1 2 /* 3 This is where the abstract matrix operations are defined 4 */ 5 6 #include <petsc-private/matimpl.h> /*I "petscmat.h" I*/ 7 #include <petsc-private/vecimpl.h> 8 9 /* Logging support */ 10 PetscClassId MAT_CLASSID; 11 PetscClassId MAT_FDCOLORING_CLASSID; 12 PetscClassId MAT_TRANSPOSECOLORING_CLASSID; 13 14 PetscLogEvent MAT_Mult, MAT_Mults, MAT_MultConstrained, MAT_MultAdd, MAT_MultTranspose; 15 PetscLogEvent MAT_MultTransposeConstrained, MAT_MultTransposeAdd, MAT_Solve, MAT_Solves, MAT_SolveAdd, MAT_SolveTranspose, MAT_MatSolve; 16 PetscLogEvent MAT_SolveTransposeAdd, MAT_SOR, MAT_ForwardSolve, MAT_BackwardSolve, MAT_LUFactor, MAT_LUFactorSymbolic; 17 PetscLogEvent MAT_LUFactorNumeric, MAT_CholeskyFactor, MAT_CholeskyFactorSymbolic, MAT_CholeskyFactorNumeric, MAT_ILUFactor; 18 PetscLogEvent MAT_ILUFactorSymbolic, MAT_ICCFactorSymbolic, MAT_Copy, MAT_Convert, MAT_Scale, MAT_AssemblyBegin; 19 PetscLogEvent MAT_AssemblyEnd, MAT_SetValues, MAT_GetValues, MAT_GetRow, MAT_GetRowIJ, MAT_GetSubMatrices, MAT_GetColoring, MAT_GetOrdering, MAT_GetRedundantMatrix, MAT_GetSeqNonzeroStructure; 20 PetscLogEvent MAT_IncreaseOverlap, MAT_Partitioning, MAT_Coarsen, MAT_ZeroEntries, MAT_Load, MAT_View, MAT_AXPY, MAT_FDColoringCreate; 21 PetscLogEvent MAT_FDColoringApply,MAT_Transpose,MAT_FDColoringFunction; 22 PetscLogEvent MAT_TransposeColoringCreate; 23 PetscLogEvent MAT_MatMult, MAT_MatMultSymbolic, MAT_MatMultNumeric; 24 PetscLogEvent MAT_PtAP, MAT_PtAPSymbolic, MAT_PtAPNumeric,MAT_RARt, MAT_RARtSymbolic, MAT_RARtNumeric; 25 PetscLogEvent MAT_MatTransposeMult, MAT_MatTransposeMultSymbolic, MAT_MatTransposeMultNumeric; 26 PetscLogEvent MAT_TransposeMatMult, MAT_TransposeMatMultSymbolic, MAT_TransposeMatMultNumeric; 27 PetscLogEvent MAT_MatMatMult, MAT_MatMatMultSymbolic, MAT_MatMatMultNumeric; 28 PetscLogEvent MAT_MultHermitianTranspose,MAT_MultHermitianTransposeAdd; 29 PetscLogEvent MAT_Getsymtranspose, MAT_Getsymtransreduced, MAT_Transpose_SeqAIJ, MAT_GetBrowsOfAcols; 30 PetscLogEvent MAT_GetBrowsOfAocols, MAT_Getlocalmat, MAT_Getlocalmatcondensed, MAT_Seqstompi, MAT_Seqstompinum, MAT_Seqstompisym; 31 PetscLogEvent MAT_Applypapt, MAT_Applypapt_numeric, MAT_Applypapt_symbolic, MAT_GetSequentialNonzeroStructure; 32 PetscLogEvent MAT_GetMultiProcBlock; 33 PetscLogEvent MAT_CUSPCopyToGPU, MAT_CUSPARSECopyToGPU, MAT_SetValuesBatch, MAT_SetValuesBatchI, MAT_SetValuesBatchII, MAT_SetValuesBatchIII, MAT_SetValuesBatchIV; 34 PetscLogEvent MAT_ViennaCLCopyToGPU; 35 PetscLogEvent MAT_Merge; 36 37 const char *const MatFactorTypes[] = {"NONE","LU","CHOLESKY","ILU","ICC","ILUDT","MatFactorType","MAT_FACTOR_",0}; 38 39 #undef __FUNCT__ 40 #define __FUNCT__ "MatSetRandom" 41 /*@ 42 MatSetRandom - Sets all components of a matrix to random numbers. For sparse matrices that have been preallocated it randomly selects appropriate locations 43 44 Logically Collective on Vec 45 46 Input Parameters: 47 + x - the vector 48 - rctx - the random number context, formed by PetscRandomCreate(), or NULL and 49 it will create one internally. 50 51 Output Parameter: 52 . x - the vector 53 54 Example of Usage: 55 .vb 56 PetscRandomCreate(PETSC_COMM_WORLD,&rctx); 57 VecSetRandom(x,rctx); 58 PetscRandomDestroy(rctx); 59 .ve 60 61 Level: intermediate 62 63 Concepts: vector^setting to random 64 Concepts: random^vector 65 66 .seealso: MatZeroEntries(), MatSetValues(), PetscRandomCreate(), PetscRandomDestroy() 67 @*/ 68 PetscErrorCode MatSetRandom(Mat x,PetscRandom rctx) 69 { 70 PetscErrorCode ierr; 71 PetscRandom randObj = NULL; 72 73 PetscFunctionBegin; 74 PetscValidHeaderSpecific(x,MAT_CLASSID,1); 75 if (rctx) PetscValidHeaderSpecific(rctx,PETSC_RANDOM_CLASSID,2); 76 PetscValidType(x,1); 77 78 if (!rctx) { 79 MPI_Comm comm; 80 ierr = PetscObjectGetComm((PetscObject)x,&comm);CHKERRQ(ierr); 81 ierr = PetscRandomCreate(comm,&randObj);CHKERRQ(ierr); 82 ierr = PetscRandomSetFromOptions(randObj);CHKERRQ(ierr); 83 rctx = randObj; 84 } 85 86 ierr = PetscLogEventBegin(VEC_SetRandom,x,rctx,0,0);CHKERRQ(ierr); 87 ierr = (*x->ops->setrandom)(x,rctx);CHKERRQ(ierr); 88 ierr = PetscLogEventEnd(VEC_SetRandom,x,rctx,0,0);CHKERRQ(ierr); 89 90 x->assembled = PETSC_TRUE; 91 ierr = PetscRandomDestroy(&randObj);CHKERRQ(ierr); 92 PetscFunctionReturn(0); 93 } 94 95 96 #undef __FUNCT__ 97 #define __FUNCT__ "MatFindNonzeroRows" 98 /*@ 99 MatFindNonzeroRows - Locate all rows that are not completely zero in the matrix 100 101 Input Parameter: 102 . A - the matrix 103 104 Output Parameter: 105 . keptrows - the rows that are not completely zero 106 107 Level: intermediate 108 109 @*/ 110 PetscErrorCode MatFindNonzeroRows(Mat mat,IS *keptrows) 111 { 112 PetscErrorCode ierr; 113 114 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 115 PetscValidType(mat,1); 116 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 117 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 118 if (!mat->ops->findnonzerorows) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not coded for this matrix type"); 119 ierr = (*mat->ops->findnonzerorows)(mat,keptrows);CHKERRQ(ierr); 120 PetscFunctionReturn(0); 121 } 122 123 #undef __FUNCT__ 124 #define __FUNCT__ "MatGetDiagonalBlock" 125 /*@ 126 MatGetDiagonalBlock - Returns the part of the matrix associated with the on-process coupling 127 128 Not Collective 129 130 Input Parameters: 131 . A - the matrix 132 133 Output Parameters: 134 . a - the diagonal part (which is a SEQUENTIAL matrix) 135 136 Notes: see the manual page for MatCreateAIJ() for more information on the "diagonal part" of the matrix. 137 138 Level: advanced 139 140 @*/ 141 PetscErrorCode MatGetDiagonalBlock(Mat A,Mat *a) 142 { 143 PetscErrorCode ierr,(*f)(Mat,Mat*); 144 PetscMPIInt size; 145 146 PetscFunctionBegin; 147 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 148 PetscValidType(A,1); 149 PetscValidPointer(a,3); 150 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 151 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 152 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 153 ierr = PetscObjectQueryFunction((PetscObject)A,"MatGetDiagonalBlock_C",&f);CHKERRQ(ierr); 154 if (f) { 155 ierr = (*f)(A,a);CHKERRQ(ierr); 156 PetscFunctionReturn(0); 157 } else if (size == 1) { 158 *a = A; 159 } else { 160 MatType mattype; 161 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 162 SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Matrix type %s does not support getting diagonal block",mattype); 163 } 164 PetscFunctionReturn(0); 165 } 166 167 #undef __FUNCT__ 168 #define __FUNCT__ "MatGetTrace" 169 /*@ 170 MatGetTrace - Gets the trace of a matrix. The sum of the diagonal entries. 171 172 Collective on Mat 173 174 Input Parameters: 175 . mat - the matrix 176 177 Output Parameter: 178 . trace - the sum of the diagonal entries 179 180 Level: advanced 181 182 @*/ 183 PetscErrorCode MatGetTrace(Mat mat,PetscScalar *trace) 184 { 185 PetscErrorCode ierr; 186 Vec diag; 187 188 PetscFunctionBegin; 189 ierr = MatGetVecs(mat,&diag,NULL);CHKERRQ(ierr); 190 ierr = MatGetDiagonal(mat,diag);CHKERRQ(ierr); 191 ierr = VecSum(diag,trace);CHKERRQ(ierr); 192 ierr = VecDestroy(&diag);CHKERRQ(ierr); 193 PetscFunctionReturn(0); 194 } 195 196 #undef __FUNCT__ 197 #define __FUNCT__ "MatRealPart" 198 /*@ 199 MatRealPart - Zeros out the imaginary part of the matrix 200 201 Logically Collective on Mat 202 203 Input Parameters: 204 . mat - the matrix 205 206 Level: advanced 207 208 209 .seealso: MatImaginaryPart() 210 @*/ 211 PetscErrorCode MatRealPart(Mat mat) 212 { 213 PetscErrorCode ierr; 214 215 PetscFunctionBegin; 216 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 217 PetscValidType(mat,1); 218 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 219 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 220 if (!mat->ops->realpart) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 221 MatCheckPreallocated(mat,1); 222 ierr = (*mat->ops->realpart)(mat);CHKERRQ(ierr); 223 #if defined(PETSC_HAVE_CUSP) 224 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 225 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 226 } 227 #endif 228 #if defined(PETSC_HAVE_VIENNACL) 229 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 230 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 231 } 232 #endif 233 PetscFunctionReturn(0); 234 } 235 236 #undef __FUNCT__ 237 #define __FUNCT__ "MatGetGhosts" 238 /*@C 239 MatGetGhosts - Get the global index of all ghost nodes defined by the sparse matrix 240 241 Collective on Mat 242 243 Input Parameter: 244 . mat - the matrix 245 246 Output Parameters: 247 + nghosts - number of ghosts (note for BAIJ matrices there is one ghost for each block) 248 - ghosts - the global indices of the ghost points 249 250 Notes: the nghosts and ghosts are suitable to pass into VecCreateGhost() 251 252 Level: advanced 253 254 @*/ 255 PetscErrorCode MatGetGhosts(Mat mat,PetscInt *nghosts,const PetscInt *ghosts[]) 256 { 257 PetscErrorCode ierr; 258 259 PetscFunctionBegin; 260 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 261 PetscValidType(mat,1); 262 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 263 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 264 if (!mat->ops->getghosts) { 265 if (nghosts) *nghosts = 0; 266 if (ghosts) *ghosts = 0; 267 } else { 268 ierr = (*mat->ops->getghosts)(mat,nghosts,ghosts);CHKERRQ(ierr); 269 } 270 PetscFunctionReturn(0); 271 } 272 273 274 #undef __FUNCT__ 275 #define __FUNCT__ "MatImaginaryPart" 276 /*@ 277 MatImaginaryPart - Moves the imaginary part of the matrix to the real part and zeros the imaginary part 278 279 Logically Collective on Mat 280 281 Input Parameters: 282 . mat - the matrix 283 284 Level: advanced 285 286 287 .seealso: MatRealPart() 288 @*/ 289 PetscErrorCode MatImaginaryPart(Mat mat) 290 { 291 PetscErrorCode ierr; 292 293 PetscFunctionBegin; 294 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 295 PetscValidType(mat,1); 296 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 297 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 298 if (!mat->ops->imaginarypart) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 299 MatCheckPreallocated(mat,1); 300 ierr = (*mat->ops->imaginarypart)(mat);CHKERRQ(ierr); 301 #if defined(PETSC_HAVE_CUSP) 302 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 303 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 304 } 305 #endif 306 #if defined(PETSC_HAVE_VIENNACL) 307 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 308 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 309 } 310 #endif 311 PetscFunctionReturn(0); 312 } 313 314 #undef __FUNCT__ 315 #define __FUNCT__ "MatMissingDiagonal" 316 /*@ 317 MatMissingDiagonal - Determine if sparse matrix is missing a diagonal entry (or block entry for BAIJ matrices) 318 319 Collective on Mat 320 321 Input Parameter: 322 . mat - the matrix 323 324 Output Parameters: 325 + missing - is any diagonal missing 326 - dd - first diagonal entry that is missing (optional) 327 328 Level: advanced 329 330 331 .seealso: MatRealPart() 332 @*/ 333 PetscErrorCode MatMissingDiagonal(Mat mat,PetscBool *missing,PetscInt *dd) 334 { 335 PetscErrorCode ierr; 336 337 PetscFunctionBegin; 338 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 339 PetscValidType(mat,1); 340 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 341 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 342 if (!mat->ops->missingdiagonal) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 343 ierr = (*mat->ops->missingdiagonal)(mat,missing,dd);CHKERRQ(ierr); 344 PetscFunctionReturn(0); 345 } 346 347 #undef __FUNCT__ 348 #define __FUNCT__ "MatGetRow" 349 /*@C 350 MatGetRow - Gets a row of a matrix. You MUST call MatRestoreRow() 351 for each row that you get to ensure that your application does 352 not bleed memory. 353 354 Not Collective 355 356 Input Parameters: 357 + mat - the matrix 358 - row - the row to get 359 360 Output Parameters: 361 + ncols - if not NULL, the number of nonzeros in the row 362 . cols - if not NULL, the column numbers 363 - vals - if not NULL, the values 364 365 Notes: 366 This routine is provided for people who need to have direct access 367 to the structure of a matrix. We hope that we provide enough 368 high-level matrix routines that few users will need it. 369 370 MatGetRow() always returns 0-based column indices, regardless of 371 whether the internal representation is 0-based (default) or 1-based. 372 373 For better efficiency, set cols and/or vals to NULL if you do 374 not wish to extract these quantities. 375 376 The user can only examine the values extracted with MatGetRow(); 377 the values cannot be altered. To change the matrix entries, one 378 must use MatSetValues(). 379 380 You can only have one call to MatGetRow() outstanding for a particular 381 matrix at a time, per processor. MatGetRow() can only obtain rows 382 associated with the given processor, it cannot get rows from the 383 other processors; for that we suggest using MatGetSubMatrices(), then 384 MatGetRow() on the submatrix. The row indix passed to MatGetRows() 385 is in the global number of rows. 386 387 Fortran Notes: 388 The calling sequence from Fortran is 389 .vb 390 MatGetRow(matrix,row,ncols,cols,values,ierr) 391 Mat matrix (input) 392 integer row (input) 393 integer ncols (output) 394 integer cols(maxcols) (output) 395 double precision (or double complex) values(maxcols) output 396 .ve 397 where maxcols >= maximum nonzeros in any row of the matrix. 398 399 400 Caution: 401 Do not try to change the contents of the output arrays (cols and vals). 402 In some cases, this may corrupt the matrix. 403 404 Level: advanced 405 406 Concepts: matrices^row access 407 408 .seealso: MatRestoreRow(), MatSetValues(), MatGetValues(), MatGetSubMatrices(), MatGetDiagonal() 409 @*/ 410 PetscErrorCode MatGetRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[]) 411 { 412 PetscErrorCode ierr; 413 PetscInt incols; 414 415 PetscFunctionBegin; 416 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 417 PetscValidType(mat,1); 418 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 419 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 420 if (!mat->ops->getrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 421 MatCheckPreallocated(mat,1); 422 ierr = PetscLogEventBegin(MAT_GetRow,mat,0,0,0);CHKERRQ(ierr); 423 ierr = (*mat->ops->getrow)(mat,row,&incols,(PetscInt**)cols,(PetscScalar**)vals);CHKERRQ(ierr); 424 if (ncols) *ncols = incols; 425 ierr = PetscLogEventEnd(MAT_GetRow,mat,0,0,0);CHKERRQ(ierr); 426 PetscFunctionReturn(0); 427 } 428 429 #undef __FUNCT__ 430 #define __FUNCT__ "MatConjugate" 431 /*@ 432 MatConjugate - replaces the matrix values with their complex conjugates 433 434 Logically Collective on Mat 435 436 Input Parameters: 437 . mat - the matrix 438 439 Level: advanced 440 441 .seealso: VecConjugate() 442 @*/ 443 PetscErrorCode MatConjugate(Mat mat) 444 { 445 #if defined(PETSC_USE_COMPLEX) 446 PetscErrorCode ierr; 447 448 PetscFunctionBegin; 449 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 450 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 451 if (!mat->ops->conjugate) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not provided for this matrix format, send email to petsc-maint@mcs.anl.gov"); 452 ierr = (*mat->ops->conjugate)(mat);CHKERRQ(ierr); 453 #if defined(PETSC_HAVE_CUSP) 454 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 455 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 456 } 457 #endif 458 #if defined(PETSC_HAVE_VIENNACL) 459 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 460 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 461 } 462 #endif 463 PetscFunctionReturn(0); 464 #else 465 return 0; 466 #endif 467 } 468 469 #undef __FUNCT__ 470 #define __FUNCT__ "MatRestoreRow" 471 /*@C 472 MatRestoreRow - Frees any temporary space allocated by MatGetRow(). 473 474 Not Collective 475 476 Input Parameters: 477 + mat - the matrix 478 . row - the row to get 479 . ncols, cols - the number of nonzeros and their columns 480 - vals - if nonzero the column values 481 482 Notes: 483 This routine should be called after you have finished examining the entries. 484 485 Fortran Notes: 486 The calling sequence from Fortran is 487 .vb 488 MatRestoreRow(matrix,row,ncols,cols,values,ierr) 489 Mat matrix (input) 490 integer row (input) 491 integer ncols (output) 492 integer cols(maxcols) (output) 493 double precision (or double complex) values(maxcols) output 494 .ve 495 Where maxcols >= maximum nonzeros in any row of the matrix. 496 497 In Fortran MatRestoreRow() MUST be called after MatGetRow() 498 before another call to MatGetRow() can be made. 499 500 Level: advanced 501 502 .seealso: MatGetRow() 503 @*/ 504 PetscErrorCode MatRestoreRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[]) 505 { 506 PetscErrorCode ierr; 507 508 PetscFunctionBegin; 509 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 510 if (ncols) PetscValidIntPointer(ncols,3); 511 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 512 if (!mat->ops->restorerow) PetscFunctionReturn(0); 513 ierr = (*mat->ops->restorerow)(mat,row,ncols,(PetscInt **)cols,(PetscScalar **)vals);CHKERRQ(ierr); 514 if (ncols) *ncols = 0; 515 if (cols) *cols = NULL; 516 if (vals) *vals = NULL; 517 PetscFunctionReturn(0); 518 } 519 520 #undef __FUNCT__ 521 #define __FUNCT__ "MatGetRowUpperTriangular" 522 /*@ 523 MatGetRowUpperTriangular - Sets a flag to enable calls to MatGetRow() for matrix in MATSBAIJ format. 524 You should call MatRestoreRowUpperTriangular() after calling MatGetRow/MatRestoreRow() to disable the flag. 525 526 Not Collective 527 528 Input Parameters: 529 + mat - the matrix 530 531 Notes: 532 The flag is to ensure that users are aware of MatGetRow() only provides the upper trianglular part of the row for the matrices in MATSBAIJ format. 533 534 Level: advanced 535 536 Concepts: matrices^row access 537 538 .seealso: MatRestoreRowRowUpperTriangular() 539 @*/ 540 PetscErrorCode MatGetRowUpperTriangular(Mat mat) 541 { 542 PetscErrorCode ierr; 543 544 PetscFunctionBegin; 545 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 546 PetscValidType(mat,1); 547 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 548 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 549 if (!mat->ops->getrowuppertriangular) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 550 MatCheckPreallocated(mat,1); 551 ierr = (*mat->ops->getrowuppertriangular)(mat);CHKERRQ(ierr); 552 PetscFunctionReturn(0); 553 } 554 555 #undef __FUNCT__ 556 #define __FUNCT__ "MatRestoreRowUpperTriangular" 557 /*@ 558 MatRestoreRowUpperTriangular - Disable calls to MatGetRow() for matrix in MATSBAIJ format. 559 560 Not Collective 561 562 Input Parameters: 563 + mat - the matrix 564 565 Notes: 566 This routine should be called after you have finished MatGetRow/MatRestoreRow(). 567 568 569 Level: advanced 570 571 .seealso: MatGetRowUpperTriangular() 572 @*/ 573 PetscErrorCode MatRestoreRowUpperTriangular(Mat mat) 574 { 575 PetscErrorCode ierr; 576 577 PetscFunctionBegin; 578 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 579 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 580 if (!mat->ops->restorerowuppertriangular) PetscFunctionReturn(0); 581 ierr = (*mat->ops->restorerowuppertriangular)(mat);CHKERRQ(ierr); 582 PetscFunctionReturn(0); 583 } 584 585 #undef __FUNCT__ 586 #define __FUNCT__ "MatSetOptionsPrefix" 587 /*@C 588 MatSetOptionsPrefix - Sets the prefix used for searching for all 589 Mat options in the database. 590 591 Logically Collective on Mat 592 593 Input Parameter: 594 + A - the Mat context 595 - prefix - the prefix to prepend to all option names 596 597 Notes: 598 A hyphen (-) must NOT be given at the beginning of the prefix name. 599 The first character of all runtime options is AUTOMATICALLY the hyphen. 600 601 Level: advanced 602 603 .keywords: Mat, set, options, prefix, database 604 605 .seealso: MatSetFromOptions() 606 @*/ 607 PetscErrorCode MatSetOptionsPrefix(Mat A,const char prefix[]) 608 { 609 PetscErrorCode ierr; 610 611 PetscFunctionBegin; 612 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 613 ierr = PetscObjectSetOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr); 614 PetscFunctionReturn(0); 615 } 616 617 #undef __FUNCT__ 618 #define __FUNCT__ "MatAppendOptionsPrefix" 619 /*@C 620 MatAppendOptionsPrefix - Appends to the prefix used for searching for all 621 Mat options in the database. 622 623 Logically Collective on Mat 624 625 Input Parameters: 626 + A - the Mat context 627 - prefix - the prefix to prepend to all option names 628 629 Notes: 630 A hyphen (-) must NOT be given at the beginning of the prefix name. 631 The first character of all runtime options is AUTOMATICALLY the hyphen. 632 633 Level: advanced 634 635 .keywords: Mat, append, options, prefix, database 636 637 .seealso: MatGetOptionsPrefix() 638 @*/ 639 PetscErrorCode MatAppendOptionsPrefix(Mat A,const char prefix[]) 640 { 641 PetscErrorCode ierr; 642 643 PetscFunctionBegin; 644 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 645 ierr = PetscObjectAppendOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr); 646 PetscFunctionReturn(0); 647 } 648 649 #undef __FUNCT__ 650 #define __FUNCT__ "MatGetOptionsPrefix" 651 /*@C 652 MatGetOptionsPrefix - Sets the prefix used for searching for all 653 Mat options in the database. 654 655 Not Collective 656 657 Input Parameter: 658 . A - the Mat context 659 660 Output Parameter: 661 . prefix - pointer to the prefix string used 662 663 Notes: On the fortran side, the user should pass in a string 'prefix' of 664 sufficient length to hold the prefix. 665 666 Level: advanced 667 668 .keywords: Mat, get, options, prefix, database 669 670 .seealso: MatAppendOptionsPrefix() 671 @*/ 672 PetscErrorCode MatGetOptionsPrefix(Mat A,const char *prefix[]) 673 { 674 PetscErrorCode ierr; 675 676 PetscFunctionBegin; 677 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 678 ierr = PetscObjectGetOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr); 679 PetscFunctionReturn(0); 680 } 681 682 #undef __FUNCT__ 683 #define __FUNCT__ "MatSetUp" 684 /*@ 685 MatSetUp - Sets up the internal matrix data structures for the later use. 686 687 Collective on Mat 688 689 Input Parameters: 690 . A - the Mat context 691 692 Notes: 693 If the user has not set preallocation for this matrix then a default preallocation that is likely to be inefficient is used. 694 695 If a suitable preallocation routine is used, this function does not need to be called. 696 697 See the Performance chapter of the PETSc users manual for how to preallocate matrices 698 699 Level: beginner 700 701 .keywords: Mat, setup 702 703 .seealso: MatCreate(), MatDestroy() 704 @*/ 705 PetscErrorCode MatSetUp(Mat A) 706 { 707 PetscMPIInt size; 708 PetscErrorCode ierr; 709 710 PetscFunctionBegin; 711 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 712 if (!((PetscObject)A)->type_name) { 713 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A), &size);CHKERRQ(ierr); 714 if (size == 1) { 715 ierr = MatSetType(A, MATSEQAIJ);CHKERRQ(ierr); 716 } else { 717 ierr = MatSetType(A, MATMPIAIJ);CHKERRQ(ierr); 718 } 719 } 720 if (!A->preallocated && A->ops->setup) { 721 ierr = PetscInfo(A,"Warning not preallocating matrix storage\n");CHKERRQ(ierr); 722 ierr = (*A->ops->setup)(A);CHKERRQ(ierr); 723 } 724 A->preallocated = PETSC_TRUE; 725 PetscFunctionReturn(0); 726 } 727 728 #if defined(PETSC_HAVE_SAWS) 729 #include <petscviewersaws.h> 730 #endif 731 #undef __FUNCT__ 732 #define __FUNCT__ "MatView" 733 /*@C 734 MatView - Visualizes a matrix object. 735 736 Collective on Mat 737 738 Input Parameters: 739 + mat - the matrix 740 - viewer - visualization context 741 742 Notes: 743 The available visualization contexts include 744 + PETSC_VIEWER_STDOUT_SELF - standard output (default) 745 . PETSC_VIEWER_STDOUT_WORLD - synchronized standard 746 output where only the first processor opens 747 the file. All other processors send their 748 data to the first processor to print. 749 - PETSC_VIEWER_DRAW_WORLD - graphical display of nonzero structure 750 751 The user can open alternative visualization contexts with 752 + PetscViewerASCIIOpen() - Outputs matrix to a specified file 753 . PetscViewerBinaryOpen() - Outputs matrix in binary to a 754 specified file; corresponding input uses MatLoad() 755 . PetscViewerDrawOpen() - Outputs nonzero matrix structure to 756 an X window display 757 - PetscViewerSocketOpen() - Outputs matrix to Socket viewer. 758 Currently only the sequential dense and AIJ 759 matrix types support the Socket viewer. 760 761 The user can call PetscViewerSetFormat() to specify the output 762 format of ASCII printed objects (when using PETSC_VIEWER_STDOUT_SELF, 763 PETSC_VIEWER_STDOUT_WORLD and PetscViewerASCIIOpen). Available formats include 764 + PETSC_VIEWER_DEFAULT - default, prints matrix contents 765 . PETSC_VIEWER_ASCII_MATLAB - prints matrix contents in Matlab format 766 . PETSC_VIEWER_ASCII_DENSE - prints entire matrix including zeros 767 . PETSC_VIEWER_ASCII_COMMON - prints matrix contents, using a sparse 768 format common among all matrix types 769 . PETSC_VIEWER_ASCII_IMPL - prints matrix contents, using an implementation-specific 770 format (which is in many cases the same as the default) 771 . PETSC_VIEWER_ASCII_INFO - prints basic information about the matrix 772 size and structure (not the matrix entries) 773 . PETSC_VIEWER_ASCII_INFO_DETAIL - prints more detailed information about 774 the matrix structure 775 776 Options Database Keys: 777 + -mat_view ::ascii_info - Prints info on matrix at conclusion of MatEndAssembly() 778 . -mat_view ::ascii_info_detail - Prints more detailed info 779 . -mat_view - Prints matrix in ASCII format 780 . -mat_view ::ascii_matlab - Prints matrix in Matlab format 781 . -mat_view draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX(). 782 . -display <name> - Sets display name (default is host) 783 . -draw_pause <sec> - Sets number of seconds to pause after display 784 . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (see the <a href="../../docs/manual.pdf">users manual</a> for details). 785 . -viewer_socket_machine <machine> 786 . -viewer_socket_port <port> 787 . -mat_view binary - save matrix to file in binary format 788 - -viewer_binary_filename <name> 789 Level: beginner 790 791 Notes: see the manual page for MatLoad() for the exact format of the binary file when the binary 792 viewer is used. 793 794 See bin/matlab/PetscBinaryRead.m for a Matlab code that can read in the binary file when the binary 795 viewer is used. 796 797 One can use '-mat_view draw -draw_pause -1' to pause the graphical display of matrix nonzero structure. 798 And then use the following mouse functions: 799 left mouse: zoom in 800 middle mouse: zoom out 801 right mouse: continue with the simulation 802 803 Concepts: matrices^viewing 804 Concepts: matrices^plotting 805 Concepts: matrices^printing 806 807 .seealso: PetscViewerSetFormat(), PetscViewerASCIIOpen(), PetscViewerDrawOpen(), 808 PetscViewerSocketOpen(), PetscViewerBinaryOpen(), MatLoad() 809 @*/ 810 PetscErrorCode MatView(Mat mat,PetscViewer viewer) 811 { 812 PetscErrorCode ierr; 813 PetscInt rows,cols,bs; 814 PetscBool iascii; 815 PetscViewerFormat format; 816 #if defined(PETSC_HAVE_SAWS) 817 PetscBool isams; 818 #endif 819 820 PetscFunctionBegin; 821 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 822 PetscValidType(mat,1); 823 if (!viewer) { 824 ierr = PetscViewerASCIIGetStdout(PetscObjectComm((PetscObject)mat),&viewer);CHKERRQ(ierr); 825 } 826 PetscValidHeaderSpecific(viewer,PETSC_VIEWER_CLASSID,2); 827 PetscCheckSameComm(mat,1,viewer,2); 828 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ORDER,"Must call MatAssemblyBegin/End() before viewing matrix"); 829 MatCheckPreallocated(mat,1); 830 831 ierr = PetscLogEventBegin(MAT_View,mat,viewer,0,0);CHKERRQ(ierr); 832 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 833 #if defined(PETSC_HAVE_SAWS) 834 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSAWS,&isams);CHKERRQ(ierr); 835 #endif 836 if (iascii) { 837 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 838 if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) { 839 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)mat,viewer);CHKERRQ(ierr); 840 ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr); 841 ierr = MatGetSize(mat,&rows,&cols);CHKERRQ(ierr); 842 ierr = MatGetBlockSize(mat,&bs);CHKERRQ(ierr); 843 if (bs != 1) { 844 ierr = PetscViewerASCIIPrintf(viewer,"rows=%D, cols=%D, bs=%D\n",rows,cols,bs);CHKERRQ(ierr); 845 } else { 846 ierr = PetscViewerASCIIPrintf(viewer,"rows=%D, cols=%D\n",rows,cols);CHKERRQ(ierr); 847 } 848 if (mat->factortype) { 849 const MatSolverPackage solver; 850 ierr = MatFactorGetSolverPackage(mat,&solver);CHKERRQ(ierr); 851 ierr = PetscViewerASCIIPrintf(viewer,"package used to perform factorization: %s\n",solver);CHKERRQ(ierr); 852 } 853 if (mat->ops->getinfo) { 854 MatInfo info; 855 ierr = MatGetInfo(mat,MAT_GLOBAL_SUM,&info);CHKERRQ(ierr); 856 ierr = PetscViewerASCIIPrintf(viewer,"total: nonzeros=%lld, allocated nonzeros=%lld\n",(Petsc64bitInt)info.nz_used,(Petsc64bitInt)info.nz_allocated);CHKERRQ(ierr); 857 ierr = PetscViewerASCIIPrintf(viewer,"total number of mallocs used during MatSetValues calls =%D\n",(PetscInt)info.mallocs);CHKERRQ(ierr); 858 } 859 } 860 #if defined(PETSC_HAVE_SAWS) 861 } else if (isams) { 862 PetscMPIInt rank; 863 864 ierr = PetscObjectName((PetscObject)mat);CHKERRQ(ierr); 865 ierr = MPI_Comm_rank(PETSC_COMM_WORLD,&rank);CHKERRQ(ierr); 866 if (!((PetscObject)mat)->amsmem && !rank) { 867 ierr = PetscObjectViewSAWs((PetscObject)mat,viewer);CHKERRQ(ierr); 868 } 869 #endif 870 } 871 if (mat->ops->view) { 872 ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr); 873 ierr = (*mat->ops->view)(mat,viewer);CHKERRQ(ierr); 874 ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr); 875 } else if (!iascii) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Viewer type %s not supported",((PetscObject)viewer)->type_name); 876 if (iascii) { 877 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 878 if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) { 879 ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr); 880 } 881 } 882 ierr = PetscLogEventEnd(MAT_View,mat,viewer,0,0);CHKERRQ(ierr); 883 PetscFunctionReturn(0); 884 } 885 886 #if defined(PETSC_USE_DEBUG) 887 #include <../src/sys/totalview/tv_data_display.h> 888 PETSC_UNUSED static int TV_display_type(const struct _p_Mat *mat) 889 { 890 TV_add_row("Local rows", "int", &mat->rmap->n); 891 TV_add_row("Local columns", "int", &mat->cmap->n); 892 TV_add_row("Global rows", "int", &mat->rmap->N); 893 TV_add_row("Global columns", "int", &mat->cmap->N); 894 TV_add_row("Typename", TV_ascii_string_type, ((PetscObject)mat)->type_name); 895 return TV_format_OK; 896 } 897 #endif 898 899 #undef __FUNCT__ 900 #define __FUNCT__ "MatLoad" 901 /*@C 902 MatLoad - Loads a matrix that has been stored in binary format 903 with MatView(). The matrix format is determined from the options database. 904 Generates a parallel MPI matrix if the communicator has more than one 905 processor. The default matrix type is AIJ. 906 907 Collective on PetscViewer 908 909 Input Parameters: 910 + newmat - the newly loaded matrix, this needs to have been created with MatCreate() 911 or some related function before a call to MatLoad() 912 - viewer - binary file viewer, created with PetscViewerBinaryOpen() 913 914 Options Database Keys: 915 Used with block matrix formats (MATSEQBAIJ, ...) to specify 916 block size 917 . -matload_block_size <bs> 918 919 Level: beginner 920 921 Notes: 922 If the Mat type has not yet been given then MATAIJ is used, call MatSetFromOptions() on the 923 Mat before calling this routine if you wish to set it from the options database. 924 925 MatLoad() automatically loads into the options database any options 926 given in the file filename.info where filename is the name of the file 927 that was passed to the PetscViewerBinaryOpen(). The options in the info 928 file will be ignored if you use the -viewer_binary_skip_info option. 929 930 If the type or size of newmat is not set before a call to MatLoad, PETSc 931 sets the default matrix type AIJ and sets the local and global sizes. 932 If type and/or size is already set, then the same are used. 933 934 In parallel, each processor can load a subset of rows (or the 935 entire matrix). This routine is especially useful when a large 936 matrix is stored on disk and only part of it is desired on each 937 processor. For example, a parallel solver may access only some of 938 the rows from each processor. The algorithm used here reads 939 relatively small blocks of data rather than reading the entire 940 matrix and then subsetting it. 941 942 Notes for advanced users: 943 Most users should not need to know the details of the binary storage 944 format, since MatLoad() and MatView() completely hide these details. 945 But for anyone who's interested, the standard binary matrix storage 946 format is 947 948 $ int MAT_FILE_CLASSID 949 $ int number of rows 950 $ int number of columns 951 $ int total number of nonzeros 952 $ int *number nonzeros in each row 953 $ int *column indices of all nonzeros (starting index is zero) 954 $ PetscScalar *values of all nonzeros 955 956 PETSc automatically does the byte swapping for 957 machines that store the bytes reversed, e.g. DEC alpha, freebsd, 958 linux, Windows and the paragon; thus if you write your own binary 959 read/write routines you have to swap the bytes; see PetscBinaryRead() 960 and PetscBinaryWrite() to see how this may be done. 961 962 .keywords: matrix, load, binary, input 963 964 .seealso: PetscViewerBinaryOpen(), MatView(), VecLoad() 965 966 @*/ 967 PetscErrorCode MatLoad(Mat newmat,PetscViewer viewer) 968 { 969 PetscErrorCode ierr; 970 PetscBool isbinary,flg; 971 972 PetscFunctionBegin; 973 PetscValidHeaderSpecific(newmat,MAT_CLASSID,1); 974 PetscValidHeaderSpecific(viewer,PETSC_VIEWER_CLASSID,2); 975 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);CHKERRQ(ierr); 976 if (!isbinary) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Invalid viewer; open viewer with PetscViewerBinaryOpen()"); 977 978 if (!((PetscObject)newmat)->type_name) { 979 ierr = MatSetType(newmat,MATAIJ);CHKERRQ(ierr); 980 } 981 982 if (!newmat->ops->load) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatLoad is not supported for type"); 983 ierr = PetscLogEventBegin(MAT_Load,viewer,0,0,0);CHKERRQ(ierr); 984 ierr = (*newmat->ops->load)(newmat,viewer);CHKERRQ(ierr); 985 ierr = PetscLogEventEnd(MAT_Load,viewer,0,0,0);CHKERRQ(ierr); 986 987 flg = PETSC_FALSE; 988 ierr = PetscOptionsGetBool(((PetscObject)newmat)->prefix,"-matload_symmetric",&flg,NULL);CHKERRQ(ierr); 989 if (flg) { 990 ierr = MatSetOption(newmat,MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr); 991 ierr = MatSetOption(newmat,MAT_SYMMETRY_ETERNAL,PETSC_TRUE);CHKERRQ(ierr); 992 } 993 flg = PETSC_FALSE; 994 ierr = PetscOptionsGetBool(((PetscObject)newmat)->prefix,"-matload_spd",&flg,NULL);CHKERRQ(ierr); 995 if (flg) { 996 ierr = MatSetOption(newmat,MAT_SPD,PETSC_TRUE);CHKERRQ(ierr); 997 } 998 PetscFunctionReturn(0); 999 } 1000 1001 #undef __FUNCT__ 1002 #define __FUNCT__ "MatDestroy" 1003 /*@ 1004 MatDestroy - Frees space taken by a matrix. 1005 1006 Collective on Mat 1007 1008 Input Parameter: 1009 . A - the matrix 1010 1011 Level: beginner 1012 1013 @*/ 1014 PetscErrorCode MatDestroy(Mat *A) 1015 { 1016 PetscErrorCode ierr; 1017 1018 PetscFunctionBegin; 1019 if (!*A) PetscFunctionReturn(0); 1020 PetscValidHeaderSpecific(*A,MAT_CLASSID,1); 1021 if (--((PetscObject)(*A))->refct > 0) {*A = NULL; PetscFunctionReturn(0);} 1022 1023 ierr = PetscViewerDestroy(&(*A)->viewonassembly);CHKERRQ(ierr); 1024 /* if memory was published with SAWs then destroy it */ 1025 ierr = PetscObjectSAWsViewOff((PetscObject)*A);CHKERRQ(ierr); 1026 if ((*A)->ops->destroy) { 1027 ierr = (*(*A)->ops->destroy)(*A);CHKERRQ(ierr); 1028 } 1029 ierr = MatNullSpaceDestroy(&(*A)->nullsp);CHKERRQ(ierr); 1030 ierr = MatNullSpaceDestroy(&(*A)->nearnullsp);CHKERRQ(ierr); 1031 ierr = PetscLayoutDestroy(&(*A)->rmap);CHKERRQ(ierr); 1032 ierr = PetscLayoutDestroy(&(*A)->cmap);CHKERRQ(ierr); 1033 ierr = PetscHeaderDestroy(A);CHKERRQ(ierr); 1034 PetscFunctionReturn(0); 1035 } 1036 1037 #undef __FUNCT__ 1038 #define __FUNCT__ "MatSetValues" 1039 /*@ 1040 MatSetValues - Inserts or adds a block of values into a matrix. 1041 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 1042 MUST be called after all calls to MatSetValues() have been completed. 1043 1044 Not Collective 1045 1046 Input Parameters: 1047 + mat - the matrix 1048 . v - a logically two-dimensional array of values 1049 . m, idxm - the number of rows and their global indices 1050 . n, idxn - the number of columns and their global indices 1051 - addv - either ADD_VALUES or INSERT_VALUES, where 1052 ADD_VALUES adds values to any existing entries, and 1053 INSERT_VALUES replaces existing entries with new values 1054 1055 Notes: 1056 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or 1057 MatSetUp() before using this routine 1058 1059 By default the values, v, are row-oriented. See MatSetOption() for other options. 1060 1061 Calls to MatSetValues() with the INSERT_VALUES and ADD_VALUES 1062 options cannot be mixed without intervening calls to the assembly 1063 routines. 1064 1065 MatSetValues() uses 0-based row and column numbers in Fortran 1066 as well as in C. 1067 1068 Negative indices may be passed in idxm and idxn, these rows and columns are 1069 simply ignored. This allows easily inserting element stiffness matrices 1070 with homogeneous Dirchlet boundary conditions that you don't want represented 1071 in the matrix. 1072 1073 Efficiency Alert: 1074 The routine MatSetValuesBlocked() may offer much better efficiency 1075 for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ). 1076 1077 Level: beginner 1078 1079 Concepts: matrices^putting entries in 1080 1081 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(), 1082 InsertMode, INSERT_VALUES, ADD_VALUES 1083 @*/ 1084 PetscErrorCode MatSetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv) 1085 { 1086 PetscErrorCode ierr; 1087 #if defined(PETSC_USE_DEBUG) 1088 PetscInt i,j; 1089 #endif 1090 1091 PetscFunctionBegin; 1092 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1093 PetscValidType(mat,1); 1094 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 1095 PetscValidIntPointer(idxm,3); 1096 PetscValidIntPointer(idxn,5); 1097 if (v) PetscValidScalarPointer(v,6); 1098 MatCheckPreallocated(mat,1); 1099 if (mat->insertmode == NOT_SET_VALUES) { 1100 mat->insertmode = addv; 1101 } 1102 #if defined(PETSC_USE_DEBUG) 1103 else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values"); 1104 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1105 if (!mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 1106 1107 if (v) { 1108 for (i=0; i<m; i++) { 1109 for (j=0; j<n; j++) { 1110 if (PetscIsInfOrNanScalar(v[i*n+j])) 1111 #if defined(PETSC_USE_COMPLEX) 1112 SETERRQ4(PETSC_COMM_SELF,PETSC_ERR_FP,"Inserting %G+iG at matrix entry (%D,%D)",PetscRealPart(v[i*n+j]),PetscImaginaryPart(v[i*n+j]),idxm[i],idxn[j]); 1113 #else 1114 SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_FP,"Inserting %G at matrix entry (%D,%D)",(PetscReal)v[i*n+j],idxm[i],idxn[j]); 1115 #endif 1116 } 1117 } 1118 } 1119 #endif 1120 1121 if (mat->assembled) { 1122 mat->was_assembled = PETSC_TRUE; 1123 mat->assembled = PETSC_FALSE; 1124 } 1125 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1126 ierr = (*mat->ops->setvalues)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr); 1127 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1128 #if defined(PETSC_HAVE_CUSP) 1129 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 1130 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 1131 } 1132 #endif 1133 #if defined(PETSC_HAVE_VIENNACL) 1134 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 1135 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 1136 } 1137 #endif 1138 PetscFunctionReturn(0); 1139 } 1140 1141 1142 #undef __FUNCT__ 1143 #define __FUNCT__ "MatSetValuesRowLocal" 1144 /*@ 1145 MatSetValuesRowLocal - Inserts a row (block row for BAIJ matrices) of nonzero 1146 values into a matrix 1147 1148 Not Collective 1149 1150 Input Parameters: 1151 + mat - the matrix 1152 . row - the (block) row to set 1153 - v - a logically two-dimensional array of values 1154 1155 Notes: 1156 By the values, v, are column-oriented (for the block version) and sorted 1157 1158 All the nonzeros in the row must be provided 1159 1160 The matrix must have previously had its column indices set 1161 1162 The row must belong to this process 1163 1164 Level: intermediate 1165 1166 Concepts: matrices^putting entries in 1167 1168 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(), 1169 InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues(), MatSetValuesRow(), MatSetLocalToGlobalMapping() 1170 @*/ 1171 PetscErrorCode MatSetValuesRowLocal(Mat mat,PetscInt row,const PetscScalar v[]) 1172 { 1173 PetscErrorCode ierr; 1174 PetscInt globalrow; 1175 1176 PetscFunctionBegin; 1177 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1178 PetscValidType(mat,1); 1179 PetscValidScalarPointer(v,2); 1180 ierr = ISLocalToGlobalMappingApply(mat->rmap->mapping,1,&row,&globalrow);CHKERRQ(ierr); 1181 ierr = MatSetValuesRow(mat,globalrow,v);CHKERRQ(ierr); 1182 #if defined(PETSC_HAVE_CUSP) 1183 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 1184 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 1185 } 1186 #endif 1187 #if defined(PETSC_HAVE_VIENNACL) 1188 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 1189 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 1190 } 1191 #endif 1192 PetscFunctionReturn(0); 1193 } 1194 1195 #undef __FUNCT__ 1196 #define __FUNCT__ "MatSetValuesRow" 1197 /*@ 1198 MatSetValuesRow - Inserts a row (block row for BAIJ matrices) of nonzero 1199 values into a matrix 1200 1201 Not Collective 1202 1203 Input Parameters: 1204 + mat - the matrix 1205 . row - the (block) row to set 1206 - v - a logically two-dimensional array of values 1207 1208 Notes: 1209 The values, v, are column-oriented for the block version. 1210 1211 All the nonzeros in the row must be provided 1212 1213 THE MATRIX MUSAT HAVE PREVIOUSLY HAD ITS COLUMN INDICES SET. IT IS RARE THAT THIS ROUTINE IS USED, usually MatSetValues() is used. 1214 1215 The row must belong to this process 1216 1217 Level: advanced 1218 1219 Concepts: matrices^putting entries in 1220 1221 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(), 1222 InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues() 1223 @*/ 1224 PetscErrorCode MatSetValuesRow(Mat mat,PetscInt row,const PetscScalar v[]) 1225 { 1226 PetscErrorCode ierr; 1227 1228 PetscFunctionBegin; 1229 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1230 PetscValidType(mat,1); 1231 MatCheckPreallocated(mat,1); 1232 PetscValidScalarPointer(v,2); 1233 #if defined(PETSC_USE_DEBUG) 1234 if (mat->insertmode == ADD_VALUES) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add and insert values"); 1235 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1236 #endif 1237 mat->insertmode = INSERT_VALUES; 1238 1239 if (mat->assembled) { 1240 mat->was_assembled = PETSC_TRUE; 1241 mat->assembled = PETSC_FALSE; 1242 } 1243 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1244 if (!mat->ops->setvaluesrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 1245 ierr = (*mat->ops->setvaluesrow)(mat,row,v);CHKERRQ(ierr); 1246 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1247 #if defined(PETSC_HAVE_CUSP) 1248 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 1249 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 1250 } 1251 #endif 1252 #if defined(PETSC_HAVE_VIENNACL) 1253 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 1254 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 1255 } 1256 #endif 1257 PetscFunctionReturn(0); 1258 } 1259 1260 #undef __FUNCT__ 1261 #define __FUNCT__ "MatSetValuesStencil" 1262 /*@ 1263 MatSetValuesStencil - Inserts or adds a block of values into a matrix. 1264 Using structured grid indexing 1265 1266 Not Collective 1267 1268 Input Parameters: 1269 + mat - the matrix 1270 . m - number of rows being entered 1271 . idxm - grid coordinates (and component number when dof > 1) for matrix rows being entered 1272 . n - number of columns being entered 1273 . idxn - grid coordinates (and component number when dof > 1) for matrix columns being entered 1274 . v - a logically two-dimensional array of values 1275 - addv - either ADD_VALUES or INSERT_VALUES, where 1276 ADD_VALUES adds values to any existing entries, and 1277 INSERT_VALUES replaces existing entries with new values 1278 1279 Notes: 1280 By default the values, v, are row-oriented. See MatSetOption() for other options. 1281 1282 Calls to MatSetValuesStencil() with the INSERT_VALUES and ADD_VALUES 1283 options cannot be mixed without intervening calls to the assembly 1284 routines. 1285 1286 The grid coordinates are across the entire grid, not just the local portion 1287 1288 MatSetValuesStencil() uses 0-based row and column numbers in Fortran 1289 as well as in C. 1290 1291 For setting/accessing vector values via array coordinates you can use the DMDAVecGetArray() routine 1292 1293 In order to use this routine you must either obtain the matrix with DMCreateMatrix() 1294 or call MatSetLocalToGlobalMapping() and MatSetStencil() first. 1295 1296 The columns and rows in the stencil passed in MUST be contained within the 1297 ghost region of the given process as set with DMDACreateXXX() or MatSetStencil(). For example, 1298 if you create a DMDA with an overlap of one grid level and on a particular process its first 1299 local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the 1300 first i index you can use in your column and row indices in MatSetStencil() is 5. 1301 1302 In Fortran idxm and idxn should be declared as 1303 $ MatStencil idxm(4,m),idxn(4,n) 1304 and the values inserted using 1305 $ idxm(MatStencil_i,1) = i 1306 $ idxm(MatStencil_j,1) = j 1307 $ idxm(MatStencil_k,1) = k 1308 $ idxm(MatStencil_c,1) = c 1309 etc 1310 1311 For periodic boundary conditions use negative indices for values to the left (below 0; that are to be 1312 obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one 1313 etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the 1314 DMDA_BOUNDARY_PERIODIC boundary type. 1315 1316 For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have 1317 a single value per point) you can skip filling those indices. 1318 1319 Inspired by the structured grid interface to the HYPRE package 1320 (http://www.llnl.gov/CASC/hypre) 1321 1322 Efficiency Alert: 1323 The routine MatSetValuesBlockedStencil() may offer much better efficiency 1324 for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ). 1325 1326 Level: beginner 1327 1328 Concepts: matrices^putting entries in 1329 1330 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal() 1331 MatSetValues(), MatSetValuesBlockedStencil(), MatSetStencil(), DMCreateMatrix(), DMDAVecGetArray(), MatStencil 1332 @*/ 1333 PetscErrorCode MatSetValuesStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv) 1334 { 1335 PetscErrorCode ierr; 1336 PetscInt buf[8192],*bufm=0,*bufn=0,*jdxm,*jdxn; 1337 PetscInt j,i,dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp; 1338 PetscInt *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc); 1339 1340 PetscFunctionBegin; 1341 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 1342 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1343 PetscValidType(mat,1); 1344 PetscValidIntPointer(idxm,3); 1345 PetscValidIntPointer(idxn,5); 1346 PetscValidScalarPointer(v,6); 1347 1348 if ((m+n) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 1349 jdxm = buf; jdxn = buf+m; 1350 } else { 1351 ierr = PetscMalloc2(m,PetscInt,&bufm,n,PetscInt,&bufn);CHKERRQ(ierr); 1352 jdxm = bufm; jdxn = bufn; 1353 } 1354 for (i=0; i<m; i++) { 1355 for (j=0; j<3-sdim; j++) dxm++; 1356 tmp = *dxm++ - starts[0]; 1357 for (j=0; j<dim-1; j++) { 1358 if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = -1; 1359 else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1]; 1360 } 1361 if (mat->stencil.noc) dxm++; 1362 jdxm[i] = tmp; 1363 } 1364 for (i=0; i<n; i++) { 1365 for (j=0; j<3-sdim; j++) dxn++; 1366 tmp = *dxn++ - starts[0]; 1367 for (j=0; j<dim-1; j++) { 1368 if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = -1; 1369 else tmp = tmp*dims[j] + *(dxn-1) - starts[j+1]; 1370 } 1371 if (mat->stencil.noc) dxn++; 1372 jdxn[i] = tmp; 1373 } 1374 ierr = MatSetValuesLocal(mat,m,jdxm,n,jdxn,v,addv);CHKERRQ(ierr); 1375 ierr = PetscFree2(bufm,bufn);CHKERRQ(ierr); 1376 PetscFunctionReturn(0); 1377 } 1378 1379 #undef __FUNCT__ 1380 #define __FUNCT__ "MatSetValuesBlockedStencil" 1381 /*@C 1382 MatSetValuesBlockedStencil - Inserts or adds a block of values into a matrix. 1383 Using structured grid indexing 1384 1385 Not Collective 1386 1387 Input Parameters: 1388 + mat - the matrix 1389 . m - number of rows being entered 1390 . idxm - grid coordinates for matrix rows being entered 1391 . n - number of columns being entered 1392 . idxn - grid coordinates for matrix columns being entered 1393 . v - a logically two-dimensional array of values 1394 - addv - either ADD_VALUES or INSERT_VALUES, where 1395 ADD_VALUES adds values to any existing entries, and 1396 INSERT_VALUES replaces existing entries with new values 1397 1398 Notes: 1399 By default the values, v, are row-oriented and unsorted. 1400 See MatSetOption() for other options. 1401 1402 Calls to MatSetValuesBlockedStencil() with the INSERT_VALUES and ADD_VALUES 1403 options cannot be mixed without intervening calls to the assembly 1404 routines. 1405 1406 The grid coordinates are across the entire grid, not just the local portion 1407 1408 MatSetValuesBlockedStencil() uses 0-based row and column numbers in Fortran 1409 as well as in C. 1410 1411 For setting/accessing vector values via array coordinates you can use the DMDAVecGetArray() routine 1412 1413 In order to use this routine you must either obtain the matrix with DMCreateMatrix() 1414 or call MatSetBlockSize(), MatSetLocalToGlobalMapping() and MatSetStencil() first. 1415 1416 The columns and rows in the stencil passed in MUST be contained within the 1417 ghost region of the given process as set with DMDACreateXXX() or MatSetStencil(). For example, 1418 if you create a DMDA with an overlap of one grid level and on a particular process its first 1419 local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the 1420 first i index you can use in your column and row indices in MatSetStencil() is 5. 1421 1422 In Fortran idxm and idxn should be declared as 1423 $ MatStencil idxm(4,m),idxn(4,n) 1424 and the values inserted using 1425 $ idxm(MatStencil_i,1) = i 1426 $ idxm(MatStencil_j,1) = j 1427 $ idxm(MatStencil_k,1) = k 1428 etc 1429 1430 Negative indices may be passed in idxm and idxn, these rows and columns are 1431 simply ignored. This allows easily inserting element stiffness matrices 1432 with homogeneous Dirchlet boundary conditions that you don't want represented 1433 in the matrix. 1434 1435 Inspired by the structured grid interface to the HYPRE package 1436 (http://www.llnl.gov/CASC/hypre) 1437 1438 Level: beginner 1439 1440 Concepts: matrices^putting entries in 1441 1442 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal() 1443 MatSetValues(), MatSetValuesStencil(), MatSetStencil(), DMCreateMatrix(), DMDAVecGetArray(), MatStencil, 1444 MatSetBlockSize(), MatSetLocalToGlobalMapping() 1445 @*/ 1446 PetscErrorCode MatSetValuesBlockedStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv) 1447 { 1448 PetscErrorCode ierr; 1449 PetscInt buf[8192],*bufm=0,*bufn=0,*jdxm,*jdxn; 1450 PetscInt j,i,dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp; 1451 PetscInt *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc); 1452 1453 PetscFunctionBegin; 1454 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 1455 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1456 PetscValidType(mat,1); 1457 PetscValidIntPointer(idxm,3); 1458 PetscValidIntPointer(idxn,5); 1459 PetscValidScalarPointer(v,6); 1460 1461 if ((m+n) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 1462 jdxm = buf; jdxn = buf+m; 1463 } else { 1464 ierr = PetscMalloc2(m,PetscInt,&bufm,n,PetscInt,&bufn);CHKERRQ(ierr); 1465 jdxm = bufm; jdxn = bufn; 1466 } 1467 for (i=0; i<m; i++) { 1468 for (j=0; j<3-sdim; j++) dxm++; 1469 tmp = *dxm++ - starts[0]; 1470 for (j=0; j<sdim-1; j++) { 1471 if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = -1; 1472 else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1]; 1473 } 1474 dxm++; 1475 jdxm[i] = tmp; 1476 } 1477 for (i=0; i<n; i++) { 1478 for (j=0; j<3-sdim; j++) dxn++; 1479 tmp = *dxn++ - starts[0]; 1480 for (j=0; j<sdim-1; j++) { 1481 if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = -1; 1482 else tmp = tmp*dims[j] + *(dxn-1) - starts[j+1]; 1483 } 1484 dxn++; 1485 jdxn[i] = tmp; 1486 } 1487 ierr = MatSetValuesBlockedLocal(mat,m,jdxm,n,jdxn,v,addv);CHKERRQ(ierr); 1488 ierr = PetscFree2(bufm,bufn);CHKERRQ(ierr); 1489 #if defined(PETSC_HAVE_CUSP) 1490 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 1491 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 1492 } 1493 #endif 1494 #if defined(PETSC_HAVE_VIENNACL) 1495 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 1496 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 1497 } 1498 #endif 1499 PetscFunctionReturn(0); 1500 } 1501 1502 #undef __FUNCT__ 1503 #define __FUNCT__ "MatSetStencil" 1504 /*@ 1505 MatSetStencil - Sets the grid information for setting values into a matrix via 1506 MatSetValuesStencil() 1507 1508 Not Collective 1509 1510 Input Parameters: 1511 + mat - the matrix 1512 . dim - dimension of the grid 1, 2, or 3 1513 . dims - number of grid points in x, y, and z direction, including ghost points on your processor 1514 . starts - starting point of ghost nodes on your processor in x, y, and z direction 1515 - dof - number of degrees of freedom per node 1516 1517 1518 Inspired by the structured grid interface to the HYPRE package 1519 (www.llnl.gov/CASC/hyper) 1520 1521 For matrices generated with DMCreateMatrix() this routine is automatically called and so not needed by the 1522 user. 1523 1524 Level: beginner 1525 1526 Concepts: matrices^putting entries in 1527 1528 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal() 1529 MatSetValues(), MatSetValuesBlockedStencil(), MatSetValuesStencil() 1530 @*/ 1531 PetscErrorCode MatSetStencil(Mat mat,PetscInt dim,const PetscInt dims[],const PetscInt starts[],PetscInt dof) 1532 { 1533 PetscInt i; 1534 1535 PetscFunctionBegin; 1536 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1537 PetscValidIntPointer(dims,3); 1538 PetscValidIntPointer(starts,4); 1539 1540 mat->stencil.dim = dim + (dof > 1); 1541 for (i=0; i<dim; i++) { 1542 mat->stencil.dims[i] = dims[dim-i-1]; /* copy the values in backwards */ 1543 mat->stencil.starts[i] = starts[dim-i-1]; 1544 } 1545 mat->stencil.dims[dim] = dof; 1546 mat->stencil.starts[dim] = 0; 1547 mat->stencil.noc = (PetscBool)(dof == 1); 1548 PetscFunctionReturn(0); 1549 } 1550 1551 #undef __FUNCT__ 1552 #define __FUNCT__ "MatSetValuesBlocked" 1553 /*@ 1554 MatSetValuesBlocked - Inserts or adds a block of values into a matrix. 1555 1556 Not Collective 1557 1558 Input Parameters: 1559 + mat - the matrix 1560 . v - a logically two-dimensional array of values 1561 . m, idxm - the number of block rows and their global block indices 1562 . n, idxn - the number of block columns and their global block indices 1563 - addv - either ADD_VALUES or INSERT_VALUES, where 1564 ADD_VALUES adds values to any existing entries, and 1565 INSERT_VALUES replaces existing entries with new values 1566 1567 Notes: 1568 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call 1569 MatXXXXSetPreallocation() or MatSetUp() before using this routine. 1570 1571 The m and n count the NUMBER of blocks in the row direction and column direction, 1572 NOT the total number of rows/columns; for example, if the block size is 2 and 1573 you are passing in values for rows 2,3,4,5 then m would be 2 (not 4). 1574 The values in idxm would be 1 2; that is the first index for each block divided by 1575 the block size. 1576 1577 Note that you must call MatSetBlockSize() when constructing this matrix (after 1578 preallocating it). 1579 1580 By default the values, v, are row-oriented, so the layout of 1581 v is the same as for MatSetValues(). See MatSetOption() for other options. 1582 1583 Calls to MatSetValuesBlocked() with the INSERT_VALUES and ADD_VALUES 1584 options cannot be mixed without intervening calls to the assembly 1585 routines. 1586 1587 MatSetValuesBlocked() uses 0-based row and column numbers in Fortran 1588 as well as in C. 1589 1590 Negative indices may be passed in idxm and idxn, these rows and columns are 1591 simply ignored. This allows easily inserting element stiffness matrices 1592 with homogeneous Dirchlet boundary conditions that you don't want represented 1593 in the matrix. 1594 1595 Each time an entry is set within a sparse matrix via MatSetValues(), 1596 internal searching must be done to determine where to place the the 1597 data in the matrix storage space. By instead inserting blocks of 1598 entries via MatSetValuesBlocked(), the overhead of matrix assembly is 1599 reduced. 1600 1601 Example: 1602 $ Suppose m=n=2 and block size(bs) = 2 The array is 1603 $ 1604 $ 1 2 | 3 4 1605 $ 5 6 | 7 8 1606 $ - - - | - - - 1607 $ 9 10 | 11 12 1608 $ 13 14 | 15 16 1609 $ 1610 $ v[] should be passed in like 1611 $ v[] = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16] 1612 $ 1613 $ If you are not using row oriented storage of v (that is you called MatSetOption(mat,MAT_ROW_ORIENTED,PETSC_FALSE)) then 1614 $ v[] = [1,5,9,13,2,6,10,14,3,7,11,15,4,8,12,16] 1615 1616 Level: intermediate 1617 1618 Concepts: matrices^putting entries in blocked 1619 1620 .seealso: MatSetBlockSize(), MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal() 1621 @*/ 1622 PetscErrorCode MatSetValuesBlocked(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv) 1623 { 1624 PetscErrorCode ierr; 1625 1626 PetscFunctionBegin; 1627 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1628 PetscValidType(mat,1); 1629 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 1630 PetscValidIntPointer(idxm,3); 1631 PetscValidIntPointer(idxn,5); 1632 PetscValidScalarPointer(v,6); 1633 MatCheckPreallocated(mat,1); 1634 if (mat->insertmode == NOT_SET_VALUES) { 1635 mat->insertmode = addv; 1636 } 1637 #if defined(PETSC_USE_DEBUG) 1638 else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values"); 1639 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1640 if (!mat->ops->setvaluesblocked && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 1641 #endif 1642 1643 if (mat->assembled) { 1644 mat->was_assembled = PETSC_TRUE; 1645 mat->assembled = PETSC_FALSE; 1646 } 1647 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1648 if (mat->ops->setvaluesblocked) { 1649 ierr = (*mat->ops->setvaluesblocked)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr); 1650 } else { 1651 PetscInt buf[8192],*bufr=0,*bufc=0,*iidxm,*iidxn; 1652 PetscInt i,j,bs=mat->rmap->bs; 1653 if ((m+n)*bs <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 1654 iidxm = buf; iidxn = buf + m*bs; 1655 } else { 1656 ierr = PetscMalloc2(m*bs,PetscInt,&bufr,n*bs,PetscInt,&bufc);CHKERRQ(ierr); 1657 iidxm = bufr; iidxn = bufc; 1658 } 1659 for (i=0; i<m; i++) { 1660 for (j=0; j<bs; j++) { 1661 iidxm[i*bs+j] = bs*idxm[i] + j; 1662 } 1663 } 1664 for (i=0; i<n; i++) { 1665 for (j=0; j<bs; j++) { 1666 iidxn[i*bs+j] = bs*idxn[i] + j; 1667 } 1668 } 1669 ierr = MatSetValues(mat,m*bs,iidxm,n*bs,iidxn,v,addv);CHKERRQ(ierr); 1670 ierr = PetscFree2(bufr,bufc);CHKERRQ(ierr); 1671 } 1672 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 1673 #if defined(PETSC_HAVE_CUSP) 1674 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 1675 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 1676 } 1677 #endif 1678 #if defined(PETSC_HAVE_VIENNACL) 1679 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 1680 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 1681 } 1682 #endif 1683 PetscFunctionReturn(0); 1684 } 1685 1686 #undef __FUNCT__ 1687 #define __FUNCT__ "MatGetValues" 1688 /*@ 1689 MatGetValues - Gets a block of values from a matrix. 1690 1691 Not Collective; currently only returns a local block 1692 1693 Input Parameters: 1694 + mat - the matrix 1695 . v - a logically two-dimensional array for storing the values 1696 . m, idxm - the number of rows and their global indices 1697 - n, idxn - the number of columns and their global indices 1698 1699 Notes: 1700 The user must allocate space (m*n PetscScalars) for the values, v. 1701 The values, v, are then returned in a row-oriented format, 1702 analogous to that used by default in MatSetValues(). 1703 1704 MatGetValues() uses 0-based row and column numbers in 1705 Fortran as well as in C. 1706 1707 MatGetValues() requires that the matrix has been assembled 1708 with MatAssemblyBegin()/MatAssemblyEnd(). Thus, calls to 1709 MatSetValues() and MatGetValues() CANNOT be made in succession 1710 without intermediate matrix assembly. 1711 1712 Negative row or column indices will be ignored and those locations in v[] will be 1713 left unchanged. 1714 1715 Level: advanced 1716 1717 Concepts: matrices^accessing values 1718 1719 .seealso: MatGetRow(), MatGetSubMatrices(), MatSetValues() 1720 @*/ 1721 PetscErrorCode MatGetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[]) 1722 { 1723 PetscErrorCode ierr; 1724 1725 PetscFunctionBegin; 1726 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1727 PetscValidType(mat,1); 1728 if (!m || !n) PetscFunctionReturn(0); 1729 PetscValidIntPointer(idxm,3); 1730 PetscValidIntPointer(idxn,5); 1731 PetscValidScalarPointer(v,6); 1732 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 1733 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1734 if (!mat->ops->getvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 1735 MatCheckPreallocated(mat,1); 1736 1737 ierr = PetscLogEventBegin(MAT_GetValues,mat,0,0,0);CHKERRQ(ierr); 1738 ierr = (*mat->ops->getvalues)(mat,m,idxm,n,idxn,v);CHKERRQ(ierr); 1739 ierr = PetscLogEventEnd(MAT_GetValues,mat,0,0,0);CHKERRQ(ierr); 1740 PetscFunctionReturn(0); 1741 } 1742 1743 #undef __FUNCT__ 1744 #define __FUNCT__ "MatSetValuesBatch" 1745 /*@ 1746 MatSetValuesBatch - Adds (ADD_VALUES) many blocks of values into a matrix at once. The blocks must all be square and 1747 the same size. Currently, this can only be called once and creates the given matrix. 1748 1749 Not Collective 1750 1751 Input Parameters: 1752 + mat - the matrix 1753 . nb - the number of blocks 1754 . bs - the number of rows (and columns) in each block 1755 . rows - a concatenation of the rows for each block 1756 - v - a concatenation of logically two-dimensional arrays of values 1757 1758 Notes: 1759 In the future, we will extend this routine to handle rectangular blocks, and to allow multiple calls for a given matrix. 1760 1761 Level: advanced 1762 1763 Concepts: matrices^putting entries in 1764 1765 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(), 1766 InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues() 1767 @*/ 1768 PetscErrorCode MatSetValuesBatch(Mat mat, PetscInt nb, PetscInt bs, PetscInt rows[], const PetscScalar v[]) 1769 { 1770 PetscErrorCode ierr; 1771 1772 PetscFunctionBegin; 1773 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1774 PetscValidType(mat,1); 1775 PetscValidScalarPointer(rows,4); 1776 PetscValidScalarPointer(v,5); 1777 #if defined(PETSC_USE_DEBUG) 1778 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1779 #endif 1780 1781 ierr = PetscLogEventBegin(MAT_SetValuesBatch,mat,0,0,0);CHKERRQ(ierr); 1782 if (mat->ops->setvaluesbatch) { 1783 ierr = (*mat->ops->setvaluesbatch)(mat,nb,bs,rows,v);CHKERRQ(ierr); 1784 } else { 1785 PetscInt b; 1786 for (b = 0; b < nb; ++b) { 1787 ierr = MatSetValues(mat, bs, &rows[b*bs], bs, &rows[b*bs], &v[b*bs*bs], ADD_VALUES);CHKERRQ(ierr); 1788 } 1789 } 1790 ierr = PetscLogEventEnd(MAT_SetValuesBatch,mat,0,0,0);CHKERRQ(ierr); 1791 PetscFunctionReturn(0); 1792 } 1793 1794 #undef __FUNCT__ 1795 #define __FUNCT__ "MatSetLocalToGlobalMapping" 1796 /*@ 1797 MatSetLocalToGlobalMapping - Sets a local-to-global numbering for use by 1798 the routine MatSetValuesLocal() to allow users to insert matrix entries 1799 using a local (per-processor) numbering. 1800 1801 Not Collective 1802 1803 Input Parameters: 1804 + x - the matrix 1805 . rmapping - row mapping created with ISLocalToGlobalMappingCreate() 1806 or ISLocalToGlobalMappingCreateIS() 1807 - cmapping - column mapping 1808 1809 Level: intermediate 1810 1811 Concepts: matrices^local to global mapping 1812 Concepts: local to global mapping^for matrices 1813 1814 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesLocal() 1815 @*/ 1816 PetscErrorCode MatSetLocalToGlobalMapping(Mat x,ISLocalToGlobalMapping rmapping,ISLocalToGlobalMapping cmapping) 1817 { 1818 PetscErrorCode ierr; 1819 1820 PetscFunctionBegin; 1821 PetscValidHeaderSpecific(x,MAT_CLASSID,1); 1822 PetscValidType(x,1); 1823 PetscValidHeaderSpecific(rmapping,IS_LTOGM_CLASSID,2); 1824 PetscValidHeaderSpecific(cmapping,IS_LTOGM_CLASSID,3); 1825 1826 if (x->ops->setlocaltoglobalmapping) { 1827 ierr = (*x->ops->setlocaltoglobalmapping)(x,rmapping,cmapping);CHKERRQ(ierr); 1828 } else { 1829 ierr = PetscLayoutSetISLocalToGlobalMapping(x->rmap,rmapping);CHKERRQ(ierr); 1830 ierr = PetscLayoutSetISLocalToGlobalMapping(x->cmap,cmapping);CHKERRQ(ierr); 1831 } 1832 PetscFunctionReturn(0); 1833 } 1834 1835 #undef __FUNCT__ 1836 #define __FUNCT__ "MatSetLocalToGlobalMappingBlock" 1837 /*@ 1838 MatSetLocalToGlobalMappingBlock - Sets a local-to-global numbering for use 1839 by the routine MatSetValuesBlockedLocal() to allow users to insert matrix 1840 entries using a local (per-processor) numbering. 1841 1842 Not Collective 1843 1844 Input Parameters: 1845 + x - the matrix 1846 . rmapping - row mapping created with ISLocalToGlobalMappingCreate() or 1847 ISLocalToGlobalMappingCreateIS() 1848 - cmapping - column mapping 1849 1850 Level: intermediate 1851 1852 Concepts: matrices^local to global mapping blocked 1853 Concepts: local to global mapping^for matrices, blocked 1854 1855 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal(), 1856 MatSetValuesBlocked(), MatSetValuesLocal() 1857 @*/ 1858 PetscErrorCode MatSetLocalToGlobalMappingBlock(Mat x,ISLocalToGlobalMapping rmapping,ISLocalToGlobalMapping cmapping) 1859 { 1860 PetscErrorCode ierr; 1861 1862 PetscFunctionBegin; 1863 PetscValidHeaderSpecific(x,MAT_CLASSID,1); 1864 PetscValidType(x,1); 1865 PetscValidHeaderSpecific(rmapping,IS_LTOGM_CLASSID,2); 1866 PetscValidHeaderSpecific(cmapping,IS_LTOGM_CLASSID,3); 1867 1868 ierr = PetscLayoutSetISLocalToGlobalMappingBlock(x->rmap,rmapping);CHKERRQ(ierr); 1869 ierr = PetscLayoutSetISLocalToGlobalMappingBlock(x->cmap,cmapping);CHKERRQ(ierr); 1870 PetscFunctionReturn(0); 1871 } 1872 1873 #undef __FUNCT__ 1874 #define __FUNCT__ "MatGetLocalToGlobalMapping" 1875 /*@ 1876 MatGetLocalToGlobalMapping - Gets the local-to-global numbering set by MatSetLocalToGlobalMapping() 1877 1878 Not Collective 1879 1880 Input Parameters: 1881 . A - the matrix 1882 1883 Output Parameters: 1884 + rmapping - row mapping 1885 - cmapping - column mapping 1886 1887 Level: advanced 1888 1889 Concepts: matrices^local to global mapping 1890 Concepts: local to global mapping^for matrices 1891 1892 .seealso: MatSetValuesLocal(), MatGetLocalToGlobalMappingBlock() 1893 @*/ 1894 PetscErrorCode MatGetLocalToGlobalMapping(Mat A,ISLocalToGlobalMapping *rmapping,ISLocalToGlobalMapping *cmapping) 1895 { 1896 PetscFunctionBegin; 1897 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 1898 PetscValidType(A,1); 1899 if (rmapping) PetscValidPointer(rmapping,2); 1900 if (cmapping) PetscValidPointer(cmapping,3); 1901 if (rmapping) *rmapping = A->rmap->mapping; 1902 if (cmapping) *cmapping = A->cmap->mapping; 1903 PetscFunctionReturn(0); 1904 } 1905 1906 #undef __FUNCT__ 1907 #define __FUNCT__ "MatGetLocalToGlobalMappingBlock" 1908 /*@ 1909 MatGetLocalToGlobalMappingBlock - Gets the local-to-global numbering set by MatSetLocalToGlobalMappingBlock() 1910 1911 Not Collective 1912 1913 Input Parameters: 1914 . A - the matrix 1915 1916 Output Parameters: 1917 + rmapping - row mapping 1918 - cmapping - column mapping 1919 1920 Level: advanced 1921 1922 Concepts: matrices^local to global mapping blocked 1923 Concepts: local to global mapping^for matrices, blocked 1924 1925 .seealso: MatSetValuesBlockedLocal(), MatGetLocalToGlobalMapping() 1926 @*/ 1927 PetscErrorCode MatGetLocalToGlobalMappingBlock(Mat A,ISLocalToGlobalMapping *rmapping,ISLocalToGlobalMapping *cmapping) 1928 { 1929 PetscFunctionBegin; 1930 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 1931 PetscValidType(A,1); 1932 if (rmapping) PetscValidPointer(rmapping,2); 1933 if (cmapping) PetscValidPointer(cmapping,3); 1934 if (rmapping) *rmapping = A->rmap->bmapping; 1935 if (cmapping) *cmapping = A->cmap->bmapping; 1936 PetscFunctionReturn(0); 1937 } 1938 1939 #undef __FUNCT__ 1940 #define __FUNCT__ "MatSetValuesLocal" 1941 /*@ 1942 MatSetValuesLocal - Inserts or adds values into certain locations of a matrix, 1943 using a local ordering of the nodes. 1944 1945 Not Collective 1946 1947 Input Parameters: 1948 + x - the matrix 1949 . nrow, irow - number of rows and their local indices 1950 . ncol, icol - number of columns and their local indices 1951 . y - a logically two-dimensional array of values 1952 - addv - either INSERT_VALUES or ADD_VALUES, where 1953 ADD_VALUES adds values to any existing entries, and 1954 INSERT_VALUES replaces existing entries with new values 1955 1956 Notes: 1957 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or 1958 MatSetUp() before using this routine 1959 1960 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatSetLocalToGlobalMapping() before using this routine 1961 1962 Calls to MatSetValuesLocal() with the INSERT_VALUES and ADD_VALUES 1963 options cannot be mixed without intervening calls to the assembly 1964 routines. 1965 1966 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 1967 MUST be called after all calls to MatSetValuesLocal() have been completed. 1968 1969 Level: intermediate 1970 1971 Concepts: matrices^putting entries in with local numbering 1972 1973 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetLocalToGlobalMapping(), 1974 MatSetValueLocal() 1975 @*/ 1976 PetscErrorCode MatSetValuesLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv) 1977 { 1978 PetscErrorCode ierr; 1979 1980 PetscFunctionBegin; 1981 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 1982 PetscValidType(mat,1); 1983 MatCheckPreallocated(mat,1); 1984 if (!nrow || !ncol) PetscFunctionReturn(0); /* no values to insert */ 1985 PetscValidIntPointer(irow,3); 1986 PetscValidIntPointer(icol,5); 1987 PetscValidScalarPointer(y,6); 1988 if (mat->insertmode == NOT_SET_VALUES) { 1989 mat->insertmode = addv; 1990 } 1991 #if defined(PETSC_USE_DEBUG) 1992 else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values"); 1993 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 1994 if (!mat->ops->setvalueslocal && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 1995 #endif 1996 1997 if (mat->assembled) { 1998 mat->was_assembled = PETSC_TRUE; 1999 mat->assembled = PETSC_FALSE; 2000 } 2001 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 2002 if (mat->ops->setvalueslocal) { 2003 ierr = (*mat->ops->setvalueslocal)(mat,nrow,irow,ncol,icol,y,addv);CHKERRQ(ierr); 2004 } else { 2005 PetscInt buf[8192],*bufr=0,*bufc=0,*irowm,*icolm; 2006 if ((nrow+ncol) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 2007 irowm = buf; icolm = buf+nrow; 2008 } else { 2009 ierr = PetscMalloc2(nrow,PetscInt,&bufr,ncol,PetscInt,&bufc);CHKERRQ(ierr); 2010 irowm = bufr; icolm = bufc; 2011 } 2012 ierr = ISLocalToGlobalMappingApply(mat->rmap->mapping,nrow,irow,irowm);CHKERRQ(ierr); 2013 ierr = ISLocalToGlobalMappingApply(mat->cmap->mapping,ncol,icol,icolm);CHKERRQ(ierr); 2014 ierr = MatSetValues(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr); 2015 ierr = PetscFree2(bufr,bufc);CHKERRQ(ierr); 2016 } 2017 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 2018 #if defined(PETSC_HAVE_CUSP) 2019 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 2020 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 2021 } 2022 #endif 2023 #if defined(PETSC_HAVE_VIENNACL) 2024 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 2025 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 2026 } 2027 #endif 2028 PetscFunctionReturn(0); 2029 } 2030 2031 #undef __FUNCT__ 2032 #define __FUNCT__ "MatSetValuesBlockedLocal" 2033 /*@ 2034 MatSetValuesBlockedLocal - Inserts or adds values into certain locations of a matrix, 2035 using a local ordering of the nodes a block at a time. 2036 2037 Not Collective 2038 2039 Input Parameters: 2040 + x - the matrix 2041 . nrow, irow - number of rows and their local indices 2042 . ncol, icol - number of columns and their local indices 2043 . y - a logically two-dimensional array of values 2044 - addv - either INSERT_VALUES or ADD_VALUES, where 2045 ADD_VALUES adds values to any existing entries, and 2046 INSERT_VALUES replaces existing entries with new values 2047 2048 Notes: 2049 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatXXXXSetPreallocation() or 2050 MatSetUp() before using this routine 2051 2052 If you create the matrix yourself (that is not with a call to DMCreateMatrix()) then you MUST call MatSetBlockSize() and MatSetLocalToGlobalMappingBlock() 2053 before using this routineBefore calling MatSetValuesLocal(), the user must first set the 2054 2055 Calls to MatSetValuesBlockedLocal() with the INSERT_VALUES and ADD_VALUES 2056 options cannot be mixed without intervening calls to the assembly 2057 routines. 2058 2059 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 2060 MUST be called after all calls to MatSetValuesBlockedLocal() have been completed. 2061 2062 Level: intermediate 2063 2064 Concepts: matrices^putting blocked values in with local numbering 2065 2066 .seealso: MatSetBlockSize(), MatSetLocalToGlobalMappingBlock(), MatAssemblyBegin(), MatAssemblyEnd(), 2067 MatSetValuesLocal(), MatSetLocalToGlobalMappingBlock(), MatSetValuesBlocked() 2068 @*/ 2069 PetscErrorCode MatSetValuesBlockedLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv) 2070 { 2071 PetscErrorCode ierr; 2072 2073 PetscFunctionBegin; 2074 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2075 PetscValidType(mat,1); 2076 MatCheckPreallocated(mat,1); 2077 if (!nrow || !ncol) PetscFunctionReturn(0); /* no values to insert */ 2078 PetscValidIntPointer(irow,3); 2079 PetscValidIntPointer(icol,5); 2080 PetscValidScalarPointer(y,6); 2081 if (mat->insertmode == NOT_SET_VALUES) { 2082 mat->insertmode = addv; 2083 } 2084 #if defined(PETSC_USE_DEBUG) 2085 else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values"); 2086 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2087 if (!mat->ops->setvaluesblockedlocal && !mat->ops->setvaluesblocked && !mat->ops->setvalueslocal && !mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2088 #endif 2089 2090 if (mat->assembled) { 2091 mat->was_assembled = PETSC_TRUE; 2092 mat->assembled = PETSC_FALSE; 2093 } 2094 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 2095 if (mat->ops->setvaluesblockedlocal) { 2096 ierr = (*mat->ops->setvaluesblockedlocal)(mat,nrow,irow,ncol,icol,y,addv);CHKERRQ(ierr); 2097 } else { 2098 PetscInt buf[8192],*bufr=0,*bufc=0,*irowm,*icolm; 2099 if (mat->rmap->bmapping && mat->cmap->bmapping) { 2100 if ((nrow+ncol) <= (PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 2101 irowm = buf; icolm = buf + nrow; 2102 } else { 2103 ierr = PetscMalloc2(nrow,PetscInt,&bufr,ncol,PetscInt,&bufc);CHKERRQ(ierr); 2104 irowm = bufr; icolm = bufc; 2105 } 2106 ierr = ISLocalToGlobalMappingApply(mat->rmap->bmapping,nrow,irow,irowm);CHKERRQ(ierr); 2107 ierr = ISLocalToGlobalMappingApply(mat->cmap->bmapping,ncol,icol,icolm);CHKERRQ(ierr); 2108 ierr = MatSetValuesBlocked(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr); 2109 ierr = PetscFree2(bufr,bufc);CHKERRQ(ierr); 2110 } else { 2111 PetscInt i,j,bs=mat->rmap->bs; 2112 if ((nrow+ncol)*bs <=(PetscInt)(sizeof(buf)/sizeof(PetscInt))) { 2113 irowm = buf; icolm = buf + nrow; 2114 } else { 2115 ierr = PetscMalloc2(nrow*bs,PetscInt,&bufr,ncol*bs,PetscInt,&bufc);CHKERRQ(ierr); 2116 irowm = bufr; icolm = bufc; 2117 } 2118 for (i=0; i<nrow; i++) { 2119 for (j=0; j<bs; j++) irowm[i*bs+j] = irow[i]*bs+j; 2120 } 2121 for (i=0; i<ncol; i++) { 2122 for (j=0; j<bs; j++) icolm[i*bs+j] = icol[i]*bs+j; 2123 } 2124 ierr = MatSetValuesLocal(mat,nrow*bs,irowm,ncol*bs,icolm,y,addv);CHKERRQ(ierr); 2125 ierr = PetscFree2(bufr,bufc);CHKERRQ(ierr); 2126 } 2127 } 2128 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 2129 #if defined(PETSC_HAVE_CUSP) 2130 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 2131 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 2132 } 2133 #endif 2134 #if defined(PETSC_HAVE_VIENNACL) 2135 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 2136 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 2137 } 2138 #endif 2139 PetscFunctionReturn(0); 2140 } 2141 2142 #undef __FUNCT__ 2143 #define __FUNCT__ "MatMultDiagonalBlock" 2144 /*@ 2145 MatMultDiagonalBlock - Computes the matrix-vector product, y = Dx. Where D is defined by the inode or block structure of the diagonal 2146 2147 Collective on Mat and Vec 2148 2149 Input Parameters: 2150 + mat - the matrix 2151 - x - the vector to be multiplied 2152 2153 Output Parameters: 2154 . y - the result 2155 2156 Notes: 2157 The vectors x and y cannot be the same. I.e., one cannot 2158 call MatMult(A,y,y). 2159 2160 Level: developer 2161 2162 Concepts: matrix-vector product 2163 2164 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 2165 @*/ 2166 PetscErrorCode MatMultDiagonalBlock(Mat mat,Vec x,Vec y) 2167 { 2168 PetscErrorCode ierr; 2169 2170 PetscFunctionBegin; 2171 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2172 PetscValidType(mat,1); 2173 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2174 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2175 2176 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2177 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2178 if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2179 MatCheckPreallocated(mat,1); 2180 2181 if (!mat->ops->multdiagonalblock) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"This matrix type does not have a multiply defined"); 2182 ierr = (*mat->ops->multdiagonalblock)(mat,x,y);CHKERRQ(ierr); 2183 ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr); 2184 PetscFunctionReturn(0); 2185 } 2186 2187 /* --------------------------------------------------------*/ 2188 #undef __FUNCT__ 2189 #define __FUNCT__ "MatMult" 2190 /*@ 2191 MatMult - Computes the matrix-vector product, y = Ax. 2192 2193 Neighbor-wise Collective on Mat and Vec 2194 2195 Input Parameters: 2196 + mat - the matrix 2197 - x - the vector to be multiplied 2198 2199 Output Parameters: 2200 . y - the result 2201 2202 Notes: 2203 The vectors x and y cannot be the same. I.e., one cannot 2204 call MatMult(A,y,y). 2205 2206 Level: beginner 2207 2208 Concepts: matrix-vector product 2209 2210 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 2211 @*/ 2212 PetscErrorCode MatMult(Mat mat,Vec x,Vec y) 2213 { 2214 PetscErrorCode ierr; 2215 2216 PetscFunctionBegin; 2217 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2218 PetscValidType(mat,1); 2219 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2220 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2221 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2222 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2223 if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2224 #if !defined(PETSC_HAVE_CONSTRAINTS) 2225 if (mat->cmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 2226 if (mat->rmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N); 2227 if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n); 2228 #endif 2229 ierr = VecValidValues(x,2,PETSC_TRUE);CHKERRQ(ierr); 2230 MatCheckPreallocated(mat,1); 2231 2232 if (!mat->ops->mult) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"This matrix type does not have a multiply defined"); 2233 ierr = PetscLogEventBegin(MAT_Mult,mat,x,y,0);CHKERRQ(ierr); 2234 ierr = (*mat->ops->mult)(mat,x,y);CHKERRQ(ierr); 2235 ierr = PetscLogEventEnd(MAT_Mult,mat,x,y,0);CHKERRQ(ierr); 2236 ierr = VecValidValues(y,3,PETSC_FALSE);CHKERRQ(ierr); 2237 PetscFunctionReturn(0); 2238 } 2239 2240 #undef __FUNCT__ 2241 #define __FUNCT__ "MatMultTranspose" 2242 /*@ 2243 MatMultTranspose - Computes matrix transpose times a vector. 2244 2245 Neighbor-wise Collective on Mat and Vec 2246 2247 Input Parameters: 2248 + mat - the matrix 2249 - x - the vector to be multilplied 2250 2251 Output Parameters: 2252 . y - the result 2253 2254 Notes: 2255 The vectors x and y cannot be the same. I.e., one cannot 2256 call MatMultTranspose(A,y,y). 2257 2258 For complex numbers this does NOT compute the Hermitian (complex conjugate) transpose multiple, 2259 use MatMultHermitianTranspose() 2260 2261 Level: beginner 2262 2263 Concepts: matrix vector product^transpose 2264 2265 .seealso: MatMult(), MatMultAdd(), MatMultTransposeAdd(), MatMultHermitianTranspose(), MatTranspose() 2266 @*/ 2267 PetscErrorCode MatMultTranspose(Mat mat,Vec x,Vec y) 2268 { 2269 PetscErrorCode ierr; 2270 2271 PetscFunctionBegin; 2272 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2273 PetscValidType(mat,1); 2274 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2275 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2276 2277 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2278 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2279 if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2280 #if !defined(PETSC_HAVE_CONSTRAINTS) 2281 if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N); 2282 if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N); 2283 #endif 2284 ierr = VecValidValues(x,2,PETSC_TRUE);CHKERRQ(ierr); 2285 MatCheckPreallocated(mat,1); 2286 2287 if (!mat->ops->multtranspose) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"This matrix type does not have a multiply tranpose defined"); 2288 ierr = PetscLogEventBegin(MAT_MultTranspose,mat,x,y,0);CHKERRQ(ierr); 2289 ierr = (*mat->ops->multtranspose)(mat,x,y);CHKERRQ(ierr); 2290 ierr = PetscLogEventEnd(MAT_MultTranspose,mat,x,y,0);CHKERRQ(ierr); 2291 ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr); 2292 ierr = VecValidValues(y,3,PETSC_FALSE);CHKERRQ(ierr); 2293 PetscFunctionReturn(0); 2294 } 2295 2296 #undef __FUNCT__ 2297 #define __FUNCT__ "MatMultHermitianTranspose" 2298 /*@ 2299 MatMultHermitianTranspose - Computes matrix Hermitian transpose times a vector. 2300 2301 Neighbor-wise Collective on Mat and Vec 2302 2303 Input Parameters: 2304 + mat - the matrix 2305 - x - the vector to be multilplied 2306 2307 Output Parameters: 2308 . y - the result 2309 2310 Notes: 2311 The vectors x and y cannot be the same. I.e., one cannot 2312 call MatMultHermitianTranspose(A,y,y). 2313 2314 Also called the conjugate transpose, complex conjugate transpose, or adjoint. 2315 2316 For real numbers MatMultTranspose() and MatMultHermitianTranspose() are identical. 2317 2318 Level: beginner 2319 2320 Concepts: matrix vector product^transpose 2321 2322 .seealso: MatMult(), MatMultAdd(), MatMultHermitianTransposeAdd(), MatMultTranspose() 2323 @*/ 2324 PetscErrorCode MatMultHermitianTranspose(Mat mat,Vec x,Vec y) 2325 { 2326 PetscErrorCode ierr; 2327 Vec w; 2328 2329 PetscFunctionBegin; 2330 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2331 PetscValidType(mat,1); 2332 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2333 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2334 2335 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2336 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2337 if (x == y) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2338 #if !defined(PETSC_HAVE_CONSTRAINTS) 2339 if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N); 2340 if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N); 2341 #endif 2342 MatCheckPreallocated(mat,1); 2343 2344 ierr = PetscLogEventBegin(MAT_MultHermitianTranspose,mat,x,y,0);CHKERRQ(ierr); 2345 if (mat->ops->multhermitiantranspose) { 2346 ierr = (*mat->ops->multhermitiantranspose)(mat,x,y);CHKERRQ(ierr); 2347 } else { 2348 ierr = VecDuplicate(x,&w);CHKERRQ(ierr); 2349 ierr = VecCopy(x,w);CHKERRQ(ierr); 2350 ierr = VecConjugate(w);CHKERRQ(ierr); 2351 ierr = MatMultTranspose(mat,w,y);CHKERRQ(ierr); 2352 ierr = VecDestroy(&w);CHKERRQ(ierr); 2353 ierr = VecConjugate(y);CHKERRQ(ierr); 2354 } 2355 ierr = PetscLogEventEnd(MAT_MultHermitianTranspose,mat,x,y,0);CHKERRQ(ierr); 2356 ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr); 2357 PetscFunctionReturn(0); 2358 } 2359 2360 #undef __FUNCT__ 2361 #define __FUNCT__ "MatMultAdd" 2362 /*@ 2363 MatMultAdd - Computes v3 = v2 + A * v1. 2364 2365 Neighbor-wise Collective on Mat and Vec 2366 2367 Input Parameters: 2368 + mat - the matrix 2369 - v1, v2 - the vectors 2370 2371 Output Parameters: 2372 . v3 - the result 2373 2374 Notes: 2375 The vectors v1 and v3 cannot be the same. I.e., one cannot 2376 call MatMultAdd(A,v1,v2,v1). 2377 2378 Level: beginner 2379 2380 Concepts: matrix vector product^addition 2381 2382 .seealso: MatMultTranspose(), MatMult(), MatMultTransposeAdd() 2383 @*/ 2384 PetscErrorCode MatMultAdd(Mat mat,Vec v1,Vec v2,Vec v3) 2385 { 2386 PetscErrorCode ierr; 2387 2388 PetscFunctionBegin; 2389 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2390 PetscValidType(mat,1); 2391 PetscValidHeaderSpecific(v1,VEC_CLASSID,2); 2392 PetscValidHeaderSpecific(v2,VEC_CLASSID,3); 2393 PetscValidHeaderSpecific(v3,VEC_CLASSID,4); 2394 2395 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2396 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2397 if (mat->cmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->cmap->N,v1->map->N); 2398 /* if (mat->rmap->N != v2->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->rmap->N,v2->map->N); 2399 if (mat->rmap->N != v3->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->rmap->N,v3->map->N); */ 2400 if (mat->rmap->n != v3->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: local dim %D %D",mat->rmap->n,v3->map->n); 2401 if (mat->rmap->n != v2->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: local dim %D %D",mat->rmap->n,v2->map->n); 2402 if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors"); 2403 MatCheckPreallocated(mat,1); 2404 2405 if (!mat->ops->multadd) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"No MatMultAdd() for matrix type '%s'",((PetscObject)mat)->type_name); 2406 ierr = PetscLogEventBegin(MAT_MultAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2407 ierr = (*mat->ops->multadd)(mat,v1,v2,v3);CHKERRQ(ierr); 2408 ierr = PetscLogEventEnd(MAT_MultAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2409 ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr); 2410 PetscFunctionReturn(0); 2411 } 2412 2413 #undef __FUNCT__ 2414 #define __FUNCT__ "MatMultTransposeAdd" 2415 /*@ 2416 MatMultTransposeAdd - Computes v3 = v2 + A' * v1. 2417 2418 Neighbor-wise Collective on Mat and Vec 2419 2420 Input Parameters: 2421 + mat - the matrix 2422 - v1, v2 - the vectors 2423 2424 Output Parameters: 2425 . v3 - the result 2426 2427 Notes: 2428 The vectors v1 and v3 cannot be the same. I.e., one cannot 2429 call MatMultTransposeAdd(A,v1,v2,v1). 2430 2431 Level: beginner 2432 2433 Concepts: matrix vector product^transpose and addition 2434 2435 .seealso: MatMultTranspose(), MatMultAdd(), MatMult() 2436 @*/ 2437 PetscErrorCode MatMultTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3) 2438 { 2439 PetscErrorCode ierr; 2440 2441 PetscFunctionBegin; 2442 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2443 PetscValidType(mat,1); 2444 PetscValidHeaderSpecific(v1,VEC_CLASSID,2); 2445 PetscValidHeaderSpecific(v2,VEC_CLASSID,3); 2446 PetscValidHeaderSpecific(v3,VEC_CLASSID,4); 2447 2448 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2449 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2450 if (!mat->ops->multtransposeadd) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2451 if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors"); 2452 if (mat->rmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N); 2453 if (mat->cmap->N != v2->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N); 2454 if (mat->cmap->N != v3->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N); 2455 MatCheckPreallocated(mat,1); 2456 2457 ierr = PetscLogEventBegin(MAT_MultTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2458 ierr = (*mat->ops->multtransposeadd)(mat,v1,v2,v3);CHKERRQ(ierr); 2459 ierr = PetscLogEventEnd(MAT_MultTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2460 ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr); 2461 PetscFunctionReturn(0); 2462 } 2463 2464 #undef __FUNCT__ 2465 #define __FUNCT__ "MatMultHermitianTransposeAdd" 2466 /*@ 2467 MatMultHermitianTransposeAdd - Computes v3 = v2 + A^H * v1. 2468 2469 Neighbor-wise Collective on Mat and Vec 2470 2471 Input Parameters: 2472 + mat - the matrix 2473 - v1, v2 - the vectors 2474 2475 Output Parameters: 2476 . v3 - the result 2477 2478 Notes: 2479 The vectors v1 and v3 cannot be the same. I.e., one cannot 2480 call MatMultHermitianTransposeAdd(A,v1,v2,v1). 2481 2482 Level: beginner 2483 2484 Concepts: matrix vector product^transpose and addition 2485 2486 .seealso: MatMultHermitianTranspose(), MatMultTranspose(), MatMultAdd(), MatMult() 2487 @*/ 2488 PetscErrorCode MatMultHermitianTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3) 2489 { 2490 PetscErrorCode ierr; 2491 2492 PetscFunctionBegin; 2493 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2494 PetscValidType(mat,1); 2495 PetscValidHeaderSpecific(v1,VEC_CLASSID,2); 2496 PetscValidHeaderSpecific(v2,VEC_CLASSID,3); 2497 PetscValidHeaderSpecific(v3,VEC_CLASSID,4); 2498 2499 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2500 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2501 if (!mat->ops->multhermitiantransposeadd) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2502 if (v1 == v3) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors"); 2503 if (mat->rmap->N != v1->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N); 2504 if (mat->cmap->N != v2->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N); 2505 if (mat->cmap->N != v3->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N); 2506 MatCheckPreallocated(mat,1); 2507 2508 ierr = PetscLogEventBegin(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2509 ierr = (*mat->ops->multhermitiantransposeadd)(mat,v1,v2,v3);CHKERRQ(ierr); 2510 ierr = PetscLogEventEnd(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr); 2511 ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr); 2512 PetscFunctionReturn(0); 2513 } 2514 2515 #undef __FUNCT__ 2516 #define __FUNCT__ "MatMultConstrained" 2517 /*@ 2518 MatMultConstrained - The inner multiplication routine for a 2519 constrained matrix P^T A P. 2520 2521 Neighbor-wise Collective on Mat and Vec 2522 2523 Input Parameters: 2524 + mat - the matrix 2525 - x - the vector to be multilplied 2526 2527 Output Parameters: 2528 . y - the result 2529 2530 Notes: 2531 The vectors x and y cannot be the same. I.e., one cannot 2532 call MatMult(A,y,y). 2533 2534 Level: beginner 2535 2536 .keywords: matrix, multiply, matrix-vector product, constraint 2537 .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 2538 @*/ 2539 PetscErrorCode MatMultConstrained(Mat mat,Vec x,Vec y) 2540 { 2541 PetscErrorCode ierr; 2542 2543 PetscFunctionBegin; 2544 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2545 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2546 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2547 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2548 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2549 if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2550 if (mat->cmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 2551 if (mat->rmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N); 2552 if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n); 2553 2554 ierr = PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr); 2555 ierr = (*mat->ops->multconstrained)(mat,x,y);CHKERRQ(ierr); 2556 ierr = PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr); 2557 ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr); 2558 PetscFunctionReturn(0); 2559 } 2560 2561 #undef __FUNCT__ 2562 #define __FUNCT__ "MatMultTransposeConstrained" 2563 /*@ 2564 MatMultTransposeConstrained - The inner multiplication routine for a 2565 constrained matrix P^T A^T P. 2566 2567 Neighbor-wise Collective on Mat and Vec 2568 2569 Input Parameters: 2570 + mat - the matrix 2571 - x - the vector to be multilplied 2572 2573 Output Parameters: 2574 . y - the result 2575 2576 Notes: 2577 The vectors x and y cannot be the same. I.e., one cannot 2578 call MatMult(A,y,y). 2579 2580 Level: beginner 2581 2582 .keywords: matrix, multiply, matrix-vector product, constraint 2583 .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 2584 @*/ 2585 PetscErrorCode MatMultTransposeConstrained(Mat mat,Vec x,Vec y) 2586 { 2587 PetscErrorCode ierr; 2588 2589 PetscFunctionBegin; 2590 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2591 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 2592 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 2593 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2594 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2595 if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors"); 2596 if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 2597 if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N); 2598 2599 ierr = PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr); 2600 ierr = (*mat->ops->multtransposeconstrained)(mat,x,y);CHKERRQ(ierr); 2601 ierr = PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr); 2602 ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr); 2603 PetscFunctionReturn(0); 2604 } 2605 2606 #undef __FUNCT__ 2607 #define __FUNCT__ "MatGetFactorType" 2608 /*@C 2609 MatGetFactorType - gets the type of factorization it is 2610 2611 Note Collective 2612 as the flag 2613 2614 Input Parameters: 2615 . mat - the matrix 2616 2617 Output Parameters: 2618 . t - the type, one of MAT_FACTOR_NONE, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ILU, MAT_FACTOR_ICC,MAT_FACTOR_ILUDT 2619 2620 Level: intermediate 2621 2622 .seealso: MatFactorType, MatGetFactor() 2623 @*/ 2624 PetscErrorCode MatGetFactorType(Mat mat,MatFactorType *t) 2625 { 2626 PetscFunctionBegin; 2627 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2628 PetscValidType(mat,1); 2629 *t = mat->factortype; 2630 PetscFunctionReturn(0); 2631 } 2632 2633 /* ------------------------------------------------------------*/ 2634 #undef __FUNCT__ 2635 #define __FUNCT__ "MatGetInfo" 2636 /*@C 2637 MatGetInfo - Returns information about matrix storage (number of 2638 nonzeros, memory, etc.). 2639 2640 Collective on Mat if MAT_GLOBAL_MAX or MAT_GLOBAL_SUM is used as the flag 2641 2642 Input Parameters: 2643 . mat - the matrix 2644 2645 Output Parameters: 2646 + flag - flag indicating the type of parameters to be returned 2647 (MAT_LOCAL - local matrix, MAT_GLOBAL_MAX - maximum over all processors, 2648 MAT_GLOBAL_SUM - sum over all processors) 2649 - info - matrix information context 2650 2651 Notes: 2652 The MatInfo context contains a variety of matrix data, including 2653 number of nonzeros allocated and used, number of mallocs during 2654 matrix assembly, etc. Additional information for factored matrices 2655 is provided (such as the fill ratio, number of mallocs during 2656 factorization, etc.). Much of this info is printed to PETSC_STDOUT 2657 when using the runtime options 2658 $ -info -mat_view ::ascii_info 2659 2660 Example for C/C++ Users: 2661 See the file ${PETSC_DIR}/include/petscmat.h for a complete list of 2662 data within the MatInfo context. For example, 2663 .vb 2664 MatInfo info; 2665 Mat A; 2666 double mal, nz_a, nz_u; 2667 2668 MatGetInfo(A,MAT_LOCAL,&info); 2669 mal = info.mallocs; 2670 nz_a = info.nz_allocated; 2671 .ve 2672 2673 Example for Fortran Users: 2674 Fortran users should declare info as a double precision 2675 array of dimension MAT_INFO_SIZE, and then extract the parameters 2676 of interest. See the file ${PETSC_DIR}/include/finclude/petscmat.h 2677 a complete list of parameter names. 2678 .vb 2679 double precision info(MAT_INFO_SIZE) 2680 double precision mal, nz_a 2681 Mat A 2682 integer ierr 2683 2684 call MatGetInfo(A,MAT_LOCAL,info,ierr) 2685 mal = info(MAT_INFO_MALLOCS) 2686 nz_a = info(MAT_INFO_NZ_ALLOCATED) 2687 .ve 2688 2689 Level: intermediate 2690 2691 Concepts: matrices^getting information on 2692 2693 Developer Note: fortran interface is not autogenerated as the f90 2694 interface defintion cannot be generated correctly [due to MatInfo] 2695 2696 .seealso: MatStashGetInfo() 2697 2698 @*/ 2699 PetscErrorCode MatGetInfo(Mat mat,MatInfoType flag,MatInfo *info) 2700 { 2701 PetscErrorCode ierr; 2702 2703 PetscFunctionBegin; 2704 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2705 PetscValidType(mat,1); 2706 PetscValidPointer(info,3); 2707 if (!mat->ops->getinfo) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2708 MatCheckPreallocated(mat,1); 2709 ierr = (*mat->ops->getinfo)(mat,flag,info);CHKERRQ(ierr); 2710 PetscFunctionReturn(0); 2711 } 2712 2713 /* ----------------------------------------------------------*/ 2714 2715 #undef __FUNCT__ 2716 #define __FUNCT__ "MatLUFactor" 2717 /*@C 2718 MatLUFactor - Performs in-place LU factorization of matrix. 2719 2720 Collective on Mat 2721 2722 Input Parameters: 2723 + mat - the matrix 2724 . row - row permutation 2725 . col - column permutation 2726 - info - options for factorization, includes 2727 $ fill - expected fill as ratio of original fill. 2728 $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting) 2729 $ Run with the option -info to determine an optimal value to use 2730 2731 Notes: 2732 Most users should employ the simplified KSP interface for linear solvers 2733 instead of working directly with matrix algebra routines such as this. 2734 See, e.g., KSPCreate(). 2735 2736 This changes the state of the matrix to a factored matrix; it cannot be used 2737 for example with MatSetValues() unless one first calls MatSetUnfactored(). 2738 2739 Level: developer 2740 2741 Concepts: matrices^LU factorization 2742 2743 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(), 2744 MatGetOrdering(), MatSetUnfactored(), MatFactorInfo, MatGetFactor() 2745 2746 Developer Note: fortran interface is not autogenerated as the f90 2747 interface defintion cannot be generated correctly [due to MatFactorInfo] 2748 2749 @*/ 2750 PetscErrorCode MatLUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info) 2751 { 2752 PetscErrorCode ierr; 2753 MatFactorInfo tinfo; 2754 2755 PetscFunctionBegin; 2756 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2757 if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2); 2758 if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3); 2759 if (info) PetscValidPointer(info,4); 2760 PetscValidType(mat,1); 2761 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2762 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2763 if (!mat->ops->lufactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2764 MatCheckPreallocated(mat,1); 2765 if (!info) { 2766 ierr = MatFactorInfoInitialize(&tinfo);CHKERRQ(ierr); 2767 info = &tinfo; 2768 } 2769 2770 ierr = PetscLogEventBegin(MAT_LUFactor,mat,row,col,0);CHKERRQ(ierr); 2771 ierr = (*mat->ops->lufactor)(mat,row,col,info);CHKERRQ(ierr); 2772 ierr = PetscLogEventEnd(MAT_LUFactor,mat,row,col,0);CHKERRQ(ierr); 2773 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 2774 PetscFunctionReturn(0); 2775 } 2776 2777 #undef __FUNCT__ 2778 #define __FUNCT__ "MatILUFactor" 2779 /*@C 2780 MatILUFactor - Performs in-place ILU factorization of matrix. 2781 2782 Collective on Mat 2783 2784 Input Parameters: 2785 + mat - the matrix 2786 . row - row permutation 2787 . col - column permutation 2788 - info - structure containing 2789 $ levels - number of levels of fill. 2790 $ expected fill - as ratio of original fill. 2791 $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices 2792 missing diagonal entries) 2793 2794 Notes: 2795 Probably really in-place only when level of fill is zero, otherwise allocates 2796 new space to store factored matrix and deletes previous memory. 2797 2798 Most users should employ the simplified KSP interface for linear solvers 2799 instead of working directly with matrix algebra routines such as this. 2800 See, e.g., KSPCreate(). 2801 2802 Level: developer 2803 2804 Concepts: matrices^ILU factorization 2805 2806 .seealso: MatILUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo 2807 2808 Developer Note: fortran interface is not autogenerated as the f90 2809 interface defintion cannot be generated correctly [due to MatFactorInfo] 2810 2811 @*/ 2812 PetscErrorCode MatILUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info) 2813 { 2814 PetscErrorCode ierr; 2815 2816 PetscFunctionBegin; 2817 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2818 if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2); 2819 if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3); 2820 PetscValidPointer(info,4); 2821 PetscValidType(mat,1); 2822 if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"matrix must be square"); 2823 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2824 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2825 if (!mat->ops->ilufactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 2826 MatCheckPreallocated(mat,1); 2827 2828 ierr = PetscLogEventBegin(MAT_ILUFactor,mat,row,col,0);CHKERRQ(ierr); 2829 ierr = (*mat->ops->ilufactor)(mat,row,col,info);CHKERRQ(ierr); 2830 ierr = PetscLogEventEnd(MAT_ILUFactor,mat,row,col,0);CHKERRQ(ierr); 2831 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 2832 PetscFunctionReturn(0); 2833 } 2834 2835 #undef __FUNCT__ 2836 #define __FUNCT__ "MatLUFactorSymbolic" 2837 /*@C 2838 MatLUFactorSymbolic - Performs symbolic LU factorization of matrix. 2839 Call this routine before calling MatLUFactorNumeric(). 2840 2841 Collective on Mat 2842 2843 Input Parameters: 2844 + fact - the factor matrix obtained with MatGetFactor() 2845 . mat - the matrix 2846 . row, col - row and column permutations 2847 - info - options for factorization, includes 2848 $ fill - expected fill as ratio of original fill. 2849 $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting) 2850 $ Run with the option -info to determine an optimal value to use 2851 2852 2853 Notes: 2854 See the <a href="../../docs/manual.pdf">users manual</a> for additional information about 2855 choosing the fill factor for better efficiency. 2856 2857 Most users should employ the simplified KSP interface for linear solvers 2858 instead of working directly with matrix algebra routines such as this. 2859 See, e.g., KSPCreate(). 2860 2861 Level: developer 2862 2863 Concepts: matrices^LU symbolic factorization 2864 2865 .seealso: MatLUFactor(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo 2866 2867 Developer Note: fortran interface is not autogenerated as the f90 2868 interface defintion cannot be generated correctly [due to MatFactorInfo] 2869 2870 @*/ 2871 PetscErrorCode MatLUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info) 2872 { 2873 PetscErrorCode ierr; 2874 2875 PetscFunctionBegin; 2876 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2877 if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2); 2878 if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3); 2879 if (info) PetscValidPointer(info,4); 2880 PetscValidType(mat,1); 2881 PetscValidPointer(fact,5); 2882 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2883 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 2884 if (!(fact)->ops->lufactorsymbolic) { 2885 const MatSolverPackage spackage; 2886 ierr = MatFactorGetSolverPackage(fact,&spackage);CHKERRQ(ierr); 2887 SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic LU using solver package %s",((PetscObject)mat)->type_name,spackage); 2888 } 2889 MatCheckPreallocated(mat,2); 2890 2891 ierr = PetscLogEventBegin(MAT_LUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr); 2892 ierr = (fact->ops->lufactorsymbolic)(fact,mat,row,col,info);CHKERRQ(ierr); 2893 ierr = PetscLogEventEnd(MAT_LUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr); 2894 ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr); 2895 PetscFunctionReturn(0); 2896 } 2897 2898 #undef __FUNCT__ 2899 #define __FUNCT__ "MatLUFactorNumeric" 2900 /*@C 2901 MatLUFactorNumeric - Performs numeric LU factorization of a matrix. 2902 Call this routine after first calling MatLUFactorSymbolic(). 2903 2904 Collective on Mat 2905 2906 Input Parameters: 2907 + fact - the factor matrix obtained with MatGetFactor() 2908 . mat - the matrix 2909 - info - options for factorization 2910 2911 Notes: 2912 See MatLUFactor() for in-place factorization. See 2913 MatCholeskyFactorNumeric() for the symmetric, positive definite case. 2914 2915 Most users should employ the simplified KSP interface for linear solvers 2916 instead of working directly with matrix algebra routines such as this. 2917 See, e.g., KSPCreate(). 2918 2919 Level: developer 2920 2921 Concepts: matrices^LU numeric factorization 2922 2923 .seealso: MatLUFactorSymbolic(), MatLUFactor(), MatCholeskyFactor() 2924 2925 Developer Note: fortran interface is not autogenerated as the f90 2926 interface defintion cannot be generated correctly [due to MatFactorInfo] 2927 2928 @*/ 2929 PetscErrorCode MatLUFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info) 2930 { 2931 PetscErrorCode ierr; 2932 2933 PetscFunctionBegin; 2934 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2935 PetscValidType(mat,1); 2936 PetscValidPointer(fact,2); 2937 PetscValidHeaderSpecific(fact,MAT_CLASSID,2); 2938 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2939 if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dimensions are different %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N); 2940 2941 if (!(fact)->ops->lufactornumeric) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s numeric LU",((PetscObject)mat)->type_name); 2942 MatCheckPreallocated(mat,2); 2943 ierr = PetscLogEventBegin(MAT_LUFactorNumeric,mat,fact,0,0);CHKERRQ(ierr); 2944 ierr = (fact->ops->lufactornumeric)(fact,mat,info);CHKERRQ(ierr); 2945 ierr = PetscLogEventEnd(MAT_LUFactorNumeric,mat,fact,0,0);CHKERRQ(ierr); 2946 2947 if (fact->viewonassembly) { 2948 ierr = PetscViewerPushFormat(fact->viewonassembly,fact->viewformatonassembly);CHKERRQ(ierr); 2949 ierr = MatView(fact,fact->viewonassembly);CHKERRQ(ierr); 2950 ierr = PetscViewerPopFormat(fact->viewonassembly);CHKERRQ(ierr); 2951 } 2952 ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr); 2953 PetscFunctionReturn(0); 2954 } 2955 2956 #undef __FUNCT__ 2957 #define __FUNCT__ "MatCholeskyFactor" 2958 /*@C 2959 MatCholeskyFactor - Performs in-place Cholesky factorization of a 2960 symmetric matrix. 2961 2962 Collective on Mat 2963 2964 Input Parameters: 2965 + mat - the matrix 2966 . perm - row and column permutations 2967 - f - expected fill as ratio of original fill 2968 2969 Notes: 2970 See MatLUFactor() for the nonsymmetric case. See also 2971 MatCholeskyFactorSymbolic(), and MatCholeskyFactorNumeric(). 2972 2973 Most users should employ the simplified KSP interface for linear solvers 2974 instead of working directly with matrix algebra routines such as this. 2975 See, e.g., KSPCreate(). 2976 2977 Level: developer 2978 2979 Concepts: matrices^Cholesky factorization 2980 2981 .seealso: MatLUFactor(), MatCholeskyFactorSymbolic(), MatCholeskyFactorNumeric() 2982 MatGetOrdering() 2983 2984 Developer Note: fortran interface is not autogenerated as the f90 2985 interface defintion cannot be generated correctly [due to MatFactorInfo] 2986 2987 @*/ 2988 PetscErrorCode MatCholeskyFactor(Mat mat,IS perm,const MatFactorInfo *info) 2989 { 2990 PetscErrorCode ierr; 2991 2992 PetscFunctionBegin; 2993 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 2994 PetscValidType(mat,1); 2995 if (perm) PetscValidHeaderSpecific(perm,IS_CLASSID,2); 2996 if (info) PetscValidPointer(info,3); 2997 if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"Matrix must be square"); 2998 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 2999 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3000 if (!mat->ops->choleskyfactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 3001 MatCheckPreallocated(mat,1); 3002 3003 ierr = PetscLogEventBegin(MAT_CholeskyFactor,mat,perm,0,0);CHKERRQ(ierr); 3004 ierr = (*mat->ops->choleskyfactor)(mat,perm,info);CHKERRQ(ierr); 3005 ierr = PetscLogEventEnd(MAT_CholeskyFactor,mat,perm,0,0);CHKERRQ(ierr); 3006 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 3007 PetscFunctionReturn(0); 3008 } 3009 3010 #undef __FUNCT__ 3011 #define __FUNCT__ "MatCholeskyFactorSymbolic" 3012 /*@C 3013 MatCholeskyFactorSymbolic - Performs symbolic Cholesky factorization 3014 of a symmetric matrix. 3015 3016 Collective on Mat 3017 3018 Input Parameters: 3019 + fact - the factor matrix obtained with MatGetFactor() 3020 . mat - the matrix 3021 . perm - row and column permutations 3022 - info - options for factorization, includes 3023 $ fill - expected fill as ratio of original fill. 3024 $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting) 3025 $ Run with the option -info to determine an optimal value to use 3026 3027 Notes: 3028 See MatLUFactorSymbolic() for the nonsymmetric case. See also 3029 MatCholeskyFactor() and MatCholeskyFactorNumeric(). 3030 3031 Most users should employ the simplified KSP interface for linear solvers 3032 instead of working directly with matrix algebra routines such as this. 3033 See, e.g., KSPCreate(). 3034 3035 Level: developer 3036 3037 Concepts: matrices^Cholesky symbolic factorization 3038 3039 .seealso: MatLUFactorSymbolic(), MatCholeskyFactor(), MatCholeskyFactorNumeric() 3040 MatGetOrdering() 3041 3042 Developer Note: fortran interface is not autogenerated as the f90 3043 interface defintion cannot be generated correctly [due to MatFactorInfo] 3044 3045 @*/ 3046 PetscErrorCode MatCholeskyFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info) 3047 { 3048 PetscErrorCode ierr; 3049 3050 PetscFunctionBegin; 3051 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3052 PetscValidType(mat,1); 3053 if (perm) PetscValidHeaderSpecific(perm,IS_CLASSID,2); 3054 if (info) PetscValidPointer(info,3); 3055 PetscValidPointer(fact,4); 3056 if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"Matrix must be square"); 3057 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3058 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3059 if (!(fact)->ops->choleskyfactorsymbolic) { 3060 const MatSolverPackage spackage; 3061 ierr = MatFactorGetSolverPackage(fact,&spackage);CHKERRQ(ierr); 3062 SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s symbolic factor Cholesky using solver package %s",((PetscObject)mat)->type_name,spackage); 3063 } 3064 MatCheckPreallocated(mat,2); 3065 3066 ierr = PetscLogEventBegin(MAT_CholeskyFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr); 3067 ierr = (fact->ops->choleskyfactorsymbolic)(fact,mat,perm,info);CHKERRQ(ierr); 3068 ierr = PetscLogEventEnd(MAT_CholeskyFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr); 3069 ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr); 3070 PetscFunctionReturn(0); 3071 } 3072 3073 #undef __FUNCT__ 3074 #define __FUNCT__ "MatCholeskyFactorNumeric" 3075 /*@C 3076 MatCholeskyFactorNumeric - Performs numeric Cholesky factorization 3077 of a symmetric matrix. Call this routine after first calling 3078 MatCholeskyFactorSymbolic(). 3079 3080 Collective on Mat 3081 3082 Input Parameters: 3083 + fact - the factor matrix obtained with MatGetFactor() 3084 . mat - the initial matrix 3085 . info - options for factorization 3086 - fact - the symbolic factor of mat 3087 3088 3089 Notes: 3090 Most users should employ the simplified KSP interface for linear solvers 3091 instead of working directly with matrix algebra routines such as this. 3092 See, e.g., KSPCreate(). 3093 3094 Level: developer 3095 3096 Concepts: matrices^Cholesky numeric factorization 3097 3098 .seealso: MatCholeskyFactorSymbolic(), MatCholeskyFactor(), MatLUFactorNumeric() 3099 3100 Developer Note: fortran interface is not autogenerated as the f90 3101 interface defintion cannot be generated correctly [due to MatFactorInfo] 3102 3103 @*/ 3104 PetscErrorCode MatCholeskyFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info) 3105 { 3106 PetscErrorCode ierr; 3107 3108 PetscFunctionBegin; 3109 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3110 PetscValidType(mat,1); 3111 PetscValidPointer(fact,2); 3112 PetscValidHeaderSpecific(fact,MAT_CLASSID,2); 3113 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3114 if (!(fact)->ops->choleskyfactornumeric) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s numeric factor Cholesky",((PetscObject)mat)->type_name); 3115 if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dim %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N); 3116 MatCheckPreallocated(mat,2); 3117 3118 ierr = PetscLogEventBegin(MAT_CholeskyFactorNumeric,mat,fact,0,0);CHKERRQ(ierr); 3119 ierr = (fact->ops->choleskyfactornumeric)(fact,mat,info);CHKERRQ(ierr); 3120 ierr = PetscLogEventEnd(MAT_CholeskyFactorNumeric,mat,fact,0,0);CHKERRQ(ierr); 3121 3122 if (fact->viewonassembly) { 3123 ierr = PetscViewerPushFormat(fact->viewonassembly,fact->viewformatonassembly);CHKERRQ(ierr); 3124 ierr = MatView(fact,fact->viewonassembly);CHKERRQ(ierr); 3125 ierr = PetscViewerPopFormat(fact->viewonassembly);CHKERRQ(ierr); 3126 } 3127 ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr); 3128 PetscFunctionReturn(0); 3129 } 3130 3131 /* ----------------------------------------------------------------*/ 3132 #undef __FUNCT__ 3133 #define __FUNCT__ "MatSolve" 3134 /*@ 3135 MatSolve - Solves A x = b, given a factored matrix. 3136 3137 Neighbor-wise Collective on Mat and Vec 3138 3139 Input Parameters: 3140 + mat - the factored matrix 3141 - b - the right-hand-side vector 3142 3143 Output Parameter: 3144 . x - the result vector 3145 3146 Notes: 3147 The vectors b and x cannot be the same. I.e., one cannot 3148 call MatSolve(A,x,x). 3149 3150 Notes: 3151 Most users should employ the simplified KSP interface for linear solvers 3152 instead of working directly with matrix algebra routines such as this. 3153 See, e.g., KSPCreate(). 3154 3155 Level: developer 3156 3157 Concepts: matrices^triangular solves 3158 3159 .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd() 3160 @*/ 3161 PetscErrorCode MatSolve(Mat mat,Vec b,Vec x) 3162 { 3163 PetscErrorCode ierr; 3164 3165 PetscFunctionBegin; 3166 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3167 PetscValidType(mat,1); 3168 PetscValidHeaderSpecific(b,VEC_CLASSID,2); 3169 PetscValidHeaderSpecific(x,VEC_CLASSID,3); 3170 PetscCheckSameComm(mat,1,b,2); 3171 PetscCheckSameComm(mat,1,x,3); 3172 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3173 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3174 if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 3175 if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N); 3176 if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n); 3177 if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(0); 3178 if (!mat->ops->solve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 3179 MatCheckPreallocated(mat,1); 3180 3181 ierr = PetscLogEventBegin(MAT_Solve,mat,b,x,0);CHKERRQ(ierr); 3182 ierr = (*mat->ops->solve)(mat,b,x);CHKERRQ(ierr); 3183 ierr = PetscLogEventEnd(MAT_Solve,mat,b,x,0);CHKERRQ(ierr); 3184 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3185 PetscFunctionReturn(0); 3186 } 3187 3188 #undef __FUNCT__ 3189 #define __FUNCT__ "MatMatSolve_Basic" 3190 PetscErrorCode MatMatSolve_Basic(Mat A,Mat B,Mat X) 3191 { 3192 PetscErrorCode ierr; 3193 Vec b,x; 3194 PetscInt m,N,i; 3195 PetscScalar *bb,*xx; 3196 PetscBool flg; 3197 3198 PetscFunctionBegin; 3199 ierr = PetscObjectTypeCompareAny((PetscObject)B,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 3200 if (!flg) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Matrix B must be MATDENSE matrix"); 3201 ierr = PetscObjectTypeCompareAny((PetscObject)X,&flg,MATSEQDENSE,MATMPIDENSE,NULL);CHKERRQ(ierr); 3202 if (!flg) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Matrix X must be MATDENSE matrix"); 3203 3204 ierr = MatDenseGetArray(B,&bb);CHKERRQ(ierr); 3205 ierr = MatDenseGetArray(X,&xx);CHKERRQ(ierr); 3206 ierr = MatGetLocalSize(B,&m,NULL);CHKERRQ(ierr); /* number local rows */ 3207 ierr = MatGetSize(B,NULL,&N);CHKERRQ(ierr); /* total columns in dense matrix */ 3208 ierr = MatGetVecs(A,&x,&b);CHKERRQ(ierr); 3209 for (i=0; i<N; i++) { 3210 ierr = VecPlaceArray(b,bb + i*m);CHKERRQ(ierr); 3211 ierr = VecPlaceArray(x,xx + i*m);CHKERRQ(ierr); 3212 ierr = MatSolve(A,b,x);CHKERRQ(ierr); 3213 ierr = VecResetArray(x);CHKERRQ(ierr); 3214 ierr = VecResetArray(b);CHKERRQ(ierr); 3215 } 3216 ierr = VecDestroy(&b);CHKERRQ(ierr); 3217 ierr = VecDestroy(&x);CHKERRQ(ierr); 3218 ierr = MatDenseRestoreArray(B,&bb);CHKERRQ(ierr); 3219 ierr = MatDenseRestoreArray(X,&xx);CHKERRQ(ierr); 3220 PetscFunctionReturn(0); 3221 } 3222 3223 #undef __FUNCT__ 3224 #define __FUNCT__ "MatMatSolve" 3225 /*@ 3226 MatMatSolve - Solves A X = B, given a factored matrix. 3227 3228 Neighbor-wise Collective on Mat 3229 3230 Input Parameters: 3231 + mat - the factored matrix 3232 - B - the right-hand-side matrix (dense matrix) 3233 3234 Output Parameter: 3235 . X - the result matrix (dense matrix) 3236 3237 Notes: 3238 The matrices b and x cannot be the same. I.e., one cannot 3239 call MatMatSolve(A,x,x). 3240 3241 Notes: 3242 Most users should usually employ the simplified KSP interface for linear solvers 3243 instead of working directly with matrix algebra routines such as this. 3244 See, e.g., KSPCreate(). However KSP can only solve for one vector (column of X) 3245 at a time. 3246 3247 When using SuperLU_Dist as a parallel solver PETSc will use the SuperLU_Dist functionality to solve multiple right hand sides simultaneously. For MUMPS 3248 it calls a separate solve for each right hand side since MUMPS does not yet support distributed right hand sides. 3249 3250 Since the resulting matrix X must always be dense we do not support sparse representation of the matrix B. 3251 3252 Level: developer 3253 3254 Concepts: matrices^triangular solves 3255 3256 .seealso: MatMatSolveAdd(), MatMatSolveTranspose(), MatMatSolveTransposeAdd(), MatLUFactor(), MatCholeskyFactor() 3257 @*/ 3258 PetscErrorCode MatMatSolve(Mat A,Mat B,Mat X) 3259 { 3260 PetscErrorCode ierr; 3261 3262 PetscFunctionBegin; 3263 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 3264 PetscValidType(A,1); 3265 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 3266 PetscValidHeaderSpecific(X,MAT_CLASSID,3); 3267 PetscCheckSameComm(A,1,B,2); 3268 PetscCheckSameComm(A,1,X,3); 3269 if (X == B) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_IDN,"X and B must be different matrices"); 3270 if (!A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3271 if (A->cmap->N != X->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat X: global dim %D %D",A->cmap->N,X->rmap->N); 3272 if (A->rmap->N != B->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D",A->rmap->N,B->rmap->N); 3273 if (A->rmap->n != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: local dim %D %D",A->rmap->n,B->rmap->n); 3274 if (!A->rmap->N && !A->cmap->N) PetscFunctionReturn(0); 3275 MatCheckPreallocated(A,1); 3276 3277 ierr = PetscLogEventBegin(MAT_MatSolve,A,B,X,0);CHKERRQ(ierr); 3278 if (!A->ops->matsolve) { 3279 ierr = PetscInfo1(A,"Mat type %s using basic MatMatSolve",((PetscObject)A)->type_name);CHKERRQ(ierr); 3280 ierr = MatMatSolve_Basic(A,B,X);CHKERRQ(ierr); 3281 } else { 3282 ierr = (*A->ops->matsolve)(A,B,X);CHKERRQ(ierr); 3283 } 3284 ierr = PetscLogEventEnd(MAT_MatSolve,A,B,X,0);CHKERRQ(ierr); 3285 ierr = PetscObjectStateIncrease((PetscObject)X);CHKERRQ(ierr); 3286 PetscFunctionReturn(0); 3287 } 3288 3289 3290 #undef __FUNCT__ 3291 #define __FUNCT__ "MatForwardSolve" 3292 /*@ 3293 MatForwardSolve - Solves L x = b, given a factored matrix, A = LU, or 3294 U^T*D^(1/2) x = b, given a factored symmetric matrix, A = U^T*D*U, 3295 3296 Neighbor-wise Collective on Mat and Vec 3297 3298 Input Parameters: 3299 + mat - the factored matrix 3300 - b - the right-hand-side vector 3301 3302 Output Parameter: 3303 . x - the result vector 3304 3305 Notes: 3306 MatSolve() should be used for most applications, as it performs 3307 a forward solve followed by a backward solve. 3308 3309 The vectors b and x cannot be the same, i.e., one cannot 3310 call MatForwardSolve(A,x,x). 3311 3312 For matrix in seqsbaij format with block size larger than 1, 3313 the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet. 3314 MatForwardSolve() solves U^T*D y = b, and 3315 MatBackwardSolve() solves U x = y. 3316 Thus they do not provide a symmetric preconditioner. 3317 3318 Most users should employ the simplified KSP interface for linear solvers 3319 instead of working directly with matrix algebra routines such as this. 3320 See, e.g., KSPCreate(). 3321 3322 Level: developer 3323 3324 Concepts: matrices^forward solves 3325 3326 .seealso: MatSolve(), MatBackwardSolve() 3327 @*/ 3328 PetscErrorCode MatForwardSolve(Mat mat,Vec b,Vec x) 3329 { 3330 PetscErrorCode ierr; 3331 3332 PetscFunctionBegin; 3333 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3334 PetscValidType(mat,1); 3335 PetscValidHeaderSpecific(b,VEC_CLASSID,2); 3336 PetscValidHeaderSpecific(x,VEC_CLASSID,3); 3337 PetscCheckSameComm(mat,1,b,2); 3338 PetscCheckSameComm(mat,1,x,3); 3339 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3340 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3341 if (!mat->ops->forwardsolve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 3342 if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 3343 if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N); 3344 if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n); 3345 MatCheckPreallocated(mat,1); 3346 ierr = PetscLogEventBegin(MAT_ForwardSolve,mat,b,x,0);CHKERRQ(ierr); 3347 ierr = (*mat->ops->forwardsolve)(mat,b,x);CHKERRQ(ierr); 3348 ierr = PetscLogEventEnd(MAT_ForwardSolve,mat,b,x,0);CHKERRQ(ierr); 3349 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3350 PetscFunctionReturn(0); 3351 } 3352 3353 #undef __FUNCT__ 3354 #define __FUNCT__ "MatBackwardSolve" 3355 /*@ 3356 MatBackwardSolve - Solves U x = b, given a factored matrix, A = LU. 3357 D^(1/2) U x = b, given a factored symmetric matrix, A = U^T*D*U, 3358 3359 Neighbor-wise Collective on Mat and Vec 3360 3361 Input Parameters: 3362 + mat - the factored matrix 3363 - b - the right-hand-side vector 3364 3365 Output Parameter: 3366 . x - the result vector 3367 3368 Notes: 3369 MatSolve() should be used for most applications, as it performs 3370 a forward solve followed by a backward solve. 3371 3372 The vectors b and x cannot be the same. I.e., one cannot 3373 call MatBackwardSolve(A,x,x). 3374 3375 For matrix in seqsbaij format with block size larger than 1, 3376 the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet. 3377 MatForwardSolve() solves U^T*D y = b, and 3378 MatBackwardSolve() solves U x = y. 3379 Thus they do not provide a symmetric preconditioner. 3380 3381 Most users should employ the simplified KSP interface for linear solvers 3382 instead of working directly with matrix algebra routines such as this. 3383 See, e.g., KSPCreate(). 3384 3385 Level: developer 3386 3387 Concepts: matrices^backward solves 3388 3389 .seealso: MatSolve(), MatForwardSolve() 3390 @*/ 3391 PetscErrorCode MatBackwardSolve(Mat mat,Vec b,Vec x) 3392 { 3393 PetscErrorCode ierr; 3394 3395 PetscFunctionBegin; 3396 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3397 PetscValidType(mat,1); 3398 PetscValidHeaderSpecific(b,VEC_CLASSID,2); 3399 PetscValidHeaderSpecific(x,VEC_CLASSID,3); 3400 PetscCheckSameComm(mat,1,b,2); 3401 PetscCheckSameComm(mat,1,x,3); 3402 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3403 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3404 if (!mat->ops->backwardsolve) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 3405 if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 3406 if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N); 3407 if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n); 3408 MatCheckPreallocated(mat,1); 3409 3410 ierr = PetscLogEventBegin(MAT_BackwardSolve,mat,b,x,0);CHKERRQ(ierr); 3411 ierr = (*mat->ops->backwardsolve)(mat,b,x);CHKERRQ(ierr); 3412 ierr = PetscLogEventEnd(MAT_BackwardSolve,mat,b,x,0);CHKERRQ(ierr); 3413 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3414 PetscFunctionReturn(0); 3415 } 3416 3417 #undef __FUNCT__ 3418 #define __FUNCT__ "MatSolveAdd" 3419 /*@ 3420 MatSolveAdd - Computes x = y + inv(A)*b, given a factored matrix. 3421 3422 Neighbor-wise Collective on Mat and Vec 3423 3424 Input Parameters: 3425 + mat - the factored matrix 3426 . b - the right-hand-side vector 3427 - y - the vector to be added to 3428 3429 Output Parameter: 3430 . x - the result vector 3431 3432 Notes: 3433 The vectors b and x cannot be the same. I.e., one cannot 3434 call MatSolveAdd(A,x,y,x). 3435 3436 Most users should employ the simplified KSP interface for linear solvers 3437 instead of working directly with matrix algebra routines such as this. 3438 See, e.g., KSPCreate(). 3439 3440 Level: developer 3441 3442 Concepts: matrices^triangular solves 3443 3444 .seealso: MatSolve(), MatSolveTranspose(), MatSolveTransposeAdd() 3445 @*/ 3446 PetscErrorCode MatSolveAdd(Mat mat,Vec b,Vec y,Vec x) 3447 { 3448 PetscScalar one = 1.0; 3449 Vec tmp; 3450 PetscErrorCode ierr; 3451 3452 PetscFunctionBegin; 3453 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3454 PetscValidType(mat,1); 3455 PetscValidHeaderSpecific(y,VEC_CLASSID,2); 3456 PetscValidHeaderSpecific(b,VEC_CLASSID,3); 3457 PetscValidHeaderSpecific(x,VEC_CLASSID,4); 3458 PetscCheckSameComm(mat,1,b,2); 3459 PetscCheckSameComm(mat,1,y,2); 3460 PetscCheckSameComm(mat,1,x,3); 3461 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3462 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3463 if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 3464 if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N); 3465 if (mat->rmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N); 3466 if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n); 3467 if (x->map->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n); 3468 MatCheckPreallocated(mat,1); 3469 3470 ierr = PetscLogEventBegin(MAT_SolveAdd,mat,b,x,y);CHKERRQ(ierr); 3471 if (mat->ops->solveadd) { 3472 ierr = (*mat->ops->solveadd)(mat,b,y,x);CHKERRQ(ierr); 3473 } else { 3474 /* do the solve then the add manually */ 3475 if (x != y) { 3476 ierr = MatSolve(mat,b,x);CHKERRQ(ierr); 3477 ierr = VecAXPY(x,one,y);CHKERRQ(ierr); 3478 } else { 3479 ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr); 3480 ierr = PetscLogObjectParent(mat,tmp);CHKERRQ(ierr); 3481 ierr = VecCopy(x,tmp);CHKERRQ(ierr); 3482 ierr = MatSolve(mat,b,x);CHKERRQ(ierr); 3483 ierr = VecAXPY(x,one,tmp);CHKERRQ(ierr); 3484 ierr = VecDestroy(&tmp);CHKERRQ(ierr); 3485 } 3486 } 3487 ierr = PetscLogEventEnd(MAT_SolveAdd,mat,b,x,y);CHKERRQ(ierr); 3488 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3489 PetscFunctionReturn(0); 3490 } 3491 3492 #undef __FUNCT__ 3493 #define __FUNCT__ "MatSolveTranspose" 3494 /*@ 3495 MatSolveTranspose - Solves A' x = b, given a factored matrix. 3496 3497 Neighbor-wise Collective on Mat and Vec 3498 3499 Input Parameters: 3500 + mat - the factored matrix 3501 - b - the right-hand-side vector 3502 3503 Output Parameter: 3504 . x - the result vector 3505 3506 Notes: 3507 The vectors b and x cannot be the same. I.e., one cannot 3508 call MatSolveTranspose(A,x,x). 3509 3510 Most users should employ the simplified KSP interface for linear solvers 3511 instead of working directly with matrix algebra routines such as this. 3512 See, e.g., KSPCreate(). 3513 3514 Level: developer 3515 3516 Concepts: matrices^triangular solves 3517 3518 .seealso: MatSolve(), MatSolveAdd(), MatSolveTransposeAdd() 3519 @*/ 3520 PetscErrorCode MatSolveTranspose(Mat mat,Vec b,Vec x) 3521 { 3522 PetscErrorCode ierr; 3523 3524 PetscFunctionBegin; 3525 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3526 PetscValidType(mat,1); 3527 PetscValidHeaderSpecific(b,VEC_CLASSID,2); 3528 PetscValidHeaderSpecific(x,VEC_CLASSID,3); 3529 PetscCheckSameComm(mat,1,b,2); 3530 PetscCheckSameComm(mat,1,x,3); 3531 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3532 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3533 if (!mat->ops->solvetranspose) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s",((PetscObject)mat)->type_name); 3534 if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N); 3535 if (mat->cmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N); 3536 MatCheckPreallocated(mat,1); 3537 ierr = PetscLogEventBegin(MAT_SolveTranspose,mat,b,x,0);CHKERRQ(ierr); 3538 ierr = (*mat->ops->solvetranspose)(mat,b,x);CHKERRQ(ierr); 3539 ierr = PetscLogEventEnd(MAT_SolveTranspose,mat,b,x,0);CHKERRQ(ierr); 3540 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3541 PetscFunctionReturn(0); 3542 } 3543 3544 #undef __FUNCT__ 3545 #define __FUNCT__ "MatSolveTransposeAdd" 3546 /*@ 3547 MatSolveTransposeAdd - Computes x = y + inv(Transpose(A)) b, given a 3548 factored matrix. 3549 3550 Neighbor-wise Collective on Mat and Vec 3551 3552 Input Parameters: 3553 + mat - the factored matrix 3554 . b - the right-hand-side vector 3555 - y - the vector to be added to 3556 3557 Output Parameter: 3558 . x - the result vector 3559 3560 Notes: 3561 The vectors b and x cannot be the same. I.e., one cannot 3562 call MatSolveTransposeAdd(A,x,y,x). 3563 3564 Most users should employ the simplified KSP interface for linear solvers 3565 instead of working directly with matrix algebra routines such as this. 3566 See, e.g., KSPCreate(). 3567 3568 Level: developer 3569 3570 Concepts: matrices^triangular solves 3571 3572 .seealso: MatSolve(), MatSolveAdd(), MatSolveTranspose() 3573 @*/ 3574 PetscErrorCode MatSolveTransposeAdd(Mat mat,Vec b,Vec y,Vec x) 3575 { 3576 PetscScalar one = 1.0; 3577 PetscErrorCode ierr; 3578 Vec tmp; 3579 3580 PetscFunctionBegin; 3581 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3582 PetscValidType(mat,1); 3583 PetscValidHeaderSpecific(y,VEC_CLASSID,2); 3584 PetscValidHeaderSpecific(b,VEC_CLASSID,3); 3585 PetscValidHeaderSpecific(x,VEC_CLASSID,4); 3586 PetscCheckSameComm(mat,1,b,2); 3587 PetscCheckSameComm(mat,1,y,3); 3588 PetscCheckSameComm(mat,1,x,4); 3589 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 3590 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 3591 if (mat->rmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N); 3592 if (mat->cmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N); 3593 if (mat->cmap->N != y->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N); 3594 if (x->map->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n); 3595 MatCheckPreallocated(mat,1); 3596 3597 ierr = PetscLogEventBegin(MAT_SolveTransposeAdd,mat,b,x,y);CHKERRQ(ierr); 3598 if (mat->ops->solvetransposeadd) { 3599 ierr = (*mat->ops->solvetransposeadd)(mat,b,y,x);CHKERRQ(ierr); 3600 } else { 3601 /* do the solve then the add manually */ 3602 if (x != y) { 3603 ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr); 3604 ierr = VecAXPY(x,one,y);CHKERRQ(ierr); 3605 } else { 3606 ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr); 3607 ierr = PetscLogObjectParent(mat,tmp);CHKERRQ(ierr); 3608 ierr = VecCopy(x,tmp);CHKERRQ(ierr); 3609 ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr); 3610 ierr = VecAXPY(x,one,tmp);CHKERRQ(ierr); 3611 ierr = VecDestroy(&tmp);CHKERRQ(ierr); 3612 } 3613 } 3614 ierr = PetscLogEventEnd(MAT_SolveTransposeAdd,mat,b,x,y);CHKERRQ(ierr); 3615 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3616 PetscFunctionReturn(0); 3617 } 3618 /* ----------------------------------------------------------------*/ 3619 3620 #undef __FUNCT__ 3621 #define __FUNCT__ "MatSOR" 3622 /*@ 3623 MatSOR - Computes relaxation (SOR, Gauss-Seidel) sweeps. 3624 3625 Neighbor-wise Collective on Mat and Vec 3626 3627 Input Parameters: 3628 + mat - the matrix 3629 . b - the right hand side 3630 . omega - the relaxation factor 3631 . flag - flag indicating the type of SOR (see below) 3632 . shift - diagonal shift 3633 . its - the number of iterations 3634 - lits - the number of local iterations 3635 3636 Output Parameters: 3637 . x - the solution (can contain an initial guess, use option SOR_ZERO_INITIAL_GUESS to indicate no guess) 3638 3639 SOR Flags: 3640 . SOR_FORWARD_SWEEP - forward SOR 3641 . SOR_BACKWARD_SWEEP - backward SOR 3642 . SOR_SYMMETRIC_SWEEP - SSOR (symmetric SOR) 3643 . SOR_LOCAL_FORWARD_SWEEP - local forward SOR 3644 . SOR_LOCAL_BACKWARD_SWEEP - local forward SOR 3645 . SOR_LOCAL_SYMMETRIC_SWEEP - local SSOR 3646 . SOR_APPLY_UPPER, SOR_APPLY_LOWER - applies 3647 upper/lower triangular part of matrix to 3648 vector (with omega) 3649 . SOR_ZERO_INITIAL_GUESS - zero initial guess 3650 3651 Notes: 3652 SOR_LOCAL_FORWARD_SWEEP, SOR_LOCAL_BACKWARD_SWEEP, and 3653 SOR_LOCAL_SYMMETRIC_SWEEP perform separate independent smoothings 3654 on each processor. 3655 3656 Application programmers will not generally use MatSOR() directly, 3657 but instead will employ the KSP/PC interface. 3658 3659 Notes: for BAIJ, SBAIJ, and AIJ matrices with Inodes this does a block SOR smoothing, otherwise it does a pointwise smoothing 3660 3661 Notes for Advanced Users: 3662 The flags are implemented as bitwise inclusive or operations. 3663 For example, use (SOR_ZERO_INITIAL_GUESS | SOR_SYMMETRIC_SWEEP) 3664 to specify a zero initial guess for SSOR. 3665 3666 Most users should employ the simplified KSP interface for linear solvers 3667 instead of working directly with matrix algebra routines such as this. 3668 See, e.g., KSPCreate(). 3669 3670 Vectors x and b CANNOT be the same 3671 3672 Developer Note: We should add block SOR support for AIJ matrices with block size set to great than one and no inodes 3673 3674 Level: developer 3675 3676 Concepts: matrices^relaxation 3677 Concepts: matrices^SOR 3678 Concepts: matrices^Gauss-Seidel 3679 3680 @*/ 3681 PetscErrorCode MatSOR(Mat mat,Vec b,PetscReal omega,MatSORType flag,PetscReal shift,PetscInt its,PetscInt lits,Vec x) 3682 { 3683 PetscErrorCode ierr; 3684 3685 PetscFunctionBegin; 3686 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3687 PetscValidType(mat,1); 3688 PetscValidHeaderSpecific(b,VEC_CLASSID,2); 3689 PetscValidHeaderSpecific(x,VEC_CLASSID,8); 3690 PetscCheckSameComm(mat,1,b,2); 3691 PetscCheckSameComm(mat,1,x,8); 3692 if (!mat->ops->sor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 3693 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3694 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3695 if (mat->cmap->N != x->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N); 3696 if (mat->rmap->N != b->map->N) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N); 3697 if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n); 3698 if (its <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D positive",its); 3699 if (lits <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires local its %D positive",lits); 3700 if (b == x) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_IDN,"b and x vector cannot be the same"); 3701 3702 MatCheckPreallocated(mat,1); 3703 ierr = PetscLogEventBegin(MAT_SOR,mat,b,x,0);CHKERRQ(ierr); 3704 ierr =(*mat->ops->sor)(mat,b,omega,flag,shift,its,lits,x);CHKERRQ(ierr); 3705 ierr = PetscLogEventEnd(MAT_SOR,mat,b,x,0);CHKERRQ(ierr); 3706 ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr); 3707 PetscFunctionReturn(0); 3708 } 3709 3710 #undef __FUNCT__ 3711 #define __FUNCT__ "MatCopy_Basic" 3712 /* 3713 Default matrix copy routine. 3714 */ 3715 PetscErrorCode MatCopy_Basic(Mat A,Mat B,MatStructure str) 3716 { 3717 PetscErrorCode ierr; 3718 PetscInt i,rstart = 0,rend = 0,nz; 3719 const PetscInt *cwork; 3720 const PetscScalar *vwork; 3721 3722 PetscFunctionBegin; 3723 if (B->assembled) { 3724 ierr = MatZeroEntries(B);CHKERRQ(ierr); 3725 } 3726 ierr = MatGetOwnershipRange(A,&rstart,&rend);CHKERRQ(ierr); 3727 for (i=rstart; i<rend; i++) { 3728 ierr = MatGetRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 3729 ierr = MatSetValues(B,1,&i,nz,cwork,vwork,INSERT_VALUES);CHKERRQ(ierr); 3730 ierr = MatRestoreRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 3731 } 3732 ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3733 ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3734 ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr); 3735 PetscFunctionReturn(0); 3736 } 3737 3738 #undef __FUNCT__ 3739 #define __FUNCT__ "MatCopy" 3740 /*@ 3741 MatCopy - Copys a matrix to another matrix. 3742 3743 Collective on Mat 3744 3745 Input Parameters: 3746 + A - the matrix 3747 - str - SAME_NONZERO_PATTERN or DIFFERENT_NONZERO_PATTERN 3748 3749 Output Parameter: 3750 . B - where the copy is put 3751 3752 Notes: 3753 If you use SAME_NONZERO_PATTERN then the two matrices had better have the 3754 same nonzero pattern or the routine will crash. 3755 3756 MatCopy() copies the matrix entries of a matrix to another existing 3757 matrix (after first zeroing the second matrix). A related routine is 3758 MatConvert(), which first creates a new matrix and then copies the data. 3759 3760 Level: intermediate 3761 3762 Concepts: matrices^copying 3763 3764 .seealso: MatConvert(), MatDuplicate() 3765 3766 @*/ 3767 PetscErrorCode MatCopy(Mat A,Mat B,MatStructure str) 3768 { 3769 PetscErrorCode ierr; 3770 PetscInt i; 3771 3772 PetscFunctionBegin; 3773 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 3774 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 3775 PetscValidType(A,1); 3776 PetscValidType(B,2); 3777 PetscCheckSameComm(A,1,B,2); 3778 MatCheckPreallocated(B,2); 3779 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3780 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3781 if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim (%D,%D) (%D,%D)",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N); 3782 MatCheckPreallocated(A,1); 3783 3784 ierr = PetscLogEventBegin(MAT_Copy,A,B,0,0);CHKERRQ(ierr); 3785 if (A->ops->copy) { 3786 ierr = (*A->ops->copy)(A,B,str);CHKERRQ(ierr); 3787 } else { /* generic conversion */ 3788 ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr); 3789 } 3790 3791 B->stencil.dim = A->stencil.dim; 3792 B->stencil.noc = A->stencil.noc; 3793 for (i=0; i<=A->stencil.dim; i++) { 3794 B->stencil.dims[i] = A->stencil.dims[i]; 3795 B->stencil.starts[i] = A->stencil.starts[i]; 3796 } 3797 3798 ierr = PetscLogEventEnd(MAT_Copy,A,B,0,0);CHKERRQ(ierr); 3799 ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr); 3800 PetscFunctionReturn(0); 3801 } 3802 3803 #undef __FUNCT__ 3804 #define __FUNCT__ "MatConvert" 3805 /*@C 3806 MatConvert - Converts a matrix to another matrix, either of the same 3807 or different type. 3808 3809 Collective on Mat 3810 3811 Input Parameters: 3812 + mat - the matrix 3813 . newtype - new matrix type. Use MATSAME to create a new matrix of the 3814 same type as the original matrix. 3815 - reuse - denotes if the destination matrix is to be created or reused. Currently 3816 MAT_REUSE_MATRIX is only supported for inplace conversion, otherwise use 3817 MAT_INITIAL_MATRIX. 3818 3819 Output Parameter: 3820 . M - pointer to place new matrix 3821 3822 Notes: 3823 MatConvert() first creates a new matrix and then copies the data from 3824 the first matrix. A related routine is MatCopy(), which copies the matrix 3825 entries of one matrix to another already existing matrix context. 3826 3827 Cannot be used to convert a sequential matrix to parallel or parallel to sequential, 3828 the MPI communicator of the generated matrix is always the same as the communicator 3829 of the input matrix. 3830 3831 Level: intermediate 3832 3833 Concepts: matrices^converting between storage formats 3834 3835 .seealso: MatCopy(), MatDuplicate() 3836 @*/ 3837 PetscErrorCode MatConvert(Mat mat, MatType newtype,MatReuse reuse,Mat *M) 3838 { 3839 PetscErrorCode ierr; 3840 PetscBool sametype,issame,flg; 3841 char convname[256],mtype[256]; 3842 Mat B; 3843 3844 PetscFunctionBegin; 3845 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3846 PetscValidType(mat,1); 3847 PetscValidPointer(M,3); 3848 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3849 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3850 MatCheckPreallocated(mat,1); 3851 ierr = MatSetOption(mat,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_FALSE);CHKERRQ(ierr); 3852 3853 ierr = PetscOptionsGetString(((PetscObject)mat)->prefix,"-matconvert_type",mtype,256,&flg);CHKERRQ(ierr); 3854 if (flg) { 3855 newtype = mtype; 3856 } 3857 ierr = PetscObjectTypeCompare((PetscObject)mat,newtype,&sametype);CHKERRQ(ierr); 3858 ierr = PetscStrcmp(newtype,"same",&issame);CHKERRQ(ierr); 3859 if ((reuse == MAT_REUSE_MATRIX) && (mat != *M)) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"MAT_REUSE_MATRIX only supported for in-place conversion currently"); 3860 3861 if ((reuse == MAT_REUSE_MATRIX) && (issame || sametype)) PetscFunctionReturn(0); 3862 3863 if ((sametype || issame) && (reuse==MAT_INITIAL_MATRIX) && mat->ops->duplicate) { 3864 ierr = (*mat->ops->duplicate)(mat,MAT_COPY_VALUES,M);CHKERRQ(ierr); 3865 } else { 3866 PetscErrorCode (*conv)(Mat, MatType,MatReuse,Mat*)=NULL; 3867 const char *prefix[3] = {"seq","mpi",""}; 3868 PetscInt i; 3869 /* 3870 Order of precedence: 3871 1) See if a specialized converter is known to the current matrix. 3872 2) See if a specialized converter is known to the desired matrix class. 3873 3) See if a good general converter is registered for the desired class 3874 (as of 6/27/03 only MATMPIADJ falls into this category). 3875 4) See if a good general converter is known for the current matrix. 3876 5) Use a really basic converter. 3877 */ 3878 3879 /* 1) See if a specialized converter is known to the current matrix and the desired class */ 3880 for (i=0; i<3; i++) { 3881 ierr = PetscStrcpy(convname,"MatConvert_");CHKERRQ(ierr); 3882 ierr = PetscStrcat(convname,((PetscObject)mat)->type_name);CHKERRQ(ierr); 3883 ierr = PetscStrcat(convname,"_");CHKERRQ(ierr); 3884 ierr = PetscStrcat(convname,prefix[i]);CHKERRQ(ierr); 3885 ierr = PetscStrcat(convname,issame ? ((PetscObject)mat)->type_name : newtype);CHKERRQ(ierr); 3886 ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr); 3887 ierr = PetscObjectQueryFunction((PetscObject)mat,convname,&conv);CHKERRQ(ierr); 3888 if (conv) goto foundconv; 3889 } 3890 3891 /* 2) See if a specialized converter is known to the desired matrix class. */ 3892 ierr = MatCreate(PetscObjectComm((PetscObject)mat),&B);CHKERRQ(ierr); 3893 ierr = MatSetSizes(B,mat->rmap->n,mat->cmap->n,mat->rmap->N,mat->cmap->N);CHKERRQ(ierr); 3894 ierr = MatSetType(B,newtype);CHKERRQ(ierr); 3895 for (i=0; i<3; i++) { 3896 ierr = PetscStrcpy(convname,"MatConvert_");CHKERRQ(ierr); 3897 ierr = PetscStrcat(convname,((PetscObject)mat)->type_name);CHKERRQ(ierr); 3898 ierr = PetscStrcat(convname,"_");CHKERRQ(ierr); 3899 ierr = PetscStrcat(convname,prefix[i]);CHKERRQ(ierr); 3900 ierr = PetscStrcat(convname,newtype);CHKERRQ(ierr); 3901 ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr); 3902 ierr = PetscObjectQueryFunction((PetscObject)B,convname,&conv);CHKERRQ(ierr); 3903 if (conv) { 3904 ierr = MatDestroy(&B);CHKERRQ(ierr); 3905 goto foundconv; 3906 } 3907 } 3908 3909 /* 3) See if a good general converter is registered for the desired class */ 3910 conv = B->ops->convertfrom; 3911 ierr = MatDestroy(&B);CHKERRQ(ierr); 3912 if (conv) goto foundconv; 3913 3914 /* 4) See if a good general converter is known for the current matrix */ 3915 if (mat->ops->convert) { 3916 conv = mat->ops->convert; 3917 } 3918 if (conv) goto foundconv; 3919 3920 /* 5) Use a really basic converter. */ 3921 conv = MatConvert_Basic; 3922 3923 foundconv: 3924 ierr = PetscLogEventBegin(MAT_Convert,mat,0,0,0);CHKERRQ(ierr); 3925 ierr = (*conv)(mat,newtype,reuse,M);CHKERRQ(ierr); 3926 ierr = PetscLogEventEnd(MAT_Convert,mat,0,0,0);CHKERRQ(ierr); 3927 } 3928 ierr = PetscObjectStateIncrease((PetscObject)*M);CHKERRQ(ierr); 3929 3930 /* Copy Mat options */ 3931 if (mat->symmetric) {ierr = MatSetOption(*M,MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr);} 3932 if (mat->hermitian) {ierr = MatSetOption(*M,MAT_HERMITIAN,PETSC_TRUE);CHKERRQ(ierr);} 3933 PetscFunctionReturn(0); 3934 } 3935 3936 #undef __FUNCT__ 3937 #define __FUNCT__ "MatFactorGetSolverPackage" 3938 /*@C 3939 MatFactorGetSolverPackage - Returns name of the package providing the factorization routines 3940 3941 Not Collective 3942 3943 Input Parameter: 3944 . mat - the matrix, must be a factored matrix 3945 3946 Output Parameter: 3947 . type - the string name of the package (do not free this string) 3948 3949 Notes: 3950 In Fortran you pass in a empty string and the package name will be copied into it. 3951 (Make sure the string is long enough) 3952 3953 Level: intermediate 3954 3955 .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor() 3956 @*/ 3957 PetscErrorCode MatFactorGetSolverPackage(Mat mat, const MatSolverPackage *type) 3958 { 3959 PetscErrorCode ierr, (*conv)(Mat,const MatSolverPackage*); 3960 3961 PetscFunctionBegin; 3962 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3963 PetscValidType(mat,1); 3964 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix"); 3965 ierr = PetscObjectQueryFunction((PetscObject)mat,"MatFactorGetSolverPackage_C",&conv);CHKERRQ(ierr); 3966 if (!conv) { 3967 *type = MATSOLVERPETSC; 3968 } else { 3969 ierr = (*conv)(mat,type);CHKERRQ(ierr); 3970 } 3971 PetscFunctionReturn(0); 3972 } 3973 3974 #undef __FUNCT__ 3975 #define __FUNCT__ "MatGetFactor" 3976 /*@C 3977 MatGetFactor - Returns a matrix suitable to calls to MatXXFactorSymbolic() 3978 3979 Collective on Mat 3980 3981 Input Parameters: 3982 + mat - the matrix 3983 . type - name of solver type, for example, superlu, petsc (to use PETSc's default) 3984 - ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU, 3985 3986 Output Parameters: 3987 . f - the factor matrix used with MatXXFactorSymbolic() calls 3988 3989 Notes: 3990 Some PETSc matrix formats have alternative solvers available that are contained in alternative packages 3991 such as pastix, superlu, mumps etc. 3992 3993 PETSc must have been ./configure to use the external solver, using the option --download-package 3994 3995 Level: intermediate 3996 3997 .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable() 3998 @*/ 3999 PetscErrorCode MatGetFactor(Mat mat, const MatSolverPackage type,MatFactorType ftype,Mat *f) 4000 { 4001 PetscErrorCode ierr,(*conv)(Mat,MatFactorType,Mat*); 4002 char convname[256]; 4003 4004 PetscFunctionBegin; 4005 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4006 PetscValidType(mat,1); 4007 4008 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4009 MatCheckPreallocated(mat,1); 4010 4011 ierr = PetscStrcpy(convname,"MatGetFactor_");CHKERRQ(ierr); 4012 ierr = PetscStrcat(convname,type);CHKERRQ(ierr); 4013 ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr); 4014 ierr = PetscObjectQueryFunction((PetscObject)mat,convname,&conv);CHKERRQ(ierr); 4015 if (!conv) { 4016 PetscBool flag; 4017 MPI_Comm comm; 4018 4019 ierr = PetscObjectGetComm((PetscObject)mat,&comm);CHKERRQ(ierr); 4020 ierr = PetscStrcasecmp(MATSOLVERPETSC,type,&flag);CHKERRQ(ierr); 4021 if (flag) SETERRQ2(comm,PETSC_ERR_SUP,"Matrix format %s does not have a built-in PETSc %s",((PetscObject)mat)->type_name,MatFactorTypes[ftype]); 4022 else SETERRQ4(comm,PETSC_ERR_SUP,"Matrix format %s does not have a solver package %s for %s. Perhaps you must ./configure with --download-%s",((PetscObject)mat)->type_name,type,MatFactorTypes[ftype],type); 4023 } 4024 ierr = (*conv)(mat,ftype,f);CHKERRQ(ierr); 4025 PetscFunctionReturn(0); 4026 } 4027 4028 #undef __FUNCT__ 4029 #define __FUNCT__ "MatGetFactorAvailable" 4030 /*@C 4031 MatGetFactorAvailable - Returns a a flag if matrix supports particular package and factor type 4032 4033 Not Collective 4034 4035 Input Parameters: 4036 + mat - the matrix 4037 . type - name of solver type, for example, superlu, petsc (to use PETSc's default) 4038 - ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU, 4039 4040 Output Parameter: 4041 . flg - PETSC_TRUE if the factorization is available 4042 4043 Notes: 4044 Some PETSc matrix formats have alternative solvers available that are contained in alternative packages 4045 such as pastix, superlu, mumps etc. 4046 4047 PETSc must have been ./configure to use the external solver, using the option --download-package 4048 4049 Level: intermediate 4050 4051 .seealso: MatCopy(), MatDuplicate(), MatGetFactor() 4052 @*/ 4053 PetscErrorCode MatGetFactorAvailable(Mat mat, const MatSolverPackage type,MatFactorType ftype,PetscBool *flg) 4054 { 4055 PetscErrorCode ierr, (*conv)(Mat,MatFactorType,PetscBool*); 4056 char convname[256]; 4057 4058 PetscFunctionBegin; 4059 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4060 PetscValidType(mat,1); 4061 4062 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4063 MatCheckPreallocated(mat,1); 4064 4065 ierr = PetscStrcpy(convname,"MatGetFactorAvailable_");CHKERRQ(ierr); 4066 ierr = PetscStrcat(convname,type);CHKERRQ(ierr); 4067 ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr); 4068 ierr = PetscObjectQueryFunction((PetscObject)mat,convname,&conv);CHKERRQ(ierr); 4069 if (!conv) { 4070 *flg = PETSC_FALSE; 4071 } else { 4072 ierr = (*conv)(mat,ftype,flg);CHKERRQ(ierr); 4073 } 4074 PetscFunctionReturn(0); 4075 } 4076 4077 4078 #undef __FUNCT__ 4079 #define __FUNCT__ "MatDuplicate" 4080 /*@ 4081 MatDuplicate - Duplicates a matrix including the non-zero structure. 4082 4083 Collective on Mat 4084 4085 Input Parameters: 4086 + mat - the matrix 4087 - op - either MAT_DO_NOT_COPY_VALUES or MAT_COPY_VALUES, cause it to copy the numerical values in the matrix 4088 MAT_SHARE_NONZERO_PATTERN to share the nonzero patterns with the previous matrix and not copy them. 4089 4090 Output Parameter: 4091 . M - pointer to place new matrix 4092 4093 Level: intermediate 4094 4095 Concepts: matrices^duplicating 4096 4097 Notes: You cannot change the nonzero pattern for the parent or child matrix if you use MAT_SHARE_NONZERO_PATTERN. 4098 4099 .seealso: MatCopy(), MatConvert() 4100 @*/ 4101 PetscErrorCode MatDuplicate(Mat mat,MatDuplicateOption op,Mat *M) 4102 { 4103 PetscErrorCode ierr; 4104 Mat B; 4105 PetscInt i; 4106 4107 PetscFunctionBegin; 4108 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4109 PetscValidType(mat,1); 4110 PetscValidPointer(M,3); 4111 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4112 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4113 MatCheckPreallocated(mat,1); 4114 4115 *M = 0; 4116 if (!mat->ops->duplicate) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not written for this matrix type"); 4117 ierr = PetscLogEventBegin(MAT_Convert,mat,0,0,0);CHKERRQ(ierr); 4118 ierr = (*mat->ops->duplicate)(mat,op,M);CHKERRQ(ierr); 4119 B = *M; 4120 4121 B->stencil.dim = mat->stencil.dim; 4122 B->stencil.noc = mat->stencil.noc; 4123 for (i=0; i<=mat->stencil.dim; i++) { 4124 B->stencil.dims[i] = mat->stencil.dims[i]; 4125 B->stencil.starts[i] = mat->stencil.starts[i]; 4126 } 4127 4128 B->nooffproczerorows = mat->nooffproczerorows; 4129 B->nooffprocentries = mat->nooffprocentries; 4130 4131 ierr = PetscLogEventEnd(MAT_Convert,mat,0,0,0);CHKERRQ(ierr); 4132 ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr); 4133 PetscFunctionReturn(0); 4134 } 4135 4136 #undef __FUNCT__ 4137 #define __FUNCT__ "MatGetDiagonal" 4138 /*@ 4139 MatGetDiagonal - Gets the diagonal of a matrix. 4140 4141 Logically Collective on Mat and Vec 4142 4143 Input Parameters: 4144 + mat - the matrix 4145 - v - the vector for storing the diagonal 4146 4147 Output Parameter: 4148 . v - the diagonal of the matrix 4149 4150 Level: intermediate 4151 4152 Note: 4153 Currently only correct in parallel for square matrices. 4154 4155 Concepts: matrices^accessing diagonals 4156 4157 .seealso: MatGetRow(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs() 4158 @*/ 4159 PetscErrorCode MatGetDiagonal(Mat mat,Vec v) 4160 { 4161 PetscErrorCode ierr; 4162 4163 PetscFunctionBegin; 4164 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4165 PetscValidType(mat,1); 4166 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4167 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4168 if (!mat->ops->getdiagonal) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4169 MatCheckPreallocated(mat,1); 4170 4171 ierr = (*mat->ops->getdiagonal)(mat,v);CHKERRQ(ierr); 4172 ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr); 4173 PetscFunctionReturn(0); 4174 } 4175 4176 #undef __FUNCT__ 4177 #define __FUNCT__ "MatGetRowMin" 4178 /*@ 4179 MatGetRowMin - Gets the minimum value (of the real part) of each 4180 row of the matrix 4181 4182 Logically Collective on Mat and Vec 4183 4184 Input Parameters: 4185 . mat - the matrix 4186 4187 Output Parameter: 4188 + v - the vector for storing the maximums 4189 - idx - the indices of the column found for each row (optional) 4190 4191 Level: intermediate 4192 4193 Notes: The result of this call are the same as if one converted the matrix to dense format 4194 and found the minimum value in each row (i.e. the implicit zeros are counted as zeros). 4195 4196 This code is only implemented for a couple of matrix formats. 4197 4198 Concepts: matrices^getting row maximums 4199 4200 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs(), 4201 MatGetRowMax() 4202 @*/ 4203 PetscErrorCode MatGetRowMin(Mat mat,Vec v,PetscInt idx[]) 4204 { 4205 PetscErrorCode ierr; 4206 4207 PetscFunctionBegin; 4208 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4209 PetscValidType(mat,1); 4210 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4211 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4212 if (!mat->ops->getrowmax) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4213 MatCheckPreallocated(mat,1); 4214 4215 ierr = (*mat->ops->getrowmin)(mat,v,idx);CHKERRQ(ierr); 4216 ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr); 4217 PetscFunctionReturn(0); 4218 } 4219 4220 #undef __FUNCT__ 4221 #define __FUNCT__ "MatGetRowMinAbs" 4222 /*@ 4223 MatGetRowMinAbs - Gets the minimum value (in absolute value) of each 4224 row of the matrix 4225 4226 Logically Collective on Mat and Vec 4227 4228 Input Parameters: 4229 . mat - the matrix 4230 4231 Output Parameter: 4232 + v - the vector for storing the minimums 4233 - idx - the indices of the column found for each row (or NULL if not needed) 4234 4235 Level: intermediate 4236 4237 Notes: if a row is completely empty or has only 0.0 values then the idx[] value for that 4238 row is 0 (the first column). 4239 4240 This code is only implemented for a couple of matrix formats. 4241 4242 Concepts: matrices^getting row maximums 4243 4244 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMaxAbs(), MatGetRowMin() 4245 @*/ 4246 PetscErrorCode MatGetRowMinAbs(Mat mat,Vec v,PetscInt idx[]) 4247 { 4248 PetscErrorCode ierr; 4249 4250 PetscFunctionBegin; 4251 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4252 PetscValidType(mat,1); 4253 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4254 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4255 if (!mat->ops->getrowminabs) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4256 MatCheckPreallocated(mat,1); 4257 if (idx) {ierr = PetscMemzero(idx,mat->rmap->n*sizeof(PetscInt));CHKERRQ(ierr);} 4258 4259 ierr = (*mat->ops->getrowminabs)(mat,v,idx);CHKERRQ(ierr); 4260 ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr); 4261 PetscFunctionReturn(0); 4262 } 4263 4264 #undef __FUNCT__ 4265 #define __FUNCT__ "MatGetRowMax" 4266 /*@ 4267 MatGetRowMax - Gets the maximum value (of the real part) of each 4268 row of the matrix 4269 4270 Logically Collective on Mat and Vec 4271 4272 Input Parameters: 4273 . mat - the matrix 4274 4275 Output Parameter: 4276 + v - the vector for storing the maximums 4277 - idx - the indices of the column found for each row (optional) 4278 4279 Level: intermediate 4280 4281 Notes: The result of this call are the same as if one converted the matrix to dense format 4282 and found the minimum value in each row (i.e. the implicit zeros are counted as zeros). 4283 4284 This code is only implemented for a couple of matrix formats. 4285 4286 Concepts: matrices^getting row maximums 4287 4288 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs(), MatGetRowMin() 4289 @*/ 4290 PetscErrorCode MatGetRowMax(Mat mat,Vec v,PetscInt idx[]) 4291 { 4292 PetscErrorCode ierr; 4293 4294 PetscFunctionBegin; 4295 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4296 PetscValidType(mat,1); 4297 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4298 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4299 if (!mat->ops->getrowmax) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4300 MatCheckPreallocated(mat,1); 4301 4302 ierr = (*mat->ops->getrowmax)(mat,v,idx);CHKERRQ(ierr); 4303 ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr); 4304 PetscFunctionReturn(0); 4305 } 4306 4307 #undef __FUNCT__ 4308 #define __FUNCT__ "MatGetRowMaxAbs" 4309 /*@ 4310 MatGetRowMaxAbs - Gets the maximum value (in absolute value) of each 4311 row of the matrix 4312 4313 Logically Collective on Mat and Vec 4314 4315 Input Parameters: 4316 . mat - the matrix 4317 4318 Output Parameter: 4319 + v - the vector for storing the maximums 4320 - idx - the indices of the column found for each row (or NULL if not needed) 4321 4322 Level: intermediate 4323 4324 Notes: if a row is completely empty or has only 0.0 values then the idx[] value for that 4325 row is 0 (the first column). 4326 4327 This code is only implemented for a couple of matrix formats. 4328 4329 Concepts: matrices^getting row maximums 4330 4331 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMin() 4332 @*/ 4333 PetscErrorCode MatGetRowMaxAbs(Mat mat,Vec v,PetscInt idx[]) 4334 { 4335 PetscErrorCode ierr; 4336 4337 PetscFunctionBegin; 4338 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4339 PetscValidType(mat,1); 4340 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4341 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4342 if (!mat->ops->getrowmaxabs) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4343 MatCheckPreallocated(mat,1); 4344 if (idx) {ierr = PetscMemzero(idx,mat->rmap->n*sizeof(PetscInt));CHKERRQ(ierr);} 4345 4346 ierr = (*mat->ops->getrowmaxabs)(mat,v,idx);CHKERRQ(ierr); 4347 ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr); 4348 PetscFunctionReturn(0); 4349 } 4350 4351 #undef __FUNCT__ 4352 #define __FUNCT__ "MatGetRowSum" 4353 /*@ 4354 MatGetRowSum - Gets the sum of each row of the matrix 4355 4356 Logically Collective on Mat and Vec 4357 4358 Input Parameters: 4359 . mat - the matrix 4360 4361 Output Parameter: 4362 . v - the vector for storing the sum of rows 4363 4364 Level: intermediate 4365 4366 Notes: This code is slow since it is not currently specialized for different formats 4367 4368 Concepts: matrices^getting row sums 4369 4370 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMin() 4371 @*/ 4372 PetscErrorCode MatGetRowSum(Mat mat, Vec v) 4373 { 4374 PetscInt start = 0, end = 0, row; 4375 PetscScalar *array; 4376 PetscErrorCode ierr; 4377 4378 PetscFunctionBegin; 4379 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4380 PetscValidType(mat,1); 4381 PetscValidHeaderSpecific(v,VEC_CLASSID,2); 4382 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4383 MatCheckPreallocated(mat,1); 4384 ierr = MatGetOwnershipRange(mat, &start, &end);CHKERRQ(ierr); 4385 ierr = VecGetArray(v, &array);CHKERRQ(ierr); 4386 for (row = start; row < end; ++row) { 4387 PetscInt ncols, col; 4388 const PetscInt *cols; 4389 const PetscScalar *vals; 4390 4391 array[row - start] = 0.0; 4392 4393 ierr = MatGetRow(mat, row, &ncols, &cols, &vals);CHKERRQ(ierr); 4394 for (col = 0; col < ncols; col++) { 4395 array[row - start] += vals[col]; 4396 } 4397 ierr = MatRestoreRow(mat, row, &ncols, &cols, &vals);CHKERRQ(ierr); 4398 } 4399 ierr = VecRestoreArray(v, &array);CHKERRQ(ierr); 4400 ierr = PetscObjectStateIncrease((PetscObject) v);CHKERRQ(ierr); 4401 PetscFunctionReturn(0); 4402 } 4403 4404 #undef __FUNCT__ 4405 #define __FUNCT__ "MatTranspose" 4406 /*@ 4407 MatTranspose - Computes an in-place or out-of-place transpose of a matrix. 4408 4409 Collective on Mat 4410 4411 Input Parameter: 4412 + mat - the matrix to transpose 4413 - reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 4414 4415 Output Parameters: 4416 . B - the transpose 4417 4418 Notes: 4419 If you pass in &mat for B the transpose will be done in place, for example MatTranspose(mat,MAT_REUSE_MATRIX,&mat); 4420 4421 Consider using MatCreateTranspose() instead if you only need a matrix that behaves like the transpose, but don't need the storage to be changed. 4422 4423 Level: intermediate 4424 4425 Concepts: matrices^transposing 4426 4427 .seealso: MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse 4428 @*/ 4429 PetscErrorCode MatTranspose(Mat mat,MatReuse reuse,Mat *B) 4430 { 4431 PetscErrorCode ierr; 4432 4433 PetscFunctionBegin; 4434 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4435 PetscValidType(mat,1); 4436 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4437 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4438 if (!mat->ops->transpose) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4439 MatCheckPreallocated(mat,1); 4440 4441 ierr = PetscLogEventBegin(MAT_Transpose,mat,0,0,0);CHKERRQ(ierr); 4442 ierr = (*mat->ops->transpose)(mat,reuse,B);CHKERRQ(ierr); 4443 ierr = PetscLogEventEnd(MAT_Transpose,mat,0,0,0);CHKERRQ(ierr); 4444 if (B) {ierr = PetscObjectStateIncrease((PetscObject)*B);CHKERRQ(ierr);} 4445 PetscFunctionReturn(0); 4446 } 4447 4448 #undef __FUNCT__ 4449 #define __FUNCT__ "MatIsTranspose" 4450 /*@ 4451 MatIsTranspose - Test whether a matrix is another one's transpose, 4452 or its own, in which case it tests symmetry. 4453 4454 Collective on Mat 4455 4456 Input Parameter: 4457 + A - the matrix to test 4458 - B - the matrix to test against, this can equal the first parameter 4459 4460 Output Parameters: 4461 . flg - the result 4462 4463 Notes: 4464 Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm 4465 has a running time of the order of the number of nonzeros; the parallel 4466 test involves parallel copies of the block-offdiagonal parts of the matrix. 4467 4468 Level: intermediate 4469 4470 Concepts: matrices^transposing, matrix^symmetry 4471 4472 .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian() 4473 @*/ 4474 PetscErrorCode MatIsTranspose(Mat A,Mat B,PetscReal tol,PetscBool *flg) 4475 { 4476 PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscBool*),(*g)(Mat,Mat,PetscReal,PetscBool*); 4477 4478 PetscFunctionBegin; 4479 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 4480 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 4481 PetscValidPointer(flg,3); 4482 ierr = PetscObjectQueryFunction((PetscObject)A,"MatIsTranspose_C",&f);CHKERRQ(ierr); 4483 ierr = PetscObjectQueryFunction((PetscObject)B,"MatIsTranspose_C",&g);CHKERRQ(ierr); 4484 *flg = PETSC_FALSE; 4485 if (f && g) { 4486 if (f == g) { 4487 ierr = (*f)(A,B,tol,flg);CHKERRQ(ierr); 4488 } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for symmetry test"); 4489 } else { 4490 MatType mattype; 4491 if (!f) { 4492 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 4493 } else { 4494 ierr = MatGetType(B,&mattype);CHKERRQ(ierr); 4495 } 4496 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for transpose",mattype); 4497 } 4498 PetscFunctionReturn(0); 4499 } 4500 4501 #undef __FUNCT__ 4502 #define __FUNCT__ "MatHermitianTranspose" 4503 /*@ 4504 MatHermitianTranspose - Computes an in-place or out-of-place transpose of a matrix in complex conjugate. 4505 4506 Collective on Mat 4507 4508 Input Parameter: 4509 + mat - the matrix to transpose and complex conjugate 4510 - reuse - store the transpose matrix in the provided B 4511 4512 Output Parameters: 4513 . B - the Hermitian 4514 4515 Notes: 4516 If you pass in &mat for B the Hermitian will be done in place 4517 4518 Level: intermediate 4519 4520 Concepts: matrices^transposing, complex conjugatex 4521 4522 .seealso: MatTranspose(), MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse 4523 @*/ 4524 PetscErrorCode MatHermitianTranspose(Mat mat,MatReuse reuse,Mat *B) 4525 { 4526 PetscErrorCode ierr; 4527 4528 PetscFunctionBegin; 4529 ierr = MatTranspose(mat,reuse,B);CHKERRQ(ierr); 4530 #if defined(PETSC_USE_COMPLEX) 4531 ierr = MatConjugate(*B);CHKERRQ(ierr); 4532 #endif 4533 PetscFunctionReturn(0); 4534 } 4535 4536 #undef __FUNCT__ 4537 #define __FUNCT__ "MatIsHermitianTranspose" 4538 /*@ 4539 MatIsHermitianTranspose - Test whether a matrix is another one's Hermitian transpose, 4540 4541 Collective on Mat 4542 4543 Input Parameter: 4544 + A - the matrix to test 4545 - B - the matrix to test against, this can equal the first parameter 4546 4547 Output Parameters: 4548 . flg - the result 4549 4550 Notes: 4551 Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm 4552 has a running time of the order of the number of nonzeros; the parallel 4553 test involves parallel copies of the block-offdiagonal parts of the matrix. 4554 4555 Level: intermediate 4556 4557 Concepts: matrices^transposing, matrix^symmetry 4558 4559 .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian(), MatIsTranspose() 4560 @*/ 4561 PetscErrorCode MatIsHermitianTranspose(Mat A,Mat B,PetscReal tol,PetscBool *flg) 4562 { 4563 PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscBool*),(*g)(Mat,Mat,PetscReal,PetscBool*); 4564 4565 PetscFunctionBegin; 4566 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 4567 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 4568 PetscValidPointer(flg,3); 4569 ierr = PetscObjectQueryFunction((PetscObject)A,"MatIsHermitianTranspose_C",&f);CHKERRQ(ierr); 4570 ierr = PetscObjectQueryFunction((PetscObject)B,"MatIsHermitianTranspose_C",&g);CHKERRQ(ierr); 4571 if (f && g) { 4572 if (f==g) { 4573 ierr = (*f)(A,B,tol,flg);CHKERRQ(ierr); 4574 } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for Hermitian test"); 4575 } 4576 PetscFunctionReturn(0); 4577 } 4578 4579 #undef __FUNCT__ 4580 #define __FUNCT__ "MatPermute" 4581 /*@ 4582 MatPermute - Creates a new matrix with rows and columns permuted from the 4583 original. 4584 4585 Collective on Mat 4586 4587 Input Parameters: 4588 + mat - the matrix to permute 4589 . row - row permutation, each processor supplies only the permutation for its rows 4590 - col - column permutation, each processor supplies only the permutation for its columns 4591 4592 Output Parameters: 4593 . B - the permuted matrix 4594 4595 Level: advanced 4596 4597 Note: 4598 The index sets map from row/col of permuted matrix to row/col of original matrix. 4599 The index sets should be on the same communicator as Mat and have the same local sizes. 4600 4601 Concepts: matrices^permuting 4602 4603 .seealso: MatGetOrdering(), ISAllGather() 4604 4605 @*/ 4606 PetscErrorCode MatPermute(Mat mat,IS row,IS col,Mat *B) 4607 { 4608 PetscErrorCode ierr; 4609 4610 PetscFunctionBegin; 4611 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4612 PetscValidType(mat,1); 4613 PetscValidHeaderSpecific(row,IS_CLASSID,2); 4614 PetscValidHeaderSpecific(col,IS_CLASSID,3); 4615 PetscValidPointer(B,4); 4616 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4617 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4618 if (!mat->ops->permute) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatPermute not available for Mat type %s",((PetscObject)mat)->type_name); 4619 MatCheckPreallocated(mat,1); 4620 4621 ierr = (*mat->ops->permute)(mat,row,col,B);CHKERRQ(ierr); 4622 ierr = PetscObjectStateIncrease((PetscObject)*B);CHKERRQ(ierr); 4623 PetscFunctionReturn(0); 4624 } 4625 4626 #undef __FUNCT__ 4627 #define __FUNCT__ "MatEqual" 4628 /*@ 4629 MatEqual - Compares two matrices. 4630 4631 Collective on Mat 4632 4633 Input Parameters: 4634 + A - the first matrix 4635 - B - the second matrix 4636 4637 Output Parameter: 4638 . flg - PETSC_TRUE if the matrices are equal; PETSC_FALSE otherwise. 4639 4640 Level: intermediate 4641 4642 Concepts: matrices^equality between 4643 @*/ 4644 PetscErrorCode MatEqual(Mat A,Mat B,PetscBool *flg) 4645 { 4646 PetscErrorCode ierr; 4647 4648 PetscFunctionBegin; 4649 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 4650 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 4651 PetscValidType(A,1); 4652 PetscValidType(B,2); 4653 PetscValidIntPointer(flg,3); 4654 PetscCheckSameComm(A,1,B,2); 4655 MatCheckPreallocated(B,2); 4656 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4657 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4658 if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D %D %D",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N); 4659 if (!A->ops->equal) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)A)->type_name); 4660 if (!B->ops->equal) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Mat type %s",((PetscObject)B)->type_name); 4661 if (A->ops->equal != B->ops->equal) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"A is type: %s\nB is type: %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name); 4662 MatCheckPreallocated(A,1); 4663 4664 ierr = (*A->ops->equal)(A,B,flg);CHKERRQ(ierr); 4665 PetscFunctionReturn(0); 4666 } 4667 4668 #undef __FUNCT__ 4669 #define __FUNCT__ "MatDiagonalScale" 4670 /*@ 4671 MatDiagonalScale - Scales a matrix on the left and right by diagonal 4672 matrices that are stored as vectors. Either of the two scaling 4673 matrices can be NULL. 4674 4675 Collective on Mat 4676 4677 Input Parameters: 4678 + mat - the matrix to be scaled 4679 . l - the left scaling vector (or NULL) 4680 - r - the right scaling vector (or NULL) 4681 4682 Notes: 4683 MatDiagonalScale() computes A = LAR, where 4684 L = a diagonal matrix (stored as a vector), R = a diagonal matrix (stored as a vector) 4685 The L scales the rows of the matrix, the R scales the columns of the matrix. 4686 4687 Level: intermediate 4688 4689 Concepts: matrices^diagonal scaling 4690 Concepts: diagonal scaling of matrices 4691 4692 .seealso: MatScale() 4693 @*/ 4694 PetscErrorCode MatDiagonalScale(Mat mat,Vec l,Vec r) 4695 { 4696 PetscErrorCode ierr; 4697 4698 PetscFunctionBegin; 4699 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4700 PetscValidType(mat,1); 4701 if (!mat->ops->diagonalscale) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4702 if (l) {PetscValidHeaderSpecific(l,VEC_CLASSID,2);PetscCheckSameComm(mat,1,l,2);} 4703 if (r) {PetscValidHeaderSpecific(r,VEC_CLASSID,3);PetscCheckSameComm(mat,1,r,3);} 4704 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4705 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4706 MatCheckPreallocated(mat,1); 4707 4708 ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 4709 ierr = (*mat->ops->diagonalscale)(mat,l,r);CHKERRQ(ierr); 4710 ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 4711 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 4712 #if defined(PETSC_HAVE_CUSP) 4713 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 4714 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 4715 } 4716 #endif 4717 #if defined(PETSC_HAVE_VIENNACL) 4718 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 4719 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 4720 } 4721 #endif 4722 PetscFunctionReturn(0); 4723 } 4724 4725 #undef __FUNCT__ 4726 #define __FUNCT__ "MatScale" 4727 /*@ 4728 MatScale - Scales all elements of a matrix by a given number. 4729 4730 Logically Collective on Mat 4731 4732 Input Parameters: 4733 + mat - the matrix to be scaled 4734 - a - the scaling value 4735 4736 Output Parameter: 4737 . mat - the scaled matrix 4738 4739 Level: intermediate 4740 4741 Concepts: matrices^scaling all entries 4742 4743 .seealso: MatDiagonalScale() 4744 @*/ 4745 PetscErrorCode MatScale(Mat mat,PetscScalar a) 4746 { 4747 PetscErrorCode ierr; 4748 4749 PetscFunctionBegin; 4750 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4751 PetscValidType(mat,1); 4752 if (a != (PetscScalar)1.0 && !mat->ops->scale) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4753 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4754 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4755 PetscValidLogicalCollectiveScalar(mat,a,2); 4756 MatCheckPreallocated(mat,1); 4757 4758 ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 4759 if (a != (PetscScalar)1.0) { 4760 ierr = (*mat->ops->scale)(mat,a);CHKERRQ(ierr); 4761 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 4762 } 4763 ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 4764 #if defined(PETSC_HAVE_CUSP) 4765 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 4766 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 4767 } 4768 #endif 4769 #if defined(PETSC_HAVE_VIENNACL) 4770 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 4771 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 4772 } 4773 #endif 4774 PetscFunctionReturn(0); 4775 } 4776 4777 #undef __FUNCT__ 4778 #define __FUNCT__ "MatNorm" 4779 /*@ 4780 MatNorm - Calculates various norms of a matrix. 4781 4782 Collective on Mat 4783 4784 Input Parameters: 4785 + mat - the matrix 4786 - type - the type of norm, NORM_1, NORM_FROBENIUS, NORM_INFINITY 4787 4788 Output Parameters: 4789 . nrm - the resulting norm 4790 4791 Level: intermediate 4792 4793 Concepts: matrices^norm 4794 Concepts: norm^of matrix 4795 @*/ 4796 PetscErrorCode MatNorm(Mat mat,NormType type,PetscReal *nrm) 4797 { 4798 PetscErrorCode ierr; 4799 4800 PetscFunctionBegin; 4801 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4802 PetscValidType(mat,1); 4803 PetscValidScalarPointer(nrm,3); 4804 4805 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 4806 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 4807 if (!mat->ops->norm) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 4808 MatCheckPreallocated(mat,1); 4809 4810 ierr = (*mat->ops->norm)(mat,type,nrm);CHKERRQ(ierr); 4811 PetscFunctionReturn(0); 4812 } 4813 4814 /* 4815 This variable is used to prevent counting of MatAssemblyBegin() that 4816 are called from within a MatAssemblyEnd(). 4817 */ 4818 static PetscInt MatAssemblyEnd_InUse = 0; 4819 #undef __FUNCT__ 4820 #define __FUNCT__ "MatAssemblyBegin" 4821 /*@ 4822 MatAssemblyBegin - Begins assembling the matrix. This routine should 4823 be called after completing all calls to MatSetValues(). 4824 4825 Collective on Mat 4826 4827 Input Parameters: 4828 + mat - the matrix 4829 - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY 4830 4831 Notes: 4832 MatSetValues() generally caches the values. The matrix is ready to 4833 use only after MatAssemblyBegin() and MatAssemblyEnd() have been called. 4834 Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES 4835 in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before 4836 using the matrix. 4837 4838 ALL processes that share a matrix MUST call MatAssemblyBegin() and MatAssemblyEnd() the SAME NUMBER of times, and each time with the 4839 same flag of MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY for all processes. Thus you CANNOT locally change from ADD_VALUES to INSERT_VALUES, that is 4840 a global collective operation requring all processes that share the matrix. 4841 4842 Space for preallocated nonzeros that is not filled by a call to MatSetValues() or a related routine are compressed 4843 out by assembly. If you intend to use that extra space on a subsequent assembly, be sure to insert explicit zeros 4844 before MAT_FINAL_ASSEMBLY so the space is not compressed out. 4845 4846 Level: beginner 4847 4848 Concepts: matrices^assembling 4849 4850 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssembled() 4851 @*/ 4852 PetscErrorCode MatAssemblyBegin(Mat mat,MatAssemblyType type) 4853 { 4854 PetscErrorCode ierr; 4855 4856 PetscFunctionBegin; 4857 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4858 PetscValidType(mat,1); 4859 MatCheckPreallocated(mat,1); 4860 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix.\nDid you forget to call MatSetUnfactored()?"); 4861 if (mat->assembled) { 4862 mat->was_assembled = PETSC_TRUE; 4863 mat->assembled = PETSC_FALSE; 4864 } 4865 if (!MatAssemblyEnd_InUse) { 4866 ierr = PetscLogEventBegin(MAT_AssemblyBegin,mat,0,0,0);CHKERRQ(ierr); 4867 if (mat->ops->assemblybegin) {ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr);} 4868 ierr = PetscLogEventEnd(MAT_AssemblyBegin,mat,0,0,0);CHKERRQ(ierr); 4869 } else if (mat->ops->assemblybegin) { 4870 ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr); 4871 } 4872 PetscFunctionReturn(0); 4873 } 4874 4875 #undef __FUNCT__ 4876 #define __FUNCT__ "MatAssembled" 4877 /*@ 4878 MatAssembled - Indicates if a matrix has been assembled and is ready for 4879 use; for example, in matrix-vector product. 4880 4881 Not Collective 4882 4883 Input Parameter: 4884 . mat - the matrix 4885 4886 Output Parameter: 4887 . assembled - PETSC_TRUE or PETSC_FALSE 4888 4889 Level: advanced 4890 4891 Concepts: matrices^assembled? 4892 4893 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssemblyBegin() 4894 @*/ 4895 PetscErrorCode MatAssembled(Mat mat,PetscBool *assembled) 4896 { 4897 PetscFunctionBegin; 4898 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4899 PetscValidType(mat,1); 4900 PetscValidPointer(assembled,2); 4901 *assembled = mat->assembled; 4902 PetscFunctionReturn(0); 4903 } 4904 4905 #undef __FUNCT__ 4906 #define __FUNCT__ "MatViewFromOptions" 4907 /* 4908 MatViewFromOptions - Processes command line options to determine if/how a matrix is to be viewed. Called from higher level packages. 4909 4910 Collective on Vec 4911 4912 Input Parameters: 4913 + mat - the matrix 4914 . prefix - prefix to use for viewing, or NULL to use prefix of 'rnd' 4915 - optionname - option to activate viewing 4916 4917 Level: intermediate 4918 4919 .keywords: Mat, view, options, database 4920 .seealso: MatViewFromOptions() 4921 */ 4922 PetscErrorCode MatViewFromOptions(Mat mat,const char prefix[],const char optionname[]) 4923 { 4924 PetscErrorCode ierr; 4925 PetscViewer viewer; 4926 PetscBool flg; 4927 static PetscBool incall = PETSC_FALSE; 4928 PetscViewerFormat format; 4929 4930 PetscFunctionBegin; 4931 if (incall) PetscFunctionReturn(0); 4932 incall = PETSC_TRUE; 4933 if (prefix) { 4934 ierr = PetscOptionsGetViewer(PetscObjectComm((PetscObject)mat),prefix,optionname,&viewer,&format,&flg);CHKERRQ(ierr); 4935 } else { 4936 ierr = PetscOptionsGetViewer(PetscObjectComm((PetscObject)mat),((PetscObject)mat)->prefix,optionname,&viewer,&format,&flg);CHKERRQ(ierr); 4937 } 4938 if (flg) { 4939 ierr = PetscViewerPushFormat(viewer,format);CHKERRQ(ierr); 4940 ierr = MatView(mat,viewer);CHKERRQ(ierr); 4941 ierr = PetscViewerPopFormat(viewer);CHKERRQ(ierr); 4942 ierr = PetscViewerDestroy(&viewer);CHKERRQ(ierr); 4943 } 4944 incall = PETSC_FALSE; 4945 PetscFunctionReturn(0); 4946 } 4947 4948 #undef __FUNCT__ 4949 #define __FUNCT__ "MatAssemblyEnd" 4950 /*@ 4951 MatAssemblyEnd - Completes assembling the matrix. This routine should 4952 be called after MatAssemblyBegin(). 4953 4954 Collective on Mat 4955 4956 Input Parameters: 4957 + mat - the matrix 4958 - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY 4959 4960 Options Database Keys: 4961 + -mat_view ::ascii_info - Prints info on matrix at conclusion of MatEndAssembly() 4962 . -mat_view ::ascii_info_detail - Prints more detailed info 4963 . -mat_view - Prints matrix in ASCII format 4964 . -mat_view ::ascii_matlab - Prints matrix in Matlab format 4965 . -mat_view draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX(). 4966 . -display <name> - Sets display name (default is host) 4967 . -draw_pause <sec> - Sets number of seconds to pause after display 4968 . -mat_view socket - Sends matrix to socket, can be accessed from Matlab (See the <a href="../../docs/manual.pdf">users manual</a>) 4969 . -viewer_socket_machine <machine> 4970 . -viewer_socket_port <port> 4971 . -mat_view binary - save matrix to file in binary format 4972 - -viewer_binary_filename <name> 4973 4974 Notes: 4975 MatSetValues() generally caches the values. The matrix is ready to 4976 use only after MatAssemblyBegin() and MatAssemblyEnd() have been called. 4977 Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES 4978 in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before 4979 using the matrix. 4980 4981 Space for preallocated nonzeros that is not filled by a call to MatSetValues() or a related routine are compressed 4982 out by assembly. If you intend to use that extra space on a subsequent assembly, be sure to insert explicit zeros 4983 before MAT_FINAL_ASSEMBLY so the space is not compressed out. 4984 4985 Level: beginner 4986 4987 .seealso: MatAssemblyBegin(), MatSetValues(), PetscDrawOpenX(), PetscDrawCreate(), MatView(), MatAssembled(), PetscViewerSocketOpen() 4988 @*/ 4989 PetscErrorCode MatAssemblyEnd(Mat mat,MatAssemblyType type) 4990 { 4991 PetscErrorCode ierr; 4992 static PetscInt inassm = 0; 4993 PetscBool flg = PETSC_FALSE; 4994 4995 PetscFunctionBegin; 4996 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 4997 PetscValidType(mat,1); 4998 4999 inassm++; 5000 MatAssemblyEnd_InUse++; 5001 if (MatAssemblyEnd_InUse == 1) { /* Do the logging only the first time through */ 5002 ierr = PetscLogEventBegin(MAT_AssemblyEnd,mat,0,0,0);CHKERRQ(ierr); 5003 if (mat->ops->assemblyend) { 5004 ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr); 5005 } 5006 ierr = PetscLogEventEnd(MAT_AssemblyEnd,mat,0,0,0);CHKERRQ(ierr); 5007 } else if (mat->ops->assemblyend) { 5008 ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr); 5009 } 5010 5011 /* Flush assembly is not a true assembly */ 5012 if (type != MAT_FLUSH_ASSEMBLY) { 5013 mat->assembled = PETSC_TRUE; mat->num_ass++; 5014 } 5015 mat->insertmode = NOT_SET_VALUES; 5016 MatAssemblyEnd_InUse--; 5017 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5018 if (!mat->symmetric_eternal) { 5019 mat->symmetric_set = PETSC_FALSE; 5020 mat->hermitian_set = PETSC_FALSE; 5021 mat->structurally_symmetric_set = PETSC_FALSE; 5022 } 5023 #if defined(PETSC_HAVE_CUSP) 5024 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5025 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5026 } 5027 #endif 5028 #if defined(PETSC_HAVE_VIENNACL) 5029 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5030 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5031 } 5032 #endif 5033 if (inassm == 1 && type != MAT_FLUSH_ASSEMBLY) { 5034 if (mat->viewonassembly) { 5035 ierr = PetscViewerPushFormat(mat->viewonassembly,mat->viewformatonassembly);CHKERRQ(ierr); 5036 ierr = MatView(mat,mat->viewonassembly);CHKERRQ(ierr); 5037 ierr = PetscViewerPopFormat(mat->viewonassembly);CHKERRQ(ierr); 5038 } 5039 5040 if (mat->checksymmetryonassembly) { 5041 ierr = MatIsSymmetric(mat,mat->checksymmetrytol,&flg);CHKERRQ(ierr); 5042 if (flg) { 5043 ierr = PetscPrintf(PetscObjectComm((PetscObject)mat),"Matrix is symmetric (tolerance %G)\n",mat->checksymmetrytol);CHKERRQ(ierr); 5044 } else { 5045 ierr = PetscPrintf(PetscObjectComm((PetscObject)mat),"Matrix is not symmetric (tolerance %G)\n",mat->checksymmetrytol);CHKERRQ(ierr); 5046 } 5047 } 5048 if (mat->nullsp && mat->checknullspaceonassembly) { 5049 ierr = MatNullSpaceTest(mat->nullsp,mat,NULL);CHKERRQ(ierr); 5050 } 5051 } 5052 inassm--; 5053 PetscFunctionReturn(0); 5054 } 5055 5056 #undef __FUNCT__ 5057 #define __FUNCT__ "MatSetOption" 5058 /*@ 5059 MatSetOption - Sets a parameter option for a matrix. Some options 5060 may be specific to certain storage formats. Some options 5061 determine how values will be inserted (or added). Sorted, 5062 row-oriented input will generally assemble the fastest. The default 5063 is row-oriented. 5064 5065 Logically Collective on Mat for certain operations, such as MAT_SPD, not collective for MAT_ROW_ORIENTED, see MatOption 5066 5067 Input Parameters: 5068 + mat - the matrix 5069 . option - the option, one of those listed below (and possibly others), 5070 - flg - turn the option on (PETSC_TRUE) or off (PETSC_FALSE) 5071 5072 Options Describing Matrix Structure: 5073 + MAT_SPD - symmetric positive definite 5074 - MAT_SYMMETRIC - symmetric in terms of both structure and value 5075 . MAT_HERMITIAN - transpose is the complex conjugation 5076 . MAT_STRUCTURALLY_SYMMETRIC - symmetric nonzero structure 5077 - MAT_SYMMETRY_ETERNAL - if you would like the symmetry/Hermitian flag 5078 you set to be kept with all future use of the matrix 5079 including after MatAssemblyBegin/End() which could 5080 potentially change the symmetry structure, i.e. you 5081 KNOW the matrix will ALWAYS have the property you set. 5082 5083 5084 Options For Use with MatSetValues(): 5085 Insert a logically dense subblock, which can be 5086 . MAT_ROW_ORIENTED - row-oriented (default) 5087 5088 Note these options reflect the data you pass in with MatSetValues(); it has 5089 nothing to do with how the data is stored internally in the matrix 5090 data structure. 5091 5092 When (re)assembling a matrix, we can restrict the input for 5093 efficiency/debugging purposes. These options include 5094 + MAT_NEW_NONZERO_LOCATIONS - additional insertions will be 5095 allowed if they generate a new nonzero 5096 . MAT_NEW_DIAGONALS - new diagonals will be allowed (for block diagonal format only) 5097 . MAT_IGNORE_OFF_PROC_ENTRIES - drops off-processor entries 5098 . MAT_NEW_NONZERO_LOCATION_ERR - generates an error for new matrix entry 5099 . MAT_USE_HASH_TABLE - uses a hash table to speed up matrix assembly 5100 + MAT_NO_OFF_PROC_ENTRIES - you know each process will only set values for its own rows, will generate an error if 5101 any process sets values for another process. This avoids all reductions in the MatAssembly routines and thus improves 5102 performance for very large process counts. 5103 5104 Notes: 5105 Some options are relevant only for particular matrix types and 5106 are thus ignored by others. Other options are not supported by 5107 certain matrix types and will generate an error message if set. 5108 5109 If using a Fortran 77 module to compute a matrix, one may need to 5110 use the column-oriented option (or convert to the row-oriented 5111 format). 5112 5113 MAT_NEW_NONZERO_LOCATIONS set to PETSC_FALSE indicates that any add or insertion 5114 that would generate a new entry in the nonzero structure is instead 5115 ignored. Thus, if memory has not alredy been allocated for this particular 5116 data, then the insertion is ignored. For dense matrices, in which 5117 the entire array is allocated, no entries are ever ignored. 5118 Set after the first MatAssemblyEnd() 5119 5120 MAT_NEW_NONZERO_LOCATION_ERR indicates that any add or insertion 5121 that would generate a new entry in the nonzero structure instead produces 5122 an error. (Currently supported for AIJ and BAIJ formats only.) 5123 This is a useful flag when using SAME_NONZERO_PATTERN in calling 5124 KSPSetOperators() to ensure that the nonzero pattern truely does 5125 remain unchanged. Set after the first MatAssemblyEnd() 5126 5127 MAT_NEW_NONZERO_ALLOCATION_ERR indicates that any add or insertion 5128 that would generate a new entry that has not been preallocated will 5129 instead produce an error. (Currently supported for AIJ and BAIJ formats 5130 only.) This is a useful flag when debugging matrix memory preallocation. 5131 5132 MAT_IGNORE_OFF_PROC_ENTRIES indicates entries destined for 5133 other processors should be dropped, rather than stashed. 5134 This is useful if you know that the "owning" processor is also 5135 always generating the correct matrix entries, so that PETSc need 5136 not transfer duplicate entries generated on another processor. 5137 5138 MAT_USE_HASH_TABLE indicates that a hash table be used to improve the 5139 searches during matrix assembly. When this flag is set, the hash table 5140 is created during the first Matrix Assembly. This hash table is 5141 used the next time through, during MatSetVaules()/MatSetVaulesBlocked() 5142 to improve the searching of indices. MAT_NEW_NONZERO_LOCATIONS flag 5143 should be used with MAT_USE_HASH_TABLE flag. This option is currently 5144 supported by MATMPIBAIJ format only. 5145 5146 MAT_KEEP_NONZERO_PATTERN indicates when MatZeroRows() is called the zeroed entries 5147 are kept in the nonzero structure 5148 5149 MAT_IGNORE_ZERO_ENTRIES - for AIJ/IS matrices this will stop zero values from creating 5150 a zero location in the matrix 5151 5152 MAT_USE_INODES - indicates using inode version of the code - works with AIJ and 5153 ROWBS matrix types 5154 5155 MAT_NO_OFF_PROC_ZERO_ROWS - you know each process will only zero its own rows. This avoids all reductions in the 5156 zero row routines and thus improves performance for very large process counts. 5157 5158 MAT_IGNORE_LOWER_TRIANGULAR - For SBAIJ matrices will ignore any insertions you make in the lower triangular 5159 part of the matrix (since they should match the upper triangular part). 5160 5161 Notes: Can only be called after MatSetSizes() and MatSetType() have been set. 5162 5163 Level: intermediate 5164 5165 Concepts: matrices^setting options 5166 5167 .seealso: MatOption, Mat 5168 5169 @*/ 5170 PetscErrorCode MatSetOption(Mat mat,MatOption op,PetscBool flg) 5171 { 5172 PetscErrorCode ierr; 5173 5174 PetscFunctionBegin; 5175 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5176 PetscValidType(mat,1); 5177 if (op > 0) PetscValidLogicalCollectiveEnum(mat,op,2); 5178 PetscValidLogicalCollectiveBool(mat,flg,3); 5179 5180 if (((int) op) <= MAT_OPTION_MIN || ((int) op) >= MAT_OPTION_MAX) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Options %d is out of range",(int)op); 5181 if (!((PetscObject)mat)->type_name) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_TYPENOTSET,"Cannot set options until type and size have been set, see MatSetType() and MatSetSizes()"); 5182 5183 switch (op) { 5184 case MAT_NO_OFF_PROC_ENTRIES: 5185 mat->nooffprocentries = flg; 5186 PetscFunctionReturn(0); 5187 break; 5188 case MAT_NO_OFF_PROC_ZERO_ROWS: 5189 mat->nooffproczerorows = flg; 5190 PetscFunctionReturn(0); 5191 break; 5192 case MAT_SPD: 5193 mat->spd_set = PETSC_TRUE; 5194 mat->spd = flg; 5195 if (flg) { 5196 mat->symmetric = PETSC_TRUE; 5197 mat->structurally_symmetric = PETSC_TRUE; 5198 mat->symmetric_set = PETSC_TRUE; 5199 mat->structurally_symmetric_set = PETSC_TRUE; 5200 } 5201 break; 5202 case MAT_SYMMETRIC: 5203 mat->symmetric = flg; 5204 if (flg) mat->structurally_symmetric = PETSC_TRUE; 5205 mat->symmetric_set = PETSC_TRUE; 5206 mat->structurally_symmetric_set = flg; 5207 break; 5208 case MAT_HERMITIAN: 5209 mat->hermitian = flg; 5210 if (flg) mat->structurally_symmetric = PETSC_TRUE; 5211 mat->hermitian_set = PETSC_TRUE; 5212 mat->structurally_symmetric_set = flg; 5213 break; 5214 case MAT_STRUCTURALLY_SYMMETRIC: 5215 mat->structurally_symmetric = flg; 5216 mat->structurally_symmetric_set = PETSC_TRUE; 5217 break; 5218 case MAT_SYMMETRY_ETERNAL: 5219 mat->symmetric_eternal = flg; 5220 break; 5221 default: 5222 break; 5223 } 5224 if (mat->ops->setoption) { 5225 ierr = (*mat->ops->setoption)(mat,op,flg);CHKERRQ(ierr); 5226 } 5227 PetscFunctionReturn(0); 5228 } 5229 5230 #undef __FUNCT__ 5231 #define __FUNCT__ "MatZeroEntries" 5232 /*@ 5233 MatZeroEntries - Zeros all entries of a matrix. For sparse matrices 5234 this routine retains the old nonzero structure. 5235 5236 Logically Collective on Mat 5237 5238 Input Parameters: 5239 . mat - the matrix 5240 5241 Level: intermediate 5242 5243 Notes: If the matrix was not preallocated then a default, likely poor preallocation will be set in the matrix, so this should be called after the preallocation phase. 5244 See the Performance chapter of the users manual for information on preallocating matrices. 5245 5246 Concepts: matrices^zeroing 5247 5248 .seealso: MatZeroRows() 5249 @*/ 5250 PetscErrorCode MatZeroEntries(Mat mat) 5251 { 5252 PetscErrorCode ierr; 5253 5254 PetscFunctionBegin; 5255 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5256 PetscValidType(mat,1); 5257 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5258 if (mat->insertmode != NOT_SET_VALUES) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for matrices where you have set values but not yet assembled"); 5259 if (!mat->ops->zeroentries) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 5260 MatCheckPreallocated(mat,1); 5261 5262 ierr = PetscLogEventBegin(MAT_ZeroEntries,mat,0,0,0);CHKERRQ(ierr); 5263 ierr = (*mat->ops->zeroentries)(mat);CHKERRQ(ierr); 5264 ierr = PetscLogEventEnd(MAT_ZeroEntries,mat,0,0,0);CHKERRQ(ierr); 5265 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5266 #if defined(PETSC_HAVE_CUSP) 5267 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5268 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5269 } 5270 #endif 5271 #if defined(PETSC_HAVE_VIENNACL) 5272 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5273 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5274 } 5275 #endif 5276 PetscFunctionReturn(0); 5277 } 5278 5279 #undef __FUNCT__ 5280 #define __FUNCT__ "MatZeroRowsColumns" 5281 /*@C 5282 MatZeroRowsColumns - Zeros all entries (except possibly the main diagonal) 5283 of a set of rows and columns of a matrix. 5284 5285 Collective on Mat 5286 5287 Input Parameters: 5288 + mat - the matrix 5289 . numRows - the number of rows to remove 5290 . rows - the global row indices 5291 . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry) 5292 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5293 - b - optional vector of right hand side, that will be adjusted by provided solution 5294 5295 Notes: 5296 This does not change the nonzero structure of the matrix, it merely zeros those entries in the matrix. 5297 5298 The user can set a value in the diagonal entry (or for the AIJ and 5299 row formats can optionally remove the main diagonal entry from the 5300 nonzero structure as well, by passing 0.0 as the final argument). 5301 5302 For the parallel case, all processes that share the matrix (i.e., 5303 those in the communicator used for matrix creation) MUST call this 5304 routine, regardless of whether any rows being zeroed are owned by 5305 them. 5306 5307 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5308 list only rows local to itself). 5309 5310 The option MAT_NO_OFF_PROC_ZERO_ROWS does not apply to this routine. 5311 5312 Level: intermediate 5313 5314 Concepts: matrices^zeroing rows 5315 5316 .seealso: MatZeroRowsIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(), MatZeroRowsColumnsIS() 5317 @*/ 5318 PetscErrorCode MatZeroRowsColumns(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 5319 { 5320 PetscErrorCode ierr; 5321 5322 PetscFunctionBegin; 5323 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5324 PetscValidType(mat,1); 5325 if (numRows) PetscValidIntPointer(rows,3); 5326 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 5327 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5328 if (!mat->ops->zerorowscolumns) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 5329 MatCheckPreallocated(mat,1); 5330 5331 ierr = (*mat->ops->zerorowscolumns)(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5332 if (mat->viewonassembly) { 5333 ierr = PetscViewerPushFormat(mat->viewonassembly,mat->viewformatonassembly);CHKERRQ(ierr); 5334 ierr = MatView(mat,mat->viewonassembly);CHKERRQ(ierr); 5335 ierr = PetscViewerPopFormat(mat->viewonassembly);CHKERRQ(ierr); 5336 } 5337 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5338 #if defined(PETSC_HAVE_CUSP) 5339 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5340 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5341 } 5342 #endif 5343 #if defined(PETSC_HAVE_VIENNACL) 5344 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5345 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5346 } 5347 #endif 5348 PetscFunctionReturn(0); 5349 } 5350 5351 #undef __FUNCT__ 5352 #define __FUNCT__ "MatZeroRowsColumnsIS" 5353 /*@C 5354 MatZeroRowsColumnsIS - Zeros all entries (except possibly the main diagonal) 5355 of a set of rows and columns of a matrix. 5356 5357 Collective on Mat 5358 5359 Input Parameters: 5360 + mat - the matrix 5361 . is - the rows to zero 5362 . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry) 5363 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5364 - b - optional vector of right hand side, that will be adjusted by provided solution 5365 5366 Notes: 5367 This does not change the nonzero structure of the matrix, it merely zeros those entries in the matrix. 5368 5369 The user can set a value in the diagonal entry (or for the AIJ and 5370 row formats can optionally remove the main diagonal entry from the 5371 nonzero structure as well, by passing 0.0 as the final argument). 5372 5373 For the parallel case, all processes that share the matrix (i.e., 5374 those in the communicator used for matrix creation) MUST call this 5375 routine, regardless of whether any rows being zeroed are owned by 5376 them. 5377 5378 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5379 list only rows local to itself). 5380 5381 The option MAT_NO_OFF_PROC_ZERO_ROWS does not apply to this routine. 5382 5383 Level: intermediate 5384 5385 Concepts: matrices^zeroing rows 5386 5387 .seealso: MatZeroRowsIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption(), MatZeroRowsColumns() 5388 @*/ 5389 PetscErrorCode MatZeroRowsColumnsIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b) 5390 { 5391 PetscErrorCode ierr; 5392 PetscInt numRows; 5393 const PetscInt *rows; 5394 5395 PetscFunctionBegin; 5396 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5397 PetscValidHeaderSpecific(is,IS_CLASSID,2); 5398 PetscValidType(mat,1); 5399 PetscValidType(is,2); 5400 ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr); 5401 ierr = ISGetIndices(is,&rows);CHKERRQ(ierr); 5402 ierr = MatZeroRowsColumns(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5403 ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr); 5404 PetscFunctionReturn(0); 5405 } 5406 5407 #undef __FUNCT__ 5408 #define __FUNCT__ "MatZeroRows" 5409 /*@C 5410 MatZeroRows - Zeros all entries (except possibly the main diagonal) 5411 of a set of rows of a matrix. 5412 5413 Collective on Mat 5414 5415 Input Parameters: 5416 + mat - the matrix 5417 . numRows - the number of rows to remove 5418 . rows - the global row indices 5419 . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry) 5420 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5421 - b - optional vector of right hand side, that will be adjusted by provided solution 5422 5423 Notes: 5424 For the AIJ and BAIJ matrix formats this removes the old nonzero structure, 5425 but does not release memory. For the dense and block diagonal 5426 formats this does not alter the nonzero structure. 5427 5428 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5429 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5430 merely zeroed. 5431 5432 The user can set a value in the diagonal entry (or for the AIJ and 5433 row formats can optionally remove the main diagonal entry from the 5434 nonzero structure as well, by passing 0.0 as the final argument). 5435 5436 For the parallel case, all processes that share the matrix (i.e., 5437 those in the communicator used for matrix creation) MUST call this 5438 routine, regardless of whether any rows being zeroed are owned by 5439 them. 5440 5441 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5442 list only rows local to itself). 5443 5444 You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it 5445 owns that are to be zeroed. This saves a global synchronization in the implementation. 5446 5447 Level: intermediate 5448 5449 Concepts: matrices^zeroing rows 5450 5451 .seealso: MatZeroRowsIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption() 5452 @*/ 5453 PetscErrorCode MatZeroRows(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 5454 { 5455 PetscErrorCode ierr; 5456 5457 PetscFunctionBegin; 5458 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5459 PetscValidType(mat,1); 5460 if (numRows) PetscValidIntPointer(rows,3); 5461 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 5462 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5463 if (!mat->ops->zerorows) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 5464 MatCheckPreallocated(mat,1); 5465 5466 ierr = (*mat->ops->zerorows)(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5467 if (mat->viewonassembly) { 5468 ierr = PetscViewerPushFormat(mat->viewonassembly,mat->viewformatonassembly);CHKERRQ(ierr); 5469 ierr = MatView(mat,mat->viewonassembly);CHKERRQ(ierr); 5470 ierr = PetscViewerPopFormat(mat->viewonassembly);CHKERRQ(ierr); 5471 } 5472 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5473 #if defined(PETSC_HAVE_CUSP) 5474 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5475 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5476 } 5477 #endif 5478 #if defined(PETSC_HAVE_VIENNACL) 5479 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5480 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5481 } 5482 #endif 5483 PetscFunctionReturn(0); 5484 } 5485 5486 #undef __FUNCT__ 5487 #define __FUNCT__ "MatZeroRowsIS" 5488 /*@C 5489 MatZeroRowsIS - Zeros all entries (except possibly the main diagonal) 5490 of a set of rows of a matrix. 5491 5492 Collective on Mat 5493 5494 Input Parameters: 5495 + mat - the matrix 5496 . is - index set of rows to remove 5497 . diag - value put in all diagonals of eliminated rows 5498 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5499 - b - optional vector of right hand side, that will be adjusted by provided solution 5500 5501 Notes: 5502 For the AIJ and BAIJ matrix formats this removes the old nonzero structure, 5503 but does not release memory. For the dense and block diagonal 5504 formats this does not alter the nonzero structure. 5505 5506 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5507 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5508 merely zeroed. 5509 5510 The user can set a value in the diagonal entry (or for the AIJ and 5511 row formats can optionally remove the main diagonal entry from the 5512 nonzero structure as well, by passing 0.0 as the final argument). 5513 5514 For the parallel case, all processes that share the matrix (i.e., 5515 those in the communicator used for matrix creation) MUST call this 5516 routine, regardless of whether any rows being zeroed are owned by 5517 them. 5518 5519 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5520 list only rows local to itself). 5521 5522 You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it 5523 owns that are to be zeroed. This saves a global synchronization in the implementation. 5524 5525 Level: intermediate 5526 5527 Concepts: matrices^zeroing rows 5528 5529 .seealso: MatZeroRows(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption() 5530 @*/ 5531 PetscErrorCode MatZeroRowsIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b) 5532 { 5533 PetscInt numRows; 5534 const PetscInt *rows; 5535 PetscErrorCode ierr; 5536 5537 PetscFunctionBegin; 5538 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5539 PetscValidType(mat,1); 5540 PetscValidHeaderSpecific(is,IS_CLASSID,2); 5541 ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr); 5542 ierr = ISGetIndices(is,&rows);CHKERRQ(ierr); 5543 ierr = MatZeroRows(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5544 ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr); 5545 PetscFunctionReturn(0); 5546 } 5547 5548 #undef __FUNCT__ 5549 #define __FUNCT__ "MatZeroRowsStencil" 5550 /*@C 5551 MatZeroRowsStencil - Zeros all entries (except possibly the main diagonal) 5552 of a set of rows of a matrix. These rows must be local to the process. 5553 5554 Collective on Mat 5555 5556 Input Parameters: 5557 + mat - the matrix 5558 . numRows - the number of rows to remove 5559 . rows - the grid coordinates (and component number when dof > 1) for matrix rows 5560 . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry) 5561 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5562 - b - optional vector of right hand side, that will be adjusted by provided solution 5563 5564 Notes: 5565 For the AIJ and BAIJ matrix formats this removes the old nonzero structure, 5566 but does not release memory. For the dense and block diagonal 5567 formats this does not alter the nonzero structure. 5568 5569 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5570 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5571 merely zeroed. 5572 5573 The user can set a value in the diagonal entry (or for the AIJ and 5574 row formats can optionally remove the main diagonal entry from the 5575 nonzero structure as well, by passing 0.0 as the final argument). 5576 5577 For the parallel case, all processes that share the matrix (i.e., 5578 those in the communicator used for matrix creation) MUST call this 5579 routine, regardless of whether any rows being zeroed are owned by 5580 them. 5581 5582 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5583 list only rows local to itself). 5584 5585 The grid coordinates are across the entire grid, not just the local portion 5586 5587 In Fortran idxm and idxn should be declared as 5588 $ MatStencil idxm(4,m) 5589 and the values inserted using 5590 $ idxm(MatStencil_i,1) = i 5591 $ idxm(MatStencil_j,1) = j 5592 $ idxm(MatStencil_k,1) = k 5593 $ idxm(MatStencil_c,1) = c 5594 etc 5595 5596 For periodic boundary conditions use negative indices for values to the left (below 0; that are to be 5597 obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one 5598 etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the 5599 DMDA_BOUNDARY_PERIODIC boundary type. 5600 5601 For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have 5602 a single value per point) you can skip filling those indices. 5603 5604 Level: intermediate 5605 5606 Concepts: matrices^zeroing rows 5607 5608 .seealso: MatZeroRows(), MatZeroRowsIS(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption() 5609 @*/ 5610 PetscErrorCode MatZeroRowsStencil(Mat mat,PetscInt numRows,const MatStencil rows[],PetscScalar diag,Vec x,Vec b) 5611 { 5612 PetscInt dim = mat->stencil.dim; 5613 PetscInt sdim = dim - (1 - (PetscInt) mat->stencil.noc); 5614 PetscInt *dims = mat->stencil.dims+1; 5615 PetscInt *starts = mat->stencil.starts; 5616 PetscInt *dxm = (PetscInt*) rows; 5617 PetscInt *jdxm, i, j, tmp, numNewRows = 0; 5618 PetscErrorCode ierr; 5619 5620 PetscFunctionBegin; 5621 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5622 PetscValidType(mat,1); 5623 if (numRows) PetscValidIntPointer(rows,3); 5624 5625 ierr = PetscMalloc(numRows*sizeof(PetscInt), &jdxm);CHKERRQ(ierr); 5626 for (i = 0; i < numRows; ++i) { 5627 /* Skip unused dimensions (they are ordered k, j, i, c) */ 5628 for (j = 0; j < 3-sdim; ++j) dxm++; 5629 /* Local index in X dir */ 5630 tmp = *dxm++ - starts[0]; 5631 /* Loop over remaining dimensions */ 5632 for (j = 0; j < dim-1; ++j) { 5633 /* If nonlocal, set index to be negative */ 5634 if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT; 5635 /* Update local index */ 5636 else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1]; 5637 } 5638 /* Skip component slot if necessary */ 5639 if (mat->stencil.noc) dxm++; 5640 /* Local row number */ 5641 if (tmp >= 0) { 5642 jdxm[numNewRows++] = tmp; 5643 } 5644 } 5645 ierr = MatZeroRowsLocal(mat,numNewRows,jdxm,diag,x,b);CHKERRQ(ierr); 5646 ierr = PetscFree(jdxm);CHKERRQ(ierr); 5647 PetscFunctionReturn(0); 5648 } 5649 5650 #undef __FUNCT__ 5651 #define __FUNCT__ "MatZeroRowsColumnsStencil" 5652 /*@C 5653 MatZeroRowsColumnsStencil - Zeros all row and column entries (except possibly the main diagonal) 5654 of a set of rows and columns of a matrix. 5655 5656 Collective on Mat 5657 5658 Input Parameters: 5659 + mat - the matrix 5660 . numRows - the number of rows/columns to remove 5661 . rows - the grid coordinates (and component number when dof > 1) for matrix rows 5662 . diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry) 5663 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5664 - b - optional vector of right hand side, that will be adjusted by provided solution 5665 5666 Notes: 5667 For the AIJ and BAIJ matrix formats this removes the old nonzero structure, 5668 but does not release memory. For the dense and block diagonal 5669 formats this does not alter the nonzero structure. 5670 5671 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5672 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5673 merely zeroed. 5674 5675 The user can set a value in the diagonal entry (or for the AIJ and 5676 row formats can optionally remove the main diagonal entry from the 5677 nonzero structure as well, by passing 0.0 as the final argument). 5678 5679 For the parallel case, all processes that share the matrix (i.e., 5680 those in the communicator used for matrix creation) MUST call this 5681 routine, regardless of whether any rows being zeroed are owned by 5682 them. 5683 5684 Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to 5685 list only rows local to itself, but the row/column numbers are given in local numbering). 5686 5687 The grid coordinates are across the entire grid, not just the local portion 5688 5689 In Fortran idxm and idxn should be declared as 5690 $ MatStencil idxm(4,m) 5691 and the values inserted using 5692 $ idxm(MatStencil_i,1) = i 5693 $ idxm(MatStencil_j,1) = j 5694 $ idxm(MatStencil_k,1) = k 5695 $ idxm(MatStencil_c,1) = c 5696 etc 5697 5698 For periodic boundary conditions use negative indices for values to the left (below 0; that are to be 5699 obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one 5700 etc to obtain values that obtained by wrapping the values from the left edge. This does not work for anything but the 5701 DMDA_BOUNDARY_PERIODIC boundary type. 5702 5703 For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have 5704 a single value per point) you can skip filling those indices. 5705 5706 Level: intermediate 5707 5708 Concepts: matrices^zeroing rows 5709 5710 .seealso: MatZeroRows(), MatZeroRowsIS(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption() 5711 @*/ 5712 PetscErrorCode MatZeroRowsColumnsStencil(Mat mat,PetscInt numRows,const MatStencil rows[],PetscScalar diag,Vec x,Vec b) 5713 { 5714 PetscInt dim = mat->stencil.dim; 5715 PetscInt sdim = dim - (1 - (PetscInt) mat->stencil.noc); 5716 PetscInt *dims = mat->stencil.dims+1; 5717 PetscInt *starts = mat->stencil.starts; 5718 PetscInt *dxm = (PetscInt*) rows; 5719 PetscInt *jdxm, i, j, tmp, numNewRows = 0; 5720 PetscErrorCode ierr; 5721 5722 PetscFunctionBegin; 5723 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5724 PetscValidType(mat,1); 5725 if (numRows) PetscValidIntPointer(rows,3); 5726 5727 ierr = PetscMalloc(numRows*sizeof(PetscInt), &jdxm);CHKERRQ(ierr); 5728 for (i = 0; i < numRows; ++i) { 5729 /* Skip unused dimensions (they are ordered k, j, i, c) */ 5730 for (j = 0; j < 3-sdim; ++j) dxm++; 5731 /* Local index in X dir */ 5732 tmp = *dxm++ - starts[0]; 5733 /* Loop over remaining dimensions */ 5734 for (j = 0; j < dim-1; ++j) { 5735 /* If nonlocal, set index to be negative */ 5736 if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT; 5737 /* Update local index */ 5738 else tmp = tmp*dims[j] + *(dxm-1) - starts[j+1]; 5739 } 5740 /* Skip component slot if necessary */ 5741 if (mat->stencil.noc) dxm++; 5742 /* Local row number */ 5743 if (tmp >= 0) { 5744 jdxm[numNewRows++] = tmp; 5745 } 5746 } 5747 ierr = MatZeroRowsColumnsLocal(mat,numNewRows,jdxm,diag,x,b);CHKERRQ(ierr); 5748 ierr = PetscFree(jdxm);CHKERRQ(ierr); 5749 PetscFunctionReturn(0); 5750 } 5751 5752 #undef __FUNCT__ 5753 #define __FUNCT__ "MatZeroRowsLocal" 5754 /*@C 5755 MatZeroRowsLocal - Zeros all entries (except possibly the main diagonal) 5756 of a set of rows of a matrix; using local numbering of rows. 5757 5758 Collective on Mat 5759 5760 Input Parameters: 5761 + mat - the matrix 5762 . numRows - the number of rows to remove 5763 . rows - the global row indices 5764 . diag - value put in all diagonals of eliminated rows 5765 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5766 - b - optional vector of right hand side, that will be adjusted by provided solution 5767 5768 Notes: 5769 Before calling MatZeroRowsLocal(), the user must first set the 5770 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 5771 5772 For the AIJ matrix formats this removes the old nonzero structure, 5773 but does not release memory. For the dense and block diagonal 5774 formats this does not alter the nonzero structure. 5775 5776 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5777 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5778 merely zeroed. 5779 5780 The user can set a value in the diagonal entry (or for the AIJ and 5781 row formats can optionally remove the main diagonal entry from the 5782 nonzero structure as well, by passing 0.0 as the final argument). 5783 5784 You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it 5785 owns that are to be zeroed. This saves a global synchronization in the implementation. 5786 5787 Level: intermediate 5788 5789 Concepts: matrices^zeroing 5790 5791 .seealso: MatZeroRows(), MatZeroRowsLocalIS(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping 5792 @*/ 5793 PetscErrorCode MatZeroRowsLocal(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 5794 { 5795 PetscErrorCode ierr; 5796 PetscMPIInt size; 5797 5798 PetscFunctionBegin; 5799 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5800 PetscValidType(mat,1); 5801 if (numRows) PetscValidIntPointer(rows,3); 5802 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 5803 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5804 MatCheckPreallocated(mat,1); 5805 5806 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);CHKERRQ(ierr); 5807 if (mat->ops->zerorowslocal) { 5808 ierr = (*mat->ops->zerorowslocal)(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5809 } else if (size == 1) { 5810 ierr = (*mat->ops->zerorows)(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5811 } else { 5812 IS is, newis; 5813 const PetscInt *newRows; 5814 5815 if (!mat->rmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Need to provide local to global mapping to matrix first"); 5816 ierr = ISCreateGeneral(PETSC_COMM_SELF,numRows,rows,PETSC_COPY_VALUES,&is);CHKERRQ(ierr); 5817 ierr = ISLocalToGlobalMappingApplyIS(mat->rmap->mapping,is,&newis);CHKERRQ(ierr); 5818 ierr = ISGetIndices(newis,&newRows);CHKERRQ(ierr); 5819 ierr = (*mat->ops->zerorows)(mat,numRows,newRows,diag,x,b);CHKERRQ(ierr); 5820 ierr = ISRestoreIndices(newis,&newRows);CHKERRQ(ierr); 5821 ierr = ISDestroy(&newis);CHKERRQ(ierr); 5822 ierr = ISDestroy(&is);CHKERRQ(ierr); 5823 } 5824 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5825 #if defined(PETSC_HAVE_CUSP) 5826 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5827 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5828 } 5829 #endif 5830 #if defined(PETSC_HAVE_VIENNACL) 5831 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5832 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5833 } 5834 #endif 5835 PetscFunctionReturn(0); 5836 } 5837 5838 #undef __FUNCT__ 5839 #define __FUNCT__ "MatZeroRowsLocalIS" 5840 /*@C 5841 MatZeroRowsLocalIS - Zeros all entries (except possibly the main diagonal) 5842 of a set of rows of a matrix; using local numbering of rows. 5843 5844 Collective on Mat 5845 5846 Input Parameters: 5847 + mat - the matrix 5848 . is - index set of rows to remove 5849 . diag - value put in all diagonals of eliminated rows 5850 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5851 - b - optional vector of right hand side, that will be adjusted by provided solution 5852 5853 Notes: 5854 Before calling MatZeroRowsLocalIS(), the user must first set the 5855 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 5856 5857 For the AIJ matrix formats this removes the old nonzero structure, 5858 but does not release memory. For the dense and block diagonal 5859 formats this does not alter the nonzero structure. 5860 5861 If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure 5862 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 5863 merely zeroed. 5864 5865 The user can set a value in the diagonal entry (or for the AIJ and 5866 row formats can optionally remove the main diagonal entry from the 5867 nonzero structure as well, by passing 0.0 as the final argument). 5868 5869 You can call MatSetOption(mat,MAT_NO_OFF_PROC_ZERO_ROWS,PETSC_TRUE) if each process indicates only rows it 5870 owns that are to be zeroed. This saves a global synchronization in the implementation. 5871 5872 Level: intermediate 5873 5874 Concepts: matrices^zeroing 5875 5876 .seealso: MatZeroRows(), MatZeroRowsLocal(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping 5877 @*/ 5878 PetscErrorCode MatZeroRowsLocalIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b) 5879 { 5880 PetscErrorCode ierr; 5881 PetscInt numRows; 5882 const PetscInt *rows; 5883 5884 PetscFunctionBegin; 5885 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5886 PetscValidType(mat,1); 5887 PetscValidHeaderSpecific(is,IS_CLASSID,2); 5888 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 5889 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5890 MatCheckPreallocated(mat,1); 5891 5892 ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr); 5893 ierr = ISGetIndices(is,&rows);CHKERRQ(ierr); 5894 ierr = MatZeroRowsLocal(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5895 ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr); 5896 PetscFunctionReturn(0); 5897 } 5898 5899 #undef __FUNCT__ 5900 #define __FUNCT__ "MatZeroRowsColumnsLocal" 5901 /*@C 5902 MatZeroRowsColumnsLocal - Zeros all entries (except possibly the main diagonal) 5903 of a set of rows and columns of a matrix; using local numbering of rows. 5904 5905 Collective on Mat 5906 5907 Input Parameters: 5908 + mat - the matrix 5909 . numRows - the number of rows to remove 5910 . rows - the global row indices 5911 . diag - value put in all diagonals of eliminated rows 5912 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5913 - b - optional vector of right hand side, that will be adjusted by provided solution 5914 5915 Notes: 5916 Before calling MatZeroRowsColumnsLocal(), the user must first set the 5917 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 5918 5919 The user can set a value in the diagonal entry (or for the AIJ and 5920 row formats can optionally remove the main diagonal entry from the 5921 nonzero structure as well, by passing 0.0 as the final argument). 5922 5923 Level: intermediate 5924 5925 Concepts: matrices^zeroing 5926 5927 .seealso: MatZeroRows(), MatZeroRowsLocalIS(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping 5928 @*/ 5929 PetscErrorCode MatZeroRowsColumnsLocal(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 5930 { 5931 PetscErrorCode ierr; 5932 PetscMPIInt size; 5933 5934 PetscFunctionBegin; 5935 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 5936 PetscValidType(mat,1); 5937 if (numRows) PetscValidIntPointer(rows,3); 5938 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 5939 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 5940 MatCheckPreallocated(mat,1); 5941 5942 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);CHKERRQ(ierr); 5943 if (size == 1) { 5944 ierr = (*mat->ops->zerorowscolumns)(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 5945 } else { 5946 IS is, newis; 5947 const PetscInt *newRows; 5948 5949 if (!mat->cmap->mapping) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Need to provide local to global mapping to matrix first"); 5950 ierr = ISCreateGeneral(PETSC_COMM_SELF,numRows,rows,PETSC_COPY_VALUES,&is);CHKERRQ(ierr); 5951 ierr = ISLocalToGlobalMappingApplyIS(mat->cmap->mapping,is,&newis);CHKERRQ(ierr); 5952 ierr = ISGetIndices(newis,&newRows);CHKERRQ(ierr); 5953 ierr = (*mat->ops->zerorowscolumns)(mat,numRows,newRows,diag,x,b);CHKERRQ(ierr); 5954 ierr = ISRestoreIndices(newis,&newRows);CHKERRQ(ierr); 5955 ierr = ISDestroy(&newis);CHKERRQ(ierr); 5956 ierr = ISDestroy(&is);CHKERRQ(ierr); 5957 } 5958 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 5959 #if defined(PETSC_HAVE_CUSP) 5960 if (mat->valid_GPU_matrix != PETSC_CUSP_UNALLOCATED) { 5961 mat->valid_GPU_matrix = PETSC_CUSP_CPU; 5962 } 5963 #endif 5964 #if defined(PETSC_HAVE_VIENNACL) 5965 if (mat->valid_GPU_matrix != PETSC_VIENNACL_UNALLOCATED) { 5966 mat->valid_GPU_matrix = PETSC_VIENNACL_CPU; 5967 } 5968 #endif 5969 PetscFunctionReturn(0); 5970 } 5971 5972 #undef __FUNCT__ 5973 #define __FUNCT__ "MatZeroRowsColumnsLocalIS" 5974 /*@C 5975 MatZeroRowsColumnsLocalIS - Zeros all entries (except possibly the main diagonal) 5976 of a set of rows and columns of a matrix; using local numbering of rows. 5977 5978 Collective on Mat 5979 5980 Input Parameters: 5981 + mat - the matrix 5982 . is - index set of rows to remove 5983 . diag - value put in all diagonals of eliminated rows 5984 . x - optional vector of solutions for zeroed rows (other entries in vector are not used) 5985 - b - optional vector of right hand side, that will be adjusted by provided solution 5986 5987 Notes: 5988 Before calling MatZeroRowsColumnsLocalIS(), the user must first set the 5989 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 5990 5991 The user can set a value in the diagonal entry (or for the AIJ and 5992 row formats can optionally remove the main diagonal entry from the 5993 nonzero structure as well, by passing 0.0 as the final argument). 5994 5995 Level: intermediate 5996 5997 Concepts: matrices^zeroing 5998 5999 .seealso: MatZeroRows(), MatZeroRowsLocal(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping 6000 @*/ 6001 PetscErrorCode MatZeroRowsColumnsLocalIS(Mat mat,IS is,PetscScalar diag,Vec x,Vec b) 6002 { 6003 PetscErrorCode ierr; 6004 PetscInt numRows; 6005 const PetscInt *rows; 6006 6007 PetscFunctionBegin; 6008 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6009 PetscValidType(mat,1); 6010 PetscValidHeaderSpecific(is,IS_CLASSID,2); 6011 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6012 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6013 MatCheckPreallocated(mat,1); 6014 6015 ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr); 6016 ierr = ISGetIndices(is,&rows);CHKERRQ(ierr); 6017 ierr = MatZeroRowsColumnsLocal(mat,numRows,rows,diag,x,b);CHKERRQ(ierr); 6018 ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr); 6019 PetscFunctionReturn(0); 6020 } 6021 6022 #undef __FUNCT__ 6023 #define __FUNCT__ "MatGetSize" 6024 /*@ 6025 MatGetSize - Returns the numbers of rows and columns in a matrix. 6026 6027 Not Collective 6028 6029 Input Parameter: 6030 . mat - the matrix 6031 6032 Output Parameters: 6033 + m - the number of global rows 6034 - n - the number of global columns 6035 6036 Note: both output parameters can be NULL on input. 6037 6038 Level: beginner 6039 6040 Concepts: matrices^size 6041 6042 .seealso: MatGetLocalSize() 6043 @*/ 6044 PetscErrorCode MatGetSize(Mat mat,PetscInt *m,PetscInt *n) 6045 { 6046 PetscFunctionBegin; 6047 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6048 if (m) *m = mat->rmap->N; 6049 if (n) *n = mat->cmap->N; 6050 PetscFunctionReturn(0); 6051 } 6052 6053 #undef __FUNCT__ 6054 #define __FUNCT__ "MatGetLocalSize" 6055 /*@ 6056 MatGetLocalSize - Returns the number of rows and columns in a matrix 6057 stored locally. This information may be implementation dependent, so 6058 use with care. 6059 6060 Not Collective 6061 6062 Input Parameters: 6063 . mat - the matrix 6064 6065 Output Parameters: 6066 + m - the number of local rows 6067 - n - the number of local columns 6068 6069 Note: both output parameters can be NULL on input. 6070 6071 Level: beginner 6072 6073 Concepts: matrices^local size 6074 6075 .seealso: MatGetSize() 6076 @*/ 6077 PetscErrorCode MatGetLocalSize(Mat mat,PetscInt *m,PetscInt *n) 6078 { 6079 PetscFunctionBegin; 6080 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6081 if (m) PetscValidIntPointer(m,2); 6082 if (n) PetscValidIntPointer(n,3); 6083 if (m) *m = mat->rmap->n; 6084 if (n) *n = mat->cmap->n; 6085 PetscFunctionReturn(0); 6086 } 6087 6088 #undef __FUNCT__ 6089 #define __FUNCT__ "MatGetOwnershipRangeColumn" 6090 /*@ 6091 MatGetOwnershipRangeColumn - Returns the range of matrix columns associated with rows of a vector one multiplies by that owned by 6092 this processor. (The columns of the "diagonal block") 6093 6094 Not Collective, unless matrix has not been allocated, then collective on Mat 6095 6096 Input Parameters: 6097 . mat - the matrix 6098 6099 Output Parameters: 6100 + m - the global index of the first local column 6101 - n - one more than the global index of the last local column 6102 6103 Notes: both output parameters can be NULL on input. 6104 6105 Level: developer 6106 6107 Concepts: matrices^column ownership 6108 6109 .seealso: MatGetOwnershipRange(), MatGetOwnershipRanges(), MatGetOwnershipRangesColumn() 6110 6111 @*/ 6112 PetscErrorCode MatGetOwnershipRangeColumn(Mat mat,PetscInt *m,PetscInt *n) 6113 { 6114 PetscFunctionBegin; 6115 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6116 PetscValidType(mat,1); 6117 if (m) PetscValidIntPointer(m,2); 6118 if (n) PetscValidIntPointer(n,3); 6119 MatCheckPreallocated(mat,1); 6120 if (m) *m = mat->cmap->rstart; 6121 if (n) *n = mat->cmap->rend; 6122 PetscFunctionReturn(0); 6123 } 6124 6125 #undef __FUNCT__ 6126 #define __FUNCT__ "MatGetOwnershipRange" 6127 /*@ 6128 MatGetOwnershipRange - Returns the range of matrix rows owned by 6129 this processor, assuming that the matrix is laid out with the first 6130 n1 rows on the first processor, the next n2 rows on the second, etc. 6131 For certain parallel layouts this range may not be well defined. 6132 6133 Not Collective 6134 6135 Input Parameters: 6136 . mat - the matrix 6137 6138 Output Parameters: 6139 + m - the global index of the first local row 6140 - n - one more than the global index of the last local row 6141 6142 Note: Both output parameters can be NULL on input. 6143 $ This function requires that the matrix be preallocated. If you have not preallocated, consider using 6144 $ PetscSplitOwnership(MPI_Comm comm, PetscInt *n, PetscInt *N) 6145 $ and then MPI_Scan() to calculate prefix sums of the local sizes. 6146 6147 Level: beginner 6148 6149 Concepts: matrices^row ownership 6150 6151 .seealso: MatGetOwnershipRanges(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn(), PetscSplitOwnership(), PetscSplitOwnershipBlock() 6152 6153 @*/ 6154 PetscErrorCode MatGetOwnershipRange(Mat mat,PetscInt *m,PetscInt *n) 6155 { 6156 PetscFunctionBegin; 6157 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6158 PetscValidType(mat,1); 6159 if (m) PetscValidIntPointer(m,2); 6160 if (n) PetscValidIntPointer(n,3); 6161 MatCheckPreallocated(mat,1); 6162 if (m) *m = mat->rmap->rstart; 6163 if (n) *n = mat->rmap->rend; 6164 PetscFunctionReturn(0); 6165 } 6166 6167 #undef __FUNCT__ 6168 #define __FUNCT__ "MatGetOwnershipRanges" 6169 /*@C 6170 MatGetOwnershipRanges - Returns the range of matrix rows owned by 6171 each process 6172 6173 Not Collective, unless matrix has not been allocated, then collective on Mat 6174 6175 Input Parameters: 6176 . mat - the matrix 6177 6178 Output Parameters: 6179 . ranges - start of each processors portion plus one more then the total length at the end 6180 6181 Level: beginner 6182 6183 Concepts: matrices^row ownership 6184 6185 .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn() 6186 6187 @*/ 6188 PetscErrorCode MatGetOwnershipRanges(Mat mat,const PetscInt **ranges) 6189 { 6190 PetscErrorCode ierr; 6191 6192 PetscFunctionBegin; 6193 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6194 PetscValidType(mat,1); 6195 MatCheckPreallocated(mat,1); 6196 ierr = PetscLayoutGetRanges(mat->rmap,ranges);CHKERRQ(ierr); 6197 PetscFunctionReturn(0); 6198 } 6199 6200 #undef __FUNCT__ 6201 #define __FUNCT__ "MatGetOwnershipRangesColumn" 6202 /*@C 6203 MatGetOwnershipRangesColumn - Returns the range of matrix columns associated with rows of a vector one multiplies by that owned by 6204 this processor. (The columns of the "diagonal blocks" for each process) 6205 6206 Not Collective, unless matrix has not been allocated, then collective on Mat 6207 6208 Input Parameters: 6209 . mat - the matrix 6210 6211 Output Parameters: 6212 . ranges - start of each processors portion plus one more then the total length at the end 6213 6214 Level: beginner 6215 6216 Concepts: matrices^column ownership 6217 6218 .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRanges() 6219 6220 @*/ 6221 PetscErrorCode MatGetOwnershipRangesColumn(Mat mat,const PetscInt **ranges) 6222 { 6223 PetscErrorCode ierr; 6224 6225 PetscFunctionBegin; 6226 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6227 PetscValidType(mat,1); 6228 MatCheckPreallocated(mat,1); 6229 ierr = PetscLayoutGetRanges(mat->cmap,ranges);CHKERRQ(ierr); 6230 PetscFunctionReturn(0); 6231 } 6232 6233 #undef __FUNCT__ 6234 #define __FUNCT__ "MatGetOwnershipIS" 6235 /*@C 6236 MatGetOwnershipIS - Get row and column ownership as index sets 6237 6238 Not Collective 6239 6240 Input Arguments: 6241 . A - matrix of type Elemental 6242 6243 Output Arguments: 6244 + rows - rows in which this process owns elements 6245 . cols - columns in which this process owns elements 6246 6247 Level: intermediate 6248 6249 .seealso: MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatSetValues(), MATELEMENTAL, MatSetValues() 6250 @*/ 6251 PetscErrorCode MatGetOwnershipIS(Mat A,IS *rows,IS *cols) 6252 { 6253 PetscErrorCode ierr,(*f)(Mat,IS*,IS*); 6254 6255 PetscFunctionBegin; 6256 MatCheckPreallocated(A,1); 6257 ierr = PetscObjectQueryFunction((PetscObject)A,"MatGetOwnershipIS_C",&f);CHKERRQ(ierr); 6258 if (f) { 6259 ierr = (*f)(A,rows,cols);CHKERRQ(ierr); 6260 } else { /* Create a standard row-based partition, each process is responsible for ALL columns in their row block */ 6261 if (rows) {ierr = ISCreateStride(PETSC_COMM_SELF,A->rmap->n,A->rmap->rstart,1,rows);CHKERRQ(ierr);} 6262 if (cols) {ierr = ISCreateStride(PETSC_COMM_SELF,A->cmap->N,0,1,cols);CHKERRQ(ierr);} 6263 } 6264 PetscFunctionReturn(0); 6265 } 6266 6267 #undef __FUNCT__ 6268 #define __FUNCT__ "MatILUFactorSymbolic" 6269 /*@C 6270 MatILUFactorSymbolic - Performs symbolic ILU factorization of a matrix. 6271 Uses levels of fill only, not drop tolerance. Use MatLUFactorNumeric() 6272 to complete the factorization. 6273 6274 Collective on Mat 6275 6276 Input Parameters: 6277 + mat - the matrix 6278 . row - row permutation 6279 . column - column permutation 6280 - info - structure containing 6281 $ levels - number of levels of fill. 6282 $ expected fill - as ratio of original fill. 6283 $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices 6284 missing diagonal entries) 6285 6286 Output Parameters: 6287 . fact - new matrix that has been symbolically factored 6288 6289 Notes: 6290 See the <a href="../../docs/manual.pdf">users manual</a> for additional information about 6291 choosing the fill factor for better efficiency. 6292 6293 Most users should employ the simplified KSP interface for linear solvers 6294 instead of working directly with matrix algebra routines such as this. 6295 See, e.g., KSPCreate(). 6296 6297 Level: developer 6298 6299 Concepts: matrices^symbolic LU factorization 6300 Concepts: matrices^factorization 6301 Concepts: LU^symbolic factorization 6302 6303 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor() 6304 MatGetOrdering(), MatFactorInfo 6305 6306 Developer Note: fortran interface is not autogenerated as the f90 6307 interface defintion cannot be generated correctly [due to MatFactorInfo] 6308 6309 @*/ 6310 PetscErrorCode MatILUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info) 6311 { 6312 PetscErrorCode ierr; 6313 6314 PetscFunctionBegin; 6315 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6316 PetscValidType(mat,1); 6317 PetscValidHeaderSpecific(row,IS_CLASSID,2); 6318 PetscValidHeaderSpecific(col,IS_CLASSID,3); 6319 PetscValidPointer(info,4); 6320 PetscValidPointer(fact,5); 6321 if (info->levels < 0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Levels of fill negative %D",(PetscInt)info->levels); 6322 if (info->fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %G",info->fill); 6323 if (!(fact)->ops->ilufactorsymbolic) { 6324 const MatSolverPackage spackage; 6325 ierr = MatFactorGetSolverPackage(fact,&spackage);CHKERRQ(ierr); 6326 SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic ILU using solver package %s",((PetscObject)mat)->type_name,spackage); 6327 } 6328 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6329 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6330 MatCheckPreallocated(mat,2); 6331 6332 ierr = PetscLogEventBegin(MAT_ILUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr); 6333 ierr = (fact->ops->ilufactorsymbolic)(fact,mat,row,col,info);CHKERRQ(ierr); 6334 ierr = PetscLogEventEnd(MAT_ILUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr); 6335 PetscFunctionReturn(0); 6336 } 6337 6338 #undef __FUNCT__ 6339 #define __FUNCT__ "MatICCFactorSymbolic" 6340 /*@C 6341 MatICCFactorSymbolic - Performs symbolic incomplete 6342 Cholesky factorization for a symmetric matrix. Use 6343 MatCholeskyFactorNumeric() to complete the factorization. 6344 6345 Collective on Mat 6346 6347 Input Parameters: 6348 + mat - the matrix 6349 . perm - row and column permutation 6350 - info - structure containing 6351 $ levels - number of levels of fill. 6352 $ expected fill - as ratio of original fill. 6353 6354 Output Parameter: 6355 . fact - the factored matrix 6356 6357 Notes: 6358 Most users should employ the KSP interface for linear solvers 6359 instead of working directly with matrix algebra routines such as this. 6360 See, e.g., KSPCreate(). 6361 6362 Level: developer 6363 6364 Concepts: matrices^symbolic incomplete Cholesky factorization 6365 Concepts: matrices^factorization 6366 Concepts: Cholsky^symbolic factorization 6367 6368 .seealso: MatCholeskyFactorNumeric(), MatCholeskyFactor(), MatFactorInfo 6369 6370 Developer Note: fortran interface is not autogenerated as the f90 6371 interface defintion cannot be generated correctly [due to MatFactorInfo] 6372 6373 @*/ 6374 PetscErrorCode MatICCFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info) 6375 { 6376 PetscErrorCode ierr; 6377 6378 PetscFunctionBegin; 6379 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6380 PetscValidType(mat,1); 6381 PetscValidHeaderSpecific(perm,IS_CLASSID,2); 6382 PetscValidPointer(info,3); 6383 PetscValidPointer(fact,4); 6384 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6385 if (info->levels < 0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Levels negative %D",(PetscInt) info->levels); 6386 if (info->fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %G",info->fill); 6387 if (!(fact)->ops->iccfactorsymbolic) { 6388 const MatSolverPackage spackage; 6389 ierr = MatFactorGetSolverPackage(fact,&spackage);CHKERRQ(ierr); 6390 SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Matrix type %s symbolic ICC using solver package %s",((PetscObject)mat)->type_name,spackage); 6391 } 6392 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6393 MatCheckPreallocated(mat,2); 6394 6395 ierr = PetscLogEventBegin(MAT_ICCFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr); 6396 ierr = (fact->ops->iccfactorsymbolic)(fact,mat,perm,info);CHKERRQ(ierr); 6397 ierr = PetscLogEventEnd(MAT_ICCFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr); 6398 PetscFunctionReturn(0); 6399 } 6400 6401 #undef __FUNCT__ 6402 #define __FUNCT__ "MatGetSubMatrices" 6403 /*@C 6404 MatGetSubMatrices - Extracts several submatrices from a matrix. If submat 6405 points to an array of valid matrices, they may be reused to store the new 6406 submatrices. 6407 6408 Collective on Mat 6409 6410 Input Parameters: 6411 + mat - the matrix 6412 . n - the number of submatrixes to be extracted (on this processor, may be zero) 6413 . irow, icol - index sets of rows and columns to extract (must be sorted) 6414 - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 6415 6416 Output Parameter: 6417 . submat - the array of submatrices 6418 6419 Notes: 6420 MatGetSubMatrices() can extract ONLY sequential submatrices 6421 (from both sequential and parallel matrices). Use MatGetSubMatrix() 6422 to extract a parallel submatrix. 6423 6424 Currently both row and column indices must be sorted to guarantee 6425 correctness with all matrix types. 6426 6427 When extracting submatrices from a parallel matrix, each processor can 6428 form a different submatrix by setting the rows and columns of its 6429 individual index sets according to the local submatrix desired. 6430 6431 When finished using the submatrices, the user should destroy 6432 them with MatDestroyMatrices(). 6433 6434 MAT_REUSE_MATRIX can only be used when the nonzero structure of the 6435 original matrix has not changed from that last call to MatGetSubMatrices(). 6436 6437 This routine creates the matrices in submat; you should NOT create them before 6438 calling it. It also allocates the array of matrix pointers submat. 6439 6440 For BAIJ matrices the index sets must respect the block structure, that is if they 6441 request one row/column in a block, they must request all rows/columns that are in 6442 that block. For example, if the block size is 2 you cannot request just row 0 and 6443 column 0. 6444 6445 Fortran Note: 6446 The Fortran interface is slightly different from that given below; it 6447 requires one to pass in as submat a Mat (integer) array of size at least m. 6448 6449 Level: advanced 6450 6451 Concepts: matrices^accessing submatrices 6452 Concepts: submatrices 6453 6454 .seealso: MatDestroyMatrices(), MatGetSubMatrix(), MatGetRow(), MatGetDiagonal(), MatReuse 6455 @*/ 6456 PetscErrorCode MatGetSubMatrices(Mat mat,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *submat[]) 6457 { 6458 PetscErrorCode ierr; 6459 PetscInt i; 6460 PetscBool eq; 6461 6462 PetscFunctionBegin; 6463 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6464 PetscValidType(mat,1); 6465 if (n) { 6466 PetscValidPointer(irow,3); 6467 PetscValidHeaderSpecific(*irow,IS_CLASSID,3); 6468 PetscValidPointer(icol,4); 6469 PetscValidHeaderSpecific(*icol,IS_CLASSID,4); 6470 } 6471 PetscValidPointer(submat,6); 6472 if (n && scall == MAT_REUSE_MATRIX) { 6473 PetscValidPointer(*submat,6); 6474 PetscValidHeaderSpecific(**submat,MAT_CLASSID,6); 6475 } 6476 if (!mat->ops->getsubmatrices) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 6477 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6478 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6479 MatCheckPreallocated(mat,1); 6480 6481 ierr = PetscLogEventBegin(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr); 6482 ierr = (*mat->ops->getsubmatrices)(mat,n,irow,icol,scall,submat);CHKERRQ(ierr); 6483 ierr = PetscLogEventEnd(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr); 6484 for (i=0; i<n; i++) { 6485 if (mat->symmetric || mat->structurally_symmetric || mat->hermitian) { 6486 ierr = ISEqual(irow[i],icol[i],&eq);CHKERRQ(ierr); 6487 if (eq) { 6488 if (mat->symmetric) { 6489 ierr = MatSetOption((*submat)[i],MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr); 6490 } else if (mat->hermitian) { 6491 ierr = MatSetOption((*submat)[i],MAT_HERMITIAN,PETSC_TRUE);CHKERRQ(ierr); 6492 } else if (mat->structurally_symmetric) { 6493 ierr = MatSetOption((*submat)[i],MAT_STRUCTURALLY_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr); 6494 } 6495 } 6496 } 6497 } 6498 PetscFunctionReturn(0); 6499 } 6500 6501 #undef __FUNCT__ 6502 #define __FUNCT__ "MatGetSubMatricesParallel" 6503 PetscErrorCode MatGetSubMatricesParallel(Mat mat,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *submat[]) 6504 { 6505 PetscErrorCode ierr; 6506 PetscInt i; 6507 PetscBool eq; 6508 6509 PetscFunctionBegin; 6510 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6511 PetscValidType(mat,1); 6512 if (n) { 6513 PetscValidPointer(irow,3); 6514 PetscValidHeaderSpecific(*irow,IS_CLASSID,3); 6515 PetscValidPointer(icol,4); 6516 PetscValidHeaderSpecific(*icol,IS_CLASSID,4); 6517 } 6518 PetscValidPointer(submat,6); 6519 if (n && scall == MAT_REUSE_MATRIX) { 6520 PetscValidPointer(*submat,6); 6521 PetscValidHeaderSpecific(**submat,MAT_CLASSID,6); 6522 } 6523 if (!mat->ops->getsubmatricesparallel) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 6524 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6525 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6526 MatCheckPreallocated(mat,1); 6527 6528 ierr = PetscLogEventBegin(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr); 6529 ierr = (*mat->ops->getsubmatricesparallel)(mat,n,irow,icol,scall,submat);CHKERRQ(ierr); 6530 ierr = PetscLogEventEnd(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr); 6531 for (i=0; i<n; i++) { 6532 if (mat->symmetric || mat->structurally_symmetric || mat->hermitian) { 6533 ierr = ISEqual(irow[i],icol[i],&eq);CHKERRQ(ierr); 6534 if (eq) { 6535 if (mat->symmetric) { 6536 ierr = MatSetOption((*submat)[i],MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr); 6537 } else if (mat->hermitian) { 6538 ierr = MatSetOption((*submat)[i],MAT_HERMITIAN,PETSC_TRUE);CHKERRQ(ierr); 6539 } else if (mat->structurally_symmetric) { 6540 ierr = MatSetOption((*submat)[i],MAT_STRUCTURALLY_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr); 6541 } 6542 } 6543 } 6544 } 6545 PetscFunctionReturn(0); 6546 } 6547 6548 #undef __FUNCT__ 6549 #define __FUNCT__ "MatDestroyMatrices" 6550 /*@C 6551 MatDestroyMatrices - Destroys a set of matrices obtained with MatGetSubMatrices(). 6552 6553 Collective on Mat 6554 6555 Input Parameters: 6556 + n - the number of local matrices 6557 - mat - the matrices (note that this is a pointer to the array of matrices, just to match the calling 6558 sequence of MatGetSubMatrices()) 6559 6560 Level: advanced 6561 6562 Notes: Frees not only the matrices, but also the array that contains the matrices 6563 In Fortran will not free the array. 6564 6565 .seealso: MatGetSubMatrices() 6566 @*/ 6567 PetscErrorCode MatDestroyMatrices(PetscInt n,Mat *mat[]) 6568 { 6569 PetscErrorCode ierr; 6570 PetscInt i; 6571 6572 PetscFunctionBegin; 6573 if (!*mat) PetscFunctionReturn(0); 6574 if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Trying to destroy negative number of matrices %D",n); 6575 PetscValidPointer(mat,2); 6576 for (i=0; i<n; i++) { 6577 ierr = MatDestroy(&(*mat)[i]);CHKERRQ(ierr); 6578 } 6579 /* memory is allocated even if n = 0 */ 6580 ierr = PetscFree(*mat);CHKERRQ(ierr); 6581 *mat = NULL; 6582 PetscFunctionReturn(0); 6583 } 6584 6585 #undef __FUNCT__ 6586 #define __FUNCT__ "MatGetSeqNonzeroStructure" 6587 /*@C 6588 MatGetSeqNonzeroStructure - Extracts the sequential nonzero structure from a matrix. 6589 6590 Collective on Mat 6591 6592 Input Parameters: 6593 . mat - the matrix 6594 6595 Output Parameter: 6596 . matstruct - the sequential matrix with the nonzero structure of mat 6597 6598 Level: intermediate 6599 6600 .seealso: MatDestroySeqNonzeroStructure(), MatGetSubMatrices(), MatDestroyMatrices() 6601 @*/ 6602 PetscErrorCode MatGetSeqNonzeroStructure(Mat mat,Mat *matstruct) 6603 { 6604 PetscErrorCode ierr; 6605 6606 PetscFunctionBegin; 6607 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6608 PetscValidPointer(matstruct,2); 6609 6610 PetscValidType(mat,1); 6611 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6612 MatCheckPreallocated(mat,1); 6613 6614 if (!mat->ops->getseqnonzerostructure) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Not for matrix type %s\n",((PetscObject)mat)->type_name); 6615 ierr = PetscLogEventBegin(MAT_GetSeqNonzeroStructure,mat,0,0,0);CHKERRQ(ierr); 6616 ierr = (*mat->ops->getseqnonzerostructure)(mat,matstruct);CHKERRQ(ierr); 6617 ierr = PetscLogEventEnd(MAT_GetSeqNonzeroStructure,mat,0,0,0);CHKERRQ(ierr); 6618 PetscFunctionReturn(0); 6619 } 6620 6621 #undef __FUNCT__ 6622 #define __FUNCT__ "MatDestroySeqNonzeroStructure" 6623 /*@C 6624 MatDestroySeqNonzeroStructure - Destroys matrix obtained with MatGetSeqNonzeroStructure(). 6625 6626 Collective on Mat 6627 6628 Input Parameters: 6629 . mat - the matrix (note that this is a pointer to the array of matrices, just to match the calling 6630 sequence of MatGetSequentialNonzeroStructure()) 6631 6632 Level: advanced 6633 6634 Notes: Frees not only the matrices, but also the array that contains the matrices 6635 6636 .seealso: MatGetSeqNonzeroStructure() 6637 @*/ 6638 PetscErrorCode MatDestroySeqNonzeroStructure(Mat *mat) 6639 { 6640 PetscErrorCode ierr; 6641 6642 PetscFunctionBegin; 6643 PetscValidPointer(mat,1); 6644 ierr = MatDestroy(mat);CHKERRQ(ierr); 6645 PetscFunctionReturn(0); 6646 } 6647 6648 #undef __FUNCT__ 6649 #define __FUNCT__ "MatIncreaseOverlap" 6650 /*@ 6651 MatIncreaseOverlap - Given a set of submatrices indicated by index sets, 6652 replaces the index sets by larger ones that represent submatrices with 6653 additional overlap. 6654 6655 Collective on Mat 6656 6657 Input Parameters: 6658 + mat - the matrix 6659 . n - the number of index sets 6660 . is - the array of index sets (these index sets will changed during the call) 6661 - ov - the additional overlap requested 6662 6663 Level: developer 6664 6665 Concepts: overlap 6666 Concepts: ASM^computing overlap 6667 6668 .seealso: MatGetSubMatrices() 6669 @*/ 6670 PetscErrorCode MatIncreaseOverlap(Mat mat,PetscInt n,IS is[],PetscInt ov) 6671 { 6672 PetscErrorCode ierr; 6673 6674 PetscFunctionBegin; 6675 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6676 PetscValidType(mat,1); 6677 if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Must have one or more domains, you have %D",n); 6678 if (n) { 6679 PetscValidPointer(is,3); 6680 PetscValidHeaderSpecific(*is,IS_CLASSID,3); 6681 } 6682 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 6683 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 6684 MatCheckPreallocated(mat,1); 6685 6686 if (!ov) PetscFunctionReturn(0); 6687 if (!mat->ops->increaseoverlap) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 6688 ierr = PetscLogEventBegin(MAT_IncreaseOverlap,mat,0,0,0);CHKERRQ(ierr); 6689 ierr = (*mat->ops->increaseoverlap)(mat,n,is,ov);CHKERRQ(ierr); 6690 ierr = PetscLogEventEnd(MAT_IncreaseOverlap,mat,0,0,0);CHKERRQ(ierr); 6691 PetscFunctionReturn(0); 6692 } 6693 6694 #undef __FUNCT__ 6695 #define __FUNCT__ "MatGetBlockSize" 6696 /*@ 6697 MatGetBlockSize - Returns the matrix block size; useful especially for the 6698 block row and block diagonal formats. 6699 6700 Not Collective 6701 6702 Input Parameter: 6703 . mat - the matrix 6704 6705 Output Parameter: 6706 . bs - block size 6707 6708 Notes: 6709 Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ 6710 6711 Level: intermediate 6712 6713 Concepts: matrices^block size 6714 6715 .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSizes() 6716 @*/ 6717 PetscErrorCode MatGetBlockSize(Mat mat,PetscInt *bs) 6718 { 6719 PetscFunctionBegin; 6720 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6721 PetscValidType(mat,1); 6722 PetscValidIntPointer(bs,2); 6723 MatCheckPreallocated(mat,1); 6724 *bs = mat->rmap->bs; 6725 PetscFunctionReturn(0); 6726 } 6727 6728 #undef __FUNCT__ 6729 #define __FUNCT__ "MatGetBlockSizes" 6730 /*@ 6731 MatGetBlockSizes - Returns the matrix block row and column sizes; 6732 useful especially for the block row and block diagonal formats. 6733 6734 Not Collective 6735 6736 Input Parameter: 6737 . mat - the matrix 6738 6739 Output Parameter: 6740 . rbs - row block size 6741 . cbs - coumn block size 6742 6743 Notes: 6744 Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ 6745 6746 Level: intermediate 6747 6748 Concepts: matrices^block size 6749 6750 .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize() 6751 @*/ 6752 PetscErrorCode MatGetBlockSizes(Mat mat,PetscInt *rbs, PetscInt *cbs) 6753 { 6754 PetscFunctionBegin; 6755 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6756 PetscValidType(mat,1); 6757 if (rbs) PetscValidIntPointer(rbs,2); 6758 if (cbs) PetscValidIntPointer(cbs,3); 6759 MatCheckPreallocated(mat,1); 6760 if (rbs) *rbs = mat->rmap->bs; 6761 if (cbs) *cbs = mat->cmap->bs; 6762 PetscFunctionReturn(0); 6763 } 6764 6765 #undef __FUNCT__ 6766 #define __FUNCT__ "MatSetBlockSize" 6767 /*@ 6768 MatSetBlockSize - Sets the matrix block size. 6769 6770 Logically Collective on Mat 6771 6772 Input Parameters: 6773 + mat - the matrix 6774 - bs - block size 6775 6776 Notes: 6777 This must be called before MatSetUp() or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later 6778 6779 Level: intermediate 6780 6781 Concepts: matrices^block size 6782 6783 .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize() 6784 @*/ 6785 PetscErrorCode MatSetBlockSize(Mat mat,PetscInt bs) 6786 { 6787 PetscErrorCode ierr; 6788 6789 PetscFunctionBegin; 6790 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6791 PetscValidLogicalCollectiveInt(mat,bs,2); 6792 ierr = PetscLayoutSetBlockSize(mat->rmap,bs);CHKERRQ(ierr); 6793 ierr = PetscLayoutSetBlockSize(mat->cmap,bs);CHKERRQ(ierr); 6794 PetscFunctionReturn(0); 6795 } 6796 6797 #undef __FUNCT__ 6798 #define __FUNCT__ "MatSetBlockSizes" 6799 /*@ 6800 MatSetBlockSizes - Sets the matrix block row and column sizes. 6801 6802 Logically Collective on Mat 6803 6804 Input Parameters: 6805 + mat - the matrix 6806 - rbs - row block size 6807 - cbs - column block size 6808 6809 Notes: 6810 This must be called before MatSetUp() or MatXXXSetPreallocation() (or will default to 1) and the block size cannot be changed later 6811 6812 Level: intermediate 6813 6814 Concepts: matrices^block size 6815 6816 .seealso: MatCreateSeqBAIJ(), MatCreateBAIJ(), MatGetBlockSize() 6817 @*/ 6818 PetscErrorCode MatSetBlockSizes(Mat mat,PetscInt rbs,PetscInt cbs) 6819 { 6820 PetscErrorCode ierr; 6821 6822 PetscFunctionBegin; 6823 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6824 PetscValidLogicalCollectiveInt(mat,rbs,2); 6825 ierr = PetscLayoutSetBlockSize(mat->rmap,rbs);CHKERRQ(ierr); 6826 ierr = PetscLayoutSetBlockSize(mat->cmap,cbs);CHKERRQ(ierr); 6827 PetscFunctionReturn(0); 6828 } 6829 6830 #undef __FUNCT__ 6831 #define __FUNCT__ "MatGetRowIJ" 6832 /*@C 6833 MatGetRowIJ - Returns the compressed row storage i and j indices for sequential matrices. 6834 6835 Collective on Mat 6836 6837 Input Parameters: 6838 + mat - the matrix 6839 . shift - 0 or 1 indicating we want the indices starting at 0 or 1 6840 . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be symmetrized 6841 - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the 6842 inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is 6843 always used. 6844 6845 Output Parameters: 6846 + n - number of rows in the (possibly compressed) matrix 6847 . ia - the row pointers [of length n+1] 6848 . ja - the column indices 6849 - done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers 6850 are responsible for handling the case when done == PETSC_FALSE and ia and ja are not set 6851 6852 Level: developer 6853 6854 Notes: You CANNOT change any of the ia[] or ja[] values. 6855 6856 Use MatRestoreRowIJ() when you are finished accessing the ia[] and ja[] values 6857 6858 Fortran Node 6859 6860 In Fortran use 6861 $ PetscInt ia(1), ja(1) 6862 $ PetscOffset iia, jja 6863 $ call MatGetRowIJ(mat,shift,symmetric,inodecompressed,n,ia,iia,ja,jja,done,ierr) 6864 $ 6865 $ or 6866 $ 6867 $ PetscScalar, pointer :: xx_v(:) 6868 $ call MatGetRowIJF90(mat,shift,symmetric,inodecompressed,n,ia,ja,done,ierr) 6869 6870 6871 Acess the ith and jth entries via ia(iia + i) and ja(jja + j) 6872 6873 .seealso: MatGetColumnIJ(), MatRestoreRowIJ(), MatSeqAIJGetArray() 6874 @*/ 6875 PetscErrorCode MatGetRowIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 6876 { 6877 PetscErrorCode ierr; 6878 6879 PetscFunctionBegin; 6880 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6881 PetscValidType(mat,1); 6882 PetscValidIntPointer(n,4); 6883 if (ia) PetscValidIntPointer(ia,5); 6884 if (ja) PetscValidIntPointer(ja,6); 6885 PetscValidIntPointer(done,7); 6886 MatCheckPreallocated(mat,1); 6887 if (!mat->ops->getrowij) *done = PETSC_FALSE; 6888 else { 6889 *done = PETSC_TRUE; 6890 ierr = PetscLogEventBegin(MAT_GetRowIJ,mat,0,0,0);CHKERRQ(ierr); 6891 ierr = (*mat->ops->getrowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr); 6892 ierr = PetscLogEventEnd(MAT_GetRowIJ,mat,0,0,0);CHKERRQ(ierr); 6893 } 6894 PetscFunctionReturn(0); 6895 } 6896 6897 #undef __FUNCT__ 6898 #define __FUNCT__ "MatGetColumnIJ" 6899 /*@C 6900 MatGetColumnIJ - Returns the compressed column storage i and j indices for sequential matrices. 6901 6902 Collective on Mat 6903 6904 Input Parameters: 6905 + mat - the matrix 6906 . shift - 1 or zero indicating we want the indices starting at 0 or 1 6907 . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 6908 symmetrized 6909 - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the 6910 inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is 6911 always used. 6912 6913 Output Parameters: 6914 + n - number of columns in the (possibly compressed) matrix 6915 . ia - the column pointers 6916 . ja - the row indices 6917 - done - PETSC_TRUE or PETSC_FALSE, indicating whether the values have been returned 6918 6919 Level: developer 6920 6921 .seealso: MatGetRowIJ(), MatRestoreColumnIJ() 6922 @*/ 6923 PetscErrorCode MatGetColumnIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 6924 { 6925 PetscErrorCode ierr; 6926 6927 PetscFunctionBegin; 6928 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6929 PetscValidType(mat,1); 6930 PetscValidIntPointer(n,4); 6931 if (ia) PetscValidIntPointer(ia,5); 6932 if (ja) PetscValidIntPointer(ja,6); 6933 PetscValidIntPointer(done,7); 6934 MatCheckPreallocated(mat,1); 6935 if (!mat->ops->getcolumnij) *done = PETSC_FALSE; 6936 else { 6937 *done = PETSC_TRUE; 6938 ierr = (*mat->ops->getcolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr); 6939 } 6940 PetscFunctionReturn(0); 6941 } 6942 6943 #undef __FUNCT__ 6944 #define __FUNCT__ "MatRestoreRowIJ" 6945 /*@C 6946 MatRestoreRowIJ - Call after you are completed with the ia,ja indices obtained with 6947 MatGetRowIJ(). 6948 6949 Collective on Mat 6950 6951 Input Parameters: 6952 + mat - the matrix 6953 . shift - 1 or zero indicating we want the indices starting at 0 or 1 6954 . symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 6955 symmetrized 6956 - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the 6957 inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is 6958 always used. 6959 6960 Output Parameters: 6961 + n - size of (possibly compressed) matrix 6962 . ia - the row pointers 6963 . ja - the column indices 6964 - done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned 6965 6966 Level: developer 6967 6968 .seealso: MatGetRowIJ(), MatRestoreColumnIJ() 6969 @*/ 6970 PetscErrorCode MatRestoreRowIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 6971 { 6972 PetscErrorCode ierr; 6973 6974 PetscFunctionBegin; 6975 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 6976 PetscValidType(mat,1); 6977 if (ia) PetscValidIntPointer(ia,5); 6978 if (ja) PetscValidIntPointer(ja,6); 6979 PetscValidIntPointer(done,7); 6980 MatCheckPreallocated(mat,1); 6981 6982 if (!mat->ops->restorerowij) *done = PETSC_FALSE; 6983 else { 6984 *done = PETSC_TRUE; 6985 ierr = (*mat->ops->restorerowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr); 6986 if (n) *n = 0; 6987 if (ia) *ia = NULL; 6988 if (ja) *ja = NULL; 6989 } 6990 PetscFunctionReturn(0); 6991 } 6992 6993 #undef __FUNCT__ 6994 #define __FUNCT__ "MatRestoreColumnIJ" 6995 /*@C 6996 MatRestoreColumnIJ - Call after you are completed with the ia,ja indices obtained with 6997 MatGetColumnIJ(). 6998 6999 Collective on Mat 7000 7001 Input Parameters: 7002 + mat - the matrix 7003 . shift - 1 or zero indicating we want the indices starting at 0 or 1 7004 - symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 7005 symmetrized 7006 - inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the 7007 inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is 7008 always used. 7009 7010 Output Parameters: 7011 + n - size of (possibly compressed) matrix 7012 . ia - the column pointers 7013 . ja - the row indices 7014 - done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned 7015 7016 Level: developer 7017 7018 .seealso: MatGetColumnIJ(), MatRestoreRowIJ() 7019 @*/ 7020 PetscErrorCode MatRestoreColumnIJ(Mat mat,PetscInt shift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 7021 { 7022 PetscErrorCode ierr; 7023 7024 PetscFunctionBegin; 7025 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7026 PetscValidType(mat,1); 7027 if (ia) PetscValidIntPointer(ia,5); 7028 if (ja) PetscValidIntPointer(ja,6); 7029 PetscValidIntPointer(done,7); 7030 MatCheckPreallocated(mat,1); 7031 7032 if (!mat->ops->restorecolumnij) *done = PETSC_FALSE; 7033 else { 7034 *done = PETSC_TRUE; 7035 ierr = (*mat->ops->restorecolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr); 7036 if (n) *n = 0; 7037 if (ia) *ia = NULL; 7038 if (ja) *ja = NULL; 7039 } 7040 PetscFunctionReturn(0); 7041 } 7042 7043 #undef __FUNCT__ 7044 #define __FUNCT__ "MatColoringPatch" 7045 /*@C 7046 MatColoringPatch -Used inside matrix coloring routines that 7047 use MatGetRowIJ() and/or MatGetColumnIJ(). 7048 7049 Collective on Mat 7050 7051 Input Parameters: 7052 + mat - the matrix 7053 . ncolors - max color value 7054 . n - number of entries in colorarray 7055 - colorarray - array indicating color for each column 7056 7057 Output Parameters: 7058 . iscoloring - coloring generated using colorarray information 7059 7060 Level: developer 7061 7062 .seealso: MatGetRowIJ(), MatGetColumnIJ() 7063 7064 @*/ 7065 PetscErrorCode MatColoringPatch(Mat mat,PetscInt ncolors,PetscInt n,ISColoringValue colorarray[],ISColoring *iscoloring) 7066 { 7067 PetscErrorCode ierr; 7068 7069 PetscFunctionBegin; 7070 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7071 PetscValidType(mat,1); 7072 PetscValidIntPointer(colorarray,4); 7073 PetscValidPointer(iscoloring,5); 7074 MatCheckPreallocated(mat,1); 7075 7076 if (!mat->ops->coloringpatch) { 7077 ierr = ISColoringCreate(PetscObjectComm((PetscObject)mat),ncolors,n,colorarray,iscoloring);CHKERRQ(ierr); 7078 } else { 7079 ierr = (*mat->ops->coloringpatch)(mat,ncolors,n,colorarray,iscoloring);CHKERRQ(ierr); 7080 } 7081 PetscFunctionReturn(0); 7082 } 7083 7084 7085 #undef __FUNCT__ 7086 #define __FUNCT__ "MatSetUnfactored" 7087 /*@ 7088 MatSetUnfactored - Resets a factored matrix to be treated as unfactored. 7089 7090 Logically Collective on Mat 7091 7092 Input Parameter: 7093 . mat - the factored matrix to be reset 7094 7095 Notes: 7096 This routine should be used only with factored matrices formed by in-place 7097 factorization via ILU(0) (or by in-place LU factorization for the MATSEQDENSE 7098 format). This option can save memory, for example, when solving nonlinear 7099 systems with a matrix-free Newton-Krylov method and a matrix-based, in-place 7100 ILU(0) preconditioner. 7101 7102 Note that one can specify in-place ILU(0) factorization by calling 7103 .vb 7104 PCType(pc,PCILU); 7105 PCFactorSeUseInPlace(pc); 7106 .ve 7107 or by using the options -pc_type ilu -pc_factor_in_place 7108 7109 In-place factorization ILU(0) can also be used as a local 7110 solver for the blocks within the block Jacobi or additive Schwarz 7111 methods (runtime option: -sub_pc_factor_in_place). See the discussion 7112 of these preconditioners in the <a href="../../docs/manual.pdf#ch_pc">PC chapter of the users manual</a> for details on setting 7113 local solver options. 7114 7115 Most users should employ the simplified KSP interface for linear solvers 7116 instead of working directly with matrix algebra routines such as this. 7117 See, e.g., KSPCreate(). 7118 7119 Level: developer 7120 7121 .seealso: PCFactorSetUseInPlace() 7122 7123 Concepts: matrices^unfactored 7124 7125 @*/ 7126 PetscErrorCode MatSetUnfactored(Mat mat) 7127 { 7128 PetscErrorCode ierr; 7129 7130 PetscFunctionBegin; 7131 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7132 PetscValidType(mat,1); 7133 MatCheckPreallocated(mat,1); 7134 mat->factortype = MAT_FACTOR_NONE; 7135 if (!mat->ops->setunfactored) PetscFunctionReturn(0); 7136 ierr = (*mat->ops->setunfactored)(mat);CHKERRQ(ierr); 7137 PetscFunctionReturn(0); 7138 } 7139 7140 /*MC 7141 MatDenseGetArrayF90 - Accesses a matrix array from Fortran90. 7142 7143 Synopsis: 7144 MatDenseGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:,:)},integer ierr) 7145 7146 Not collective 7147 7148 Input Parameter: 7149 . x - matrix 7150 7151 Output Parameters: 7152 + xx_v - the Fortran90 pointer to the array 7153 - ierr - error code 7154 7155 Example of Usage: 7156 .vb 7157 PetscScalar, pointer xx_v(:,:) 7158 .... 7159 call MatDenseGetArrayF90(x,xx_v,ierr) 7160 a = xx_v(3) 7161 call MatDenseRestoreArrayF90(x,xx_v,ierr) 7162 .ve 7163 7164 Level: advanced 7165 7166 .seealso: MatDenseRestoreArrayF90(), MatDenseGetArray(), MatDenseRestoreArray() 7167 7168 Concepts: matrices^accessing array 7169 7170 M*/ 7171 7172 /*MC 7173 MatDenseRestoreArrayF90 - Restores a matrix array that has been 7174 accessed with MatGetArrayF90(). 7175 7176 Synopsis: 7177 MatDenseRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr) 7178 7179 Not collective 7180 7181 Input Parameters: 7182 + x - matrix 7183 - xx_v - the Fortran90 pointer to the array 7184 7185 Output Parameter: 7186 . ierr - error code 7187 7188 Example of Usage: 7189 .vb 7190 PetscScalar, pointer xx_v(:) 7191 .... 7192 call MatDenseGetArrayF90(x,xx_v,ierr) 7193 a = xx_v(3) 7194 call MatDenseRestoreArrayF90(x,xx_v,ierr) 7195 .ve 7196 7197 Level: advanced 7198 7199 .seealso: MatDenseGetArrayF90(), MatDenseGetArray(), MatDenseRestoreArray() 7200 7201 M*/ 7202 7203 7204 #undef __FUNCT__ 7205 #define __FUNCT__ "MatGetSubMatrix" 7206 /*@ 7207 MatGetSubMatrix - Gets a single submatrix on the same number of processors 7208 as the original matrix. 7209 7210 Collective on Mat 7211 7212 Input Parameters: 7213 + mat - the original matrix 7214 . isrow - parallel IS containing the rows this processor should obtain 7215 . iscol - parallel IS containing all columns you wish to keep. Each process should list the columns that will be in IT's "diagonal part" in the new matrix. 7216 - cll - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 7217 7218 Output Parameter: 7219 . newmat - the new submatrix, of the same type as the old 7220 7221 Level: advanced 7222 7223 Notes: 7224 The submatrix will be able to be multiplied with vectors using the same layout as iscol. 7225 7226 The rows in isrow will be sorted into the same order as the original matrix on each process. 7227 7228 The first time this is called you should use a cll of MAT_INITIAL_MATRIX, 7229 the MatGetSubMatrix() routine will create the newmat for you. Any additional calls 7230 to this routine with a mat of the same nonzero structure and with a call of MAT_REUSE_MATRIX 7231 will reuse the matrix generated the first time. You should call MatDestroy() on newmat when 7232 you are finished using it. 7233 7234 The communicator of the newly obtained matrix is ALWAYS the same as the communicator of 7235 the input matrix. 7236 7237 If iscol is NULL then all columns are obtained (not supported in Fortran). 7238 7239 Example usage: 7240 Consider the following 8x8 matrix with 34 non-zero values, that is 7241 assembled across 3 processors. Let's assume that proc0 owns 3 rows, 7242 proc1 owns 3 rows, proc2 owns 2 rows. This division can be shown 7243 as follows: 7244 7245 .vb 7246 1 2 0 | 0 3 0 | 0 4 7247 Proc0 0 5 6 | 7 0 0 | 8 0 7248 9 0 10 | 11 0 0 | 12 0 7249 ------------------------------------- 7250 13 0 14 | 15 16 17 | 0 0 7251 Proc1 0 18 0 | 19 20 21 | 0 0 7252 0 0 0 | 22 23 0 | 24 0 7253 ------------------------------------- 7254 Proc2 25 26 27 | 0 0 28 | 29 0 7255 30 0 0 | 31 32 33 | 0 34 7256 .ve 7257 7258 Suppose isrow = [0 1 | 4 | 6 7] and iscol = [1 2 | 3 4 5 | 6]. The resulting submatrix is 7259 7260 .vb 7261 2 0 | 0 3 0 | 0 7262 Proc0 5 6 | 7 0 0 | 8 7263 ------------------------------- 7264 Proc1 18 0 | 19 20 21 | 0 7265 ------------------------------- 7266 Proc2 26 27 | 0 0 28 | 29 7267 0 0 | 31 32 33 | 0 7268 .ve 7269 7270 7271 Concepts: matrices^submatrices 7272 7273 .seealso: MatGetSubMatrices() 7274 @*/ 7275 PetscErrorCode MatGetSubMatrix(Mat mat,IS isrow,IS iscol,MatReuse cll,Mat *newmat) 7276 { 7277 PetscErrorCode ierr; 7278 PetscMPIInt size; 7279 Mat *local; 7280 IS iscoltmp; 7281 7282 PetscFunctionBegin; 7283 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7284 PetscValidHeaderSpecific(isrow,IS_CLASSID,2); 7285 if (iscol) PetscValidHeaderSpecific(iscol,IS_CLASSID,3); 7286 PetscValidPointer(newmat,5); 7287 if (cll == MAT_REUSE_MATRIX) PetscValidHeaderSpecific(*newmat,MAT_CLASSID,5); 7288 PetscValidType(mat,1); 7289 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 7290 MatCheckPreallocated(mat,1); 7291 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);CHKERRQ(ierr); 7292 7293 if (!iscol) { 7294 ierr = ISCreateStride(PetscObjectComm((PetscObject)mat),mat->cmap->n,mat->cmap->rstart,1,&iscoltmp);CHKERRQ(ierr); 7295 } else { 7296 iscoltmp = iscol; 7297 } 7298 7299 /* if original matrix is on just one processor then use submatrix generated */ 7300 if (mat->ops->getsubmatrices && !mat->ops->getsubmatrix && size == 1 && cll == MAT_REUSE_MATRIX) { 7301 ierr = MatGetSubMatrices(mat,1,&isrow,&iscoltmp,MAT_REUSE_MATRIX,&newmat);CHKERRQ(ierr); 7302 if (!iscol) {ierr = ISDestroy(&iscoltmp);CHKERRQ(ierr);} 7303 PetscFunctionReturn(0); 7304 } else if (mat->ops->getsubmatrices && !mat->ops->getsubmatrix && size == 1) { 7305 ierr = MatGetSubMatrices(mat,1,&isrow,&iscoltmp,MAT_INITIAL_MATRIX,&local);CHKERRQ(ierr); 7306 *newmat = *local; 7307 ierr = PetscFree(local);CHKERRQ(ierr); 7308 if (!iscol) {ierr = ISDestroy(&iscoltmp);CHKERRQ(ierr);} 7309 PetscFunctionReturn(0); 7310 } else if (!mat->ops->getsubmatrix) { 7311 /* Create a new matrix type that implements the operation using the full matrix */ 7312 switch (cll) { 7313 case MAT_INITIAL_MATRIX: 7314 ierr = MatCreateSubMatrix(mat,isrow,iscoltmp,newmat);CHKERRQ(ierr); 7315 break; 7316 case MAT_REUSE_MATRIX: 7317 ierr = MatSubMatrixUpdate(*newmat,mat,isrow,iscoltmp);CHKERRQ(ierr); 7318 break; 7319 default: SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"Invalid MatReuse, must be either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX"); 7320 } 7321 if (!iscol) {ierr = ISDestroy(&iscoltmp);CHKERRQ(ierr);} 7322 PetscFunctionReturn(0); 7323 } 7324 7325 if (!mat->ops->getsubmatrix) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 7326 ierr = (*mat->ops->getsubmatrix)(mat,isrow,iscoltmp,cll,newmat);CHKERRQ(ierr); 7327 if (!iscol) {ierr = ISDestroy(&iscoltmp);CHKERRQ(ierr);} 7328 if (*newmat && cll == MAT_INITIAL_MATRIX) {ierr = PetscObjectStateIncrease((PetscObject)*newmat);CHKERRQ(ierr);} 7329 PetscFunctionReturn(0); 7330 } 7331 7332 #undef __FUNCT__ 7333 #define __FUNCT__ "MatStashSetInitialSize" 7334 /*@ 7335 MatStashSetInitialSize - sets the sizes of the matrix stash, that is 7336 used during the assembly process to store values that belong to 7337 other processors. 7338 7339 Not Collective 7340 7341 Input Parameters: 7342 + mat - the matrix 7343 . size - the initial size of the stash. 7344 - bsize - the initial size of the block-stash(if used). 7345 7346 Options Database Keys: 7347 + -matstash_initial_size <size> or <size0,size1,...sizep-1> 7348 - -matstash_block_initial_size <bsize> or <bsize0,bsize1,...bsizep-1> 7349 7350 Level: intermediate 7351 7352 Notes: 7353 The block-stash is used for values set with MatSetValuesBlocked() while 7354 the stash is used for values set with MatSetValues() 7355 7356 Run with the option -info and look for output of the form 7357 MatAssemblyBegin_MPIXXX:Stash has MM entries, uses nn mallocs. 7358 to determine the appropriate value, MM, to use for size and 7359 MatAssemblyBegin_MPIXXX:Block-Stash has BMM entries, uses nn mallocs. 7360 to determine the value, BMM to use for bsize 7361 7362 Concepts: stash^setting matrix size 7363 Concepts: matrices^stash 7364 7365 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashGetInfo() 7366 7367 @*/ 7368 PetscErrorCode MatStashSetInitialSize(Mat mat,PetscInt size, PetscInt bsize) 7369 { 7370 PetscErrorCode ierr; 7371 7372 PetscFunctionBegin; 7373 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7374 PetscValidType(mat,1); 7375 ierr = MatStashSetInitialSize_Private(&mat->stash,size);CHKERRQ(ierr); 7376 ierr = MatStashSetInitialSize_Private(&mat->bstash,bsize);CHKERRQ(ierr); 7377 PetscFunctionReturn(0); 7378 } 7379 7380 #undef __FUNCT__ 7381 #define __FUNCT__ "MatInterpolateAdd" 7382 /*@ 7383 MatInterpolateAdd - w = y + A*x or A'*x depending on the shape of 7384 the matrix 7385 7386 Neighbor-wise Collective on Mat 7387 7388 Input Parameters: 7389 + mat - the matrix 7390 . x,y - the vectors 7391 - w - where the result is stored 7392 7393 Level: intermediate 7394 7395 Notes: 7396 w may be the same vector as y. 7397 7398 This allows one to use either the restriction or interpolation (its transpose) 7399 matrix to do the interpolation 7400 7401 Concepts: interpolation 7402 7403 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict() 7404 7405 @*/ 7406 PetscErrorCode MatInterpolateAdd(Mat A,Vec x,Vec y,Vec w) 7407 { 7408 PetscErrorCode ierr; 7409 PetscInt M,N,Ny; 7410 7411 PetscFunctionBegin; 7412 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 7413 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 7414 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 7415 PetscValidHeaderSpecific(w,VEC_CLASSID,4); 7416 PetscValidType(A,1); 7417 MatCheckPreallocated(A,1); 7418 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 7419 ierr = VecGetSize(y,&Ny);CHKERRQ(ierr); 7420 if (M == Ny) { 7421 ierr = MatMultAdd(A,x,y,w);CHKERRQ(ierr); 7422 } else { 7423 ierr = MatMultTransposeAdd(A,x,y,w);CHKERRQ(ierr); 7424 } 7425 PetscFunctionReturn(0); 7426 } 7427 7428 #undef __FUNCT__ 7429 #define __FUNCT__ "MatInterpolate" 7430 /*@ 7431 MatInterpolate - y = A*x or A'*x depending on the shape of 7432 the matrix 7433 7434 Neighbor-wise Collective on Mat 7435 7436 Input Parameters: 7437 + mat - the matrix 7438 - x,y - the vectors 7439 7440 Level: intermediate 7441 7442 Notes: 7443 This allows one to use either the restriction or interpolation (its transpose) 7444 matrix to do the interpolation 7445 7446 Concepts: matrices^interpolation 7447 7448 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict() 7449 7450 @*/ 7451 PetscErrorCode MatInterpolate(Mat A,Vec x,Vec y) 7452 { 7453 PetscErrorCode ierr; 7454 PetscInt M,N,Ny; 7455 7456 PetscFunctionBegin; 7457 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 7458 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 7459 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 7460 PetscValidType(A,1); 7461 MatCheckPreallocated(A,1); 7462 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 7463 ierr = VecGetSize(y,&Ny);CHKERRQ(ierr); 7464 if (M == Ny) { 7465 ierr = MatMult(A,x,y);CHKERRQ(ierr); 7466 } else { 7467 ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr); 7468 } 7469 PetscFunctionReturn(0); 7470 } 7471 7472 #undef __FUNCT__ 7473 #define __FUNCT__ "MatRestrict" 7474 /*@ 7475 MatRestrict - y = A*x or A'*x 7476 7477 Neighbor-wise Collective on Mat 7478 7479 Input Parameters: 7480 + mat - the matrix 7481 - x,y - the vectors 7482 7483 Level: intermediate 7484 7485 Notes: 7486 This allows one to use either the restriction or interpolation (its transpose) 7487 matrix to do the restriction 7488 7489 Concepts: matrices^restriction 7490 7491 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatInterpolate() 7492 7493 @*/ 7494 PetscErrorCode MatRestrict(Mat A,Vec x,Vec y) 7495 { 7496 PetscErrorCode ierr; 7497 PetscInt M,N,Ny; 7498 7499 PetscFunctionBegin; 7500 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 7501 PetscValidHeaderSpecific(x,VEC_CLASSID,2); 7502 PetscValidHeaderSpecific(y,VEC_CLASSID,3); 7503 PetscValidType(A,1); 7504 MatCheckPreallocated(A,1); 7505 7506 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 7507 ierr = VecGetSize(y,&Ny);CHKERRQ(ierr); 7508 if (M == Ny) { 7509 ierr = MatMult(A,x,y);CHKERRQ(ierr); 7510 } else { 7511 ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr); 7512 } 7513 PetscFunctionReturn(0); 7514 } 7515 7516 #undef __FUNCT__ 7517 #define __FUNCT__ "MatGetNullSpace" 7518 /*@ 7519 MatGetNullSpace - retrieves the null space to a matrix. 7520 7521 Logically Collective on Mat and MatNullSpace 7522 7523 Input Parameters: 7524 + mat - the matrix 7525 - nullsp - the null space object 7526 7527 Level: developer 7528 7529 Notes: 7530 This null space is used by solvers. Overwrites any previous null space that may have been attached 7531 7532 Concepts: null space^attaching to matrix 7533 7534 .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace() 7535 @*/ 7536 PetscErrorCode MatGetNullSpace(Mat mat, MatNullSpace *nullsp) 7537 { 7538 PetscFunctionBegin; 7539 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7540 PetscValidType(mat,1); 7541 PetscValidPointer(nullsp,2); 7542 *nullsp = mat->nullsp; 7543 PetscFunctionReturn(0); 7544 } 7545 7546 #undef __FUNCT__ 7547 #define __FUNCT__ "MatSetNullSpace" 7548 /*@ 7549 MatSetNullSpace - attaches a null space to a matrix. 7550 This null space will be removed from the resulting vector whenever 7551 MatMult() is called 7552 7553 Logically Collective on Mat and MatNullSpace 7554 7555 Input Parameters: 7556 + mat - the matrix 7557 - nullsp - the null space object 7558 7559 Level: advanced 7560 7561 Notes: 7562 This null space is used by solvers. Overwrites any previous null space that may have been attached 7563 7564 Concepts: null space^attaching to matrix 7565 7566 .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNearNullSpace() 7567 @*/ 7568 PetscErrorCode MatSetNullSpace(Mat mat,MatNullSpace nullsp) 7569 { 7570 PetscErrorCode ierr; 7571 7572 PetscFunctionBegin; 7573 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7574 PetscValidType(mat,1); 7575 PetscValidHeaderSpecific(nullsp,MAT_NULLSPACE_CLASSID,2); 7576 MatCheckPreallocated(mat,1); 7577 ierr = PetscObjectReference((PetscObject)nullsp);CHKERRQ(ierr); 7578 ierr = MatNullSpaceDestroy(&mat->nullsp);CHKERRQ(ierr); 7579 7580 mat->nullsp = nullsp; 7581 PetscFunctionReturn(0); 7582 } 7583 7584 #undef __FUNCT__ 7585 #define __FUNCT__ "MatSetNearNullSpace" 7586 /*@ 7587 MatSetNearNullSpace - attaches a null space to a matrix. 7588 This null space will be used to provide near null space vectors to a multigrid preconditioner built from this matrix. 7589 7590 Logically Collective on Mat and MatNullSpace 7591 7592 Input Parameters: 7593 + mat - the matrix 7594 - nullsp - the null space object 7595 7596 Level: advanced 7597 7598 Notes: 7599 Overwrites any previous near null space that may have been attached 7600 7601 Concepts: null space^attaching to matrix 7602 7603 .seealso: MatCreate(), MatNullSpaceCreate(), MatSetNullSpace() 7604 @*/ 7605 PetscErrorCode MatSetNearNullSpace(Mat mat,MatNullSpace nullsp) 7606 { 7607 PetscErrorCode ierr; 7608 7609 PetscFunctionBegin; 7610 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7611 PetscValidType(mat,1); 7612 PetscValidHeaderSpecific(nullsp,MAT_NULLSPACE_CLASSID,2); 7613 MatCheckPreallocated(mat,1); 7614 ierr = PetscObjectReference((PetscObject)nullsp);CHKERRQ(ierr); 7615 ierr = MatNullSpaceDestroy(&mat->nearnullsp);CHKERRQ(ierr); 7616 7617 mat->nearnullsp = nullsp; 7618 PetscFunctionReturn(0); 7619 } 7620 7621 #undef __FUNCT__ 7622 #define __FUNCT__ "MatGetNearNullSpace" 7623 /*@ 7624 MatGetNearNullSpace -Get null space attached with MatSetNearNullSpace() 7625 7626 Not Collective 7627 7628 Input Parameters: 7629 . mat - the matrix 7630 7631 Output Parameters: 7632 . nullsp - the null space object, NULL if not set 7633 7634 Level: developer 7635 7636 Concepts: null space^attaching to matrix 7637 7638 .seealso: MatSetNearNullSpace(), MatGetNullSpace() 7639 @*/ 7640 PetscErrorCode MatGetNearNullSpace(Mat mat,MatNullSpace *nullsp) 7641 { 7642 PetscFunctionBegin; 7643 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7644 PetscValidType(mat,1); 7645 PetscValidPointer(nullsp,2); 7646 MatCheckPreallocated(mat,1); 7647 *nullsp = mat->nearnullsp; 7648 PetscFunctionReturn(0); 7649 } 7650 7651 #undef __FUNCT__ 7652 #define __FUNCT__ "MatICCFactor" 7653 /*@C 7654 MatICCFactor - Performs in-place incomplete Cholesky factorization of matrix. 7655 7656 Collective on Mat 7657 7658 Input Parameters: 7659 + mat - the matrix 7660 . row - row/column permutation 7661 . fill - expected fill factor >= 1.0 7662 - level - level of fill, for ICC(k) 7663 7664 Notes: 7665 Probably really in-place only when level of fill is zero, otherwise allocates 7666 new space to store factored matrix and deletes previous memory. 7667 7668 Most users should employ the simplified KSP interface for linear solvers 7669 instead of working directly with matrix algebra routines such as this. 7670 See, e.g., KSPCreate(). 7671 7672 Level: developer 7673 7674 Concepts: matrices^incomplete Cholesky factorization 7675 Concepts: Cholesky factorization 7676 7677 .seealso: MatICCFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor() 7678 7679 Developer Note: fortran interface is not autogenerated as the f90 7680 interface defintion cannot be generated correctly [due to MatFactorInfo] 7681 7682 @*/ 7683 PetscErrorCode MatICCFactor(Mat mat,IS row,const MatFactorInfo *info) 7684 { 7685 PetscErrorCode ierr; 7686 7687 PetscFunctionBegin; 7688 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7689 PetscValidType(mat,1); 7690 if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2); 7691 PetscValidPointer(info,3); 7692 if (mat->rmap->N != mat->cmap->N) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONG,"matrix must be square"); 7693 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 7694 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 7695 if (!mat->ops->iccfactor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 7696 MatCheckPreallocated(mat,1); 7697 ierr = (*mat->ops->iccfactor)(mat,row,info);CHKERRQ(ierr); 7698 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 7699 PetscFunctionReturn(0); 7700 } 7701 7702 #undef __FUNCT__ 7703 #define __FUNCT__ "MatSetValuesAdifor" 7704 /*@ 7705 MatSetValuesAdifor - Sets values computed with automatic differentiation into a matrix. 7706 7707 Not Collective 7708 7709 Input Parameters: 7710 + mat - the matrix 7711 . nl - leading dimension of v 7712 - v - the values compute with ADIFOR 7713 7714 Level: developer 7715 7716 Notes: 7717 Must call MatSetColoring() before using this routine. Also this matrix must already 7718 have its nonzero pattern determined. 7719 7720 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(), 7721 MatSetValues(), MatSetColoring() 7722 @*/ 7723 PetscErrorCode MatSetValuesAdifor(Mat mat,PetscInt nl,void *v) 7724 { 7725 PetscErrorCode ierr; 7726 7727 PetscFunctionBegin; 7728 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7729 PetscValidType(mat,1); 7730 PetscValidPointer(v,3); 7731 7732 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled"); 7733 ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 7734 if (!mat->ops->setvaluesadifor) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 7735 ierr = (*mat->ops->setvaluesadifor)(mat,nl,v);CHKERRQ(ierr); 7736 ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr); 7737 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 7738 PetscFunctionReturn(0); 7739 } 7740 7741 #undef __FUNCT__ 7742 #define __FUNCT__ "MatDiagonalScaleLocal" 7743 /*@ 7744 MatDiagonalScaleLocal - Scales columns of a matrix given the scaling values including the 7745 ghosted ones. 7746 7747 Not Collective 7748 7749 Input Parameters: 7750 + mat - the matrix 7751 - diag = the diagonal values, including ghost ones 7752 7753 Level: developer 7754 7755 Notes: Works only for MPIAIJ and MPIBAIJ matrices 7756 7757 .seealso: MatDiagonalScale() 7758 @*/ 7759 PetscErrorCode MatDiagonalScaleLocal(Mat mat,Vec diag) 7760 { 7761 PetscErrorCode ierr; 7762 PetscMPIInt size; 7763 7764 PetscFunctionBegin; 7765 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7766 PetscValidHeaderSpecific(diag,VEC_CLASSID,2); 7767 PetscValidType(mat,1); 7768 7769 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled"); 7770 ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 7771 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);CHKERRQ(ierr); 7772 if (size == 1) { 7773 PetscInt n,m; 7774 ierr = VecGetSize(diag,&n);CHKERRQ(ierr); 7775 ierr = MatGetSize(mat,0,&m);CHKERRQ(ierr); 7776 if (m == n) { 7777 ierr = MatDiagonalScale(mat,0,diag);CHKERRQ(ierr); 7778 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only supported for sequential matrices when no ghost points/periodic conditions"); 7779 } else { 7780 ierr = PetscUseMethod(mat,"MatDiagonalScaleLocal_C",(Mat,Vec),(mat,diag));CHKERRQ(ierr); 7781 } 7782 ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr); 7783 ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr); 7784 PetscFunctionReturn(0); 7785 } 7786 7787 #undef __FUNCT__ 7788 #define __FUNCT__ "MatGetInertia" 7789 /*@ 7790 MatGetInertia - Gets the inertia from a factored matrix 7791 7792 Collective on Mat 7793 7794 Input Parameter: 7795 . mat - the matrix 7796 7797 Output Parameters: 7798 + nneg - number of negative eigenvalues 7799 . nzero - number of zero eigenvalues 7800 - npos - number of positive eigenvalues 7801 7802 Level: advanced 7803 7804 Notes: Matrix must have been factored by MatCholeskyFactor() 7805 7806 7807 @*/ 7808 PetscErrorCode MatGetInertia(Mat mat,PetscInt *nneg,PetscInt *nzero,PetscInt *npos) 7809 { 7810 PetscErrorCode ierr; 7811 7812 PetscFunctionBegin; 7813 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7814 PetscValidType(mat,1); 7815 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 7816 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Numeric factor mat is not assembled"); 7817 if (!mat->ops->getinertia) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 7818 ierr = (*mat->ops->getinertia)(mat,nneg,nzero,npos);CHKERRQ(ierr); 7819 PetscFunctionReturn(0); 7820 } 7821 7822 /* ----------------------------------------------------------------*/ 7823 #undef __FUNCT__ 7824 #define __FUNCT__ "MatSolves" 7825 /*@C 7826 MatSolves - Solves A x = b, given a factored matrix, for a collection of vectors 7827 7828 Neighbor-wise Collective on Mat and Vecs 7829 7830 Input Parameters: 7831 + mat - the factored matrix 7832 - b - the right-hand-side vectors 7833 7834 Output Parameter: 7835 . x - the result vectors 7836 7837 Notes: 7838 The vectors b and x cannot be the same. I.e., one cannot 7839 call MatSolves(A,x,x). 7840 7841 Notes: 7842 Most users should employ the simplified KSP interface for linear solvers 7843 instead of working directly with matrix algebra routines such as this. 7844 See, e.g., KSPCreate(). 7845 7846 Level: developer 7847 7848 Concepts: matrices^triangular solves 7849 7850 .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd(), MatSolve() 7851 @*/ 7852 PetscErrorCode MatSolves(Mat mat,Vecs b,Vecs x) 7853 { 7854 PetscErrorCode ierr; 7855 7856 PetscFunctionBegin; 7857 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 7858 PetscValidType(mat,1); 7859 if (x == b) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_IDN,"x and b must be different vectors"); 7860 if (!mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix"); 7861 if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(0); 7862 7863 if (!mat->ops->solves) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 7864 MatCheckPreallocated(mat,1); 7865 ierr = PetscLogEventBegin(MAT_Solves,mat,0,0,0);CHKERRQ(ierr); 7866 ierr = (*mat->ops->solves)(mat,b,x);CHKERRQ(ierr); 7867 ierr = PetscLogEventEnd(MAT_Solves,mat,0,0,0);CHKERRQ(ierr); 7868 PetscFunctionReturn(0); 7869 } 7870 7871 #undef __FUNCT__ 7872 #define __FUNCT__ "MatIsSymmetric" 7873 /*@ 7874 MatIsSymmetric - Test whether a matrix is symmetric 7875 7876 Collective on Mat 7877 7878 Input Parameter: 7879 + A - the matrix to test 7880 - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact transpose) 7881 7882 Output Parameters: 7883 . flg - the result 7884 7885 Notes: For real numbers MatIsSymmetric() and MatIsHermitian() return identical results 7886 7887 Level: intermediate 7888 7889 Concepts: matrix^symmetry 7890 7891 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetricKnown() 7892 @*/ 7893 PetscErrorCode MatIsSymmetric(Mat A,PetscReal tol,PetscBool *flg) 7894 { 7895 PetscErrorCode ierr; 7896 7897 PetscFunctionBegin; 7898 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 7899 PetscValidPointer(flg,2); 7900 7901 if (!A->symmetric_set) { 7902 if (!A->ops->issymmetric) { 7903 MatType mattype; 7904 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 7905 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for symmetric",mattype); 7906 } 7907 ierr = (*A->ops->issymmetric)(A,tol,flg);CHKERRQ(ierr); 7908 if (!tol) { 7909 A->symmetric_set = PETSC_TRUE; 7910 A->symmetric = *flg; 7911 if (A->symmetric) { 7912 A->structurally_symmetric_set = PETSC_TRUE; 7913 A->structurally_symmetric = PETSC_TRUE; 7914 } 7915 } 7916 } else if (A->symmetric) { 7917 *flg = PETSC_TRUE; 7918 } else if (!tol) { 7919 *flg = PETSC_FALSE; 7920 } else { 7921 if (!A->ops->issymmetric) { 7922 MatType mattype; 7923 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 7924 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for symmetric",mattype); 7925 } 7926 ierr = (*A->ops->issymmetric)(A,tol,flg);CHKERRQ(ierr); 7927 } 7928 PetscFunctionReturn(0); 7929 } 7930 7931 #undef __FUNCT__ 7932 #define __FUNCT__ "MatIsHermitian" 7933 /*@ 7934 MatIsHermitian - Test whether a matrix is Hermitian 7935 7936 Collective on Mat 7937 7938 Input Parameter: 7939 + A - the matrix to test 7940 - tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact Hermitian) 7941 7942 Output Parameters: 7943 . flg - the result 7944 7945 Level: intermediate 7946 7947 Concepts: matrix^symmetry 7948 7949 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), 7950 MatIsSymmetricKnown(), MatIsSymmetric() 7951 @*/ 7952 PetscErrorCode MatIsHermitian(Mat A,PetscReal tol,PetscBool *flg) 7953 { 7954 PetscErrorCode ierr; 7955 7956 PetscFunctionBegin; 7957 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 7958 PetscValidPointer(flg,2); 7959 7960 if (!A->hermitian_set) { 7961 if (!A->ops->ishermitian) { 7962 MatType mattype; 7963 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 7964 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for hermitian",mattype); 7965 } 7966 ierr = (*A->ops->ishermitian)(A,tol,flg);CHKERRQ(ierr); 7967 if (!tol) { 7968 A->hermitian_set = PETSC_TRUE; 7969 A->hermitian = *flg; 7970 if (A->hermitian) { 7971 A->structurally_symmetric_set = PETSC_TRUE; 7972 A->structurally_symmetric = PETSC_TRUE; 7973 } 7974 } 7975 } else if (A->hermitian) { 7976 *flg = PETSC_TRUE; 7977 } else if (!tol) { 7978 *flg = PETSC_FALSE; 7979 } else { 7980 if (!A->ops->ishermitian) { 7981 MatType mattype; 7982 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 7983 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for hermitian",mattype); 7984 } 7985 ierr = (*A->ops->ishermitian)(A,tol,flg);CHKERRQ(ierr); 7986 } 7987 PetscFunctionReturn(0); 7988 } 7989 7990 #undef __FUNCT__ 7991 #define __FUNCT__ "MatIsSymmetricKnown" 7992 /*@ 7993 MatIsSymmetricKnown - Checks the flag on the matrix to see if it is symmetric. 7994 7995 Not Collective 7996 7997 Input Parameter: 7998 . A - the matrix to check 7999 8000 Output Parameters: 8001 + set - if the symmetric flag is set (this tells you if the next flag is valid) 8002 - flg - the result 8003 8004 Level: advanced 8005 8006 Concepts: matrix^symmetry 8007 8008 Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsSymmetric() 8009 if you want it explicitly checked 8010 8011 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric() 8012 @*/ 8013 PetscErrorCode MatIsSymmetricKnown(Mat A,PetscBool *set,PetscBool *flg) 8014 { 8015 PetscFunctionBegin; 8016 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8017 PetscValidPointer(set,2); 8018 PetscValidPointer(flg,3); 8019 if (A->symmetric_set) { 8020 *set = PETSC_TRUE; 8021 *flg = A->symmetric; 8022 } else { 8023 *set = PETSC_FALSE; 8024 } 8025 PetscFunctionReturn(0); 8026 } 8027 8028 #undef __FUNCT__ 8029 #define __FUNCT__ "MatIsHermitianKnown" 8030 /*@ 8031 MatIsHermitianKnown - Checks the flag on the matrix to see if it is hermitian. 8032 8033 Not Collective 8034 8035 Input Parameter: 8036 . A - the matrix to check 8037 8038 Output Parameters: 8039 + set - if the hermitian flag is set (this tells you if the next flag is valid) 8040 - flg - the result 8041 8042 Level: advanced 8043 8044 Concepts: matrix^symmetry 8045 8046 Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsHermitian() 8047 if you want it explicitly checked 8048 8049 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric() 8050 @*/ 8051 PetscErrorCode MatIsHermitianKnown(Mat A,PetscBool *set,PetscBool *flg) 8052 { 8053 PetscFunctionBegin; 8054 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8055 PetscValidPointer(set,2); 8056 PetscValidPointer(flg,3); 8057 if (A->hermitian_set) { 8058 *set = PETSC_TRUE; 8059 *flg = A->hermitian; 8060 } else { 8061 *set = PETSC_FALSE; 8062 } 8063 PetscFunctionReturn(0); 8064 } 8065 8066 #undef __FUNCT__ 8067 #define __FUNCT__ "MatIsStructurallySymmetric" 8068 /*@ 8069 MatIsStructurallySymmetric - Test whether a matrix is structurally symmetric 8070 8071 Collective on Mat 8072 8073 Input Parameter: 8074 . A - the matrix to test 8075 8076 Output Parameters: 8077 . flg - the result 8078 8079 Level: intermediate 8080 8081 Concepts: matrix^symmetry 8082 8083 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsSymmetric(), MatSetOption() 8084 @*/ 8085 PetscErrorCode MatIsStructurallySymmetric(Mat A,PetscBool *flg) 8086 { 8087 PetscErrorCode ierr; 8088 8089 PetscFunctionBegin; 8090 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8091 PetscValidPointer(flg,2); 8092 if (!A->structurally_symmetric_set) { 8093 if (!A->ops->isstructurallysymmetric) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Matrix does not support checking for structural symmetric"); 8094 ierr = (*A->ops->isstructurallysymmetric)(A,&A->structurally_symmetric);CHKERRQ(ierr); 8095 8096 A->structurally_symmetric_set = PETSC_TRUE; 8097 } 8098 *flg = A->structurally_symmetric; 8099 PetscFunctionReturn(0); 8100 } 8101 8102 #undef __FUNCT__ 8103 #define __FUNCT__ "MatStashGetInfo" 8104 extern PetscErrorCode MatStashGetInfo_Private(MatStash*,PetscInt*,PetscInt*); 8105 /*@ 8106 MatStashGetInfo - Gets how many values are currently in the matrix stash, i.e. need 8107 to be communicated to other processors during the MatAssemblyBegin/End() process 8108 8109 Not collective 8110 8111 Input Parameter: 8112 . vec - the vector 8113 8114 Output Parameters: 8115 + nstash - the size of the stash 8116 . reallocs - the number of additional mallocs incurred. 8117 . bnstash - the size of the block stash 8118 - breallocs - the number of additional mallocs incurred.in the block stash 8119 8120 Level: advanced 8121 8122 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashSetInitialSize() 8123 8124 @*/ 8125 PetscErrorCode MatStashGetInfo(Mat mat,PetscInt *nstash,PetscInt *reallocs,PetscInt *bnstash,PetscInt *breallocs) 8126 { 8127 PetscErrorCode ierr; 8128 8129 PetscFunctionBegin; 8130 ierr = MatStashGetInfo_Private(&mat->stash,nstash,reallocs);CHKERRQ(ierr); 8131 ierr = MatStashGetInfo_Private(&mat->bstash,bnstash,breallocs);CHKERRQ(ierr); 8132 PetscFunctionReturn(0); 8133 } 8134 8135 #undef __FUNCT__ 8136 #define __FUNCT__ "MatGetVecs" 8137 /*@C 8138 MatGetVecs - Get vector(s) compatible with the matrix, i.e. with the same 8139 parallel layout 8140 8141 Collective on Mat 8142 8143 Input Parameter: 8144 . mat - the matrix 8145 8146 Output Parameter: 8147 + right - (optional) vector that the matrix can be multiplied against 8148 - left - (optional) vector that the matrix vector product can be stored in 8149 8150 Level: advanced 8151 8152 .seealso: MatCreate() 8153 @*/ 8154 PetscErrorCode MatGetVecs(Mat mat,Vec *right,Vec *left) 8155 { 8156 PetscErrorCode ierr; 8157 8158 PetscFunctionBegin; 8159 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 8160 PetscValidType(mat,1); 8161 MatCheckPreallocated(mat,1); 8162 if (mat->ops->getvecs) { 8163 ierr = (*mat->ops->getvecs)(mat,right,left);CHKERRQ(ierr); 8164 } else { 8165 PetscMPIInt size; 8166 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat), &size);CHKERRQ(ierr); 8167 if (right) { 8168 ierr = VecCreate(PetscObjectComm((PetscObject)mat),right);CHKERRQ(ierr); 8169 ierr = VecSetSizes(*right,mat->cmap->n,PETSC_DETERMINE);CHKERRQ(ierr); 8170 ierr = VecSetBlockSize(*right,mat->rmap->bs);CHKERRQ(ierr); 8171 ierr = VecSetType(*right,VECSTANDARD);CHKERRQ(ierr); 8172 ierr = PetscLayoutReference(mat->cmap,&(*right)->map);CHKERRQ(ierr); 8173 } 8174 if (left) { 8175 ierr = VecCreate(PetscObjectComm((PetscObject)mat),left);CHKERRQ(ierr); 8176 ierr = VecSetSizes(*left,mat->rmap->n,PETSC_DETERMINE);CHKERRQ(ierr); 8177 ierr = VecSetBlockSize(*left,mat->rmap->bs);CHKERRQ(ierr); 8178 ierr = VecSetType(*left,VECSTANDARD);CHKERRQ(ierr); 8179 ierr = PetscLayoutReference(mat->rmap,&(*left)->map);CHKERRQ(ierr); 8180 } 8181 } 8182 PetscFunctionReturn(0); 8183 } 8184 8185 #undef __FUNCT__ 8186 #define __FUNCT__ "MatFactorInfoInitialize" 8187 /*@C 8188 MatFactorInfoInitialize - Initializes a MatFactorInfo data structure 8189 with default values. 8190 8191 Not Collective 8192 8193 Input Parameters: 8194 . info - the MatFactorInfo data structure 8195 8196 8197 Notes: The solvers are generally used through the KSP and PC objects, for example 8198 PCLU, PCILU, PCCHOLESKY, PCICC 8199 8200 Level: developer 8201 8202 .seealso: MatFactorInfo 8203 8204 Developer Note: fortran interface is not autogenerated as the f90 8205 interface defintion cannot be generated correctly [due to MatFactorInfo] 8206 8207 @*/ 8208 8209 PetscErrorCode MatFactorInfoInitialize(MatFactorInfo *info) 8210 { 8211 PetscErrorCode ierr; 8212 8213 PetscFunctionBegin; 8214 ierr = PetscMemzero(info,sizeof(MatFactorInfo));CHKERRQ(ierr); 8215 PetscFunctionReturn(0); 8216 } 8217 8218 #undef __FUNCT__ 8219 #define __FUNCT__ "MatPtAP" 8220 /*@ 8221 MatPtAP - Creates the matrix product C = P^T * A * P 8222 8223 Neighbor-wise Collective on Mat 8224 8225 Input Parameters: 8226 + A - the matrix 8227 . P - the projection matrix 8228 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8229 - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(P)) 8230 8231 Output Parameters: 8232 . C - the product matrix 8233 8234 Notes: 8235 C will be created and must be destroyed by the user with MatDestroy(). 8236 8237 This routine is currently only implemented for pairs of AIJ matrices and classes 8238 which inherit from AIJ. 8239 8240 Level: intermediate 8241 8242 .seealso: MatPtAPSymbolic(), MatPtAPNumeric(), MatMatMult(), MatRARt() 8243 @*/ 8244 PetscErrorCode MatPtAP(Mat A,Mat P,MatReuse scall,PetscReal fill,Mat *C) 8245 { 8246 PetscErrorCode ierr; 8247 PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*); 8248 PetscErrorCode (*fP)(Mat,Mat,MatReuse,PetscReal,Mat*); 8249 PetscErrorCode (*ptap)(Mat,Mat,MatReuse,PetscReal,Mat*)=NULL; 8250 PetscBool viatranspose=PETSC_FALSE,viamatmatmatmult=PETSC_FALSE; 8251 8252 PetscFunctionBegin; 8253 ierr = PetscOptionsGetBool(((PetscObject)A)->prefix,"-matptap_viatranspose",&viatranspose,NULL);CHKERRQ(ierr); 8254 ierr = PetscOptionsGetBool(((PetscObject)A)->prefix,"-matptap_viamatmatmatmult",&viamatmatmatmult,NULL);CHKERRQ(ierr); 8255 8256 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8257 PetscValidType(A,1); 8258 MatCheckPreallocated(A,1); 8259 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8260 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8261 PetscValidHeaderSpecific(P,MAT_CLASSID,2); 8262 PetscValidType(P,2); 8263 MatCheckPreallocated(P,2); 8264 if (!P->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8265 if (P->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8266 8267 if (P->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N); 8268 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8269 8270 if (scall == MAT_REUSE_MATRIX) { 8271 PetscValidPointer(*C,5); 8272 PetscValidHeaderSpecific(*C,MAT_CLASSID,5); 8273 if (viatranspose || viamatmatmatmult) { 8274 Mat Pt; 8275 ierr = MatTranspose(P,MAT_INITIAL_MATRIX,&Pt);CHKERRQ(ierr); 8276 if (viamatmatmatmult) { 8277 ierr = MatMatMatMult(Pt,A,P,scall,fill,C);CHKERRQ(ierr); 8278 } else { 8279 Mat AP; 8280 ierr = MatMatMult(A,P,MAT_INITIAL_MATRIX,fill,&AP);CHKERRQ(ierr); 8281 ierr = MatMatMult(Pt,AP,scall,fill,C);CHKERRQ(ierr); 8282 ierr = MatDestroy(&AP);CHKERRQ(ierr); 8283 } 8284 ierr = MatDestroy(&Pt);CHKERRQ(ierr); 8285 } else { 8286 ierr = PetscLogEventBegin(MAT_PtAP,A,P,0,0);CHKERRQ(ierr); 8287 ierr = PetscLogEventBegin(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr); 8288 ierr = (*(*C)->ops->ptapnumeric)(A,P,*C);CHKERRQ(ierr); 8289 ierr = PetscLogEventEnd(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr); 8290 ierr = PetscLogEventEnd(MAT_PtAP,A,P,0,0);CHKERRQ(ierr); 8291 } 8292 PetscFunctionReturn(0); 8293 } 8294 8295 if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0; 8296 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8297 8298 fA = A->ops->ptap; 8299 fP = P->ops->ptap; 8300 if (fP == fA) { 8301 if (!fA) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatPtAP not supported for A of type %s",((PetscObject)A)->type_name); 8302 ptap = fA; 8303 } else { 8304 /* dispatch based on the type of A and P from their PetscObject's PetscFunctionLists. */ 8305 char ptapname[256]; 8306 ierr = PetscStrcpy(ptapname,"MatPtAP_");CHKERRQ(ierr); 8307 ierr = PetscStrcat(ptapname,((PetscObject)A)->type_name);CHKERRQ(ierr); 8308 ierr = PetscStrcat(ptapname,"_");CHKERRQ(ierr); 8309 ierr = PetscStrcat(ptapname,((PetscObject)P)->type_name);CHKERRQ(ierr); 8310 ierr = PetscStrcat(ptapname,"_C");CHKERRQ(ierr); /* e.g., ptapname = "MatPtAP_seqdense_seqaij_C" */ 8311 ierr = PetscObjectQueryFunction((PetscObject)P,ptapname,&ptap);CHKERRQ(ierr); 8312 if (!ptap) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatPtAP requires A, %s, to be compatible with P, %s",((PetscObject)A)->type_name,((PetscObject)P)->type_name); 8313 } 8314 8315 if (viatranspose || viamatmatmatmult) { 8316 Mat Pt; 8317 ierr = MatTranspose(P,MAT_INITIAL_MATRIX,&Pt);CHKERRQ(ierr); 8318 if (viamatmatmatmult) { 8319 ierr = MatMatMatMult(Pt,A,P,scall,fill,C);CHKERRQ(ierr); 8320 ierr = PetscInfo(*C,"MatPtAP via MatMatMatMult\n");CHKERRQ(ierr); 8321 } else { 8322 Mat AP; 8323 ierr = MatMatMult(A,P,MAT_INITIAL_MATRIX,fill,&AP);CHKERRQ(ierr); 8324 ierr = MatMatMult(Pt,AP,scall,fill,C);CHKERRQ(ierr); 8325 ierr = MatDestroy(&AP);CHKERRQ(ierr); 8326 ierr = PetscInfo(*C,"MatPtAP via MatTranspose and MatMatMult\n");CHKERRQ(ierr); 8327 } 8328 ierr = MatDestroy(&Pt);CHKERRQ(ierr); 8329 } else { 8330 ierr = PetscLogEventBegin(MAT_PtAP,A,P,0,0);CHKERRQ(ierr); 8331 ierr = (*ptap)(A,P,scall,fill,C);CHKERRQ(ierr); 8332 ierr = PetscLogEventEnd(MAT_PtAP,A,P,0,0);CHKERRQ(ierr); 8333 } 8334 PetscFunctionReturn(0); 8335 } 8336 8337 #undef __FUNCT__ 8338 #define __FUNCT__ "MatPtAPNumeric" 8339 /*@ 8340 MatPtAPNumeric - Computes the matrix product C = P^T * A * P 8341 8342 Neighbor-wise Collective on Mat 8343 8344 Input Parameters: 8345 + A - the matrix 8346 - P - the projection matrix 8347 8348 Output Parameters: 8349 . C - the product matrix 8350 8351 Notes: 8352 C must have been created by calling MatPtAPSymbolic and must be destroyed by 8353 the user using MatDeatroy(). 8354 8355 This routine is currently only implemented for pairs of AIJ matrices and classes 8356 which inherit from AIJ. C will be of type MATAIJ. 8357 8358 Level: intermediate 8359 8360 .seealso: MatPtAP(), MatPtAPSymbolic(), MatMatMultNumeric() 8361 @*/ 8362 PetscErrorCode MatPtAPNumeric(Mat A,Mat P,Mat C) 8363 { 8364 PetscErrorCode ierr; 8365 8366 PetscFunctionBegin; 8367 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8368 PetscValidType(A,1); 8369 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8370 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8371 PetscValidHeaderSpecific(P,MAT_CLASSID,2); 8372 PetscValidType(P,2); 8373 MatCheckPreallocated(P,2); 8374 if (!P->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8375 if (P->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8376 PetscValidHeaderSpecific(C,MAT_CLASSID,3); 8377 PetscValidType(C,3); 8378 MatCheckPreallocated(C,3); 8379 if (C->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8380 if (P->cmap->N!=C->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->cmap->N,C->rmap->N); 8381 if (P->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N); 8382 if (A->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N); 8383 if (P->cmap->N!=C->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->cmap->N,C->cmap->N); 8384 MatCheckPreallocated(A,1); 8385 8386 ierr = PetscLogEventBegin(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr); 8387 ierr = (*C->ops->ptapnumeric)(A,P,C);CHKERRQ(ierr); 8388 ierr = PetscLogEventEnd(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr); 8389 PetscFunctionReturn(0); 8390 } 8391 8392 #undef __FUNCT__ 8393 #define __FUNCT__ "MatPtAPSymbolic" 8394 /*@ 8395 MatPtAPSymbolic - Creates the (i,j) structure of the matrix product C = P^T * A * P 8396 8397 Neighbor-wise Collective on Mat 8398 8399 Input Parameters: 8400 + A - the matrix 8401 - P - the projection matrix 8402 8403 Output Parameters: 8404 . C - the (i,j) structure of the product matrix 8405 8406 Notes: 8407 C will be created and must be destroyed by the user with MatDestroy(). 8408 8409 This routine is currently only implemented for pairs of SeqAIJ matrices and classes 8410 which inherit from SeqAIJ. C will be of type MATSEQAIJ. The product is computed using 8411 this (i,j) structure by calling MatPtAPNumeric(). 8412 8413 Level: intermediate 8414 8415 .seealso: MatPtAP(), MatPtAPNumeric(), MatMatMultSymbolic() 8416 @*/ 8417 PetscErrorCode MatPtAPSymbolic(Mat A,Mat P,PetscReal fill,Mat *C) 8418 { 8419 PetscErrorCode ierr; 8420 8421 PetscFunctionBegin; 8422 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8423 PetscValidType(A,1); 8424 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8425 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8426 if (fill <1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8427 PetscValidHeaderSpecific(P,MAT_CLASSID,2); 8428 PetscValidType(P,2); 8429 MatCheckPreallocated(P,2); 8430 if (!P->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8431 if (P->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8432 PetscValidPointer(C,3); 8433 8434 if (P->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N); 8435 if (A->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N); 8436 MatCheckPreallocated(A,1); 8437 ierr = PetscLogEventBegin(MAT_PtAPSymbolic,A,P,0,0);CHKERRQ(ierr); 8438 ierr = (*A->ops->ptapsymbolic)(A,P,fill,C);CHKERRQ(ierr); 8439 ierr = PetscLogEventEnd(MAT_PtAPSymbolic,A,P,0,0);CHKERRQ(ierr); 8440 8441 /* ierr = MatSetBlockSize(*C,A->rmap->bs);CHKERRQ(ierr); NO! this is not always true -ma */ 8442 PetscFunctionReturn(0); 8443 } 8444 8445 #undef __FUNCT__ 8446 #define __FUNCT__ "MatRARt" 8447 /*@ 8448 MatRARt - Creates the matrix product C = R * A * R^T 8449 8450 Neighbor-wise Collective on Mat 8451 8452 Input Parameters: 8453 + A - the matrix 8454 . R - the projection matrix 8455 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8456 - fill - expected fill as ratio of nnz(C)/nnz(A) 8457 8458 Output Parameters: 8459 . C - the product matrix 8460 8461 Notes: 8462 C will be created and must be destroyed by the user with MatDestroy(). 8463 8464 This routine is currently only implemented for pairs of AIJ matrices and classes 8465 which inherit from AIJ. 8466 8467 Level: intermediate 8468 8469 .seealso: MatRARtSymbolic(), MatRARtNumeric(), MatMatMult(), MatPtAP() 8470 @*/ 8471 PetscErrorCode MatRARt(Mat A,Mat R,MatReuse scall,PetscReal fill,Mat *C) 8472 { 8473 PetscErrorCode ierr; 8474 8475 PetscFunctionBegin; 8476 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8477 PetscValidType(A,1); 8478 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8479 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8480 PetscValidHeaderSpecific(R,MAT_CLASSID,2); 8481 PetscValidType(R,2); 8482 MatCheckPreallocated(R,2); 8483 if (!R->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8484 if (R->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8485 PetscValidPointer(C,3); 8486 if (R->cmap->N!=A->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)R),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",R->cmap->N,A->rmap->N); 8487 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8488 MatCheckPreallocated(A,1); 8489 8490 if (!A->ops->rart) { 8491 MatType mattype; 8492 ierr = MatGetType(A,&mattype);CHKERRQ(ierr); 8493 SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Matrix of type <%s> does not support RARt",mattype); 8494 } 8495 ierr = PetscLogEventBegin(MAT_RARt,A,R,0,0);CHKERRQ(ierr); 8496 ierr = (*A->ops->rart)(A,R,scall,fill,C);CHKERRQ(ierr); 8497 ierr = PetscLogEventEnd(MAT_RARt,A,R,0,0);CHKERRQ(ierr); 8498 PetscFunctionReturn(0); 8499 } 8500 8501 #undef __FUNCT__ 8502 #define __FUNCT__ "MatRARtNumeric" 8503 /*@ 8504 MatRARtNumeric - Computes the matrix product C = R * A * R^T 8505 8506 Neighbor-wise Collective on Mat 8507 8508 Input Parameters: 8509 + A - the matrix 8510 - R - the projection matrix 8511 8512 Output Parameters: 8513 . C - the product matrix 8514 8515 Notes: 8516 C must have been created by calling MatRARtSymbolic and must be destroyed by 8517 the user using MatDeatroy(). 8518 8519 This routine is currently only implemented for pairs of AIJ matrices and classes 8520 which inherit from AIJ. C will be of type MATAIJ. 8521 8522 Level: intermediate 8523 8524 .seealso: MatRARt(), MatRARtSymbolic(), MatMatMultNumeric() 8525 @*/ 8526 PetscErrorCode MatRARtNumeric(Mat A,Mat R,Mat C) 8527 { 8528 PetscErrorCode ierr; 8529 8530 PetscFunctionBegin; 8531 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8532 PetscValidType(A,1); 8533 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8534 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8535 PetscValidHeaderSpecific(R,MAT_CLASSID,2); 8536 PetscValidType(R,2); 8537 MatCheckPreallocated(R,2); 8538 if (!R->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8539 if (R->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8540 PetscValidHeaderSpecific(C,MAT_CLASSID,3); 8541 PetscValidType(C,3); 8542 MatCheckPreallocated(C,3); 8543 if (C->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8544 if (R->rmap->N!=C->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",R->rmap->N,C->rmap->N); 8545 if (R->cmap->N!=A->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",R->cmap->N,A->rmap->N); 8546 if (A->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N); 8547 if (R->rmap->N!=C->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",R->rmap->N,C->cmap->N); 8548 MatCheckPreallocated(A,1); 8549 8550 ierr = PetscLogEventBegin(MAT_RARtNumeric,A,R,0,0);CHKERRQ(ierr); 8551 ierr = (*A->ops->rartnumeric)(A,R,C);CHKERRQ(ierr); 8552 ierr = PetscLogEventEnd(MAT_RARtNumeric,A,R,0,0);CHKERRQ(ierr); 8553 PetscFunctionReturn(0); 8554 } 8555 8556 #undef __FUNCT__ 8557 #define __FUNCT__ "MatRARtSymbolic" 8558 /*@ 8559 MatRARtSymbolic - Creates the (i,j) structure of the matrix product C = R * A * R^T 8560 8561 Neighbor-wise Collective on Mat 8562 8563 Input Parameters: 8564 + A - the matrix 8565 - R - the projection matrix 8566 8567 Output Parameters: 8568 . C - the (i,j) structure of the product matrix 8569 8570 Notes: 8571 C will be created and must be destroyed by the user with MatDestroy(). 8572 8573 This routine is currently only implemented for pairs of SeqAIJ matrices and classes 8574 which inherit from SeqAIJ. C will be of type MATSEQAIJ. The product is computed using 8575 this (i,j) structure by calling MatRARtNumeric(). 8576 8577 Level: intermediate 8578 8579 .seealso: MatRARt(), MatRARtNumeric(), MatMatMultSymbolic() 8580 @*/ 8581 PetscErrorCode MatRARtSymbolic(Mat A,Mat R,PetscReal fill,Mat *C) 8582 { 8583 PetscErrorCode ierr; 8584 8585 PetscFunctionBegin; 8586 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8587 PetscValidType(A,1); 8588 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8589 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8590 if (fill <1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8591 PetscValidHeaderSpecific(R,MAT_CLASSID,2); 8592 PetscValidType(R,2); 8593 MatCheckPreallocated(R,2); 8594 if (!R->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8595 if (R->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8596 PetscValidPointer(C,3); 8597 8598 if (R->cmap->N!=A->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",R->cmap->N,A->rmap->N); 8599 if (A->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N); 8600 MatCheckPreallocated(A,1); 8601 ierr = PetscLogEventBegin(MAT_RARtSymbolic,A,R,0,0);CHKERRQ(ierr); 8602 ierr = (*A->ops->rartsymbolic)(A,R,fill,C);CHKERRQ(ierr); 8603 ierr = PetscLogEventEnd(MAT_RARtSymbolic,A,R,0,0);CHKERRQ(ierr); 8604 8605 ierr = MatSetBlockSize(*C,A->rmap->bs);CHKERRQ(ierr); 8606 PetscFunctionReturn(0); 8607 } 8608 8609 #undef __FUNCT__ 8610 #define __FUNCT__ "MatMatMult" 8611 /*@ 8612 MatMatMult - Performs Matrix-Matrix Multiplication C=A*B. 8613 8614 Neighbor-wise Collective on Mat 8615 8616 Input Parameters: 8617 + A - the left matrix 8618 . B - the right matrix 8619 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8620 - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if you do not have a good estimate 8621 if the result is a dense matrix this is irrelevent 8622 8623 Output Parameters: 8624 . C - the product matrix 8625 8626 Notes: 8627 Unless scall is MAT_REUSE_MATRIX C will be created. 8628 8629 MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call 8630 8631 To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value 8632 actually needed. 8633 8634 If you have many matrices with the same non-zero structure to multiply, you 8635 should either 8636 $ 1) use MAT_REUSE_MATRIX in all calls but the first or 8637 $ 2) call MatMatMultSymbolic() once and then MatMatMultNumeric() for each product needed 8638 8639 Level: intermediate 8640 8641 .seealso: MatMatMultSymbolic(), MatMatMultNumeric(), MatTransposeMatMult(), MatMatTransposeMult(), MatPtAP() 8642 @*/ 8643 PetscErrorCode MatMatMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C) 8644 { 8645 PetscErrorCode ierr; 8646 PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*); 8647 PetscErrorCode (*fB)(Mat,Mat,MatReuse,PetscReal,Mat*); 8648 PetscErrorCode (*mult)(Mat,Mat,MatReuse,PetscReal,Mat*)=NULL; 8649 8650 PetscFunctionBegin; 8651 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8652 PetscValidType(A,1); 8653 MatCheckPreallocated(A,1); 8654 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8655 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8656 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 8657 PetscValidType(B,2); 8658 MatCheckPreallocated(B,2); 8659 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8660 if (B->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8661 PetscValidPointer(C,3); 8662 if (B->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N); 8663 if (scall == MAT_REUSE_MATRIX) { 8664 PetscValidPointer(*C,5); 8665 PetscValidHeaderSpecific(*C,MAT_CLASSID,5); 8666 ierr = PetscLogEventBegin(MAT_MatMult,A,B,0,0);CHKERRQ(ierr); 8667 ierr = PetscLogEventBegin(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr); 8668 ierr = (*(*C)->ops->matmultnumeric)(A,B,*C);CHKERRQ(ierr); 8669 ierr = PetscLogEventEnd(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr); 8670 ierr = PetscLogEventEnd(MAT_MatMult,A,B,0,0);CHKERRQ(ierr); 8671 PetscFunctionReturn(0); 8672 } 8673 if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0; 8674 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 8675 8676 fA = A->ops->matmult; 8677 fB = B->ops->matmult; 8678 if (fB == fA) { 8679 if (!fB) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatMatMult not supported for B of type %s",((PetscObject)B)->type_name); 8680 mult = fB; 8681 } else { 8682 /* dispatch based on the type of A and B from their PetscObject's PetscFunctionLists. */ 8683 char multname[256]; 8684 ierr = PetscStrcpy(multname,"MatMatMult_");CHKERRQ(ierr); 8685 ierr = PetscStrcat(multname,((PetscObject)A)->type_name);CHKERRQ(ierr); 8686 ierr = PetscStrcat(multname,"_");CHKERRQ(ierr); 8687 ierr = PetscStrcat(multname,((PetscObject)B)->type_name);CHKERRQ(ierr); 8688 ierr = PetscStrcat(multname,"_C");CHKERRQ(ierr); /* e.g., multname = "MatMatMult_seqdense_seqaij_C" */ 8689 ierr = PetscObjectQueryFunction((PetscObject)B,multname,&mult);CHKERRQ(ierr); 8690 if (!mult) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatMatMult requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name); 8691 } 8692 ierr = PetscLogEventBegin(MAT_MatMult,A,B,0,0);CHKERRQ(ierr); 8693 ierr = (*mult)(A,B,scall,fill,C);CHKERRQ(ierr); 8694 ierr = PetscLogEventEnd(MAT_MatMult,A,B,0,0);CHKERRQ(ierr); 8695 PetscFunctionReturn(0); 8696 } 8697 8698 #undef __FUNCT__ 8699 #define __FUNCT__ "MatMatMultSymbolic" 8700 /*@ 8701 MatMatMultSymbolic - Performs construction, preallocation, and computes the ij structure 8702 of the matrix-matrix product C=A*B. Call this routine before calling MatMatMultNumeric(). 8703 8704 Neighbor-wise Collective on Mat 8705 8706 Input Parameters: 8707 + A - the left matrix 8708 . B - the right matrix 8709 - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if you do not have a good estimate, 8710 if C is a dense matrix this is irrelevent 8711 8712 Output Parameters: 8713 . C - the product matrix 8714 8715 Notes: 8716 Unless scall is MAT_REUSE_MATRIX C will be created. 8717 8718 To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value 8719 actually needed. 8720 8721 This routine is currently implemented for 8722 - pairs of AIJ matrices and classes which inherit from AIJ, C will be of type AIJ 8723 - pairs of AIJ (A) and Dense (B) matrix, C will be of type Dense. 8724 - pairs of Dense (A) and AIJ (B) matrix, C will be of type Dense. 8725 8726 Level: intermediate 8727 8728 Developers Note: There are ways to estimate the number of nonzeros in the resulting product, see for example, http://arxiv.org/abs/1006.4173 8729 We should incorporate them into PETSc. 8730 8731 .seealso: MatMatMult(), MatMatMultNumeric() 8732 @*/ 8733 PetscErrorCode MatMatMultSymbolic(Mat A,Mat B,PetscReal fill,Mat *C) 8734 { 8735 PetscErrorCode ierr; 8736 PetscErrorCode (*Asymbolic)(Mat,Mat,PetscReal,Mat*); 8737 PetscErrorCode (*Bsymbolic)(Mat,Mat,PetscReal,Mat*); 8738 PetscErrorCode (*symbolic)(Mat,Mat,PetscReal,Mat*)=NULL; 8739 8740 PetscFunctionBegin; 8741 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8742 PetscValidType(A,1); 8743 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8744 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8745 8746 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 8747 PetscValidType(B,2); 8748 MatCheckPreallocated(B,2); 8749 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8750 if (B->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8751 PetscValidPointer(C,3); 8752 8753 if (B->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N); 8754 if (fill == PETSC_DEFAULT) fill = 2.0; 8755 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be > 1.0",fill); 8756 MatCheckPreallocated(A,1); 8757 8758 Asymbolic = A->ops->matmultsymbolic; 8759 Bsymbolic = B->ops->matmultsymbolic; 8760 if (Asymbolic == Bsymbolic) { 8761 if (!Bsymbolic) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"C=A*B not implemented for B of type %s",((PetscObject)B)->type_name); 8762 symbolic = Bsymbolic; 8763 } else { /* dispatch based on the type of A and B */ 8764 char symbolicname[256]; 8765 ierr = PetscStrcpy(symbolicname,"MatMatMultSymbolic_");CHKERRQ(ierr); 8766 ierr = PetscStrcat(symbolicname,((PetscObject)A)->type_name);CHKERRQ(ierr); 8767 ierr = PetscStrcat(symbolicname,"_");CHKERRQ(ierr); 8768 ierr = PetscStrcat(symbolicname,((PetscObject)B)->type_name);CHKERRQ(ierr); 8769 ierr = PetscStrcat(symbolicname,"_C");CHKERRQ(ierr); 8770 ierr = PetscObjectQueryFunction((PetscObject)B,symbolicname,&symbolic);CHKERRQ(ierr); 8771 if (!symbolic) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatMatMultSymbolic requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name); 8772 } 8773 ierr = PetscLogEventBegin(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr); 8774 ierr = (*symbolic)(A,B,fill,C);CHKERRQ(ierr); 8775 ierr = PetscLogEventEnd(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr); 8776 PetscFunctionReturn(0); 8777 } 8778 8779 #undef __FUNCT__ 8780 #define __FUNCT__ "MatMatMultNumeric" 8781 /*@ 8782 MatMatMultNumeric - Performs the numeric matrix-matrix product. 8783 Call this routine after first calling MatMatMultSymbolic(). 8784 8785 Neighbor-wise Collective on Mat 8786 8787 Input Parameters: 8788 + A - the left matrix 8789 - B - the right matrix 8790 8791 Output Parameters: 8792 . C - the product matrix, which was created by from MatMatMultSymbolic() or a call to MatMatMult(). 8793 8794 Notes: 8795 C must have been created with MatMatMultSymbolic(). 8796 8797 This routine is currently implemented for 8798 - pairs of AIJ matrices and classes which inherit from AIJ, C will be of type MATAIJ. 8799 - pairs of AIJ (A) and Dense (B) matrix, C will be of type Dense. 8800 - pairs of Dense (A) and AIJ (B) matrix, C will be of type Dense. 8801 8802 Level: intermediate 8803 8804 .seealso: MatMatMult(), MatMatMultSymbolic() 8805 @*/ 8806 PetscErrorCode MatMatMultNumeric(Mat A,Mat B,Mat C) 8807 { 8808 PetscErrorCode ierr; 8809 8810 PetscFunctionBegin; 8811 ierr = MatMatMult(A,B,MAT_REUSE_MATRIX,0.0,&C);CHKERRQ(ierr); 8812 PetscFunctionReturn(0); 8813 } 8814 8815 #undef __FUNCT__ 8816 #define __FUNCT__ "MatMatTransposeMult" 8817 /*@ 8818 MatMatTransposeMult - Performs Matrix-Matrix Multiplication C=A*B^T. 8819 8820 Neighbor-wise Collective on Mat 8821 8822 Input Parameters: 8823 + A - the left matrix 8824 . B - the right matrix 8825 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8826 - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if not known 8827 8828 Output Parameters: 8829 . C - the product matrix 8830 8831 Notes: 8832 C will be created if MAT_INITIAL_MATRIX and must be destroyed by the user with MatDestroy(). 8833 8834 MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call 8835 8836 To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value 8837 actually needed. 8838 8839 This routine is currently only implemented for pairs of SeqAIJ matrices. C will be of type MATSEQAIJ. 8840 8841 Level: intermediate 8842 8843 .seealso: MatMatTransposeMultSymbolic(), MatMatTransposeMultNumeric(), MatMatMult(), MatTransposeMatMult() MatPtAP() 8844 @*/ 8845 PetscErrorCode MatMatTransposeMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C) 8846 { 8847 PetscErrorCode ierr; 8848 PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*); 8849 PetscErrorCode (*fB)(Mat,Mat,MatReuse,PetscReal,Mat*); 8850 8851 PetscFunctionBegin; 8852 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8853 PetscValidType(A,1); 8854 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8855 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8856 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 8857 PetscValidType(B,2); 8858 MatCheckPreallocated(B,2); 8859 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8860 if (B->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8861 PetscValidPointer(C,3); 8862 if (B->cmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, AN %D != BN %D",A->cmap->N,B->cmap->N); 8863 if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0; 8864 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be > 1.0",fill); 8865 MatCheckPreallocated(A,1); 8866 8867 fA = A->ops->mattransposemult; 8868 if (!fA) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatMatTransposeMult not supported for A of type %s",((PetscObject)A)->type_name); 8869 fB = B->ops->mattransposemult; 8870 if (!fB) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatMatTransposeMult not supported for B of type %s",((PetscObject)B)->type_name); 8871 if (fB!=fA) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatMatTransposeMult requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name); 8872 8873 ierr = PetscLogEventBegin(MAT_MatTransposeMult,A,B,0,0);CHKERRQ(ierr); 8874 if (scall == MAT_INITIAL_MATRIX) { 8875 ierr = PetscLogEventBegin(MAT_MatTransposeMultSymbolic,A,B,0,0);CHKERRQ(ierr); 8876 ierr = (*A->ops->mattransposemultsymbolic)(A,B,fill,C);CHKERRQ(ierr); 8877 ierr = PetscLogEventEnd(MAT_MatTransposeMultSymbolic,A,B,0,0);CHKERRQ(ierr); 8878 } 8879 ierr = PetscLogEventBegin(MAT_MatTransposeMultNumeric,A,B,0,0);CHKERRQ(ierr); 8880 ierr = (*A->ops->mattransposemultnumeric)(A,B,*C);CHKERRQ(ierr); 8881 ierr = PetscLogEventEnd(MAT_MatTransposeMultNumeric,A,B,0,0);CHKERRQ(ierr); 8882 ierr = PetscLogEventEnd(MAT_MatTransposeMult,A,B,0,0);CHKERRQ(ierr); 8883 PetscFunctionReturn(0); 8884 } 8885 8886 #undef __FUNCT__ 8887 #define __FUNCT__ "MatTransposeMatMult" 8888 /*@ 8889 MatTransposeMatMult - Performs Matrix-Matrix Multiplication C=A^T*B. 8890 8891 Neighbor-wise Collective on Mat 8892 8893 Input Parameters: 8894 + A - the left matrix 8895 . B - the right matrix 8896 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8897 - fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if not known 8898 8899 Output Parameters: 8900 . C - the product matrix 8901 8902 Notes: 8903 C will be created if MAT_INITIAL_MATRIX and must be destroyed by the user with MatDestroy(). 8904 8905 MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call 8906 8907 To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value 8908 actually needed. 8909 8910 This routine is currently implemented for pairs of AIJ matrices and pairs of SeqDense matrices and classes 8911 which inherit from SeqAIJ. C will be of same type as the input matrices. 8912 8913 Level: intermediate 8914 8915 .seealso: MatTransposeMatMultSymbolic(), MatTransposeMatMultNumeric(), MatMatMult(), MatMatTransposeMult(), MatPtAP() 8916 @*/ 8917 PetscErrorCode MatTransposeMatMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C) 8918 { 8919 PetscErrorCode ierr; 8920 PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*); 8921 PetscErrorCode (*fB)(Mat,Mat,MatReuse,PetscReal,Mat*); 8922 PetscErrorCode (*transposematmult)(Mat,Mat,MatReuse,PetscReal,Mat*) = NULL; 8923 8924 PetscFunctionBegin; 8925 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 8926 PetscValidType(A,1); 8927 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8928 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8929 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 8930 PetscValidType(B,2); 8931 MatCheckPreallocated(B,2); 8932 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 8933 if (B->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 8934 PetscValidPointer(C,3); 8935 if (B->rmap->N!=A->rmap->N) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->rmap->N); 8936 if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0; 8937 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be > 1.0",fill); 8938 MatCheckPreallocated(A,1); 8939 8940 fA = A->ops->transposematmult; 8941 fB = B->ops->transposematmult; 8942 if (fB==fA) { 8943 if (!fA) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatTransposeMatMult not supported for A of type %s",((PetscObject)A)->type_name); 8944 transposematmult = fA; 8945 } else { 8946 /* dispatch based on the type of A and B from their PetscObject's PetscFunctionLists. */ 8947 char multname[256]; 8948 ierr = PetscStrcpy(multname,"MatTransposeMatMult_");CHKERRQ(ierr); 8949 ierr = PetscStrcat(multname,((PetscObject)A)->type_name);CHKERRQ(ierr); 8950 ierr = PetscStrcat(multname,"_");CHKERRQ(ierr); 8951 ierr = PetscStrcat(multname,((PetscObject)B)->type_name);CHKERRQ(ierr); 8952 ierr = PetscStrcat(multname,"_C");CHKERRQ(ierr); /* e.g., multname = "MatMatMult_seqdense_seqaij_C" */ 8953 ierr = PetscObjectQueryFunction((PetscObject)B,multname,&transposematmult);CHKERRQ(ierr); 8954 if (!transposematmult) SETERRQ2(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatTransposeMatMult requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name); 8955 } 8956 ierr = PetscLogEventBegin(MAT_TransposeMatMult,A,B,0,0);CHKERRQ(ierr); 8957 ierr = (*transposematmult)(A,B,scall,fill,C);CHKERRQ(ierr); 8958 ierr = PetscLogEventEnd(MAT_TransposeMatMult,A,B,0,0);CHKERRQ(ierr); 8959 PetscFunctionReturn(0); 8960 } 8961 8962 #undef __FUNCT__ 8963 #define __FUNCT__ "MatMatMatMult" 8964 /*@ 8965 MatMatMatMult - Performs Matrix-Matrix-Matrix Multiplication D=A*B*C. 8966 8967 Neighbor-wise Collective on Mat 8968 8969 Input Parameters: 8970 + A - the left matrix 8971 . B - the middle matrix 8972 . C - the right matrix 8973 . scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 8974 - fill - expected fill as ratio of nnz(D)/(nnz(A) + nnz(B)+nnz(C)), use PETSC_DEFAULT if you do not have a good estimate 8975 if the result is a dense matrix this is irrelevent 8976 8977 Output Parameters: 8978 . D - the product matrix 8979 8980 Notes: 8981 Unless scall is MAT_REUSE_MATRIX D will be created. 8982 8983 MAT_REUSE_MATRIX can only be used if the matrices A, B and C have the same nonzero pattern as in the previous call 8984 8985 To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value 8986 actually needed. 8987 8988 If you have many matrices with the same non-zero structure to multiply, you 8989 should either 8990 $ 1) use MAT_REUSE_MATRIX in all calls but the first or 8991 $ 2) call MatMatMatMultSymbolic() once and then MatMatMatMultNumeric() for each product needed 8992 8993 Level: intermediate 8994 8995 .seealso: MatMatMult, MatPtAP() 8996 @*/ 8997 PetscErrorCode MatMatMatMult(Mat A,Mat B,Mat C,MatReuse scall,PetscReal fill,Mat *D) 8998 { 8999 PetscErrorCode ierr; 9000 PetscErrorCode (*fA)(Mat,Mat,Mat,MatReuse,PetscReal,Mat*); 9001 PetscErrorCode (*fB)(Mat,Mat,Mat,MatReuse,PetscReal,Mat*); 9002 PetscErrorCode (*fC)(Mat,Mat,Mat,MatReuse,PetscReal,Mat*); 9003 PetscErrorCode (*mult)(Mat,Mat,Mat,MatReuse,PetscReal,Mat*)=NULL; 9004 9005 PetscFunctionBegin; 9006 PetscValidHeaderSpecific(A,MAT_CLASSID,1); 9007 PetscValidType(A,1); 9008 MatCheckPreallocated(A,1); 9009 if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9010 if (A->factortype) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9011 PetscValidHeaderSpecific(B,MAT_CLASSID,2); 9012 PetscValidType(B,2); 9013 MatCheckPreallocated(B,2); 9014 if (!B->assembled) SETERRQ(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9015 if (B->factortype) SETERRQ(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9016 PetscValidHeaderSpecific(C,MAT_CLASSID,3); 9017 PetscValidPointer(C,3); 9018 MatCheckPreallocated(C,3); 9019 if (!C->assembled) SETERRQ(PetscObjectComm((PetscObject)C),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9020 if (C->factortype) SETERRQ(PetscObjectComm((PetscObject)C),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9021 if (B->rmap->N!=A->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N); 9022 if (C->rmap->N!=B->cmap->N) SETERRQ2(PetscObjectComm((PetscObject)C),PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",C->rmap->N,B->cmap->N); 9023 if (scall == MAT_REUSE_MATRIX) { 9024 PetscValidPointer(*D,6); 9025 PetscValidHeaderSpecific(*D,MAT_CLASSID,6); 9026 ierr = PetscLogEventBegin(MAT_MatMatMult,A,B,0,0);CHKERRQ(ierr); 9027 ierr = (*(*D)->ops->matmatmult)(A,B,C,scall,fill,D);CHKERRQ(ierr); 9028 ierr = PetscLogEventEnd(MAT_MatMatMult,A,B,0,0);CHKERRQ(ierr); 9029 PetscFunctionReturn(0); 9030 } 9031 if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0; 9032 if (fill < 1.0) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill); 9033 9034 fA = A->ops->matmatmult; 9035 fB = B->ops->matmatmult; 9036 fC = C->ops->matmatmult; 9037 if (fA == fB && fA == fC) { 9038 if (!fA) SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatMatMatMult not supported for A of type %s",((PetscObject)A)->type_name); 9039 mult = fA; 9040 } else { 9041 /* dispatch based on the type of A, B and C from their PetscObject's PetscFunctionLists. */ 9042 char multname[256]; 9043 ierr = PetscStrcpy(multname,"MatMatMatMult_");CHKERRQ(ierr); 9044 ierr = PetscStrcat(multname,((PetscObject)A)->type_name);CHKERRQ(ierr); 9045 ierr = PetscStrcat(multname,"_");CHKERRQ(ierr); 9046 ierr = PetscStrcat(multname,((PetscObject)B)->type_name);CHKERRQ(ierr); 9047 ierr = PetscStrcat(multname,"_");CHKERRQ(ierr); 9048 ierr = PetscStrcat(multname,((PetscObject)C)->type_name);CHKERRQ(ierr); 9049 ierr = PetscStrcat(multname,"_C");CHKERRQ(ierr); 9050 ierr = PetscObjectQueryFunction((PetscObject)B,multname,&mult);CHKERRQ(ierr); 9051 if (!mult) SETERRQ3(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_INCOMP,"MatMatMatMult requires A, %s, to be compatible with B, %s, C, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name,((PetscObject)C)->type_name); 9052 } 9053 ierr = PetscLogEventBegin(MAT_MatMatMult,A,B,0,0);CHKERRQ(ierr); 9054 ierr = (*mult)(A,B,C,scall,fill,D);CHKERRQ(ierr); 9055 ierr = PetscLogEventEnd(MAT_MatMatMult,A,B,0,0);CHKERRQ(ierr); 9056 PetscFunctionReturn(0); 9057 } 9058 9059 #undef __FUNCT__ 9060 #define __FUNCT__ "MatGetRedundantMatrix" 9061 /*@C 9062 MatGetRedundantMatrix - Create redundant matrices and put them into processors of subcommunicators. 9063 9064 Collective on Mat 9065 9066 Input Parameters: 9067 + mat - the matrix 9068 . nsubcomm - the number of subcommunicators (= number of redundant parallel or sequential matrices) 9069 . subcomm - MPI communicator split from the communicator where mat resides in 9070 . mlocal_red - number of local rows of the redundant matrix 9071 - reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 9072 9073 Output Parameter: 9074 . matredundant - redundant matrix 9075 9076 Notes: 9077 MAT_REUSE_MATRIX can only be used when the nonzero structure of the 9078 original matrix has not changed from that last call to MatGetRedundantMatrix(). 9079 9080 This routine creates the duplicated matrices in subcommunicators; you should NOT create them before 9081 calling it. 9082 9083 Only MPIAIJ matrix is supported. 9084 9085 Level: advanced 9086 9087 Concepts: subcommunicator 9088 Concepts: duplicate matrix 9089 9090 .seealso: MatDestroy() 9091 @*/ 9092 PetscErrorCode MatGetRedundantMatrix(Mat mat,PetscInt nsubcomm,MPI_Comm subcomm,PetscInt mlocal_red,MatReuse reuse,Mat *matredundant) 9093 { 9094 PetscErrorCode ierr; 9095 9096 PetscFunctionBegin; 9097 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 9098 if (nsubcomm && reuse == MAT_REUSE_MATRIX) { 9099 PetscValidPointer(*matredundant,6); 9100 PetscValidHeaderSpecific(*matredundant,MAT_CLASSID,6); 9101 } 9102 if (!mat->ops->getredundantmatrix) SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name); 9103 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9104 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9105 MatCheckPreallocated(mat,1); 9106 9107 ierr = PetscLogEventBegin(MAT_GetRedundantMatrix,mat,0,0,0);CHKERRQ(ierr); 9108 ierr = (*mat->ops->getredundantmatrix)(mat,nsubcomm,subcomm,mlocal_red,reuse,matredundant);CHKERRQ(ierr); 9109 ierr = PetscLogEventEnd(MAT_GetRedundantMatrix,mat,0,0,0);CHKERRQ(ierr); 9110 PetscFunctionReturn(0); 9111 } 9112 9113 #undef __FUNCT__ 9114 #define __FUNCT__ "MatGetMultiProcBlock" 9115 /*@C 9116 MatGetMultiProcBlock - Create multiple [bjacobi] 'parallel submatrices' from 9117 a given 'mat' object. Each submatrix can span multiple procs. 9118 9119 Collective on Mat 9120 9121 Input Parameters: 9122 + mat - the matrix 9123 . subcomm - the subcommunicator obtained by com_split(comm) 9124 - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 9125 9126 Output Parameter: 9127 . subMat - 'parallel submatrices each spans a given subcomm 9128 9129 Notes: 9130 The submatrix partition across processors is dictated by 'subComm' a 9131 communicator obtained by com_split(comm). The comm_split 9132 is not restriced to be grouped with consecutive original ranks. 9133 9134 Due the comm_split() usage, the parallel layout of the submatrices 9135 map directly to the layout of the original matrix [wrt the local 9136 row,col partitioning]. So the original 'DiagonalMat' naturally maps 9137 into the 'DiagonalMat' of the subMat, hence it is used directly from 9138 the subMat. However the offDiagMat looses some columns - and this is 9139 reconstructed with MatSetValues() 9140 9141 Level: advanced 9142 9143 Concepts: subcommunicator 9144 Concepts: submatrices 9145 9146 .seealso: MatGetSubMatrices() 9147 @*/ 9148 PetscErrorCode MatGetMultiProcBlock(Mat mat, MPI_Comm subComm, MatReuse scall,Mat *subMat) 9149 { 9150 PetscErrorCode ierr; 9151 PetscMPIInt commsize,subCommSize; 9152 9153 PetscFunctionBegin; 9154 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)mat),&commsize);CHKERRQ(ierr); 9155 ierr = MPI_Comm_size(subComm,&subCommSize);CHKERRQ(ierr); 9156 if (subCommSize > commsize) SETERRQ2(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_OUTOFRANGE,"CommSize %D < SubCommZize %D",commsize,subCommSize); 9157 9158 ierr = PetscLogEventBegin(MAT_GetMultiProcBlock,mat,0,0,0);CHKERRQ(ierr); 9159 ierr = (*mat->ops->getmultiprocblock)(mat,subComm,scall,subMat);CHKERRQ(ierr); 9160 ierr = PetscLogEventEnd(MAT_GetMultiProcBlock,mat,0,0,0);CHKERRQ(ierr); 9161 PetscFunctionReturn(0); 9162 } 9163 9164 #undef __FUNCT__ 9165 #define __FUNCT__ "MatGetLocalSubMatrix" 9166 /*@ 9167 MatGetLocalSubMatrix - Gets a reference to a submatrix specified in local numbering 9168 9169 Not Collective 9170 9171 Input Arguments: 9172 mat - matrix to extract local submatrix from 9173 isrow - local row indices for submatrix 9174 iscol - local column indices for submatrix 9175 9176 Output Arguments: 9177 submat - the submatrix 9178 9179 Level: intermediate 9180 9181 Notes: 9182 The submat should be returned with MatRestoreLocalSubMatrix(). 9183 9184 Depending on the format of mat, the returned submat may not implement MatMult(). Its communicator may be 9185 the same as mat, it may be PETSC_COMM_SELF, or some other subcomm of mat's. 9186 9187 The submat always implements MatSetValuesLocal(). If isrow and iscol have the same block size, then 9188 MatSetValuesBlockedLocal() will also be implemented. 9189 9190 .seealso: MatRestoreLocalSubMatrix(), MatCreateLocalRef() 9191 @*/ 9192 PetscErrorCode MatGetLocalSubMatrix(Mat mat,IS isrow,IS iscol,Mat *submat) 9193 { 9194 PetscErrorCode ierr; 9195 9196 PetscFunctionBegin; 9197 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 9198 PetscValidHeaderSpecific(isrow,IS_CLASSID,2); 9199 PetscValidHeaderSpecific(iscol,IS_CLASSID,3); 9200 PetscCheckSameComm(isrow,2,iscol,3); 9201 PetscValidPointer(submat,4); 9202 9203 if (mat->ops->getlocalsubmatrix) { 9204 ierr = (*mat->ops->getlocalsubmatrix)(mat,isrow,iscol,submat);CHKERRQ(ierr); 9205 } else { 9206 ierr = MatCreateLocalRef(mat,isrow,iscol,submat);CHKERRQ(ierr); 9207 } 9208 PetscFunctionReturn(0); 9209 } 9210 9211 #undef __FUNCT__ 9212 #define __FUNCT__ "MatRestoreLocalSubMatrix" 9213 /*@ 9214 MatRestoreLocalSubMatrix - Restores a reference to a submatrix specified in local numbering 9215 9216 Not Collective 9217 9218 Input Arguments: 9219 mat - matrix to extract local submatrix from 9220 isrow - local row indices for submatrix 9221 iscol - local column indices for submatrix 9222 submat - the submatrix 9223 9224 Level: intermediate 9225 9226 .seealso: MatGetLocalSubMatrix() 9227 @*/ 9228 PetscErrorCode MatRestoreLocalSubMatrix(Mat mat,IS isrow,IS iscol,Mat *submat) 9229 { 9230 PetscErrorCode ierr; 9231 9232 PetscFunctionBegin; 9233 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 9234 PetscValidHeaderSpecific(isrow,IS_CLASSID,2); 9235 PetscValidHeaderSpecific(iscol,IS_CLASSID,3); 9236 PetscCheckSameComm(isrow,2,iscol,3); 9237 PetscValidPointer(submat,4); 9238 if (*submat) { 9239 PetscValidHeaderSpecific(*submat,MAT_CLASSID,4); 9240 } 9241 9242 if (mat->ops->restorelocalsubmatrix) { 9243 ierr = (*mat->ops->restorelocalsubmatrix)(mat,isrow,iscol,submat);CHKERRQ(ierr); 9244 } else { 9245 ierr = MatDestroy(submat);CHKERRQ(ierr); 9246 } 9247 *submat = NULL; 9248 PetscFunctionReturn(0); 9249 } 9250 9251 /* --------------------------------------------------------*/ 9252 #undef __FUNCT__ 9253 #define __FUNCT__ "MatFindZeroDiagonals" 9254 /*@ 9255 MatFindZeroDiagonals - Finds all the rows of a matrix that have zero or no entry in the matrix 9256 9257 Collective on Mat 9258 9259 Input Parameter: 9260 . mat - the matrix 9261 9262 Output Parameter: 9263 . is - if any rows have zero diagonals this contains the list of them 9264 9265 Level: developer 9266 9267 Concepts: matrix-vector product 9268 9269 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 9270 @*/ 9271 PetscErrorCode MatFindZeroDiagonals(Mat mat,IS *is) 9272 { 9273 PetscErrorCode ierr; 9274 9275 PetscFunctionBegin; 9276 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 9277 PetscValidType(mat,1); 9278 if (!mat->assembled) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9279 if (mat->factortype) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9280 9281 if (!mat->ops->findzerodiagonals) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"This matrix type does not have a find zero diagonals defined"); 9282 ierr = (*mat->ops->findzerodiagonals)(mat,is);CHKERRQ(ierr); 9283 PetscFunctionReturn(0); 9284 } 9285 9286 #undef __FUNCT__ 9287 #define __FUNCT__ "MatInvertBlockDiagonal" 9288 /*@C 9289 MatInvertBlockDiagonal - Inverts the block diagonal entries. 9290 9291 Collective on Mat 9292 9293 Input Parameters: 9294 . mat - the matrix 9295 9296 Output Parameters: 9297 . values - the block inverses in column major order (FORTRAN-like) 9298 9299 Note: 9300 This routine is not available from Fortran. 9301 9302 Level: advanced 9303 @*/ 9304 PetscErrorCode MatInvertBlockDiagonal(Mat mat,const PetscScalar **values) 9305 { 9306 PetscErrorCode ierr; 9307 9308 PetscFunctionBegin; 9309 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 9310 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 9311 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 9312 if (!mat->ops->invertblockdiagonal) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported"); 9313 ierr = (*mat->ops->invertblockdiagonal)(mat,values);CHKERRQ(ierr); 9314 PetscFunctionReturn(0); 9315 } 9316 9317 #undef __FUNCT__ 9318 #define __FUNCT__ "MatTransposeColoringDestroy" 9319 /*@C 9320 MatTransposeColoringDestroy - Destroys a coloring context for matrix product C=A*B^T that was created 9321 via MatTransposeColoringCreate(). 9322 9323 Collective on MatTransposeColoring 9324 9325 Input Parameter: 9326 . c - coloring context 9327 9328 Level: intermediate 9329 9330 .seealso: MatTransposeColoringCreate() 9331 @*/ 9332 PetscErrorCode MatTransposeColoringDestroy(MatTransposeColoring *c) 9333 { 9334 PetscErrorCode ierr; 9335 MatTransposeColoring matcolor=*c; 9336 9337 PetscFunctionBegin; 9338 if (!matcolor) PetscFunctionReturn(0); 9339 if (--((PetscObject)matcolor)->refct > 0) {matcolor = 0; PetscFunctionReturn(0);} 9340 9341 ierr = PetscFree3(matcolor->ncolumns,matcolor->nrows,matcolor->colorforrow);CHKERRQ(ierr); 9342 ierr = PetscFree(matcolor->rows);CHKERRQ(ierr); 9343 ierr = PetscFree(matcolor->den2sp);CHKERRQ(ierr); 9344 ierr = PetscFree(matcolor->colorforcol);CHKERRQ(ierr); 9345 ierr = PetscFree(matcolor->columns);CHKERRQ(ierr); 9346 if (matcolor->brows>0) { 9347 ierr = PetscFree(matcolor->lstart);CHKERRQ(ierr); 9348 } 9349 ierr = PetscHeaderDestroy(c);CHKERRQ(ierr); 9350 PetscFunctionReturn(0); 9351 } 9352 9353 #undef __FUNCT__ 9354 #define __FUNCT__ "MatTransColoringApplySpToDen" 9355 /*@C 9356 MatTransColoringApplySpToDen - Given a symbolic matrix product C=A*B^T for which 9357 a MatTransposeColoring context has been created, computes a dense B^T by Apply 9358 MatTransposeColoring to sparse B. 9359 9360 Collective on MatTransposeColoring 9361 9362 Input Parameters: 9363 + B - sparse matrix B 9364 . Btdense - symbolic dense matrix B^T 9365 - coloring - coloring context created with MatTransposeColoringCreate() 9366 9367 Output Parameter: 9368 . Btdense - dense matrix B^T 9369 9370 Options Database Keys: 9371 + -mat_transpose_coloring_view - Activates basic viewing or coloring 9372 . -mat_transpose_coloring_view_draw - Activates drawing of coloring 9373 - -mat_transpose_coloring_view_info - Activates viewing of coloring info 9374 9375 Level: intermediate 9376 9377 .seealso: MatTransposeColoringCreate(), MatTransposeColoringDestroy() 9378 9379 .keywords: coloring 9380 @*/ 9381 PetscErrorCode MatTransColoringApplySpToDen(MatTransposeColoring coloring,Mat B,Mat Btdense) 9382 { 9383 PetscErrorCode ierr; 9384 9385 PetscFunctionBegin; 9386 PetscValidHeaderSpecific(B,MAT_CLASSID,1); 9387 PetscValidHeaderSpecific(Btdense,MAT_CLASSID,2); 9388 PetscValidHeaderSpecific(coloring,MAT_TRANSPOSECOLORING_CLASSID,3); 9389 9390 if (!B->ops->transcoloringapplysptoden) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for this matrix type %s",((PetscObject)B)->type_name); 9391 ierr = (B->ops->transcoloringapplysptoden)(coloring,B,Btdense);CHKERRQ(ierr); 9392 PetscFunctionReturn(0); 9393 } 9394 9395 #undef __FUNCT__ 9396 #define __FUNCT__ "MatTransColoringApplyDenToSp" 9397 /*@C 9398 MatTransColoringApplyDenToSp - Given a symbolic matrix product Csp=A*B^T for which 9399 a MatTransposeColoring context has been created and a dense matrix Cden=A*Btdense 9400 in which Btdens is obtained from MatTransColoringApplySpToDen(), recover sparse matrix 9401 Csp from Cden. 9402 9403 Collective on MatTransposeColoring 9404 9405 Input Parameters: 9406 + coloring - coloring context created with MatTransposeColoringCreate() 9407 - Cden - matrix product of a sparse matrix and a dense matrix Btdense 9408 9409 Output Parameter: 9410 . Csp - sparse matrix 9411 9412 Options Database Keys: 9413 + -mat_multtranspose_coloring_view - Activates basic viewing or coloring 9414 . -mat_multtranspose_coloring_view_draw - Activates drawing of coloring 9415 - -mat_multtranspose_coloring_view_info - Activates viewing of coloring info 9416 9417 Level: intermediate 9418 9419 .seealso: MatTransposeColoringCreate(), MatTransposeColoringDestroy(), MatTransColoringApplySpToDen() 9420 9421 .keywords: coloring 9422 @*/ 9423 PetscErrorCode MatTransColoringApplyDenToSp(MatTransposeColoring matcoloring,Mat Cden,Mat Csp) 9424 { 9425 PetscErrorCode ierr; 9426 9427 PetscFunctionBegin; 9428 PetscValidHeaderSpecific(matcoloring,MAT_TRANSPOSECOLORING_CLASSID,1); 9429 PetscValidHeaderSpecific(Cden,MAT_CLASSID,2); 9430 PetscValidHeaderSpecific(Csp,MAT_CLASSID,3); 9431 9432 if (!Csp->ops->transcoloringapplydentosp) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not supported for this matrix type %s",((PetscObject)Csp)->type_name); 9433 ierr = (Csp->ops->transcoloringapplydentosp)(matcoloring,Cden,Csp);CHKERRQ(ierr); 9434 PetscFunctionReturn(0); 9435 } 9436 9437 #undef __FUNCT__ 9438 #define __FUNCT__ "MatTransposeColoringCreate" 9439 /*@C 9440 MatTransposeColoringCreate - Creates a matrix coloring context for matrix product C=A*B^T. 9441 9442 Collective on Mat 9443 9444 Input Parameters: 9445 + mat - the matrix product C 9446 - iscoloring - the coloring of the matrix; usually obtained with MatGetColoring() or DMCreateColoring() 9447 9448 Output Parameter: 9449 . color - the new coloring context 9450 9451 Level: intermediate 9452 9453 .seealso: MatTransposeColoringDestroy(), MatTransposeColoringSetFromOptions(), MatTransColoringApplySpToDen(), 9454 MatTransColoringApplyDenToSp(), MatTransposeColoringView(), 9455 @*/ 9456 PetscErrorCode MatTransposeColoringCreate(Mat mat,ISColoring iscoloring,MatTransposeColoring *color) 9457 { 9458 MatTransposeColoring c; 9459 MPI_Comm comm; 9460 PetscErrorCode ierr; 9461 9462 PetscFunctionBegin; 9463 ierr = PetscLogEventBegin(MAT_TransposeColoringCreate,mat,0,0,0);CHKERRQ(ierr); 9464 ierr = PetscObjectGetComm((PetscObject)mat,&comm);CHKERRQ(ierr); 9465 ierr = PetscHeaderCreate(c,_p_MatTransposeColoring,int,MAT_TRANSPOSECOLORING_CLASSID,"MatTransposeColoring","Matrix product C=A*B^T via coloring","Mat",comm,MatTransposeColoringDestroy,0);CHKERRQ(ierr); 9466 9467 c->ctype = iscoloring->ctype; 9468 if (mat->ops->transposecoloringcreate) { 9469 ierr = (*mat->ops->transposecoloringcreate)(mat,iscoloring,c);CHKERRQ(ierr); 9470 } else SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Code not yet written for this matrix type"); 9471 9472 *color = c; 9473 ierr = PetscLogEventEnd(MAT_TransposeColoringCreate,mat,0,0,0);CHKERRQ(ierr); 9474 PetscFunctionReturn(0); 9475 } 9476