xref: /petsc/src/mat/interface/matrix.c (revision 4cb17eb5d24fda53f1a75322daaa24f590374b40)
1 #define PETSCMAT_DLL
2 
3 /*
4    This is where the abstract matrix operations are defined
5 */
6 
7 #include "private/matimpl.h"        /*I "petscmat.h" I*/
8 #include "private/vecimpl.h"
9 
10 /* Logging support */
11 PetscClassId PETSCMAT_DLLEXPORT MAT_CLASSID;
12 PetscClassId PETSCMAT_DLLEXPORT MAT_FDCOLORING_CLASSID;
13 
14 PetscLogEvent  MAT_Mult, MAT_Mults, MAT_MultConstrained, MAT_MultAdd, MAT_MultTranspose;
15 PetscLogEvent  MAT_MultTransposeConstrained, MAT_MultTransposeAdd, MAT_Solve, MAT_Solves, MAT_SolveAdd, MAT_SolveTranspose, MAT_MatSolve;
16 PetscLogEvent  MAT_SolveTransposeAdd, MAT_SOR, MAT_ForwardSolve, MAT_BackwardSolve, MAT_LUFactor, MAT_LUFactorSymbolic;
17 PetscLogEvent  MAT_LUFactorNumeric, MAT_CholeskyFactor, MAT_CholeskyFactorSymbolic, MAT_CholeskyFactorNumeric, MAT_ILUFactor;
18 PetscLogEvent  MAT_ILUFactorSymbolic, MAT_ICCFactorSymbolic, MAT_Copy, MAT_Convert, MAT_Scale, MAT_AssemblyBegin;
19 PetscLogEvent  MAT_AssemblyEnd, MAT_SetValues, MAT_GetValues, MAT_GetRow, MAT_GetRowIJ, MAT_GetSubMatrices, MAT_GetColoring, MAT_GetOrdering, MAT_GetRedundantMatrix, MAT_GetSeqNonzeroStructure;
20 PetscLogEvent  MAT_IncreaseOverlap, MAT_Partitioning, MAT_ZeroEntries, MAT_Load, MAT_View, MAT_AXPY, MAT_FDColoringCreate;
21 PetscLogEvent  MAT_FDColoringApply,MAT_Transpose,MAT_FDColoringFunction;
22 PetscLogEvent  MAT_MatMult, MAT_MatMultSymbolic, MAT_MatMultNumeric;
23 PetscLogEvent  MAT_PtAP, MAT_PtAPSymbolic, MAT_PtAPNumeric;
24 PetscLogEvent  MAT_MatMultTranspose, MAT_MatMultTransposeSymbolic, MAT_MatMultTransposeNumeric;
25 PetscLogEvent  MAT_MultHermitianTranspose,MAT_MultHermitianTransposeAdd;
26 PetscLogEvent  MAT_Getsymtranspose, MAT_Getsymtransreduced, MAT_Transpose_SeqAIJ, MAT_GetBrowsOfAcols;
27 PetscLogEvent  MAT_GetBrowsOfAocols, MAT_Getlocalmat, MAT_Getlocalmatcondensed, MAT_Seqstompi, MAT_Seqstompinum, MAT_Seqstompisym;
28 PetscLogEvent  MAT_Applypapt, MAT_Applypapt_numeric, MAT_Applypapt_symbolic, MAT_GetSequentialNonzeroStructure;
29 PetscLogEvent  MAT_GetMultiProcBlock;
30 
31 /* nasty global values for MatSetValue() */
32 PetscInt    PETSCMAT_DLLEXPORT MatSetValue_Row = 0;
33 PetscInt    PETSCMAT_DLLEXPORT MatSetValue_Column = 0;
34 PetscScalar PETSCMAT_DLLEXPORT MatSetValue_Value = 0.0;
35 
36 const char *const MatFactorTypes[] = {"NONE","LU","CHOLESKY","ILU","ICC","ILUDT","MatFactorType","MAT_FACTOR_",0};
37 
38 #undef __FUNCT__
39 #define __FUNCT__ "MatGetDiagonalBlock"
40 /*@
41    MatGetDiagonalBlock - Returns the part of the matrix associated with the on-process coupling
42 
43    Not Collective
44 
45    Input Parameters:
46 +  mat - the matrix
47 -  reuse - indicates you are passing in the a matrix and want it reused
48 
49    Output Parameters:
50 +   iscopy - indicates a copy of the diagonal matrix was created and you should use MatDestroy() on it
51 -   a - the diagonal part (which is a SEQUENTIAL matrix)
52 
53    Notes: see the manual page for MatCreateMPIAIJ() for more information on the "diagonal part" of the matrix
54 
55    Level: advanced
56 
57 @*/
58 PetscErrorCode PETSCMAT_DLLEXPORT MatGetDiagonalBlock(Mat A,PetscTruth *iscopy,MatReuse reuse,Mat *a)
59 {
60   PetscErrorCode ierr,(*f)(Mat,PetscTruth*,MatReuse,Mat*);
61   PetscMPIInt    size;
62 
63   PetscFunctionBegin;
64   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
65   PetscValidType(A,1);
66   PetscValidPointer(iscopy,2);
67   PetscValidPointer(a,3);
68   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
69   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
70   ierr = MPI_Comm_size(((PetscObject)A)->comm,&size);CHKERRQ(ierr);
71   ierr = PetscObjectQueryFunction((PetscObject)A,"MatGetDiagonalBlock_C",(void (**)(void))&f);CHKERRQ(ierr);
72   if (f) {
73     ierr = (*f)(A,iscopy,reuse,a);CHKERRQ(ierr);
74   } else if (size == 1) {
75     *a = A;
76     *iscopy = PETSC_FALSE;
77   } else SETERRQ(((PetscObject)A)->comm,PETSC_ERR_SUP,"Cannot get diagonal part for this matrix");
78   PetscFunctionReturn(0);
79 }
80 
81 #undef __FUNCT__
82 #define __FUNCT__ "MatGetTrace"
83 /*@
84    MatGetTrace - Gets the trace of a matrix. The sum of the diagonal entries.
85 
86    Collective on Mat
87 
88    Input Parameters:
89 .  mat - the matrix
90 
91    Output Parameter:
92 .   trace - the sum of the diagonal entries
93 
94    Level: advanced
95 
96 @*/
97 PetscErrorCode  MatGetTrace(Mat mat,PetscScalar *trace)
98 {
99    PetscErrorCode ierr;
100    Vec            diag;
101 
102    PetscFunctionBegin;
103    ierr = MatGetVecs(mat,&diag,PETSC_NULL);CHKERRQ(ierr);
104    ierr = MatGetDiagonal(mat,diag);CHKERRQ(ierr);
105    ierr = VecSum(diag,trace);CHKERRQ(ierr);
106    ierr = VecDestroy(diag);CHKERRQ(ierr);
107    PetscFunctionReturn(0);
108 }
109 
110 #undef __FUNCT__
111 #define __FUNCT__ "MatRealPart"
112 /*@
113    MatRealPart - Zeros out the imaginary part of the matrix
114 
115    Logically Collective on Mat
116 
117    Input Parameters:
118 .  mat - the matrix
119 
120    Level: advanced
121 
122 
123 .seealso: MatImaginaryPart()
124 @*/
125 PetscErrorCode PETSCMAT_DLLEXPORT MatRealPart(Mat mat)
126 {
127   PetscErrorCode ierr;
128 
129   PetscFunctionBegin;
130   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
131   PetscValidType(mat,1);
132   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
133   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
134   if (!mat->ops->realpart) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
135   ierr = MatPreallocated(mat);CHKERRQ(ierr);
136   ierr = (*mat->ops->realpart)(mat);CHKERRQ(ierr);
137   PetscFunctionReturn(0);
138 }
139 
140 #undef __FUNCT__
141 #define __FUNCT__ "MatGetGhosts"
142 /*@C
143    MatGetGhosts - Get the global index of all ghost nodes defined by the sparse matrix
144 
145    Collective on Mat
146 
147    Input Parameter:
148 .  mat - the matrix
149 
150    Output Parameters:
151 +   nghosts - number of ghosts (note for BAIJ matrices there is one ghost for each block)
152 -   ghosts - the global indices of the ghost points
153 
154    Notes: the nghosts and ghosts are suitable to pass into VecCreateGhost()
155 
156    Level: advanced
157 
158 @*/
159 PetscErrorCode PETSCMAT_DLLEXPORT MatGetGhosts(Mat mat,PetscInt *nghosts,const PetscInt *ghosts[])
160 {
161   PetscErrorCode ierr;
162 
163   PetscFunctionBegin;
164   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
165   PetscValidType(mat,1);
166   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
167   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
168   if (!mat->ops->getghosts) {
169     if (nghosts) *nghosts = 0;
170     if (ghosts) *ghosts = 0;
171   } else {
172     ierr = (*mat->ops->getghosts)(mat,nghosts,ghosts);CHKERRQ(ierr);
173   }
174   PetscFunctionReturn(0);
175 }
176 
177 
178 #undef __FUNCT__
179 #define __FUNCT__ "MatImaginaryPart"
180 /*@
181    MatImaginaryPart - Moves the imaginary part of the matrix to the real part and zeros the imaginary part
182 
183    Logically Collective on Mat
184 
185    Input Parameters:
186 .  mat - the matrix
187 
188    Level: advanced
189 
190 
191 .seealso: MatRealPart()
192 @*/
193 PetscErrorCode PETSCMAT_DLLEXPORT MatImaginaryPart(Mat mat)
194 {
195   PetscErrorCode ierr;
196 
197   PetscFunctionBegin;
198   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
199   PetscValidType(mat,1);
200   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
201   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
202   if (!mat->ops->imaginarypart) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
203   ierr = MatPreallocated(mat);CHKERRQ(ierr);
204   ierr = (*mat->ops->imaginarypart)(mat);CHKERRQ(ierr);
205   PetscFunctionReturn(0);
206 }
207 
208 #undef __FUNCT__
209 #define __FUNCT__ "MatMissingDiagonal"
210 /*@
211    MatMissingDiagonal - Determine if sparse matrix is missing a diagonal entry (or block entry for BAIJ matrices)
212 
213    Collective on Mat
214 
215    Input Parameter:
216 .  mat - the matrix
217 
218    Output Parameters:
219 +  missing - is any diagonal missing
220 -  dd - first diagonal entry that is missing (optional)
221 
222    Level: advanced
223 
224 
225 .seealso: MatRealPart()
226 @*/
227 PetscErrorCode PETSCMAT_DLLEXPORT MatMissingDiagonal(Mat mat,PetscTruth *missing,PetscInt *dd)
228 {
229   PetscErrorCode ierr;
230 
231   PetscFunctionBegin;
232   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
233   PetscValidType(mat,1);
234   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
235   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
236   if (!mat->ops->missingdiagonal) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
237   ierr = (*mat->ops->missingdiagonal)(mat,missing,dd);CHKERRQ(ierr);
238   PetscFunctionReturn(0);
239 }
240 
241 #undef __FUNCT__
242 #define __FUNCT__ "MatGetRow"
243 /*@C
244    MatGetRow - Gets a row of a matrix.  You MUST call MatRestoreRow()
245    for each row that you get to ensure that your application does
246    not bleed memory.
247 
248    Not Collective
249 
250    Input Parameters:
251 +  mat - the matrix
252 -  row - the row to get
253 
254    Output Parameters:
255 +  ncols -  if not NULL, the number of nonzeros in the row
256 .  cols - if not NULL, the column numbers
257 -  vals - if not NULL, the values
258 
259    Notes:
260    This routine is provided for people who need to have direct access
261    to the structure of a matrix.  We hope that we provide enough
262    high-level matrix routines that few users will need it.
263 
264    MatGetRow() always returns 0-based column indices, regardless of
265    whether the internal representation is 0-based (default) or 1-based.
266 
267    For better efficiency, set cols and/or vals to PETSC_NULL if you do
268    not wish to extract these quantities.
269 
270    The user can only examine the values extracted with MatGetRow();
271    the values cannot be altered.  To change the matrix entries, one
272    must use MatSetValues().
273 
274    You can only have one call to MatGetRow() outstanding for a particular
275    matrix at a time, per processor. MatGetRow() can only obtain rows
276    associated with the given processor, it cannot get rows from the
277    other processors; for that we suggest using MatGetSubMatrices(), then
278    MatGetRow() on the submatrix. The row indix passed to MatGetRows()
279    is in the global number of rows.
280 
281    Fortran Notes:
282    The calling sequence from Fortran is
283 .vb
284    MatGetRow(matrix,row,ncols,cols,values,ierr)
285          Mat     matrix (input)
286          integer row    (input)
287          integer ncols  (output)
288          integer cols(maxcols) (output)
289          double precision (or double complex) values(maxcols) output
290 .ve
291    where maxcols >= maximum nonzeros in any row of the matrix.
292 
293 
294    Caution:
295    Do not try to change the contents of the output arrays (cols and vals).
296    In some cases, this may corrupt the matrix.
297 
298    Level: advanced
299 
300    Concepts: matrices^row access
301 
302 .seealso: MatRestoreRow(), MatSetValues(), MatGetValues(), MatGetSubMatrices(), MatGetDiagonal()
303 @*/
304 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[])
305 {
306   PetscErrorCode ierr;
307   PetscInt       incols;
308 
309   PetscFunctionBegin;
310   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
311   PetscValidType(mat,1);
312   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
313   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
314   if (!mat->ops->getrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
315   ierr = MatPreallocated(mat);CHKERRQ(ierr);
316   ierr = PetscLogEventBegin(MAT_GetRow,mat,0,0,0);CHKERRQ(ierr);
317   ierr = (*mat->ops->getrow)(mat,row,&incols,(PetscInt **)cols,(PetscScalar **)vals);CHKERRQ(ierr);
318   if (ncols) *ncols = incols;
319   ierr = PetscLogEventEnd(MAT_GetRow,mat,0,0,0);CHKERRQ(ierr);
320   PetscFunctionReturn(0);
321 }
322 
323 #undef __FUNCT__
324 #define __FUNCT__ "MatConjugate"
325 /*@
326    MatConjugate - replaces the matrix values with their complex conjugates
327 
328    Logically Collective on Mat
329 
330    Input Parameters:
331 .  mat - the matrix
332 
333    Level: advanced
334 
335 .seealso:  VecConjugate()
336 @*/
337 PetscErrorCode PETSCMAT_DLLEXPORT MatConjugate(Mat mat)
338 {
339   PetscErrorCode ierr;
340 
341   PetscFunctionBegin;
342   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
343   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
344   if (!mat->ops->conjugate) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Not provided for this matrix format, send email to petsc-maint@mcs.anl.gov");
345   ierr = (*mat->ops->conjugate)(mat);CHKERRQ(ierr);
346   PetscFunctionReturn(0);
347 }
348 
349 #undef __FUNCT__
350 #define __FUNCT__ "MatRestoreRow"
351 /*@C
352    MatRestoreRow - Frees any temporary space allocated by MatGetRow().
353 
354    Not Collective
355 
356    Input Parameters:
357 +  mat - the matrix
358 .  row - the row to get
359 .  ncols, cols - the number of nonzeros and their columns
360 -  vals - if nonzero the column values
361 
362    Notes:
363    This routine should be called after you have finished examining the entries.
364 
365    Fortran Notes:
366    The calling sequence from Fortran is
367 .vb
368    MatRestoreRow(matrix,row,ncols,cols,values,ierr)
369       Mat     matrix (input)
370       integer row    (input)
371       integer ncols  (output)
372       integer cols(maxcols) (output)
373       double precision (or double complex) values(maxcols) output
374 .ve
375    Where maxcols >= maximum nonzeros in any row of the matrix.
376 
377    In Fortran MatRestoreRow() MUST be called after MatGetRow()
378    before another call to MatGetRow() can be made.
379 
380    Level: advanced
381 
382 .seealso:  MatGetRow()
383 @*/
384 PetscErrorCode PETSCMAT_DLLEXPORT MatRestoreRow(Mat mat,PetscInt row,PetscInt *ncols,const PetscInt *cols[],const PetscScalar *vals[])
385 {
386   PetscErrorCode ierr;
387 
388   PetscFunctionBegin;
389   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
390   PetscValidIntPointer(ncols,3);
391   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
392   if (!mat->ops->restorerow) PetscFunctionReturn(0);
393   ierr = (*mat->ops->restorerow)(mat,row,ncols,(PetscInt **)cols,(PetscScalar **)vals);CHKERRQ(ierr);
394   PetscFunctionReturn(0);
395 }
396 
397 #undef __FUNCT__
398 #define __FUNCT__ "MatGetRowUpperTriangular"
399 /*@
400    MatGetRowUpperTriangular - Sets a flag to enable calls to MatGetRow() for matrix in MATSBAIJ format.
401    You should call MatRestoreRowUpperTriangular() after calling MatGetRow/MatRestoreRow() to disable the flag.
402 
403    Not Collective
404 
405    Input Parameters:
406 +  mat - the matrix
407 
408    Notes:
409    The flag is to ensure that users are aware of MatGetRow() only provides the upper trianglular part of the row for the matrices in MATSBAIJ format.
410 
411    Level: advanced
412 
413    Concepts: matrices^row access
414 
415 .seealso: MatRestoreRowRowUpperTriangular()
416 @*/
417 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowUpperTriangular(Mat mat)
418 {
419   PetscErrorCode ierr;
420 
421   PetscFunctionBegin;
422   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
423   PetscValidType(mat,1);
424   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
425   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
426   if (!mat->ops->getrowuppertriangular) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
427   ierr = MatPreallocated(mat);CHKERRQ(ierr);
428   ierr = (*mat->ops->getrowuppertriangular)(mat);CHKERRQ(ierr);
429   PetscFunctionReturn(0);
430 }
431 
432 #undef __FUNCT__
433 #define __FUNCT__ "MatRestoreRowUpperTriangular"
434 /*@
435    MatRestoreRowUpperTriangular - Disable calls to MatGetRow() for matrix in MATSBAIJ format.
436 
437    Not Collective
438 
439    Input Parameters:
440 +  mat - the matrix
441 
442    Notes:
443    This routine should be called after you have finished MatGetRow/MatRestoreRow().
444 
445 
446    Level: advanced
447 
448 .seealso:  MatGetRowUpperTriangular()
449 @*/
450 PetscErrorCode PETSCMAT_DLLEXPORT MatRestoreRowUpperTriangular(Mat mat)
451 {
452   PetscErrorCode ierr;
453 
454   PetscFunctionBegin;
455   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
456   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
457   if (!mat->ops->restorerowuppertriangular) PetscFunctionReturn(0);
458   ierr = (*mat->ops->restorerowuppertriangular)(mat);CHKERRQ(ierr);
459   PetscFunctionReturn(0);
460 }
461 
462 #undef __FUNCT__
463 #define __FUNCT__ "MatSetOptionsPrefix"
464 /*@C
465    MatSetOptionsPrefix - Sets the prefix used for searching for all
466    Mat options in the database.
467 
468    Logically Collective on Mat
469 
470    Input Parameter:
471 +  A - the Mat context
472 -  prefix - the prefix to prepend to all option names
473 
474    Notes:
475    A hyphen (-) must NOT be given at the beginning of the prefix name.
476    The first character of all runtime options is AUTOMATICALLY the hyphen.
477 
478    Level: advanced
479 
480 .keywords: Mat, set, options, prefix, database
481 
482 .seealso: MatSetFromOptions()
483 @*/
484 PetscErrorCode PETSCMAT_DLLEXPORT MatSetOptionsPrefix(Mat A,const char prefix[])
485 {
486   PetscErrorCode ierr;
487 
488   PetscFunctionBegin;
489   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
490   ierr = PetscObjectSetOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr);
491   PetscFunctionReturn(0);
492 }
493 
494 #undef __FUNCT__
495 #define __FUNCT__ "MatAppendOptionsPrefix"
496 /*@C
497    MatAppendOptionsPrefix - Appends to the prefix used for searching for all
498    Mat options in the database.
499 
500    Logically Collective on Mat
501 
502    Input Parameters:
503 +  A - the Mat context
504 -  prefix - the prefix to prepend to all option names
505 
506    Notes:
507    A hyphen (-) must NOT be given at the beginning of the prefix name.
508    The first character of all runtime options is AUTOMATICALLY the hyphen.
509 
510    Level: advanced
511 
512 .keywords: Mat, append, options, prefix, database
513 
514 .seealso: MatGetOptionsPrefix()
515 @*/
516 PetscErrorCode PETSCMAT_DLLEXPORT MatAppendOptionsPrefix(Mat A,const char prefix[])
517 {
518   PetscErrorCode ierr;
519 
520   PetscFunctionBegin;
521   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
522   ierr = PetscObjectAppendOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr);
523   PetscFunctionReturn(0);
524 }
525 
526 #undef __FUNCT__
527 #define __FUNCT__ "MatGetOptionsPrefix"
528 /*@C
529    MatGetOptionsPrefix - Sets the prefix used for searching for all
530    Mat options in the database.
531 
532    Not Collective
533 
534    Input Parameter:
535 .  A - the Mat context
536 
537    Output Parameter:
538 .  prefix - pointer to the prefix string used
539 
540    Notes: On the fortran side, the user should pass in a string 'prefix' of
541    sufficient length to hold the prefix.
542 
543    Level: advanced
544 
545 .keywords: Mat, get, options, prefix, database
546 
547 .seealso: MatAppendOptionsPrefix()
548 @*/
549 PetscErrorCode PETSCMAT_DLLEXPORT MatGetOptionsPrefix(Mat A,const char *prefix[])
550 {
551   PetscErrorCode ierr;
552 
553   PetscFunctionBegin;
554   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
555   ierr = PetscObjectGetOptionsPrefix((PetscObject)A,prefix);CHKERRQ(ierr);
556   PetscFunctionReturn(0);
557 }
558 
559 #undef __FUNCT__
560 #define __FUNCT__ "MatSetUp"
561 /*@
562    MatSetUp - Sets up the internal matrix data structures for the later use.
563 
564    Collective on Mat
565 
566    Input Parameters:
567 .  A - the Mat context
568 
569    Notes:
570    For basic use of the Mat classes the user need not explicitly call
571    MatSetUp(), since these actions will happen automatically.
572 
573    Level: advanced
574 
575 .keywords: Mat, setup
576 
577 .seealso: MatCreate(), MatDestroy()
578 @*/
579 PetscErrorCode PETSCMAT_DLLEXPORT MatSetUp(Mat A)
580 {
581   PetscMPIInt    size;
582   PetscErrorCode ierr;
583 
584   PetscFunctionBegin;
585   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
586   if (!((PetscObject)A)->type_name) {
587     ierr = MPI_Comm_size(((PetscObject)A)->comm, &size);CHKERRQ(ierr);
588     if (size == 1) {
589       ierr = MatSetType(A, MATSEQAIJ);CHKERRQ(ierr);
590     } else {
591       ierr = MatSetType(A, MATMPIAIJ);CHKERRQ(ierr);
592     }
593   }
594   ierr = MatSetUpPreallocation(A);CHKERRQ(ierr);
595   PetscFunctionReturn(0);
596 }
597 
598 #undef __FUNCT__
599 #define __FUNCT__ "MatView"
600 /*@C
601    MatView - Visualizes a matrix object.
602 
603    Collective on Mat
604 
605    Input Parameters:
606 +  mat - the matrix
607 -  viewer - visualization context
608 
609   Notes:
610   The available visualization contexts include
611 +    PETSC_VIEWER_STDOUT_SELF - standard output (default)
612 .    PETSC_VIEWER_STDOUT_WORLD - synchronized standard
613         output where only the first processor opens
614         the file.  All other processors send their
615         data to the first processor to print.
616 -     PETSC_VIEWER_DRAW_WORLD - graphical display of nonzero structure
617 
618    The user can open alternative visualization contexts with
619 +    PetscViewerASCIIOpen() - Outputs matrix to a specified file
620 .    PetscViewerBinaryOpen() - Outputs matrix in binary to a
621          specified file; corresponding input uses MatLoad()
622 .    PetscViewerDrawOpen() - Outputs nonzero matrix structure to
623          an X window display
624 -    PetscViewerSocketOpen() - Outputs matrix to Socket viewer.
625          Currently only the sequential dense and AIJ
626          matrix types support the Socket viewer.
627 
628    The user can call PetscViewerSetFormat() to specify the output
629    format of ASCII printed objects (when using PETSC_VIEWER_STDOUT_SELF,
630    PETSC_VIEWER_STDOUT_WORLD and PetscViewerASCIIOpen).  Available formats include
631 +    PETSC_VIEWER_DEFAULT - default, prints matrix contents
632 .    PETSC_VIEWER_ASCII_MATLAB - prints matrix contents in Matlab format
633 .    PETSC_VIEWER_ASCII_DENSE - prints entire matrix including zeros
634 .    PETSC_VIEWER_ASCII_COMMON - prints matrix contents, using a sparse
635          format common among all matrix types
636 .    PETSC_VIEWER_ASCII_IMPL - prints matrix contents, using an implementation-specific
637          format (which is in many cases the same as the default)
638 .    PETSC_VIEWER_ASCII_INFO - prints basic information about the matrix
639          size and structure (not the matrix entries)
640 .    PETSC_VIEWER_ASCII_INFO_DETAIL - prints more detailed information about
641          the matrix structure
642 
643    Options Database Keys:
644 +  -mat_view_info - Prints info on matrix at conclusion of MatEndAssembly()
645 .  -mat_view_info_detailed - Prints more detailed info
646 .  -mat_view - Prints matrix in ASCII format
647 .  -mat_view_matlab - Prints matrix in Matlab format
648 .  -mat_view_draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX().
649 .  -display <name> - Sets display name (default is host)
650 .  -draw_pause <sec> - Sets number of seconds to pause after display
651 .  -mat_view_socket - Sends matrix to socket, can be accessed from Matlab (see the <a href="../../docs/manual.pdf">users manual</a> for details).
652 .  -viewer_socket_machine <machine>
653 .  -viewer_socket_port <port>
654 .  -mat_view_binary - save matrix to file in binary format
655 -  -viewer_binary_filename <name>
656    Level: beginner
657 
658    Notes: see the manual page for MatLoad() for the exact format of the binary file when the binary
659       viewer is used.
660 
661       See bin/matlab/PetscBinaryRead.m for a Matlab code that can read in the binary file when the binary
662       viewer is used.
663 
664    Concepts: matrices^viewing
665    Concepts: matrices^plotting
666    Concepts: matrices^printing
667 
668 .seealso: PetscViewerSetFormat(), PetscViewerASCIIOpen(), PetscViewerDrawOpen(),
669           PetscViewerSocketOpen(), PetscViewerBinaryOpen(), MatLoad()
670 @*/
671 PetscErrorCode PETSCMAT_DLLEXPORT MatView(Mat mat,PetscViewer viewer)
672 {
673   PetscErrorCode    ierr;
674   PetscInt          rows,cols;
675   PetscTruth        iascii;
676   const MatType     cstr;
677   PetscViewerFormat format;
678 
679   PetscFunctionBegin;
680   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
681   PetscValidType(mat,1);
682   if (!viewer) {
683     ierr = PetscViewerASCIIGetStdout(((PetscObject)mat)->comm,&viewer);CHKERRQ(ierr);
684   }
685   PetscValidHeaderSpecific(viewer,PETSC_VIEWER_CLASSID,2);
686   PetscCheckSameComm(mat,1,viewer,2);
687   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ORDER,"Must call MatAssemblyBegin/End() before viewing matrix");
688   ierr = MatPreallocated(mat);CHKERRQ(ierr);
689 
690   ierr = PetscLogEventBegin(MAT_View,mat,viewer,0,0);CHKERRQ(ierr);
691   ierr = PetscTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr);
692   if (iascii) {
693     ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr);
694     if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
695       if (((PetscObject)mat)->prefix) {
696         ierr = PetscViewerASCIIPrintf(viewer,"Matrix Object:(%s)\n",((PetscObject)mat)->prefix);CHKERRQ(ierr);
697       } else {
698         ierr = PetscViewerASCIIPrintf(viewer,"Matrix Object:\n");CHKERRQ(ierr);
699       }
700       ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
701       ierr = MatGetType(mat,&cstr);CHKERRQ(ierr);
702       ierr = MatGetSize(mat,&rows,&cols);CHKERRQ(ierr);
703       ierr = PetscViewerASCIIPrintf(viewer,"type=%s, rows=%D, cols=%D\n",cstr,rows,cols);CHKERRQ(ierr);
704       if (mat->factortype) {
705         const MatSolverPackage solver;
706         ierr = MatFactorGetSolverPackage(mat,&solver);CHKERRQ(ierr);
707         ierr = PetscViewerASCIIPrintf(viewer,"package used to perform factorization: %s\n",solver);CHKERRQ(ierr);
708       }
709       if (mat->ops->getinfo) {
710         MatInfo info;
711         ierr = MatGetInfo(mat,MAT_GLOBAL_SUM,&info);CHKERRQ(ierr);
712         ierr = PetscViewerASCIIPrintf(viewer,"total: nonzeros=%D, allocated nonzeros=%D\n",(PetscInt)info.nz_used,(PetscInt)info.nz_allocated);CHKERRQ(ierr);
713         ierr = PetscViewerASCIIPrintf(viewer,"total number of mallocs used during MatSetValues calls =%D\n",(PetscInt)info.mallocs);CHKERRQ(ierr);
714       }
715     }
716   }
717   if (mat->ops->view) {
718     ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr);
719     ierr = (*mat->ops->view)(mat,viewer);CHKERRQ(ierr);
720     ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
721   } else if (!iascii) {
722     SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Viewer type %s not supported",((PetscObject)viewer)->type_name);
723   }
724   if (iascii) {
725     ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr);
726     if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
727       ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr);
728     }
729   }
730   ierr = PetscLogEventEnd(MAT_View,mat,viewer,0,0);CHKERRQ(ierr);
731   PetscFunctionReturn(0);
732 }
733 
734 #if defined(PETSC_USE_DEBUG)
735 #include "../src/sys/totalview/tv_data_display.h"
736 PETSC_UNUSED static int TV_display_type(const struct _p_Mat *mat)
737 {
738   TV_add_row("Local rows", "int", &mat->rmap->n);
739   TV_add_row("Local columns", "int", &mat->cmap->n);
740   TV_add_row("Global rows", "int", &mat->rmap->N);
741   TV_add_row("Global columns", "int", &mat->cmap->N);
742   TV_add_row("Typename", TV_ascii_string_type, ((PetscObject)mat)->type_name);
743   return TV_format_OK;
744 }
745 #endif
746 
747 #undef __FUNCT__
748 #define __FUNCT__ "MatLoad"
749 /*@C
750    MatLoad - Loads a matrix that has been stored in binary format
751    with MatView().  The matrix format is determined from the options database.
752    Generates a parallel MPI matrix if the communicator has more than one
753    processor.  The default matrix type is AIJ.
754 
755    Collective on PetscViewer
756 
757    Input Parameters:
758 +  newmat - the newly loaded matrix, this needs to have been created with MatCreate()
759             or some related function before a call to MatLoad()
760 -  viewer - binary file viewer, created with PetscViewerBinaryOpen()
761 
762    Basic Options Database Keys:
763 +    -mat_type seqaij   - AIJ type
764 .    -mat_type mpiaij   - parallel AIJ type
765 .    -mat_type seqbaij  - block AIJ type
766 .    -mat_type mpibaij  - parallel block AIJ type
767 .    -mat_type seqsbaij - block symmetric AIJ type
768 .    -mat_type mpisbaij - parallel block symmetric AIJ type
769 .    -mat_type seqdense - dense type
770 .    -mat_type mpidense - parallel dense type
771 .    -mat_type blockmat - sequential blockmat type
772 .    -matload_symmetric - matrix in file is symmetric
773 -    -matload_spd       - matrix in file is symmetric positive definite
774 
775    More Options Database Keys:
776    Used with block matrix formats (MATSEQBAIJ,  ...) to specify
777    block size
778 .    -matload_block_size <bs>
779 
780    Level: beginner
781 
782    Notes:
783    MatLoad() automatically loads into the options database any options
784    given in the file filename.info where filename is the name of the file
785    that was passed to the PetscViewerBinaryOpen(). The options in the info
786    file will be ignored if you use the -viewer_binary_skip_info option.
787 
788    If the type or size of newmat is not set before a call to MatLoad, PETSc
789    sets the default matrix type AIJ and sets the local and global sizes.
790    If type and/or size is already set, then the same are used.
791 
792    In parallel, each processor can load a subset of rows (or the
793    entire matrix).  This routine is especially useful when a large
794    matrix is stored on disk and only part of it is desired on each
795    processor.  For example, a parallel solver may access only some of
796    the rows from each processor.  The algorithm used here reads
797    relatively small blocks of data rather than reading the entire
798    matrix and then subsetting it.
799 
800    Notes for advanced users:
801    Most users should not need to know the details of the binary storage
802    format, since MatLoad() and MatView() completely hide these details.
803    But for anyone who's interested, the standard binary matrix storage
804    format is
805 
806 $    int    MAT_FILE_CLASSID
807 $    int    number of rows
808 $    int    number of columns
809 $    int    total number of nonzeros
810 $    int    *number nonzeros in each row
811 $    int    *column indices of all nonzeros (starting index is zero)
812 $    PetscScalar *values of all nonzeros
813 
814    PETSc automatically does the byte swapping for
815 machines that store the bytes reversed, e.g.  DEC alpha, freebsd,
816 linux, Windows and the paragon; thus if you write your own binary
817 read/write routines you have to swap the bytes; see PetscBinaryRead()
818 and PetscBinaryWrite() to see how this may be done.
819 
820 .keywords: matrix, load, binary, input
821 
822 .seealso: PetscViewerBinaryOpen(), MatView(), VecLoad()
823 
824  @*/
825 PetscErrorCode PETSCMAT_DLLEXPORT MatLoad(Mat newmat,PetscViewer viewer)
826 {
827   PetscErrorCode ierr;
828   PetscTruth     isbinary,flg;
829   const MatType  outtype=0;
830 
831   PetscFunctionBegin;
832   PetscValidHeaderSpecific(newmat,MAT_CLASSID,1);
833   PetscValidHeaderSpecific(viewer,PETSC_VIEWER_CLASSID,2);
834   ierr = PetscTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);CHKERRQ(ierr);
835   if (!isbinary) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Invalid viewer; open viewer with PetscViewerBinaryOpen()");
836 
837 
838   if (((PetscObject)newmat)->type_name) outtype = ((PetscObject)newmat)->type_name;
839   if (!outtype) {
840     ierr = MatSetFromOptions(newmat);CHKERRQ(ierr);
841   }
842 
843   if (!newmat->ops->load) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatLoad is not supported for type: %s",outtype);
844 
845   ierr = PetscLogEventBegin(MAT_Load,viewer,0,0,0);CHKERRQ(ierr);
846   ierr = (*newmat->ops->load)(newmat,viewer);CHKERRQ(ierr);
847   ierr = PetscLogEventEnd(MAT_Load,viewer,0,0,0);CHKERRQ(ierr);
848 
849   flg  = PETSC_FALSE;
850   ierr = PetscOptionsGetTruth(((PetscObject)newmat)->prefix,"-matload_symmetric",&flg,PETSC_NULL);CHKERRQ(ierr);
851   if (flg) {
852     ierr = MatSetOption(newmat,MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr);
853     ierr = MatSetOption(newmat,MAT_SYMMETRY_ETERNAL,PETSC_TRUE);CHKERRQ(ierr);
854   }
855   flg  = PETSC_FALSE;
856   ierr = PetscOptionsGetTruth(((PetscObject)newmat)->prefix,"-matload_spd",&flg,PETSC_NULL);CHKERRQ(ierr);
857   if (flg) {
858     ierr = MatSetOption(newmat,MAT_SPD,PETSC_TRUE);CHKERRQ(ierr);
859   }
860   PetscFunctionReturn(0);
861 }
862 
863 #undef __FUNCT__
864 #define __FUNCT__ "MatScaleSystem"
865 /*@
866    MatScaleSystem - Scale a vector solution and right hand side to
867    match the scaling of a scaled matrix.
868 
869    Collective on Mat
870 
871    Input Parameter:
872 +  mat - the matrix
873 .  b - right hand side vector (or PETSC_NULL)
874 -  x - solution vector (or PETSC_NULL)
875 
876 
877    Notes:
878    For AIJ, and BAIJ matrix formats, the matrices are not
879    internally scaled, so this does nothing.
880 
881    The KSP methods automatically call this routine when required
882    (via PCPreSolve()) so it is rarely used directly.
883 
884    Level: Developer
885 
886    Concepts: matrices^scaling
887 
888 .seealso: MatUseScaledForm(), MatUnScaleSystem()
889 @*/
890 PetscErrorCode PETSCMAT_DLLEXPORT MatScaleSystem(Mat mat,Vec b,Vec x)
891 {
892   PetscErrorCode ierr;
893 
894   PetscFunctionBegin;
895   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
896   PetscValidType(mat,1);
897   ierr = MatPreallocated(mat);CHKERRQ(ierr);
898   if (x) {PetscValidHeaderSpecific(x,VEC_CLASSID,3);PetscCheckSameComm(mat,1,x,3);}
899   if (b) {PetscValidHeaderSpecific(b,VEC_CLASSID,2);PetscCheckSameComm(mat,1,b,2);}
900 
901   if (mat->ops->scalesystem) {
902     ierr = (*mat->ops->scalesystem)(mat,b,x);CHKERRQ(ierr);
903   }
904   PetscFunctionReturn(0);
905 }
906 
907 #undef __FUNCT__
908 #define __FUNCT__ "MatUnScaleSystem"
909 /*@
910    MatUnScaleSystem - Unscales a vector solution and right hand side to
911    match the original scaling of a scaled matrix.
912 
913    Collective on Mat
914 
915    Input Parameter:
916 +  mat - the matrix
917 .  b - right hand side vector (or PETSC_NULL)
918 -  x - solution vector (or PETSC_NULL)
919 
920 
921    Notes:
922    For AIJ and BAIJ matrix formats, the matrices are not
923    internally scaled, so this does nothing.
924 
925    The KSP methods automatically call this routine when required
926    (via PCPreSolve()) so it is rarely used directly.
927 
928    Level: Developer
929 
930 .seealso: MatUseScaledForm(), MatScaleSystem()
931 @*/
932 PetscErrorCode PETSCMAT_DLLEXPORT MatUnScaleSystem(Mat mat,Vec b,Vec x)
933 {
934   PetscErrorCode ierr;
935 
936   PetscFunctionBegin;
937   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
938   PetscValidType(mat,1);
939   ierr = MatPreallocated(mat);CHKERRQ(ierr);
940   if (x) {PetscValidHeaderSpecific(x,VEC_CLASSID,3);PetscCheckSameComm(mat,1,x,3);}
941   if (b) {PetscValidHeaderSpecific(b,VEC_CLASSID,2);PetscCheckSameComm(mat,1,b,2);}
942   if (mat->ops->unscalesystem) {
943     ierr = (*mat->ops->unscalesystem)(mat,b,x);CHKERRQ(ierr);
944   }
945   PetscFunctionReturn(0);
946 }
947 
948 #undef __FUNCT__
949 #define __FUNCT__ "MatUseScaledForm"
950 /*@
951    MatUseScaledForm - For matrix storage formats that scale the
952    matrix indicates matrix operations (MatMult() etc) are
953    applied using the scaled matrix.
954 
955    Logically Collective on Mat
956 
957    Input Parameter:
958 +  mat - the matrix
959 -  scaled - PETSC_TRUE for applying the scaled, PETSC_FALSE for
960             applying the original matrix
961 
962    Notes:
963    For scaled matrix formats, applying the original, unscaled matrix
964    will be slightly more expensive
965 
966    Level: Developer
967 
968 .seealso: MatScaleSystem(), MatUnScaleSystem()
969 @*/
970 PetscErrorCode PETSCMAT_DLLEXPORT MatUseScaledForm(Mat mat,PetscTruth scaled)
971 {
972   PetscErrorCode ierr;
973 
974   PetscFunctionBegin;
975   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
976   PetscValidType(mat,1);
977   PetscValidLogicalCollectiveTruth(mat,scaled,2);
978   ierr = MatPreallocated(mat);CHKERRQ(ierr);
979   if (mat->ops->usescaledform) {
980     ierr = (*mat->ops->usescaledform)(mat,scaled);CHKERRQ(ierr);
981   }
982   PetscFunctionReturn(0);
983 }
984 
985 #undef __FUNCT__
986 #define __FUNCT__ "MatDestroy"
987 /*@
988    MatDestroy - Frees space taken by a matrix.
989 
990    Collective on Mat
991 
992    Input Parameter:
993 .  A - the matrix
994 
995    Level: beginner
996 
997 @*/
998 PetscErrorCode PETSCMAT_DLLEXPORT MatDestroy(Mat A)
999 {
1000   PetscErrorCode ierr;
1001   PetscFunctionBegin;
1002   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
1003   if (--((PetscObject)A)->refct > 0) PetscFunctionReturn(0);
1004   ierr = MatPreallocated(A);CHKERRQ(ierr);
1005   /* if memory was published with AMS then destroy it */
1006   ierr = PetscObjectDepublish(A);CHKERRQ(ierr);
1007   if (A->ops->destroy) {
1008     ierr = (*A->ops->destroy)(A);CHKERRQ(ierr);
1009   }
1010   if (A->mapping) {
1011     ierr = ISLocalToGlobalMappingDestroy(A->mapping);CHKERRQ(ierr);
1012   }
1013   if (A->bmapping) {
1014     ierr = ISLocalToGlobalMappingDestroy(A->bmapping);CHKERRQ(ierr);
1015   }
1016 
1017   if (A->spptr){ierr = PetscFree(A->spptr);CHKERRQ(ierr);}
1018   ierr = PetscLayoutDestroy(A->rmap);CHKERRQ(ierr);
1019   ierr = PetscLayoutDestroy(A->cmap);CHKERRQ(ierr);
1020   ierr = PetscHeaderDestroy(A);CHKERRQ(ierr);
1021   PetscFunctionReturn(0);
1022 }
1023 
1024 #undef __FUNCT__
1025 #define __FUNCT__ "MatSetValues"
1026 /*@
1027    MatSetValues - Inserts or adds a block of values into a matrix.
1028    These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
1029    MUST be called after all calls to MatSetValues() have been completed.
1030 
1031    Not Collective
1032 
1033    Input Parameters:
1034 +  mat - the matrix
1035 .  v - a logically two-dimensional array of values
1036 .  m, idxm - the number of rows and their global indices
1037 .  n, idxn - the number of columns and their global indices
1038 -  addv - either ADD_VALUES or INSERT_VALUES, where
1039    ADD_VALUES adds values to any existing entries, and
1040    INSERT_VALUES replaces existing entries with new values
1041 
1042    Notes:
1043    By default the values, v, are row-oriented. See MatSetOption() for other options.
1044 
1045    Calls to MatSetValues() with the INSERT_VALUES and ADD_VALUES
1046    options cannot be mixed without intervening calls to the assembly
1047    routines.
1048 
1049    MatSetValues() uses 0-based row and column numbers in Fortran
1050    as well as in C.
1051 
1052    Negative indices may be passed in idxm and idxn, these rows and columns are
1053    simply ignored. This allows easily inserting element stiffness matrices
1054    with homogeneous Dirchlet boundary conditions that you don't want represented
1055    in the matrix.
1056 
1057    Efficiency Alert:
1058    The routine MatSetValuesBlocked() may offer much better efficiency
1059    for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ).
1060 
1061    Level: beginner
1062 
1063    Concepts: matrices^putting entries in
1064 
1065 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1066           InsertMode, INSERT_VALUES, ADD_VALUES
1067 @*/
1068 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv)
1069 {
1070   PetscErrorCode ierr;
1071 
1072   PetscFunctionBegin;
1073   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1074   PetscValidType(mat,1);
1075   if (!m || !n) PetscFunctionReturn(0); /* no values to insert */
1076   PetscValidIntPointer(idxm,3);
1077   PetscValidIntPointer(idxn,5);
1078   if (v) PetscValidDoublePointer(v,6);
1079   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1080   if (mat->insertmode == NOT_SET_VALUES) {
1081     mat->insertmode = addv;
1082   }
1083 #if defined(PETSC_USE_DEBUG)
1084   else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1085   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1086 #endif
1087 
1088   if (mat->assembled) {
1089     mat->was_assembled = PETSC_TRUE;
1090     mat->assembled     = PETSC_FALSE;
1091   }
1092   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1093   if (!mat->ops->setvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1094   ierr = (*mat->ops->setvalues)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr);
1095   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1096   PetscFunctionReturn(0);
1097 }
1098 
1099 
1100 #undef __FUNCT__
1101 #define __FUNCT__ "MatSetValuesRowLocal"
1102 /*@
1103    MatSetValuesRowLocal - Inserts a row (block row for BAIJ matrices) of nonzero
1104         values into a matrix
1105 
1106    Not Collective
1107 
1108    Input Parameters:
1109 +  mat - the matrix
1110 .  row - the (block) row to set
1111 -  v - a logically two-dimensional array of values
1112 
1113    Notes:
1114    By the values, v, are column-oriented (for the block version) and sorted
1115 
1116    All the nonzeros in the row must be provided
1117 
1118    The matrix must have previously had its column indices set
1119 
1120    The row must belong to this process
1121 
1122    Level: intermediate
1123 
1124    Concepts: matrices^putting entries in
1125 
1126 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1127           InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues(), MatSetValuesRow(), MatSetLocalToGlobalMapping()
1128 @*/
1129 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesRowLocal(Mat mat,PetscInt row,const PetscScalar v[])
1130 {
1131   PetscErrorCode ierr;
1132 
1133   PetscFunctionBegin;
1134   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1135   PetscValidType(mat,1);
1136   PetscValidScalarPointer(v,2);
1137   ierr = MatSetValuesRow(mat, mat->mapping->indices[row],v);CHKERRQ(ierr);
1138   PetscFunctionReturn(0);
1139 }
1140 
1141 #undef __FUNCT__
1142 #define __FUNCT__ "MatSetValuesRow"
1143 /*@
1144    MatSetValuesRow - Inserts a row (block row for BAIJ matrices) of nonzero
1145         values into a matrix
1146 
1147    Not Collective
1148 
1149    Input Parameters:
1150 +  mat - the matrix
1151 .  row - the (block) row to set
1152 -  v - a logically two-dimensional array of values
1153 
1154    Notes:
1155    The values, v, are column-oriented for the block version.
1156 
1157    All the nonzeros in the row must be provided
1158 
1159    THE MATRIX MUSAT HAVE PREVIOUSLY HAD ITS COLUMN INDICES SET. IT IS RARE THAT THIS ROUTINE IS USED, usually MatSetValues() is used.
1160 
1161    The row must belong to this process
1162 
1163    Level: advanced
1164 
1165    Concepts: matrices^putting entries in
1166 
1167 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
1168           InsertMode, INSERT_VALUES, ADD_VALUES, MatSetValues()
1169 @*/
1170 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesRow(Mat mat,PetscInt row,const PetscScalar v[])
1171 {
1172   PetscErrorCode ierr;
1173 
1174   PetscFunctionBegin;
1175   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1176   PetscValidType(mat,1);
1177   PetscValidScalarPointer(v,2);
1178 #if defined(PETSC_USE_DEBUG)
1179   if (mat->insertmode == ADD_VALUES) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add and insert values");
1180   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1181 #endif
1182   mat->insertmode = INSERT_VALUES;
1183 
1184   if (mat->assembled) {
1185     mat->was_assembled = PETSC_TRUE;
1186     mat->assembled     = PETSC_FALSE;
1187   }
1188   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1189   if (!mat->ops->setvaluesrow) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1190   ierr = (*mat->ops->setvaluesrow)(mat,row,v);CHKERRQ(ierr);
1191   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1192   PetscFunctionReturn(0);
1193 }
1194 
1195 #undef __FUNCT__
1196 #define __FUNCT__ "MatSetValuesStencil"
1197 /*@
1198    MatSetValuesStencil - Inserts or adds a block of values into a matrix.
1199      Using structured grid indexing
1200 
1201    Not Collective
1202 
1203    Input Parameters:
1204 +  mat - the matrix
1205 .  m - number of rows being entered
1206 .  idxm - grid coordinates (and component number when dof > 1) for matrix rows being entered
1207 .  n - number of columns being entered
1208 .  idxn - grid coordinates (and component number when dof > 1) for matrix columns being entered
1209 .  v - a logically two-dimensional array of values
1210 -  addv - either ADD_VALUES or INSERT_VALUES, where
1211    ADD_VALUES adds values to any existing entries, and
1212    INSERT_VALUES replaces existing entries with new values
1213 
1214    Notes:
1215    By default the values, v, are row-oriented.  See MatSetOption() for other options.
1216 
1217    Calls to MatSetValuesStencil() with the INSERT_VALUES and ADD_VALUES
1218    options cannot be mixed without intervening calls to the assembly
1219    routines.
1220 
1221    The grid coordinates are across the entire grid, not just the local portion
1222 
1223    MatSetValuesStencil() uses 0-based row and column numbers in Fortran
1224    as well as in C.
1225 
1226    For setting/accessing vector values via array coordinates you can use the DAVecGetArray() routine
1227 
1228    In order to use this routine you must either obtain the matrix with DAGetMatrix()
1229    or call MatSetLocalToGlobalMapping() and MatSetStencil() first.
1230 
1231    The columns and rows in the stencil passed in MUST be contained within the
1232    ghost region of the given process as set with DACreateXXX() or MatSetStencil(). For example,
1233    if you create a DA with an overlap of one grid level and on a particular process its first
1234    local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1235    first i index you can use in your column and row indices in MatSetStencil() is 5.
1236 
1237    In Fortran idxm and idxn should be declared as
1238 $     MatStencil idxm(4,m),idxn(4,n)
1239    and the values inserted using
1240 $    idxm(MatStencil_i,1) = i
1241 $    idxm(MatStencil_j,1) = j
1242 $    idxm(MatStencil_k,1) = k
1243 $    idxm(MatStencil_c,1) = c
1244    etc
1245 
1246    For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
1247    obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
1248    etc to obtain values that obtained by wrapping the values from the left edge. This does not work for the DA_NONPERIODIC
1249    wrap.
1250 
1251    For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have
1252    a single value per point) you can skip filling those indices.
1253 
1254    Inspired by the structured grid interface to the HYPRE package
1255    (http://www.llnl.gov/CASC/hypre)
1256 
1257    Efficiency Alert:
1258    The routine MatSetValuesBlockedStencil() may offer much better efficiency
1259    for users of block sparse formats (MATSEQBAIJ and MATMPIBAIJ).
1260 
1261    Level: beginner
1262 
1263    Concepts: matrices^putting entries in
1264 
1265 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1266           MatSetValues(), MatSetValuesBlockedStencil(), MatSetStencil(), DAGetMatrix(), DAVecGetArray(), MatStencil
1267 @*/
1268 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv)
1269 {
1270   PetscErrorCode ierr;
1271   PetscInt       j,i,jdxm[128],jdxn[256],dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp;
1272   PetscInt       *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1273 
1274   PetscFunctionBegin;
1275   if (!m || !n) PetscFunctionReturn(0); /* no values to insert */
1276   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1277   PetscValidType(mat,1);
1278   PetscValidIntPointer(idxm,3);
1279   PetscValidIntPointer(idxn,5);
1280   PetscValidScalarPointer(v,6);
1281 
1282   if (m > 128) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only set 128 rows at a time; trying to set %D",m);
1283   if (n > 256) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only set 256 columns at a time; trying to set %D",n);
1284 
1285   for (i=0; i<m; i++) {
1286     for (j=0; j<3-sdim; j++) dxm++;
1287     tmp = *dxm++ - starts[0];
1288     for (j=0; j<dim-1; j++) {
1289       if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
1290       else                                       tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
1291     }
1292     if (mat->stencil.noc) dxm++;
1293     jdxm[i] = tmp;
1294   }
1295   for (i=0; i<n; i++) {
1296     for (j=0; j<3-sdim; j++) dxn++;
1297     tmp = *dxn++ - starts[0];
1298     for (j=0; j<dim-1; j++) {
1299       if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
1300       else                                       tmp = tmp*dims[j] + *(dxn-1) - starts[j+1];
1301     }
1302     if (mat->stencil.noc) dxn++;
1303     jdxn[i] = tmp;
1304   }
1305   ierr = MatSetValuesLocal(mat,m,jdxm,n,jdxn,v,addv);CHKERRQ(ierr);
1306   PetscFunctionReturn(0);
1307 }
1308 
1309 #undef __FUNCT__
1310 #define __FUNCT__ "MatSetValuesBlockedStencil"
1311 /*@C
1312    MatSetValuesBlockedStencil - Inserts or adds a block of values into a matrix.
1313      Using structured grid indexing
1314 
1315    Not Collective
1316 
1317    Input Parameters:
1318 +  mat - the matrix
1319 .  m - number of rows being entered
1320 .  idxm - grid coordinates for matrix rows being entered
1321 .  n - number of columns being entered
1322 .  idxn - grid coordinates for matrix columns being entered
1323 .  v - a logically two-dimensional array of values
1324 -  addv - either ADD_VALUES or INSERT_VALUES, where
1325    ADD_VALUES adds values to any existing entries, and
1326    INSERT_VALUES replaces existing entries with new values
1327 
1328    Notes:
1329    By default the values, v, are row-oriented and unsorted.
1330    See MatSetOption() for other options.
1331 
1332    Calls to MatSetValuesBlockedStencil() with the INSERT_VALUES and ADD_VALUES
1333    options cannot be mixed without intervening calls to the assembly
1334    routines.
1335 
1336    The grid coordinates are across the entire grid, not just the local portion
1337 
1338    MatSetValuesBlockedStencil() uses 0-based row and column numbers in Fortran
1339    as well as in C.
1340 
1341    For setting/accessing vector values via array coordinates you can use the DAVecGetArray() routine
1342 
1343    In order to use this routine you must either obtain the matrix with DAGetMatrix()
1344    or call MatSetBlockSize(), MatSetLocalToGlobalMapping() and MatSetStencil() first.
1345 
1346    The columns and rows in the stencil passed in MUST be contained within the
1347    ghost region of the given process as set with DACreateXXX() or MatSetStencil(). For example,
1348    if you create a DA with an overlap of one grid level and on a particular process its first
1349    local nonghost x logical coordinate is 6 (so its first ghost x logical coordinate is 5) the
1350    first i index you can use in your column and row indices in MatSetStencil() is 5.
1351 
1352    In Fortran idxm and idxn should be declared as
1353 $     MatStencil idxm(4,m),idxn(4,n)
1354    and the values inserted using
1355 $    idxm(MatStencil_i,1) = i
1356 $    idxm(MatStencil_j,1) = j
1357 $    idxm(MatStencil_k,1) = k
1358    etc
1359 
1360    Negative indices may be passed in idxm and idxn, these rows and columns are
1361    simply ignored. This allows easily inserting element stiffness matrices
1362    with homogeneous Dirchlet boundary conditions that you don't want represented
1363    in the matrix.
1364 
1365    Inspired by the structured grid interface to the HYPRE package
1366    (http://www.llnl.gov/CASC/hypre)
1367 
1368    Level: beginner
1369 
1370    Concepts: matrices^putting entries in
1371 
1372 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1373           MatSetValues(), MatSetValuesStencil(), MatSetStencil(), DAGetMatrix(), DAVecGetArray(), MatStencil,
1374           MatSetBlockSize(), MatSetLocalToGlobalMapping()
1375 @*/
1376 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesBlockedStencil(Mat mat,PetscInt m,const MatStencil idxm[],PetscInt n,const MatStencil idxn[],const PetscScalar v[],InsertMode addv)
1377 {
1378   PetscErrorCode ierr;
1379   PetscInt       j,i,jdxm[128],jdxn[256],dim = mat->stencil.dim,*dims = mat->stencil.dims+1,tmp;
1380   PetscInt       *starts = mat->stencil.starts,*dxm = (PetscInt*)idxm,*dxn = (PetscInt*)idxn,sdim = dim - (1 - (PetscInt)mat->stencil.noc);
1381 
1382   PetscFunctionBegin;
1383   if (!m || !n) PetscFunctionReturn(0); /* no values to insert */
1384   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1385   PetscValidType(mat,1);
1386   PetscValidIntPointer(idxm,3);
1387   PetscValidIntPointer(idxn,5);
1388   PetscValidScalarPointer(v,6);
1389 
1390   if (m > 128) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only set 128 rows at a time; trying to set %D",m);
1391   if (n > 128) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only set 256 columns at a time; trying to set %D",n);
1392 
1393   for (i=0; i<m; i++) {
1394     for (j=0; j<3-sdim; j++) dxm++;
1395     tmp = *dxm++ - starts[0];
1396     for (j=0; j<sdim-1; j++) {
1397       if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
1398       else                                      tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
1399     }
1400     dxm++;
1401     jdxm[i] = tmp;
1402   }
1403   for (i=0; i<n; i++) {
1404     for (j=0; j<3-sdim; j++) dxn++;
1405     tmp = *dxn++ - starts[0];
1406     for (j=0; j<sdim-1; j++) {
1407       if ((*dxn++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
1408       else                                       tmp = tmp*dims[j] + *(dxn-1) - starts[j+1];
1409     }
1410     dxn++;
1411     jdxn[i] = tmp;
1412   }
1413   ierr = MatSetValuesBlockedLocal(mat,m,jdxm,n,jdxn,v,addv);CHKERRQ(ierr);
1414   PetscFunctionReturn(0);
1415 }
1416 
1417 #undef __FUNCT__
1418 #define __FUNCT__ "MatSetStencil"
1419 /*@
1420    MatSetStencil - Sets the grid information for setting values into a matrix via
1421         MatSetValuesStencil()
1422 
1423    Not Collective
1424 
1425    Input Parameters:
1426 +  mat - the matrix
1427 .  dim - dimension of the grid 1, 2, or 3
1428 .  dims - number of grid points in x, y, and z direction, including ghost points on your processor
1429 .  starts - starting point of ghost nodes on your processor in x, y, and z direction
1430 -  dof - number of degrees of freedom per node
1431 
1432 
1433    Inspired by the structured grid interface to the HYPRE package
1434    (www.llnl.gov/CASC/hyper)
1435 
1436    For matrices generated with DAGetMatrix() this routine is automatically called and so not needed by the
1437    user.
1438 
1439    Level: beginner
1440 
1441    Concepts: matrices^putting entries in
1442 
1443 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal()
1444           MatSetValues(), MatSetValuesBlockedStencil(), MatSetValuesStencil()
1445 @*/
1446 PetscErrorCode PETSCMAT_DLLEXPORT MatSetStencil(Mat mat,PetscInt dim,const PetscInt dims[],const PetscInt starts[],PetscInt dof)
1447 {
1448   PetscInt i;
1449 
1450   PetscFunctionBegin;
1451   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1452   PetscValidIntPointer(dims,3);
1453   PetscValidIntPointer(starts,4);
1454 
1455   mat->stencil.dim = dim + (dof > 1);
1456   for (i=0; i<dim; i++) {
1457     mat->stencil.dims[i]   = dims[dim-i-1];      /* copy the values in backwards */
1458     mat->stencil.starts[i] = starts[dim-i-1];
1459   }
1460   mat->stencil.dims[dim]   = dof;
1461   mat->stencil.starts[dim] = 0;
1462   mat->stencil.noc         = (PetscTruth)(dof == 1);
1463   PetscFunctionReturn(0);
1464 }
1465 
1466 #undef __FUNCT__
1467 #define __FUNCT__ "MatSetValuesBlocked"
1468 /*@
1469    MatSetValuesBlocked - Inserts or adds a block of values into a matrix.
1470 
1471    Not Collective
1472 
1473    Input Parameters:
1474 +  mat - the matrix
1475 .  v - a logically two-dimensional array of values
1476 .  m, idxm - the number of block rows and their global block indices
1477 .  n, idxn - the number of block columns and their global block indices
1478 -  addv - either ADD_VALUES or INSERT_VALUES, where
1479    ADD_VALUES adds values to any existing entries, and
1480    INSERT_VALUES replaces existing entries with new values
1481 
1482    Notes:
1483    The m and n count the NUMBER of blocks in the row direction and column direction,
1484    NOT the total number of rows/columns; for example, if the block size is 2 and
1485    you are passing in values for rows 2,3,4,5  then m would be 2 (not 4).
1486    The values in idxm would be 1 2; that is the first index for each block divided by
1487    the block size.
1488 
1489    Note that you must call MatSetBlockSize() when constructing this matrix (after
1490    preallocating it).
1491 
1492    By default the values, v, are row-oriented, so the layout of
1493    v is the same as for MatSetValues(). See MatSetOption() for other options.
1494 
1495    Calls to MatSetValuesBlocked() with the INSERT_VALUES and ADD_VALUES
1496    options cannot be mixed without intervening calls to the assembly
1497    routines.
1498 
1499    MatSetValuesBlocked() uses 0-based row and column numbers in Fortran
1500    as well as in C.
1501 
1502    Negative indices may be passed in idxm and idxn, these rows and columns are
1503    simply ignored. This allows easily inserting element stiffness matrices
1504    with homogeneous Dirchlet boundary conditions that you don't want represented
1505    in the matrix.
1506 
1507    Each time an entry is set within a sparse matrix via MatSetValues(),
1508    internal searching must be done to determine where to place the the
1509    data in the matrix storage space.  By instead inserting blocks of
1510    entries via MatSetValuesBlocked(), the overhead of matrix assembly is
1511    reduced.
1512 
1513    Example:
1514 $   Suppose m=n=2 and block size(bs) = 2 The array is
1515 $
1516 $   1  2  | 3  4
1517 $   5  6  | 7  8
1518 $   - - - | - - -
1519 $   9  10 | 11 12
1520 $   13 14 | 15 16
1521 $
1522 $   v[] should be passed in like
1523 $   v[] = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]
1524 $
1525 $  If you are not using row oriented storage of v (that is you called MatSetOption(mat,MAT_ROW_ORIENTED,PETSC_FALSE)) then
1526 $   v[] = [1,5,9,13,2,6,10,14,3,7,11,15,4,8,12,16]
1527 
1528    Level: intermediate
1529 
1530    Concepts: matrices^putting entries in blocked
1531 
1532 .seealso: MatSetBlockSize(), MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal()
1533 @*/
1534 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesBlocked(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],const PetscScalar v[],InsertMode addv)
1535 {
1536   PetscErrorCode ierr;
1537 
1538   PetscFunctionBegin;
1539   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1540   PetscValidType(mat,1);
1541   if (!m || !n) PetscFunctionReturn(0); /* no values to insert */
1542   PetscValidIntPointer(idxm,3);
1543   PetscValidIntPointer(idxn,5);
1544   PetscValidScalarPointer(v,6);
1545   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1546   if (mat->insertmode == NOT_SET_VALUES) {
1547     mat->insertmode = addv;
1548   }
1549 #if defined(PETSC_USE_DEBUG)
1550   else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1551   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1552 #endif
1553 
1554   if (mat->assembled) {
1555     mat->was_assembled = PETSC_TRUE;
1556     mat->assembled     = PETSC_FALSE;
1557   }
1558   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1559   if (mat->ops->setvaluesblocked) {
1560     ierr = (*mat->ops->setvaluesblocked)(mat,m,idxm,n,idxn,v,addv);CHKERRQ(ierr);
1561   } else {
1562     PetscInt buf[4096],*ibufm=0,*ibufn=0;
1563     PetscInt i,j,*iidxm,*iidxn,bs=mat->rmap->bs;
1564     if ((m+n)*bs <= 4096) {
1565       iidxm = buf; iidxn = buf + m*bs;
1566     } else {
1567       ierr = PetscMalloc2(m*bs,PetscInt,&ibufm,n*bs,PetscInt,&ibufn);CHKERRQ(ierr);
1568       iidxm = ibufm; iidxn = ibufn;
1569     }
1570     for (i=0; i<m; i++) {
1571       for (j=0; j<bs; j++) {
1572 	iidxm[i*bs+j] = bs*idxm[i] + j;
1573       }
1574     }
1575     for (i=0; i<n; i++) {
1576       for (j=0; j<bs; j++) {
1577 	iidxn[i*bs+j] = bs*idxn[i] + j;
1578       }
1579     }
1580     ierr = MatSetValues(mat,bs*m,iidxm,bs*n,iidxn,v,addv);CHKERRQ(ierr);
1581     ierr = PetscFree2(ibufm,ibufn);CHKERRQ(ierr);
1582   }
1583   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1584   PetscFunctionReturn(0);
1585 }
1586 
1587 #undef __FUNCT__
1588 #define __FUNCT__ "MatGetValues"
1589 /*@
1590    MatGetValues - Gets a block of values from a matrix.
1591 
1592    Not Collective; currently only returns a local block
1593 
1594    Input Parameters:
1595 +  mat - the matrix
1596 .  v - a logically two-dimensional array for storing the values
1597 .  m, idxm - the number of rows and their global indices
1598 -  n, idxn - the number of columns and their global indices
1599 
1600    Notes:
1601    The user must allocate space (m*n PetscScalars) for the values, v.
1602    The values, v, are then returned in a row-oriented format,
1603    analogous to that used by default in MatSetValues().
1604 
1605    MatGetValues() uses 0-based row and column numbers in
1606    Fortran as well as in C.
1607 
1608    MatGetValues() requires that the matrix has been assembled
1609    with MatAssemblyBegin()/MatAssemblyEnd().  Thus, calls to
1610    MatSetValues() and MatGetValues() CANNOT be made in succession
1611    without intermediate matrix assembly.
1612 
1613    Negative row or column indices will be ignored and those locations in v[] will be
1614    left unchanged.
1615 
1616    Level: advanced
1617 
1618    Concepts: matrices^accessing values
1619 
1620 .seealso: MatGetRow(), MatGetSubMatrices(), MatSetValues()
1621 @*/
1622 PetscErrorCode PETSCMAT_DLLEXPORT MatGetValues(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[])
1623 {
1624   PetscErrorCode ierr;
1625 
1626   PetscFunctionBegin;
1627   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1628   PetscValidType(mat,1);
1629   if (!m || !n) PetscFunctionReturn(0);
1630   PetscValidIntPointer(idxm,3);
1631   PetscValidIntPointer(idxn,5);
1632   PetscValidScalarPointer(v,6);
1633   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
1634   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1635   if (!mat->ops->getvalues) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
1636   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1637 
1638   ierr = PetscLogEventBegin(MAT_GetValues,mat,0,0,0);CHKERRQ(ierr);
1639   ierr = (*mat->ops->getvalues)(mat,m,idxm,n,idxn,v);CHKERRQ(ierr);
1640   ierr = PetscLogEventEnd(MAT_GetValues,mat,0,0,0);CHKERRQ(ierr);
1641   PetscFunctionReturn(0);
1642 }
1643 
1644 #undef __FUNCT__
1645 #define __FUNCT__ "MatSetLocalToGlobalMapping"
1646 /*@
1647    MatSetLocalToGlobalMapping - Sets a local-to-global numbering for use by
1648    the routine MatSetValuesLocal() to allow users to insert matrix entries
1649    using a local (per-processor) numbering.
1650 
1651    Not Collective
1652 
1653    Input Parameters:
1654 +  x - the matrix
1655 -  mapping - mapping created with ISLocalToGlobalMappingCreate()
1656              or ISLocalToGlobalMappingCreateIS()
1657 
1658    Level: intermediate
1659 
1660    Concepts: matrices^local to global mapping
1661    Concepts: local to global mapping^for matrices
1662 
1663 .seealso:  MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesLocal()
1664 @*/
1665 PetscErrorCode PETSCMAT_DLLEXPORT MatSetLocalToGlobalMapping(Mat x,ISLocalToGlobalMapping mapping)
1666 {
1667   PetscErrorCode ierr;
1668   PetscFunctionBegin;
1669   PetscValidHeaderSpecific(x,MAT_CLASSID,1);
1670   PetscValidType(x,1);
1671   PetscValidHeaderSpecific(mapping,IS_LTOGM_CLASSID,2);
1672   if (x->mapping) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Mapping already set for matrix");
1673   ierr = MatPreallocated(x);CHKERRQ(ierr);
1674 
1675   if (x->ops->setlocaltoglobalmapping) {
1676     ierr = (*x->ops->setlocaltoglobalmapping)(x,mapping);CHKERRQ(ierr);
1677   } else {
1678     ierr = PetscObjectReference((PetscObject)mapping);CHKERRQ(ierr);
1679     if (x->mapping) { ierr = ISLocalToGlobalMappingDestroy(x->mapping);CHKERRQ(ierr); }
1680     x->mapping = mapping;
1681   }
1682   PetscFunctionReturn(0);
1683 }
1684 
1685 #undef __FUNCT__
1686 #define __FUNCT__ "MatSetLocalToGlobalMappingBlock"
1687 /*@
1688    MatSetLocalToGlobalMappingBlock - Sets a local-to-global numbering for use
1689    by the routine MatSetValuesBlockedLocal() to allow users to insert matrix
1690    entries using a local (per-processor) numbering.
1691 
1692    Not Collective
1693 
1694    Input Parameters:
1695 +  x - the matrix
1696 -  mapping - mapping created with ISLocalToGlobalMappingCreate() or
1697              ISLocalToGlobalMappingCreateIS()
1698 
1699    Level: intermediate
1700 
1701    Concepts: matrices^local to global mapping blocked
1702    Concepts: local to global mapping^for matrices, blocked
1703 
1704 .seealso:  MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetValuesBlockedLocal(),
1705            MatSetValuesBlocked(), MatSetValuesLocal()
1706 @*/
1707 PetscErrorCode PETSCMAT_DLLEXPORT MatSetLocalToGlobalMappingBlock(Mat x,ISLocalToGlobalMapping mapping)
1708 {
1709   PetscErrorCode ierr;
1710   PetscFunctionBegin;
1711   PetscValidHeaderSpecific(x,MAT_CLASSID,1);
1712   PetscValidType(x,1);
1713   PetscValidHeaderSpecific(mapping,IS_LTOGM_CLASSID,2);
1714   if (x->bmapping) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Mapping already set for matrix");
1715   ierr = PetscObjectReference((PetscObject)mapping);CHKERRQ(ierr);
1716   if (x->bmapping) { ierr = ISLocalToGlobalMappingDestroy(x->bmapping);CHKERRQ(ierr); }
1717   x->bmapping = mapping;
1718   PetscFunctionReturn(0);
1719 }
1720 
1721 #undef __FUNCT__
1722 #define __FUNCT__ "MatSetValuesLocal"
1723 /*@
1724    MatSetValuesLocal - Inserts or adds values into certain locations of a matrix,
1725    using a local ordering of the nodes.
1726 
1727    Not Collective
1728 
1729    Input Parameters:
1730 +  x - the matrix
1731 .  nrow, irow - number of rows and their local indices
1732 .  ncol, icol - number of columns and their local indices
1733 .  y -  a logically two-dimensional array of values
1734 -  addv - either INSERT_VALUES or ADD_VALUES, where
1735    ADD_VALUES adds values to any existing entries, and
1736    INSERT_VALUES replaces existing entries with new values
1737 
1738    Notes:
1739    Before calling MatSetValuesLocal(), the user must first set the
1740    local-to-global mapping by calling MatSetLocalToGlobalMapping().
1741 
1742    Calls to MatSetValuesLocal() with the INSERT_VALUES and ADD_VALUES
1743    options cannot be mixed without intervening calls to the assembly
1744    routines.
1745 
1746    These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
1747    MUST be called after all calls to MatSetValuesLocal() have been completed.
1748 
1749    Level: intermediate
1750 
1751    Concepts: matrices^putting entries in with local numbering
1752 
1753 .seealso:  MatAssemblyBegin(), MatAssemblyEnd(), MatSetValues(), MatSetLocalToGlobalMapping(),
1754            MatSetValueLocal()
1755 @*/
1756 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv)
1757 {
1758   PetscErrorCode ierr;
1759   PetscInt       irowm[2048],icolm[2048];
1760 
1761   PetscFunctionBegin;
1762   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1763   PetscValidType(mat,1);
1764   if (!nrow || !ncol) PetscFunctionReturn(0); /* no values to insert */
1765   PetscValidIntPointer(irow,3);
1766   PetscValidIntPointer(icol,5);
1767   PetscValidScalarPointer(y,6);
1768   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1769   if (mat->insertmode == NOT_SET_VALUES) {
1770     mat->insertmode = addv;
1771   }
1772 #if defined(PETSC_USE_DEBUG)
1773   else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1774   if (!mat->ops->setvalueslocal && (nrow > 2048 || ncol > 2048)) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_SUP,"Number column/row indices must be <= 2048: are %D %D",nrow,ncol);
1775   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1776 #endif
1777 
1778   if (mat->assembled) {
1779     mat->was_assembled = PETSC_TRUE;
1780     mat->assembled     = PETSC_FALSE;
1781   }
1782   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1783   if (!mat->ops->setvalueslocal) {
1784     ierr = ISLocalToGlobalMappingApply(mat->mapping,nrow,irow,irowm);CHKERRQ(ierr);
1785     ierr = ISLocalToGlobalMappingApply(mat->mapping,ncol,icol,icolm);CHKERRQ(ierr);
1786     ierr = (*mat->ops->setvalues)(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr);
1787   } else {
1788     ierr = (*mat->ops->setvalueslocal)(mat,nrow,irow,ncol,icol,y,addv);CHKERRQ(ierr);
1789   }
1790   mat->same_nonzero = PETSC_FALSE;
1791   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1792   PetscFunctionReturn(0);
1793 }
1794 
1795 #undef __FUNCT__
1796 #define __FUNCT__ "MatSetValuesBlockedLocal"
1797 /*@
1798    MatSetValuesBlockedLocal - Inserts or adds values into certain locations of a matrix,
1799    using a local ordering of the nodes a block at a time.
1800 
1801    Not Collective
1802 
1803    Input Parameters:
1804 +  x - the matrix
1805 .  nrow, irow - number of rows and their local indices
1806 .  ncol, icol - number of columns and their local indices
1807 .  y -  a logically two-dimensional array of values
1808 -  addv - either INSERT_VALUES or ADD_VALUES, where
1809    ADD_VALUES adds values to any existing entries, and
1810    INSERT_VALUES replaces existing entries with new values
1811 
1812    Notes:
1813    Before calling MatSetValuesBlockedLocal(), the user must first set the
1814    block size using MatSetBlockSize(), and the local-to-global mapping by
1815    calling MatSetLocalToGlobalMappingBlock(), where the mapping MUST be
1816    set for matrix blocks, not for matrix elements.
1817 
1818    Calls to MatSetValuesBlockedLocal() with the INSERT_VALUES and ADD_VALUES
1819    options cannot be mixed without intervening calls to the assembly
1820    routines.
1821 
1822    These values may be cached, so MatAssemblyBegin() and MatAssemblyEnd()
1823    MUST be called after all calls to MatSetValuesBlockedLocal() have been completed.
1824 
1825    Level: intermediate
1826 
1827    Concepts: matrices^putting blocked values in with local numbering
1828 
1829 .seealso:  MatSetBlockSize(), MatSetLocalToGlobalMappingBlock(), MatAssemblyBegin(), MatAssemblyEnd(),
1830            MatSetValuesLocal(), MatSetLocalToGlobalMappingBlock(), MatSetValuesBlocked()
1831 @*/
1832 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesBlockedLocal(Mat mat,PetscInt nrow,const PetscInt irow[],PetscInt ncol,const PetscInt icol[],const PetscScalar y[],InsertMode addv)
1833 {
1834   PetscErrorCode ierr;
1835   PetscInt       irowm[2048],icolm[2048];
1836 
1837   PetscFunctionBegin;
1838   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1839   PetscValidType(mat,1);
1840   if (!nrow || !ncol) PetscFunctionReturn(0); /* no values to insert */
1841   PetscValidIntPointer(irow,3);
1842   PetscValidIntPointer(icol,5);
1843   PetscValidScalarPointer(y,6);
1844   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1845   if (mat->insertmode == NOT_SET_VALUES) {
1846     mat->insertmode = addv;
1847   }
1848 #if defined(PETSC_USE_DEBUG)
1849   else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
1850   if (!mat->bmapping) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Local to global never set with MatSetLocalToGlobalMappingBlock()");
1851   if (nrow > 2048 || ncol > 2048) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_SUP,"Number column/row indices must be <= 2048: are %D %D",nrow,ncol);
1852   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1853 #endif
1854 
1855   if (mat->assembled) {
1856     mat->was_assembled = PETSC_TRUE;
1857     mat->assembled     = PETSC_FALSE;
1858   }
1859   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1860   ierr = ISLocalToGlobalMappingApply(mat->bmapping,nrow,irow,irowm);CHKERRQ(ierr);
1861   ierr = ISLocalToGlobalMappingApply(mat->bmapping,ncol,icol,icolm);CHKERRQ(ierr);
1862   if (mat->ops->setvaluesblocked) {
1863     ierr = (*mat->ops->setvaluesblocked)(mat,nrow,irowm,ncol,icolm,y,addv);CHKERRQ(ierr);
1864   } else {
1865     PetscInt buf[4096],*ibufm=0,*ibufn=0;
1866     PetscInt i,j,*iirowm,*iicolm,bs=mat->rmap->bs;
1867     if ((nrow+ncol)*bs <= 4096) {
1868       iirowm = buf; iicolm = buf + nrow*bs;
1869     } else {
1870       ierr = PetscMalloc2(nrow*bs,PetscInt,&ibufm,ncol*bs,PetscInt,&ibufn);CHKERRQ(ierr);
1871       iirowm = ibufm; iicolm = ibufn;
1872     }
1873     for (i=0; i<nrow; i++) {
1874       for (j=0; j<bs; j++) {
1875 	iirowm[i*bs+j] = bs*irowm[i] + j;
1876       }
1877     }
1878     for (i=0; i<ncol; i++) {
1879       for (j=0; j<bs; j++) {
1880 	iicolm[i*bs+j] = bs*icolm[i] + j;
1881       }
1882     }
1883     ierr = MatSetValues(mat,bs*nrow,iirowm,bs*ncol,iicolm,y,addv);CHKERRQ(ierr);
1884     ierr = PetscFree2(ibufm,ibufn);CHKERRQ(ierr);
1885   }
1886   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
1887   PetscFunctionReturn(0);
1888 }
1889 
1890 #undef __FUNCT__
1891 #define __FUNCT__ "MatMultDiagonalBlock"
1892 /*@
1893    MatMultDiagonalBlock - Computes the matrix-vector product, y = Dx. Where D is defined by the inode or block structure of the diagonal
1894 
1895    Collective on Mat and Vec
1896 
1897    Input Parameters:
1898 +  mat - the matrix
1899 -  x   - the vector to be multiplied
1900 
1901    Output Parameters:
1902 .  y - the result
1903 
1904    Notes:
1905    The vectors x and y cannot be the same.  I.e., one cannot
1906    call MatMult(A,y,y).
1907 
1908    Level: developer
1909 
1910    Concepts: matrix-vector product
1911 
1912 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
1913 @*/
1914 PetscErrorCode PETSCMAT_DLLEXPORT MatMultDiagonalBlock(Mat mat,Vec x,Vec y)
1915 {
1916   PetscErrorCode ierr;
1917 
1918   PetscFunctionBegin;
1919   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1920   PetscValidType(mat,1);
1921   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
1922   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
1923 
1924   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
1925   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1926   if (x == y) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
1927   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1928 
1929   if (!mat->ops->multdiagonalblock) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"This matrix type does not have a multiply defined");
1930   ierr = (*mat->ops->multdiagonalblock)(mat,x,y);CHKERRQ(ierr);
1931   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
1932   PetscFunctionReturn(0);
1933 }
1934 
1935 /* --------------------------------------------------------*/
1936 #undef __FUNCT__
1937 #define __FUNCT__ "MatMult"
1938 /*@
1939    MatMult - Computes the matrix-vector product, y = Ax.
1940 
1941    Neighbor-wise Collective on Mat and Vec
1942 
1943    Input Parameters:
1944 +  mat - the matrix
1945 -  x   - the vector to be multiplied
1946 
1947    Output Parameters:
1948 .  y - the result
1949 
1950    Notes:
1951    The vectors x and y cannot be the same.  I.e., one cannot
1952    call MatMult(A,y,y).
1953 
1954    Level: beginner
1955 
1956    Concepts: matrix-vector product
1957 
1958 .seealso: MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
1959 @*/
1960 PetscErrorCode PETSCMAT_DLLEXPORT MatMult(Mat mat,Vec x,Vec y)
1961 {
1962   PetscErrorCode ierr;
1963 
1964   PetscFunctionBegin;
1965   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
1966   PetscValidType(mat,1);
1967   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
1968   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
1969 
1970   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
1971   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1972   if (x == y) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
1973 #ifndef PETSC_HAVE_CONSTRAINTS
1974   if (mat->cmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
1975   if (mat->rmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
1976   if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n);
1977 #endif
1978   ierr = MatPreallocated(mat);CHKERRQ(ierr);
1979 
1980   if (mat->nullsp) {
1981     ierr = MatNullSpaceRemove(mat->nullsp,x,&x);CHKERRQ(ierr);
1982   }
1983 
1984   if (!mat->ops->mult) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"This matrix type does not have a multiply defined");
1985   ierr = PetscLogEventBegin(MAT_Mult,mat,x,y,0);CHKERRQ(ierr);
1986   ierr = (*mat->ops->mult)(mat,x,y);CHKERRQ(ierr);
1987   ierr = PetscLogEventEnd(MAT_Mult,mat,x,y,0);CHKERRQ(ierr);
1988 
1989   if (mat->nullsp) {
1990     ierr = MatNullSpaceRemove(mat->nullsp,y,PETSC_NULL);CHKERRQ(ierr);
1991   }
1992   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
1993   PetscFunctionReturn(0);
1994 }
1995 
1996 #undef __FUNCT__
1997 #define __FUNCT__ "MatMultTranspose"
1998 /*@
1999    MatMultTranspose - Computes matrix transpose times a vector.
2000 
2001    Neighbor-wise Collective on Mat and Vec
2002 
2003    Input Parameters:
2004 +  mat - the matrix
2005 -  x   - the vector to be multilplied
2006 
2007    Output Parameters:
2008 .  y - the result
2009 
2010    Notes:
2011    The vectors x and y cannot be the same.  I.e., one cannot
2012    call MatMultTranspose(A,y,y).
2013 
2014    Level: beginner
2015 
2016    Concepts: matrix vector product^transpose
2017 
2018 .seealso: MatMult(), MatMultAdd(), MatMultTransposeAdd()
2019 @*/
2020 PetscErrorCode PETSCMAT_DLLEXPORT MatMultTranspose(Mat mat,Vec x,Vec y)
2021 {
2022   PetscErrorCode ierr;
2023 
2024   PetscFunctionBegin;
2025   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2026   PetscValidType(mat,1);
2027   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
2028   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
2029 
2030   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2031   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2032   if (x == y) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2033 #ifndef PETSC_HAVE_CONSTRAINTS
2034   if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
2035   if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
2036 #endif
2037   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2038 
2039   if (!mat->ops->multtranspose) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"This matrix type does not have a multiply tranpose defined");
2040   ierr = PetscLogEventBegin(MAT_MultTranspose,mat,x,y,0);CHKERRQ(ierr);
2041   ierr = (*mat->ops->multtranspose)(mat,x,y);CHKERRQ(ierr);
2042   ierr = PetscLogEventEnd(MAT_MultTranspose,mat,x,y,0);CHKERRQ(ierr);
2043   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
2044   PetscFunctionReturn(0);
2045 }
2046 
2047 #undef __FUNCT__
2048 #define __FUNCT__ "MatMultHermitianTranspose"
2049 /*@
2050    MatMultHermitianTranspose - Computes matrix Hermitian transpose times a vector.
2051 
2052    Neighbor-wise Collective on Mat and Vec
2053 
2054    Input Parameters:
2055 +  mat - the matrix
2056 -  x   - the vector to be multilplied
2057 
2058    Output Parameters:
2059 .  y - the result
2060 
2061    Notes:
2062    The vectors x and y cannot be the same.  I.e., one cannot
2063    call MatMultHermitianTranspose(A,y,y).
2064 
2065    Level: beginner
2066 
2067    Concepts: matrix vector product^transpose
2068 
2069 .seealso: MatMult(), MatMultAdd(), MatMultHermitianTransposeAdd(), MatMultTranspose()
2070 @*/
2071 PetscErrorCode PETSCMAT_DLLEXPORT MatMultHermitianTranspose(Mat mat,Vec x,Vec y)
2072 {
2073   PetscErrorCode ierr;
2074 
2075   PetscFunctionBegin;
2076   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2077   PetscValidType(mat,1);
2078   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
2079   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
2080 
2081   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2082   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2083   if (x == y) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2084 #ifndef PETSC_HAVE_CONSTRAINTS
2085   if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
2086   if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
2087 #endif
2088   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2089 
2090   if (!mat->ops->multhermitiantranspose) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2091   ierr = PetscLogEventBegin(MAT_MultHermitianTranspose,mat,x,y,0);CHKERRQ(ierr);
2092   ierr = (*mat->ops->multhermitiantranspose)(mat,x,y);CHKERRQ(ierr);
2093   ierr = PetscLogEventEnd(MAT_MultHermitianTranspose,mat,x,y,0);CHKERRQ(ierr);
2094   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
2095   PetscFunctionReturn(0);
2096 }
2097 
2098 #undef __FUNCT__
2099 #define __FUNCT__ "MatMultAdd"
2100 /*@
2101     MatMultAdd -  Computes v3 = v2 + A * v1.
2102 
2103     Neighbor-wise Collective on Mat and Vec
2104 
2105     Input Parameters:
2106 +   mat - the matrix
2107 -   v1, v2 - the vectors
2108 
2109     Output Parameters:
2110 .   v3 - the result
2111 
2112     Notes:
2113     The vectors v1 and v3 cannot be the same.  I.e., one cannot
2114     call MatMultAdd(A,v1,v2,v1).
2115 
2116     Level: beginner
2117 
2118     Concepts: matrix vector product^addition
2119 
2120 .seealso: MatMultTranspose(), MatMult(), MatMultTransposeAdd()
2121 @*/
2122 PetscErrorCode PETSCMAT_DLLEXPORT MatMultAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2123 {
2124   PetscErrorCode ierr;
2125 
2126   PetscFunctionBegin;
2127   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2128   PetscValidType(mat,1);
2129   PetscValidHeaderSpecific(v1,VEC_CLASSID,2);
2130   PetscValidHeaderSpecific(v2,VEC_CLASSID,3);
2131   PetscValidHeaderSpecific(v3,VEC_CLASSID,4);
2132 
2133   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2134   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2135   if (mat->cmap->N != v1->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->cmap->N,v1->map->N);
2136   /* if (mat->rmap->N != v2->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->rmap->N,v2->map->N);
2137      if (mat->rmap->N != v3->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->rmap->N,v3->map->N); */
2138   if (mat->rmap->n != v3->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: local dim %D %D",mat->rmap->n,v3->map->n);
2139   if (mat->rmap->n != v2->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: local dim %D %D",mat->rmap->n,v2->map->n);
2140   if (v1 == v3) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2141   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2142 
2143   ierr = PetscLogEventBegin(MAT_MultAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2144   ierr = (*mat->ops->multadd)(mat,v1,v2,v3);CHKERRQ(ierr);
2145   ierr = PetscLogEventEnd(MAT_MultAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2146   ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr);
2147   PetscFunctionReturn(0);
2148 }
2149 
2150 #undef __FUNCT__
2151 #define __FUNCT__ "MatMultTransposeAdd"
2152 /*@
2153    MatMultTransposeAdd - Computes v3 = v2 + A' * v1.
2154 
2155    Neighbor-wise Collective on Mat and Vec
2156 
2157    Input Parameters:
2158 +  mat - the matrix
2159 -  v1, v2 - the vectors
2160 
2161    Output Parameters:
2162 .  v3 - the result
2163 
2164    Notes:
2165    The vectors v1 and v3 cannot be the same.  I.e., one cannot
2166    call MatMultTransposeAdd(A,v1,v2,v1).
2167 
2168    Level: beginner
2169 
2170    Concepts: matrix vector product^transpose and addition
2171 
2172 .seealso: MatMultTranspose(), MatMultAdd(), MatMult()
2173 @*/
2174 PetscErrorCode PETSCMAT_DLLEXPORT MatMultTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2175 {
2176   PetscErrorCode ierr;
2177 
2178   PetscFunctionBegin;
2179   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2180   PetscValidType(mat,1);
2181   PetscValidHeaderSpecific(v1,VEC_CLASSID,2);
2182   PetscValidHeaderSpecific(v2,VEC_CLASSID,3);
2183   PetscValidHeaderSpecific(v3,VEC_CLASSID,4);
2184 
2185   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2186   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2187   if (!mat->ops->multtransposeadd) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2188   if (v1 == v3) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2189   if (mat->rmap->N != v1->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N);
2190   if (mat->cmap->N != v2->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N);
2191   if (mat->cmap->N != v3->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N);
2192   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2193 
2194   ierr = PetscLogEventBegin(MAT_MultTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2195   ierr = (*mat->ops->multtransposeadd)(mat,v1,v2,v3);CHKERRQ(ierr);
2196   ierr = PetscLogEventEnd(MAT_MultTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2197   ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr);
2198   PetscFunctionReturn(0);
2199 }
2200 
2201 #undef __FUNCT__
2202 #define __FUNCT__ "MatMultHermitianTransposeAdd"
2203 /*@
2204    MatMultHermitianTransposeAdd - Computes v3 = v2 + A^H * v1.
2205 
2206    Neighbor-wise Collective on Mat and Vec
2207 
2208    Input Parameters:
2209 +  mat - the matrix
2210 -  v1, v2 - the vectors
2211 
2212    Output Parameters:
2213 .  v3 - the result
2214 
2215    Notes:
2216    The vectors v1 and v3 cannot be the same.  I.e., one cannot
2217    call MatMultHermitianTransposeAdd(A,v1,v2,v1).
2218 
2219    Level: beginner
2220 
2221    Concepts: matrix vector product^transpose and addition
2222 
2223 .seealso: MatMultHermitianTranspose(), MatMultTranspose(), MatMultAdd(), MatMult()
2224 @*/
2225 PetscErrorCode PETSCMAT_DLLEXPORT MatMultHermitianTransposeAdd(Mat mat,Vec v1,Vec v2,Vec v3)
2226 {
2227   PetscErrorCode ierr;
2228 
2229   PetscFunctionBegin;
2230   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2231   PetscValidType(mat,1);
2232   PetscValidHeaderSpecific(v1,VEC_CLASSID,2);
2233   PetscValidHeaderSpecific(v2,VEC_CLASSID,3);
2234   PetscValidHeaderSpecific(v3,VEC_CLASSID,4);
2235 
2236   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2237   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2238   if (!mat->ops->multhermitiantransposeadd) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2239   if (v1 == v3) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"v1 and v3 must be different vectors");
2240   if (mat->rmap->N != v1->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v1: global dim %D %D",mat->rmap->N,v1->map->N);
2241   if (mat->cmap->N != v2->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v2: global dim %D %D",mat->cmap->N,v2->map->N);
2242   if (mat->cmap->N != v3->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec v3: global dim %D %D",mat->cmap->N,v3->map->N);
2243   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2244 
2245   ierr = PetscLogEventBegin(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2246   ierr = (*mat->ops->multhermitiantransposeadd)(mat,v1,v2,v3);CHKERRQ(ierr);
2247   ierr = PetscLogEventEnd(MAT_MultHermitianTransposeAdd,mat,v1,v2,v3);CHKERRQ(ierr);
2248   ierr = PetscObjectStateIncrease((PetscObject)v3);CHKERRQ(ierr);
2249   PetscFunctionReturn(0);
2250 }
2251 
2252 #undef __FUNCT__
2253 #define __FUNCT__ "MatMultConstrained"
2254 /*@
2255    MatMultConstrained - The inner multiplication routine for a
2256    constrained matrix P^T A P.
2257 
2258    Neighbor-wise Collective on Mat and Vec
2259 
2260    Input Parameters:
2261 +  mat - the matrix
2262 -  x   - the vector to be multilplied
2263 
2264    Output Parameters:
2265 .  y - the result
2266 
2267    Notes:
2268    The vectors x and y cannot be the same.  I.e., one cannot
2269    call MatMult(A,y,y).
2270 
2271    Level: beginner
2272 
2273 .keywords: matrix, multiply, matrix-vector product, constraint
2274 .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2275 @*/
2276 PetscErrorCode PETSCMAT_DLLEXPORT MatMultConstrained(Mat mat,Vec x,Vec y)
2277 {
2278   PetscErrorCode ierr;
2279 
2280   PetscFunctionBegin;
2281   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2282   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
2283   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
2284   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2285   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2286   if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2287   if (mat->cmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2288   if (mat->rmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
2289   if (mat->rmap->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: local dim %D %D",mat->rmap->n,y->map->n);
2290 
2291   ierr = PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr);
2292   ierr = (*mat->ops->multconstrained)(mat,x,y);CHKERRQ(ierr);
2293   ierr = PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr);
2294   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
2295 
2296   PetscFunctionReturn(0);
2297 }
2298 
2299 #undef __FUNCT__
2300 #define __FUNCT__ "MatMultTransposeConstrained"
2301 /*@
2302    MatMultTransposeConstrained - The inner multiplication routine for a
2303    constrained matrix P^T A^T P.
2304 
2305    Neighbor-wise Collective on Mat and Vec
2306 
2307    Input Parameters:
2308 +  mat - the matrix
2309 -  x   - the vector to be multilplied
2310 
2311    Output Parameters:
2312 .  y - the result
2313 
2314    Notes:
2315    The vectors x and y cannot be the same.  I.e., one cannot
2316    call MatMult(A,y,y).
2317 
2318    Level: beginner
2319 
2320 .keywords: matrix, multiply, matrix-vector product, constraint
2321 .seealso: MatMult(), MatMultTranspose(), MatMultAdd(), MatMultTransposeAdd()
2322 @*/
2323 PetscErrorCode PETSCMAT_DLLEXPORT MatMultTransposeConstrained(Mat mat,Vec x,Vec y)
2324 {
2325   PetscErrorCode ierr;
2326 
2327   PetscFunctionBegin;
2328   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2329   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
2330   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
2331   if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2332   if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2333   if (x == y) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"x and y must be different vectors");
2334   if (mat->rmap->N != x->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2335   if (mat->cmap->N != y->map->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
2336 
2337   ierr = PetscLogEventBegin(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr);
2338   ierr = (*mat->ops->multtransposeconstrained)(mat,x,y);CHKERRQ(ierr);
2339   ierr = PetscLogEventEnd(MAT_MultConstrained,mat,x,y,0);CHKERRQ(ierr);
2340   ierr = PetscObjectStateIncrease((PetscObject)y);CHKERRQ(ierr);
2341 
2342   PetscFunctionReturn(0);
2343 }
2344 
2345 #undef __FUNCT__
2346 #define __FUNCT__ "MatGetFactorType"
2347 /*@C
2348    MatGetFactorType - gets the type of factorization it is
2349 
2350    Note Collective
2351    as the flag
2352 
2353    Input Parameters:
2354 .  mat - the matrix
2355 
2356    Output Parameters:
2357 .  t - the type, one of MAT_FACTOR_NONE, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ILU, MAT_FACTOR_ICC,MAT_FACTOR_ILUDT
2358 
2359     Level: intermediate
2360 
2361 .seealso:    MatFactorType, MatGetFactor()
2362 @*/
2363 PetscErrorCode PETSCMAT_DLLEXPORT MatGetFactorType(Mat mat,MatFactorType *t)
2364 {
2365   PetscFunctionBegin;
2366   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2367   PetscValidType(mat,1);
2368   *t = mat->factortype;
2369   PetscFunctionReturn(0);
2370 }
2371 
2372 /* ------------------------------------------------------------*/
2373 #undef __FUNCT__
2374 #define __FUNCT__ "MatGetInfo"
2375 /*@C
2376    MatGetInfo - Returns information about matrix storage (number of
2377    nonzeros, memory, etc.).
2378 
2379    Collective on Mat if MAT_GLOBAL_MAX or MAT_GLOBAL_SUM is used as the flag
2380 
2381    Input Parameters:
2382 .  mat - the matrix
2383 
2384    Output Parameters:
2385 +  flag - flag indicating the type of parameters to be returned
2386    (MAT_LOCAL - local matrix, MAT_GLOBAL_MAX - maximum over all processors,
2387    MAT_GLOBAL_SUM - sum over all processors)
2388 -  info - matrix information context
2389 
2390    Notes:
2391    The MatInfo context contains a variety of matrix data, including
2392    number of nonzeros allocated and used, number of mallocs during
2393    matrix assembly, etc.  Additional information for factored matrices
2394    is provided (such as the fill ratio, number of mallocs during
2395    factorization, etc.).  Much of this info is printed to PETSC_STDOUT
2396    when using the runtime options
2397 $       -info -mat_view_info
2398 
2399    Example for C/C++ Users:
2400    See the file ${PETSC_DIR}/include/petscmat.h for a complete list of
2401    data within the MatInfo context.  For example,
2402 .vb
2403       MatInfo info;
2404       Mat     A;
2405       double  mal, nz_a, nz_u;
2406 
2407       MatGetInfo(A,MAT_LOCAL,&info);
2408       mal  = info.mallocs;
2409       nz_a = info.nz_allocated;
2410 .ve
2411 
2412    Example for Fortran Users:
2413    Fortran users should declare info as a double precision
2414    array of dimension MAT_INFO_SIZE, and then extract the parameters
2415    of interest.  See the file ${PETSC_DIR}/include/finclude/petscmat.h
2416    a complete list of parameter names.
2417 .vb
2418       double  precision info(MAT_INFO_SIZE)
2419       double  precision mal, nz_a
2420       Mat     A
2421       integer ierr
2422 
2423       call MatGetInfo(A,MAT_LOCAL,info,ierr)
2424       mal = info(MAT_INFO_MALLOCS)
2425       nz_a = info(MAT_INFO_NZ_ALLOCATED)
2426 .ve
2427 
2428     Level: intermediate
2429 
2430     Concepts: matrices^getting information on
2431 
2432     Developer Note: fortran interface is not autogenerated as the f90
2433     interface defintion cannot be generated correctly [due to MatInfo]
2434 
2435 @*/
2436 PetscErrorCode PETSCMAT_DLLEXPORT MatGetInfo(Mat mat,MatInfoType flag,MatInfo *info)
2437 {
2438   PetscErrorCode ierr;
2439 
2440   PetscFunctionBegin;
2441   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2442   PetscValidType(mat,1);
2443   PetscValidPointer(info,3);
2444   if (!mat->ops->getinfo) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2445   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2446   ierr = (*mat->ops->getinfo)(mat,flag,info);CHKERRQ(ierr);
2447   PetscFunctionReturn(0);
2448 }
2449 
2450 /* ----------------------------------------------------------*/
2451 
2452 #undef __FUNCT__
2453 #define __FUNCT__ "MatLUFactor"
2454 /*@C
2455    MatLUFactor - Performs in-place LU factorization of matrix.
2456 
2457    Collective on Mat
2458 
2459    Input Parameters:
2460 +  mat - the matrix
2461 .  row - row permutation
2462 .  col - column permutation
2463 -  info - options for factorization, includes
2464 $          fill - expected fill as ratio of original fill.
2465 $          dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
2466 $                   Run with the option -info to determine an optimal value to use
2467 
2468    Notes:
2469    Most users should employ the simplified KSP interface for linear solvers
2470    instead of working directly with matrix algebra routines such as this.
2471    See, e.g., KSPCreate().
2472 
2473    This changes the state of the matrix to a factored matrix; it cannot be used
2474    for example with MatSetValues() unless one first calls MatSetUnfactored().
2475 
2476    Level: developer
2477 
2478    Concepts: matrices^LU factorization
2479 
2480 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(),
2481           MatGetOrdering(), MatSetUnfactored(), MatFactorInfo
2482 
2483     Developer Note: fortran interface is not autogenerated as the f90
2484     interface defintion cannot be generated correctly [due to MatFactorInfo]
2485 
2486 @*/
2487 PetscErrorCode PETSCMAT_DLLEXPORT MatLUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info)
2488 {
2489   PetscErrorCode ierr;
2490 
2491   PetscFunctionBegin;
2492   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2493   if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2);
2494   if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3);
2495   PetscValidPointer(info,4);
2496   PetscValidType(mat,1);
2497   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2498   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2499   if (!mat->ops->lufactor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2500   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2501 
2502   ierr = PetscLogEventBegin(MAT_LUFactor,mat,row,col,0);CHKERRQ(ierr);
2503   ierr = (*mat->ops->lufactor)(mat,row,col,info);CHKERRQ(ierr);
2504   ierr = PetscLogEventEnd(MAT_LUFactor,mat,row,col,0);CHKERRQ(ierr);
2505   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
2506   PetscFunctionReturn(0);
2507 }
2508 
2509 #undef __FUNCT__
2510 #define __FUNCT__ "MatILUFactor"
2511 /*@C
2512    MatILUFactor - Performs in-place ILU factorization of matrix.
2513 
2514    Collective on Mat
2515 
2516    Input Parameters:
2517 +  mat - the matrix
2518 .  row - row permutation
2519 .  col - column permutation
2520 -  info - structure containing
2521 $      levels - number of levels of fill.
2522 $      expected fill - as ratio of original fill.
2523 $      1 or 0 - indicating force fill on diagonal (improves robustness for matrices
2524                 missing diagonal entries)
2525 
2526    Notes:
2527    Probably really in-place only when level of fill is zero, otherwise allocates
2528    new space to store factored matrix and deletes previous memory.
2529 
2530    Most users should employ the simplified KSP interface for linear solvers
2531    instead of working directly with matrix algebra routines such as this.
2532    See, e.g., KSPCreate().
2533 
2534    Level: developer
2535 
2536    Concepts: matrices^ILU factorization
2537 
2538 .seealso: MatILUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo
2539 
2540     Developer Note: fortran interface is not autogenerated as the f90
2541     interface defintion cannot be generated correctly [due to MatFactorInfo]
2542 
2543 @*/
2544 PetscErrorCode PETSCMAT_DLLEXPORT MatILUFactor(Mat mat,IS row,IS col,const MatFactorInfo *info)
2545 {
2546   PetscErrorCode ierr;
2547 
2548   PetscFunctionBegin;
2549   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2550   if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2);
2551   if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3);
2552   PetscValidPointer(info,4);
2553   PetscValidType(mat,1);
2554   if (mat->rmap->N != mat->cmap->N) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONG,"matrix must be square");
2555   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2556   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2557   if (!mat->ops->ilufactor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2558   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2559 
2560   ierr = PetscLogEventBegin(MAT_ILUFactor,mat,row,col,0);CHKERRQ(ierr);
2561   ierr = (*mat->ops->ilufactor)(mat,row,col,info);CHKERRQ(ierr);
2562   ierr = PetscLogEventEnd(MAT_ILUFactor,mat,row,col,0);CHKERRQ(ierr);
2563   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
2564   PetscFunctionReturn(0);
2565 }
2566 
2567 #undef __FUNCT__
2568 #define __FUNCT__ "MatLUFactorSymbolic"
2569 /*@C
2570    MatLUFactorSymbolic - Performs symbolic LU factorization of matrix.
2571    Call this routine before calling MatLUFactorNumeric().
2572 
2573    Collective on Mat
2574 
2575    Input Parameters:
2576 +  fact - the factor matrix obtained with MatGetFactor()
2577 .  mat - the matrix
2578 .  row, col - row and column permutations
2579 -  info - options for factorization, includes
2580 $          fill - expected fill as ratio of original fill.
2581 $          dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
2582 $                   Run with the option -info to determine an optimal value to use
2583 
2584 
2585    Notes:
2586    See the <a href="../../docs/manual.pdf">users manual</a> for additional information about
2587    choosing the fill factor for better efficiency.
2588 
2589    Most users should employ the simplified KSP interface for linear solvers
2590    instead of working directly with matrix algebra routines such as this.
2591    See, e.g., KSPCreate().
2592 
2593    Level: developer
2594 
2595    Concepts: matrices^LU symbolic factorization
2596 
2597 .seealso: MatLUFactor(), MatLUFactorNumeric(), MatCholeskyFactor(), MatFactorInfo
2598 
2599     Developer Note: fortran interface is not autogenerated as the f90
2600     interface defintion cannot be generated correctly [due to MatFactorInfo]
2601 
2602 @*/
2603 PetscErrorCode PETSCMAT_DLLEXPORT MatLUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info)
2604 {
2605   PetscErrorCode ierr;
2606 
2607   PetscFunctionBegin;
2608   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2609   if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2);
2610   if (col) PetscValidHeaderSpecific(col,IS_CLASSID,3);
2611   PetscValidPointer(info,4);
2612   PetscValidType(mat,1);
2613   PetscValidPointer(fact,5);
2614   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2615   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2616   if (!(fact)->ops->lufactorsymbolic) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Matrix type %s  symbolic LU",((PetscObject)mat)->type_name);
2617   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2618 
2619   ierr = PetscLogEventBegin(MAT_LUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr);
2620   ierr = (fact->ops->lufactorsymbolic)(fact,mat,row,col,info);CHKERRQ(ierr);
2621   ierr = PetscLogEventEnd(MAT_LUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr);
2622   ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr);
2623   PetscFunctionReturn(0);
2624 }
2625 
2626 #undef __FUNCT__
2627 #define __FUNCT__ "MatLUFactorNumeric"
2628 /*@C
2629    MatLUFactorNumeric - Performs numeric LU factorization of a matrix.
2630    Call this routine after first calling MatLUFactorSymbolic().
2631 
2632    Collective on Mat
2633 
2634    Input Parameters:
2635 +  fact - the factor matrix obtained with MatGetFactor()
2636 .  mat - the matrix
2637 -  info - options for factorization
2638 
2639    Notes:
2640    See MatLUFactor() for in-place factorization.  See
2641    MatCholeskyFactorNumeric() for the symmetric, positive definite case.
2642 
2643    Most users should employ the simplified KSP interface for linear solvers
2644    instead of working directly with matrix algebra routines such as this.
2645    See, e.g., KSPCreate().
2646 
2647    Level: developer
2648 
2649    Concepts: matrices^LU numeric factorization
2650 
2651 .seealso: MatLUFactorSymbolic(), MatLUFactor(), MatCholeskyFactor()
2652 
2653     Developer Note: fortran interface is not autogenerated as the f90
2654     interface defintion cannot be generated correctly [due to MatFactorInfo]
2655 
2656 @*/
2657 PetscErrorCode PETSCMAT_DLLEXPORT MatLUFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info)
2658 {
2659   PetscErrorCode ierr;
2660 
2661   PetscFunctionBegin;
2662   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2663   PetscValidType(mat,1);
2664   PetscValidPointer(fact,2);
2665   PetscValidHeaderSpecific(fact,MAT_CLASSID,2);
2666   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2667   if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) {
2668     SETERRQ4(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dimensions are different %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N);
2669   }
2670   if (!(fact)->ops->lufactornumeric) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2671   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2672   ierr = PetscLogEventBegin(MAT_LUFactorNumeric,mat,fact,0,0);CHKERRQ(ierr);
2673   ierr = (fact->ops->lufactornumeric)(fact,mat,info);CHKERRQ(ierr);
2674   ierr = PetscLogEventEnd(MAT_LUFactorNumeric,mat,fact,0,0);CHKERRQ(ierr);
2675 
2676   ierr = MatView_Private(fact);CHKERRQ(ierr);
2677   ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr);
2678   PetscFunctionReturn(0);
2679 }
2680 
2681 #undef __FUNCT__
2682 #define __FUNCT__ "MatCholeskyFactor"
2683 /*@C
2684    MatCholeskyFactor - Performs in-place Cholesky factorization of a
2685    symmetric matrix.
2686 
2687    Collective on Mat
2688 
2689    Input Parameters:
2690 +  mat - the matrix
2691 .  perm - row and column permutations
2692 -  f - expected fill as ratio of original fill
2693 
2694    Notes:
2695    See MatLUFactor() for the nonsymmetric case.  See also
2696    MatCholeskyFactorSymbolic(), and MatCholeskyFactorNumeric().
2697 
2698    Most users should employ the simplified KSP interface for linear solvers
2699    instead of working directly with matrix algebra routines such as this.
2700    See, e.g., KSPCreate().
2701 
2702    Level: developer
2703 
2704    Concepts: matrices^Cholesky factorization
2705 
2706 .seealso: MatLUFactor(), MatCholeskyFactorSymbolic(), MatCholeskyFactorNumeric()
2707           MatGetOrdering()
2708 
2709     Developer Note: fortran interface is not autogenerated as the f90
2710     interface defintion cannot be generated correctly [due to MatFactorInfo]
2711 
2712 @*/
2713 PetscErrorCode PETSCMAT_DLLEXPORT MatCholeskyFactor(Mat mat,IS perm,const MatFactorInfo *info)
2714 {
2715   PetscErrorCode ierr;
2716 
2717   PetscFunctionBegin;
2718   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2719   PetscValidType(mat,1);
2720   PetscValidHeaderSpecific(perm,IS_CLASSID,2);
2721   PetscValidPointer(info,3);
2722   if (mat->rmap->N != mat->cmap->N) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONG,"Matrix must be square");
2723   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2724   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2725   if (!mat->ops->choleskyfactor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2726   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2727 
2728   ierr = PetscLogEventBegin(MAT_CholeskyFactor,mat,perm,0,0);CHKERRQ(ierr);
2729   ierr = (*mat->ops->choleskyfactor)(mat,perm,info);CHKERRQ(ierr);
2730   ierr = PetscLogEventEnd(MAT_CholeskyFactor,mat,perm,0,0);CHKERRQ(ierr);
2731   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
2732   PetscFunctionReturn(0);
2733 }
2734 
2735 #undef __FUNCT__
2736 #define __FUNCT__ "MatCholeskyFactorSymbolic"
2737 /*@C
2738    MatCholeskyFactorSymbolic - Performs symbolic Cholesky factorization
2739    of a symmetric matrix.
2740 
2741    Collective on Mat
2742 
2743    Input Parameters:
2744 +  fact - the factor matrix obtained with MatGetFactor()
2745 .  mat - the matrix
2746 .  perm - row and column permutations
2747 -  info - options for factorization, includes
2748 $          fill - expected fill as ratio of original fill.
2749 $          dtcol - pivot tolerance (0 no pivot, 1 full column pivoting)
2750 $                   Run with the option -info to determine an optimal value to use
2751 
2752    Notes:
2753    See MatLUFactorSymbolic() for the nonsymmetric case.  See also
2754    MatCholeskyFactor() and MatCholeskyFactorNumeric().
2755 
2756    Most users should employ the simplified KSP interface for linear solvers
2757    instead of working directly with matrix algebra routines such as this.
2758    See, e.g., KSPCreate().
2759 
2760    Level: developer
2761 
2762    Concepts: matrices^Cholesky symbolic factorization
2763 
2764 .seealso: MatLUFactorSymbolic(), MatCholeskyFactor(), MatCholeskyFactorNumeric()
2765           MatGetOrdering()
2766 
2767     Developer Note: fortran interface is not autogenerated as the f90
2768     interface defintion cannot be generated correctly [due to MatFactorInfo]
2769 
2770 @*/
2771 PetscErrorCode PETSCMAT_DLLEXPORT MatCholeskyFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info)
2772 {
2773   PetscErrorCode ierr;
2774 
2775   PetscFunctionBegin;
2776   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2777   PetscValidType(mat,1);
2778   if (perm) PetscValidHeaderSpecific(perm,IS_CLASSID,2);
2779   PetscValidPointer(info,3);
2780   PetscValidPointer(fact,4);
2781   if (mat->rmap->N != mat->cmap->N) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONG,"Matrix must be square");
2782   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2783   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2784   if (!(fact)->ops->choleskyfactorsymbolic) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2785   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2786 
2787   ierr = PetscLogEventBegin(MAT_CholeskyFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr);
2788   ierr = (fact->ops->choleskyfactorsymbolic)(fact,mat,perm,info);CHKERRQ(ierr);
2789   ierr = PetscLogEventEnd(MAT_CholeskyFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr);
2790   ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr);
2791   PetscFunctionReturn(0);
2792 }
2793 
2794 #undef __FUNCT__
2795 #define __FUNCT__ "MatCholeskyFactorNumeric"
2796 /*@C
2797    MatCholeskyFactorNumeric - Performs numeric Cholesky factorization
2798    of a symmetric matrix. Call this routine after first calling
2799    MatCholeskyFactorSymbolic().
2800 
2801    Collective on Mat
2802 
2803    Input Parameters:
2804 +  fact - the factor matrix obtained with MatGetFactor()
2805 .  mat - the initial matrix
2806 .  info - options for factorization
2807 -  fact - the symbolic factor of mat
2808 
2809 
2810    Notes:
2811    Most users should employ the simplified KSP interface for linear solvers
2812    instead of working directly with matrix algebra routines such as this.
2813    See, e.g., KSPCreate().
2814 
2815    Level: developer
2816 
2817    Concepts: matrices^Cholesky numeric factorization
2818 
2819 .seealso: MatCholeskyFactorSymbolic(), MatCholeskyFactor(), MatLUFactorNumeric()
2820 
2821     Developer Note: fortran interface is not autogenerated as the f90
2822     interface defintion cannot be generated correctly [due to MatFactorInfo]
2823 
2824 @*/
2825 PetscErrorCode PETSCMAT_DLLEXPORT MatCholeskyFactorNumeric(Mat fact,Mat mat,const MatFactorInfo *info)
2826 {
2827   PetscErrorCode ierr;
2828 
2829   PetscFunctionBegin;
2830   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2831   PetscValidType(mat,1);
2832   PetscValidPointer(fact,2);
2833   PetscValidHeaderSpecific(fact,MAT_CLASSID,2);
2834   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2835   if (!(fact)->ops->choleskyfactornumeric) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2836   if (mat->rmap->N != (fact)->rmap->N || mat->cmap->N != (fact)->cmap->N) {
2837     SETERRQ4(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Mat fact: global dim %D should = %D %D should = %D",mat->rmap->N,(fact)->rmap->N,mat->cmap->N,(fact)->cmap->N);
2838   }
2839   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2840 
2841   ierr = PetscLogEventBegin(MAT_CholeskyFactorNumeric,mat,fact,0,0);CHKERRQ(ierr);
2842   ierr = (fact->ops->choleskyfactornumeric)(fact,mat,info);CHKERRQ(ierr);
2843   ierr = PetscLogEventEnd(MAT_CholeskyFactorNumeric,mat,fact,0,0);CHKERRQ(ierr);
2844 
2845   ierr = MatView_Private(fact);CHKERRQ(ierr);
2846   ierr = PetscObjectStateIncrease((PetscObject)fact);CHKERRQ(ierr);
2847   PetscFunctionReturn(0);
2848 }
2849 
2850 /* ----------------------------------------------------------------*/
2851 #undef __FUNCT__
2852 #define __FUNCT__ "MatSolve"
2853 /*@
2854    MatSolve - Solves A x = b, given a factored matrix.
2855 
2856    Neighbor-wise Collective on Mat and Vec
2857 
2858    Input Parameters:
2859 +  mat - the factored matrix
2860 -  b - the right-hand-side vector
2861 
2862    Output Parameter:
2863 .  x - the result vector
2864 
2865    Notes:
2866    The vectors b and x cannot be the same.  I.e., one cannot
2867    call MatSolve(A,x,x).
2868 
2869    Notes:
2870    Most users should employ the simplified KSP interface for linear solvers
2871    instead of working directly with matrix algebra routines such as this.
2872    See, e.g., KSPCreate().
2873 
2874    Level: developer
2875 
2876    Concepts: matrices^triangular solves
2877 
2878 .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd()
2879 @*/
2880 PetscErrorCode PETSCMAT_DLLEXPORT MatSolve(Mat mat,Vec b,Vec x)
2881 {
2882   PetscErrorCode ierr;
2883 
2884   PetscFunctionBegin;
2885   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
2886   PetscValidType(mat,1);
2887   PetscValidHeaderSpecific(b,VEC_CLASSID,2);
2888   PetscValidHeaderSpecific(x,VEC_CLASSID,3);
2889   PetscCheckSameComm(mat,1,b,2);
2890   PetscCheckSameComm(mat,1,x,3);
2891   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
2892   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
2893   if (mat->cmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
2894   if (mat->rmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
2895   if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
2896   if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(0);
2897   if (!mat->ops->solve) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
2898   ierr = MatPreallocated(mat);CHKERRQ(ierr);
2899 
2900   ierr = PetscLogEventBegin(MAT_Solve,mat,b,x,0);CHKERRQ(ierr);
2901   ierr = (*mat->ops->solve)(mat,b,x);CHKERRQ(ierr);
2902   ierr = PetscLogEventEnd(MAT_Solve,mat,b,x,0);CHKERRQ(ierr);
2903   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
2904   PetscFunctionReturn(0);
2905 }
2906 
2907 #undef __FUNCT__
2908 #define __FUNCT__ "MatMatSolve_Basic"
2909 PetscErrorCode PETSCMAT_DLLEXPORT MatMatSolve_Basic(Mat A,Mat B,Mat X)
2910 {
2911   PetscErrorCode ierr;
2912   Vec            b,x;
2913   PetscInt       m,N,i;
2914   PetscScalar    *bb,*xx;
2915 
2916   PetscFunctionBegin;
2917   ierr = MatGetArray(B,&bb);CHKERRQ(ierr);
2918   ierr = MatGetArray(X,&xx);CHKERRQ(ierr);
2919   ierr = MatGetLocalSize(B,&m,PETSC_NULL);CHKERRQ(ierr);  /* number local rows */
2920   ierr = MatGetSize(B,PETSC_NULL,&N);CHKERRQ(ierr);       /* total columns in dense matrix */
2921   ierr = MatGetVecs(A,&x,&b);CHKERRQ(ierr);
2922   for (i=0; i<N; i++) {
2923     ierr = VecPlaceArray(b,bb + i*m);CHKERRQ(ierr);
2924     ierr = VecPlaceArray(x,xx + i*m);CHKERRQ(ierr);
2925     ierr = MatSolve(A,b,x);CHKERRQ(ierr);
2926     ierr = VecResetArray(x);CHKERRQ(ierr);
2927     ierr = VecResetArray(b);CHKERRQ(ierr);
2928   }
2929   ierr = VecDestroy(b);CHKERRQ(ierr);
2930   ierr = VecDestroy(x);CHKERRQ(ierr);
2931   ierr = MatRestoreArray(B,&bb);CHKERRQ(ierr);
2932   ierr = MatRestoreArray(X,&xx);CHKERRQ(ierr);
2933   PetscFunctionReturn(0);
2934 }
2935 
2936 #undef __FUNCT__
2937 #define __FUNCT__ "MatMatSolve"
2938 /*@
2939    MatMatSolve - Solves A X = B, given a factored matrix.
2940 
2941    Neighbor-wise Collective on Mat
2942 
2943    Input Parameters:
2944 +  mat - the factored matrix
2945 -  B - the right-hand-side matrix  (dense matrix)
2946 
2947    Output Parameter:
2948 .  X - the result matrix (dense matrix)
2949 
2950    Notes:
2951    The matrices b and x cannot be the same.  I.e., one cannot
2952    call MatMatSolve(A,x,x).
2953 
2954    Notes:
2955    Most users should usually employ the simplified KSP interface for linear solvers
2956    instead of working directly with matrix algebra routines such as this.
2957    See, e.g., KSPCreate(). However KSP can only solve for one vector (column of X)
2958    at a time.
2959 
2960    Level: developer
2961 
2962    Concepts: matrices^triangular solves
2963 
2964 .seealso: MatMatSolveAdd(), MatMatSolveTranspose(), MatMatSolveTransposeAdd(), MatLUFactor(), MatCholeskyFactor()
2965 @*/
2966 PetscErrorCode PETSCMAT_DLLEXPORT MatMatSolve(Mat A,Mat B,Mat X)
2967 {
2968   PetscErrorCode ierr;
2969 
2970   PetscFunctionBegin;
2971   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
2972   PetscValidType(A,1);
2973   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
2974   PetscValidHeaderSpecific(X,MAT_CLASSID,3);
2975   PetscCheckSameComm(A,1,B,2);
2976   PetscCheckSameComm(A,1,X,3);
2977   if (X == B) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_IDN,"X and B must be different matrices");
2978   if (!A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
2979   if (A->cmap->N != X->rmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Mat A,Mat X: global dim %D %D",A->cmap->N,X->rmap->N);
2980   if (A->rmap->N != B->rmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D",A->rmap->N,B->rmap->N);
2981   if (A->rmap->n != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: local dim %D %D",A->rmap->n,B->rmap->n);
2982   if (!A->rmap->N && !A->cmap->N) PetscFunctionReturn(0);
2983   ierr = MatPreallocated(A);CHKERRQ(ierr);
2984 
2985   ierr = PetscLogEventBegin(MAT_MatSolve,A,B,X,0);CHKERRQ(ierr);
2986   if (!A->ops->matsolve) {
2987     ierr = PetscInfo1(A,"Mat type %s using basic MatMatSolve",((PetscObject)A)->type_name);CHKERRQ(ierr);
2988     ierr = MatMatSolve_Basic(A,B,X);CHKERRQ(ierr);
2989   } else {
2990     ierr = (*A->ops->matsolve)(A,B,X);CHKERRQ(ierr);
2991   }
2992   ierr = PetscLogEventEnd(MAT_MatSolve,A,B,X,0);CHKERRQ(ierr);
2993   ierr = PetscObjectStateIncrease((PetscObject)X);CHKERRQ(ierr);
2994   PetscFunctionReturn(0);
2995 }
2996 
2997 
2998 #undef __FUNCT__
2999 #define __FUNCT__ "MatForwardSolve"
3000 /*@
3001    MatForwardSolve - Solves L x = b, given a factored matrix, A = LU, or
3002                             U^T*D^(1/2) x = b, given a factored symmetric matrix, A = U^T*D*U,
3003 
3004    Neighbor-wise Collective on Mat and Vec
3005 
3006    Input Parameters:
3007 +  mat - the factored matrix
3008 -  b - the right-hand-side vector
3009 
3010    Output Parameter:
3011 .  x - the result vector
3012 
3013    Notes:
3014    MatSolve() should be used for most applications, as it performs
3015    a forward solve followed by a backward solve.
3016 
3017    The vectors b and x cannot be the same,  i.e., one cannot
3018    call MatForwardSolve(A,x,x).
3019 
3020    For matrix in seqsbaij format with block size larger than 1,
3021    the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3022    MatForwardSolve() solves U^T*D y = b, and
3023    MatBackwardSolve() solves U x = y.
3024    Thus they do not provide a symmetric preconditioner.
3025 
3026    Most users should employ the simplified KSP interface for linear solvers
3027    instead of working directly with matrix algebra routines such as this.
3028    See, e.g., KSPCreate().
3029 
3030    Level: developer
3031 
3032    Concepts: matrices^forward solves
3033 
3034 .seealso: MatSolve(), MatBackwardSolve()
3035 @*/
3036 PetscErrorCode PETSCMAT_DLLEXPORT MatForwardSolve(Mat mat,Vec b,Vec x)
3037 {
3038   PetscErrorCode ierr;
3039 
3040   PetscFunctionBegin;
3041   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3042   PetscValidType(mat,1);
3043   PetscValidHeaderSpecific(b,VEC_CLASSID,2);
3044   PetscValidHeaderSpecific(x,VEC_CLASSID,3);
3045   PetscCheckSameComm(mat,1,b,2);
3046   PetscCheckSameComm(mat,1,x,3);
3047   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3048   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3049   if (!mat->ops->forwardsolve) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3050   if (mat->cmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3051   if (mat->rmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3052   if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3053   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3054   ierr = PetscLogEventBegin(MAT_ForwardSolve,mat,b,x,0);CHKERRQ(ierr);
3055   ierr = (*mat->ops->forwardsolve)(mat,b,x);CHKERRQ(ierr);
3056   ierr = PetscLogEventEnd(MAT_ForwardSolve,mat,b,x,0);CHKERRQ(ierr);
3057   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3058   PetscFunctionReturn(0);
3059 }
3060 
3061 #undef __FUNCT__
3062 #define __FUNCT__ "MatBackwardSolve"
3063 /*@
3064    MatBackwardSolve - Solves U x = b, given a factored matrix, A = LU.
3065                              D^(1/2) U x = b, given a factored symmetric matrix, A = U^T*D*U,
3066 
3067    Neighbor-wise Collective on Mat and Vec
3068 
3069    Input Parameters:
3070 +  mat - the factored matrix
3071 -  b - the right-hand-side vector
3072 
3073    Output Parameter:
3074 .  x - the result vector
3075 
3076    Notes:
3077    MatSolve() should be used for most applications, as it performs
3078    a forward solve followed by a backward solve.
3079 
3080    The vectors b and x cannot be the same.  I.e., one cannot
3081    call MatBackwardSolve(A,x,x).
3082 
3083    For matrix in seqsbaij format with block size larger than 1,
3084    the diagonal blocks are not implemented as D = D^(1/2) * D^(1/2) yet.
3085    MatForwardSolve() solves U^T*D y = b, and
3086    MatBackwardSolve() solves U x = y.
3087    Thus they do not provide a symmetric preconditioner.
3088 
3089    Most users should employ the simplified KSP interface for linear solvers
3090    instead of working directly with matrix algebra routines such as this.
3091    See, e.g., KSPCreate().
3092 
3093    Level: developer
3094 
3095    Concepts: matrices^backward solves
3096 
3097 .seealso: MatSolve(), MatForwardSolve()
3098 @*/
3099 PetscErrorCode PETSCMAT_DLLEXPORT MatBackwardSolve(Mat mat,Vec b,Vec x)
3100 {
3101   PetscErrorCode ierr;
3102 
3103   PetscFunctionBegin;
3104   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3105   PetscValidType(mat,1);
3106   PetscValidHeaderSpecific(b,VEC_CLASSID,2);
3107   PetscValidHeaderSpecific(x,VEC_CLASSID,3);
3108   PetscCheckSameComm(mat,1,b,2);
3109   PetscCheckSameComm(mat,1,x,3);
3110   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3111   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3112   if (!mat->ops->backwardsolve) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3113   if (mat->cmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3114   if (mat->rmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3115   if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3116   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3117 
3118   ierr = PetscLogEventBegin(MAT_BackwardSolve,mat,b,x,0);CHKERRQ(ierr);
3119   ierr = (*mat->ops->backwardsolve)(mat,b,x);CHKERRQ(ierr);
3120   ierr = PetscLogEventEnd(MAT_BackwardSolve,mat,b,x,0);CHKERRQ(ierr);
3121   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3122   PetscFunctionReturn(0);
3123 }
3124 
3125 #undef __FUNCT__
3126 #define __FUNCT__ "MatSolveAdd"
3127 /*@
3128    MatSolveAdd - Computes x = y + inv(A)*b, given a factored matrix.
3129 
3130    Neighbor-wise Collective on Mat and Vec
3131 
3132    Input Parameters:
3133 +  mat - the factored matrix
3134 .  b - the right-hand-side vector
3135 -  y - the vector to be added to
3136 
3137    Output Parameter:
3138 .  x - the result vector
3139 
3140    Notes:
3141    The vectors b and x cannot be the same.  I.e., one cannot
3142    call MatSolveAdd(A,x,y,x).
3143 
3144    Most users should employ the simplified KSP interface for linear solvers
3145    instead of working directly with matrix algebra routines such as this.
3146    See, e.g., KSPCreate().
3147 
3148    Level: developer
3149 
3150    Concepts: matrices^triangular solves
3151 
3152 .seealso: MatSolve(), MatSolveTranspose(), MatSolveTransposeAdd()
3153 @*/
3154 PetscErrorCode PETSCMAT_DLLEXPORT MatSolveAdd(Mat mat,Vec b,Vec y,Vec x)
3155 {
3156   PetscScalar    one = 1.0;
3157   Vec            tmp;
3158   PetscErrorCode ierr;
3159 
3160   PetscFunctionBegin;
3161   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3162   PetscValidType(mat,1);
3163   PetscValidHeaderSpecific(y,VEC_CLASSID,2);
3164   PetscValidHeaderSpecific(b,VEC_CLASSID,3);
3165   PetscValidHeaderSpecific(x,VEC_CLASSID,4);
3166   PetscCheckSameComm(mat,1,b,2);
3167   PetscCheckSameComm(mat,1,y,2);
3168   PetscCheckSameComm(mat,1,x,3);
3169   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3170   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3171   if (mat->cmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3172   if (mat->rmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3173   if (mat->rmap->N != y->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->rmap->N,y->map->N);
3174   if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3175   if (x->map->n != y->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n);
3176   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3177 
3178   ierr = PetscLogEventBegin(MAT_SolveAdd,mat,b,x,y);CHKERRQ(ierr);
3179   if (mat->ops->solveadd)  {
3180     ierr = (*mat->ops->solveadd)(mat,b,y,x);CHKERRQ(ierr);
3181   } else {
3182     /* do the solve then the add manually */
3183     if (x != y) {
3184       ierr = MatSolve(mat,b,x);CHKERRQ(ierr);
3185       ierr = VecAXPY(x,one,y);CHKERRQ(ierr);
3186     } else {
3187       ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr);
3188       ierr = PetscLogObjectParent(mat,tmp);CHKERRQ(ierr);
3189       ierr = VecCopy(x,tmp);CHKERRQ(ierr);
3190       ierr = MatSolve(mat,b,x);CHKERRQ(ierr);
3191       ierr = VecAXPY(x,one,tmp);CHKERRQ(ierr);
3192       ierr = VecDestroy(tmp);CHKERRQ(ierr);
3193     }
3194   }
3195   ierr = PetscLogEventEnd(MAT_SolveAdd,mat,b,x,y);CHKERRQ(ierr);
3196   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3197   PetscFunctionReturn(0);
3198 }
3199 
3200 #undef __FUNCT__
3201 #define __FUNCT__ "MatSolveTranspose"
3202 /*@
3203    MatSolveTranspose - Solves A' x = b, given a factored matrix.
3204 
3205    Neighbor-wise Collective on Mat and Vec
3206 
3207    Input Parameters:
3208 +  mat - the factored matrix
3209 -  b - the right-hand-side vector
3210 
3211    Output Parameter:
3212 .  x - the result vector
3213 
3214    Notes:
3215    The vectors b and x cannot be the same.  I.e., one cannot
3216    call MatSolveTranspose(A,x,x).
3217 
3218    Most users should employ the simplified KSP interface for linear solvers
3219    instead of working directly with matrix algebra routines such as this.
3220    See, e.g., KSPCreate().
3221 
3222    Level: developer
3223 
3224    Concepts: matrices^triangular solves
3225 
3226 .seealso: MatSolve(), MatSolveAdd(), MatSolveTransposeAdd()
3227 @*/
3228 PetscErrorCode PETSCMAT_DLLEXPORT MatSolveTranspose(Mat mat,Vec b,Vec x)
3229 {
3230   PetscErrorCode ierr;
3231 
3232   PetscFunctionBegin;
3233   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3234   PetscValidType(mat,1);
3235   PetscValidHeaderSpecific(b,VEC_CLASSID,2);
3236   PetscValidHeaderSpecific(x,VEC_CLASSID,3);
3237   PetscCheckSameComm(mat,1,b,2);
3238   PetscCheckSameComm(mat,1,x,3);
3239   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3240   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3241   if (!mat->ops->solvetranspose) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Matrix type %s",((PetscObject)mat)->type_name);
3242   if (mat->rmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
3243   if (mat->cmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N);
3244   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3245   ierr = PetscLogEventBegin(MAT_SolveTranspose,mat,b,x,0);CHKERRQ(ierr);
3246   ierr = (*mat->ops->solvetranspose)(mat,b,x);CHKERRQ(ierr);
3247   ierr = PetscLogEventEnd(MAT_SolveTranspose,mat,b,x,0);CHKERRQ(ierr);
3248   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3249   PetscFunctionReturn(0);
3250 }
3251 
3252 #undef __FUNCT__
3253 #define __FUNCT__ "MatSolveTransposeAdd"
3254 /*@
3255    MatSolveTransposeAdd - Computes x = y + inv(Transpose(A)) b, given a
3256                       factored matrix.
3257 
3258    Neighbor-wise Collective on Mat and Vec
3259 
3260    Input Parameters:
3261 +  mat - the factored matrix
3262 .  b - the right-hand-side vector
3263 -  y - the vector to be added to
3264 
3265    Output Parameter:
3266 .  x - the result vector
3267 
3268    Notes:
3269    The vectors b and x cannot be the same.  I.e., one cannot
3270    call MatSolveTransposeAdd(A,x,y,x).
3271 
3272    Most users should employ the simplified KSP interface for linear solvers
3273    instead of working directly with matrix algebra routines such as this.
3274    See, e.g., KSPCreate().
3275 
3276    Level: developer
3277 
3278    Concepts: matrices^triangular solves
3279 
3280 .seealso: MatSolve(), MatSolveAdd(), MatSolveTranspose()
3281 @*/
3282 PetscErrorCode PETSCMAT_DLLEXPORT MatSolveTransposeAdd(Mat mat,Vec b,Vec y,Vec x)
3283 {
3284   PetscScalar    one = 1.0;
3285   PetscErrorCode ierr;
3286   Vec            tmp;
3287 
3288   PetscFunctionBegin;
3289   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3290   PetscValidType(mat,1);
3291   PetscValidHeaderSpecific(y,VEC_CLASSID,2);
3292   PetscValidHeaderSpecific(b,VEC_CLASSID,3);
3293   PetscValidHeaderSpecific(x,VEC_CLASSID,4);
3294   PetscCheckSameComm(mat,1,b,2);
3295   PetscCheckSameComm(mat,1,y,3);
3296   PetscCheckSameComm(mat,1,x,4);
3297   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
3298   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
3299   if (mat->rmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->rmap->N,x->map->N);
3300   if (mat->cmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->cmap->N,b->map->N);
3301   if (mat->cmap->N != y->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec y: global dim %D %D",mat->cmap->N,y->map->N);
3302   if (x->map->n != y->map->n)   SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Vec x,Vec y: local dim %D %D",x->map->n,y->map->n);
3303   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3304 
3305   ierr = PetscLogEventBegin(MAT_SolveTransposeAdd,mat,b,x,y);CHKERRQ(ierr);
3306   if (mat->ops->solvetransposeadd) {
3307     ierr = (*mat->ops->solvetransposeadd)(mat,b,y,x);CHKERRQ(ierr);
3308   } else {
3309     /* do the solve then the add manually */
3310     if (x != y) {
3311       ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr);
3312       ierr = VecAXPY(x,one,y);CHKERRQ(ierr);
3313     } else {
3314       ierr = VecDuplicate(x,&tmp);CHKERRQ(ierr);
3315       ierr = PetscLogObjectParent(mat,tmp);CHKERRQ(ierr);
3316       ierr = VecCopy(x,tmp);CHKERRQ(ierr);
3317       ierr = MatSolveTranspose(mat,b,x);CHKERRQ(ierr);
3318       ierr = VecAXPY(x,one,tmp);CHKERRQ(ierr);
3319       ierr = VecDestroy(tmp);CHKERRQ(ierr);
3320     }
3321   }
3322   ierr = PetscLogEventEnd(MAT_SolveTransposeAdd,mat,b,x,y);CHKERRQ(ierr);
3323   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3324   PetscFunctionReturn(0);
3325 }
3326 /* ----------------------------------------------------------------*/
3327 
3328 #undef __FUNCT__
3329 #define __FUNCT__ "MatSOR"
3330 /*@
3331    MatSOR - Computes relaxation (SOR, Gauss-Seidel) sweeps.
3332 
3333    Neighbor-wise Collective on Mat and Vec
3334 
3335    Input Parameters:
3336 +  mat - the matrix
3337 .  b - the right hand side
3338 .  omega - the relaxation factor
3339 .  flag - flag indicating the type of SOR (see below)
3340 .  shift -  diagonal shift
3341 .  its - the number of iterations
3342 -  lits - the number of local iterations
3343 
3344    Output Parameters:
3345 .  x - the solution (can contain an initial guess, use option SOR_ZERO_INITIAL_GUESS to indicate no guess)
3346 
3347    SOR Flags:
3348 .     SOR_FORWARD_SWEEP - forward SOR
3349 .     SOR_BACKWARD_SWEEP - backward SOR
3350 .     SOR_SYMMETRIC_SWEEP - SSOR (symmetric SOR)
3351 .     SOR_LOCAL_FORWARD_SWEEP - local forward SOR
3352 .     SOR_LOCAL_BACKWARD_SWEEP - local forward SOR
3353 .     SOR_LOCAL_SYMMETRIC_SWEEP - local SSOR
3354 .     SOR_APPLY_UPPER, SOR_APPLY_LOWER - applies
3355          upper/lower triangular part of matrix to
3356          vector (with omega)
3357 .     SOR_ZERO_INITIAL_GUESS - zero initial guess
3358 
3359    Notes:
3360    SOR_LOCAL_FORWARD_SWEEP, SOR_LOCAL_BACKWARD_SWEEP, and
3361    SOR_LOCAL_SYMMETRIC_SWEEP perform separate independent smoothings
3362    on each processor.
3363 
3364    Application programmers will not generally use MatSOR() directly,
3365    but instead will employ the KSP/PC interface.
3366 
3367    Notes: for BAIJ, SBAIJ, and AIJ matrices with Inodes this does a block SOR smoothing, otherwise it does a pointwise smoothing
3368 
3369    Notes for Advanced Users:
3370    The flags are implemented as bitwise inclusive or operations.
3371    For example, use (SOR_ZERO_INITIAL_GUESS | SOR_SYMMETRIC_SWEEP)
3372    to specify a zero initial guess for SSOR.
3373 
3374    Most users should employ the simplified KSP interface for linear solvers
3375    instead of working directly with matrix algebra routines such as this.
3376    See, e.g., KSPCreate().
3377 
3378    Vectors x and b CANNOT be the same
3379 
3380    Level: developer
3381 
3382    Concepts: matrices^relaxation
3383    Concepts: matrices^SOR
3384    Concepts: matrices^Gauss-Seidel
3385 
3386 @*/
3387 PetscErrorCode PETSCMAT_DLLEXPORT MatSOR(Mat mat,Vec b,PetscReal omega,MatSORType flag,PetscReal shift,PetscInt its,PetscInt lits,Vec x)
3388 {
3389   PetscErrorCode ierr;
3390 
3391   PetscFunctionBegin;
3392   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3393   PetscValidType(mat,1);
3394   PetscValidHeaderSpecific(b,VEC_CLASSID,2);
3395   PetscValidHeaderSpecific(x,VEC_CLASSID,8);
3396   PetscCheckSameComm(mat,1,b,2);
3397   PetscCheckSameComm(mat,1,x,8);
3398   if (!mat->ops->sor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3399   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3400   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3401   if (mat->cmap->N != x->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec x: global dim %D %D",mat->cmap->N,x->map->N);
3402   if (mat->rmap->N != b->map->N) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: global dim %D %D",mat->rmap->N,b->map->N);
3403   if (mat->rmap->n != b->map->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Mat mat,Vec b: local dim %D %D",mat->rmap->n,b->map->n);
3404   if (its <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D positive",its);
3405   if (lits <= 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires local its %D positive",lits);
3406   if (b == x) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_IDN,"b and x vector cannot be the same");
3407 
3408   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3409   ierr = PetscLogEventBegin(MAT_SOR,mat,b,x,0);CHKERRQ(ierr);
3410   ierr =(*mat->ops->sor)(mat,b,omega,flag,shift,its,lits,x);CHKERRQ(ierr);
3411   ierr = PetscLogEventEnd(MAT_SOR,mat,b,x,0);CHKERRQ(ierr);
3412   ierr = PetscObjectStateIncrease((PetscObject)x);CHKERRQ(ierr);
3413   PetscFunctionReturn(0);
3414 }
3415 
3416 #undef __FUNCT__
3417 #define __FUNCT__ "MatCopy_Basic"
3418 /*
3419       Default matrix copy routine.
3420 */
3421 PetscErrorCode MatCopy_Basic(Mat A,Mat B,MatStructure str)
3422 {
3423   PetscErrorCode    ierr;
3424   PetscInt          i,rstart = 0,rend = 0,nz;
3425   const PetscInt    *cwork;
3426   const PetscScalar *vwork;
3427 
3428   PetscFunctionBegin;
3429   if (B->assembled) {
3430     ierr = MatZeroEntries(B);CHKERRQ(ierr);
3431   }
3432   ierr = MatGetOwnershipRange(A,&rstart,&rend);CHKERRQ(ierr);
3433   for (i=rstart; i<rend; i++) {
3434     ierr = MatGetRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr);
3435     ierr = MatSetValues(B,1,&i,nz,cwork,vwork,INSERT_VALUES);CHKERRQ(ierr);
3436     ierr = MatRestoreRow(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr);
3437   }
3438   ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
3439   ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
3440   ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr);
3441   PetscFunctionReturn(0);
3442 }
3443 
3444 #undef __FUNCT__
3445 #define __FUNCT__ "MatCopy"
3446 /*@
3447    MatCopy - Copys a matrix to another matrix.
3448 
3449    Collective on Mat
3450 
3451    Input Parameters:
3452 +  A - the matrix
3453 -  str - SAME_NONZERO_PATTERN or DIFFERENT_NONZERO_PATTERN
3454 
3455    Output Parameter:
3456 .  B - where the copy is put
3457 
3458    Notes:
3459    If you use SAME_NONZERO_PATTERN then the two matrices had better have the
3460    same nonzero pattern or the routine will crash.
3461 
3462    MatCopy() copies the matrix entries of a matrix to another existing
3463    matrix (after first zeroing the second matrix).  A related routine is
3464    MatConvert(), which first creates a new matrix and then copies the data.
3465 
3466    Level: intermediate
3467 
3468    Concepts: matrices^copying
3469 
3470 .seealso: MatConvert(), MatDuplicate()
3471 
3472 @*/
3473 PetscErrorCode PETSCMAT_DLLEXPORT MatCopy(Mat A,Mat B,MatStructure str)
3474 {
3475   PetscErrorCode ierr;
3476   PetscInt       i;
3477 
3478   PetscFunctionBegin;
3479   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
3480   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
3481   PetscValidType(A,1);
3482   PetscValidType(B,2);
3483   PetscCheckSameComm(A,1,B,2);
3484   ierr = MatPreallocated(B);CHKERRQ(ierr);
3485   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3486   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3487   if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim (%D,%D) (%D,%D)",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N);
3488   ierr = MatPreallocated(A);CHKERRQ(ierr);
3489 
3490   ierr = PetscLogEventBegin(MAT_Copy,A,B,0,0);CHKERRQ(ierr);
3491   if (A->ops->copy) {
3492     ierr = (*A->ops->copy)(A,B,str);CHKERRQ(ierr);
3493   } else { /* generic conversion */
3494     ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr);
3495   }
3496   if (A->mapping) {
3497     if (B->mapping) {ierr = ISLocalToGlobalMappingDestroy(B->mapping);CHKERRQ(ierr);B->mapping = 0;}
3498     ierr = MatSetLocalToGlobalMapping(B,A->mapping);CHKERRQ(ierr);
3499   }
3500   if (A->bmapping) {
3501     if (B->bmapping) {ierr = ISLocalToGlobalMappingDestroy(B->bmapping);CHKERRQ(ierr);B->bmapping = 0;}
3502     ierr = MatSetLocalToGlobalMappingBlock(B,A->mapping);CHKERRQ(ierr);
3503   }
3504 
3505   B->stencil.dim = A->stencil.dim;
3506   B->stencil.noc = A->stencil.noc;
3507   for (i=0; i<=A->stencil.dim; i++) {
3508     B->stencil.dims[i]   = A->stencil.dims[i];
3509     B->stencil.starts[i] = A->stencil.starts[i];
3510   }
3511 
3512   ierr = PetscLogEventEnd(MAT_Copy,A,B,0,0);CHKERRQ(ierr);
3513   ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr);
3514   PetscFunctionReturn(0);
3515 }
3516 
3517 #undef __FUNCT__
3518 #define __FUNCT__ "MatConvert"
3519 /*@C
3520    MatConvert - Converts a matrix to another matrix, either of the same
3521    or different type.
3522 
3523    Collective on Mat
3524 
3525    Input Parameters:
3526 +  mat - the matrix
3527 .  newtype - new matrix type.  Use MATSAME to create a new matrix of the
3528    same type as the original matrix.
3529 -  reuse - denotes if the destination matrix is to be created or reused.  Currently
3530    MAT_REUSE_MATRIX is only supported for inplace conversion, otherwise use
3531    MAT_INITIAL_MATRIX.
3532 
3533    Output Parameter:
3534 .  M - pointer to place new matrix
3535 
3536    Notes:
3537    MatConvert() first creates a new matrix and then copies the data from
3538    the first matrix.  A related routine is MatCopy(), which copies the matrix
3539    entries of one matrix to another already existing matrix context.
3540 
3541    Cannot be used to convert a sequential matrix to parallel or parallel to sequential,
3542    the MPI communicator of the generated matrix is always the same as the communicator
3543    of the input matrix.
3544 
3545    Level: intermediate
3546 
3547    Concepts: matrices^converting between storage formats
3548 
3549 .seealso: MatCopy(), MatDuplicate()
3550 @*/
3551 PetscErrorCode PETSCMAT_DLLEXPORT MatConvert(Mat mat, const MatType newtype,MatReuse reuse,Mat *M)
3552 {
3553   PetscErrorCode         ierr;
3554   PetscTruth             sametype,issame,flg;
3555   char                   convname[256],mtype[256];
3556   Mat                    B;
3557 
3558   PetscFunctionBegin;
3559   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3560   PetscValidType(mat,1);
3561   PetscValidPointer(M,3);
3562   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3563   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3564   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3565 
3566   ierr = PetscOptionsGetString(((PetscObject)mat)->prefix,"-matconvert_type",mtype,256,&flg);CHKERRQ(ierr);
3567   if (flg) {
3568     newtype = mtype;
3569   }
3570   ierr = PetscTypeCompare((PetscObject)mat,newtype,&sametype);CHKERRQ(ierr);
3571   ierr = PetscStrcmp(newtype,"same",&issame);CHKERRQ(ierr);
3572   if ((reuse == MAT_REUSE_MATRIX) && (mat != *M)) {
3573     SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"MAT_REUSE_MATRIX only supported for in-place conversion currently");
3574   }
3575 
3576   if ((reuse == MAT_REUSE_MATRIX) && (issame || sametype)) PetscFunctionReturn(0);
3577 
3578   if ((sametype || issame) && (reuse==MAT_INITIAL_MATRIX) && mat->ops->duplicate) {
3579     ierr = (*mat->ops->duplicate)(mat,MAT_COPY_VALUES,M);CHKERRQ(ierr);
3580   } else {
3581     PetscErrorCode (*conv)(Mat, const MatType,MatReuse,Mat*)=PETSC_NULL;
3582     const char     *prefix[3] = {"seq","mpi",""};
3583     PetscInt       i;
3584     /*
3585        Order of precedence:
3586        1) See if a specialized converter is known to the current matrix.
3587        2) See if a specialized converter is known to the desired matrix class.
3588        3) See if a good general converter is registered for the desired class
3589           (as of 6/27/03 only MATMPIADJ falls into this category).
3590        4) See if a good general converter is known for the current matrix.
3591        5) Use a really basic converter.
3592     */
3593 
3594     /* 1) See if a specialized converter is known to the current matrix and the desired class */
3595     for (i=0; i<3; i++) {
3596       ierr = PetscStrcpy(convname,"MatConvert_");CHKERRQ(ierr);
3597       ierr = PetscStrcat(convname,((PetscObject)mat)->type_name);CHKERRQ(ierr);
3598       ierr = PetscStrcat(convname,"_");CHKERRQ(ierr);
3599       ierr = PetscStrcat(convname,prefix[i]);CHKERRQ(ierr);
3600       ierr = PetscStrcat(convname,newtype);CHKERRQ(ierr);
3601       ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr);
3602       ierr = PetscObjectQueryFunction((PetscObject)mat,convname,(void (**)(void))&conv);CHKERRQ(ierr);
3603       if (conv) goto foundconv;
3604     }
3605 
3606     /* 2)  See if a specialized converter is known to the desired matrix class. */
3607     ierr = MatCreate(((PetscObject)mat)->comm,&B);CHKERRQ(ierr);
3608     ierr = MatSetSizes(B,mat->rmap->n,mat->cmap->n,mat->rmap->N,mat->cmap->N);CHKERRQ(ierr);
3609     ierr = MatSetType(B,newtype);CHKERRQ(ierr);
3610     for (i=0; i<3; i++) {
3611       ierr = PetscStrcpy(convname,"MatConvert_");CHKERRQ(ierr);
3612       ierr = PetscStrcat(convname,((PetscObject)mat)->type_name);CHKERRQ(ierr);
3613       ierr = PetscStrcat(convname,"_");CHKERRQ(ierr);
3614       ierr = PetscStrcat(convname,prefix[i]);CHKERRQ(ierr);
3615       ierr = PetscStrcat(convname,newtype);CHKERRQ(ierr);
3616       ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr);
3617       ierr = PetscObjectQueryFunction((PetscObject)B,convname,(void (**)(void))&conv);CHKERRQ(ierr);
3618       if (conv) {
3619         ierr = MatDestroy(B);CHKERRQ(ierr);
3620         goto foundconv;
3621       }
3622     }
3623 
3624     /* 3) See if a good general converter is registered for the desired class */
3625     conv = B->ops->convertfrom;
3626     ierr = MatDestroy(B);CHKERRQ(ierr);
3627     if (conv) goto foundconv;
3628 
3629     /* 4) See if a good general converter is known for the current matrix */
3630     if (mat->ops->convert) {
3631       conv = mat->ops->convert;
3632     }
3633     if (conv) goto foundconv;
3634 
3635     /* 5) Use a really basic converter. */
3636     conv = MatConvert_Basic;
3637 
3638     foundconv:
3639     ierr = PetscLogEventBegin(MAT_Convert,mat,0,0,0);CHKERRQ(ierr);
3640     ierr = (*conv)(mat,newtype,reuse,M);CHKERRQ(ierr);
3641     ierr = PetscLogEventEnd(MAT_Convert,mat,0,0,0);CHKERRQ(ierr);
3642   }
3643   ierr = PetscObjectStateIncrease((PetscObject)*M);CHKERRQ(ierr);
3644   PetscFunctionReturn(0);
3645 }
3646 
3647 #undef __FUNCT__
3648 #define __FUNCT__ "MatFactorGetSolverPackage"
3649 /*@C
3650    MatFactorGetSolverPackage - Returns name of the package providing the factorization routines
3651 
3652    Not Collective
3653 
3654    Input Parameter:
3655 .  mat - the matrix, must be a factored matrix
3656 
3657    Output Parameter:
3658 .   type - the string name of the package (do not free this string)
3659 
3660    Notes:
3661       In Fortran you pass in a empty string and the package name will be copied into it.
3662     (Make sure the string is long enough)
3663 
3664    Level: intermediate
3665 
3666 .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable(), MatGetFactor()
3667 @*/
3668 PetscErrorCode PETSCMAT_DLLEXPORT MatFactorGetSolverPackage(Mat mat, const MatSolverPackage *type)
3669 {
3670   PetscErrorCode         ierr;
3671   PetscErrorCode         (*conv)(Mat,const MatSolverPackage*);
3672 
3673   PetscFunctionBegin;
3674   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3675   PetscValidType(mat,1);
3676   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Only for factored matrix");
3677   ierr = PetscObjectQueryFunction((PetscObject)mat,"MatFactorGetSolverPackage_C",(void (**)(void))&conv);CHKERRQ(ierr);
3678   if (!conv) {
3679     *type = MATSOLVERPETSC;
3680   } else {
3681     ierr = (*conv)(mat,type);CHKERRQ(ierr);
3682   }
3683   PetscFunctionReturn(0);
3684 }
3685 
3686 #undef __FUNCT__
3687 #define __FUNCT__ "MatGetFactor"
3688 /*@C
3689    MatGetFactor - Returns a matrix suitable to calls to MatXXFactorSymbolic()
3690 
3691    Collective on Mat
3692 
3693    Input Parameters:
3694 +  mat - the matrix
3695 .  type - name of solver type, for example, spooles, superlu, plapack, petsc (to use PETSc's default)
3696 -  ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU,
3697 
3698    Output Parameters:
3699 .  f - the factor matrix used with MatXXFactorSymbolic() calls
3700 
3701    Notes:
3702       Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
3703      such as pastix, superlu, mumps, spooles etc.
3704 
3705       PETSc must have been ./configure to use the external solver, using the option --download-package
3706 
3707    Level: intermediate
3708 
3709 .seealso: MatCopy(), MatDuplicate(), MatGetFactorAvailable()
3710 @*/
3711 PetscErrorCode PETSCMAT_DLLEXPORT MatGetFactor(Mat mat, const MatSolverPackage type,MatFactorType ftype,Mat *f)
3712 {
3713   PetscErrorCode  ierr,(*conv)(Mat,MatFactorType,Mat*);
3714   char            convname[256];
3715 
3716   PetscFunctionBegin;
3717   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3718   PetscValidType(mat,1);
3719 
3720   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3721   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3722 
3723   ierr = PetscStrcpy(convname,"MatGetFactor_");CHKERRQ(ierr);
3724   ierr = PetscStrcat(convname,type);CHKERRQ(ierr);
3725   ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr);
3726   ierr = PetscObjectQueryFunction((PetscObject)mat,convname,(void (**)(void))&conv);CHKERRQ(ierr);
3727   if (!conv) {
3728     PetscTruth flag;
3729     MPI_Comm   comm;
3730 
3731     ierr = PetscObjectGetComm((PetscObject)mat,&comm);CHKERRQ(ierr);
3732     ierr = PetscStrcasecmp(MATSOLVERPETSC,type,&flag);CHKERRQ(ierr);
3733     if (flag) {
3734       SETERRQ2(comm,PETSC_ERR_SUP,"Matrix format %s does not have a built-in PETSc %s",((PetscObject)mat)->type_name,MatFactorTypes[ftype]);
3735     } else {
3736       SETERRQ4(comm,PETSC_ERR_SUP,"Matrix format %s does not have a solver package %s for %s. Perhaps you must ./configure with --download-%s",((PetscObject)mat)->type_name,type,MatFactorTypes[ftype],type);
3737     }
3738   }
3739   ierr = (*conv)(mat,ftype,f);CHKERRQ(ierr);
3740   PetscFunctionReturn(0);
3741 }
3742 
3743 #undef __FUNCT__
3744 #define __FUNCT__ "MatGetFactorAvailable"
3745 /*@C
3746    MatGetFactorAvailable - Returns a a flag if matrix supports particular package and factor type
3747 
3748    Not Collective
3749 
3750    Input Parameters:
3751 +  mat - the matrix
3752 .  type - name of solver type, for example, spooles, superlu, plapack, petsc (to use PETSc's default)
3753 -  ftype - factor type, MAT_FACTOR_LU, MAT_FACTOR_CHOLESKY, MAT_FACTOR_ICC, MAT_FACTOR_ILU,
3754 
3755    Output Parameter:
3756 .    flg - PETSC_TRUE if the factorization is available
3757 
3758    Notes:
3759       Some PETSc matrix formats have alternative solvers available that are contained in alternative packages
3760      such as pastix, superlu, mumps, spooles etc.
3761 
3762       PETSc must have been ./configure to use the external solver, using the option --download-package
3763 
3764    Level: intermediate
3765 
3766 .seealso: MatCopy(), MatDuplicate(), MatGetFactor()
3767 @*/
3768 PetscErrorCode PETSCMAT_DLLEXPORT MatGetFactorAvailable(Mat mat, const MatSolverPackage type,MatFactorType ftype,PetscTruth *flg)
3769 {
3770   PetscErrorCode         ierr;
3771   char                   convname[256];
3772   PetscErrorCode         (*conv)(Mat,MatFactorType,PetscTruth*);
3773 
3774   PetscFunctionBegin;
3775   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3776   PetscValidType(mat,1);
3777 
3778   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3779   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3780 
3781   ierr = PetscStrcpy(convname,"MatGetFactorAvailable_");CHKERRQ(ierr);
3782   ierr = PetscStrcat(convname,type);CHKERRQ(ierr);
3783   ierr = PetscStrcat(convname,"_C");CHKERRQ(ierr);
3784   ierr = PetscObjectQueryFunction((PetscObject)mat,convname,(void (**)(void))&conv);CHKERRQ(ierr);
3785   if (!conv) {
3786     *flg = PETSC_FALSE;
3787   } else {
3788     ierr = (*conv)(mat,ftype,flg);CHKERRQ(ierr);
3789   }
3790   PetscFunctionReturn(0);
3791 }
3792 
3793 
3794 #undef __FUNCT__
3795 #define __FUNCT__ "MatDuplicate"
3796 /*@
3797    MatDuplicate - Duplicates a matrix including the non-zero structure.
3798 
3799    Collective on Mat
3800 
3801    Input Parameters:
3802 +  mat - the matrix
3803 -  op - either MAT_DO_NOT_COPY_VALUES or MAT_COPY_VALUES, cause it to copy the numerical values in the matrix
3804         MAT_SHARE_NONZERO_PATTERN to share the nonzero patterns with the previous matrix and not copy them.
3805 
3806    Output Parameter:
3807 .  M - pointer to place new matrix
3808 
3809    Level: intermediate
3810 
3811    Concepts: matrices^duplicating
3812 
3813     Notes: You cannot change the nonzero pattern for the parent or child matrix if you use MAT_SHARE_NONZERO_PATTERN.
3814 
3815 .seealso: MatCopy(), MatConvert()
3816 @*/
3817 PetscErrorCode PETSCMAT_DLLEXPORT MatDuplicate(Mat mat,MatDuplicateOption op,Mat *M)
3818 {
3819   PetscErrorCode ierr;
3820   Mat            B;
3821   PetscInt       i;
3822 
3823   PetscFunctionBegin;
3824   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3825   PetscValidType(mat,1);
3826   PetscValidPointer(M,3);
3827   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3828   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3829   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3830 
3831   *M  = 0;
3832   if (!mat->ops->duplicate) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Not written for this matrix type");
3833   ierr = PetscLogEventBegin(MAT_Convert,mat,0,0,0);CHKERRQ(ierr);
3834   ierr = (*mat->ops->duplicate)(mat,op,M);CHKERRQ(ierr);
3835   B = *M;
3836   if (mat->mapping) {
3837     ierr = MatSetLocalToGlobalMapping(B,mat->mapping);CHKERRQ(ierr);
3838   }
3839   if (mat->bmapping) {
3840     ierr = MatSetLocalToGlobalMappingBlock(B,mat->bmapping);CHKERRQ(ierr);
3841   }
3842   ierr = PetscLayoutCopy(mat->rmap,&B->rmap);CHKERRQ(ierr);
3843   ierr = PetscLayoutCopy(mat->cmap,&B->cmap);CHKERRQ(ierr);
3844 
3845   B->stencil.dim = mat->stencil.dim;
3846   B->stencil.noc = mat->stencil.noc;
3847   for (i=0; i<=mat->stencil.dim; i++) {
3848     B->stencil.dims[i]   = mat->stencil.dims[i];
3849     B->stencil.starts[i] = mat->stencil.starts[i];
3850   }
3851 
3852   ierr = PetscLogEventEnd(MAT_Convert,mat,0,0,0);CHKERRQ(ierr);
3853   ierr = PetscObjectStateIncrease((PetscObject)B);CHKERRQ(ierr);
3854   PetscFunctionReturn(0);
3855 }
3856 
3857 #undef __FUNCT__
3858 #define __FUNCT__ "MatGetDiagonal"
3859 /*@
3860    MatGetDiagonal - Gets the diagonal of a matrix.
3861 
3862    Logically Collective on Mat and Vec
3863 
3864    Input Parameters:
3865 +  mat - the matrix
3866 -  v - the vector for storing the diagonal
3867 
3868    Output Parameter:
3869 .  v - the diagonal of the matrix
3870 
3871    Level: intermediate
3872 
3873    Concepts: matrices^accessing diagonals
3874 
3875 .seealso: MatGetRow(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs()
3876 @*/
3877 PetscErrorCode PETSCMAT_DLLEXPORT MatGetDiagonal(Mat mat,Vec v)
3878 {
3879   PetscErrorCode ierr;
3880 
3881   PetscFunctionBegin;
3882   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3883   PetscValidType(mat,1);
3884   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
3885   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3886   if (!mat->ops->getdiagonal) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3887   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3888 
3889   ierr = (*mat->ops->getdiagonal)(mat,v);CHKERRQ(ierr);
3890   ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr);
3891   PetscFunctionReturn(0);
3892 }
3893 
3894 #undef __FUNCT__
3895 #define __FUNCT__ "MatGetRowMin"
3896 /*@
3897    MatGetRowMin - Gets the minimum value (of the real part) of each
3898         row of the matrix
3899 
3900    Logically Collective on Mat and Vec
3901 
3902    Input Parameters:
3903 .  mat - the matrix
3904 
3905    Output Parameter:
3906 +  v - the vector for storing the maximums
3907 -  idx - the indices of the column found for each row (optional)
3908 
3909    Level: intermediate
3910 
3911    Notes: The result of this call are the same as if one converted the matrix to dense format
3912       and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
3913 
3914     This code is only implemented for a couple of matrix formats.
3915 
3916    Concepts: matrices^getting row maximums
3917 
3918 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs(),
3919           MatGetRowMax()
3920 @*/
3921 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowMin(Mat mat,Vec v,PetscInt idx[])
3922 {
3923   PetscErrorCode ierr;
3924 
3925   PetscFunctionBegin;
3926   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3927   PetscValidType(mat,1);
3928   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
3929   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3930   if (!mat->ops->getrowmax) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3931   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3932 
3933   ierr = (*mat->ops->getrowmin)(mat,v,idx);CHKERRQ(ierr);
3934   ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr);
3935   PetscFunctionReturn(0);
3936 }
3937 
3938 #undef __FUNCT__
3939 #define __FUNCT__ "MatGetRowMinAbs"
3940 /*@
3941    MatGetRowMinAbs - Gets the minimum value (in absolute value) of each
3942         row of the matrix
3943 
3944    Logically Collective on Mat and Vec
3945 
3946    Input Parameters:
3947 .  mat - the matrix
3948 
3949    Output Parameter:
3950 +  v - the vector for storing the minimums
3951 -  idx - the indices of the column found for each row (optional)
3952 
3953    Level: intermediate
3954 
3955    Notes: if a row is completely empty or has only 0.0 values then the idx[] value for that
3956     row is 0 (the first column).
3957 
3958     This code is only implemented for a couple of matrix formats.
3959 
3960    Concepts: matrices^getting row maximums
3961 
3962 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMaxAbs(), MatGetRowMin()
3963 @*/
3964 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowMinAbs(Mat mat,Vec v,PetscInt idx[])
3965 {
3966   PetscErrorCode ierr;
3967 
3968   PetscFunctionBegin;
3969   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
3970   PetscValidType(mat,1);
3971   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
3972   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3973   if (!mat->ops->getrowminabs) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
3974   ierr = MatPreallocated(mat);CHKERRQ(ierr);
3975   if (idx) {ierr = PetscMemzero(idx,mat->rmap->n*sizeof(PetscInt));CHKERRQ(ierr);}
3976 
3977   ierr = (*mat->ops->getrowminabs)(mat,v,idx);CHKERRQ(ierr);
3978   ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr);
3979   PetscFunctionReturn(0);
3980 }
3981 
3982 #undef __FUNCT__
3983 #define __FUNCT__ "MatGetRowMax"
3984 /*@
3985    MatGetRowMax - Gets the maximum value (of the real part) of each
3986         row of the matrix
3987 
3988    Logically Collective on Mat and Vec
3989 
3990    Input Parameters:
3991 .  mat - the matrix
3992 
3993    Output Parameter:
3994 +  v - the vector for storing the maximums
3995 -  idx - the indices of the column found for each row (optional)
3996 
3997    Level: intermediate
3998 
3999    Notes: The result of this call are the same as if one converted the matrix to dense format
4000       and found the minimum value in each row (i.e. the implicit zeros are counted as zeros).
4001 
4002     This code is only implemented for a couple of matrix formats.
4003 
4004    Concepts: matrices^getting row maximums
4005 
4006 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMaxAbs(), MatGetRowMin()
4007 @*/
4008 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowMax(Mat mat,Vec v,PetscInt idx[])
4009 {
4010   PetscErrorCode ierr;
4011 
4012   PetscFunctionBegin;
4013   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4014   PetscValidType(mat,1);
4015   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
4016   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4017   if (!mat->ops->getrowmax) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4018   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4019 
4020   ierr = (*mat->ops->getrowmax)(mat,v,idx);CHKERRQ(ierr);
4021   ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr);
4022   PetscFunctionReturn(0);
4023 }
4024 
4025 #undef __FUNCT__
4026 #define __FUNCT__ "MatGetRowMaxAbs"
4027 /*@
4028    MatGetRowMaxAbs - Gets the maximum value (in absolute value) of each
4029         row of the matrix
4030 
4031    Logically Collective on Mat and Vec
4032 
4033    Input Parameters:
4034 .  mat - the matrix
4035 
4036    Output Parameter:
4037 +  v - the vector for storing the maximums
4038 -  idx - the indices of the column found for each row (optional)
4039 
4040    Level: intermediate
4041 
4042    Notes: if a row is completely empty or has only 0.0 values then the idx[] value for that
4043     row is 0 (the first column).
4044 
4045     This code is only implemented for a couple of matrix formats.
4046 
4047    Concepts: matrices^getting row maximums
4048 
4049 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMin()
4050 @*/
4051 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowMaxAbs(Mat mat,Vec v,PetscInt idx[])
4052 {
4053   PetscErrorCode ierr;
4054 
4055   PetscFunctionBegin;
4056   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4057   PetscValidType(mat,1);
4058   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
4059   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4060   if (!mat->ops->getrowmaxabs) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4061   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4062   if (idx) {ierr = PetscMemzero(idx,mat->rmap->n*sizeof(PetscInt));CHKERRQ(ierr);}
4063 
4064   ierr = (*mat->ops->getrowmaxabs)(mat,v,idx);CHKERRQ(ierr);
4065   ierr = PetscObjectStateIncrease((PetscObject)v);CHKERRQ(ierr);
4066   PetscFunctionReturn(0);
4067 }
4068 
4069 #undef __FUNCT__
4070 #define __FUNCT__ "MatGetRowSum"
4071 /*@
4072    MatGetRowSum - Gets the sum of each row of the matrix
4073 
4074    Logically Collective on Mat and Vec
4075 
4076    Input Parameters:
4077 .  mat - the matrix
4078 
4079    Output Parameter:
4080 .  v - the vector for storing the sum of rows
4081 
4082    Level: intermediate
4083 
4084    Notes: This code is slow since it is not currently specialized for different formats
4085 
4086    Concepts: matrices^getting row sums
4087 
4088 .seealso: MatGetDiagonal(), MatGetSubMatrices(), MatGetSubmatrix(), MatGetRowMax(), MatGetRowMin()
4089 @*/
4090 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowSum(Mat mat, Vec v)
4091 {
4092   PetscInt       start = 0, end = 0, row;
4093   PetscScalar   *array;
4094   PetscErrorCode ierr;
4095 
4096   PetscFunctionBegin;
4097   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4098   PetscValidType(mat,1);
4099   PetscValidHeaderSpecific(v,VEC_CLASSID,2);
4100   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4101   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4102   ierr = MatGetOwnershipRange(mat, &start, &end);CHKERRQ(ierr);
4103   ierr = VecGetArray(v, &array);CHKERRQ(ierr);
4104   for(row = start; row < end; ++row) {
4105     PetscInt           ncols, col;
4106     const PetscInt    *cols;
4107     const PetscScalar *vals;
4108 
4109     array[row - start] = 0.0;
4110     ierr = MatGetRow(mat, row, &ncols, &cols, &vals);CHKERRQ(ierr);
4111     for(col = 0; col < ncols; col++) {
4112       array[row - start] += vals[col];
4113     }
4114     ierr = MatRestoreRow(mat, row, &ncols, &cols, &vals);CHKERRQ(ierr);
4115   }
4116   ierr = VecRestoreArray(v, &array);CHKERRQ(ierr);
4117   ierr = PetscObjectStateIncrease((PetscObject) v);CHKERRQ(ierr);
4118   PetscFunctionReturn(0);
4119 }
4120 
4121 #undef __FUNCT__
4122 #define __FUNCT__ "MatTranspose"
4123 /*@
4124    MatTranspose - Computes an in-place or out-of-place transpose of a matrix.
4125 
4126    Collective on Mat
4127 
4128    Input Parameter:
4129 +  mat - the matrix to transpose
4130 -  reuse - store the transpose matrix in the provided B
4131 
4132    Output Parameters:
4133 .  B - the transpose
4134 
4135    Notes:
4136      If you  pass in &mat for B the transpose will be done in place
4137 
4138    Level: intermediate
4139 
4140    Concepts: matrices^transposing
4141 
4142 .seealso: MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse
4143 @*/
4144 PetscErrorCode PETSCMAT_DLLEXPORT MatTranspose(Mat mat,MatReuse reuse,Mat *B)
4145 {
4146   PetscErrorCode ierr;
4147 
4148   PetscFunctionBegin;
4149   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4150   PetscValidType(mat,1);
4151   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4152   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4153   if (!mat->ops->transpose) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4154   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4155 
4156   ierr = PetscLogEventBegin(MAT_Transpose,mat,0,0,0);CHKERRQ(ierr);
4157   ierr = (*mat->ops->transpose)(mat,reuse,B);CHKERRQ(ierr);
4158   ierr = PetscLogEventEnd(MAT_Transpose,mat,0,0,0);CHKERRQ(ierr);
4159   if (B) {ierr = PetscObjectStateIncrease((PetscObject)*B);CHKERRQ(ierr);}
4160   PetscFunctionReturn(0);
4161 }
4162 
4163 #undef __FUNCT__
4164 #define __FUNCT__ "MatIsTranspose"
4165 /*@
4166    MatIsTranspose - Test whether a matrix is another one's transpose,
4167         or its own, in which case it tests symmetry.
4168 
4169    Collective on Mat
4170 
4171    Input Parameter:
4172 +  A - the matrix to test
4173 -  B - the matrix to test against, this can equal the first parameter
4174 
4175    Output Parameters:
4176 .  flg - the result
4177 
4178    Notes:
4179    Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm
4180    has a running time of the order of the number of nonzeros; the parallel
4181    test involves parallel copies of the block-offdiagonal parts of the matrix.
4182 
4183    Level: intermediate
4184 
4185    Concepts: matrices^transposing, matrix^symmetry
4186 
4187 .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian()
4188 @*/
4189 PetscErrorCode PETSCMAT_DLLEXPORT MatIsTranspose(Mat A,Mat B,PetscReal tol,PetscTruth *flg)
4190 {
4191   PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscTruth*),(*g)(Mat,Mat,PetscReal,PetscTruth*);
4192 
4193   PetscFunctionBegin;
4194   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
4195   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
4196   PetscValidPointer(flg,3);
4197   ierr = PetscObjectQueryFunction((PetscObject)A,"MatIsTranspose_C",(void (**)(void))&f);CHKERRQ(ierr);
4198   ierr = PetscObjectQueryFunction((PetscObject)B,"MatIsTranspose_C",(void (**)(void))&g);CHKERRQ(ierr);
4199   if (f && g) {
4200     if (f==g) {
4201       ierr = (*f)(A,B,tol,flg);CHKERRQ(ierr);
4202     } else {
4203       SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for symmetry test");
4204     }
4205   }
4206   PetscFunctionReturn(0);
4207 }
4208 
4209 #undef __FUNCT__
4210 #define __FUNCT__ "MatHermitianTranspose"
4211 /*@
4212    MatHermitianTranspose - Computes an in-place or out-of-place transpose of a matrix in complex conjugate.
4213 
4214    Collective on Mat
4215 
4216    Input Parameter:
4217 +  mat - the matrix to transpose and complex conjugate
4218 -  reuse - store the transpose matrix in the provided B
4219 
4220    Output Parameters:
4221 .  B - the Hermitian
4222 
4223    Notes:
4224      If you  pass in &mat for B the Hermitian will be done in place
4225 
4226    Level: intermediate
4227 
4228    Concepts: matrices^transposing, complex conjugatex
4229 
4230 .seealso: MatTranspose(), MatMultTranspose(), MatMultTransposeAdd(), MatIsTranspose(), MatReuse
4231 @*/
4232 PetscErrorCode PETSCMAT_DLLEXPORT MatHermitianTranspose(Mat mat,MatReuse reuse,Mat *B)
4233 {
4234   PetscErrorCode ierr;
4235 
4236   PetscFunctionBegin;
4237   ierr = MatTranspose(mat,reuse,B);CHKERRQ(ierr);
4238 #if defined(PETSC_USE_COMPLEX)
4239   ierr = MatConjugate(*B);CHKERRQ(ierr);
4240 #endif
4241   PetscFunctionReturn(0);
4242 }
4243 
4244 #undef __FUNCT__
4245 #define __FUNCT__ "MatIsHermitianTranspose"
4246 /*@
4247    MatIsHermitianTranspose - Test whether a matrix is another one's Hermitian transpose,
4248 
4249    Collective on Mat
4250 
4251    Input Parameter:
4252 +  A - the matrix to test
4253 -  B - the matrix to test against, this can equal the first parameter
4254 
4255    Output Parameters:
4256 .  flg - the result
4257 
4258    Notes:
4259    Only available for SeqAIJ/MPIAIJ matrices. The sequential algorithm
4260    has a running time of the order of the number of nonzeros; the parallel
4261    test involves parallel copies of the block-offdiagonal parts of the matrix.
4262 
4263    Level: intermediate
4264 
4265    Concepts: matrices^transposing, matrix^symmetry
4266 
4267 .seealso: MatTranspose(), MatIsSymmetric(), MatIsHermitian(), MatIsTranspose()
4268 @*/
4269 PetscErrorCode PETSCMAT_DLLEXPORT MatIsHermitianTranspose(Mat A,Mat B,PetscReal tol,PetscTruth *flg)
4270 {
4271   PetscErrorCode ierr,(*f)(Mat,Mat,PetscReal,PetscTruth*),(*g)(Mat,Mat,PetscReal,PetscTruth*);
4272 
4273   PetscFunctionBegin;
4274   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
4275   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
4276   PetscValidPointer(flg,3);
4277   ierr = PetscObjectQueryFunction((PetscObject)A,"MatIsHermitianTranspose_C",(void (**)(void))&f);CHKERRQ(ierr);
4278   ierr = PetscObjectQueryFunction((PetscObject)B,"MatIsHermitianTranspose_C",(void (**)(void))&g);CHKERRQ(ierr);
4279   if (f && g) {
4280     if (f==g) {
4281       ierr = (*f)(A,B,tol,flg);CHKERRQ(ierr);
4282     } else {
4283       SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_NOTSAMETYPE,"Matrices do not have the same comparator for Hermitian test");
4284     }
4285   }
4286   PetscFunctionReturn(0);
4287 }
4288 
4289 #undef __FUNCT__
4290 #define __FUNCT__ "MatPermute"
4291 /*@
4292    MatPermute - Creates a new matrix with rows and columns permuted from the
4293    original.
4294 
4295    Collective on Mat
4296 
4297    Input Parameters:
4298 +  mat - the matrix to permute
4299 .  row - row permutation, each processor supplies only the permutation for its rows
4300 -  col - column permutation, each processor needs the entire column permutation, that is
4301          this is the same size as the total number of columns in the matrix. It can often
4302          be obtained with ISAllGather() on the row permutation
4303 
4304    Output Parameters:
4305 .  B - the permuted matrix
4306 
4307    Level: advanced
4308 
4309    Concepts: matrices^permuting
4310 
4311 .seealso: MatGetOrdering(), ISAllGather()
4312 
4313 @*/
4314 PetscErrorCode PETSCMAT_DLLEXPORT MatPermute(Mat mat,IS row,IS col,Mat *B)
4315 {
4316   PetscErrorCode ierr;
4317 
4318   PetscFunctionBegin;
4319   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4320   PetscValidType(mat,1);
4321   PetscValidHeaderSpecific(row,IS_CLASSID,2);
4322   PetscValidHeaderSpecific(col,IS_CLASSID,3);
4323   PetscValidPointer(B,4);
4324   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4325   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4326   if (!mat->ops->permute) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatPermute not available for Mat type %s",((PetscObject)mat)->type_name);
4327   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4328 
4329   ierr = (*mat->ops->permute)(mat,row,col,B);CHKERRQ(ierr);
4330   ierr = PetscObjectStateIncrease((PetscObject)*B);CHKERRQ(ierr);
4331   PetscFunctionReturn(0);
4332 }
4333 
4334 #undef __FUNCT__
4335 #define __FUNCT__ "MatPermuteSparsify"
4336 /*@
4337   MatPermuteSparsify - Creates a new matrix with rows and columns permuted from the
4338   original and sparsified to the prescribed tolerance.
4339 
4340   Collective on Mat
4341 
4342   Input Parameters:
4343 + A    - The matrix to permute
4344 . band - The half-bandwidth of the sparsified matrix, or PETSC_DECIDE
4345 . frac - The half-bandwidth as a fraction of the total size, or 0.0
4346 . tol  - The drop tolerance
4347 . rowp - The row permutation
4348 - colp - The column permutation
4349 
4350   Output Parameter:
4351 . B    - The permuted, sparsified matrix
4352 
4353   Level: advanced
4354 
4355   Note:
4356   The default behavior (band = PETSC_DECIDE and frac = 0.0) is to
4357   restrict the half-bandwidth of the resulting matrix to 5% of the
4358   total matrix size.
4359 
4360 .keywords: matrix, permute, sparsify
4361 
4362 .seealso: MatGetOrdering(), MatPermute()
4363 @*/
4364 PetscErrorCode PETSCMAT_DLLEXPORT MatPermuteSparsify(Mat A, PetscInt band, PetscReal frac, PetscReal tol, IS rowp, IS colp, Mat *B)
4365 {
4366   IS                irowp, icolp;
4367   const PetscInt    *rows, *cols;
4368   PetscInt          M, N, locRowStart = 0, locRowEnd = 0;
4369   PetscInt          nz, newNz;
4370   const PetscInt    *cwork;
4371   PetscInt          *cnew;
4372   const PetscScalar *vwork;
4373   PetscScalar       *vnew;
4374   PetscInt          bw, issize;
4375   PetscInt          row, locRow, newRow, col, newCol;
4376   PetscErrorCode    ierr;
4377 
4378   PetscFunctionBegin;
4379   PetscValidHeaderSpecific(A,    MAT_CLASSID,1);
4380   PetscValidHeaderSpecific(rowp, IS_CLASSID,5);
4381   PetscValidHeaderSpecific(colp, IS_CLASSID,6);
4382   PetscValidPointer(B,7);
4383   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4384   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4385   if (!A->ops->permutesparsify) {
4386     ierr = MatGetSize(A, &M, &N);CHKERRQ(ierr);
4387     ierr = MatGetOwnershipRange(A, &locRowStart, &locRowEnd);CHKERRQ(ierr);
4388     ierr = ISGetSize(rowp, &issize);CHKERRQ(ierr);
4389     if (issize != M) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG, "Wrong size %D for row permutation, should be %D", issize, M);
4390     ierr = ISGetSize(colp, &issize);CHKERRQ(ierr);
4391     if (issize != N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG, "Wrong size %D for column permutation, should be %D", issize, N);
4392     ierr = ISInvertPermutation(rowp, 0, &irowp);CHKERRQ(ierr);
4393     ierr = ISGetIndices(irowp, &rows);CHKERRQ(ierr);
4394     ierr = ISInvertPermutation(colp, 0, &icolp);CHKERRQ(ierr);
4395     ierr = ISGetIndices(icolp, &cols);CHKERRQ(ierr);
4396     ierr = PetscMalloc(N*sizeof(PetscInt),&cnew);CHKERRQ(ierr);
4397     ierr = PetscMalloc(N*sizeof(PetscScalar),&vnew);CHKERRQ(ierr);
4398 
4399     /* Setup bandwidth to include */
4400     if (band == PETSC_DECIDE) {
4401       if (frac <= 0.0)
4402         bw = (PetscInt) (M * 0.05);
4403       else
4404         bw = (PetscInt) (M * frac);
4405     } else {
4406       if (band <= 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG, "Bandwidth must be a positive integer");
4407       bw = band;
4408     }
4409 
4410     /* Put values into new matrix */
4411     ierr = MatDuplicate(A, MAT_DO_NOT_COPY_VALUES, B);CHKERRQ(ierr);
4412     for(row = locRowStart, locRow = 0; row < locRowEnd; row++, locRow++) {
4413       ierr = MatGetRow(A, row, &nz, &cwork, &vwork);CHKERRQ(ierr);
4414       newRow   = rows[locRow]+locRowStart;
4415       for(col = 0, newNz = 0; col < nz; col++) {
4416         newCol = cols[cwork[col]];
4417         if ((newCol >= newRow - bw) && (newCol < newRow + bw) && (PetscAbsScalar(vwork[col]) >= tol)) {
4418           cnew[newNz] = newCol;
4419           vnew[newNz] = vwork[col];
4420           newNz++;
4421         }
4422       }
4423       ierr = MatSetValues(*B, 1, &newRow, newNz, cnew, vnew, INSERT_VALUES);CHKERRQ(ierr);
4424       ierr = MatRestoreRow(A, row, &nz, &cwork, &vwork);CHKERRQ(ierr);
4425     }
4426     ierr = PetscFree(cnew);CHKERRQ(ierr);
4427     ierr = PetscFree(vnew);CHKERRQ(ierr);
4428     ierr = MatAssemblyBegin(*B, MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
4429     ierr = MatAssemblyEnd(*B, MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
4430     ierr = ISRestoreIndices(irowp, &rows);CHKERRQ(ierr);
4431     ierr = ISRestoreIndices(icolp, &cols);CHKERRQ(ierr);
4432     ierr = ISDestroy(irowp);CHKERRQ(ierr);
4433     ierr = ISDestroy(icolp);CHKERRQ(ierr);
4434   } else {
4435     ierr = (*A->ops->permutesparsify)(A, band, frac, tol, rowp, colp, B);CHKERRQ(ierr);
4436   }
4437   ierr = PetscObjectStateIncrease((PetscObject)*B);CHKERRQ(ierr);
4438   PetscFunctionReturn(0);
4439 }
4440 
4441 #undef __FUNCT__
4442 #define __FUNCT__ "MatEqual"
4443 /*@
4444    MatEqual - Compares two matrices.
4445 
4446    Collective on Mat
4447 
4448    Input Parameters:
4449 +  A - the first matrix
4450 -  B - the second matrix
4451 
4452    Output Parameter:
4453 .  flg - PETSC_TRUE if the matrices are equal; PETSC_FALSE otherwise.
4454 
4455    Level: intermediate
4456 
4457    Concepts: matrices^equality between
4458 @*/
4459 PetscErrorCode PETSCMAT_DLLEXPORT MatEqual(Mat A,Mat B,PetscTruth *flg)
4460 {
4461   PetscErrorCode ierr;
4462 
4463   PetscFunctionBegin;
4464   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
4465   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
4466   PetscValidType(A,1);
4467   PetscValidType(B,2);
4468   PetscValidIntPointer(flg,3);
4469   PetscCheckSameComm(A,1,B,2);
4470   ierr = MatPreallocated(B);CHKERRQ(ierr);
4471   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4472   if (!B->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4473   if (A->rmap->N != B->rmap->N || A->cmap->N != B->cmap->N) SETERRQ4(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Mat A,Mat B: global dim %D %D %D %D",A->rmap->N,B->rmap->N,A->cmap->N,B->cmap->N);
4474   if (!A->ops->equal) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)A)->type_name);
4475   if (!B->ops->equal) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)B)->type_name);
4476   if (A->ops->equal != B->ops->equal) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_INCOMP,"A is type: %s\nB is type: %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
4477   ierr = MatPreallocated(A);CHKERRQ(ierr);
4478 
4479   ierr = (*A->ops->equal)(A,B,flg);CHKERRQ(ierr);
4480   PetscFunctionReturn(0);
4481 }
4482 
4483 #undef __FUNCT__
4484 #define __FUNCT__ "MatDiagonalScale"
4485 /*@
4486    MatDiagonalScale - Scales a matrix on the left and right by diagonal
4487    matrices that are stored as vectors.  Either of the two scaling
4488    matrices can be PETSC_NULL.
4489 
4490    Collective on Mat
4491 
4492    Input Parameters:
4493 +  mat - the matrix to be scaled
4494 .  l - the left scaling vector (or PETSC_NULL)
4495 -  r - the right scaling vector (or PETSC_NULL)
4496 
4497    Notes:
4498    MatDiagonalScale() computes A = LAR, where
4499    L = a diagonal matrix (stored as a vector), R = a diagonal matrix (stored as a vector)
4500    The L scales the rows of the matrix, the R scales the columns of the matrix.
4501 
4502    Level: intermediate
4503 
4504    Concepts: matrices^diagonal scaling
4505    Concepts: diagonal scaling of matrices
4506 
4507 .seealso: MatScale()
4508 @*/
4509 PetscErrorCode PETSCMAT_DLLEXPORT MatDiagonalScale(Mat mat,Vec l,Vec r)
4510 {
4511   PetscErrorCode ierr;
4512 
4513   PetscFunctionBegin;
4514   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4515   PetscValidType(mat,1);
4516   if (!mat->ops->diagonalscale) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4517   if (l) {PetscValidHeaderSpecific(l,VEC_CLASSID,2);PetscCheckSameComm(mat,1,l,2);}
4518   if (r) {PetscValidHeaderSpecific(r,VEC_CLASSID,3);PetscCheckSameComm(mat,1,r,3);}
4519   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4520   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4521   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4522 
4523   ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
4524   ierr = (*mat->ops->diagonalscale)(mat,l,r);CHKERRQ(ierr);
4525   ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
4526   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
4527   PetscFunctionReturn(0);
4528 }
4529 
4530 #undef __FUNCT__
4531 #define __FUNCT__ "MatScale"
4532 /*@
4533     MatScale - Scales all elements of a matrix by a given number.
4534 
4535     Logically Collective on Mat
4536 
4537     Input Parameters:
4538 +   mat - the matrix to be scaled
4539 -   a  - the scaling value
4540 
4541     Output Parameter:
4542 .   mat - the scaled matrix
4543 
4544     Level: intermediate
4545 
4546     Concepts: matrices^scaling all entries
4547 
4548 .seealso: MatDiagonalScale()
4549 @*/
4550 PetscErrorCode PETSCMAT_DLLEXPORT MatScale(Mat mat,PetscScalar a)
4551 {
4552   PetscErrorCode ierr;
4553 
4554   PetscFunctionBegin;
4555   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4556   PetscValidType(mat,1);
4557   if (a != 1.0 && !mat->ops->scale) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4558   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4559   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4560   PetscValidLogicalCollectiveScalar(mat,a,2);
4561   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4562 
4563   ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
4564   if (a != 1.0) {
4565     ierr = (*mat->ops->scale)(mat,a);CHKERRQ(ierr);
4566     ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
4567   }
4568   ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
4569   PetscFunctionReturn(0);
4570 }
4571 
4572 #undef __FUNCT__
4573 #define __FUNCT__ "MatNorm"
4574 /*@
4575    MatNorm - Calculates various norms of a matrix.
4576 
4577    Collective on Mat
4578 
4579    Input Parameters:
4580 +  mat - the matrix
4581 -  type - the type of norm, NORM_1, NORM_FROBENIUS, NORM_INFINITY
4582 
4583    Output Parameters:
4584 .  nrm - the resulting norm
4585 
4586    Level: intermediate
4587 
4588    Concepts: matrices^norm
4589    Concepts: norm^of matrix
4590 @*/
4591 PetscErrorCode PETSCMAT_DLLEXPORT MatNorm(Mat mat,NormType type,PetscReal *nrm)
4592 {
4593   PetscErrorCode ierr;
4594 
4595   PetscFunctionBegin;
4596   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4597   PetscValidType(mat,1);
4598   PetscValidScalarPointer(nrm,3);
4599 
4600   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
4601   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
4602   if (!mat->ops->norm) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
4603   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4604 
4605   ierr = (*mat->ops->norm)(mat,type,nrm);CHKERRQ(ierr);
4606   PetscFunctionReturn(0);
4607 }
4608 
4609 /*
4610      This variable is used to prevent counting of MatAssemblyBegin() that
4611    are called from within a MatAssemblyEnd().
4612 */
4613 static PetscInt MatAssemblyEnd_InUse = 0;
4614 #undef __FUNCT__
4615 #define __FUNCT__ "MatAssemblyBegin"
4616 /*@
4617    MatAssemblyBegin - Begins assembling the matrix.  This routine should
4618    be called after completing all calls to MatSetValues().
4619 
4620    Collective on Mat
4621 
4622    Input Parameters:
4623 +  mat - the matrix
4624 -  type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY
4625 
4626    Notes:
4627    MatSetValues() generally caches the values.  The matrix is ready to
4628    use only after MatAssemblyBegin() and MatAssemblyEnd() have been called.
4629    Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES
4630    in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before
4631    using the matrix.
4632 
4633    Level: beginner
4634 
4635    Concepts: matrices^assembling
4636 
4637 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssembled()
4638 @*/
4639 PetscErrorCode PETSCMAT_DLLEXPORT MatAssemblyBegin(Mat mat,MatAssemblyType type)
4640 {
4641   PetscErrorCode ierr;
4642 
4643   PetscFunctionBegin;
4644   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4645   PetscValidType(mat,1);
4646   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4647   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix.\nDid you forget to call MatSetUnfactored()?");
4648   if (mat->assembled) {
4649     mat->was_assembled = PETSC_TRUE;
4650     mat->assembled     = PETSC_FALSE;
4651   }
4652   if (!MatAssemblyEnd_InUse) {
4653     ierr = PetscLogEventBegin(MAT_AssemblyBegin,mat,0,0,0);CHKERRQ(ierr);
4654     if (mat->ops->assemblybegin){ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr);}
4655     ierr = PetscLogEventEnd(MAT_AssemblyBegin,mat,0,0,0);CHKERRQ(ierr);
4656   } else {
4657     if (mat->ops->assemblybegin){ierr = (*mat->ops->assemblybegin)(mat,type);CHKERRQ(ierr);}
4658   }
4659   PetscFunctionReturn(0);
4660 }
4661 
4662 #undef __FUNCT__
4663 #define __FUNCT__ "MatAssembled"
4664 /*@
4665    MatAssembled - Indicates if a matrix has been assembled and is ready for
4666      use; for example, in matrix-vector product.
4667 
4668    Not Collective
4669 
4670    Input Parameter:
4671 .  mat - the matrix
4672 
4673    Output Parameter:
4674 .  assembled - PETSC_TRUE or PETSC_FALSE
4675 
4676    Level: advanced
4677 
4678    Concepts: matrices^assembled?
4679 
4680 .seealso: MatAssemblyEnd(), MatSetValues(), MatAssemblyBegin()
4681 @*/
4682 PetscErrorCode PETSCMAT_DLLEXPORT MatAssembled(Mat mat,PetscTruth *assembled)
4683 {
4684   PetscFunctionBegin;
4685   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4686   PetscValidType(mat,1);
4687   PetscValidPointer(assembled,2);
4688   *assembled = mat->assembled;
4689   PetscFunctionReturn(0);
4690 }
4691 
4692 #undef __FUNCT__
4693 #define __FUNCT__ "MatView_Private"
4694 /*
4695     Processes command line options to determine if/how a matrix
4696   is to be viewed. Called by MatAssemblyEnd() and MatLoad().
4697 */
4698 PetscErrorCode MatView_Private(Mat mat)
4699 {
4700   PetscErrorCode    ierr;
4701   PetscTruth        flg1 = PETSC_FALSE,flg2 = PETSC_FALSE,flg3 = PETSC_FALSE,flg4 = PETSC_FALSE,flg6 = PETSC_FALSE,flg7 = PETSC_FALSE,flg8 = PETSC_FALSE;
4702   static PetscTruth incall = PETSC_FALSE;
4703 #if defined(PETSC_USE_SOCKET_VIEWER)
4704   PetscTruth        flg5 = PETSC_FALSE;
4705 #endif
4706 
4707   PetscFunctionBegin;
4708   if (incall) PetscFunctionReturn(0);
4709   incall = PETSC_TRUE;
4710   ierr = PetscOptionsBegin(((PetscObject)mat)->comm,((PetscObject)mat)->prefix,"Matrix Options","Mat");CHKERRQ(ierr);
4711     ierr = PetscOptionsTruth("-mat_view_info","Information on matrix size","MatView",flg1,&flg1,PETSC_NULL);CHKERRQ(ierr);
4712     ierr = PetscOptionsTruth("-mat_view_info_detailed","Nonzeros in the matrix","MatView",flg2,&flg2,PETSC_NULL);CHKERRQ(ierr);
4713     ierr = PetscOptionsTruth("-mat_view","Print matrix to stdout","MatView",flg3,&flg3,PETSC_NULL);CHKERRQ(ierr);
4714     ierr = PetscOptionsTruth("-mat_view_matlab","Print matrix to stdout in a format Matlab can read","MatView",flg4,&flg4,PETSC_NULL);CHKERRQ(ierr);
4715 #if defined(PETSC_USE_SOCKET_VIEWER)
4716     ierr = PetscOptionsTruth("-mat_view_socket","Send matrix to socket (can be read from matlab)","MatView",flg5,&flg5,PETSC_NULL);CHKERRQ(ierr);
4717 #endif
4718     ierr = PetscOptionsTruth("-mat_view_binary","Save matrix to file in binary format","MatView",flg6,&flg6,PETSC_NULL);CHKERRQ(ierr);
4719     ierr = PetscOptionsTruth("-mat_view_draw","Draw the matrix nonzero structure","MatView",flg7,&flg7,PETSC_NULL);CHKERRQ(ierr);
4720   ierr = PetscOptionsEnd();CHKERRQ(ierr);
4721 
4722   if (flg1) {
4723     PetscViewer viewer;
4724 
4725     ierr = PetscViewerASCIIGetStdout(((PetscObject)mat)->comm,&viewer);CHKERRQ(ierr);
4726     ierr = PetscViewerPushFormat(viewer,PETSC_VIEWER_ASCII_INFO);CHKERRQ(ierr);
4727     ierr = MatView(mat,viewer);CHKERRQ(ierr);
4728     ierr = PetscViewerPopFormat(viewer);CHKERRQ(ierr);
4729   }
4730   if (flg2) {
4731     PetscViewer viewer;
4732 
4733     ierr = PetscViewerASCIIGetStdout(((PetscObject)mat)->comm,&viewer);CHKERRQ(ierr);
4734     ierr = PetscViewerPushFormat(viewer,PETSC_VIEWER_ASCII_INFO_DETAIL);CHKERRQ(ierr);
4735     ierr = MatView(mat,viewer);CHKERRQ(ierr);
4736     ierr = PetscViewerPopFormat(viewer);CHKERRQ(ierr);
4737   }
4738   if (flg3) {
4739     PetscViewer viewer;
4740 
4741     ierr = PetscViewerASCIIGetStdout(((PetscObject)mat)->comm,&viewer);CHKERRQ(ierr);
4742     ierr = MatView(mat,viewer);CHKERRQ(ierr);
4743   }
4744   if (flg4) {
4745     PetscViewer viewer;
4746 
4747     ierr = PetscViewerASCIIGetStdout(((PetscObject)mat)->comm,&viewer);CHKERRQ(ierr);
4748     ierr = PetscViewerPushFormat(viewer,PETSC_VIEWER_ASCII_MATLAB);CHKERRQ(ierr);
4749     ierr = MatView(mat,viewer);CHKERRQ(ierr);
4750     ierr = PetscViewerPopFormat(viewer);CHKERRQ(ierr);
4751   }
4752 #if defined(PETSC_USE_SOCKET_VIEWER)
4753   if (flg5) {
4754     ierr = MatView(mat,PETSC_VIEWER_SOCKET_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4755     ierr = PetscViewerFlush(PETSC_VIEWER_SOCKET_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4756   }
4757 #endif
4758   if (flg6) {
4759     ierr = MatView(mat,PETSC_VIEWER_BINARY_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4760     ierr = PetscViewerFlush(PETSC_VIEWER_BINARY_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4761   }
4762   if (flg7) {
4763     ierr = PetscOptionsGetTruth(((PetscObject)mat)->prefix,"-mat_view_contour",&flg8,PETSC_NULL);CHKERRQ(ierr);
4764     if (flg8) {
4765       PetscViewerPushFormat(PETSC_VIEWER_DRAW_(((PetscObject)mat)->comm),PETSC_VIEWER_DRAW_CONTOUR);CHKERRQ(ierr);
4766     }
4767     ierr = MatView(mat,PETSC_VIEWER_DRAW_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4768     ierr = PetscViewerFlush(PETSC_VIEWER_DRAW_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4769     if (flg8) {
4770       PetscViewerPopFormat(PETSC_VIEWER_DRAW_(((PetscObject)mat)->comm));CHKERRQ(ierr);
4771     }
4772   }
4773   incall = PETSC_FALSE;
4774   PetscFunctionReturn(0);
4775 }
4776 
4777 #undef __FUNCT__
4778 #define __FUNCT__ "MatAssemblyEnd"
4779 /*@
4780    MatAssemblyEnd - Completes assembling the matrix.  This routine should
4781    be called after MatAssemblyBegin().
4782 
4783    Collective on Mat
4784 
4785    Input Parameters:
4786 +  mat - the matrix
4787 -  type - type of assembly, either MAT_FLUSH_ASSEMBLY or MAT_FINAL_ASSEMBLY
4788 
4789    Options Database Keys:
4790 +  -mat_view_info - Prints info on matrix at conclusion of MatEndAssembly()
4791 .  -mat_view_info_detailed - Prints more detailed info
4792 .  -mat_view - Prints matrix in ASCII format
4793 .  -mat_view_matlab - Prints matrix in Matlab format
4794 .  -mat_view_draw - PetscDraws nonzero structure of matrix, using MatView() and PetscDrawOpenX().
4795 .  -display <name> - Sets display name (default is host)
4796 .  -draw_pause <sec> - Sets number of seconds to pause after display
4797 .  -mat_view_socket - Sends matrix to socket, can be accessed from Matlab (See the <a href="../../docs/manual.pdf">users manual</a>)
4798 .  -viewer_socket_machine <machine>
4799 .  -viewer_socket_port <port>
4800 .  -mat_view_binary - save matrix to file in binary format
4801 -  -viewer_binary_filename <name>
4802 
4803    Notes:
4804    MatSetValues() generally caches the values.  The matrix is ready to
4805    use only after MatAssemblyBegin() and MatAssemblyEnd() have been called.
4806    Use MAT_FLUSH_ASSEMBLY when switching between ADD_VALUES and INSERT_VALUES
4807    in MatSetValues(); use MAT_FINAL_ASSEMBLY for the final assembly before
4808    using the matrix.
4809 
4810    Level: beginner
4811 
4812 .seealso: MatAssemblyBegin(), MatSetValues(), PetscDrawOpenX(), MatView(), MatAssembled(), PetscViewerSocketOpen()
4813 @*/
4814 PetscErrorCode PETSCMAT_DLLEXPORT MatAssemblyEnd(Mat mat,MatAssemblyType type)
4815 {
4816   PetscErrorCode  ierr;
4817   static PetscInt inassm = 0;
4818   PetscTruth      flg = PETSC_FALSE;
4819 
4820   PetscFunctionBegin;
4821   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4822   PetscValidType(mat,1);
4823 
4824   inassm++;
4825   MatAssemblyEnd_InUse++;
4826   if (MatAssemblyEnd_InUse == 1) { /* Do the logging only the first time through */
4827     ierr = PetscLogEventBegin(MAT_AssemblyEnd,mat,0,0,0);CHKERRQ(ierr);
4828     if (mat->ops->assemblyend) {
4829       ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr);
4830     }
4831     ierr = PetscLogEventEnd(MAT_AssemblyEnd,mat,0,0,0);CHKERRQ(ierr);
4832   } else {
4833     if (mat->ops->assemblyend) {
4834       ierr = (*mat->ops->assemblyend)(mat,type);CHKERRQ(ierr);
4835     }
4836   }
4837 
4838   /* Flush assembly is not a true assembly */
4839   if (type != MAT_FLUSH_ASSEMBLY) {
4840     mat->assembled  = PETSC_TRUE; mat->num_ass++;
4841   }
4842   mat->insertmode = NOT_SET_VALUES;
4843   MatAssemblyEnd_InUse--;
4844   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
4845   if (!mat->symmetric_eternal) {
4846     mat->symmetric_set              = PETSC_FALSE;
4847     mat->hermitian_set              = PETSC_FALSE;
4848     mat->structurally_symmetric_set = PETSC_FALSE;
4849   }
4850   if (inassm == 1 && type != MAT_FLUSH_ASSEMBLY) {
4851     ierr = MatView_Private(mat);CHKERRQ(ierr);
4852     ierr = PetscOptionsGetTruth(((PetscObject)mat)->prefix,"-mat_is_symmetric",&flg,PETSC_NULL);CHKERRQ(ierr);
4853     if (flg) {
4854       PetscReal tol = 0.0;
4855       ierr = PetscOptionsGetReal(((PetscObject)mat)->prefix,"-mat_is_symmetric",&tol,PETSC_NULL);CHKERRQ(ierr);
4856       ierr = MatIsSymmetric(mat,tol,&flg);CHKERRQ(ierr);
4857       if (flg) {
4858         ierr = PetscPrintf(((PetscObject)mat)->comm,"Matrix is symmetric (tolerance %G)\n",tol);CHKERRQ(ierr);
4859       } else {
4860         ierr = PetscPrintf(((PetscObject)mat)->comm,"Matrix is not symmetric (tolerance %G)\n",tol);CHKERRQ(ierr);
4861       }
4862     }
4863   }
4864   inassm--;
4865   PetscFunctionReturn(0);
4866 }
4867 
4868 #undef __FUNCT__
4869 #define __FUNCT__ "MatSetOption"
4870 /*@
4871    MatSetOption - Sets a parameter option for a matrix. Some options
4872    may be specific to certain storage formats.  Some options
4873    determine how values will be inserted (or added). Sorted,
4874    row-oriented input will generally assemble the fastest. The default
4875    is row-oriented, nonsorted input.
4876 
4877    Logically Collective on Mat
4878 
4879    Input Parameters:
4880 +  mat - the matrix
4881 .  option - the option, one of those listed below (and possibly others),
4882 -  flg - turn the option on (PETSC_TRUE) or off (PETSC_FALSE)
4883 
4884   Options Describing Matrix Structure:
4885 +    MAT_SPD - symmetric positive definite
4886 -    MAT_SYMMETRIC - symmetric in terms of both structure and value
4887 .    MAT_HERMITIAN - transpose is the complex conjugation
4888 .    MAT_STRUCTURALLY_SYMMETRIC - symmetric nonzero structure
4889 -    MAT_SYMMETRY_ETERNAL - if you would like the symmetry/Hermitian flag
4890                             you set to be kept with all future use of the matrix
4891                             including after MatAssemblyBegin/End() which could
4892                             potentially change the symmetry structure, i.e. you
4893                             KNOW the matrix will ALWAYS have the property you set.
4894 
4895 
4896    Options For Use with MatSetValues():
4897    Insert a logically dense subblock, which can be
4898 .    MAT_ROW_ORIENTED - row-oriented (default)
4899 
4900    Note these options reflect the data you pass in with MatSetValues(); it has
4901    nothing to do with how the data is stored internally in the matrix
4902    data structure.
4903 
4904    When (re)assembling a matrix, we can restrict the input for
4905    efficiency/debugging purposes.  These options include
4906 +    MAT_NEW_NONZERO_LOCATIONS - additional insertions will be
4907         allowed if they generate a new nonzero
4908 .    MAT_NEW_DIAGONALS - new diagonals will be allowed (for block diagonal format only)
4909 .    MAT_IGNORE_OFF_PROC_ENTRIES - drops off-processor entries
4910 .    MAT_NEW_NONZERO_LOCATION_ERR - generates an error for new matrix entry
4911 .    MAT_USE_HASH_TABLE - uses a hash table to speed up matrix assembly
4912 +    MAT_NO_OFF_PROC_ENTRIES - you know each process will only set values for its own rows, will generate an error if
4913         any process sets values for another process. This avoids all reductions in the MatAssembly routines and thus improves
4914         performance for very large process counts.
4915 
4916    Notes:
4917    Some options are relevant only for particular matrix types and
4918    are thus ignored by others.  Other options are not supported by
4919    certain matrix types and will generate an error message if set.
4920 
4921    If using a Fortran 77 module to compute a matrix, one may need to
4922    use the column-oriented option (or convert to the row-oriented
4923    format).
4924 
4925    MAT_NEW_NONZERO_LOCATIONS set to PETSC_FALSE indicates that any add or insertion
4926    that would generate a new entry in the nonzero structure is instead
4927    ignored.  Thus, if memory has not alredy been allocated for this particular
4928    data, then the insertion is ignored. For dense matrices, in which
4929    the entire array is allocated, no entries are ever ignored.
4930    Set after the first MatAssemblyEnd()
4931 
4932    MAT_NEW_NONZERO_LOCATION_ERR indicates that any add or insertion
4933    that would generate a new entry in the nonzero structure instead produces
4934    an error. (Currently supported for AIJ and BAIJ formats only.)
4935    This is a useful flag when using SAME_NONZERO_PATTERN in calling
4936    KSPSetOperators() to ensure that the nonzero pattern truely does
4937    remain unchanged. Set after the first MatAssemblyEnd()
4938 
4939    MAT_NEW_NONZERO_ALLOCATION_ERR indicates that any add or insertion
4940    that would generate a new entry that has not been preallocated will
4941    instead produce an error. (Currently supported for AIJ and BAIJ formats
4942    only.) This is a useful flag when debugging matrix memory preallocation.
4943 
4944    MAT_IGNORE_OFF_PROC_ENTRIES indicates entries destined for
4945    other processors should be dropped, rather than stashed.
4946    This is useful if you know that the "owning" processor is also
4947    always generating the correct matrix entries, so that PETSc need
4948    not transfer duplicate entries generated on another processor.
4949 
4950    MAT_USE_HASH_TABLE indicates that a hash table be used to improve the
4951    searches during matrix assembly. When this flag is set, the hash table
4952    is created during the first Matrix Assembly. This hash table is
4953    used the next time through, during MatSetVaules()/MatSetVaulesBlocked()
4954    to improve the searching of indices. MAT_NEW_NONZERO_LOCATIONS flag
4955    should be used with MAT_USE_HASH_TABLE flag. This option is currently
4956    supported by MATMPIBAIJ format only.
4957 
4958    MAT_KEEP_NONZERO_PATTERN indicates when MatZeroRows() is called the zeroed entries
4959    are kept in the nonzero structure
4960 
4961    MAT_IGNORE_ZERO_ENTRIES - for AIJ/IS matrices this will stop zero values from creating
4962    a zero location in the matrix
4963 
4964    MAT_USE_INODES - indicates using inode version of the code - works with AIJ and
4965    ROWBS matrix types
4966 
4967   MAT_NO_OFF_PROC_ZERO_ROWS - you know each process will only zero its own rows. This avoids all reductions in the
4968         zero row routines and thus improves performance for very large process counts.
4969 
4970    Level: intermediate
4971 
4972    Concepts: matrices^setting options
4973 
4974 @*/
4975 PetscErrorCode PETSCMAT_DLLEXPORT MatSetOption(Mat mat,MatOption op,PetscTruth flg)
4976 {
4977   PetscErrorCode ierr;
4978 
4979   PetscFunctionBegin;
4980   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
4981   PetscValidType(mat,1);
4982   PetscValidLogicalCollectiveEnum(mat,op,2);
4983   PetscValidLogicalCollectiveTruth(mat,flg,3);
4984 
4985   if (((int) op) < 0 || ((int) op) >= NUM_MAT_OPTIONS) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Options %d is out of range",(int)op);
4986   if (!((PetscObject)mat)->type_name) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_TYPENOTSET,"Cannot set options until type and size have been set, see MatSetType() and MatSetSizes()");
4987   ierr = MatPreallocated(mat);CHKERRQ(ierr);
4988   switch (op) {
4989   case MAT_NO_OFF_PROC_ENTRIES:
4990     mat->nooffprocentries                = flg;
4991     PetscFunctionReturn(0);
4992     break;
4993   case MAT_NO_OFF_PROC_ZERO_ROWS:
4994     mat->nooffproczerorows               = flg;
4995     PetscFunctionReturn(0);
4996     break;
4997   case MAT_SPD:
4998     mat->spd_set                         = PETSC_TRUE;
4999     mat->spd                             = flg;
5000     if (flg) {
5001       mat->symmetric                     = PETSC_TRUE;
5002       mat->structurally_symmetric        = PETSC_TRUE;
5003       mat->symmetric_set                 = PETSC_TRUE;
5004       mat->structurally_symmetric_set    = PETSC_TRUE;
5005     }
5006     break;
5007   case MAT_SYMMETRIC:
5008     mat->symmetric                       = flg;
5009     if (flg) mat->structurally_symmetric = PETSC_TRUE;
5010     mat->symmetric_set                   = PETSC_TRUE;
5011     mat->structurally_symmetric_set      = flg;
5012     break;
5013   case MAT_HERMITIAN:
5014     mat->hermitian                       = flg;
5015     if (flg) mat->structurally_symmetric = PETSC_TRUE;
5016     mat->hermitian_set                   = PETSC_TRUE;
5017     mat->structurally_symmetric_set      = flg;
5018     break;
5019   case MAT_STRUCTURALLY_SYMMETRIC:
5020     mat->structurally_symmetric          = flg;
5021     mat->structurally_symmetric_set      = PETSC_TRUE;
5022     break;
5023   case MAT_SYMMETRY_ETERNAL:
5024     mat->symmetric_eternal               = flg;
5025     break;
5026   default:
5027     break;
5028   }
5029   if (mat->ops->setoption) {
5030     ierr = (*mat->ops->setoption)(mat,op,flg);CHKERRQ(ierr);
5031   }
5032   PetscFunctionReturn(0);
5033 }
5034 
5035 #undef __FUNCT__
5036 #define __FUNCT__ "MatZeroEntries"
5037 /*@
5038    MatZeroEntries - Zeros all entries of a matrix.  For sparse matrices
5039    this routine retains the old nonzero structure.
5040 
5041    Logically Collective on Mat
5042 
5043    Input Parameters:
5044 .  mat - the matrix
5045 
5046    Level: intermediate
5047 
5048    Concepts: matrices^zeroing
5049 
5050 .seealso: MatZeroRows()
5051 @*/
5052 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroEntries(Mat mat)
5053 {
5054   PetscErrorCode ierr;
5055 
5056   PetscFunctionBegin;
5057   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5058   PetscValidType(mat,1);
5059   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5060   if (mat->insertmode != NOT_SET_VALUES) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for matrices where you have set values but not yet assembled");
5061   if (!mat->ops->zeroentries) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5062   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5063 
5064   ierr = PetscLogEventBegin(MAT_ZeroEntries,mat,0,0,0);CHKERRQ(ierr);
5065   ierr = (*mat->ops->zeroentries)(mat);CHKERRQ(ierr);
5066   ierr = PetscLogEventEnd(MAT_ZeroEntries,mat,0,0,0);CHKERRQ(ierr);
5067   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
5068   PetscFunctionReturn(0);
5069 }
5070 
5071 #undef __FUNCT__
5072 #define __FUNCT__ "MatZeroRows"
5073 /*@C
5074    MatZeroRows - Zeros all entries (except possibly the main diagonal)
5075    of a set of rows of a matrix.
5076 
5077    Collective on Mat
5078 
5079    Input Parameters:
5080 +  mat - the matrix
5081 .  numRows - the number of rows to remove
5082 .  rows - the global row indices
5083 -  diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
5084 
5085    Notes:
5086    For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
5087    but does not release memory.  For the dense and block diagonal
5088    formats this does not alter the nonzero structure.
5089 
5090    If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
5091    of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
5092    merely zeroed.
5093 
5094    The user can set a value in the diagonal entry (or for the AIJ and
5095    row formats can optionally remove the main diagonal entry from the
5096    nonzero structure as well, by passing 0.0 as the final argument).
5097 
5098    For the parallel case, all processes that share the matrix (i.e.,
5099    those in the communicator used for matrix creation) MUST call this
5100    routine, regardless of whether any rows being zeroed are owned by
5101    them.
5102 
5103    Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
5104    list only rows local to itself).
5105 
5106    Level: intermediate
5107 
5108    Concepts: matrices^zeroing rows
5109 
5110 .seealso: MatZeroRowsIS(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption()
5111 @*/
5112 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroRows(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag)
5113 {
5114   PetscErrorCode ierr;
5115 
5116   PetscFunctionBegin;
5117   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5118   PetscValidType(mat,1);
5119   if (numRows) PetscValidIntPointer(rows,3);
5120   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5121   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5122   if (!mat->ops->zerorows) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5123   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5124 
5125   ierr = (*mat->ops->zerorows)(mat,numRows,rows,diag);CHKERRQ(ierr);
5126   ierr = MatView_Private(mat);CHKERRQ(ierr);
5127   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
5128   PetscFunctionReturn(0);
5129 }
5130 
5131 #undef __FUNCT__
5132 #define __FUNCT__ "MatZeroRowsIS"
5133 /*@C
5134    MatZeroRowsIS - Zeros all entries (except possibly the main diagonal)
5135    of a set of rows of a matrix.
5136 
5137    Collective on Mat
5138 
5139    Input Parameters:
5140 +  mat - the matrix
5141 .  is - index set of rows to remove
5142 -  diag - value put in all diagonals of eliminated rows
5143 
5144    Notes:
5145    For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
5146    but does not release memory.  For the dense and block diagonal
5147    formats this does not alter the nonzero structure.
5148 
5149    If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
5150    of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
5151    merely zeroed.
5152 
5153    The user can set a value in the diagonal entry (or for the AIJ and
5154    row formats can optionally remove the main diagonal entry from the
5155    nonzero structure as well, by passing 0.0 as the final argument).
5156 
5157    For the parallel case, all processes that share the matrix (i.e.,
5158    those in the communicator used for matrix creation) MUST call this
5159    routine, regardless of whether any rows being zeroed are owned by
5160    them.
5161 
5162    Each processor should list the rows that IT wants zeroed
5163 
5164    Level: intermediate
5165 
5166    Concepts: matrices^zeroing rows
5167 
5168 .seealso: MatZeroRows(), MatZeroRowsStencil(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption()
5169 @*/
5170 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroRowsIS(Mat mat,IS is,PetscScalar diag)
5171 {
5172   PetscInt       numRows;
5173   const PetscInt *rows;
5174   PetscErrorCode ierr;
5175 
5176   PetscFunctionBegin;
5177   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5178   PetscValidType(mat,1);
5179   PetscValidHeaderSpecific(is,IS_CLASSID,2);
5180   ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr);
5181   ierr = ISGetIndices(is,&rows);CHKERRQ(ierr);
5182   ierr = MatZeroRows(mat,numRows,rows,diag);CHKERRQ(ierr);
5183   ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr);
5184   PetscFunctionReturn(0);
5185 }
5186 
5187 #undef __FUNCT__
5188 #define __FUNCT__ "MatZeroRowsStencil"
5189 /*@C
5190    MatZeroRowsStencil - Zeros all entries (except possibly the main diagonal)
5191    of a set of rows of a matrix. These rows must be local to the process.
5192 
5193    Collective on Mat
5194 
5195    Input Parameters:
5196 +  mat - the matrix
5197 .  numRows - the number of rows to remove
5198 .  rows - the grid coordinates (and component number when dof > 1) for matrix rows
5199 -  diag - value put in all diagonals of eliminated rows (0.0 will even eliminate diagonal entry)
5200 
5201    Notes:
5202    For the AIJ and BAIJ matrix formats this removes the old nonzero structure,
5203    but does not release memory.  For the dense and block diagonal
5204    formats this does not alter the nonzero structure.
5205 
5206    If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
5207    of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
5208    merely zeroed.
5209 
5210    The user can set a value in the diagonal entry (or for the AIJ and
5211    row formats can optionally remove the main diagonal entry from the
5212    nonzero structure as well, by passing 0.0 as the final argument).
5213 
5214    For the parallel case, all processes that share the matrix (i.e.,
5215    those in the communicator used for matrix creation) MUST call this
5216    routine, regardless of whether any rows being zeroed are owned by
5217    them.
5218 
5219    Each processor can indicate any rows in the entire matrix to be zeroed (i.e. each process does NOT have to
5220    list only rows local to itself).
5221 
5222    The grid coordinates are across the entire grid, not just the local portion
5223 
5224    In Fortran idxm and idxn should be declared as
5225 $     MatStencil idxm(4,m)
5226    and the values inserted using
5227 $    idxm(MatStencil_i,1) = i
5228 $    idxm(MatStencil_j,1) = j
5229 $    idxm(MatStencil_k,1) = k
5230 $    idxm(MatStencil_c,1) = c
5231    etc
5232 
5233    For periodic boundary conditions use negative indices for values to the left (below 0; that are to be
5234    obtained by wrapping values from right edge). For values to the right of the last entry using that index plus one
5235    etc to obtain values that obtained by wrapping the values from the left edge. This does not work for the DA_NONPERIODIC
5236    wrap.
5237 
5238    For indices that don't mean anything for your case (like the k index when working in 2d) or the c index when you have
5239    a single value per point) you can skip filling those indices.
5240 
5241    Level: intermediate
5242 
5243    Concepts: matrices^zeroing rows
5244 
5245 .seealso: MatZeroRows(), MatZeroRowsIS(), MatZeroEntries(), MatZeroRowsLocal(), MatSetOption()
5246 @*/
5247 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroRowsStencil(Mat mat,PetscInt numRows,const MatStencil rows[],PetscScalar diag)
5248 {
5249   PetscInt       dim    = mat->stencil.dim;
5250   PetscInt       sdim   = dim - (1 - (PetscInt) mat->stencil.noc);
5251   PetscInt      *dims   = mat->stencil.dims+1;
5252   PetscInt      *starts = mat->stencil.starts;
5253   PetscInt      *dxm    = (PetscInt *) rows;
5254   PetscInt      *jdxm, i, j, tmp, numNewRows = 0;
5255   PetscErrorCode ierr;
5256 
5257   PetscFunctionBegin;
5258   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5259   PetscValidType(mat,1);
5260   if (numRows) PetscValidIntPointer(rows,3);
5261 
5262   ierr = PetscMalloc(numRows * sizeof(PetscInt), &jdxm);CHKERRQ(ierr);
5263   for(i = 0; i < numRows; ++i) {
5264     /* Skip unused dimensions (they are ordered k, j, i, c) */
5265     for(j = 0; j < 3-sdim; ++j) dxm++;
5266     /* Local index in X dir */
5267     tmp = *dxm++ - starts[0];
5268     /* Loop over remaining dimensions */
5269     for(j = 0; j < dim-1; ++j) {
5270       /* If nonlocal, set index to be negative */
5271       if ((*dxm++ - starts[j+1]) < 0 || tmp < 0) tmp = PETSC_MIN_INT;
5272       /* Update local index */
5273       else                                       tmp = tmp*dims[j] + *(dxm-1) - starts[j+1];
5274     }
5275     /* Skip component slot if necessary */
5276     if (mat->stencil.noc) dxm++;
5277     /* Local row number */
5278     if (tmp >= 0) {
5279       jdxm[numNewRows++] = tmp;
5280     }
5281   }
5282   ierr = MatZeroRowsLocal(mat,numNewRows,jdxm,diag);CHKERRQ(ierr);
5283   ierr = PetscFree(jdxm);CHKERRQ(ierr);
5284   PetscFunctionReturn(0);
5285 }
5286 
5287 #undef __FUNCT__
5288 #define __FUNCT__ "MatZeroRowsLocal"
5289 /*@C
5290    MatZeroRowsLocal - Zeros all entries (except possibly the main diagonal)
5291    of a set of rows of a matrix; using local numbering of rows.
5292 
5293    Logically Collective on Mat
5294 
5295    Input Parameters:
5296 +  mat - the matrix
5297 .  numRows - the number of rows to remove
5298 .  rows - the global row indices
5299 -  diag - value put in all diagonals of eliminated rows
5300 
5301    Notes:
5302    Before calling MatZeroRowsLocal(), the user must first set the
5303    local-to-global mapping by calling MatSetLocalToGlobalMapping().
5304 
5305    For the AIJ matrix formats this removes the old nonzero structure,
5306    but does not release memory.  For the dense and block diagonal
5307    formats this does not alter the nonzero structure.
5308 
5309    If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
5310    of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
5311    merely zeroed.
5312 
5313    The user can set a value in the diagonal entry (or for the AIJ and
5314    row formats can optionally remove the main diagonal entry from the
5315    nonzero structure as well, by passing 0.0 as the final argument).
5316 
5317    Level: intermediate
5318 
5319    Concepts: matrices^zeroing
5320 
5321 .seealso: MatZeroRows(), MatZeroRowsLocalIS(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping
5322 @*/
5323 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroRowsLocal(Mat mat,PetscInt numRows,const PetscInt rows[],PetscScalar diag)
5324 {
5325   PetscErrorCode ierr;
5326 
5327   PetscFunctionBegin;
5328   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5329   PetscValidType(mat,1);
5330   if (numRows) PetscValidIntPointer(rows,3);
5331   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5332   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5333   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5334 
5335   if (mat->ops->zerorowslocal) {
5336     ierr = (*mat->ops->zerorowslocal)(mat,numRows,rows,diag);CHKERRQ(ierr);
5337   } else {
5338     IS             is, newis;
5339     const PetscInt *newRows;
5340 
5341     if (!mat->mapping) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Need to provide local to global mapping to matrix first");
5342     ierr = ISCreateGeneral(PETSC_COMM_SELF,numRows,rows,&is);CHKERRQ(ierr);
5343     ierr = ISLocalToGlobalMappingApplyIS(mat->mapping,is,&newis);CHKERRQ(ierr);
5344     ierr = ISGetIndices(newis,&newRows);CHKERRQ(ierr);
5345     ierr = (*mat->ops->zerorows)(mat,numRows,newRows,diag);CHKERRQ(ierr);
5346     ierr = ISRestoreIndices(newis,&newRows);CHKERRQ(ierr);
5347     ierr = ISDestroy(newis);CHKERRQ(ierr);
5348     ierr = ISDestroy(is);CHKERRQ(ierr);
5349   }
5350   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
5351   PetscFunctionReturn(0);
5352 }
5353 
5354 #undef __FUNCT__
5355 #define __FUNCT__ "MatZeroRowsLocalIS"
5356 /*@C
5357    MatZeroRowsLocalIS - Zeros all entries (except possibly the main diagonal)
5358    of a set of rows of a matrix; using local numbering of rows.
5359 
5360    Logically Collective on Mat
5361 
5362    Input Parameters:
5363 +  mat - the matrix
5364 .  is - index set of rows to remove
5365 -  diag - value put in all diagonals of eliminated rows
5366 
5367    Notes:
5368    Before calling MatZeroRowsLocalIS(), the user must first set the
5369    local-to-global mapping by calling MatSetLocalToGlobalMapping().
5370 
5371    For the AIJ matrix formats this removes the old nonzero structure,
5372    but does not release memory.  For the dense and block diagonal
5373    formats this does not alter the nonzero structure.
5374 
5375    If the option MatSetOption(mat,MAT_KEEP_NONZERO_PATTERN,PETSC_TRUE) the nonzero structure
5376    of the matrix is not changed (even for AIJ and BAIJ matrices) the values are
5377    merely zeroed.
5378 
5379    The user can set a value in the diagonal entry (or for the AIJ and
5380    row formats can optionally remove the main diagonal entry from the
5381    nonzero structure as well, by passing 0.0 as the final argument).
5382 
5383    Level: intermediate
5384 
5385    Concepts: matrices^zeroing
5386 
5387 .seealso: MatZeroRows(), MatZeroRowsLocal(), MatZeroEntries(), MatZeroRows(), MatSetLocalToGlobalMapping
5388 @*/
5389 PetscErrorCode PETSCMAT_DLLEXPORT MatZeroRowsLocalIS(Mat mat,IS is,PetscScalar diag)
5390 {
5391   PetscErrorCode ierr;
5392   PetscInt       numRows;
5393   const PetscInt *rows;
5394 
5395   PetscFunctionBegin;
5396   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5397   PetscValidType(mat,1);
5398   PetscValidHeaderSpecific(is,IS_CLASSID,2);
5399   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5400   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5401   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5402 
5403   ierr = ISGetLocalSize(is,&numRows);CHKERRQ(ierr);
5404   ierr = ISGetIndices(is,&rows);CHKERRQ(ierr);
5405   ierr = MatZeroRowsLocal(mat,numRows,rows,diag);CHKERRQ(ierr);
5406   ierr = ISRestoreIndices(is,&rows);CHKERRQ(ierr);
5407   PetscFunctionReturn(0);
5408 }
5409 
5410 #undef __FUNCT__
5411 #define __FUNCT__ "MatGetSize"
5412 /*@
5413    MatGetSize - Returns the numbers of rows and columns in a matrix.
5414 
5415    Not Collective
5416 
5417    Input Parameter:
5418 .  mat - the matrix
5419 
5420    Output Parameters:
5421 +  m - the number of global rows
5422 -  n - the number of global columns
5423 
5424    Note: both output parameters can be PETSC_NULL on input.
5425 
5426    Level: beginner
5427 
5428    Concepts: matrices^size
5429 
5430 .seealso: MatGetLocalSize()
5431 @*/
5432 PetscErrorCode PETSCMAT_DLLEXPORT MatGetSize(Mat mat,PetscInt *m,PetscInt* n)
5433 {
5434   PetscFunctionBegin;
5435   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5436   if (m) *m = mat->rmap->N;
5437   if (n) *n = mat->cmap->N;
5438   PetscFunctionReturn(0);
5439 }
5440 
5441 #undef __FUNCT__
5442 #define __FUNCT__ "MatGetLocalSize"
5443 /*@
5444    MatGetLocalSize - Returns the number of rows and columns in a matrix
5445    stored locally.  This information may be implementation dependent, so
5446    use with care.
5447 
5448    Not Collective
5449 
5450    Input Parameters:
5451 .  mat - the matrix
5452 
5453    Output Parameters:
5454 +  m - the number of local rows
5455 -  n - the number of local columns
5456 
5457    Note: both output parameters can be PETSC_NULL on input.
5458 
5459    Level: beginner
5460 
5461    Concepts: matrices^local size
5462 
5463 .seealso: MatGetSize()
5464 @*/
5465 PetscErrorCode PETSCMAT_DLLEXPORT MatGetLocalSize(Mat mat,PetscInt *m,PetscInt* n)
5466 {
5467   PetscFunctionBegin;
5468   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5469   if (m) PetscValidIntPointer(m,2);
5470   if (n) PetscValidIntPointer(n,3);
5471   if (m) *m = mat->rmap->n;
5472   if (n) *n = mat->cmap->n;
5473   PetscFunctionReturn(0);
5474 }
5475 
5476 #undef __FUNCT__
5477 #define __FUNCT__ "MatGetOwnershipRangeColumn"
5478 /*@
5479    MatGetOwnershipRangeColumn - Returns the range of matrix columns owned by
5480    this processor.
5481 
5482    Not Collective, unless matrix has not been allocated, then collective on Mat
5483 
5484    Input Parameters:
5485 .  mat - the matrix
5486 
5487    Output Parameters:
5488 +  m - the global index of the first local column
5489 -  n - one more than the global index of the last local column
5490 
5491    Notes: both output parameters can be PETSC_NULL on input.
5492 
5493    Level: developer
5494 
5495    Concepts: matrices^column ownership
5496 
5497 .seealso:  MatGetOwnershipRange(), MatGetOwnershipRanges(), MatGetOwnershipRangesColumn()
5498 
5499 @*/
5500 PetscErrorCode PETSCMAT_DLLEXPORT MatGetOwnershipRangeColumn(Mat mat,PetscInt *m,PetscInt* n)
5501 {
5502   PetscErrorCode ierr;
5503 
5504   PetscFunctionBegin;
5505   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5506   PetscValidType(mat,1);
5507   if (m) PetscValidIntPointer(m,2);
5508   if (n) PetscValidIntPointer(n,3);
5509   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5510   if (m) *m = mat->cmap->rstart;
5511   if (n) *n = mat->cmap->rend;
5512   PetscFunctionReturn(0);
5513 }
5514 
5515 #undef __FUNCT__
5516 #define __FUNCT__ "MatGetOwnershipRange"
5517 /*@
5518    MatGetOwnershipRange - Returns the range of matrix rows owned by
5519    this processor, assuming that the matrix is laid out with the first
5520    n1 rows on the first processor, the next n2 rows on the second, etc.
5521    For certain parallel layouts this range may not be well defined.
5522 
5523    Not Collective, unless matrix has not been allocated, then collective on Mat
5524 
5525    Input Parameters:
5526 .  mat - the matrix
5527 
5528    Output Parameters:
5529 +  m - the global index of the first local row
5530 -  n - one more than the global index of the last local row
5531 
5532    Note: both output parameters can be PETSC_NULL on input.
5533 
5534    Level: beginner
5535 
5536    Concepts: matrices^row ownership
5537 
5538 .seealso:   MatGetOwnershipRanges(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn()
5539 
5540 @*/
5541 PetscErrorCode PETSCMAT_DLLEXPORT MatGetOwnershipRange(Mat mat,PetscInt *m,PetscInt* n)
5542 {
5543   PetscErrorCode ierr;
5544 
5545   PetscFunctionBegin;
5546   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5547   PetscValidType(mat,1);
5548   if (m) PetscValidIntPointer(m,2);
5549   if (n) PetscValidIntPointer(n,3);
5550   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5551   if (m) *m = mat->rmap->rstart;
5552   if (n) *n = mat->rmap->rend;
5553   PetscFunctionReturn(0);
5554 }
5555 
5556 #undef __FUNCT__
5557 #define __FUNCT__ "MatGetOwnershipRanges"
5558 /*@C
5559    MatGetOwnershipRanges - Returns the range of matrix rows owned by
5560    each process
5561 
5562    Not Collective, unless matrix has not been allocated, then collective on Mat
5563 
5564    Input Parameters:
5565 .  mat - the matrix
5566 
5567    Output Parameters:
5568 .  ranges - start of each processors portion plus one more then the total length at the end
5569 
5570    Level: beginner
5571 
5572    Concepts: matrices^row ownership
5573 
5574 .seealso:   MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRangesColumn()
5575 
5576 @*/
5577 PetscErrorCode PETSCMAT_DLLEXPORT MatGetOwnershipRanges(Mat mat,const PetscInt **ranges)
5578 {
5579   PetscErrorCode ierr;
5580 
5581   PetscFunctionBegin;
5582   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5583   PetscValidType(mat,1);
5584   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5585   ierr = PetscLayoutGetRanges(mat->rmap,ranges);CHKERRQ(ierr);
5586   PetscFunctionReturn(0);
5587 }
5588 
5589 #undef __FUNCT__
5590 #define __FUNCT__ "MatGetOwnershipRangesColumn"
5591 /*@C
5592    MatGetOwnershipRangesColumn - Returns the range of local columns for each process
5593 
5594    Not Collective, unless matrix has not been allocated, then collective on Mat
5595 
5596    Input Parameters:
5597 .  mat - the matrix
5598 
5599    Output Parameters:
5600 .  ranges - start of each processors portion plus one more then the total length at the end
5601 
5602    Level: beginner
5603 
5604    Concepts: matrices^column ownership
5605 
5606 .seealso:   MatGetOwnershipRange(), MatGetOwnershipRangeColumn(), MatGetOwnershipRanges()
5607 
5608 @*/
5609 PetscErrorCode PETSCMAT_DLLEXPORT MatGetOwnershipRangesColumn(Mat mat,const PetscInt **ranges)
5610 {
5611   PetscErrorCode ierr;
5612 
5613   PetscFunctionBegin;
5614   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5615   PetscValidType(mat,1);
5616   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5617   ierr = PetscLayoutGetRanges(mat->cmap,ranges);CHKERRQ(ierr);
5618   PetscFunctionReturn(0);
5619 }
5620 
5621 #undef __FUNCT__
5622 #define __FUNCT__ "MatILUFactorSymbolic"
5623 /*@C
5624    MatILUFactorSymbolic - Performs symbolic ILU factorization of a matrix.
5625    Uses levels of fill only, not drop tolerance. Use MatLUFactorNumeric()
5626    to complete the factorization.
5627 
5628    Collective on Mat
5629 
5630    Input Parameters:
5631 +  mat - the matrix
5632 .  row - row permutation
5633 .  column - column permutation
5634 -  info - structure containing
5635 $      levels - number of levels of fill.
5636 $      expected fill - as ratio of original fill.
5637 $      1 or 0 - indicating force fill on diagonal (improves robustness for matrices
5638                 missing diagonal entries)
5639 
5640    Output Parameters:
5641 .  fact - new matrix that has been symbolically factored
5642 
5643    Notes:
5644    See the <a href="../../docs/manual.pdf">users manual</a>  for additional information about
5645    choosing the fill factor for better efficiency.
5646 
5647    Most users should employ the simplified KSP interface for linear solvers
5648    instead of working directly with matrix algebra routines such as this.
5649    See, e.g., KSPCreate().
5650 
5651    Level: developer
5652 
5653   Concepts: matrices^symbolic LU factorization
5654   Concepts: matrices^factorization
5655   Concepts: LU^symbolic factorization
5656 
5657 .seealso: MatLUFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor()
5658           MatGetOrdering(), MatFactorInfo
5659 
5660     Developer Note: fortran interface is not autogenerated as the f90
5661     interface defintion cannot be generated correctly [due to MatFactorInfo]
5662 
5663 @*/
5664 PetscErrorCode PETSCMAT_DLLEXPORT MatILUFactorSymbolic(Mat fact,Mat mat,IS row,IS col,const MatFactorInfo *info)
5665 {
5666   PetscErrorCode ierr;
5667 
5668   PetscFunctionBegin;
5669   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5670   PetscValidType(mat,1);
5671   PetscValidHeaderSpecific(row,IS_CLASSID,2);
5672   PetscValidHeaderSpecific(col,IS_CLASSID,3);
5673   PetscValidPointer(info,4);
5674   PetscValidPointer(fact,5);
5675   if (info->levels < 0) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Levels of fill negative %D",(PetscInt)info->levels);
5676   if (info->fill < 1.0) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %G",info->fill);
5677   if (!(fact)->ops->ilufactorsymbolic) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Matrix type %s  symbolic ILU",((PetscObject)mat)->type_name);
5678   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5679   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5680   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5681 
5682   ierr = PetscLogEventBegin(MAT_ILUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr);
5683   ierr = (fact->ops->ilufactorsymbolic)(fact,mat,row,col,info);CHKERRQ(ierr);
5684   ierr = PetscLogEventEnd(MAT_ILUFactorSymbolic,mat,row,col,0);CHKERRQ(ierr);
5685   PetscFunctionReturn(0);
5686 }
5687 
5688 #undef __FUNCT__
5689 #define __FUNCT__ "MatICCFactorSymbolic"
5690 /*@C
5691    MatICCFactorSymbolic - Performs symbolic incomplete
5692    Cholesky factorization for a symmetric matrix.  Use
5693    MatCholeskyFactorNumeric() to complete the factorization.
5694 
5695    Collective on Mat
5696 
5697    Input Parameters:
5698 +  mat - the matrix
5699 .  perm - row and column permutation
5700 -  info - structure containing
5701 $      levels - number of levels of fill.
5702 $      expected fill - as ratio of original fill.
5703 
5704    Output Parameter:
5705 .  fact - the factored matrix
5706 
5707    Notes:
5708    Most users should employ the KSP interface for linear solvers
5709    instead of working directly with matrix algebra routines such as this.
5710    See, e.g., KSPCreate().
5711 
5712    Level: developer
5713 
5714   Concepts: matrices^symbolic incomplete Cholesky factorization
5715   Concepts: matrices^factorization
5716   Concepts: Cholsky^symbolic factorization
5717 
5718 .seealso: MatCholeskyFactorNumeric(), MatCholeskyFactor(), MatFactorInfo
5719 
5720     Developer Note: fortran interface is not autogenerated as the f90
5721     interface defintion cannot be generated correctly [due to MatFactorInfo]
5722 
5723 @*/
5724 PetscErrorCode PETSCMAT_DLLEXPORT MatICCFactorSymbolic(Mat fact,Mat mat,IS perm,const MatFactorInfo *info)
5725 {
5726   PetscErrorCode ierr;
5727 
5728   PetscFunctionBegin;
5729   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5730   PetscValidType(mat,1);
5731   PetscValidHeaderSpecific(perm,IS_CLASSID,2);
5732   PetscValidPointer(info,3);
5733   PetscValidPointer(fact,4);
5734   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5735   if (info->levels < 0) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Levels negative %D",(PetscInt) info->levels);
5736   if (info->fill < 1.0) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Expected fill less than 1.0 %G",info->fill);
5737   if (!(fact)->ops->iccfactorsymbolic) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Matrix type %s  symbolic ICC",((PetscObject)mat)->type_name);
5738   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5739   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5740 
5741   ierr = PetscLogEventBegin(MAT_ICCFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr);
5742   ierr = (fact->ops->iccfactorsymbolic)(fact,mat,perm,info);CHKERRQ(ierr);
5743   ierr = PetscLogEventEnd(MAT_ICCFactorSymbolic,mat,perm,0,0);CHKERRQ(ierr);
5744   PetscFunctionReturn(0);
5745 }
5746 
5747 #undef __FUNCT__
5748 #define __FUNCT__ "MatGetArray"
5749 /*@C
5750    MatGetArray - Returns a pointer to the element values in the matrix.
5751    The result of this routine is dependent on the underlying matrix data
5752    structure, and may not even work for certain matrix types.  You MUST
5753    call MatRestoreArray() when you no longer need to access the array.
5754 
5755    Not Collective
5756 
5757    Input Parameter:
5758 .  mat - the matrix
5759 
5760    Output Parameter:
5761 .  v - the location of the values
5762 
5763 
5764    Fortran Note:
5765    This routine is used differently from Fortran, e.g.,
5766 .vb
5767         Mat         mat
5768         PetscScalar mat_array(1)
5769         PetscOffset i_mat
5770         PetscErrorCode ierr
5771         call MatGetArray(mat,mat_array,i_mat,ierr)
5772 
5773   C  Access first local entry in matrix; note that array is
5774   C  treated as one dimensional
5775         value = mat_array(i_mat + 1)
5776 
5777         [... other code ...]
5778         call MatRestoreArray(mat,mat_array,i_mat,ierr)
5779 .ve
5780 
5781    See the <a href="../../docs/manual.pdf#ch_fortran">Fortran chapter of the users manual</a> and
5782    src/mat/examples/tests for details.
5783 
5784    Level: advanced
5785 
5786    Concepts: matrices^access array
5787 
5788 .seealso: MatRestoreArray(), MatGetArrayF90(), MatGetRowIJ()
5789 @*/
5790 PetscErrorCode PETSCMAT_DLLEXPORT MatGetArray(Mat mat,PetscScalar *v[])
5791 {
5792   PetscErrorCode ierr;
5793 
5794   PetscFunctionBegin;
5795   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5796   PetscValidType(mat,1);
5797   PetscValidPointer(v,2);
5798   if (!mat->ops->getarray) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5799   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5800   ierr = (*mat->ops->getarray)(mat,v);CHKERRQ(ierr);
5801   CHKMEMQ;
5802   PetscFunctionReturn(0);
5803 }
5804 
5805 #undef __FUNCT__
5806 #define __FUNCT__ "MatRestoreArray"
5807 /*@C
5808    MatRestoreArray - Restores the matrix after MatGetArray() has been called.
5809 
5810    Not Collective
5811 
5812    Input Parameter:
5813 +  mat - the matrix
5814 -  v - the location of the values
5815 
5816    Fortran Note:
5817    This routine is used differently from Fortran, e.g.,
5818 .vb
5819         Mat         mat
5820         PetscScalar mat_array(1)
5821         PetscOffset i_mat
5822         PetscErrorCode ierr
5823         call MatGetArray(mat,mat_array,i_mat,ierr)
5824 
5825   C  Access first local entry in matrix; note that array is
5826   C  treated as one dimensional
5827         value = mat_array(i_mat + 1)
5828 
5829         [... other code ...]
5830         call MatRestoreArray(mat,mat_array,i_mat,ierr)
5831 .ve
5832 
5833    See the <a href="../../docs/manual.pdf#ch_fortran">Fortran chapter of the users manual</a>
5834    src/mat/examples/tests for details
5835 
5836    Level: advanced
5837 
5838 .seealso: MatGetArray(), MatRestoreArrayF90()
5839 @*/
5840 PetscErrorCode PETSCMAT_DLLEXPORT MatRestoreArray(Mat mat,PetscScalar *v[])
5841 {
5842   PetscErrorCode ierr;
5843 
5844   PetscFunctionBegin;
5845   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5846   PetscValidType(mat,1);
5847   PetscValidPointer(v,2);
5848 #if defined(PETSC_USE_DEBUG)
5849   CHKMEMQ;
5850 #endif
5851   if (!mat->ops->restorearray) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5852   ierr = (*mat->ops->restorearray)(mat,v);CHKERRQ(ierr);
5853   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
5854   PetscFunctionReturn(0);
5855 }
5856 
5857 #undef __FUNCT__
5858 #define __FUNCT__ "MatGetSubMatrices"
5859 /*@C
5860    MatGetSubMatrices - Extracts several submatrices from a matrix. If submat
5861    points to an array of valid matrices, they may be reused to store the new
5862    submatrices.
5863 
5864    Collective on Mat
5865 
5866    Input Parameters:
5867 +  mat - the matrix
5868 .  n   - the number of submatrixes to be extracted (on this processor, may be zero)
5869 .  irow, icol - index sets of rows and columns to extract
5870 -  scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
5871 
5872    Output Parameter:
5873 .  submat - the array of submatrices
5874 
5875    Notes:
5876    MatGetSubMatrices() can extract ONLY sequential submatrices
5877    (from both sequential and parallel matrices). Use MatGetSubMatrix()
5878    to extract a parallel submatrix.
5879 
5880    When extracting submatrices from a parallel matrix, each processor can
5881    form a different submatrix by setting the rows and columns of its
5882    individual index sets according to the local submatrix desired.
5883 
5884    When finished using the submatrices, the user should destroy
5885    them with MatDestroyMatrices().
5886 
5887    MAT_REUSE_MATRIX can only be used when the nonzero structure of the
5888    original matrix has not changed from that last call to MatGetSubMatrices().
5889 
5890    This routine creates the matrices in submat; you should NOT create them before
5891    calling it. It also allocates the array of matrix pointers submat.
5892 
5893    For BAIJ matrices the index sets must respect the block structure, that is if they
5894    request one row/column in a block, they must request all rows/columns that are in
5895    that block. For example, if the block size is 2 you cannot request just row 0 and
5896    column 0.
5897 
5898    Fortran Note:
5899    The Fortran interface is slightly different from that given below; it
5900    requires one to pass in  as submat a Mat (integer) array of size at least m.
5901 
5902    Level: advanced
5903 
5904    Concepts: matrices^accessing submatrices
5905    Concepts: submatrices
5906 
5907 .seealso: MatDestroyMatrices(), MatGetSubMatrix(), MatGetRow(), MatGetDiagonal(), MatReuse
5908 @*/
5909 PetscErrorCode PETSCMAT_DLLEXPORT MatGetSubMatrices(Mat mat,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *submat[])
5910 {
5911   PetscErrorCode ierr;
5912   PetscInt        i;
5913   PetscTruth      eq;
5914 
5915   PetscFunctionBegin;
5916   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
5917   PetscValidType(mat,1);
5918   if (n) {
5919     PetscValidPointer(irow,3);
5920     PetscValidHeaderSpecific(*irow,IS_CLASSID,3);
5921     PetscValidPointer(icol,4);
5922     PetscValidHeaderSpecific(*icol,IS_CLASSID,4);
5923   }
5924   PetscValidPointer(submat,6);
5925   if (n && scall == MAT_REUSE_MATRIX) {
5926     PetscValidPointer(*submat,6);
5927     PetscValidHeaderSpecific(**submat,MAT_CLASSID,6);
5928   }
5929   if (!mat->ops->getsubmatrices) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
5930   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
5931   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
5932   ierr = MatPreallocated(mat);CHKERRQ(ierr);
5933 
5934   ierr = PetscLogEventBegin(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr);
5935   ierr = (*mat->ops->getsubmatrices)(mat,n,irow,icol,scall,submat);CHKERRQ(ierr);
5936   ierr = PetscLogEventEnd(MAT_GetSubMatrices,mat,0,0,0);CHKERRQ(ierr);
5937   for (i=0; i<n; i++) {
5938     if (mat->symmetric || mat->structurally_symmetric || mat->hermitian) {
5939       ierr = ISEqual(irow[i],icol[i],&eq);CHKERRQ(ierr);
5940       if (eq) {
5941 	if (mat->symmetric){
5942 	  ierr = MatSetOption((*submat)[i],MAT_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr);
5943 	} else if (mat->hermitian) {
5944 	  ierr = MatSetOption((*submat)[i],MAT_HERMITIAN,PETSC_TRUE);CHKERRQ(ierr);
5945 	} else if (mat->structurally_symmetric) {
5946 	  ierr = MatSetOption((*submat)[i],MAT_STRUCTURALLY_SYMMETRIC,PETSC_TRUE);CHKERRQ(ierr);
5947 	}
5948       }
5949     }
5950   }
5951   PetscFunctionReturn(0);
5952 }
5953 
5954 #undef __FUNCT__
5955 #define __FUNCT__ "MatDestroyMatrices"
5956 /*@C
5957    MatDestroyMatrices - Destroys a set of matrices obtained with MatGetSubMatrices().
5958 
5959    Collective on Mat
5960 
5961    Input Parameters:
5962 +  n - the number of local matrices
5963 -  mat - the matrices (note that this is a pointer to the array of matrices, just to match the calling
5964                        sequence of MatGetSubMatrices())
5965 
5966    Level: advanced
5967 
5968     Notes: Frees not only the matrices, but also the array that contains the matrices
5969            In Fortran will not free the array.
5970 
5971 .seealso: MatGetSubMatrices()
5972 @*/
5973 PetscErrorCode PETSCMAT_DLLEXPORT MatDestroyMatrices(PetscInt n,Mat *mat[])
5974 {
5975   PetscErrorCode ierr;
5976   PetscInt       i;
5977 
5978   PetscFunctionBegin;
5979   if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Trying to destroy negative number of matrices %D",n);
5980   PetscValidPointer(mat,2);
5981   for (i=0; i<n; i++) {
5982     ierr = MatDestroy((*mat)[i]);CHKERRQ(ierr);
5983   }
5984   /* memory is allocated even if n = 0 */
5985   ierr = PetscFree(*mat);CHKERRQ(ierr);
5986   PetscFunctionReturn(0);
5987 }
5988 
5989 #undef __FUNCT__
5990 #define __FUNCT__ "MatGetSeqNonzeroStructure"
5991 /*@C
5992    MatGetSeqNonzeroStructure - Extracts the sequential nonzero structure from a matrix.
5993 
5994    Collective on Mat
5995 
5996    Input Parameters:
5997 .  mat - the matrix
5998 
5999    Output Parameter:
6000 .  matstruct - the sequential matrix with the nonzero structure of mat
6001 
6002   Level: intermediate
6003 
6004 .seealso: MatDestroySeqNonzeroStructure(), MatGetSubMatrices(), MatDestroyMatrices()
6005 @*/
6006 PetscErrorCode PETSCMAT_DLLEXPORT MatGetSeqNonzeroStructure(Mat mat,Mat *matstruct)
6007 {
6008   PetscErrorCode ierr;
6009 
6010   PetscFunctionBegin;
6011   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6012   PetscValidPointer(matstruct,2);
6013 
6014   PetscValidType(mat,1);
6015   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6016   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6017 
6018   if (!mat->ops->getseqnonzerostructure) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Not for matrix type %s\n",((PetscObject)mat)->type_name);
6019   ierr = PetscLogEventBegin(MAT_GetSeqNonzeroStructure,mat,0,0,0);CHKERRQ(ierr);
6020   ierr = (*mat->ops->getseqnonzerostructure)(mat,matstruct);CHKERRQ(ierr);
6021   ierr = PetscLogEventEnd(MAT_GetSeqNonzeroStructure,mat,0,0,0);CHKERRQ(ierr);
6022   PetscFunctionReturn(0);
6023 }
6024 
6025 #undef __FUNCT__
6026 #define __FUNCT__ "MatDestroySeqNonzeroStructure"
6027 /*@C
6028    MatDestroySeqNonzeroStructure - Destroys matrix obtained with MatGetSeqNonzeroStructure().
6029 
6030    Collective on Mat
6031 
6032    Input Parameters:
6033 .  mat - the matrix (note that this is a pointer to the array of matrices, just to match the calling
6034                        sequence of MatGetSequentialNonzeroStructure())
6035 
6036    Level: advanced
6037 
6038     Notes: Frees not only the matrices, but also the array that contains the matrices
6039 
6040 .seealso: MatGetSeqNonzeroStructure()
6041 @*/
6042 PetscErrorCode PETSCMAT_DLLEXPORT MatDestroySeqNonzeroStructure(Mat *mat)
6043 {
6044   PetscErrorCode ierr;
6045 
6046   PetscFunctionBegin;
6047   PetscValidPointer(mat,1);
6048   ierr = MatDestroy(*mat);CHKERRQ(ierr);
6049   PetscFunctionReturn(0);
6050 }
6051 
6052 #undef __FUNCT__
6053 #define __FUNCT__ "MatIncreaseOverlap"
6054 /*@
6055    MatIncreaseOverlap - Given a set of submatrices indicated by index sets,
6056    replaces the index sets by larger ones that represent submatrices with
6057    additional overlap.
6058 
6059    Collective on Mat
6060 
6061    Input Parameters:
6062 +  mat - the matrix
6063 .  n   - the number of index sets
6064 .  is  - the array of index sets (these index sets will changed during the call)
6065 -  ov  - the additional overlap requested
6066 
6067    Level: developer
6068 
6069    Concepts: overlap
6070    Concepts: ASM^computing overlap
6071 
6072 .seealso: MatGetSubMatrices()
6073 @*/
6074 PetscErrorCode PETSCMAT_DLLEXPORT MatIncreaseOverlap(Mat mat,PetscInt n,IS is[],PetscInt ov)
6075 {
6076   PetscErrorCode ierr;
6077 
6078   PetscFunctionBegin;
6079   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6080   PetscValidType(mat,1);
6081   if (n < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Must have one or more domains, you have %D",n);
6082   if (n) {
6083     PetscValidPointer(is,3);
6084     PetscValidHeaderSpecific(*is,IS_CLASSID,3);
6085   }
6086   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6087   if (mat->factortype)     SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6088   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6089 
6090   if (!ov) PetscFunctionReturn(0);
6091   if (!mat->ops->increaseoverlap) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6092   ierr = PetscLogEventBegin(MAT_IncreaseOverlap,mat,0,0,0);CHKERRQ(ierr);
6093   ierr = (*mat->ops->increaseoverlap)(mat,n,is,ov);CHKERRQ(ierr);
6094   ierr = PetscLogEventEnd(MAT_IncreaseOverlap,mat,0,0,0);CHKERRQ(ierr);
6095   PetscFunctionReturn(0);
6096 }
6097 
6098 #undef __FUNCT__
6099 #define __FUNCT__ "MatGetBlockSize"
6100 /*@
6101    MatGetBlockSize - Returns the matrix block size; useful especially for the
6102    block row and block diagonal formats.
6103 
6104    Not Collective
6105 
6106    Input Parameter:
6107 .  mat - the matrix
6108 
6109    Output Parameter:
6110 .  bs - block size
6111 
6112    Notes:
6113    Block row formats are MATSEQBAIJ, MATMPIBAIJ, MATSEQSBAIJ, MATMPISBAIJ
6114 
6115    Level: intermediate
6116 
6117    Concepts: matrices^block size
6118 
6119 .seealso: MatCreateSeqBAIJ(), MatCreateMPIBAIJ()
6120 @*/
6121 PetscErrorCode PETSCMAT_DLLEXPORT MatGetBlockSize(Mat mat,PetscInt *bs)
6122 {
6123   PetscErrorCode ierr;
6124 
6125   PetscFunctionBegin;
6126   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6127   PetscValidType(mat,1);
6128   PetscValidIntPointer(bs,2);
6129   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6130   *bs = mat->rmap->bs;
6131   PetscFunctionReturn(0);
6132 }
6133 
6134 #undef __FUNCT__
6135 #define __FUNCT__ "MatSetBlockSize"
6136 /*@
6137    MatSetBlockSize - Sets the matrix block size; for many matrix types you
6138      cannot use this and MUST set the blocksize when you preallocate the matrix
6139 
6140    Logically Collective on Mat
6141 
6142    Input Parameters:
6143 +  mat - the matrix
6144 -  bs - block size
6145 
6146    Notes:
6147      For BAIJ matrices, this just checks that the block size agrees with the BAIJ size,
6148      it is not possible to change BAIJ block sizes after preallocation.
6149 
6150    Level: intermediate
6151 
6152    Concepts: matrices^block size
6153 
6154 .seealso: MatCreateSeqBAIJ(), MatCreateMPIBAIJ(), MatGetBlockSize()
6155 @*/
6156 PetscErrorCode PETSCMAT_DLLEXPORT MatSetBlockSize(Mat mat,PetscInt bs)
6157 {
6158   PetscErrorCode ierr;
6159 
6160   PetscFunctionBegin;
6161   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6162   PetscValidType(mat,1);
6163   PetscValidLogicalCollectiveInt(mat,bs,2);
6164   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6165   if (bs < 1) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block size %d, must be positive",bs);
6166   if (mat->ops->setblocksize) {
6167     ierr = (*mat->ops->setblocksize)(mat,bs);CHKERRQ(ierr);
6168   } else {
6169     SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Cannot set the blocksize for matrix type %s",((PetscObject)mat)->type_name);
6170   }
6171   PetscFunctionReturn(0);
6172 }
6173 
6174 #undef __FUNCT__
6175 #define __FUNCT__ "MatGetRowIJ"
6176 /*@C
6177     MatGetRowIJ - Returns the compressed row storage i and j indices for sequential matrices.
6178 
6179    Collective on Mat
6180 
6181     Input Parameters:
6182 +   mat - the matrix
6183 .   shift -  0 or 1 indicating we want the indices starting at 0 or 1
6184 .   symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
6185                 symmetrized
6186 -   inodecompressed - PETSC_TRUE or PETSC_FALSE  indicating if the nonzero structure of the
6187                  inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
6188                  always used.
6189 
6190     Output Parameters:
6191 +   n - number of rows in the (possibly compressed) matrix
6192 .   ia - the row pointers [of length n+1]
6193 .   ja - the column indices
6194 -   done - indicates if the routine actually worked and returned appropriate ia[] and ja[] arrays; callers
6195            are responsible for handling the case when done == PETSC_FALSE and ia and ja are not set
6196 
6197     Level: developer
6198 
6199     Notes: You CANNOT change any of the ia[] or ja[] values.
6200 
6201            Use MatRestoreRowIJ() when you are finished accessing the ia[] and ja[] values
6202 
6203     Fortran Node
6204 
6205            In Fortran use
6206 $           PetscInt ia(1), ja(1)
6207 $           PetscOffset iia, jja
6208 $      call MatGetRowIJ(mat,shift,symmetric,inodecompressed,n,ia,iia,ja,jja,done,ierr)
6209 $
6210 $          or
6211 $
6212 $           PetscScalar, pointer :: xx_v(:)
6213 $    call  MatGetRowIJF90(mat,shift,symmetric,inodecompressed,n,ia,ja,done,ierr)
6214 
6215 
6216        Acess the ith and jth entries via ia(iia + i) and ja(jja + j)
6217 
6218 .seealso: MatGetColumnIJ(), MatRestoreRowIJ(), MatGetArray()
6219 @*/
6220 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRowIJ(Mat mat,PetscInt shift,PetscTruth symmetric,PetscTruth inodecompressed,PetscInt *n,PetscInt *ia[],PetscInt* ja[],PetscTruth *done)
6221 {
6222   PetscErrorCode ierr;
6223 
6224   PetscFunctionBegin;
6225   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6226   PetscValidType(mat,1);
6227   PetscValidIntPointer(n,4);
6228   if (ia) PetscValidIntPointer(ia,5);
6229   if (ja) PetscValidIntPointer(ja,6);
6230   PetscValidIntPointer(done,7);
6231   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6232   if (!mat->ops->getrowij) *done = PETSC_FALSE;
6233   else {
6234     *done = PETSC_TRUE;
6235     ierr = PetscLogEventBegin(MAT_GetRowIJ,mat,0,0,0);CHKERRQ(ierr);
6236     ierr  = (*mat->ops->getrowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr);
6237     ierr = PetscLogEventEnd(MAT_GetRowIJ,mat,0,0,0);CHKERRQ(ierr);
6238   }
6239   PetscFunctionReturn(0);
6240 }
6241 
6242 #undef __FUNCT__
6243 #define __FUNCT__ "MatGetColumnIJ"
6244 /*@C
6245     MatGetColumnIJ - Returns the compressed column storage i and j indices for sequential matrices.
6246 
6247     Collective on Mat
6248 
6249     Input Parameters:
6250 +   mat - the matrix
6251 .   shift - 1 or zero indicating we want the indices starting at 0 or 1
6252 .   symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
6253                 symmetrized
6254 -   inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
6255                  inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
6256                  always used.
6257 
6258     Output Parameters:
6259 +   n - number of columns in the (possibly compressed) matrix
6260 .   ia - the column pointers
6261 .   ja - the row indices
6262 -   done - PETSC_TRUE or PETSC_FALSE, indicating whether the values have been returned
6263 
6264     Level: developer
6265 
6266 .seealso: MatGetRowIJ(), MatRestoreColumnIJ()
6267 @*/
6268 PetscErrorCode PETSCMAT_DLLEXPORT MatGetColumnIJ(Mat mat,PetscInt shift,PetscTruth symmetric,PetscTruth inodecompressed,PetscInt *n,PetscInt *ia[],PetscInt* ja[],PetscTruth *done)
6269 {
6270   PetscErrorCode ierr;
6271 
6272   PetscFunctionBegin;
6273   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6274   PetscValidType(mat,1);
6275   PetscValidIntPointer(n,4);
6276   if (ia) PetscValidIntPointer(ia,5);
6277   if (ja) PetscValidIntPointer(ja,6);
6278   PetscValidIntPointer(done,7);
6279   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6280   if (!mat->ops->getcolumnij) *done = PETSC_FALSE;
6281   else {
6282     *done = PETSC_TRUE;
6283     ierr  = (*mat->ops->getcolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr);
6284   }
6285   PetscFunctionReturn(0);
6286 }
6287 
6288 #undef __FUNCT__
6289 #define __FUNCT__ "MatRestoreRowIJ"
6290 /*@C
6291     MatRestoreRowIJ - Call after you are completed with the ia,ja indices obtained with
6292     MatGetRowIJ().
6293 
6294     Collective on Mat
6295 
6296     Input Parameters:
6297 +   mat - the matrix
6298 .   shift - 1 or zero indicating we want the indices starting at 0 or 1
6299 .   symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
6300                 symmetrized
6301 -   inodecompressed -  PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
6302                  inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
6303                  always used.
6304 
6305     Output Parameters:
6306 +   n - size of (possibly compressed) matrix
6307 .   ia - the row pointers
6308 .   ja - the column indices
6309 -   done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned
6310 
6311     Level: developer
6312 
6313 .seealso: MatGetRowIJ(), MatRestoreColumnIJ()
6314 @*/
6315 PetscErrorCode PETSCMAT_DLLEXPORT MatRestoreRowIJ(Mat mat,PetscInt shift,PetscTruth symmetric,PetscTruth inodecompressed,PetscInt *n,PetscInt *ia[],PetscInt* ja[],PetscTruth *done)
6316 {
6317   PetscErrorCode ierr;
6318 
6319   PetscFunctionBegin;
6320   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6321   PetscValidType(mat,1);
6322   if (ia) PetscValidIntPointer(ia,5);
6323   if (ja) PetscValidIntPointer(ja,6);
6324   PetscValidIntPointer(done,7);
6325   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6326 
6327   if (!mat->ops->restorerowij) *done = PETSC_FALSE;
6328   else {
6329     *done = PETSC_TRUE;
6330     ierr  = (*mat->ops->restorerowij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr);
6331   }
6332   PetscFunctionReturn(0);
6333 }
6334 
6335 #undef __FUNCT__
6336 #define __FUNCT__ "MatRestoreColumnIJ"
6337 /*@C
6338     MatRestoreColumnIJ - Call after you are completed with the ia,ja indices obtained with
6339     MatGetColumnIJ().
6340 
6341     Collective on Mat
6342 
6343     Input Parameters:
6344 +   mat - the matrix
6345 .   shift - 1 or zero indicating we want the indices starting at 0 or 1
6346 -   symmetric - PETSC_TRUE or PETSC_FALSE indicating the matrix data structure should be
6347                 symmetrized
6348 -   inodecompressed - PETSC_TRUE or PETSC_FALSE indicating if the nonzero structure of the
6349                  inodes or the nonzero elements is wanted. For BAIJ matrices the compressed version is
6350                  always used.
6351 
6352     Output Parameters:
6353 +   n - size of (possibly compressed) matrix
6354 .   ia - the column pointers
6355 .   ja - the row indices
6356 -   done - PETSC_TRUE or PETSC_FALSE indicated that the values have been returned
6357 
6358     Level: developer
6359 
6360 .seealso: MatGetColumnIJ(), MatRestoreRowIJ()
6361 @*/
6362 PetscErrorCode PETSCMAT_DLLEXPORT MatRestoreColumnIJ(Mat mat,PetscInt shift,PetscTruth symmetric,PetscTruth inodecompressed,PetscInt *n,PetscInt *ia[],PetscInt* ja[],PetscTruth *done)
6363 {
6364   PetscErrorCode ierr;
6365 
6366   PetscFunctionBegin;
6367   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6368   PetscValidType(mat,1);
6369   if (ia) PetscValidIntPointer(ia,5);
6370   if (ja) PetscValidIntPointer(ja,6);
6371   PetscValidIntPointer(done,7);
6372   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6373 
6374   if (!mat->ops->restorecolumnij) *done = PETSC_FALSE;
6375   else {
6376     *done = PETSC_TRUE;
6377     ierr  = (*mat->ops->restorecolumnij)(mat,shift,symmetric,inodecompressed,n,ia,ja,done);CHKERRQ(ierr);
6378   }
6379   PetscFunctionReturn(0);
6380 }
6381 
6382 #undef __FUNCT__
6383 #define __FUNCT__ "MatColoringPatch"
6384 /*@C
6385     MatColoringPatch -Used inside matrix coloring routines that
6386     use MatGetRowIJ() and/or MatGetColumnIJ().
6387 
6388     Collective on Mat
6389 
6390     Input Parameters:
6391 +   mat - the matrix
6392 .   ncolors - max color value
6393 .   n   - number of entries in colorarray
6394 -   colorarray - array indicating color for each column
6395 
6396     Output Parameters:
6397 .   iscoloring - coloring generated using colorarray information
6398 
6399     Level: developer
6400 
6401 .seealso: MatGetRowIJ(), MatGetColumnIJ()
6402 
6403 @*/
6404 PetscErrorCode PETSCMAT_DLLEXPORT MatColoringPatch(Mat mat,PetscInt ncolors,PetscInt n,ISColoringValue colorarray[],ISColoring *iscoloring)
6405 {
6406   PetscErrorCode ierr;
6407 
6408   PetscFunctionBegin;
6409   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6410   PetscValidType(mat,1);
6411   PetscValidIntPointer(colorarray,4);
6412   PetscValidPointer(iscoloring,5);
6413   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6414 
6415   if (!mat->ops->coloringpatch){
6416     ierr = ISColoringCreate(((PetscObject)mat)->comm,ncolors,n,colorarray,iscoloring);CHKERRQ(ierr);
6417   } else {
6418     ierr = (*mat->ops->coloringpatch)(mat,ncolors,n,colorarray,iscoloring);CHKERRQ(ierr);
6419   }
6420   PetscFunctionReturn(0);
6421 }
6422 
6423 
6424 #undef __FUNCT__
6425 #define __FUNCT__ "MatSetUnfactored"
6426 /*@
6427    MatSetUnfactored - Resets a factored matrix to be treated as unfactored.
6428 
6429    Logically Collective on Mat
6430 
6431    Input Parameter:
6432 .  mat - the factored matrix to be reset
6433 
6434    Notes:
6435    This routine should be used only with factored matrices formed by in-place
6436    factorization via ILU(0) (or by in-place LU factorization for the MATSEQDENSE
6437    format).  This option can save memory, for example, when solving nonlinear
6438    systems with a matrix-free Newton-Krylov method and a matrix-based, in-place
6439    ILU(0) preconditioner.
6440 
6441    Note that one can specify in-place ILU(0) factorization by calling
6442 .vb
6443      PCType(pc,PCILU);
6444      PCFactorSeUseInPlace(pc);
6445 .ve
6446    or by using the options -pc_type ilu -pc_factor_in_place
6447 
6448    In-place factorization ILU(0) can also be used as a local
6449    solver for the blocks within the block Jacobi or additive Schwarz
6450    methods (runtime option: -sub_pc_factor_in_place).  See the discussion
6451    of these preconditioners in the <a href="../../docs/manual.pdf#ch_pc">PC chapter of the users manual</a> for details on setting
6452    local solver options.
6453 
6454    Most users should employ the simplified KSP interface for linear solvers
6455    instead of working directly with matrix algebra routines such as this.
6456    See, e.g., KSPCreate().
6457 
6458    Level: developer
6459 
6460 .seealso: PCFactorSetUseInPlace()
6461 
6462    Concepts: matrices^unfactored
6463 
6464 @*/
6465 PetscErrorCode PETSCMAT_DLLEXPORT MatSetUnfactored(Mat mat)
6466 {
6467   PetscErrorCode ierr;
6468 
6469   PetscFunctionBegin;
6470   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6471   PetscValidType(mat,1);
6472   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6473   mat->factortype = MAT_FACTOR_NONE;
6474   if (!mat->ops->setunfactored) PetscFunctionReturn(0);
6475   ierr = (*mat->ops->setunfactored)(mat);CHKERRQ(ierr);
6476   PetscFunctionReturn(0);
6477 }
6478 
6479 /*MC
6480     MatGetArrayF90 - Accesses a matrix array from Fortran90.
6481 
6482     Synopsis:
6483     MatGetArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
6484 
6485     Not collective
6486 
6487     Input Parameter:
6488 .   x - matrix
6489 
6490     Output Parameters:
6491 +   xx_v - the Fortran90 pointer to the array
6492 -   ierr - error code
6493 
6494     Example of Usage:
6495 .vb
6496       PetscScalar, pointer xx_v(:)
6497       ....
6498       call MatGetArrayF90(x,xx_v,ierr)
6499       a = xx_v(3)
6500       call MatRestoreArrayF90(x,xx_v,ierr)
6501 .ve
6502 
6503     Notes:
6504     Not yet supported for all F90 compilers
6505 
6506     Level: advanced
6507 
6508 .seealso:  MatRestoreArrayF90(), MatGetArray(), MatRestoreArray()
6509 
6510     Concepts: matrices^accessing array
6511 
6512 M*/
6513 
6514 /*MC
6515     MatRestoreArrayF90 - Restores a matrix array that has been
6516     accessed with MatGetArrayF90().
6517 
6518     Synopsis:
6519     MatRestoreArrayF90(Mat x,{Scalar, pointer :: xx_v(:)},integer ierr)
6520 
6521     Not collective
6522 
6523     Input Parameters:
6524 +   x - matrix
6525 -   xx_v - the Fortran90 pointer to the array
6526 
6527     Output Parameter:
6528 .   ierr - error code
6529 
6530     Example of Usage:
6531 .vb
6532        PetscScalar, pointer xx_v(:)
6533        ....
6534        call MatGetArrayF90(x,xx_v,ierr)
6535        a = xx_v(3)
6536        call MatRestoreArrayF90(x,xx_v,ierr)
6537 .ve
6538 
6539     Notes:
6540     Not yet supported for all F90 compilers
6541 
6542     Level: advanced
6543 
6544 .seealso:  MatGetArrayF90(), MatGetArray(), MatRestoreArray()
6545 
6546 M*/
6547 
6548 
6549 #undef __FUNCT__
6550 #define __FUNCT__ "MatGetSubMatrix"
6551 /*@
6552     MatGetSubMatrix - Gets a single submatrix on the same number of processors
6553                       as the original matrix.
6554 
6555     Collective on Mat
6556 
6557     Input Parameters:
6558 +   mat - the original matrix
6559 .   isrow - parallel IS containing the rows this processor should obtain
6560 .   iscol - parallel IS containing all columns you wish to keep. Each process should list the columns that will be in IT's "diagonal part" in the new matrix.
6561 -   cll - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
6562 
6563     Output Parameter:
6564 .   newmat - the new submatrix, of the same type as the old
6565 
6566     Level: advanced
6567 
6568     Notes:
6569     The submatrix will be able to be multiplied with vectors using the same layout as iscol.
6570 
6571     The rows in isrow will be sorted into the same order as the original matrix on each process.
6572 
6573       The first time this is called you should use a cll of MAT_INITIAL_MATRIX,
6574    the MatGetSubMatrix() routine will create the newmat for you. Any additional calls
6575    to this routine with a mat of the same nonzero structure and with a call of MAT_REUSE_MATRIX
6576    will reuse the matrix generated the first time.  You should call MatDestroy() on newmat when
6577    you are finished using it.
6578 
6579     The communicator of the newly obtained matrix is ALWAYS the same as the communicator of
6580     the input matrix.
6581 
6582     If iscol is PETSC_NULL then all columns are obtained (not supported in Fortran).
6583 
6584    Example usage:
6585    Consider the following 8x8 matrix with 34 non-zero values, that is
6586    assembled across 3 processors. Lets assume that proc0 owns 3 rows,
6587    proc1 owns 3 rows, proc2 owns 2 rows. This division can be shown
6588    as follows:
6589 
6590 .vb
6591             1  2  0  |  0  3  0  |  0  4
6592     Proc0   0  5  6  |  7  0  0  |  8  0
6593             9  0 10  | 11  0  0  | 12  0
6594     -------------------------------------
6595            13  0 14  | 15 16 17  |  0  0
6596     Proc1   0 18  0  | 19 20 21  |  0  0
6597             0  0  0  | 22 23  0  | 24  0
6598     -------------------------------------
6599     Proc2  25 26 27  |  0  0 28  | 29  0
6600            30  0  0  | 31 32 33  |  0 34
6601 .ve
6602 
6603     Suppose isrow = [0 1 | 4 | 6 7] and iscol = [1 2 | 3 4 5 | 6].  The resulting submatrix is
6604 
6605 .vb
6606             2  0  |  0  3  0  |  0
6607     Proc0   5  6  |  7  0  0  |  8
6608     -------------------------------
6609     Proc1  18  0  | 19 20 21  |  0
6610     -------------------------------
6611     Proc2  26 27  |  0  0 28  | 29
6612             0  0  | 31 32 33  |  0
6613 .ve
6614 
6615 
6616     Concepts: matrices^submatrices
6617 
6618 .seealso: MatGetSubMatrices()
6619 @*/
6620 PetscErrorCode PETSCMAT_DLLEXPORT MatGetSubMatrix(Mat mat,IS isrow,IS iscol,MatReuse cll,Mat *newmat)
6621 {
6622   PetscErrorCode ierr;
6623   PetscMPIInt    size;
6624   Mat            *local;
6625   IS             iscoltmp;
6626 
6627   PetscFunctionBegin;
6628   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6629   PetscValidHeaderSpecific(isrow,IS_CLASSID,2);
6630   if (iscol) PetscValidHeaderSpecific(iscol,IS_CLASSID,3);
6631   PetscValidPointer(newmat,5);
6632   if (cll == MAT_REUSE_MATRIX) PetscValidHeaderSpecific(*newmat,MAT_CLASSID,5);
6633   PetscValidType(mat,1);
6634   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6635   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6636   ierr = MPI_Comm_size(((PetscObject)mat)->comm,&size);CHKERRQ(ierr);
6637 
6638   if (!iscol) {
6639     ierr = ISCreateStride(((PetscObject)mat)->comm,mat->cmap->n,mat->cmap->rstart,1,&iscoltmp);CHKERRQ(ierr);
6640   } else {
6641     iscoltmp = iscol;
6642   }
6643 
6644   /* if original matrix is on just one processor then use submatrix generated */
6645   if (mat->ops->getsubmatrices && !mat->ops->getsubmatrix && size == 1 && cll == MAT_REUSE_MATRIX) {
6646     ierr = MatGetSubMatrices(mat,1,&isrow,&iscoltmp,MAT_REUSE_MATRIX,&newmat);CHKERRQ(ierr);
6647     if (!iscol) {ierr = ISDestroy(iscoltmp);CHKERRQ(ierr);}
6648     PetscFunctionReturn(0);
6649   } else if (mat->ops->getsubmatrices && !mat->ops->getsubmatrix && size == 1) {
6650     ierr    = MatGetSubMatrices(mat,1,&isrow,&iscoltmp,MAT_INITIAL_MATRIX,&local);CHKERRQ(ierr);
6651     *newmat = *local;
6652     ierr    = PetscFree(local);CHKERRQ(ierr);
6653     if (!iscol) {ierr = ISDestroy(iscoltmp);CHKERRQ(ierr);}
6654     PetscFunctionReturn(0);
6655   } else if (!mat->ops->getsubmatrix) {
6656     /* Create a new matrix type that implements the operation using the full matrix */
6657     switch (cll) {
6658       case MAT_INITIAL_MATRIX:
6659         ierr = MatCreateSubMatrix(mat,isrow,iscoltmp,newmat);CHKERRQ(ierr);
6660         break;
6661       case MAT_REUSE_MATRIX:
6662         ierr = MatSubMatrixUpdate(*newmat,mat,isrow,iscoltmp);CHKERRQ(ierr);
6663         break;
6664       default: SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"Invalid MatReuse, must be either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX");
6665     }
6666     if (!iscol) {ierr = ISDestroy(iscoltmp);CHKERRQ(ierr);}
6667     PetscFunctionReturn(0);
6668   }
6669 
6670   if (!mat->ops->getsubmatrix) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6671   ierr = (*mat->ops->getsubmatrix)(mat,isrow,iscoltmp,cll,newmat);CHKERRQ(ierr);
6672   if (!iscol) {ierr = ISDestroy(iscoltmp);CHKERRQ(ierr);}
6673   ierr = PetscObjectStateIncrease((PetscObject)*newmat);CHKERRQ(ierr);
6674   PetscFunctionReturn(0);
6675 }
6676 
6677 #undef __FUNCT__
6678 #define __FUNCT__ "MatStashSetInitialSize"
6679 /*@
6680    MatStashSetInitialSize - sets the sizes of the matrix stash, that is
6681    used during the assembly process to store values that belong to
6682    other processors.
6683 
6684    Not Collective
6685 
6686    Input Parameters:
6687 +  mat   - the matrix
6688 .  size  - the initial size of the stash.
6689 -  bsize - the initial size of the block-stash(if used).
6690 
6691    Options Database Keys:
6692 +   -matstash_initial_size <size> or <size0,size1,...sizep-1>
6693 -   -matstash_block_initial_size <bsize>  or <bsize0,bsize1,...bsizep-1>
6694 
6695    Level: intermediate
6696 
6697    Notes:
6698      The block-stash is used for values set with MatSetValuesBlocked() while
6699      the stash is used for values set with MatSetValues()
6700 
6701      Run with the option -info and look for output of the form
6702      MatAssemblyBegin_MPIXXX:Stash has MM entries, uses nn mallocs.
6703      to determine the appropriate value, MM, to use for size and
6704      MatAssemblyBegin_MPIXXX:Block-Stash has BMM entries, uses nn mallocs.
6705      to determine the value, BMM to use for bsize
6706 
6707    Concepts: stash^setting matrix size
6708    Concepts: matrices^stash
6709 
6710 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashGetInfo()
6711 
6712 @*/
6713 PetscErrorCode PETSCMAT_DLLEXPORT MatStashSetInitialSize(Mat mat,PetscInt size, PetscInt bsize)
6714 {
6715   PetscErrorCode ierr;
6716 
6717   PetscFunctionBegin;
6718   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6719   PetscValidType(mat,1);
6720   ierr = MatStashSetInitialSize_Private(&mat->stash,size);CHKERRQ(ierr);
6721   ierr = MatStashSetInitialSize_Private(&mat->bstash,bsize);CHKERRQ(ierr);
6722   PetscFunctionReturn(0);
6723 }
6724 
6725 #undef __FUNCT__
6726 #define __FUNCT__ "MatInterpolateAdd"
6727 /*@
6728    MatInterpolateAdd - w = y + A*x or A'*x depending on the shape of
6729      the matrix
6730 
6731    Neighbor-wise Collective on Mat
6732 
6733    Input Parameters:
6734 +  mat   - the matrix
6735 .  x,y - the vectors
6736 -  w - where the result is stored
6737 
6738    Level: intermediate
6739 
6740    Notes:
6741     w may be the same vector as y.
6742 
6743     This allows one to use either the restriction or interpolation (its transpose)
6744     matrix to do the interpolation
6745 
6746     Concepts: interpolation
6747 
6748 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict()
6749 
6750 @*/
6751 PetscErrorCode PETSCMAT_DLLEXPORT MatInterpolateAdd(Mat A,Vec x,Vec y,Vec w)
6752 {
6753   PetscErrorCode ierr;
6754   PetscInt       M,N;
6755 
6756   PetscFunctionBegin;
6757   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
6758   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
6759   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
6760   PetscValidHeaderSpecific(w,VEC_CLASSID,4);
6761   PetscValidType(A,1);
6762   ierr = MatPreallocated(A);CHKERRQ(ierr);
6763   ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr);
6764   if (N > M) {
6765     ierr = MatMultTransposeAdd(A,x,y,w);CHKERRQ(ierr);
6766   } else {
6767     ierr = MatMultAdd(A,x,y,w);CHKERRQ(ierr);
6768   }
6769   PetscFunctionReturn(0);
6770 }
6771 
6772 #undef __FUNCT__
6773 #define __FUNCT__ "MatInterpolate"
6774 /*@
6775    MatInterpolate - y = A*x or A'*x depending on the shape of
6776      the matrix
6777 
6778    Neighbor-wise Collective on Mat
6779 
6780    Input Parameters:
6781 +  mat   - the matrix
6782 -  x,y - the vectors
6783 
6784    Level: intermediate
6785 
6786    Notes:
6787     This allows one to use either the restriction or interpolation (its transpose)
6788     matrix to do the interpolation
6789 
6790    Concepts: matrices^interpolation
6791 
6792 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatRestrict()
6793 
6794 @*/
6795 PetscErrorCode PETSCMAT_DLLEXPORT MatInterpolate(Mat A,Vec x,Vec y)
6796 {
6797   PetscErrorCode ierr;
6798   PetscInt       M,N;
6799 
6800   PetscFunctionBegin;
6801   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
6802   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
6803   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
6804   PetscValidType(A,1);
6805   ierr = MatPreallocated(A);CHKERRQ(ierr);
6806   ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr);
6807   if (N > M) {
6808     ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr);
6809   } else {
6810     ierr = MatMult(A,x,y);CHKERRQ(ierr);
6811   }
6812   PetscFunctionReturn(0);
6813 }
6814 
6815 #undef __FUNCT__
6816 #define __FUNCT__ "MatRestrict"
6817 /*@
6818    MatRestrict - y = A*x or A'*x
6819 
6820    Neighbor-wise Collective on Mat
6821 
6822    Input Parameters:
6823 +  mat   - the matrix
6824 -  x,y - the vectors
6825 
6826    Level: intermediate
6827 
6828    Notes:
6829     This allows one to use either the restriction or interpolation (its transpose)
6830     matrix to do the restriction
6831 
6832    Concepts: matrices^restriction
6833 
6834 .seealso: MatMultAdd(), MatMultTransposeAdd(), MatInterpolate()
6835 
6836 @*/
6837 PetscErrorCode PETSCMAT_DLLEXPORT MatRestrict(Mat A,Vec x,Vec y)
6838 {
6839   PetscErrorCode ierr;
6840   PetscInt       M,N;
6841 
6842   PetscFunctionBegin;
6843   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
6844   PetscValidHeaderSpecific(x,VEC_CLASSID,2);
6845   PetscValidHeaderSpecific(y,VEC_CLASSID,3);
6846   PetscValidType(A,1);
6847   ierr = MatPreallocated(A);CHKERRQ(ierr);
6848 
6849   ierr = MatGetSize(A,&M,&N);CHKERRQ(ierr);
6850   if (N > M) {
6851     ierr = MatMult(A,x,y);CHKERRQ(ierr);
6852   } else {
6853     ierr = MatMultTranspose(A,x,y);CHKERRQ(ierr);
6854   }
6855   PetscFunctionReturn(0);
6856 }
6857 
6858 #undef __FUNCT__
6859 #define __FUNCT__ "MatNullSpaceAttach"
6860 /*@
6861    MatNullSpaceAttach - attaches a null space to a matrix.
6862         This null space will be removed from the resulting vector whenever
6863         MatMult() is called
6864 
6865    Logically Collective on Mat and MatNullSpace
6866 
6867    Input Parameters:
6868 +  mat - the matrix
6869 -  nullsp - the null space object
6870 
6871    Level: developer
6872 
6873    Notes:
6874       Overwrites any previous null space that may have been attached
6875 
6876    Concepts: null space^attaching to matrix
6877 
6878 .seealso: MatCreate(), MatNullSpaceCreate()
6879 @*/
6880 PetscErrorCode PETSCMAT_DLLEXPORT MatNullSpaceAttach(Mat mat,MatNullSpace nullsp)
6881 {
6882   PetscErrorCode ierr;
6883 
6884   PetscFunctionBegin;
6885   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6886   PetscValidType(mat,1);
6887   PetscValidHeaderSpecific(nullsp,MAT_NULLSPACE_CLASSID,2);
6888   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6889   ierr = PetscObjectReference((PetscObject)nullsp);CHKERRQ(ierr);
6890   if (mat->nullsp) { ierr = MatNullSpaceDestroy(mat->nullsp);CHKERRQ(ierr); }
6891   mat->nullsp = nullsp;
6892   PetscFunctionReturn(0);
6893 }
6894 
6895 #undef __FUNCT__
6896 #define __FUNCT__ "MatICCFactor"
6897 /*@C
6898    MatICCFactor - Performs in-place incomplete Cholesky factorization of matrix.
6899 
6900    Collective on Mat
6901 
6902    Input Parameters:
6903 +  mat - the matrix
6904 .  row - row/column permutation
6905 .  fill - expected fill factor >= 1.0
6906 -  level - level of fill, for ICC(k)
6907 
6908    Notes:
6909    Probably really in-place only when level of fill is zero, otherwise allocates
6910    new space to store factored matrix and deletes previous memory.
6911 
6912    Most users should employ the simplified KSP interface for linear solvers
6913    instead of working directly with matrix algebra routines such as this.
6914    See, e.g., KSPCreate().
6915 
6916    Level: developer
6917 
6918    Concepts: matrices^incomplete Cholesky factorization
6919    Concepts: Cholesky factorization
6920 
6921 .seealso: MatICCFactorSymbolic(), MatLUFactorNumeric(), MatCholeskyFactor()
6922 
6923     Developer Note: fortran interface is not autogenerated as the f90
6924     interface defintion cannot be generated correctly [due to MatFactorInfo]
6925 
6926 @*/
6927 PetscErrorCode PETSCMAT_DLLEXPORT MatICCFactor(Mat mat,IS row,const MatFactorInfo* info)
6928 {
6929   PetscErrorCode ierr;
6930 
6931   PetscFunctionBegin;
6932   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6933   PetscValidType(mat,1);
6934   if (row) PetscValidHeaderSpecific(row,IS_CLASSID,2);
6935   PetscValidPointer(info,3);
6936   if (mat->rmap->N != mat->cmap->N) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONG,"matrix must be square");
6937   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
6938   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
6939   if (!mat->ops->iccfactor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6940   ierr = MatPreallocated(mat);CHKERRQ(ierr);
6941   ierr = (*mat->ops->iccfactor)(mat,row,info);CHKERRQ(ierr);
6942   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
6943   PetscFunctionReturn(0);
6944 }
6945 
6946 #undef __FUNCT__
6947 #define __FUNCT__ "MatSetValuesAdic"
6948 /*@
6949    MatSetValuesAdic - Sets values computed with ADIC automatic differentiation into a matrix.
6950 
6951    Not Collective
6952 
6953    Input Parameters:
6954 +  mat - the matrix
6955 -  v - the values compute with ADIC
6956 
6957    Level: developer
6958 
6959    Notes:
6960      Must call MatSetColoring() before using this routine. Also this matrix must already
6961      have its nonzero pattern determined.
6962 
6963 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
6964           MatSetValues(), MatSetColoring(), MatSetValuesAdifor()
6965 @*/
6966 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesAdic(Mat mat,void *v)
6967 {
6968   PetscErrorCode ierr;
6969 
6970   PetscFunctionBegin;
6971   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
6972   PetscValidType(mat,1);
6973   PetscValidPointer(mat,2);
6974 
6975   if (!mat->assembled) {
6976     SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled");
6977   }
6978   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
6979   if (!mat->ops->setvaluesadic) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
6980   ierr = (*mat->ops->setvaluesadic)(mat,v);CHKERRQ(ierr);
6981   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
6982   ierr = MatView_Private(mat);CHKERRQ(ierr);
6983   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
6984   PetscFunctionReturn(0);
6985 }
6986 
6987 
6988 #undef __FUNCT__
6989 #define __FUNCT__ "MatSetColoring"
6990 /*@
6991    MatSetColoring - Sets a coloring used by calls to MatSetValuesAdic()
6992 
6993    Not Collective
6994 
6995    Input Parameters:
6996 +  mat - the matrix
6997 -  coloring - the coloring
6998 
6999    Level: developer
7000 
7001 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
7002           MatSetValues(), MatSetValuesAdic()
7003 @*/
7004 PetscErrorCode PETSCMAT_DLLEXPORT MatSetColoring(Mat mat,ISColoring coloring)
7005 {
7006   PetscErrorCode ierr;
7007 
7008   PetscFunctionBegin;
7009   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7010   PetscValidType(mat,1);
7011   PetscValidPointer(coloring,2);
7012 
7013   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled");
7014   if (!mat->ops->setcoloring) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7015   ierr = (*mat->ops->setcoloring)(mat,coloring);CHKERRQ(ierr);
7016   PetscFunctionReturn(0);
7017 }
7018 
7019 #undef __FUNCT__
7020 #define __FUNCT__ "MatSetValuesAdifor"
7021 /*@
7022    MatSetValuesAdifor - Sets values computed with automatic differentiation into a matrix.
7023 
7024    Not Collective
7025 
7026    Input Parameters:
7027 +  mat - the matrix
7028 .  nl - leading dimension of v
7029 -  v - the values compute with ADIFOR
7030 
7031    Level: developer
7032 
7033    Notes:
7034      Must call MatSetColoring() before using this routine. Also this matrix must already
7035      have its nonzero pattern determined.
7036 
7037 .seealso: MatSetOption(), MatAssemblyBegin(), MatAssemblyEnd(), MatSetValuesBlocked(), MatSetValuesLocal(),
7038           MatSetValues(), MatSetColoring()
7039 @*/
7040 PetscErrorCode PETSCMAT_DLLEXPORT MatSetValuesAdifor(Mat mat,PetscInt nl,void *v)
7041 {
7042   PetscErrorCode ierr;
7043 
7044   PetscFunctionBegin;
7045   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7046   PetscValidType(mat,1);
7047   PetscValidPointer(v,3);
7048 
7049   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled");
7050   ierr = PetscLogEventBegin(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
7051   if (!mat->ops->setvaluesadifor) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7052   ierr = (*mat->ops->setvaluesadifor)(mat,nl,v);CHKERRQ(ierr);
7053   ierr = PetscLogEventEnd(MAT_SetValues,mat,0,0,0);CHKERRQ(ierr);
7054   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
7055   PetscFunctionReturn(0);
7056 }
7057 
7058 #undef __FUNCT__
7059 #define __FUNCT__ "MatDiagonalScaleLocal"
7060 /*@
7061    MatDiagonalScaleLocal - Scales columns of a matrix given the scaling values including the
7062          ghosted ones.
7063 
7064    Not Collective
7065 
7066    Input Parameters:
7067 +  mat - the matrix
7068 -  diag = the diagonal values, including ghost ones
7069 
7070    Level: developer
7071 
7072    Notes: Works only for MPIAIJ and MPIBAIJ matrices
7073 
7074 .seealso: MatDiagonalScale()
7075 @*/
7076 PetscErrorCode PETSCMAT_DLLEXPORT MatDiagonalScaleLocal(Mat mat,Vec diag)
7077 {
7078   PetscErrorCode ierr;
7079   PetscMPIInt    size;
7080 
7081   PetscFunctionBegin;
7082   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7083   PetscValidHeaderSpecific(diag,VEC_CLASSID,2);
7084   PetscValidType(mat,1);
7085 
7086   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Matrix must be already assembled");
7087   ierr = PetscLogEventBegin(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
7088   ierr = MPI_Comm_size(((PetscObject)mat)->comm,&size);CHKERRQ(ierr);
7089   if (size == 1) {
7090     PetscInt n,m;
7091     ierr = VecGetSize(diag,&n);CHKERRQ(ierr);
7092     ierr = MatGetSize(mat,0,&m);CHKERRQ(ierr);
7093     if (m == n) {
7094       ierr = MatDiagonalScale(mat,0,diag);CHKERRQ(ierr);
7095     } else {
7096       SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only supported for sequential matrices when no ghost points/periodic conditions");
7097     }
7098   } else {
7099     PetscErrorCode (*f)(Mat,Vec);
7100     ierr = PetscObjectQueryFunction((PetscObject)mat,"MatDiagonalScaleLocal_C",(void (**)(void))&f);CHKERRQ(ierr);
7101     if (f) {
7102       ierr = (*f)(mat,diag);CHKERRQ(ierr);
7103     } else {
7104       SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only supported for MPIAIJ and MPIBAIJ parallel matrices");
7105     }
7106   }
7107   ierr = PetscLogEventEnd(MAT_Scale,mat,0,0,0);CHKERRQ(ierr);
7108   ierr = PetscObjectStateIncrease((PetscObject)mat);CHKERRQ(ierr);
7109   PetscFunctionReturn(0);
7110 }
7111 
7112 #undef __FUNCT__
7113 #define __FUNCT__ "MatGetInertia"
7114 /*@
7115    MatGetInertia - Gets the inertia from a factored matrix
7116 
7117    Collective on Mat
7118 
7119    Input Parameter:
7120 .  mat - the matrix
7121 
7122    Output Parameters:
7123 +   nneg - number of negative eigenvalues
7124 .   nzero - number of zero eigenvalues
7125 -   npos - number of positive eigenvalues
7126 
7127    Level: advanced
7128 
7129    Notes: Matrix must have been factored by MatCholeskyFactor()
7130 
7131 
7132 @*/
7133 PetscErrorCode PETSCMAT_DLLEXPORT MatGetInertia(Mat mat,PetscInt *nneg,PetscInt *nzero,PetscInt *npos)
7134 {
7135   PetscErrorCode ierr;
7136 
7137   PetscFunctionBegin;
7138   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7139   PetscValidType(mat,1);
7140   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
7141   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Numeric factor mat is not assembled");
7142   if (!mat->ops->getinertia) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7143   ierr = (*mat->ops->getinertia)(mat,nneg,nzero,npos);CHKERRQ(ierr);
7144   PetscFunctionReturn(0);
7145 }
7146 
7147 /* ----------------------------------------------------------------*/
7148 #undef __FUNCT__
7149 #define __FUNCT__ "MatSolves"
7150 /*@C
7151    MatSolves - Solves A x = b, given a factored matrix, for a collection of vectors
7152 
7153    Neighbor-wise Collective on Mat and Vecs
7154 
7155    Input Parameters:
7156 +  mat - the factored matrix
7157 -  b - the right-hand-side vectors
7158 
7159    Output Parameter:
7160 .  x - the result vectors
7161 
7162    Notes:
7163    The vectors b and x cannot be the same.  I.e., one cannot
7164    call MatSolves(A,x,x).
7165 
7166    Notes:
7167    Most users should employ the simplified KSP interface for linear solvers
7168    instead of working directly with matrix algebra routines such as this.
7169    See, e.g., KSPCreate().
7170 
7171    Level: developer
7172 
7173    Concepts: matrices^triangular solves
7174 
7175 .seealso: MatSolveAdd(), MatSolveTranspose(), MatSolveTransposeAdd(), MatSolve()
7176 @*/
7177 PetscErrorCode PETSCMAT_DLLEXPORT MatSolves(Mat mat,Vecs b,Vecs x)
7178 {
7179   PetscErrorCode ierr;
7180 
7181   PetscFunctionBegin;
7182   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7183   PetscValidType(mat,1);
7184   if (x == b) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_IDN,"x and b must be different vectors");
7185   if (!mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Unfactored matrix");
7186   if (!mat->rmap->N && !mat->cmap->N) PetscFunctionReturn(0);
7187 
7188   if (!mat->ops->solves) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
7189   ierr = MatPreallocated(mat);CHKERRQ(ierr);
7190   ierr = PetscLogEventBegin(MAT_Solves,mat,0,0,0);CHKERRQ(ierr);
7191   ierr = (*mat->ops->solves)(mat,b,x);CHKERRQ(ierr);
7192   ierr = PetscLogEventEnd(MAT_Solves,mat,0,0,0);CHKERRQ(ierr);
7193   PetscFunctionReturn(0);
7194 }
7195 
7196 #undef __FUNCT__
7197 #define __FUNCT__ "MatIsSymmetric"
7198 /*@
7199    MatIsSymmetric - Test whether a matrix is symmetric
7200 
7201    Collective on Mat
7202 
7203    Input Parameter:
7204 +  A - the matrix to test
7205 -  tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact transpose)
7206 
7207    Output Parameters:
7208 .  flg - the result
7209 
7210    Level: intermediate
7211 
7212    Concepts: matrix^symmetry
7213 
7214 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetricKnown()
7215 @*/
7216 PetscErrorCode PETSCMAT_DLLEXPORT MatIsSymmetric(Mat A,PetscReal tol,PetscTruth *flg)
7217 {
7218   PetscErrorCode ierr;
7219 
7220   PetscFunctionBegin;
7221   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7222   PetscValidPointer(flg,2);
7223 
7224   if (!A->symmetric_set) {
7225     if (!A->ops->issymmetric) {
7226       const MatType mattype;
7227       ierr = MatGetType(A,&mattype);CHKERRQ(ierr);
7228       SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for symmetric",mattype);
7229     }
7230     ierr = (*A->ops->issymmetric)(A,tol,flg);CHKERRQ(ierr);
7231     if (!tol) {
7232       A->symmetric_set = PETSC_TRUE;
7233       A->symmetric = *flg;
7234       if (A->symmetric) {
7235 	A->structurally_symmetric_set = PETSC_TRUE;
7236 	A->structurally_symmetric     = PETSC_TRUE;
7237       }
7238     }
7239   } else if (A->symmetric) {
7240     *flg = PETSC_TRUE;
7241   } else if (!tol) {
7242     *flg = PETSC_FALSE;
7243   } else {
7244     if (!A->ops->issymmetric) {
7245       const MatType mattype;
7246       ierr = MatGetType(A,&mattype);CHKERRQ(ierr);
7247       SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for symmetric",mattype);
7248     }
7249     ierr = (*A->ops->issymmetric)(A,tol,flg);CHKERRQ(ierr);
7250   }
7251   PetscFunctionReturn(0);
7252 }
7253 
7254 #undef __FUNCT__
7255 #define __FUNCT__ "MatIsHermitian"
7256 /*@
7257    MatIsHermitian - Test whether a matrix is Hermitian
7258 
7259    Collective on Mat
7260 
7261    Input Parameter:
7262 +  A - the matrix to test
7263 -  tol - difference between value and its transpose less than this amount counts as equal (use 0.0 for exact Hermitian)
7264 
7265    Output Parameters:
7266 .  flg - the result
7267 
7268    Level: intermediate
7269 
7270    Concepts: matrix^symmetry
7271 
7272 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetricKnown()
7273 @*/
7274 PetscErrorCode PETSCMAT_DLLEXPORT MatIsHermitian(Mat A,PetscReal tol,PetscTruth *flg)
7275 {
7276   PetscErrorCode ierr;
7277 
7278   PetscFunctionBegin;
7279   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7280   PetscValidPointer(flg,2);
7281 
7282   if (!A->hermitian_set) {
7283     if (!A->ops->ishermitian) {
7284       const MatType mattype;
7285       ierr = MatGetType(A,&mattype);CHKERRQ(ierr);
7286       SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for hermitian",mattype);
7287     }
7288     ierr = (*A->ops->ishermitian)(A,tol,flg);CHKERRQ(ierr);
7289     if (!tol) {
7290       A->hermitian_set = PETSC_TRUE;
7291       A->hermitian = *flg;
7292       if (A->hermitian) {
7293 	A->structurally_symmetric_set = PETSC_TRUE;
7294 	A->structurally_symmetric     = PETSC_TRUE;
7295       }
7296     }
7297   } else if (A->hermitian) {
7298     *flg = PETSC_TRUE;
7299   } else if (!tol) {
7300     *flg = PETSC_FALSE;
7301   } else {
7302     if (!A->ops->ishermitian) {
7303       const MatType mattype;
7304       ierr = MatGetType(A,&mattype);CHKERRQ(ierr);
7305       SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix of type <%s> does not support checking for hermitian",mattype);
7306     }
7307     ierr = (*A->ops->ishermitian)(A,tol,flg);CHKERRQ(ierr);
7308   }
7309   PetscFunctionReturn(0);
7310 }
7311 
7312 #undef __FUNCT__
7313 #define __FUNCT__ "MatIsSymmetricKnown"
7314 /*@
7315    MatIsSymmetricKnown - Checks the flag on the matrix to see if it is symmetric.
7316 
7317    Not Collective
7318 
7319    Input Parameter:
7320 .  A - the matrix to check
7321 
7322    Output Parameters:
7323 +  set - if the symmetric flag is set (this tells you if the next flag is valid)
7324 -  flg - the result
7325 
7326    Level: advanced
7327 
7328    Concepts: matrix^symmetry
7329 
7330    Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsSymmetric()
7331          if you want it explicitly checked
7332 
7333 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric()
7334 @*/
7335 PetscErrorCode PETSCMAT_DLLEXPORT MatIsSymmetricKnown(Mat A,PetscTruth *set,PetscTruth *flg)
7336 {
7337   PetscFunctionBegin;
7338   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7339   PetscValidPointer(set,2);
7340   PetscValidPointer(flg,3);
7341   if (A->symmetric_set) {
7342     *set = PETSC_TRUE;
7343     *flg = A->symmetric;
7344   } else {
7345     *set = PETSC_FALSE;
7346   }
7347   PetscFunctionReturn(0);
7348 }
7349 
7350 #undef __FUNCT__
7351 #define __FUNCT__ "MatIsHermitianKnown"
7352 /*@
7353    MatIsHermitianKnown - Checks the flag on the matrix to see if it is hermitian.
7354 
7355    Not Collective
7356 
7357    Input Parameter:
7358 .  A - the matrix to check
7359 
7360    Output Parameters:
7361 +  set - if the hermitian flag is set (this tells you if the next flag is valid)
7362 -  flg - the result
7363 
7364    Level: advanced
7365 
7366    Concepts: matrix^symmetry
7367 
7368    Note: Does not check the matrix values directly, so this may return unknown (set = PETSC_FALSE). Use MatIsHermitian()
7369          if you want it explicitly checked
7370 
7371 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsStructurallySymmetric(), MatSetOption(), MatIsSymmetric()
7372 @*/
7373 PetscErrorCode PETSCMAT_DLLEXPORT MatIsHermitianKnown(Mat A,PetscTruth *set,PetscTruth *flg)
7374 {
7375   PetscFunctionBegin;
7376   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7377   PetscValidPointer(set,2);
7378   PetscValidPointer(flg,3);
7379   if (A->hermitian_set) {
7380     *set = PETSC_TRUE;
7381     *flg = A->hermitian;
7382   } else {
7383     *set = PETSC_FALSE;
7384   }
7385   PetscFunctionReturn(0);
7386 }
7387 
7388 #undef __FUNCT__
7389 #define __FUNCT__ "MatIsStructurallySymmetric"
7390 /*@
7391    MatIsStructurallySymmetric - Test whether a matrix is structurally symmetric
7392 
7393    Collective on Mat
7394 
7395    Input Parameter:
7396 .  A - the matrix to test
7397 
7398    Output Parameters:
7399 .  flg - the result
7400 
7401    Level: intermediate
7402 
7403    Concepts: matrix^symmetry
7404 
7405 .seealso: MatTranspose(), MatIsTranspose(), MatIsHermitian(), MatIsSymmetric(), MatSetOption()
7406 @*/
7407 PetscErrorCode PETSCMAT_DLLEXPORT MatIsStructurallySymmetric(Mat A,PetscTruth *flg)
7408 {
7409   PetscErrorCode ierr;
7410 
7411   PetscFunctionBegin;
7412   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7413   PetscValidPointer(flg,2);
7414   if (!A->structurally_symmetric_set) {
7415     if (!A->ops->isstructurallysymmetric) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_SUP,"Matrix does not support checking for structural symmetric");
7416     ierr = (*A->ops->isstructurallysymmetric)(A,&A->structurally_symmetric);CHKERRQ(ierr);
7417     A->structurally_symmetric_set = PETSC_TRUE;
7418   }
7419   *flg = A->structurally_symmetric;
7420   PetscFunctionReturn(0);
7421 }
7422 
7423 #undef __FUNCT__
7424 #define __FUNCT__ "MatStashGetInfo"
7425 extern PetscErrorCode MatStashGetInfo_Private(MatStash*,PetscInt*,PetscInt*);
7426 /*@
7427    MatStashGetInfo - Gets how many values are currently in the vector stash, i.e. need
7428        to be communicated to other processors during the MatAssemblyBegin/End() process
7429 
7430     Not collective
7431 
7432    Input Parameter:
7433 .   vec - the vector
7434 
7435    Output Parameters:
7436 +   nstash   - the size of the stash
7437 .   reallocs - the number of additional mallocs incurred.
7438 .   bnstash   - the size of the block stash
7439 -   breallocs - the number of additional mallocs incurred.in the block stash
7440 
7441    Level: advanced
7442 
7443 .seealso: MatAssemblyBegin(), MatAssemblyEnd(), Mat, MatStashSetInitialSize()
7444 
7445 @*/
7446 PetscErrorCode PETSCMAT_DLLEXPORT MatStashGetInfo(Mat mat,PetscInt *nstash,PetscInt *reallocs,PetscInt *bnstash,PetscInt *breallocs)
7447 {
7448   PetscErrorCode ierr;
7449   PetscFunctionBegin;
7450   ierr = MatStashGetInfo_Private(&mat->stash,nstash,reallocs);CHKERRQ(ierr);
7451   ierr = MatStashGetInfo_Private(&mat->bstash,bnstash,breallocs);CHKERRQ(ierr);
7452   PetscFunctionReturn(0);
7453 }
7454 
7455 #undef __FUNCT__
7456 #define __FUNCT__ "MatGetVecs"
7457 /*@C
7458    MatGetVecs - Get vector(s) compatible with the matrix, i.e. with the same
7459      parallel layout
7460 
7461    Collective on Mat
7462 
7463    Input Parameter:
7464 .  mat - the matrix
7465 
7466    Output Parameter:
7467 +   right - (optional) vector that the matrix can be multiplied against
7468 -   left - (optional) vector that the matrix vector product can be stored in
7469 
7470   Level: advanced
7471 
7472 .seealso: MatCreate()
7473 @*/
7474 PetscErrorCode PETSCMAT_DLLEXPORT MatGetVecs(Mat mat,Vec *right,Vec *left)
7475 {
7476   PetscErrorCode ierr;
7477 
7478   PetscFunctionBegin;
7479   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
7480   PetscValidType(mat,1);
7481   ierr = MatPreallocated(mat);CHKERRQ(ierr);
7482   if (mat->ops->getvecs) {
7483     ierr = (*mat->ops->getvecs)(mat,right,left);CHKERRQ(ierr);
7484   } else {
7485     PetscMPIInt size;
7486     ierr = MPI_Comm_size(((PetscObject)mat)->comm, &size);CHKERRQ(ierr);
7487     if (right) {
7488       ierr = VecCreate(((PetscObject)mat)->comm,right);CHKERRQ(ierr);
7489       ierr = VecSetSizes(*right,mat->cmap->n,PETSC_DETERMINE);CHKERRQ(ierr);
7490       ierr = VecSetBlockSize(*right,mat->rmap->bs);CHKERRQ(ierr);
7491       if (size > 1) {
7492         /* New vectors uses Mat cmap and does not create a new one */
7493 	ierr = PetscLayoutDestroy((*right)->map);CHKERRQ(ierr);
7494 	(*right)->map = mat->cmap;
7495 	mat->cmap->refcnt++;
7496 
7497         ierr = VecSetType(*right,VECMPI);CHKERRQ(ierr);
7498       } else {ierr = VecSetType(*right,VECSEQ);CHKERRQ(ierr);}
7499     }
7500     if (left) {
7501       ierr = VecCreate(((PetscObject)mat)->comm,left);CHKERRQ(ierr);
7502       ierr = VecSetSizes(*left,mat->rmap->n,PETSC_DETERMINE);CHKERRQ(ierr);
7503       ierr = VecSetBlockSize(*left,mat->rmap->bs);CHKERRQ(ierr);
7504       if (size > 1) {
7505         /* New vectors uses Mat rmap and does not create a new one */
7506 	ierr = PetscLayoutDestroy((*left)->map);CHKERRQ(ierr);
7507 	(*left)->map = mat->rmap;
7508 	mat->rmap->refcnt++;
7509 
7510         ierr = VecSetType(*left,VECMPI);CHKERRQ(ierr);
7511       } else {ierr = VecSetType(*left,VECSEQ);CHKERRQ(ierr);}
7512     }
7513   }
7514   if (mat->mapping) {
7515     if (right) {ierr = VecSetLocalToGlobalMapping(*right,mat->mapping);CHKERRQ(ierr);}
7516     if (left) {ierr = VecSetLocalToGlobalMapping(*left,mat->mapping);CHKERRQ(ierr);}
7517   }
7518   if (mat->bmapping) {
7519     if (right) {ierr = VecSetLocalToGlobalMappingBlock(*right,mat->bmapping);CHKERRQ(ierr);}
7520     if (left) {ierr = VecSetLocalToGlobalMappingBlock(*left,mat->bmapping);CHKERRQ(ierr);}
7521   }
7522   PetscFunctionReturn(0);
7523 }
7524 
7525 #undef __FUNCT__
7526 #define __FUNCT__ "MatFactorInfoInitialize"
7527 /*@C
7528    MatFactorInfoInitialize - Initializes a MatFactorInfo data structure
7529      with default values.
7530 
7531    Not Collective
7532 
7533    Input Parameters:
7534 .    info - the MatFactorInfo data structure
7535 
7536 
7537    Notes: The solvers are generally used through the KSP and PC objects, for example
7538           PCLU, PCILU, PCCHOLESKY, PCICC
7539 
7540    Level: developer
7541 
7542 .seealso: MatFactorInfo
7543 
7544     Developer Note: fortran interface is not autogenerated as the f90
7545     interface defintion cannot be generated correctly [due to MatFactorInfo]
7546 
7547 @*/
7548 
7549 PetscErrorCode PETSCMAT_DLLEXPORT MatFactorInfoInitialize(MatFactorInfo *info)
7550 {
7551   PetscErrorCode ierr;
7552 
7553   PetscFunctionBegin;
7554   ierr = PetscMemzero(info,sizeof(MatFactorInfo));CHKERRQ(ierr);
7555   PetscFunctionReturn(0);
7556 }
7557 
7558 #undef __FUNCT__
7559 #define __FUNCT__ "MatPtAP"
7560 /*@
7561    MatPtAP - Creates the matrix product C = P^T * A * P
7562 
7563    Neighbor-wise Collective on Mat
7564 
7565    Input Parameters:
7566 +  A - the matrix
7567 .  P - the projection matrix
7568 .  scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
7569 -  fill - expected fill as ratio of nnz(C)/nnz(A)
7570 
7571    Output Parameters:
7572 .  C - the product matrix
7573 
7574    Notes:
7575    C will be created and must be destroyed by the user with MatDestroy().
7576 
7577    This routine is currently only implemented for pairs of AIJ matrices and classes
7578    which inherit from AIJ.
7579 
7580    Level: intermediate
7581 
7582 .seealso: MatPtAPSymbolic(), MatPtAPNumeric(), MatMatMult()
7583 @*/
7584 PetscErrorCode PETSCMAT_DLLEXPORT MatPtAP(Mat A,Mat P,MatReuse scall,PetscReal fill,Mat *C)
7585 {
7586   PetscErrorCode ierr;
7587 
7588   PetscFunctionBegin;
7589   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7590   PetscValidType(A,1);
7591   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7592   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7593   PetscValidHeaderSpecific(P,MAT_CLASSID,2);
7594   PetscValidType(P,2);
7595   ierr = MatPreallocated(P);CHKERRQ(ierr);
7596   if (!P->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7597   if (P->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7598   PetscValidPointer(C,3);
7599   if (P->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N);
7600   if (fill < 1.0) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill);
7601   ierr = MatPreallocated(A);CHKERRQ(ierr);
7602 
7603   if (!A->ops->ptap) {
7604     const MatType mattype;
7605     ierr = MatGetType(A,&mattype);CHKERRQ(ierr);
7606     SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"Matrix of type <%s> does not support PtAP",mattype);
7607   }
7608   ierr = PetscLogEventBegin(MAT_PtAP,A,P,0,0);CHKERRQ(ierr);
7609   ierr = (*A->ops->ptap)(A,P,scall,fill,C);CHKERRQ(ierr);
7610   ierr = PetscLogEventEnd(MAT_PtAP,A,P,0,0);CHKERRQ(ierr);
7611 
7612   PetscFunctionReturn(0);
7613 }
7614 
7615 #undef __FUNCT__
7616 #define __FUNCT__ "MatPtAPNumeric"
7617 /*@
7618    MatPtAPNumeric - Computes the matrix product C = P^T * A * P
7619 
7620    Neighbor-wise Collective on Mat
7621 
7622    Input Parameters:
7623 +  A - the matrix
7624 -  P - the projection matrix
7625 
7626    Output Parameters:
7627 .  C - the product matrix
7628 
7629    Notes:
7630    C must have been created by calling MatPtAPSymbolic and must be destroyed by
7631    the user using MatDeatroy().
7632 
7633    This routine is currently only implemented for pairs of AIJ matrices and classes
7634    which inherit from AIJ.  C will be of type MATAIJ.
7635 
7636    Level: intermediate
7637 
7638 .seealso: MatPtAP(), MatPtAPSymbolic(), MatMatMultNumeric()
7639 @*/
7640 PetscErrorCode PETSCMAT_DLLEXPORT MatPtAPNumeric(Mat A,Mat P,Mat C)
7641 {
7642   PetscErrorCode ierr;
7643 
7644   PetscFunctionBegin;
7645   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7646   PetscValidType(A,1);
7647   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7648   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7649   PetscValidHeaderSpecific(P,MAT_CLASSID,2);
7650   PetscValidType(P,2);
7651   ierr = MatPreallocated(P);CHKERRQ(ierr);
7652   if (!P->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7653   if (P->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7654   PetscValidHeaderSpecific(C,MAT_CLASSID,3);
7655   PetscValidType(C,3);
7656   ierr = MatPreallocated(C);CHKERRQ(ierr);
7657   if (C->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7658   if (P->cmap->N!=C->rmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->cmap->N,C->rmap->N);
7659   if (P->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N);
7660   if (A->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N);
7661   if (P->cmap->N!=C->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->cmap->N,C->cmap->N);
7662   ierr = MatPreallocated(A);CHKERRQ(ierr);
7663 
7664   ierr = PetscLogEventBegin(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr);
7665   ierr = (*A->ops->ptapnumeric)(A,P,C);CHKERRQ(ierr);
7666   ierr = PetscLogEventEnd(MAT_PtAPNumeric,A,P,0,0);CHKERRQ(ierr);
7667   PetscFunctionReturn(0);
7668 }
7669 
7670 #undef __FUNCT__
7671 #define __FUNCT__ "MatPtAPSymbolic"
7672 /*@
7673    MatPtAPSymbolic - Creates the (i,j) structure of the matrix product C = P^T * A * P
7674 
7675    Neighbor-wise Collective on Mat
7676 
7677    Input Parameters:
7678 +  A - the matrix
7679 -  P - the projection matrix
7680 
7681    Output Parameters:
7682 .  C - the (i,j) structure of the product matrix
7683 
7684    Notes:
7685    C will be created and must be destroyed by the user with MatDestroy().
7686 
7687    This routine is currently only implemented for pairs of SeqAIJ matrices and classes
7688    which inherit from SeqAIJ.  C will be of type MATSEQAIJ.  The product is computed using
7689    this (i,j) structure by calling MatPtAPNumeric().
7690 
7691    Level: intermediate
7692 
7693 .seealso: MatPtAP(), MatPtAPNumeric(), MatMatMultSymbolic()
7694 @*/
7695 PetscErrorCode PETSCMAT_DLLEXPORT MatPtAPSymbolic(Mat A,Mat P,PetscReal fill,Mat *C)
7696 {
7697   PetscErrorCode ierr;
7698 
7699   PetscFunctionBegin;
7700   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7701   PetscValidType(A,1);
7702   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7703   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7704   if (fill <1.0) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill);
7705   PetscValidHeaderSpecific(P,MAT_CLASSID,2);
7706   PetscValidType(P,2);
7707   ierr = MatPreallocated(P);CHKERRQ(ierr);
7708   if (!P->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7709   if (P->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7710   PetscValidPointer(C,3);
7711 
7712   if (P->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",P->rmap->N,A->cmap->N);
7713   if (A->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix 'A' must be square, %D != %D",A->rmap->N,A->cmap->N);
7714   ierr = MatPreallocated(A);CHKERRQ(ierr);
7715   ierr = PetscLogEventBegin(MAT_PtAPSymbolic,A,P,0,0);CHKERRQ(ierr);
7716   ierr = (*A->ops->ptapsymbolic)(A,P,fill,C);CHKERRQ(ierr);
7717   ierr = PetscLogEventEnd(MAT_PtAPSymbolic,A,P,0,0);CHKERRQ(ierr);
7718 
7719   ierr = MatSetBlockSize(*C,A->rmap->bs);CHKERRQ(ierr);
7720 
7721   PetscFunctionReturn(0);
7722 }
7723 
7724 #undef __FUNCT__
7725 #define __FUNCT__ "MatMatMult"
7726 /*@
7727    MatMatMult - Performs Matrix-Matrix Multiplication C=A*B.
7728 
7729    Neighbor-wise Collective on Mat
7730 
7731    Input Parameters:
7732 +  A - the left matrix
7733 .  B - the right matrix
7734 .  scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
7735 -  fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if you do not have a good estimate
7736           if the result is a dense matrix this is irrelevent
7737 
7738    Output Parameters:
7739 .  C - the product matrix
7740 
7741    Notes:
7742    Unless scall is MAT_REUSE_MATRIX C will be created.
7743 
7744    MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call
7745 
7746    To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
7747    actually needed.
7748 
7749    If you have many matrices with the same non-zero structure to multiply, you
7750    should either
7751 $   1) use MAT_REUSE_MATRIX in all calls but the first or
7752 $   2) call MatMatMultSymbolic() once and then MatMatMultNumeric() for each product needed
7753 
7754    Level: intermediate
7755 
7756 .seealso: MatMatMultSymbolic(), MatMatMultNumeric(), MatPtAP()
7757 @*/
7758 PetscErrorCode PETSCMAT_DLLEXPORT MatMatMult(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
7759 {
7760   PetscErrorCode ierr;
7761   PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*);
7762   PetscErrorCode (*fB)(Mat,Mat,MatReuse,PetscReal,Mat*);
7763   PetscErrorCode (*mult)(Mat,Mat,MatReuse,PetscReal,Mat *)=PETSC_NULL;
7764 
7765   PetscFunctionBegin;
7766   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7767   PetscValidType(A,1);
7768   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7769   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7770   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
7771   PetscValidType(B,2);
7772   ierr = MatPreallocated(B);CHKERRQ(ierr);
7773   if (!B->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7774   if (B->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7775   PetscValidPointer(C,3);
7776   if (B->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N);
7777   if (scall == MAT_REUSE_MATRIX){
7778     PetscValidPointer(*C,5);
7779     PetscValidHeaderSpecific(*C,MAT_CLASSID,5);
7780   }
7781   if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0;
7782   if (fill < 1.0) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Expected fill=%G must be >= 1.0",fill);
7783   ierr = MatPreallocated(A);CHKERRQ(ierr);
7784 
7785   fA = A->ops->matmult;
7786   fB = B->ops->matmult;
7787   if (fB == fA) {
7788     if (!fB) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"MatMatMult not supported for B of type %s",((PetscObject)B)->type_name);
7789     mult = fB;
7790   } else {
7791     /* dispatch based on the type of A and B */
7792     char  multname[256];
7793     ierr = PetscStrcpy(multname,"MatMatMult_");CHKERRQ(ierr);
7794     ierr = PetscStrcat(multname,((PetscObject)A)->type_name);CHKERRQ(ierr);
7795     ierr = PetscStrcat(multname,"_");CHKERRQ(ierr);
7796     ierr = PetscStrcat(multname,((PetscObject)B)->type_name);CHKERRQ(ierr);
7797     ierr = PetscStrcat(multname,"_C");CHKERRQ(ierr); /* e.g., multname = "MatMatMult_seqdense_seqaij_C" */
7798     ierr = PetscObjectQueryFunction((PetscObject)B,multname,(void (**)(void))&mult);CHKERRQ(ierr);
7799     if (!mult) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_INCOMP,"MatMatMult requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
7800   }
7801   ierr = PetscLogEventBegin(MAT_MatMult,A,B,0,0);CHKERRQ(ierr);
7802   ierr = (*mult)(A,B,scall,fill,C);CHKERRQ(ierr);
7803   ierr = PetscLogEventEnd(MAT_MatMult,A,B,0,0);CHKERRQ(ierr);
7804   PetscFunctionReturn(0);
7805 }
7806 
7807 #undef __FUNCT__
7808 #define __FUNCT__ "MatMatMultSymbolic"
7809 /*@
7810    MatMatMultSymbolic - Performs construction, preallocation, and computes the ij structure
7811    of the matrix-matrix product C=A*B.  Call this routine before calling MatMatMultNumeric().
7812 
7813    Neighbor-wise Collective on Mat
7814 
7815    Input Parameters:
7816 +  A - the left matrix
7817 .  B - the right matrix
7818 -  fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if you do not have a good estimate,
7819       if C is a dense matrix this is irrelevent
7820 
7821    Output Parameters:
7822 .  C - the product matrix
7823 
7824    Notes:
7825    Unless scall is MAT_REUSE_MATRIX C will be created.
7826 
7827    To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
7828    actually needed.
7829 
7830    This routine is currently implemented for
7831     - pairs of AIJ matrices and classes which inherit from AIJ, C will be of type AIJ
7832     - pairs of AIJ (A) and Dense (B) matrix, C will be of type Dense.
7833     - pairs of Dense (A) and AIJ (B) matrix, C will be of type Dense.
7834 
7835    Level: intermediate
7836 
7837    Developers Note: There are ways to estimate the number of nonzeros in the resulting product, see for example, http://arxiv.org/abs/1006.4173
7838      We should incorporate them into PETSc.
7839 
7840 .seealso: MatMatMult(), MatMatMultNumeric()
7841 @*/
7842 PetscErrorCode PETSCMAT_DLLEXPORT MatMatMultSymbolic(Mat A,Mat B,PetscReal fill,Mat *C)
7843 {
7844   PetscErrorCode ierr;
7845   PetscErrorCode (*Asymbolic)(Mat,Mat,PetscReal,Mat *);
7846   PetscErrorCode (*Bsymbolic)(Mat,Mat,PetscReal,Mat *);
7847   PetscErrorCode (*symbolic)(Mat,Mat,PetscReal,Mat *)=PETSC_NULL;
7848 
7849   PetscFunctionBegin;
7850   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7851   PetscValidType(A,1);
7852   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7853   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7854 
7855   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
7856   PetscValidType(B,2);
7857   ierr = MatPreallocated(B);CHKERRQ(ierr);
7858   if (!B->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7859   if (B->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7860   PetscValidPointer(C,3);
7861 
7862   if (B->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N);
7863   if (fill == PETSC_DEFAULT) fill = 2.0;
7864   if (fill < 1.0) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Expected fill=%G must be > 1.0",fill);
7865   ierr = MatPreallocated(A);CHKERRQ(ierr);
7866 
7867   Asymbolic = A->ops->matmultsymbolic;
7868   Bsymbolic = B->ops->matmultsymbolic;
7869   if (Asymbolic == Bsymbolic){
7870     if (!Bsymbolic) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"C=A*B not implemented for B of type %s",((PetscObject)B)->type_name);
7871     symbolic = Bsymbolic;
7872   } else { /* dispatch based on the type of A and B */
7873     char  symbolicname[256];
7874     ierr = PetscStrcpy(symbolicname,"MatMatMultSymbolic_");CHKERRQ(ierr);
7875     ierr = PetscStrcat(symbolicname,((PetscObject)A)->type_name);CHKERRQ(ierr);
7876     ierr = PetscStrcat(symbolicname,"_");CHKERRQ(ierr);
7877     ierr = PetscStrcat(symbolicname,((PetscObject)B)->type_name);CHKERRQ(ierr);
7878     ierr = PetscStrcat(symbolicname,"_C");CHKERRQ(ierr);
7879     ierr = PetscObjectQueryFunction((PetscObject)B,symbolicname,(void (**)(void))&symbolic);CHKERRQ(ierr);
7880     if (!symbolic) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_INCOMP,"MatMatMultSymbolic requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
7881   }
7882   ierr = PetscLogEventBegin(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr);
7883   ierr = (*symbolic)(A,B,fill,C);CHKERRQ(ierr);
7884   ierr = PetscLogEventEnd(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr);
7885   PetscFunctionReturn(0);
7886 }
7887 
7888 #undef __FUNCT__
7889 #define __FUNCT__ "MatMatMultNumeric"
7890 /*@
7891    MatMatMultNumeric - Performs the numeric matrix-matrix product.
7892    Call this routine after first calling MatMatMultSymbolic().
7893 
7894    Neighbor-wise Collective on Mat
7895 
7896    Input Parameters:
7897 +  A - the left matrix
7898 -  B - the right matrix
7899 
7900    Output Parameters:
7901 .  C - the product matrix, which was created by from MatMatMultSymbolic() or a call to MatMatMult().
7902 
7903    Notes:
7904    C must have been created with MatMatMultSymbolic().
7905 
7906    This routine is currently implemented for
7907     - pairs of AIJ matrices and classes which inherit from AIJ, C will be of type MATAIJ.
7908     - pairs of AIJ (A) and Dense (B) matrix, C will be of type Dense.
7909     - pairs of Dense (A) and AIJ (B) matrix, C will be of type Dense.
7910 
7911    Level: intermediate
7912 
7913 .seealso: MatMatMult(), MatMatMultSymbolic()
7914 @*/
7915 PetscErrorCode PETSCMAT_DLLEXPORT MatMatMultNumeric(Mat A,Mat B,Mat C)
7916 {
7917   PetscErrorCode ierr;
7918   PetscErrorCode (*Anumeric)(Mat,Mat,Mat);
7919   PetscErrorCode (*Bnumeric)(Mat,Mat,Mat);
7920   PetscErrorCode (*numeric)(Mat,Mat,Mat)=PETSC_NULL;
7921 
7922   PetscFunctionBegin;
7923   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
7924   PetscValidType(A,1);
7925   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7926   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7927 
7928   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
7929   PetscValidType(B,2);
7930   ierr = MatPreallocated(B);CHKERRQ(ierr);
7931   if (!B->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7932   if (B->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7933 
7934   PetscValidHeaderSpecific(C,MAT_CLASSID,3);
7935   PetscValidType(C,3);
7936   ierr = MatPreallocated(C);CHKERRQ(ierr);
7937   if (!C->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
7938   if (C->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
7939 
7940   if (B->cmap->N!=C->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->cmap->N,C->cmap->N);
7941   if (B->rmap->N!=A->cmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->cmap->N);
7942   if (A->rmap->N!=C->rmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",A->rmap->N,C->rmap->N);
7943   ierr = MatPreallocated(A);CHKERRQ(ierr);
7944 
7945   Anumeric = A->ops->matmultnumeric;
7946   Bnumeric = B->ops->matmultnumeric;
7947   if (Anumeric == Bnumeric){
7948     if (!Bnumeric) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"MatMatMultNumeric not supported for B of type %s",((PetscObject)B)->type_name);
7949     numeric = Bnumeric;
7950   } else {
7951     char  numericname[256];
7952     ierr = PetscStrcpy(numericname,"MatMatMultNumeric_");CHKERRQ(ierr);
7953     ierr = PetscStrcat(numericname,((PetscObject)A)->type_name);CHKERRQ(ierr);
7954     ierr = PetscStrcat(numericname,"_");CHKERRQ(ierr);
7955     ierr = PetscStrcat(numericname,((PetscObject)B)->type_name);CHKERRQ(ierr);
7956     ierr = PetscStrcat(numericname,"_C");CHKERRQ(ierr);
7957     ierr = PetscObjectQueryFunction((PetscObject)B,numericname,(void (**)(void))&numeric);CHKERRQ(ierr);
7958     if (!numeric)
7959       SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_INCOMP,"MatMatMultNumeric requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
7960   }
7961   ierr = PetscLogEventBegin(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr);
7962   ierr = (*numeric)(A,B,C);CHKERRQ(ierr);
7963   ierr = PetscLogEventEnd(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr);
7964   PetscFunctionReturn(0);
7965 }
7966 
7967 #undef __FUNCT__
7968 #define __FUNCT__ "MatMatMultTranspose"
7969 /*@
7970    MatMatMultTranspose - Performs Matrix-Matrix Multiplication C=A^T*B.
7971 
7972    Neighbor-wise Collective on Mat
7973 
7974    Input Parameters:
7975 +  A - the left matrix
7976 .  B - the right matrix
7977 .  scall - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
7978 -  fill - expected fill as ratio of nnz(C)/(nnz(A) + nnz(B)), use PETSC_DEFAULT if not known
7979 
7980    Output Parameters:
7981 .  C - the product matrix
7982 
7983    Notes:
7984    C will be created if MAT_INITIAL_MATRIX and must be destroyed by the user with MatDestroy().
7985 
7986    MAT_REUSE_MATRIX can only be used if the matrices A and B have the same nonzero pattern as in the previous call
7987 
7988   To determine the correct fill value, run with -info and search for the string "Fill ratio" to see the value
7989    actually needed.
7990 
7991    This routine is currently only implemented for pairs of SeqAIJ matrices and pairs of SeqDense matrices and classes
7992    which inherit from SeqAIJ.  C will be of type MATSEQAIJ.
7993 
7994    Level: intermediate
7995 
7996 .seealso: MatMatMultTransposeSymbolic(), MatMatMultTransposeNumeric(), MatPtAP()
7997 @*/
7998 PetscErrorCode PETSCMAT_DLLEXPORT MatMatMultTranspose(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
7999 {
8000   PetscErrorCode ierr;
8001   PetscErrorCode (*fA)(Mat,Mat,MatReuse,PetscReal,Mat*);
8002   PetscErrorCode (*fB)(Mat,Mat,MatReuse,PetscReal,Mat*);
8003 
8004   PetscFunctionBegin;
8005   PetscValidHeaderSpecific(A,MAT_CLASSID,1);
8006   PetscValidType(A,1);
8007   if (!A->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
8008   if (A->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
8009   PetscValidHeaderSpecific(B,MAT_CLASSID,2);
8010   PetscValidType(B,2);
8011   ierr = MatPreallocated(B);CHKERRQ(ierr);
8012   if (!B->assembled) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
8013   if (B->factortype) SETERRQ(((PetscObject)A)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
8014   PetscValidPointer(C,3);
8015   if (B->rmap->N!=A->rmap->N) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Matrix dimensions are incompatible, %D != %D",B->rmap->N,A->rmap->N);
8016   if (fill == PETSC_DEFAULT || fill == PETSC_DECIDE) fill = 2.0;
8017   if (fill < 1.0) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_ARG_SIZ,"Expected fill=%G must be > 1.0",fill);
8018   ierr = MatPreallocated(A);CHKERRQ(ierr);
8019 
8020   fA = A->ops->matmulttranspose;
8021   if (!fA) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"MatMatMultTranspose not supported for A of type %s",((PetscObject)A)->type_name);
8022   fB = B->ops->matmulttranspose;
8023   if (!fB) SETERRQ1(((PetscObject)A)->comm,PETSC_ERR_SUP,"MatMatMultTranspose not supported for B of type %s",((PetscObject)B)->type_name);
8024   if (fB!=fA) SETERRQ2(((PetscObject)A)->comm,PETSC_ERR_ARG_INCOMP,"MatMatMultTranspose requires A, %s, to be compatible with B, %s",((PetscObject)A)->type_name,((PetscObject)B)->type_name);
8025 
8026   ierr = PetscLogEventBegin(MAT_MatMultTranspose,A,B,0,0);CHKERRQ(ierr);
8027   ierr = (*A->ops->matmulttranspose)(A,B,scall,fill,C);CHKERRQ(ierr);
8028   ierr = PetscLogEventEnd(MAT_MatMultTranspose,A,B,0,0);CHKERRQ(ierr);
8029 
8030   PetscFunctionReturn(0);
8031 }
8032 
8033 #undef __FUNCT__
8034 #define __FUNCT__ "MatGetRedundantMatrix"
8035 /*@C
8036    MatGetRedundantMatrix - Create redundant matrices and put them into processors of subcommunicators.
8037 
8038    Collective on Mat
8039 
8040    Input Parameters:
8041 +  mat - the matrix
8042 .  nsubcomm - the number of subcommunicators (= number of redundant pareallel or sequential matrices)
8043 .  subcomm - MPI communicator split from the communicator where mat resides in
8044 .  mlocal_red - number of local rows of the redundant matrix
8045 -  reuse - either MAT_INITIAL_MATRIX or MAT_REUSE_MATRIX
8046 
8047    Output Parameter:
8048 .  matredundant - redundant matrix
8049 
8050    Notes:
8051    MAT_REUSE_MATRIX can only be used when the nonzero structure of the
8052    original matrix has not changed from that last call to MatGetRedundantMatrix().
8053 
8054    This routine creates the duplicated matrices in subcommunicators; you should NOT create them before
8055    calling it.
8056 
8057    Only MPIAIJ matrix is supported.
8058 
8059    Level: advanced
8060 
8061    Concepts: subcommunicator
8062    Concepts: duplicate matrix
8063 
8064 .seealso: MatDestroy()
8065 @*/
8066 PetscErrorCode PETSCMAT_DLLEXPORT MatGetRedundantMatrix(Mat mat,PetscInt nsubcomm,MPI_Comm subcomm,PetscInt mlocal_red,MatReuse reuse,Mat *matredundant)
8067 {
8068   PetscErrorCode ierr;
8069 
8070   PetscFunctionBegin;
8071   PetscValidHeaderSpecific(mat,MAT_CLASSID,1);
8072   if (nsubcomm && reuse == MAT_REUSE_MATRIX) {
8073     PetscValidPointer(*matredundant,6);
8074     PetscValidHeaderSpecific(*matredundant,MAT_CLASSID,6);
8075   }
8076   if (!mat->ops->getredundantmatrix) SETERRQ1(((PetscObject)mat)->comm,PETSC_ERR_SUP,"Mat type %s",((PetscObject)mat)->type_name);
8077   if (!mat->assembled) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
8078   if (mat->factortype) SETERRQ(((PetscObject)mat)->comm,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
8079   ierr = MatPreallocated(mat);CHKERRQ(ierr);
8080 
8081   ierr = PetscLogEventBegin(MAT_GetRedundantMatrix,mat,0,0,0);CHKERRQ(ierr);
8082   ierr = (*mat->ops->getredundantmatrix)(mat,nsubcomm,subcomm,mlocal_red,reuse,matredundant);CHKERRQ(ierr);
8083   ierr = PetscLogEventEnd(MAT_GetRedundantMatrix,mat,0,0,0);CHKERRQ(ierr);
8084   PetscFunctionReturn(0);
8085 }
8086 
8087 #undef __FUNCT__
8088 #define __FUNCT__ "MatGetMultiProcBlock"
8089 /*@C
8090    MatGetMultiProcBlock - Create multiple [bjacobi] 'parallel submatrices' from
8091    a given 'mat' object. Each submatrix can span multiple procs.
8092 
8093    Collective on Mat
8094 
8095    Input Parameters:
8096 +  mat - the matrix
8097 -  subcomm - the subcommunicator obtained by com_split(comm)
8098 
8099    Output Parameter:
8100 .  subMat - 'parallel submatrices each spans a given subcomm
8101 
8102   Notes:
8103   The submatrix partition across processors is dicated by 'subComm' a
8104   communicator obtained by com_split(comm). The comm_split
8105   is not restriced to be grouped with consequitive original ranks.
8106 
8107   Due the comm_split() usage, the parallel layout of the submatrices
8108   map directly to the layout of the original matrix [wrt the local
8109   row,col partitioning]. So the original 'DiagonalMat' naturally maps
8110   into the 'DiagonalMat' of the subMat, hence it is used directly from
8111   the subMat. However the offDiagMat looses some columns - and this is
8112   reconstructed with MatSetValues()
8113 
8114   Level: advanced
8115 
8116   Concepts: subcommunicator
8117   Concepts: submatrices
8118 
8119 .seealso: MatGetSubMatrices()
8120 @*/
8121 PetscErrorCode  PETSCMAT_DLLEXPORT MatGetMultiProcBlock(Mat mat, MPI_Comm subComm, Mat* subMat)
8122 {
8123   PetscErrorCode ierr;
8124   PetscMPIInt    commsize,subCommSize;
8125 
8126   PetscFunctionBegin;
8127   ierr = MPI_Comm_size(((PetscObject)mat)->comm,&commsize);CHKERRQ(ierr);
8128   ierr = MPI_Comm_size(subComm,&subCommSize);CHKERRQ(ierr);
8129   if (subCommSize > commsize) SETERRQ2(((PetscObject)mat)->comm,PETSC_ERR_ARG_OUTOFRANGE,"CommSize %D < SubCommZize %D",commsize,subCommSize);
8130 
8131   ierr = PetscLogEventBegin(MAT_GetMultiProcBlock,mat,0,0,0);CHKERRQ(ierr);
8132   ierr = (*mat->ops->getmultiprocblock)(mat,subComm,subMat);CHKERRQ(ierr);
8133   ierr = PetscLogEventEnd(MAT_GetMultiProcBlock,mat,0,0,0);CHKERRQ(ierr);
8134   PetscFunctionReturn(0);
8135 }
8136