErrorCode PEPCISSSetSizes_CISS(PEP pep,PetscInt ip,PetscInt bs,PetscInt ms,PetscInt npart,PetscInt bsmax,PetscBool realmats)
475: {
476: PEP_CISS *ctx = (PEP_CISS*)pep->data;
477: PetscInt oN,oL,oM,oLmax,onpart;
478: PetscMPIInt size;
480: oN = ctx->N;
481: if (ip == PETSC_DECIDE || ip == PETSC_DEFAULT) {
482: if (ctx->N!=32) { ctx->N =32; ctx->M = ctx->N/4; }
483: } else {
486: if (ctx->N!=ip) { ctx->N = ip; ctx->M = ctx->N/4; }
487: }
488: oL = ctx->L;
489: if (bs == PETSC_DECIDE || bs == PETSC_DEFAULT) {
490: ctx->L = 16;
491: } else {
493: ctx->L = bs;
494: }
495: oM = ctx->M;
496: if (ms == PETSC_DECIDE || ms == PETSC_DEFAULT) {
497: ctx->M = ctx->N/4;
498: } else {
501: ctx->M = PetscMax(ms,2);
502: }
503: onpart = ctx->npart;
504: if (npart == PETSC_DECIDE || npart == PETSC_DEFAULT) {
505: ctx->npart = 1;
506: } else {
507: MPI_Comm_size(PetscObjectComm((PetscObject)pep),&size);
509: ctx->npart = npart;
510: }
511: oLmax = ctx->L_max;
512: if (bsmax == PETSC_DECIDE || bsmax == PETSC_DEFAULT) {
513: ctx->L_max = 64;
514: } else {
516: ctx->L_max = PetscMax(bsmax,ctx->L);
517: }
518: if (onpart != ctx->npart || oN != ctx->N || realmats != ctx->isreal) {
519: SlepcContourDataDestroy(&ctx->contour);
520: PetscInfo(pep,"Resetting the contour data structure due to a change of parameters\n");
521: pep->state = PEP_STATE_INITIAL;
522: }
523: ctx->isreal = realmats;
524: if (oL != ctx->L || oM != ctx->M || oLmax != ctx->L_max) pep->state = PEP_STATE_INITIAL;
525: PetscFunctionReturn(0);
526: }
528: /*@
529: PEPCISSSetSizes - Sets the values of various size parameters in the CISS solver.
531: Logically Collective on pep
533: Input Parameters:
534: + pep - the polynomial eigensolver context
535: . ip - number of integration points
536: . bs - block size
537: . ms - moment size
538: . npart - number of partitions when splitting the communicator
539: . bsmax - max block size
540: - realmats - all coefficient matrices of P(.) are real
542: Options Database Keys:
543: + -pep_ciss_integration_points - Sets the number of integration points
544: . -pep_ciss_blocksize - Sets the block size
545: . -pep_ciss_moments - Sets the moment size
546: . -pep_ciss_partitions - Sets the number of partitions
547: . -pep_ciss_maxblocksize - Sets the maximum block size
548: - -pep_ciss_realmats - all coefficient matrices of P(.) are real
550: Notes:
551: The default number of partitions is 1. This means the internal KSP object is shared
552: among all processes of the PEP communicator. Otherwise, the communicator is split
553: into npart communicators, so that npart KSP solves proceed simultaneously.
555: Level: advanced
557: .seealso: PEPCISSGetSizes()
558: @*/
559: PetscErrorCode PEPCISSSetSizes(PEP pep,PetscInt ip,PetscInt bs,PetscInt ms,PetscInt npart,PetscInt bsmax,PetscBool realmats)
560: {
568: PetscTryMethod(pep,"PEPCISSSetSizes_C",(PEP,PetscInt,PetscInt,PetscInt,PetscInt,PetscInt,PetscBool),(pep,ip,bs,ms,npart,bsmax,realmats));
569: PetscFunctionReturn(0);
570: }
572: static PetscErrorCode PEPCISSGetSizes_CISS(PEP pep,PetscInt *ip,PetscInt *bs,PetscInt *ms,PetscInt *npart,PetscInt *bsmax,PetscBool *realmats)
573: {
574: PEP_CISS *ctx = (PEP_CISS*)pep->data;
576: if (ip) *ip = ctx->N;
577: if (bs) *bs = ctx->L;
578: if (ms) *ms = ctx->M;
579: if (npart) *npart = ctx->npart;
580: if (bsmax) *bsmax = ctx->L_max;
581: if (realmats) *realmats = ctx->isreal;
582: PetscFunctionReturn(0);
583: }
585: /*@
586: PEPCISSGetSizes - Gets the values of various size parameters in the CISS solver.
588: Not Collective
590: Input Parameter:
591: . pep - the polynomial eigensolver context
593: Output Parameters:
594: + ip - number of integration points
595: . bs - block size
596: . ms - moment size
597: . npart - number of partitions when splitting the communicator
598: . bsmax - max block size
599: - realmats - all coefficient matrices of P(.) are real
601: Level: advanced
603: .seealso: PEPCISSSetSizes()
604: @*/
605: PetscErrorCode PEPCISSGetSizes(PEP pep,PetscInt *ip,PetscInt *bs,PetscInt *ms,PetscInt *npart,PetscInt *bsmax,PetscBool *realmats)
606: {
608: PetscUseMethod(pep,"PEPCISSGetSizes_C",(PEP,PetscInt*,PetscInt*,PetscInt*,PetscInt*,PetscInt*,PetscBool*),(pep,ip,bs,ms,npart,bsmax,realmats));
609: PetscFunctionReturn(0);
610: }
612: static PetscErrorCode PEPCISSSetThreshold_CISS(PEP pep,PetscReal delta,PetscReal spur)
613: {
614: PEP_CISS *ctx = (PEP_CISS*)pep->data;
616: if (delta == PETSC_DEFAULT) {
617: ctx->delta = SLEPC_DEFAULT_TOL*1e-4;
618: } else {
620: ctx->delta = delta;
621: }
622: if (spur == PETSC_DEFAULT) {
623: ctx->spurious_threshold = PetscSqrtReal(SLEPC_DEFAULT_TOL);
624: } else {
626: ctx->spurious_threshold = spur;
627: }
628: PetscFunctionReturn(0);
629: }
631: /*@
632: PEPCISSSetThreshold - Sets the values of various threshold parameters in
633: the CISS solver.
635: Logically Collective on pep
637: Input Parameters:
638: + pep - the polynomial eigensolver context
639: . delta - threshold for numerical rank
640: - spur - spurious threshold (to discard spurious eigenpairs)
642: Options Database Keys:
643: + -pep_ciss_delta - Sets the delta
644: - -pep_ciss_spurious_threshold - Sets the spurious threshold
646: Level: advanced
648: .seealso: PEPCISSGetThreshold()
649: @*/
650: PetscErrorCode PEPCISSSetThreshold(PEP pep,PetscReal delta,PetscReal spur)
651: {
655: PetscTryMethod(pep,"PEPCISSSetThreshold_C",(PEP,PetscReal,PetscReal),(pep,delta,spur));
656: PetscFunctionReturn(0);
657: }
659: static PetscErrorCode PEPCISSGetThreshold_CISS(PEP pep,PetscReal *delta,PetscReal *spur)
660: {
661: PEP_CISS *ctx = (PEP_CISS*)pep->data;
663: if (delta) *delta = ctx->delta;
664: if (spur) *spur = ctx->spurious_threshold;
665: PetscFunctionReturn(0);
666: }
668: /*@
669: PEPCISSGetThreshold - Gets the values of various threshold parameters in
670: the CISS solver.
672: Not Collective
674: Input Parameter:
675: . pep - the polynomial eigensolver context
677: Output Parameters:
678: + delta - threshold for numerical rank
679: - spur - spurious threshold (to discard spurious eigenpairs)
681: Level: advanced
683: .seealso: PEPCISSSetThreshold()
684: @*/
685: PetscErrorCode PEPCISSGetThreshold(PEP pep,PetscReal *delta,PetscReal *spur)
686: {
688: PetscUseMethod(pep,"PEPCISSGetThreshold_C",(PEP,PetscReal*,PetscReal*),(pep,delta,spur));
689: PetscFunctionReturn(0);
690: }
692: static PetscErrorCode PEPCISSSetRefinement_CISS(PEP pep,PetscInt inner,PetscInt blsize)
693: {
694: PEP_CISS *ctx = (PEP_CISS*)pep->data;
696: if (inner == PETSC_DEFAULT) {
697: ctx->refine_inner = 0;
698: } else {
700: ctx->refine_inner = inner;
701: }
702: if (blsize == PETSC_DEFAULT) {
703: ctx->refine_blocksize = 0;
704: } else {
706: ctx->refine_blocksize = blsize;
707: }
708: PetscFunctionReturn(0);
709: }
711: /*@
712: PEPCISSSetRefinement - Sets the values of various refinement parameters
713: in the CISS solver.
715: Logically Collective on pep
717: Input Parameters:
718: + pep - the polynomial eigensolver context
719: . inner - number of iterative refinement iterations (inner loop)
720: - blsize - number of iterative refinement iterations (blocksize loop)
722: Options Database Keys:
723: + -pep_ciss_refine_inner - Sets number of inner iterations
724: - -pep_ciss_refine_blocksize - Sets number of blocksize iterations
726: Level: advanced
728: .seealso: PEPCISSGetRefinement()
729: @*/
730: PetscErrorCode PEPCISSSetRefinement(PEP pep,PetscInt inner,PetscInt blsize)
731: {
735: PetscTryMethod(pep,"PEPCISSSetRefinement_C",(PEP,PetscInt,PetscInt),(pep,inner,blsize));
736: PetscFunctionReturn(0);
737: }
739: static PetscErrorCode PEPCISSGetRefinement_CISS(PEP pep,PetscInt *inner,PetscInt *blsize)
740: {
741: PEP_CISS *ctx = (PEP_CISS*)pep->data;
743: if (inner) *inner = ctx->refine_inner;
744: if (blsize) *blsize = ctx->refine_blocksize;
745: PetscFunctionReturn(0);
746: }
748: /*@
749: PEPCISSGetRefinement - Gets the values of various refinement parameters
750: in the CISS solver.
752: Not Collective
754: Input Parameter:
755: . pep - the polynomial eigensolver context
757: Output Parameters:
758: + inner - number of iterative refinement iterations (inner loop)
759: - blsize - number of iterative refinement iterations (blocksize loop)
761: Level: advanced
763: .seealso: PEPCISSSetRefinement()
764: @*/
765: PetscErrorCode PEPCISSGetRefinement(PEP pep, PetscInt *inner, PetscInt *blsize)
766: {
768: PetscUseMethod(pep,"PEPCISSGetRefinement_C",(PEP,PetscInt*,PetscInt*),(pep,inner,blsize));
769: PetscFunctionReturn(0);
770: }
772: static PetscErrorCode PEPCISSSetExtraction_CISS(PEP pep,PEPCISSExtraction extraction)
773: {
774: PEP_CISS *ctx = (PEP_CISS*)pep->data;
776: if (ctx->extraction != extraction) {
777: ctx->extraction = extraction;
778: pep->state = PEP_STATE_INITIAL;
779: }
780: PetscFunctionReturn(0);
781: }
783: /*@
784: PEPCISSSetExtraction - Sets the extraction technique used in the CISS solver.
786: Logically Collective on pep
788: Input Parameters:
789: + pep - the polynomial eigensolver context
790: - extraction - the extraction technique
792: Options Database Key:
793: . -pep_ciss_extraction - Sets the extraction technique (either 'ritz', 'hankel' or 'caa')
795: Notes:
796: By default, the Rayleigh-Ritz extraction is used (PEP_CISS_EXTRACTION_RITZ).
798: If the 'hankel' or the 'caa' option is specified (PEP_CISS_EXTRACTION_HANKEL or
799: PEP_CISS_EXTRACTION_CAA), then the Block Hankel method, or the Communication-avoiding
800: Arnoldi method, respectively, is used for extracting eigenpairs.
802: Level: advanced
804: .seealso: PEPCISSGetExtraction(), PEPCISSExtraction
805: @*/
806: PetscErrorCode PEPCISSSetExtraction(PEP pep,PEPCISSExtraction extraction)
807: {
810: PetscTryMethod(pep,"PEPCISSSetExtraction_C",(PEP,PEPCISSExtraction),(pep,extraction));
811: PetscFunctionReturn(0);
812: }
814: static PetscErrorCode PEPCISSGetExtraction_CISS(PEP pep,PEPCISSExtraction *extraction)
815: {
816: PEP_CISS *ctx = (PEP_CISS*)pep->data;
818: *extraction = ctx->extraction;
819: PetscFunctionReturn(0);
820: }
822: /*@
823: PEPCISSGetExtraction - Gets the extraction technique used in the CISS solver.
825: Not Collective
827: Input Parameter:
828: . pep - the polynomial eigensolver context
830: Output Parameters:
831: . extraction - extraction technique
833: Level: advanced
835: .seealso: PEPCISSSetExtraction() PEPCISSExtraction
836: @*/
837: PetscErrorCode PEPCISSGetExtraction(PEP pep,PEPCISSExtraction *extraction)
838: {
841: PetscUseMethod(pep,"PEPCISSGetExtraction_C",(PEP,PEPCISSExtraction*),(pep,extraction));
842: PetscFunctionReturn(0);
843: }
845: static PetscErrorCode PEPCISSGetKSPs_CISS(PEP pep,PetscInt *nsolve,KSP **ksp)
846: {
847: PEP_CISS *ctx = (PEP_CISS*)pep->data;
848: SlepcContourData contour;
849: PetscInt i,nsplit;
850: PC pc;
851: MPI_Comm child;
853: if (!ctx->contour) { /* initialize contour data structure first */
854: RGCanUseConjugates(pep->rg,ctx->isreal,&ctx->useconj);
855: SlepcContourDataCreate(ctx->useconj?ctx->N/2:ctx->N,ctx->npart,(PetscObject)pep,&ctx->contour);
856: }
857: contour = ctx->contour;
858: if (!contour->ksp) {
859: PetscMalloc1(contour->npoints,&contour->ksp);
860: PEPGetST(pep,&pep->st);
861: STGetSplitPreconditionerInfo(pep->st,&nsplit,NULL);
862: PetscSubcommGetChild(contour->subcomm,&child);
863: for (i=0;i<contour->npoints;i++) {
864: KSPCreate(child,&contour->ksp[i]);
865: PetscObjectIncrementTabLevel((PetscObject)contour->ksp[i],(PetscObject)pep,1);
866: KSPSetOptionsPrefix(contour->ksp[i],((PetscObject)pep)->prefix);
867: KSPAppendOptionsPrefix(contour->ksp[i],"pep_ciss_");
868: PetscLogObjectParent((PetscObject)pep,(PetscObject)contour->ksp[i]);
869: PetscObjectSetOptions((PetscObject)contour->ksp[i],((PetscObject)pep)->options);
870: KSPSetErrorIfNotConverged(contour->ksp[i],PETSC_TRUE);
871: KSPSetTolerances(contour->ksp[i],SlepcDefaultTol(pep->tol),PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT);
872: KSPGetPC(contour->ksp[i],&pc);
873: if (nsplit) {
874: KSPSetType(contour->ksp[i],KSPBCGS);
875: PCSetType(pc,PCBJACOBI);
876: } else {
877: KSPSetType(contour->ksp[i],KSPPREONLY);
878: PCSetType(pc,PCLU);
879: }
880: }
881: }
882: if (nsolve) *nsolve = contour->npoints;
883: if (ksp) *ksp = contour->ksp;
884: PetscFunctionReturn(0);
885: }
887: /*@C
888: PEPCISSGetKSPs - Retrieve the array of linear solver objects associated with
889: the CISS solver.
891: Not Collective
893: Input Parameter:
894: . pep - polynomial eigenvalue solver
896: Output Parameters:
897: + nsolve - number of solver objects
898: - ksp - array of linear solver object
900: Notes:
901: The number of KSP solvers is equal to the number of integration points divided by
902: the number of partitions. This value is halved in the case of real matrices with
903: a region centered at the real axis.
905: Level: advanced
907: .seealso: PEPCISSSetSizes()
908: @*/
909: PetscErrorCode PEPCISSGetKSPs(PEP pep,PetscInt *nsolve,KSP **ksp)
910: {
912: PetscUseMethod(pep,"PEPCISSGetKSPs_C",(PEP,PetscInt*,KSP**),(pep,nsolve,ksp));
913: PetscFunctionReturn(0);
914: }
916: PetscErrorCode PEPReset_CISS(PEP pep)
917: {
918: PEP_CISS *ctx = (PEP_CISS*)pep->data;
920: BVDestroy(&ctx->S);
921: BVDestroy(&ctx->V);
922: BVDestroy(&ctx->Y);
923: SlepcContourDataReset(ctx->contour);
924: MatDestroy(&ctx->J);
925: BVDestroy(&ctx->pV);
926: PetscFree(ctx->Psplit);
927: PetscFunctionReturn(0);
928: }
930: PetscErrorCode PEPSetFromOptions_CISS(PetscOptionItems *PetscOptionsObject,PEP pep)
931: {
932: PEP_CISS *ctx = (PEP_CISS*)pep->data;
933: PetscReal r1,r2;
934: PetscInt i,i1,i2,i3,i4,i5,i6,i7;
935: PetscBool b1,flg,flg2,flg3,flg4,flg5,flg6;
936: PEPCISSExtraction extraction;
938: PetscOptionsHead(PetscOptionsObject,"PEP CISS Options");
940: PEPCISSGetSizes(pep,&i1,&i2,&i3,&i4,&i5,&b1);
941: PetscOptionsInt("-pep_ciss_integration_points","Number of integration points","PEPCISSSetSizes",i1,&i1,&flg);
942: PetscOptionsInt("-pep_ciss_blocksize","Block size","PEPCISSSetSizes",i2,&i2,&flg2);
943: PetscOptionsInt("-pep_ciss_moments","Moment size","PEPCISSSetSizes",i3,&i3,&flg3);
944: PetscOptionsInt("-pep_ciss_partitions","Number of partitions","PEPCISSSetSizes",i4,&i4,&flg4);
945: PetscOptionsInt("-pep_ciss_maxblocksize","Maximum block size","PEPCISSSetSizes",i5,&i5,&flg5);
946: PetscOptionsBool("-pep_ciss_realmats","True if all coefficient matrices of P(.) are real","PEPCISSSetSizes",b1,&b1,&flg6);
947: if (flg || flg2 || flg3 || flg4 || flg5 || flg6) PEPCISSSetSizes(pep,i1,i2,i3,i4,i5,b1);
949: PEPCISSGetThreshold(pep,&r1,&r2);
950: PetscOptionsReal("-pep_ciss_delta","Threshold for numerical rank","PEPCISSSetThreshold",r1,&r1,&flg);
951: PetscOptionsReal("-pep_ciss_spurious_threshold","Threshold for the spurious eigenpairs","PEPCISSSetThreshold",r2,&r2,&flg2);
952: if (flg || flg2) PEPCISSSetThreshold(pep,r1,r2);
954: PEPCISSGetRefinement(pep,&i6,&i7);
955: PetscOptionsInt("-pep_ciss_refine_inner","Number of inner iterative refinement iterations","PEPCISSSetRefinement",i6,&i6,&flg);
956: PetscOptionsInt("-pep_ciss_refine_blocksize","Number of blocksize iterative refinement iterations","PEPCISSSetRefinement",i7,&i7,&flg2);
957: if (flg || flg2) PEPCISSSetRefinement(pep,i6,i7);
959: PetscOptionsEnum("-pep_ciss_extraction","Extraction technique","PEPCISSSetExtraction",PEPCISSExtractions,(PetscEnum)ctx->extraction,(PetscEnum*)&extraction,&flg);
960: if (flg) PEPCISSSetExtraction(pep,extraction);
962: PetscOptionsTail();
964: if (!pep->rg) PEPGetRG(pep,&pep->rg);
965: RGSetFromOptions(pep->rg); /* this is necessary here to set useconj */
966: if (!ctx->contour || !ctx->contour->ksp) PEPCISSGetKSPs(pep,NULL,NULL);
967: for (i=0;i<ctx->contour->npoints;i++) KSPSetFromOptions(ctx->contour->ksp[i]);
968: PetscSubcommSetFromOptions(ctx->contour->subcomm);
969: PetscFunctionReturn(0);
970: }
972: PetscErrorCode PEPDestroy_CISS(PEP pep)
973: {
974: PEP_CISS *ctx = (PEP_CISS*)pep->data;
976: SlepcContourDataDestroy(&ctx->contour);
977: PetscFree4(ctx->weight,ctx->omega,ctx->pp,ctx->sigma);
978: PetscFree(pep->data);
979: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetSizes_C",NULL);
980: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetSizes_C",NULL);
981: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetThreshold_C",NULL);
982: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetThreshold_C",NULL);
983: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetRefinement_C",NULL);
984: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetRefinement_C",NULL);
985: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetExtraction_C",NULL);
986: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetExtraction_C",NULL);
987: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetKSPs_C",NULL);
988: PetscFunctionReturn(0);
989: }
991: PetscErrorCode PEPView_CISS(PEP pep,PetscViewer viewer)
992: {
993: PEP_CISS *ctx = (PEP_CISS*)pep->data;
994: PetscBool isascii;
995: PetscViewer sviewer;
997: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isascii);
998: if (isascii) {
999: PetscViewerASCIIPrintf(viewer," sizes { integration points: %" PetscInt_FMT ", block size: %" PetscInt_FMT ", moment size: %" PetscInt_FMT ", partitions: %" PetscInt_FMT ", maximum block size: %" PetscInt_FMT " }\n",ctx->N,ctx->L,ctx->M,ctx->npart,ctx->L_max);
1000: if (ctx->isreal) PetscViewerASCIIPrintf(viewer," exploiting symmetry of integration points\n");
1001: PetscViewerASCIIPrintf(viewer," threshold { delta: %g, spurious threshold: %g }\n",(double)ctx->delta,(double)ctx->spurious_threshold);
1002: PetscViewerASCIIPrintf(viewer," iterative refinement { inner: %" PetscInt_FMT ", blocksize: %" PetscInt_FMT " }\n",ctx->refine_inner, ctx->refine_blocksize);
1003: PetscViewerASCIIPrintf(viewer," extraction: %s\n",PEPCISSExtractions[ctx->extraction]);
1004: if (!ctx->contour || !ctx->contour->ksp) PEPCISSGetKSPs(pep,NULL,NULL);
1005: PetscViewerASCIIPushTab(viewer);
1006: if (ctx->npart>1 && ctx->contour->subcomm) {
1007: PetscViewerGetSubViewer(viewer,ctx->contour->subcomm->child,&sviewer);
1008: if (!ctx->contour->subcomm->color) KSPView(ctx->contour->ksp[0],sviewer);
1009: PetscViewerFlush(sviewer);
1010: PetscViewerRestoreSubViewer(viewer,ctx->contour->subcomm->child,&sviewer);
1011: PetscViewerFlush(viewer);
1012: /* extra call needed because of the two calls to PetscViewerASCIIPushSynchronized() in PetscViewerGetSubViewer() */
1013: PetscViewerASCIIPopSynchronized(viewer);
1014: } else KSPView(ctx->contour->ksp[0],viewer);
1015: PetscViewerASCIIPopTab(viewer);
1016: }
1017: PetscFunctionReturn(0);
1018: }
1020: SLEPC_EXTERN PetscErrorCode PEPCreate_CISS(PEP pep)
1021: {
1022: PEP_CISS *ctx = (PEP_CISS*)pep->data;
1024: PetscNewLog(pep,&ctx);
1025: pep->data = ctx;
1026: /* set default values of parameters */
1027: ctx->N = 32;
1028: ctx->L = 16;
1029: ctx->M = ctx->N/4;
1030: ctx->delta = SLEPC_DEFAULT_TOL*1e-4;
1031: ctx->L_max = 64;
1032: ctx->spurious_threshold = PetscSqrtReal(SLEPC_DEFAULT_TOL);
1033: ctx->isreal = PETSC_FALSE;
1034: ctx->npart = 1;
1036: pep->ops->solve = PEPSolve_CISS;
1037: pep->ops->setup = PEPSetUp_CISS;
1038: pep->ops->setfromoptions = PEPSetFromOptions_CISS;
1039: pep->ops->reset = PEPReset_CISS;
1040: pep->ops->destroy = PEPDestroy_CISS;
1041: pep->ops->view = PEPView_CISS;
1043: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetSizes_C",PEPCISSSetSizes_CISS);
1044: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetSizes_C",PEPCISSGetSizes_CISS);
1045: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetThreshold_C",PEPCISSSetThreshold_CISS);
1046: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetThreshold_C",PEPCISSGetThreshold_CISS);
1047: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetRefinement_C",PEPCISSSetRefinement_CISS);
1048: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetRefinement_C",PEPCISSGetRefinement_CISS);
1049: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSSetExtraction_C",PEPCISSSetExtraction_CISS);
1050: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetExtraction_C",PEPCISSGetExtraction_CISS);
1051: PetscObjectComposeFunction((PetscObject)pep,"PEPCISSGetKSPs_C",PEPCISSGetKSPs_CISS);
1052: PetscFunctionReturn(0);
1053: }