1 2 /* 3 Provides an interface to the SuperLU_DIST_2.2 sparse solver 4 */ 5 6 #include <../src/mat/impls/aij/seq/aij.h> 7 #include <../src/mat/impls/aij/mpi/mpiaij.h> 8 #include <petsctime.h> 9 #if defined(PETSC_HAVE_STDLIB_H) /* This is to get around weird problem with SuperLU on cray */ 10 #include <stdlib.h> 11 #endif 12 13 #if defined(PETSC_USE_64BIT_INDICES) 14 /* ugly SuperLU_Dist variable telling it to use long long int */ 15 #define _LONGINT 16 #endif 17 18 EXTERN_C_BEGIN 19 #if defined(PETSC_USE_COMPLEX) 20 #include <superlu_zdefs.h> 21 #else 22 #include <superlu_ddefs.h> 23 #endif 24 EXTERN_C_END 25 26 /* 27 GLOBAL - The sparse matrix and right hand side are all stored initially on process 0. Should be called centralized 28 DISTRIBUTED - The sparse matrix and right hand size are initially stored across the entire MPI communicator. 29 */ 30 typedef enum {GLOBAL,DISTRIBUTED} SuperLU_MatInputMode; 31 const char *SuperLU_MatInputModes[] = {"GLOBAL","DISTRIBUTED","SuperLU_MatInputMode","PETSC_",0}; 32 33 typedef struct { 34 int_t nprow,npcol,*row,*col; 35 gridinfo_t grid; 36 superlu_options_t options; 37 SuperMatrix A_sup; 38 ScalePermstruct_t ScalePermstruct; 39 LUstruct_t LUstruct; 40 int StatPrint; 41 SuperLU_MatInputMode MatInputMode; 42 SOLVEstruct_t SOLVEstruct; 43 fact_t FactPattern; 44 MPI_Comm comm_superlu; 45 #if defined(PETSC_USE_COMPLEX) 46 doublecomplex *val; 47 #else 48 double *val; 49 #endif 50 PetscBool matsolve_iscalled,matmatsolve_iscalled; 51 PetscBool CleanUpSuperLU_Dist; /* Flag to clean up (non-global) SuperLU objects during Destroy */ 52 } Mat_SuperLU_DIST; 53 54 extern PetscErrorCode MatFactorInfo_SuperLU_DIST(Mat,PetscViewer); 55 extern PetscErrorCode MatLUFactorNumeric_SuperLU_DIST(Mat,Mat,const MatFactorInfo*); 56 extern PetscErrorCode MatDestroy_SuperLU_DIST(Mat); 57 extern PetscErrorCode MatView_SuperLU_DIST(Mat,PetscViewer); 58 extern PetscErrorCode MatSolve_SuperLU_DIST(Mat,Vec,Vec); 59 extern PetscErrorCode MatLUFactorSymbolic_SuperLU_DIST(Mat,Mat,IS,IS,const MatFactorInfo*); 60 extern PetscErrorCode MatDestroy_MPIAIJ(Mat); 61 62 #undef __FUNCT__ 63 #define __FUNCT__ "MatDestroy_SuperLU_DIST" 64 PetscErrorCode MatDestroy_SuperLU_DIST(Mat A) 65 { 66 PetscErrorCode ierr; 67 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->spptr; 68 PetscBool flg; 69 70 PetscFunctionBegin; 71 if (lu && lu->CleanUpSuperLU_Dist) { 72 /* Deallocate SuperLU_DIST storage */ 73 if (lu->MatInputMode == GLOBAL) { 74 PetscStackCall("SuperLU_DIST:Destroy_CompCol_Matrix_dist",Destroy_CompCol_Matrix_dist(&lu->A_sup)); 75 } else { 76 PetscStackCall("SuperLU_DIST:Destroy_CompRowLoc_Matrix_dist",Destroy_CompRowLoc_Matrix_dist(&lu->A_sup)); 77 if (lu->options.SolveInitialized) { 78 #if defined(PETSC_USE_COMPLEX) 79 PetscStackCall("SuperLU_DIST:zSolveFinalize",zSolveFinalize(&lu->options, &lu->SOLVEstruct)); 80 #else 81 PetscStackCall("SuperLU_DIST:dSolveFinalize",dSolveFinalize(&lu->options, &lu->SOLVEstruct)); 82 #endif 83 } 84 } 85 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(A->cmap->N, &lu->grid, &lu->LUstruct)); 86 PetscStackCall("SuperLU_DIST:ScalePermstructFree",ScalePermstructFree(&lu->ScalePermstruct)); 87 PetscStackCall("SuperLU_DIST:LUstructFree",LUstructFree(&lu->LUstruct)); 88 89 /* Release the SuperLU_DIST process grid. */ 90 PetscStackCall("SuperLU_DIST:superlu_gridexit",superlu_gridexit(&lu->grid)); 91 ierr = MPI_Comm_free(&(lu->comm_superlu));CHKERRQ(ierr); 92 } 93 ierr = PetscFree(A->spptr);CHKERRQ(ierr); 94 95 ierr = PetscObjectTypeCompare((PetscObject)A,MATSEQAIJ,&flg);CHKERRQ(ierr); 96 if (flg) { 97 ierr = MatDestroy_SeqAIJ(A);CHKERRQ(ierr); 98 } else { 99 ierr = MatDestroy_MPIAIJ(A);CHKERRQ(ierr); 100 } 101 PetscFunctionReturn(0); 102 } 103 104 #undef __FUNCT__ 105 #define __FUNCT__ "MatSolve_SuperLU_DIST" 106 PetscErrorCode MatSolve_SuperLU_DIST(Mat A,Vec b_mpi,Vec x) 107 { 108 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->spptr; 109 PetscErrorCode ierr; 110 PetscMPIInt size; 111 PetscInt m=A->rmap->n,M=A->rmap->N,N=A->cmap->N; 112 SuperLUStat_t stat; 113 double berr[1]; 114 PetscScalar *bptr; 115 PetscInt nrhs=1; 116 Vec x_seq; 117 IS iden; 118 VecScatter scat; 119 int info; /* SuperLU_Dist info code is ALWAYS an int, even with long long indices */ 120 static PetscBool cite = PETSC_FALSE; 121 122 PetscFunctionBegin; 123 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); 124 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 125 if (size > 1 && lu->MatInputMode == GLOBAL) { 126 /* global mat input, convert b to x_seq */ 127 ierr = VecCreateSeq(PETSC_COMM_SELF,N,&x_seq);CHKERRQ(ierr); 128 ierr = ISCreateStride(PETSC_COMM_SELF,N,0,1,&iden);CHKERRQ(ierr); 129 ierr = VecScatterCreate(b_mpi,iden,x_seq,iden,&scat);CHKERRQ(ierr); 130 ierr = ISDestroy(&iden);CHKERRQ(ierr); 131 132 ierr = VecScatterBegin(scat,b_mpi,x_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 133 ierr = VecScatterEnd(scat,b_mpi,x_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 134 ierr = VecGetArray(x_seq,&bptr);CHKERRQ(ierr); 135 } else { /* size==1 || distributed mat input */ 136 if (lu->options.SolveInitialized && !lu->matsolve_iscalled) { 137 /* see comments in MatMatSolve() */ 138 #if defined(PETSC_USE_COMPLEX) 139 PetscStackCall("SuperLU_DIST:zSolveFinalize",zSolveFinalize(&lu->options, &lu->SOLVEstruct)); 140 #else 141 PetscStackCall("SuperLU_DIST:dSolveFinalize",dSolveFinalize(&lu->options, &lu->SOLVEstruct)); 142 #endif 143 lu->options.SolveInitialized = NO; 144 } 145 ierr = VecCopy(b_mpi,x);CHKERRQ(ierr); 146 ierr = VecGetArray(x,&bptr);CHKERRQ(ierr); 147 } 148 149 if (lu->options.Fact != FACTORED) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"SuperLU_DIST options.Fact mush equal FACTORED"); 150 151 PetscStackCall("SuperLU_DIST:PStatInit",PStatInit(&stat)); /* Initialize the statistics variables. */ 152 if (lu->MatInputMode == GLOBAL) { 153 #if defined(PETSC_USE_COMPLEX) 154 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)); 155 #else 156 PetscStackCall("SuperLU_DIST:pdgssvx_ABglobal",pdgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct,bptr,M,nrhs,&lu->grid,&lu->LUstruct,berr,&stat,&info)); 157 #endif 158 } else { /* distributed mat input */ 159 #if defined(PETSC_USE_COMPLEX) 160 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)); 161 #else 162 PetscStackCall("SuperLU_DIST:pdgssvx",pdgssvx(&lu->options,&lu->A_sup,&lu->ScalePermstruct,bptr,m,nrhs,&lu->grid,&lu->LUstruct,&lu->SOLVEstruct,berr,&stat,&info)); 163 #endif 164 } 165 if (info) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"pdgssvx fails, info: %d\n",info); 166 167 if (lu->options.PrintStat) PStatPrint(&lu->options, &stat, &lu->grid); /* Print the statistics. */ 168 PetscStackCall("SuperLU_DIST:PStatFree",PStatFree(&stat)); 169 170 if (size > 1 && lu->MatInputMode == GLOBAL) { 171 /* convert seq x to mpi x */ 172 ierr = VecRestoreArray(x_seq,&bptr);CHKERRQ(ierr); 173 ierr = VecScatterBegin(scat,x_seq,x,INSERT_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 174 ierr = VecScatterEnd(scat,x_seq,x,INSERT_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 175 ierr = VecScatterDestroy(&scat);CHKERRQ(ierr); 176 ierr = VecDestroy(&x_seq);CHKERRQ(ierr); 177 } else { 178 ierr = VecRestoreArray(x,&bptr);CHKERRQ(ierr); 179 180 lu->matsolve_iscalled = PETSC_TRUE; 181 lu->matmatsolve_iscalled = PETSC_FALSE; 182 } 183 PetscFunctionReturn(0); 184 } 185 186 #undef __FUNCT__ 187 #define __FUNCT__ "MatMatSolve_SuperLU_DIST" 188 PetscErrorCode MatMatSolve_SuperLU_DIST(Mat A,Mat B_mpi,Mat X) 189 { 190 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)A->spptr; 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 #undef __FUNCT__ 253 #define __FUNCT__ "MatLUFactorNumeric_SuperLU_DIST" 254 PetscErrorCode MatLUFactorNumeric_SuperLU_DIST(Mat F,Mat A,const MatFactorInfo *info) 255 { 256 Mat *tseq,A_seq = NULL; 257 Mat_SeqAIJ *aa,*bb; 258 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)(F)->spptr; 259 PetscErrorCode ierr; 260 PetscInt M=A->rmap->N,N=A->cmap->N,i,*ai,*aj,*bi,*bj,nz,rstart,*garray, 261 m=A->rmap->n, colA_start,j,jcol,jB,countA,countB,*bjj,*ajj; 262 int sinfo; /* SuperLU_Dist info flag is always an int even with long long indices */ 263 PetscMPIInt size; 264 SuperLUStat_t stat; 265 double *berr=0; 266 IS isrow; 267 PetscLogDouble time0,time,time_min,time_max; 268 Mat F_diag=NULL; 269 #if defined(PETSC_USE_COMPLEX) 270 doublecomplex *av, *bv; 271 #else 272 double *av, *bv; 273 #endif 274 275 PetscFunctionBegin; 276 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 277 278 if (lu->options.PrintStat) { /* collect time for mat conversion */ 279 ierr = MPI_Barrier(PetscObjectComm((PetscObject)A));CHKERRQ(ierr); 280 ierr = PetscTime(&time0);CHKERRQ(ierr); 281 } 282 283 if (lu->MatInputMode == GLOBAL) { /* global mat input */ 284 if (size > 1) { /* convert mpi A to seq mat A */ 285 ierr = ISCreateStride(PETSC_COMM_SELF,M,0,1,&isrow);CHKERRQ(ierr); 286 ierr = MatGetSubMatrices(A,1,&isrow,&isrow,MAT_INITIAL_MATRIX,&tseq);CHKERRQ(ierr); 287 ierr = ISDestroy(&isrow);CHKERRQ(ierr); 288 289 A_seq = *tseq; 290 ierr = PetscFree(tseq);CHKERRQ(ierr); 291 aa = (Mat_SeqAIJ*)A_seq->data; 292 } else { 293 PetscBool flg; 294 ierr = PetscObjectTypeCompare((PetscObject)A,MATMPIAIJ,&flg);CHKERRQ(ierr); 295 if (flg) { 296 Mat_MPIAIJ *At = (Mat_MPIAIJ*)A->data; 297 A = At->A; 298 } 299 aa = (Mat_SeqAIJ*)A->data; 300 } 301 302 /* Convert Petsc NR matrix to SuperLU_DIST NC. 303 Note: memories of lu->val, col and row are allocated by CompRow_to_CompCol_dist()! */ 304 if (lu->options.Fact != DOFACT) {/* successive numeric factorization, sparsity pattern is reused. */ 305 PetscStackCall("SuperLU_DIST:Destroy_CompCol_Matrix_dist",Destroy_CompCol_Matrix_dist(&lu->A_sup)); 306 if (lu->FactPattern == SamePattern_SameRowPerm) { 307 lu->options.Fact = SamePattern_SameRowPerm; /* matrix has similar numerical values */ 308 } else { /* lu->FactPattern == SamePattern */ 309 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(N, &lu->grid, &lu->LUstruct)); 310 lu->options.Fact = SamePattern; 311 } 312 } 313 #if defined(PETSC_USE_COMPLEX) 314 PetscStackCall("SuperLU_DIST:zCompRow_to_CompCol_dist",zCompRow_to_CompCol_dist(M,N,aa->nz,(doublecomplex*)aa->a,aa->j,aa->i,&lu->val,&lu->col, &lu->row)); 315 #else 316 PetscStackCall("SuperLU_DIST:dCompRow_to_CompCol_dist",dCompRow_to_CompCol_dist(M,N,aa->nz,aa->a,aa->j,aa->i,&lu->val, &lu->col, &lu->row)); 317 #endif 318 319 /* Create compressed column matrix A_sup. */ 320 #if defined(PETSC_USE_COMPLEX) 321 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)); 322 #else 323 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)); 324 #endif 325 } else { /* distributed mat input */ 326 Mat_MPIAIJ *mat = (Mat_MPIAIJ*)A->data; 327 aa=(Mat_SeqAIJ*)(mat->A)->data; 328 bb=(Mat_SeqAIJ*)(mat->B)->data; 329 ai=aa->i; aj=aa->j; 330 bi=bb->i; bj=bb->j; 331 #if defined(PETSC_USE_COMPLEX) 332 av=(doublecomplex*)aa->a; 333 bv=(doublecomplex*)bb->a; 334 #else 335 av=aa->a; 336 bv=bb->a; 337 #endif 338 rstart = A->rmap->rstart; 339 nz = aa->nz + bb->nz; 340 garray = mat->garray; 341 342 if (lu->options.Fact == DOFACT) { /* first numeric factorization */ 343 #if defined(PETSC_USE_COMPLEX) 344 PetscStackCall("SuperLU_DIST:zallocateA_dist",zallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 345 #else 346 PetscStackCall("SuperLU_DIST:dallocateA_dist",dallocateA_dist(m, nz, &lu->val, &lu->col, &lu->row)); 347 #endif 348 } else { /* successive numeric factorization, sparsity pattern and perm_c are reused. */ 349 /* Destroy_CompRowLoc_Matrix_dist(&lu->A_sup); */ /* this leads to crash! However, see SuperLU_DIST_2.5/EXAMPLE/pzdrive2.c */ 350 if (lu->FactPattern == SamePattern_SameRowPerm) { 351 lu->options.Fact = SamePattern_SameRowPerm; /* matrix has similar numerical values */ 352 } else { 353 PetscStackCall("SuperLU_DIST:Destroy_LU",Destroy_LU(N, &lu->grid, &lu->LUstruct)); /* Deallocate storage associated with the L and U matrices. */ 354 lu->options.Fact = SamePattern; 355 } 356 } 357 nz = 0; 358 for (i=0; i<m; i++) { 359 lu->row[i] = nz; 360 countA = ai[i+1] - ai[i]; 361 countB = bi[i+1] - bi[i]; 362 ajj = aj + ai[i]; /* ptr to the beginning of this row */ 363 bjj = bj + bi[i]; 364 365 /* B part, smaller col index */ 366 colA_start = rstart + ajj[0]; /* the smallest global col index of A */ 367 jB = 0; 368 for (j=0; j<countB; j++) { 369 jcol = garray[bjj[j]]; 370 if (jcol > colA_start) { 371 jB = j; 372 break; 373 } 374 lu->col[nz] = jcol; 375 lu->val[nz++] = *bv++; 376 if (j==countB-1) jB = countB; 377 } 378 379 /* A part */ 380 for (j=0; j<countA; j++) { 381 lu->col[nz] = rstart + ajj[j]; 382 lu->val[nz++] = *av++; 383 } 384 385 /* B part, larger col index */ 386 for (j=jB; j<countB; j++) { 387 lu->col[nz] = garray[bjj[j]]; 388 lu->val[nz++] = *bv++; 389 } 390 } 391 lu->row[m] = nz; 392 #if defined(PETSC_USE_COMPLEX) 393 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)); 394 #else 395 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)); 396 #endif 397 } 398 if (lu->options.PrintStat) { 399 ierr = PetscTime(&time);CHKERRQ(ierr); 400 time0 = time - time0; 401 } 402 403 /* Factor the matrix. */ 404 PetscStackCall("SuperLU_DIST:PStatInit",PStatInit(&stat)); /* Initialize the statistics variables. */ 405 if (lu->MatInputMode == GLOBAL) { /* global mat input */ 406 #if defined(PETSC_USE_COMPLEX) 407 PetscStackCall("SuperLU_DIST:pzgssvx_ABglobal",pzgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, M, 0,&lu->grid, &lu->LUstruct, berr, &stat, &sinfo)); 408 #else 409 PetscStackCall("SuperLU_DIST:pdgssvx_ABglobal",pdgssvx_ABglobal(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, M, 0,&lu->grid, &lu->LUstruct, berr, &stat, &sinfo)); 410 #endif 411 } else { /* distributed mat input */ 412 #if defined(PETSC_USE_COMPLEX) 413 PetscStackCall("SuperLU_DIST:pzgssvx",pzgssvx(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, m, 0, &lu->grid,&lu->LUstruct, &lu->SOLVEstruct, berr, &stat, &sinfo)); 414 if (sinfo) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"pzgssvx fails, info: %d\n",sinfo); 415 #else 416 PetscStackCall("SuperLU_DIST:pdgssvx",pdgssvx(&lu->options, &lu->A_sup, &lu->ScalePermstruct, 0, m, 0, &lu->grid,&lu->LUstruct, &lu->SOLVEstruct, berr, &stat, &sinfo)); 417 if (sinfo) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"pdgssvx fails, info: %d\n",sinfo); 418 #endif 419 } 420 421 if (lu->MatInputMode == GLOBAL && size > 1) { 422 ierr = MatDestroy(&A_seq);CHKERRQ(ierr); 423 } 424 425 if (lu->options.PrintStat) { 426 ierr = MPI_Reduce(&time0,&time_max,1,MPI_DOUBLE,MPI_MAX,0,PetscObjectComm((PetscObject)A)); 427 ierr = MPI_Reduce(&time0,&time_min,1,MPI_DOUBLE,MPI_MIN,0,PetscObjectComm((PetscObject)A)); 428 ierr = MPI_Reduce(&time0,&time,1,MPI_DOUBLE,MPI_SUM,0,PetscObjectComm((PetscObject)A)); 429 time = time/size; /* average time */ 430 ierr = PetscPrintf(PetscObjectComm((PetscObject)A), " Mat conversion(PETSc->SuperLU_DIST) time (max/min/avg): \n %g / %g / %g\n",time_max,time_min,time);CHKERRQ(ierr); 431 PStatPrint(&lu->options, &stat, &lu->grid); /* Print the statistics. */ 432 } 433 PStatFree(&stat); 434 if (size > 1) { 435 F_diag = ((Mat_MPIAIJ*)(F)->data)->A; 436 F_diag->assembled = PETSC_TRUE; 437 } 438 (F)->assembled = PETSC_TRUE; 439 (F)->preallocated = PETSC_TRUE; 440 lu->options.Fact = FACTORED; /* The factored form of A is supplied. Local option used by this func. only */ 441 PetscFunctionReturn(0); 442 } 443 444 /* Note the Petsc r and c permutations are ignored */ 445 #undef __FUNCT__ 446 #define __FUNCT__ "MatLUFactorSymbolic_SuperLU_DIST" 447 PetscErrorCode MatLUFactorSymbolic_SuperLU_DIST(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info) 448 { 449 Mat_SuperLU_DIST *lu = (Mat_SuperLU_DIST*)F->spptr; 450 PetscInt M = A->rmap->N,N=A->cmap->N; 451 452 PetscFunctionBegin; 453 /* Initialize the SuperLU process grid. */ 454 PetscStackCall("SuperLU_DIST:superlu_gridinit",superlu_gridinit(lu->comm_superlu, lu->nprow, lu->npcol, &lu->grid)); 455 456 /* Initialize ScalePermstruct and LUstruct. */ 457 PetscStackCall("SuperLU_DIST:ScalePermstructInit",ScalePermstructInit(M, N, &lu->ScalePermstruct)); 458 PetscStackCall("SuperLU_DIST:LUstructInit",LUstructInit(M, N, &lu->LUstruct)); 459 F->ops->lufactornumeric = MatLUFactorNumeric_SuperLU_DIST; 460 F->ops->solve = MatSolve_SuperLU_DIST; 461 F->ops->matsolve = MatMatSolve_SuperLU_DIST; 462 lu->CleanUpSuperLU_Dist = PETSC_TRUE; 463 PetscFunctionReturn(0); 464 } 465 466 #undef __FUNCT__ 467 #define __FUNCT__ "MatFactorGetSolverPackage_aij_superlu_dist" 468 PetscErrorCode MatFactorGetSolverPackage_aij_superlu_dist(Mat A,const MatSolverPackage *type) 469 { 470 PetscFunctionBegin; 471 *type = MATSOLVERSUPERLU_DIST; 472 PetscFunctionReturn(0); 473 } 474 475 #undef __FUNCT__ 476 #define __FUNCT__ "MatGetFactor_aij_superlu_dist" 477 PETSC_EXTERN PetscErrorCode MatGetFactor_aij_superlu_dist(Mat A,MatFactorType ftype,Mat *F) 478 { 479 Mat B; 480 Mat_SuperLU_DIST *lu; 481 PetscErrorCode ierr; 482 PetscInt M=A->rmap->N,N=A->cmap->N,indx; 483 PetscMPIInt size; 484 superlu_options_t options; 485 PetscBool flg; 486 const char *colperm[] = {"NATURAL","MMD_AT_PLUS_A","MMD_ATA","METIS_AT_PLUS_A","PARMETIS"}; 487 const char *rowperm[] = {"LargeDiag","NATURAL"}; 488 const char *factPattern[] = {"SamePattern","SamePattern_SameRowPerm"}; 489 490 PetscFunctionBegin; 491 /* Create the factorization matrix */ 492 ierr = MatCreate(PetscObjectComm((PetscObject)A),&B);CHKERRQ(ierr); 493 ierr = MatSetSizes(B,A->rmap->n,A->cmap->n,M,N);CHKERRQ(ierr); 494 ierr = MatSetType(B,((PetscObject)A)->type_name);CHKERRQ(ierr); 495 ierr = MatSeqAIJSetPreallocation(B,0,NULL); 496 ierr = MatMPIAIJSetPreallocation(B,0,NULL,0,NULL);CHKERRQ(ierr); 497 498 B->ops->lufactorsymbolic = MatLUFactorSymbolic_SuperLU_DIST; 499 B->ops->view = MatView_SuperLU_DIST; 500 B->ops->destroy = MatDestroy_SuperLU_DIST; 501 502 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverPackage_C",MatFactorGetSolverPackage_aij_superlu_dist);CHKERRQ(ierr); 503 504 B->factortype = MAT_FACTOR_LU; 505 506 ierr = PetscNewLog(B,Mat_SuperLU_DIST,&lu);CHKERRQ(ierr); 507 B->spptr = lu; 508 509 /* Set the default input options: 510 options.Fact = DOFACT; 511 options.Equil = YES; 512 options.ParSymbFact = NO; 513 options.ColPerm = METIS_AT_PLUS_A; 514 options.RowPerm = LargeDiag; 515 options.ReplaceTinyPivot = YES; 516 options.IterRefine = DOUBLE; 517 options.Trans = NOTRANS; 518 options.SolveInitialized = NO; -hold the communication pattern used MatSolve() and MatMatSolve() 519 options.RefineInitialized = NO; 520 options.PrintStat = YES; 521 */ 522 set_default_options_dist(&options); 523 524 ierr = MPI_Comm_dup(PetscObjectComm((PetscObject)A),&(lu->comm_superlu));CHKERRQ(ierr); 525 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 526 /* Default num of process columns and rows */ 527 ierr = PetscMPIIntCast((PetscInt)(0.5 + PetscSqrtReal((PetscReal)size)),&lu->npcol);CHKERRQ(ierr); 528 if (!lu->npcol) lu->npcol = 1; 529 while (lu->npcol > 0) { 530 ierr = PetscMPIIntCast(size/lu->npcol,&lu->nprow);CHKERRQ(ierr); 531 if (size == lu->nprow * lu->npcol) break; 532 lu->npcol--; 533 } 534 535 ierr = PetscOptionsBegin(PetscObjectComm((PetscObject)A),((PetscObject)A)->prefix,"SuperLU_Dist Options","Mat");CHKERRQ(ierr); 536 ierr = PetscOptionsInt("-mat_superlu_dist_r","Number rows in processor partition","None",lu->nprow,&lu->nprow,NULL);CHKERRQ(ierr); 537 ierr = PetscOptionsInt("-mat_superlu_dist_c","Number columns in processor partition","None",lu->npcol,&lu->npcol,NULL);CHKERRQ(ierr); 538 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); 539 540 lu->MatInputMode = DISTRIBUTED; 541 542 ierr = PetscOptionsEnum("-mat_superlu_dist_matinput","Matrix input mode (global or distributed)","None",SuperLU_MatInputModes,(PetscEnum)lu->MatInputMode,(PetscEnum*)&lu->MatInputMode,NULL);CHKERRQ(ierr); 543 if (lu->MatInputMode == DISTRIBUTED && size == 1) lu->MatInputMode = GLOBAL; 544 545 ierr = PetscOptionsBool("-mat_superlu_dist_equil","Equilibrate matrix","None",PETSC_TRUE,&flg,0);CHKERRQ(ierr); 546 if (!flg) options.Equil = NO; 547 548 ierr = PetscOptionsEList("-mat_superlu_dist_rowperm","Row permutation","None",rowperm,2,rowperm[0],&indx,&flg);CHKERRQ(ierr); 549 if (flg) { 550 switch (indx) { 551 case 0: 552 options.RowPerm = LargeDiag; 553 break; 554 case 1: 555 options.RowPerm = NOROWPERM; 556 break; 557 } 558 } 559 560 ierr = PetscOptionsEList("-mat_superlu_dist_colperm","Column permutation","None",colperm,5,colperm[3],&indx,&flg);CHKERRQ(ierr); 561 if (flg) { 562 switch (indx) { 563 case 0: 564 options.ColPerm = NATURAL; 565 break; 566 case 1: 567 options.ColPerm = MMD_AT_PLUS_A; 568 break; 569 case 2: 570 options.ColPerm = MMD_ATA; 571 break; 572 case 3: 573 options.ColPerm = METIS_AT_PLUS_A; 574 break; 575 case 4: 576 options.ColPerm = PARMETIS; /* only works for np>1 */ 577 break; 578 default: 579 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown column permutation"); 580 } 581 } 582 583 ierr = PetscOptionsBool("-mat_superlu_dist_replacetinypivot","Replace tiny pivots","None",PETSC_TRUE,&flg,0);CHKERRQ(ierr); 584 if (!flg) options.ReplaceTinyPivot = NO; 585 586 options.ParSymbFact = NO; 587 588 ierr = PetscOptionsBool("-mat_superlu_dist_parsymbfact","Parallel symbolic factorization","None",PETSC_FALSE,&flg,0);CHKERRQ(ierr); 589 if (flg) { 590 #if defined(PETSC_HAVE_PARMETIS) 591 options.ParSymbFact = YES; 592 options.ColPerm = PARMETIS; /* in v2.2, PARMETIS is forced for ParSymbFact regardless of user ordering setting */ 593 #else 594 printf("parsymbfact needs PARMETIS"); 595 #endif 596 } 597 598 lu->FactPattern = SamePattern_SameRowPerm; 599 600 ierr = PetscOptionsEList("-mat_superlu_dist_fact","Sparsity pattern for repeated matrix factorization","None",factPattern,2,factPattern[1],&indx,&flg);CHKERRQ(ierr); 601 if (flg) { 602 switch (indx) { 603 case 0: 604 lu->FactPattern = SamePattern; 605 break; 606 case 1: 607 lu->FactPattern = SamePattern_SameRowPerm; 608 break; 609 } 610 } 611 612 options.IterRefine = NOREFINE; 613 ierr = PetscOptionsBool("-mat_superlu_dist_iterrefine","Use iterative refinement","None",PETSC_FALSE,&flg,0);CHKERRQ(ierr); 614 if (flg) options.IterRefine = SLU_DOUBLE; 615 616 if (PetscLogPrintInfo) options.PrintStat = YES; 617 else options.PrintStat = NO; 618 ierr = PetscOptionsBool("-mat_superlu_dist_statprint","Print factorization information","None", 619 (PetscBool)options.PrintStat,(PetscBool*)&options.PrintStat,0);CHKERRQ(ierr); 620 PetscOptionsEnd(); 621 622 lu->options = options; 623 lu->options.Fact = DOFACT; 624 lu->matsolve_iscalled = PETSC_FALSE; 625 lu->matmatsolve_iscalled = PETSC_FALSE; 626 627 *F = B; 628 PetscFunctionReturn(0); 629 } 630 631 #undef __FUNCT__ 632 #define __FUNCT__ "MatGetFactor_seqaij_superlu_dist" 633 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_superlu_dist(Mat A,MatFactorType ftype,Mat *F) 634 { 635 PetscErrorCode ierr; 636 637 PetscFunctionBegin; 638 ierr = MatGetFactor_aij_superlu_dist(A,ftype,F);CHKERRQ(ierr); 639 PetscFunctionReturn(0); 640 } 641 642 #undef __FUNCT__ 643 #define __FUNCT__ "MatGetFactor_mpiaij_superlu_dist" 644 PETSC_EXTERN PetscErrorCode MatGetFactor_mpiaij_superlu_dist(Mat A,MatFactorType ftype,Mat *F) 645 { 646 PetscErrorCode ierr; 647 648 PetscFunctionBegin; 649 ierr = MatGetFactor_aij_superlu_dist(A,ftype,F);CHKERRQ(ierr); 650 PetscFunctionReturn(0); 651 } 652 653 #undef __FUNCT__ 654 #define __FUNCT__ "MatFactorInfo_SuperLU_DIST" 655 PetscErrorCode MatFactorInfo_SuperLU_DIST(Mat A,PetscViewer viewer) 656 { 657 Mat_SuperLU_DIST *lu=(Mat_SuperLU_DIST*)A->spptr; 658 superlu_options_t options; 659 PetscErrorCode ierr; 660 661 PetscFunctionBegin; 662 /* check if matrix is superlu_dist type */ 663 if (A->ops->solve != MatSolve_SuperLU_DIST) PetscFunctionReturn(0); 664 665 options = lu->options; 666 ierr = PetscViewerASCIIPrintf(viewer,"SuperLU_DIST run parameters:\n");CHKERRQ(ierr); 667 ierr = PetscViewerASCIIPrintf(viewer," Process grid nprow %D x npcol %D \n",lu->nprow,lu->npcol);CHKERRQ(ierr); 668 ierr = PetscViewerASCIIPrintf(viewer," Equilibrate matrix %s \n",PetscBools[options.Equil != NO]);CHKERRQ(ierr); 669 ierr = PetscViewerASCIIPrintf(viewer," Matrix input mode %d \n",lu->MatInputMode);CHKERRQ(ierr); 670 ierr = PetscViewerASCIIPrintf(viewer," Replace tiny pivots %s \n",PetscBools[options.ReplaceTinyPivot != NO]);CHKERRQ(ierr); 671 ierr = PetscViewerASCIIPrintf(viewer," Use iterative refinement %s \n",PetscBools[options.IterRefine == SLU_DOUBLE]);CHKERRQ(ierr); 672 ierr = PetscViewerASCIIPrintf(viewer," Processors in row %d col partition %d \n",lu->nprow,lu->npcol);CHKERRQ(ierr); 673 ierr = PetscViewerASCIIPrintf(viewer," Row permutation %s \n",(options.RowPerm == NOROWPERM) ? "NATURAL" : "LargeDiag");CHKERRQ(ierr); 674 675 switch (options.ColPerm) { 676 case NATURAL: 677 ierr = PetscViewerASCIIPrintf(viewer," Column permutation NATURAL\n");CHKERRQ(ierr); 678 break; 679 case MMD_AT_PLUS_A: 680 ierr = PetscViewerASCIIPrintf(viewer," Column permutation MMD_AT_PLUS_A\n");CHKERRQ(ierr); 681 break; 682 case MMD_ATA: 683 ierr = PetscViewerASCIIPrintf(viewer," Column permutation MMD_ATA\n");CHKERRQ(ierr); 684 break; 685 case METIS_AT_PLUS_A: 686 ierr = PetscViewerASCIIPrintf(viewer," Column permutation METIS_AT_PLUS_A\n");CHKERRQ(ierr); 687 break; 688 case PARMETIS: 689 ierr = PetscViewerASCIIPrintf(viewer," Column permutation PARMETIS\n");CHKERRQ(ierr); 690 break; 691 default: 692 SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown column permutation"); 693 } 694 695 ierr = PetscViewerASCIIPrintf(viewer," Parallel symbolic factorization %s \n",PetscBools[options.ParSymbFact != NO]);CHKERRQ(ierr); 696 697 if (lu->FactPattern == SamePattern) { 698 ierr = PetscViewerASCIIPrintf(viewer," Repeated factorization SamePattern\n");CHKERRQ(ierr); 699 } else { 700 ierr = PetscViewerASCIIPrintf(viewer," Repeated factorization SamePattern_SameRowPerm\n");CHKERRQ(ierr); 701 } 702 PetscFunctionReturn(0); 703 } 704 705 #undef __FUNCT__ 706 #define __FUNCT__ "MatView_SuperLU_DIST" 707 PetscErrorCode MatView_SuperLU_DIST(Mat A,PetscViewer viewer) 708 { 709 PetscErrorCode ierr; 710 PetscBool iascii; 711 PetscViewerFormat format; 712 713 PetscFunctionBegin; 714 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 715 if (iascii) { 716 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 717 if (format == PETSC_VIEWER_ASCII_INFO) { 718 ierr = MatFactorInfo_SuperLU_DIST(A,viewer);CHKERRQ(ierr); 719 } 720 } 721 PetscFunctionReturn(0); 722 } 723 724 725 /*MC 726 MATSOLVERSUPERLU_DIST - Parallel direct solver package for LU factorization 727 728 Works with AIJ matrices 729 730 Options Database Keys: 731 + -mat_superlu_dist_r <n> - number of rows in processor partition 732 . -mat_superlu_dist_c <n> - number of columns in processor partition 733 . -mat_superlu_dist_matinput <0,1> - matrix input mode; 0=global, 1=distributed 734 . -mat_superlu_dist_equil - equilibrate the matrix 735 . -mat_superlu_dist_rowperm <LargeDiag,NATURAL> - row permutation 736 . -mat_superlu_dist_colperm <MMD_AT_PLUS_A,MMD_ATA,NATURAL> - column permutation 737 . -mat_superlu_dist_replacetinypivot - replace tiny pivots 738 . -mat_superlu_dist_fact <SamePattern> - (choose one of) SamePattern SamePattern_SameRowPerm 739 . -mat_superlu_dist_iterrefine - use iterative refinement 740 - -mat_superlu_dist_statprint - print factorization information 741 742 Level: beginner 743 744 .seealso: PCLU 745 746 .seealso: PCFactorSetMatSolverPackage(), MatSolverPackage 747 748 M*/ 749 750