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