xref: /petsc/src/mat/impls/aij/seq/aijfact.c (revision d2fd7bfc6f0fd2e1d083decbb7cc7d77e16824f0)
1 
2 #include <../src/mat/impls/aij/seq/aij.h>
3 #include <../src/mat/impls/sbaij/seq/sbaij.h>
4 #include <petscbt.h>
5 #include <../src/mat/utils/freespace.h>
6 
7 /*
8       Computes an ordering to get most of the large numerical values in the lower triangular part of the matrix
9 
10       This code does not work and is not called anywhere. It would be registered with MatOrderingRegisterAll()
11 */
12 PetscErrorCode MatGetOrdering_Flow_SeqAIJ(Mat mat,MatOrderingType type,IS *irow,IS *icol)
13 {
14   Mat_SeqAIJ        *a = (Mat_SeqAIJ*)mat->data;
15   PetscErrorCode    ierr;
16   PetscInt          i,j,jj,k, kk,n = mat->rmap->n, current = 0, newcurrent = 0,*order;
17   const PetscInt    *ai = a->i, *aj = a->j;
18   const PetscScalar *aa = a->a;
19   PetscBool         *done;
20   PetscReal         best,past = 0,future;
21 
22   PetscFunctionBegin;
23   /* pick initial row */
24   best = -1;
25   for (i=0; i<n; i++) {
26     future = 0.0;
27     for (j=ai[i]; j<ai[i+1]; j++) {
28       if (aj[j] != i) future += PetscAbsScalar(aa[j]);
29       else              past  = PetscAbsScalar(aa[j]);
30     }
31     if (!future) future = 1.e-10; /* if there is zero in the upper diagonal part want to rank this row high */
32     if (past/future > best) {
33       best    = past/future;
34       current = i;
35     }
36   }
37 
38   ierr     = PetscMalloc1(n,&done);CHKERRQ(ierr);
39   ierr     = PetscArrayzero(done,n);CHKERRQ(ierr);
40   ierr     = PetscMalloc1(n,&order);CHKERRQ(ierr);
41   order[0] = current;
42   for (i=0; i<n-1; i++) {
43     done[current] = PETSC_TRUE;
44     best          = -1;
45     /* loop over all neighbors of current pivot */
46     for (j=ai[current]; j<ai[current+1]; j++) {
47       jj = aj[j];
48       if (done[jj]) continue;
49       /* loop over columns of potential next row computing weights for below and above diagonal */
50       past = future = 0.0;
51       for (k=ai[jj]; k<ai[jj+1]; k++) {
52         kk = aj[k];
53         if (done[kk]) past += PetscAbsScalar(aa[k]);
54         else if (kk != jj) future += PetscAbsScalar(aa[k]);
55       }
56       if (!future) future = 1.e-10; /* if there is zero in the upper diagonal part want to rank this row high */
57       if (past/future > best) {
58         best       = past/future;
59         newcurrent = jj;
60       }
61     }
62     if (best == -1) { /* no neighbors to select from so select best of all that remain */
63       best = -1;
64       for (k=0; k<n; k++) {
65         if (done[k]) continue;
66         future = 0.0;
67         past   = 0.0;
68         for (j=ai[k]; j<ai[k+1]; j++) {
69           kk = aj[j];
70           if (done[kk])       past += PetscAbsScalar(aa[j]);
71           else if (kk != k) future += PetscAbsScalar(aa[j]);
72         }
73         if (!future) future = 1.e-10; /* if there is zero in the upper diagonal part want to rank this row high */
74         if (past/future > best) {
75           best       = past/future;
76           newcurrent = k;
77         }
78       }
79     }
80     if (current == newcurrent) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"newcurrent cannot be current");
81     current    = newcurrent;
82     order[i+1] = current;
83   }
84   ierr  = ISCreateGeneral(PETSC_COMM_SELF,n,order,PETSC_COPY_VALUES,irow);CHKERRQ(ierr);
85   *icol = *irow;
86   ierr  = PetscObjectReference((PetscObject)*irow);CHKERRQ(ierr);
87   ierr  = PetscFree(done);CHKERRQ(ierr);
88   ierr  = PetscFree(order);CHKERRQ(ierr);
89   PetscFunctionReturn(0);
90 }
91 
92 PETSC_INTERN PetscErrorCode MatGetFactor_seqaij_petsc(Mat A,MatFactorType ftype,Mat *B)
93 {
94   PetscInt       n = A->rmap->n;
95   PetscErrorCode ierr;
96 
97   PetscFunctionBegin;
98 #if defined(PETSC_USE_COMPLEX)
99   if (A->hermitian && (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Hermitian Factor is not supported");
100 #endif
101   ierr = MatCreate(PetscObjectComm((PetscObject)A),B);CHKERRQ(ierr);
102   ierr = MatSetSizes(*B,n,n,n,n);CHKERRQ(ierr);
103   if (ftype == MAT_FACTOR_LU || ftype == MAT_FACTOR_ILU || ftype == MAT_FACTOR_ILUDT) {
104     ierr = MatSetType(*B,MATSEQAIJ);CHKERRQ(ierr);
105 
106     (*B)->ops->ilufactorsymbolic = MatILUFactorSymbolic_SeqAIJ;
107     (*B)->ops->lufactorsymbolic  = MatLUFactorSymbolic_SeqAIJ;
108 
109     ierr = MatSetBlockSizesFromMats(*B,A,A);CHKERRQ(ierr);
110   } else if (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) {
111     ierr = MatSetType(*B,MATSEQSBAIJ);CHKERRQ(ierr);
112     ierr = MatSeqSBAIJSetPreallocation(*B,1,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
113 
114     (*B)->ops->iccfactorsymbolic      = MatICCFactorSymbolic_SeqAIJ;
115     (*B)->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SeqAIJ;
116   } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Factor type not supported");
117   (*B)->factortype = ftype;
118 
119   ierr = PetscFree((*B)->solvertype);CHKERRQ(ierr);
120   ierr = PetscStrallocpy(MATSOLVERPETSC,&(*B)->solvertype);CHKERRQ(ierr);
121   PetscFunctionReturn(0);
122 }
123 
124 PetscErrorCode MatLUFactorSymbolic_SeqAIJ_inplace(Mat B,Mat A,IS isrow,IS iscol,const MatFactorInfo *info)
125 {
126   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data,*b;
127   IS                 isicol;
128   PetscErrorCode     ierr;
129   const PetscInt     *r,*ic;
130   PetscInt           i,n=A->rmap->n,*ai=a->i,*aj=a->j;
131   PetscInt           *bi,*bj,*ajtmp;
132   PetscInt           *bdiag,row,nnz,nzi,reallocs=0,nzbd,*im;
133   PetscReal          f;
134   PetscInt           nlnk,*lnk,k,**bi_ptr;
135   PetscFreeSpaceList free_space=NULL,current_space=NULL;
136   PetscBT            lnkbt;
137   PetscBool          missing;
138 
139   PetscFunctionBegin;
140   if (A->rmap->N != A->cmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"matrix must be square");
141   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
142   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
143 
144   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
145   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
146   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
147 
148   /* get new row pointers */
149   ierr  = PetscMalloc1(n+1,&bi);CHKERRQ(ierr);
150   bi[0] = 0;
151 
152   /* bdiag is location of diagonal in factor */
153   ierr     = PetscMalloc1(n+1,&bdiag);CHKERRQ(ierr);
154   bdiag[0] = 0;
155 
156   /* linked list for storing column indices of the active row */
157   nlnk = n + 1;
158   ierr = PetscLLCreate(n,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
159 
160   ierr = PetscMalloc2(n+1,&bi_ptr,n+1,&im);CHKERRQ(ierr);
161 
162   /* initial FreeSpace size is f*(ai[n]+1) */
163   f             = info->fill;
164   ierr          = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space);CHKERRQ(ierr);
165   current_space = free_space;
166 
167   for (i=0; i<n; i++) {
168     /* copy previous fill into linked list */
169     nzi = 0;
170     nnz = ai[r[i]+1] - ai[r[i]];
171     ajtmp = aj + ai[r[i]];
172     ierr  = PetscLLAddPerm(nnz,ajtmp,ic,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
173     nzi  += nlnk;
174 
175     /* add pivot rows into linked list */
176     row = lnk[n];
177     while (row < i) {
178       nzbd  = bdiag[row] - bi[row] + 1;   /* num of entries in the row with column index <= row */
179       ajtmp = bi_ptr[row] + nzbd;   /* points to the entry next to the diagonal */
180       ierr  = PetscLLAddSortedLU(ajtmp,row,nlnk,lnk,lnkbt,i,nzbd,im);CHKERRQ(ierr);
181       nzi  += nlnk;
182       row   = lnk[row];
183     }
184     bi[i+1] = bi[i] + nzi;
185     im[i]   = nzi;
186 
187     /* mark bdiag */
188     nzbd = 0;
189     nnz  = nzi;
190     k    = lnk[n];
191     while (nnz-- && k < i) {
192       nzbd++;
193       k = lnk[k];
194     }
195     bdiag[i] = bi[i] + nzbd;
196 
197     /* if free space is not available, make more free space */
198     if (current_space->local_remaining<nzi) {
199       nnz  = PetscIntMultTruncate(n - i,nzi); /* estimated and max additional space needed */
200       ierr = PetscFreeSpaceGet(nnz,&current_space);CHKERRQ(ierr);
201       reallocs++;
202     }
203 
204     /* copy data into free space, then initialize lnk */
205     ierr = PetscLLClean(n,n,nzi,lnk,current_space->array,lnkbt);CHKERRQ(ierr);
206 
207     bi_ptr[i]                       = current_space->array;
208     current_space->array           += nzi;
209     current_space->local_used      += nzi;
210     current_space->local_remaining -= nzi;
211   }
212 #if defined(PETSC_USE_INFO)
213   if (ai[n] != 0) {
214     PetscReal af = ((PetscReal)bi[n])/((PetscReal)ai[n]);
215     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)f,(double)af);CHKERRQ(ierr);
216     ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
217     ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g);\n",(double)af);CHKERRQ(ierr);
218     ierr = PetscInfo(A,"for best performance.\n");CHKERRQ(ierr);
219   } else {
220     ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
221   }
222 #endif
223 
224   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
225   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
226 
227   /* destroy list of free space and other temporary array(s) */
228   ierr = PetscMalloc1(bi[n]+1,&bj);CHKERRQ(ierr);
229   ierr = PetscFreeSpaceContiguous(&free_space,bj);CHKERRQ(ierr);
230   ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
231   ierr = PetscFree2(bi_ptr,im);CHKERRQ(ierr);
232 
233   /* put together the new matrix */
234   ierr = MatSeqAIJSetPreallocation_SeqAIJ(B,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
235   ierr = PetscLogObjectParent((PetscObject)B,(PetscObject)isicol);CHKERRQ(ierr);
236   b    = (Mat_SeqAIJ*)(B)->data;
237 
238   b->free_a       = PETSC_TRUE;
239   b->free_ij      = PETSC_TRUE;
240   b->singlemalloc = PETSC_FALSE;
241 
242   ierr    = PetscMalloc1(bi[n]+1,&b->a);CHKERRQ(ierr);
243   b->j    = bj;
244   b->i    = bi;
245   b->diag = bdiag;
246   b->ilen = 0;
247   b->imax = 0;
248   b->row  = isrow;
249   b->col  = iscol;
250   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
251   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
252   b->icol = isicol;
253   ierr    = PetscMalloc1(n+1,&b->solve_work);CHKERRQ(ierr);
254 
255   /* In b structure:  Free imax, ilen, old a, old j.  Allocate solve_work, new a, new j */
256   ierr     = PetscLogObjectMemory((PetscObject)B,(bi[n]-n)*(sizeof(PetscInt)+sizeof(PetscScalar)));CHKERRQ(ierr);
257   b->maxnz = b->nz = bi[n];
258 
259   (B)->factortype            = MAT_FACTOR_LU;
260   (B)->info.factor_mallocs   = reallocs;
261   (B)->info.fill_ratio_given = f;
262 
263   if (ai[n]) {
264     (B)->info.fill_ratio_needed = ((PetscReal)bi[n])/((PetscReal)ai[n]);
265   } else {
266     (B)->info.fill_ratio_needed = 0.0;
267   }
268   (B)->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_inplace;
269   if (a->inode.size) {
270     (B)->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode_inplace;
271   }
272   PetscFunctionReturn(0);
273 }
274 
275 PetscErrorCode MatLUFactorSymbolic_SeqAIJ(Mat B,Mat A,IS isrow,IS iscol,const MatFactorInfo *info)
276 {
277   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data,*b;
278   IS                 isicol;
279   PetscErrorCode     ierr;
280   const PetscInt     *r,*ic,*ai=a->i,*aj=a->j,*ajtmp;
281   PetscInt           i,n=A->rmap->n;
282   PetscInt           *bi,*bj;
283   PetscInt           *bdiag,row,nnz,nzi,reallocs=0,nzbd,*im;
284   PetscReal          f;
285   PetscInt           nlnk,*lnk,k,**bi_ptr;
286   PetscFreeSpaceList free_space=NULL,current_space=NULL;
287   PetscBT            lnkbt;
288   PetscBool          missing;
289 
290   PetscFunctionBegin;
291   if (A->rmap->N != A->cmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"matrix must be square");
292   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
293   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
294 
295   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
296   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
297   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
298 
299   /* get new row and diagonal pointers, must be allocated separately because they will be given to the Mat_SeqAIJ and freed separately */
300   ierr  = PetscMalloc1(n+1,&bi);CHKERRQ(ierr);
301   ierr  = PetscMalloc1(n+1,&bdiag);CHKERRQ(ierr);
302   bi[0] = bdiag[0] = 0;
303 
304   /* linked list for storing column indices of the active row */
305   nlnk = n + 1;
306   ierr = PetscLLCreate(n,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
307 
308   ierr = PetscMalloc2(n+1,&bi_ptr,n+1,&im);CHKERRQ(ierr);
309 
310   /* initial FreeSpace size is f*(ai[n]+1) */
311   f             = info->fill;
312   ierr          = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space);CHKERRQ(ierr);
313   current_space = free_space;
314 
315   for (i=0; i<n; i++) {
316     /* copy previous fill into linked list */
317     nzi = 0;
318     nnz = ai[r[i]+1] - ai[r[i]];
319     ajtmp = aj + ai[r[i]];
320     ierr  = PetscLLAddPerm(nnz,ajtmp,ic,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
321     nzi  += nlnk;
322 
323     /* add pivot rows into linked list */
324     row = lnk[n];
325     while (row < i) {
326       nzbd  = bdiag[row] + 1; /* num of entries in the row with column index <= row */
327       ajtmp = bi_ptr[row] + nzbd; /* points to the entry next to the diagonal */
328       ierr  = PetscLLAddSortedLU(ajtmp,row,nlnk,lnk,lnkbt,i,nzbd,im);CHKERRQ(ierr);
329       nzi  += nlnk;
330       row   = lnk[row];
331     }
332     bi[i+1] = bi[i] + nzi;
333     im[i]   = nzi;
334 
335     /* mark bdiag */
336     nzbd = 0;
337     nnz  = nzi;
338     k    = lnk[n];
339     while (nnz-- && k < i) {
340       nzbd++;
341       k = lnk[k];
342     }
343     bdiag[i] = nzbd; /* note: bdiag[i] = nnzL as input for PetscFreeSpaceContiguous_LU() */
344 
345     /* if free space is not available, make more free space */
346     if (current_space->local_remaining<nzi) {
347       /* estimated additional space needed */
348       nnz  = PetscIntMultTruncate(2,PetscIntMultTruncate(n-1,nzi));
349       ierr = PetscFreeSpaceGet(nnz,&current_space);CHKERRQ(ierr);
350       reallocs++;
351     }
352 
353     /* copy data into free space, then initialize lnk */
354     ierr = PetscLLClean(n,n,nzi,lnk,current_space->array,lnkbt);CHKERRQ(ierr);
355 
356     bi_ptr[i]                       = current_space->array;
357     current_space->array           += nzi;
358     current_space->local_used      += nzi;
359     current_space->local_remaining -= nzi;
360   }
361 
362   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
363   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
364 
365   /*   copy free_space into bj and free free_space; set bi, bj, bdiag in new datastructure; */
366   ierr = PetscMalloc1(bi[n]+1,&bj);CHKERRQ(ierr);
367   ierr = PetscFreeSpaceContiguous_LU(&free_space,bj,n,bi,bdiag);CHKERRQ(ierr);
368   ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
369   ierr = PetscFree2(bi_ptr,im);CHKERRQ(ierr);
370 
371   /* put together the new matrix */
372   ierr = MatSeqAIJSetPreallocation_SeqAIJ(B,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
373   ierr = PetscLogObjectParent((PetscObject)B,(PetscObject)isicol);CHKERRQ(ierr);
374   b    = (Mat_SeqAIJ*)(B)->data;
375 
376   b->free_a       = PETSC_TRUE;
377   b->free_ij      = PETSC_TRUE;
378   b->singlemalloc = PETSC_FALSE;
379 
380   ierr = PetscMalloc1(bdiag[0]+1,&b->a);CHKERRQ(ierr);
381 
382   b->j    = bj;
383   b->i    = bi;
384   b->diag = bdiag;
385   b->ilen = 0;
386   b->imax = 0;
387   b->row  = isrow;
388   b->col  = iscol;
389   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
390   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
391   b->icol = isicol;
392   ierr    = PetscMalloc1(n+1,&b->solve_work);CHKERRQ(ierr);
393 
394   /* In b structure:  Free imax, ilen, old a, old j.  Allocate solve_work, new a, new j */
395   ierr     = PetscLogObjectMemory((PetscObject)B,(bdiag[0]+1)*(sizeof(PetscInt)+sizeof(PetscScalar)));CHKERRQ(ierr);
396   b->maxnz = b->nz = bdiag[0]+1;
397 
398   B->factortype            = MAT_FACTOR_LU;
399   B->info.factor_mallocs   = reallocs;
400   B->info.fill_ratio_given = f;
401 
402   if (ai[n]) {
403     B->info.fill_ratio_needed = ((PetscReal)(bdiag[0]+1))/((PetscReal)ai[n]);
404   } else {
405     B->info.fill_ratio_needed = 0.0;
406   }
407 #if defined(PETSC_USE_INFO)
408   if (ai[n] != 0) {
409     PetscReal af = B->info.fill_ratio_needed;
410     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)f,(double)af);CHKERRQ(ierr);
411     ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
412     ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g);\n",(double)af);CHKERRQ(ierr);
413     ierr = PetscInfo(A,"for best performance.\n");CHKERRQ(ierr);
414   } else {
415     ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
416   }
417 #endif
418   B->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ;
419   if (a->inode.size) {
420     B->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode;
421   }
422   ierr = MatSeqAIJCheckInode_FactorLU(B);CHKERRQ(ierr);
423   PetscFunctionReturn(0);
424 }
425 
426 /*
427     Trouble in factorization, should we dump the original matrix?
428 */
429 PetscErrorCode MatFactorDumpMatrix(Mat A)
430 {
431   PetscErrorCode ierr;
432   PetscBool      flg = PETSC_FALSE;
433 
434   PetscFunctionBegin;
435   ierr = PetscOptionsGetBool(((PetscObject)A)->options,NULL,"-mat_factor_dump_on_error",&flg,NULL);CHKERRQ(ierr);
436   if (flg) {
437     PetscViewer viewer;
438     char        filename[PETSC_MAX_PATH_LEN];
439 
440     ierr = PetscSNPrintf(filename,PETSC_MAX_PATH_LEN,"matrix_factor_error.%d",PetscGlobalRank);CHKERRQ(ierr);
441     ierr = PetscViewerBinaryOpen(PetscObjectComm((PetscObject)A),filename,FILE_MODE_WRITE,&viewer);CHKERRQ(ierr);
442     ierr = MatView(A,viewer);CHKERRQ(ierr);
443     ierr = PetscViewerDestroy(&viewer);CHKERRQ(ierr);
444   }
445   PetscFunctionReturn(0);
446 }
447 
448 PetscErrorCode MatLUFactorNumeric_SeqAIJ(Mat B,Mat A,const MatFactorInfo *info)
449 {
450   Mat             C     =B;
451   Mat_SeqAIJ      *a    =(Mat_SeqAIJ*)A->data,*b=(Mat_SeqAIJ*)C->data;
452   IS              isrow = b->row,isicol = b->icol;
453   PetscErrorCode  ierr;
454   const PetscInt  *r,*ic,*ics;
455   const PetscInt  n=A->rmap->n,*ai=a->i,*aj=a->j,*bi=b->i,*bj=b->j,*bdiag=b->diag;
456   PetscInt        i,j,k,nz,nzL,row,*pj;
457   const PetscInt  *ajtmp,*bjtmp;
458   MatScalar       *rtmp,*pc,multiplier,*pv;
459   const MatScalar *aa=a->a,*v;
460   PetscBool       row_identity,col_identity;
461   FactorShiftCtx  sctx;
462   const PetscInt  *ddiag;
463   PetscReal       rs;
464   MatScalar       d;
465 
466   PetscFunctionBegin;
467   /* MatPivotSetUp(): initialize shift context sctx */
468   ierr = PetscMemzero(&sctx,sizeof(FactorShiftCtx));CHKERRQ(ierr);
469 
470   if (info->shifttype == (PetscReal) MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
471     ddiag          = a->diag;
472     sctx.shift_top = info->zeropivot;
473     for (i=0; i<n; i++) {
474       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
475       d  = (aa)[ddiag[i]];
476       rs = -PetscAbsScalar(d) - PetscRealPart(d);
477       v  = aa+ai[i];
478       nz = ai[i+1] - ai[i];
479       for (j=0; j<nz; j++) rs += PetscAbsScalar(v[j]);
480       if (rs>sctx.shift_top) sctx.shift_top = rs;
481     }
482     sctx.shift_top *= 1.1;
483     sctx.nshift_max = 5;
484     sctx.shift_lo   = 0.;
485     sctx.shift_hi   = 1.;
486   }
487 
488   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
489   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
490   ierr = PetscMalloc1(n+1,&rtmp);CHKERRQ(ierr);
491   ics  = ic;
492 
493   do {
494     sctx.newshift = PETSC_FALSE;
495     for (i=0; i<n; i++) {
496       /* zero rtmp */
497       /* L part */
498       nz    = bi[i+1] - bi[i];
499       bjtmp = bj + bi[i];
500       for  (j=0; j<nz; j++) rtmp[bjtmp[j]] = 0.0;
501 
502       /* U part */
503       nz    = bdiag[i]-bdiag[i+1];
504       bjtmp = bj + bdiag[i+1]+1;
505       for  (j=0; j<nz; j++) rtmp[bjtmp[j]] = 0.0;
506 
507       /* load in initial (unfactored row) */
508       nz    = ai[r[i]+1] - ai[r[i]];
509       ajtmp = aj + ai[r[i]];
510       v     = aa + ai[r[i]];
511       for (j=0; j<nz; j++) {
512         rtmp[ics[ajtmp[j]]] = v[j];
513       }
514       /* ZeropivotApply() */
515       rtmp[i] += sctx.shift_amount;  /* shift the diagonal of the matrix */
516 
517       /* elimination */
518       bjtmp = bj + bi[i];
519       row   = *bjtmp++;
520       nzL   = bi[i+1] - bi[i];
521       for (k=0; k < nzL; k++) {
522         pc = rtmp + row;
523         if (*pc != 0.0) {
524           pv         = b->a + bdiag[row];
525           multiplier = *pc * (*pv);
526           *pc        = multiplier;
527 
528           pj = b->j + bdiag[row+1]+1; /* beginning of U(row,:) */
529           pv = b->a + bdiag[row+1]+1;
530           nz = bdiag[row]-bdiag[row+1]-1; /* num of entries in U(row,:) excluding diag */
531 
532           for (j=0; j<nz; j++) rtmp[pj[j]] -= multiplier * pv[j];
533           ierr = PetscLogFlops(1+2*nz);CHKERRQ(ierr);
534         }
535         row = *bjtmp++;
536       }
537 
538       /* finished row so stick it into b->a */
539       rs = 0.0;
540       /* L part */
541       pv = b->a + bi[i];
542       pj = b->j + bi[i];
543       nz = bi[i+1] - bi[i];
544       for (j=0; j<nz; j++) {
545         pv[j] = rtmp[pj[j]]; rs += PetscAbsScalar(pv[j]);
546       }
547 
548       /* U part */
549       pv = b->a + bdiag[i+1]+1;
550       pj = b->j + bdiag[i+1]+1;
551       nz = bdiag[i] - bdiag[i+1]-1;
552       for (j=0; j<nz; j++) {
553         pv[j] = rtmp[pj[j]]; rs += PetscAbsScalar(pv[j]);
554       }
555 
556       sctx.rs = rs;
557       sctx.pv = rtmp[i];
558       ierr    = MatPivotCheck(B,A,info,&sctx,i);CHKERRQ(ierr);
559       if (sctx.newshift) break; /* break for-loop */
560       rtmp[i] = sctx.pv; /* sctx.pv might be updated in the case of MAT_SHIFT_INBLOCKS */
561 
562       /* Mark diagonal and invert diagonal for simplier triangular solves */
563       pv  = b->a + bdiag[i];
564       *pv = 1.0/rtmp[i];
565 
566     } /* endof for (i=0; i<n; i++) { */
567 
568     /* MatPivotRefine() */
569     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE && !sctx.newshift && sctx.shift_fraction>0 && sctx.nshift<sctx.nshift_max) {
570       /*
571        * if no shift in this attempt & shifting & started shifting & can refine,
572        * then try lower shift
573        */
574       sctx.shift_hi       = sctx.shift_fraction;
575       sctx.shift_fraction = (sctx.shift_hi+sctx.shift_lo)/2.;
576       sctx.shift_amount   = sctx.shift_fraction * sctx.shift_top;
577       sctx.newshift       = PETSC_TRUE;
578       sctx.nshift++;
579     }
580   } while (sctx.newshift);
581 
582   ierr = PetscFree(rtmp);CHKERRQ(ierr);
583   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
584   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
585 
586   ierr = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
587   ierr = ISIdentity(isicol,&col_identity);CHKERRQ(ierr);
588   if (b->inode.size) {
589     C->ops->solve = MatSolve_SeqAIJ_Inode;
590   } else if (row_identity && col_identity) {
591     C->ops->solve = MatSolve_SeqAIJ_NaturalOrdering;
592   } else {
593     C->ops->solve = MatSolve_SeqAIJ;
594   }
595   C->ops->solveadd          = MatSolveAdd_SeqAIJ;
596   C->ops->solvetranspose    = MatSolveTranspose_SeqAIJ;
597   C->ops->solvetransposeadd = MatSolveTransposeAdd_SeqAIJ;
598   C->ops->matsolve          = MatMatSolve_SeqAIJ;
599   C->assembled              = PETSC_TRUE;
600   C->preallocated           = PETSC_TRUE;
601 
602   ierr = PetscLogFlops(C->cmap->n);CHKERRQ(ierr);
603 
604   /* MatShiftView(A,info,&sctx) */
605   if (sctx.nshift) {
606     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) {
607       ierr = PetscInfo4(A,"number of shift_pd tries %D, shift_amount %g, diagonal shifted up by %e fraction top_value %e\n",sctx.nshift,(double)sctx.shift_amount,(double)sctx.shift_fraction,(double)sctx.shift_top);CHKERRQ(ierr);
608     } else if (info->shifttype == (PetscReal)MAT_SHIFT_NONZERO) {
609       ierr = PetscInfo2(A,"number of shift_nz tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
610     } else if (info->shifttype == (PetscReal)MAT_SHIFT_INBLOCKS) {
611       ierr = PetscInfo2(A,"number of shift_inblocks applied %D, each shift_amount %g\n",sctx.nshift,(double)info->shiftamount);CHKERRQ(ierr);
612     }
613   }
614   PetscFunctionReturn(0);
615 }
616 
617 PetscErrorCode MatLUFactorNumeric_SeqAIJ_inplace(Mat B,Mat A,const MatFactorInfo *info)
618 {
619   Mat             C     =B;
620   Mat_SeqAIJ      *a    =(Mat_SeqAIJ*)A->data,*b=(Mat_SeqAIJ*)C->data;
621   IS              isrow = b->row,isicol = b->icol;
622   PetscErrorCode  ierr;
623   const PetscInt  *r,*ic,*ics;
624   PetscInt        nz,row,i,j,n=A->rmap->n,diag;
625   const PetscInt  *ai=a->i,*aj=a->j,*bi=b->i,*bj=b->j;
626   const PetscInt  *ajtmp,*bjtmp,*diag_offset = b->diag,*pj;
627   MatScalar       *pv,*rtmp,*pc,multiplier,d;
628   const MatScalar *v,*aa=a->a;
629   PetscReal       rs=0.0;
630   FactorShiftCtx  sctx;
631   const PetscInt  *ddiag;
632   PetscBool       row_identity, col_identity;
633 
634   PetscFunctionBegin;
635   /* MatPivotSetUp(): initialize shift context sctx */
636   ierr = PetscMemzero(&sctx,sizeof(FactorShiftCtx));CHKERRQ(ierr);
637 
638   if (info->shifttype == (PetscReal) MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
639     ddiag          = a->diag;
640     sctx.shift_top = info->zeropivot;
641     for (i=0; i<n; i++) {
642       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
643       d  = (aa)[ddiag[i]];
644       rs = -PetscAbsScalar(d) - PetscRealPart(d);
645       v  = aa+ai[i];
646       nz = ai[i+1] - ai[i];
647       for (j=0; j<nz; j++) rs += PetscAbsScalar(v[j]);
648       if (rs>sctx.shift_top) sctx.shift_top = rs;
649     }
650     sctx.shift_top *= 1.1;
651     sctx.nshift_max = 5;
652     sctx.shift_lo   = 0.;
653     sctx.shift_hi   = 1.;
654   }
655 
656   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
657   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
658   ierr = PetscMalloc1(n+1,&rtmp);CHKERRQ(ierr);
659   ics  = ic;
660 
661   do {
662     sctx.newshift = PETSC_FALSE;
663     for (i=0; i<n; i++) {
664       nz    = bi[i+1] - bi[i];
665       bjtmp = bj + bi[i];
666       for  (j=0; j<nz; j++) rtmp[bjtmp[j]] = 0.0;
667 
668       /* load in initial (unfactored row) */
669       nz    = ai[r[i]+1] - ai[r[i]];
670       ajtmp = aj + ai[r[i]];
671       v     = aa + ai[r[i]];
672       for (j=0; j<nz; j++) {
673         rtmp[ics[ajtmp[j]]] = v[j];
674       }
675       rtmp[ics[r[i]]] += sctx.shift_amount; /* shift the diagonal of the matrix */
676 
677       row = *bjtmp++;
678       while  (row < i) {
679         pc = rtmp + row;
680         if (*pc != 0.0) {
681           pv         = b->a + diag_offset[row];
682           pj         = b->j + diag_offset[row] + 1;
683           multiplier = *pc / *pv++;
684           *pc        = multiplier;
685           nz         = bi[row+1] - diag_offset[row] - 1;
686           for (j=0; j<nz; j++) rtmp[pj[j]] -= multiplier * pv[j];
687           ierr = PetscLogFlops(1+2*nz);CHKERRQ(ierr);
688         }
689         row = *bjtmp++;
690       }
691       /* finished row so stick it into b->a */
692       pv   = b->a + bi[i];
693       pj   = b->j + bi[i];
694       nz   = bi[i+1] - bi[i];
695       diag = diag_offset[i] - bi[i];
696       rs   = 0.0;
697       for (j=0; j<nz; j++) {
698         pv[j] = rtmp[pj[j]];
699         rs   += PetscAbsScalar(pv[j]);
700       }
701       rs -= PetscAbsScalar(pv[diag]);
702 
703       sctx.rs = rs;
704       sctx.pv = pv[diag];
705       ierr    = MatPivotCheck(B,A,info,&sctx,i);CHKERRQ(ierr);
706       if (sctx.newshift) break;
707       pv[diag] = sctx.pv;
708     }
709 
710     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE && !sctx.newshift && sctx.shift_fraction>0 && sctx.nshift<sctx.nshift_max) {
711       /*
712        * if no shift in this attempt & shifting & started shifting & can refine,
713        * then try lower shift
714        */
715       sctx.shift_hi       = sctx.shift_fraction;
716       sctx.shift_fraction = (sctx.shift_hi+sctx.shift_lo)/2.;
717       sctx.shift_amount   = sctx.shift_fraction * sctx.shift_top;
718       sctx.newshift       = PETSC_TRUE;
719       sctx.nshift++;
720     }
721   } while (sctx.newshift);
722 
723   /* invert diagonal entries for simplier triangular solves */
724   for (i=0; i<n; i++) {
725     b->a[diag_offset[i]] = 1.0/b->a[diag_offset[i]];
726   }
727   ierr = PetscFree(rtmp);CHKERRQ(ierr);
728   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
729   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
730 
731   ierr = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
732   ierr = ISIdentity(isicol,&col_identity);CHKERRQ(ierr);
733   if (row_identity && col_identity) {
734     C->ops->solve = MatSolve_SeqAIJ_NaturalOrdering_inplace;
735   } else {
736     C->ops->solve = MatSolve_SeqAIJ_inplace;
737   }
738   C->ops->solveadd          = MatSolveAdd_SeqAIJ_inplace;
739   C->ops->solvetranspose    = MatSolveTranspose_SeqAIJ_inplace;
740   C->ops->solvetransposeadd = MatSolveTransposeAdd_SeqAIJ_inplace;
741   C->ops->matsolve          = MatMatSolve_SeqAIJ_inplace;
742 
743   C->assembled    = PETSC_TRUE;
744   C->preallocated = PETSC_TRUE;
745 
746   ierr = PetscLogFlops(C->cmap->n);CHKERRQ(ierr);
747   if (sctx.nshift) {
748     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) {
749       ierr = PetscInfo4(A,"number of shift_pd tries %D, shift_amount %g, diagonal shifted up by %e fraction top_value %e\n",sctx.nshift,(double)sctx.shift_amount,(double)sctx.shift_fraction,(double)sctx.shift_top);CHKERRQ(ierr);
750     } else if (info->shifttype == (PetscReal)MAT_SHIFT_NONZERO) {
751       ierr = PetscInfo2(A,"number of shift_nz tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
752     }
753   }
754   (C)->ops->solve          = MatSolve_SeqAIJ_inplace;
755   (C)->ops->solvetranspose = MatSolveTranspose_SeqAIJ_inplace;
756 
757   ierr = MatSeqAIJCheckInode(C);CHKERRQ(ierr);
758   PetscFunctionReturn(0);
759 }
760 
761 /*
762    This routine implements inplace ILU(0) with row or/and column permutations.
763    Input:
764      A - original matrix
765    Output;
766      A - a->i (rowptr) is same as original rowptr, but factored i-the row is stored in rowperm[i]
767          a->j (col index) is permuted by the inverse of colperm, then sorted
768          a->a reordered accordingly with a->j
769          a->diag (ptr to diagonal elements) is updated.
770 */
771 PetscErrorCode MatLUFactorNumeric_SeqAIJ_InplaceWithPerm(Mat B,Mat A,const MatFactorInfo *info)
772 {
773   Mat_SeqAIJ      *a    =(Mat_SeqAIJ*)A->data;
774   IS              isrow = a->row,isicol = a->icol;
775   PetscErrorCode  ierr;
776   const PetscInt  *r,*ic,*ics;
777   PetscInt        i,j,n=A->rmap->n,*ai=a->i,*aj=a->j;
778   PetscInt        *ajtmp,nz,row;
779   PetscInt        *diag = a->diag,nbdiag,*pj;
780   PetscScalar     *rtmp,*pc,multiplier,d;
781   MatScalar       *pv,*v;
782   PetscReal       rs;
783   FactorShiftCtx  sctx;
784   const MatScalar *aa=a->a,*vtmp;
785 
786   PetscFunctionBegin;
787   if (A != B) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"input and output matrix must have same address");
788 
789   /* MatPivotSetUp(): initialize shift context sctx */
790   ierr = PetscMemzero(&sctx,sizeof(FactorShiftCtx));CHKERRQ(ierr);
791 
792   if (info->shifttype == (PetscReal) MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
793     const PetscInt *ddiag = a->diag;
794     sctx.shift_top = info->zeropivot;
795     for (i=0; i<n; i++) {
796       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
797       d    = (aa)[ddiag[i]];
798       rs   = -PetscAbsScalar(d) - PetscRealPart(d);
799       vtmp = aa+ai[i];
800       nz   = ai[i+1] - ai[i];
801       for (j=0; j<nz; j++) rs += PetscAbsScalar(vtmp[j]);
802       if (rs>sctx.shift_top) sctx.shift_top = rs;
803     }
804     sctx.shift_top *= 1.1;
805     sctx.nshift_max = 5;
806     sctx.shift_lo   = 0.;
807     sctx.shift_hi   = 1.;
808   }
809 
810   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
811   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
812   ierr = PetscMalloc1(n+1,&rtmp);CHKERRQ(ierr);
813   ierr = PetscArrayzero(rtmp,n+1);CHKERRQ(ierr);
814   ics  = ic;
815 
816 #if defined(MV)
817   sctx.shift_top      = 0.;
818   sctx.nshift_max     = 0;
819   sctx.shift_lo       = 0.;
820   sctx.shift_hi       = 0.;
821   sctx.shift_fraction = 0.;
822 
823   if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
824     sctx.shift_top = 0.;
825     for (i=0; i<n; i++) {
826       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
827       d  = (a->a)[diag[i]];
828       rs = -PetscAbsScalar(d) - PetscRealPart(d);
829       v  = a->a+ai[i];
830       nz = ai[i+1] - ai[i];
831       for (j=0; j<nz; j++) rs += PetscAbsScalar(v[j]);
832       if (rs>sctx.shift_top) sctx.shift_top = rs;
833     }
834     if (sctx.shift_top < info->zeropivot) sctx.shift_top = info->zeropivot;
835     sctx.shift_top *= 1.1;
836     sctx.nshift_max = 5;
837     sctx.shift_lo   = 0.;
838     sctx.shift_hi   = 1.;
839   }
840 
841   sctx.shift_amount = 0.;
842   sctx.nshift       = 0;
843 #endif
844 
845   do {
846     sctx.newshift = PETSC_FALSE;
847     for (i=0; i<n; i++) {
848       /* load in initial unfactored row */
849       nz    = ai[r[i]+1] - ai[r[i]];
850       ajtmp = aj + ai[r[i]];
851       v     = a->a + ai[r[i]];
852       /* sort permuted ajtmp and values v accordingly */
853       for (j=0; j<nz; j++) ajtmp[j] = ics[ajtmp[j]];
854       ierr = PetscSortIntWithScalarArray(nz,ajtmp,v);CHKERRQ(ierr);
855 
856       diag[r[i]] = ai[r[i]];
857       for (j=0; j<nz; j++) {
858         rtmp[ajtmp[j]] = v[j];
859         if (ajtmp[j] < i) diag[r[i]]++; /* update a->diag */
860       }
861       rtmp[r[i]] += sctx.shift_amount; /* shift the diagonal of the matrix */
862 
863       row = *ajtmp++;
864       while  (row < i) {
865         pc = rtmp + row;
866         if (*pc != 0.0) {
867           pv = a->a + diag[r[row]];
868           pj = aj + diag[r[row]] + 1;
869 
870           multiplier = *pc / *pv++;
871           *pc        = multiplier;
872           nz         = ai[r[row]+1] - diag[r[row]] - 1;
873           for (j=0; j<nz; j++) rtmp[pj[j]] -= multiplier * pv[j];
874           ierr = PetscLogFlops(1+2*nz);CHKERRQ(ierr);
875         }
876         row = *ajtmp++;
877       }
878       /* finished row so overwrite it onto a->a */
879       pv     = a->a + ai[r[i]];
880       pj     = aj + ai[r[i]];
881       nz     = ai[r[i]+1] - ai[r[i]];
882       nbdiag = diag[r[i]] - ai[r[i]]; /* num of entries before the diagonal */
883 
884       rs = 0.0;
885       for (j=0; j<nz; j++) {
886         pv[j] = rtmp[pj[j]];
887         if (j != nbdiag) rs += PetscAbsScalar(pv[j]);
888       }
889 
890       sctx.rs = rs;
891       sctx.pv = pv[nbdiag];
892       ierr    = MatPivotCheck(B,A,info,&sctx,i);CHKERRQ(ierr);
893       if (sctx.newshift) break;
894       pv[nbdiag] = sctx.pv;
895     }
896 
897     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE && !sctx.newshift && sctx.shift_fraction>0 && sctx.nshift<sctx.nshift_max) {
898       /*
899        * if no shift in this attempt & shifting & started shifting & can refine,
900        * then try lower shift
901        */
902       sctx.shift_hi       = sctx.shift_fraction;
903       sctx.shift_fraction = (sctx.shift_hi+sctx.shift_lo)/2.;
904       sctx.shift_amount   = sctx.shift_fraction * sctx.shift_top;
905       sctx.newshift       = PETSC_TRUE;
906       sctx.nshift++;
907     }
908   } while (sctx.newshift);
909 
910   /* invert diagonal entries for simplier triangular solves */
911   for (i=0; i<n; i++) {
912     a->a[diag[r[i]]] = 1.0/a->a[diag[r[i]]];
913   }
914 
915   ierr = PetscFree(rtmp);CHKERRQ(ierr);
916   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
917   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
918 
919   A->ops->solve             = MatSolve_SeqAIJ_InplaceWithPerm;
920   A->ops->solveadd          = MatSolveAdd_SeqAIJ_inplace;
921   A->ops->solvetranspose    = MatSolveTranspose_SeqAIJ_inplace;
922   A->ops->solvetransposeadd = MatSolveTransposeAdd_SeqAIJ_inplace;
923 
924   A->assembled    = PETSC_TRUE;
925   A->preallocated = PETSC_TRUE;
926 
927   ierr = PetscLogFlops(A->cmap->n);CHKERRQ(ierr);
928   if (sctx.nshift) {
929     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) {
930       ierr = PetscInfo4(A,"number of shift_pd tries %D, shift_amount %g, diagonal shifted up by %e fraction top_value %e\n",sctx.nshift,(double)sctx.shift_amount,(double)sctx.shift_fraction,(double)sctx.shift_top);CHKERRQ(ierr);
931     } else if (info->shifttype == (PetscReal)MAT_SHIFT_NONZERO) {
932       ierr = PetscInfo2(A,"number of shift_nz tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
933     }
934   }
935   PetscFunctionReturn(0);
936 }
937 
938 /* ----------------------------------------------------------- */
939 PetscErrorCode MatLUFactor_SeqAIJ(Mat A,IS row,IS col,const MatFactorInfo *info)
940 {
941   PetscErrorCode ierr;
942   Mat            C;
943 
944   PetscFunctionBegin;
945   ierr = MatGetFactor(A,MATSOLVERPETSC,MAT_FACTOR_LU,&C);CHKERRQ(ierr);
946   ierr = MatLUFactorSymbolic(C,A,row,col,info);CHKERRQ(ierr);
947   ierr = MatLUFactorNumeric(C,A,info);CHKERRQ(ierr);
948 
949   A->ops->solve          = C->ops->solve;
950   A->ops->solvetranspose = C->ops->solvetranspose;
951 
952   ierr = MatHeaderMerge(A,&C);CHKERRQ(ierr);
953   ierr = PetscLogObjectParent((PetscObject)A,(PetscObject)((Mat_SeqAIJ*)(A->data))->icol);CHKERRQ(ierr);
954   PetscFunctionReturn(0);
955 }
956 /* ----------------------------------------------------------- */
957 
958 
959 PetscErrorCode MatSolve_SeqAIJ_inplace(Mat A,Vec bb,Vec xx)
960 {
961   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
962   IS                iscol = a->col,isrow = a->row;
963   PetscErrorCode    ierr;
964   PetscInt          i, n = A->rmap->n,*vi,*ai = a->i,*aj = a->j;
965   PetscInt          nz;
966   const PetscInt    *rout,*cout,*r,*c;
967   PetscScalar       *x,*tmp,*tmps,sum;
968   const PetscScalar *b;
969   const MatScalar   *aa = a->a,*v;
970 
971   PetscFunctionBegin;
972   if (!n) PetscFunctionReturn(0);
973 
974   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
975   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
976   tmp  = a->solve_work;
977 
978   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
979   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout + (n-1);
980 
981   /* forward solve the lower triangular */
982   tmp[0] = b[*r++];
983   tmps   = tmp;
984   for (i=1; i<n; i++) {
985     v   = aa + ai[i];
986     vi  = aj + ai[i];
987     nz  = a->diag[i] - ai[i];
988     sum = b[*r++];
989     PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
990     tmp[i] = sum;
991   }
992 
993   /* backward solve the upper triangular */
994   for (i=n-1; i>=0; i--) {
995     v   = aa + a->diag[i] + 1;
996     vi  = aj + a->diag[i] + 1;
997     nz  = ai[i+1] - a->diag[i] - 1;
998     sum = tmp[i];
999     PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
1000     x[*c--] = tmp[i] = sum*aa[a->diag[i]];
1001   }
1002 
1003   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1004   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1005   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1006   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
1007   ierr = PetscLogFlops(2.0*a->nz - A->cmap->n);CHKERRQ(ierr);
1008   PetscFunctionReturn(0);
1009 }
1010 
1011 PetscErrorCode MatMatSolve_SeqAIJ_inplace(Mat A,Mat B,Mat X)
1012 {
1013   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1014   IS                iscol = a->col,isrow = a->row;
1015   PetscErrorCode    ierr;
1016   PetscInt          i, n = A->rmap->n,*vi,*ai = a->i,*aj = a->j;
1017   PetscInt          nz,neq;
1018   const PetscInt    *rout,*cout,*r,*c;
1019   PetscScalar       *x,*tmp,*tmps,sum;
1020   const PetscScalar *aa = a->a,*v;
1021   const PetscScalar *b;
1022   PetscBool         bisdense,xisdense;
1023 
1024   PetscFunctionBegin;
1025   if (!n) PetscFunctionReturn(0);
1026 
1027   ierr = PetscObjectTypeCompare((PetscObject)B,MATSEQDENSE,&bisdense);CHKERRQ(ierr);
1028   if (!bisdense) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"B matrix must be a SeqDense matrix");
1029   ierr = PetscObjectTypeCompare((PetscObject)X,MATSEQDENSE,&xisdense);CHKERRQ(ierr);
1030   if (!xisdense) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"X matrix must be a SeqDense matrix");
1031 
1032   ierr = MatDenseGetArrayRead(B,&b);CHKERRQ(ierr);
1033   ierr = MatDenseGetArray(X,&x);CHKERRQ(ierr);
1034 
1035   tmp  = a->solve_work;
1036   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1037   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1038 
1039   for (neq=0; neq<B->cmap->n; neq++) {
1040     /* forward solve the lower triangular */
1041     tmp[0] = b[r[0]];
1042     tmps   = tmp;
1043     for (i=1; i<n; i++) {
1044       v   = aa + ai[i];
1045       vi  = aj + ai[i];
1046       nz  = a->diag[i] - ai[i];
1047       sum = b[r[i]];
1048       PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
1049       tmp[i] = sum;
1050     }
1051     /* backward solve the upper triangular */
1052     for (i=n-1; i>=0; i--) {
1053       v   = aa + a->diag[i] + 1;
1054       vi  = aj + a->diag[i] + 1;
1055       nz  = ai[i+1] - a->diag[i] - 1;
1056       sum = tmp[i];
1057       PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
1058       x[c[i]] = tmp[i] = sum*aa[a->diag[i]];
1059     }
1060 
1061     b += n;
1062     x += n;
1063   }
1064   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1065   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1066   ierr = MatDenseRestoreArrayRead(B,&b);CHKERRQ(ierr);
1067   ierr = MatDenseRestoreArray(X,&x);CHKERRQ(ierr);
1068   ierr = PetscLogFlops(B->cmap->n*(2.0*a->nz - n));CHKERRQ(ierr);
1069   PetscFunctionReturn(0);
1070 }
1071 
1072 PetscErrorCode MatMatSolve_SeqAIJ(Mat A,Mat B,Mat X)
1073 {
1074   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1075   IS                iscol = a->col,isrow = a->row;
1076   PetscErrorCode    ierr;
1077   PetscInt          i, n = A->rmap->n,*vi,*ai = a->i,*aj = a->j,*adiag = a->diag;
1078   PetscInt          nz,neq;
1079   const PetscInt    *rout,*cout,*r,*c;
1080   PetscScalar       *x,*tmp,sum;
1081   const PetscScalar *b;
1082   const PetscScalar *aa = a->a,*v;
1083   PetscBool         bisdense,xisdense;
1084 
1085   PetscFunctionBegin;
1086   if (!n) PetscFunctionReturn(0);
1087 
1088   ierr = PetscObjectTypeCompare((PetscObject)B,MATSEQDENSE,&bisdense);CHKERRQ(ierr);
1089   if (!bisdense) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"B matrix must be a SeqDense matrix");
1090   ierr = PetscObjectTypeCompare((PetscObject)X,MATSEQDENSE,&xisdense);CHKERRQ(ierr);
1091   if (!xisdense) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"X matrix must be a SeqDense matrix");
1092 
1093   ierr = MatDenseGetArrayRead(B,&b);CHKERRQ(ierr);
1094   ierr = MatDenseGetArray(X,&x);CHKERRQ(ierr);
1095 
1096   tmp  = a->solve_work;
1097   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1098   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1099 
1100   for (neq=0; neq<B->cmap->n; neq++) {
1101     /* forward solve the lower triangular */
1102     tmp[0] = b[r[0]];
1103     v      = aa;
1104     vi     = aj;
1105     for (i=1; i<n; i++) {
1106       nz  = ai[i+1] - ai[i];
1107       sum = b[r[i]];
1108       PetscSparseDenseMinusDot(sum,tmp,v,vi,nz);
1109       tmp[i] = sum;
1110       v     += nz; vi += nz;
1111     }
1112 
1113     /* backward solve the upper triangular */
1114     for (i=n-1; i>=0; i--) {
1115       v   = aa + adiag[i+1]+1;
1116       vi  = aj + adiag[i+1]+1;
1117       nz  = adiag[i]-adiag[i+1]-1;
1118       sum = tmp[i];
1119       PetscSparseDenseMinusDot(sum,tmp,v,vi,nz);
1120       x[c[i]] = tmp[i] = sum*v[nz]; /* v[nz] = aa[adiag[i]] */
1121     }
1122 
1123     b += n;
1124     x += n;
1125   }
1126   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1127   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1128   ierr = MatDenseRestoreArrayRead(B,&b);CHKERRQ(ierr);
1129   ierr = MatDenseRestoreArray(X,&x);CHKERRQ(ierr);
1130   ierr = PetscLogFlops(B->cmap->n*(2.0*a->nz - n));CHKERRQ(ierr);
1131   PetscFunctionReturn(0);
1132 }
1133 
1134 PetscErrorCode MatSolve_SeqAIJ_InplaceWithPerm(Mat A,Vec bb,Vec xx)
1135 {
1136   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1137   IS                iscol = a->col,isrow = a->row;
1138   PetscErrorCode    ierr;
1139   const PetscInt    *r,*c,*rout,*cout;
1140   PetscInt          i, n = A->rmap->n,*vi,*ai = a->i,*aj = a->j;
1141   PetscInt          nz,row;
1142   PetscScalar       *x,*tmp,*tmps,sum;
1143   const PetscScalar *b;
1144   const MatScalar   *aa = a->a,*v;
1145 
1146   PetscFunctionBegin;
1147   if (!n) PetscFunctionReturn(0);
1148 
1149   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1150   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
1151   tmp  = a->solve_work;
1152 
1153   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1154   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout + (n-1);
1155 
1156   /* forward solve the lower triangular */
1157   tmp[0] = b[*r++];
1158   tmps   = tmp;
1159   for (row=1; row<n; row++) {
1160     i   = rout[row]; /* permuted row */
1161     v   = aa + ai[i];
1162     vi  = aj + ai[i];
1163     nz  = a->diag[i] - ai[i];
1164     sum = b[*r++];
1165     PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
1166     tmp[row] = sum;
1167   }
1168 
1169   /* backward solve the upper triangular */
1170   for (row=n-1; row>=0; row--) {
1171     i   = rout[row]; /* permuted row */
1172     v   = aa + a->diag[i] + 1;
1173     vi  = aj + a->diag[i] + 1;
1174     nz  = ai[i+1] - a->diag[i] - 1;
1175     sum = tmp[row];
1176     PetscSparseDenseMinusDot(sum,tmps,v,vi,nz);
1177     x[*c--] = tmp[row] = sum*aa[a->diag[i]];
1178   }
1179 
1180   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1181   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1182   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1183   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
1184   ierr = PetscLogFlops(2.0*a->nz - A->cmap->n);CHKERRQ(ierr);
1185   PetscFunctionReturn(0);
1186 }
1187 
1188 /* ----------------------------------------------------------- */
1189 #include <../src/mat/impls/aij/seq/ftn-kernels/fsolve.h>
1190 PetscErrorCode MatSolve_SeqAIJ_NaturalOrdering_inplace(Mat A,Vec bb,Vec xx)
1191 {
1192   Mat_SeqAIJ        *a = (Mat_SeqAIJ*)A->data;
1193   PetscErrorCode    ierr;
1194   PetscInt          n   = A->rmap->n;
1195   const PetscInt    *ai = a->i,*aj = a->j,*adiag = a->diag;
1196   PetscScalar       *x;
1197   const PetscScalar *b;
1198   const MatScalar   *aa = a->a;
1199 #if !defined(PETSC_USE_FORTRAN_KERNEL_SOLVEAIJ)
1200   PetscInt        adiag_i,i,nz,ai_i;
1201   const PetscInt  *vi;
1202   const MatScalar *v;
1203   PetscScalar     sum;
1204 #endif
1205 
1206   PetscFunctionBegin;
1207   if (!n) PetscFunctionReturn(0);
1208 
1209   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1210   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
1211 
1212 #if defined(PETSC_USE_FORTRAN_KERNEL_SOLVEAIJ)
1213   fortransolveaij_(&n,x,ai,aj,adiag,aa,b);
1214 #else
1215   /* forward solve the lower triangular */
1216   x[0] = b[0];
1217   for (i=1; i<n; i++) {
1218     ai_i = ai[i];
1219     v    = aa + ai_i;
1220     vi   = aj + ai_i;
1221     nz   = adiag[i] - ai_i;
1222     sum  = b[i];
1223     PetscSparseDenseMinusDot(sum,x,v,vi,nz);
1224     x[i] = sum;
1225   }
1226 
1227   /* backward solve the upper triangular */
1228   for (i=n-1; i>=0; i--) {
1229     adiag_i = adiag[i];
1230     v       = aa + adiag_i + 1;
1231     vi      = aj + adiag_i + 1;
1232     nz      = ai[i+1] - adiag_i - 1;
1233     sum     = x[i];
1234     PetscSparseDenseMinusDot(sum,x,v,vi,nz);
1235     x[i] = sum*aa[adiag_i];
1236   }
1237 #endif
1238   ierr = PetscLogFlops(2.0*a->nz - A->cmap->n);CHKERRQ(ierr);
1239   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1240   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
1241   PetscFunctionReturn(0);
1242 }
1243 
1244 PetscErrorCode MatSolveAdd_SeqAIJ_inplace(Mat A,Vec bb,Vec yy,Vec xx)
1245 {
1246   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1247   IS                iscol = a->col,isrow = a->row;
1248   PetscErrorCode    ierr;
1249   PetscInt          i, n = A->rmap->n,j;
1250   PetscInt          nz;
1251   const PetscInt    *rout,*cout,*r,*c,*vi,*ai = a->i,*aj = a->j;
1252   PetscScalar       *x,*tmp,sum;
1253   const PetscScalar *b;
1254   const MatScalar   *aa = a->a,*v;
1255 
1256   PetscFunctionBegin;
1257   if (yy != xx) {ierr = VecCopy(yy,xx);CHKERRQ(ierr);}
1258 
1259   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1260   ierr = VecGetArray(xx,&x);CHKERRQ(ierr);
1261   tmp  = a->solve_work;
1262 
1263   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1264   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout + (n-1);
1265 
1266   /* forward solve the lower triangular */
1267   tmp[0] = b[*r++];
1268   for (i=1; i<n; i++) {
1269     v   = aa + ai[i];
1270     vi  = aj + ai[i];
1271     nz  = a->diag[i] - ai[i];
1272     sum = b[*r++];
1273     for (j=0; j<nz; j++) sum -= v[j]*tmp[vi[j]];
1274     tmp[i] = sum;
1275   }
1276 
1277   /* backward solve the upper triangular */
1278   for (i=n-1; i>=0; i--) {
1279     v   = aa + a->diag[i] + 1;
1280     vi  = aj + a->diag[i] + 1;
1281     nz  = ai[i+1] - a->diag[i] - 1;
1282     sum = tmp[i];
1283     for (j=0; j<nz; j++) sum -= v[j]*tmp[vi[j]];
1284     tmp[i]   = sum*aa[a->diag[i]];
1285     x[*c--] += tmp[i];
1286   }
1287 
1288   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1289   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1290   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1291   ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr);
1292   ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr);
1293   PetscFunctionReturn(0);
1294 }
1295 
1296 PetscErrorCode MatSolveAdd_SeqAIJ(Mat A,Vec bb,Vec yy,Vec xx)
1297 {
1298   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1299   IS                iscol = a->col,isrow = a->row;
1300   PetscErrorCode    ierr;
1301   PetscInt          i, n = A->rmap->n,j;
1302   PetscInt          nz;
1303   const PetscInt    *rout,*cout,*r,*c,*vi,*ai = a->i,*aj = a->j,*adiag = a->diag;
1304   PetscScalar       *x,*tmp,sum;
1305   const PetscScalar *b;
1306   const MatScalar   *aa = a->a,*v;
1307 
1308   PetscFunctionBegin;
1309   if (yy != xx) {ierr = VecCopy(yy,xx);CHKERRQ(ierr);}
1310 
1311   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1312   ierr = VecGetArray(xx,&x);CHKERRQ(ierr);
1313   tmp  = a->solve_work;
1314 
1315   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1316   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1317 
1318   /* forward solve the lower triangular */
1319   tmp[0] = b[r[0]];
1320   v      = aa;
1321   vi     = aj;
1322   for (i=1; i<n; i++) {
1323     nz  = ai[i+1] - ai[i];
1324     sum = b[r[i]];
1325     for (j=0; j<nz; j++) sum -= v[j]*tmp[vi[j]];
1326     tmp[i] = sum;
1327     v     += nz;
1328     vi    += nz;
1329   }
1330 
1331   /* backward solve the upper triangular */
1332   v  = aa + adiag[n-1];
1333   vi = aj + adiag[n-1];
1334   for (i=n-1; i>=0; i--) {
1335     nz  = adiag[i] - adiag[i+1] - 1;
1336     sum = tmp[i];
1337     for (j=0; j<nz; j++) sum -= v[j]*tmp[vi[j]];
1338     tmp[i]   = sum*v[nz];
1339     x[c[i]] += tmp[i];
1340     v       += nz+1; vi += nz+1;
1341   }
1342 
1343   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1344   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1345   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1346   ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr);
1347   ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr);
1348   PetscFunctionReturn(0);
1349 }
1350 
1351 PetscErrorCode MatSolveTranspose_SeqAIJ_inplace(Mat A,Vec bb,Vec xx)
1352 {
1353   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1354   IS                iscol = a->col,isrow = a->row;
1355   PetscErrorCode    ierr;
1356   const PetscInt    *rout,*cout,*r,*c,*diag = a->diag,*ai = a->i,*aj = a->j,*vi;
1357   PetscInt          i,n = A->rmap->n,j;
1358   PetscInt          nz;
1359   PetscScalar       *x,*tmp,s1;
1360   const MatScalar   *aa = a->a,*v;
1361   const PetscScalar *b;
1362 
1363   PetscFunctionBegin;
1364   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1365   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
1366   tmp  = a->solve_work;
1367 
1368   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1369   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1370 
1371   /* copy the b into temp work space according to permutation */
1372   for (i=0; i<n; i++) tmp[i] = b[c[i]];
1373 
1374   /* forward solve the U^T */
1375   for (i=0; i<n; i++) {
1376     v   = aa + diag[i];
1377     vi  = aj + diag[i] + 1;
1378     nz  = ai[i+1] - diag[i] - 1;
1379     s1  = tmp[i];
1380     s1 *= (*v++);  /* multiply by inverse of diagonal entry */
1381     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1382     tmp[i] = s1;
1383   }
1384 
1385   /* backward solve the L^T */
1386   for (i=n-1; i>=0; i--) {
1387     v  = aa + diag[i] - 1;
1388     vi = aj + diag[i] - 1;
1389     nz = diag[i] - ai[i];
1390     s1 = tmp[i];
1391     for (j=0; j>-nz; j--) tmp[vi[j]] -= s1*v[j];
1392   }
1393 
1394   /* copy tmp into x according to permutation */
1395   for (i=0; i<n; i++) x[r[i]] = tmp[i];
1396 
1397   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1398   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1399   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1400   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
1401 
1402   ierr = PetscLogFlops(2.0*a->nz-A->cmap->n);CHKERRQ(ierr);
1403   PetscFunctionReturn(0);
1404 }
1405 
1406 PetscErrorCode MatSolveTranspose_SeqAIJ(Mat A,Vec bb,Vec xx)
1407 {
1408   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1409   IS                iscol = a->col,isrow = a->row;
1410   PetscErrorCode    ierr;
1411   const PetscInt    *rout,*cout,*r,*c,*adiag = a->diag,*ai = a->i,*aj = a->j,*vi;
1412   PetscInt          i,n = A->rmap->n,j;
1413   PetscInt          nz;
1414   PetscScalar       *x,*tmp,s1;
1415   const MatScalar   *aa = a->a,*v;
1416   const PetscScalar *b;
1417 
1418   PetscFunctionBegin;
1419   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1420   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
1421   tmp  = a->solve_work;
1422 
1423   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1424   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1425 
1426   /* copy the b into temp work space according to permutation */
1427   for (i=0; i<n; i++) tmp[i] = b[c[i]];
1428 
1429   /* forward solve the U^T */
1430   for (i=0; i<n; i++) {
1431     v   = aa + adiag[i+1] + 1;
1432     vi  = aj + adiag[i+1] + 1;
1433     nz  = adiag[i] - adiag[i+1] - 1;
1434     s1  = tmp[i];
1435     s1 *= v[nz];  /* multiply by inverse of diagonal entry */
1436     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1437     tmp[i] = s1;
1438   }
1439 
1440   /* backward solve the L^T */
1441   for (i=n-1; i>=0; i--) {
1442     v  = aa + ai[i];
1443     vi = aj + ai[i];
1444     nz = ai[i+1] - ai[i];
1445     s1 = tmp[i];
1446     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1447   }
1448 
1449   /* copy tmp into x according to permutation */
1450   for (i=0; i<n; i++) x[r[i]] = tmp[i];
1451 
1452   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1453   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1454   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1455   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
1456 
1457   ierr = PetscLogFlops(2.0*a->nz-A->cmap->n);CHKERRQ(ierr);
1458   PetscFunctionReturn(0);
1459 }
1460 
1461 PetscErrorCode MatSolveTransposeAdd_SeqAIJ_inplace(Mat A,Vec bb,Vec zz,Vec xx)
1462 {
1463   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1464   IS                iscol = a->col,isrow = a->row;
1465   PetscErrorCode    ierr;
1466   const PetscInt    *rout,*cout,*r,*c,*diag = a->diag,*ai = a->i,*aj = a->j,*vi;
1467   PetscInt          i,n = A->rmap->n,j;
1468   PetscInt          nz;
1469   PetscScalar       *x,*tmp,s1;
1470   const MatScalar   *aa = a->a,*v;
1471   const PetscScalar *b;
1472 
1473   PetscFunctionBegin;
1474   if (zz != xx) {ierr = VecCopy(zz,xx);CHKERRQ(ierr);}
1475   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1476   ierr = VecGetArray(xx,&x);CHKERRQ(ierr);
1477   tmp  = a->solve_work;
1478 
1479   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1480   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1481 
1482   /* copy the b into temp work space according to permutation */
1483   for (i=0; i<n; i++) tmp[i] = b[c[i]];
1484 
1485   /* forward solve the U^T */
1486   for (i=0; i<n; i++) {
1487     v   = aa + diag[i];
1488     vi  = aj + diag[i] + 1;
1489     nz  = ai[i+1] - diag[i] - 1;
1490     s1  = tmp[i];
1491     s1 *= (*v++);  /* multiply by inverse of diagonal entry */
1492     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1493     tmp[i] = s1;
1494   }
1495 
1496   /* backward solve the L^T */
1497   for (i=n-1; i>=0; i--) {
1498     v  = aa + diag[i] - 1;
1499     vi = aj + diag[i] - 1;
1500     nz = diag[i] - ai[i];
1501     s1 = tmp[i];
1502     for (j=0; j>-nz; j--) tmp[vi[j]] -= s1*v[j];
1503   }
1504 
1505   /* copy tmp into x according to permutation */
1506   for (i=0; i<n; i++) x[r[i]] += tmp[i];
1507 
1508   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1509   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1510   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1511   ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr);
1512 
1513   ierr = PetscLogFlops(2.0*a->nz-A->cmap->n);CHKERRQ(ierr);
1514   PetscFunctionReturn(0);
1515 }
1516 
1517 PetscErrorCode MatSolveTransposeAdd_SeqAIJ(Mat A,Vec bb,Vec zz,Vec xx)
1518 {
1519   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
1520   IS                iscol = a->col,isrow = a->row;
1521   PetscErrorCode    ierr;
1522   const PetscInt    *rout,*cout,*r,*c,*adiag = a->diag,*ai = a->i,*aj = a->j,*vi;
1523   PetscInt          i,n = A->rmap->n,j;
1524   PetscInt          nz;
1525   PetscScalar       *x,*tmp,s1;
1526   const MatScalar   *aa = a->a,*v;
1527   const PetscScalar *b;
1528 
1529   PetscFunctionBegin;
1530   if (zz != xx) {ierr = VecCopy(zz,xx);CHKERRQ(ierr);}
1531   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
1532   ierr = VecGetArray(xx,&x);CHKERRQ(ierr);
1533   tmp  = a->solve_work;
1534 
1535   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
1536   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
1537 
1538   /* copy the b into temp work space according to permutation */
1539   for (i=0; i<n; i++) tmp[i] = b[c[i]];
1540 
1541   /* forward solve the U^T */
1542   for (i=0; i<n; i++) {
1543     v   = aa + adiag[i+1] + 1;
1544     vi  = aj + adiag[i+1] + 1;
1545     nz  = adiag[i] - adiag[i+1] - 1;
1546     s1  = tmp[i];
1547     s1 *= v[nz];  /* multiply by inverse of diagonal entry */
1548     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1549     tmp[i] = s1;
1550   }
1551 
1552 
1553   /* backward solve the L^T */
1554   for (i=n-1; i>=0; i--) {
1555     v  = aa + ai[i];
1556     vi = aj + ai[i];
1557     nz = ai[i+1] - ai[i];
1558     s1 = tmp[i];
1559     for (j=0; j<nz; j++) tmp[vi[j]] -= s1*v[j];
1560   }
1561 
1562   /* copy tmp into x according to permutation */
1563   for (i=0; i<n; i++) x[r[i]] += tmp[i];
1564 
1565   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
1566   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
1567   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
1568   ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr);
1569 
1570   ierr = PetscLogFlops(2.0*a->nz-A->cmap->n);CHKERRQ(ierr);
1571   PetscFunctionReturn(0);
1572 }
1573 
1574 /* ----------------------------------------------------------------*/
1575 
1576 /*
1577    ilu() under revised new data structure.
1578    Factored arrays bj and ba are stored as
1579      L(0,:), L(1,:), ...,L(n-1,:),  U(n-1,:),...,U(i,:),U(i-1,:),...,U(0,:)
1580 
1581    bi=fact->i is an array of size n+1, in which
1582    bi+
1583      bi[i]:  points to 1st entry of L(i,:),i=0,...,n-1
1584      bi[n]:  points to L(n-1,n-1)+1
1585 
1586   bdiag=fact->diag is an array of size n+1,in which
1587      bdiag[i]: points to diagonal of U(i,:), i=0,...,n-1
1588      bdiag[n]: points to entry of U(n-1,0)-1
1589 
1590    U(i,:) contains bdiag[i] as its last entry, i.e.,
1591     U(i,:) = (u[i,i+1],...,u[i,n-1],diag[i])
1592 */
1593 PetscErrorCode MatILUFactorSymbolic_SeqAIJ_ilu0(Mat fact,Mat A,IS isrow,IS iscol,const MatFactorInfo *info)
1594 {
1595   Mat_SeqAIJ     *a = (Mat_SeqAIJ*)A->data,*b;
1596   PetscErrorCode ierr;
1597   const PetscInt n=A->rmap->n,*ai=a->i,*aj,*adiag=a->diag;
1598   PetscInt       i,j,k=0,nz,*bi,*bj,*bdiag;
1599   IS             isicol;
1600 
1601   PetscFunctionBegin;
1602   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
1603   ierr = MatDuplicateNoCreate_SeqAIJ(fact,A,MAT_DO_NOT_COPY_VALUES,PETSC_FALSE);CHKERRQ(ierr);
1604   b    = (Mat_SeqAIJ*)(fact)->data;
1605 
1606   /* allocate matrix arrays for new data structure */
1607   ierr = PetscMalloc3(ai[n]+1,&b->a,ai[n]+1,&b->j,n+1,&b->i);CHKERRQ(ierr);
1608   ierr = PetscLogObjectMemory((PetscObject)fact,ai[n]*(sizeof(PetscScalar)+sizeof(PetscInt))+(n+1)*sizeof(PetscInt));CHKERRQ(ierr);
1609 
1610   b->singlemalloc = PETSC_TRUE;
1611   if (!b->diag) {
1612     ierr = PetscMalloc1(n+1,&b->diag);CHKERRQ(ierr);
1613     ierr = PetscLogObjectMemory((PetscObject)fact,(n+1)*sizeof(PetscInt));CHKERRQ(ierr);
1614   }
1615   bdiag = b->diag;
1616 
1617   if (n > 0) {
1618     ierr = PetscArrayzero(b->a,ai[n]);CHKERRQ(ierr);
1619   }
1620 
1621   /* set bi and bj with new data structure */
1622   bi = b->i;
1623   bj = b->j;
1624 
1625   /* L part */
1626   bi[0] = 0;
1627   for (i=0; i<n; i++) {
1628     nz      = adiag[i] - ai[i];
1629     bi[i+1] = bi[i] + nz;
1630     aj      = a->j + ai[i];
1631     for (j=0; j<nz; j++) {
1632       /*   *bj = aj[j]; bj++; */
1633       bj[k++] = aj[j];
1634     }
1635   }
1636 
1637   /* U part */
1638   bdiag[n] = bi[n]-1;
1639   for (i=n-1; i>=0; i--) {
1640     nz = ai[i+1] - adiag[i] - 1;
1641     aj = a->j + adiag[i] + 1;
1642     for (j=0; j<nz; j++) {
1643       /*      *bj = aj[j]; bj++; */
1644       bj[k++] = aj[j];
1645     }
1646     /* diag[i] */
1647     /*    *bj = i; bj++; */
1648     bj[k++]  = i;
1649     bdiag[i] = bdiag[i+1] + nz + 1;
1650   }
1651 
1652   fact->factortype             = MAT_FACTOR_ILU;
1653   fact->info.factor_mallocs    = 0;
1654   fact->info.fill_ratio_given  = info->fill;
1655   fact->info.fill_ratio_needed = 1.0;
1656   fact->ops->lufactornumeric   = MatLUFactorNumeric_SeqAIJ;
1657   ierr = MatSeqAIJCheckInode_FactorLU(fact);CHKERRQ(ierr);
1658 
1659   b       = (Mat_SeqAIJ*)(fact)->data;
1660   b->row  = isrow;
1661   b->col  = iscol;
1662   b->icol = isicol;
1663   ierr    = PetscMalloc1(fact->rmap->n+1,&b->solve_work);CHKERRQ(ierr);
1664   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
1665   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
1666   PetscFunctionReturn(0);
1667 }
1668 
1669 PetscErrorCode MatILUFactorSymbolic_SeqAIJ(Mat fact,Mat A,IS isrow,IS iscol,const MatFactorInfo *info)
1670 {
1671   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data,*b;
1672   IS                 isicol;
1673   PetscErrorCode     ierr;
1674   const PetscInt     *r,*ic;
1675   PetscInt           n=A->rmap->n,*ai=a->i,*aj=a->j;
1676   PetscInt           *bi,*cols,nnz,*cols_lvl;
1677   PetscInt           *bdiag,prow,fm,nzbd,reallocs=0,dcount=0;
1678   PetscInt           i,levels,diagonal_fill;
1679   PetscBool          col_identity,row_identity,missing;
1680   PetscReal          f;
1681   PetscInt           nlnk,*lnk,*lnk_lvl=NULL;
1682   PetscBT            lnkbt;
1683   PetscInt           nzi,*bj,**bj_ptr,**bjlvl_ptr;
1684   PetscFreeSpaceList free_space    =NULL,current_space=NULL;
1685   PetscFreeSpaceList free_space_lvl=NULL,current_space_lvl=NULL;
1686 
1687   PetscFunctionBegin;
1688   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
1689   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
1690   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
1691 
1692   levels = (PetscInt)info->levels;
1693   ierr   = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
1694   ierr   = ISIdentity(iscol,&col_identity);CHKERRQ(ierr);
1695   if (!levels && row_identity && col_identity) {
1696     /* special case: ilu(0) with natural ordering */
1697     ierr = MatILUFactorSymbolic_SeqAIJ_ilu0(fact,A,isrow,iscol,info);CHKERRQ(ierr);
1698     if (a->inode.size) {
1699       fact->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode;
1700     }
1701     PetscFunctionReturn(0);
1702   }
1703 
1704   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
1705   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
1706   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
1707 
1708   /* get new row and diagonal pointers, must be allocated separately because they will be given to the Mat_SeqAIJ and freed separately */
1709   ierr  = PetscMalloc1(n+1,&bi);CHKERRQ(ierr);
1710   ierr  = PetscMalloc1(n+1,&bdiag);CHKERRQ(ierr);
1711   bi[0] = bdiag[0] = 0;
1712   ierr  = PetscMalloc2(n,&bj_ptr,n,&bjlvl_ptr);CHKERRQ(ierr);
1713 
1714   /* create a linked list for storing column indices of the active row */
1715   nlnk = n + 1;
1716   ierr = PetscIncompleteLLCreate(n,n,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
1717 
1718   /* initial FreeSpace size is f*(ai[n]+1) */
1719   f                 = info->fill;
1720   diagonal_fill     = (PetscInt)info->diagonal_fill;
1721   ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space);CHKERRQ(ierr);
1722   current_space     = free_space;
1723   ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space_lvl);CHKERRQ(ierr);
1724   current_space_lvl = free_space_lvl;
1725   for (i=0; i<n; i++) {
1726     nzi = 0;
1727     /* copy current row into linked list */
1728     nnz = ai[r[i]+1] - ai[r[i]];
1729     if (!nnz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",r[i],i);
1730     cols   = aj + ai[r[i]];
1731     lnk[i] = -1; /* marker to indicate if diagonal exists */
1732     ierr   = PetscIncompleteLLInit(nnz,cols,n,ic,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
1733     nzi   += nlnk;
1734 
1735     /* make sure diagonal entry is included */
1736     if (diagonal_fill && lnk[i] == -1) {
1737       fm = n;
1738       while (lnk[fm] < i) fm = lnk[fm];
1739       lnk[i]     = lnk[fm]; /* insert diagonal into linked list */
1740       lnk[fm]    = i;
1741       lnk_lvl[i] = 0;
1742       nzi++; dcount++;
1743     }
1744 
1745     /* add pivot rows into the active row */
1746     nzbd = 0;
1747     prow = lnk[n];
1748     while (prow < i) {
1749       nnz      = bdiag[prow];
1750       cols     = bj_ptr[prow] + nnz + 1;
1751       cols_lvl = bjlvl_ptr[prow] + nnz + 1;
1752       nnz      = bi[prow+1] - bi[prow] - nnz - 1;
1753       ierr     = PetscILULLAddSorted(nnz,cols,levels,cols_lvl,prow,nlnk,lnk,lnk_lvl,lnkbt,prow);CHKERRQ(ierr);
1754       nzi     += nlnk;
1755       prow     = lnk[prow];
1756       nzbd++;
1757     }
1758     bdiag[i] = nzbd;
1759     bi[i+1]  = bi[i] + nzi;
1760     /* if free space is not available, make more free space */
1761     if (current_space->local_remaining<nzi) {
1762       nnz  = PetscIntMultTruncate(2,PetscIntMultTruncate(nzi,n - i)); /* estimated and max additional space needed */
1763       ierr = PetscFreeSpaceGet(nnz,&current_space);CHKERRQ(ierr);
1764       ierr = PetscFreeSpaceGet(nnz,&current_space_lvl);CHKERRQ(ierr);
1765       reallocs++;
1766     }
1767 
1768     /* copy data into free_space and free_space_lvl, then initialize lnk */
1769     ierr         = PetscIncompleteLLClean(n,n,nzi,lnk,lnk_lvl,current_space->array,current_space_lvl->array,lnkbt);CHKERRQ(ierr);
1770     bj_ptr[i]    = current_space->array;
1771     bjlvl_ptr[i] = current_space_lvl->array;
1772 
1773     /* make sure the active row i has diagonal entry */
1774     if (*(bj_ptr[i]+bdiag[i]) != i) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Row %D has missing diagonal in factored matrix\ntry running with -pc_factor_nonzeros_along_diagonal or -pc_factor_diagonal_fill",i);
1775 
1776     current_space->array               += nzi;
1777     current_space->local_used          += nzi;
1778     current_space->local_remaining     -= nzi;
1779     current_space_lvl->array           += nzi;
1780     current_space_lvl->local_used      += nzi;
1781     current_space_lvl->local_remaining -= nzi;
1782   }
1783 
1784   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
1785   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
1786   /* copy free_space into bj and free free_space; set bi, bj, bdiag in new datastructure; */
1787   ierr = PetscMalloc1(bi[n]+1,&bj);CHKERRQ(ierr);
1788   ierr = PetscFreeSpaceContiguous_LU(&free_space,bj,n,bi,bdiag);CHKERRQ(ierr);
1789 
1790   ierr = PetscIncompleteLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
1791   ierr = PetscFreeSpaceDestroy(free_space_lvl);CHKERRQ(ierr);
1792   ierr = PetscFree2(bj_ptr,bjlvl_ptr);CHKERRQ(ierr);
1793 
1794 #if defined(PETSC_USE_INFO)
1795   {
1796     PetscReal af = ((PetscReal)(bdiag[0]+1))/((PetscReal)ai[n]);
1797     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)f,(double)af);CHKERRQ(ierr);
1798     ierr = PetscInfo1(A,"Run with -[sub_]pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
1799     ierr = PetscInfo1(A,"PCFactorSetFill([sub]pc,%g);\n",(double)af);CHKERRQ(ierr);
1800     ierr = PetscInfo(A,"for best performance.\n");CHKERRQ(ierr);
1801     if (diagonal_fill) {
1802       ierr = PetscInfo1(A,"Detected and replaced %D missing diagonals\n",dcount);CHKERRQ(ierr);
1803     }
1804   }
1805 #endif
1806   /* put together the new matrix */
1807   ierr = MatSeqAIJSetPreallocation_SeqAIJ(fact,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
1808   ierr = PetscLogObjectParent((PetscObject)fact,(PetscObject)isicol);CHKERRQ(ierr);
1809   b    = (Mat_SeqAIJ*)(fact)->data;
1810 
1811   b->free_a       = PETSC_TRUE;
1812   b->free_ij      = PETSC_TRUE;
1813   b->singlemalloc = PETSC_FALSE;
1814 
1815   ierr = PetscMalloc1(bdiag[0]+1,&b->a);CHKERRQ(ierr);
1816 
1817   b->j    = bj;
1818   b->i    = bi;
1819   b->diag = bdiag;
1820   b->ilen = 0;
1821   b->imax = 0;
1822   b->row  = isrow;
1823   b->col  = iscol;
1824   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
1825   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
1826   b->icol = isicol;
1827 
1828   ierr = PetscMalloc1(n+1,&b->solve_work);CHKERRQ(ierr);
1829   /* In b structure:  Free imax, ilen, old a, old j.
1830      Allocate bdiag, solve_work, new a, new j */
1831   ierr     = PetscLogObjectMemory((PetscObject)fact,(bdiag[0]+1)*(sizeof(PetscInt)+sizeof(PetscScalar)));CHKERRQ(ierr);
1832   b->maxnz = b->nz = bdiag[0]+1;
1833 
1834   (fact)->info.factor_mallocs    = reallocs;
1835   (fact)->info.fill_ratio_given  = f;
1836   (fact)->info.fill_ratio_needed = ((PetscReal)(bdiag[0]+1))/((PetscReal)ai[n]);
1837   (fact)->ops->lufactornumeric   = MatLUFactorNumeric_SeqAIJ;
1838   if (a->inode.size) {
1839     (fact)->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode;
1840   }
1841   ierr = MatSeqAIJCheckInode_FactorLU(fact);CHKERRQ(ierr);
1842   PetscFunctionReturn(0);
1843 }
1844 
1845 PetscErrorCode MatILUFactorSymbolic_SeqAIJ_inplace(Mat fact,Mat A,IS isrow,IS iscol,const MatFactorInfo *info)
1846 {
1847   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data,*b;
1848   IS                 isicol;
1849   PetscErrorCode     ierr;
1850   const PetscInt     *r,*ic;
1851   PetscInt           n=A->rmap->n,*ai=a->i,*aj=a->j;
1852   PetscInt           *bi,*cols,nnz,*cols_lvl;
1853   PetscInt           *bdiag,prow,fm,nzbd,reallocs=0,dcount=0;
1854   PetscInt           i,levels,diagonal_fill;
1855   PetscBool          col_identity,row_identity;
1856   PetscReal          f;
1857   PetscInt           nlnk,*lnk,*lnk_lvl=NULL;
1858   PetscBT            lnkbt;
1859   PetscInt           nzi,*bj,**bj_ptr,**bjlvl_ptr;
1860   PetscFreeSpaceList free_space    =NULL,current_space=NULL;
1861   PetscFreeSpaceList free_space_lvl=NULL,current_space_lvl=NULL;
1862   PetscBool          missing;
1863 
1864   PetscFunctionBegin;
1865   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
1866   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
1867   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
1868 
1869   f             = info->fill;
1870   levels        = (PetscInt)info->levels;
1871   diagonal_fill = (PetscInt)info->diagonal_fill;
1872 
1873   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
1874 
1875   ierr = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
1876   ierr = ISIdentity(iscol,&col_identity);CHKERRQ(ierr);
1877   if (!levels && row_identity && col_identity) { /* special case: ilu(0) with natural ordering */
1878     ierr = MatDuplicateNoCreate_SeqAIJ(fact,A,MAT_DO_NOT_COPY_VALUES,PETSC_TRUE);CHKERRQ(ierr);
1879 
1880     (fact)->ops->lufactornumeric =  MatLUFactorNumeric_SeqAIJ_inplace;
1881     if (a->inode.size) {
1882       (fact)->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode_inplace;
1883     }
1884     fact->factortype               = MAT_FACTOR_ILU;
1885     (fact)->info.factor_mallocs    = 0;
1886     (fact)->info.fill_ratio_given  = info->fill;
1887     (fact)->info.fill_ratio_needed = 1.0;
1888 
1889     b    = (Mat_SeqAIJ*)(fact)->data;
1890     b->row  = isrow;
1891     b->col  = iscol;
1892     b->icol = isicol;
1893     ierr    = PetscMalloc1((fact)->rmap->n+1,&b->solve_work);CHKERRQ(ierr);
1894     ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
1895     ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
1896     PetscFunctionReturn(0);
1897   }
1898 
1899   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
1900   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
1901 
1902   /* get new row and diagonal pointers, must be allocated separately because they will be given to the Mat_SeqAIJ and freed separately */
1903   ierr  = PetscMalloc1(n+1,&bi);CHKERRQ(ierr);
1904   ierr  = PetscMalloc1(n+1,&bdiag);CHKERRQ(ierr);
1905   bi[0] = bdiag[0] = 0;
1906 
1907   ierr = PetscMalloc2(n,&bj_ptr,n,&bjlvl_ptr);CHKERRQ(ierr);
1908 
1909   /* create a linked list for storing column indices of the active row */
1910   nlnk = n + 1;
1911   ierr = PetscIncompleteLLCreate(n,n,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
1912 
1913   /* initial FreeSpace size is f*(ai[n]+1) */
1914   ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space);CHKERRQ(ierr);
1915   current_space     = free_space;
1916   ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(f,ai[n]+1),&free_space_lvl);CHKERRQ(ierr);
1917   current_space_lvl = free_space_lvl;
1918 
1919   for (i=0; i<n; i++) {
1920     nzi = 0;
1921     /* copy current row into linked list */
1922     nnz = ai[r[i]+1] - ai[r[i]];
1923     if (!nnz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",r[i],i);
1924     cols   = aj + ai[r[i]];
1925     lnk[i] = -1; /* marker to indicate if diagonal exists */
1926     ierr   = PetscIncompleteLLInit(nnz,cols,n,ic,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
1927     nzi   += nlnk;
1928 
1929     /* make sure diagonal entry is included */
1930     if (diagonal_fill && lnk[i] == -1) {
1931       fm = n;
1932       while (lnk[fm] < i) fm = lnk[fm];
1933       lnk[i]     = lnk[fm]; /* insert diagonal into linked list */
1934       lnk[fm]    = i;
1935       lnk_lvl[i] = 0;
1936       nzi++; dcount++;
1937     }
1938 
1939     /* add pivot rows into the active row */
1940     nzbd = 0;
1941     prow = lnk[n];
1942     while (prow < i) {
1943       nnz      = bdiag[prow];
1944       cols     = bj_ptr[prow] + nnz + 1;
1945       cols_lvl = bjlvl_ptr[prow] + nnz + 1;
1946       nnz      = bi[prow+1] - bi[prow] - nnz - 1;
1947       ierr     = PetscILULLAddSorted(nnz,cols,levels,cols_lvl,prow,nlnk,lnk,lnk_lvl,lnkbt,prow);CHKERRQ(ierr);
1948       nzi     += nlnk;
1949       prow     = lnk[prow];
1950       nzbd++;
1951     }
1952     bdiag[i] = nzbd;
1953     bi[i+1]  = bi[i] + nzi;
1954 
1955     /* if free space is not available, make more free space */
1956     if (current_space->local_remaining<nzi) {
1957       nnz  = PetscIntMultTruncate(nzi,n - i); /* estimated and max additional space needed */
1958       ierr = PetscFreeSpaceGet(nnz,&current_space);CHKERRQ(ierr);
1959       ierr = PetscFreeSpaceGet(nnz,&current_space_lvl);CHKERRQ(ierr);
1960       reallocs++;
1961     }
1962 
1963     /* copy data into free_space and free_space_lvl, then initialize lnk */
1964     ierr         = PetscIncompleteLLClean(n,n,nzi,lnk,lnk_lvl,current_space->array,current_space_lvl->array,lnkbt);CHKERRQ(ierr);
1965     bj_ptr[i]    = current_space->array;
1966     bjlvl_ptr[i] = current_space_lvl->array;
1967 
1968     /* make sure the active row i has diagonal entry */
1969     if (*(bj_ptr[i]+bdiag[i]) != i) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Row %D has missing diagonal in factored matrix\ntry running with -pc_factor_nonzeros_along_diagonal or -pc_factor_diagonal_fill",i);
1970 
1971     current_space->array               += nzi;
1972     current_space->local_used          += nzi;
1973     current_space->local_remaining     -= nzi;
1974     current_space_lvl->array           += nzi;
1975     current_space_lvl->local_used      += nzi;
1976     current_space_lvl->local_remaining -= nzi;
1977   }
1978 
1979   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
1980   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
1981 
1982   /* destroy list of free space and other temporary arrays */
1983   ierr = PetscMalloc1(bi[n]+1,&bj);CHKERRQ(ierr);
1984   ierr = PetscFreeSpaceContiguous(&free_space,bj);CHKERRQ(ierr); /* copy free_space -> bj */
1985   ierr = PetscIncompleteLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
1986   ierr = PetscFreeSpaceDestroy(free_space_lvl);CHKERRQ(ierr);
1987   ierr = PetscFree2(bj_ptr,bjlvl_ptr);CHKERRQ(ierr);
1988 
1989 #if defined(PETSC_USE_INFO)
1990   {
1991     PetscReal af = ((PetscReal)bi[n])/((PetscReal)ai[n]);
1992     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)f,(double)af);CHKERRQ(ierr);
1993     ierr = PetscInfo1(A,"Run with -[sub_]pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
1994     ierr = PetscInfo1(A,"PCFactorSetFill([sub]pc,%g);\n",(double)af);CHKERRQ(ierr);
1995     ierr = PetscInfo(A,"for best performance.\n");CHKERRQ(ierr);
1996     if (diagonal_fill) {
1997       ierr = PetscInfo1(A,"Detected and replaced %D missing diagonals\n",dcount);CHKERRQ(ierr);
1998     }
1999   }
2000 #endif
2001 
2002   /* put together the new matrix */
2003   ierr = MatSeqAIJSetPreallocation_SeqAIJ(fact,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
2004   ierr = PetscLogObjectParent((PetscObject)fact,(PetscObject)isicol);CHKERRQ(ierr);
2005   b    = (Mat_SeqAIJ*)(fact)->data;
2006 
2007   b->free_a       = PETSC_TRUE;
2008   b->free_ij      = PETSC_TRUE;
2009   b->singlemalloc = PETSC_FALSE;
2010 
2011   ierr = PetscMalloc1(bi[n],&b->a);CHKERRQ(ierr);
2012   b->j = bj;
2013   b->i = bi;
2014   for (i=0; i<n; i++) bdiag[i] += bi[i];
2015   b->diag = bdiag;
2016   b->ilen = 0;
2017   b->imax = 0;
2018   b->row  = isrow;
2019   b->col  = iscol;
2020   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
2021   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
2022   b->icol = isicol;
2023   ierr    = PetscMalloc1(n+1,&b->solve_work);CHKERRQ(ierr);
2024   /* In b structure:  Free imax, ilen, old a, old j.
2025      Allocate bdiag, solve_work, new a, new j */
2026   ierr     = PetscLogObjectMemory((PetscObject)fact,(bi[n]-n) * (sizeof(PetscInt)+sizeof(PetscScalar)));CHKERRQ(ierr);
2027   b->maxnz = b->nz = bi[n];
2028 
2029   (fact)->info.factor_mallocs    = reallocs;
2030   (fact)->info.fill_ratio_given  = f;
2031   (fact)->info.fill_ratio_needed = ((PetscReal)bi[n])/((PetscReal)ai[n]);
2032   (fact)->ops->lufactornumeric   =  MatLUFactorNumeric_SeqAIJ_inplace;
2033   if (a->inode.size) {
2034     (fact)->ops->lufactornumeric = MatLUFactorNumeric_SeqAIJ_Inode_inplace;
2035   }
2036   PetscFunctionReturn(0);
2037 }
2038 
2039 PetscErrorCode MatCholeskyFactorNumeric_SeqAIJ(Mat B,Mat A,const MatFactorInfo *info)
2040 {
2041   Mat            C = B;
2042   Mat_SeqAIJ     *a=(Mat_SeqAIJ*)A->data;
2043   Mat_SeqSBAIJ   *b=(Mat_SeqSBAIJ*)C->data;
2044   IS             ip=b->row,iip = b->icol;
2045   PetscErrorCode ierr;
2046   const PetscInt *rip,*riip;
2047   PetscInt       i,j,mbs=A->rmap->n,*bi=b->i,*bj=b->j,*bdiag=b->diag,*bjtmp;
2048   PetscInt       *ai=a->i,*aj=a->j;
2049   PetscInt       k,jmin,jmax,*c2r,*il,col,nexti,ili,nz;
2050   MatScalar      *rtmp,*ba=b->a,*bval,*aa=a->a,dk,uikdi;
2051   PetscBool      perm_identity;
2052   FactorShiftCtx sctx;
2053   PetscReal      rs;
2054   MatScalar      d,*v;
2055 
2056   PetscFunctionBegin;
2057   /* MatPivotSetUp(): initialize shift context sctx */
2058   ierr = PetscMemzero(&sctx,sizeof(FactorShiftCtx));CHKERRQ(ierr);
2059 
2060   if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
2061     sctx.shift_top = info->zeropivot;
2062     for (i=0; i<mbs; i++) {
2063       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
2064       d  = (aa)[a->diag[i]];
2065       rs = -PetscAbsScalar(d) - PetscRealPart(d);
2066       v  = aa+ai[i];
2067       nz = ai[i+1] - ai[i];
2068       for (j=0; j<nz; j++) rs += PetscAbsScalar(v[j]);
2069       if (rs>sctx.shift_top) sctx.shift_top = rs;
2070     }
2071     sctx.shift_top *= 1.1;
2072     sctx.nshift_max = 5;
2073     sctx.shift_lo   = 0.;
2074     sctx.shift_hi   = 1.;
2075   }
2076 
2077   ierr = ISGetIndices(ip,&rip);CHKERRQ(ierr);
2078   ierr = ISGetIndices(iip,&riip);CHKERRQ(ierr);
2079 
2080   /* allocate working arrays
2081      c2r: linked list, keep track of pivot rows for a given column. c2r[col]: head of the list for a given col
2082      il:  for active k row, il[i] gives the index of the 1st nonzero entry in U[i,k:n-1] in bj and ba arrays
2083   */
2084   ierr = PetscMalloc3(mbs,&rtmp,mbs,&il,mbs,&c2r);CHKERRQ(ierr);
2085 
2086   do {
2087     sctx.newshift = PETSC_FALSE;
2088 
2089     for (i=0; i<mbs; i++) c2r[i] = mbs;
2090     if (mbs) il[0] = 0;
2091 
2092     for (k = 0; k<mbs; k++) {
2093       /* zero rtmp */
2094       nz    = bi[k+1] - bi[k];
2095       bjtmp = bj + bi[k];
2096       for (j=0; j<nz; j++) rtmp[bjtmp[j]] = 0.0;
2097 
2098       /* load in initial unfactored row */
2099       bval = ba + bi[k];
2100       jmin = ai[rip[k]]; jmax = ai[rip[k]+1];
2101       for (j = jmin; j < jmax; j++) {
2102         col = riip[aj[j]];
2103         if (col >= k) { /* only take upper triangular entry */
2104           rtmp[col] = aa[j];
2105           *bval++   = 0.0; /* for in-place factorization */
2106         }
2107       }
2108       /* shift the diagonal of the matrix: ZeropivotApply() */
2109       rtmp[k] += sctx.shift_amount;  /* shift the diagonal of the matrix */
2110 
2111       /* modify k-th row by adding in those rows i with U(i,k)!=0 */
2112       dk = rtmp[k];
2113       i  = c2r[k]; /* first row to be added to k_th row  */
2114 
2115       while (i < k) {
2116         nexti = c2r[i]; /* next row to be added to k_th row */
2117 
2118         /* compute multiplier, update diag(k) and U(i,k) */
2119         ili     = il[i]; /* index of first nonzero element in U(i,k:bms-1) */
2120         uikdi   = -ba[ili]*ba[bdiag[i]]; /* diagonal(k) */
2121         dk     += uikdi*ba[ili]; /* update diag[k] */
2122         ba[ili] = uikdi; /* -U(i,k) */
2123 
2124         /* add multiple of row i to k-th row */
2125         jmin = ili + 1; jmax = bi[i+1];
2126         if (jmin < jmax) {
2127           for (j=jmin; j<jmax; j++) rtmp[bj[j]] += uikdi*ba[j];
2128           /* update il and c2r for row i */
2129           il[i] = jmin;
2130           j     = bj[jmin]; c2r[i] = c2r[j]; c2r[j] = i;
2131         }
2132         i = nexti;
2133       }
2134 
2135       /* copy data into U(k,:) */
2136       rs   = 0.0;
2137       jmin = bi[k]; jmax = bi[k+1]-1;
2138       if (jmin < jmax) {
2139         for (j=jmin; j<jmax; j++) {
2140           col = bj[j]; ba[j] = rtmp[col]; rs += PetscAbsScalar(ba[j]);
2141         }
2142         /* add the k-th row into il and c2r */
2143         il[k] = jmin;
2144         i     = bj[jmin]; c2r[k] = c2r[i]; c2r[i] = k;
2145       }
2146 
2147       /* MatPivotCheck() */
2148       sctx.rs = rs;
2149       sctx.pv = dk;
2150       ierr    = MatPivotCheck(B,A,info,&sctx,i);CHKERRQ(ierr);
2151       if (sctx.newshift) break;
2152       dk = sctx.pv;
2153 
2154       ba[bdiag[k]] = 1.0/dk; /* U(k,k) */
2155     }
2156   } while (sctx.newshift);
2157 
2158   ierr = PetscFree3(rtmp,il,c2r);CHKERRQ(ierr);
2159   ierr = ISRestoreIndices(ip,&rip);CHKERRQ(ierr);
2160   ierr = ISRestoreIndices(iip,&riip);CHKERRQ(ierr);
2161 
2162   ierr = ISIdentity(ip,&perm_identity);CHKERRQ(ierr);
2163   if (perm_identity) {
2164     B->ops->solve          = MatSolve_SeqSBAIJ_1_NaturalOrdering;
2165     B->ops->solvetranspose = MatSolve_SeqSBAIJ_1_NaturalOrdering;
2166     B->ops->forwardsolve   = MatForwardSolve_SeqSBAIJ_1_NaturalOrdering;
2167     B->ops->backwardsolve  = MatBackwardSolve_SeqSBAIJ_1_NaturalOrdering;
2168   } else {
2169     B->ops->solve          = MatSolve_SeqSBAIJ_1;
2170     B->ops->solvetranspose = MatSolve_SeqSBAIJ_1;
2171     B->ops->forwardsolve   = MatForwardSolve_SeqSBAIJ_1;
2172     B->ops->backwardsolve  = MatBackwardSolve_SeqSBAIJ_1;
2173   }
2174 
2175   C->assembled    = PETSC_TRUE;
2176   C->preallocated = PETSC_TRUE;
2177 
2178   ierr = PetscLogFlops(C->rmap->n);CHKERRQ(ierr);
2179 
2180   /* MatPivotView() */
2181   if (sctx.nshift) {
2182     if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) {
2183       ierr = PetscInfo4(A,"number of shift_pd tries %D, shift_amount %g, diagonal shifted up by %e fraction top_value %e\n",sctx.nshift,(double)sctx.shift_amount,(double)sctx.shift_fraction,(double)sctx.shift_top);CHKERRQ(ierr);
2184     } else if (info->shifttype == (PetscReal)MAT_SHIFT_NONZERO) {
2185       ierr = PetscInfo2(A,"number of shift_nz tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
2186     } else if (info->shifttype == (PetscReal)MAT_SHIFT_INBLOCKS) {
2187       ierr = PetscInfo2(A,"number of shift_inblocks applied %D, each shift_amount %g\n",sctx.nshift,(double)info->shiftamount);CHKERRQ(ierr);
2188     }
2189   }
2190   PetscFunctionReturn(0);
2191 }
2192 
2193 PetscErrorCode MatCholeskyFactorNumeric_SeqAIJ_inplace(Mat B,Mat A,const MatFactorInfo *info)
2194 {
2195   Mat            C = B;
2196   Mat_SeqAIJ     *a=(Mat_SeqAIJ*)A->data;
2197   Mat_SeqSBAIJ   *b=(Mat_SeqSBAIJ*)C->data;
2198   IS             ip=b->row,iip = b->icol;
2199   PetscErrorCode ierr;
2200   const PetscInt *rip,*riip;
2201   PetscInt       i,j,mbs=A->rmap->n,*bi=b->i,*bj=b->j,*bcol,*bjtmp;
2202   PetscInt       *ai=a->i,*aj=a->j;
2203   PetscInt       k,jmin,jmax,*jl,*il,col,nexti,ili,nz;
2204   MatScalar      *rtmp,*ba=b->a,*bval,*aa=a->a,dk,uikdi;
2205   PetscBool      perm_identity;
2206   FactorShiftCtx sctx;
2207   PetscReal      rs;
2208   MatScalar      d,*v;
2209 
2210   PetscFunctionBegin;
2211   /* MatPivotSetUp(): initialize shift context sctx */
2212   ierr = PetscMemzero(&sctx,sizeof(FactorShiftCtx));CHKERRQ(ierr);
2213 
2214   if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) { /* set sctx.shift_top=max{rs} */
2215     sctx.shift_top = info->zeropivot;
2216     for (i=0; i<mbs; i++) {
2217       /* calculate sum(|aij|)-RealPart(aii), amt of shift needed for this row */
2218       d  = (aa)[a->diag[i]];
2219       rs = -PetscAbsScalar(d) - PetscRealPart(d);
2220       v  = aa+ai[i];
2221       nz = ai[i+1] - ai[i];
2222       for (j=0; j<nz; j++) rs += PetscAbsScalar(v[j]);
2223       if (rs>sctx.shift_top) sctx.shift_top = rs;
2224     }
2225     sctx.shift_top *= 1.1;
2226     sctx.nshift_max = 5;
2227     sctx.shift_lo   = 0.;
2228     sctx.shift_hi   = 1.;
2229   }
2230 
2231   ierr = ISGetIndices(ip,&rip);CHKERRQ(ierr);
2232   ierr = ISGetIndices(iip,&riip);CHKERRQ(ierr);
2233 
2234   /* initialization */
2235   ierr = PetscMalloc3(mbs,&rtmp,mbs,&il,mbs,&jl);CHKERRQ(ierr);
2236 
2237   do {
2238     sctx.newshift = PETSC_FALSE;
2239 
2240     for (i=0; i<mbs; i++) jl[i] = mbs;
2241     il[0] = 0;
2242 
2243     for (k = 0; k<mbs; k++) {
2244       /* zero rtmp */
2245       nz    = bi[k+1] - bi[k];
2246       bjtmp = bj + bi[k];
2247       for (j=0; j<nz; j++) rtmp[bjtmp[j]] = 0.0;
2248 
2249       bval = ba + bi[k];
2250       /* initialize k-th row by the perm[k]-th row of A */
2251       jmin = ai[rip[k]]; jmax = ai[rip[k]+1];
2252       for (j = jmin; j < jmax; j++) {
2253         col = riip[aj[j]];
2254         if (col >= k) { /* only take upper triangular entry */
2255           rtmp[col] = aa[j];
2256           *bval++   = 0.0; /* for in-place factorization */
2257         }
2258       }
2259       /* shift the diagonal of the matrix */
2260       if (sctx.nshift) rtmp[k] += sctx.shift_amount;
2261 
2262       /* modify k-th row by adding in those rows i with U(i,k)!=0 */
2263       dk = rtmp[k];
2264       i  = jl[k]; /* first row to be added to k_th row  */
2265 
2266       while (i < k) {
2267         nexti = jl[i]; /* next row to be added to k_th row */
2268 
2269         /* compute multiplier, update diag(k) and U(i,k) */
2270         ili     = il[i]; /* index of first nonzero element in U(i,k:bms-1) */
2271         uikdi   = -ba[ili]*ba[bi[i]]; /* diagonal(k) */
2272         dk     += uikdi*ba[ili];
2273         ba[ili] = uikdi; /* -U(i,k) */
2274 
2275         /* add multiple of row i to k-th row */
2276         jmin = ili + 1; jmax = bi[i+1];
2277         if (jmin < jmax) {
2278           for (j=jmin; j<jmax; j++) rtmp[bj[j]] += uikdi*ba[j];
2279           /* update il and jl for row i */
2280           il[i] = jmin;
2281           j     = bj[jmin]; jl[i] = jl[j]; jl[j] = i;
2282         }
2283         i = nexti;
2284       }
2285 
2286       /* shift the diagonals when zero pivot is detected */
2287       /* compute rs=sum of abs(off-diagonal) */
2288       rs   = 0.0;
2289       jmin = bi[k]+1;
2290       nz   = bi[k+1] - jmin;
2291       bcol = bj + jmin;
2292       for (j=0; j<nz; j++) {
2293         rs += PetscAbsScalar(rtmp[bcol[j]]);
2294       }
2295 
2296       sctx.rs = rs;
2297       sctx.pv = dk;
2298       ierr    = MatPivotCheck(B,A,info,&sctx,k);CHKERRQ(ierr);
2299       if (sctx.newshift) break;
2300       dk = sctx.pv;
2301 
2302       /* copy data into U(k,:) */
2303       ba[bi[k]] = 1.0/dk; /* U(k,k) */
2304       jmin      = bi[k]+1; jmax = bi[k+1];
2305       if (jmin < jmax) {
2306         for (j=jmin; j<jmax; j++) {
2307           col = bj[j]; ba[j] = rtmp[col];
2308         }
2309         /* add the k-th row into il and jl */
2310         il[k] = jmin;
2311         i     = bj[jmin]; jl[k] = jl[i]; jl[i] = k;
2312       }
2313     }
2314   } while (sctx.newshift);
2315 
2316   ierr = PetscFree3(rtmp,il,jl);CHKERRQ(ierr);
2317   ierr = ISRestoreIndices(ip,&rip);CHKERRQ(ierr);
2318   ierr = ISRestoreIndices(iip,&riip);CHKERRQ(ierr);
2319 
2320   ierr = ISIdentity(ip,&perm_identity);CHKERRQ(ierr);
2321   if (perm_identity) {
2322     B->ops->solve          = MatSolve_SeqSBAIJ_1_NaturalOrdering_inplace;
2323     B->ops->solvetranspose = MatSolve_SeqSBAIJ_1_NaturalOrdering_inplace;
2324     B->ops->forwardsolve   = MatForwardSolve_SeqSBAIJ_1_NaturalOrdering_inplace;
2325     B->ops->backwardsolve  = MatBackwardSolve_SeqSBAIJ_1_NaturalOrdering_inplace;
2326   } else {
2327     B->ops->solve          = MatSolve_SeqSBAIJ_1_inplace;
2328     B->ops->solvetranspose = MatSolve_SeqSBAIJ_1_inplace;
2329     B->ops->forwardsolve   = MatForwardSolve_SeqSBAIJ_1_inplace;
2330     B->ops->backwardsolve  = MatBackwardSolve_SeqSBAIJ_1_inplace;
2331   }
2332 
2333   C->assembled    = PETSC_TRUE;
2334   C->preallocated = PETSC_TRUE;
2335 
2336   ierr = PetscLogFlops(C->rmap->n);CHKERRQ(ierr);
2337   if (sctx.nshift) {
2338     if (info->shifttype == (PetscReal)MAT_SHIFT_NONZERO) {
2339       ierr = PetscInfo2(A,"number of shiftnz tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
2340     } else if (info->shifttype == (PetscReal)MAT_SHIFT_POSITIVE_DEFINITE) {
2341       ierr = PetscInfo2(A,"number of shiftpd tries %D, shift_amount %g\n",sctx.nshift,(double)sctx.shift_amount);CHKERRQ(ierr);
2342     }
2343   }
2344   PetscFunctionReturn(0);
2345 }
2346 
2347 /*
2348    icc() under revised new data structure.
2349    Factored arrays bj and ba are stored as
2350      U(0,:),...,U(i,:),U(n-1,:)
2351 
2352    ui=fact->i is an array of size n+1, in which
2353    ui+
2354      ui[i]:  points to 1st entry of U(i,:),i=0,...,n-1
2355      ui[n]:  points to U(n-1,n-1)+1
2356 
2357   udiag=fact->diag is an array of size n,in which
2358      udiag[i]: points to diagonal of U(i,:), i=0,...,n-1
2359 
2360    U(i,:) contains udiag[i] as its last entry, i.e.,
2361     U(i,:) = (u[i,i+1],...,u[i,n-1],diag[i])
2362 */
2363 
2364 PetscErrorCode MatICCFactorSymbolic_SeqAIJ(Mat fact,Mat A,IS perm,const MatFactorInfo *info)
2365 {
2366   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data;
2367   Mat_SeqSBAIJ       *b;
2368   PetscErrorCode     ierr;
2369   PetscBool          perm_identity,missing;
2370   PetscInt           reallocs=0,i,*ai=a->i,*aj=a->j,am=A->rmap->n,*ui,*udiag;
2371   const PetscInt     *rip,*riip;
2372   PetscInt           jmin,jmax,nzk,k,j,*jl,prow,*il,nextprow;
2373   PetscInt           nlnk,*lnk,*lnk_lvl=NULL,d;
2374   PetscInt           ncols,ncols_upper,*cols,*ajtmp,*uj,**uj_ptr,**uj_lvl_ptr;
2375   PetscReal          fill          =info->fill,levels=info->levels;
2376   PetscFreeSpaceList free_space    =NULL,current_space=NULL;
2377   PetscFreeSpaceList free_space_lvl=NULL,current_space_lvl=NULL;
2378   PetscBT            lnkbt;
2379   IS                 iperm;
2380 
2381   PetscFunctionBegin;
2382   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
2383   ierr = MatMissingDiagonal(A,&missing,&d);CHKERRQ(ierr);
2384   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",d);
2385   ierr = ISIdentity(perm,&perm_identity);CHKERRQ(ierr);
2386   ierr = ISInvertPermutation(perm,PETSC_DECIDE,&iperm);CHKERRQ(ierr);
2387 
2388   ierr  = PetscMalloc1(am+1,&ui);CHKERRQ(ierr);
2389   ierr  = PetscMalloc1(am+1,&udiag);CHKERRQ(ierr);
2390   ui[0] = 0;
2391 
2392   /* ICC(0) without matrix ordering: simply rearrange column indices */
2393   if (!levels && perm_identity) {
2394     for (i=0; i<am; i++) {
2395       ncols    = ai[i+1] - a->diag[i];
2396       ui[i+1]  = ui[i] + ncols;
2397       udiag[i] = ui[i+1] - 1; /* points to the last entry of U(i,:) */
2398     }
2399     ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
2400     cols = uj;
2401     for (i=0; i<am; i++) {
2402       aj    = a->j + a->diag[i] + 1; /* 1st entry of U(i,:) without diagonal */
2403       ncols = ai[i+1] - a->diag[i] -1;
2404       for (j=0; j<ncols; j++) *cols++ = aj[j];
2405       *cols++ = i; /* diagoanl is located as the last entry of U(i,:) */
2406     }
2407   } else { /* case: levels>0 || (levels=0 && !perm_identity) */
2408     ierr = ISGetIndices(iperm,&riip);CHKERRQ(ierr);
2409     ierr = ISGetIndices(perm,&rip);CHKERRQ(ierr);
2410 
2411     /* initialization */
2412     ierr = PetscMalloc1(am+1,&ajtmp);CHKERRQ(ierr);
2413 
2414     /* jl: linked list for storing indices of the pivot rows
2415        il: il[i] points to the 1st nonzero entry of U(i,k:am-1) */
2416     ierr = PetscMalloc4(am,&uj_ptr,am,&uj_lvl_ptr,am,&jl,am,&il);CHKERRQ(ierr);
2417     for (i=0; i<am; i++) {
2418       jl[i] = am; il[i] = 0;
2419     }
2420 
2421     /* create and initialize a linked list for storing column indices of the active row k */
2422     nlnk = am + 1;
2423     ierr = PetscIncompleteLLCreate(am,am,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
2424 
2425     /* initial FreeSpace size is fill*(ai[am]+am)/2 */
2426     ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,(ai[am]+am)/2),&free_space);CHKERRQ(ierr);
2427     current_space     = free_space;
2428     ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,(ai[am]+am)/2),&free_space_lvl);CHKERRQ(ierr);
2429     current_space_lvl = free_space_lvl;
2430 
2431     for (k=0; k<am; k++) {  /* for each active row k */
2432       /* initialize lnk by the column indices of row rip[k] of A */
2433       nzk   = 0;
2434       ncols = ai[rip[k]+1] - ai[rip[k]];
2435       if (!ncols) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_CH_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",rip[k],k);
2436       ncols_upper = 0;
2437       for (j=0; j<ncols; j++) {
2438         i = *(aj + ai[rip[k]] + j); /* unpermuted column index */
2439         if (riip[i] >= k) { /* only take upper triangular entry */
2440           ajtmp[ncols_upper] = i;
2441           ncols_upper++;
2442         }
2443       }
2444       ierr = PetscIncompleteLLInit(ncols_upper,ajtmp,am,riip,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
2445       nzk += nlnk;
2446 
2447       /* update lnk by computing fill-in for each pivot row to be merged in */
2448       prow = jl[k]; /* 1st pivot row */
2449 
2450       while (prow < k) {
2451         nextprow = jl[prow];
2452 
2453         /* merge prow into k-th row */
2454         jmin  = il[prow] + 1; /* index of the 2nd nzero entry in U(prow,k:am-1) */
2455         jmax  = ui[prow+1];
2456         ncols = jmax-jmin;
2457         i     = jmin - ui[prow];
2458         cols  = uj_ptr[prow] + i; /* points to the 2nd nzero entry in U(prow,k:am-1) */
2459         uj    = uj_lvl_ptr[prow] + i; /* levels of cols */
2460         j     = *(uj - 1);
2461         ierr  = PetscICCLLAddSorted(ncols,cols,levels,uj,am,nlnk,lnk,lnk_lvl,lnkbt,j);CHKERRQ(ierr);
2462         nzk  += nlnk;
2463 
2464         /* update il and jl for prow */
2465         if (jmin < jmax) {
2466           il[prow] = jmin;
2467           j        = *cols; jl[prow] = jl[j]; jl[j] = prow;
2468         }
2469         prow = nextprow;
2470       }
2471 
2472       /* if free space is not available, make more free space */
2473       if (current_space->local_remaining<nzk) {
2474         i    = am - k + 1; /* num of unfactored rows */
2475         i    = PetscIntMultTruncate(i,PetscMin(nzk, i-1)); /* i*nzk, i*(i-1): estimated and max additional space needed */
2476         ierr = PetscFreeSpaceGet(i,&current_space);CHKERRQ(ierr);
2477         ierr = PetscFreeSpaceGet(i,&current_space_lvl);CHKERRQ(ierr);
2478         reallocs++;
2479       }
2480 
2481       /* copy data into free_space and free_space_lvl, then initialize lnk */
2482       if (nzk == 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Empty row %D in ICC matrix factor",k);
2483       ierr = PetscIncompleteLLClean(am,am,nzk,lnk,lnk_lvl,current_space->array,current_space_lvl->array,lnkbt);CHKERRQ(ierr);
2484 
2485       /* add the k-th row into il and jl */
2486       if (nzk > 1) {
2487         i     = current_space->array[1]; /* col value of the first nonzero element in U(k, k+1:am-1) */
2488         jl[k] = jl[i]; jl[i] = k;
2489         il[k] = ui[k] + 1;
2490       }
2491       uj_ptr[k]     = current_space->array;
2492       uj_lvl_ptr[k] = current_space_lvl->array;
2493 
2494       current_space->array           += nzk;
2495       current_space->local_used      += nzk;
2496       current_space->local_remaining -= nzk;
2497 
2498       current_space_lvl->array           += nzk;
2499       current_space_lvl->local_used      += nzk;
2500       current_space_lvl->local_remaining -= nzk;
2501 
2502       ui[k+1] = ui[k] + nzk;
2503     }
2504 
2505     ierr = ISRestoreIndices(perm,&rip);CHKERRQ(ierr);
2506     ierr = ISRestoreIndices(iperm,&riip);CHKERRQ(ierr);
2507     ierr = PetscFree4(uj_ptr,uj_lvl_ptr,jl,il);CHKERRQ(ierr);
2508     ierr = PetscFree(ajtmp);CHKERRQ(ierr);
2509 
2510     /* copy free_space into uj and free free_space; set ui, uj, udiag in new datastructure; */
2511     ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
2512     ierr = PetscFreeSpaceContiguous_Cholesky(&free_space,uj,am,ui,udiag);CHKERRQ(ierr); /* store matrix factor  */
2513     ierr = PetscIncompleteLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
2514     ierr = PetscFreeSpaceDestroy(free_space_lvl);CHKERRQ(ierr);
2515 
2516   } /* end of case: levels>0 || (levels=0 && !perm_identity) */
2517 
2518   /* put together the new matrix in MATSEQSBAIJ format */
2519   b               = (Mat_SeqSBAIJ*)(fact)->data;
2520   b->singlemalloc = PETSC_FALSE;
2521 
2522   ierr = PetscMalloc1(ui[am]+1,&b->a);CHKERRQ(ierr);
2523 
2524   b->j             = uj;
2525   b->i             = ui;
2526   b->diag          = udiag;
2527   b->free_diag     = PETSC_TRUE;
2528   b->ilen          = 0;
2529   b->imax          = 0;
2530   b->row           = perm;
2531   b->col           = perm;
2532   ierr             = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2533   ierr             = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2534   b->icol          = iperm;
2535   b->pivotinblocks = PETSC_FALSE; /* need to get from MatFactorInfo */
2536 
2537   ierr = PetscMalloc1(am+1,&b->solve_work);CHKERRQ(ierr);
2538   ierr = PetscLogObjectMemory((PetscObject)fact,ui[am]*(sizeof(PetscInt)+sizeof(MatScalar)));CHKERRQ(ierr);
2539 
2540   b->maxnz   = b->nz = ui[am];
2541   b->free_a  = PETSC_TRUE;
2542   b->free_ij = PETSC_TRUE;
2543 
2544   fact->info.factor_mallocs   = reallocs;
2545   fact->info.fill_ratio_given = fill;
2546   if (ai[am] != 0) {
2547     /* nonzeros in lower triangular part of A (including diagonals) = (ai[am]+am)/2 */
2548     fact->info.fill_ratio_needed = ((PetscReal)2*ui[am])/(ai[am]+am);
2549   } else {
2550     fact->info.fill_ratio_needed = 0.0;
2551   }
2552 #if defined(PETSC_USE_INFO)
2553   if (ai[am] != 0) {
2554     PetscReal af = fact->info.fill_ratio_needed;
2555     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)fill,(double)af);CHKERRQ(ierr);
2556     ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
2557     ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g) for best performance.\n",(double)af);CHKERRQ(ierr);
2558   } else {
2559     ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
2560   }
2561 #endif
2562   fact->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqAIJ;
2563   PetscFunctionReturn(0);
2564 }
2565 
2566 PetscErrorCode MatICCFactorSymbolic_SeqAIJ_inplace(Mat fact,Mat A,IS perm,const MatFactorInfo *info)
2567 {
2568   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data;
2569   Mat_SeqSBAIJ       *b;
2570   PetscErrorCode     ierr;
2571   PetscBool          perm_identity,missing;
2572   PetscInt           reallocs=0,i,*ai=a->i,*aj=a->j,am=A->rmap->n,*ui,*udiag;
2573   const PetscInt     *rip,*riip;
2574   PetscInt           jmin,jmax,nzk,k,j,*jl,prow,*il,nextprow;
2575   PetscInt           nlnk,*lnk,*lnk_lvl=NULL,d;
2576   PetscInt           ncols,ncols_upper,*cols,*ajtmp,*uj,**uj_ptr,**uj_lvl_ptr;
2577   PetscReal          fill          =info->fill,levels=info->levels;
2578   PetscFreeSpaceList free_space    =NULL,current_space=NULL;
2579   PetscFreeSpaceList free_space_lvl=NULL,current_space_lvl=NULL;
2580   PetscBT            lnkbt;
2581   IS                 iperm;
2582 
2583   PetscFunctionBegin;
2584   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
2585   ierr = MatMissingDiagonal(A,&missing,&d);CHKERRQ(ierr);
2586   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",d);
2587   ierr = ISIdentity(perm,&perm_identity);CHKERRQ(ierr);
2588   ierr = ISInvertPermutation(perm,PETSC_DECIDE,&iperm);CHKERRQ(ierr);
2589 
2590   ierr  = PetscMalloc1(am+1,&ui);CHKERRQ(ierr);
2591   ierr  = PetscMalloc1(am+1,&udiag);CHKERRQ(ierr);
2592   ui[0] = 0;
2593 
2594   /* ICC(0) without matrix ordering: simply copies fill pattern */
2595   if (!levels && perm_identity) {
2596 
2597     for (i=0; i<am; i++) {
2598       ui[i+1]  = ui[i] + ai[i+1] - a->diag[i];
2599       udiag[i] = ui[i];
2600     }
2601     ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
2602     cols = uj;
2603     for (i=0; i<am; i++) {
2604       aj    = a->j + a->diag[i];
2605       ncols = ui[i+1] - ui[i];
2606       for (j=0; j<ncols; j++) *cols++ = *aj++;
2607     }
2608   } else { /* case: levels>0 || (levels=0 && !perm_identity) */
2609     ierr = ISGetIndices(iperm,&riip);CHKERRQ(ierr);
2610     ierr = ISGetIndices(perm,&rip);CHKERRQ(ierr);
2611 
2612     /* initialization */
2613     ierr = PetscMalloc1(am+1,&ajtmp);CHKERRQ(ierr);
2614 
2615     /* jl: linked list for storing indices of the pivot rows
2616        il: il[i] points to the 1st nonzero entry of U(i,k:am-1) */
2617     ierr = PetscMalloc4(am,&uj_ptr,am,&uj_lvl_ptr,am,&jl,am,&il);CHKERRQ(ierr);
2618     for (i=0; i<am; i++) {
2619       jl[i] = am; il[i] = 0;
2620     }
2621 
2622     /* create and initialize a linked list for storing column indices of the active row k */
2623     nlnk = am + 1;
2624     ierr = PetscIncompleteLLCreate(am,am,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
2625 
2626     /* initial FreeSpace size is fill*(ai[am]+1) */
2627     ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,ai[am]+1),&free_space);CHKERRQ(ierr);
2628     current_space     = free_space;
2629     ierr              = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,ai[am]+1),&free_space_lvl);CHKERRQ(ierr);
2630     current_space_lvl = free_space_lvl;
2631 
2632     for (k=0; k<am; k++) {  /* for each active row k */
2633       /* initialize lnk by the column indices of row rip[k] of A */
2634       nzk   = 0;
2635       ncols = ai[rip[k]+1] - ai[rip[k]];
2636       if (!ncols) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_CH_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",rip[k],k);
2637       ncols_upper = 0;
2638       for (j=0; j<ncols; j++) {
2639         i = *(aj + ai[rip[k]] + j); /* unpermuted column index */
2640         if (riip[i] >= k) { /* only take upper triangular entry */
2641           ajtmp[ncols_upper] = i;
2642           ncols_upper++;
2643         }
2644       }
2645       ierr = PetscIncompleteLLInit(ncols_upper,ajtmp,am,riip,nlnk,lnk,lnk_lvl,lnkbt);CHKERRQ(ierr);
2646       nzk += nlnk;
2647 
2648       /* update lnk by computing fill-in for each pivot row to be merged in */
2649       prow = jl[k]; /* 1st pivot row */
2650 
2651       while (prow < k) {
2652         nextprow = jl[prow];
2653 
2654         /* merge prow into k-th row */
2655         jmin  = il[prow] + 1; /* index of the 2nd nzero entry in U(prow,k:am-1) */
2656         jmax  = ui[prow+1];
2657         ncols = jmax-jmin;
2658         i     = jmin - ui[prow];
2659         cols  = uj_ptr[prow] + i; /* points to the 2nd nzero entry in U(prow,k:am-1) */
2660         uj    = uj_lvl_ptr[prow] + i; /* levels of cols */
2661         j     = *(uj - 1);
2662         ierr  = PetscICCLLAddSorted(ncols,cols,levels,uj,am,nlnk,lnk,lnk_lvl,lnkbt,j);CHKERRQ(ierr);
2663         nzk  += nlnk;
2664 
2665         /* update il and jl for prow */
2666         if (jmin < jmax) {
2667           il[prow] = jmin;
2668           j        = *cols; jl[prow] = jl[j]; jl[j] = prow;
2669         }
2670         prow = nextprow;
2671       }
2672 
2673       /* if free space is not available, make more free space */
2674       if (current_space->local_remaining<nzk) {
2675         i    = am - k + 1; /* num of unfactored rows */
2676         i    = PetscIntMultTruncate(i,PetscMin(nzk, i-1)); /* i*nzk, i*(i-1): estimated and max additional space needed */
2677         ierr = PetscFreeSpaceGet(i,&current_space);CHKERRQ(ierr);
2678         ierr = PetscFreeSpaceGet(i,&current_space_lvl);CHKERRQ(ierr);
2679         reallocs++;
2680       }
2681 
2682       /* copy data into free_space and free_space_lvl, then initialize lnk */
2683       if (!nzk) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Empty row %D in ICC matrix factor",k);
2684       ierr = PetscIncompleteLLClean(am,am,nzk,lnk,lnk_lvl,current_space->array,current_space_lvl->array,lnkbt);CHKERRQ(ierr);
2685 
2686       /* add the k-th row into il and jl */
2687       if (nzk > 1) {
2688         i     = current_space->array[1]; /* col value of the first nonzero element in U(k, k+1:am-1) */
2689         jl[k] = jl[i]; jl[i] = k;
2690         il[k] = ui[k] + 1;
2691       }
2692       uj_ptr[k]     = current_space->array;
2693       uj_lvl_ptr[k] = current_space_lvl->array;
2694 
2695       current_space->array           += nzk;
2696       current_space->local_used      += nzk;
2697       current_space->local_remaining -= nzk;
2698 
2699       current_space_lvl->array           += nzk;
2700       current_space_lvl->local_used      += nzk;
2701       current_space_lvl->local_remaining -= nzk;
2702 
2703       ui[k+1] = ui[k] + nzk;
2704     }
2705 
2706 #if defined(PETSC_USE_INFO)
2707     if (ai[am] != 0) {
2708       PetscReal af = (PetscReal)ui[am]/((PetscReal)ai[am]);
2709       ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)fill,(double)af);CHKERRQ(ierr);
2710       ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
2711       ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g) for best performance.\n",(double)af);CHKERRQ(ierr);
2712     } else {
2713       ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
2714     }
2715 #endif
2716 
2717     ierr = ISRestoreIndices(perm,&rip);CHKERRQ(ierr);
2718     ierr = ISRestoreIndices(iperm,&riip);CHKERRQ(ierr);
2719     ierr = PetscFree4(uj_ptr,uj_lvl_ptr,jl,il);CHKERRQ(ierr);
2720     ierr = PetscFree(ajtmp);CHKERRQ(ierr);
2721 
2722     /* destroy list of free space and other temporary array(s) */
2723     ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
2724     ierr = PetscFreeSpaceContiguous(&free_space,uj);CHKERRQ(ierr);
2725     ierr = PetscIncompleteLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
2726     ierr = PetscFreeSpaceDestroy(free_space_lvl);CHKERRQ(ierr);
2727 
2728   } /* end of case: levels>0 || (levels=0 && !perm_identity) */
2729 
2730   /* put together the new matrix in MATSEQSBAIJ format */
2731 
2732   b               = (Mat_SeqSBAIJ*)fact->data;
2733   b->singlemalloc = PETSC_FALSE;
2734 
2735   ierr = PetscMalloc1(ui[am]+1,&b->a);CHKERRQ(ierr);
2736 
2737   b->j         = uj;
2738   b->i         = ui;
2739   b->diag      = udiag;
2740   b->free_diag = PETSC_TRUE;
2741   b->ilen      = 0;
2742   b->imax      = 0;
2743   b->row       = perm;
2744   b->col       = perm;
2745 
2746   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2747   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2748 
2749   b->icol          = iperm;
2750   b->pivotinblocks = PETSC_FALSE; /* need to get from MatFactorInfo */
2751   ierr             = PetscMalloc1(am+1,&b->solve_work);CHKERRQ(ierr);
2752   ierr             = PetscLogObjectMemory((PetscObject)fact,(ui[am]-am)*(sizeof(PetscInt)+sizeof(MatScalar)));CHKERRQ(ierr);
2753   b->maxnz         = b->nz = ui[am];
2754   b->free_a        = PETSC_TRUE;
2755   b->free_ij       = PETSC_TRUE;
2756 
2757   fact->info.factor_mallocs   = reallocs;
2758   fact->info.fill_ratio_given = fill;
2759   if (ai[am] != 0) {
2760     fact->info.fill_ratio_needed = ((PetscReal)ui[am])/((PetscReal)ai[am]);
2761   } else {
2762     fact->info.fill_ratio_needed = 0.0;
2763   }
2764   fact->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqAIJ_inplace;
2765   PetscFunctionReturn(0);
2766 }
2767 
2768 PetscErrorCode MatCholeskyFactorSymbolic_SeqAIJ(Mat fact,Mat A,IS perm,const MatFactorInfo *info)
2769 {
2770   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data;
2771   Mat_SeqSBAIJ       *b;
2772   PetscErrorCode     ierr;
2773   PetscBool          perm_identity,missing;
2774   PetscReal          fill = info->fill;
2775   const PetscInt     *rip,*riip;
2776   PetscInt           i,am=A->rmap->n,*ai=a->i,*aj=a->j,reallocs=0,prow;
2777   PetscInt           *jl,jmin,jmax,nzk,*ui,k,j,*il,nextprow;
2778   PetscInt           nlnk,*lnk,ncols,ncols_upper,*cols,*uj,**ui_ptr,*uj_ptr,*udiag;
2779   PetscFreeSpaceList free_space=NULL,current_space=NULL;
2780   PetscBT            lnkbt;
2781   IS                 iperm;
2782 
2783   PetscFunctionBegin;
2784   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
2785   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
2786   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
2787 
2788   /* check whether perm is the identity mapping */
2789   ierr = ISIdentity(perm,&perm_identity);CHKERRQ(ierr);
2790   ierr = ISInvertPermutation(perm,PETSC_DECIDE,&iperm);CHKERRQ(ierr);
2791   ierr = ISGetIndices(iperm,&riip);CHKERRQ(ierr);
2792   ierr = ISGetIndices(perm,&rip);CHKERRQ(ierr);
2793 
2794   /* initialization */
2795   ierr  = PetscMalloc1(am+1,&ui);CHKERRQ(ierr);
2796   ierr  = PetscMalloc1(am+1,&udiag);CHKERRQ(ierr);
2797   ui[0] = 0;
2798 
2799   /* jl: linked list for storing indices of the pivot rows
2800      il: il[i] points to the 1st nonzero entry of U(i,k:am-1) */
2801   ierr = PetscMalloc4(am,&ui_ptr,am,&jl,am,&il,am,&cols);CHKERRQ(ierr);
2802   for (i=0; i<am; i++) {
2803     jl[i] = am; il[i] = 0;
2804   }
2805 
2806   /* create and initialize a linked list for storing column indices of the active row k */
2807   nlnk = am + 1;
2808   ierr = PetscLLCreate(am,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
2809 
2810   /* initial FreeSpace size is fill*(ai[am]+am)/2 */
2811   ierr          = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,(ai[am]+am)/2),&free_space);CHKERRQ(ierr);
2812   current_space = free_space;
2813 
2814   for (k=0; k<am; k++) {  /* for each active row k */
2815     /* initialize lnk by the column indices of row rip[k] of A */
2816     nzk   = 0;
2817     ncols = ai[rip[k]+1] - ai[rip[k]];
2818     if (!ncols) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_CH_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",rip[k],k);
2819     ncols_upper = 0;
2820     for (j=0; j<ncols; j++) {
2821       i = riip[*(aj + ai[rip[k]] + j)];
2822       if (i >= k) { /* only take upper triangular entry */
2823         cols[ncols_upper] = i;
2824         ncols_upper++;
2825       }
2826     }
2827     ierr = PetscLLAdd(ncols_upper,cols,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
2828     nzk += nlnk;
2829 
2830     /* update lnk by computing fill-in for each pivot row to be merged in */
2831     prow = jl[k]; /* 1st pivot row */
2832 
2833     while (prow < k) {
2834       nextprow = jl[prow];
2835       /* merge prow into k-th row */
2836       jmin   = il[prow] + 1; /* index of the 2nd nzero entry in U(prow,k:am-1) */
2837       jmax   = ui[prow+1];
2838       ncols  = jmax-jmin;
2839       uj_ptr = ui_ptr[prow] + jmin - ui[prow]; /* points to the 2nd nzero entry in U(prow,k:am-1) */
2840       ierr   = PetscLLAddSorted(ncols,uj_ptr,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
2841       nzk   += nlnk;
2842 
2843       /* update il and jl for prow */
2844       if (jmin < jmax) {
2845         il[prow] = jmin;
2846         j        = *uj_ptr;
2847         jl[prow] = jl[j];
2848         jl[j]    = prow;
2849       }
2850       prow = nextprow;
2851     }
2852 
2853     /* if free space is not available, make more free space */
2854     if (current_space->local_remaining<nzk) {
2855       i    = am - k + 1; /* num of unfactored rows */
2856       i    = PetscIntMultTruncate(i,PetscMin(nzk,i-1)); /* i*nzk, i*(i-1): estimated and max additional space needed */
2857       ierr = PetscFreeSpaceGet(i,&current_space);CHKERRQ(ierr);
2858       reallocs++;
2859     }
2860 
2861     /* copy data into free space, then initialize lnk */
2862     ierr = PetscLLClean(am,am,nzk,lnk,current_space->array,lnkbt);CHKERRQ(ierr);
2863 
2864     /* add the k-th row into il and jl */
2865     if (nzk > 1) {
2866       i     = current_space->array[1]; /* col value of the first nonzero element in U(k, k+1:am-1) */
2867       jl[k] = jl[i]; jl[i] = k;
2868       il[k] = ui[k] + 1;
2869     }
2870     ui_ptr[k] = current_space->array;
2871 
2872     current_space->array           += nzk;
2873     current_space->local_used      += nzk;
2874     current_space->local_remaining -= nzk;
2875 
2876     ui[k+1] = ui[k] + nzk;
2877   }
2878 
2879   ierr = ISRestoreIndices(perm,&rip);CHKERRQ(ierr);
2880   ierr = ISRestoreIndices(iperm,&riip);CHKERRQ(ierr);
2881   ierr = PetscFree4(ui_ptr,jl,il,cols);CHKERRQ(ierr);
2882 
2883   /* copy free_space into uj and free free_space; set ui, uj, udiag in new datastructure; */
2884   ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
2885   ierr = PetscFreeSpaceContiguous_Cholesky(&free_space,uj,am,ui,udiag);CHKERRQ(ierr); /* store matrix factor */
2886   ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
2887 
2888   /* put together the new matrix in MATSEQSBAIJ format */
2889 
2890   b               = (Mat_SeqSBAIJ*)fact->data;
2891   b->singlemalloc = PETSC_FALSE;
2892   b->free_a       = PETSC_TRUE;
2893   b->free_ij      = PETSC_TRUE;
2894 
2895   ierr = PetscMalloc1(ui[am]+1,&b->a);CHKERRQ(ierr);
2896 
2897   b->j         = uj;
2898   b->i         = ui;
2899   b->diag      = udiag;
2900   b->free_diag = PETSC_TRUE;
2901   b->ilen      = 0;
2902   b->imax      = 0;
2903   b->row       = perm;
2904   b->col       = perm;
2905 
2906   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2907   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
2908 
2909   b->icol          = iperm;
2910   b->pivotinblocks = PETSC_FALSE; /* need to get from MatFactorInfo */
2911 
2912   ierr = PetscMalloc1(am+1,&b->solve_work);CHKERRQ(ierr);
2913   ierr = PetscLogObjectMemory((PetscObject)fact,ui[am]*(sizeof(PetscInt)+sizeof(MatScalar)));CHKERRQ(ierr);
2914 
2915   b->maxnz = b->nz = ui[am];
2916 
2917   fact->info.factor_mallocs   = reallocs;
2918   fact->info.fill_ratio_given = fill;
2919   if (ai[am] != 0) {
2920     /* nonzeros in lower triangular part of A (including diagonals) = (ai[am]+am)/2 */
2921     fact->info.fill_ratio_needed = ((PetscReal)2*ui[am])/(ai[am]+am);
2922   } else {
2923     fact->info.fill_ratio_needed = 0.0;
2924   }
2925 #if defined(PETSC_USE_INFO)
2926   if (ai[am] != 0) {
2927     PetscReal af = fact->info.fill_ratio_needed;
2928     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)fill,(double)af);CHKERRQ(ierr);
2929     ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
2930     ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g) for best performance.\n",(double)af);CHKERRQ(ierr);
2931   } else {
2932     ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
2933   }
2934 #endif
2935   fact->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqAIJ;
2936   PetscFunctionReturn(0);
2937 }
2938 
2939 PetscErrorCode MatCholeskyFactorSymbolic_SeqAIJ_inplace(Mat fact,Mat A,IS perm,const MatFactorInfo *info)
2940 {
2941   Mat_SeqAIJ         *a = (Mat_SeqAIJ*)A->data;
2942   Mat_SeqSBAIJ       *b;
2943   PetscErrorCode     ierr;
2944   PetscBool          perm_identity,missing;
2945   PetscReal          fill = info->fill;
2946   const PetscInt     *rip,*riip;
2947   PetscInt           i,am=A->rmap->n,*ai=a->i,*aj=a->j,reallocs=0,prow;
2948   PetscInt           *jl,jmin,jmax,nzk,*ui,k,j,*il,nextprow;
2949   PetscInt           nlnk,*lnk,ncols,ncols_upper,*cols,*uj,**ui_ptr,*uj_ptr;
2950   PetscFreeSpaceList free_space=NULL,current_space=NULL;
2951   PetscBT            lnkbt;
2952   IS                 iperm;
2953 
2954   PetscFunctionBegin;
2955   if (A->rmap->n != A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Must be square matrix, rows %D columns %D",A->rmap->n,A->cmap->n);
2956   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
2957   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
2958 
2959   /* check whether perm is the identity mapping */
2960   ierr = ISIdentity(perm,&perm_identity);CHKERRQ(ierr);
2961   ierr = ISInvertPermutation(perm,PETSC_DECIDE,&iperm);CHKERRQ(ierr);
2962   ierr = ISGetIndices(iperm,&riip);CHKERRQ(ierr);
2963   ierr = ISGetIndices(perm,&rip);CHKERRQ(ierr);
2964 
2965   /* initialization */
2966   ierr  = PetscMalloc1(am+1,&ui);CHKERRQ(ierr);
2967   ui[0] = 0;
2968 
2969   /* jl: linked list for storing indices of the pivot rows
2970      il: il[i] points to the 1st nonzero entry of U(i,k:am-1) */
2971   ierr = PetscMalloc4(am,&ui_ptr,am,&jl,am,&il,am,&cols);CHKERRQ(ierr);
2972   for (i=0; i<am; i++) {
2973     jl[i] = am; il[i] = 0;
2974   }
2975 
2976   /* create and initialize a linked list for storing column indices of the active row k */
2977   nlnk = am + 1;
2978   ierr = PetscLLCreate(am,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
2979 
2980   /* initial FreeSpace size is fill*(ai[am]+1) */
2981   ierr          = PetscFreeSpaceGet(PetscRealIntMultTruncate(fill,ai[am]+1),&free_space);CHKERRQ(ierr);
2982   current_space = free_space;
2983 
2984   for (k=0; k<am; k++) {  /* for each active row k */
2985     /* initialize lnk by the column indices of row rip[k] of A */
2986     nzk   = 0;
2987     ncols = ai[rip[k]+1] - ai[rip[k]];
2988     if (!ncols) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_CH_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",rip[k],k);
2989     ncols_upper = 0;
2990     for (j=0; j<ncols; j++) {
2991       i = riip[*(aj + ai[rip[k]] + j)];
2992       if (i >= k) { /* only take upper triangular entry */
2993         cols[ncols_upper] = i;
2994         ncols_upper++;
2995       }
2996     }
2997     ierr = PetscLLAdd(ncols_upper,cols,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
2998     nzk += nlnk;
2999 
3000     /* update lnk by computing fill-in for each pivot row to be merged in */
3001     prow = jl[k]; /* 1st pivot row */
3002 
3003     while (prow < k) {
3004       nextprow = jl[prow];
3005       /* merge prow into k-th row */
3006       jmin   = il[prow] + 1; /* index of the 2nd nzero entry in U(prow,k:am-1) */
3007       jmax   = ui[prow+1];
3008       ncols  = jmax-jmin;
3009       uj_ptr = ui_ptr[prow] + jmin - ui[prow]; /* points to the 2nd nzero entry in U(prow,k:am-1) */
3010       ierr   = PetscLLAddSorted(ncols,uj_ptr,am,nlnk,lnk,lnkbt);CHKERRQ(ierr);
3011       nzk   += nlnk;
3012 
3013       /* update il and jl for prow */
3014       if (jmin < jmax) {
3015         il[prow] = jmin;
3016         j        = *uj_ptr; jl[prow] = jl[j]; jl[j] = prow;
3017       }
3018       prow = nextprow;
3019     }
3020 
3021     /* if free space is not available, make more free space */
3022     if (current_space->local_remaining<nzk) {
3023       i    = am - k + 1; /* num of unfactored rows */
3024       i    = PetscMin(i*nzk, i*(i-1)); /* i*nzk, i*(i-1): estimated and max additional space needed */
3025       ierr = PetscFreeSpaceGet(i,&current_space);CHKERRQ(ierr);
3026       reallocs++;
3027     }
3028 
3029     /* copy data into free space, then initialize lnk */
3030     ierr = PetscLLClean(am,am,nzk,lnk,current_space->array,lnkbt);CHKERRQ(ierr);
3031 
3032     /* add the k-th row into il and jl */
3033     if (nzk-1 > 0) {
3034       i     = current_space->array[1]; /* col value of the first nonzero element in U(k, k+1:am-1) */
3035       jl[k] = jl[i]; jl[i] = k;
3036       il[k] = ui[k] + 1;
3037     }
3038     ui_ptr[k] = current_space->array;
3039 
3040     current_space->array           += nzk;
3041     current_space->local_used      += nzk;
3042     current_space->local_remaining -= nzk;
3043 
3044     ui[k+1] = ui[k] + nzk;
3045   }
3046 
3047 #if defined(PETSC_USE_INFO)
3048   if (ai[am] != 0) {
3049     PetscReal af = (PetscReal)(ui[am])/((PetscReal)ai[am]);
3050     ierr = PetscInfo3(A,"Reallocs %D Fill ratio:given %g needed %g\n",reallocs,(double)fill,(double)af);CHKERRQ(ierr);
3051     ierr = PetscInfo1(A,"Run with -pc_factor_fill %g or use \n",(double)af);CHKERRQ(ierr);
3052     ierr = PetscInfo1(A,"PCFactorSetFill(pc,%g) for best performance.\n",(double)af);CHKERRQ(ierr);
3053   } else {
3054     ierr = PetscInfo(A,"Empty matrix\n");CHKERRQ(ierr);
3055   }
3056 #endif
3057 
3058   ierr = ISRestoreIndices(perm,&rip);CHKERRQ(ierr);
3059   ierr = ISRestoreIndices(iperm,&riip);CHKERRQ(ierr);
3060   ierr = PetscFree4(ui_ptr,jl,il,cols);CHKERRQ(ierr);
3061 
3062   /* destroy list of free space and other temporary array(s) */
3063   ierr = PetscMalloc1(ui[am]+1,&uj);CHKERRQ(ierr);
3064   ierr = PetscFreeSpaceContiguous(&free_space,uj);CHKERRQ(ierr);
3065   ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
3066 
3067   /* put together the new matrix in MATSEQSBAIJ format */
3068 
3069   b               = (Mat_SeqSBAIJ*)fact->data;
3070   b->singlemalloc = PETSC_FALSE;
3071   b->free_a       = PETSC_TRUE;
3072   b->free_ij      = PETSC_TRUE;
3073 
3074   ierr = PetscMalloc1(ui[am]+1,&b->a);CHKERRQ(ierr);
3075 
3076   b->j    = uj;
3077   b->i    = ui;
3078   b->diag = 0;
3079   b->ilen = 0;
3080   b->imax = 0;
3081   b->row  = perm;
3082   b->col  = perm;
3083 
3084   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
3085   ierr = PetscObjectReference((PetscObject)perm);CHKERRQ(ierr);
3086 
3087   b->icol          = iperm;
3088   b->pivotinblocks = PETSC_FALSE; /* need to get from MatFactorInfo */
3089 
3090   ierr     = PetscMalloc1(am+1,&b->solve_work);CHKERRQ(ierr);
3091   ierr     = PetscLogObjectMemory((PetscObject)fact,(ui[am]-am)*(sizeof(PetscInt)+sizeof(MatScalar)));CHKERRQ(ierr);
3092   b->maxnz = b->nz = ui[am];
3093 
3094   fact->info.factor_mallocs   = reallocs;
3095   fact->info.fill_ratio_given = fill;
3096   if (ai[am] != 0) {
3097     fact->info.fill_ratio_needed = ((PetscReal)ui[am])/((PetscReal)ai[am]);
3098   } else {
3099     fact->info.fill_ratio_needed = 0.0;
3100   }
3101   fact->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqAIJ_inplace;
3102   PetscFunctionReturn(0);
3103 }
3104 
3105 PetscErrorCode MatSolve_SeqAIJ_NaturalOrdering(Mat A,Vec bb,Vec xx)
3106 {
3107   Mat_SeqAIJ        *a = (Mat_SeqAIJ*)A->data;
3108   PetscErrorCode    ierr;
3109   PetscInt          n   = A->rmap->n;
3110   const PetscInt    *ai = a->i,*aj = a->j,*adiag = a->diag,*vi;
3111   PetscScalar       *x,sum;
3112   const PetscScalar *b;
3113   const MatScalar   *aa = a->a,*v;
3114   PetscInt          i,nz;
3115 
3116   PetscFunctionBegin;
3117   if (!n) PetscFunctionReturn(0);
3118 
3119   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
3120   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
3121 
3122   /* forward solve the lower triangular */
3123   x[0] = b[0];
3124   v    = aa;
3125   vi   = aj;
3126   for (i=1; i<n; i++) {
3127     nz  = ai[i+1] - ai[i];
3128     sum = b[i];
3129     PetscSparseDenseMinusDot(sum,x,v,vi,nz);
3130     v   += nz;
3131     vi  += nz;
3132     x[i] = sum;
3133   }
3134 
3135   /* backward solve the upper triangular */
3136   for (i=n-1; i>=0; i--) {
3137     v   = aa + adiag[i+1] + 1;
3138     vi  = aj + adiag[i+1] + 1;
3139     nz  = adiag[i] - adiag[i+1]-1;
3140     sum = x[i];
3141     PetscSparseDenseMinusDot(sum,x,v,vi,nz);
3142     x[i] = sum*v[nz]; /* x[i]=aa[adiag[i]]*sum; v++; */
3143   }
3144 
3145   ierr = PetscLogFlops(2.0*a->nz - A->cmap->n);CHKERRQ(ierr);
3146   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
3147   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
3148   PetscFunctionReturn(0);
3149 }
3150 
3151 PetscErrorCode MatSolve_SeqAIJ(Mat A,Vec bb,Vec xx)
3152 {
3153   Mat_SeqAIJ        *a    = (Mat_SeqAIJ*)A->data;
3154   IS                iscol = a->col,isrow = a->row;
3155   PetscErrorCode    ierr;
3156   PetscInt          i,n=A->rmap->n,*vi,*ai=a->i,*aj=a->j,*adiag = a->diag,nz;
3157   const PetscInt    *rout,*cout,*r,*c;
3158   PetscScalar       *x,*tmp,sum;
3159   const PetscScalar *b;
3160   const MatScalar   *aa = a->a,*v;
3161 
3162   PetscFunctionBegin;
3163   if (!n) PetscFunctionReturn(0);
3164 
3165   ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr);
3166   ierr = VecGetArrayWrite(xx,&x);CHKERRQ(ierr);
3167   tmp  = a->solve_work;
3168 
3169   ierr = ISGetIndices(isrow,&rout);CHKERRQ(ierr); r = rout;
3170   ierr = ISGetIndices(iscol,&cout);CHKERRQ(ierr); c = cout;
3171 
3172   /* forward solve the lower triangular */
3173   tmp[0] = b[r[0]];
3174   v      = aa;
3175   vi     = aj;
3176   for (i=1; i<n; i++) {
3177     nz  = ai[i+1] - ai[i];
3178     sum = b[r[i]];
3179     PetscSparseDenseMinusDot(sum,tmp,v,vi,nz);
3180     tmp[i] = sum;
3181     v     += nz; vi += nz;
3182   }
3183 
3184   /* backward solve the upper triangular */
3185   for (i=n-1; i>=0; i--) {
3186     v   = aa + adiag[i+1]+1;
3187     vi  = aj + adiag[i+1]+1;
3188     nz  = adiag[i]-adiag[i+1]-1;
3189     sum = tmp[i];
3190     PetscSparseDenseMinusDot(sum,tmp,v,vi,nz);
3191     x[c[i]] = tmp[i] = sum*v[nz]; /* v[nz] = aa[adiag[i]] */
3192   }
3193 
3194   ierr = ISRestoreIndices(isrow,&rout);CHKERRQ(ierr);
3195   ierr = ISRestoreIndices(iscol,&cout);CHKERRQ(ierr);
3196   ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr);
3197   ierr = VecRestoreArrayWrite(xx,&x);CHKERRQ(ierr);
3198   ierr = PetscLogFlops(2*a->nz - A->cmap->n);CHKERRQ(ierr);
3199   PetscFunctionReturn(0);
3200 }
3201 
3202 /*
3203     This will get a new name and become a varient of MatILUFactor_SeqAIJ() there is no longer separate functions in the matrix function table for dt factors
3204 */
3205 PetscErrorCode MatILUDTFactor_SeqAIJ(Mat A,IS isrow,IS iscol,const MatFactorInfo *info,Mat *fact)
3206 {
3207   Mat            B = *fact;
3208   Mat_SeqAIJ     *a=(Mat_SeqAIJ*)A->data,*b;
3209   IS             isicol;
3210   PetscErrorCode ierr;
3211   const PetscInt *r,*ic;
3212   PetscInt       i,n=A->rmap->n,*ai=a->i,*aj=a->j,*ajtmp,*adiag;
3213   PetscInt       *bi,*bj,*bdiag,*bdiag_rev;
3214   PetscInt       row,nzi,nzi_bl,nzi_bu,*im,nzi_al,nzi_au;
3215   PetscInt       nlnk,*lnk;
3216   PetscBT        lnkbt;
3217   PetscBool      row_identity,icol_identity;
3218   MatScalar      *aatmp,*pv,*batmp,*ba,*rtmp,*pc,multiplier,*vtmp,diag_tmp;
3219   const PetscInt *ics;
3220   PetscInt       j,nz,*pj,*bjtmp,k,ncut,*jtmp;
3221   PetscReal      dt     =info->dt,shift=info->shiftamount;
3222   PetscInt       dtcount=(PetscInt)info->dtcount,nnz_max;
3223   PetscBool      missing;
3224 
3225   PetscFunctionBegin;
3226   if (dt      == PETSC_DEFAULT) dt = 0.005;
3227   if (dtcount == PETSC_DEFAULT) dtcount = (PetscInt)(1.5*a->rmax);
3228 
3229   /* ------- symbolic factorization, can be reused ---------*/
3230   ierr = MatMissingDiagonal(A,&missing,&i);CHKERRQ(ierr);
3231   if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry %D",i);
3232   adiag=a->diag;
3233 
3234   ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr);
3235 
3236   /* bdiag is location of diagonal in factor */
3237   ierr = PetscMalloc1(n+1,&bdiag);CHKERRQ(ierr);     /* becomes b->diag */
3238   ierr = PetscMalloc1(n+1,&bdiag_rev);CHKERRQ(ierr); /* temporary */
3239 
3240   /* allocate row pointers bi */
3241   ierr = PetscMalloc1(2*n+2,&bi);CHKERRQ(ierr);
3242 
3243   /* allocate bj and ba; max num of nonzero entries is (ai[n]+2*n*dtcount+2) */
3244   if (dtcount > n-1) dtcount = n-1; /* diagonal is excluded */
3245   nnz_max = ai[n]+2*n*dtcount+2;
3246 
3247   ierr = PetscMalloc1(nnz_max+1,&bj);CHKERRQ(ierr);
3248   ierr = PetscMalloc1(nnz_max+1,&ba);CHKERRQ(ierr);
3249 
3250   /* put together the new matrix */
3251   ierr = MatSeqAIJSetPreallocation_SeqAIJ(B,MAT_SKIP_ALLOCATION,NULL);CHKERRQ(ierr);
3252   ierr = PetscLogObjectParent((PetscObject)B,(PetscObject)isicol);CHKERRQ(ierr);
3253   b    = (Mat_SeqAIJ*)B->data;
3254 
3255   b->free_a       = PETSC_TRUE;
3256   b->free_ij      = PETSC_TRUE;
3257   b->singlemalloc = PETSC_FALSE;
3258 
3259   b->a    = ba;
3260   b->j    = bj;
3261   b->i    = bi;
3262   b->diag = bdiag;
3263   b->ilen = 0;
3264   b->imax = 0;
3265   b->row  = isrow;
3266   b->col  = iscol;
3267   ierr    = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr);
3268   ierr    = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr);
3269   b->icol = isicol;
3270 
3271   ierr     = PetscMalloc1(n+1,&b->solve_work);CHKERRQ(ierr);
3272   ierr     = PetscLogObjectMemory((PetscObject)B,nnz_max*(sizeof(PetscInt)+sizeof(MatScalar)));CHKERRQ(ierr);
3273   b->maxnz = nnz_max;
3274 
3275   B->factortype            = MAT_FACTOR_ILUDT;
3276   B->info.factor_mallocs   = 0;
3277   B->info.fill_ratio_given = ((PetscReal)nnz_max)/((PetscReal)ai[n]);
3278   /* ------- end of symbolic factorization ---------*/
3279 
3280   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
3281   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
3282   ics  = ic;
3283 
3284   /* linked list for storing column indices of the active row */
3285   nlnk = n + 1;
3286   ierr = PetscLLCreate(n,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
3287 
3288   /* im: used by PetscLLAddSortedLU(); jtmp: working array for column indices of active row */
3289   ierr = PetscMalloc2(n,&im,n,&jtmp);CHKERRQ(ierr);
3290   /* rtmp, vtmp: working arrays for sparse and contiguous row entries of active row */
3291   ierr = PetscMalloc2(n,&rtmp,n,&vtmp);CHKERRQ(ierr);
3292   ierr = PetscArrayzero(rtmp,n);CHKERRQ(ierr);
3293 
3294   bi[0]        = 0;
3295   bdiag[0]     = nnz_max-1; /* location of diag[0] in factor B */
3296   bdiag_rev[n] = bdiag[0];
3297   bi[2*n+1]    = bdiag[0]+1; /* endof bj and ba array */
3298   for (i=0; i<n; i++) {
3299     /* copy initial fill into linked list */
3300     nzi = ai[r[i]+1] - ai[r[i]];
3301     if (!nzi) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Empty row in matrix: row in original ordering %D in permuted ordering %D",r[i],i);
3302     nzi_al = adiag[r[i]] - ai[r[i]];
3303     nzi_au = ai[r[i]+1] - adiag[r[i]] -1;
3304     ajtmp  = aj + ai[r[i]];
3305     ierr   = PetscLLAddPerm(nzi,ajtmp,ic,n,nlnk,lnk,lnkbt);CHKERRQ(ierr);
3306 
3307     /* load in initial (unfactored row) */
3308     aatmp = a->a + ai[r[i]];
3309     for (j=0; j<nzi; j++) {
3310       rtmp[ics[*ajtmp++]] = *aatmp++;
3311     }
3312 
3313     /* add pivot rows into linked list */
3314     row = lnk[n];
3315     while (row < i) {
3316       nzi_bl = bi[row+1] - bi[row] + 1;
3317       bjtmp  = bj + bdiag[row+1]+1; /* points to 1st column next to the diagonal in U */
3318       ierr   = PetscLLAddSortedLU(bjtmp,row,nlnk,lnk,lnkbt,i,nzi_bl,im);CHKERRQ(ierr);
3319       nzi   += nlnk;
3320       row    = lnk[row];
3321     }
3322 
3323     /* copy data from lnk into jtmp, then initialize lnk */
3324     ierr = PetscLLClean(n,n,nzi,lnk,jtmp,lnkbt);CHKERRQ(ierr);
3325 
3326     /* numerical factorization */
3327     bjtmp = jtmp;
3328     row   = *bjtmp++; /* 1st pivot row */
3329     while (row < i) {
3330       pc         = rtmp + row;
3331       pv         = ba + bdiag[row]; /* 1./(diag of the pivot row) */
3332       multiplier = (*pc) * (*pv);
3333       *pc        = multiplier;
3334       if (PetscAbsScalar(*pc) > dt) { /* apply tolerance dropping rule */
3335         pj = bj + bdiag[row+1] + 1;         /* point to 1st entry of U(row,:) */
3336         pv = ba + bdiag[row+1] + 1;
3337         /* if (multiplier < -1.0 or multiplier >1.0) printf("row/prow %d, %d, multiplier %g\n",i,row,multiplier); */
3338         nz = bdiag[row] - bdiag[row+1] - 1;         /* num of entries in U(row,:), excluding diagonal */
3339         for (j=0; j<nz; j++) rtmp[*pj++] -= multiplier * (*pv++);
3340         ierr = PetscLogFlops(1+2*nz);CHKERRQ(ierr);
3341       }
3342       row = *bjtmp++;
3343     }
3344 
3345     /* copy sparse rtmp into contiguous vtmp; separate L and U part */
3346     diag_tmp = rtmp[i];  /* save diagonal value - may not needed?? */
3347     nzi_bl   = 0; j = 0;
3348     while (jtmp[j] < i) { /* Note: jtmp is sorted */
3349       vtmp[j] = rtmp[jtmp[j]]; rtmp[jtmp[j]]=0.0;
3350       nzi_bl++; j++;
3351     }
3352     nzi_bu = nzi - nzi_bl -1;
3353     while (j < nzi) {
3354       vtmp[j] = rtmp[jtmp[j]]; rtmp[jtmp[j]]=0.0;
3355       j++;
3356     }
3357 
3358     bjtmp = bj + bi[i];
3359     batmp = ba + bi[i];
3360     /* apply level dropping rule to L part */
3361     ncut = nzi_al + dtcount;
3362     if (ncut < nzi_bl) {
3363       ierr = PetscSortSplit(ncut,nzi_bl,vtmp,jtmp);CHKERRQ(ierr);
3364       ierr = PetscSortIntWithScalarArray(ncut,jtmp,vtmp);CHKERRQ(ierr);
3365     } else {
3366       ncut = nzi_bl;
3367     }
3368     for (j=0; j<ncut; j++) {
3369       bjtmp[j] = jtmp[j];
3370       batmp[j] = vtmp[j];
3371       /* printf(" (%d,%g),",bjtmp[j],batmp[j]); */
3372     }
3373     bi[i+1] = bi[i] + ncut;
3374     nzi     = ncut + 1;
3375 
3376     /* apply level dropping rule to U part */
3377     ncut = nzi_au + dtcount;
3378     if (ncut < nzi_bu) {
3379       ierr = PetscSortSplit(ncut,nzi_bu,vtmp+nzi_bl+1,jtmp+nzi_bl+1);CHKERRQ(ierr);
3380       ierr = PetscSortIntWithScalarArray(ncut,jtmp+nzi_bl+1,vtmp+nzi_bl+1);CHKERRQ(ierr);
3381     } else {
3382       ncut = nzi_bu;
3383     }
3384     nzi += ncut;
3385 
3386     /* mark bdiagonal */
3387     bdiag[i+1]       = bdiag[i] - (ncut + 1);
3388     bdiag_rev[n-i-1] = bdiag[i+1];
3389     bi[2*n - i]      = bi[2*n - i +1] - (ncut + 1);
3390     bjtmp            = bj + bdiag[i];
3391     batmp            = ba + bdiag[i];
3392     *bjtmp           = i;
3393     *batmp           = diag_tmp; /* rtmp[i]; */
3394     if (*batmp == 0.0) {
3395       *batmp = dt+shift;
3396       /* printf(" row %d add shift %g\n",i,shift); */
3397     }
3398     *batmp = 1.0/(*batmp); /* invert diagonal entries for simplier triangular solves */
3399     /* printf(" (%d,%g),",*bjtmp,*batmp); */
3400 
3401     bjtmp = bj + bdiag[i+1]+1;
3402     batmp = ba + bdiag[i+1]+1;
3403     for (k=0; k<ncut; k++) {
3404       bjtmp[k] = jtmp[nzi_bl+1+k];
3405       batmp[k] = vtmp[nzi_bl+1+k];
3406       /* printf(" (%d,%g),",bjtmp[k],batmp[k]); */
3407     }
3408     /* printf("\n"); */
3409 
3410     im[i] = nzi;   /* used by PetscLLAddSortedLU() */
3411     /*
3412     printf("row %d: bi %d, bdiag %d\n",i,bi[i],bdiag[i]);
3413     printf(" ----------------------------\n");
3414     */
3415   } /* for (i=0; i<n; i++) */
3416     /* printf("end of L %d, beginning of U %d\n",bi[n],bdiag[n]); */
3417   if (bi[n] >= bdiag[n]) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"end of L array %d cannot >= the beginning of U array %d",bi[n],bdiag[n]);
3418 
3419   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
3420   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
3421 
3422   ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr);
3423   ierr = PetscFree2(im,jtmp);CHKERRQ(ierr);
3424   ierr = PetscFree2(rtmp,vtmp);CHKERRQ(ierr);
3425   ierr = PetscFree(bdiag_rev);CHKERRQ(ierr);
3426 
3427   ierr     = PetscLogFlops(B->cmap->n);CHKERRQ(ierr);
3428   b->maxnz = b->nz = bi[n] + bdiag[0] - bdiag[n];
3429 
3430   ierr = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
3431   ierr = ISIdentity(isicol,&icol_identity);CHKERRQ(ierr);
3432   if (row_identity && icol_identity) {
3433     B->ops->solve = MatSolve_SeqAIJ_NaturalOrdering;
3434   } else {
3435     B->ops->solve = MatSolve_SeqAIJ;
3436   }
3437 
3438   B->ops->solveadd          = 0;
3439   B->ops->solvetranspose    = 0;
3440   B->ops->solvetransposeadd = 0;
3441   B->ops->matsolve          = 0;
3442   B->assembled              = PETSC_TRUE;
3443   B->preallocated           = PETSC_TRUE;
3444   PetscFunctionReturn(0);
3445 }
3446 
3447 /* a wraper of MatILUDTFactor_SeqAIJ() */
3448 /*
3449     This will get a new name and become a varient of MatILUFactor_SeqAIJ() there is no longer separate functions in the matrix function table for dt factors
3450 */
3451 
3452 PetscErrorCode  MatILUDTFactorSymbolic_SeqAIJ(Mat fact,Mat A,IS row,IS col,const MatFactorInfo *info)
3453 {
3454   PetscErrorCode ierr;
3455 
3456   PetscFunctionBegin;
3457   ierr = MatILUDTFactor_SeqAIJ(A,row,col,info,&fact);CHKERRQ(ierr);
3458   PetscFunctionReturn(0);
3459 }
3460 
3461 /*
3462    same as MatLUFactorNumeric_SeqAIJ(), except using contiguous array matrix factors
3463    - intend to replace existing MatLUFactorNumeric_SeqAIJ()
3464 */
3465 /*
3466     This will get a new name and become a varient of MatILUFactor_SeqAIJ() there is no longer separate functions in the matrix function table for dt factors
3467 */
3468 
3469 PetscErrorCode  MatILUDTFactorNumeric_SeqAIJ(Mat fact,Mat A,const MatFactorInfo *info)
3470 {
3471   Mat            C     =fact;
3472   Mat_SeqAIJ     *a    =(Mat_SeqAIJ*)A->data,*b=(Mat_SeqAIJ*)C->data;
3473   IS             isrow = b->row,isicol = b->icol;
3474   PetscErrorCode ierr;
3475   const PetscInt *r,*ic,*ics;
3476   PetscInt       i,j,k,n=A->rmap->n,*ai=a->i,*aj=a->j,*bi=b->i,*bj=b->j;
3477   PetscInt       *ajtmp,*bjtmp,nz,nzl,nzu,row,*bdiag = b->diag,*pj;
3478   MatScalar      *rtmp,*pc,multiplier,*v,*pv,*aa=a->a;
3479   PetscReal      dt=info->dt,shift=info->shiftamount;
3480   PetscBool      row_identity, col_identity;
3481 
3482   PetscFunctionBegin;
3483   ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr);
3484   ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr);
3485   ierr = PetscMalloc1(n+1,&rtmp);CHKERRQ(ierr);
3486   ics  = ic;
3487 
3488   for (i=0; i<n; i++) {
3489     /* initialize rtmp array */
3490     nzl   = bi[i+1] - bi[i];       /* num of nozeros in L(i,:) */
3491     bjtmp = bj + bi[i];
3492     for  (j=0; j<nzl; j++) rtmp[*bjtmp++] = 0.0;
3493     rtmp[i] = 0.0;
3494     nzu     = bdiag[i] - bdiag[i+1]; /* num of nozeros in U(i,:) */
3495     bjtmp   = bj + bdiag[i+1] + 1;
3496     for  (j=0; j<nzu; j++) rtmp[*bjtmp++] = 0.0;
3497 
3498     /* load in initial unfactored row of A */
3499     /* printf("row %d\n",i); */
3500     nz    = ai[r[i]+1] - ai[r[i]];
3501     ajtmp = aj + ai[r[i]];
3502     v     = aa + ai[r[i]];
3503     for (j=0; j<nz; j++) {
3504       rtmp[ics[*ajtmp++]] = v[j];
3505       /* printf(" (%d,%g),",ics[ajtmp[j]],rtmp[ics[ajtmp[j]]]); */
3506     }
3507     /* printf("\n"); */
3508 
3509     /* numerical factorization */
3510     bjtmp = bj + bi[i]; /* point to 1st entry of L(i,:) */
3511     nzl   = bi[i+1] - bi[i]; /* num of entries in L(i,:) */
3512     k     = 0;
3513     while (k < nzl) {
3514       row = *bjtmp++;
3515       /* printf("  prow %d\n",row); */
3516       pc         = rtmp + row;
3517       pv         = b->a + bdiag[row]; /* 1./(diag of the pivot row) */
3518       multiplier = (*pc) * (*pv);
3519       *pc        = multiplier;
3520       if (PetscAbsScalar(multiplier) > dt) {
3521         pj = bj + bdiag[row+1] + 1;         /* point to 1st entry of U(row,:) */
3522         pv = b->a + bdiag[row+1] + 1;
3523         nz = bdiag[row] - bdiag[row+1] - 1;         /* num of entries in U(row,:), excluding diagonal */
3524         for (j=0; j<nz; j++) rtmp[*pj++] -= multiplier * (*pv++);
3525         ierr = PetscLogFlops(1+2*nz);CHKERRQ(ierr);
3526       }
3527       k++;
3528     }
3529 
3530     /* finished row so stick it into b->a */
3531     /* L-part */
3532     pv  = b->a + bi[i];
3533     pj  = bj + bi[i];
3534     nzl = bi[i+1] - bi[i];
3535     for (j=0; j<nzl; j++) {
3536       pv[j] = rtmp[pj[j]];
3537       /* printf(" (%d,%g),",pj[j],pv[j]); */
3538     }
3539 
3540     /* diagonal: invert diagonal entries for simplier triangular solves */
3541     if (rtmp[i] == 0.0) rtmp[i] = dt+shift;
3542     b->a[bdiag[i]] = 1.0/rtmp[i];
3543     /* printf(" (%d,%g),",i,b->a[bdiag[i]]); */
3544 
3545     /* U-part */
3546     pv  = b->a + bdiag[i+1] + 1;
3547     pj  = bj + bdiag[i+1] + 1;
3548     nzu = bdiag[i] - bdiag[i+1] - 1;
3549     for (j=0; j<nzu; j++) {
3550       pv[j] = rtmp[pj[j]];
3551       /* printf(" (%d,%g),",pj[j],pv[j]); */
3552     }
3553     /* printf("\n"); */
3554   }
3555 
3556   ierr = PetscFree(rtmp);CHKERRQ(ierr);
3557   ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr);
3558   ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr);
3559 
3560   ierr = ISIdentity(isrow,&row_identity);CHKERRQ(ierr);
3561   ierr = ISIdentity(isicol,&col_identity);CHKERRQ(ierr);
3562   if (row_identity && col_identity) {
3563     C->ops->solve = MatSolve_SeqAIJ_NaturalOrdering;
3564   } else {
3565     C->ops->solve = MatSolve_SeqAIJ;
3566   }
3567   C->ops->solveadd          = 0;
3568   C->ops->solvetranspose    = 0;
3569   C->ops->solvetransposeadd = 0;
3570   C->ops->matsolve          = 0;
3571   C->assembled              = PETSC_TRUE;
3572   C->preallocated           = PETSC_TRUE;
3573 
3574   ierr = PetscLogFlops(C->cmap->n);CHKERRQ(ierr);
3575   PetscFunctionReturn(0);
3576 }
3577