1 2 /* 3 Provides an interface to the SuperLU_DIST sparse solver 4 */ 5 6 #include <../src/mat/impls/aij/seq/aij.h> 7 #include <../src/mat/impls/aij/mpi/mpiaij.h> 8 #if defined(PETSC_HAVE_STDLIB_H) /* This is to get around weird problem with SuperLU on cray */ 9 #include <stdlib.h> 10 #endif 11 12 EXTERN_C_BEGIN 13 #if defined(PETSC_USE_COMPLEX) 14 #include <superlu_zdefs.h> 15 #else 16 #include <superlu_ddefs.h> 17 #endif 18 EXTERN_C_END 19 20 /* 21 GLOBAL - The sparse matrix and right hand side are all stored initially on process 0. Should be called centralized 22 DISTRIBUTED - The sparse matrix and right hand size are initially stored across the entire MPI communicator. 23 */ 24 typedef enum {GLOBAL,DISTRIBUTED} SuperLU_MatInputMode; 25 const char *SuperLU_MatInputModes[] = {"GLOBAL","DISTRIBUTED","SuperLU_MatInputMode","PETSC_",0}; 26 27 typedef struct { 28 int_t nprow,npcol,*row,*col; 29 gridinfo_t grid; 30 superlu_dist_options_t options; 31 SuperMatrix A_sup; 32 ScalePermstruct_t ScalePermstruct; 33 LUstruct_t LUstruct; 34 int StatPrint; 35 SuperLU_MatInputMode MatInputMode; 36 SOLVEstruct_t SOLVEstruct; 37 fact_t FactPattern; 38 MPI_Comm comm_superlu; 39 #if defined(PETSC_USE_COMPLEX) 40 doublecomplex *val; 41 #else 42 double *val; 43 #endif 44 PetscBool matsolve_iscalled,matmatsolve_iscalled; 45 PetscBool CleanUpSuperLU_Dist; /* Flag to clean up (non-global) SuperLU objects during Destroy */ 46 } Mat_SuperLU_DIST; 47 48 49 PetscErrorCode MatSuperluDistGetDiagU_SuperLU_DIST(Mat F,PetscScalar *diagU) 50 { 51 Mat_SuperLU_DIST *lu= (Mat_SuperLU_DIST*)F->data; 52 53 PetscFunctionBegin; 54 #if defined(PETSC_USE_COMPLEX) 55 PetscStackCall("SuperLU_DIST:pzGetDiagU",pzGetDiagU(F->rmap->N,&lu->LUstruct,&lu->grid,(doublecomplex*)diagU)); 56 #else 57 PetscStackCall("SuperLU_DIST:pdGetDiagU",pdGetDiagU(F->rmap->N,&lu->LUstruct,&lu->grid,diagU)); 58 #endif 59 PetscFunctionReturn(0); 60 } 61 62 PetscErrorCode MatSuperluDistGetDiagU(Mat F,PetscScalar *diagU) 63 { 64 PetscErrorCode ierr; 65 66 PetscFunctionBegin; 67 PetscValidHeaderSpecific(F,MAT_CLASSID,1); 68 ierr = PetscTryMethod(F,"MatSuperluDistGetDiagU_C",(Mat,PetscScalar*),(F,diagU));CHKERRQ(ierr); 69 PetscFunctionReturn(0); 70 } 71 72 static PetscErrorCode MatDestroy_SuperLU_DIST(Mat A) 73 { 74 PetscErrorCode ierr; 75 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->data; 76 77 PetscFunctionBegin; 78 if (lu->CleanUpSuperLU_Dist) { 79 /* Deallocate SuperLU_DIST storage */ 80 if (lu->MatInputMode == GLOBAL) { 81 PetscStackCall("SuperLU_DIST:Destroy_CompCol_Matrix_dist",Destroy_CompCol_Matrix_dist(&lu->A_sup)); 82 } else { 83 PetscStackCall("SuperLU_DIST:Destroy_CompRowLoc_Matrix_dist",Destroy_CompRowLoc_Matrix_dist(&lu->A_sup)); 84 if (lu->options.SolveInitialized) { 85 #if defined(PETSC_USE_COMPLEX) 86 PetscStackCall("SuperLU_DIST:zSolveFinalize",zSolveFinalize(&lu->options, &lu->SOLVEstruct)); 87 #else 88 PetscStackCall("SuperLU_DIST:dSolveFinalize",dSolveFinalize(&lu->options, &lu->SOLVEstruct)); 89 #endif 90 } 91 } 92 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(A->cmap->N, &lu->grid, &lu->LUstruct)); 93 PetscStackCall("SuperLU_DIST:ScalePermstructFree",ScalePermstructFree(&lu->ScalePermstruct)); 94 PetscStackCall("SuperLU_DIST:LUstructFree",LUstructFree(&lu->LUstruct)); 95 96 /* Release the SuperLU_DIST process grid. */ 97 PetscStackCall("SuperLU_DIST:superlu_gridexit",superlu_gridexit(&lu->grid)); 98 ierr = MPI_Comm_free(&(lu->comm_superlu));CHKERRQ(ierr); 99 } 100 ierr = PetscFree(A->data);CHKERRQ(ierr); 101 /* clear composed functions */ 102 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorGetSolverType_C",NULL);CHKERRQ(ierr); 103 ierr = PetscObjectComposeFunction((PetscObject)A,"MatSuperluDistGetDiagU_C",NULL);CHKERRQ(ierr); 104 105 PetscFunctionReturn(0); 106 } 107 108 static PetscErrorCode MatSolve_SuperLU_DIST(Mat A,Vec b_mpi,Vec x) 109 { 110 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->data; 111 PetscErrorCode ierr; 112 PetscMPIInt size; 113 PetscInt m=A->rmap->n,M=A->rmap->N,N=A->cmap->N; 114 SuperLUStat_t stat; 115 double berr[1]; 116 PetscScalar *bptr=NULL; 117 PetscInt nrhs=1; 118 Vec x_seq; 119 IS iden; 120 VecScatter scat; 121 int info; /* SuperLU_Dist info code is ALWAYS an int, even with long long indices */ 122 static PetscBool cite = PETSC_FALSE; 123 124 PetscFunctionBegin; 125 if (lu->options.Fact != FACTORED) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"SuperLU_DIST options.Fact mush equal FACTORED"); 126 ierr = PetscCitationsRegister("@article{lidemmel03,\n author = {Xiaoye S. Li and James W. Demmel},\n title = {{SuperLU_DIST}: A Scalable Distributed-Memory Sparse Direct\n Solver for Unsymmetric Linear Systems},\n journal = {ACM Trans. Mathematical Software},\n volume = {29},\n number = {2},\n pages = {110-140},\n year = 2003\n}\n",&cite);CHKERRQ(ierr); 127 128 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 129 if (size > 1 && lu->MatInputMode == GLOBAL) { 130 /* global mat input, convert b to x_seq */ 131 ierr = VecCreateSeq(PETSC_COMM_SELF,N,&x_seq);CHKERRQ(ierr); 132 ierr = ISCreateStride(PETSC_COMM_SELF,N,0,1,&iden);CHKERRQ(ierr); 133 ierr = VecScatterCreateWithData(b_mpi,iden,x_seq,iden,&scat);CHKERRQ(ierr); 134 ierr = ISDestroy(&iden);CHKERRQ(ierr); 135 136 ierr = VecScatterBegin(scat,b_mpi,x_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 137 ierr = VecScatterEnd(scat,b_mpi,x_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 138 ierr = VecGetArray(x_seq,&bptr);CHKERRQ(ierr); 139 } else { /* size==1 || distributed mat input */ 140 if (lu->options.SolveInitialized && !lu->matsolve_iscalled) { 141 /* see comments in MatMatSolve() */ 142 #if defined(PETSC_USE_COMPLEX) 143 PetscStackCall("SuperLU_DIST:zSolveFinalize",zSolveFinalize(&lu->options, &lu->SOLVEstruct)); 144 #else 145 PetscStackCall("SuperLU_DIST:dSolveFinalize",dSolveFinalize(&lu->options, &lu->SOLVEstruct)); 146 #endif 147 lu->options.SolveInitialized = NO; 148 } 149 ierr = VecCopy(b_mpi,x);CHKERRQ(ierr); 150 ierr = VecGetArray(x,&bptr);CHKERRQ(ierr); 151 } 152 153 PetscStackCall("SuperLU_DIST:PStatInit",PStatInit(&stat)); /* Initialize the statistics variables. */ 154 if (lu->MatInputMode == GLOBAL) { 155 #if defined(PETSC_USE_COMPLEX) 156 PetscStackCall("SuperLU_DIST:pzgssvx_ABglobal",pzgssvx_ABglobal(&lu->options,&lu->A_sup,&lu->ScalePermstruct,(doublecomplex*)bptr,M,nrhs,&lu->grid,&lu->LUstruct,berr,&stat,&info)); 157 #else 158 PetscStackCall("SuperLU_DIST:pdgssvx_ABglobal",pdgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct,bptr,M,nrhs,&lu->grid,&lu->LUstruct,berr,&stat,&info)); 159 #endif 160 } else { /* distributed mat input */ 161 #if defined(PETSC_USE_COMPLEX) 162 PetscStackCall("SuperLU_DIST:pzgssvx",pzgssvx(&lu->options,&lu->A_sup,&lu->ScalePermstruct,(doublecomplex*)bptr,m,nrhs,&lu->grid,&lu->LUstruct,&lu->SOLVEstruct,berr,&stat,&info)); 163 #else 164 PetscStackCall("SuperLU_DIST:pdgssvx",pdgssvx(&lu->options,&lu->A_sup,&lu->ScalePermstruct,bptr,m,nrhs,&lu->grid,&lu->LUstruct,&lu->SOLVEstruct,berr,&stat,&info)); 165 #endif 166 } 167 if (info) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"pdgssvx fails, info: %d\n",info); 168 169 if (lu->options.PrintStat) PStatPrint(&lu->options, &stat, &lu->grid); /* Print the statistics. */ 170 PetscStackCall("SuperLU_DIST:PStatFree",PStatFree(&stat)); 171 172 if (size > 1 && lu->MatInputMode == GLOBAL) { 173 /* convert seq x to mpi x */ 174 ierr = VecRestoreArray(x_seq,&bptr);CHKERRQ(ierr); 175 ierr = VecScatterBegin(scat,x_seq,x,INSERT_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 176 ierr = VecScatterEnd(scat,x_seq,x,INSERT_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 177 ierr = VecScatterDestroy(&scat);CHKERRQ(ierr); 178 ierr = VecDestroy(&x_seq);CHKERRQ(ierr); 179 } else { 180 ierr = VecRestoreArray(x,&bptr);CHKERRQ(ierr); 181 182 lu->matsolve_iscalled = PETSC_TRUE; 183 lu->matmatsolve_iscalled = PETSC_FALSE; 184 } 185 PetscFunctionReturn(0); 186 } 187 188 static PetscErrorCode MatMatSolve_SuperLU_DIST(Mat A,Mat B_mpi,Mat X) 189 { 190 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->data; 191 PetscErrorCode ierr; 192 PetscMPIInt size; 193 PetscInt M=A->rmap->N,m=A->rmap->n,nrhs; 194 SuperLUStat_t stat; 195 double berr[1]; 196 PetscScalar *bptr; 197 int info; /* SuperLU_Dist info code is ALWAYS an int, even with long long indices */ 198 PetscBool flg; 199 200 PetscFunctionBegin; 201 if (lu->options.Fact != FACTORED) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"SuperLU_DIST options.Fact mush equal FACTORED"); 202 ierr = PetscObjectTypeCompareAny((PetscObject)B_mpi,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 203 if (!flg) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Matrix B must be MATDENSE matrix"); 204 ierr = PetscObjectTypeCompareAny((PetscObject)X,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 205 if (!flg) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Matrix X must be MATDENSE matrix"); 206 207 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 208 if (size > 1 && lu->MatInputMode == GLOBAL) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatInputMode=GLOBAL for nproc %d>1 is not supported",size); 209 /* size==1 or distributed mat input */ 210 if (lu->options.SolveInitialized && !lu->matmatsolve_iscalled) { 211 /* communication pattern of SOLVEstruct is unlikely created for matmatsolve, 212 thus destroy it and create a new SOLVEstruct. 213 Otherwise it may result in memory corruption or incorrect solution 214 See src/mat/examples/tests/ex125.c */ 215 #if defined(PETSC_USE_COMPLEX) 216 PetscStackCall("SuperLU_DIST:zSolveFinalize",zSolveFinalize(&lu->options, &lu->SOLVEstruct)); 217 #else 218 PetscStackCall("SuperLU_DIST:dSolveFinalize",dSolveFinalize(&lu->options, &lu->SOLVEstruct)); 219 #endif 220 lu->options.SolveInitialized = NO; 221 } 222 ierr = MatCopy(B_mpi,X,SAME_NONZERO_PATTERN);CHKERRQ(ierr); 223 224 ierr = MatGetSize(B_mpi,NULL,&nrhs);CHKERRQ(ierr); 225 226 PetscStackCall("SuperLU_DIST:PStatInit",PStatInit(&stat)); /* Initialize the statistics variables. */ 227 ierr = MatDenseGetArray(X,&bptr);CHKERRQ(ierr); 228 if (lu->MatInputMode == GLOBAL) { /* size == 1 */ 229 #if defined(PETSC_USE_COMPLEX) 230 PetscStackCall("SuperLU_DIST:pzgssvx_ABglobal",pzgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct,(doublecomplex*)bptr, M, nrhs,&lu->grid, &lu->LUstruct, berr, &stat, &info)); 231 #else 232 PetscStackCall("SuperLU_DIST:pdgssvx_ABglobal",pdgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct,bptr, M, nrhs, &lu->grid, &lu->LUstruct, berr, &stat, &info)); 233 #endif 234 } else { /* distributed mat input */ 235 #if defined(PETSC_USE_COMPLEX) 236 PetscStackCall("SuperLU_DIST:pzgssvx",pzgssvx(&lu->options,&lu->A_sup,&lu->ScalePermstruct,(doublecomplex*)bptr,m,nrhs,&lu->grid, &lu->LUstruct,&lu->SOLVEstruct,berr,&stat,&info)); 237 #else 238 PetscStackCall("SuperLU_DIST:pdgssvx",pdgssvx(&lu->options,&lu->A_sup,&lu->ScalePermstruct,bptr,m,nrhs,&lu->grid,&lu->LUstruct,&lu->SOLVEstruct,berr,&stat,&info)); 239 #endif 240 } 241 if (info) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"pdgssvx fails, info: %d\n",info); 242 ierr = MatDenseRestoreArray(X,&bptr);CHKERRQ(ierr); 243 244 if (lu->options.PrintStat) PStatPrint(&lu->options, &stat, &lu->grid); /* Print the statistics. */ 245 PetscStackCall("SuperLU_DIST:PStatFree",PStatFree(&stat)); 246 lu->matsolve_iscalled = PETSC_FALSE; 247 lu->matmatsolve_iscalled = PETSC_TRUE; 248 PetscFunctionReturn(0); 249 } 250 251 /* 252 input: 253 F: numeric Cholesky factor 254 output: 255 nneg: total number of negative pivots 256 nzero: total number of zero pivots 257 npos: (global dimension of F) - nneg - nzero 258 */ 259 static PetscErrorCode MatGetInertia_SuperLU_DIST(Mat F,PetscInt *nneg,PetscInt *nzero,PetscInt *npos) 260 { 261 PetscErrorCode ierr; 262 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)F->data; 263 PetscScalar *diagU=NULL; 264 PetscInt M,i,neg=0,zero=0,pos=0; 265 PetscReal r; 266 267 PetscFunctionBegin; 268 if (!F->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Matrix factor F is not assembled"); 269 if (lu->options.RowPerm != NOROWPERM) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must set NOROWPERM"); 270 ierr = MatGetSize(F,&M,NULL);CHKERRQ(ierr); 271 ierr = PetscMalloc1(M,&diagU);CHKERRQ(ierr); 272 ierr = MatSuperluDistGetDiagU(F,diagU);CHKERRQ(ierr); 273 for (i=0; i<M; i++) { 274 #if defined(PETSC_USE_COMPLEX) 275 r = PetscImaginaryPart(diagU[i])/10.0; 276 if (r< -PETSC_MACHINE_EPSILON || r>PETSC_MACHINE_EPSILON) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"diagU[%d]=%g + i %g is non-real",i,PetscRealPart(diagU[i]),r*10.0); 277 r = PetscRealPart(diagU[i]); 278 #else 279 r = diagU[i]; 280 #endif 281 if (r > 0) { 282 pos++; 283 } else if (r < 0) { 284 neg++; 285 } else zero++; 286 } 287 288 ierr = PetscFree(diagU);CHKERRQ(ierr); 289 if (nneg) *nneg = neg; 290 if (nzero) *nzero = zero; 291 if (npos) *npos = pos; 292 PetscFunctionReturn(0); 293 } 294 295 static PetscErrorCode MatLUFactorNumeric_SuperLU_DIST(Mat F,Mat A,const MatFactorInfo *info) 296 { 297 Mat *tseq,A_seq = NULL; 298 Mat_SeqAIJ *aa,*bb; 299 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)F->data; 300 PetscErrorCode ierr; 301 PetscInt M=A->rmap->N,N=A->cmap->N,i,*ai,*aj,*bi,*bj,nz,rstart,*garray, 302 m=A->rmap->n, colA_start,j,jcol,jB,countA,countB,*bjj,*ajj=NULL; 303 int sinfo; /* SuperLU_Dist info flag is always an int even with long long indices */ 304 PetscMPIInt size; 305 SuperLUStat_t stat; 306 double *berr=0; 307 IS isrow; 308 #if defined(PETSC_USE_COMPLEX) 309 doublecomplex *av, *bv; 310 #else 311 double *av, *bv; 312 #endif 313 314 PetscFunctionBegin; 315 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 316 317 if (lu->MatInputMode == GLOBAL) { /* global mat input */ 318 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"deprecate"); 319 //------------------------------- rm! -------------------- 320 321 if (size > 1) { /* convert mpi A to seq mat A */ 322 ierr = ISCreateStride(PETSC_COMM_SELF,M,0,1,&isrow);CHKERRQ(ierr); 323 ierr = MatCreateSubMatrices(A,1,&isrow,&isrow,MAT_INITIAL_MATRIX,&tseq);CHKERRQ(ierr); 324 ierr = ISDestroy(&isrow);CHKERRQ(ierr); 325 326 A_seq = *tseq; 327 ierr = PetscFree(tseq);CHKERRQ(ierr); 328 aa = (Mat_SeqAIJ*)A_seq->data; 329 } else { 330 PetscBool flg; 331 ierr = PetscObjectTypeCompare((PetscObject)A,MATMPIAIJ,&flg);CHKERRQ(ierr); 332 if (flg) { 333 Mat_MPIAIJ *At = (Mat_MPIAIJ*)A->data; 334 A = At->A; 335 } 336 aa = (Mat_SeqAIJ*)A->data; 337 } 338 339 /* Convert Petsc NR matrix to SuperLU_DIST NC. 340 Note: memories of lu->val, col and row are allocated by CompRow_to_CompCol_dist()! */ 341 if (lu->options.Fact != DOFACT) {/* successive numeric factorization, sparsity pattern is reused. */ 342 PetscStackCall("SuperLU_DIST:Destroy_CompCol_Matrix_dist",Destroy_CompCol_Matrix_dist(&lu->A_sup)); 343 if (lu->FactPattern == SamePattern_SameRowPerm) { 344 lu->options.Fact = SamePattern_SameRowPerm; /* matrix has similar numerical values */ 345 } else { /* lu->FactPattern == SamePattern */ 346 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(N, &lu->grid, &lu->LUstruct)); 347 lu->options.Fact = SamePattern; 348 } 349 } 350 #if defined(PETSC_USE_COMPLEX) 351 PetscStackCall("SuperLU_DIST:zCompRow_to_CompCol_dist",zCompRow_to_CompCol_dist(M,N,aa->nz,(doublecomplex*)aa->a,(int_t*)aa->j,(int_t*)aa->i,&lu->val,&lu->col, &lu->row)); 352 #else 353 PetscStackCall("SuperLU_DIST:dCompRow_to_CompCol_dist",dCompRow_to_CompCol_dist(M,N,aa->nz,aa->a,(int_t*)aa->j,(int_t*)aa->i,&lu->val, &lu->col, &lu->row)); 354 #endif 355 356 /* Create compressed column matrix A_sup. */ 357 #if defined(PETSC_USE_COMPLEX) 358 PetscStackCall("SuperLU_DIST:zCreate_CompCol_Matrix_dist",zCreate_CompCol_Matrix_dist(&lu->A_sup, M, N, aa->nz, lu->val, lu->col, lu->row, SLU_NC, SLU_Z, SLU_GE)); 359 #else 360 PetscStackCall("SuperLU_DIST:dCreate_CompCol_Matrix_dist",dCreate_CompCol_Matrix_dist(&lu->A_sup, M, N, aa->nz, lu->val, lu->col, lu->row, SLU_NC, SLU_D, SLU_GE)); 361 #endif 362 //------------------------------- rm! -------------------- 363 } else { /* distributed mat input */ 364 Mat_MPIAIJ *mat = (Mat_MPIAIJ*)A->data; 365 aa=(Mat_SeqAIJ*)(mat->A)->data; 366 bb=(Mat_SeqAIJ*)(mat->B)->data; 367 ai=aa->i; aj=aa->j; 368 bi=bb->i; bj=bb->j; 369 #if defined(PETSC_USE_COMPLEX) 370 av=(doublecomplex*)aa->a; 371 bv=(doublecomplex*)bb->a; 372 #else 373 av=aa->a; 374 bv=bb->a; 375 #endif 376 rstart = A->rmap->rstart; 377 nz = aa->nz + bb->nz; 378 garray = mat->garray; 379 380 if (lu->options.Fact == DOFACT) { /* first numeric factorization */ 381 #if defined(PETSC_USE_COMPLEX) 382 PetscStackCall("SuperLU_DIST:zallocateA_dist",zallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 383 #else 384 PetscStackCall("SuperLU_DIST:dallocateA_dist",dallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 385 #endif 386 } else { /* successive numeric factorization, sparsity pattern and perm_c are reused. */ 387 if (lu->FactPattern == SamePattern_SameRowPerm) { 388 lu->options.Fact = SamePattern_SameRowPerm; /* matrix has similar numerical values */ 389 } else if (lu->FactPattern == SamePattern) { 390 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(N, &lu->grid, &lu->LUstruct)); /* Deallocate L and U matrices. */ 391 lu->options.Fact = SamePattern; 392 } else if (lu->FactPattern == DOFACT) { 393 PetscStackCall("SuperLU_DIST:Destroy_CompRowLoc_Matrix_dist",Destroy_CompRowLoc_Matrix_dist(&lu->A_sup)); 394 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(N, &lu->grid, &lu->LUstruct)); 395 lu->options.Fact = DOFACT; 396 397 #if defined(PETSC_USE_COMPLEX) 398 PetscStackCall("SuperLU_DIST:zallocateA_dist",zallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 399 #else 400 PetscStackCall("SuperLU_DIST:dallocateA_dist",dallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 401 #endif 402 } else { 403 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"options.Fact must be one of SamePattern SamePattern_SameRowPerm DOFACT"); 404 } 405 } 406 nz = 0; 407 for (i=0; i<m; i++) { 408 lu->row[i] = nz; 409 countA = ai[i+1] - ai[i]; 410 countB = bi[i+1] - bi[i]; 411 if (aj) { 412 ajj = aj + ai[i]; /* ptr to the beginning of this row */ 413 } else { 414 ajj = NULL; 415 } 416 bjj = bj + bi[i]; 417 418 /* B part, smaller col index */ 419 if (aj) { 420 colA_start = rstart + ajj[0]; /* the smallest global col index of A */ 421 } else { /* superlu_dist does not require matrix has diagonal entries, thus aj=NULL would work */ 422 colA_start = rstart; 423 } 424 jB = 0; 425 for (j=0; j<countB; j++) { 426 jcol = garray[bjj[j]]; 427 if (jcol > colA_start) { 428 jB = j; 429 break; 430 } 431 lu->col[nz] = jcol; 432 lu->val[nz++] = *bv++; 433 if (j==countB-1) jB = countB; 434 } 435 436 /* A part */ 437 for (j=0; j<countA; j++) { 438 lu->col[nz] = rstart + ajj[j]; 439 lu->val[nz++] = *av++; 440 } 441 442 /* B part, larger col index */ 443 for (j=jB; j<countB; j++) { 444 lu->col[nz] = garray[bjj[j]]; 445 lu->val[nz++] = *bv++; 446 } 447 } 448 lu->row[m] = nz; 449 450 if (lu->options.Fact == DOFACT) { 451 #if defined(PETSC_USE_COMPLEX) 452 PetscStackCall("SuperLU_DIST:zCreate_CompRowLoc_Matrix_dist",zCreate_CompRowLoc_Matrix_dist(&lu->A_sup, M, N, nz, m, rstart,lu->val, lu->col, lu->row, SLU_NR_loc, SLU_Z, SLU_GE)); 453 #else 454 PetscStackCall("SuperLU_DIST:dCreate_CompRowLoc_Matrix_dist",dCreate_CompRowLoc_Matrix_dist(&lu->A_sup, M, N, nz, m, rstart,lu->val, lu->col, lu->row, SLU_NR_loc, SLU_D, SLU_GE)); 455 #endif 456 } 457 } 458 459 /* Factor the matrix. */ 460 PetscStackCall("SuperLU_DIST:PStatInit",PStatInit(&stat)); /* Initialize the statistics variables. */ 461 if (lu->MatInputMode == GLOBAL) { /* global mat input */ 462 #if defined(PETSC_USE_COMPLEX) 463 PetscStackCall("SuperLU_DIST:pzgssvx_ABglobal",pzgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, M, 0,&lu->grid, &lu->LUstruct, berr, &stat, &sinfo)); 464 #else 465 PetscStackCall("SuperLU_DIST:pdgssvx_ABglobal",pdgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, M, 0,&lu->grid, &lu->LUstruct, berr, &stat, &sinfo)); 466 #endif 467 } else { /* distributed mat input */ 468 #if defined(PETSC_USE_COMPLEX) 469 PetscStackCall("SuperLU_DIST:pzgssvx",pzgssvx(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, m, 0, &lu->grid,&lu->LUstruct, &lu->SOLVEstruct, berr, &stat, &sinfo)); 470 #else 471 PetscStackCall("SuperLU_DIST:pdgssvx",pdgssvx(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, m, 0, &lu->grid,&lu->LUstruct, &lu->SOLVEstruct, berr, &stat, &sinfo)); 472 #endif 473 } 474 475 if (sinfo > 0) { 476 if (A->erroriffailure) { 477 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Zero pivot in row %D",sinfo); 478 } else { 479 if (sinfo <= lu->A_sup.ncol) { 480 F->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT; 481 ierr = PetscInfo1(F,"U(i,i) is exactly zero, i= %D\n",sinfo);CHKERRQ(ierr); 482 } else if (sinfo > lu->A_sup.ncol) { 483 /* 484 number of bytes allocated when memory allocation 485 failure occurred, plus A->ncol. 486 */ 487 F->factorerrortype = MAT_FACTOR_OUTMEMORY; 488 ierr = PetscInfo1(F,"Number of bytes allocated when memory allocation fails %D\n",sinfo);CHKERRQ(ierr); 489 } 490 } 491 } else if (sinfo < 0) { 492 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB, "info = %D, argument in p*gssvx() had an illegal value", sinfo); 493 } 494 495 if (lu->MatInputMode == GLOBAL && size > 1) { 496 ierr = MatDestroy(&A_seq);CHKERRQ(ierr); 497 } 498 499 if (lu->options.PrintStat) { 500 PStatPrint(&lu->options, &stat, &lu->grid); /* Print the statistics. */ 501 } 502 PetscStackCall("SuperLU_DIST:PStatFree",PStatFree(&stat)); 503 F->assembled = PETSC_TRUE; 504 F->preallocated = PETSC_TRUE; 505 lu->options.Fact = FACTORED; /* The factored form of A is supplied. Local option used by this func. only */ 506 PetscFunctionReturn(0); 507 } 508 509 /* Note the Petsc r and c permutations are ignored */ 510 static PetscErrorCode MatLUFactorSymbolic_SuperLU_DIST(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info) 511 { 512 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)F->data; 513 PetscInt M = A->rmap->N,N=A->cmap->N; 514 515 PetscFunctionBegin; 516 /* Initialize the SuperLU process grid. */ 517 PetscStackCall("SuperLU_DIST:superlu_gridinit",superlu_gridinit(lu->comm_superlu, lu->nprow, lu->npcol, &lu->grid)); 518 519 /* Initialize ScalePermstruct and LUstruct. */ 520 PetscStackCall("SuperLU_DIST:ScalePermstructInit",ScalePermstructInit(M, N, &lu->ScalePermstruct)); 521 PetscStackCall("SuperLU_DIST:LUstructInit",LUstructInit(N, &lu->LUstruct)); 522 F->ops->lufactornumeric = MatLUFactorNumeric_SuperLU_DIST; 523 F->ops->solve = MatSolve_SuperLU_DIST; 524 F->ops->matsolve = MatMatSolve_SuperLU_DIST; 525 F->ops->getinertia = NULL; 526 527 if (A->symmetric || A->hermitian) { 528 F->ops->getinertia = MatGetInertia_SuperLU_DIST; 529 } 530 lu->CleanUpSuperLU_Dist = PETSC_TRUE; 531 PetscFunctionReturn(0); 532 } 533 534 static PetscErrorCode MatCholeskyFactorSymbolic_SuperLU_DIST(Mat F,Mat A,IS r,const MatFactorInfo *info) 535 { 536 PetscErrorCode ierr; 537 538 PetscFunctionBegin; 539 if (!A->symmetric) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Input matrix must be symmetric\n"); 540 ierr = MatLUFactorSymbolic_SuperLU_DIST(F,A,r,r,info);CHKERRQ(ierr); 541 F->ops->choleskyfactornumeric = MatLUFactorNumeric_SuperLU_DIST; 542 PetscFunctionReturn(0); 543 } 544 545 static PetscErrorCode MatFactorGetSolverType_aij_superlu_dist(Mat A,MatSolverType *type) 546 { 547 PetscFunctionBegin; 548 *type = MATSOLVERSUPERLU_DIST; 549 PetscFunctionReturn(0); 550 } 551 552 static PetscErrorCode MatView_Info_SuperLU_DIST(Mat A,PetscViewer viewer) 553 { 554 Mat_SuperLU_DIST *lu=(Mat_SuperLU_DIST*)A->data; 555 superlu_dist_options_t options; 556 PetscErrorCode ierr; 557 558 PetscFunctionBegin; 559 /* check if matrix is superlu_dist type */ 560 if (A->ops->solve != MatSolve_SuperLU_DIST) PetscFunctionReturn(0); 561 562 options = lu->options; 563 ierr = PetscViewerASCIIPrintf(viewer,"SuperLU_DIST run parameters:\n");CHKERRQ(ierr); 564 ierr = PetscViewerASCIIPrintf(viewer," Process grid nprow %D x npcol %D \n",lu->nprow,lu->npcol);CHKERRQ(ierr); 565 ierr = PetscViewerASCIIPrintf(viewer," Equilibrate matrix %s \n",PetscBools[options.Equil != NO]);CHKERRQ(ierr); 566 ierr = PetscViewerASCIIPrintf(viewer," Matrix input mode %d \n",lu->MatInputMode);CHKERRQ(ierr); 567 ierr = PetscViewerASCIIPrintf(viewer," Replace tiny pivots %s \n",PetscBools[options.ReplaceTinyPivot != NO]);CHKERRQ(ierr); 568 ierr = PetscViewerASCIIPrintf(viewer," Use iterative refinement %s \n",PetscBools[options.IterRefine == SLU_DOUBLE]);CHKERRQ(ierr); 569 ierr = PetscViewerASCIIPrintf(viewer," Processors in row %d col partition %d \n",lu->nprow,lu->npcol);CHKERRQ(ierr); 570 571 switch (options.RowPerm) { 572 case NOROWPERM: 573 ierr = PetscViewerASCIIPrintf(viewer," Row permutation NOROWPERM\n");CHKERRQ(ierr); 574 break; 575 case LargeDiag_MC64: 576 ierr = PetscViewerASCIIPrintf(viewer," Row permutation LargeDiag_MC64\n");CHKERRQ(ierr); 577 break; 578 case LargeDiag_AWPM: 579 ierr = PetscViewerASCIIPrintf(viewer," Row permutation LargeDiag_AWPM\n");CHKERRQ(ierr); 580 break; 581 case MY_PERMR: 582 ierr = PetscViewerASCIIPrintf(viewer," Row permutation MY_PERMR\n");CHKERRQ(ierr); 583 break; 584 default: 585 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown column permutation"); 586 } 587 588 switch (options.ColPerm) { 589 case NATURAL: 590 ierr = PetscViewerASCIIPrintf(viewer," Column permutation NATURAL\n");CHKERRQ(ierr); 591 break; 592 case MMD_AT_PLUS_A: 593 ierr = PetscViewerASCIIPrintf(viewer," Column permutation MMD_AT_PLUS_A\n");CHKERRQ(ierr); 594 break; 595 case MMD_ATA: 596 ierr = PetscViewerASCIIPrintf(viewer," Column permutation MMD_ATA\n");CHKERRQ(ierr); 597 break; 598 case METIS_AT_PLUS_A: 599 ierr = PetscViewerASCIIPrintf(viewer," Column permutation METIS_AT_PLUS_A\n");CHKERRQ(ierr); 600 break; 601 case PARMETIS: 602 ierr = PetscViewerASCIIPrintf(viewer," Column permutation PARMETIS\n");CHKERRQ(ierr); 603 break; 604 default: 605 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown column permutation"); 606 } 607 608 ierr = PetscViewerASCIIPrintf(viewer," Parallel symbolic factorization %s \n",PetscBools[options.ParSymbFact != NO]);CHKERRQ(ierr); 609 610 if (lu->FactPattern == SamePattern) { 611 ierr = PetscViewerASCIIPrintf(viewer," Repeated factorization SamePattern\n");CHKERRQ(ierr); 612 } else if (lu->FactPattern == SamePattern_SameRowPerm) { 613 ierr = PetscViewerASCIIPrintf(viewer," Repeated factorization SamePattern_SameRowPerm\n");CHKERRQ(ierr); 614 } else if (lu->FactPattern == DOFACT) { 615 ierr = PetscViewerASCIIPrintf(viewer," Repeated factorization DOFACT\n");CHKERRQ(ierr); 616 } else { 617 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown factorization pattern"); 618 } 619 PetscFunctionReturn(0); 620 } 621 622 static PetscErrorCode MatView_SuperLU_DIST(Mat A,PetscViewer viewer) 623 { 624 PetscErrorCode ierr; 625 PetscBool iascii; 626 PetscViewerFormat format; 627 628 PetscFunctionBegin; 629 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 630 if (iascii) { 631 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 632 if (format == PETSC_VIEWER_ASCII_INFO) { 633 ierr = MatView_Info_SuperLU_DIST(A,viewer);CHKERRQ(ierr); 634 } 635 } 636 PetscFunctionReturn(0); 637 } 638 639 static PetscErrorCode MatGetFactor_aij_superlu_dist(Mat A,MatFactorType ftype,Mat *F) 640 { 641 Mat B; 642 Mat_SuperLU_DIST *lu; 643 PetscErrorCode ierr; 644 PetscInt M=A->rmap->N,N=A->cmap->N,indx; 645 PetscMPIInt size; 646 superlu_dist_options_t options; 647 PetscBool flg; 648 const char *colperm[] = {"NATURAL","MMD_AT_PLUS_A","MMD_ATA","METIS_AT_PLUS_A","PARMETIS"}; 649 const char *rowperm[] = {"NOROWPERM","LargeDiag_MC64","LargeDiag_AWPM","MY_PERMR"}; 650 const char *factPattern[] = {"SamePattern","SamePattern_SameRowPerm","DOFACT"}; 651 PetscBool set; 652 653 PetscFunctionBegin; 654 /* Create the factorization matrix */ 655 ierr = MatCreate(PetscObjectComm((PetscObject)A),&B);CHKERRQ(ierr); 656 ierr = MatSetSizes(B,A->rmap->n,A->cmap->n,M,N);CHKERRQ(ierr); 657 ierr = PetscStrallocpy("superlu_dist",&((PetscObject)B)->type_name);CHKERRQ(ierr); 658 ierr = MatSetUp(B);CHKERRQ(ierr); 659 B->ops->getinfo = MatGetInfo_External; 660 B->ops->view = MatView_SuperLU_DIST; 661 B->ops->destroy = MatDestroy_SuperLU_DIST; 662 663 if (ftype == MAT_FACTOR_LU) { 664 B->factortype = MAT_FACTOR_LU; 665 B->ops->lufactorsymbolic = MatLUFactorSymbolic_SuperLU_DIST; 666 } else { 667 B->factortype = MAT_FACTOR_CHOLESKY; 668 B->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SuperLU_DIST; 669 } 670 671 /* set solvertype */ 672 ierr = PetscFree(B->solvertype);CHKERRQ(ierr); 673 ierr = PetscStrallocpy(MATSOLVERSUPERLU_DIST,&B->solvertype);CHKERRQ(ierr); 674 675 ierr = PetscNewLog(B,&lu);CHKERRQ(ierr); 676 B->data = lu; 677 678 /* Set the default input options: 679 options.Fact = DOFACT; 680 options.Equil = YES; 681 options.ParSymbFact = NO; 682 options.ColPerm = METIS_AT_PLUS_A; 683 options.RowPerm = LargeDiag_MC64; 684 options.ReplaceTinyPivot = YES; 685 options.IterRefine = DOUBLE; 686 options.Trans = NOTRANS; 687 options.SolveInitialized = NO; -hold the communication pattern used MatSolve() and MatMatSolve() 688 options.RefineInitialized = NO; 689 options.PrintStat = YES; 690 */ 691 set_default_options_dist(&options); 692 693 ierr = MPI_Comm_dup(PetscObjectComm((PetscObject)A),&(lu->comm_superlu));CHKERRQ(ierr); 694 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 695 /* Default num of process columns and rows */ 696 lu->nprow = (int_t) (0.5 + PetscSqrtReal((PetscReal)size)); 697 if (!lu->nprow) lu->nprow = 1; 698 while (lu->nprow > 0) { 699 lu->npcol = (int_t) (size/lu->nprow); 700 if (size == lu->nprow * lu->npcol) break; 701 lu->nprow--; 702 } 703 704 ierr = PetscOptionsBegin(PetscObjectComm((PetscObject)A),((PetscObject)A)->prefix,"SuperLU_Dist Options","Mat");CHKERRQ(ierr); 705 ierr = PetscOptionsInt("-mat_superlu_dist_r","Number rows in processor partition","None",lu->nprow,(PetscInt*)&lu->nprow,NULL);CHKERRQ(ierr); 706 ierr = PetscOptionsInt("-mat_superlu_dist_c","Number columns in processor partition","None",lu->npcol,(PetscInt*)&lu->npcol,NULL);CHKERRQ(ierr); 707 if (size != lu->nprow * lu->npcol) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Number of processes %d must equal to nprow %d * npcol %d",size,lu->nprow,lu->npcol); 708 709 lu->MatInputMode = DISTRIBUTED; 710 711 ierr = PetscOptionsEnum("-mat_superlu_dist_matinput","Matrix input mode (global or distributed)","None",SuperLU_MatInputModes,(PetscEnum)lu->MatInputMode,(PetscEnum*)&lu->MatInputMode,NULL);CHKERRQ(ierr); 712 //if (lu->MatInputMode == DISTRIBUTED && size == 1) lu->MatInputMode = GLOBAL; 713 714 ierr = PetscOptionsBool("-mat_superlu_dist_equil","Equilibrate matrix","None",options.Equil ? PETSC_TRUE : PETSC_FALSE,&flg,&set);CHKERRQ(ierr); 715 if (set && !flg) options.Equil = NO; 716 717 ierr = PetscOptionsEList("-mat_superlu_dist_rowperm","Row permutation","None",rowperm,4,rowperm[1],&indx,&flg);CHKERRQ(ierr); 718 if (flg) { 719 switch (indx) { 720 case 0: 721 options.RowPerm = NOROWPERM; 722 break; 723 case 1: 724 options.RowPerm = LargeDiag_MC64; 725 break; 726 case 2: 727 options.RowPerm = LargeDiag_AWPM; 728 break; 729 case 3: 730 options.RowPerm = MY_PERMR; 731 break; 732 default: 733 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown row permutation"); 734 } 735 } 736 737 ierr = PetscOptionsEList("-mat_superlu_dist_colperm","Column permutation","None",colperm,5,colperm[3],&indx,&flg);CHKERRQ(ierr); 738 if (flg) { 739 switch (indx) { 740 case 0: 741 options.ColPerm = NATURAL; 742 break; 743 case 1: 744 options.ColPerm = MMD_AT_PLUS_A; 745 break; 746 case 2: 747 options.ColPerm = MMD_ATA; 748 break; 749 case 3: 750 options.ColPerm = METIS_AT_PLUS_A; 751 break; 752 case 4: 753 options.ColPerm = PARMETIS; /* only works for np>1 */ 754 break; 755 default: 756 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown column permutation"); 757 } 758 } 759 760 options.ReplaceTinyPivot = NO; 761 ierr = PetscOptionsBool("-mat_superlu_dist_replacetinypivot","Replace tiny pivots","None",options.ReplaceTinyPivot ? PETSC_TRUE : PETSC_FALSE,&flg,&set);CHKERRQ(ierr); 762 if (set && flg) options.ReplaceTinyPivot = YES; 763 764 options.ParSymbFact = NO; 765 ierr = PetscOptionsBool("-mat_superlu_dist_parsymbfact","Parallel symbolic factorization","None",PETSC_FALSE,&flg,&set);CHKERRQ(ierr); 766 if (set && flg && size>1) { 767 if (lu->MatInputMode == GLOBAL) { 768 #if defined(PETSC_USE_INFO) 769 ierr = PetscInfo(A,"Warning: '-mat_superlu_dist_parsymbfact' is ignored because MatInputMode=GLOBAL\n");CHKERRQ(ierr); 770 #endif 771 } else { 772 #if defined(PETSC_HAVE_PARMETIS) 773 options.ParSymbFact = YES; 774 options.ColPerm = PARMETIS; /* in v2.2, PARMETIS is forced for ParSymbFact regardless of user ordering setting */ 775 #else 776 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"parsymbfact needs PARMETIS"); 777 #endif 778 } 779 } 780 781 lu->FactPattern = SamePattern; 782 ierr = PetscOptionsEList("-mat_superlu_dist_fact","Sparsity pattern for repeated matrix factorization","None",factPattern,3,factPattern[0],&indx,&flg);CHKERRQ(ierr); 783 if (flg) { 784 switch (indx) { 785 case 0: 786 lu->FactPattern = SamePattern; 787 break; 788 case 1: 789 lu->FactPattern = SamePattern_SameRowPerm; 790 break; 791 case 2: 792 lu->FactPattern = DOFACT; 793 break; 794 } 795 } 796 797 options.IterRefine = NOREFINE; 798 ierr = PetscOptionsBool("-mat_superlu_dist_iterrefine","Use iterative refinement","None",options.IterRefine == NOREFINE ? PETSC_FALSE : PETSC_TRUE ,&flg,&set);CHKERRQ(ierr); 799 if (set) { 800 if (flg) options.IterRefine = SLU_DOUBLE; 801 else options.IterRefine = NOREFINE; 802 } 803 804 if (PetscLogPrintInfo) options.PrintStat = YES; 805 else options.PrintStat = NO; 806 ierr = PetscOptionsBool("-mat_superlu_dist_statprint","Print factorization information","None",(PetscBool)options.PrintStat,(PetscBool*)&options.PrintStat,NULL);CHKERRQ(ierr); 807 ierr = PetscOptionsEnd();CHKERRQ(ierr); 808 809 lu->options = options; 810 lu->options.Fact = DOFACT; 811 lu->matsolve_iscalled = PETSC_FALSE; 812 lu->matmatsolve_iscalled = PETSC_FALSE; 813 814 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverType_C",MatFactorGetSolverType_aij_superlu_dist);CHKERRQ(ierr); 815 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSuperluDistGetDiagU_C",MatSuperluDistGetDiagU_SuperLU_DIST);CHKERRQ(ierr); 816 817 *F = B; 818 PetscFunctionReturn(0); 819 } 820 821 PETSC_EXTERN PetscErrorCode MatSolverTypeRegister_SuperLU_DIST(void) 822 { 823 PetscErrorCode ierr; 824 PetscFunctionBegin; 825 ierr = MatSolverTypeRegister(MATSOLVERSUPERLU_DIST,MATMPIAIJ, MAT_FACTOR_LU,MatGetFactor_aij_superlu_dist);CHKERRQ(ierr); 826 ierr = MatSolverTypeRegister(MATSOLVERSUPERLU_DIST,MATSEQAIJ, MAT_FACTOR_LU,MatGetFactor_aij_superlu_dist);CHKERRQ(ierr); 827 ierr = MatSolverTypeRegister(MATSOLVERSUPERLU_DIST,MATMPIAIJ, MAT_FACTOR_CHOLESKY,MatGetFactor_aij_superlu_dist);CHKERRQ(ierr); 828 ierr = MatSolverTypeRegister(MATSOLVERSUPERLU_DIST,MATSEQAIJ, MAT_FACTOR_CHOLESKY,MatGetFactor_aij_superlu_dist);CHKERRQ(ierr); 829 PetscFunctionReturn(0); 830 } 831 832 /*MC 833 MATSOLVERSUPERLU_DIST - Parallel direct solver package for LU factorization 834 835 Use ./configure --download-superlu_dist --download-parmetis --download-metis --download-ptscotch to have PETSc installed with SuperLU_DIST 836 837 Use -pc_type lu -pc_factor_mat_solver_type superlu_dist to use this direct solver 838 839 Works with AIJ matrices 840 841 Options Database Keys: 842 + -mat_superlu_dist_r <n> - number of rows in processor partition 843 . -mat_superlu_dist_c <n> - number of columns in processor partition 844 . -mat_superlu_dist_matinput <0,1> - matrix input mode; 0=global, 1=distributed 845 . -mat_superlu_dist_equil - equilibrate the matrix 846 . -mat_superlu_dist_rowperm <NOROWPERM,LargeDiag_MC64,LargeDiag_AWPM,MY_PERMR> - row permutation 847 . -mat_superlu_dist_colperm <MMD_AT_PLUS_A,MMD_ATA,NATURAL> - column permutation 848 . -mat_superlu_dist_replacetinypivot - replace tiny pivots 849 . -mat_superlu_dist_fact <SamePattern> - (choose one of) SamePattern SamePattern_SameRowPerm DOFACT 850 . -mat_superlu_dist_iterrefine - use iterative refinement 851 - -mat_superlu_dist_statprint - print factorization information 852 853 Level: beginner 854 855 .seealso: PCLU 856 857 .seealso: PCFactorSetMatSolverType(), MatSolverType 858 859 M*/ 860