1 2 /* 3 Provides an interface to the MUMPS sparse solver 4 */ 5 6 #include <../src/mat/impls/aij/mpi/mpiaij.h> /*I "petscmat.h" I*/ 7 #include <../src/mat/impls/sbaij/mpi/mpisbaij.h> 8 #include <petscblaslapack.h> 9 10 EXTERN_C_BEGIN 11 #if defined(PETSC_USE_COMPLEX) 12 #if defined(PETSC_USE_REAL_SINGLE) 13 #include <cmumps_c.h> 14 #else 15 #include <zmumps_c.h> 16 #endif 17 #else 18 #if defined(PETSC_USE_REAL_SINGLE) 19 #include <smumps_c.h> 20 #else 21 #include <dmumps_c.h> 22 #endif 23 #endif 24 EXTERN_C_END 25 #define JOB_INIT -1 26 #define JOB_FACTSYMBOLIC 1 27 #define JOB_FACTNUMERIC 2 28 #define JOB_SOLVE 3 29 #define JOB_END -2 30 31 /* calls to MUMPS */ 32 #if defined(PETSC_USE_COMPLEX) 33 #if defined(PETSC_USE_REAL_SINGLE) 34 #define PetscMUMPS_c cmumps_c 35 #else 36 #define PetscMUMPS_c zmumps_c 37 #endif 38 #else 39 #if defined(PETSC_USE_REAL_SINGLE) 40 #define PetscMUMPS_c smumps_c 41 #else 42 #define PetscMUMPS_c dmumps_c 43 #endif 44 #endif 45 46 /* declare MumpsScalar */ 47 #if defined(PETSC_USE_COMPLEX) 48 #if defined(PETSC_USE_REAL_SINGLE) 49 #define MumpsScalar mumps_complex 50 #else 51 #define MumpsScalar mumps_double_complex 52 #endif 53 #else 54 #define MumpsScalar PetscScalar 55 #endif 56 57 /* macros s.t. indices match MUMPS documentation */ 58 #define ICNTL(I) icntl[(I)-1] 59 #define CNTL(I) cntl[(I)-1] 60 #define INFOG(I) infog[(I)-1] 61 #define INFO(I) info[(I)-1] 62 #define RINFOG(I) rinfog[(I)-1] 63 #define RINFO(I) rinfo[(I)-1] 64 65 typedef struct { 66 #if defined(PETSC_USE_COMPLEX) 67 #if defined(PETSC_USE_REAL_SINGLE) 68 CMUMPS_STRUC_C id; 69 #else 70 ZMUMPS_STRUC_C id; 71 #endif 72 #else 73 #if defined(PETSC_USE_REAL_SINGLE) 74 SMUMPS_STRUC_C id; 75 #else 76 DMUMPS_STRUC_C id; 77 #endif 78 #endif 79 80 MatStructure matstruc; 81 PetscMPIInt myid,size; 82 PetscInt *irn,*jcn,nz,sym; 83 PetscScalar *val; 84 MPI_Comm comm_mumps; 85 PetscBool isAIJ; 86 PetscInt ICNTL9_pre; /* check if ICNTL(9) is changed from previous MatSolve */ 87 VecScatter scat_rhs, scat_sol; /* used by MatSolve() */ 88 Vec b_seq,x_seq; 89 PetscInt ninfo,*info; /* display INFO */ 90 PetscInt sizeredrhs; 91 PetscInt *schur_pivots; 92 PetscInt schur_B_lwork; 93 PetscScalar *schur_work; 94 PetscScalar *schur_sol; 95 PetscInt schur_sizesol; 96 PetscBool schur_factored; 97 PetscBool schur_inverted; 98 PetscInt schur_sym; 99 100 PetscErrorCode (*ConvertToTriples)(Mat, int, MatReuse, int*, int**, int**, PetscScalar**); 101 } Mat_MUMPS; 102 103 extern PetscErrorCode MatDuplicate_MUMPS(Mat,MatDuplicateOption,Mat*); 104 105 static PetscErrorCode MatMumpsResetSchur_Private(Mat_MUMPS* mumps) 106 { 107 PetscErrorCode ierr; 108 109 PetscFunctionBegin; 110 ierr = PetscFree2(mumps->id.listvar_schur,mumps->id.schur);CHKERRQ(ierr); 111 ierr = PetscFree(mumps->id.redrhs);CHKERRQ(ierr); 112 ierr = PetscFree(mumps->schur_sol);CHKERRQ(ierr); 113 ierr = PetscFree(mumps->schur_pivots);CHKERRQ(ierr); 114 ierr = PetscFree(mumps->schur_work);CHKERRQ(ierr); 115 mumps->id.size_schur = 0; 116 mumps->id.ICNTL(19) = 0; 117 PetscFunctionReturn(0); 118 } 119 120 static PetscErrorCode MatMumpsFactorSchur_Private(Mat_MUMPS* mumps) 121 { 122 PetscBLASInt B_N,B_ierr,B_slda; 123 PetscErrorCode ierr; 124 125 PetscFunctionBegin; 126 if (mumps->schur_factored) { 127 PetscFunctionReturn(0); 128 } 129 ierr = PetscBLASIntCast(mumps->id.size_schur,&B_N);CHKERRQ(ierr); 130 ierr = PetscBLASIntCast(mumps->id.schur_lld,&B_slda);CHKERRQ(ierr); 131 if (!mumps->sym) { /* MUMPS always return a full Schur matrix */ 132 if (!mumps->schur_pivots) { 133 ierr = PetscMalloc1(B_N,&mumps->schur_pivots);CHKERRQ(ierr); 134 } 135 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 136 PetscStackCallBLAS("LAPACKgetrf",LAPACKgetrf_(&B_N,&B_N,(PetscScalar*)mumps->id.schur,&B_slda,mumps->schur_pivots,&B_ierr)); 137 ierr = PetscFPTrapPop();CHKERRQ(ierr); 138 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in GETRF Lapack routine %d",(int)B_ierr); 139 } else { /* either full or lower-triangular (not packed) */ 140 char ord[2]; 141 if (mumps->id.ICNTL(19) == 2 || mumps->id.ICNTL(19) == 3) { /* lower triangular stored by columns or full matrix */ 142 sprintf(ord,"L"); 143 } else { /* ICNTL(19) == 1 lower triangular stored by rows */ 144 sprintf(ord,"U"); 145 } 146 if (mumps->schur_sym == 2) { 147 if (!mumps->schur_pivots) { 148 PetscScalar lwork; 149 150 ierr = PetscMalloc1(B_N,&mumps->schur_pivots);CHKERRQ(ierr); 151 mumps->schur_B_lwork=-1; 152 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 153 PetscStackCallBLAS("LAPACKsytrf",LAPACKsytrf_(ord,&B_N,(PetscScalar*)mumps->id.schur,&B_slda,mumps->schur_pivots,&lwork,&mumps->schur_B_lwork,&B_ierr)); 154 ierr = PetscFPTrapPop();CHKERRQ(ierr); 155 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in query to SYTRF Lapack routine %d",(int)B_ierr); 156 ierr = PetscBLASIntCast((PetscInt)PetscRealPart(lwork),&mumps->schur_B_lwork);CHKERRQ(ierr); 157 ierr = PetscMalloc1(mumps->schur_B_lwork,&mumps->schur_work);CHKERRQ(ierr); 158 } 159 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 160 PetscStackCallBLAS("LAPACKsytrf",LAPACKsytrf_(ord,&B_N,(PetscScalar*)mumps->id.schur,&B_slda,mumps->schur_pivots,mumps->schur_work,&mumps->schur_B_lwork,&B_ierr)); 161 ierr = PetscFPTrapPop();CHKERRQ(ierr); 162 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in SYTRF Lapack routine %d",(int)B_ierr); 163 } else { 164 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 165 PetscStackCallBLAS("LAPACKpotrf",LAPACKpotrf_(ord,&B_N,(PetscScalar*)mumps->id.schur,&B_slda,&B_ierr)); 166 ierr = PetscFPTrapPop();CHKERRQ(ierr); 167 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in POTRF Lapack routine %d",(int)B_ierr); 168 } 169 } 170 mumps->schur_factored = PETSC_TRUE; 171 PetscFunctionReturn(0); 172 } 173 174 static PetscErrorCode MatMumpsInvertSchur_Private(Mat_MUMPS* mumps) 175 { 176 PetscBLASInt B_N,B_ierr,B_slda; 177 PetscErrorCode ierr; 178 179 PetscFunctionBegin; 180 ierr = MatMumpsFactorSchur_Private(mumps);CHKERRQ(ierr); 181 ierr = PetscBLASIntCast(mumps->id.size_schur,&B_N);CHKERRQ(ierr); 182 ierr = PetscBLASIntCast(mumps->id.schur_lld,&B_slda);CHKERRQ(ierr); 183 if (!mumps->sym) { /* MUMPS always return a full Schur matrix */ 184 if (!mumps->schur_work) { 185 PetscScalar lwork; 186 187 mumps->schur_B_lwork = -1; 188 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 189 PetscStackCallBLAS("LAPACKgetri",LAPACKgetri_(&B_N,(PetscScalar*)mumps->id.schur,&B_N,mumps->schur_pivots,&lwork,&mumps->schur_B_lwork,&B_ierr)); 190 ierr = PetscFPTrapPop();CHKERRQ(ierr); 191 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in query to GETRI Lapack routine %d",(int)B_ierr); 192 ierr = PetscBLASIntCast((PetscInt)PetscRealPart(lwork),&mumps->schur_B_lwork);CHKERRQ(ierr); 193 ierr = PetscMalloc1(mumps->schur_B_lwork,&mumps->schur_work);CHKERRQ(ierr); 194 } 195 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 196 PetscStackCallBLAS("LAPACKgetri",LAPACKgetri_(&B_N,(PetscScalar*)mumps->id.schur,&B_N,mumps->schur_pivots,mumps->schur_work,&mumps->schur_B_lwork,&B_ierr)); 197 ierr = PetscFPTrapPop();CHKERRQ(ierr); 198 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in GETRI Lapack routine %d",(int)B_ierr); 199 } else { /* either full or lower-triangular (not packed) */ 200 char ord[2]; 201 if (mumps->id.ICNTL(19) == 2 || mumps->id.ICNTL(19) == 3) { /* lower triangular stored by columns or full matrix */ 202 sprintf(ord,"L"); 203 } else { /* ICNTL(19) == 1 lower triangular stored by rows */ 204 sprintf(ord,"U"); 205 } 206 if (mumps->schur_sym == 2) { 207 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 208 PetscStackCallBLAS("LAPACKsytri",LAPACKsytri_(ord,&B_N,(PetscScalar*)mumps->id.schur,&B_N,mumps->schur_pivots,mumps->schur_work,&B_ierr)); 209 ierr = PetscFPTrapPop();CHKERRQ(ierr); 210 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in SYTRI Lapack routine %d",(int)B_ierr); 211 } else { 212 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 213 PetscStackCallBLAS("LAPACKpotri",LAPACKpotri_(ord,&B_N,(PetscScalar*)mumps->id.schur,&B_N,&B_ierr)); 214 ierr = PetscFPTrapPop();CHKERRQ(ierr); 215 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in POTRI Lapack routine %d",(int)B_ierr); 216 } 217 } 218 mumps->schur_inverted = PETSC_TRUE; 219 PetscFunctionReturn(0); 220 } 221 222 static PetscErrorCode MatMumpsSolveSchur_Private(Mat_MUMPS* mumps, PetscBool sol_in_redrhs) 223 { 224 PetscBLASInt B_N,B_Nrhs,B_ierr,B_slda,B_rlda; 225 PetscScalar one=1.,zero=0.; 226 PetscErrorCode ierr; 227 228 PetscFunctionBegin; 229 ierr = MatMumpsFactorSchur_Private(mumps);CHKERRQ(ierr); 230 ierr = PetscBLASIntCast(mumps->id.size_schur,&B_N);CHKERRQ(ierr); 231 ierr = PetscBLASIntCast(mumps->id.schur_lld,&B_slda);CHKERRQ(ierr); 232 ierr = PetscBLASIntCast(mumps->id.nrhs,&B_Nrhs);CHKERRQ(ierr); 233 ierr = PetscBLASIntCast(mumps->id.lredrhs,&B_rlda);CHKERRQ(ierr); 234 if (mumps->schur_inverted) { 235 PetscInt sizesol = B_Nrhs*B_N; 236 if (!mumps->schur_sol || sizesol > mumps->schur_sizesol) { 237 ierr = PetscFree(mumps->schur_sol);CHKERRQ(ierr); 238 ierr = PetscMalloc1(sizesol,&mumps->schur_sol);CHKERRQ(ierr); 239 mumps->schur_sizesol = sizesol; 240 } 241 if (!mumps->sym) { 242 char type[2]; 243 if (mumps->id.ICNTL(19) == 1) { /* stored by rows */ 244 if (!mumps->id.ICNTL(9)) { /* transpose solve */ 245 sprintf(type,"N"); 246 } else { 247 sprintf(type,"T"); 248 } 249 } else { /* stored by columns */ 250 if (!mumps->id.ICNTL(9)) { /* transpose solve */ 251 sprintf(type,"T"); 252 } else { 253 sprintf(type,"N"); 254 } 255 } 256 PetscStackCallBLAS("BLASgemm",BLASgemm_(type,"N",&B_N,&B_Nrhs,&B_N,&one,(PetscScalar*)mumps->id.schur,&B_slda,(PetscScalar*)mumps->id.redrhs,&B_rlda,&zero,mumps->schur_sol,&B_rlda)); 257 } else { 258 char ord[2]; 259 if (mumps->id.ICNTL(19) == 2 || mumps->id.ICNTL(19) == 3) { /* lower triangular stored by columns or full matrix */ 260 sprintf(ord,"L"); 261 } else { /* ICNTL(19) == 1 lower triangular stored by rows */ 262 sprintf(ord,"U"); 263 } 264 PetscStackCallBLAS("BLASsymm",BLASsymm_("L",ord,&B_N,&B_Nrhs,&one,(PetscScalar*)mumps->id.schur,&B_slda,(PetscScalar*)mumps->id.redrhs,&B_rlda,&zero,mumps->schur_sol,&B_rlda)); 265 } 266 if (sol_in_redrhs) { 267 ierr = PetscMemcpy(mumps->id.redrhs,mumps->schur_sol,sizesol*sizeof(PetscScalar));CHKERRQ(ierr); 268 } 269 } else { /* Schur complement has not yet been inverted */ 270 MumpsScalar *orhs=NULL; 271 272 if (!sol_in_redrhs) { 273 PetscInt sizesol = B_Nrhs*B_N; 274 if (!mumps->schur_sol || sizesol > mumps->schur_sizesol) { 275 ierr = PetscFree(mumps->schur_sol);CHKERRQ(ierr); 276 ierr = PetscMalloc1(sizesol,&mumps->schur_sol);CHKERRQ(ierr); 277 mumps->schur_sizesol = sizesol; 278 } 279 orhs = mumps->id.redrhs; 280 ierr = PetscMemcpy(mumps->schur_sol,mumps->id.redrhs,sizesol*sizeof(PetscScalar));CHKERRQ(ierr); 281 mumps->id.redrhs = (MumpsScalar*)mumps->schur_sol; 282 } 283 if (!mumps->sym) { /* MUMPS always return a full Schur matrix */ 284 char type[2]; 285 if (mumps->id.ICNTL(19) == 1) { /* stored by rows */ 286 if (!mumps->id.ICNTL(9)) { /* transpose solve */ 287 sprintf(type,"N"); 288 } else { 289 sprintf(type,"T"); 290 } 291 } else { /* stored by columns */ 292 if (!mumps->id.ICNTL(9)) { /* transpose solve */ 293 sprintf(type,"T"); 294 } else { 295 sprintf(type,"N"); 296 } 297 } 298 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 299 PetscStackCallBLAS("LAPACKgetrs",LAPACKgetrs_(type,&B_N,&B_Nrhs,(PetscScalar*)mumps->id.schur,&B_slda,mumps->schur_pivots,(PetscScalar*)mumps->id.redrhs,&B_rlda,&B_ierr)); 300 ierr = PetscFPTrapPop();CHKERRQ(ierr); 301 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in GETRS Lapack routine %d",(int)B_ierr); 302 } else { /* either full or lower-triangular (not packed) */ 303 char ord[2]; 304 if (mumps->id.ICNTL(19) == 2 || mumps->id.ICNTL(19) == 3) { /* lower triangular stored by columns or full matrix */ 305 sprintf(ord,"L"); 306 } else { /* ICNTL(19) == 1 lower triangular stored by rows */ 307 sprintf(ord,"U"); 308 } 309 if (mumps->schur_sym == 2) { 310 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 311 PetscStackCallBLAS("LAPACKsytrs",LAPACKsytrs_(ord,&B_N,&B_Nrhs,(PetscScalar*)mumps->id.schur,&B_slda,mumps->schur_pivots,(PetscScalar*)mumps->id.redrhs,&B_rlda,&B_ierr)); 312 ierr = PetscFPTrapPop();CHKERRQ(ierr); 313 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in SYTRS Lapack routine %d",(int)B_ierr); 314 } else { 315 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 316 PetscStackCallBLAS("LAPACKpotrs",LAPACKpotrs_(ord,&B_N,&B_Nrhs,(PetscScalar*)mumps->id.schur,&B_slda,(PetscScalar*)mumps->id.redrhs,&B_rlda,&B_ierr)); 317 ierr = PetscFPTrapPop();CHKERRQ(ierr); 318 if (B_ierr) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error in POTRS Lapack routine %d",(int)B_ierr); 319 } 320 } 321 if (!sol_in_redrhs) { 322 mumps->id.redrhs = orhs; 323 } 324 } 325 PetscFunctionReturn(0); 326 } 327 328 static PetscErrorCode MatMumpsHandleSchur_Private(Mat_MUMPS* mumps, PetscBool expansion) 329 { 330 PetscErrorCode ierr; 331 332 PetscFunctionBegin; 333 if (!mumps->id.ICNTL(19)) { /* do nothing when Schur complement has not been computed */ 334 PetscFunctionReturn(0); 335 } 336 if (!expansion) { /* prepare for the condensation step */ 337 PetscInt sizeredrhs = mumps->id.nrhs*mumps->id.size_schur; 338 /* allocate MUMPS internal array to store reduced right-hand sides */ 339 if (!mumps->id.redrhs || sizeredrhs > mumps->sizeredrhs) { 340 ierr = PetscFree(mumps->id.redrhs);CHKERRQ(ierr); 341 mumps->id.lredrhs = mumps->id.size_schur; 342 ierr = PetscMalloc1(mumps->id.nrhs*mumps->id.lredrhs,&mumps->id.redrhs);CHKERRQ(ierr); 343 mumps->sizeredrhs = mumps->id.nrhs*mumps->id.lredrhs; 344 } 345 mumps->id.ICNTL(26) = 1; /* condensation phase */ 346 } else { /* prepare for the expansion step */ 347 /* solve Schur complement (this has to be done by the MUMPS user, so basically us) */ 348 ierr = MatMumpsSolveSchur_Private(mumps,PETSC_TRUE);CHKERRQ(ierr); 349 mumps->id.ICNTL(26) = 2; /* expansion phase */ 350 PetscMUMPS_c(&mumps->id); 351 if (mumps->id.INFOG(1) < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in solve phase: INFOG(1)=%d\n",mumps->id.INFOG(1)); 352 /* restore defaults */ 353 mumps->id.ICNTL(26) = -1; 354 /* free MUMPS internal array for redrhs if we have solved for multiple rhs in order to save memory space */ 355 if (mumps->id.nrhs > 1) { 356 ierr = PetscFree(mumps->id.redrhs);CHKERRQ(ierr); 357 mumps->id.lredrhs = 0; 358 mumps->sizeredrhs = 0; 359 } 360 } 361 PetscFunctionReturn(0); 362 } 363 364 /* 365 MatConvertToTriples_A_B - convert Petsc matrix to triples: row[nz], col[nz], val[nz] 366 367 input: 368 A - matrix in aij,baij or sbaij (bs=1) format 369 shift - 0: C style output triple; 1: Fortran style output triple. 370 reuse - MAT_INITIAL_MATRIX: spaces are allocated and values are set for the triple 371 MAT_REUSE_MATRIX: only the values in v array are updated 372 output: 373 nnz - dim of r, c, and v (number of local nonzero entries of A) 374 r, c, v - row and col index, matrix values (matrix triples) 375 376 The returned values r, c, and sometimes v are obtained in a single PetscMalloc(). Then in MatDestroy_MUMPS() it is 377 freed with PetscFree((mumps->irn); This is not ideal code, the fact that v is ONLY sometimes part of mumps->irn means 378 that the PetscMalloc() cannot easily be replaced with a PetscMalloc3(). 379 380 */ 381 382 PetscErrorCode MatConvertToTriples_seqaij_seqaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 383 { 384 const PetscInt *ai,*aj,*ajj,M=A->rmap->n; 385 PetscInt nz,rnz,i,j; 386 PetscErrorCode ierr; 387 PetscInt *row,*col; 388 Mat_SeqAIJ *aa=(Mat_SeqAIJ*)A->data; 389 390 PetscFunctionBegin; 391 *v=aa->a; 392 if (reuse == MAT_INITIAL_MATRIX) { 393 nz = aa->nz; 394 ai = aa->i; 395 aj = aa->j; 396 *nnz = nz; 397 ierr = PetscMalloc1(2*nz, &row);CHKERRQ(ierr); 398 col = row + nz; 399 400 nz = 0; 401 for (i=0; i<M; i++) { 402 rnz = ai[i+1] - ai[i]; 403 ajj = aj + ai[i]; 404 for (j=0; j<rnz; j++) { 405 row[nz] = i+shift; col[nz++] = ajj[j] + shift; 406 } 407 } 408 *r = row; *c = col; 409 } 410 PetscFunctionReturn(0); 411 } 412 413 PetscErrorCode MatConvertToTriples_seqbaij_seqaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 414 { 415 Mat_SeqBAIJ *aa=(Mat_SeqBAIJ*)A->data; 416 const PetscInt *ai,*aj,*ajj,bs2 = aa->bs2; 417 PetscInt bs,M,nz,idx=0,rnz,i,j,k,m; 418 PetscErrorCode ierr; 419 PetscInt *row,*col; 420 421 PetscFunctionBegin; 422 ierr = MatGetBlockSize(A,&bs);CHKERRQ(ierr); 423 M = A->rmap->N/bs; 424 *v = aa->a; 425 if (reuse == MAT_INITIAL_MATRIX) { 426 ai = aa->i; aj = aa->j; 427 nz = bs2*aa->nz; 428 *nnz = nz; 429 ierr = PetscMalloc1(2*nz, &row);CHKERRQ(ierr); 430 col = row + nz; 431 432 for (i=0; i<M; i++) { 433 ajj = aj + ai[i]; 434 rnz = ai[i+1] - ai[i]; 435 for (k=0; k<rnz; k++) { 436 for (j=0; j<bs; j++) { 437 for (m=0; m<bs; m++) { 438 row[idx] = i*bs + m + shift; 439 col[idx++] = bs*(ajj[k]) + j + shift; 440 } 441 } 442 } 443 } 444 *r = row; *c = col; 445 } 446 PetscFunctionReturn(0); 447 } 448 449 PetscErrorCode MatConvertToTriples_seqsbaij_seqsbaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 450 { 451 const PetscInt *ai, *aj,*ajj,M=A->rmap->n; 452 PetscInt nz,rnz,i,j; 453 PetscErrorCode ierr; 454 PetscInt *row,*col; 455 Mat_SeqSBAIJ *aa=(Mat_SeqSBAIJ*)A->data; 456 457 PetscFunctionBegin; 458 *v = aa->a; 459 if (reuse == MAT_INITIAL_MATRIX) { 460 nz = aa->nz; 461 ai = aa->i; 462 aj = aa->j; 463 *v = aa->a; 464 *nnz = nz; 465 ierr = PetscMalloc1(2*nz, &row);CHKERRQ(ierr); 466 col = row + nz; 467 468 nz = 0; 469 for (i=0; i<M; i++) { 470 rnz = ai[i+1] - ai[i]; 471 ajj = aj + ai[i]; 472 for (j=0; j<rnz; j++) { 473 row[nz] = i+shift; col[nz++] = ajj[j] + shift; 474 } 475 } 476 *r = row; *c = col; 477 } 478 PetscFunctionReturn(0); 479 } 480 481 PetscErrorCode MatConvertToTriples_seqaij_seqsbaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 482 { 483 const PetscInt *ai,*aj,*ajj,*adiag,M=A->rmap->n; 484 PetscInt nz,rnz,i,j; 485 const PetscScalar *av,*v1; 486 PetscScalar *val; 487 PetscErrorCode ierr; 488 PetscInt *row,*col; 489 Mat_SeqAIJ *aa=(Mat_SeqAIJ*)A->data; 490 PetscBool missing; 491 492 PetscFunctionBegin; 493 ai =aa->i; aj=aa->j;av=aa->a; 494 adiag=aa->diag; 495 ierr = MatMissingDiagonal_SeqAIJ(A,&missing,&i);CHKERRQ(ierr); 496 if (reuse == MAT_INITIAL_MATRIX) { 497 /* count nz in the uppper triangular part of A */ 498 nz = 0; 499 if (missing) { 500 for (i=0; i<M; i++) { 501 if (PetscUnlikely(adiag[i] >= ai[i+1])) { 502 for (j=ai[i];j<ai[i+1];j++) { 503 if (aj[j] < i) continue; 504 nz++; 505 } 506 } else { 507 nz += ai[i+1] - adiag[i]; 508 } 509 } 510 } else { 511 for (i=0; i<M; i++) nz += ai[i+1] - adiag[i]; 512 } 513 *nnz = nz; 514 515 ierr = PetscMalloc((2*nz*sizeof(PetscInt)+nz*sizeof(PetscScalar)), &row);CHKERRQ(ierr); 516 col = row + nz; 517 val = (PetscScalar*)(col + nz); 518 519 nz = 0; 520 if (missing) { 521 for (i=0; i<M; i++) { 522 if (PetscUnlikely(adiag[i] >= ai[i+1])) { 523 for (j=ai[i];j<ai[i+1];j++) { 524 if (aj[j] < i) continue; 525 row[nz] = i+shift; 526 col[nz] = aj[j]+shift; 527 val[nz] = av[j]; 528 nz++; 529 } 530 } else { 531 rnz = ai[i+1] - adiag[i]; 532 ajj = aj + adiag[i]; 533 v1 = av + adiag[i]; 534 for (j=0; j<rnz; j++) { 535 row[nz] = i+shift; col[nz] = ajj[j] + shift; val[nz++] = v1[j]; 536 } 537 } 538 } 539 } else { 540 for (i=0; i<M; i++) { 541 rnz = ai[i+1] - adiag[i]; 542 ajj = aj + adiag[i]; 543 v1 = av + adiag[i]; 544 for (j=0; j<rnz; j++) { 545 row[nz] = i+shift; col[nz] = ajj[j] + shift; val[nz++] = v1[j]; 546 } 547 } 548 } 549 *r = row; *c = col; *v = val; 550 } else { 551 nz = 0; val = *v; 552 if (missing) { 553 for (i=0; i <M; i++) { 554 if (PetscUnlikely(adiag[i] >= ai[i+1])) { 555 for (j=ai[i];j<ai[i+1];j++) { 556 if (aj[j] < i) continue; 557 val[nz++] = av[j]; 558 } 559 } else { 560 rnz = ai[i+1] - adiag[i]; 561 v1 = av + adiag[i]; 562 for (j=0; j<rnz; j++) { 563 val[nz++] = v1[j]; 564 } 565 } 566 } 567 } else { 568 for (i=0; i <M; i++) { 569 rnz = ai[i+1] - adiag[i]; 570 v1 = av + adiag[i]; 571 for (j=0; j<rnz; j++) { 572 val[nz++] = v1[j]; 573 } 574 } 575 } 576 } 577 PetscFunctionReturn(0); 578 } 579 580 PetscErrorCode MatConvertToTriples_mpisbaij_mpisbaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 581 { 582 const PetscInt *ai, *aj, *bi, *bj,*garray,m=A->rmap->n,*ajj,*bjj; 583 PetscErrorCode ierr; 584 PetscInt rstart,nz,i,j,jj,irow,countA,countB; 585 PetscInt *row,*col; 586 const PetscScalar *av, *bv,*v1,*v2; 587 PetscScalar *val; 588 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)A->data; 589 Mat_SeqSBAIJ *aa = (Mat_SeqSBAIJ*)(mat->A)->data; 590 Mat_SeqBAIJ *bb = (Mat_SeqBAIJ*)(mat->B)->data; 591 592 PetscFunctionBegin; 593 ai=aa->i; aj=aa->j; bi=bb->i; bj=bb->j; rstart= A->rmap->rstart; 594 av=aa->a; bv=bb->a; 595 596 garray = mat->garray; 597 598 if (reuse == MAT_INITIAL_MATRIX) { 599 nz = aa->nz + bb->nz; 600 *nnz = nz; 601 ierr = PetscMalloc((2*nz*sizeof(PetscInt)+nz*sizeof(PetscScalar)), &row);CHKERRQ(ierr); 602 col = row + nz; 603 val = (PetscScalar*)(col + nz); 604 605 *r = row; *c = col; *v = val; 606 } else { 607 row = *r; col = *c; val = *v; 608 } 609 610 jj = 0; irow = rstart; 611 for (i=0; i<m; i++) { 612 ajj = aj + ai[i]; /* ptr to the beginning of this row */ 613 countA = ai[i+1] - ai[i]; 614 countB = bi[i+1] - bi[i]; 615 bjj = bj + bi[i]; 616 v1 = av + ai[i]; 617 v2 = bv + bi[i]; 618 619 /* A-part */ 620 for (j=0; j<countA; j++) { 621 if (reuse == MAT_INITIAL_MATRIX) { 622 row[jj] = irow + shift; col[jj] = rstart + ajj[j] + shift; 623 } 624 val[jj++] = v1[j]; 625 } 626 627 /* B-part */ 628 for (j=0; j < countB; j++) { 629 if (reuse == MAT_INITIAL_MATRIX) { 630 row[jj] = irow + shift; col[jj] = garray[bjj[j]] + shift; 631 } 632 val[jj++] = v2[j]; 633 } 634 irow++; 635 } 636 PetscFunctionReturn(0); 637 } 638 639 PetscErrorCode MatConvertToTriples_mpiaij_mpiaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 640 { 641 const PetscInt *ai, *aj, *bi, *bj,*garray,m=A->rmap->n,*ajj,*bjj; 642 PetscErrorCode ierr; 643 PetscInt rstart,nz,i,j,jj,irow,countA,countB; 644 PetscInt *row,*col; 645 const PetscScalar *av, *bv,*v1,*v2; 646 PetscScalar *val; 647 Mat_MPIAIJ *mat = (Mat_MPIAIJ*)A->data; 648 Mat_SeqAIJ *aa = (Mat_SeqAIJ*)(mat->A)->data; 649 Mat_SeqAIJ *bb = (Mat_SeqAIJ*)(mat->B)->data; 650 651 PetscFunctionBegin; 652 ai=aa->i; aj=aa->j; bi=bb->i; bj=bb->j; rstart= A->rmap->rstart; 653 av=aa->a; bv=bb->a; 654 655 garray = mat->garray; 656 657 if (reuse == MAT_INITIAL_MATRIX) { 658 nz = aa->nz + bb->nz; 659 *nnz = nz; 660 ierr = PetscMalloc((2*nz*sizeof(PetscInt)+nz*sizeof(PetscScalar)), &row);CHKERRQ(ierr); 661 col = row + nz; 662 val = (PetscScalar*)(col + nz); 663 664 *r = row; *c = col; *v = val; 665 } else { 666 row = *r; col = *c; val = *v; 667 } 668 669 jj = 0; irow = rstart; 670 for (i=0; i<m; i++) { 671 ajj = aj + ai[i]; /* ptr to the beginning of this row */ 672 countA = ai[i+1] - ai[i]; 673 countB = bi[i+1] - bi[i]; 674 bjj = bj + bi[i]; 675 v1 = av + ai[i]; 676 v2 = bv + bi[i]; 677 678 /* A-part */ 679 for (j=0; j<countA; j++) { 680 if (reuse == MAT_INITIAL_MATRIX) { 681 row[jj] = irow + shift; col[jj] = rstart + ajj[j] + shift; 682 } 683 val[jj++] = v1[j]; 684 } 685 686 /* B-part */ 687 for (j=0; j < countB; j++) { 688 if (reuse == MAT_INITIAL_MATRIX) { 689 row[jj] = irow + shift; col[jj] = garray[bjj[j]] + shift; 690 } 691 val[jj++] = v2[j]; 692 } 693 irow++; 694 } 695 PetscFunctionReturn(0); 696 } 697 698 PetscErrorCode MatConvertToTriples_mpibaij_mpiaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 699 { 700 Mat_MPIBAIJ *mat = (Mat_MPIBAIJ*)A->data; 701 Mat_SeqBAIJ *aa = (Mat_SeqBAIJ*)(mat->A)->data; 702 Mat_SeqBAIJ *bb = (Mat_SeqBAIJ*)(mat->B)->data; 703 const PetscInt *ai = aa->i, *bi = bb->i, *aj = aa->j, *bj = bb->j,*ajj, *bjj; 704 const PetscInt *garray = mat->garray,mbs=mat->mbs,rstart=A->rmap->rstart; 705 const PetscInt bs2=mat->bs2; 706 PetscErrorCode ierr; 707 PetscInt bs,nz,i,j,k,n,jj,irow,countA,countB,idx; 708 PetscInt *row,*col; 709 const PetscScalar *av=aa->a, *bv=bb->a,*v1,*v2; 710 PetscScalar *val; 711 712 PetscFunctionBegin; 713 ierr = MatGetBlockSize(A,&bs);CHKERRQ(ierr); 714 if (reuse == MAT_INITIAL_MATRIX) { 715 nz = bs2*(aa->nz + bb->nz); 716 *nnz = nz; 717 ierr = PetscMalloc((2*nz*sizeof(PetscInt)+nz*sizeof(PetscScalar)), &row);CHKERRQ(ierr); 718 col = row + nz; 719 val = (PetscScalar*)(col + nz); 720 721 *r = row; *c = col; *v = val; 722 } else { 723 row = *r; col = *c; val = *v; 724 } 725 726 jj = 0; irow = rstart; 727 for (i=0; i<mbs; i++) { 728 countA = ai[i+1] - ai[i]; 729 countB = bi[i+1] - bi[i]; 730 ajj = aj + ai[i]; 731 bjj = bj + bi[i]; 732 v1 = av + bs2*ai[i]; 733 v2 = bv + bs2*bi[i]; 734 735 idx = 0; 736 /* A-part */ 737 for (k=0; k<countA; k++) { 738 for (j=0; j<bs; j++) { 739 for (n=0; n<bs; n++) { 740 if (reuse == MAT_INITIAL_MATRIX) { 741 row[jj] = irow + n + shift; 742 col[jj] = rstart + bs*ajj[k] + j + shift; 743 } 744 val[jj++] = v1[idx++]; 745 } 746 } 747 } 748 749 idx = 0; 750 /* B-part */ 751 for (k=0; k<countB; k++) { 752 for (j=0; j<bs; j++) { 753 for (n=0; n<bs; n++) { 754 if (reuse == MAT_INITIAL_MATRIX) { 755 row[jj] = irow + n + shift; 756 col[jj] = bs*garray[bjj[k]] + j + shift; 757 } 758 val[jj++] = v2[idx++]; 759 } 760 } 761 } 762 irow += bs; 763 } 764 PetscFunctionReturn(0); 765 } 766 767 PetscErrorCode MatConvertToTriples_mpiaij_mpisbaij(Mat A,int shift,MatReuse reuse,int *nnz,int **r, int **c, PetscScalar **v) 768 { 769 const PetscInt *ai, *aj,*adiag, *bi, *bj,*garray,m=A->rmap->n,*ajj,*bjj; 770 PetscErrorCode ierr; 771 PetscInt rstart,nz,nza,nzb,i,j,jj,irow,countA,countB; 772 PetscInt *row,*col; 773 const PetscScalar *av, *bv,*v1,*v2; 774 PetscScalar *val; 775 Mat_MPIAIJ *mat = (Mat_MPIAIJ*)A->data; 776 Mat_SeqAIJ *aa =(Mat_SeqAIJ*)(mat->A)->data; 777 Mat_SeqAIJ *bb =(Mat_SeqAIJ*)(mat->B)->data; 778 779 PetscFunctionBegin; 780 ai=aa->i; aj=aa->j; adiag=aa->diag; 781 bi=bb->i; bj=bb->j; garray = mat->garray; 782 av=aa->a; bv=bb->a; 783 784 rstart = A->rmap->rstart; 785 786 if (reuse == MAT_INITIAL_MATRIX) { 787 nza = 0; /* num of upper triangular entries in mat->A, including diagonals */ 788 nzb = 0; /* num of upper triangular entries in mat->B */ 789 for (i=0; i<m; i++) { 790 nza += (ai[i+1] - adiag[i]); 791 countB = bi[i+1] - bi[i]; 792 bjj = bj + bi[i]; 793 for (j=0; j<countB; j++) { 794 if (garray[bjj[j]] > rstart) nzb++; 795 } 796 } 797 798 nz = nza + nzb; /* total nz of upper triangular part of mat */ 799 *nnz = nz; 800 ierr = PetscMalloc((2*nz*sizeof(PetscInt)+nz*sizeof(PetscScalar)), &row);CHKERRQ(ierr); 801 col = row + nz; 802 val = (PetscScalar*)(col + nz); 803 804 *r = row; *c = col; *v = val; 805 } else { 806 row = *r; col = *c; val = *v; 807 } 808 809 jj = 0; irow = rstart; 810 for (i=0; i<m; i++) { 811 ajj = aj + adiag[i]; /* ptr to the beginning of the diagonal of this row */ 812 v1 = av + adiag[i]; 813 countA = ai[i+1] - adiag[i]; 814 countB = bi[i+1] - bi[i]; 815 bjj = bj + bi[i]; 816 v2 = bv + bi[i]; 817 818 /* A-part */ 819 for (j=0; j<countA; j++) { 820 if (reuse == MAT_INITIAL_MATRIX) { 821 row[jj] = irow + shift; col[jj] = rstart + ajj[j] + shift; 822 } 823 val[jj++] = v1[j]; 824 } 825 826 /* B-part */ 827 for (j=0; j < countB; j++) { 828 if (garray[bjj[j]] > rstart) { 829 if (reuse == MAT_INITIAL_MATRIX) { 830 row[jj] = irow + shift; col[jj] = garray[bjj[j]] + shift; 831 } 832 val[jj++] = v2[j]; 833 } 834 } 835 irow++; 836 } 837 PetscFunctionReturn(0); 838 } 839 840 PetscErrorCode MatDestroy_MUMPS(Mat A) 841 { 842 Mat_MUMPS *mumps=(Mat_MUMPS*)A->data; 843 PetscErrorCode ierr; 844 845 PetscFunctionBegin; 846 ierr = PetscFree2(mumps->id.sol_loc,mumps->id.isol_loc);CHKERRQ(ierr); 847 ierr = VecScatterDestroy(&mumps->scat_rhs);CHKERRQ(ierr); 848 ierr = VecScatterDestroy(&mumps->scat_sol);CHKERRQ(ierr); 849 ierr = VecDestroy(&mumps->b_seq);CHKERRQ(ierr); 850 ierr = VecDestroy(&mumps->x_seq);CHKERRQ(ierr); 851 ierr = PetscFree(mumps->id.perm_in);CHKERRQ(ierr); 852 ierr = PetscFree(mumps->irn);CHKERRQ(ierr); 853 ierr = PetscFree(mumps->info);CHKERRQ(ierr); 854 ierr = MatMumpsResetSchur_Private(mumps);CHKERRQ(ierr); 855 mumps->id.job = JOB_END; 856 PetscMUMPS_c(&mumps->id); 857 ierr = MPI_Comm_free(&mumps->comm_mumps);CHKERRQ(ierr); 858 ierr = PetscFree(A->data);CHKERRQ(ierr); 859 860 /* clear composed functions */ 861 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorGetSolverPackage_C",NULL);CHKERRQ(ierr); 862 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorSetSchurIS_C",NULL);CHKERRQ(ierr); 863 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorInvertSchurComplement_C",NULL);CHKERRQ(ierr); 864 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorCreateSchurComplement_C",NULL);CHKERRQ(ierr); 865 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorGetSchurComplement_C",NULL);CHKERRQ(ierr); 866 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorSolveSchurComplement_C",NULL);CHKERRQ(ierr); 867 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorSolveSchurComplementTranspose_C",NULL);CHKERRQ(ierr); 868 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorFactorizeSchurComplement_C",NULL);CHKERRQ(ierr); 869 ierr = PetscObjectComposeFunction((PetscObject)A,"MatFactorSetSchurComplementSolverType_C",NULL);CHKERRQ(ierr); 870 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsSetIcntl_C",NULL);CHKERRQ(ierr); 871 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetIcntl_C",NULL);CHKERRQ(ierr); 872 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsSetCntl_C",NULL);CHKERRQ(ierr); 873 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetCntl_C",NULL);CHKERRQ(ierr); 874 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetInfo_C",NULL);CHKERRQ(ierr); 875 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetInfog_C",NULL);CHKERRQ(ierr); 876 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetRinfo_C",NULL);CHKERRQ(ierr); 877 ierr = PetscObjectComposeFunction((PetscObject)A,"MatMumpsGetRinfog_C",NULL);CHKERRQ(ierr); 878 PetscFunctionReturn(0); 879 } 880 881 PetscErrorCode MatSolve_MUMPS(Mat A,Vec b,Vec x) 882 { 883 Mat_MUMPS *mumps=(Mat_MUMPS*)A->data; 884 PetscScalar *array; 885 Vec b_seq; 886 IS is_iden,is_petsc; 887 PetscErrorCode ierr; 888 PetscInt i; 889 PetscBool second_solve = PETSC_FALSE; 890 static PetscBool cite1 = PETSC_FALSE,cite2 = PETSC_FALSE; 891 892 PetscFunctionBegin; 893 ierr = PetscCitationsRegister("@article{MUMPS01,\n author = {P.~R. Amestoy and I.~S. Duff and J.-Y. L'Excellent and J. Koster},\n title = {A fully asynchronous multifrontal solver using distributed dynamic scheduling},\n journal = {SIAM Journal on Matrix Analysis and Applications},\n volume = {23},\n number = {1},\n pages = {15--41},\n year = {2001}\n}\n",&cite1);CHKERRQ(ierr); 894 ierr = PetscCitationsRegister("@article{MUMPS02,\n author = {P.~R. Amestoy and A. Guermouche and J.-Y. L'Excellent and S. Pralet},\n title = {Hybrid scheduling for the parallel solution of linear systems},\n journal = {Parallel Computing},\n volume = {32},\n number = {2},\n pages = {136--156},\n year = {2006}\n}\n",&cite2);CHKERRQ(ierr); 895 896 if (A->factorerrortype) { 897 ierr = PetscInfo2(A,"MatSolve is called with singular matrix factor, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 898 ierr = VecSetInf(x);CHKERRQ(ierr); 899 PetscFunctionReturn(0); 900 } 901 902 mumps->id.nrhs = 1; 903 b_seq = mumps->b_seq; 904 if (mumps->size > 1) { 905 /* MUMPS only supports centralized rhs. Scatter b into a seqential rhs vector */ 906 ierr = VecScatterBegin(mumps->scat_rhs,b,b_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 907 ierr = VecScatterEnd(mumps->scat_rhs,b,b_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 908 if (!mumps->myid) {ierr = VecGetArray(b_seq,&array);CHKERRQ(ierr);} 909 } else { /* size == 1 */ 910 ierr = VecCopy(b,x);CHKERRQ(ierr); 911 ierr = VecGetArray(x,&array);CHKERRQ(ierr); 912 } 913 if (!mumps->myid) { /* define rhs on the host */ 914 mumps->id.nrhs = 1; 915 mumps->id.rhs = (MumpsScalar*)array; 916 } 917 918 /* 919 handle condensation step of Schur complement (if any) 920 We set by default ICNTL(26) == -1 when Schur indices have been provided by the user. 921 According to MUMPS (5.0.0) manual, any value should be harmful during the factorization phase 922 Unless the user provides a valid value for ICNTL(26), MatSolve and MatMatSolve routines solve the full system. 923 This requires an extra call to PetscMUMPS_c and the computation of the factors for S 924 */ 925 if (mumps->id.ICNTL(26) < 0 || mumps->id.ICNTL(26) > 2) { 926 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Parallel Schur complements not yet supported from PETSc\n"); 927 second_solve = PETSC_TRUE; 928 ierr = MatMumpsHandleSchur_Private(mumps,PETSC_FALSE);CHKERRQ(ierr); 929 } 930 /* solve phase */ 931 /*-------------*/ 932 mumps->id.job = JOB_SOLVE; 933 PetscMUMPS_c(&mumps->id); 934 if (mumps->id.INFOG(1) < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in solve phase: INFOG(1)=%d\n",mumps->id.INFOG(1)); 935 936 /* handle expansion step of Schur complement (if any) */ 937 if (second_solve) { 938 ierr = MatMumpsHandleSchur_Private(mumps,PETSC_TRUE);CHKERRQ(ierr); 939 } 940 941 if (mumps->size > 1) { /* convert mumps distributed solution to petsc mpi x */ 942 if (mumps->scat_sol && mumps->ICNTL9_pre != mumps->id.ICNTL(9)) { 943 /* when id.ICNTL(9) changes, the contents of lsol_loc may change (not its size, lsol_loc), recreates scat_sol */ 944 ierr = VecScatterDestroy(&mumps->scat_sol);CHKERRQ(ierr); 945 } 946 if (!mumps->scat_sol) { /* create scatter scat_sol */ 947 ierr = ISCreateStride(PETSC_COMM_SELF,mumps->id.lsol_loc,0,1,&is_iden);CHKERRQ(ierr); /* from */ 948 for (i=0; i<mumps->id.lsol_loc; i++) { 949 mumps->id.isol_loc[i] -= 1; /* change Fortran style to C style */ 950 } 951 ierr = ISCreateGeneral(PETSC_COMM_SELF,mumps->id.lsol_loc,mumps->id.isol_loc,PETSC_COPY_VALUES,&is_petsc);CHKERRQ(ierr); /* to */ 952 ierr = VecScatterCreate(mumps->x_seq,is_iden,x,is_petsc,&mumps->scat_sol);CHKERRQ(ierr); 953 ierr = ISDestroy(&is_iden);CHKERRQ(ierr); 954 ierr = ISDestroy(&is_petsc);CHKERRQ(ierr); 955 956 mumps->ICNTL9_pre = mumps->id.ICNTL(9); /* save current value of id.ICNTL(9) */ 957 } 958 959 ierr = VecScatterBegin(mumps->scat_sol,mumps->x_seq,x,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 960 ierr = VecScatterEnd(mumps->scat_sol,mumps->x_seq,x,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 961 } 962 PetscFunctionReturn(0); 963 } 964 965 PetscErrorCode MatSolveTranspose_MUMPS(Mat A,Vec b,Vec x) 966 { 967 Mat_MUMPS *mumps=(Mat_MUMPS*)A->data; 968 PetscErrorCode ierr; 969 970 PetscFunctionBegin; 971 mumps->id.ICNTL(9) = 0; 972 ierr = MatSolve_MUMPS(A,b,x);CHKERRQ(ierr); 973 mumps->id.ICNTL(9) = 1; 974 PetscFunctionReturn(0); 975 } 976 977 PetscErrorCode MatMatSolve_MUMPS(Mat A,Mat B,Mat X) 978 { 979 PetscErrorCode ierr; 980 Mat Bt = NULL; 981 PetscBool flg, flgT; 982 Mat_MUMPS *mumps=(Mat_MUMPS*)A->data; 983 PetscInt i,nrhs,M; 984 PetscScalar *array,*bray; 985 986 PetscFunctionBegin; 987 ierr = PetscObjectTypeCompareAny((PetscObject)B,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 988 ierr = PetscObjectTypeCompare((PetscObject)B,MATTRANSPOSEMAT,&flgT);CHKERRQ(ierr); 989 if (flgT) { 990 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_WRONG,"Matrix B must be MATDENSE matrix"); 991 ierr = MatTransposeGetMat(B,&Bt);CHKERRQ(ierr); 992 } else { 993 if (!flg) SETERRQ(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_WRONG,"Matrix B must be MATDENSE matrix"); 994 } 995 ierr = PetscObjectTypeCompareAny((PetscObject)X,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 996 if (!flg) SETERRQ(PetscObjectComm((PetscObject)X),PETSC_ERR_ARG_WRONG,"Matrix X must be MATDENSE matrix"); 997 if (B->rmap->n != X->rmap->n) SETERRQ(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_WRONG,"Matrix B and X must have same row distribution"); 998 999 ierr = MatGetSize(B,&M,&nrhs);CHKERRQ(ierr); 1000 mumps->id.nrhs = nrhs; 1001 mumps->id.lrhs = M; 1002 1003 if (mumps->size == 1) { 1004 PetscScalar *aa; 1005 PetscInt spnr,*ia,*ja; 1006 PetscBool second_solve = PETSC_FALSE; 1007 1008 /* copy B to X */ 1009 ierr = MatDenseGetArray(X,&array);CHKERRQ(ierr); 1010 mumps->id.rhs = (MumpsScalar*)array; 1011 if (!Bt) { 1012 ierr = MatDenseGetArray(B,&bray);CHKERRQ(ierr); 1013 ierr = PetscMemcpy(array,bray,M*nrhs*sizeof(PetscScalar));CHKERRQ(ierr); 1014 ierr = MatDenseRestoreArray(B,&bray);CHKERRQ(ierr); 1015 } else { 1016 PetscBool done; 1017 1018 ierr = MatSeqAIJGetArray(Bt,&aa);CHKERRQ(ierr); 1019 ierr = MatGetRowIJ(Bt,1,PETSC_FALSE,PETSC_FALSE,&spnr,(const PetscInt**)&ia,(const PetscInt**)&ja,&done);CHKERRQ(ierr); 1020 if (!done) SETERRQ(PetscObjectComm((PetscObject)Bt),PETSC_ERR_ARG_WRONG,"Cannot get IJ structure"); 1021 mumps->id.irhs_ptr = ia; 1022 mumps->id.irhs_sparse = ja; 1023 mumps->id.nz_rhs = ia[spnr] - 1; 1024 mumps->id.rhs_sparse = (MumpsScalar*)aa; 1025 mumps->id.ICNTL(20) = 1; 1026 } 1027 /* handle condensation step of Schur complement (if any) */ 1028 if (mumps->id.ICNTL(26) < 0 || mumps->id.ICNTL(26) > 2) { 1029 second_solve = PETSC_TRUE; 1030 ierr = MatMumpsHandleSchur_Private(mumps,PETSC_FALSE);CHKERRQ(ierr); 1031 } 1032 /* solve phase */ 1033 /*-------------*/ 1034 mumps->id.job = JOB_SOLVE; 1035 PetscMUMPS_c(&mumps->id); 1036 if (mumps->id.INFOG(1) < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in solve phase: INFOG(1)=%d\n",mumps->id.INFOG(1)); 1037 1038 /* handle expansion step of Schur complement (if any) */ 1039 if (second_solve) { 1040 ierr = MatMumpsHandleSchur_Private(mumps,PETSC_TRUE);CHKERRQ(ierr); 1041 } 1042 if (Bt) { 1043 PetscBool done; 1044 1045 ierr = MatSeqAIJRestoreArray(Bt,&aa);CHKERRQ(ierr); 1046 ierr = MatRestoreRowIJ(Bt,1,PETSC_FALSE,PETSC_FALSE,&spnr,(const PetscInt**)&ia,(const PetscInt**)&ja,&done);CHKERRQ(ierr); 1047 if (!done) SETERRQ(PetscObjectComm((PetscObject)Bt),PETSC_ERR_ARG_WRONG,"Cannot restore IJ structure"); 1048 mumps->id.ICNTL(20) = 0; 1049 } 1050 ierr = MatDenseRestoreArray(X,&array);CHKERRQ(ierr); 1051 } else { /*--------- parallel case --------*/ 1052 PetscInt lsol_loc,nlsol_loc,*isol_loc,*idx,*iidx,*idxx,*isol_loc_save; 1053 MumpsScalar *sol_loc,*sol_loc_save; 1054 IS is_to,is_from; 1055 PetscInt k,proc,j,m; 1056 const PetscInt *rstart; 1057 Vec v_mpi,b_seq,x_seq; 1058 VecScatter scat_rhs,scat_sol; 1059 1060 if (mumps->size > 1 && mumps->id.ICNTL(19)) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Parallel Schur complements not yet supported from PETSc\n"); 1061 1062 /* create x_seq to hold local solution */ 1063 isol_loc_save = mumps->id.isol_loc; /* save it for MatSovle() */ 1064 sol_loc_save = mumps->id.sol_loc; 1065 1066 lsol_loc = mumps->id.INFO(23); 1067 nlsol_loc = nrhs*lsol_loc; /* length of sol_loc */ 1068 ierr = PetscMalloc2(nlsol_loc,&sol_loc,nlsol_loc,&isol_loc);CHKERRQ(ierr); 1069 mumps->id.sol_loc = (MumpsScalar*)sol_loc; 1070 mumps->id.isol_loc = isol_loc; 1071 1072 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,1,nlsol_loc,(PetscScalar*)sol_loc,&x_seq);CHKERRQ(ierr); 1073 1074 /* copy rhs matrix B into vector v_mpi */ 1075 ierr = MatGetLocalSize(B,&m,NULL);CHKERRQ(ierr); 1076 ierr = MatDenseGetArray(B,&bray);CHKERRQ(ierr); 1077 ierr = VecCreateMPIWithArray(PetscObjectComm((PetscObject)B),1,nrhs*m,nrhs*M,(const PetscScalar*)bray,&v_mpi);CHKERRQ(ierr); 1078 ierr = MatDenseRestoreArray(B,&bray);CHKERRQ(ierr); 1079 1080 /* scatter v_mpi to b_seq because MUMPS only supports centralized rhs */ 1081 /* idx: maps from k-th index of v_mpi to (i,j)-th global entry of B; 1082 iidx: inverse of idx, will be used by scattering xx_seq -> X */ 1083 ierr = PetscMalloc2(nrhs*M,&idx,nrhs*M,&iidx);CHKERRQ(ierr); 1084 ierr = MatGetOwnershipRanges(B,&rstart);CHKERRQ(ierr); 1085 k = 0; 1086 for (proc=0; proc<mumps->size; proc++){ 1087 for (j=0; j<nrhs; j++){ 1088 for (i=rstart[proc]; i<rstart[proc+1]; i++){ 1089 iidx[j*M + i] = k; 1090 idx[k++] = j*M + i; 1091 } 1092 } 1093 } 1094 1095 if (!mumps->myid) { 1096 ierr = VecCreateSeq(PETSC_COMM_SELF,nrhs*M,&b_seq);CHKERRQ(ierr); 1097 ierr = ISCreateGeneral(PETSC_COMM_SELF,nrhs*M,idx,PETSC_COPY_VALUES,&is_to);CHKERRQ(ierr); 1098 ierr = ISCreateStride(PETSC_COMM_SELF,nrhs*M,0,1,&is_from);CHKERRQ(ierr); 1099 } else { 1100 ierr = VecCreateSeq(PETSC_COMM_SELF,0,&b_seq);CHKERRQ(ierr); 1101 ierr = ISCreateStride(PETSC_COMM_SELF,0,0,1,&is_to);CHKERRQ(ierr); 1102 ierr = ISCreateStride(PETSC_COMM_SELF,0,0,1,&is_from);CHKERRQ(ierr); 1103 } 1104 ierr = VecScatterCreate(v_mpi,is_from,b_seq,is_to,&scat_rhs);CHKERRQ(ierr); 1105 ierr = VecScatterBegin(scat_rhs,v_mpi,b_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1106 ierr = ISDestroy(&is_to);CHKERRQ(ierr); 1107 ierr = ISDestroy(&is_from);CHKERRQ(ierr); 1108 ierr = VecScatterEnd(scat_rhs,v_mpi,b_seq,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1109 1110 if (!mumps->myid) { /* define rhs on the host */ 1111 ierr = VecGetArray(b_seq,&bray);CHKERRQ(ierr); 1112 mumps->id.rhs = (MumpsScalar*)bray; 1113 ierr = VecRestoreArray(b_seq,&bray);CHKERRQ(ierr); 1114 } 1115 1116 /* solve phase */ 1117 /*-------------*/ 1118 mumps->id.job = JOB_SOLVE; 1119 PetscMUMPS_c(&mumps->id); 1120 if (mumps->id.INFOG(1) < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in solve phase: INFOG(1)=%d\n",mumps->id.INFOG(1)); 1121 1122 /* scatter mumps distributed solution to petsc vector v_mpi, which shares local arrays with solution matrix X */ 1123 ierr = MatDenseGetArray(X,&array);CHKERRQ(ierr); 1124 ierr = VecPlaceArray(v_mpi,array);CHKERRQ(ierr); 1125 1126 /* create scatter scat_sol */ 1127 ierr = PetscMalloc1(nlsol_loc,&idxx);CHKERRQ(ierr); 1128 ierr = ISCreateStride(PETSC_COMM_SELF,nlsol_loc,0,1,&is_from);CHKERRQ(ierr); 1129 for (i=0; i<lsol_loc; i++) { 1130 isol_loc[i] -= 1; /* change Fortran style to C style */ 1131 idxx[i] = iidx[isol_loc[i]]; 1132 for (j=1; j<nrhs; j++){ 1133 idxx[j*lsol_loc+i] = iidx[isol_loc[i]+j*M]; 1134 } 1135 } 1136 ierr = ISCreateGeneral(PETSC_COMM_SELF,nlsol_loc,idxx,PETSC_COPY_VALUES,&is_to);CHKERRQ(ierr); 1137 ierr = VecScatterCreate(x_seq,is_from,v_mpi,is_to,&scat_sol);CHKERRQ(ierr); 1138 ierr = VecScatterBegin(scat_sol,x_seq,v_mpi,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1139 ierr = ISDestroy(&is_from);CHKERRQ(ierr); 1140 ierr = ISDestroy(&is_to);CHKERRQ(ierr); 1141 ierr = VecScatterEnd(scat_sol,x_seq,v_mpi,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1142 ierr = MatDenseRestoreArray(X,&array);CHKERRQ(ierr); 1143 1144 /* free spaces */ 1145 mumps->id.sol_loc = sol_loc_save; 1146 mumps->id.isol_loc = isol_loc_save; 1147 1148 ierr = PetscFree2(sol_loc,isol_loc);CHKERRQ(ierr); 1149 ierr = PetscFree2(idx,iidx);CHKERRQ(ierr); 1150 ierr = PetscFree(idxx);CHKERRQ(ierr); 1151 ierr = VecDestroy(&x_seq);CHKERRQ(ierr); 1152 ierr = VecDestroy(&v_mpi);CHKERRQ(ierr); 1153 ierr = VecDestroy(&b_seq);CHKERRQ(ierr); 1154 ierr = VecScatterDestroy(&scat_rhs);CHKERRQ(ierr); 1155 ierr = VecScatterDestroy(&scat_sol);CHKERRQ(ierr); 1156 } 1157 PetscFunctionReturn(0); 1158 } 1159 1160 #if !defined(PETSC_USE_COMPLEX) 1161 /* 1162 input: 1163 F: numeric factor 1164 output: 1165 nneg: total number of negative pivots 1166 nzero: total number of zero pivots 1167 npos: (global dimension of F) - nneg - nzero 1168 */ 1169 PetscErrorCode MatGetInertia_SBAIJMUMPS(Mat F,int *nneg,int *nzero,int *npos) 1170 { 1171 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1172 PetscErrorCode ierr; 1173 PetscMPIInt size; 1174 1175 PetscFunctionBegin; 1176 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)F),&size);CHKERRQ(ierr); 1177 /* MUMPS 4.3.1 calls ScaLAPACK when ICNTL(13)=0 (default), which does not offer the possibility to compute the inertia of a dense matrix. Set ICNTL(13)=1 to skip ScaLAPACK */ 1178 if (size > 1 && mumps->id.ICNTL(13) != 1) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"ICNTL(13)=%d. -mat_mumps_icntl_13 must be set as 1 for correct global matrix inertia\n",mumps->id.INFOG(13)); 1179 1180 if (nneg) *nneg = mumps->id.INFOG(12); 1181 if (nzero || npos) { 1182 if (mumps->id.ICNTL(24) != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"-mat_mumps_icntl_24 must be set as 1 for null pivot row detection"); 1183 if (nzero) *nzero = mumps->id.INFOG(28); 1184 if (npos) *npos = F->rmap->N - (mumps->id.INFOG(12) + mumps->id.INFOG(28)); 1185 } 1186 PetscFunctionReturn(0); 1187 } 1188 #endif 1189 1190 PetscErrorCode MatFactorNumeric_MUMPS(Mat F,Mat A,const MatFactorInfo *info) 1191 { 1192 Mat_MUMPS *mumps =(Mat_MUMPS*)(F)->data; 1193 PetscErrorCode ierr; 1194 PetscBool isMPIAIJ; 1195 1196 PetscFunctionBegin; 1197 if (mumps->id.INFOG(1) < 0) { 1198 if (mumps->id.INFOG(1) == -6) { 1199 ierr = PetscInfo2(A,"MatFactorNumeric is called with singular matrix structure, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1200 } 1201 ierr = PetscInfo2(A,"MatFactorNumeric is called after analysis phase fails, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1202 PetscFunctionReturn(0); 1203 } 1204 1205 ierr = (*mumps->ConvertToTriples)(A, 1, MAT_REUSE_MATRIX, &mumps->nz, &mumps->irn, &mumps->jcn, &mumps->val);CHKERRQ(ierr); 1206 1207 /* numerical factorization phase */ 1208 /*-------------------------------*/ 1209 mumps->id.job = JOB_FACTNUMERIC; 1210 if (!mumps->id.ICNTL(18)) { /* A is centralized */ 1211 if (!mumps->myid) { 1212 mumps->id.a = (MumpsScalar*)mumps->val; 1213 } 1214 } else { 1215 mumps->id.a_loc = (MumpsScalar*)mumps->val; 1216 } 1217 PetscMUMPS_c(&mumps->id); 1218 if (mumps->id.INFOG(1) < 0) { 1219 if (A->erroriffailure) { 1220 SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in numerical factorization phase: INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2)); 1221 } else { 1222 if (mumps->id.INFOG(1) == -10) { /* numerically singular matrix */ 1223 ierr = PetscInfo2(F,"matrix is numerically singular, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1224 F->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT; 1225 } else if (mumps->id.INFOG(1) == -13) { 1226 ierr = PetscInfo2(F,"MUMPS in numerical factorization phase: INFOG(1)=%d, cannot allocate required memory %d megabytes\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1227 F->factorerrortype = MAT_FACTOR_OUTMEMORY; 1228 } else if (mumps->id.INFOG(1) == -8 || mumps->id.INFOG(1) == -9 || (-16 < mumps->id.INFOG(1) && mumps->id.INFOG(1) < -10) ) { 1229 ierr = PetscInfo2(F,"MUMPS in numerical factorization phase: INFOG(1)=%d, INFO(2)=%d, problem with workarray \n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1230 F->factorerrortype = MAT_FACTOR_OUTMEMORY; 1231 } else { 1232 ierr = PetscInfo2(F,"MUMPS in numerical factorization phase: INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1233 F->factorerrortype = MAT_FACTOR_OTHER; 1234 } 1235 } 1236 } 1237 if (!mumps->myid && mumps->id.ICNTL(16) > 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB," mumps->id.ICNTL(16):=%d\n",mumps->id.INFOG(16)); 1238 1239 (F)->assembled = PETSC_TRUE; 1240 mumps->matstruc = SAME_NONZERO_PATTERN; 1241 mumps->schur_factored = PETSC_FALSE; 1242 mumps->schur_inverted = PETSC_FALSE; 1243 1244 /* just to be sure that ICNTL(19) value returned by a call from MatMumpsGetIcntl is always consistent */ 1245 if (!mumps->sym && mumps->id.ICNTL(19) && mumps->id.ICNTL(19) != 1) mumps->id.ICNTL(19) = 3; 1246 1247 if (mumps->size > 1) { 1248 PetscInt lsol_loc; 1249 PetscScalar *sol_loc; 1250 1251 ierr = PetscObjectTypeCompare((PetscObject)A,MATMPIAIJ,&isMPIAIJ);CHKERRQ(ierr); 1252 1253 /* distributed solution; Create x_seq=sol_loc for repeated use */ 1254 if (mumps->x_seq) { 1255 ierr = VecScatterDestroy(&mumps->scat_sol);CHKERRQ(ierr); 1256 ierr = PetscFree2(mumps->id.sol_loc,mumps->id.isol_loc);CHKERRQ(ierr); 1257 ierr = VecDestroy(&mumps->x_seq);CHKERRQ(ierr); 1258 } 1259 lsol_loc = mumps->id.INFO(23); /* length of sol_loc */ 1260 ierr = PetscMalloc2(lsol_loc,&sol_loc,lsol_loc,&mumps->id.isol_loc);CHKERRQ(ierr); 1261 mumps->id.lsol_loc = lsol_loc; 1262 mumps->id.sol_loc = (MumpsScalar*)sol_loc; 1263 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,1,lsol_loc,sol_loc,&mumps->x_seq);CHKERRQ(ierr); 1264 } 1265 PetscFunctionReturn(0); 1266 } 1267 1268 /* Sets MUMPS options from the options database */ 1269 PetscErrorCode PetscSetMUMPSFromOptions(Mat F, Mat A) 1270 { 1271 Mat_MUMPS *mumps = (Mat_MUMPS*)F->data; 1272 PetscErrorCode ierr; 1273 PetscInt icntl,info[40],i,ninfo=40; 1274 PetscBool flg; 1275 1276 PetscFunctionBegin; 1277 ierr = PetscOptionsBegin(PetscObjectComm((PetscObject)A),((PetscObject)A)->prefix,"MUMPS Options","Mat");CHKERRQ(ierr); 1278 ierr = PetscOptionsInt("-mat_mumps_icntl_1","ICNTL(1): output stream for error messages","None",mumps->id.ICNTL(1),&icntl,&flg);CHKERRQ(ierr); 1279 if (flg) mumps->id.ICNTL(1) = icntl; 1280 ierr = PetscOptionsInt("-mat_mumps_icntl_2","ICNTL(2): output stream for diagnostic printing, statistics, and warning","None",mumps->id.ICNTL(2),&icntl,&flg);CHKERRQ(ierr); 1281 if (flg) mumps->id.ICNTL(2) = icntl; 1282 ierr = PetscOptionsInt("-mat_mumps_icntl_3","ICNTL(3): output stream for global information, collected on the host","None",mumps->id.ICNTL(3),&icntl,&flg);CHKERRQ(ierr); 1283 if (flg) mumps->id.ICNTL(3) = icntl; 1284 1285 ierr = PetscOptionsInt("-mat_mumps_icntl_4","ICNTL(4): level of printing (0 to 4)","None",mumps->id.ICNTL(4),&icntl,&flg);CHKERRQ(ierr); 1286 if (flg) mumps->id.ICNTL(4) = icntl; 1287 if (mumps->id.ICNTL(4) || PetscLogPrintInfo) mumps->id.ICNTL(3) = 6; /* resume MUMPS default id.ICNTL(3) = 6 */ 1288 1289 ierr = PetscOptionsInt("-mat_mumps_icntl_6","ICNTL(6): permutes to a zero-free diagonal and/or scale the matrix (0 to 7)","None",mumps->id.ICNTL(6),&icntl,&flg);CHKERRQ(ierr); 1290 if (flg) mumps->id.ICNTL(6) = icntl; 1291 1292 ierr = PetscOptionsInt("-mat_mumps_icntl_7","ICNTL(7): computes a symmetric permutation in sequential analysis (0 to 7). 3=Scotch, 4=PORD, 5=Metis","None",mumps->id.ICNTL(7),&icntl,&flg);CHKERRQ(ierr); 1293 if (flg) { 1294 if (icntl== 1 && mumps->size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"pivot order be set by the user in PERM_IN -- not supported by the PETSc/MUMPS interface\n"); 1295 else mumps->id.ICNTL(7) = icntl; 1296 } 1297 1298 ierr = PetscOptionsInt("-mat_mumps_icntl_8","ICNTL(8): scaling strategy (-2 to 8 or 77)","None",mumps->id.ICNTL(8),&mumps->id.ICNTL(8),NULL);CHKERRQ(ierr); 1299 /* ierr = PetscOptionsInt("-mat_mumps_icntl_9","ICNTL(9): computes the solution using A or A^T","None",mumps->id.ICNTL(9),&mumps->id.ICNTL(9),NULL);CHKERRQ(ierr); handled by MatSolveTranspose_MUMPS() */ 1300 ierr = PetscOptionsInt("-mat_mumps_icntl_10","ICNTL(10): max num of refinements","None",mumps->id.ICNTL(10),&mumps->id.ICNTL(10),NULL);CHKERRQ(ierr); 1301 ierr = PetscOptionsInt("-mat_mumps_icntl_11","ICNTL(11): statistics related to an error analysis (via -ksp_view)","None",mumps->id.ICNTL(11),&mumps->id.ICNTL(11),NULL);CHKERRQ(ierr); 1302 ierr = PetscOptionsInt("-mat_mumps_icntl_12","ICNTL(12): an ordering strategy for symmetric matrices (0 to 3)","None",mumps->id.ICNTL(12),&mumps->id.ICNTL(12),NULL);CHKERRQ(ierr); 1303 ierr = PetscOptionsInt("-mat_mumps_icntl_13","ICNTL(13): parallelism of the root node (enable ScaLAPACK) and its splitting","None",mumps->id.ICNTL(13),&mumps->id.ICNTL(13),NULL);CHKERRQ(ierr); 1304 ierr = PetscOptionsInt("-mat_mumps_icntl_14","ICNTL(14): percentage increase in the estimated working space","None",mumps->id.ICNTL(14),&mumps->id.ICNTL(14),NULL);CHKERRQ(ierr); 1305 ierr = PetscOptionsInt("-mat_mumps_icntl_19","ICNTL(19): computes the Schur complement","None",mumps->id.ICNTL(19),&mumps->id.ICNTL(19),NULL);CHKERRQ(ierr); 1306 if (mumps->id.ICNTL(19) <= 0 || mumps->id.ICNTL(19) > 3) { /* reset any schur data (if any) */ 1307 ierr = MatMumpsResetSchur_Private(mumps);CHKERRQ(ierr); 1308 } 1309 /* ierr = PetscOptionsInt("-mat_mumps_icntl_20","ICNTL(20): the format (dense or sparse) of the right-hand sides","None",mumps->id.ICNTL(20),&mumps->id.ICNTL(20),NULL);CHKERRQ(ierr); -- sparse rhs is not supported in PETSc API */ 1310 /* ierr = PetscOptionsInt("-mat_mumps_icntl_21","ICNTL(21): the distribution (centralized or distributed) of the solution vectors","None",mumps->id.ICNTL(21),&mumps->id.ICNTL(21),NULL);CHKERRQ(ierr); we only use distributed solution vector */ 1311 1312 ierr = PetscOptionsInt("-mat_mumps_icntl_22","ICNTL(22): in-core/out-of-core factorization and solve (0 or 1)","None",mumps->id.ICNTL(22),&mumps->id.ICNTL(22),NULL);CHKERRQ(ierr); 1313 ierr = PetscOptionsInt("-mat_mumps_icntl_23","ICNTL(23): max size of the working memory (MB) that can allocate per processor","None",mumps->id.ICNTL(23),&mumps->id.ICNTL(23),NULL);CHKERRQ(ierr); 1314 ierr = PetscOptionsInt("-mat_mumps_icntl_24","ICNTL(24): detection of null pivot rows (0 or 1)","None",mumps->id.ICNTL(24),&mumps->id.ICNTL(24),NULL);CHKERRQ(ierr); 1315 if (mumps->id.ICNTL(24)) { 1316 mumps->id.ICNTL(13) = 1; /* turn-off ScaLAPACK to help with the correct detection of null pivots */ 1317 } 1318 1319 ierr = PetscOptionsInt("-mat_mumps_icntl_25","ICNTL(25): compute a solution of a deficient matrix and a null space basis","None",mumps->id.ICNTL(25),&mumps->id.ICNTL(25),NULL);CHKERRQ(ierr); 1320 ierr = PetscOptionsInt("-mat_mumps_icntl_26","ICNTL(26): drives the solution phase if a Schur complement matrix","None",mumps->id.ICNTL(26),&mumps->id.ICNTL(26),NULL);CHKERRQ(ierr); 1321 ierr = PetscOptionsInt("-mat_mumps_icntl_27","ICNTL(27): the blocking size for multiple right-hand sides","None",mumps->id.ICNTL(27),&mumps->id.ICNTL(27),NULL);CHKERRQ(ierr); 1322 ierr = PetscOptionsInt("-mat_mumps_icntl_28","ICNTL(28): use 1 for sequential analysis and ictnl(7) ordering, or 2 for parallel analysis and ictnl(29) ordering","None",mumps->id.ICNTL(28),&mumps->id.ICNTL(28),NULL);CHKERRQ(ierr); 1323 ierr = PetscOptionsInt("-mat_mumps_icntl_29","ICNTL(29): parallel ordering 1 = ptscotch, 2 = parmetis","None",mumps->id.ICNTL(29),&mumps->id.ICNTL(29),NULL);CHKERRQ(ierr); 1324 ierr = PetscOptionsInt("-mat_mumps_icntl_30","ICNTL(30): compute user-specified set of entries in inv(A)","None",mumps->id.ICNTL(30),&mumps->id.ICNTL(30),NULL);CHKERRQ(ierr); 1325 ierr = PetscOptionsInt("-mat_mumps_icntl_31","ICNTL(31): indicates which factors may be discarded during factorization","None",mumps->id.ICNTL(31),&mumps->id.ICNTL(31),NULL);CHKERRQ(ierr); 1326 /* ierr = PetscOptionsInt("-mat_mumps_icntl_32","ICNTL(32): performs the forward elemination of the right-hand sides during factorization","None",mumps->id.ICNTL(32),&mumps->id.ICNTL(32),NULL);CHKERRQ(ierr); -- not supported by PETSc API */ 1327 ierr = PetscOptionsInt("-mat_mumps_icntl_33","ICNTL(33): compute determinant","None",mumps->id.ICNTL(33),&mumps->id.ICNTL(33),NULL);CHKERRQ(ierr); 1328 1329 ierr = PetscOptionsReal("-mat_mumps_cntl_1","CNTL(1): relative pivoting threshold","None",mumps->id.CNTL(1),&mumps->id.CNTL(1),NULL);CHKERRQ(ierr); 1330 ierr = PetscOptionsReal("-mat_mumps_cntl_2","CNTL(2): stopping criterion of refinement","None",mumps->id.CNTL(2),&mumps->id.CNTL(2),NULL);CHKERRQ(ierr); 1331 ierr = PetscOptionsReal("-mat_mumps_cntl_3","CNTL(3): absolute pivoting threshold","None",mumps->id.CNTL(3),&mumps->id.CNTL(3),NULL);CHKERRQ(ierr); 1332 ierr = PetscOptionsReal("-mat_mumps_cntl_4","CNTL(4): value for static pivoting","None",mumps->id.CNTL(4),&mumps->id.CNTL(4),NULL);CHKERRQ(ierr); 1333 ierr = PetscOptionsReal("-mat_mumps_cntl_5","CNTL(5): fixation for null pivots","None",mumps->id.CNTL(5),&mumps->id.CNTL(5),NULL);CHKERRQ(ierr); 1334 1335 ierr = PetscOptionsString("-mat_mumps_ooc_tmpdir", "out of core directory", "None", mumps->id.ooc_tmpdir, mumps->id.ooc_tmpdir, 256, NULL);CHKERRQ(ierr); 1336 1337 ierr = PetscOptionsIntArray("-mat_mumps_view_info","request INFO local to each processor","",info,&ninfo,NULL);CHKERRQ(ierr); 1338 if (ninfo) { 1339 if (ninfo > 40) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_USER,"number of INFO %d must <= 40\n",ninfo); 1340 ierr = PetscMalloc1(ninfo,&mumps->info);CHKERRQ(ierr); 1341 mumps->ninfo = ninfo; 1342 for (i=0; i<ninfo; i++) { 1343 if (info[i] < 0 || info[i]>40) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_USER,"index of INFO %d must between 1 and 40\n",ninfo); 1344 else mumps->info[i] = info[i]; 1345 } 1346 } 1347 1348 ierr = PetscOptionsEnd();CHKERRQ(ierr); 1349 PetscFunctionReturn(0); 1350 } 1351 1352 PetscErrorCode PetscInitializeMUMPS(Mat A,Mat_MUMPS *mumps) 1353 { 1354 PetscErrorCode ierr; 1355 1356 PetscFunctionBegin; 1357 ierr = MPI_Comm_rank(PetscObjectComm((PetscObject)A), &mumps->myid);CHKERRQ(ierr); 1358 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&mumps->size);CHKERRQ(ierr); 1359 ierr = MPI_Comm_dup(PetscObjectComm((PetscObject)A),&(mumps->comm_mumps));CHKERRQ(ierr); 1360 1361 mumps->id.comm_fortran = MPI_Comm_c2f(mumps->comm_mumps); 1362 1363 mumps->id.job = JOB_INIT; 1364 mumps->id.par = 1; /* host participates factorizaton and solve */ 1365 mumps->id.sym = mumps->sym; 1366 PetscMUMPS_c(&mumps->id); 1367 1368 mumps->scat_rhs = NULL; 1369 mumps->scat_sol = NULL; 1370 1371 /* set PETSc-MUMPS default options - override MUMPS default */ 1372 mumps->id.ICNTL(3) = 0; 1373 mumps->id.ICNTL(4) = 0; 1374 if (mumps->size == 1) { 1375 mumps->id.ICNTL(18) = 0; /* centralized assembled matrix input */ 1376 } else { 1377 mumps->id.ICNTL(18) = 3; /* distributed assembled matrix input */ 1378 mumps->id.ICNTL(20) = 0; /* rhs is in dense format */ 1379 mumps->id.ICNTL(21) = 1; /* distributed solution */ 1380 } 1381 1382 /* schur */ 1383 mumps->id.size_schur = 0; 1384 mumps->id.listvar_schur = NULL; 1385 mumps->id.schur = NULL; 1386 mumps->sizeredrhs = 0; 1387 mumps->schur_pivots = NULL; 1388 mumps->schur_work = NULL; 1389 mumps->schur_sol = NULL; 1390 mumps->schur_sizesol = 0; 1391 mumps->schur_factored = PETSC_FALSE; 1392 mumps->schur_inverted = PETSC_FALSE; 1393 mumps->schur_sym = mumps->id.sym; 1394 PetscFunctionReturn(0); 1395 } 1396 1397 PetscErrorCode MatFactorSymbolic_MUMPS_ReportIfError(Mat F,Mat A,const MatFactorInfo *info,Mat_MUMPS *mumps) 1398 { 1399 PetscErrorCode ierr; 1400 1401 PetscFunctionBegin; 1402 if (mumps->id.INFOG(1) < 0) { 1403 if (A->erroriffailure) { 1404 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_LIB,"Error reported by MUMPS in analysis phase: INFOG(1)=%d\n",mumps->id.INFOG(1)); 1405 } else { 1406 if (mumps->id.INFOG(1) == -6) { 1407 ierr = PetscInfo2(F,"matrix is singular in structure, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1408 F->factorerrortype = MAT_FACTOR_STRUCT_ZEROPIVOT; 1409 } else if (mumps->id.INFOG(1) == -5 || mumps->id.INFOG(1) == -7) { 1410 ierr = PetscInfo2(F,"problem of workspace, INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1411 F->factorerrortype = MAT_FACTOR_OUTMEMORY; 1412 } else { 1413 ierr = PetscInfo2(F,"Error reported by MUMPS in analysis phase: INFOG(1)=%d, INFO(2)=%d\n",mumps->id.INFOG(1),mumps->id.INFO(2));CHKERRQ(ierr); 1414 F->factorerrortype = MAT_FACTOR_OTHER; 1415 } 1416 } 1417 } 1418 PetscFunctionReturn(0); 1419 } 1420 1421 /* Note Petsc r(=c) permutation is used when mumps->id.ICNTL(7)==1 with centralized assembled matrix input; otherwise r and c are ignored */ 1422 PetscErrorCode MatLUFactorSymbolic_AIJMUMPS(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info) 1423 { 1424 Mat_MUMPS *mumps = (Mat_MUMPS*)F->data; 1425 PetscErrorCode ierr; 1426 Vec b; 1427 IS is_iden; 1428 const PetscInt M = A->rmap->N; 1429 1430 PetscFunctionBegin; 1431 mumps->matstruc = DIFFERENT_NONZERO_PATTERN; 1432 1433 /* Set MUMPS options from the options database */ 1434 ierr = PetscSetMUMPSFromOptions(F,A);CHKERRQ(ierr); 1435 1436 ierr = (*mumps->ConvertToTriples)(A, 1, MAT_INITIAL_MATRIX, &mumps->nz, &mumps->irn, &mumps->jcn, &mumps->val);CHKERRQ(ierr); 1437 1438 /* analysis phase */ 1439 /*----------------*/ 1440 mumps->id.job = JOB_FACTSYMBOLIC; 1441 mumps->id.n = M; 1442 switch (mumps->id.ICNTL(18)) { 1443 case 0: /* centralized assembled matrix input */ 1444 if (!mumps->myid) { 1445 mumps->id.nz =mumps->nz; mumps->id.irn=mumps->irn; mumps->id.jcn=mumps->jcn; 1446 if (mumps->id.ICNTL(6)>1) { 1447 mumps->id.a = (MumpsScalar*)mumps->val; 1448 } 1449 if (mumps->id.ICNTL(7) == 1) { /* use user-provide matrix ordering - assuming r = c ordering */ 1450 /* 1451 PetscBool flag; 1452 ierr = ISEqual(r,c,&flag);CHKERRQ(ierr); 1453 if (!flag) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER,"row_perm != col_perm"); 1454 ierr = ISView(r,PETSC_VIEWER_STDOUT_SELF); 1455 */ 1456 if (!mumps->myid) { 1457 const PetscInt *idx; 1458 PetscInt i,*perm_in; 1459 1460 ierr = PetscMalloc1(M,&perm_in);CHKERRQ(ierr); 1461 ierr = ISGetIndices(r,&idx);CHKERRQ(ierr); 1462 1463 mumps->id.perm_in = perm_in; 1464 for (i=0; i<M; i++) perm_in[i] = idx[i]+1; /* perm_in[]: start from 1, not 0! */ 1465 ierr = ISRestoreIndices(r,&idx);CHKERRQ(ierr); 1466 } 1467 } 1468 } 1469 break; 1470 case 3: /* distributed assembled matrix input (size>1) */ 1471 mumps->id.nz_loc = mumps->nz; 1472 mumps->id.irn_loc=mumps->irn; mumps->id.jcn_loc=mumps->jcn; 1473 if (mumps->id.ICNTL(6)>1) { 1474 mumps->id.a_loc = (MumpsScalar*)mumps->val; 1475 } 1476 /* MUMPS only supports centralized rhs. Create scatter scat_rhs for repeated use in MatSolve() */ 1477 if (!mumps->myid) { 1478 ierr = VecCreateSeq(PETSC_COMM_SELF,A->rmap->N,&mumps->b_seq);CHKERRQ(ierr); 1479 ierr = ISCreateStride(PETSC_COMM_SELF,A->rmap->N,0,1,&is_iden);CHKERRQ(ierr); 1480 } else { 1481 ierr = VecCreateSeq(PETSC_COMM_SELF,0,&mumps->b_seq);CHKERRQ(ierr); 1482 ierr = ISCreateStride(PETSC_COMM_SELF,0,0,1,&is_iden);CHKERRQ(ierr); 1483 } 1484 ierr = MatCreateVecs(A,NULL,&b);CHKERRQ(ierr); 1485 ierr = VecScatterCreate(b,is_iden,mumps->b_seq,is_iden,&mumps->scat_rhs);CHKERRQ(ierr); 1486 ierr = ISDestroy(&is_iden);CHKERRQ(ierr); 1487 ierr = VecDestroy(&b);CHKERRQ(ierr); 1488 break; 1489 } 1490 PetscMUMPS_c(&mumps->id); 1491 ierr = MatFactorSymbolic_MUMPS_ReportIfError(F,A,info,mumps);CHKERRQ(ierr); 1492 1493 F->ops->lufactornumeric = MatFactorNumeric_MUMPS; 1494 F->ops->solve = MatSolve_MUMPS; 1495 F->ops->solvetranspose = MatSolveTranspose_MUMPS; 1496 F->ops->matsolve = MatMatSolve_MUMPS; 1497 PetscFunctionReturn(0); 1498 } 1499 1500 /* Note the Petsc r and c permutations are ignored */ 1501 PetscErrorCode MatLUFactorSymbolic_BAIJMUMPS(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info) 1502 { 1503 Mat_MUMPS *mumps = (Mat_MUMPS*)F->data; 1504 PetscErrorCode ierr; 1505 Vec b; 1506 IS is_iden; 1507 const PetscInt M = A->rmap->N; 1508 1509 PetscFunctionBegin; 1510 mumps->matstruc = DIFFERENT_NONZERO_PATTERN; 1511 1512 /* Set MUMPS options from the options database */ 1513 ierr = PetscSetMUMPSFromOptions(F,A);CHKERRQ(ierr); 1514 1515 ierr = (*mumps->ConvertToTriples)(A, 1, MAT_INITIAL_MATRIX, &mumps->nz, &mumps->irn, &mumps->jcn, &mumps->val);CHKERRQ(ierr); 1516 1517 /* analysis phase */ 1518 /*----------------*/ 1519 mumps->id.job = JOB_FACTSYMBOLIC; 1520 mumps->id.n = M; 1521 switch (mumps->id.ICNTL(18)) { 1522 case 0: /* centralized assembled matrix input */ 1523 if (!mumps->myid) { 1524 mumps->id.nz =mumps->nz; mumps->id.irn=mumps->irn; mumps->id.jcn=mumps->jcn; 1525 if (mumps->id.ICNTL(6)>1) { 1526 mumps->id.a = (MumpsScalar*)mumps->val; 1527 } 1528 } 1529 break; 1530 case 3: /* distributed assembled matrix input (size>1) */ 1531 mumps->id.nz_loc = mumps->nz; 1532 mumps->id.irn_loc=mumps->irn; mumps->id.jcn_loc=mumps->jcn; 1533 if (mumps->id.ICNTL(6)>1) { 1534 mumps->id.a_loc = (MumpsScalar*)mumps->val; 1535 } 1536 /* MUMPS only supports centralized rhs. Create scatter scat_rhs for repeated use in MatSolve() */ 1537 if (!mumps->myid) { 1538 ierr = VecCreateSeq(PETSC_COMM_SELF,A->cmap->N,&mumps->b_seq);CHKERRQ(ierr); 1539 ierr = ISCreateStride(PETSC_COMM_SELF,A->cmap->N,0,1,&is_iden);CHKERRQ(ierr); 1540 } else { 1541 ierr = VecCreateSeq(PETSC_COMM_SELF,0,&mumps->b_seq);CHKERRQ(ierr); 1542 ierr = ISCreateStride(PETSC_COMM_SELF,0,0,1,&is_iden);CHKERRQ(ierr); 1543 } 1544 ierr = MatCreateVecs(A,NULL,&b);CHKERRQ(ierr); 1545 ierr = VecScatterCreate(b,is_iden,mumps->b_seq,is_iden,&mumps->scat_rhs);CHKERRQ(ierr); 1546 ierr = ISDestroy(&is_iden);CHKERRQ(ierr); 1547 ierr = VecDestroy(&b);CHKERRQ(ierr); 1548 break; 1549 } 1550 PetscMUMPS_c(&mumps->id); 1551 ierr = MatFactorSymbolic_MUMPS_ReportIfError(F,A,info,mumps);CHKERRQ(ierr); 1552 1553 F->ops->lufactornumeric = MatFactorNumeric_MUMPS; 1554 F->ops->solve = MatSolve_MUMPS; 1555 F->ops->solvetranspose = MatSolveTranspose_MUMPS; 1556 PetscFunctionReturn(0); 1557 } 1558 1559 /* Note the Petsc r permutation and factor info are ignored */ 1560 PetscErrorCode MatCholeskyFactorSymbolic_MUMPS(Mat F,Mat A,IS r,const MatFactorInfo *info) 1561 { 1562 Mat_MUMPS *mumps = (Mat_MUMPS*)F->data; 1563 PetscErrorCode ierr; 1564 Vec b; 1565 IS is_iden; 1566 const PetscInt M = A->rmap->N; 1567 1568 PetscFunctionBegin; 1569 mumps->matstruc = DIFFERENT_NONZERO_PATTERN; 1570 1571 /* Set MUMPS options from the options database */ 1572 ierr = PetscSetMUMPSFromOptions(F,A);CHKERRQ(ierr); 1573 1574 ierr = (*mumps->ConvertToTriples)(A, 1, MAT_INITIAL_MATRIX, &mumps->nz, &mumps->irn, &mumps->jcn, &mumps->val);CHKERRQ(ierr); 1575 1576 /* analysis phase */ 1577 /*----------------*/ 1578 mumps->id.job = JOB_FACTSYMBOLIC; 1579 mumps->id.n = M; 1580 switch (mumps->id.ICNTL(18)) { 1581 case 0: /* centralized assembled matrix input */ 1582 if (!mumps->myid) { 1583 mumps->id.nz =mumps->nz; mumps->id.irn=mumps->irn; mumps->id.jcn=mumps->jcn; 1584 if (mumps->id.ICNTL(6)>1) { 1585 mumps->id.a = (MumpsScalar*)mumps->val; 1586 } 1587 } 1588 break; 1589 case 3: /* distributed assembled matrix input (size>1) */ 1590 mumps->id.nz_loc = mumps->nz; 1591 mumps->id.irn_loc=mumps->irn; mumps->id.jcn_loc=mumps->jcn; 1592 if (mumps->id.ICNTL(6)>1) { 1593 mumps->id.a_loc = (MumpsScalar*)mumps->val; 1594 } 1595 /* MUMPS only supports centralized rhs. Create scatter scat_rhs for repeated use in MatSolve() */ 1596 if (!mumps->myid) { 1597 ierr = VecCreateSeq(PETSC_COMM_SELF,A->cmap->N,&mumps->b_seq);CHKERRQ(ierr); 1598 ierr = ISCreateStride(PETSC_COMM_SELF,A->cmap->N,0,1,&is_iden);CHKERRQ(ierr); 1599 } else { 1600 ierr = VecCreateSeq(PETSC_COMM_SELF,0,&mumps->b_seq);CHKERRQ(ierr); 1601 ierr = ISCreateStride(PETSC_COMM_SELF,0,0,1,&is_iden);CHKERRQ(ierr); 1602 } 1603 ierr = MatCreateVecs(A,NULL,&b);CHKERRQ(ierr); 1604 ierr = VecScatterCreate(b,is_iden,mumps->b_seq,is_iden,&mumps->scat_rhs);CHKERRQ(ierr); 1605 ierr = ISDestroy(&is_iden);CHKERRQ(ierr); 1606 ierr = VecDestroy(&b);CHKERRQ(ierr); 1607 break; 1608 } 1609 PetscMUMPS_c(&mumps->id); 1610 ierr = MatFactorSymbolic_MUMPS_ReportIfError(F,A,info,mumps);CHKERRQ(ierr); 1611 1612 F->ops->choleskyfactornumeric = MatFactorNumeric_MUMPS; 1613 F->ops->solve = MatSolve_MUMPS; 1614 F->ops->solvetranspose = MatSolve_MUMPS; 1615 F->ops->matsolve = MatMatSolve_MUMPS; 1616 #if defined(PETSC_USE_COMPLEX) 1617 F->ops->getinertia = NULL; 1618 #else 1619 F->ops->getinertia = MatGetInertia_SBAIJMUMPS; 1620 #endif 1621 PetscFunctionReturn(0); 1622 } 1623 1624 PetscErrorCode MatView_MUMPS(Mat A,PetscViewer viewer) 1625 { 1626 PetscErrorCode ierr; 1627 PetscBool iascii; 1628 PetscViewerFormat format; 1629 Mat_MUMPS *mumps=(Mat_MUMPS*)A->data; 1630 1631 PetscFunctionBegin; 1632 /* check if matrix is mumps type */ 1633 if (A->ops->solve != MatSolve_MUMPS) PetscFunctionReturn(0); 1634 1635 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 1636 if (iascii) { 1637 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 1638 if (format == PETSC_VIEWER_ASCII_INFO) { 1639 ierr = PetscViewerASCIIPrintf(viewer,"MUMPS run parameters:\n");CHKERRQ(ierr); 1640 ierr = PetscViewerASCIIPrintf(viewer," SYM (matrix type): %d \n",mumps->id.sym);CHKERRQ(ierr); 1641 ierr = PetscViewerASCIIPrintf(viewer," PAR (host participation): %d \n",mumps->id.par);CHKERRQ(ierr); 1642 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(1) (output for error): %d \n",mumps->id.ICNTL(1));CHKERRQ(ierr); 1643 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(2) (output of diagnostic msg): %d \n",mumps->id.ICNTL(2));CHKERRQ(ierr); 1644 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(3) (output for global info): %d \n",mumps->id.ICNTL(3));CHKERRQ(ierr); 1645 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(4) (level of printing): %d \n",mumps->id.ICNTL(4));CHKERRQ(ierr); 1646 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(5) (input mat struct): %d \n",mumps->id.ICNTL(5));CHKERRQ(ierr); 1647 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(6) (matrix prescaling): %d \n",mumps->id.ICNTL(6));CHKERRQ(ierr); 1648 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(7) (sequential matrix ordering):%d \n",mumps->id.ICNTL(7));CHKERRQ(ierr); 1649 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(8) (scaling strategy): %d \n",mumps->id.ICNTL(8));CHKERRQ(ierr); 1650 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(10) (max num of refinements): %d \n",mumps->id.ICNTL(10));CHKERRQ(ierr); 1651 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(11) (error analysis): %d \n",mumps->id.ICNTL(11));CHKERRQ(ierr); 1652 if (mumps->id.ICNTL(11)>0) { 1653 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(4) (inf norm of input mat): %g\n",mumps->id.RINFOG(4));CHKERRQ(ierr); 1654 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(5) (inf norm of solution): %g\n",mumps->id.RINFOG(5));CHKERRQ(ierr); 1655 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(6) (inf norm of residual): %g\n",mumps->id.RINFOG(6));CHKERRQ(ierr); 1656 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(7),RINFOG(8) (backward error est): %g, %g\n",mumps->id.RINFOG(7),mumps->id.RINFOG(8));CHKERRQ(ierr); 1657 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(9) (error estimate): %g \n",mumps->id.RINFOG(9));CHKERRQ(ierr); 1658 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(10),RINFOG(11)(condition numbers): %g, %g\n",mumps->id.RINFOG(10),mumps->id.RINFOG(11));CHKERRQ(ierr); 1659 } 1660 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(12) (efficiency control): %d \n",mumps->id.ICNTL(12));CHKERRQ(ierr); 1661 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(13) (efficiency control): %d \n",mumps->id.ICNTL(13));CHKERRQ(ierr); 1662 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(14) (percentage of estimated workspace increase): %d \n",mumps->id.ICNTL(14));CHKERRQ(ierr); 1663 /* ICNTL(15-17) not used */ 1664 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(18) (input mat struct): %d \n",mumps->id.ICNTL(18));CHKERRQ(ierr); 1665 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(19) (Schur complement info): %d \n",mumps->id.ICNTL(19));CHKERRQ(ierr); 1666 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(20) (rhs sparse pattern): %d \n",mumps->id.ICNTL(20));CHKERRQ(ierr); 1667 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(21) (solution struct): %d \n",mumps->id.ICNTL(21));CHKERRQ(ierr); 1668 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(22) (in-core/out-of-core facility): %d \n",mumps->id.ICNTL(22));CHKERRQ(ierr); 1669 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(23) (max size of memory can be allocated locally):%d \n",mumps->id.ICNTL(23));CHKERRQ(ierr); 1670 1671 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(24) (detection of null pivot rows): %d \n",mumps->id.ICNTL(24));CHKERRQ(ierr); 1672 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(25) (computation of a null space basis): %d \n",mumps->id.ICNTL(25));CHKERRQ(ierr); 1673 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(26) (Schur options for rhs or solution): %d \n",mumps->id.ICNTL(26));CHKERRQ(ierr); 1674 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(27) (experimental parameter): %d \n",mumps->id.ICNTL(27));CHKERRQ(ierr); 1675 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(28) (use parallel or sequential ordering): %d \n",mumps->id.ICNTL(28));CHKERRQ(ierr); 1676 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(29) (parallel ordering): %d \n",mumps->id.ICNTL(29));CHKERRQ(ierr); 1677 1678 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(30) (user-specified set of entries in inv(A)): %d \n",mumps->id.ICNTL(30));CHKERRQ(ierr); 1679 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(31) (factors is discarded in the solve phase): %d \n",mumps->id.ICNTL(31));CHKERRQ(ierr); 1680 ierr = PetscViewerASCIIPrintf(viewer," ICNTL(33) (compute determinant): %d \n",mumps->id.ICNTL(33));CHKERRQ(ierr); 1681 1682 ierr = PetscViewerASCIIPrintf(viewer," CNTL(1) (relative pivoting threshold): %g \n",mumps->id.CNTL(1));CHKERRQ(ierr); 1683 ierr = PetscViewerASCIIPrintf(viewer," CNTL(2) (stopping criterion of refinement): %g \n",mumps->id.CNTL(2));CHKERRQ(ierr); 1684 ierr = PetscViewerASCIIPrintf(viewer," CNTL(3) (absolute pivoting threshold): %g \n",mumps->id.CNTL(3));CHKERRQ(ierr); 1685 ierr = PetscViewerASCIIPrintf(viewer," CNTL(4) (value of static pivoting): %g \n",mumps->id.CNTL(4));CHKERRQ(ierr); 1686 ierr = PetscViewerASCIIPrintf(viewer," CNTL(5) (fixation for null pivots): %g \n",mumps->id.CNTL(5));CHKERRQ(ierr); 1687 1688 /* infomation local to each processor */ 1689 ierr = PetscViewerASCIIPrintf(viewer, " RINFO(1) (local estimated flops for the elimination after analysis): \n");CHKERRQ(ierr); 1690 ierr = PetscViewerASCIIPushSynchronized(viewer);CHKERRQ(ierr); 1691 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %g \n",mumps->myid,mumps->id.RINFO(1));CHKERRQ(ierr); 1692 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1693 ierr = PetscViewerASCIIPrintf(viewer, " RINFO(2) (local estimated flops for the assembly after factorization): \n");CHKERRQ(ierr); 1694 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %g \n",mumps->myid,mumps->id.RINFO(2));CHKERRQ(ierr); 1695 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1696 ierr = PetscViewerASCIIPrintf(viewer, " RINFO(3) (local estimated flops for the elimination after factorization): \n");CHKERRQ(ierr); 1697 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %g \n",mumps->myid,mumps->id.RINFO(3));CHKERRQ(ierr); 1698 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1699 1700 ierr = PetscViewerASCIIPrintf(viewer, " INFO(15) (estimated size of (in MB) MUMPS internal data for running numerical factorization): \n");CHKERRQ(ierr); 1701 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %d \n",mumps->myid,mumps->id.INFO(15));CHKERRQ(ierr); 1702 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1703 1704 ierr = PetscViewerASCIIPrintf(viewer, " INFO(16) (size of (in MB) MUMPS internal data used during numerical factorization): \n");CHKERRQ(ierr); 1705 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %d \n",mumps->myid,mumps->id.INFO(16));CHKERRQ(ierr); 1706 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1707 1708 ierr = PetscViewerASCIIPrintf(viewer, " INFO(23) (num of pivots eliminated on this processor after factorization): \n");CHKERRQ(ierr); 1709 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %d \n",mumps->myid,mumps->id.INFO(23));CHKERRQ(ierr); 1710 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1711 1712 if (mumps->ninfo && mumps->ninfo <= 40){ 1713 PetscInt i; 1714 for (i=0; i<mumps->ninfo; i++){ 1715 ierr = PetscViewerASCIIPrintf(viewer, " INFO(%d): \n",mumps->info[i]);CHKERRQ(ierr); 1716 ierr = PetscViewerASCIISynchronizedPrintf(viewer," [%d] %d \n",mumps->myid,mumps->id.INFO(mumps->info[i]));CHKERRQ(ierr); 1717 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 1718 } 1719 } 1720 1721 1722 ierr = PetscViewerASCIIPopSynchronized(viewer);CHKERRQ(ierr); 1723 1724 if (!mumps->myid) { /* information from the host */ 1725 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(1) (global estimated flops for the elimination after analysis): %g \n",mumps->id.RINFOG(1));CHKERRQ(ierr); 1726 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(2) (global estimated flops for the assembly after factorization): %g \n",mumps->id.RINFOG(2));CHKERRQ(ierr); 1727 ierr = PetscViewerASCIIPrintf(viewer," RINFOG(3) (global estimated flops for the elimination after factorization): %g \n",mumps->id.RINFOG(3));CHKERRQ(ierr); 1728 ierr = PetscViewerASCIIPrintf(viewer," (RINFOG(12) RINFOG(13))*2^INFOG(34) (determinant): (%g,%g)*(2^%d)\n",mumps->id.RINFOG(12),mumps->id.RINFOG(13),mumps->id.INFOG(34));CHKERRQ(ierr); 1729 1730 ierr = PetscViewerASCIIPrintf(viewer," INFOG(3) (estimated real workspace for factors on all processors after analysis): %d \n",mumps->id.INFOG(3));CHKERRQ(ierr); 1731 ierr = PetscViewerASCIIPrintf(viewer," INFOG(4) (estimated integer workspace for factors on all processors after analysis): %d \n",mumps->id.INFOG(4));CHKERRQ(ierr); 1732 ierr = PetscViewerASCIIPrintf(viewer," INFOG(5) (estimated maximum front size in the complete tree): %d \n",mumps->id.INFOG(5));CHKERRQ(ierr); 1733 ierr = PetscViewerASCIIPrintf(viewer," INFOG(6) (number of nodes in the complete tree): %d \n",mumps->id.INFOG(6));CHKERRQ(ierr); 1734 ierr = PetscViewerASCIIPrintf(viewer," INFOG(7) (ordering option effectively use after analysis): %d \n",mumps->id.INFOG(7));CHKERRQ(ierr); 1735 ierr = PetscViewerASCIIPrintf(viewer," INFOG(8) (structural symmetry in percent of the permuted matrix after analysis): %d \n",mumps->id.INFOG(8));CHKERRQ(ierr); 1736 ierr = PetscViewerASCIIPrintf(viewer," INFOG(9) (total real/complex workspace to store the matrix factors after factorization): %d \n",mumps->id.INFOG(9));CHKERRQ(ierr); 1737 ierr = PetscViewerASCIIPrintf(viewer," INFOG(10) (total integer space store the matrix factors after factorization): %d \n",mumps->id.INFOG(10));CHKERRQ(ierr); 1738 ierr = PetscViewerASCIIPrintf(viewer," INFOG(11) (order of largest frontal matrix after factorization): %d \n",mumps->id.INFOG(11));CHKERRQ(ierr); 1739 ierr = PetscViewerASCIIPrintf(viewer," INFOG(12) (number of off-diagonal pivots): %d \n",mumps->id.INFOG(12));CHKERRQ(ierr); 1740 ierr = PetscViewerASCIIPrintf(viewer," INFOG(13) (number of delayed pivots after factorization): %d \n",mumps->id.INFOG(13));CHKERRQ(ierr); 1741 ierr = PetscViewerASCIIPrintf(viewer," INFOG(14) (number of memory compress after factorization): %d \n",mumps->id.INFOG(14));CHKERRQ(ierr); 1742 ierr = PetscViewerASCIIPrintf(viewer," INFOG(15) (number of steps of iterative refinement after solution): %d \n",mumps->id.INFOG(15));CHKERRQ(ierr); 1743 ierr = PetscViewerASCIIPrintf(viewer," INFOG(16) (estimated size (in MB) of all MUMPS internal data for factorization after analysis: value on the most memory consuming processor): %d \n",mumps->id.INFOG(16));CHKERRQ(ierr); 1744 ierr = PetscViewerASCIIPrintf(viewer," INFOG(17) (estimated size of all MUMPS internal data for factorization after analysis: sum over all processors): %d \n",mumps->id.INFOG(17));CHKERRQ(ierr); 1745 ierr = PetscViewerASCIIPrintf(viewer," INFOG(18) (size of all MUMPS internal data allocated during factorization: value on the most memory consuming processor): %d \n",mumps->id.INFOG(18));CHKERRQ(ierr); 1746 ierr = PetscViewerASCIIPrintf(viewer," INFOG(19) (size of all MUMPS internal data allocated during factorization: sum over all processors): %d \n",mumps->id.INFOG(19));CHKERRQ(ierr); 1747 ierr = PetscViewerASCIIPrintf(viewer," INFOG(20) (estimated number of entries in the factors): %d \n",mumps->id.INFOG(20));CHKERRQ(ierr); 1748 ierr = PetscViewerASCIIPrintf(viewer," INFOG(21) (size in MB of memory effectively used during factorization - value on the most memory consuming processor): %d \n",mumps->id.INFOG(21));CHKERRQ(ierr); 1749 ierr = PetscViewerASCIIPrintf(viewer," INFOG(22) (size in MB of memory effectively used during factorization - sum over all processors): %d \n",mumps->id.INFOG(22));CHKERRQ(ierr); 1750 ierr = PetscViewerASCIIPrintf(viewer," INFOG(23) (after analysis: value of ICNTL(6) effectively used): %d \n",mumps->id.INFOG(23));CHKERRQ(ierr); 1751 ierr = PetscViewerASCIIPrintf(viewer," INFOG(24) (after analysis: value of ICNTL(12) effectively used): %d \n",mumps->id.INFOG(24));CHKERRQ(ierr); 1752 ierr = PetscViewerASCIIPrintf(viewer," INFOG(25) (after factorization: number of pivots modified by static pivoting): %d \n",mumps->id.INFOG(25));CHKERRQ(ierr); 1753 ierr = PetscViewerASCIIPrintf(viewer," INFOG(28) (after factorization: number of null pivots encountered): %d\n",mumps->id.INFOG(28));CHKERRQ(ierr); 1754 ierr = PetscViewerASCIIPrintf(viewer," INFOG(29) (after factorization: effective number of entries in the factors (sum over all processors)): %d\n",mumps->id.INFOG(29));CHKERRQ(ierr); 1755 ierr = PetscViewerASCIIPrintf(viewer," INFOG(30, 31) (after solution: size in Mbytes of memory used during solution phase): %d, %d\n",mumps->id.INFOG(30),mumps->id.INFOG(31));CHKERRQ(ierr); 1756 ierr = PetscViewerASCIIPrintf(viewer," INFOG(32) (after analysis: type of analysis done): %d\n",mumps->id.INFOG(32));CHKERRQ(ierr); 1757 ierr = PetscViewerASCIIPrintf(viewer," INFOG(33) (value used for ICNTL(8)): %d\n",mumps->id.INFOG(33));CHKERRQ(ierr); 1758 ierr = PetscViewerASCIIPrintf(viewer," INFOG(34) (exponent of the determinant if determinant is requested): %d\n",mumps->id.INFOG(34));CHKERRQ(ierr); 1759 } 1760 } 1761 } 1762 PetscFunctionReturn(0); 1763 } 1764 1765 PetscErrorCode MatGetInfo_MUMPS(Mat A,MatInfoType flag,MatInfo *info) 1766 { 1767 Mat_MUMPS *mumps =(Mat_MUMPS*)A->data; 1768 1769 PetscFunctionBegin; 1770 info->block_size = 1.0; 1771 info->nz_allocated = mumps->id.INFOG(20); 1772 info->nz_used = mumps->id.INFOG(20); 1773 info->nz_unneeded = 0.0; 1774 info->assemblies = 0.0; 1775 info->mallocs = 0.0; 1776 info->memory = 0.0; 1777 info->fill_ratio_given = 0; 1778 info->fill_ratio_needed = 0; 1779 info->factor_mallocs = 0; 1780 PetscFunctionReturn(0); 1781 } 1782 1783 /* -------------------------------------------------------------------------------------------*/ 1784 PetscErrorCode MatFactorSetSchurIS_MUMPS(Mat F, IS is) 1785 { 1786 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1787 const PetscInt *idxs; 1788 PetscInt size,i; 1789 PetscErrorCode ierr; 1790 1791 PetscFunctionBegin; 1792 if (mumps->size > 1) { 1793 PetscBool ls,gs; 1794 1795 ierr = ISGetLocalSize(is,&size);CHKERRQ(ierr); 1796 ls = mumps->myid ? (size ? PETSC_FALSE : PETSC_TRUE) : PETSC_TRUE; 1797 ierr = MPI_Allreduce(&ls,&gs,1,MPIU_BOOL,MPI_LAND,mumps->comm_mumps);CHKERRQ(ierr); 1798 if (!gs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MUMPS distributed parallel Schur complements not yet supported from PETSc\n"); 1799 } 1800 ierr = ISGetLocalSize(is,&size);CHKERRQ(ierr); 1801 if (mumps->id.size_schur != size) { 1802 ierr = PetscFree2(mumps->id.listvar_schur,mumps->id.schur);CHKERRQ(ierr); 1803 mumps->id.size_schur = size; 1804 mumps->id.schur_lld = size; 1805 ierr = PetscMalloc2(size,&mumps->id.listvar_schur,size*size,&mumps->id.schur);CHKERRQ(ierr); 1806 } 1807 ierr = ISGetIndices(is,&idxs);CHKERRQ(ierr); 1808 ierr = PetscMemcpy(mumps->id.listvar_schur,idxs,size*sizeof(PetscInt));CHKERRQ(ierr); 1809 /* MUMPS expects Fortran style indices */ 1810 for (i=0;i<size;i++) mumps->id.listvar_schur[i]++; 1811 ierr = ISRestoreIndices(is,&idxs);CHKERRQ(ierr); 1812 if (mumps->size > 1) { 1813 mumps->id.ICNTL(19) = 1; /* MUMPS returns Schur centralized on the host */ 1814 } else { 1815 if (F->factortype == MAT_FACTOR_LU) { 1816 mumps->id.ICNTL(19) = 3; /* MUMPS returns full matrix */ 1817 } else { 1818 mumps->id.ICNTL(19) = 2; /* MUMPS returns lower triangular part */ 1819 } 1820 } 1821 /* set a special value of ICNTL (not handled my MUMPS) to be used in the solve phase by PETSc */ 1822 mumps->id.ICNTL(26) = -1; 1823 PetscFunctionReturn(0); 1824 } 1825 1826 /* -------------------------------------------------------------------------------------------*/ 1827 PetscErrorCode MatFactorCreateSchurComplement_MUMPS(Mat F,Mat* S) 1828 { 1829 Mat St; 1830 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1831 PetscScalar *array; 1832 #if defined(PETSC_USE_COMPLEX) 1833 PetscScalar im = PetscSqrtScalar((PetscScalar)-1.0); 1834 #endif 1835 PetscErrorCode ierr; 1836 1837 PetscFunctionBegin; 1838 if (!mumps->id.ICNTL(19)) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ORDER,"Schur complement mode not selected! You should call MatFactorSetSchurIS to enable it"); 1839 ierr = MatCreate(PETSC_COMM_SELF,&St);CHKERRQ(ierr); 1840 ierr = MatSetSizes(St,PETSC_DECIDE,PETSC_DECIDE,mumps->id.size_schur,mumps->id.size_schur);CHKERRQ(ierr); 1841 ierr = MatSetType(St,MATDENSE);CHKERRQ(ierr); 1842 ierr = MatSetUp(St);CHKERRQ(ierr); 1843 ierr = MatDenseGetArray(St,&array);CHKERRQ(ierr); 1844 if (!mumps->sym) { /* MUMPS always return a full matrix */ 1845 if (mumps->id.ICNTL(19) == 1) { /* stored by rows */ 1846 PetscInt i,j,N=mumps->id.size_schur; 1847 for (i=0;i<N;i++) { 1848 for (j=0;j<N;j++) { 1849 #if !defined(PETSC_USE_COMPLEX) 1850 PetscScalar val = mumps->id.schur[i*N+j]; 1851 #else 1852 PetscScalar val = mumps->id.schur[i*N+j].r + im*mumps->id.schur[i*N+j].i; 1853 #endif 1854 array[j*N+i] = val; 1855 } 1856 } 1857 } else { /* stored by columns */ 1858 ierr = PetscMemcpy(array,mumps->id.schur,mumps->id.size_schur*mumps->id.size_schur*sizeof(PetscScalar));CHKERRQ(ierr); 1859 } 1860 } else { /* either full or lower-triangular (not packed) */ 1861 if (mumps->id.ICNTL(19) == 2) { /* lower triangular stored by columns */ 1862 PetscInt i,j,N=mumps->id.size_schur; 1863 for (i=0;i<N;i++) { 1864 for (j=i;j<N;j++) { 1865 #if !defined(PETSC_USE_COMPLEX) 1866 PetscScalar val = mumps->id.schur[i*N+j]; 1867 #else 1868 PetscScalar val = mumps->id.schur[i*N+j].r + im*mumps->id.schur[i*N+j].i; 1869 #endif 1870 array[i*N+j] = val; 1871 array[j*N+i] = val; 1872 } 1873 } 1874 } else if (mumps->id.ICNTL(19) == 3) { /* full matrix */ 1875 ierr = PetscMemcpy(array,mumps->id.schur,mumps->id.size_schur*mumps->id.size_schur*sizeof(PetscScalar));CHKERRQ(ierr); 1876 } else { /* ICNTL(19) == 1 lower triangular stored by rows */ 1877 PetscInt i,j,N=mumps->id.size_schur; 1878 for (i=0;i<N;i++) { 1879 for (j=0;j<i+1;j++) { 1880 #if !defined(PETSC_USE_COMPLEX) 1881 PetscScalar val = mumps->id.schur[i*N+j]; 1882 #else 1883 PetscScalar val = mumps->id.schur[i*N+j].r + im*mumps->id.schur[i*N+j].i; 1884 #endif 1885 array[i*N+j] = val; 1886 array[j*N+i] = val; 1887 } 1888 } 1889 } 1890 } 1891 ierr = MatDenseRestoreArray(St,&array);CHKERRQ(ierr); 1892 *S = St; 1893 PetscFunctionReturn(0); 1894 } 1895 1896 /* -------------------------------------------------------------------------------------------*/ 1897 PetscErrorCode MatFactorGetSchurComplement_MUMPS(Mat F,Mat* S) 1898 { 1899 Mat St; 1900 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1901 PetscErrorCode ierr; 1902 1903 PetscFunctionBegin; 1904 if (!mumps->id.ICNTL(19)) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ORDER,"Schur complement mode not selected! You should call MatFactorSetSchurIS to enable it"); 1905 /* It should be the responsibility of the user to handle different ICNTL(19) cases and factorization stages if they want to work with the raw data */ 1906 ierr = MatCreateSeqDense(PETSC_COMM_SELF,mumps->id.size_schur,mumps->id.size_schur,(PetscScalar*)mumps->id.schur,&St);CHKERRQ(ierr); 1907 *S = St; 1908 PetscFunctionReturn(0); 1909 } 1910 1911 /* -------------------------------------------------------------------------------------------*/ 1912 PetscErrorCode MatFactorFactorizeSchurComplement_MUMPS(Mat F) 1913 { 1914 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1915 PetscErrorCode ierr; 1916 1917 PetscFunctionBegin; 1918 if (!mumps->id.ICNTL(19)) { /* do nothing */ 1919 PetscFunctionReturn(0); 1920 } 1921 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Parallel Schur complement not yet supported from PETSc"); 1922 ierr = MatMumpsFactorSchur_Private(mumps);CHKERRQ(ierr); 1923 PetscFunctionReturn(0); 1924 } 1925 1926 PetscErrorCode MatFactorInvertSchurComplement_MUMPS(Mat F) 1927 { 1928 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1929 PetscErrorCode ierr; 1930 1931 PetscFunctionBegin; 1932 if (!mumps->id.ICNTL(19)) { /* do nothing */ 1933 PetscFunctionReturn(0); 1934 } 1935 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Parallel Schur complement not yet supported from PETSc"); 1936 ierr = MatMumpsInvertSchur_Private(mumps);CHKERRQ(ierr); 1937 PetscFunctionReturn(0); 1938 } 1939 1940 /* -------------------------------------------------------------------------------------------*/ 1941 PetscErrorCode MatFactorSolveSchurComplement_MUMPS(Mat F, Vec rhs, Vec sol) 1942 { 1943 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1944 MumpsScalar *orhs; 1945 PetscScalar *osol,*nrhs,*nsol; 1946 PetscInt orhs_size,osol_size,olrhs_size; 1947 PetscErrorCode ierr; 1948 1949 PetscFunctionBegin; 1950 if (!mumps->id.ICNTL(19)) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ORDER,"Schur complement mode not selected! You should call MatFactorSetSchurIS to enable it"); 1951 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Parallel Schur complement not yet supported from PETSc"); 1952 1953 /* swap pointers */ 1954 orhs = mumps->id.redrhs; 1955 olrhs_size = mumps->id.lredrhs; 1956 orhs_size = mumps->sizeredrhs; 1957 osol = mumps->schur_sol; 1958 osol_size = mumps->schur_sizesol; 1959 ierr = VecGetArray(rhs,&nrhs);CHKERRQ(ierr); 1960 ierr = VecGetArray(sol,&nsol);CHKERRQ(ierr); 1961 mumps->id.redrhs = (MumpsScalar*)nrhs; 1962 ierr = VecGetLocalSize(rhs,&mumps->sizeredrhs);CHKERRQ(ierr); 1963 mumps->id.lredrhs = mumps->sizeredrhs; 1964 mumps->schur_sol = nsol; 1965 ierr = VecGetLocalSize(sol,&mumps->schur_sizesol);CHKERRQ(ierr); 1966 1967 /* solve Schur complement */ 1968 mumps->id.nrhs = 1; 1969 ierr = MatMumpsSolveSchur_Private(mumps,PETSC_FALSE);CHKERRQ(ierr); 1970 /* restore pointers */ 1971 ierr = VecRestoreArray(rhs,&nrhs);CHKERRQ(ierr); 1972 ierr = VecRestoreArray(sol,&nsol);CHKERRQ(ierr); 1973 mumps->id.redrhs = orhs; 1974 mumps->id.lredrhs = olrhs_size; 1975 mumps->sizeredrhs = orhs_size; 1976 mumps->schur_sol = osol; 1977 mumps->schur_sizesol = osol_size; 1978 PetscFunctionReturn(0); 1979 } 1980 1981 /* -------------------------------------------------------------------------------------------*/ 1982 PetscErrorCode MatFactorSolveSchurComplementTranspose_MUMPS(Mat F, Vec rhs, Vec sol) 1983 { 1984 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 1985 MumpsScalar *orhs; 1986 PetscScalar *osol,*nrhs,*nsol; 1987 PetscInt orhs_size,osol_size; 1988 PetscErrorCode ierr; 1989 1990 PetscFunctionBegin; 1991 if (!mumps->id.ICNTL(19)) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ORDER,"Schur complement mode not selected! You should call MatFactorSetSchurIS to enable it"); 1992 if (mumps->size > 1) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_SUP,"Parallel Schur complement not yet supported from PETSc"); 1993 1994 /* swap pointers */ 1995 orhs = mumps->id.redrhs; 1996 orhs_size = mumps->sizeredrhs; 1997 osol = mumps->schur_sol; 1998 osol_size = mumps->schur_sizesol; 1999 ierr = VecGetArray(rhs,&nrhs);CHKERRQ(ierr); 2000 ierr = VecGetArray(sol,&nsol);CHKERRQ(ierr); 2001 mumps->id.redrhs = (MumpsScalar*)nrhs; 2002 ierr = VecGetLocalSize(rhs,&mumps->sizeredrhs);CHKERRQ(ierr); 2003 mumps->schur_sol = nsol; 2004 ierr = VecGetLocalSize(sol,&mumps->schur_sizesol);CHKERRQ(ierr); 2005 2006 /* solve Schur complement */ 2007 mumps->id.nrhs = 1; 2008 mumps->id.ICNTL(9) = 0; 2009 ierr = MatMumpsSolveSchur_Private(mumps,PETSC_FALSE);CHKERRQ(ierr); 2010 mumps->id.ICNTL(9) = 1; 2011 /* restore pointers */ 2012 ierr = VecRestoreArray(rhs,&nrhs);CHKERRQ(ierr); 2013 ierr = VecRestoreArray(sol,&nsol);CHKERRQ(ierr); 2014 mumps->id.redrhs = orhs; 2015 mumps->sizeredrhs = orhs_size; 2016 mumps->schur_sol = osol; 2017 mumps->schur_sizesol = osol_size; 2018 PetscFunctionReturn(0); 2019 } 2020 2021 /* -------------------------------------------------------------------------------------------*/ 2022 PetscErrorCode MatFactorSetSchurComplementSolverType_MUMPS(Mat F, PetscInt sym) 2023 { 2024 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2025 2026 PetscFunctionBegin; 2027 if (mumps->schur_factored && mumps->sym != mumps->schur_sym) { 2028 SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONG,"Cannot change the Schur solver! Schur complement data has been already factored"); 2029 } 2030 mumps->schur_sym = sym; 2031 PetscFunctionReturn(0); 2032 } 2033 2034 /* -------------------------------------------------------------------------------------------*/ 2035 PetscErrorCode MatMumpsSetIcntl_MUMPS(Mat F,PetscInt icntl,PetscInt ival) 2036 { 2037 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2038 2039 PetscFunctionBegin; 2040 mumps->id.ICNTL(icntl) = ival; 2041 PetscFunctionReturn(0); 2042 } 2043 2044 PetscErrorCode MatMumpsGetIcntl_MUMPS(Mat F,PetscInt icntl,PetscInt *ival) 2045 { 2046 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2047 2048 PetscFunctionBegin; 2049 *ival = mumps->id.ICNTL(icntl); 2050 PetscFunctionReturn(0); 2051 } 2052 2053 /*@ 2054 MatMumpsSetIcntl - Set MUMPS parameter ICNTL() 2055 2056 Logically Collective on Mat 2057 2058 Input Parameters: 2059 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2060 . icntl - index of MUMPS parameter array ICNTL() 2061 - ival - value of MUMPS ICNTL(icntl) 2062 2063 Options Database: 2064 . -mat_mumps_icntl_<icntl> <ival> 2065 2066 Level: beginner 2067 2068 References: 2069 . MUMPS Users' Guide 2070 2071 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2072 @*/ 2073 PetscErrorCode MatMumpsSetIcntl(Mat F,PetscInt icntl,PetscInt ival) 2074 { 2075 PetscErrorCode ierr; 2076 2077 PetscFunctionBegin; 2078 PetscValidType(F,1); 2079 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2080 PetscValidLogicalCollectiveInt(F,icntl,2); 2081 PetscValidLogicalCollectiveInt(F,ival,3); 2082 ierr = PetscTryMethod(F,"MatMumpsSetIcntl_C",(Mat,PetscInt,PetscInt),(F,icntl,ival));CHKERRQ(ierr); 2083 PetscFunctionReturn(0); 2084 } 2085 2086 /*@ 2087 MatMumpsGetIcntl - Get MUMPS parameter ICNTL() 2088 2089 Logically Collective on Mat 2090 2091 Input Parameters: 2092 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2093 - icntl - index of MUMPS parameter array ICNTL() 2094 2095 Output Parameter: 2096 . ival - value of MUMPS ICNTL(icntl) 2097 2098 Level: beginner 2099 2100 References: 2101 . MUMPS Users' Guide 2102 2103 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2104 @*/ 2105 PetscErrorCode MatMumpsGetIcntl(Mat F,PetscInt icntl,PetscInt *ival) 2106 { 2107 PetscErrorCode ierr; 2108 2109 PetscFunctionBegin; 2110 PetscValidType(F,1); 2111 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2112 PetscValidLogicalCollectiveInt(F,icntl,2); 2113 PetscValidIntPointer(ival,3); 2114 ierr = PetscUseMethod(F,"MatMumpsGetIcntl_C",(Mat,PetscInt,PetscInt*),(F,icntl,ival));CHKERRQ(ierr); 2115 PetscFunctionReturn(0); 2116 } 2117 2118 /* -------------------------------------------------------------------------------------------*/ 2119 PetscErrorCode MatMumpsSetCntl_MUMPS(Mat F,PetscInt icntl,PetscReal val) 2120 { 2121 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2122 2123 PetscFunctionBegin; 2124 mumps->id.CNTL(icntl) = val; 2125 PetscFunctionReturn(0); 2126 } 2127 2128 PetscErrorCode MatMumpsGetCntl_MUMPS(Mat F,PetscInt icntl,PetscReal *val) 2129 { 2130 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2131 2132 PetscFunctionBegin; 2133 *val = mumps->id.CNTL(icntl); 2134 PetscFunctionReturn(0); 2135 } 2136 2137 /*@ 2138 MatMumpsSetCntl - Set MUMPS parameter CNTL() 2139 2140 Logically Collective on Mat 2141 2142 Input Parameters: 2143 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2144 . icntl - index of MUMPS parameter array CNTL() 2145 - val - value of MUMPS CNTL(icntl) 2146 2147 Options Database: 2148 . -mat_mumps_cntl_<icntl> <val> 2149 2150 Level: beginner 2151 2152 References: 2153 . MUMPS Users' Guide 2154 2155 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2156 @*/ 2157 PetscErrorCode MatMumpsSetCntl(Mat F,PetscInt icntl,PetscReal val) 2158 { 2159 PetscErrorCode ierr; 2160 2161 PetscFunctionBegin; 2162 PetscValidType(F,1); 2163 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2164 PetscValidLogicalCollectiveInt(F,icntl,2); 2165 PetscValidLogicalCollectiveReal(F,val,3); 2166 ierr = PetscTryMethod(F,"MatMumpsSetCntl_C",(Mat,PetscInt,PetscReal),(F,icntl,val));CHKERRQ(ierr); 2167 PetscFunctionReturn(0); 2168 } 2169 2170 /*@ 2171 MatMumpsGetCntl - Get MUMPS parameter CNTL() 2172 2173 Logically Collective on Mat 2174 2175 Input Parameters: 2176 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2177 - icntl - index of MUMPS parameter array CNTL() 2178 2179 Output Parameter: 2180 . val - value of MUMPS CNTL(icntl) 2181 2182 Level: beginner 2183 2184 References: 2185 . MUMPS Users' Guide 2186 2187 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2188 @*/ 2189 PetscErrorCode MatMumpsGetCntl(Mat F,PetscInt icntl,PetscReal *val) 2190 { 2191 PetscErrorCode ierr; 2192 2193 PetscFunctionBegin; 2194 PetscValidType(F,1); 2195 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2196 PetscValidLogicalCollectiveInt(F,icntl,2); 2197 PetscValidRealPointer(val,3); 2198 ierr = PetscUseMethod(F,"MatMumpsGetCntl_C",(Mat,PetscInt,PetscReal*),(F,icntl,val));CHKERRQ(ierr); 2199 PetscFunctionReturn(0); 2200 } 2201 2202 PetscErrorCode MatMumpsGetInfo_MUMPS(Mat F,PetscInt icntl,PetscInt *info) 2203 { 2204 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2205 2206 PetscFunctionBegin; 2207 *info = mumps->id.INFO(icntl); 2208 PetscFunctionReturn(0); 2209 } 2210 2211 PetscErrorCode MatMumpsGetInfog_MUMPS(Mat F,PetscInt icntl,PetscInt *infog) 2212 { 2213 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2214 2215 PetscFunctionBegin; 2216 *infog = mumps->id.INFOG(icntl); 2217 PetscFunctionReturn(0); 2218 } 2219 2220 PetscErrorCode MatMumpsGetRinfo_MUMPS(Mat F,PetscInt icntl,PetscReal *rinfo) 2221 { 2222 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2223 2224 PetscFunctionBegin; 2225 *rinfo = mumps->id.RINFO(icntl); 2226 PetscFunctionReturn(0); 2227 } 2228 2229 PetscErrorCode MatMumpsGetRinfog_MUMPS(Mat F,PetscInt icntl,PetscReal *rinfog) 2230 { 2231 Mat_MUMPS *mumps =(Mat_MUMPS*)F->data; 2232 2233 PetscFunctionBegin; 2234 *rinfog = mumps->id.RINFOG(icntl); 2235 PetscFunctionReturn(0); 2236 } 2237 2238 /*@ 2239 MatMumpsGetInfo - Get MUMPS parameter INFO() 2240 2241 Logically Collective on Mat 2242 2243 Input Parameters: 2244 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2245 - icntl - index of MUMPS parameter array INFO() 2246 2247 Output Parameter: 2248 . ival - value of MUMPS INFO(icntl) 2249 2250 Level: beginner 2251 2252 References: 2253 . MUMPS Users' Guide 2254 2255 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2256 @*/ 2257 PetscErrorCode MatMumpsGetInfo(Mat F,PetscInt icntl,PetscInt *ival) 2258 { 2259 PetscErrorCode ierr; 2260 2261 PetscFunctionBegin; 2262 PetscValidType(F,1); 2263 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2264 PetscValidIntPointer(ival,3); 2265 ierr = PetscUseMethod(F,"MatMumpsGetInfo_C",(Mat,PetscInt,PetscInt*),(F,icntl,ival));CHKERRQ(ierr); 2266 PetscFunctionReturn(0); 2267 } 2268 2269 /*@ 2270 MatMumpsGetInfog - Get MUMPS parameter INFOG() 2271 2272 Logically Collective on Mat 2273 2274 Input Parameters: 2275 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2276 - icntl - index of MUMPS parameter array INFOG() 2277 2278 Output Parameter: 2279 . ival - value of MUMPS INFOG(icntl) 2280 2281 Level: beginner 2282 2283 References: 2284 . MUMPS Users' Guide 2285 2286 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2287 @*/ 2288 PetscErrorCode MatMumpsGetInfog(Mat F,PetscInt icntl,PetscInt *ival) 2289 { 2290 PetscErrorCode ierr; 2291 2292 PetscFunctionBegin; 2293 PetscValidType(F,1); 2294 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2295 PetscValidIntPointer(ival,3); 2296 ierr = PetscUseMethod(F,"MatMumpsGetInfog_C",(Mat,PetscInt,PetscInt*),(F,icntl,ival));CHKERRQ(ierr); 2297 PetscFunctionReturn(0); 2298 } 2299 2300 /*@ 2301 MatMumpsGetRinfo - Get MUMPS parameter RINFO() 2302 2303 Logically Collective on Mat 2304 2305 Input Parameters: 2306 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2307 - icntl - index of MUMPS parameter array RINFO() 2308 2309 Output Parameter: 2310 . val - value of MUMPS RINFO(icntl) 2311 2312 Level: beginner 2313 2314 References: 2315 . MUMPS Users' Guide 2316 2317 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2318 @*/ 2319 PetscErrorCode MatMumpsGetRinfo(Mat F,PetscInt icntl,PetscReal *val) 2320 { 2321 PetscErrorCode ierr; 2322 2323 PetscFunctionBegin; 2324 PetscValidType(F,1); 2325 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2326 PetscValidRealPointer(val,3); 2327 ierr = PetscUseMethod(F,"MatMumpsGetRinfo_C",(Mat,PetscInt,PetscReal*),(F,icntl,val));CHKERRQ(ierr); 2328 PetscFunctionReturn(0); 2329 } 2330 2331 /*@ 2332 MatMumpsGetRinfog - Get MUMPS parameter RINFOG() 2333 2334 Logically Collective on Mat 2335 2336 Input Parameters: 2337 + F - the factored matrix obtained by calling MatGetFactor() from PETSc-MUMPS interface 2338 - icntl - index of MUMPS parameter array RINFOG() 2339 2340 Output Parameter: 2341 . val - value of MUMPS RINFOG(icntl) 2342 2343 Level: beginner 2344 2345 References: 2346 . MUMPS Users' Guide 2347 2348 .seealso: MatGetFactor(), MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog() 2349 @*/ 2350 PetscErrorCode MatMumpsGetRinfog(Mat F,PetscInt icntl,PetscReal *val) 2351 { 2352 PetscErrorCode ierr; 2353 2354 PetscFunctionBegin; 2355 PetscValidType(F,1); 2356 if (!F->factortype) SETERRQ(PetscObjectComm((PetscObject)F),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 2357 PetscValidRealPointer(val,3); 2358 ierr = PetscUseMethod(F,"MatMumpsGetRinfog_C",(Mat,PetscInt,PetscReal*),(F,icntl,val));CHKERRQ(ierr); 2359 PetscFunctionReturn(0); 2360 } 2361 2362 /*MC 2363 MATSOLVERMUMPS - A matrix type providing direct solvers (LU and Cholesky) for 2364 distributed and sequential matrices via the external package MUMPS. 2365 2366 Works with MATAIJ and MATSBAIJ matrices 2367 2368 Use ./configure --download-mumps --download-scalapack --download-parmetis --download-metis --download-ptscotch to have PETSc installed with MUMPS 2369 2370 Use -pc_type cholesky or lu -pc_factor_mat_solver_package mumps to us this direct solver 2371 2372 Options Database Keys: 2373 + -mat_mumps_icntl_1 - ICNTL(1): output stream for error messages 2374 . -mat_mumps_icntl_2 - ICNTL(2): output stream for diagnostic printing, statistics, and warning 2375 . -mat_mumps_icntl_3 - ICNTL(3): output stream for global information, collected on the host 2376 . -mat_mumps_icntl_4 - ICNTL(4): level of printing (0 to 4) 2377 . -mat_mumps_icntl_6 - ICNTL(6): permutes to a zero-free diagonal and/or scale the matrix (0 to 7) 2378 . -mat_mumps_icntl_7 - ICNTL(7): computes a symmetric permutation in sequential analysis (0 to 7). 3=Scotch, 4=PORD, 5=Metis 2379 . -mat_mumps_icntl_8 - ICNTL(8): scaling strategy (-2 to 8 or 77) 2380 . -mat_mumps_icntl_10 - ICNTL(10): max num of refinements 2381 . -mat_mumps_icntl_11 - ICNTL(11): statistics related to an error analysis (via -ksp_view) 2382 . -mat_mumps_icntl_12 - ICNTL(12): an ordering strategy for symmetric matrices (0 to 3) 2383 . -mat_mumps_icntl_13 - ICNTL(13): parallelism of the root node (enable ScaLAPACK) and its splitting 2384 . -mat_mumps_icntl_14 - ICNTL(14): percentage increase in the estimated working space 2385 . -mat_mumps_icntl_19 - ICNTL(19): computes the Schur complement 2386 . -mat_mumps_icntl_22 - ICNTL(22): in-core/out-of-core factorization and solve (0 or 1) 2387 . -mat_mumps_icntl_23 - ICNTL(23): max size of the working memory (MB) that can allocate per processor 2388 . -mat_mumps_icntl_24 - ICNTL(24): detection of null pivot rows (0 or 1) 2389 . -mat_mumps_icntl_25 - ICNTL(25): compute a solution of a deficient matrix and a null space basis 2390 . -mat_mumps_icntl_26 - ICNTL(26): drives the solution phase if a Schur complement matrix 2391 . -mat_mumps_icntl_28 - ICNTL(28): use 1 for sequential analysis and ictnl(7) ordering, or 2 for parallel analysis and ictnl(29) ordering 2392 . -mat_mumps_icntl_29 - ICNTL(29): parallel ordering 1 = ptscotch, 2 = parmetis 2393 . -mat_mumps_icntl_30 - ICNTL(30): compute user-specified set of entries in inv(A) 2394 . -mat_mumps_icntl_31 - ICNTL(31): indicates which factors may be discarded during factorization 2395 . -mat_mumps_icntl_33 - ICNTL(33): compute determinant 2396 . -mat_mumps_cntl_1 - CNTL(1): relative pivoting threshold 2397 . -mat_mumps_cntl_2 - CNTL(2): stopping criterion of refinement 2398 . -mat_mumps_cntl_3 - CNTL(3): absolute pivoting threshold 2399 . -mat_mumps_cntl_4 - CNTL(4): value for static pivoting 2400 - -mat_mumps_cntl_5 - CNTL(5): fixation for null pivots 2401 2402 Level: beginner 2403 2404 Notes: When a MUMPS factorization fails inside a KSP solve, for example with a KSP_DIVERGED_PCSETUP_FAILED, one can find the MUMPS information about the failure by calling 2405 $ KSPGetPC(ksp,&pc); 2406 $ PCFactorGetMatrix(pc,&mat); 2407 $ MatMumpsGetInfo(mat,....); 2408 $ MatMumpsGetInfog(mat,....); etc. 2409 Or you can run with -ksp_error_if_not_converged and the program will be stopped and the information printed in the error message. 2410 2411 .seealso: PCFactorSetMatSolverPackage(), MatSolverPackage, MatMumpsSetICntl(), MatMumpsGetIcntl(), MatMumpsSetCntl(), MatMumpsGetCntl(), MatMumpsGetInfo(), MatMumpsGetInfog(), MatMumpsGetRinfo(), MatMumpsGetRinfog(), KSPGetPC(), PCGetFactor(), PCFactorGetMatrix() 2412 2413 M*/ 2414 2415 static PetscErrorCode MatFactorGetSolverPackage_mumps(Mat A,const MatSolverPackage *type) 2416 { 2417 PetscFunctionBegin; 2418 *type = MATSOLVERMUMPS; 2419 PetscFunctionReturn(0); 2420 } 2421 2422 /* MatGetFactor for Seq and MPI AIJ matrices */ 2423 static PetscErrorCode MatGetFactor_aij_mumps(Mat A,MatFactorType ftype,Mat *F) 2424 { 2425 Mat B; 2426 PetscErrorCode ierr; 2427 Mat_MUMPS *mumps; 2428 PetscBool isSeqAIJ; 2429 2430 PetscFunctionBegin; 2431 /* Create the factorization matrix */ 2432 ierr = PetscObjectTypeCompare((PetscObject)A,MATSEQAIJ,&isSeqAIJ);CHKERRQ(ierr); 2433 ierr = MatCreate(PetscObjectComm((PetscObject)A),&B);CHKERRQ(ierr); 2434 ierr = MatSetSizes(B,A->rmap->n,A->cmap->n,A->rmap->N,A->cmap->N);CHKERRQ(ierr); 2435 ierr = PetscStrallocpy("mumps",&((PetscObject)B)->type_name);CHKERRQ(ierr); 2436 ierr = MatSetUp(B);CHKERRQ(ierr); 2437 2438 ierr = PetscNewLog(B,&mumps);CHKERRQ(ierr); 2439 2440 B->ops->view = MatView_MUMPS; 2441 B->ops->getinfo = MatGetInfo_MUMPS; 2442 2443 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverPackage_C",MatFactorGetSolverPackage_mumps);CHKERRQ(ierr); 2444 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurIS_C",MatFactorSetSchurIS_MUMPS);CHKERRQ(ierr); 2445 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorInvertSchurComplement_C",MatFactorInvertSchurComplement_MUMPS);CHKERRQ(ierr); 2446 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorCreateSchurComplement_C",MatFactorCreateSchurComplement_MUMPS);CHKERRQ(ierr); 2447 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSchurComplement_C",MatFactorGetSchurComplement_MUMPS);CHKERRQ(ierr); 2448 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplement_C",MatFactorSolveSchurComplement_MUMPS);CHKERRQ(ierr); 2449 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplementTranspose_C",MatFactorSolveSchurComplementTranspose_MUMPS);CHKERRQ(ierr); 2450 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorFactorizeSchurComplement_C",MatFactorFactorizeSchurComplement_MUMPS);CHKERRQ(ierr); 2451 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurComplementSolverType_C",MatFactorSetSchurComplementSolverType_MUMPS);CHKERRQ(ierr); 2452 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetIcntl_C",MatMumpsSetIcntl_MUMPS);CHKERRQ(ierr); 2453 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetIcntl_C",MatMumpsGetIcntl_MUMPS);CHKERRQ(ierr); 2454 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetCntl_C",MatMumpsSetCntl_MUMPS);CHKERRQ(ierr); 2455 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetCntl_C",MatMumpsGetCntl_MUMPS);CHKERRQ(ierr); 2456 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfo_C",MatMumpsGetInfo_MUMPS);CHKERRQ(ierr); 2457 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfog_C",MatMumpsGetInfog_MUMPS);CHKERRQ(ierr); 2458 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfo_C",MatMumpsGetRinfo_MUMPS);CHKERRQ(ierr); 2459 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfog_C",MatMumpsGetRinfog_MUMPS);CHKERRQ(ierr); 2460 2461 if (ftype == MAT_FACTOR_LU) { 2462 B->ops->lufactorsymbolic = MatLUFactorSymbolic_AIJMUMPS; 2463 B->factortype = MAT_FACTOR_LU; 2464 if (isSeqAIJ) mumps->ConvertToTriples = MatConvertToTriples_seqaij_seqaij; 2465 else mumps->ConvertToTriples = MatConvertToTriples_mpiaij_mpiaij; 2466 mumps->sym = 0; 2467 } else { 2468 B->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_MUMPS; 2469 B->factortype = MAT_FACTOR_CHOLESKY; 2470 if (isSeqAIJ) mumps->ConvertToTriples = MatConvertToTriples_seqaij_seqsbaij; 2471 else mumps->ConvertToTriples = MatConvertToTriples_mpiaij_mpisbaij; 2472 #if defined(PETSC_USE_COMPLEX) 2473 mumps->sym = 2; 2474 #else 2475 if (A->spd_set && A->spd) mumps->sym = 1; 2476 else mumps->sym = 2; 2477 #endif 2478 } 2479 2480 /* set solvertype */ 2481 ierr = PetscFree(B->solvertype);CHKERRQ(ierr); 2482 ierr = PetscStrallocpy(MATSOLVERMUMPS,&B->solvertype);CHKERRQ(ierr); 2483 2484 mumps->isAIJ = PETSC_TRUE; 2485 B->ops->destroy = MatDestroy_MUMPS; 2486 B->data = (void*)mumps; 2487 2488 ierr = PetscInitializeMUMPS(A,mumps);CHKERRQ(ierr); 2489 2490 *F = B; 2491 PetscFunctionReturn(0); 2492 } 2493 2494 /* MatGetFactor for Seq and MPI SBAIJ matrices */ 2495 static PetscErrorCode MatGetFactor_sbaij_mumps(Mat A,MatFactorType ftype,Mat *F) 2496 { 2497 Mat B; 2498 PetscErrorCode ierr; 2499 Mat_MUMPS *mumps; 2500 PetscBool isSeqSBAIJ; 2501 2502 PetscFunctionBegin; 2503 if (ftype != MAT_FACTOR_CHOLESKY) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Cannot use PETSc SBAIJ matrices with MUMPS LU, use AIJ matrix"); 2504 if (A->rmap->bs > 1) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Cannot use PETSc SBAIJ matrices with block size > 1 with MUMPS Cholesky, use AIJ matrix instead"); 2505 ierr = PetscObjectTypeCompare((PetscObject)A,MATSEQSBAIJ,&isSeqSBAIJ);CHKERRQ(ierr); 2506 /* Create the factorization matrix */ 2507 ierr = MatCreate(PetscObjectComm((PetscObject)A),&B);CHKERRQ(ierr); 2508 ierr = MatSetSizes(B,A->rmap->n,A->cmap->n,A->rmap->N,A->cmap->N);CHKERRQ(ierr); 2509 ierr = PetscStrallocpy("mumps",&((PetscObject)B)->type_name);CHKERRQ(ierr); 2510 ierr = MatSetUp(B);CHKERRQ(ierr); 2511 2512 ierr = PetscNewLog(B,&mumps);CHKERRQ(ierr); 2513 if (isSeqSBAIJ) { 2514 mumps->ConvertToTriples = MatConvertToTriples_seqsbaij_seqsbaij; 2515 } else { 2516 mumps->ConvertToTriples = MatConvertToTriples_mpisbaij_mpisbaij; 2517 } 2518 2519 B->ops->getinfo = MatGetInfo_External; 2520 B->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_MUMPS; 2521 B->ops->view = MatView_MUMPS; 2522 2523 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverPackage_C",MatFactorGetSolverPackage_mumps);CHKERRQ(ierr); 2524 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurIS_C",MatFactorSetSchurIS_MUMPS);CHKERRQ(ierr); 2525 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorInvertSchurComplement_C",MatFactorInvertSchurComplement_MUMPS);CHKERRQ(ierr); 2526 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorCreateSchurComplement_C",MatFactorCreateSchurComplement_MUMPS);CHKERRQ(ierr); 2527 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSchurComplement_C",MatFactorGetSchurComplement_MUMPS);CHKERRQ(ierr); 2528 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplement_C",MatFactorSolveSchurComplement_MUMPS);CHKERRQ(ierr); 2529 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplementTranspose_C",MatFactorSolveSchurComplementTranspose_MUMPS);CHKERRQ(ierr); 2530 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorFactorizeSchurComplement_C",MatFactorFactorizeSchurComplement_MUMPS);CHKERRQ(ierr); 2531 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurComplementSolverType_C",MatFactorSetSchurComplementSolverType_MUMPS);CHKERRQ(ierr); 2532 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetIcntl_C",MatMumpsSetIcntl_MUMPS);CHKERRQ(ierr); 2533 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetIcntl_C",MatMumpsGetIcntl_MUMPS);CHKERRQ(ierr); 2534 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetCntl_C",MatMumpsSetCntl_MUMPS);CHKERRQ(ierr); 2535 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetCntl_C",MatMumpsGetCntl_MUMPS);CHKERRQ(ierr); 2536 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfo_C",MatMumpsGetInfo_MUMPS);CHKERRQ(ierr); 2537 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfog_C",MatMumpsGetInfog_MUMPS);CHKERRQ(ierr); 2538 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfo_C",MatMumpsGetRinfo_MUMPS);CHKERRQ(ierr); 2539 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfog_C",MatMumpsGetRinfog_MUMPS);CHKERRQ(ierr); 2540 2541 B->factortype = MAT_FACTOR_CHOLESKY; 2542 #if defined(PETSC_USE_COMPLEX) 2543 mumps->sym = 2; 2544 #else 2545 if (A->spd_set && A->spd) mumps->sym = 1; 2546 else mumps->sym = 2; 2547 #endif 2548 2549 /* set solvertype */ 2550 ierr = PetscFree(B->solvertype);CHKERRQ(ierr); 2551 ierr = PetscStrallocpy(MATSOLVERMUMPS,&B->solvertype);CHKERRQ(ierr); 2552 2553 mumps->isAIJ = PETSC_FALSE; 2554 B->ops->destroy = MatDestroy_MUMPS; 2555 B->data = (void*)mumps; 2556 2557 ierr = PetscInitializeMUMPS(A,mumps);CHKERRQ(ierr); 2558 2559 *F = B; 2560 PetscFunctionReturn(0); 2561 } 2562 2563 static PetscErrorCode MatGetFactor_baij_mumps(Mat A,MatFactorType ftype,Mat *F) 2564 { 2565 Mat B; 2566 PetscErrorCode ierr; 2567 Mat_MUMPS *mumps; 2568 PetscBool isSeqBAIJ; 2569 2570 PetscFunctionBegin; 2571 /* Create the factorization matrix */ 2572 ierr = PetscObjectTypeCompare((PetscObject)A,MATSEQBAIJ,&isSeqBAIJ);CHKERRQ(ierr); 2573 ierr = MatCreate(PetscObjectComm((PetscObject)A),&B);CHKERRQ(ierr); 2574 ierr = MatSetSizes(B,A->rmap->n,A->cmap->n,A->rmap->N,A->cmap->N);CHKERRQ(ierr); 2575 ierr = PetscStrallocpy("mumps",&((PetscObject)B)->type_name);CHKERRQ(ierr); 2576 ierr = MatSetUp(B);CHKERRQ(ierr); 2577 2578 ierr = PetscNewLog(B,&mumps);CHKERRQ(ierr); 2579 if (ftype == MAT_FACTOR_LU) { 2580 B->ops->lufactorsymbolic = MatLUFactorSymbolic_BAIJMUMPS; 2581 B->factortype = MAT_FACTOR_LU; 2582 if (isSeqBAIJ) mumps->ConvertToTriples = MatConvertToTriples_seqbaij_seqaij; 2583 else mumps->ConvertToTriples = MatConvertToTriples_mpibaij_mpiaij; 2584 mumps->sym = 0; 2585 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Cannot use PETSc BAIJ matrices with MUMPS Cholesky, use SBAIJ or AIJ matrix instead\n"); 2586 2587 B->ops->getinfo = MatGetInfo_External; 2588 B->ops->view = MatView_MUMPS; 2589 2590 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverPackage_C",MatFactorGetSolverPackage_mumps);CHKERRQ(ierr); 2591 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurIS_C",MatFactorSetSchurIS_MUMPS);CHKERRQ(ierr); 2592 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorInvertSchurComplement_C",MatFactorInvertSchurComplement_MUMPS);CHKERRQ(ierr); 2593 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorCreateSchurComplement_C",MatFactorCreateSchurComplement_MUMPS);CHKERRQ(ierr); 2594 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSchurComplement_C",MatFactorGetSchurComplement_MUMPS);CHKERRQ(ierr); 2595 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplement_C",MatFactorSolveSchurComplement_MUMPS);CHKERRQ(ierr); 2596 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSolveSchurComplementTranspose_C",MatFactorSolveSchurComplementTranspose_MUMPS);CHKERRQ(ierr); 2597 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorFactorizeSchurComplement_C",MatFactorFactorizeSchurComplement_MUMPS);CHKERRQ(ierr); 2598 ierr = PetscObjectComposeFunction((PetscObject)B,"MatFactorSetSchurComplementSolverType_C",MatFactorSetSchurComplementSolverType_MUMPS);CHKERRQ(ierr); 2599 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetIcntl_C",MatMumpsSetIcntl_MUMPS);CHKERRQ(ierr); 2600 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetIcntl_C",MatMumpsGetIcntl_MUMPS);CHKERRQ(ierr); 2601 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsSetCntl_C",MatMumpsSetCntl_MUMPS);CHKERRQ(ierr); 2602 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetCntl_C",MatMumpsGetCntl_MUMPS);CHKERRQ(ierr); 2603 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfo_C",MatMumpsGetInfo_MUMPS);CHKERRQ(ierr); 2604 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetInfog_C",MatMumpsGetInfog_MUMPS);CHKERRQ(ierr); 2605 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfo_C",MatMumpsGetRinfo_MUMPS);CHKERRQ(ierr); 2606 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMumpsGetRinfog_C",MatMumpsGetRinfog_MUMPS);CHKERRQ(ierr); 2607 2608 /* set solvertype */ 2609 ierr = PetscFree(B->solvertype);CHKERRQ(ierr); 2610 ierr = PetscStrallocpy(MATSOLVERMUMPS,&B->solvertype);CHKERRQ(ierr); 2611 2612 mumps->isAIJ = PETSC_TRUE; 2613 B->ops->destroy = MatDestroy_MUMPS; 2614 B->data = (void*)mumps; 2615 2616 ierr = PetscInitializeMUMPS(A,mumps);CHKERRQ(ierr); 2617 2618 *F = B; 2619 PetscFunctionReturn(0); 2620 } 2621 2622 PETSC_EXTERN PetscErrorCode MatSolverPackageRegister_MUMPS(void) 2623 { 2624 PetscErrorCode ierr; 2625 2626 PetscFunctionBegin; 2627 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATMPIAIJ,MAT_FACTOR_LU,MatGetFactor_aij_mumps);CHKERRQ(ierr); 2628 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATMPIAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_aij_mumps);CHKERRQ(ierr); 2629 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATMPIBAIJ,MAT_FACTOR_LU,MatGetFactor_baij_mumps);CHKERRQ(ierr); 2630 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATMPIBAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_baij_mumps);CHKERRQ(ierr); 2631 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATMPISBAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_sbaij_mumps);CHKERRQ(ierr); 2632 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATSEQAIJ,MAT_FACTOR_LU,MatGetFactor_aij_mumps);CHKERRQ(ierr); 2633 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATSEQAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_aij_mumps);CHKERRQ(ierr); 2634 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATSEQBAIJ,MAT_FACTOR_LU,MatGetFactor_baij_mumps);CHKERRQ(ierr); 2635 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATSEQBAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_baij_mumps);CHKERRQ(ierr); 2636 ierr = MatSolverPackageRegister(MATSOLVERMUMPS,MATSEQSBAIJ,MAT_FACTOR_CHOLESKY,MatGetFactor_sbaij_mumps);CHKERRQ(ierr); 2637 PetscFunctionReturn(0); 2638 } 2639 2640