Actual source code: mg.c

petsc-3.6.3 2015-12-03
Report Typos and Errors
  2: /*
  3:     Defines the multigrid preconditioner interface.
  4: */
  5: #include <petsc/private/pcmgimpl.h>                    /*I "petscksp.h" I*/
  6: #include <petscdm.h>

 10: PetscErrorCode PCMGMCycle_Private(PC pc,PC_MG_Levels **mglevelsin,PCRichardsonConvergedReason *reason)
 11: {
 12:   PC_MG          *mg = (PC_MG*)pc->data;
 13:   PC_MG_Levels   *mgc,*mglevels = *mglevelsin;
 15:   PetscInt       cycles = (mglevels->level == 1) ? 1 : (PetscInt) mglevels->cycles;

 18:   if (mglevels->eventsmoothsolve) {PetscLogEventBegin(mglevels->eventsmoothsolve,0,0,0,0);}
 19:   KSPSolve(mglevels->smoothd,mglevels->b,mglevels->x);  /* pre-smooth */
 20:   if (mglevels->eventsmoothsolve) {PetscLogEventEnd(mglevels->eventsmoothsolve,0,0,0,0);}
 21:   if (mglevels->level) {  /* not the coarsest grid */
 22:     if (mglevels->eventresidual) {PetscLogEventBegin(mglevels->eventresidual,0,0,0,0);}
 23:     (*mglevels->residual)(mglevels->A,mglevels->b,mglevels->x,mglevels->r);
 24:     if (mglevels->eventresidual) {PetscLogEventEnd(mglevels->eventresidual,0,0,0,0);}

 26:     /* if on finest level and have convergence criteria set */
 27:     if (mglevels->level == mglevels->levels-1 && mg->ttol && reason) {
 28:       PetscReal rnorm;
 29:       VecNorm(mglevels->r,NORM_2,&rnorm);
 30:       if (rnorm <= mg->ttol) {
 31:         if (rnorm < mg->abstol) {
 32:           *reason = PCRICHARDSON_CONVERGED_ATOL;
 33:           PetscInfo2(pc,"Linear solver has converged. Residual norm %g is less than absolute tolerance %g\n",(double)rnorm,(double)mg->abstol);
 34:         } else {
 35:           *reason = PCRICHARDSON_CONVERGED_RTOL;
 36:           PetscInfo2(pc,"Linear solver has converged. Residual norm %g is less than relative tolerance times initial residual norm %g\n",(double)rnorm,(double)mg->ttol);
 37:         }
 38:         return(0);
 39:       }
 40:     }

 42:     mgc = *(mglevelsin - 1);
 43:     if (mglevels->eventinterprestrict) {PetscLogEventBegin(mglevels->eventinterprestrict,0,0,0,0);}
 44:     MatRestrict(mglevels->restrct,mglevels->r,mgc->b);
 45:     if (mglevels->eventinterprestrict) {PetscLogEventEnd(mglevels->eventinterprestrict,0,0,0,0);}
 46:     VecSet(mgc->x,0.0);
 47:     while (cycles--) {
 48:       PCMGMCycle_Private(pc,mglevelsin-1,reason);
 49:     }
 50:     if (mglevels->eventinterprestrict) {PetscLogEventBegin(mglevels->eventinterprestrict,0,0,0,0);}
 51:     MatInterpolateAdd(mglevels->interpolate,mgc->x,mglevels->x,mglevels->x);
 52:     if (mglevels->eventinterprestrict) {PetscLogEventEnd(mglevels->eventinterprestrict,0,0,0,0);}
 53:     if (mglevels->eventsmoothsolve) {PetscLogEventBegin(mglevels->eventsmoothsolve,0,0,0,0);}
 54:     KSPSolve(mglevels->smoothu,mglevels->b,mglevels->x);    /* post smooth */
 55:     if (mglevels->eventsmoothsolve) {PetscLogEventEnd(mglevels->eventsmoothsolve,0,0,0,0);}
 56:   }
 57:   return(0);
 58: }

 62: static PetscErrorCode PCApplyRichardson_MG(PC pc,Vec b,Vec x,Vec w,PetscReal rtol,PetscReal abstol, PetscReal dtol,PetscInt its,PetscBool zeroguess,PetscInt *outits,PCRichardsonConvergedReason *reason)
 63: {
 64:   PC_MG          *mg        = (PC_MG*)pc->data;
 65:   PC_MG_Levels   **mglevels = mg->levels;
 67:   PetscInt       levels = mglevels[0]->levels,i;

 70:   /* When the DM is supplying the matrix then it will not exist until here */
 71:   for (i=0; i<levels; i++) {
 72:     if (!mglevels[i]->A) {
 73:       KSPGetOperators(mglevels[i]->smoothu,&mglevels[i]->A,NULL);
 74:       PetscObjectReference((PetscObject)mglevels[i]->A);
 75:     }
 76:   }
 77:   mglevels[levels-1]->b = b;
 78:   mglevels[levels-1]->x = x;

 80:   mg->rtol   = rtol;
 81:   mg->abstol = abstol;
 82:   mg->dtol   = dtol;
 83:   if (rtol) {
 84:     /* compute initial residual norm for relative convergence test */
 85:     PetscReal rnorm;
 86:     if (zeroguess) {
 87:       VecNorm(b,NORM_2,&rnorm);
 88:     } else {
 89:       (*mglevels[levels-1]->residual)(mglevels[levels-1]->A,b,x,w);
 90:       VecNorm(w,NORM_2,&rnorm);
 91:     }
 92:     mg->ttol = PetscMax(rtol*rnorm,abstol);
 93:   } else if (abstol) mg->ttol = abstol;
 94:   else mg->ttol = 0.0;

 96:   /* since smoother is applied to full system, not just residual we need to make sure that smoothers don't
 97:      stop prematurely due to small residual */
 98:   for (i=1; i<levels; i++) {
 99:     KSPSetTolerances(mglevels[i]->smoothu,0,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT);
100:     if (mglevels[i]->smoothu != mglevels[i]->smoothd) {
101:       KSPSetTolerances(mglevels[i]->smoothd,0,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT);
102:     }
103:   }

105:   *reason = (PCRichardsonConvergedReason)0;
106:   for (i=0; i<its; i++) {
107:     PCMGMCycle_Private(pc,mglevels+levels-1,reason);
108:     if (*reason) break;
109:   }
110:   if (!*reason) *reason = PCRICHARDSON_CONVERGED_ITS;
111:   *outits = i;
112:   return(0);
113: }

117: PetscErrorCode PCReset_MG(PC pc)
118: {
119:   PC_MG          *mg        = (PC_MG*)pc->data;
120:   PC_MG_Levels   **mglevels = mg->levels;
122:   PetscInt       i,n;

125:   if (mglevels) {
126:     n = mglevels[0]->levels;
127:     for (i=0; i<n-1; i++) {
128:       VecDestroy(&mglevels[i+1]->r);
129:       VecDestroy(&mglevels[i]->b);
130:       VecDestroy(&mglevels[i]->x);
131:       MatDestroy(&mglevels[i+1]->restrct);
132:       MatDestroy(&mglevels[i+1]->interpolate);
133:       VecDestroy(&mglevels[i+1]->rscale);
134:     }

136:     for (i=0; i<n; i++) {
137:       MatDestroy(&mglevels[i]->A);
138:       if (mglevels[i]->smoothd != mglevels[i]->smoothu) {
139:         KSPReset(mglevels[i]->smoothd);
140:       }
141:       KSPReset(mglevels[i]->smoothu);
142:     }
143:   }
144:   return(0);
145: }

149: /*@C
150:    PCMGSetLevels - Sets the number of levels to use with MG.
151:    Must be called before any other MG routine.

153:    Logically Collective on PC

155:    Input Parameters:
156: +  pc - the preconditioner context
157: .  levels - the number of levels
158: -  comms - optional communicators for each level; this is to allow solving the coarser problems
159:            on smaller sets of processors. Use NULL_OBJECT for default in Fortran

161:    Level: intermediate

163:    Notes:
164:      If the number of levels is one then the multigrid uses the -mg_levels prefix
165:   for setting the level options rather than the -mg_coarse prefix.

167: .keywords: MG, set, levels, multigrid

169: .seealso: PCMGSetType(), PCMGGetLevels()
170: @*/
171: PetscErrorCode  PCMGSetLevels(PC pc,PetscInt levels,MPI_Comm *comms)
172: {
174:   PC_MG          *mg        = (PC_MG*)pc->data;
175:   MPI_Comm       comm;
176:   PC_MG_Levels   **mglevels = mg->levels;
177:   PCMGType       mgtype     = mg->am;
178:   PetscInt       mgctype    = (PetscInt) PC_MG_CYCLE_V;
179:   PetscInt       i;
180:   PetscMPIInt    size;
181:   const char     *prefix;
182:   PC             ipc;
183:   PetscInt       n;

188:   PetscObjectGetComm((PetscObject)pc,&comm);
189:   if (mg->nlevels == levels) return(0);
190:   if (mglevels) {
191:     mgctype = mglevels[0]->cycles;
192:     /* changing the number of levels so free up the previous stuff */
193:     PCReset_MG(pc);
194:     n    = mglevels[0]->levels;
195:     for (i=0; i<n; i++) {
196:       if (mglevels[i]->smoothd != mglevels[i]->smoothu) {
197:         KSPDestroy(&mglevels[i]->smoothd);
198:       }
199:       KSPDestroy(&mglevels[i]->smoothu);
200:       PetscFree(mglevels[i]);
201:     }
202:     PetscFree(mg->levels);
203:   }

205:   mg->nlevels = levels;

207:   PetscMalloc1(levels,&mglevels);
208:   PetscLogObjectMemory((PetscObject)pc,levels*(sizeof(PC_MG*)));

210:   PCGetOptionsPrefix(pc,&prefix);

212:   mg->stageApply = 0;
213:   for (i=0; i<levels; i++) {
214:     PetscNewLog(pc,&mglevels[i]);

216:     mglevels[i]->level               = i;
217:     mglevels[i]->levels              = levels;
218:     mglevels[i]->cycles              = mgctype;
219:     mg->default_smoothu              = 2;
220:     mg->default_smoothd              = 2;
221:     mglevels[i]->eventsmoothsetup    = 0;
222:     mglevels[i]->eventsmoothsolve    = 0;
223:     mglevels[i]->eventresidual       = 0;
224:     mglevels[i]->eventinterprestrict = 0;

226:     if (comms) comm = comms[i];
227:     KSPCreate(comm,&mglevels[i]->smoothd);
228:     KSPSetErrorIfNotConverged(mglevels[i]->smoothd,pc->erroriffailure);
229:     KSPSetType(mglevels[i]->smoothd,KSPCHEBYSHEV);
230:     KSPSetConvergenceTest(mglevels[i]->smoothd,KSPConvergedSkip,NULL,NULL);
231:     KSPSetNormType(mglevels[i]->smoothd,KSP_NORM_NONE);
232:     KSPGetPC(mglevels[i]->smoothd,&ipc);
233:     PCSetType(ipc,PCSOR);
234:     PetscObjectIncrementTabLevel((PetscObject)mglevels[i]->smoothd,(PetscObject)pc,levels-i);
235:     KSPSetTolerances(mglevels[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT, i ? mg->default_smoothd : 1);
236:     KSPSetOptionsPrefix(mglevels[i]->smoothd,prefix);

238:     /* do special stuff for coarse grid */
239:     if (!i && levels > 1) {
240:       KSPAppendOptionsPrefix(mglevels[0]->smoothd,"mg_coarse_");

242:       /* coarse solve is (redundant) LU by default; set shifttype NONZERO to avoid annoying zero-pivot in LU preconditioner */
243:       KSPSetType(mglevels[0]->smoothd,KSPPREONLY);
244:       KSPGetPC(mglevels[0]->smoothd,&ipc);
245:       MPI_Comm_size(comm,&size);
246:       if (size > 1) {
247:         KSP innerksp;
248:         PC  innerpc;
249:         PCSetType(ipc,PCREDUNDANT);
250:         PCRedundantGetKSP(ipc,&innerksp);
251:         KSPGetPC(innerksp,&innerpc);
252:         PCFactorSetShiftType(innerpc,MAT_SHIFT_INBLOCKS);
253:       } else {
254:         PCSetType(ipc,PCLU);
255:         PCFactorSetShiftType(ipc,MAT_SHIFT_INBLOCKS);
256:       }
257:     } else {
258:       char tprefix[128];
259:       sprintf(tprefix,"mg_levels_%d_",(int)i);
260:       KSPAppendOptionsPrefix(mglevels[i]->smoothd,tprefix);
261:     }
262:     PetscLogObjectParent((PetscObject)pc,(PetscObject)mglevels[i]->smoothd);

264:     mglevels[i]->smoothu = mglevels[i]->smoothd;
265:     mg->rtol             = 0.0;
266:     mg->abstol           = 0.0;
267:     mg->dtol             = 0.0;
268:     mg->ttol             = 0.0;
269:     mg->cyclesperpcapply = 1;
270:   }
271:   mg->levels = mglevels;
272:   PCMGSetType(pc,mgtype);
273:   return(0);
274: }


279: PetscErrorCode PCDestroy_MG(PC pc)
280: {
282:   PC_MG          *mg        = (PC_MG*)pc->data;
283:   PC_MG_Levels   **mglevels = mg->levels;
284:   PetscInt       i,n;

287:   PCReset_MG(pc);
288:   if (mglevels) {
289:     n = mglevels[0]->levels;
290:     for (i=0; i<n; i++) {
291:       if (mglevels[i]->smoothd != mglevels[i]->smoothu) {
292:         KSPDestroy(&mglevels[i]->smoothd);
293:       }
294:       KSPDestroy(&mglevels[i]->smoothu);
295:       PetscFree(mglevels[i]);
296:     }
297:     PetscFree(mg->levels);
298:   }
299:   PetscFree(pc->data);
300:   return(0);
301: }



305: extern PetscErrorCode PCMGACycle_Private(PC,PC_MG_Levels**);
306: extern PetscErrorCode PCMGFCycle_Private(PC,PC_MG_Levels**);
307: extern PetscErrorCode PCMGKCycle_Private(PC,PC_MG_Levels**);

309: /*
310:    PCApply_MG - Runs either an additive, multiplicative, Kaskadic
311:              or full cycle of multigrid.

313:   Note:
314:   A simple wrapper which calls PCMGMCycle(),PCMGACycle(), or PCMGFCycle().
315: */
318: static PetscErrorCode PCApply_MG(PC pc,Vec b,Vec x)
319: {
320:   PC_MG          *mg        = (PC_MG*)pc->data;
321:   PC_MG_Levels   **mglevels = mg->levels;
323:   PetscInt       levels = mglevels[0]->levels,i;

326:   if (mg->stageApply) {PetscLogStagePush(mg->stageApply);}
327:   /* When the DM is supplying the matrix then it will not exist until here */
328:   for (i=0; i<levels; i++) {
329:     if (!mglevels[i]->A) {
330:       KSPGetOperators(mglevels[i]->smoothu,&mglevels[i]->A,NULL);
331:       PetscObjectReference((PetscObject)mglevels[i]->A);
332:     }
333:   }

335:   mglevels[levels-1]->b = b;
336:   mglevels[levels-1]->x = x;
337:   if (mg->am == PC_MG_MULTIPLICATIVE) {
338:     VecSet(x,0.0);
339:     for (i=0; i<mg->cyclesperpcapply; i++) {
340:       PCMGMCycle_Private(pc,mglevels+levels-1,NULL);
341:     }
342:   } else if (mg->am == PC_MG_ADDITIVE) {
343:     PCMGACycle_Private(pc,mglevels);
344:   } else if (mg->am == PC_MG_KASKADE) {
345:     PCMGKCycle_Private(pc,mglevels);
346:   } else {
347:     PCMGFCycle_Private(pc,mglevels);
348:   }
349:   if (mg->stageApply) {PetscLogStagePop();}
350:   return(0);
351: }


356: PetscErrorCode PCSetFromOptions_MG(PetscOptions *PetscOptionsObject,PC pc)
357: {
359:   PetscInt       m,levels = 1,cycles;
360:   PetscBool      flg,set;
361:   PC_MG          *mg        = (PC_MG*)pc->data;
362:   PC_MG_Levels   **mglevels = mg->levels;
363:   PCMGType       mgtype;
364:   PCMGCycleType  mgctype;

367:   PetscOptionsHead(PetscOptionsObject,"Multigrid options");
368:   if (!mg->levels) {
369:     PetscOptionsInt("-pc_mg_levels","Number of Levels","PCMGSetLevels",levels,&levels,&flg);
370:     if (!flg && pc->dm) {
371:       DMGetRefineLevel(pc->dm,&levels);
372:       levels++;
373:       mg->usedmfornumberoflevels = PETSC_TRUE;
374:     }
375:     PCMGSetLevels(pc,levels,NULL);
376:   }
377:   mglevels = mg->levels;

379:   mgctype = (PCMGCycleType) mglevels[0]->cycles;
380:   PetscOptionsEnum("-pc_mg_cycle_type","V cycle or for W-cycle","PCMGSetCycleType",PCMGCycleTypes,(PetscEnum)mgctype,(PetscEnum*)&mgctype,&flg);
381:   if (flg) {
382:     PCMGSetCycleType(pc,mgctype);
383:   }
384:   flg  = PETSC_FALSE;
385:   PetscOptionsBool("-pc_mg_galerkin","Use Galerkin process to compute coarser operators","PCMGSetGalerkin",flg,&flg,&set);
386:   if (set) {
387:     PCMGSetGalerkin(pc,flg);
388:   }
389:   PetscOptionsInt("-pc_mg_smoothup","Number of post-smoothing steps","PCMGSetNumberSmoothUp",mg->default_smoothu,&m,&flg);
390:   if (flg) {
391:     PCMGSetNumberSmoothUp(pc,m);
392:   }
393:   PetscOptionsInt("-pc_mg_smoothdown","Number of pre-smoothing steps","PCMGSetNumberSmoothDown",mg->default_smoothd,&m,&flg);
394:   if (flg) {
395:     PCMGSetNumberSmoothDown(pc,m);
396:   }
397:   mgtype = mg->am;
398:   PetscOptionsEnum("-pc_mg_type","Multigrid type","PCMGSetType",PCMGTypes,(PetscEnum)mgtype,(PetscEnum*)&mgtype,&flg);
399:   if (flg) {
400:     PCMGSetType(pc,mgtype);
401:   }
402:   if (mg->am == PC_MG_MULTIPLICATIVE) {
403:     PetscOptionsInt("-pc_mg_multiplicative_cycles","Number of cycles for each preconditioner step","PCMGMultiplicativeSetCycles",mg->cyclesperpcapply,&cycles,&flg);
404:     if (flg) {
405:       PCMGMultiplicativeSetCycles(pc,cycles);
406:     }
407:   }
408:   flg  = PETSC_FALSE;
409:   PetscOptionsBool("-pc_mg_log","Log times for each multigrid level","None",flg,&flg,NULL);
410:   if (flg) {
411:     PetscInt i;
412:     char     eventname[128];
413:     if (!mglevels) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
414:     levels = mglevels[0]->levels;
415:     for (i=0; i<levels; i++) {
416:       sprintf(eventname,"MGSetup Level %d",(int)i);
417:       PetscLogEventRegister(eventname,((PetscObject)pc)->classid,&mglevels[i]->eventsmoothsetup);
418:       sprintf(eventname,"MGSmooth Level %d",(int)i);
419:       PetscLogEventRegister(eventname,((PetscObject)pc)->classid,&mglevels[i]->eventsmoothsolve);
420:       if (i) {
421:         sprintf(eventname,"MGResid Level %d",(int)i);
422:         PetscLogEventRegister(eventname,((PetscObject)pc)->classid,&mglevels[i]->eventresidual);
423:         sprintf(eventname,"MGInterp Level %d",(int)i);
424:         PetscLogEventRegister(eventname,((PetscObject)pc)->classid,&mglevels[i]->eventinterprestrict);
425:       }
426:     }

428: #if defined(PETSC_USE_LOG)
429:     {
430:       const char    *sname = "MG Apply";
431:       PetscStageLog stageLog;
432:       PetscInt      st;

435:       PetscLogGetStageLog(&stageLog);
436:       for (st = 0; st < stageLog->numStages; ++st) {
437:         PetscBool same;

439:         PetscStrcmp(stageLog->stageInfo[st].name, sname, &same);
440:         if (same) mg->stageApply = st;
441:       }
442:       if (!mg->stageApply) {
443:         PetscLogStageRegister(sname, &mg->stageApply);
444:       }
445:     }
446: #endif
447:   }
448:   PetscOptionsTail();
449:   return(0);
450: }

452: const char *const PCMGTypes[] = {"MULTIPLICATIVE","ADDITIVE","FULL","KASKADE","PCMGType","PC_MG",0};
453: const char *const PCMGCycleTypes[] = {"invalid","v","w","PCMGCycleType","PC_MG_CYCLE",0};

455: #include <petscdraw.h>
458: PetscErrorCode PCView_MG(PC pc,PetscViewer viewer)
459: {
460:   PC_MG          *mg        = (PC_MG*)pc->data;
461:   PC_MG_Levels   **mglevels = mg->levels;
463:   PetscInt       levels = mglevels ? mglevels[0]->levels : 0,i;
464:   PetscBool      iascii,isbinary,isdraw;

467:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
468:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);
469:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
470:   if (iascii) {
471:     const char *cyclename = levels ? (mglevels[0]->cycles == PC_MG_CYCLE_V ? "v" : "w") : "unknown";
472:     PetscViewerASCIIPrintf(viewer,"  MG: type is %s, levels=%D cycles=%s\n", PCMGTypes[mg->am],levels,cyclename);
473:     if (mg->am == PC_MG_MULTIPLICATIVE) {
474:       PetscViewerASCIIPrintf(viewer,"    Cycles per PCApply=%d\n",mg->cyclesperpcapply);
475:     }
476:     if (mg->galerkin) {
477:       PetscViewerASCIIPrintf(viewer,"    Using Galerkin computed coarse grid matrices\n");
478:     } else {
479:       PetscViewerASCIIPrintf(viewer,"    Not using Galerkin computed coarse grid matrices\n");
480:     }
481:     if (mg->view){
482:       (*mg->view)(pc,viewer);
483:     }
484:     for (i=0; i<levels; i++) {
485:       if (!i) {
486:         PetscViewerASCIIPrintf(viewer,"Coarse grid solver -- level -------------------------------\n",i);
487:       } else {
488:         PetscViewerASCIIPrintf(viewer,"Down solver (pre-smoother) on level %D -------------------------------\n",i);
489:       }
490:       PetscViewerASCIIPushTab(viewer);
491:       KSPView(mglevels[i]->smoothd,viewer);
492:       PetscViewerASCIIPopTab(viewer);
493:       if (i && mglevels[i]->smoothd == mglevels[i]->smoothu) {
494:         PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) same as down solver (pre-smoother)\n");
495:       } else if (i) {
496:         PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) on level %D -------------------------------\n",i);
497:         PetscViewerASCIIPushTab(viewer);
498:         KSPView(mglevels[i]->smoothu,viewer);
499:         PetscViewerASCIIPopTab(viewer);
500:       }
501:     }
502:   } else if (isbinary) {
503:     for (i=levels-1; i>=0; i--) {
504:       KSPView(mglevels[i]->smoothd,viewer);
505:       if (i && mglevels[i]->smoothd != mglevels[i]->smoothu) {
506:         KSPView(mglevels[i]->smoothu,viewer);
507:       }
508:     }
509:   } else if (isdraw) {
510:     PetscDraw draw;
511:     PetscReal x,w,y,bottom,th;
512:     PetscViewerDrawGetDraw(viewer,0,&draw);
513:     PetscDrawGetCurrentPoint(draw,&x,&y);
514:     PetscDrawStringGetSize(draw,NULL,&th);
515:     bottom = y - th;
516:     for (i=levels-1; i>=0; i--) {
517:       if (!mglevels[i]->smoothu || (mglevels[i]->smoothu == mglevels[i]->smoothd)) {
518:         PetscDrawPushCurrentPoint(draw,x,bottom);
519:         KSPView(mglevels[i]->smoothd,viewer);
520:         PetscDrawPopCurrentPoint(draw);
521:       } else {
522:         w    = 0.5*PetscMin(1.0-x,x);
523:         PetscDrawPushCurrentPoint(draw,x+w,bottom);
524:         KSPView(mglevels[i]->smoothd,viewer);
525:         PetscDrawPopCurrentPoint(draw);
526:         PetscDrawPushCurrentPoint(draw,x-w,bottom);
527:         KSPView(mglevels[i]->smoothu,viewer);
528:         PetscDrawPopCurrentPoint(draw);
529:       }
530:       PetscDrawGetBoundingBox(draw,NULL,&bottom,NULL,NULL);
531:       bottom -= th;
532:     }
533:   }
534:   return(0);
535: }

537: #include <petsc/private/dmimpl.h>
538: #include <petsc/private/kspimpl.h>

540: /*
541:     Calls setup for the KSP on each level
542: */
545: PetscErrorCode PCSetUp_MG(PC pc)
546: {
547:   PC_MG          *mg        = (PC_MG*)pc->data;
548:   PC_MG_Levels   **mglevels = mg->levels;
550:   PetscInt       i,n = mglevels[0]->levels;
551:   PC             cpc;
552:   PetscBool      preonly,lu,redundant,cholesky,svd,dump = PETSC_FALSE,opsset,use_amat,missinginterpolate = PETSC_FALSE;
553:   Mat            dA,dB;
554:   Vec            tvec;
555:   DM             *dms;
556:   PetscViewer    viewer = 0;

559:   /* FIX: Move this to PCSetFromOptions_MG? */
560:   if (mg->usedmfornumberoflevels) {
561:     PetscInt levels;
562:     DMGetRefineLevel(pc->dm,&levels);
563:     levels++;
564:     if (levels > n) { /* the problem is now being solved on a finer grid */
565:       PCMGSetLevels(pc,levels,NULL);
566:       n        = levels;
567:       PCSetFromOptions(pc); /* it is bad to call this here, but otherwise will never be called for the new hierarchy */
568:       mglevels =  mg->levels;
569:     }
570:   }
571:   KSPGetPC(mglevels[0]->smoothd,&cpc);


574:   /* If user did not provide fine grid operators OR operator was not updated since last global KSPSetOperators() */
575:   /* so use those from global PC */
576:   /* Is this what we always want? What if user wants to keep old one? */
577:   KSPGetOperatorsSet(mglevels[n-1]->smoothd,NULL,&opsset);
578:   if (opsset) {
579:     Mat mmat;
580:     KSPGetOperators(mglevels[n-1]->smoothd,NULL,&mmat);
581:     if (mmat == pc->pmat) opsset = PETSC_FALSE;
582:   }

584:   if (!opsset) {
585:     PCGetUseAmat(pc,&use_amat);
586:     if(use_amat){
587:       PetscInfo(pc,"Using outer operators to define finest grid operator \n  because PCMGGetSmoother(pc,nlevels-1,&ksp);KSPSetOperators(ksp,...); was not called.\n");
588:       KSPSetOperators(mglevels[n-1]->smoothd,pc->mat,pc->pmat);
589:     }
590:     else {
591:       PetscInfo(pc,"Using matrix (pmat) operators to define finest grid operator \n  because PCMGGetSmoother(pc,nlevels-1,&ksp);KSPSetOperators(ksp,...); was not called.\n");
592:       KSPSetOperators(mglevels[n-1]->smoothd,pc->pmat,pc->pmat);
593:     }
594:   }

596:   for (i=n-1; i>0; i--) {
597:     if (!(mglevels[i]->interpolate || mglevels[i]->restrct)) {
598:       missinginterpolate = PETSC_TRUE;
599:       continue;
600:     }
601:   }
602:   /*
603:    Skipping if user has provided all interpolation/restriction needed (since DM might not be able to produce them (when coming from SNES/TS)
604:    Skipping for galerkin==2 (externally managed hierarchy such as ML and GAMG). Cleaner logic here would be great. Wrap ML/GAMG as DMs?
605:   */
606:   if (missinginterpolate && pc->dm && mg->galerkin != 2 && !pc->setupcalled) {
607:     /* construct the interpolation from the DMs */
608:     Mat p;
609:     Vec rscale;
610:     PetscMalloc1(n,&dms);
611:     dms[n-1] = pc->dm;
612:     /* Separately create them so we do not get DMKSP interference between levels */
613:     for (i=n-2; i>-1; i--) {DMCoarsen(dms[i+1],MPI_COMM_NULL,&dms[i]);}
614:     for (i=n-2; i>-1; i--) {
615:       DMKSP kdm;
616:       KSPSetDM(mglevels[i]->smoothd,dms[i]);
617:       if (mg->galerkin) {KSPSetDMActive(mglevels[i]->smoothd,PETSC_FALSE);}
618:       DMGetDMKSPWrite(dms[i],&kdm);
619:       /* Ugly hack so that the next KSPSetUp() will use the RHS that we set. A better fix is to change dmActive to take
620:        * a bitwise OR of computing the matrix, RHS, and initial iterate. */
621:       kdm->ops->computerhs = NULL;
622:       kdm->rhsctx          = NULL;
623:       if (!mglevels[i+1]->interpolate) {
624:         DMCreateInterpolation(dms[i],dms[i+1],&p,&rscale);
625:         PCMGSetInterpolation(pc,i+1,p);
626:         if (rscale) {PCMGSetRScale(pc,i+1,rscale);}
627:         VecDestroy(&rscale);
628:         MatDestroy(&p);
629:       }
630:     }

632:     for (i=n-2; i>-1; i--) {DMDestroy(&dms[i]);}
633:     PetscFree(dms);
634:   }

636:   if (pc->dm && !pc->setupcalled) {
637:     /* finest smoother also gets DM but it is not active, independent of whether galerkin==2 */
638:     KSPSetDM(mglevels[n-1]->smoothd,pc->dm);
639:     KSPSetDMActive(mglevels[n-1]->smoothd,PETSC_FALSE);
640:   }

642:   if (mg->galerkin == 1) {
643:     Mat B;
644:     /* currently only handle case where mat and pmat are the same on coarser levels */
645:     KSPGetOperators(mglevels[n-1]->smoothd,&dA,&dB);
646:     if (!pc->setupcalled) {
647:       for (i=n-2; i>-1; i--) {
648:         if (!mglevels[i+1]->restrct && !mglevels[i+1]->interpolate) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must provide interpolation or restriction for each MG level except level 0");
649:         if (!mglevels[i+1]->interpolate) {
650:           PCMGSetInterpolation(pc,i+1,mglevels[i+1]->restrct);
651:         }
652:         if (!mglevels[i+1]->restrct) {
653:           PCMGSetRestriction(pc,i+1,mglevels[i+1]->interpolate);
654:         }
655:         if (mglevels[i+1]->interpolate == mglevels[i+1]->restrct) {
656:           MatPtAP(dB,mglevels[i+1]->interpolate,MAT_INITIAL_MATRIX,1.0,&B);
657:         } else {
658:           MatMatMatMult(mglevels[i+1]->restrct,dB,mglevels[i+1]->interpolate,MAT_INITIAL_MATRIX,1.0,&B);
659:         }
660:         KSPSetOperators(mglevels[i]->smoothd,B,B);
661:         if (i != n-2) {PetscObjectDereference((PetscObject)dB);}
662:         dB = B;
663:       }
664:       if (n > 1) {PetscObjectDereference((PetscObject)dB);}
665:     } else {
666:       for (i=n-2; i>-1; i--) {
667:         if (!mglevels[i+1]->restrct && !mglevels[i+1]->interpolate) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must provide interpolation or restriction for each MG level except level 0");
668:         if (!mglevels[i+1]->interpolate) {
669:           PCMGSetInterpolation(pc,i+1,mglevels[i+1]->restrct);
670:         }
671:         if (!mglevels[i+1]->restrct) {
672:           PCMGSetRestriction(pc,i+1,mglevels[i+1]->interpolate);
673:         }
674:         KSPGetOperators(mglevels[i]->smoothd,NULL,&B);
675:         if (mglevels[i+1]->interpolate == mglevels[i+1]->restrct) {
676:           MatPtAP(dB,mglevels[i+1]->interpolate,MAT_REUSE_MATRIX,1.0,&B);
677:         } else {
678:           MatMatMatMult(mglevels[i+1]->restrct,dB,mglevels[i+1]->interpolate,MAT_REUSE_MATRIX,1.0,&B);
679:         }
680:         KSPSetOperators(mglevels[i]->smoothd,B,B);
681:         dB   = B;
682:       }
683:     }
684:   } else if (!mg->galerkin && pc->dm && pc->dm->x) {
685:     /* need to restrict Jacobian location to coarser meshes for evaluation */
686:     for (i=n-2; i>-1; i--) {
687:       Mat R;
688:       Vec rscale;
689:       if (!mglevels[i]->smoothd->dm->x) {
690:         Vec *vecs;
691:         KSPCreateVecs(mglevels[i]->smoothd,1,&vecs,0,NULL);

693:         mglevels[i]->smoothd->dm->x = vecs[0];

695:         PetscFree(vecs);
696:       }
697:       PCMGGetRestriction(pc,i+1,&R);
698:       PCMGGetRScale(pc,i+1,&rscale);
699:       MatRestrict(R,mglevels[i+1]->smoothd->dm->x,mglevels[i]->smoothd->dm->x);
700:       VecPointwiseMult(mglevels[i]->smoothd->dm->x,mglevels[i]->smoothd->dm->x,rscale);
701:     }
702:   }
703:   if (!mg->galerkin && pc->dm) {
704:     for (i=n-2; i>=0; i--) {
705:       DM  dmfine,dmcoarse;
706:       Mat Restrict,Inject;
707:       Vec rscale;
708:       KSPGetDM(mglevels[i+1]->smoothd,&dmfine);
709:       KSPGetDM(mglevels[i]->smoothd,&dmcoarse);
710:       PCMGGetRestriction(pc,i+1,&Restrict);
711:       PCMGGetRScale(pc,i+1,&rscale);
712:       Inject = NULL;      /* Callback should create it if it needs Injection */
713:       DMRestrict(dmfine,Restrict,rscale,Inject,dmcoarse);
714:     }
715:   }

717:   if (!pc->setupcalled) {
718:     for (i=0; i<n; i++) {
719:       KSPSetFromOptions(mglevels[i]->smoothd);
720:     }
721:     for (i=1; i<n; i++) {
722:       if (mglevels[i]->smoothu && (mglevels[i]->smoothu != mglevels[i]->smoothd)) {
723:         KSPSetFromOptions(mglevels[i]->smoothu);
724:       }
725:     }
726:     for (i=1; i<n; i++) {
727:       PCMGGetInterpolation(pc,i,&mglevels[i]->interpolate);
728:       PCMGGetRestriction(pc,i,&mglevels[i]->restrct);
729:     }
730:     for (i=0; i<n-1; i++) {
731:       if (!mglevels[i]->b) {
732:         Vec *vec;
733:         KSPCreateVecs(mglevels[i]->smoothd,1,&vec,0,NULL);
734:         PCMGSetRhs(pc,i,*vec);
735:         VecDestroy(vec);
736:         PetscFree(vec);
737:       }
738:       if (!mglevels[i]->r && i) {
739:         VecDuplicate(mglevels[i]->b,&tvec);
740:         PCMGSetR(pc,i,tvec);
741:         VecDestroy(&tvec);
742:       }
743:       if (!mglevels[i]->x) {
744:         VecDuplicate(mglevels[i]->b,&tvec);
745:         PCMGSetX(pc,i,tvec);
746:         VecDestroy(&tvec);
747:       }
748:     }
749:     if (n != 1 && !mglevels[n-1]->r) {
750:       /* PCMGSetR() on the finest level if user did not supply it */
751:       Vec *vec;
752:       KSPCreateVecs(mglevels[n-1]->smoothd,1,&vec,0,NULL);
753:       PCMGSetR(pc,n-1,*vec);
754:       VecDestroy(vec);
755:       PetscFree(vec);
756:     }
757:   }

759:   if (pc->dm) {
760:     /* need to tell all the coarser levels to rebuild the matrix using the DM for that level */
761:     for (i=0; i<n-1; i++) {
762:       if (mglevels[i]->smoothd->setupstage != KSP_SETUP_NEW) mglevels[i]->smoothd->setupstage = KSP_SETUP_NEWMATRIX;
763:     }
764:   }

766:   for (i=1; i<n; i++) {
767:     if (mglevels[i]->smoothu == mglevels[i]->smoothd || mg->am == PC_MG_FULL || mg->am == PC_MG_KASKADE || mg->cyclesperpcapply > 1){
768:       /* if doing only down then initial guess is zero */
769:       KSPSetInitialGuessNonzero(mglevels[i]->smoothd,PETSC_TRUE);
770:     }
771:     if (mglevels[i]->eventsmoothsetup) {PetscLogEventBegin(mglevels[i]->eventsmoothsetup,0,0,0,0);}
772:     KSPSetUp(mglevels[i]->smoothd);
773:     if (mglevels[i]->eventsmoothsetup) {PetscLogEventEnd(mglevels[i]->eventsmoothsetup,0,0,0,0);}
774:     if (!mglevels[i]->residual) {
775:       Mat mat;
776:       KSPGetOperators(mglevels[i]->smoothd,NULL,&mat);
777:       PCMGSetResidual(pc,i,PCMGResidualDefault,mat);
778:     }
779:   }
780:   for (i=1; i<n; i++) {
781:     if (mglevels[i]->smoothu && mglevels[i]->smoothu != mglevels[i]->smoothd) {
782:       Mat          downmat,downpmat;

784:       /* check if operators have been set for up, if not use down operators to set them */
785:       KSPGetOperatorsSet(mglevels[i]->smoothu,&opsset,NULL);
786:       if (!opsset) {
787:         KSPGetOperators(mglevels[i]->smoothd,&downmat,&downpmat);
788:         KSPSetOperators(mglevels[i]->smoothu,downmat,downpmat);
789:       }

791:       KSPSetInitialGuessNonzero(mglevels[i]->smoothu,PETSC_TRUE);
792:       if (mglevels[i]->eventsmoothsetup) {PetscLogEventBegin(mglevels[i]->eventsmoothsetup,0,0,0,0);}
793:       KSPSetUp(mglevels[i]->smoothu);
794:       if (mglevels[i]->eventsmoothsetup) {PetscLogEventEnd(mglevels[i]->eventsmoothsetup,0,0,0,0);}
795:     }
796:   }

798:   /*
799:       If coarse solver is not direct method then DO NOT USE preonly
800:   */
801:   PetscObjectTypeCompare((PetscObject)mglevels[0]->smoothd,KSPPREONLY,&preonly);
802:   if (preonly) {
803:     PetscObjectTypeCompare((PetscObject)cpc,PCLU,&lu);
804:     PetscObjectTypeCompare((PetscObject)cpc,PCREDUNDANT,&redundant);
805:     PetscObjectTypeCompare((PetscObject)cpc,PCCHOLESKY,&cholesky);
806:     PetscObjectTypeCompare((PetscObject)cpc,PCSVD,&svd);
807:     if (!lu && !redundant && !cholesky && !svd) {
808:       KSPSetType(mglevels[0]->smoothd,KSPGMRES);
809:     }
810:   }

812:   if (mglevels[0]->eventsmoothsetup) {PetscLogEventBegin(mglevels[0]->eventsmoothsetup,0,0,0,0);}
813:   KSPSetUp(mglevels[0]->smoothd);
814:   if (mglevels[0]->eventsmoothsetup) {PetscLogEventEnd(mglevels[0]->eventsmoothsetup,0,0,0,0);}

816:   /*
817:      Dump the interpolation/restriction matrices plus the
818:    Jacobian/stiffness on each level. This allows MATLAB users to
819:    easily check if the Galerkin condition A_c = R A_f R^T is satisfied.

821:    Only support one or the other at the same time.
822:   */
823: #if defined(PETSC_USE_SOCKET_VIEWER)
824:   PetscOptionsGetBool(((PetscObject)pc)->prefix,"-pc_mg_dump_matlab",&dump,NULL);
825:   if (dump) viewer = PETSC_VIEWER_SOCKET_(PetscObjectComm((PetscObject)pc));
826:   dump = PETSC_FALSE;
827: #endif
828:   PetscOptionsGetBool(((PetscObject)pc)->prefix,"-pc_mg_dump_binary",&dump,NULL);
829:   if (dump) viewer = PETSC_VIEWER_BINARY_(PetscObjectComm((PetscObject)pc));

831:   if (viewer) {
832:     for (i=1; i<n; i++) {
833:       MatView(mglevels[i]->restrct,viewer);
834:     }
835:     for (i=0; i<n; i++) {
836:       KSPGetPC(mglevels[i]->smoothd,&pc);
837:       MatView(pc->mat,viewer);
838:     }
839:   }
840:   return(0);
841: }

843: /* -------------------------------------------------------------------------------------*/

847: /*@
848:    PCMGGetLevels - Gets the number of levels to use with MG.

850:    Not Collective

852:    Input Parameter:
853: .  pc - the preconditioner context

855:    Output parameter:
856: .  levels - the number of levels

858:    Level: advanced

860: .keywords: MG, get, levels, multigrid

862: .seealso: PCMGSetLevels()
863: @*/
864: PetscErrorCode  PCMGGetLevels(PC pc,PetscInt *levels)
865: {
866:   PC_MG *mg = (PC_MG*)pc->data;

871:   *levels = mg->nlevels;
872:   return(0);
873: }

877: /*@
878:    PCMGSetType - Determines the form of multigrid to use:
879:    multiplicative, additive, full, or the Kaskade algorithm.

881:    Logically Collective on PC

883:    Input Parameters:
884: +  pc - the preconditioner context
885: -  form - multigrid form, one of PC_MG_MULTIPLICATIVE, PC_MG_ADDITIVE,
886:    PC_MG_FULL, PC_MG_KASKADE

888:    Options Database Key:
889: .  -pc_mg_type <form> - Sets <form>, one of multiplicative,
890:    additive, full, kaskade

892:    Level: advanced

894: .keywords: MG, set, method, multiplicative, additive, full, Kaskade, multigrid

896: .seealso: PCMGSetLevels()
897: @*/
898: PetscErrorCode  PCMGSetType(PC pc,PCMGType form)
899: {
900:   PC_MG *mg = (PC_MG*)pc->data;

905:   mg->am = form;
906:   if (form == PC_MG_MULTIPLICATIVE) pc->ops->applyrichardson = PCApplyRichardson_MG;
907:   else pc->ops->applyrichardson = NULL;
908:   return(0);
909: }

913: /*@
914:    PCMGGetType - Determines the form of multigrid to use:
915:    multiplicative, additive, full, or the Kaskade algorithm.

917:    Logically Collective on PC

919:    Input Parameter:
920: .  pc - the preconditioner context

922:    Output Parameter:
923: .  type - one of PC_MG_MULTIPLICATIVE, PC_MG_ADDITIVE,PC_MG_FULL, PC_MG_KASKADE


926:    Level: advanced

928: .keywords: MG, set, method, multiplicative, additive, full, Kaskade, multigrid

930: .seealso: PCMGSetLevels()
931: @*/
932: PetscErrorCode  PCMGGetType(PC pc,PCMGType *type)
933: {
934:   PC_MG *mg = (PC_MG*)pc->data;

938:   *type = mg->am;
939:   return(0);
940: }

944: /*@
945:    PCMGSetCycleType - Sets the type cycles to use.  Use PCMGSetCycleTypeOnLevel() for more
946:    complicated cycling.

948:    Logically Collective on PC

950:    Input Parameters:
951: +  pc - the multigrid context
952: -  PC_MG_CYCLE_V or PC_MG_CYCLE_W

954:    Options Database Key:
955: .  -pc_mg_cycle_type <v,w>

957:    Level: advanced

959: .keywords: MG, set, cycles, V-cycle, W-cycle, multigrid

961: .seealso: PCMGSetCycleTypeOnLevel()
962: @*/
963: PetscErrorCode  PCMGSetCycleType(PC pc,PCMGCycleType n)
964: {
965:   PC_MG        *mg        = (PC_MG*)pc->data;
966:   PC_MG_Levels **mglevels = mg->levels;
967:   PetscInt     i,levels;

971:   if (!mglevels) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
973:   levels = mglevels[0]->levels;

975:   for (i=0; i<levels; i++) mglevels[i]->cycles = n;
976:   return(0);
977: }

981: /*@
982:    PCMGMultiplicativeSetCycles - Sets the number of cycles to use for each preconditioner step
983:          of multigrid when PCMGType of PC_MG_MULTIPLICATIVE is used

985:    Logically Collective on PC

987:    Input Parameters:
988: +  pc - the multigrid context
989: -  n - number of cycles (default is 1)

991:    Options Database Key:
992: .  -pc_mg_multiplicative_cycles n

994:    Level: advanced

996:    Notes: This is not associated with setting a v or w cycle, that is set with PCMGSetCycleType()

998: .keywords: MG, set, cycles, V-cycle, W-cycle, multigrid

1000: .seealso: PCMGSetCycleTypeOnLevel(), PCMGSetCycleType()
1001: @*/
1002: PetscErrorCode  PCMGMultiplicativeSetCycles(PC pc,PetscInt n)
1003: {
1004:   PC_MG        *mg        = (PC_MG*)pc->data;

1009:   mg->cyclesperpcapply = n;
1010:   return(0);
1011: }

1015: PetscErrorCode PCMGSetGalerkin_MG(PC pc,PetscBool use)
1016: {
1017:   PC_MG *mg = (PC_MG*)pc->data;

1020:   mg->galerkin = use ? 1 : 0;
1021:   return(0);
1022: }

1026: /*@
1027:    PCMGSetGalerkin - Causes the coarser grid matrices to be computed from the
1028:       finest grid via the Galerkin process: A_i-1 = r_i * A_i * p_i

1030:    Logically Collective on PC

1032:    Input Parameters:
1033: +  pc - the multigrid context
1034: -  use - PETSC_TRUE to use the Galerkin process to compute coarse-level operators

1036:    Options Database Key:
1037: .  -pc_mg_galerkin <true,false>

1039:    Level: intermediate

1041:    Notes: Some codes that use PCMG such as PCGAMG use Galerkin internally while constructing the hierarchy and thus do not
1042:      use the PCMG construction of the coarser grids.

1044: .keywords: MG, set, Galerkin

1046: .seealso: PCMGGetGalerkin()

1048: @*/
1049: PetscErrorCode PCMGSetGalerkin(PC pc,PetscBool use)
1050: {

1055:   PetscTryMethod(pc,"PCMGSetGalerkin_C",(PC,PetscBool),(pc,use));
1056:   return(0);
1057: }

1061: /*@
1062:    PCMGGetGalerkin - Checks if Galerkin multigrid is being used, i.e.
1063:       A_i-1 = r_i * A_i * p_i

1065:    Not Collective

1067:    Input Parameter:
1068: .  pc - the multigrid context

1070:    Output Parameter:
1071: .  galerkin - PETSC_TRUE or PETSC_FALSE

1073:    Options Database Key:
1074: .  -pc_mg_galerkin

1076:    Level: intermediate

1078: .keywords: MG, set, Galerkin

1080: .seealso: PCMGSetGalerkin()

1082: @*/
1083: PetscErrorCode  PCMGGetGalerkin(PC pc,PetscBool  *galerkin)
1084: {
1085:   PC_MG *mg = (PC_MG*)pc->data;

1089:   *galerkin = (PetscBool)mg->galerkin;
1090:   return(0);
1091: }

1095: /*@
1096:    PCMGSetNumberSmoothDown - Sets the number of pre-smoothing steps to
1097:    use on all levels. Use PCMGGetSmootherDown() to set different
1098:    pre-smoothing steps on different levels.

1100:    Logically Collective on PC

1102:    Input Parameters:
1103: +  mg - the multigrid context
1104: -  n - the number of smoothing steps

1106:    Options Database Key:
1107: .  -pc_mg_smoothdown <n> - Sets number of pre-smoothing steps

1109:    Level: advanced

1111: .keywords: MG, smooth, down, pre-smoothing, steps, multigrid

1113: .seealso: PCMGSetNumberSmoothUp()
1114: @*/
1115: PetscErrorCode  PCMGSetNumberSmoothDown(PC pc,PetscInt n)
1116: {
1117:   PC_MG          *mg        = (PC_MG*)pc->data;
1118:   PC_MG_Levels   **mglevels = mg->levels;
1120:   PetscInt       i,levels;

1124:   if (!mglevels) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
1126:   levels = mglevels[0]->levels;

1128:   for (i=1; i<levels; i++) {
1129:     /* make sure smoother up and down are different */
1130:     PCMGGetSmootherUp(pc,i,NULL);
1131:     KSPSetTolerances(mglevels[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);

1133:     mg->default_smoothd = n;
1134:   }
1135:   return(0);
1136: }

1140: /*@
1141:    PCMGSetNumberSmoothUp - Sets the number of post-smoothing steps to use
1142:    on all levels. Use PCMGGetSmootherUp() to set different numbers of
1143:    post-smoothing steps on different levels.

1145:    Logically Collective on PC

1147:    Input Parameters:
1148: +  mg - the multigrid context
1149: -  n - the number of smoothing steps

1151:    Options Database Key:
1152: .  -pc_mg_smoothup <n> - Sets number of post-smoothing steps

1154:    Level: advanced

1156:    Note: this does not set a value on the coarsest grid, since we assume that
1157:     there is no separate smooth up on the coarsest grid.

1159: .keywords: MG, smooth, up, post-smoothing, steps, multigrid

1161: .seealso: PCMGSetNumberSmoothDown()
1162: @*/
1163: PetscErrorCode  PCMGSetNumberSmoothUp(PC pc,PetscInt n)
1164: {
1165:   PC_MG          *mg        = (PC_MG*)pc->data;
1166:   PC_MG_Levels   **mglevels = mg->levels;
1168:   PetscInt       i,levels;

1172:   if (!mglevels) SETERRQ(PetscObjectComm((PetscObject)pc),PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
1174:   levels = mglevels[0]->levels;

1176:   for (i=1; i<levels; i++) {
1177:     /* make sure smoother up and down are different */
1178:     PCMGGetSmootherUp(pc,i,NULL);
1179:     KSPSetTolerances(mglevels[i]->smoothu,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);

1181:     mg->default_smoothu = n;
1182:   }
1183:   return(0);
1184: }

1186: /* ----------------------------------------------------------------------------------------*/

1188: /*MC
1189:    PCMG - Use multigrid preconditioning. This preconditioner requires you provide additional
1190:     information about the coarser grid matrices and restriction/interpolation operators.

1192:    Options Database Keys:
1193: +  -pc_mg_levels <nlevels> - number of levels including finest
1194: .  -pc_mg_cycles <v,w> -
1195: .  -pc_mg_smoothup <n> - number of smoothing steps after interpolation
1196: .  -pc_mg_smoothdown <n> - number of smoothing steps before applying restriction operator
1197: .  -pc_mg_type <additive,multiplicative,full,kaskade> - multiplicative is the default
1198: .  -pc_mg_log - log information about time spent on each level of the solver
1199: .  -pc_mg_galerkin - use Galerkin process to compute coarser operators, i.e. Acoarse = R A R'
1200: .  -pc_mg_multiplicative_cycles - number of cycles to use as the preconditioner (defaults to 1)
1201: .  -pc_mg_dump_matlab - dumps the matrices for each level and the restriction/interpolation matrices
1202:                         to the Socket viewer for reading from MATLAB.
1203: -  -pc_mg_dump_binary - dumps the matrices for each level and the restriction/interpolation matrices
1204:                         to the binary output file called binaryoutput

1206:    Notes: By default this uses GMRES on the fine grid smoother so this should be used with KSPFGMRES or the smoother changed to not use GMRES

1208:        When run with a single level the smoother options are used on that level NOT the coarse grid solver options

1210:    Level: intermediate

1212:    Concepts: multigrid/multilevel

1214: .seealso:  PCCreate(), PCSetType(), PCType (for list of available types), PC, PCMGType, PCEXOTIC, PCGAMG, PCML, PCHYPRE
1215:            PCMGSetLevels(), PCMGGetLevels(), PCMGSetType(), PCMGSetCycleType(), PCMGSetNumberSmoothDown(),
1216:            PCMGSetNumberSmoothUp(), PCMGGetCoarseSolve(), PCMGSetResidual(), PCMGSetInterpolation(),
1217:            PCMGSetRestriction(), PCMGGetSmoother(), PCMGGetSmootherUp(), PCMGGetSmootherDown(),
1218:            PCMGSetCycleTypeOnLevel(), PCMGSetRhs(), PCMGSetX(), PCMGSetR()
1219: M*/

1223: PETSC_EXTERN PetscErrorCode PCCreate_MG(PC pc)
1224: {
1225:   PC_MG          *mg;

1229:   PetscNewLog(pc,&mg);
1230:   pc->data    = (void*)mg;
1231:   mg->nlevels = -1;
1232:   mg->am      = PC_MG_MULTIPLICATIVE;

1234:   pc->useAmat = PETSC_TRUE;

1236:   pc->ops->apply          = PCApply_MG;
1237:   pc->ops->setup          = PCSetUp_MG;
1238:   pc->ops->reset          = PCReset_MG;
1239:   pc->ops->destroy        = PCDestroy_MG;
1240:   pc->ops->setfromoptions = PCSetFromOptions_MG;
1241:   pc->ops->view           = PCView_MG;

1243:   PetscObjectComposeFunction((PetscObject)pc,"PCMGSetGalerkin_C",PCMGSetGalerkin_MG);
1244:   return(0);
1245: }