xref: /petsc/src/ksp/pc/impls/mg/mg.c (revision 38f152fed54d3fbfd94b4b1def17d9130b5826b8)
1 #define PETSCKSP_DLL
2 
3 /*
4     Defines the multigrid preconditioner interface.
5 */
6 #include "src/ksp/pc/impls/mg/mgimpl.h"                    /*I "petscmg.h" I*/
7 
8 
9 /*
10        MGMCycle_Private - Given an MG structure created with MGCreate() runs
11                   one multiplicative cycle down through the levels and
12                   back up.
13 
14     Input Parameter:
15 .   mg - structure created with  MGCreate().
16 */
17 #undef __FUNCT__
18 #define __FUNCT__ "MGMCycle_Private"
19 PetscErrorCode MGMCycle_Private(MG *mglevels,PetscTruth *converged)
20 {
21   MG             mg = *mglevels,mgc;
22   PetscErrorCode ierr;
23   PetscInt       cycles = mg->cycles;
24   PetscScalar    zero = 0.0;
25 
26   PetscFunctionBegin;
27   if (converged) *converged = PETSC_FALSE;
28 
29   if (mg->eventsolve) {ierr = PetscLogEventBegin(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
30   ierr = KSPSolve(mg->smoothd,mg->b,mg->x);CHKERRQ(ierr);
31   if (mg->eventsolve) {ierr = PetscLogEventEnd(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
32   if (mg->level) {  /* not the coarsest grid */
33     ierr = (*mg->residual)(mg->A,mg->b,mg->x,mg->r);CHKERRQ(ierr);
34 
35     /* if on finest level and have convergence criteria set */
36     if (mg->level == mg->levels-1 && mg->ttol) {
37       PetscReal rnorm;
38       ierr = VecNorm(mg->r,NORM_2,&rnorm);CHKERRQ(ierr);
39       if (rnorm <= mg->ttol) {
40         *converged = PETSC_TRUE;
41         if (rnorm < mg->abstol) {
42           ierr = PetscLogInfo((0,"MGMCycle_Private:Linear solver has converged. Residual norm %g is less than absolute tolerance %g\n",rnorm,mg->abstol));CHKERRQ(ierr);
43         } else {
44           ierr = PetscLogInfo((0,"MGMCycle_Private:Linear solver has converged. Residual norm %g is less than relative tolerance times initial residual norm %g\n",rnorm,mg->ttol));CHKERRQ(ierr);
45         }
46         PetscFunctionReturn(0);
47       }
48     }
49 
50     mgc = *(mglevels - 1);
51     ierr = MatRestrict(mg->restrct,mg->r,mgc->b);CHKERRQ(ierr);
52     ierr = VecSet(&zero,mgc->x);CHKERRQ(ierr);
53     while (cycles--) {
54       ierr = MGMCycle_Private(mglevels-1,converged);CHKERRQ(ierr);
55     }
56     ierr = MatInterpolateAdd(mg->interpolate,mgc->x,mg->x,mg->x);CHKERRQ(ierr);
57     if (mg->eventsolve) {ierr = PetscLogEventBegin(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
58     ierr = KSPSolve(mg->smoothu,mg->b,mg->x);CHKERRQ(ierr);
59     if (mg->eventsolve) {ierr = PetscLogEventEnd(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
60   }
61   PetscFunctionReturn(0);
62 }
63 
64 /*
65        MGCreate_Private - Creates a MG structure for use with the
66                multigrid code. Level 0 is the coarsest. (But the
67                finest level is stored first in the array).
68 
69 */
70 #undef __FUNCT__
71 #define __FUNCT__ "MGCreate_Private"
72 static PetscErrorCode MGCreate_Private(MPI_Comm comm,PetscInt levels,PC pc,MPI_Comm *comms,MG **result)
73 {
74   MG             *mg;
75   PetscErrorCode ierr;
76   PetscInt       i;
77   PetscMPIInt    size;
78   char           *prefix;
79   PC             ipc;
80 
81   PetscFunctionBegin;
82   ierr = PetscMalloc(levels*sizeof(MG),&mg);CHKERRQ(ierr);
83   ierr = PetscLogObjectMemory(pc,levels*(sizeof(MG)+sizeof(struct _MG)));CHKERRQ(ierr);
84 
85   ierr = PCGetOptionsPrefix(pc,&prefix);CHKERRQ(ierr);
86 
87   for (i=0; i<levels; i++) {
88     ierr = PetscNew(struct _MG,&mg[i]);CHKERRQ(ierr);
89     mg[i]->level           = i;
90     mg[i]->levels          = levels;
91     mg[i]->cycles          = 1;
92     mg[i]->galerkin        = PETSC_FALSE;
93     mg[i]->galerkinused    = PETSC_FALSE;
94     mg[i]->default_smoothu = 1;
95     mg[i]->default_smoothd = 1;
96 
97     if (comms) comm = comms[i];
98     ierr = KSPCreate(comm,&mg[i]->smoothd);CHKERRQ(ierr);
99     ierr = KSPSetTolerances(mg[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT, mg[i]->default_smoothd);CHKERRQ(ierr);
100     ierr = KSPSetOptionsPrefix(mg[i]->smoothd,prefix);CHKERRQ(ierr);
101 
102     /* do special stuff for coarse grid */
103     if (!i && levels > 1) {
104       ierr = KSPAppendOptionsPrefix(mg[0]->smoothd,"mg_coarse_");CHKERRQ(ierr);
105 
106       /* coarse solve is (redundant) LU by default */
107       ierr = KSPSetType(mg[0]->smoothd,KSPPREONLY);CHKERRQ(ierr);
108       ierr = KSPGetPC(mg[0]->smoothd,&ipc);CHKERRQ(ierr);
109       ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr);
110       if (size > 1) {
111         ierr = PCSetType(ipc,PCREDUNDANT);CHKERRQ(ierr);
112         ierr = PCRedundantGetPC(ipc,&ipc);CHKERRQ(ierr);
113       }
114       ierr = PCSetType(ipc,PCLU);CHKERRQ(ierr);
115 
116     } else {
117       char tprefix[128];
118       sprintf(tprefix,"mg_levels_%d_",(int)i);
119       ierr = KSPAppendOptionsPrefix(mg[i]->smoothd,tprefix);CHKERRQ(ierr);
120     }
121     ierr = PetscLogObjectParent(pc,mg[i]->smoothd);CHKERRQ(ierr);
122     mg[i]->smoothu         = mg[i]->smoothd;
123     mg[i]->rtol = 0.0;
124     mg[i]->abstol = 0.0;
125     mg[i]->dtol = 0.0;
126     mg[i]->ttol = 0.0;
127     mg[i]->eventsetup = 0;
128     mg[i]->eventsolve = 0;
129   }
130   *result = mg;
131   PetscFunctionReturn(0);
132 }
133 
134 #undef __FUNCT__
135 #define __FUNCT__ "PCDestroy_MG"
136 static PetscErrorCode PCDestroy_MG(PC pc)
137 {
138   MG             *mg = (MG*)pc->data;
139   PetscErrorCode ierr;
140   PetscInt       i,n = mg[0]->levels;
141 
142   PetscFunctionBegin;
143   if (mg[0]->galerkinused) {
144     Mat B;
145     for (i=0; i<n-1; i++) {
146       ierr = KSPGetOperators(mg[i]->smoothd,0,&B,0);CHKERRQ(ierr);
147       ierr = MatDestroy(B);CHKERRQ(ierr);
148     }
149   }
150 
151   for (i=0; i<n-1; i++) {
152     if (mg[i]->r) {ierr = VecDestroy(mg[i]->r);CHKERRQ(ierr);}
153     if (mg[i]->b) {ierr = VecDestroy(mg[i]->b);CHKERRQ(ierr);}
154     if (mg[i]->x) {ierr = VecDestroy(mg[i]->x);CHKERRQ(ierr);}
155     if (mg[i+1]->restrct) {ierr = MatDestroy(mg[i+1]->restrct);CHKERRQ(ierr);}
156     if (mg[i+1]->interpolate) {ierr = MatDestroy(mg[i+1]->interpolate);CHKERRQ(ierr);}
157   }
158   if (mg[n-1]->r) {ierr = VecDestroy(mg[n-1]->r);CHKERRQ(ierr);}
159 
160   for (i=0; i<n; i++) {
161     if (mg[i]->smoothd != mg[i]->smoothu) {
162       ierr = KSPDestroy(mg[i]->smoothd);CHKERRQ(ierr);
163     }
164     ierr = KSPDestroy(mg[i]->smoothu);CHKERRQ(ierr);
165     ierr = PetscFree(mg[i]);CHKERRQ(ierr);
166   }
167   ierr = PetscFree(mg);CHKERRQ(ierr);
168   PetscFunctionReturn(0);
169 }
170 
171 
172 
173 EXTERN PetscErrorCode MGACycle_Private(MG*);
174 EXTERN PetscErrorCode MGFCycle_Private(MG*);
175 EXTERN PetscErrorCode MGKCycle_Private(MG*);
176 
177 /*
178    PCApply_MG - Runs either an additive, multiplicative, Kaskadic
179              or full cycle of multigrid.
180 
181   Note:
182   A simple wrapper which calls MGMCycle(),MGACycle(), or MGFCycle().
183 */
184 #undef __FUNCT__
185 #define __FUNCT__ "PCApply_MG"
186 static PetscErrorCode PCApply_MG(PC pc,Vec b,Vec x)
187 {
188   MG             *mg = (MG*)pc->data;
189   PetscScalar    zero = 0.0;
190   PetscErrorCode ierr;
191   PetscInt       levels = mg[0]->levels;
192 
193   PetscFunctionBegin;
194   mg[levels-1]->b = b;
195   mg[levels-1]->x = x;
196   if (!mg[levels-1]->r && mg[0]->am == MGADDITIVE) {
197     Vec tvec;
198     ierr = VecDuplicate(mg[levels-1]->b,&tvec);CHKERRQ(ierr);
199     ierr = MGSetR(pc,levels-1,tvec);CHKERRQ(ierr);
200     ierr = VecDestroy(tvec);CHKERRQ(ierr);
201   }
202   if (mg[0]->am == MGMULTIPLICATIVE) {
203     ierr = VecSet(&zero,x);CHKERRQ(ierr);
204     ierr = MGMCycle_Private(mg+levels-1,PETSC_NULL);CHKERRQ(ierr);
205   }
206   else if (mg[0]->am == MGADDITIVE) {
207     ierr = MGACycle_Private(mg);CHKERRQ(ierr);
208   }
209   else if (mg[0]->am == MGKASKADE) {
210     ierr = MGKCycle_Private(mg);CHKERRQ(ierr);
211   }
212   else {
213     ierr = MGFCycle_Private(mg);CHKERRQ(ierr);
214   }
215   PetscFunctionReturn(0);
216 }
217 
218 #undef __FUNCT__
219 #define __FUNCT__ "PCApplyRichardson_MG"
220 static PetscErrorCode PCApplyRichardson_MG(PC pc,Vec b,Vec x,Vec w,PetscReal rtol,PetscReal abstol, PetscReal dtol,PetscInt its)
221 {
222   MG             *mg = (MG*)pc->data;
223   PetscErrorCode ierr;
224   PetscInt       levels = mg[0]->levels;
225   PetscTruth     converged = PETSC_FALSE;
226 
227   PetscFunctionBegin;
228   mg[levels-1]->b    = b;
229   mg[levels-1]->x    = x;
230 
231   mg[levels-1]->rtol = rtol;
232   mg[levels-1]->abstol = abstol;
233   mg[levels-1]->dtol = dtol;
234   if (rtol) {
235     /* compute initial residual norm for relative convergence test */
236     PetscReal rnorm;
237     ierr               = (*mg[levels-1]->residual)(mg[levels-1]->A,b,x,w);CHKERRQ(ierr);
238     ierr               = VecNorm(w,NORM_2,&rnorm);CHKERRQ(ierr);
239     mg[levels-1]->ttol = PetscMax(rtol*rnorm,abstol);
240   } else if (abstol) {
241     mg[levels-1]->ttol = abstol;
242   } else {
243     mg[levels-1]->ttol = 0.0;
244   }
245 
246   while (its-- && !converged) {
247     ierr = MGMCycle_Private(mg+levels-1,&converged);CHKERRQ(ierr);
248   }
249   PetscFunctionReturn(0);
250 }
251 
252 #undef __FUNCT__
253 #define __FUNCT__ "PCSetFromOptions_MG"
254 static PetscErrorCode PCSetFromOptions_MG(PC pc)
255 {
256   PetscErrorCode ierr;
257   PetscInt       indx,m,levels = 1;
258   PetscTruth     flg;
259   const char     *type[] = {"additive","multiplicative","full","cascade","kascade"};
260 
261   PetscFunctionBegin;
262 
263   ierr = PetscOptionsHead("Multigrid options");CHKERRQ(ierr);
264     if (!pc->data) {
265       ierr = PetscOptionsInt("-pc_mg_levels","Number of Levels","MGSetLevels",levels,&levels,&flg);CHKERRQ(ierr);
266       ierr = MGSetLevels(pc,levels,PETSC_NULL);CHKERRQ(ierr);
267     }
268     ierr = PetscOptionsInt("-pc_mg_cycles","1 for V cycle, 2 for W-cycle","MGSetCycles",1,&m,&flg);CHKERRQ(ierr);
269     if (flg) {
270       ierr = MGSetCycles(pc,m);CHKERRQ(ierr);
271     }
272     ierr = PetscOptionsName("-pc_mg_galerkin","Use Galerkin process to compute coarser operators","MGSetGalerkin",&flg);CHKERRQ(ierr);
273     if (flg) {
274       ierr = MGSetGalerkin(pc);CHKERRQ(ierr);
275     }
276     ierr = PetscOptionsInt("-pc_mg_smoothup","Number of post-smoothing steps","MGSetNumberSmoothUp",1,&m,&flg);CHKERRQ(ierr);
277     if (flg) {
278       ierr = MGSetNumberSmoothUp(pc,m);CHKERRQ(ierr);
279     }
280     ierr = PetscOptionsInt("-pc_mg_smoothdown","Number of pre-smoothing steps","MGSetNumberSmoothDown",1,&m,&flg);CHKERRQ(ierr);
281     if (flg) {
282       ierr = MGSetNumberSmoothDown(pc,m);CHKERRQ(ierr);
283     }
284     ierr = PetscOptionsEList("-pc_mg_type","Multigrid type","MGSetType",type,5,type[1],&indx,&flg);CHKERRQ(ierr);
285     if (flg) {
286       MGType mg = (MGType) 0;
287       switch (indx) {
288       case 0:
289         mg = MGADDITIVE;
290         break;
291       case 1:
292         mg = MGMULTIPLICATIVE;
293         break;
294       case 2:
295         mg = MGFULL;
296         break;
297       case 3:
298         mg = MGKASKADE;
299         break;
300       case 4:
301         mg = MGKASKADE;
302         break;
303       }
304       ierr = MGSetType(pc,mg);CHKERRQ(ierr);
305     }
306     ierr = PetscOptionsName("-pc_mg_log","Log times for each multigrid level","None",&flg);CHKERRQ(ierr);
307     if (flg) {
308       MG   *mg = (MG*)pc->data;
309       PetscInt i;
310       char eventname[128];
311       if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
312       levels = mg[0]->levels;
313       for (i=0; i<levels; i++) {
314         sprintf(eventname,"MSetup Level %d",(int)i);
315         ierr = PetscLogEventRegister(&mg[i]->eventsetup,eventname,pc->cookie);CHKERRQ(ierr);
316         sprintf(eventname,"MGSolve Level %d to 0",(int)i);
317         ierr = PetscLogEventRegister(&mg[i]->eventsolve,eventname,pc->cookie);CHKERRQ(ierr);
318       }
319     }
320   ierr = PetscOptionsTail();CHKERRQ(ierr);
321   PetscFunctionReturn(0);
322 }
323 
324 #undef __FUNCT__
325 #define __FUNCT__ "PCView_MG"
326 static PetscErrorCode PCView_MG(PC pc,PetscViewer viewer)
327 {
328   MG             *mg = (MG*)pc->data;
329   PetscErrorCode ierr;
330   PetscInt       levels = mg[0]->levels,i;
331   const char     *cstring;
332   PetscTruth     iascii;
333 
334   PetscFunctionBegin;
335   ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);CHKERRQ(ierr);
336   if (iascii) {
337     if (mg[0]->am == MGMULTIPLICATIVE) cstring = "multiplicative";
338     else if (mg[0]->am == MGADDITIVE)  cstring = "additive";
339     else if (mg[0]->am == MGFULL)      cstring = "full";
340     else if (mg[0]->am == MGKASKADE)   cstring = "Kaskade";
341     else cstring = "unknown";
342     ierr = PetscViewerASCIIPrintf(viewer,"  MG: type is %s, levels=%D cycles=%D, pre-smooths=%D, post-smooths=%D\n",
343                       cstring,levels,mg[0]->cycles,mg[0]->default_smoothd,mg[0]->default_smoothu);CHKERRQ(ierr);
344     if (mg[0]->galerkin) {
345       ierr = PetscViewerASCIIPrintf(viewer,"    Using Galerkin computed coarse grid matrices\n");CHKERRQ(ierr);
346     }
347     for (i=0; i<levels; i++) {
348       if (!i) {
349         ierr = PetscViewerASCIIPrintf(viewer,"Coarse gride solver -- level %D -------------------------------\n",i);CHKERRQ(ierr);
350       } else {
351         ierr = PetscViewerASCIIPrintf(viewer,"Down solver (pre-smoother) on level %D -------------------------------\n",i);CHKERRQ(ierr);
352       }
353       ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
354       ierr = KSPView(mg[i]->smoothd,viewer);CHKERRQ(ierr);
355       ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
356       if (i && mg[i]->smoothd == mg[i]->smoothu) {
357         ierr = PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) same as down solver (pre-smoother)\n");CHKERRQ(ierr);
358       } else if (i){
359         ierr = PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) on level %D -------------------------------\n",i);CHKERRQ(ierr);
360         ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
361         ierr = KSPView(mg[i]->smoothu,viewer);CHKERRQ(ierr);
362         ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
363       }
364     }
365   } else {
366     SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported for PCMG",((PetscObject)viewer)->type_name);
367   }
368   PetscFunctionReturn(0);
369 }
370 
371 /*
372     Calls setup for the KSP on each level
373 */
374 #undef __FUNCT__
375 #define __FUNCT__ "PCSetUp_MG"
376 static PetscErrorCode PCSetUp_MG(PC pc)
377 {
378   MG             *mg = (MG*)pc->data;
379   PetscErrorCode ierr;
380   PetscInt       i,n = mg[0]->levels;
381   PC             cpc;
382   PetscTruth     preonly,lu,redundant,monitor = PETSC_FALSE,dump;
383   PetscViewer    ascii;
384   MPI_Comm       comm;
385   Mat            dA,dB;
386   MatStructure   uflag;
387   Vec            tvec;
388 
389   PetscFunctionBegin;
390   if (!pc->setupcalled) {
391     ierr = PetscOptionsHasName(0,"-pc_mg_monitor",&monitor);CHKERRQ(ierr);
392 
393     for (i=0; i<n; i++) {
394       if (monitor) {
395         ierr = PetscObjectGetComm((PetscObject)mg[i]->smoothd,&comm);CHKERRQ(ierr);
396         ierr = PetscViewerASCIIOpen(comm,"stdout",&ascii);CHKERRQ(ierr);
397         ierr = PetscViewerASCIISetTab(ascii,n-i);CHKERRQ(ierr);
398         ierr = KSPSetMonitor(mg[i]->smoothd,KSPDefaultMonitor,ascii,(PetscErrorCode(*)(void*))PetscViewerDestroy);CHKERRQ(ierr);
399       }
400       ierr = KSPSetFromOptions(mg[i]->smoothd);CHKERRQ(ierr);
401     }
402     for (i=1; i<n; i++) {
403       if (mg[i]->smoothu && mg[i]->smoothu != mg[i]->smoothd) {
404         if (monitor) {
405           ierr = PetscObjectGetComm((PetscObject)mg[i]->smoothu,&comm);CHKERRQ(ierr);
406           ierr = PetscViewerASCIIOpen(comm,"stdout",&ascii);CHKERRQ(ierr);
407           ierr = PetscViewerASCIISetTab(ascii,n-i);CHKERRQ(ierr);
408           ierr = KSPSetMonitor(mg[i]->smoothu,KSPDefaultMonitor,ascii,(PetscErrorCode(*)(void*))PetscViewerDestroy);CHKERRQ(ierr);
409         }
410         ierr = KSPSetFromOptions(mg[i]->smoothu);CHKERRQ(ierr);
411       }
412     }
413     for (i=1; i<n; i++) {
414       if (mg[i]->restrct && !mg[i]->interpolate) {
415         ierr = MGSetInterpolate(pc,i,mg[i]->restrct);CHKERRQ(ierr);
416       }
417       if (!mg[i]->restrct && mg[i]->interpolate) {
418         ierr = MGSetRestriction(pc,i,mg[i]->interpolate);CHKERRQ(ierr);
419       }
420 #if defined(PETSC_USE_DEBUG)
421       if (!mg[i]->restrct || !mg[i]->interpolate) {
422         SETERRQ1(PETSC_ERR_ARG_WRONGSTATE,"Need to set restriction or interpolation on level %d",(int)i);
423       }
424 #endif
425     }
426     for (i=0; i<n-1; i++) {
427       if (!mg[i]->r) {
428         ierr = VecDuplicate(mg[i]->b,&tvec);CHKERRQ(ierr);
429         ierr = MGSetR(pc,i,tvec);CHKERRQ(ierr);
430         ierr = VecDestroy(tvec);CHKERRQ(ierr);
431       }
432       if (!mg[i]->x) {
433         ierr = VecDuplicate(mg[i]->b,&tvec);CHKERRQ(ierr);
434         ierr = MGSetX(pc,i,tvec);CHKERRQ(ierr);
435         ierr = VecDestroy(tvec);CHKERRQ(ierr);
436       }
437     }
438   }
439 
440   /* If user did not provide fine grid operators, use those from PC */
441   /* BUG BUG BUG This will work ONLY the first time called: hence if the user changes
442      the PC matrices between solves PCMG will continue to use first set provided */
443   ierr = KSPGetOperators(mg[n-1]->smoothd,&dA,&dB,&uflag);CHKERRQ(ierr);
444   if (!dA  && !dB) {
445     ierr = PetscLogInfo((pc,"PCSetUp_MG: Using outer operators to define finest grid operator \n  because MGGetSmoother(pc,nlevels-1,&ksp);KSPSetOperators(ksp,...); was not called.\n"));
446     ierr = KSPSetOperators(mg[n-1]->smoothd,pc->mat,pc->pmat,uflag);CHKERRQ(ierr);
447   }
448 
449   if (mg[0]->galerkin) {
450     mg[0]->galerkinused = PETSC_TRUE;
451     /* currently only handle case where mat and pmat are the same on coarser levels */
452     Mat B;
453     ierr = KSPGetOperators(mg[n-1]->smoothd,&dA,&dB,&uflag);CHKERRQ(ierr);
454     if (!pc->setupcalled) {
455       for (i=n-2; i>-1; i--) {
456         ierr = MatPtAP(dB,mg[i+1]->interpolate,MAT_INITIAL_MATRIX,1.0,&B);CHKERRQ(ierr);
457         ierr = KSPSetOperators(mg[i]->smoothd,B,B,uflag);CHKERRQ(ierr);
458         dB   = B;
459       }
460     } else {
461       for (i=n-2; i>-1; i--) {
462         ierr = KSPGetOperators(mg[i]->smoothd,0,&B,0);CHKERRQ(ierr);
463         ierr = MatPtAP(dB,mg[i]->interpolate,MAT_REUSE_MATRIX,1.0,&B);CHKERRQ(ierr);
464         ierr = KSPSetOperators(mg[i]->smoothd,B,B,uflag);CHKERRQ(ierr);
465         dB   = B;
466       }
467     }
468   }
469 
470   for (i=1; i<n; i++) {
471     if (mg[i]->smoothu == mg[i]->smoothd) {
472       /* if doing only down then initial guess is zero */
473       ierr = KSPSetInitialGuessNonzero(mg[i]->smoothd,PETSC_TRUE);CHKERRQ(ierr);
474     }
475     if (mg[i]->eventsetup) {ierr = PetscLogEventBegin(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
476     ierr = KSPSetUp(mg[i]->smoothd);CHKERRQ(ierr);
477     if (mg[i]->eventsetup) {ierr = PetscLogEventEnd(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
478   }
479   for (i=1; i<n; i++) {
480     if (mg[i]->smoothu && mg[i]->smoothu != mg[i]->smoothd) {
481       PC           uppc,downpc;
482       Mat          downmat,downpmat,upmat,uppmat;
483       MatStructure matflag;
484 
485       /* check if operators have been set for up, if not use down operators to set them */
486       ierr = KSPGetPC(mg[i]->smoothu,&uppc);CHKERRQ(ierr);
487       ierr = PCGetOperators(uppc,&upmat,&uppmat,PETSC_NULL);CHKERRQ(ierr);
488       if (!upmat) {
489         ierr = KSPGetPC(mg[i]->smoothd,&downpc);CHKERRQ(ierr);
490         ierr = PCGetOperators(downpc,&downmat,&downpmat,&matflag);CHKERRQ(ierr);
491         ierr = KSPSetOperators(mg[i]->smoothu,downmat,downpmat,matflag);CHKERRQ(ierr);
492       }
493 
494       ierr = KSPSetInitialGuessNonzero(mg[i]->smoothu,PETSC_TRUE);CHKERRQ(ierr);
495       if (mg[i]->eventsetup) {ierr = PetscLogEventBegin(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
496       ierr = KSPSetUp(mg[i]->smoothu);CHKERRQ(ierr);
497       if (mg[i]->eventsetup) {ierr = PetscLogEventEnd(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
498     }
499   }
500 
501   /*
502       If coarse solver is not direct method then DO NOT USE preonly
503   */
504   ierr = PetscTypeCompare((PetscObject)mg[0]->smoothd,KSPPREONLY,&preonly);CHKERRQ(ierr);
505   if (preonly) {
506     ierr = KSPGetPC(mg[0]->smoothd,&cpc);CHKERRQ(ierr);
507     ierr = PetscTypeCompare((PetscObject)cpc,PCLU,&lu);CHKERRQ(ierr);
508     ierr = PetscTypeCompare((PetscObject)cpc,PCREDUNDANT,&redundant);CHKERRQ(ierr);
509     if (!lu && !redundant) {
510       ierr = KSPSetType(mg[0]->smoothd,KSPGMRES);CHKERRQ(ierr);
511     }
512   }
513 
514   if (!pc->setupcalled) {
515     if (monitor) {
516       ierr = PetscObjectGetComm((PetscObject)mg[0]->smoothd,&comm);CHKERRQ(ierr);
517       ierr = PetscViewerASCIIOpen(comm,"stdout",&ascii);CHKERRQ(ierr);
518       ierr = PetscViewerASCIISetTab(ascii,n);CHKERRQ(ierr);
519       ierr = KSPSetMonitor(mg[0]->smoothd,KSPDefaultMonitor,ascii,(PetscErrorCode(*)(void*))PetscViewerDestroy);CHKERRQ(ierr);
520     }
521     ierr = KSPSetFromOptions(mg[0]->smoothd);CHKERRQ(ierr);
522   }
523 
524   if (mg[0]->eventsetup) {ierr = PetscLogEventBegin(mg[0]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
525   ierr = KSPSetUp(mg[0]->smoothd);CHKERRQ(ierr);
526   if (mg[0]->eventsetup) {ierr = PetscLogEventEnd(mg[0]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
527 
528   /*
529      Dump the interpolation/restriction matrices to matlab plus the
530    Jacobian/stiffness on each level. This allows Matlab users to
531    easily check if the Galerkin condition A_c = R A_f R^T is satisfied */
532   ierr = PetscOptionsHasName(pc->prefix,"-pc_mg_dump_matlab",&dump);CHKERRQ(ierr);
533   if (dump) {
534     for (i=1; i<n; i++) {
535       ierr = MatView(mg[i]->restrct,PETSC_VIEWER_SOCKET_(pc->comm));CHKERRQ(ierr);
536     }
537     for (i=0; i<n; i++) {
538       ierr = KSPGetPC(mg[i]->smoothd,&pc);CHKERRQ(ierr);
539       ierr = MatView(pc->mat,PETSC_VIEWER_SOCKET_(pc->comm));CHKERRQ(ierr);
540     }
541   }
542   ierr = PetscOptionsHasName(pc->prefix,"-pc_mg_dump_binary",&dump);CHKERRQ(ierr);
543   if (dump) {
544     for (i=1; i<n; i++) {
545       ierr = MatView(mg[i]->restrct,PETSC_VIEWER_BINARY_(pc->comm));CHKERRQ(ierr);
546     }
547     for (i=0; i<n; i++) {
548       ierr = KSPGetPC(mg[i]->smoothd,&pc);CHKERRQ(ierr);
549       ierr = MatView(pc->mat,PETSC_VIEWER_BINARY_(pc->comm));CHKERRQ(ierr);
550     }
551   }
552   PetscFunctionReturn(0);
553 }
554 
555 /* -------------------------------------------------------------------------------------*/
556 
557 #undef __FUNCT__
558 #define __FUNCT__ "MGSetLevels"
559 /*@C
560    MGSetLevels - Sets the number of levels to use with MG.
561    Must be called before any other MG routine.
562 
563    Collective on PC
564 
565    Input Parameters:
566 +  pc - the preconditioner context
567 .  levels - the number of levels
568 -  comms - optional communicators for each level; this is to allow solving the coarser problems
569            on smaller sets of processors. Use PETSC_NULL_OBJECT for default in Fortran
570 
571    Level: intermediate
572 
573    Notes:
574      If the number of levels is one then the multigrid uses the -mg_levels prefix
575   for setting the level options rather than the -mg_coarse prefix.
576 
577 .keywords: MG, set, levels, multigrid
578 
579 .seealso: MGSetType(), MGGetLevels()
580 @*/
581 PetscErrorCode PETSCKSP_DLLEXPORT MGSetLevels(PC pc,PetscInt levels,MPI_Comm *comms)
582 {
583   PetscErrorCode ierr;
584   MG             *mg;
585 
586   PetscFunctionBegin;
587   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
588 
589   if (pc->data) {
590     SETERRQ(PETSC_ERR_ORDER,"Number levels already set for MG\n\
591     make sure that you call MGSetLevels() before KSPSetFromOptions()");
592   }
593   ierr                     = MGCreate_Private(pc->comm,levels,pc,comms,&mg);CHKERRQ(ierr);
594   mg[0]->am                = MGMULTIPLICATIVE;
595   pc->data                 = (void*)mg;
596   pc->ops->applyrichardson = PCApplyRichardson_MG;
597   PetscFunctionReturn(0);
598 }
599 
600 #undef __FUNCT__
601 #define __FUNCT__ "MGGetLevels"
602 /*@
603    MGGetLevels - Gets the number of levels to use with MG.
604 
605    Not Collective
606 
607    Input Parameter:
608 .  pc - the preconditioner context
609 
610    Output parameter:
611 .  levels - the number of levels
612 
613    Level: advanced
614 
615 .keywords: MG, get, levels, multigrid
616 
617 .seealso: MGSetLevels()
618 @*/
619 PetscErrorCode PETSCKSP_DLLEXPORT MGGetLevels(PC pc,PetscInt *levels)
620 {
621   MG  *mg;
622 
623   PetscFunctionBegin;
624   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
625   PetscValidIntPointer(levels,2);
626 
627   mg      = (MG*)pc->data;
628   *levels = mg[0]->levels;
629   PetscFunctionReturn(0);
630 }
631 
632 #undef __FUNCT__
633 #define __FUNCT__ "MGSetType"
634 /*@
635    MGSetType - Determines the form of multigrid to use:
636    multiplicative, additive, full, or the Kaskade algorithm.
637 
638    Collective on PC
639 
640    Input Parameters:
641 +  pc - the preconditioner context
642 -  form - multigrid form, one of MGMULTIPLICATIVE, MGADDITIVE,
643    MGFULL, MGKASKADE
644 
645    Options Database Key:
646 .  -pc_mg_type <form> - Sets <form>, one of multiplicative,
647    additive, full, kaskade
648 
649    Level: advanced
650 
651 .keywords: MG, set, method, multiplicative, additive, full, Kaskade, multigrid
652 
653 .seealso: MGSetLevels()
654 @*/
655 PetscErrorCode PETSCKSP_DLLEXPORT MGSetType(PC pc,MGType form)
656 {
657   MG *mg;
658 
659   PetscFunctionBegin;
660   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
661   mg = (MG*)pc->data;
662 
663   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
664   mg[0]->am = form;
665   if (form == MGMULTIPLICATIVE) pc->ops->applyrichardson = PCApplyRichardson_MG;
666   else pc->ops->applyrichardson = 0;
667   PetscFunctionReturn(0);
668 }
669 
670 #undef __FUNCT__
671 #define __FUNCT__ "MGSetCycles"
672 /*@
673    MGSetCycles - Sets the type cycles to use.  Use MGSetCyclesOnLevel() for more
674    complicated cycling.
675 
676    Collective on PC
677 
678    Input Parameters:
679 +  pc - the multigrid context
680 -  n - the number of cycles
681 
682    Options Database Key:
683 $  -pc_mg_cycles n - 1 denotes a V-cycle; 2 denotes a W-cycle.
684 
685    Level: advanced
686 
687 .keywords: MG, set, cycles, V-cycle, W-cycle, multigrid
688 
689 .seealso: MGSetCyclesOnLevel()
690 @*/
691 PetscErrorCode PETSCKSP_DLLEXPORT MGSetCycles(PC pc,PetscInt n)
692 {
693   MG       *mg;
694   PetscInt i,levels;
695 
696   PetscFunctionBegin;
697   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
698   mg     = (MG*)pc->data;
699   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
700   levels = mg[0]->levels;
701 
702   for (i=0; i<levels; i++) {
703     mg[i]->cycles  = n;
704   }
705   PetscFunctionReturn(0);
706 }
707 
708 #undef __FUNCT__
709 #define __FUNCT__ "MGSetGalerkin"
710 /*@
711    MGSetGalerkin - Causes the coarser grid matrices to be computed from the
712       finest grid via the Galerkin process: A_i-1 = r_i * A_i * r_i^t
713 
714    Collective on PC
715 
716    Input Parameters:
717 +  pc - the multigrid context
718 -  n - the number of cycles
719 
720    Options Database Key:
721 $  -pc_mg_galerkin
722 
723    Level: intermediate
724 
725 .keywords: MG, set, Galerkin
726 
727 @*/
728 PetscErrorCode PETSCKSP_DLLEXPORT MGSetGalerkin(PC pc)
729 {
730   MG       *mg;
731   PetscInt i,levels;
732 
733   PetscFunctionBegin;
734   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
735   mg     = (MG*)pc->data;
736   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
737   levels = mg[0]->levels;
738 
739   for (i=0; i<levels; i++) {
740     mg[i]->galerkin = PETSC_TRUE;
741   }
742   PetscFunctionReturn(0);
743 }
744 
745 #undef __FUNCT__
746 #define __FUNCT__ "MGCheck"
747 /*@
748    MGCheck - Checks that all components of the MG structure have
749    been set.
750 
751    Collective on PC
752 
753    Input Parameters:
754 .  mg - the MG structure
755 
756    Level: advanced
757 
758 .keywords: MG, check, set, multigrid
759 @*/
760 PetscErrorCode PETSCKSP_DLLEXPORT MGCheck(PC pc)
761 {
762   MG       *mg;
763   PetscInt i,n,count = 0;
764 
765   PetscFunctionBegin;
766   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
767   mg = (MG*)pc->data;
768 
769   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
770 
771   n = mg[0]->levels;
772 
773   for (i=1; i<n; i++) {
774     if (!mg[i]->restrct) {
775       (*PetscErrorPrintf)("No restrict set level %D \n",n-i); count++;
776     }
777     if (!mg[i]->interpolate) {
778       (*PetscErrorPrintf)("No interpolate set level %D \n",n-i); count++;
779     }
780     if (!mg[i]->r) {
781       (*PetscErrorPrintf)("No r set level %D \n",n-i); count++;
782     }
783     if (!mg[i-1]->x) {
784       (*PetscErrorPrintf)("No x set level %D \n",n-i); count++;
785     }
786     if (!mg[i-1]->b) {
787       (*PetscErrorPrintf)("No b set level %D \n",n-i); count++;
788     }
789   }
790   PetscFunctionReturn(count);
791 }
792 
793 
794 #undef __FUNCT__
795 #define __FUNCT__ "MGSetNumberSmoothDown"
796 /*@
797    MGSetNumberSmoothDown - Sets the number of pre-smoothing steps to
798    use on all levels. Use MGGetSmootherDown() to set different
799    pre-smoothing steps on different levels.
800 
801    Collective on PC
802 
803    Input Parameters:
804 +  mg - the multigrid context
805 -  n - the number of smoothing steps
806 
807    Options Database Key:
808 .  -pc_mg_smoothdown <n> - Sets number of pre-smoothing steps
809 
810    Level: advanced
811 
812 .keywords: MG, smooth, down, pre-smoothing, steps, multigrid
813 
814 .seealso: MGSetNumberSmoothUp()
815 @*/
816 PetscErrorCode PETSCKSP_DLLEXPORT MGSetNumberSmoothDown(PC pc,PetscInt n)
817 {
818   MG             *mg;
819   PetscErrorCode ierr;
820   PetscInt       i,levels;
821 
822   PetscFunctionBegin;
823   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
824   mg     = (MG*)pc->data;
825   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
826   levels = mg[0]->levels;
827 
828   for (i=0; i<levels; i++) {
829     /* make sure smoother up and down are different */
830     ierr = MGGetSmootherUp(pc,i,PETSC_NULL);CHKERRQ(ierr);
831     ierr = KSPSetTolerances(mg[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);CHKERRQ(ierr);
832     mg[i]->default_smoothd = n;
833   }
834   PetscFunctionReturn(0);
835 }
836 
837 #undef __FUNCT__
838 #define __FUNCT__ "MGSetNumberSmoothUp"
839 /*@
840    MGSetNumberSmoothUp - Sets the number of post-smoothing steps to use
841    on all levels. Use MGGetSmootherUp() to set different numbers of
842    post-smoothing steps on different levels.
843 
844    Collective on PC
845 
846    Input Parameters:
847 +  mg - the multigrid context
848 -  n - the number of smoothing steps
849 
850    Options Database Key:
851 .  -pc_mg_smoothup <n> - Sets number of post-smoothing steps
852 
853    Level: advanced
854 
855    Note: this does not set a value on the coarsest grid, since we assume that
856     there is no seperate smooth up on the coarsest grid.
857 
858 .keywords: MG, smooth, up, post-smoothing, steps, multigrid
859 
860 .seealso: MGSetNumberSmoothDown()
861 @*/
862 PetscErrorCode PETSCKSP_DLLEXPORT MGSetNumberSmoothUp(PC pc,PetscInt n)
863 {
864   MG             *mg;
865   PetscErrorCode ierr;
866   PetscInt       i,levels;
867 
868   PetscFunctionBegin;
869   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
870   mg     = (MG*)pc->data;
871   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
872   levels = mg[0]->levels;
873 
874   for (i=1; i<levels; i++) {
875     /* make sure smoother up and down are different */
876     ierr = MGGetSmootherUp(pc,i,PETSC_NULL);CHKERRQ(ierr);
877     ierr = KSPSetTolerances(mg[i]->smoothu,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);CHKERRQ(ierr);
878     mg[i]->default_smoothu = n;
879   }
880   PetscFunctionReturn(0);
881 }
882 
883 /* ----------------------------------------------------------------------------------------*/
884 
885 /*MC
886    PCMG - Use geometric multigrid preconditioning. This preconditioner requires you provide additional
887     information about the coarser grid matrices and restriction/interpolation operators.
888 
889    Options Database Keys:
890 +  -pc_mg_levels <nlevels> - number of levels including finest
891 .  -pc_mg_cycles 1 or 2 - for V or W-cycle
892 .  -pc_mg_smoothup <n> - number of smoothing steps after interpolation
893 .  -pc_mg_smoothdown <n> - number of smoothing steps before applying restriction operator
894 .  -pc_mg_type <additive,multiplicative,full,cascade> - multiplicative is the default
895 .  -pc_mg_log - log information about time spent on each level of the solver
896 .  -pc_mg_monitor - print information on the multigrid convergence
897 -  -pc_mg_dump_matlab - dumps the matrices for each level and the restriction/interpolation matrices
898                         to the Socket viewer for reading from Matlab.
899 
900    Notes:
901 
902    Level: intermediate
903 
904    Concepts: multigrid
905 
906 .seealso:  PCCreate(), PCSetType(), PCType (for list of available types), PC, PCMGType,
907            MGSetLevels(), MGGetLevels(), MGSetType(), MPSetCycles(), MGSetNumberSmoothDown(),
908            MGSetNumberSmoothUp(), MGGetCoarseSolve(), MGSetResidual(), MGSetInterpolation(),
909            MGSetRestriction(), MGGetSmoother(), MGGetSmootherUp(), MGGetSmootherDown(),
910            MGSetCyclesOnLevel(), MGSetRhs(), MGSetX(), MGSetR()
911 M*/
912 
913 EXTERN_C_BEGIN
914 #undef __FUNCT__
915 #define __FUNCT__ "PCCreate_MG"
916 PetscErrorCode PETSCKSP_DLLEXPORT PCCreate_MG(PC pc)
917 {
918   PetscFunctionBegin;
919   pc->ops->apply          = PCApply_MG;
920   pc->ops->setup          = PCSetUp_MG;
921   pc->ops->destroy        = PCDestroy_MG;
922   pc->ops->setfromoptions = PCSetFromOptions_MG;
923   pc->ops->view           = PCView_MG;
924 
925   pc->data                = (void*)0;
926   PetscFunctionReturn(0);
927 }
928 EXTERN_C_END
929