1 /*$Id: matrix.c,v 1.367 2000/05/16 17:02:11 bsmith Exp bsmith $*/ 2 3 /* 4 This is where the abstract matrix operations are defined 5 */ 6 7 #include "src/mat/matimpl.h" /*I "petscmat.h" I*/ 8 #include "src/vec/vecimpl.h" 9 10 #undef __FUNC__ 11 #define __FUNC__ /*<a name=""></a>*/"MatGetRow" 12 /*@C 13 MatGetRow - Gets a row of a matrix. You MUST call MatRestoreRow() 14 for each row that you get to ensure that your application does 15 not bleed memory. 16 17 Not Collective 18 19 Input Parameters: 20 + mat - the matrix 21 - row - the row to get 22 23 Output Parameters: 24 + ncols - the number of nonzeros in the row 25 . cols - if not NULL, the column numbers 26 - vals - if not NULL, the values 27 28 Notes: 29 This routine is provided for people who need to have direct access 30 to the structure of a matrix. We hope that we provide enough 31 high-level matrix routines that few users will need it. 32 33 MatGetRow() always returns 0-based column indices, regardless of 34 whether the internal representation is 0-based (default) or 1-based. 35 36 For better efficiency, set cols and/or vals to PETSC_NULL if you do 37 not wish to extract these quantities. 38 39 The user can only examine the values extracted with MatGetRow(); 40 the values cannot be altered. To change the matrix entries, one 41 must use MatSetValues(). 42 43 You can only have one call to MatGetRow() outstanding for a particular 44 matrix at a time, per processor. MatGetRow() can only obtained rows 45 associated with the given processor, it cannot get rows from the 46 other processors; for that we suggest using MatGetSubMatrices(), then 47 MatGetRow() on the submatrix. The row indix passed to MatGetRows() 48 is in the global number of rows. 49 50 Fortran Notes: 51 The calling sequence from Fortran is 52 .vb 53 MatGetRow(matrix,row,ncols,cols,values,ierr) 54 Mat matrix (input) 55 integer row (input) 56 integer ncols (output) 57 integer cols(maxcols) (output) 58 double precision (or double complex) values(maxcols) output 59 .ve 60 where maxcols >= maximum nonzeros in any row of the matrix. 61 62 Caution: 63 Do not try to change the contents of the output arrays (cols and vals). 64 In some cases, this may corrupt the matrix. 65 66 Level: advanced 67 68 .keywords: matrix, row, get, extract 69 70 .seealso: MatRestoreRow(), MatSetValues(), MatGetValues(), MatGetSubmatrices(), MatGetDiagonal() 71 @*/ 72 int MatGetRow(Mat mat,int row,int *ncols,int **cols,Scalar **vals) 73 { 74 int ierr; 75 76 PetscFunctionBegin; 77 PetscValidHeaderSpecific(mat,MAT_COOKIE); 78 PetscValidIntPointer(ncols); 79 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 80 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 81 if (!mat->ops->getrow) SETERRQ(PETSC_ERR_SUP,0,""); 82 PLogEventBegin(MAT_GetRow,mat,0,0,0); 83 ierr = (*mat->ops->getrow)(mat,row,ncols,cols,vals);CHKERRQ(ierr); 84 PLogEventEnd(MAT_GetRow,mat,0,0,0); 85 PetscFunctionReturn(0); 86 } 87 88 #undef __FUNC__ 89 #define __FUNC__ /*<a name=""></a>*/"MatRestoreRow" 90 /*@C 91 MatRestoreRow - Frees any temporary space allocated by MatGetRow(). 92 93 Not Collective 94 95 Input Parameters: 96 + mat - the matrix 97 . row - the row to get 98 . ncols, cols - the number of nonzeros and their columns 99 - vals - if nonzero the column values 100 101 Notes: 102 This routine should be called after you have finished examining the entries. 103 104 Fortran Notes: 105 The calling sequence from Fortran is 106 .vb 107 MatRestoreRow(matrix,row,ncols,cols,values,ierr) 108 Mat matrix (input) 109 integer row (input) 110 integer ncols (output) 111 integer cols(maxcols) (output) 112 double precision (or double complex) values(maxcols) output 113 .ve 114 Where maxcols >= maximum nonzeros in any row of the matrix. 115 116 In Fortran MatRestoreRow() MUST be called after MatGetRow() 117 before another call to MatGetRow() can be made. 118 119 Level: advanced 120 121 .keywords: matrix, row, restore 122 123 .seealso: MatGetRow() 124 @*/ 125 int MatRestoreRow(Mat mat,int row,int *ncols,int **cols,Scalar **vals) 126 { 127 int ierr; 128 129 PetscFunctionBegin; 130 PetscValidHeaderSpecific(mat,MAT_COOKIE); 131 PetscValidIntPointer(ncols); 132 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 133 if (!mat->ops->restorerow) PetscFunctionReturn(0); 134 ierr = (*mat->ops->restorerow)(mat,row,ncols,cols,vals);CHKERRQ(ierr); 135 PetscFunctionReturn(0); 136 } 137 138 #undef __FUNC__ 139 #define __FUNC__ /*<a name=""></a>*/"MatView" 140 /*@C 141 MatView - Visualizes a matrix object. 142 143 Collective on Mat 144 145 Input Parameters: 146 + mat - the matrix 147 - ptr - visualization context 148 149 Notes: 150 The available visualization contexts include 151 + VIEWER_STDOUT_SELF - standard output (default) 152 . VIEWER_STDOUT_WORLD - synchronized standard 153 output where only the first processor opens 154 the file. All other processors send their 155 data to the first processor to print. 156 - VIEWER_DRAW_WORLD - graphical display of nonzero structure 157 158 The user can open alternative visualization contexts with 159 + ViewerASCIIOpen() - Outputs matrix to a specified file 160 . ViewerBinaryOpen() - Outputs matrix in binary to a 161 specified file; corresponding input uses MatLoad() 162 . ViewerDrawOpen() - Outputs nonzero matrix structure to 163 an X window display 164 - ViewerSocketOpen() - Outputs matrix to Socket viewer. 165 Currently only the sequential dense and AIJ 166 matrix types support the Socket viewer. 167 168 The user can call ViewerSetFormat() to specify the output 169 format of ASCII printed objects (when using VIEWER_STDOUT_SELF, 170 VIEWER_STDOUT_WORLD and ViewerASCIIOpen). Available formats include 171 + VIEWER_FORMAT_ASCII_DEFAULT - default, prints matrix contents 172 . VIEWER_FORMAT_ASCII_MATLAB - prints matrix contents in Matlab format 173 . VIEWER_FORMAT_ASCII_DENSE - prints entire matrix including zeros 174 . VIEWER_FORMAT_ASCII_COMMON - prints matrix contents, using a sparse 175 format common among all matrix types 176 . VIEWER_FORMAT_ASCII_IMPL - prints matrix contents, using an implementation-specific 177 format (which is in many cases the same as the default) 178 . VIEWER_FORMAT_ASCII_INFO - prints basic information about the matrix 179 size and structure (not the matrix entries) 180 - VIEWER_FORMAT_ASCII_INFO_LONG - prints more detailed information about 181 the matrix structure 182 183 Level: beginner 184 185 .keywords: matrix, view, visualize, output, print, write, draw 186 187 .seealso: ViewerSetFormat(), ViewerASCIIOpen(), ViewerDrawOpen(), 188 ViewerSocketOpen(), ViewerBinaryOpen(), MatLoad() 189 @*/ 190 int MatView(Mat mat,Viewer viewer) 191 { 192 int format,ierr,rows,cols; 193 PetscTruth isascii; 194 char *cstr; 195 196 PetscFunctionBegin; 197 PetscValidHeaderSpecific(mat,MAT_COOKIE); 198 if (!viewer) viewer = VIEWER_STDOUT_(mat->comm); 199 PetscValidHeaderSpecific(viewer,VIEWER_COOKIE); 200 PetscCheckSameComm(mat,viewer); 201 if (!mat->assembled) SETERRQ(1,1,"Must call MatAssemblyBegin/End() before viewing matrix"); 202 203 ierr = PetscTypeCompare((PetscObject)viewer,ASCII_VIEWER,&isascii);CHKERRQ(ierr); 204 if (isascii) { 205 ierr = ViewerGetFormat(viewer,&format);CHKERRQ(ierr); 206 if (format == VIEWER_FORMAT_ASCII_INFO || format == VIEWER_FORMAT_ASCII_INFO_LONG) { 207 ierr = ViewerASCIIPrintf(viewer,"Matrix Object:\n");CHKERRQ(ierr); 208 ierr = ViewerASCIIPushTab(viewer);CHKERRQ(ierr); 209 ierr = MatGetType(mat,PETSC_NULL,&cstr);CHKERRQ(ierr); 210 ierr = MatGetSize(mat,&rows,&cols);CHKERRQ(ierr); 211 ierr = ViewerASCIIPrintf(viewer,"type=%s, rows=%d, cols=%d\n",cstr,rows,cols);CHKERRQ(ierr); 212 if (mat->ops->getinfo) { 213 MatInfo info; 214 ierr = MatGetInfo(mat,MAT_GLOBAL_SUM,&info);CHKERRQ(ierr); 215 ierr = ViewerASCIIPrintf(viewer,"total: nonzeros=%d, allocated nonzeros=%d\n", 216 (int)info.nz_used,(int)info.nz_allocated);CHKERRQ(ierr); 217 } 218 } 219 } 220 if (mat->ops->view) { 221 ierr = ViewerASCIIPushTab(viewer);CHKERRQ(ierr); 222 ierr = (*mat->ops->view)(mat,viewer);CHKERRQ(ierr); 223 ierr = ViewerASCIIPopTab(viewer);CHKERRQ(ierr); 224 } else if (!isascii) { 225 SETERRQ1(1,1,"Viewer type %s not supported",((PetscObject)viewer)->type_name); 226 } 227 if (isascii) { 228 ierr = ViewerGetFormat(viewer,&format);CHKERRQ(ierr); 229 if (format == VIEWER_FORMAT_ASCII_INFO || format == VIEWER_FORMAT_ASCII_INFO_LONG) { 230 ierr = ViewerASCIIPopTab(viewer);CHKERRQ(ierr); 231 } 232 } 233 PetscFunctionReturn(0); 234 } 235 236 #undef __FUNC__ 237 #define __FUNC__ /*<a name=""></a>*/"MatScaleSystem" 238 /*@C 239 MatScaleSystem - Scale a vector solution and right hand side to 240 match the scaling of a scaled matrix. 241 242 Collective on Mat 243 244 Input Parameter: 245 + mat - the matrix 246 . x - solution vector (or PETSC_NULL) 247 + b - right hand side vector (or PETSC_NULL) 248 249 250 Notes: 251 For AIJ, BAIJ, and BDiag matrix formats, the matrices are not 252 internally scaled, so this does nothing. For MPIROWBS it 253 permutes and diagonally scales. 254 255 The SLES methods automatically call this routine when required 256 (via PCPreSolve()) so it is rarely used directly. 257 258 Level: Developer 259 260 .keywords: matrix, scale 261 262 .seealso: MatUseScaledForm(), MatUnScaleSystem() 263 @*/ 264 int MatScaleSystem(Mat mat,Vec x,Vec b) 265 { 266 int ierr; 267 268 PetscFunctionBegin; 269 PetscValidHeaderSpecific(mat,MAT_COOKIE); 270 if (x) {PetscValidHeaderSpecific(x,VEC_COOKIE);PetscCheckSameComm(mat,x);} 271 if (b) {PetscValidHeaderSpecific(b,VEC_COOKIE);PetscCheckSameComm(mat,b);} 272 273 if (mat->ops->scalesystem) { 274 ierr = (*mat->ops->scalesystem)(mat,x,b);CHKERRQ(ierr); 275 } 276 PetscFunctionReturn(0); 277 } 278 279 #undef __FUNC__ 280 #define __FUNC__ /*<a name=""></a>*/"MatUnScaleSystem" 281 /*@C 282 MatUnScaleSystem - Unscales a vector solution and right hand side to 283 match the original scaling of a scaled matrix. 284 285 Collective on Mat 286 287 Input Parameter: 288 + mat - the matrix 289 . x - solution vector (or PETSC_NULL) 290 + b - right hand side vector (or PETSC_NULL) 291 292 293 Notes: 294 For AIJ, BAIJ, and BDiag matrix formats, the matrices are not 295 internally scaled, so this does nothing. For MPIROWBS it 296 permutes and diagonally scales. 297 298 The SLES methods automatically call this routine when required 299 (via PCPreSolve()) so it is rarely used directly. 300 301 Level: Developer 302 303 .keywords: matrix, scale 304 305 .seealso: MatUseScaledForm(), MatScaleSystem() 306 @*/ 307 int MatUnScaleSystem(Mat mat,Vec x,Vec b) 308 { 309 int ierr; 310 311 PetscFunctionBegin; 312 PetscValidHeaderSpecific(mat,MAT_COOKIE); 313 if (x) {PetscValidHeaderSpecific(x,VEC_COOKIE);PetscCheckSameComm(mat,x);} 314 if (b) {PetscValidHeaderSpecific(b,VEC_COOKIE);PetscCheckSameComm(mat,b);} 315 if (mat->ops->unscalesystem) { 316 ierr = (*mat->ops->unscalesystem)(mat,x,b);CHKERRQ(ierr); 317 } 318 PetscFunctionReturn(0); 319 } 320 321 #undef __FUNC__ 322 #define __FUNC__ /*<a name=""></a>*/"MatUseScaledForm" 323 /*@C 324 MatUseScaledForm - For matrix storage formats that scale the 325 matrix (for example MPIRowBS matrices are diagonally scaled on 326 assembly) indicates matrix operations (MatMult() etc) are 327 applied using the scaled matrix. 328 329 Collective on Mat 330 331 Input Parameter: 332 + mat - the matrix 333 - scaled - PETSC_TRUE for applying the scaled, PETSC_FALSE for 334 applying the original matrix 335 336 Notes: 337 For scaled matrix formats, applying the original, unscaled matrix 338 will be slightly more expensive 339 340 Level: Developer 341 342 .keywords: matrix, scale 343 344 .seealso: MatScaleSystem(), MatUnScaleSystem() 345 @*/ 346 int MatUseScaledForm(Mat mat,PetscTruth scaled) 347 { 348 int ierr; 349 350 PetscFunctionBegin; 351 PetscValidHeaderSpecific(mat,MAT_COOKIE); 352 if (mat->ops->usescaledform) { 353 ierr = (*mat->ops->usescaledform)(mat,scaled);CHKERRQ(ierr); 354 } 355 PetscFunctionReturn(0); 356 } 357 358 #undef __FUNC__ 359 #define __FUNC__ /*<a name=""></a>*/"MatDestroy" 360 /*@C 361 MatDestroy - Frees space taken by a matrix. 362 363 Collective on Mat 364 365 Input Parameter: 366 . mat - the matrix 367 368 Level: beginner 369 370 .keywords: matrix, destroy 371 @*/ 372 int MatDestroy(Mat mat) 373 { 374 int ierr; 375 376 PetscFunctionBegin; 377 PetscValidHeaderSpecific(mat,MAT_COOKIE); 378 if (--mat->refct > 0) PetscFunctionReturn(0); 379 380 /* if memory was published with AMS then destroy it */ 381 ierr = PetscObjectDepublish(mat);CHKERRQ(ierr); 382 383 ierr = (*mat->ops->destroy)(mat);CHKERRQ(ierr); 384 PetscFunctionReturn(0); 385 } 386 387 #undef __FUNC__ 388 #define __FUNC__ /*<a name=""></a>*/"MatValid" 389 /*@ 390 MatValid - Checks whether a matrix object is valid. 391 392 Collective on Mat 393 394 Input Parameter: 395 . m - the matrix to check 396 397 Output Parameter: 398 flg - flag indicating matrix status, either 399 PETSC_TRUE if matrix is valid, or PETSC_FALSE otherwise. 400 401 Level: developer 402 403 .keywords: matrix, valid 404 @*/ 405 int MatValid(Mat m,PetscTruth *flg) 406 { 407 PetscFunctionBegin; 408 PetscValidIntPointer(flg); 409 if (!m) *flg = PETSC_FALSE; 410 else if (m->cookie != MAT_COOKIE) *flg = PETSC_FALSE; 411 else *flg = PETSC_TRUE; 412 PetscFunctionReturn(0); 413 } 414 415 #undef __FUNC__ 416 #define __FUNC__ /*<a name=""></a>*/"MatSetValues" 417 /*@ 418 MatSetValues - Inserts or adds a block of values into a matrix. 419 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 420 MUST be called after all calls to MatSetValues() have been completed. 421 422 Not Collective 423 424 Input Parameters: 425 + mat - the matrix 426 . v - a logically two-dimensional array of values 427 . m, idxm - the number of rows and their global indices 428 . n, idxn - the number of columns and their global indices 429 - addv - either ADD_VALUES or INSERT_VALUES, where 430 ADD_VALUES adds values to any existing entries, and 431 INSERT_VALUES replaces existing entries with new values 432 433 Notes: 434 By default the values, v, are row-oriented and unsorted. 435 See MatSetOption() for other options. 436 437 Calls to MatSetValues() with the INSERT_VALUES and ADD_VALUES 438 options cannot be mixed without intervening calls to the assembly 439 routines. 440 441 MatSetValues() uses 0-based row and column numbers in Fortran 442 as well as in C. 443 444 Negative indices may be passed in idxm and idxn, these rows and columns are 445 simply ignored. This allows easily inserting element stiffness matrices 446 with homogeneous Dirchlet boundary conditions that you don't want represented 447 in the matrix. 448 449 Efficiency Alert: 450 The routine MatSetValuesBlocked() may offer much better efficiency 451 for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ). 452 453 Level: beginner 454 455 .keywords: matrix, insert, add, set, values 456 457 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal() 458 @*/ 459 int MatSetValues(Mat mat,int m,int *idxm,int n,int *idxn,Scalar *v,InsertMode addv) 460 { 461 int ierr; 462 463 PetscFunctionBegin; 464 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 465 PetscValidHeaderSpecific(mat,MAT_COOKIE); 466 PetscValidIntPointer(idxm); 467 PetscValidIntPointer(idxn); 468 PetscValidScalarPointer(v); 469 if (mat->insertmode == NOT_SET_VALUES) { 470 mat->insertmode = addv; 471 } 472 #if defined(PETSC_USE_BOPT_g) 473 else if (mat->insertmode != addv) { 474 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,1,"Cannot mix add values and insert values"); 475 } 476 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 477 #endif 478 479 if (mat->assembled) { 480 mat->was_assembled = PETSC_TRUE; 481 mat->assembled = PETSC_FALSE; 482 } 483 PLogEventBegin(MAT_SetValues,mat,0,0,0); 484 if (!mat->ops->setvalues) SETERRQ(PETSC_ERR_SUP,1,"Not supported for this matrix type"); 485 ierr = (*mat->ops->setvalues)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr); 486 PLogEventEnd(MAT_SetValues,mat,0,0,0); 487 PetscFunctionReturn(0); 488 } 489 490 #undef __FUNC__ 491 #define __FUNC__ /*<a name=""></a>*/"MatSetValuesBlocked" 492 /*@ 493 MatSetValuesBlocked - Inserts or adds a block of values into a matrix. 494 495 Not Collective 496 497 Input Parameters: 498 + mat - the matrix 499 . v - a logically two-dimensional array of values 500 . m, idxm - the number of block rows and their global block indices 501 . n, idxn - the number of block columns and their global block indices 502 - addv - either ADD_VALUES or INSERT_VALUES, where 503 ADD_VALUES adds values to any existing entries, and 504 INSERT_VALUES replaces existing entries with new values 505 506 Notes: 507 By default the values, v, are row-oriented and unsorted. So the layout of 508 v is the same as for MatSetValues(). See MatSetOption() for other options. 509 510 Calls to MatSetValuesBlocked() with the INSERT_VALUES and ADD_VALUES 511 options cannot be mixed without intervening calls to the assembly 512 routines. 513 514 MatSetValuesBlocked() uses 0-based row and column numbers in Fortran 515 as well as in C. 516 517 Negative indices may be passed in idxm and idxn, these rows and columns are 518 simply ignored. This allows easily inserting element stiffness matrices 519 with homogeneous Dirchlet boundary conditions that you don't want represented 520 in the matrix. 521 522 Each time an entry is set within a sparse matrix via MatSetValues(), 523 internal searching must be done to determine where to place the the 524 data in the matrix storage space. By instead inserting blocks of 525 entries via MatSetValuesBlocked(), the overhead of matrix assembly is 526 reduced. 527 528 Restrictions: 529 MatSetValuesBlocked() is currently supported only for the block AIJ 530 matrix format (MATSEQBAIJ and MATMPIBAIJ, which are created via 531 MatCreateSeqBAIJ() and MatCreateMPIBAIJ()). 532 533 Level: intermediate 534 535 .keywords: matrix, insert, add, set, values 536 537 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal() 538 @*/ 539 int MatSetValuesBlocked(Mat mat,int m,int *idxm,int n,int *idxn,Scalar *v,InsertMode addv) 540 { 541 int ierr; 542 543 PetscFunctionBegin; 544 if (!m || !n) PetscFunctionReturn(0); /* no values to insert */ 545 PetscValidHeaderSpecific(mat,MAT_COOKIE); 546 PetscValidIntPointer(idxm); 547 PetscValidIntPointer(idxn); 548 PetscValidScalarPointer(v); 549 if (mat->insertmode == NOT_SET_VALUES) { 550 mat->insertmode = addv; 551 } 552 #if defined(PETSC_USE_BOPT_g) 553 else if (mat->insertmode != addv) { 554 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,1,"Cannot mix add values and insert values"); 555 } 556 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 557 #endif 558 559 if (mat->assembled) { 560 mat->was_assembled = PETSC_TRUE; 561 mat->assembled = PETSC_FALSE; 562 } 563 PLogEventBegin(MAT_SetValues,mat,0,0,0); 564 if (!mat->ops->setvaluesblocked) SETERRQ(PETSC_ERR_SUP,1,"Not supported for this matrix type"); 565 ierr = (*mat->ops->setvaluesblocked)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr); 566 PLogEventEnd(MAT_SetValues,mat,0,0,0); 567 PetscFunctionReturn(0); 568 } 569 570 /*MC 571 MatSetValue - Set a single entry into a matrix. 572 573 Synopsis: 574 void MatSetValue(Mat m,int row,int col,Scalar value,InsertMode mode); 575 576 Not collective 577 578 Input Parameters: 579 + m - the matrix 580 . row - the row location of the entry 581 . col - the column location of the entry 582 . value - the value to insert 583 - mode - either INSERT_VALUES or ADD_VALUES 584 585 Notes: 586 For efficiency one should use MatSetValues() and set several or many 587 values simultaneously if possible. 588 589 Note that VecSetValue() does NOT return an error code (since this 590 is checked internally). 591 592 Level: beginner 593 594 .seealso: MatSetValues() 595 M*/ 596 597 #undef __FUNC__ 598 #define __FUNC__ /*<a name=""></a>*/"MatGetValues" 599 /*@ 600 MatGetValues - Gets a block of values from a matrix. 601 602 Not Collective; currently only returns a local block 603 604 Input Parameters: 605 + mat - the matrix 606 . v - a logically two-dimensional array for storing the values 607 . m, idxm - the number of rows and their global indices 608 - n, idxn - the number of columns and their global indices 609 610 Notes: 611 The user must allocate space (m*n Scalars) for the values, v. 612 The values, v, are then returned in a row-oriented format, 613 analogous to that used by default in MatSetValues(). 614 615 MatGetValues() uses 0-based row and column numbers in 616 Fortran as well as in C. 617 618 MatGetValues() requires that the matrix has been assembled 619 with MatAssemblyBegin()/MatAssemblyEnd(). Thus, calls to 620 MatSetValues() and MatGetValues() CANNOT be made in succession 621 without intermediate matrix assembly. 622 623 Level: advanced 624 625 .keywords: matrix, get, values 626 627 .seealso: MatGetRow(), MatGetSubmatrices(), MatSetValues() 628 @*/ 629 int MatGetValues(Mat mat,int m,int *idxm,int n,int *idxn,Scalar *v) 630 { 631 int ierr; 632 633 PetscFunctionBegin; 634 PetscValidHeaderSpecific(mat,MAT_COOKIE); 635 PetscValidIntPointer(idxm); 636 PetscValidIntPointer(idxn); 637 PetscValidScalarPointer(v); 638 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 639 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 640 if (!mat->ops->getvalues) SETERRQ(PETSC_ERR_SUP,0,""); 641 642 PLogEventBegin(MAT_GetValues,mat,0,0,0); 643 ierr = (*mat->ops->getvalues)(mat,m,idxm,n,idxn,v);CHKERRQ(ierr); 644 PLogEventEnd(MAT_GetValues,mat,0,0,0); 645 PetscFunctionReturn(0); 646 } 647 648 #undef __FUNC__ 649 #define __FUNC__ /*<a name=""></a>*/"MatSetLocalToGlobalMapping" 650 /*@ 651 MatSetLocalToGlobalMapping - Sets a local-to-global numbering for use by 652 the routine MatSetValuesLocal() to allow users to insert matrix entries 653 using a local (per-processor) numbering. 654 655 Not Collective 656 657 Input Parameters: 658 + x - the matrix 659 - mapping - mapping created with ISLocalToGlobalMappingCreate() 660 or ISLocalToGlobalMappingCreateIS() 661 662 Level: intermediate 663 664 .keywords: matrix, set, values, local, global, mapping 665 666 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesLocal() 667 @*/ 668 int MatSetLocalToGlobalMapping(Mat x,ISLocalToGlobalMapping mapping) 669 { 670 int ierr; 671 PetscFunctionBegin; 672 PetscValidHeaderSpecific(x,MAT_COOKIE); 673 PetscValidHeaderSpecific(mapping,IS_LTOGM_COOKIE); 674 if (x->mapping) { 675 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Mapping already set for matrix"); 676 } 677 678 x->mapping = mapping; 679 ierr = PetscObjectReference((PetscObject)mapping);CHKERRQ(ierr); 680 PetscFunctionReturn(0); 681 } 682 683 #undef __FUNC__ 684 #define __FUNC__ /*<a name=""></a>*/"MatSetLocalToGlobalMappingBlock" 685 /*@ 686 MatSetLocalToGlobalMappingBlock - Sets a local-to-global numbering for use 687 by the routine MatSetValuesBlockedLocal() to allow users to insert matrix 688 entries using a local (per-processor) numbering. 689 690 Not Collective 691 692 Input Parameters: 693 + x - the matrix 694 - mapping - mapping created with ISLocalToGlobalMappingCreate() or 695 ISLocalToGlobalMappingCreateIS() 696 697 Level: intermediate 698 699 .keywords: matrix, set, values, local ordering 700 701 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal(), 702 MatSetValuesBlocked(), MatSetValuesLocal() 703 @*/ 704 int MatSetLocalToGlobalMappingBlock(Mat x,ISLocalToGlobalMapping mapping) 705 { 706 int ierr; 707 PetscFunctionBegin; 708 PetscValidHeaderSpecific(x,MAT_COOKIE); 709 PetscValidHeaderSpecific(mapping,IS_LTOGM_COOKIE); 710 if (x->bmapping) { 711 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Mapping already set for matrix"); 712 } 713 714 x->bmapping = mapping; 715 ierr = PetscObjectReference((PetscObject)mapping);CHKERRQ(ierr); 716 PetscFunctionReturn(0); 717 } 718 719 #undef __FUNC__ 720 #define __FUNC__ /*<a name=""></a>*/"MatSetValuesLocal" 721 /*@ 722 MatSetValuesLocal - Inserts or adds values into certain locations of a matrix, 723 using a local ordering of the nodes. 724 725 Not Collective 726 727 Input Parameters: 728 + x - the matrix 729 . nrow, irow - number of rows and their local indices 730 . ncol, icol - number of columns and their local indices 731 . y - a logically two-dimensional array of values 732 - addv - either INSERT_VALUES or ADD_VALUES, where 733 ADD_VALUES adds values to any existing entries, and 734 INSERT_VALUES replaces existing entries with new values 735 736 Notes: 737 Before calling MatSetValuesLocal(), the user must first set the 738 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 739 740 Calls to MatSetValuesLocal() with the INSERT_VALUES and ADD_VALUES 741 options cannot be mixed without intervening calls to the assembly 742 routines. 743 744 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 745 MUST be called after all calls to MatSetValuesLocal() have been completed. 746 747 Level: intermediate 748 749 .keywords: matrix, set, values, local ordering 750 751 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetLocalToGlobalMapping() 752 @*/ 753 int MatSetValuesLocal(Mat mat,int nrow,int *irow,int ncol,int *icol,Scalar *y,InsertMode addv) 754 { 755 int ierr,irowm[2048],icolm[2048]; 756 757 PetscFunctionBegin; 758 PetscValidHeaderSpecific(mat,MAT_COOKIE); 759 PetscValidIntPointer(irow); 760 PetscValidIntPointer(icol); 761 PetscValidScalarPointer(y); 762 763 if (mat->insertmode == NOT_SET_VALUES) { 764 mat->insertmode = addv; 765 } 766 #if defined(PETSC_USE_BOPT_g) 767 else if (mat->insertmode != addv) { 768 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,1,"Cannot mix add values and insert values"); 769 } 770 if (!mat->mapping) { 771 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Local to global never set with MatSetLocalToGlobalMapping()"); 772 } 773 if (nrow > 2048 || ncol > 2048) { 774 SETERRQ2(PETSC_ERR_SUP,0,"Number column/row indices must be <= 2048: are %d %d",nrow,ncol); 775 } 776 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 777 #endif 778 779 if (mat->assembled) { 780 mat->was_assembled = PETSC_TRUE; 781 mat->assembled = PETSC_FALSE; 782 } 783 PLogEventBegin(MAT_SetValues,mat,0,0,0); 784 ierr = ISLocalToGlobalMappingApply(mat->mapping,nrow,irow,irowm);CHKERRQ(ierr); 785 ierr = ISLocalToGlobalMappingApply(mat->mapping,ncol,icol,icolm);CHKERRQ(ierr); 786 ierr = (*mat->ops->setvalues)(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr); 787 PLogEventEnd(MAT_SetValues,mat,0,0,0); 788 PetscFunctionReturn(0); 789 } 790 791 #undef __FUNC__ 792 #define __FUNC__ /*<a name=""></a>*/"MatSetValuesBlockedLocal" 793 /*@ 794 MatSetValuesBlockedLocal - Inserts or adds values into certain locations of a matrix, 795 using a local ordering of the nodes a block at a time. 796 797 Not Collective 798 799 Input Parameters: 800 + x - the matrix 801 . nrow, irow - number of rows and their local indices 802 . ncol, icol - number of columns and their local indices 803 . y - a logically two-dimensional array of values 804 - addv - either INSERT_VALUES or ADD_VALUES, where 805 ADD_VALUES adds values to any existing entries, and 806 INSERT_VALUES replaces existing entries with new values 807 808 Notes: 809 Before calling MatSetValuesBlockedLocal(), the user must first set the 810 local-to-global mapping by calling MatSetLocalToGlobalMappingBlock(), 811 where the mapping MUST be set for matrix blocks, not for matrix elements. 812 813 Calls to MatSetValuesBlockedLocal() with the INSERT_VALUES and ADD_VALUES 814 options cannot be mixed without intervening calls to the assembly 815 routines. 816 817 These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd() 818 MUST be called after all calls to MatSetValuesBlockedLocal() have been completed. 819 820 Level: intermediate 821 822 .keywords: matrix, set, values, blocked, local 823 824 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesLocal(), MatSetLocalToGlobalMappingBlock(), MatSetValuesBlocked() 825 @*/ 826 int MatSetValuesBlockedLocal(Mat mat,int nrow,int *irow,int ncol,int *icol,Scalar *y,InsertMode addv) 827 { 828 int ierr,irowm[2048],icolm[2048]; 829 830 PetscFunctionBegin; 831 PetscValidHeaderSpecific(mat,MAT_COOKIE); 832 PetscValidIntPointer(irow); 833 PetscValidIntPointer(icol); 834 PetscValidScalarPointer(y); 835 if (mat->insertmode == NOT_SET_VALUES) { 836 mat->insertmode = addv; 837 } 838 #if defined(PETSC_USE_BOPT_g) 839 else if (mat->insertmode != addv) { 840 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,1,"Cannot mix add values and insert values"); 841 } 842 if (!mat->bmapping) { 843 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Local to global never set with MatSetLocalToGlobalMappingBlock()"); 844 } 845 if (nrow > 2048 || ncol > 2048) { 846 SETERRQ2(PETSC_ERR_SUP,0,"Number column/row indices must be <= 2048: are %d %d",nrow,ncol); 847 } 848 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 849 #endif 850 851 if (mat->assembled) { 852 mat->was_assembled = PETSC_TRUE; 853 mat->assembled = PETSC_FALSE; 854 } 855 PLogEventBegin(MAT_SetValues,mat,0,0,0); 856 ierr = ISLocalToGlobalMappingApply(mat->bmapping,nrow,irow,irowm);CHKERRQ(ierr); 857 ierr = ISLocalToGlobalMappingApply(mat->bmapping,ncol,icol,icolm);CHKERRQ(ierr); 858 ierr = (*mat->ops->setvaluesblocked)(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr); 859 PLogEventEnd(MAT_SetValues,mat,0,0,0); 860 PetscFunctionReturn(0); 861 } 862 863 /* --------------------------------------------------------*/ 864 #undef __FUNC__ 865 #define __FUNC__ /*<a name=""></a>*/"MatMult" 866 /*@ 867 MatMult - Computes the matrix-vector product, y = Ax. 868 869 Collective on Mat and Vec 870 871 Input Parameters: 872 + mat - the matrix 873 - x - the vector to be multilplied 874 875 Output Parameters: 876 . y - the result 877 878 Notes: 879 The vectors x and y cannot be the same. I.e., one cannot 880 call MatMult(A,y,y). 881 882 Level: beginner 883 884 .keywords: matrix, multiply, matrix-vector product 885 886 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd() 887 @*/ 888 int MatMult(Mat mat,Vec x,Vec y) 889 { 890 int ierr; 891 892 PetscFunctionBegin; 893 PetscValidHeaderSpecific(mat,MAT_COOKIE); 894 PetscValidHeaderSpecific(x,VEC_COOKIE); 895 PetscValidHeaderSpecific(y,VEC_COOKIE); 896 PetscCheckSameComm(mat,x); 897 PetscCheckSameComm(mat,y); 898 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 899 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 900 if (x == y) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"x and y must be different vectors"); 901 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 902 if (mat->M != y->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec y: global dim %d %d",mat->M,y->N); 903 if (mat->m != y->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec y: local dim %d %d",mat->m,y->n); 904 905 PLogEventBegin(MAT_Mult,mat,x,y,0); 906 ierr = (*mat->ops->mult)(mat,x,y);CHKERRQ(ierr); 907 PLogEventEnd(MAT_Mult,mat,x,y,0); 908 909 PetscFunctionReturn(0); 910 } 911 912 #undef __FUNC__ 913 #define __FUNC__ /*<a name=""></a>*/"MatMultTranspose" 914 /*@ 915 MatMultTranspose - Computes matrix transpose times a vector. 916 917 Collective on Mat and Vec 918 919 Input Parameters: 920 + mat - the matrix 921 - x - the vector to be multilplied 922 923 Output Parameters: 924 . y - the result 925 926 Notes: 927 The vectors x and y cannot be the same. I.e., one cannot 928 call MatMultTranspose(A,y,y). 929 930 Level: beginner 931 932 .keywords: matrix, multiply, matrix-vector product, transpose 933 934 .seealso: MatMult(), MatMultAdd(), MatMultTransposeAdd() 935 @*/ 936 int MatMultTranspose(Mat mat,Vec x,Vec y) 937 { 938 int ierr; 939 940 PetscFunctionBegin; 941 PetscValidHeaderSpecific(mat,MAT_COOKIE); 942 PetscValidHeaderSpecific(x,VEC_COOKIE); 943 PetscValidHeaderSpecific(y,VEC_COOKIE); 944 PetscCheckSameComm(mat,x); 945 PetscCheckSameComm(mat,y); 946 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 947 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 948 if (x == y) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"x and y must be different vectors"); 949 if (mat->M != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->M,x->N); 950 if (mat->N != y->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec y: global dim %d %d",mat->N,y->N); 951 952 PLogEventBegin(MAT_MultTranspose,mat,x,y,0); 953 ierr = (*mat->ops->multtranspose)(mat,x,y);CHKERRQ(ierr); 954 PLogEventEnd(MAT_MultTranspose,mat,x,y,0); 955 PetscFunctionReturn(0); 956 } 957 958 #undef __FUNC__ 959 #define __FUNC__ /*<a name=""></a>*/"MatMultAdd" 960 /*@ 961 MatMultAdd - Computes v3 = v2 + A * v1. 962 963 Collective on Mat and Vec 964 965 Input Parameters: 966 + mat - the matrix 967 - v1, v2 - the vectors 968 969 Output Parameters: 970 . v3 - the result 971 972 Notes: 973 The vectors v1 and v3 cannot be the same. I.e., one cannot 974 call MatMultAdd(A,v1,v2,v1). 975 976 Level: beginner 977 978 .keywords: matrix, multiply, matrix-vector product, add 979 980 .seealso: MatMultTranspose(), MatMult(), MatMultTransposeAdd() 981 @*/ 982 int MatMultAdd(Mat mat,Vec v1,Vec v2,Vec v3) 983 { 984 int ierr; 985 986 PetscFunctionBegin; 987 PetscValidHeaderSpecific(mat,MAT_COOKIE); 988 PetscValidHeaderSpecific(v1,VEC_COOKIE); 989 PetscValidHeaderSpecific(v2,VEC_COOKIE); 990 PetscValidHeaderSpecific(v3,VEC_COOKIE); 991 PetscCheckSameComm(mat,v1); 992 PetscCheckSameComm(mat,v2); 993 PetscCheckSameComm(mat,v3); 994 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 995 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 996 if (mat->N != v1->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v1: global dim %d %d",mat->N,v1->N); 997 if (mat->M != v2->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v2: global dim %d %d",mat->M,v2->N); 998 if (mat->M != v3->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v3: global dim %d %d",mat->M,v3->N); 999 if (mat->m != v3->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v3: local dim %d %d",mat->m,v3->n); 1000 if (mat->m != v2->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v2: local dim %d %d",mat->m,v2->n); 1001 if (v1 == v3) SETERRQ(PETSC_ERR_ARG_IDN,0,"v1 and v3 must be different vectors"); 1002 1003 PLogEventBegin(MAT_MultAdd,mat,v1,v2,v3); 1004 ierr = (*mat->ops->multadd)(mat,v1,v2,v3);CHKERRQ(ierr); 1005 PLogEventEnd(MAT_MultAdd,mat,v1,v2,v3); 1006 PetscFunctionReturn(0); 1007 } 1008 1009 #undef __FUNC__ 1010 #define __FUNC__ /*<a name=""></a>*/"MatMultTransposeAdd" 1011 /*@ 1012 MatMultTransposeAdd - Computes v3 = v2 + A' * v1. 1013 1014 Collective on Mat and Vec 1015 1016 Input Parameters: 1017 + mat - the matrix 1018 - v1, v2 - the vectors 1019 1020 Output Parameters: 1021 . v3 - the result 1022 1023 Notes: 1024 The vectors v1 and v3 cannot be the same. I.e., one cannot 1025 call MatMultTransposeAdd(A,v1,v2,v1). 1026 1027 Level: beginner 1028 1029 .keywords: matrix, multiply, matrix-vector product, transpose, add 1030 1031 .seealso: MatMultTranspose(), MatMultAdd(), MatMult() 1032 @*/ 1033 int MatMultTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3) 1034 { 1035 int ierr; 1036 1037 PetscFunctionBegin; 1038 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1039 PetscValidHeaderSpecific(v1,VEC_COOKIE); 1040 PetscValidHeaderSpecific(v2,VEC_COOKIE); 1041 PetscValidHeaderSpecific(v3,VEC_COOKIE); 1042 PetscCheckSameComm(mat,v1); 1043 PetscCheckSameComm(mat,v2); 1044 PetscCheckSameComm(mat,v3); 1045 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1046 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1047 if (!mat->ops->multtransposeadd) SETERRQ(PETSC_ERR_SUP,0,""); 1048 if (v1 == v3) SETERRQ(PETSC_ERR_ARG_IDN,0,"v1 and v3 must be different vectors"); 1049 if (mat->M != v1->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v1: global dim %d %d",mat->M,v1->N); 1050 if (mat->N != v2->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v2: global dim %d %d",mat->N,v2->N); 1051 if (mat->N != v3->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec v3: global dim %d %d",mat->N,v3->N); 1052 1053 PLogEventBegin(MAT_MultTransposeAdd,mat,v1,v2,v3); 1054 ierr = (*mat->ops->multtransposeadd)(mat,v1,v2,v3);CHKERRQ(ierr); 1055 PLogEventEnd(MAT_MultTransposeAdd,mat,v1,v2,v3); 1056 PetscFunctionReturn(0); 1057 } 1058 /* ------------------------------------------------------------*/ 1059 #undef __FUNC__ 1060 #define __FUNC__ /*<a name=""></a>*/"MatGetInfo" 1061 /*@C 1062 MatGetInfo - Returns information about matrix storage (number of 1063 nonzeros, memory, etc.). 1064 1065 Collective on Mat if MAT_GLOBAL_MAX or MAT_GLOBAL_SUM is used 1066 as the flag 1067 1068 Input Parameters: 1069 . mat - the matrix 1070 1071 Output Parameters: 1072 + flag - flag indicating the type of parameters to be returned 1073 (MAT_LOCAL - local matrix, MAT_GLOBAL_MAX - maximum over all processors, 1074 MAT_GLOBAL_SUM - sum over all processors) 1075 - info - matrix information context 1076 1077 Notes: 1078 The MatInfo context contains a variety of matrix data, including 1079 number of nonzeros allocated and used, number of mallocs during 1080 matrix assembly, etc. Additional information for factored matrices 1081 is provided (such as the fill ratio, number of mallocs during 1082 factorization, etc.). Much of this info is printed to STDOUT 1083 when using the runtime options 1084 $ -log_info -mat_view_info 1085 1086 Example for C/C++ Users: 1087 See the file ${PETSC_DIR}/include/petscmat.h for a complete list of 1088 data within the MatInfo context. For example, 1089 .vb 1090 MatInfo info; 1091 Mat A; 1092 double mal, nz_a, nz_u; 1093 1094 MatGetInfo(A,MAT_LOCAL,&info); 1095 mal = info.mallocs; 1096 nz_a = info.nz_allocated; 1097 .ve 1098 1099 Example for Fortran Users: 1100 Fortran users should declare info as a double precision 1101 array of dimension MAT_INFO_SIZE, and then extract the parameters 1102 of interest. See the file ${PETSC_DIR}/include/finclude/petscmat.h 1103 a complete list of parameter names. 1104 .vb 1105 double precision info(MAT_INFO_SIZE) 1106 double precision mal, nz_a 1107 Mat A 1108 integer ierr 1109 1110 call MatGetInfo(A,MAT_LOCAL,info,ierr) 1111 mal = info(MAT_INFO_MALLOCS) 1112 nz_a = info(MAT_INFO_NZ_ALLOCATED) 1113 .ve 1114 1115 Level: intermediate 1116 1117 .keywords: matrix, get, info, storage, nonzeros, memory, fill 1118 @*/ 1119 int MatGetInfo(Mat mat,MatInfoType flag,MatInfo *info) 1120 { 1121 int ierr; 1122 1123 PetscFunctionBegin; 1124 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1125 PetscValidPointer(info); 1126 if (!mat->ops->getinfo) SETERRQ(PETSC_ERR_SUP,0,""); 1127 ierr = (*mat->ops->getinfo)(mat,flag,info);CHKERRQ(ierr); 1128 PetscFunctionReturn(0); 1129 } 1130 1131 /* ----------------------------------------------------------*/ 1132 #undef __FUNC__ 1133 #define __FUNC__ /*<a name=""></a>*/"MatILUDTFactor" 1134 /*@C 1135 MatILUDTFactor - Performs a drop tolerance ILU factorization. 1136 1137 Collective on Mat 1138 1139 Input Parameters: 1140 + mat - the matrix 1141 . info - information about the factorization to be done 1142 . row - row permutation 1143 - col - column permutation 1144 1145 Output Parameters: 1146 . fact - the factored matrix 1147 1148 Level: developer 1149 1150 Notes: 1151 Most users should employ the simplified SLES interface for linear solvers 1152 instead of working directly with matrix algebra routines such as this. 1153 See, e.g., SLESCreate(). 1154 1155 This is currently only supported for the SeqAIJ matrix format using code 1156 from Yousef Saad's SPARSEKIT2 package. That code is copyright by Yousef 1157 Saad with the GNU copyright. 1158 1159 .keywords: matrix, factor, LU, in-place 1160 1161 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor() 1162 @*/ 1163 int MatILUDTFactor(Mat mat,MatILUInfo *info,IS row,IS col,Mat *fact) 1164 { 1165 int ierr; 1166 1167 PetscFunctionBegin; 1168 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1169 PetscValidPointer(fact); 1170 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1171 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1172 if (!mat->ops->iludtfactor) SETERRQ(PETSC_ERR_SUP,0,""); 1173 1174 PLogEventBegin(MAT_ILUFactor,mat,row,col,0); 1175 ierr = (*mat->ops->iludtfactor)(mat,info,row,col,fact);CHKERRQ(ierr); 1176 PLogEventEnd(MAT_ILUFactor,mat,row,col,0); 1177 1178 PetscFunctionReturn(0); 1179 } 1180 1181 #undef __FUNC__ 1182 #define __FUNC__ /*<a name=""></a>*/"MatLUFactor" 1183 /*@ 1184 MatLUFactor - Performs in-place LU factorization of matrix. 1185 1186 Collective on Mat 1187 1188 Input Parameters: 1189 + mat - the matrix 1190 . row - row permutation 1191 . col - column permutation 1192 - info - options for factorization, includes 1193 $ fill - expected fill as ratio of original fill. 1194 $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting) 1195 $ Run with the option -log_info to determine an optimal value to use 1196 1197 Notes: 1198 Most users should employ the simplified SLES interface for linear solvers 1199 instead of working directly with matrix algebra routines such as this. 1200 See, e.g., SLESCreate(). 1201 1202 This changes the state of the matrix to a factored matrix; it cannot be used 1203 for example with MatSetValues() unless one first calls MatSetUnfactored(). 1204 1205 Level: developer 1206 1207 .keywords: matrix, factor, LU, in-place 1208 1209 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(), 1210 MatGetOrdering(), MatSetUnfactored() 1211 1212 @*/ 1213 int MatLUFactor(Mat mat,IS row,IS col,MatLUInfo *info) 1214 { 1215 int ierr; 1216 1217 PetscFunctionBegin; 1218 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1219 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1220 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1221 if (!mat->ops->lufactor) SETERRQ(PETSC_ERR_SUP,0,""); 1222 1223 PLogEventBegin(MAT_LUFactor,mat,row,col,0); 1224 ierr = (*mat->ops->lufactor)(mat,row,col,info);CHKERRQ(ierr); 1225 PLogEventEnd(MAT_LUFactor,mat,row,col,0); 1226 PetscFunctionReturn(0); 1227 } 1228 1229 #undef __FUNC__ 1230 #define __FUNC__ /*<a name=""></a>*/"MatILUFactor" 1231 /*@ 1232 MatILUFactor - Performs in-place ILU factorization of matrix. 1233 1234 Collective on Mat 1235 1236 Input Parameters: 1237 + mat - the matrix 1238 . row - row permutation 1239 . col - column permutation 1240 - info - structure containing 1241 $ levels - number of levels of fill. 1242 $ expected fill - as ratio of original fill. 1243 $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices 1244 missing diagonal entries) 1245 1246 Notes: 1247 Probably really in-place only when level of fill is zero, otherwise allocates 1248 new space to store factored matrix and deletes previous memory. 1249 1250 Most users should employ the simplified SLES interface for linear solvers 1251 instead of working directly with matrix algebra routines such as this. 1252 See, e.g., SLESCreate(). 1253 1254 Level: developer 1255 1256 .keywords: matrix, factor, ILU, in-place 1257 1258 .seealso: MatILUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor() 1259 @*/ 1260 int MatILUFactor(Mat mat,IS row,IS col,MatILUInfo *info) 1261 { 1262 int ierr; 1263 1264 PetscFunctionBegin; 1265 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1266 if (mat->M != mat->N) SETERRQ(PETSC_ERR_ARG_WRONG,0,"matrix must be square"); 1267 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1268 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1269 if (!mat->ops->ilufactor) SETERRQ(PETSC_ERR_SUP,0,""); 1270 1271 PLogEventBegin(MAT_ILUFactor,mat,row,col,0); 1272 ierr = (*mat->ops->ilufactor)(mat,row,col,info);CHKERRQ(ierr); 1273 PLogEventEnd(MAT_ILUFactor,mat,row,col,0); 1274 PetscFunctionReturn(0); 1275 } 1276 1277 #undef __FUNC__ 1278 #define __FUNC__ /*<a name=""></a>*/"MatLUFactorSymbolic" 1279 /*@ 1280 MatLUFactorSymbolic - Performs symbolic LU factorization of matrix. 1281 Call this routine before calling MatLUFactorNumeric(). 1282 1283 Collective on Mat 1284 1285 Input Parameters: 1286 + mat - the matrix 1287 . row, col - row and column permutations 1288 - info - options for factorization, includes 1289 $ fill - expected fill as ratio of original fill. 1290 $ dtcol - pivot tolerance (0 no pivot, 1 full column pivoting) 1291 $ Run with the option -log_info to determine an optimal value to use 1292 1293 Output Parameter: 1294 . fact - new matrix that has been symbolically factored 1295 1296 Notes: 1297 See the users manual for additional information about 1298 choosing the fill factor for better efficiency. 1299 1300 Most users should employ the simplified SLES interface for linear solvers 1301 instead of working directly with matrix algebra routines such as this. 1302 See, e.g., SLESCreate(). 1303 1304 Level: developer 1305 1306 .keywords: matrix, factor, LU, symbolic, fill 1307 1308 .seealso: MatLUFactor(), MatLUFactorNumeric(), MatCholeskyFactor() 1309 @*/ 1310 int MatLUFactorSymbolic(Mat mat,IS row,IS col,MatLUInfo *info,Mat *fact) 1311 { 1312 int ierr; 1313 1314 PetscFunctionBegin; 1315 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1316 PetscValidPointer(fact); 1317 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1318 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1319 if (!mat->ops->lufactorsymbolic) SETERRQ(PETSC_ERR_SUP,0,""); 1320 1321 PLogEventBegin(MAT_LUFactorSymbolic,mat,row,col,0); 1322 ierr = (*mat->ops->lufactorsymbolic)(mat,row,col,info,fact);CHKERRQ(ierr); 1323 PLogEventEnd(MAT_LUFactorSymbolic,mat,row,col,0); 1324 PetscFunctionReturn(0); 1325 } 1326 1327 #undef __FUNC__ 1328 #define __FUNC__ /*<a name=""></a>*/"MatLUFactorNumeric" 1329 /*@ 1330 MatLUFactorNumeric - Performs numeric LU factorization of a matrix. 1331 Call this routine after first calling MatLUFactorSymbolic(). 1332 1333 Collective on Mat 1334 1335 Input Parameters: 1336 + mat - the matrix 1337 - fact - the matrix generated for the factor, from MatLUFactorSymbolic() 1338 1339 Notes: 1340 See MatLUFactor() for in-place factorization. See 1341 MatCholeskyFactorNumeric() for the symmetric, positive definite case. 1342 1343 Most users should employ the simplified SLES interface for linear solvers 1344 instead of working directly with matrix algebra routines such as this. 1345 See, e.g., SLESCreate(). 1346 1347 Level: developer 1348 1349 .keywords: matrix, factor, LU, numeric 1350 1351 .seealso: MatLUFactorSymbolic(), MatLUFactor(), MatCholeskyFactor() 1352 @*/ 1353 int MatLUFactorNumeric(Mat mat,Mat *fact) 1354 { 1355 int ierr; 1356 PetscTruth flg; 1357 1358 PetscFunctionBegin; 1359 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1360 PetscValidPointer(fact); 1361 PetscValidHeaderSpecific(*fact,MAT_COOKIE); 1362 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1363 if (mat->M != (*fact)->M || mat->N != (*fact)->N) { 1364 SETERRQ4(PETSC_ERR_ARG_SIZ,0,"Mat mat,Mat *fact: global dimensions are different %d should = %d %d should = %d", 1365 mat->M,(*fact)->M,mat->N,(*fact)->N); 1366 } 1367 if (!(*fact)->ops->lufactornumeric) SETERRQ(PETSC_ERR_SUP,0,""); 1368 1369 PLogEventBegin(MAT_LUFactorNumeric,mat,*fact,0,0); 1370 ierr = (*(*fact)->ops->lufactornumeric)(mat,fact);CHKERRQ(ierr); 1371 PLogEventEnd(MAT_LUFactorNumeric,mat,*fact,0,0); 1372 ierr = OptionsHasName(PETSC_NULL,"-mat_view_draw",&flg);CHKERRQ(ierr); 1373 if (flg) { 1374 ierr = OptionsHasName(PETSC_NULL,"-mat_view_contour",&flg);CHKERRQ(ierr); 1375 if (flg) { 1376 ierr = ViewerPushFormat(VIEWER_DRAW_(mat->comm),VIEWER_FORMAT_DRAW_CONTOUR,0);CHKERRQ(ierr); 1377 } 1378 ierr = MatView(*fact,VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 1379 ierr = ViewerFlush(VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 1380 if (flg) { 1381 ierr = ViewerPopFormat(VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 1382 } 1383 } 1384 PetscFunctionReturn(0); 1385 } 1386 1387 #undef __FUNC__ 1388 #define __FUNC__ /*<a name=""></a>*/"MatCholeskyFactor" 1389 /*@ 1390 MatCholeskyFactor - Performs in-place Cholesky factorization of a 1391 symmetric matrix. 1392 1393 Collective on Mat 1394 1395 Input Parameters: 1396 + mat - the matrix 1397 . perm - row and column permutations 1398 - f - expected fill as ratio of original fill 1399 1400 Notes: 1401 See MatLUFactor() for the nonsymmetric case. See also 1402 MatCholeskyFactorSymbolic(), and MatCholeskyFactorNumeric(). 1403 1404 Most users should employ the simplified SLES interface for linear solvers 1405 instead of working directly with matrix algebra routines such as this. 1406 See, e.g., SLESCreate(). 1407 1408 Level: developer 1409 1410 .keywords: matrix, factor, in-place, Cholesky 1411 1412 .seealso: MatLUFactor(), MatCholeskyFactorSymbolic(), MatCholeskyFactorNumeric() 1413 MatGetOrdering() 1414 1415 @*/ 1416 int MatCholeskyFactor(Mat mat,IS perm,PetscReal f) 1417 { 1418 int ierr; 1419 1420 PetscFunctionBegin; 1421 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1422 if (mat->M != mat->N) SETERRQ(PETSC_ERR_ARG_WRONG,0,"Matrix must be square"); 1423 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1424 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1425 if (!mat->ops->choleskyfactor) SETERRQ(PETSC_ERR_SUP,0,""); 1426 1427 PLogEventBegin(MAT_CholeskyFactor,mat,perm,0,0); 1428 ierr = (*mat->ops->choleskyfactor)(mat,perm,f);CHKERRQ(ierr); 1429 PLogEventEnd(MAT_CholeskyFactor,mat,perm,0,0); 1430 PetscFunctionReturn(0); 1431 } 1432 1433 #undef __FUNC__ 1434 #define __FUNC__ /*<a name=""></a>*/"MatCholeskyFactorSymbolic" 1435 /*@ 1436 MatCholeskyFactorSymbolic - Performs symbolic Cholesky factorization 1437 of a symmetric matrix. 1438 1439 Collective on Mat 1440 1441 Input Parameters: 1442 + mat - the matrix 1443 . perm - row and column permutations 1444 - f - expected fill as ratio of original 1445 1446 Output Parameter: 1447 . fact - the factored matrix 1448 1449 Notes: 1450 See MatLUFactorSymbolic() for the nonsymmetric case. See also 1451 MatCholeskyFactor() and MatCholeskyFactorNumeric(). 1452 1453 Most users should employ the simplified SLES interface for linear solvers 1454 instead of working directly with matrix algebra routines such as this. 1455 See, e.g., SLESCreate(). 1456 1457 Level: developer 1458 1459 .keywords: matrix, factor, factorization, symbolic, Cholesky 1460 1461 .seealso: MatLUFactorSymbolic(), MatCholeskyFactor(), MatCholeskyFactorNumeric() 1462 MatGetOrdering() 1463 1464 @*/ 1465 int MatCholeskyFactorSymbolic(Mat mat,IS perm,PetscReal f,Mat *fact) 1466 { 1467 int ierr; 1468 1469 PetscFunctionBegin; 1470 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1471 PetscValidPointer(fact); 1472 if (mat->M != mat->N) SETERRQ(PETSC_ERR_ARG_WRONG,0,"Matrix must be square"); 1473 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1474 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1475 if (!mat->ops->choleskyfactorsymbolic) SETERRQ(PETSC_ERR_SUP,0,""); 1476 1477 PLogEventBegin(MAT_CholeskyFactorSymbolic,mat,perm,0,0); 1478 ierr = (*mat->ops->choleskyfactorsymbolic)(mat,perm,f,fact);CHKERRQ(ierr); 1479 PLogEventEnd(MAT_CholeskyFactorSymbolic,mat,perm,0,0); 1480 PetscFunctionReturn(0); 1481 } 1482 1483 #undef __FUNC__ 1484 #define __FUNC__ /*<a name=""></a>*/"MatCholeskyFactorNumeric" 1485 /*@ 1486 MatCholeskyFactorNumeric - Performs numeric Cholesky factorization 1487 of a symmetric matrix. Call this routine after first calling 1488 MatCholeskyFactorSymbolic(). 1489 1490 Collective on Mat 1491 1492 Input Parameter: 1493 . mat - the initial matrix 1494 1495 Output Parameter: 1496 . fact - the factored matrix 1497 1498 Notes: 1499 Most users should employ the simplified SLES interface for linear solvers 1500 instead of working directly with matrix algebra routines such as this. 1501 See, e.g., SLESCreate(). 1502 1503 Level: developer 1504 1505 .keywords: matrix, factor, numeric, Cholesky 1506 1507 .seealso: MatCholeskyFactorSymbolic(), MatCholeskyFactor(), MatLUFactorNumeric() 1508 @*/ 1509 int MatCholeskyFactorNumeric(Mat mat,Mat *fact) 1510 { 1511 int ierr; 1512 1513 PetscFunctionBegin; 1514 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1515 PetscValidPointer(fact); 1516 if (!mat->ops->choleskyfactornumeric) SETERRQ(PETSC_ERR_SUP,0,""); 1517 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1518 if (mat->M != (*fact)->M || mat->N != (*fact)->N) { 1519 SETERRQ4(PETSC_ERR_ARG_SIZ,0,"Mat mat,Mat *fact: global dim %d should = %d %d should = %d", 1520 mat->M,(*fact)->M,mat->N,(*fact)->N); 1521 } 1522 1523 PLogEventBegin(MAT_CholeskyFactorNumeric,mat,*fact,0,0); 1524 ierr = (*mat->ops->choleskyfactornumeric)(mat,fact);CHKERRQ(ierr); 1525 PLogEventEnd(MAT_CholeskyFactorNumeric,mat,*fact,0,0); 1526 PetscFunctionReturn(0); 1527 } 1528 1529 /* ----------------------------------------------------------------*/ 1530 #undef __FUNC__ 1531 #define __FUNC__ /*<a name=""></a>*/"MatSolve" 1532 /*@ 1533 MatSolve - Solves A x = b, given a factored matrix. 1534 1535 Collective on Mat and Vec 1536 1537 Input Parameters: 1538 + mat - the factored matrix 1539 - b - the right-hand-side vector 1540 1541 Output Parameter: 1542 . x - the result vector 1543 1544 Notes: 1545 The vectors b and x cannot be the same. I.e., one cannot 1546 call MatSolve(A,x,x). 1547 1548 Notes: 1549 Most users should employ the simplified SLES interface for linear solvers 1550 instead of working directly with matrix algebra routines such as this. 1551 See, e.g., SLESCreate(). 1552 1553 Level: developer 1554 1555 .keywords: matrix, linear system, solve, LU, Cholesky, triangular solve 1556 1557 .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd() 1558 @*/ 1559 int MatSolve(Mat mat,Vec b,Vec x) 1560 { 1561 int ierr; 1562 1563 PetscFunctionBegin; 1564 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1565 PetscValidHeaderSpecific(b,VEC_COOKIE); 1566 PetscValidHeaderSpecific(x,VEC_COOKIE); 1567 PetscCheckSameComm(mat,b); 1568 PetscCheckSameComm(mat,x); 1569 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1570 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1571 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 1572 if (mat->M != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->M,b->N); 1573 if (mat->m != b->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: local dim %d %d",mat->m,b->n); 1574 if (mat->M == 0 && mat->N == 0) PetscFunctionReturn(0); 1575 1576 if (!mat->ops->solve) SETERRQ(PETSC_ERR_SUP,0,""); 1577 PLogEventBegin(MAT_Solve,mat,b,x,0); 1578 ierr = (*mat->ops->solve)(mat,b,x);CHKERRQ(ierr); 1579 PLogEventEnd(MAT_Solve,mat,b,x,0); 1580 PetscFunctionReturn(0); 1581 } 1582 1583 #undef __FUNC__ 1584 #define __FUNC__ /*<a name=""></a>*/"MatForwardSolve" 1585 /* @ 1586 MatForwardSolve - Solves L x = b, given a factored matrix, A = LU. 1587 1588 Collective on Mat and Vec 1589 1590 Input Parameters: 1591 + mat - the factored matrix 1592 - b - the right-hand-side vector 1593 1594 Output Parameter: 1595 . x - the result vector 1596 1597 Notes: 1598 MatSolve() should be used for most applications, as it performs 1599 a forward solve followed by a backward solve. 1600 1601 The vectors b and x cannot be the same. I.e., one cannot 1602 call MatForwardSolve(A,x,x). 1603 1604 Most users should employ the simplified SLES interface for linear solvers 1605 instead of working directly with matrix algebra routines such as this. 1606 See, e.g., SLESCreate(). 1607 1608 Level: developer 1609 1610 .keywords: matrix, forward, LU, Cholesky, triangular solve 1611 1612 .seealso: MatSolve(), MatBackwardSolve() 1613 @ */ 1614 int MatForwardSolve(Mat mat,Vec b,Vec x) 1615 { 1616 int ierr; 1617 1618 PetscFunctionBegin; 1619 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1620 PetscValidHeaderSpecific(b,VEC_COOKIE); 1621 PetscValidHeaderSpecific(x,VEC_COOKIE); 1622 PetscCheckSameComm(mat,b); 1623 PetscCheckSameComm(mat,x); 1624 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1625 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1626 if (!mat->ops->forwardsolve) SETERRQ(PETSC_ERR_SUP,0,""); 1627 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 1628 if (mat->M != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->M,b->N); 1629 if (mat->m != b->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: local dim %d %d",mat->m,b->n); 1630 1631 PLogEventBegin(MAT_ForwardSolve,mat,b,x,0); 1632 ierr = (*mat->ops->forwardsolve)(mat,b,x);CHKERRQ(ierr); 1633 PLogEventEnd(MAT_ForwardSolve,mat,b,x,0); 1634 PetscFunctionReturn(0); 1635 } 1636 1637 #undef __FUNC__ 1638 #define __FUNC__ /*<a name=""></a>*/"MatBackwardSolve" 1639 /* @ 1640 MatBackwardSolve - Solves U x = b, given a factored matrix, A = LU. 1641 1642 Collective on Mat and Vec 1643 1644 Input Parameters: 1645 + mat - the factored matrix 1646 - b - the right-hand-side vector 1647 1648 Output Parameter: 1649 . x - the result vector 1650 1651 Notes: 1652 MatSolve() should be used for most applications, as it performs 1653 a forward solve followed by a backward solve. 1654 1655 The vectors b and x cannot be the same. I.e., one cannot 1656 call MatBackwardSolve(A,x,x). 1657 1658 Most users should employ the simplified SLES interface for linear solvers 1659 instead of working directly with matrix algebra routines such as this. 1660 See, e.g., SLESCreate(). 1661 1662 Level: developer 1663 1664 .keywords: matrix, backward, LU, Cholesky, triangular solve 1665 1666 .seealso: MatSolve(), MatForwardSolve() 1667 @ */ 1668 int MatBackwardSolve(Mat mat,Vec b,Vec x) 1669 { 1670 int ierr; 1671 1672 PetscFunctionBegin; 1673 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1674 PetscValidHeaderSpecific(b,VEC_COOKIE); 1675 PetscValidHeaderSpecific(x,VEC_COOKIE); 1676 PetscCheckSameComm(mat,b); 1677 PetscCheckSameComm(mat,x); 1678 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1679 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1680 if (!mat->ops->backwardsolve) SETERRQ(PETSC_ERR_SUP,0,""); 1681 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 1682 if (mat->M != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->M,b->N); 1683 if (mat->m != b->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: local dim %d %d",mat->m,b->n); 1684 1685 PLogEventBegin(MAT_BackwardSolve,mat,b,x,0); 1686 ierr = (*mat->ops->backwardsolve)(mat,b,x);CHKERRQ(ierr); 1687 PLogEventEnd(MAT_BackwardSolve,mat,b,x,0); 1688 PetscFunctionReturn(0); 1689 } 1690 1691 #undef __FUNC__ 1692 #define __FUNC__ /*<a name=""></a>*/"MatSolveAdd" 1693 /*@ 1694 MatSolveAdd - Computes x = y + inv(A)*b, given a factored matrix. 1695 1696 Collective on Mat and Vec 1697 1698 Input Parameters: 1699 + mat - the factored matrix 1700 . b - the right-hand-side vector 1701 - y - the vector to be added to 1702 1703 Output Parameter: 1704 . x - the result vector 1705 1706 Notes: 1707 The vectors b and x cannot be the same. I.e., one cannot 1708 call MatSolveAdd(A,x,y,x). 1709 1710 Most users should employ the simplified SLES interface for linear solvers 1711 instead of working directly with matrix algebra routines such as this. 1712 See, e.g., SLESCreate(). 1713 1714 Level: developer 1715 1716 .keywords: matrix, linear system, solve, LU, Cholesky, add 1717 1718 .seealso: MatSolve(), MatSolveTranspose(), MatSolveTransposeAdd() 1719 @*/ 1720 int MatSolveAdd(Mat mat,Vec b,Vec y,Vec x) 1721 { 1722 Scalar one = 1.0; 1723 Vec tmp; 1724 int ierr; 1725 1726 PetscFunctionBegin; 1727 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1728 PetscValidHeaderSpecific(y,VEC_COOKIE); 1729 PetscValidHeaderSpecific(b,VEC_COOKIE); 1730 PetscValidHeaderSpecific(x,VEC_COOKIE); 1731 PetscCheckSameComm(mat,b); 1732 PetscCheckSameComm(mat,y); 1733 PetscCheckSameComm(mat,x); 1734 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1735 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1736 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 1737 if (mat->M != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->M,b->N); 1738 if (mat->M != y->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec y: global dim %d %d",mat->M,y->N); 1739 if (mat->m != b->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: local dim %d %d",mat->m,b->n); 1740 if (x->n != y->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Vec x,Vec y: local dim %d %d",x->n,y->n); 1741 1742 PLogEventBegin(MAT_SolveAdd,mat,b,x,y); 1743 if (mat->ops->solveadd) { 1744 ierr = (*mat->ops->solveadd)(mat,b,y,x);CHKERRQ(ierr); 1745 } else { 1746 /* do the solve then the add manually */ 1747 if (x != y) { 1748 ierr = MatSolve(mat,b,x);CHKERRQ(ierr); 1749 ierr = VecAXPY(&one,y,x);CHKERRQ(ierr); 1750 } else { 1751 ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr); 1752 PLogObjectParent(mat,tmp); 1753 ierr = VecCopy(x,tmp);CHKERRQ(ierr); 1754 ierr = MatSolve(mat,b,x);CHKERRQ(ierr); 1755 ierr = VecAXPY(&one,tmp,x);CHKERRQ(ierr); 1756 ierr = VecDestroy(tmp);CHKERRQ(ierr); 1757 } 1758 } 1759 PLogEventEnd(MAT_SolveAdd,mat,b,x,y); 1760 PetscFunctionReturn(0); 1761 } 1762 1763 #undef __FUNC__ 1764 #define __FUNC__ /*<a name=""></a>*/"MatSolveTranspose" 1765 /*@ 1766 MatSolveTranspose - Solves A' x = b, given a factored matrix. 1767 1768 Collective on Mat and Vec 1769 1770 Input Parameters: 1771 + mat - the factored matrix 1772 - b - the right-hand-side vector 1773 1774 Output Parameter: 1775 . x - the result vector 1776 1777 Notes: 1778 The vectors b and x cannot be the same. I.e., one cannot 1779 call MatSolveTranspose(A,x,x). 1780 1781 Most users should employ the simplified SLES interface for linear solvers 1782 instead of working directly with matrix algebra routines such as this. 1783 See, e.g., SLESCreate(). 1784 1785 Level: developer 1786 1787 .keywords: matrix, linear system, solve, LU, Cholesky, transpose 1788 1789 .seealso: MatSolve(), MatSolveAdd(), MatSolveTransposeAdd() 1790 @*/ 1791 int MatSolveTranspose(Mat mat,Vec b,Vec x) 1792 { 1793 int ierr; 1794 1795 PetscFunctionBegin; 1796 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1797 PetscValidHeaderSpecific(b,VEC_COOKIE); 1798 PetscValidHeaderSpecific(x,VEC_COOKIE); 1799 PetscCheckSameComm(mat,b); 1800 PetscCheckSameComm(mat,x); 1801 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1802 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1803 if (!mat->ops->solvetranspose) SETERRQ(PETSC_ERR_SUP,0,""); 1804 if (mat->M != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->M,x->N); 1805 if (mat->N != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->N,b->N); 1806 1807 PLogEventBegin(MAT_SolveTranspose,mat,b,x,0); 1808 ierr = (*mat->ops->solvetranspose)(mat,b,x);CHKERRQ(ierr); 1809 PLogEventEnd(MAT_SolveTranspose,mat,b,x,0); 1810 PetscFunctionReturn(0); 1811 } 1812 1813 #undef __FUNC__ 1814 #define __FUNC__ /*<a name=""></a>*/"MatSolveTransposeAdd" 1815 /*@ 1816 MatSolveTransposeAdd - Computes x = y + inv(Transpose(A)) b, given a 1817 factored matrix. 1818 1819 Collective on Mat and Vec 1820 1821 Input Parameters: 1822 + mat - the factored matrix 1823 . b - the right-hand-side vector 1824 - y - the vector to be added to 1825 1826 Output Parameter: 1827 . x - the result vector 1828 1829 Notes: 1830 The vectors b and x cannot be the same. I.e., one cannot 1831 call MatSolveTransposeAdd(A,x,y,x). 1832 1833 Most users should employ the simplified SLES interface for linear solvers 1834 instead of working directly with matrix algebra routines such as this. 1835 See, e.g., SLESCreate(). 1836 1837 Level: developer 1838 1839 .keywords: matrix, linear system, solve, LU, Cholesky, transpose, add 1840 1841 .seealso: MatSolve(), MatSolveAdd(), MatSolveTranspose() 1842 @*/ 1843 int MatSolveTransposeAdd(Mat mat,Vec b,Vec y,Vec x) 1844 { 1845 Scalar one = 1.0; 1846 int ierr; 1847 Vec tmp; 1848 1849 PetscFunctionBegin; 1850 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1851 PetscValidHeaderSpecific(y,VEC_COOKIE); 1852 PetscValidHeaderSpecific(b,VEC_COOKIE); 1853 PetscValidHeaderSpecific(x,VEC_COOKIE); 1854 PetscCheckSameComm(mat,b); 1855 PetscCheckSameComm(mat,y); 1856 PetscCheckSameComm(mat,x); 1857 if (x == b) SETERRQ(PETSC_ERR_ARG_IDN,0,"x and b must be different vectors"); 1858 if (!mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Unfactored matrix"); 1859 if (mat->M != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->M,x->N); 1860 if (mat->N != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->N,b->N); 1861 if (mat->N != y->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec y: global dim %d %d",mat->N,y->N); 1862 if (x->n != y->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Vec x,Vec y: local dim %d %d",x->n,y->n); 1863 1864 PLogEventBegin(MAT_SolveTransposeAdd,mat,b,x,y); 1865 if (mat->ops->solvetransposeadd) { 1866 ierr = (*mat->ops->solvetransposeadd)(mat,b,y,x);CHKERRQ(ierr); 1867 } else { 1868 /* do the solve then the add manually */ 1869 if (x != y) { 1870 ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr); 1871 ierr = VecAXPY(&one,y,x);CHKERRQ(ierr); 1872 } else { 1873 ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr); 1874 PLogObjectParent(mat,tmp); 1875 ierr = VecCopy(x,tmp);CHKERRQ(ierr); 1876 ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr); 1877 ierr = VecAXPY(&one,tmp,x);CHKERRQ(ierr); 1878 ierr = VecDestroy(tmp);CHKERRQ(ierr); 1879 } 1880 } 1881 PLogEventEnd(MAT_SolveTransposeAdd,mat,b,x,y); 1882 PetscFunctionReturn(0); 1883 } 1884 /* ----------------------------------------------------------------*/ 1885 1886 #undef __FUNC__ 1887 #define __FUNC__ /*<a name=""></a>*/"MatRelax" 1888 /*@ 1889 MatRelax - Computes one relaxation sweep. 1890 1891 Collective on Mat and Vec 1892 1893 Input Parameters: 1894 + mat - the matrix 1895 . b - the right hand side 1896 . omega - the relaxation factor 1897 . flag - flag indicating the type of SOR (see below) 1898 . shift - diagonal shift 1899 - its - the number of iterations 1900 1901 Output Parameters: 1902 . x - the solution (can contain an initial guess) 1903 1904 SOR Flags: 1905 . SOR_FORWARD_SWEEP - forward SOR 1906 . SOR_BACKWARD_SWEEP - backward SOR 1907 . SOR_SYMMETRIC_SWEEP - SSOR (symmetric SOR) 1908 . SOR_LOCAL_FORWARD_SWEEP - local forward SOR 1909 . SOR_LOCAL_BACKWARD_SWEEP - local forward SOR 1910 . SOR_LOCAL_SYMMETRIC_SWEEP - local SSOR 1911 . SOR_APPLY_UPPER, SOR_APPLY_LOWER - applies 1912 upper/lower triangular part of matrix to 1913 vector (with omega) 1914 . SOR_ZERO_INITIAL_GUESS - zero initial guess 1915 1916 Notes: 1917 SOR_LOCAL_FORWARD_SWEEP, SOR_LOCAL_BACKWARD_SWEEP, and 1918 SOR_LOCAL_SYMMETRIC_SWEEP perform seperate independent smoothings 1919 on each processor. 1920 1921 Application programmers will not generally use MatRelax() directly, 1922 but instead will employ the SLES/PC interface. 1923 1924 Notes for Advanced Users: 1925 The flags are implemented as bitwise inclusive or operations. 1926 For example, use (SOR_ZERO_INITIAL_GUESS | SOR_SYMMETRIC_SWEEP) 1927 to specify a zero initial guess for SSOR. 1928 1929 Most users should employ the simplified SLES interface for linear solvers 1930 instead of working directly with matrix algebra routines such as this. 1931 See, e.g., SLESCreate(). 1932 1933 Level: developer 1934 1935 .keywords: matrix, relax, relaxation, sweep 1936 @*/ 1937 int MatRelax(Mat mat,Vec b,PetscReal omega,MatSORType flag,PetscReal shift,int its,Vec x) 1938 { 1939 int ierr; 1940 1941 PetscFunctionBegin; 1942 PetscValidHeaderSpecific(mat,MAT_COOKIE); 1943 PetscValidHeaderSpecific(b,VEC_COOKIE); 1944 PetscValidHeaderSpecific(x,VEC_COOKIE); 1945 PetscCheckSameComm(mat,b); 1946 PetscCheckSameComm(mat,x); 1947 if (!mat->ops->relax) SETERRQ(PETSC_ERR_SUP,0,""); 1948 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 1949 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 1950 if (mat->N != x->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec x: global dim %d %d",mat->N,x->N); 1951 if (mat->M != b->N) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: global dim %d %d",mat->M,b->N); 1952 if (mat->m != b->n) SETERRQ2(PETSC_ERR_ARG_SIZ,0,"Mat mat,Vec b: local dim %d %d",mat->m,b->n); 1953 1954 PLogEventBegin(MAT_Relax,mat,b,x,0); 1955 ierr =(*mat->ops->relax)(mat,b,omega,flag,shift,its,x);CHKERRQ(ierr); 1956 PLogEventEnd(MAT_Relax,mat,b,x,0); 1957 PetscFunctionReturn(0); 1958 } 1959 1960 #undef __FUNC__ 1961 #define __FUNC__ /*<a name=""></a>*/"MatCopy_Basic" 1962 /* 1963 Default matrix copy routine. 1964 */ 1965 int MatCopy_Basic(Mat A,Mat B,MatStructure str) 1966 { 1967 int ierr,i,rstart,rend,nz,*cwork; 1968 Scalar *vwork; 1969 1970 PetscFunctionBegin; 1971 ierr = MatZeroEntries(B);CHKERRQ(ierr); 1972 ierr = MatGetOwnershipRange(A,&rstart,&rend);CHKERRQ(ierr); 1973 for (i=rstart; i<rend; i++) { 1974 ierr = MatGetRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 1975 ierr = MatSetValues(B,1,&i,nz,cwork,vwork,INSERT_VALUES);CHKERRQ(ierr); 1976 ierr = MatRestoreRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 1977 } 1978 ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1979 ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1980 PetscFunctionReturn(0); 1981 } 1982 1983 #undef __FUNC__ 1984 #define __FUNC__ /*<a name=""></a>*/"MatCopy" 1985 /*@C 1986 MatCopy - Copys a matrix to another matrix. 1987 1988 Collective on Mat 1989 1990 Input Parameters: 1991 + A - the matrix 1992 - str - SAME_NONZERO_PATTERN or DIFFERENT_NONZERO_PATTERN 1993 1994 Output Parameter: 1995 . B - where the copy is put 1996 1997 Notes: 1998 If you use SAME_NONZERO_PATTERN then the zero matrices had better have the 1999 same nonzero pattern or the routine will crash. 2000 2001 MatCopy() copies the matrix entries of a matrix to another existing 2002 matrix (after first zeroing the second matrix). A related routine is 2003 MatConvert(), which first creates a new matrix and then copies the data. 2004 2005 Level: intermediate 2006 2007 .keywords: matrix, copy, convert 2008 2009 .seealso: MatConvert() 2010 @*/ 2011 int MatCopy(Mat A,Mat B,MatStructure str) 2012 { 2013 int ierr; 2014 2015 PetscFunctionBegin; 2016 PetscValidHeaderSpecific(A,MAT_COOKIE); 2017 PetscValidHeaderSpecific(B,MAT_COOKIE); 2018 PetscCheckSameComm(A,B); 2019 if (!A->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2020 if (A->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2021 if (A->M != B->M || A->N != B->N) SETERRQ4(PETSC_ERR_ARG_SIZ,0,"Mat A,Mat B: global dim %d %d",A->M,B->M, 2022 A->N,B->N); 2023 2024 PLogEventBegin(MAT_Copy,A,B,0,0); 2025 if (A->ops->copy) { 2026 ierr = (*A->ops->copy)(A,B,str);CHKERRQ(ierr); 2027 } else { /* generic conversion */ 2028 ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr); 2029 } 2030 PLogEventEnd(MAT_Copy,A,B,0,0); 2031 PetscFunctionReturn(0); 2032 } 2033 2034 static int MatConvertersSet = 0; 2035 static int (*MatConverters[MAX_MATRIX_TYPES][MAX_MATRIX_TYPES])(Mat,MatType,Mat*) = 2036 {{0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}, 2037 {0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}, 2038 {0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}, 2039 {0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}, 2040 {0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}, 2041 {0,0,0,0,0,0,0,0,0,0,0,0},{0,0,0,0,0,0,0,0,0,0,0,0}}; 2042 2043 #undef __FUNC__ 2044 #define __FUNC__ /*<a name=""></a>*/"MatConvertRegister" 2045 /*@C 2046 MatConvertRegister - Allows one to register a routine that converts between 2047 two matrix types. 2048 2049 Not Collective 2050 2051 Input Parameters: 2052 + intype - the type of matrix (defined in include/petscmat.h), for example, MATSEQAIJ. 2053 - outtype - new matrix type, or MATSAME 2054 2055 Level: advanced 2056 2057 .seealso: MatConvertRegisterAll() 2058 @*/ 2059 int MatConvertRegister(MatType intype,MatType outtype,int (*converter)(Mat,MatType,Mat*)) 2060 { 2061 PetscFunctionBegin; 2062 MatConverters[intype][outtype] = converter; 2063 MatConvertersSet = 1; 2064 PetscFunctionReturn(0); 2065 } 2066 2067 #undef __FUNC__ 2068 #define __FUNC__ /*<a name=""></a>*/"MatConvert" 2069 /*@C 2070 MatConvert - Converts a matrix to another matrix, either of the same 2071 or different type. 2072 2073 Collective on Mat 2074 2075 Input Parameters: 2076 + mat - the matrix 2077 - newtype - new matrix type. Use MATSAME to create a new matrix of the 2078 same type as the original matrix. 2079 2080 Output Parameter: 2081 . M - pointer to place new matrix 2082 2083 Notes: 2084 MatConvert() first creates a new matrix and then copies the data from 2085 the first matrix. A related routine is MatCopy(), which copies the matrix 2086 entries of one matrix to another already existing matrix context. 2087 2088 Level: intermediate 2089 2090 .keywords: matrix, copy, convert 2091 2092 .seealso: MatCopy(), MatDuplicate() 2093 @*/ 2094 int MatConvert(Mat mat,MatType newtype,Mat *M) 2095 { 2096 int ierr; 2097 2098 PetscFunctionBegin; 2099 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2100 PetscValidPointer(M); 2101 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2102 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2103 2104 if (newtype > MAX_MATRIX_TYPES || newtype < -1) { 2105 SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,1,"Not a valid matrix type"); 2106 } 2107 *M = 0; 2108 2109 if (!MatConvertersSet) { 2110 ierr = MatLoadRegisterAll();CHKERRQ(ierr); 2111 } 2112 2113 PLogEventBegin(MAT_Convert,mat,0,0,0); 2114 if ((newtype == mat->type || newtype == MATSAME) && mat->ops->duplicate) { 2115 ierr = (*mat->ops->duplicate)(mat,MAT_COPY_VALUES,M);CHKERRQ(ierr); 2116 } else { 2117 if (!MatConvertersSet) { 2118 ierr = MatConvertRegisterAll();CHKERRQ(ierr); 2119 } 2120 if (!MatConverters[mat->type][newtype]) { 2121 SETERRQ(PETSC_ERR_ARG_WRONG,1,"Invalid matrix type, or matrix converter not registered"); 2122 } 2123 ierr = (*MatConverters[mat->type][newtype])(mat,newtype,M);CHKERRQ(ierr); 2124 } 2125 PLogEventEnd(MAT_Convert,mat,0,0,0); 2126 PetscFunctionReturn(0); 2127 } 2128 2129 #undef __FUNC__ 2130 #define __FUNC__ /*<a name=""></a>*/"MatDuplicate" 2131 /*@C 2132 MatDuplicate - Duplicates a matrix including the non-zero structure. 2133 2134 Collective on Mat 2135 2136 Input Parameters: 2137 + mat - the matrix 2138 - op - either MAT_DO_NOT_COPY_VALUES or MAT_COPY_VALUES, cause it to copy nonzero 2139 values as well or not 2140 2141 Output Parameter: 2142 . M - pointer to place new matrix 2143 2144 Level: intermediate 2145 2146 .keywords: matrix, copy, convert, duplicate 2147 2148 .seealso: MatCopy(), MatConvert() 2149 @*/ 2150 int MatDuplicate(Mat mat,MatDuplicateOption op,Mat *M) 2151 { 2152 int ierr; 2153 2154 PetscFunctionBegin; 2155 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2156 PetscValidPointer(M); 2157 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2158 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2159 2160 *M = 0; 2161 PLogEventBegin(MAT_Convert,mat,0,0,0); 2162 if (!mat->ops->duplicate) { 2163 SETERRQ(PETSC_ERR_SUP,1,"Not written for this matrix type"); 2164 } 2165 ierr = (*mat->ops->duplicate)(mat,op,M);CHKERRQ(ierr); 2166 PLogEventEnd(MAT_Convert,mat,0,0,0); 2167 PetscFunctionReturn(0); 2168 } 2169 2170 #undef __FUNC__ 2171 #define __FUNC__ /*<a name=""></a>*/"MatGetDiagonal" 2172 /*@ 2173 MatGetDiagonal - Gets the diagonal of a matrix. 2174 2175 Collective on Mat and Vec 2176 2177 Input Parameters: 2178 + mat - the matrix 2179 - v - the vector for storing the diagonal 2180 2181 Output Parameter: 2182 . v - the diagonal of the matrix 2183 2184 Notes: 2185 For the SeqAIJ matrix format, this routine may also be called 2186 on a LU factored matrix; in that case it routines the reciprocal of 2187 the diagonal entries in U. It returns the entries permuted by the 2188 row and column permutation used during the symbolic factorization. 2189 2190 Level: intermediate 2191 2192 .keywords: matrix, get, diagonal 2193 2194 .seealso: MatGetRow(), MatGetSubmatrices(), MatGetSubmatrix() 2195 @*/ 2196 int MatGetDiagonal(Mat mat,Vec v) 2197 { 2198 int ierr; 2199 2200 PetscFunctionBegin; 2201 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2202 PetscValidHeaderSpecific(v,VEC_COOKIE); 2203 /* PetscCheckSameComm(mat,v); Could be MPI vector but Seq matrix cause of two submatrix storage */ 2204 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2205 if (!mat->ops->getdiagonal) SETERRQ(PETSC_ERR_SUP,0,""); 2206 2207 ierr = (*mat->ops->getdiagonal)(mat,v);CHKERRQ(ierr); 2208 PetscFunctionReturn(0); 2209 } 2210 2211 #undef __FUNC__ 2212 #define __FUNC__ /*<a name=""></a>*/"MatTranspose" 2213 /*@C 2214 MatTranspose - Computes an in-place or out-of-place transpose of a matrix. 2215 2216 Collective on Mat 2217 2218 Input Parameter: 2219 . mat - the matrix to transpose 2220 2221 Output Parameters: 2222 . B - the transpose (or pass in PETSC_NULL for an in-place transpose) 2223 2224 Level: intermediate 2225 2226 .keywords: matrix, transpose 2227 2228 .seealso: MatMultTranspose(), MatMultTransposeAdd() 2229 @*/ 2230 int MatTranspose(Mat mat,Mat *B) 2231 { 2232 int ierr; 2233 2234 PetscFunctionBegin; 2235 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2236 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2237 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2238 if (!mat->ops->transpose) SETERRQ(PETSC_ERR_SUP,0,""); 2239 ierr = (*mat->ops->transpose)(mat,B);CHKERRQ(ierr); 2240 PetscFunctionReturn(0); 2241 } 2242 2243 #undef __FUNC__ 2244 #define __FUNC__ /*<a name=""></a>*/"MatPermute" 2245 /*@C 2246 MatPermute - Creates a new matrix with rows and columns permuted from the 2247 original. 2248 2249 Collective on Mat 2250 2251 Input Parameters: 2252 + mat - the matrix to permute 2253 . row - row permutation, each processor supplies only the permutation for its rows 2254 - col - column permutation, each processor needs the entire column permutation, that is 2255 this is the same size as the total number of columns in the matrix 2256 2257 Output Parameters: 2258 . B - the permuted matrix 2259 2260 Level: advanced 2261 2262 .keywords: matrix, transpose 2263 2264 .seealso: MatGetOrdering() 2265 @*/ 2266 int MatPermute(Mat mat,IS row,IS col,Mat *B) 2267 { 2268 int ierr; 2269 2270 PetscFunctionBegin; 2271 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2272 PetscValidHeaderSpecific(row,IS_COOKIE); 2273 PetscValidHeaderSpecific(col,IS_COOKIE); 2274 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2275 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2276 if (!mat->ops->permute) SETERRQ(PETSC_ERR_SUP,0,""); 2277 ierr = (*mat->ops->permute)(mat,row,col,B);CHKERRQ(ierr); 2278 PetscFunctionReturn(0); 2279 } 2280 2281 #undef __FUNC__ 2282 #define __FUNC__ /*<a name=""></a>*/"MatEqual" 2283 /*@ 2284 MatEqual - Compares two matrices. 2285 2286 Collective on Mat 2287 2288 Input Parameters: 2289 + A - the first matrix 2290 - B - the second matrix 2291 2292 Output Parameter: 2293 . flg - PETSC_TRUE if the matrices are equal; PETSC_FALSE otherwise. 2294 2295 Level: intermediate 2296 2297 .keywords: matrix, equal, equivalent 2298 @*/ 2299 int MatEqual(Mat A,Mat B,PetscTruth *flg) 2300 { 2301 int ierr; 2302 2303 PetscFunctionBegin; 2304 PetscValidHeaderSpecific(A,MAT_COOKIE); 2305 PetscValidHeaderSpecific(B,MAT_COOKIE); 2306 PetscValidIntPointer(flg); 2307 PetscCheckSameComm(A,B); 2308 if (!A->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2309 if (!B->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2310 if (A->M != B->M || A->N != B->N) SETERRQ4(PETSC_ERR_ARG_SIZ,0,"Mat A,Mat B: global dim %d %d %d %d", 2311 A->M,B->M,A->N,B->N); 2312 if (!A->ops->equal) SETERRQ(PETSC_ERR_SUP,0,""); 2313 ierr = (*A->ops->equal)(A,B,flg);CHKERRQ(ierr); 2314 PetscFunctionReturn(0); 2315 } 2316 2317 #undef __FUNC__ 2318 #define __FUNC__ /*<a name=""></a>*/"MatDiagonalScale" 2319 /*@ 2320 MatDiagonalScale - Scales a matrix on the left and right by diagonal 2321 matrices that are stored as vectors. Either of the two scaling 2322 matrices can be PETSC_NULL. 2323 2324 Collective on Mat 2325 2326 Input Parameters: 2327 + mat - the matrix to be scaled 2328 . l - the left scaling vector (or PETSC_NULL) 2329 - r - the right scaling vector (or PETSC_NULL) 2330 2331 Notes: 2332 MatDiagonalScale() computes A = LAR, where 2333 L = a diagonal matrix, R = a diagonal matrix 2334 2335 Level: intermediate 2336 2337 .keywords: matrix, diagonal, scale 2338 2339 .seealso: MatScale() 2340 @*/ 2341 int MatDiagonalScale(Mat mat,Vec l,Vec r) 2342 { 2343 int ierr; 2344 2345 PetscFunctionBegin; 2346 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2347 if (!mat->ops->diagonalscale) SETERRQ(PETSC_ERR_SUP,0,""); 2348 if (l) {PetscValidHeaderSpecific(l,VEC_COOKIE);PetscCheckSameComm(mat,l);} 2349 if (r) {PetscValidHeaderSpecific(r,VEC_COOKIE);PetscCheckSameComm(mat,r);} 2350 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2351 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2352 2353 PLogEventBegin(MAT_Scale,mat,0,0,0); 2354 ierr = (*mat->ops->diagonalscale)(mat,l,r);CHKERRQ(ierr); 2355 PLogEventEnd(MAT_Scale,mat,0,0,0); 2356 PetscFunctionReturn(0); 2357 } 2358 2359 #undef __FUNC__ 2360 #define __FUNC__ /*<a name=""></a>*/"MatScale" 2361 /*@ 2362 MatScale - Scales all elements of a matrix by a given number. 2363 2364 Collective on Mat 2365 2366 Input Parameters: 2367 + mat - the matrix to be scaled 2368 - a - the scaling value 2369 2370 Output Parameter: 2371 . mat - the scaled matrix 2372 2373 Level: intermediate 2374 2375 .keywords: matrix, scale 2376 2377 .seealso: MatDiagonalScale() 2378 @*/ 2379 int MatScale(Scalar *a,Mat mat) 2380 { 2381 int ierr; 2382 2383 PetscFunctionBegin; 2384 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2385 PetscValidScalarPointer(a); 2386 if (!mat->ops->scale) SETERRQ(PETSC_ERR_SUP,0,""); 2387 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2388 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2389 2390 PLogEventBegin(MAT_Scale,mat,0,0,0); 2391 ierr = (*mat->ops->scale)(a,mat);CHKERRQ(ierr); 2392 PLogEventEnd(MAT_Scale,mat,0,0,0); 2393 PetscFunctionReturn(0); 2394 } 2395 2396 #undef __FUNC__ 2397 #define __FUNC__ /*<a name=""></a>*/"MatNorm" 2398 /*@ 2399 MatNorm - Calculates various norms of a matrix. 2400 2401 Collective on Mat 2402 2403 Input Parameters: 2404 + mat - the matrix 2405 - type - the type of norm, NORM_1, NORM_2, NORM_FROBENIUS, NORM_INFINITY 2406 2407 Output Parameters: 2408 . norm - the resulting norm 2409 2410 Level: intermediate 2411 2412 .keywords: matrix, norm, Frobenius 2413 @*/ 2414 int MatNorm(Mat mat,NormType type,PetscReal *norm) 2415 { 2416 int ierr; 2417 2418 PetscFunctionBegin; 2419 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2420 PetscValidScalarPointer(norm); 2421 2422 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2423 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2424 if (!mat->ops->norm) SETERRQ(PETSC_ERR_SUP,0,"Not for this matrix type"); 2425 ierr = (*mat->ops->norm)(mat,type,norm);CHKERRQ(ierr); 2426 PetscFunctionReturn(0); 2427 } 2428 2429 /* 2430 This variable is used to prevent counting of MatAssemblyBegin() that 2431 are called from within a MatAssemblyEnd(). 2432 */ 2433 static int MatAssemblyEnd_InUse = 0; 2434 #undef __FUNC__ 2435 #define __FUNC__ /*<a name=""></a>*/"MatAssemblyBegin" 2436 /*@ 2437 MatAssemblyBegin - Begins assembling the matrix. This routine should 2438 be called after completing all calls to MatSetValues(). 2439 2440 Collective on Mat 2441 2442 Input Parameters: 2443 + mat - the matrix 2444 - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY 2445 2446 Notes: 2447 MatSetValues() generally caches the values. The matrix is ready to 2448 use only after MatAssemblyBegin() and MatAssemblyEnd() have been called. 2449 Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES 2450 in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before 2451 using the matrix. 2452 2453 Level: beginner 2454 2455 .keywords: matrix, assembly, assemble, begin 2456 2457 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssembled() 2458 @*/ 2459 int MatAssemblyBegin(Mat mat,MatAssemblyType type) 2460 { 2461 int ierr; 2462 2463 PetscFunctionBegin; 2464 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2465 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix.\n did you forget to call MatSetUnfactored()?"); 2466 if (mat->assembled) { 2467 mat->was_assembled = PETSC_TRUE; 2468 mat->assembled = PETSC_FALSE; 2469 } 2470 if (!MatAssemblyEnd_InUse) { 2471 PLogEventBegin(MAT_AssemblyBegin,mat,0,0,0); 2472 if (mat->ops->assemblybegin){ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr);} 2473 PLogEventEnd(MAT_AssemblyBegin,mat,0,0,0); 2474 } else { 2475 if (mat->ops->assemblybegin){ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr);} 2476 } 2477 PetscFunctionReturn(0); 2478 } 2479 2480 #undef __FUNC__ 2481 #define __FUNC__ /*<a name=""></a>*/"MatAssembed" 2482 /*@ 2483 MatAssembled - Indicates if a matrix has been assembled and is ready for 2484 use; for example, in matrix-vector product. 2485 2486 Collective on Mat 2487 2488 Input Parameter: 2489 . mat - the matrix 2490 2491 Output Parameter: 2492 . assembled - PETSC_TRUE or PETSC_FALSE 2493 2494 Level: advanced 2495 2496 .keywords: matrix, assembly, assemble, begin 2497 2498 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssemblyBegin() 2499 @*/ 2500 int MatAssembled(Mat mat,PetscTruth *assembled) 2501 { 2502 PetscFunctionBegin; 2503 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2504 *assembled = mat->assembled; 2505 PetscFunctionReturn(0); 2506 } 2507 2508 #undef __FUNC__ 2509 #define __FUNC__ /*<a name=""></a>*/"MatView_Private" 2510 /* 2511 Processes command line options to determine if/how a matrix 2512 is to be viewed. Called by MatAssemblyEnd() and MatLoad(). 2513 */ 2514 int MatView_Private(Mat mat) 2515 { 2516 int ierr; 2517 PetscTruth flg; 2518 2519 PetscFunctionBegin; 2520 ierr = OptionsHasName(mat->prefix,"-mat_view_info",&flg);CHKERRQ(ierr); 2521 if (flg) { 2522 ierr = ViewerPushFormat(VIEWER_STDOUT_(mat->comm),VIEWER_FORMAT_ASCII_INFO,0);CHKERRQ(ierr); 2523 ierr = MatView(mat,VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2524 ierr = ViewerPopFormat(VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2525 } 2526 ierr = OptionsHasName(mat->prefix,"-mat_view_info_detailed",&flg);CHKERRQ(ierr); 2527 if (flg) { 2528 ierr = ViewerPushFormat(VIEWER_STDOUT_(mat->comm),VIEWER_FORMAT_ASCII_INFO_LONG,0);CHKERRQ(ierr); 2529 ierr = MatView(mat,VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2530 ierr = ViewerPopFormat(VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2531 } 2532 ierr = OptionsHasName(mat->prefix,"-mat_view",&flg);CHKERRQ(ierr); 2533 if (flg) { 2534 ierr = MatView(mat,VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2535 } 2536 ierr = OptionsHasName(mat->prefix,"-mat_view_matlab",&flg);CHKERRQ(ierr); 2537 if (flg) { 2538 ierr = ViewerPushFormat(VIEWER_STDOUT_(mat->comm),VIEWER_FORMAT_ASCII_MATLAB,"M");CHKERRQ(ierr); 2539 ierr = MatView(mat,VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2540 ierr = ViewerPopFormat(VIEWER_STDOUT_(mat->comm));CHKERRQ(ierr); 2541 } 2542 ierr = OptionsHasName(mat->prefix,"-mat_view_draw",&flg);CHKERRQ(ierr); 2543 if (flg) { 2544 ierr = OptionsHasName(mat->prefix,"-mat_view_contour",&flg);CHKERRQ(ierr); 2545 if (flg) { 2546 ViewerPushFormat(VIEWER_DRAW_(mat->comm),VIEWER_FORMAT_DRAW_CONTOUR,0);CHKERRQ(ierr); 2547 } 2548 ierr = MatView(mat,VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 2549 ierr = ViewerFlush(VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 2550 if (flg) { 2551 ViewerPopFormat(VIEWER_DRAW_(mat->comm));CHKERRQ(ierr); 2552 } 2553 } 2554 ierr = OptionsHasName(mat->prefix,"-mat_view_socket",&flg);CHKERRQ(ierr); 2555 if (flg) { 2556 ierr = MatView(mat,VIEWER_SOCKET_(mat->comm));CHKERRQ(ierr); 2557 ierr = ViewerFlush(VIEWER_SOCKET_(mat->comm));CHKERRQ(ierr); 2558 } 2559 PetscFunctionReturn(0); 2560 } 2561 2562 #undef __FUNC__ 2563 #define __FUNC__ /*<a name=""></a>*/"MatAssemblyEnd" 2564 /*@ 2565 MatAssemblyEnd - Completes assembling the matrix. This routine should 2566 be called after MatAssemblyBegin(). 2567 2568 Collective on Mat 2569 2570 Input Parameters: 2571 + mat - the matrix 2572 - type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY 2573 2574 Options Database Keys: 2575 + -mat_view_info - Prints info on matrix at conclusion of MatEndAssembly() 2576 . -mat_view_info_detailed - Prints more detailed info 2577 . -mat_view - Prints matrix in ASCII format 2578 . -mat_view_matlab - Prints matrix in Matlab format 2579 . -mat_view_draw - Draws nonzero structure of matrix, using MatView() and DrawOpenX(). 2580 . -display <name> - Sets display name (default is host) 2581 - -draw_pause <sec> - Sets number of seconds to pause after display 2582 2583 Notes: 2584 MatSetValues() generally caches the values. The matrix is ready to 2585 use only after MatAssemblyBegin() and MatAssemblyEnd() have been called. 2586 Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES 2587 in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before 2588 using the matrix. 2589 2590 Level: beginner 2591 2592 .keywords: matrix, assembly, assemble, end 2593 2594 .seealso: MatAssemblyBegin(), MatSetValues(), DrawOpenX(), MatView(), MatAssembled() 2595 @*/ 2596 int MatAssemblyEnd(Mat mat,MatAssemblyType type) 2597 { 2598 int ierr; 2599 static int inassm = 0; 2600 2601 PetscFunctionBegin; 2602 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2603 2604 inassm++; 2605 MatAssemblyEnd_InUse++; 2606 if (MatAssemblyEnd_InUse == 1) { /* Do the logging only the first time through */ 2607 PLogEventBegin(MAT_AssemblyEnd,mat,0,0,0); 2608 if (mat->ops->assemblyend) { 2609 ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr); 2610 } 2611 PLogEventEnd(MAT_AssemblyEnd,mat,0,0,0); 2612 } else { 2613 if (mat->ops->assemblyend) { 2614 ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr); 2615 } 2616 } 2617 2618 /* Flush assembly is not a true assembly */ 2619 if (type != MAT_FLUSH_ASSEMBLY) { 2620 mat->assembled = PETSC_TRUE; mat->num_ass++; 2621 } 2622 mat->insertmode = NOT_SET_VALUES; 2623 MatAssemblyEnd_InUse--; 2624 2625 if (inassm == 1 && type != MAT_FLUSH_ASSEMBLY) { 2626 ierr = MatView_Private(mat);CHKERRQ(ierr); 2627 } 2628 inassm--; 2629 PetscFunctionReturn(0); 2630 } 2631 2632 2633 #undef __FUNC__ 2634 #define __FUNC__ /*<a name=""></a>*/"MatCompress" 2635 /*@ 2636 MatCompress - Tries to store the matrix in as little space as 2637 possible. May fail if memory is already fully used, since it 2638 tries to allocate new space. 2639 2640 Collective on Mat 2641 2642 Input Parameters: 2643 . mat - the matrix 2644 2645 Level: advanced 2646 2647 .keywords: matrix, compress 2648 @*/ 2649 int MatCompress(Mat mat) 2650 { 2651 int ierr; 2652 2653 PetscFunctionBegin; 2654 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2655 if (mat->ops->compress) {ierr = (*mat->ops->compress)(mat);CHKERRQ(ierr);} 2656 PetscFunctionReturn(0); 2657 } 2658 2659 #undef __FUNC__ 2660 #define __FUNC__ /*<a name=""></a>*/"MatSetOption" 2661 /*@ 2662 MatSetOption - Sets a parameter option for a matrix. Some options 2663 may be specific to certain storage formats. Some options 2664 determine how values will be inserted (or added). Sorted, 2665 row-oriented input will generally assemble the fastest. The default 2666 is row-oriented, nonsorted input. 2667 2668 Collective on Mat 2669 2670 Input Parameters: 2671 + mat - the matrix 2672 - option - the option, one of those listed below (and possibly others), 2673 e.g., MAT_ROWS_SORTED, MAT_NEW_NONZERO_LOCATION_ERR 2674 2675 Options Describing Matrix Structure: 2676 + MAT_SYMMETRIC - symmetric in terms of both structure and value 2677 - MAT_STRUCTURALLY_SYMMETRIC - symmetric nonzero structure 2678 2679 Options For Use with MatSetValues(): 2680 Insert a logically dense subblock, which can be 2681 + MAT_ROW_ORIENTED - row-oriented 2682 . MAT_COLUMN_ORIENTED - column-oriented 2683 . MAT_ROWS_SORTED - sorted by row 2684 . MAT_ROWS_UNSORTED - not sorted by row 2685 . MAT_COLUMNS_SORTED - sorted by column 2686 - MAT_COLUMNS_UNSORTED - not sorted by column 2687 2688 Not these options reflect the data you pass in with MatSetValues(); it has 2689 nothing to do with how the data is stored internally in the matrix 2690 data structure. 2691 2692 When (re)assembling a matrix, we can restrict the input for 2693 efficiency/debugging purposes. These options include 2694 + MAT_NO_NEW_NONZERO_LOCATIONS - additional insertions will not be 2695 allowed if they generate a new nonzero 2696 . MAT_YES_NEW_NONZERO_LOCATIONS - additional insertions will be allowed 2697 . MAT_NO_NEW_DIAGONALS - additional insertions will not be allowed if 2698 they generate a nonzero in a new diagonal (for block diagonal format only) 2699 . MAT_YES_NEW_DIAGONALS - new diagonals will be allowed (for block diagonal format only) 2700 . MAT_IGNORE_OFF_PROC_ENTRIES - drops off-processor entries 2701 . MAT_NEW_NONZERO_LOCATION_ERR - generates an error for new matrix entry 2702 - MAT_USE_HASH_TABLE - uses a hash table to speed up matrix assembly 2703 2704 Notes: 2705 Some options are relevant only for particular matrix types and 2706 are thus ignored by others. Other options are not supported by 2707 certain matrix types and will generate an error message if set. 2708 2709 If using a Fortran 77 module to compute a matrix, one may need to 2710 use the column-oriented option (or convert to the row-oriented 2711 format). 2712 2713 MAT_NO_NEW_NONZERO_LOCATIONS indicates that any add or insertion 2714 that would generate a new entry in the nonzero structure is instead 2715 ignored. Thus, if memory has not alredy been allocated for this particular 2716 data, then the insertion is ignored. For dense matrices, in which 2717 the entire array is allocated, no entries are ever ignored. 2718 2719 MAT_NEW_NONZERO_LOCATION_ERR indicates that any add or insertion 2720 that would generate a new entry in the nonzero structure instead produces 2721 an error. (Currently supported for AIJ and BAIJ formats only.) 2722 This is a useful flag when using SAME_NONZERO_PATTERN in calling 2723 SLESSetOperators() to ensure that the nonzero pattern truely does 2724 remain unchanged. 2725 2726 MAT_NEW_NONZERO_ALLOCATION_ERR indicates that any add or insertion 2727 that would generate a new entry that has not been preallocated will 2728 instead produce an error. (Currently supported for AIJ and BAIJ formats 2729 only.) This is a useful flag when debugging matrix memory preallocation. 2730 2731 MAT_IGNORE_OFF_PROC_ENTRIES indicates entries destined for 2732 other processors should be dropped, rather than stashed. 2733 This is useful if you know that the "owning" processor is also 2734 always generating the correct matrix entries, so that PETSc need 2735 not transfer duplicate entries generated on another processor. 2736 2737 MAT_USE_HASH_TABLE indicates that a hash table be used to improve the 2738 searches during matrix assembly. When this flag is set, the hash table 2739 is created during the first Matrix Assembly. This hash table is 2740 used the next time through, during MatSetVaules()/MatSetVaulesBlocked() 2741 to improve the searching of indices. MAT_NO_NEW_NONZERO_LOCATIONS flag 2742 should be used with MAT_USE_HASH_TABLE flag. This option is currently 2743 supported by MATMPIBAIJ format only. 2744 2745 MAT_KEEP_ZEROED_ROWS indicates when MatZeroRows() is called the zeroed entries 2746 are kept in the nonzero structure 2747 2748 MAT_IGNORE_ZERO_ENTRIES - when using ADD_VALUES for AIJ matrices this will stop 2749 zero values from creating a zero location in the matrix 2750 2751 Level: intermediate 2752 2753 .keywords: matrix, option, row-oriented, column-oriented, sorted, nonzero 2754 @*/ 2755 int MatSetOption(Mat mat,MatOption op) 2756 { 2757 int ierr; 2758 2759 PetscFunctionBegin; 2760 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2761 if (mat->ops->setoption) {ierr = (*mat->ops->setoption)(mat,op);CHKERRQ(ierr);} 2762 PetscFunctionReturn(0); 2763 } 2764 2765 #undef __FUNC__ 2766 #define __FUNC__ /*<a name=""></a>*/"MatZeroEntries" 2767 /*@ 2768 MatZeroEntries - Zeros all entries of a matrix. For sparse matrices 2769 this routine retains the old nonzero structure. 2770 2771 Collective on Mat 2772 2773 Input Parameters: 2774 . mat - the matrix 2775 2776 Level: intermediate 2777 2778 .keywords: matrix, zero, entries 2779 2780 .seealso: MatZeroRows() 2781 @*/ 2782 int MatZeroEntries(Mat mat) 2783 { 2784 int ierr; 2785 2786 PetscFunctionBegin; 2787 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2788 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2789 if (!mat->ops->zeroentries) SETERRQ(PETSC_ERR_SUP,0,""); 2790 2791 PLogEventBegin(MAT_ZeroEntries,mat,0,0,0); 2792 ierr = (*mat->ops->zeroentries)(mat);CHKERRQ(ierr); 2793 PLogEventEnd(MAT_ZeroEntries,mat,0,0,0); 2794 PetscFunctionReturn(0); 2795 } 2796 2797 #undef __FUNC__ 2798 #define __FUNC__ /*<a name=""></a>*/"MatZeroRows" 2799 /*@C 2800 MatZeroRows - Zeros all entries (except possibly the main diagonal) 2801 of a set of rows of a matrix. 2802 2803 Collective on Mat 2804 2805 Input Parameters: 2806 + mat - the matrix 2807 . is - index set of rows to remove 2808 - diag - pointer to value put in all diagonals of eliminated rows. 2809 Note that diag is not a pointer to an array, but merely a 2810 pointer to a single value. 2811 2812 Notes: 2813 For the AIJ and BAIJ matrix formats this removes the old nonzero structure, 2814 but does not release memory. For the dense and block diagonal 2815 formats this does not alter the nonzero structure. 2816 2817 If the option MatSetOption(mat,MAT_KEEP_ZEROED_ROWS) the nonzero structure 2818 of the matrix is not changed (even for AIJ and BAIJ matrices) the values are 2819 merely zeroed. 2820 2821 The user can set a value in the diagonal entry (or for the AIJ and 2822 row formats can optionally remove the main diagonal entry from the 2823 nonzero structure as well, by passing a null pointer (PETSC_NULL 2824 in C or PETSC_NULL_SCALAR in Fortran) as the final argument). 2825 2826 For the parallel case, all processes that share the matrix (i.e., 2827 those in the communicator used for matrix creation) MUST call this 2828 routine, regardless of whether any rows being zeroed are owned by 2829 them. 2830 2831 2832 Level: intermediate 2833 2834 .keywords: matrix, zero, rows, boundary conditions 2835 2836 .seealso: MatZeroEntries(), MatZeroRowsLocal(), MatSetOption() 2837 @*/ 2838 int MatZeroRows(Mat mat,IS is,Scalar *diag) 2839 { 2840 int ierr; 2841 2842 PetscFunctionBegin; 2843 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2844 PetscValidHeaderSpecific(is,IS_COOKIE); 2845 if (diag) PetscValidScalarPointer(diag); 2846 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2847 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2848 if (!mat->ops->zerorows) SETERRQ(PETSC_ERR_SUP,0,""); 2849 2850 ierr = (*mat->ops->zerorows)(mat,is,diag);CHKERRQ(ierr); 2851 ierr = MatView_Private(mat);CHKERRQ(ierr); 2852 PetscFunctionReturn(0); 2853 } 2854 2855 #undef __FUNC__ 2856 #define __FUNC__ /*<a name=""></a>*/"MatZeroRowsLocal" 2857 /*@C 2858 MatZeroRowsLocal - Zeros all entries (except possibly the main diagonal) 2859 of a set of rows of a matrix; using local numbering of rows. 2860 2861 Collective on Mat 2862 2863 Input Parameters: 2864 + mat - the matrix 2865 . is - index set of rows to remove 2866 - diag - pointer to value put in all diagonals of eliminated rows. 2867 Note that diag is not a pointer to an array, but merely a 2868 pointer to a single value. 2869 2870 Notes: 2871 Before calling MatZeroRowsLocal(), the user must first set the 2872 local-to-global mapping by calling MatSetLocalToGlobalMapping(). 2873 2874 For the AIJ matrix formats this removes the old nonzero structure, 2875 but does not release memory. For the dense and block diagonal 2876 formats this does not alter the nonzero structure. 2877 2878 The user can set a value in the diagonal entry (or for the AIJ and 2879 row formats can optionally remove the main diagonal entry from the 2880 nonzero structure as well, by passing a null pointer (PETSC_NULL 2881 in C or PETSC_NULL_SCALAR in Fortran) as the final argument). 2882 2883 Level: intermediate 2884 2885 .keywords: matrix, zero, rows, boundary conditions 2886 2887 .seealso: MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping 2888 @*/ 2889 int MatZeroRowsLocal(Mat mat,IS is,Scalar *diag) 2890 { 2891 int ierr; 2892 IS newis; 2893 2894 PetscFunctionBegin; 2895 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2896 PetscValidHeaderSpecific(is,IS_COOKIE); 2897 if (diag) PetscValidScalarPointer(diag); 2898 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 2899 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 2900 if (!mat->ops->zerorows) SETERRQ(PETSC_ERR_SUP,0,""); 2901 if (!mat->mapping) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Need to provide local to global mapping to matrix first"); 2902 2903 ierr = ISLocalToGlobalMappingApplyIS(mat->mapping,is,&newis);CHKERRQ(ierr); 2904 ierr = (*mat->ops->zerorows)(mat,newis,diag);CHKERRQ(ierr); 2905 ierr = ISDestroy(newis);CHKERRQ(ierr); 2906 PetscFunctionReturn(0); 2907 } 2908 2909 #undef __FUNC__ 2910 #define __FUNC__ /*<a name=""></a>*/"MatGetSize" 2911 /*@ 2912 MatGetSize - Returns the numbers of rows and columns in a matrix. 2913 2914 Not Collective 2915 2916 Input Parameter: 2917 . mat - the matrix 2918 2919 Output Parameters: 2920 + m - the number of global rows 2921 - n - the number of global columns 2922 2923 Level: beginner 2924 2925 .keywords: matrix, dimension, size, rows, columns, global, get 2926 2927 .seealso: MatGetLocalSize() 2928 @*/ 2929 int MatGetSize(Mat mat,int *m,int* n) 2930 { 2931 int ierr; 2932 2933 PetscFunctionBegin; 2934 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2935 ierr = (*mat->ops->getsize)(mat,m,n);CHKERRQ(ierr); 2936 PetscFunctionReturn(0); 2937 } 2938 2939 #undef __FUNC__ 2940 #define __FUNC__ /*<a name=""></a>*/"MatGetLocalSize" 2941 /*@ 2942 MatGetLocalSize - Returns the number of rows and columns in a matrix 2943 stored locally. This information may be implementation dependent, so 2944 use with care. 2945 2946 Not Collective 2947 2948 Input Parameters: 2949 . mat - the matrix 2950 2951 Output Parameters: 2952 + m - the number of local rows 2953 - n - the number of local columns 2954 2955 Level: beginner 2956 2957 .keywords: matrix, dimension, size, local, rows, columns, get 2958 2959 .seealso: MatGetSize() 2960 @*/ 2961 int MatGetLocalSize(Mat mat,int *m,int* n) 2962 { 2963 int ierr; 2964 2965 PetscFunctionBegin; 2966 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2967 ierr = (*mat->ops->getlocalsize)(mat,m,n);CHKERRQ(ierr); 2968 PetscFunctionReturn(0); 2969 } 2970 2971 #undef __FUNC__ 2972 #define __FUNC__ /*<a name=""></a>*/"MatGetOwnershipRange" 2973 /*@ 2974 MatGetOwnershipRange - Returns the range of matrix rows owned by 2975 this processor, assuming that the matrix is laid out with the first 2976 n1 rows on the first processor, the next n2 rows on the second, etc. 2977 For certain parallel layouts this range may not be well defined. 2978 2979 Not Collective 2980 2981 Input Parameters: 2982 . mat - the matrix 2983 2984 Output Parameters: 2985 + m - the global index of the first local row 2986 - n - one more than the global index of the last local row 2987 2988 Level: beginner 2989 2990 .keywords: matrix, get, range, ownership 2991 @*/ 2992 int MatGetOwnershipRange(Mat mat,int *m,int* n) 2993 { 2994 int ierr; 2995 2996 PetscFunctionBegin; 2997 PetscValidHeaderSpecific(mat,MAT_COOKIE); 2998 if (m) PetscValidIntPointer(m); 2999 if (n) PetscValidIntPointer(n); 3000 if (!mat->ops->getownershiprange) SETERRQ(PETSC_ERR_SUP,0,""); 3001 ierr = (*mat->ops->getownershiprange)(mat,m,n);CHKERRQ(ierr); 3002 PetscFunctionReturn(0); 3003 } 3004 3005 #undef __FUNC__ 3006 #define __FUNC__ /*<a name=""></a>*/"MatILUFactorSymbolic" 3007 /*@ 3008 MatILUFactorSymbolic - Performs symbolic ILU factorization of a matrix. 3009 Uses levels of fill only, not drop tolerance. Use MatLUFactorNumeric() 3010 to complete the factorization. 3011 3012 Collective on Mat 3013 3014 Input Parameters: 3015 + mat - the matrix 3016 . row - row permutation 3017 . column - column permutation 3018 - info - structure containing 3019 $ levels - number of levels of fill. 3020 $ expected fill - as ratio of original fill. 3021 $ 1 or 0 - indicating force fill on diagonal (improves robustness for matrices 3022 missing diagonal entries) 3023 3024 Output Parameters: 3025 . fact - new matrix that has been symbolically factored 3026 3027 Notes: 3028 See the users manual for additional information about 3029 choosing the fill factor for better efficiency. 3030 3031 Most users should employ the simplified SLES interface for linear solvers 3032 instead of working directly with matrix algebra routines such as this. 3033 See, e.g., SLESCreate(). 3034 3035 Level: developer 3036 3037 .keywords: matrix, factor, incomplete, ILU, symbolic, fill 3038 3039 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor() 3040 MatGetOrdering() 3041 3042 @*/ 3043 int MatILUFactorSymbolic(Mat mat,IS row,IS col,MatILUInfo *info,Mat *fact) 3044 { 3045 int ierr; 3046 3047 PetscFunctionBegin; 3048 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3049 PetscValidPointer(fact); 3050 if (info && info->levels < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,0,"Levels of fill negative %d",info->levels); 3051 if (info && info->fill < 1.0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,0,"Expected fill less than 1.0 %g",info->fill); 3052 if (!mat->ops->ilufactorsymbolic) SETERRQ(PETSC_ERR_SUP,0,"Only MatCreateMPIRowbs() matrices support parallel ILU"); 3053 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 3054 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 3055 3056 PLogEventBegin(MAT_ILUFactorSymbolic,mat,row,col,0); 3057 ierr = (*mat->ops->ilufactorsymbolic)(mat,row,col,info,fact);CHKERRQ(ierr); 3058 PLogEventEnd(MAT_ILUFactorSymbolic,mat,row,col,0); 3059 PetscFunctionReturn(0); 3060 } 3061 3062 #undef __FUNC__ 3063 #define __FUNC__ /*<a name=""></a>*/"MatIncompleteCholeskyFactorSymbolic" 3064 /*@ 3065 MatIncompleteCholeskyFactorSymbolic - Performs symbolic incomplete 3066 Cholesky factorization for a symmetric matrix. Use 3067 MatCholeskyFactorNumeric() to complete the factorization. 3068 3069 Collective on Mat 3070 3071 Input Parameters: 3072 + mat - the matrix 3073 . perm - row and column permutation 3074 . fill - levels of fill 3075 - f - expected fill as ratio of original fill 3076 3077 Output Parameter: 3078 . fact - the factored matrix 3079 3080 Notes: 3081 Currently only no-fill factorization is supported. 3082 3083 Most users should employ the simplified SLES interface for linear solvers 3084 instead of working directly with matrix algebra routines such as this. 3085 See, e.g., SLESCreate(). 3086 3087 Level: developer 3088 3089 .keywords: matrix, factor, incomplete, ICC, Cholesky, symbolic, fill 3090 3091 .seealso: MatCholeskyFactorNumeric(), MatCholeskyFactor() 3092 @*/ 3093 int MatIncompleteCholeskyFactorSymbolic(Mat mat,IS perm,PetscReal f,int fill,Mat *fact) 3094 { 3095 int ierr; 3096 3097 PetscFunctionBegin; 3098 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3099 PetscValidPointer(fact); 3100 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 3101 if (fill < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,0,"Fill negative %d",fill); 3102 if (f < 1.0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,0,"Expected fill less than 1.0 %g",f); 3103 if (!mat->ops->incompletecholeskyfactorsymbolic) SETERRQ(PETSC_ERR_SUP,0,"Currently only MatCreateMPIRowbs() matrices support ICC in parallel"); 3104 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 3105 3106 PLogEventBegin(MAT_IncompleteCholeskyFactorSymbolic,mat,perm,0,0); 3107 ierr = (*mat->ops->incompletecholeskyfactorsymbolic)(mat,perm,f,fill,fact);CHKERRQ(ierr); 3108 PLogEventEnd(MAT_IncompleteCholeskyFactorSymbolic,mat,perm,0,0); 3109 PetscFunctionReturn(0); 3110 } 3111 3112 #undef __FUNC__ 3113 #define __FUNC__ /*<a name=""></a>*/"MatGetArray" 3114 /*@C 3115 MatGetArray - Returns a pointer to the element values in the matrix. 3116 The result of this routine is dependent on the underlying matrix data 3117 structure, and may not even work for certain matrix types. You MUST 3118 call MatRestoreArray() when you no longer need to access the array. 3119 3120 Not Collective 3121 3122 Input Parameter: 3123 . mat - the matrix 3124 3125 Output Parameter: 3126 . v - the location of the values 3127 3128 Currently returns an array only for the dense formats, giving access to 3129 the local portion of the matrix in the usual Fortran column-oriented format. 3130 3131 Fortran Note: 3132 This routine is used differently from Fortran, e.g., 3133 .vb 3134 Mat mat 3135 Scalar mat_array(1) 3136 PetscOffset i_mat 3137 int ierr 3138 call MatGetArray(mat,mat_array,i_mat,ierr) 3139 3140 C Access first local entry in matrix; note that array is 3141 C treated as one dimensional 3142 value = mat_array(i_mat + 1) 3143 3144 [... other code ...] 3145 call MatRestoreArray(mat,mat_array,i_mat,ierr) 3146 .ve 3147 3148 See the Fortran chapter of the users manual and 3149 petsc/src/mat/examples/tests for details. 3150 3151 Level: advanced 3152 3153 .keywords: matrix, array, elements, values 3154 3155 .seealso: MatRestoreArray(), MatGetArrayF90() 3156 @*/ 3157 int MatGetArray(Mat mat,Scalar **v) 3158 { 3159 int ierr; 3160 3161 PetscFunctionBegin; 3162 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3163 PetscValidPointer(v); 3164 if (!mat->ops->getarray) SETERRQ(PETSC_ERR_SUP,0,""); 3165 ierr = (*mat->ops->getarray)(mat,v);CHKERRQ(ierr); 3166 PetscFunctionReturn(0); 3167 } 3168 3169 #undef __FUNC__ 3170 #define __FUNC__ /*<a name=""></a>*/"MatRestoreArray" 3171 /*@C 3172 MatRestoreArray - Restores the matrix after MatGetArray() has been called. 3173 3174 Not Collective 3175 3176 Input Parameter: 3177 + mat - the matrix 3178 - v - the location of the values 3179 3180 Fortran Note: 3181 This routine is used differently from Fortran, e.g., 3182 .vb 3183 Mat mat 3184 Scalar mat_array(1) 3185 PetscOffset i_mat 3186 int ierr 3187 call MatGetArray(mat,mat_array,i_mat,ierr) 3188 3189 C Access first local entry in matrix; note that array is 3190 C treated as one dimensional 3191 value = mat_array(i_mat + 1) 3192 3193 [... other code ...] 3194 call MatRestoreArray(mat,mat_array,i_mat,ierr) 3195 .ve 3196 3197 See the Fortran chapter of the users manual and 3198 petsc/src/mat/examples/tests for details 3199 3200 Level: advanced 3201 3202 .keywords: matrix, array, elements, values, restore 3203 3204 .seealso: MatGetArray(), MatRestoreArrayF90() 3205 @*/ 3206 int MatRestoreArray(Mat mat,Scalar **v) 3207 { 3208 int ierr; 3209 3210 PetscFunctionBegin; 3211 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3212 PetscValidPointer(v); 3213 if (!mat->ops->restorearray) SETERRQ(PETSC_ERR_SUP,0,""); 3214 ierr = (*mat->ops->restorearray)(mat,v);CHKERRQ(ierr); 3215 PetscFunctionReturn(0); 3216 } 3217 3218 #undef __FUNC__ 3219 #define __FUNC__ /*<a name=""></a>*/"MatGetSubMatrices" 3220 /*@C 3221 MatGetSubMatrices - Extracts several submatrices from a matrix. If submat 3222 points to an array of valid matrices, they may be reused to store the new 3223 submatrices. 3224 3225 Collective on Mat 3226 3227 Input Parameters: 3228 + mat - the matrix 3229 . n - the number of submatrixes to be extracted (on this processor, may be zero) 3230 . irow, icol - index sets of rows and columns to extract 3231 - scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 3232 3233 Output Parameter: 3234 . submat - the array of submatrices 3235 3236 Notes: 3237 MatGetSubMatrices() can extract only sequential submatrices 3238 (from both sequential and parallel matrices). Use MatGetSubMatrix() 3239 to extract a parallel submatrix. 3240 3241 When extracting submatrices from a parallel matrix, each processor can 3242 form a different submatrix by setting the rows and columns of its 3243 individual index sets according to the local submatrix desired. 3244 3245 When finished using the submatrices, the user should destroy 3246 them with MatDestroySubMatrices(). 3247 3248 MAT_REUSE_MATRIX can only be used when the nonzero structure of the 3249 original matrix has not changed from that last call to MatGetSubMatrices(). 3250 3251 Fortran Note: 3252 The Fortran interface is slightly different from that given below; it 3253 requires one to pass in as submat a Mat (integer) array of size at least m. 3254 3255 Level: advanced 3256 3257 .keywords: matrix, get, submatrix, submatrices 3258 3259 .seealso: MatDestroyMatrices(), MatGetSubMatrix(), MatGetRow(), MatGetDiagonal() 3260 @*/ 3261 int MatGetSubMatrices(Mat mat,int n,IS *irow,IS *icol,MatReuse scall,Mat **submat) 3262 { 3263 int ierr; 3264 3265 PetscFunctionBegin; 3266 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3267 if (!mat->ops->getsubmatrices) SETERRQ(PETSC_ERR_SUP,0,""); 3268 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 3269 3270 PLogEventBegin(MAT_GetSubMatrices,mat,0,0,0); 3271 ierr = (*mat->ops->getsubmatrices)(mat,n,irow,icol,scall,submat);CHKERRQ(ierr); 3272 PLogEventEnd(MAT_GetSubMatrices,mat,0,0,0); 3273 3274 PetscFunctionReturn(0); 3275 } 3276 3277 #undef __FUNC__ 3278 #define __FUNC__ /*<a name=""></a>*/"MatDestroyMatrices" 3279 /*@C 3280 MatDestroyMatrices - Destroys a set of matrices obtained with MatGetSubMatrices(). 3281 3282 Collective on Mat 3283 3284 Input Parameters: 3285 + n - the number of local matrices 3286 - mat - the matrices 3287 3288 Level: advanced 3289 3290 .keywords: matrix, destroy, submatrix, submatrices 3291 3292 .seealso: MatGetSubMatrices() 3293 @*/ 3294 int MatDestroyMatrices(int n,Mat **mat) 3295 { 3296 int ierr,i; 3297 3298 PetscFunctionBegin; 3299 if (n < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,1,"Trying to destroy negative number of matrices %d",n); 3300 PetscValidPointer(mat); 3301 for (i=0; i<n; i++) { 3302 ierr = MatDestroy((*mat)[i]);CHKERRQ(ierr); 3303 } 3304 if (n) {ierr = PetscFree(*mat);CHKERRQ(ierr);} 3305 PetscFunctionReturn(0); 3306 } 3307 3308 #undef __FUNC__ 3309 #define __FUNC__ /*<a name=""></a>*/"MatIncreaseOverlap" 3310 /*@ 3311 MatIncreaseOverlap - Given a set of submatrices indicated by index sets, 3312 replaces the index sets by larger ones that represent submatrices with 3313 additional overlap. 3314 3315 Collective on Mat 3316 3317 Input Parameters: 3318 + mat - the matrix 3319 . n - the number of index sets 3320 . is - the array of pointers to index sets 3321 - ov - the additional overlap requested 3322 3323 Level: developer 3324 3325 .keywords: matrix, overlap, Schwarz 3326 3327 .seealso: MatGetSubMatrices() 3328 @*/ 3329 int MatIncreaseOverlap(Mat mat,int n,IS *is,int ov) 3330 { 3331 int ierr; 3332 3333 PetscFunctionBegin; 3334 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3335 if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for unassembled matrix"); 3336 if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,0,"Not for factored matrix"); 3337 3338 if (!ov) PetscFunctionReturn(0); 3339 if (!mat->ops->increaseoverlap) SETERRQ(PETSC_ERR_SUP,0,""); 3340 PLogEventBegin(MAT_IncreaseOverlap,mat,0,0,0); 3341 ierr = (*mat->ops->increaseoverlap)(mat,n,is,ov);CHKERRQ(ierr); 3342 PLogEventEnd(MAT_IncreaseOverlap,mat,0,0,0); 3343 PetscFunctionReturn(0); 3344 } 3345 3346 #undef __FUNC__ 3347 #define __FUNC__ /*<a name=""></a>*/"MatPrintHelp" 3348 /*@ 3349 MatPrintHelp - Prints all the options for the matrix. 3350 3351 Collective on Mat 3352 3353 Input Parameter: 3354 . mat - the matrix 3355 3356 Options Database Keys: 3357 + -help - Prints matrix options 3358 - -h - Prints matrix options 3359 3360 Level: developer 3361 3362 .keywords: mat, help 3363 3364 .seealso: MatCreate(), MatCreateXXX() 3365 @*/ 3366 int MatPrintHelp(Mat mat) 3367 { 3368 static PetscTruth called = PETSC_FALSE; 3369 int ierr; 3370 MPI_Comm comm; 3371 3372 PetscFunctionBegin; 3373 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3374 3375 comm = mat->comm; 3376 if (!called) { 3377 ierr = (*PetscHelpPrintf)(comm,"General matrix options:\n");CHKERRQ(ierr); 3378 ierr = (*PetscHelpPrintf)(comm," -mat_view_info: view basic matrix info during MatAssemblyEnd()\n");CHKERRQ(ierr); 3379 ierr = (*PetscHelpPrintf)(comm," -mat_view_info_detailed: view detailed matrix info during MatAssemblyEnd()\n");CHKERRQ(ierr); 3380 ierr = (*PetscHelpPrintf)(comm," -mat_view_draw: draw nonzero matrix structure during MatAssemblyEnd()\n");CHKERRQ(ierr); 3381 ierr = (*PetscHelpPrintf)(comm," -draw_pause <sec>: set seconds of display pause\n");CHKERRQ(ierr); 3382 ierr = (*PetscHelpPrintf)(comm," -display <name>: set alternate display\n");CHKERRQ(ierr); 3383 called = PETSC_TRUE; 3384 } 3385 if (mat->ops->printhelp) { 3386 ierr = (*mat->ops->printhelp)(mat);CHKERRQ(ierr); 3387 } 3388 PetscFunctionReturn(0); 3389 } 3390 3391 #undef __FUNC__ 3392 #define __FUNC__ /*<a name=""></a>*/"MatGetBlockSize" 3393 /*@ 3394 MatGetBlockSize - Returns the matrix block size; useful especially for the 3395 block row and block diagonal formats. 3396 3397 Not Collective 3398 3399 Input Parameter: 3400 . mat - the matrix 3401 3402 Output Parameter: 3403 . bs - block size 3404 3405 Notes: 3406 Block diagonal formats are MATSEQBDIAG, MATMPIBDIAG. 3407 Block row formats are MATSEQBAIJ, MATMPIBAIJ 3408 3409 Level: intermediate 3410 3411 .keywords: matrix, get, block, size 3412 3413 .seealso: MatCreateSeqBAIJ(), MatCreateMPIBAIJ(), MatCreateSeqBDiag(), MatCreateMPIBDiag() 3414 @*/ 3415 int MatGetBlockSize(Mat mat,int *bs) 3416 { 3417 int ierr; 3418 3419 PetscFunctionBegin; 3420 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3421 PetscValidIntPointer(bs); 3422 if (!mat->ops->getblocksize) SETERRQ(PETSC_ERR_SUP,0,""); 3423 ierr = (*mat->ops->getblocksize)(mat,bs);CHKERRQ(ierr); 3424 PetscFunctionReturn(0); 3425 } 3426 3427 #undef __FUNC__ 3428 #define __FUNC__ /*<a name=""></a>*/"MatGetRowIJ" 3429 /*@C 3430 MatGetRowIJ - Returns the compressed row storage i and j indices for sequential matrices. 3431 3432 Collective on Mat 3433 3434 Input Parameters: 3435 + mat - the matrix 3436 . shift - 0 or 1 indicating we want the indices starting at 0 or 1 3437 - symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 3438 symmetrized 3439 3440 Output Parameters: 3441 + n - number of rows in the (possibly compressed) matrix 3442 . ia - the row pointers 3443 . ja - the column indices 3444 - done - PETSC_TRUE or PETSC_FALSE, indicating whether the values have been returned 3445 3446 Level: developer 3447 3448 .seealso: MatGetColumnIJ(), MatRestoreRowIJ() 3449 @*/ 3450 int MatGetRowIJ(Mat mat,int shift,PetscTruth symmetric,int *n,int **ia,int** ja,PetscTruth *done) 3451 { 3452 int ierr; 3453 3454 PetscFunctionBegin; 3455 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3456 if (ia) PetscValidIntPointer(ia); 3457 if (ja) PetscValidIntPointer(ja); 3458 PetscValidIntPointer(done); 3459 if (!mat->ops->getrowij) *done = PETSC_FALSE; 3460 else { 3461 *done = PETSC_TRUE; 3462 ierr = (*mat->ops->getrowij)(mat,shift,symmetric,n,ia,ja,done);CHKERRQ(ierr); 3463 } 3464 PetscFunctionReturn(0); 3465 } 3466 3467 #undef __FUNC__ 3468 #define __FUNC__ /*<a name=""></a>*/"MatGetColumnIJ" 3469 /*@C 3470 MatGetColumnIJ - Returns the compressed column storage i and j indices for sequential matrices. 3471 3472 Collective on Mat 3473 3474 Input Parameters: 3475 + mat - the matrix 3476 . shift - 1 or zero indicating we want the indices starting at 0 or 1 3477 - symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 3478 symmetrized 3479 3480 Output Parameters: 3481 + n - number of columns in the (possibly compressed) matrix 3482 . ia - the column pointers 3483 . ja - the row indices 3484 - done - PETSC_TRUE or PETSC_FALSE, indicating whether the values have been returned 3485 3486 Level: developer 3487 3488 .seealso: MatGetRowIJ(), MatRestoreColumnIJ() 3489 @*/ 3490 int MatGetColumnIJ(Mat mat,int shift,PetscTruth symmetric,int *n,int **ia,int** ja,PetscTruth *done) 3491 { 3492 int ierr; 3493 3494 PetscFunctionBegin; 3495 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3496 if (ia) PetscValidIntPointer(ia); 3497 if (ja) PetscValidIntPointer(ja); 3498 PetscValidIntPointer(done); 3499 3500 if (!mat->ops->getcolumnij) *done = PETSC_FALSE; 3501 else { 3502 *done = PETSC_TRUE; 3503 ierr = (*mat->ops->getcolumnij)(mat,shift,symmetric,n,ia,ja,done);CHKERRQ(ierr); 3504 } 3505 PetscFunctionReturn(0); 3506 } 3507 3508 #undef __FUNC__ 3509 #define __FUNC__ /*<a name=""></a>*/"MatRestoreRowIJ" 3510 /*@C 3511 MatRestoreRowIJ - Call after you are completed with the ia,ja indices obtained with 3512 MatGetRowIJ(). 3513 3514 Collective on Mat 3515 3516 Input Parameters: 3517 + mat - the matrix 3518 . shift - 1 or zero indicating we want the indices starting at 0 or 1 3519 - symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 3520 symmetrized 3521 3522 Output Parameters: 3523 + n - size of (possibly compressed) matrix 3524 . ia - the row pointers 3525 . ja - the column indices 3526 - done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned 3527 3528 Level: developer 3529 3530 .seealso: MatGetRowIJ(), MatRestoreColumnIJ() 3531 @*/ 3532 int MatRestoreRowIJ(Mat mat,int shift,PetscTruth symmetric,int *n,int **ia,int** ja,PetscTruth *done) 3533 { 3534 int ierr; 3535 3536 PetscFunctionBegin; 3537 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3538 if (ia) PetscValidIntPointer(ia); 3539 if (ja) PetscValidIntPointer(ja); 3540 PetscValidIntPointer(done); 3541 3542 if (!mat->ops->restorerowij) *done = PETSC_FALSE; 3543 else { 3544 *done = PETSC_TRUE; 3545 ierr = (*mat->ops->restorerowij)(mat,shift,symmetric,n,ia,ja,done);CHKERRQ(ierr); 3546 } 3547 PetscFunctionReturn(0); 3548 } 3549 3550 #undef __FUNC__ 3551 #define __FUNC__ /*<a name=""></a>*/"MatRestoreColumnIJ" 3552 /*@C 3553 MatRestoreColumnIJ - Call after you are completed with the ia,ja indices obtained with 3554 MatGetColumnIJ(). 3555 3556 Collective on Mat 3557 3558 Input Parameters: 3559 + mat - the matrix 3560 . shift - 1 or zero indicating we want the indices starting at 0 or 1 3561 - symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be 3562 symmetrized 3563 3564 Output Parameters: 3565 + n - size of (possibly compressed) matrix 3566 . ia - the column pointers 3567 . ja - the row indices 3568 - done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned 3569 3570 Level: developer 3571 3572 .seealso: MatGetColumnIJ(), MatRestoreRowIJ() 3573 @*/ 3574 int MatRestoreColumnIJ(Mat mat,int shift,PetscTruth symmetric,int *n,int **ia,int** ja,PetscTruth *done) 3575 { 3576 int ierr; 3577 3578 PetscFunctionBegin; 3579 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3580 if (ia) PetscValidIntPointer(ia); 3581 if (ja) PetscValidIntPointer(ja); 3582 PetscValidIntPointer(done); 3583 3584 if (!mat->ops->restorecolumnij) *done = PETSC_FALSE; 3585 else { 3586 *done = PETSC_TRUE; 3587 ierr = (*mat->ops->restorecolumnij)(mat,shift,symmetric,n,ia,ja,done);CHKERRQ(ierr); 3588 } 3589 PetscFunctionReturn(0); 3590 } 3591 3592 #undef __FUNC__ 3593 #define __FUNC__ /*<a name=""></a>*/"MatColoringPatch" 3594 /*@C 3595 MatColoringPatch -Used inside matrix coloring routines that 3596 use MatGetRowIJ() and/or MatGetColumnIJ(). 3597 3598 Collective on Mat 3599 3600 Input Parameters: 3601 + mat - the matrix 3602 . n - number of colors 3603 - colorarray - array indicating color for each column 3604 3605 Output Parameters: 3606 . iscoloring - coloring generated using colorarray information 3607 3608 Level: developer 3609 3610 .seealso: MatGetRowIJ(), MatGetColumnIJ() 3611 3612 .keywords: mat, coloring, patch 3613 @*/ 3614 int MatColoringPatch(Mat mat,int n,int *colorarray,ISColoring *iscoloring) 3615 { 3616 int ierr; 3617 3618 PetscFunctionBegin; 3619 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3620 PetscValidIntPointer(colorarray); 3621 3622 if (!mat->ops->coloringpatch) {SETERRQ(PETSC_ERR_SUP,0,"");} 3623 else { 3624 ierr = (*mat->ops->coloringpatch)(mat,n,colorarray,iscoloring);CHKERRQ(ierr); 3625 } 3626 PetscFunctionReturn(0); 3627 } 3628 3629 3630 #undef __FUNC__ 3631 #define __FUNC__ /*<a name=""></a>*/"MatSetUnfactored" 3632 /*@ 3633 MatSetUnfactored - Resets a factored matrix to be treated as unfactored. 3634 3635 Collective on Mat 3636 3637 Input Parameter: 3638 . mat - the factored matrix to be reset 3639 3640 Notes: 3641 This routine should be used only with factored matrices formed by in-place 3642 factorization via ILU(0) (or by in-place LU factorization for the MATSEQDENSE 3643 format). This option can save memory, for example, when solving nonlinear 3644 systems with a matrix-free Newton-Krylov method and a matrix-based, in-place 3645 ILU(0) preconditioner. 3646 3647 Note that one can specify in-place ILU(0) factorization by calling 3648 .vb 3649 PCType(pc,PCILU); 3650 PCILUSeUseInPlace(pc); 3651 .ve 3652 or by using the options -pc_type ilu -pc_ilu_in_place 3653 3654 In-place factorization ILU(0) can also be used as a local 3655 solver for the blocks within the block Jacobi or additive Schwarz 3656 methods (runtime option: -sub_pc_ilu_in_place). See the discussion 3657 of these preconditioners in the users manual for details on setting 3658 local solver options. 3659 3660 Most users should employ the simplified SLES interface for linear solvers 3661 instead of working directly with matrix algebra routines such as this. 3662 See, e.g., SLESCreate(). 3663 3664 Level: developer 3665 3666 .seealso: PCILUSetUseInPlace(), PCLUSetUseInPlace() 3667 3668 .keywords: matrix-free, in-place ILU, in-place LU 3669 @*/ 3670 int MatSetUnfactored(Mat mat) 3671 { 3672 int ierr; 3673 3674 PetscFunctionBegin; 3675 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3676 mat->factor = 0; 3677 if (!mat->ops->setunfactored) PetscFunctionReturn(0); 3678 ierr = (*mat->ops->setunfactored)(mat);CHKERRQ(ierr); 3679 PetscFunctionReturn(0); 3680 } 3681 3682 #undef __FUNC__ 3683 #define __FUNC__ /*<a name=""></a>*/"MatGetType" 3684 /*@C 3685 MatGetType - Gets the matrix type and name (as a string) from the matrix. 3686 3687 Not Collective 3688 3689 Input Parameter: 3690 . mat - the matrix 3691 3692 Output Parameter: 3693 + type - the matrix type (or use PETSC_NULL) 3694 - name - name of matrix type (or use PETSC_NULL) 3695 3696 Level: intermediate 3697 3698 .keywords: matrix, get, type, name 3699 @*/ 3700 int MatGetType(Mat mat,MatType *type,char **name) 3701 { 3702 int itype = (int)mat->type; 3703 char *matname[10]; 3704 3705 PetscFunctionBegin; 3706 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3707 3708 if (type) *type = (MatType) mat->type; 3709 if (name) { 3710 /* Note: Be sure that this list corresponds to the enum in petscmat.h */ 3711 matname[0] = "MATSEQDENSE"; 3712 matname[1] = "MATSEQAIJ"; 3713 matname[2] = "MATMPIAIJ"; 3714 matname[3] = "MATSHELL"; 3715 matname[4] = "MATMPIROWBS"; 3716 matname[5] = "MATSEQBDIAG"; 3717 matname[6] = "MATMPIBDIAG"; 3718 matname[7] = "MATMPIDENSE"; 3719 matname[8] = "MATSEQBAIJ"; 3720 matname[9] = "MATMPIBAIJ"; 3721 3722 if (itype < 0 || itype > 9) *name = "Unknown matrix type"; 3723 else *name = matname[itype]; 3724 } 3725 PetscFunctionReturn(0); 3726 } 3727 3728 /*MC 3729 MatGetArrayF90 - Accesses a matrix array from Fortran90. 3730 3731 Synopsis: 3732 MatGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr) 3733 3734 Not collective 3735 3736 Input Parameter: 3737 . x - matrix 3738 3739 Output Parameters: 3740 + xx_v - the Fortran90 pointer to the array 3741 - ierr - error code 3742 3743 Example of Usage: 3744 .vb 3745 Scalar, pointer xx_v(:) 3746 .... 3747 call MatGetArrayF90(x,xx_v,ierr) 3748 a = xx_v(3) 3749 call MatRestoreArrayF90(x,xx_v,ierr) 3750 .ve 3751 3752 Notes: 3753 Not yet supported for all F90 compilers 3754 3755 Level: advanced 3756 3757 .seealso: MatRestoreArrayF90(), MatGetArray(), MatRestoreArray() 3758 3759 .keywords: matrix, array, f90 3760 M*/ 3761 3762 /*MC 3763 MatRestoreArrayF90 - Restores a matrix array that has been 3764 accessed with MatGetArrayF90(). 3765 3766 Synopsis: 3767 MatRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr) 3768 3769 Not collective 3770 3771 Input Parameters: 3772 + x - matrix 3773 - xx_v - the Fortran90 pointer to the array 3774 3775 Output Parameter: 3776 . ierr - error code 3777 3778 Example of Usage: 3779 .vb 3780 Scalar, pointer xx_v(:) 3781 .... 3782 call MatGetArrayF90(x,xx_v,ierr) 3783 a = xx_v(3) 3784 call MatRestoreArrayF90(x,xx_v,ierr) 3785 .ve 3786 3787 Notes: 3788 Not yet supported for all F90 compilers 3789 3790 Level: advanced 3791 3792 .seealso: MatGetArrayF90(), MatGetArray(), MatRestoreArray() 3793 3794 .keywords: matrix, array, f90 3795 M*/ 3796 3797 3798 #undef __FUNC__ 3799 #define __FUNC__ /*<a name=""></a>*/"MatGetSubMatrix" 3800 /*@ 3801 MatGetSubMatrix - Gets a single submatrix on the same number of processors 3802 as the original matrix. 3803 3804 Collective on Mat 3805 3806 Input Parameters: 3807 + mat - the original matrix 3808 . isrow - rows this processor should obtain 3809 . iscol - columns for all processors you wish to keep 3810 . csize - number of columns "local" to this processor (does nothing for sequential 3811 matrices). This should match the result from VecGetLocalSize(x,...) if you 3812 plan to use the matrix in a A*x; alternatively, you can use PETSC_DECIDE 3813 - cll - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX 3814 3815 Output Parameter: 3816 . newmat - the new submatrix, of the same type as the old 3817 3818 Level: advanced 3819 3820 .keywords: matrix, get, submatrix, submatrices 3821 3822 .seealso: MatGetSubMatrices() 3823 @*/ 3824 int MatGetSubMatrix(Mat mat,IS isrow,IS iscol,int csize,MatReuse cll,Mat *newmat) 3825 { 3826 int ierr, size; 3827 Mat *local; 3828 3829 PetscFunctionBegin; 3830 ierr = MPI_Comm_size(mat->comm,&size);CHKERRQ(ierr); 3831 3832 /* if original matrix is on just one processor then use submatrix generated */ 3833 if (!mat->ops->getsubmatrix && size == 1 && cll == MAT_REUSE_MATRIX) { 3834 ierr = MatGetSubMatrices(mat,1,&isrow,&iscol,MAT_REUSE_MATRIX,&newmat);CHKERRQ(ierr); 3835 PetscFunctionReturn(0); 3836 } else if (!mat->ops->getsubmatrix && size == 1) { 3837 ierr = MatGetSubMatrices(mat,1,&isrow,&iscol,MAT_INITIAL_MATRIX,&local);CHKERRQ(ierr); 3838 *newmat = *local; 3839 ierr = PetscFree(local);CHKERRQ(ierr); 3840 PetscFunctionReturn(0); 3841 } 3842 3843 if (!mat->ops->getsubmatrix) SETERRQ(PETSC_ERR_SUP,0,"Not currently implemented"); 3844 ierr = (*mat->ops->getsubmatrix)(mat,isrow,iscol,csize,cll,newmat);CHKERRQ(ierr); 3845 PetscFunctionReturn(0); 3846 } 3847 3848 #undef __FUNC__ 3849 #define __FUNC__ /*<a name=""></a>*/"MatGetMaps" 3850 /*@C 3851 MatGetMaps - Returns the maps associated with the matrix. 3852 3853 Not Collective 3854 3855 Input Parameter: 3856 . mat - the matrix 3857 3858 Output Parameters: 3859 + rmap - the row (right) map 3860 - cmap - the column (left) map 3861 3862 Level: developer 3863 3864 .keywords: matrix, get, map 3865 @*/ 3866 int MatGetMaps(Mat mat,Map *rmap,Map *cmap) 3867 { 3868 int ierr; 3869 3870 PetscFunctionBegin; 3871 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3872 ierr = (*mat->ops->getmaps)(mat,rmap,cmap);CHKERRQ(ierr); 3873 PetscFunctionReturn(0); 3874 } 3875 3876 /* 3877 Version that works for all PETSc matrices 3878 */ 3879 #undef __FUNC__ 3880 #define __FUNC__ /*<a name=""></a>*/"MatGetMaps_Petsc" 3881 int MatGetMaps_Petsc(Mat mat,Map *rmap,Map *cmap) 3882 { 3883 PetscFunctionBegin; 3884 if (rmap) *rmap = mat->rmap; 3885 if (cmap) *cmap = mat->cmap; 3886 PetscFunctionReturn(0); 3887 } 3888 3889 #undef __FUNC__ 3890 #define __FUNC__ /*<a name=""></a>*/"MatSetStashInitialSize" 3891 /*@ 3892 MatSetStashInitialSize - sets the sizes of the matrix stash, that is 3893 used during the assembly process to store values that belong to 3894 other processors. 3895 3896 Not Collective 3897 3898 Input Parameters: 3899 + mat - the matrix 3900 . size - the initial size of the stash. 3901 - bsize - the initial size of the block-stash(if used). 3902 3903 Options Database Keys: 3904 + -matstash_initial_size <size> or <size0,size1,...sizep-1> 3905 - -matstash_block_initial_size <bsize> or <bsize0,bsize1,...bsizep-1> 3906 3907 Level: intermediate 3908 3909 Notes: 3910 The block-stash is used for values set with VecSetValuesBlocked() while 3911 the stash is used for values set with VecSetValues() 3912 3913 Run with the option -log_info and look for output of the form 3914 MatAssemblyBegin_MPIXXX:Stash has MM entries, uses nn mallocs. 3915 to determine the appropriate value, MM, to use for size and 3916 MatAssemblyBegin_MPIXXX:Block-Stash has BMM entries, uses nn mallocs. 3917 to determine the value, BMM to use for bsize 3918 3919 .keywords: matrix, stash, assembly 3920 @*/ 3921 int MatSetStashInitialSize(Mat mat,int size, int bsize) 3922 { 3923 int ierr; 3924 3925 PetscFunctionBegin; 3926 PetscValidHeaderSpecific(mat,MAT_COOKIE); 3927 ierr = MatStashSetInitialSize_Private(&mat->stash,size);CHKERRQ(ierr); 3928 ierr = MatStashSetInitialSize_Private(&mat->bstash,bsize);CHKERRQ(ierr); 3929 PetscFunctionReturn(0); 3930 } 3931 3932 #undef __FUNC__ 3933 #define __FUNC__ /*<a name=""></a>*/"MatInterpolateAdd" 3934 /*@ 3935 MatInterpolateAdd - w = y + A*x or A'*x depending on the shape of 3936 the matrix 3937 3938 Collective on Mat 3939 3940 Input Parameters: 3941 + mat - the matrix 3942 . x,y - the vectors 3943 - w - where the result is stored 3944 3945 Level: intermediate 3946 3947 Notes: 3948 w may be the same vector as y. 3949 3950 This allows one to use either the restriction or interpolation (its transpose) 3951 matrix to do the interpolation 3952 3953 .keywords: interpolate, 3954 3955 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict() 3956 3957 @*/ 3958 int MatInterpolateAdd(Mat A,Vec x,Vec y,Vec w) 3959 { 3960 int M,N,ierr; 3961 3962 PetscFunctionBegin; 3963 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 3964 if (N > M) { 3965 ierr = MatMultTransposeAdd(A,x,y,w);CHKERRQ(ierr); 3966 } else { 3967 ierr = MatMultAdd(A,x,y,w);CHKERRQ(ierr); 3968 } 3969 PetscFunctionReturn(0); 3970 } 3971 3972 #undef __FUNC__ 3973 #define __FUNC__ /*<a name=""></a>*/"MatInterpolate" 3974 /*@ 3975 MatInterpolate - y = A*x or A'*x depending on the shape of 3976 the matrix 3977 3978 Collective on Mat 3979 3980 Input Parameters: 3981 + mat - the matrix 3982 - x,y - the vectors 3983 3984 Level: intermediate 3985 3986 Notes: 3987 This allows one to use either the restriction or interpolation (its transpose) 3988 matrix to do the interpolation 3989 3990 .keywords: interpolate, 3991 3992 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict() 3993 3994 @*/ 3995 int MatInterpolate(Mat A,Vec x,Vec y) 3996 { 3997 int M,N,ierr; 3998 3999 PetscFunctionBegin; 4000 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 4001 if (N > M) { 4002 ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr); 4003 } else { 4004 ierr = MatMult(A,x,y);CHKERRQ(ierr); 4005 } 4006 PetscFunctionReturn(0); 4007 } 4008 4009 #undef __FUNC__ 4010 #define __FUNC__ /*<a name=""></a>*/"MatRestrict" 4011 /*@ 4012 MatRestrict - y = A*x or A'*x 4013 4014 Collective on Mat 4015 4016 Input Parameters: 4017 + mat - the matrix 4018 - x,y - the vectors 4019 4020 Level: intermediate 4021 4022 Notes: 4023 This allows one to use either the restriction or interpolation (its transpose) 4024 matrix to do the restriction 4025 4026 .keywords: interpolate, 4027 4028 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatInterpolate() 4029 4030 @*/ 4031 int MatRestrict(Mat A,Vec x,Vec y) 4032 { 4033 int M,N,ierr; 4034 4035 PetscFunctionBegin; 4036 ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr); 4037 if (N > M) { 4038 ierr = MatMult(A,x,y);CHKERRQ(ierr); 4039 } else { 4040 ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr); 4041 } 4042 PetscFunctionReturn(0); 4043 } 4044