xref: /petsc/src/ksp/pc/impls/mg/mg.c (revision f3f8564e68dacfa3f5888fed8abe34f553e62bf2)
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 #undef __FUNCT__
10 #define __FUNCT__ "PCMGMCycle_Private"
11 PetscErrorCode PCMGMCycle_Private(PC_MG **mglevels,PetscTruth *converged)
12 {
13   PC_MG          *mg = *mglevels,*mgc;
14   PetscErrorCode ierr;
15   PetscInt       cycles = (PetscInt) mg->cycles;
16 
17   PetscFunctionBegin;
18   if (converged) *converged = PETSC_FALSE;
19 
20   if (mg->eventsolve) {ierr = PetscLogEventBegin(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
21   ierr = KSPSolve(mg->smoothd,mg->b,mg->x);CHKERRQ(ierr);  /* pre-smooth */
22   if (mg->eventsolve) {ierr = PetscLogEventEnd(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
23   if (mg->level) {  /* not the coarsest grid */
24     ierr = (*mg->residual)(mg->A,mg->b,mg->x,mg->r);CHKERRQ(ierr);
25 
26     /* if on finest level and have convergence criteria set */
27     if (mg->level == mg->levels-1 && mg->ttol) {
28       PetscReal rnorm;
29       ierr = VecNorm(mg->r,NORM_2,&rnorm);CHKERRQ(ierr);
30       if (rnorm <= mg->ttol) {
31         *converged = PETSC_TRUE;
32         if (rnorm < mg->abstol) {
33           ierr = PetscInfo2(0,"Linear solver has converged. Residual norm %G is less than absolute tolerance %G\n",rnorm,mg->abstol);CHKERRQ(ierr);
34         } else {
35           ierr = PetscInfo2(0,"Linear solver has converged. Residual norm %G is less than relative tolerance times initial residual norm %G\n",rnorm,mg->ttol);CHKERRQ(ierr);
36         }
37         PetscFunctionReturn(0);
38       }
39     }
40 
41     mgc = *(mglevels - 1);
42     ierr = MatRestrict(mg->restrct,mg->r,mgc->b);CHKERRQ(ierr);
43     ierr = VecSet(mgc->x,0.0);CHKERRQ(ierr);
44     while (cycles--) {
45       ierr = PCMGMCycle_Private(mglevels-1,converged);CHKERRQ(ierr);
46     }
47     ierr = MatInterpolateAdd(mg->interpolate,mgc->x,mg->x,mg->x);CHKERRQ(ierr);
48     if (mg->eventsolve) {ierr = PetscLogEventBegin(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
49     ierr = KSPSolve(mg->smoothu,mg->b,mg->x);CHKERRQ(ierr);    /* post smooth */
50     if (mg->eventsolve) {ierr = PetscLogEventEnd(mg->eventsolve,0,0,0,0);CHKERRQ(ierr);}
51   }
52   PetscFunctionReturn(0);
53 }
54 
55 /*
56        PCMGCreate_Private - Creates a PC_MG structure for use with the
57                multigrid code. Level 0 is the coarsest. (But the
58                finest level is stored first in the array).
59 
60 */
61 #undef __FUNCT__
62 #define __FUNCT__ "PCMGCreate_Private"
63 static PetscErrorCode PCMGCreate_Private(MPI_Comm comm,PetscInt levels,PC pc,MPI_Comm *comms,PC_MG ***result)
64 {
65   PC_MG          **mg;
66   PetscErrorCode ierr;
67   PetscInt       i;
68   PetscMPIInt    size;
69   const char     *prefix;
70   PC             ipc;
71 
72   PetscFunctionBegin;
73   ierr = PetscMalloc(levels*sizeof(PC_MG*),&mg);CHKERRQ(ierr);
74   ierr = PetscLogObjectMemory(pc,levels*(sizeof(PC_MG*)+sizeof(PC_MG)));CHKERRQ(ierr);
75 
76   ierr = PCGetOptionsPrefix(pc,&prefix);CHKERRQ(ierr);
77 
78   for (i=0; i<levels; i++) {
79     ierr = PetscNew(PC_MG,&mg[i]);CHKERRQ(ierr);
80     mg[i]->level           = i;
81     mg[i]->levels          = levels;
82     mg[i]->cycles          = PC_MG_CYCLE_V;
83     mg[i]->galerkin        = PETSC_FALSE;
84     mg[i]->galerkinused    = PETSC_FALSE;
85     mg[i]->default_smoothu = 1;
86     mg[i]->default_smoothd = 1;
87 
88     if (comms) comm = comms[i];
89     ierr = KSPCreate(comm,&mg[i]->smoothd);CHKERRQ(ierr);
90     ierr = KSPSetTolerances(mg[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT, mg[i]->default_smoothd);CHKERRQ(ierr);
91     ierr = KSPSetOptionsPrefix(mg[i]->smoothd,prefix);CHKERRQ(ierr);
92 
93     /* do special stuff for coarse grid */
94     if (!i && levels > 1) {
95       ierr = KSPAppendOptionsPrefix(mg[0]->smoothd,"mg_coarse_");CHKERRQ(ierr);
96 
97       /* coarse solve is (redundant) LU by default */
98       ierr = KSPSetType(mg[0]->smoothd,KSPPREONLY);CHKERRQ(ierr);
99       ierr = KSPGetPC(mg[0]->smoothd,&ipc);CHKERRQ(ierr);
100       ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr);
101       if (size > 1) {
102         ierr = PCSetType(ipc,PCREDUNDANT);CHKERRQ(ierr);
103         ierr = PCRedundantGetPC(ipc,&ipc);CHKERRQ(ierr);
104       }
105       ierr = PCSetType(ipc,PCLU);CHKERRQ(ierr);
106 
107     } else {
108       char tprefix[128];
109       sprintf(tprefix,"mg_levels_%d_",(int)i);
110       ierr = KSPAppendOptionsPrefix(mg[i]->smoothd,tprefix);CHKERRQ(ierr);
111     }
112     ierr = PetscLogObjectParent(pc,mg[i]->smoothd);CHKERRQ(ierr);
113     mg[i]->smoothu           = mg[i]->smoothd;
114     mg[i]->rtol              = 0.0;
115     mg[i]->abstol            = 0.0;
116     mg[i]->dtol              = 0.0;
117     mg[i]->ttol              = 0.0;
118     mg[i]->eventsetup        = 0;
119     mg[i]->eventsolve        = 0;
120     mg[i]->cyclesperpcapply  = 1;
121   }
122   *result = mg;
123   PetscFunctionReturn(0);
124 }
125 
126 #undef __FUNCT__
127 #define __FUNCT__ "PCDestroy_MG"
128 static PetscErrorCode PCDestroy_MG(PC pc)
129 {
130   PC_MG          **mg = (PC_MG**)pc->data;
131   PetscErrorCode ierr;
132   PetscInt       i,n = mg[0]->levels;
133 
134   PetscFunctionBegin;
135   for (i=0; i<n-1; i++) {
136     if (mg[i+1]->r) {ierr = VecDestroy(mg[i+1]->r);CHKERRQ(ierr);}
137     if (mg[i]->b) {ierr = VecDestroy(mg[i]->b);CHKERRQ(ierr);}
138     if (mg[i]->x) {ierr = VecDestroy(mg[i]->x);CHKERRQ(ierr);}
139     if (mg[i+1]->restrct) {ierr = MatDestroy(mg[i+1]->restrct);CHKERRQ(ierr);}
140     if (mg[i+1]->interpolate) {ierr = MatDestroy(mg[i+1]->interpolate);CHKERRQ(ierr);}
141   }
142 
143   for (i=0; i<n; i++) {
144     if (mg[i]->smoothd != mg[i]->smoothu) {
145       ierr = KSPDestroy(mg[i]->smoothd);CHKERRQ(ierr);
146     }
147     ierr = KSPDestroy(mg[i]->smoothu);CHKERRQ(ierr);
148     ierr = PetscFree(mg[i]);CHKERRQ(ierr);
149   }
150   ierr = PetscFree(mg);CHKERRQ(ierr);
151   PetscFunctionReturn(0);
152 }
153 
154 
155 
156 EXTERN PetscErrorCode PCMGACycle_Private(PC_MG**);
157 EXTERN PetscErrorCode PCMGFCycle_Private(PC_MG**);
158 EXTERN PetscErrorCode PCMGKCycle_Private(PC_MG**);
159 
160 /*
161    PCApply_MG - Runs either an additive, multiplicative, Kaskadic
162              or full cycle of multigrid.
163 
164   Note:
165   A simple wrapper which calls PCMGMCycle(),PCMGACycle(), or PCMGFCycle().
166 */
167 #undef __FUNCT__
168 #define __FUNCT__ "PCApply_MG"
169 static PetscErrorCode PCApply_MG(PC pc,Vec b,Vec x)
170 {
171   PC_MG          **mg = (PC_MG**)pc->data;
172   PetscErrorCode ierr;
173   PetscInt       levels = mg[0]->levels,i;
174 
175   PetscFunctionBegin;
176   mg[levels-1]->b = b;
177   mg[levels-1]->x = x;
178   if (!mg[levels-1]->r && mg[0]->am != PC_MG_ADDITIVE && levels > 1) {
179     Vec tvec;
180     ierr = VecDuplicate(mg[levels-1]->b,&tvec);CHKERRQ(ierr);
181     ierr = PCMGSetR(pc,levels-1,tvec);CHKERRQ(ierr);
182     ierr = VecDestroy(tvec);CHKERRQ(ierr);
183   }
184   if (mg[0]->am == PC_MG_MULTIPLICATIVE) {
185     ierr = VecSet(x,0.0);CHKERRQ(ierr);
186     for (i=0; i<mg[0]->cyclesperpcapply; i++) {
187       ierr = PCMGMCycle_Private(mg+levels-1,PETSC_NULL);CHKERRQ(ierr);
188     }
189   }
190   else if (mg[0]->am == PC_MG_ADDITIVE) {
191     ierr = PCMGACycle_Private(mg);CHKERRQ(ierr);
192   }
193   else if (mg[0]->am == PC_MG_KASKADE) {
194     ierr = PCMGKCycle_Private(mg);CHKERRQ(ierr);
195   }
196   else {
197     ierr = PCMGFCycle_Private(mg);CHKERRQ(ierr);
198   }
199   PetscFunctionReturn(0);
200 }
201 
202 #undef __FUNCT__
203 #define __FUNCT__ "PCApplyRichardson_MG"
204 static PetscErrorCode PCApplyRichardson_MG(PC pc,Vec b,Vec x,Vec w,PetscReal rtol,PetscReal abstol, PetscReal dtol,PetscInt its)
205 {
206   PC_MG          **mg = (PC_MG**)pc->data;
207   PetscErrorCode ierr;
208   PetscInt       levels = mg[0]->levels;
209   PetscTruth     converged = PETSC_FALSE;
210 
211   PetscFunctionBegin;
212   mg[levels-1]->b    = b;
213   mg[levels-1]->x    = x;
214 
215   mg[levels-1]->rtol = rtol;
216   mg[levels-1]->abstol = abstol;
217   mg[levels-1]->dtol = dtol;
218   if (rtol) {
219     /* compute initial residual norm for relative convergence test */
220     PetscReal rnorm;
221     ierr               = (*mg[levels-1]->residual)(mg[levels-1]->A,b,x,w);CHKERRQ(ierr);
222     ierr               = VecNorm(w,NORM_2,&rnorm);CHKERRQ(ierr);
223     mg[levels-1]->ttol = PetscMax(rtol*rnorm,abstol);
224   } else if (abstol) {
225     mg[levels-1]->ttol = abstol;
226   } else {
227     mg[levels-1]->ttol = 0.0;
228   }
229 
230   while (its-- && !converged) {
231     ierr = PCMGMCycle_Private(mg+levels-1,&converged);CHKERRQ(ierr);
232   }
233   PetscFunctionReturn(0);
234 }
235 
236 #undef __FUNCT__
237 #define __FUNCT__ "PCSetFromOptions_MG"
238 PetscErrorCode PCSetFromOptions_MG(PC pc)
239 {
240   PetscErrorCode ierr;
241   PetscInt       m,levels = 1,cycles;
242   PetscTruth     flg;
243   PC_MG          **mg = (PC_MG**)pc->data;
244   PCMGType       mgtype=PC_MG_MULTIPLICATIVE;
245   PCMGCycleType  mgctype;
246 
247   PetscFunctionBegin;
248   ierr = PetscOptionsHead("Multigrid options");CHKERRQ(ierr);
249     if (!pc->data) {
250       ierr = PetscOptionsInt("-pc_mg_levels","Number of Levels","PCMGSetLevels",levels,&levels,&flg);CHKERRQ(ierr);
251       ierr = PCMGSetLevels(pc,levels,PETSC_NULL);CHKERRQ(ierr);
252       mg = (PC_MG**)pc->data;
253     }
254     mgctype = (PCMGCycleType)mg[0]->cycles;
255     ierr = PetscOptionsEnum("-pc_mg_cycle_type","V cycle or for W-cycle","PCMGSetCycleType",PCMGCycleTypes,(PetscEnum)mgctype,(PetscEnum*)&mgctype,&flg);CHKERRQ(ierr);
256     if (flg) {
257       ierr = PCMGSetCycleType(pc,mgctype);CHKERRQ(ierr);
258     };
259     ierr = PetscOptionsName("-pc_mg_galerkin","Use Galerkin process to compute coarser operators","PCMGSetGalerkin",&flg);CHKERRQ(ierr);
260     if (flg) {
261       ierr = PCMGSetGalerkin(pc);CHKERRQ(ierr);
262     }
263     ierr = PetscOptionsInt("-pc_mg_smoothup","Number of post-smoothing steps","PCMGSetNumberSmoothUp",1,&m,&flg);CHKERRQ(ierr);
264     if (flg) {
265       ierr = PCMGSetNumberSmoothUp(pc,m);CHKERRQ(ierr);
266     }
267     ierr = PetscOptionsInt("-pc_mg_smoothdown","Number of pre-smoothing steps","PCMGSetNumberSmoothDown",1,&m,&flg);CHKERRQ(ierr);
268     if (flg) {
269       ierr = PCMGSetNumberSmoothDown(pc,m);CHKERRQ(ierr);
270     }
271     ierr = PetscOptionsEnum("-pc_mg_type","Multigrid type","PCMGSetType",PCMGTypes,(PetscEnum)mgtype,(PetscEnum*)&mgtype,&flg);CHKERRQ(ierr);
272     if (flg) {
273       ierr = PCMGSetType(pc,mgtype);CHKERRQ(ierr);
274     }
275     if (mg[0]->am == PC_MG_MULTIPLICATIVE) {
276       ierr = PetscOptionsInt("-pc_mg_multiplicative_cycles","Number of cycles for each preconditioner step","PCMGSetLevels",mg[0]->cyclesperpcapply,&cycles,&flg);CHKERRQ(ierr);
277       if (flg) {
278 	ierr = PCMGMultiplicativeSetCycles(pc,cycles);CHKERRQ(ierr);
279       }
280     }
281     ierr = PetscOptionsName("-pc_mg_log","Log times for each multigrid level","None",&flg);CHKERRQ(ierr);
282     if (flg) {
283       PetscInt i;
284       char     eventname[128];
285       if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
286       levels = mg[0]->levels;
287       for (i=0; i<levels; i++) {
288         sprintf(eventname,"MSetup Level %d",(int)i);
289         ierr = PetscLogEventRegister(&mg[i]->eventsetup,eventname,pc->cookie);CHKERRQ(ierr);
290         sprintf(eventname,"MGSolve Level %d to 0",(int)i);
291         ierr = PetscLogEventRegister(&mg[i]->eventsolve,eventname,pc->cookie);CHKERRQ(ierr);
292       }
293     }
294   ierr = PetscOptionsTail();CHKERRQ(ierr);
295   PetscFunctionReturn(0);
296 }
297 
298 const char *PCMGTypes[] = {"MULTIPLICATIVE","ADDITIVE","FULL","KASKADE","PCMGType","PC_MG",0};
299 const char *PCMGCycleTypes[] = {"invalid","v","w","PCMGCycleType","PC_MG_CYCLE",0};
300 
301 #undef __FUNCT__
302 #define __FUNCT__ "PCView_MG"
303 static PetscErrorCode PCView_MG(PC pc,PetscViewer viewer)
304 {
305   PC_MG          **mg = (PC_MG**)pc->data;
306   PetscErrorCode ierr;
307   PetscInt       levels = mg[0]->levels,i;
308   PetscTruth     iascii;
309 
310   PetscFunctionBegin;
311   ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);CHKERRQ(ierr);
312   if (iascii) {
313     ierr = PetscViewerASCIIPrintf(viewer,"  MG: type is %s, levels=%D cycles=%s, pre-smooths=%D, post-smooths=%D\n",
314 				  PCMGTypes[mg[0]->am],levels,(mg[0]->cycles == PC_MG_CYCLE_V) ? "v" : "w",
315                                   mg[0]->default_smoothd,mg[0]->default_smoothu);CHKERRQ(ierr);
316     if (mg[0]->galerkin) {
317       ierr = PetscViewerASCIIPrintf(viewer,"    Using Galerkin computed coarse grid matrices\n");CHKERRQ(ierr);
318     }
319     for (i=0; i<levels; i++) {
320       if (!i) {
321         ierr = PetscViewerASCIIPrintf(viewer,"Coarse gride solver -- level %D -------------------------------\n",i);CHKERRQ(ierr);
322       } else {
323         ierr = PetscViewerASCIIPrintf(viewer,"Down solver (pre-smoother) on level %D -------------------------------\n",i);CHKERRQ(ierr);
324       }
325       ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
326       ierr = KSPView(mg[i]->smoothd,viewer);CHKERRQ(ierr);
327       ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
328       if (i && mg[i]->smoothd == mg[i]->smoothu) {
329         ierr = PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) same as down solver (pre-smoother)\n");CHKERRQ(ierr);
330       } else if (i){
331         ierr = PetscViewerASCIIPrintf(viewer,"Up solver (post-smoother) on level %D -------------------------------\n",i);CHKERRQ(ierr);
332         ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
333         ierr = KSPView(mg[i]->smoothu,viewer);CHKERRQ(ierr);
334         ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
335       }
336     }
337   } else {
338     SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported for PCMG",((PetscObject)viewer)->type_name);
339   }
340   PetscFunctionReturn(0);
341 }
342 
343 /*
344     Calls setup for the KSP on each level
345 */
346 #undef __FUNCT__
347 #define __FUNCT__ "PCSetUp_MG"
348 static PetscErrorCode PCSetUp_MG(PC pc)
349 {
350   PC_MG                   **mg = (PC_MG**)pc->data;
351   PetscErrorCode          ierr;
352   PetscInt                i,n = mg[0]->levels;
353   PC                      cpc;
354   PetscTruth              preonly,lu,redundant,cholesky,monitor = PETSC_FALSE,dump,opsset;
355   PetscViewerASCIIMonitor ascii;
356   PetscViewer             viewer = PETSC_NULL;
357   MPI_Comm                comm;
358   Mat                     dA,dB;
359   MatStructure            uflag;
360   Vec                     tvec;
361 
362   PetscFunctionBegin;
363 
364   /* If user did not provide fine grid operators OR operator was not updated since last global KSPSetOperators() */
365   /* so use those from global PC */
366   /* Is this what we always want? What if user wants to keep old one? */
367   ierr = KSPGetOperatorsSet(mg[n-1]->smoothd,PETSC_NULL,&opsset);CHKERRQ(ierr);
368   ierr = KSPGetPC(mg[0]->smoothd,&cpc);CHKERRQ(ierr);
369   if (!opsset || cpc->setupcalled == 2) {
370     ierr = PetscInfo(pc,"Using outer operators to define finest grid operator \n  because PCMGGetSmoother(pc,nlevels-1,&ksp);KSPSetOperators(ksp,...); was not called.\n");CHKERRQ(ierr);
371     ierr = KSPSetOperators(mg[n-1]->smoothd,pc->mat,pc->pmat,pc->flag);CHKERRQ(ierr);
372   }
373 
374   if (mg[0]->galerkin) {
375     Mat B;
376     mg[0]->galerkinused = PETSC_TRUE;
377     /* currently only handle case where mat and pmat are the same on coarser levels */
378     ierr = KSPGetOperators(mg[n-1]->smoothd,&dA,&dB,&uflag);CHKERRQ(ierr);
379     if (!pc->setupcalled) {
380       for (i=n-2; i>-1; i--) {
381         ierr = MatPtAP(dB,mg[i+1]->interpolate,MAT_INITIAL_MATRIX,1.0,&B);CHKERRQ(ierr);
382         ierr = KSPSetOperators(mg[i]->smoothd,B,B,uflag);CHKERRQ(ierr);
383 	if (i != n-2) {ierr = PetscObjectDereference((PetscObject)dB);CHKERRQ(ierr);}
384         dB   = B;
385       }
386       ierr = PetscObjectDereference((PetscObject)dB);CHKERRQ(ierr);
387     } else {
388       for (i=n-2; i>-1; i--) {
389         ierr = KSPGetOperators(mg[i]->smoothd,PETSC_NULL,&B,PETSC_NULL);CHKERRQ(ierr);
390         ierr = MatPtAP(dB,mg[i+1]->interpolate,MAT_REUSE_MATRIX,1.0,&B);CHKERRQ(ierr);
391         ierr = KSPSetOperators(mg[i]->smoothd,B,B,uflag);CHKERRQ(ierr);
392         dB   = B;
393       }
394     }
395   }
396 
397   if (!pc->setupcalled) {
398     ierr = PetscOptionsHasName(0,"-pc_mg_monitor",&monitor);CHKERRQ(ierr);
399 
400     for (i=0; i<n; i++) {
401       if (monitor) {
402         ierr = PetscObjectGetComm((PetscObject)mg[i]->smoothd,&comm);CHKERRQ(ierr);
403         ierr = PetscViewerASCIIMonitorCreate(comm,"stdout",n-i,&ascii);CHKERRQ(ierr);
404         ierr = KSPMonitorSet(mg[i]->smoothd,KSPMonitorDefault,ascii,(PetscErrorCode(*)(void*))PetscViewerASCIIMonitorDestroy);CHKERRQ(ierr);
405       }
406       ierr = KSPSetFromOptions(mg[i]->smoothd);CHKERRQ(ierr);
407     }
408     for (i=1; i<n; i++) {
409       if (mg[i]->smoothu && (mg[i]->smoothu != mg[i]->smoothd)) {
410         if (monitor) {
411           ierr = PetscObjectGetComm((PetscObject)mg[i]->smoothu,&comm);CHKERRQ(ierr);
412           ierr = PetscViewerASCIIMonitorCreate(comm,"stdout",n-i,&ascii);CHKERRQ(ierr);
413           ierr = KSPMonitorSet(mg[i]->smoothu,KSPMonitorDefault,ascii,(PetscErrorCode(*)(void*))PetscViewerASCIIMonitorDestroy);CHKERRQ(ierr);
414         }
415         ierr = KSPSetFromOptions(mg[i]->smoothu);CHKERRQ(ierr);
416       }
417     }
418     for (i=1; i<n; i++) {
419       if (!mg[i]->residual) {
420         Mat mat;
421         ierr = KSPGetOperators(mg[i]->smoothd,PETSC_NULL,&mat,PETSC_NULL);CHKERRQ(ierr);
422         ierr = PCMGSetResidual(pc,i,PCMGDefaultResidual,mat);CHKERRQ(ierr);
423       }
424       if (mg[i]->restrct && !mg[i]->interpolate) {
425         ierr = PCMGSetInterpolate(pc,i,mg[i]->restrct);CHKERRQ(ierr);
426       }
427       if (!mg[i]->restrct && mg[i]->interpolate) {
428         ierr = PCMGSetRestriction(pc,i,mg[i]->interpolate);CHKERRQ(ierr);
429       }
430 #if defined(PETSC_USE_DEBUG)
431       if (!mg[i]->restrct || !mg[i]->interpolate) {
432         SETERRQ1(PETSC_ERR_ARG_WRONGSTATE,"Need to set restriction or interpolation on level %d",(int)i);
433       }
434 #endif
435     }
436     for (i=0; i<n-1; i++) {
437       if (!mg[i]->b) {
438         Vec *vec;
439         ierr = KSPGetVecs(mg[i]->smoothd,1,&vec,0,PETSC_NULL);CHKERRQ(ierr);
440         ierr = PCMGSetRhs(pc,i,*vec);CHKERRQ(ierr);
441         ierr = PetscFree(vec);CHKERRQ(ierr);
442       }
443       if (!mg[i]->r && i) {
444         ierr = VecDuplicate(mg[i]->b,&tvec);CHKERRQ(ierr);
445         ierr = PCMGSetR(pc,i,tvec);CHKERRQ(ierr);
446         ierr = VecDestroy(tvec);CHKERRQ(ierr);
447       }
448       if (!mg[i]->x) {
449         ierr = VecDuplicate(mg[i]->b,&tvec);CHKERRQ(ierr);
450         ierr = PCMGSetX(pc,i,tvec);CHKERRQ(ierr);
451         ierr = VecDestroy(tvec);CHKERRQ(ierr);
452       }
453     }
454   }
455 
456 
457   for (i=1; i<n; i++) {
458     if (mg[i]->smoothu == mg[i]->smoothd) {
459       /* if doing only down then initial guess is zero */
460       ierr = KSPSetInitialGuessNonzero(mg[i]->smoothd,PETSC_TRUE);CHKERRQ(ierr);
461     }
462     if (mg[i]->eventsetup) {ierr = PetscLogEventBegin(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
463     ierr = KSPSetUp(mg[i]->smoothd);CHKERRQ(ierr);
464     if (mg[i]->eventsetup) {ierr = PetscLogEventEnd(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
465   }
466   for (i=1; i<n; i++) {
467     if (mg[i]->smoothu && mg[i]->smoothu != mg[i]->smoothd) {
468       Mat          downmat,downpmat;
469       MatStructure matflag;
470       PetscTruth   opsset;
471 
472       /* check if operators have been set for up, if not use down operators to set them */
473       ierr = KSPGetOperatorsSet(mg[i]->smoothu,&opsset,PETSC_NULL);CHKERRQ(ierr);
474       if (!opsset) {
475         ierr = KSPGetOperators(mg[i]->smoothd,&downmat,&downpmat,&matflag);CHKERRQ(ierr);
476         ierr = KSPSetOperators(mg[i]->smoothu,downmat,downpmat,matflag);CHKERRQ(ierr);
477       }
478 
479       ierr = KSPSetInitialGuessNonzero(mg[i]->smoothu,PETSC_TRUE);CHKERRQ(ierr);
480       if (mg[i]->eventsetup) {ierr = PetscLogEventBegin(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
481       ierr = KSPSetUp(mg[i]->smoothu);CHKERRQ(ierr);
482       if (mg[i]->eventsetup) {ierr = PetscLogEventEnd(mg[i]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
483     }
484   }
485 
486   /*
487       If coarse solver is not direct method then DO NOT USE preonly
488   */
489   ierr = PetscTypeCompare((PetscObject)mg[0]->smoothd,KSPPREONLY,&preonly);CHKERRQ(ierr);
490   if (preonly) {
491     ierr = PetscTypeCompare((PetscObject)cpc,PCLU,&lu);CHKERRQ(ierr);
492     ierr = PetscTypeCompare((PetscObject)cpc,PCREDUNDANT,&redundant);CHKERRQ(ierr);
493     ierr = PetscTypeCompare((PetscObject)cpc,PCCHOLESKY,&cholesky);CHKERRQ(ierr);
494     if (!lu && !redundant && !cholesky) {
495       ierr = KSPSetType(mg[0]->smoothd,KSPGMRES);CHKERRQ(ierr);
496     }
497   }
498 
499   if (!pc->setupcalled) {
500     if (monitor) {
501       ierr = PetscObjectGetComm((PetscObject)mg[0]->smoothd,&comm);CHKERRQ(ierr);
502       ierr = PetscViewerASCIIMonitorCreate(comm,"stdout",n,&ascii);CHKERRQ(ierr);
503       ierr = KSPMonitorSet(mg[0]->smoothd,KSPMonitorDefault,ascii,(PetscErrorCode(*)(void*))PetscViewerASCIIMonitorDestroy);CHKERRQ(ierr);
504     }
505     ierr = KSPSetFromOptions(mg[0]->smoothd);CHKERRQ(ierr);
506   }
507 
508   if (mg[0]->eventsetup) {ierr = PetscLogEventBegin(mg[0]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
509   ierr = KSPSetUp(mg[0]->smoothd);CHKERRQ(ierr);
510   if (mg[0]->eventsetup) {ierr = PetscLogEventEnd(mg[0]->eventsetup,0,0,0,0);CHKERRQ(ierr);}
511 
512   /*
513      Dump the interpolation/restriction matrices plus the
514    Jacobian/stiffness on each level. This allows Matlab users to
515    easily check if the Galerkin condition A_c = R A_f R^T is satisfied.
516 
517    Only support one or the other at the same time.
518   */
519 #if defined(PETSC_USE_SOCKET_VIEWER)
520   ierr = PetscOptionsHasName(pc->prefix,"-pc_mg_dump_matlab",&dump);CHKERRQ(ierr);
521   if (dump) {
522     viewer = PETSC_VIEWER_SOCKET_(pc->comm);
523   }
524 #endif
525   ierr = PetscOptionsHasName(pc->prefix,"-pc_mg_dump_binary",&dump);CHKERRQ(ierr);
526   if (dump) {
527     viewer = PETSC_VIEWER_BINARY_(pc->comm);
528   }
529 
530   if (viewer) {
531     for (i=1; i<n; i++) {
532       ierr = MatView(mg[i]->restrct,viewer);CHKERRQ(ierr);
533     }
534     for (i=0; i<n; i++) {
535       ierr = KSPGetPC(mg[i]->smoothd,&pc);CHKERRQ(ierr);
536       ierr = MatView(pc->mat,viewer);CHKERRQ(ierr);
537     }
538   }
539   PetscFunctionReturn(0);
540 }
541 
542 /* -------------------------------------------------------------------------------------*/
543 
544 #undef __FUNCT__
545 #define __FUNCT__ "PCMGSetLevels"
546 /*@C
547    PCMGSetLevels - Sets the number of levels to use with MG.
548    Must be called before any other MG routine.
549 
550    Collective on PC
551 
552    Input Parameters:
553 +  pc - the preconditioner context
554 .  levels - the number of levels
555 -  comms - optional communicators for each level; this is to allow solving the coarser problems
556            on smaller sets of processors. Use PETSC_NULL_OBJECT for default in Fortran
557 
558    Level: intermediate
559 
560    Notes:
561      If the number of levels is one then the multigrid uses the -mg_levels prefix
562   for setting the level options rather than the -mg_coarse prefix.
563 
564 .keywords: MG, set, levels, multigrid
565 
566 .seealso: PCMGSetType(), PCMGGetLevels()
567 @*/
568 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetLevels(PC pc,PetscInt levels,MPI_Comm *comms)
569 {
570   PetscErrorCode ierr;
571   PC_MG          **mg=0;
572 
573   PetscFunctionBegin;
574   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
575 
576   if (pc->data) {
577     SETERRQ(PETSC_ERR_ORDER,"Number levels already set for MG\n\
578     make sure that you call PCMGSetLevels() before KSPSetFromOptions()");
579   }
580   ierr                     = PCMGCreate_Private(pc->comm,levels,pc,comms,&mg);CHKERRQ(ierr);
581   mg[0]->am                = PC_MG_MULTIPLICATIVE;
582   pc->data                 = (void*)mg;
583   pc->ops->applyrichardson = PCApplyRichardson_MG;
584   PetscFunctionReturn(0);
585 }
586 
587 #undef __FUNCT__
588 #define __FUNCT__ "PCMGGetLevels"
589 /*@
590    PCMGGetLevels - Gets the number of levels to use with MG.
591 
592    Not Collective
593 
594    Input Parameter:
595 .  pc - the preconditioner context
596 
597    Output parameter:
598 .  levels - the number of levels
599 
600    Level: advanced
601 
602 .keywords: MG, get, levels, multigrid
603 
604 .seealso: PCMGSetLevels()
605 @*/
606 PetscErrorCode PETSCKSP_DLLEXPORT PCMGGetLevels(PC pc,PetscInt *levels)
607 {
608   PC_MG  **mg;
609 
610   PetscFunctionBegin;
611   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
612   PetscValidIntPointer(levels,2);
613 
614   mg      = (PC_MG**)pc->data;
615   *levels = mg[0]->levels;
616   PetscFunctionReturn(0);
617 }
618 
619 #undef __FUNCT__
620 #define __FUNCT__ "PCMGSetType"
621 /*@
622    PCMGSetType - Determines the form of multigrid to use:
623    multiplicative, additive, full, or the Kaskade algorithm.
624 
625    Collective on PC
626 
627    Input Parameters:
628 +  pc - the preconditioner context
629 -  form - multigrid form, one of PC_MG_MULTIPLICATIVE, PC_MG_ADDITIVE,
630    PC_MG_FULL, PC_MG_KASKADE
631 
632    Options Database Key:
633 .  -pc_mg_type <form> - Sets <form>, one of multiplicative,
634    additive, full, kaskade
635 
636    Level: advanced
637 
638 .keywords: MG, set, method, multiplicative, additive, full, Kaskade, multigrid
639 
640 .seealso: PCMGSetLevels()
641 @*/
642 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetType(PC pc,PCMGType form)
643 {
644   PC_MG **mg;
645 
646   PetscFunctionBegin;
647   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
648   mg = (PC_MG**)pc->data;
649 
650   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
651   mg[0]->am = form;
652   if (form == PC_MG_MULTIPLICATIVE) pc->ops->applyrichardson = PCApplyRichardson_MG;
653   else pc->ops->applyrichardson = 0;
654   PetscFunctionReturn(0);
655 }
656 
657 #undef __FUNCT__
658 #define __FUNCT__ "PCMGSetCycleType"
659 /*@
660    PCMGSetCycleType - Sets the type cycles to use.  Use PCMGSetCycleTypeOnLevel() for more
661    complicated cycling.
662 
663    Collective on PC
664 
665    Input Parameters:
666 +  pc - the multigrid context
667 -  PC_MG_CYCLE_V or PC_MG_CYCLE_W
668 
669    Options Database Key:
670 $  -pc_mg_cycle_type v or w
671 
672    Level: advanced
673 
674 .keywords: MG, set, cycles, V-cycle, W-cycle, multigrid
675 
676 .seealso: PCMGSetCycleTypeOnLevel()
677 @*/
678 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetCycleType(PC pc,PCMGCycleType n)
679 {
680   PC_MG    **mg;
681   PetscInt i,levels;
682 
683   PetscFunctionBegin;
684   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
685   mg     = (PC_MG**)pc->data;
686   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
687   levels = mg[0]->levels;
688 
689   for (i=0; i<levels; i++) {
690     mg[i]->cycles  = n;
691   }
692   PetscFunctionReturn(0);
693 }
694 
695 #undef __FUNCT__
696 #define __FUNCT__ "PCMGMultiplicativeSetCycles"
697 /*@
698    PCMGMultiplicativeSetCycles - Sets the number of cycles to use for each preconditioner step
699          of multigrid when PCMGType of PC_MG_MULTIPLICATIVE is used
700 
701    Collective on PC
702 
703    Input Parameters:
704 +  pc - the multigrid context
705 -  n - number of cycles (default is 1)
706 
707    Options Database Key:
708 $  -pc_mg_multiplicative_cycles n
709 
710    Level: advanced
711 
712    Notes: This is not associated with setting a v or w cycle, that is set with PCMGSetCycleType()
713 
714 .keywords: MG, set, cycles, V-cycle, W-cycle, multigrid
715 
716 .seealso: PCMGSetCycleTypeOnLevel(), PCMGSetCycleType()
717 @*/
718 PetscErrorCode PETSCKSP_DLLEXPORT PCMGMultiplicativeSetCycles(PC pc,PetscInt n)
719 {
720   PC_MG    **mg;
721   PetscInt i,levels;
722 
723   PetscFunctionBegin;
724   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
725   mg     = (PC_MG**)pc->data;
726   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
727   levels = mg[0]->levels;
728 
729   for (i=0; i<levels; i++) {
730     mg[i]->cyclesperpcapply  = n;
731   }
732   PetscFunctionReturn(0);
733 }
734 
735 #undef __FUNCT__
736 #define __FUNCT__ "PCMGSetGalerkin"
737 /*@
738    PCMGSetGalerkin - Causes the coarser grid matrices to be computed from the
739       finest grid via the Galerkin process: A_i-1 = r_i * A_i * r_i^t
740 
741    Collective on PC
742 
743    Input Parameters:
744 .  pc - the multigrid context
745 
746    Options Database Key:
747 $  -pc_mg_galerkin
748 
749    Level: intermediate
750 
751 .keywords: MG, set, Galerkin
752 
753 .seealso: PCMGGetGalerkin()
754 
755 @*/
756 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetGalerkin(PC pc)
757 {
758   PC_MG    **mg;
759   PetscInt i,levels;
760 
761   PetscFunctionBegin;
762   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
763   mg     = (PC_MG**)pc->data;
764   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
765   levels = mg[0]->levels;
766 
767   for (i=0; i<levels; i++) {
768     mg[i]->galerkin = PETSC_TRUE;
769   }
770   PetscFunctionReturn(0);
771 }
772 
773 #undef __FUNCT__
774 #define __FUNCT__ "PCMGGetGalerkin"
775 /*@
776    PCMGGetGalerkin - Checks if Galerkin multigrid is being used, i.e.
777       A_i-1 = r_i * A_i * r_i^t
778 
779    Not Collective
780 
781    Input Parameter:
782 .  pc - the multigrid context
783 
784    Output Parameter:
785 .  gelerkin - PETSC_TRUE or PETSC_FALSE
786 
787    Options Database Key:
788 $  -pc_mg_galerkin
789 
790    Level: intermediate
791 
792 .keywords: MG, set, Galerkin
793 
794 .seealso: PCMGSetGalerkin()
795 
796 @*/
797 PetscErrorCode PETSCKSP_DLLEXPORT PCMGGetGalerkin(PC pc,PetscTruth *galerkin)
798 {
799   PC_MG    **mg;
800 
801   PetscFunctionBegin;
802   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
803   mg     = (PC_MG**)pc->data;
804   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
805   *galerkin = mg[0]->galerkin;
806   PetscFunctionReturn(0);
807 }
808 
809 #undef __FUNCT__
810 #define __FUNCT__ "PCMGSetNumberSmoothDown"
811 /*@
812    PCMGSetNumberSmoothDown - Sets the number of pre-smoothing steps to
813    use on all levels. Use PCMGGetSmootherDown() to set different
814    pre-smoothing steps on different levels.
815 
816    Collective on PC
817 
818    Input Parameters:
819 +  mg - the multigrid context
820 -  n - the number of smoothing steps
821 
822    Options Database Key:
823 .  -pc_mg_smoothdown <n> - Sets number of pre-smoothing steps
824 
825    Level: advanced
826 
827 .keywords: MG, smooth, down, pre-smoothing, steps, multigrid
828 
829 .seealso: PCMGSetNumberSmoothUp()
830 @*/
831 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetNumberSmoothDown(PC pc,PetscInt n)
832 {
833   PC_MG          **mg;
834   PetscErrorCode ierr;
835   PetscInt       i,levels;
836 
837   PetscFunctionBegin;
838   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
839   mg     = (PC_MG**)pc->data;
840   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
841   levels = mg[0]->levels;
842 
843   for (i=1; i<levels; i++) {
844     /* make sure smoother up and down are different */
845     ierr = PCMGGetSmootherUp(pc,i,PETSC_NULL);CHKERRQ(ierr);
846     ierr = KSPSetTolerances(mg[i]->smoothd,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);CHKERRQ(ierr);
847     mg[i]->default_smoothd = n;
848   }
849   PetscFunctionReturn(0);
850 }
851 
852 #undef __FUNCT__
853 #define __FUNCT__ "PCMGSetNumberSmoothUp"
854 /*@
855    PCMGSetNumberSmoothUp - Sets the number of post-smoothing steps to use
856    on all levels. Use PCMGGetSmootherUp() to set different numbers of
857    post-smoothing steps on different levels.
858 
859    Collective on PC
860 
861    Input Parameters:
862 +  mg - the multigrid context
863 -  n - the number of smoothing steps
864 
865    Options Database Key:
866 .  -pc_mg_smoothup <n> - Sets number of post-smoothing steps
867 
868    Level: advanced
869 
870    Note: this does not set a value on the coarsest grid, since we assume that
871     there is no separate smooth up on the coarsest grid.
872 
873 .keywords: MG, smooth, up, post-smoothing, steps, multigrid
874 
875 .seealso: PCMGSetNumberSmoothDown()
876 @*/
877 PetscErrorCode PETSCKSP_DLLEXPORT PCMGSetNumberSmoothUp(PC pc,PetscInt n)
878 {
879   PC_MG          **mg;
880   PetscErrorCode ierr;
881   PetscInt       i,levels;
882 
883   PetscFunctionBegin;
884   PetscValidHeaderSpecific(pc,PC_COOKIE,1);
885   mg     = (PC_MG**)pc->data;
886   if (!mg) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must set MG levels before calling");
887   levels = mg[0]->levels;
888 
889   for (i=1; i<levels; i++) {
890     /* make sure smoother up and down are different */
891     ierr = PCMGGetSmootherUp(pc,i,PETSC_NULL);CHKERRQ(ierr);
892     ierr = KSPSetTolerances(mg[i]->smoothu,PETSC_DEFAULT,PETSC_DEFAULT,PETSC_DEFAULT,n);CHKERRQ(ierr);
893     mg[i]->default_smoothu = n;
894   }
895   PetscFunctionReturn(0);
896 }
897 
898 /* ----------------------------------------------------------------------------------------*/
899 
900 /*MC
901    PCMG - Use multigrid preconditioning. This preconditioner requires you provide additional
902     information about the coarser grid matrices and restriction/interpolation operators.
903 
904    Options Database Keys:
905 +  -pc_mg_levels <nlevels> - number of levels including finest
906 .  -pc_mg_cycles v or w
907 .  -pc_mg_smoothup <n> - number of smoothing steps after interpolation
908 .  -pc_mg_smoothdown <n> - number of smoothing steps before applying restriction operator
909 .  -pc_mg_type <additive,multiplicative,full,cascade> - multiplicative is the default
910 .  -pc_mg_log - log information about time spent on each level of the solver
911 .  -pc_mg_monitor - print information on the multigrid convergence
912 .  -pc_mg_galerkin - use Galerkin process to compute coarser operators
913 -  -pc_mg_dump_matlab - dumps the matrices for each level and the restriction/interpolation matrices
914                         to the Socket viewer for reading from Matlab.
915 
916    Notes:
917 
918    Level: intermediate
919 
920    Concepts: multigrid
921 
922 .seealso:  PCCreate(), PCSetType(), PCType (for list of available types), PC, PCMGType,
923            PCMGSetLevels(), PCMGGetLevels(), PCMGSetType(), PCMGSetCycleType(), PCMGSetNumberSmoothDown(),
924            PCMGSetNumberSmoothUp(), PCMGGetCoarseSolve(), PCMGSetResidual(), PCMGSetInterpolation(),
925            PCMGSetRestriction(), PCMGGetSmoother(), PCMGGetSmootherUp(), PCMGGetSmootherDown(),
926            PCMGSetCycleTypeOnLevel(), PCMGSetRhs(), PCMGSetX(), PCMGSetR()
927 M*/
928 
929 EXTERN_C_BEGIN
930 #undef __FUNCT__
931 #define __FUNCT__ "PCCreate_MG"
932 PetscErrorCode PETSCKSP_DLLEXPORT PCCreate_MG(PC pc)
933 {
934   PetscFunctionBegin;
935   pc->ops->apply          = PCApply_MG;
936   pc->ops->setup          = PCSetUp_MG;
937   pc->ops->destroy        = PCDestroy_MG;
938   pc->ops->setfromoptions = PCSetFromOptions_MG;
939   pc->ops->view           = PCView_MG;
940 
941   pc->data                = (void*)0;
942   PetscFunctionReturn(0);
943 }
944 EXTERN_C_END
945