Actual source code: aij.c
petsc-master 2019-07-18
2: /*
3: Defines the basic matrix operations for the AIJ (compressed row)
4: matrix storage format.
5: */
8: #include <../src/mat/impls/aij/seq/aij.h>
9: #include <petscblaslapack.h>
10: #include <petscbt.h>
11: #include <petsc/private/kernels/blocktranspose.h>
13: PetscErrorCode MatSeqAIJSetTypeFromOptions(Mat A)
14: {
15: PetscErrorCode ierr;
16: PetscBool flg;
17: char type[256];
20: PetscObjectOptionsBegin((PetscObject)A);
21: PetscOptionsFList("-mat_seqaij_type","Matrix SeqAIJ type","MatSeqAIJSetType",MatSeqAIJList,"seqaij",type,256,&flg);
22: if (flg) {
23: MatSeqAIJSetType(A,type);
24: }
25: PetscOptionsEnd();
26: return(0);
27: }
29: PetscErrorCode MatGetColumnNorms_SeqAIJ(Mat A,NormType type,PetscReal *norms)
30: {
32: PetscInt i,m,n;
33: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)A->data;
36: MatGetSize(A,&m,&n);
37: PetscArrayzero(norms,n);
38: if (type == NORM_2) {
39: for (i=0; i<aij->i[m]; i++) {
40: norms[aij->j[i]] += PetscAbsScalar(aij->a[i]*aij->a[i]);
41: }
42: } else if (type == NORM_1) {
43: for (i=0; i<aij->i[m]; i++) {
44: norms[aij->j[i]] += PetscAbsScalar(aij->a[i]);
45: }
46: } else if (type == NORM_INFINITY) {
47: for (i=0; i<aij->i[m]; i++) {
48: norms[aij->j[i]] = PetscMax(PetscAbsScalar(aij->a[i]),norms[aij->j[i]]);
49: }
50: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown NormType");
52: if (type == NORM_2) {
53: for (i=0; i<n; i++) norms[i] = PetscSqrtReal(norms[i]);
54: }
55: return(0);
56: }
58: PetscErrorCode MatFindOffBlockDiagonalEntries_SeqAIJ(Mat A,IS *is)
59: {
60: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
61: PetscInt i,m=A->rmap->n,cnt = 0, bs = A->rmap->bs;
62: const PetscInt *jj = a->j,*ii = a->i;
63: PetscInt *rows;
64: PetscErrorCode ierr;
67: for (i=0; i<m; i++) {
68: if ((ii[i] != ii[i+1]) && ((jj[ii[i]] < bs*(i/bs)) || (jj[ii[i+1]-1] > bs*((i+bs)/bs)-1))) {
69: cnt++;
70: }
71: }
72: PetscMalloc1(cnt,&rows);
73: cnt = 0;
74: for (i=0; i<m; i++) {
75: if ((ii[i] != ii[i+1]) && ((jj[ii[i]] < bs*(i/bs)) || (jj[ii[i+1]-1] > bs*((i+bs)/bs)-1))) {
76: rows[cnt] = i;
77: cnt++;
78: }
79: }
80: ISCreateGeneral(PETSC_COMM_SELF,cnt,rows,PETSC_OWN_POINTER,is);
81: return(0);
82: }
84: PetscErrorCode MatFindZeroDiagonals_SeqAIJ_Private(Mat A,PetscInt *nrows,PetscInt **zrows)
85: {
86: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
87: const MatScalar *aa = a->a;
88: PetscInt i,m=A->rmap->n,cnt = 0;
89: const PetscInt *ii = a->i,*jj = a->j,*diag;
90: PetscInt *rows;
91: PetscErrorCode ierr;
94: MatMarkDiagonal_SeqAIJ(A);
95: diag = a->diag;
96: for (i=0; i<m; i++) {
97: if ((diag[i] >= ii[i+1]) || (jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) {
98: cnt++;
99: }
100: }
101: PetscMalloc1(cnt,&rows);
102: cnt = 0;
103: for (i=0; i<m; i++) {
104: if ((diag[i] >= ii[i+1]) || (jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) {
105: rows[cnt++] = i;
106: }
107: }
108: *nrows = cnt;
109: *zrows = rows;
110: return(0);
111: }
113: PetscErrorCode MatFindZeroDiagonals_SeqAIJ(Mat A,IS *zrows)
114: {
115: PetscInt nrows,*rows;
119: *zrows = NULL;
120: MatFindZeroDiagonals_SeqAIJ_Private(A,&nrows,&rows);
121: ISCreateGeneral(PetscObjectComm((PetscObject)A),nrows,rows,PETSC_OWN_POINTER,zrows);
122: return(0);
123: }
125: PetscErrorCode MatFindNonzeroRows_SeqAIJ(Mat A,IS *keptrows)
126: {
127: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
128: const MatScalar *aa;
129: PetscInt m=A->rmap->n,cnt = 0;
130: const PetscInt *ii;
131: PetscInt n,i,j,*rows;
132: PetscErrorCode ierr;
135: *keptrows = 0;
136: ii = a->i;
137: for (i=0; i<m; i++) {
138: n = ii[i+1] - ii[i];
139: if (!n) {
140: cnt++;
141: goto ok1;
142: }
143: aa = a->a + ii[i];
144: for (j=0; j<n; j++) {
145: if (aa[j] != 0.0) goto ok1;
146: }
147: cnt++;
148: ok1:;
149: }
150: if (!cnt) return(0);
151: PetscMalloc1(A->rmap->n-cnt,&rows);
152: cnt = 0;
153: for (i=0; i<m; i++) {
154: n = ii[i+1] - ii[i];
155: if (!n) continue;
156: aa = a->a + ii[i];
157: for (j=0; j<n; j++) {
158: if (aa[j] != 0.0) {
159: rows[cnt++] = i;
160: break;
161: }
162: }
163: }
164: ISCreateGeneral(PETSC_COMM_SELF,cnt,rows,PETSC_OWN_POINTER,keptrows);
165: return(0);
166: }
168: PetscErrorCode MatDiagonalSet_SeqAIJ(Mat Y,Vec D,InsertMode is)
169: {
170: PetscErrorCode ierr;
171: Mat_SeqAIJ *aij = (Mat_SeqAIJ*) Y->data;
172: PetscInt i,m = Y->rmap->n;
173: const PetscInt *diag;
174: MatScalar *aa = aij->a;
175: const PetscScalar *v;
176: PetscBool missing;
179: if (Y->assembled) {
180: MatMissingDiagonal_SeqAIJ(Y,&missing,NULL);
181: if (!missing) {
182: diag = aij->diag;
183: VecGetArrayRead(D,&v);
184: if (is == INSERT_VALUES) {
185: for (i=0; i<m; i++) {
186: aa[diag[i]] = v[i];
187: }
188: } else {
189: for (i=0; i<m; i++) {
190: aa[diag[i]] += v[i];
191: }
192: }
193: VecRestoreArrayRead(D,&v);
194: return(0);
195: }
196: MatSeqAIJInvalidateDiagonal(Y);
197: }
198: MatDiagonalSet_Default(Y,D,is);
199: return(0);
200: }
202: PetscErrorCode MatGetRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *m,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
203: {
204: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
206: PetscInt i,ishift;
209: *m = A->rmap->n;
210: if (!ia) return(0);
211: ishift = 0;
212: if (symmetric && !A->structurally_symmetric) {
213: MatToSymmetricIJ_SeqAIJ(A->rmap->n,a->i,a->j,PETSC_TRUE,ishift,oshift,(PetscInt**)ia,(PetscInt**)ja);
214: } else if (oshift == 1) {
215: PetscInt *tia;
216: PetscInt nz = a->i[A->rmap->n];
217: /* malloc space and add 1 to i and j indices */
218: PetscMalloc1(A->rmap->n+1,&tia);
219: for (i=0; i<A->rmap->n+1; i++) tia[i] = a->i[i] + 1;
220: *ia = tia;
221: if (ja) {
222: PetscInt *tja;
223: PetscMalloc1(nz+1,&tja);
224: for (i=0; i<nz; i++) tja[i] = a->j[i] + 1;
225: *ja = tja;
226: }
227: } else {
228: *ia = a->i;
229: if (ja) *ja = a->j;
230: }
231: return(0);
232: }
234: PetscErrorCode MatRestoreRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
235: {
239: if (!ia) return(0);
240: if ((symmetric && !A->structurally_symmetric) || oshift == 1) {
241: PetscFree(*ia);
242: if (ja) {PetscFree(*ja);}
243: }
244: return(0);
245: }
247: PetscErrorCode MatGetColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
248: {
249: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
251: PetscInt i,*collengths,*cia,*cja,n = A->cmap->n,m = A->rmap->n;
252: PetscInt nz = a->i[m],row,*jj,mr,col;
255: *nn = n;
256: if (!ia) return(0);
257: if (symmetric) {
258: MatToSymmetricIJ_SeqAIJ(A->rmap->n,a->i,a->j,PETSC_TRUE,0,oshift,(PetscInt**)ia,(PetscInt**)ja);
259: } else {
260: PetscCalloc1(n,&collengths);
261: PetscMalloc1(n+1,&cia);
262: PetscMalloc1(nz,&cja);
263: jj = a->j;
264: for (i=0; i<nz; i++) {
265: collengths[jj[i]]++;
266: }
267: cia[0] = oshift;
268: for (i=0; i<n; i++) {
269: cia[i+1] = cia[i] + collengths[i];
270: }
271: PetscArrayzero(collengths,n);
272: jj = a->j;
273: for (row=0; row<m; row++) {
274: mr = a->i[row+1] - a->i[row];
275: for (i=0; i<mr; i++) {
276: col = *jj++;
278: cja[cia[col] + collengths[col]++ - oshift] = row + oshift;
279: }
280: }
281: PetscFree(collengths);
282: *ia = cia; *ja = cja;
283: }
284: return(0);
285: }
287: PetscErrorCode MatRestoreColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
288: {
292: if (!ia) return(0);
294: PetscFree(*ia);
295: PetscFree(*ja);
296: return(0);
297: }
299: /*
300: MatGetColumnIJ_SeqAIJ_Color() and MatRestoreColumnIJ_SeqAIJ_Color() are customized from
301: MatGetColumnIJ_SeqAIJ() and MatRestoreColumnIJ_SeqAIJ() by adding an output
302: spidx[], index of a->a, to be used in MatTransposeColoringCreate_SeqAIJ() and MatFDColoringCreate_SeqXAIJ()
303: */
304: PetscErrorCode MatGetColumnIJ_SeqAIJ_Color(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscInt *spidx[],PetscBool *done)
305: {
306: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
308: PetscInt i,*collengths,*cia,*cja,n = A->cmap->n,m = A->rmap->n;
309: PetscInt nz = a->i[m],row,mr,col,tmp;
310: PetscInt *cspidx;
311: const PetscInt *jj;
314: *nn = n;
315: if (!ia) return(0);
317: PetscCalloc1(n,&collengths);
318: PetscMalloc1(n+1,&cia);
319: PetscMalloc1(nz,&cja);
320: PetscMalloc1(nz,&cspidx);
321: jj = a->j;
322: for (i=0; i<nz; i++) {
323: collengths[jj[i]]++;
324: }
325: cia[0] = oshift;
326: for (i=0; i<n; i++) {
327: cia[i+1] = cia[i] + collengths[i];
328: }
329: PetscArrayzero(collengths,n);
330: jj = a->j;
331: for (row=0; row<m; row++) {
332: mr = a->i[row+1] - a->i[row];
333: for (i=0; i<mr; i++) {
334: col = *jj++;
335: tmp = cia[col] + collengths[col]++ - oshift;
336: cspidx[tmp] = a->i[row] + i; /* index of a->j */
337: cja[tmp] = row + oshift;
338: }
339: }
340: PetscFree(collengths);
341: *ia = cia;
342: *ja = cja;
343: *spidx = cspidx;
344: return(0);
345: }
347: PetscErrorCode MatRestoreColumnIJ_SeqAIJ_Color(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscInt *spidx[],PetscBool *done)
348: {
352: MatRestoreColumnIJ_SeqAIJ(A,oshift,symmetric,inodecompressed,n,ia,ja,done);
353: PetscFree(*spidx);
354: return(0);
355: }
357: PetscErrorCode MatSetValuesRow_SeqAIJ(Mat A,PetscInt row,const PetscScalar v[])
358: {
359: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
360: PetscInt *ai = a->i;
364: PetscArraycpy(a->a+ai[row],v,ai[row+1]-ai[row]);
365: return(0);
366: }
368: /*
369: MatSeqAIJSetValuesLocalFast - An optimized version of MatSetValuesLocal() for SeqAIJ matrices with several assumptions
371: - a single row of values is set with each call
372: - no row or column indices are negative or (in error) larger than the number of rows or columns
373: - the values are always added to the matrix, not set
374: - no new locations are introduced in the nonzero structure of the matrix
376: This does NOT assume the global column indices are sorted
378: */
380: #include <petsc/private/isimpl.h>
381: PetscErrorCode MatSeqAIJSetValuesLocalFast(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
382: {
383: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
384: PetscInt low,high,t,row,nrow,i,col,l;
385: const PetscInt *rp,*ai = a->i,*ailen = a->ilen,*aj = a->j;
386: PetscInt lastcol = -1;
387: MatScalar *ap,value,*aa = a->a;
388: const PetscInt *ridx = A->rmap->mapping->indices,*cidx = A->cmap->mapping->indices;
390: row = ridx[im[0]];
391: rp = aj + ai[row];
392: ap = aa + ai[row];
393: nrow = ailen[row];
394: low = 0;
395: high = nrow;
396: for (l=0; l<n; l++) { /* loop over added columns */
397: col = cidx[in[l]];
398: value = v[l];
400: if (col <= lastcol) low = 0;
401: else high = nrow;
402: lastcol = col;
403: while (high-low > 5) {
404: t = (low+high)/2;
405: if (rp[t] > col) high = t;
406: else low = t;
407: }
408: for (i=low; i<high; i++) {
409: if (rp[i] == col) {
410: ap[i] += value;
411: low = i + 1;
412: break;
413: }
414: }
415: }
416: return 0;
417: }
419: PetscErrorCode MatSetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
420: {
421: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
422: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N;
423: PetscInt *imax = a->imax,*ai = a->i,*ailen = a->ilen;
425: PetscInt *aj = a->j,nonew = a->nonew,lastcol = -1;
426: MatScalar *ap=NULL,value=0.0,*aa = a->a;
427: PetscBool ignorezeroentries = a->ignorezeroentries;
428: PetscBool roworiented = a->roworiented;
431: for (k=0; k<m; k++) { /* loop over added rows */
432: row = im[k];
433: if (row < 0) continue;
434: #if defined(PETSC_USE_DEBUG)
435: if (row >= A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap->n-1);
436: #endif
437: rp = aj + ai[row];
438: if (!A->structure_only) ap = aa + ai[row];
439: rmax = imax[row]; nrow = ailen[row];
440: low = 0;
441: high = nrow;
442: for (l=0; l<n; l++) { /* loop over added columns */
443: if (in[l] < 0) continue;
444: #if defined(PETSC_USE_DEBUG)
445: if (in[l] >= A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap->n-1);
446: #endif
447: col = in[l];
448: if (v && !A->structure_only) value = roworiented ? v[l + k*n] : v[k + l*m];
449: if (!A->structure_only && value == 0.0 && ignorezeroentries && is == ADD_VALUES && row != col) continue;
451: if (col <= lastcol) low = 0;
452: else high = nrow;
453: lastcol = col;
454: while (high-low > 5) {
455: t = (low+high)/2;
456: if (rp[t] > col) high = t;
457: else low = t;
458: }
459: for (i=low; i<high; i++) {
460: if (rp[i] > col) break;
461: if (rp[i] == col) {
462: if (!A->structure_only) {
463: if (is == ADD_VALUES) {
464: ap[i] += value;
465: (void)PetscLogFlops(1.0);
466: }
467: else ap[i] = value;
468: }
469: low = i + 1;
470: goto noinsert;
471: }
472: }
473: if (value == 0.0 && ignorezeroentries && row != col) goto noinsert;
474: if (nonew == 1) goto noinsert;
475: if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero at (%D,%D) in the matrix",row,col);
476: if (A->structure_only) {
477: MatSeqXAIJReallocateAIJ_structure_only(A,A->rmap->n,1,nrow,row,col,rmax,ai,aj,rp,imax,nonew,MatScalar);
478: } else {
479: MatSeqXAIJReallocateAIJ(A,A->rmap->n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
480: }
481: N = nrow++ - 1; a->nz++; high++;
482: /* shift up all the later entries in this row */
483: PetscArraymove(rp+i+1,rp+i,N-i+1);
484: rp[i] = col;
485: if (!A->structure_only){
486: PetscArraymove(ap+i+1,ap+i,N-i+1);
487: ap[i] = value;
488: }
489: low = i + 1;
490: A->nonzerostate++;
491: noinsert:;
492: }
493: ailen[row] = nrow;
494: }
495: return(0);
496: }
498: PetscErrorCode MatSetValues_SeqAIJ_SortedFull(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
499: {
500: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
501: PetscInt *rp,k,row;
502: PetscInt *ai = a->i,*ailen = a->ilen;
504: PetscInt *aj = a->j;
505: MatScalar *aa = a->a,*ap;
508: for (k=0; k<m; k++) { /* loop over added rows */
509: row = im[k];
510: rp = aj + ai[row];
511: ap = aa + ai[row];
512: if (!A->was_assembled) {
513: PetscMemcpy(rp,in,n*sizeof(PetscInt));
514: }
515: if (!A->structure_only) {
516: if (v) {
517: PetscMemcpy(ap,v,n*sizeof(PetscScalar));
518: v += n;
519: } else {
520: PetscMemzero(ap,n*sizeof(PetscScalar));
521: }
522: }
523: ailen[row] = n;
524: a->nz += n;
525: }
526: return(0);
527: }
530: PetscErrorCode MatGetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],PetscScalar v[])
531: {
532: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
533: PetscInt *rp,k,low,high,t,row,nrow,i,col,l,*aj = a->j;
534: PetscInt *ai = a->i,*ailen = a->ilen;
535: MatScalar *ap,*aa = a->a;
538: for (k=0; k<m; k++) { /* loop over rows */
539: row = im[k];
540: if (row < 0) {v += n; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",row); */
541: if (row >= A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap->n-1);
542: rp = aj + ai[row]; ap = aa + ai[row];
543: nrow = ailen[row];
544: for (l=0; l<n; l++) { /* loop over columns */
545: if (in[l] < 0) {v++; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",in[l]); */
546: if (in[l] >= A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap->n-1);
547: col = in[l];
548: high = nrow; low = 0; /* assume unsorted */
549: while (high-low > 5) {
550: t = (low+high)/2;
551: if (rp[t] > col) high = t;
552: else low = t;
553: }
554: for (i=low; i<high; i++) {
555: if (rp[i] > col) break;
556: if (rp[i] == col) {
557: *v++ = ap[i];
558: goto finished;
559: }
560: }
561: *v++ = 0.0;
562: finished:;
563: }
564: }
565: return(0);
566: }
569: PetscErrorCode MatView_SeqAIJ_Binary(Mat A,PetscViewer viewer)
570: {
571: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
573: PetscInt i,*col_lens;
574: int fd;
575: FILE *file;
578: PetscViewerBinaryGetDescriptor(viewer,&fd);
579: PetscMalloc1(4+A->rmap->n,&col_lens);
581: col_lens[0] = MAT_FILE_CLASSID;
582: col_lens[1] = A->rmap->n;
583: col_lens[2] = A->cmap->n;
584: col_lens[3] = a->nz;
586: /* store lengths of each row and write (including header) to file */
587: for (i=0; i<A->rmap->n; i++) {
588: col_lens[4+i] = a->i[i+1] - a->i[i];
589: }
590: PetscBinaryWrite(fd,col_lens,4+A->rmap->n,PETSC_INT,PETSC_TRUE);
591: PetscFree(col_lens);
593: /* store column indices (zero start index) */
594: PetscBinaryWrite(fd,a->j,a->nz,PETSC_INT,PETSC_FALSE);
596: /* store nonzero values */
597: PetscBinaryWrite(fd,a->a,a->nz,PETSC_SCALAR,PETSC_FALSE);
599: PetscViewerBinaryGetInfoPointer(viewer,&file);
600: if (file) {
601: fprintf(file,"-matload_block_size %d\n",(int)PetscAbs(A->rmap->bs));
602: }
603: return(0);
604: }
606: static PetscErrorCode MatView_SeqAIJ_ASCII_structonly(Mat A,PetscViewer viewer)
607: {
609: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
610: PetscInt i,k,m=A->rmap->N;
613: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
614: for (i=0; i<m; i++) {
615: PetscViewerASCIIPrintf(viewer,"row %D:",i);
616: for (k=a->i[i]; k<a->i[i+1]; k++) {
617: PetscViewerASCIIPrintf(viewer," (%D) ",a->j[k]);
618: }
619: PetscViewerASCIIPrintf(viewer,"\n");
620: }
621: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
622: return(0);
623: }
625: extern PetscErrorCode MatSeqAIJFactorInfo_Matlab(Mat,PetscViewer);
627: PetscErrorCode MatView_SeqAIJ_ASCII(Mat A,PetscViewer viewer)
628: {
629: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
630: PetscErrorCode ierr;
631: PetscInt i,j,m = A->rmap->n;
632: const char *name;
633: PetscViewerFormat format;
636: if (A->structure_only) {
637: MatView_SeqAIJ_ASCII_structonly(A,viewer);
638: return(0);
639: }
641: PetscViewerGetFormat(viewer,&format);
642: if (format == PETSC_VIEWER_ASCII_MATLAB) {
643: PetscInt nofinalvalue = 0;
644: if (m && ((a->i[m] == a->i[m-1]) || (a->j[a->nz-1] != A->cmap->n-1))) {
645: /* Need a dummy value to ensure the dimension of the matrix. */
646: nofinalvalue = 1;
647: }
648: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
649: PetscViewerASCIIPrintf(viewer,"%% Size = %D %D \n",m,A->cmap->n);
650: PetscViewerASCIIPrintf(viewer,"%% Nonzeros = %D \n",a->nz);
651: #if defined(PETSC_USE_COMPLEX)
652: PetscViewerASCIIPrintf(viewer,"zzz = zeros(%D,4);\n",a->nz+nofinalvalue);
653: #else
654: PetscViewerASCIIPrintf(viewer,"zzz = zeros(%D,3);\n",a->nz+nofinalvalue);
655: #endif
656: PetscViewerASCIIPrintf(viewer,"zzz = [\n");
658: for (i=0; i<m; i++) {
659: for (j=a->i[i]; j<a->i[i+1]; j++) {
660: #if defined(PETSC_USE_COMPLEX)
661: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e %18.16e\n",i+1,a->j[j]+1,(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
662: #else
663: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",i+1,a->j[j]+1,(double)a->a[j]);
664: #endif
665: }
666: }
667: if (nofinalvalue) {
668: #if defined(PETSC_USE_COMPLEX)
669: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e %18.16e\n",m,A->cmap->n,0.,0.);
670: #else
671: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",m,A->cmap->n,0.0);
672: #endif
673: }
674: PetscObjectGetName((PetscObject)A,&name);
675: PetscViewerASCIIPrintf(viewer,"];\n %s = spconvert(zzz);\n",name);
676: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
677: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO || format == PETSC_VIEWER_ASCII_INFO) {
678: return(0);
679: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
680: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
681: for (i=0; i<m; i++) {
682: PetscViewerASCIIPrintf(viewer,"row %D:",i);
683: for (j=a->i[i]; j<a->i[i+1]; j++) {
684: #if defined(PETSC_USE_COMPLEX)
685: if (PetscImaginaryPart(a->a[j]) > 0.0 && PetscRealPart(a->a[j]) != 0.0) {
686: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
687: } else if (PetscImaginaryPart(a->a[j]) < 0.0 && PetscRealPart(a->a[j]) != 0.0) {
688: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)-PetscImaginaryPart(a->a[j]));
689: } else if (PetscRealPart(a->a[j]) != 0.0) {
690: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)PetscRealPart(a->a[j]));
691: }
692: #else
693: if (a->a[j] != 0.0) {PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)a->a[j]);}
694: #endif
695: }
696: PetscViewerASCIIPrintf(viewer,"\n");
697: }
698: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
699: } else if (format == PETSC_VIEWER_ASCII_SYMMODU) {
700: PetscInt nzd=0,fshift=1,*sptr;
701: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
702: PetscMalloc1(m+1,&sptr);
703: for (i=0; i<m; i++) {
704: sptr[i] = nzd+1;
705: for (j=a->i[i]; j<a->i[i+1]; j++) {
706: if (a->j[j] >= i) {
707: #if defined(PETSC_USE_COMPLEX)
708: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) nzd++;
709: #else
710: if (a->a[j] != 0.0) nzd++;
711: #endif
712: }
713: }
714: }
715: sptr[m] = nzd+1;
716: PetscViewerASCIIPrintf(viewer," %D %D\n\n",m,nzd);
717: for (i=0; i<m+1; i+=6) {
718: if (i+4<m) {
719: PetscViewerASCIIPrintf(viewer," %D %D %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3],sptr[i+4],sptr[i+5]);
720: } else if (i+3<m) {
721: PetscViewerASCIIPrintf(viewer," %D %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3],sptr[i+4]);
722: } else if (i+2<m) {
723: PetscViewerASCIIPrintf(viewer," %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3]);
724: } else if (i+1<m) {
725: PetscViewerASCIIPrintf(viewer," %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2]);
726: } else if (i<m) {
727: PetscViewerASCIIPrintf(viewer," %D %D\n",sptr[i],sptr[i+1]);
728: } else {
729: PetscViewerASCIIPrintf(viewer," %D\n",sptr[i]);
730: }
731: }
732: PetscViewerASCIIPrintf(viewer,"\n");
733: PetscFree(sptr);
734: for (i=0; i<m; i++) {
735: for (j=a->i[i]; j<a->i[i+1]; j++) {
736: if (a->j[j] >= i) {PetscViewerASCIIPrintf(viewer," %D ",a->j[j]+fshift);}
737: }
738: PetscViewerASCIIPrintf(viewer,"\n");
739: }
740: PetscViewerASCIIPrintf(viewer,"\n");
741: for (i=0; i<m; i++) {
742: for (j=a->i[i]; j<a->i[i+1]; j++) {
743: if (a->j[j] >= i) {
744: #if defined(PETSC_USE_COMPLEX)
745: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) {
746: PetscViewerASCIIPrintf(viewer," %18.16e %18.16e ",(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
747: }
748: #else
749: if (a->a[j] != 0.0) {PetscViewerASCIIPrintf(viewer," %18.16e ",(double)a->a[j]);}
750: #endif
751: }
752: }
753: PetscViewerASCIIPrintf(viewer,"\n");
754: }
755: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
756: } else if (format == PETSC_VIEWER_ASCII_DENSE) {
757: PetscInt cnt = 0,jcnt;
758: PetscScalar value;
759: #if defined(PETSC_USE_COMPLEX)
760: PetscBool realonly = PETSC_TRUE;
762: for (i=0; i<a->i[m]; i++) {
763: if (PetscImaginaryPart(a->a[i]) != 0.0) {
764: realonly = PETSC_FALSE;
765: break;
766: }
767: }
768: #endif
770: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
771: for (i=0; i<m; i++) {
772: jcnt = 0;
773: for (j=0; j<A->cmap->n; j++) {
774: if (jcnt < a->i[i+1]-a->i[i] && j == a->j[cnt]) {
775: value = a->a[cnt++];
776: jcnt++;
777: } else {
778: value = 0.0;
779: }
780: #if defined(PETSC_USE_COMPLEX)
781: if (realonly) {
782: PetscViewerASCIIPrintf(viewer," %7.5e ",(double)PetscRealPart(value));
783: } else {
784: PetscViewerASCIIPrintf(viewer," %7.5e+%7.5e i ",(double)PetscRealPart(value),(double)PetscImaginaryPart(value));
785: }
786: #else
787: PetscViewerASCIIPrintf(viewer," %7.5e ",(double)value);
788: #endif
789: }
790: PetscViewerASCIIPrintf(viewer,"\n");
791: }
792: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
793: } else if (format == PETSC_VIEWER_ASCII_MATRIXMARKET) {
794: PetscInt fshift=1;
795: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
796: #if defined(PETSC_USE_COMPLEX)
797: PetscViewerASCIIPrintf(viewer,"%%%%MatrixMarket matrix coordinate complex general\n");
798: #else
799: PetscViewerASCIIPrintf(viewer,"%%%%MatrixMarket matrix coordinate real general\n");
800: #endif
801: PetscViewerASCIIPrintf(viewer,"%D %D %D\n", m, A->cmap->n, a->nz);
802: for (i=0; i<m; i++) {
803: for (j=a->i[i]; j<a->i[i+1]; j++) {
804: #if defined(PETSC_USE_COMPLEX)
805: PetscViewerASCIIPrintf(viewer,"%D %D %g %g\n", i+fshift,a->j[j]+fshift,(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
806: #else
807: PetscViewerASCIIPrintf(viewer,"%D %D %g\n", i+fshift, a->j[j]+fshift, (double)a->a[j]);
808: #endif
809: }
810: }
811: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
812: } else {
813: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
814: if (A->factortype) {
815: for (i=0; i<m; i++) {
816: PetscViewerASCIIPrintf(viewer,"row %D:",i);
817: /* L part */
818: for (j=a->i[i]; j<a->i[i+1]; j++) {
819: #if defined(PETSC_USE_COMPLEX)
820: if (PetscImaginaryPart(a->a[j]) > 0.0) {
821: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
822: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
823: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)(-PetscImaginaryPart(a->a[j])));
824: } else {
825: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)PetscRealPart(a->a[j]));
826: }
827: #else
828: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)a->a[j]);
829: #endif
830: }
831: /* diagonal */
832: j = a->diag[i];
833: #if defined(PETSC_USE_COMPLEX)
834: if (PetscImaginaryPart(a->a[j]) > 0.0) {
835: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j],(double)PetscRealPart(1.0/a->a[j]),(double)PetscImaginaryPart(1.0/a->a[j]));
836: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
837: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j],(double)PetscRealPart(1.0/a->a[j]),(double)(-PetscImaginaryPart(1.0/a->a[j])));
838: } else {
839: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)PetscRealPart(1.0/a->a[j]));
840: }
841: #else
842: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)(1.0/a->a[j]));
843: #endif
845: /* U part */
846: for (j=a->diag[i+1]+1; j<a->diag[i]; j++) {
847: #if defined(PETSC_USE_COMPLEX)
848: if (PetscImaginaryPart(a->a[j]) > 0.0) {
849: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
850: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
851: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)(-PetscImaginaryPart(a->a[j])));
852: } else {
853: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)PetscRealPart(a->a[j]));
854: }
855: #else
856: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)a->a[j]);
857: #endif
858: }
859: PetscViewerASCIIPrintf(viewer,"\n");
860: }
861: } else {
862: for (i=0; i<m; i++) {
863: PetscViewerASCIIPrintf(viewer,"row %D:",i);
864: for (j=a->i[i]; j<a->i[i+1]; j++) {
865: #if defined(PETSC_USE_COMPLEX)
866: if (PetscImaginaryPart(a->a[j]) > 0.0) {
867: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)PetscImaginaryPart(a->a[j]));
868: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
869: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j],(double)PetscRealPart(a->a[j]),(double)-PetscImaginaryPart(a->a[j]));
870: } else {
871: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)PetscRealPart(a->a[j]));
872: }
873: #else
874: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j],(double)a->a[j]);
875: #endif
876: }
877: PetscViewerASCIIPrintf(viewer,"\n");
878: }
879: }
880: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
881: }
882: PetscViewerFlush(viewer);
883: return(0);
884: }
886: #include <petscdraw.h>
887: PetscErrorCode MatView_SeqAIJ_Draw_Zoom(PetscDraw draw,void *Aa)
888: {
889: Mat A = (Mat) Aa;
890: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
891: PetscErrorCode ierr;
892: PetscInt i,j,m = A->rmap->n;
893: int color;
894: PetscReal xl,yl,xr,yr,x_l,x_r,y_l,y_r;
895: PetscViewer viewer;
896: PetscViewerFormat format;
899: PetscObjectQuery((PetscObject)A,"Zoomviewer",(PetscObject*)&viewer);
900: PetscViewerGetFormat(viewer,&format);
901: PetscDrawGetCoordinates(draw,&xl,&yl,&xr,&yr);
903: /* loop over matrix elements drawing boxes */
905: if (format != PETSC_VIEWER_DRAW_CONTOUR) {
906: PetscDrawCollectiveBegin(draw);
907: /* Blue for negative, Cyan for zero and Red for positive */
908: color = PETSC_DRAW_BLUE;
909: for (i=0; i<m; i++) {
910: y_l = m - i - 1.0; y_r = y_l + 1.0;
911: for (j=a->i[i]; j<a->i[i+1]; j++) {
912: x_l = a->j[j]; x_r = x_l + 1.0;
913: if (PetscRealPart(a->a[j]) >= 0.) continue;
914: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
915: }
916: }
917: color = PETSC_DRAW_CYAN;
918: for (i=0; i<m; i++) {
919: y_l = m - i - 1.0; y_r = y_l + 1.0;
920: for (j=a->i[i]; j<a->i[i+1]; j++) {
921: x_l = a->j[j]; x_r = x_l + 1.0;
922: if (a->a[j] != 0.) continue;
923: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
924: }
925: }
926: color = PETSC_DRAW_RED;
927: for (i=0; i<m; i++) {
928: y_l = m - i - 1.0; y_r = y_l + 1.0;
929: for (j=a->i[i]; j<a->i[i+1]; j++) {
930: x_l = a->j[j]; x_r = x_l + 1.0;
931: if (PetscRealPart(a->a[j]) <= 0.) continue;
932: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
933: }
934: }
935: PetscDrawCollectiveEnd(draw);
936: } else {
937: /* use contour shading to indicate magnitude of values */
938: /* first determine max of all nonzero values */
939: PetscReal minv = 0.0, maxv = 0.0;
940: PetscInt nz = a->nz, count = 0;
941: PetscDraw popup;
943: for (i=0; i<nz; i++) {
944: if (PetscAbsScalar(a->a[i]) > maxv) maxv = PetscAbsScalar(a->a[i]);
945: }
946: if (minv >= maxv) maxv = minv + PETSC_SMALL;
947: PetscDrawGetPopup(draw,&popup);
948: PetscDrawScalePopup(popup,minv,maxv);
950: PetscDrawCollectiveBegin(draw);
951: for (i=0; i<m; i++) {
952: y_l = m - i - 1.0;
953: y_r = y_l + 1.0;
954: for (j=a->i[i]; j<a->i[i+1]; j++) {
955: x_l = a->j[j];
956: x_r = x_l + 1.0;
957: color = PetscDrawRealToColor(PetscAbsScalar(a->a[count]),minv,maxv);
958: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
959: count++;
960: }
961: }
962: PetscDrawCollectiveEnd(draw);
963: }
964: return(0);
965: }
967: #include <petscdraw.h>
968: PetscErrorCode MatView_SeqAIJ_Draw(Mat A,PetscViewer viewer)
969: {
971: PetscDraw draw;
972: PetscReal xr,yr,xl,yl,h,w;
973: PetscBool isnull;
976: PetscViewerDrawGetDraw(viewer,0,&draw);
977: PetscDrawIsNull(draw,&isnull);
978: if (isnull) return(0);
980: xr = A->cmap->n; yr = A->rmap->n; h = yr/10.0; w = xr/10.0;
981: xr += w; yr += h; xl = -w; yl = -h;
982: PetscDrawSetCoordinates(draw,xl,yl,xr,yr);
983: PetscObjectCompose((PetscObject)A,"Zoomviewer",(PetscObject)viewer);
984: PetscDrawZoom(draw,MatView_SeqAIJ_Draw_Zoom,A);
985: PetscObjectCompose((PetscObject)A,"Zoomviewer",NULL);
986: PetscDrawSave(draw);
987: return(0);
988: }
990: PetscErrorCode MatView_SeqAIJ(Mat A,PetscViewer viewer)
991: {
993: PetscBool iascii,isbinary,isdraw;
996: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
997: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);
998: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
999: if (iascii) {
1000: MatView_SeqAIJ_ASCII(A,viewer);
1001: } else if (isbinary) {
1002: MatView_SeqAIJ_Binary(A,viewer);
1003: } else if (isdraw) {
1004: MatView_SeqAIJ_Draw(A,viewer);
1005: }
1006: MatView_SeqAIJ_Inode(A,viewer);
1007: return(0);
1008: }
1010: PetscErrorCode MatAssemblyEnd_SeqAIJ(Mat A,MatAssemblyType mode)
1011: {
1012: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1014: PetscInt fshift = 0,i,*ai = a->i,*aj = a->j,*imax = a->imax;
1015: PetscInt m = A->rmap->n,*ip,N,*ailen = a->ilen,rmax = 0;
1016: MatScalar *aa = a->a,*ap;
1017: PetscReal ratio = 0.6;
1020: if (mode == MAT_FLUSH_ASSEMBLY) return(0);
1021: MatSeqAIJInvalidateDiagonal(A);
1022: if (A->was_assembled && A->ass_nonzerostate == A->nonzerostate) return(0);
1024: if (m) rmax = ailen[0]; /* determine row with most nonzeros */
1025: for (i=1; i<m; i++) {
1026: /* move each row back by the amount of empty slots (fshift) before it*/
1027: fshift += imax[i-1] - ailen[i-1];
1028: rmax = PetscMax(rmax,ailen[i]);
1029: if (fshift) {
1030: ip = aj + ai[i];
1031: ap = aa + ai[i];
1032: N = ailen[i];
1033: PetscArraymove(ip-fshift,ip,N);
1034: if (!A->structure_only) {
1035: PetscArraymove(ap-fshift,ap,N);
1036: }
1037: }
1038: ai[i] = ai[i-1] + ailen[i-1];
1039: }
1040: if (m) {
1041: fshift += imax[m-1] - ailen[m-1];
1042: ai[m] = ai[m-1] + ailen[m-1];
1043: }
1045: /* reset ilen and imax for each row */
1046: a->nonzerorowcnt = 0;
1047: if (A->structure_only) {
1048: PetscFree(a->imax);
1049: PetscFree(a->ilen);
1050: } else { /* !A->structure_only */
1051: for (i=0; i<m; i++) {
1052: ailen[i] = imax[i] = ai[i+1] - ai[i];
1053: a->nonzerorowcnt += ((ai[i+1] - ai[i]) > 0);
1054: }
1055: }
1056: a->nz = ai[m];
1057: if (fshift && a->nounused == -1) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_PLIB, "Unused space detected in matrix: %D X %D, %D unneeded", m, A->cmap->n, fshift);
1059: MatMarkDiagonal_SeqAIJ(A);
1060: PetscInfo4(A,"Matrix size: %D X %D; storage space: %D unneeded,%D used\n",m,A->cmap->n,fshift,a->nz);
1061: PetscInfo1(A,"Number of mallocs during MatSetValues() is %D\n",a->reallocs);
1062: PetscInfo1(A,"Maximum nonzeros in any row is %D\n",rmax);
1064: A->info.mallocs += a->reallocs;
1065: a->reallocs = 0;
1066: A->info.nz_unneeded = (PetscReal)fshift;
1067: a->rmax = rmax;
1069: if (!A->structure_only) {
1070: MatCheckCompressedRow(A,a->nonzerorowcnt,&a->compressedrow,a->i,m,ratio);
1071: }
1072: MatAssemblyEnd_SeqAIJ_Inode(A,mode);
1073: return(0);
1074: }
1076: PetscErrorCode MatRealPart_SeqAIJ(Mat A)
1077: {
1078: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1079: PetscInt i,nz = a->nz;
1080: MatScalar *aa = a->a;
1084: for (i=0; i<nz; i++) aa[i] = PetscRealPart(aa[i]);
1085: MatSeqAIJInvalidateDiagonal(A);
1086: return(0);
1087: }
1089: PetscErrorCode MatImaginaryPart_SeqAIJ(Mat A)
1090: {
1091: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1092: PetscInt i,nz = a->nz;
1093: MatScalar *aa = a->a;
1097: for (i=0; i<nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1098: MatSeqAIJInvalidateDiagonal(A);
1099: return(0);
1100: }
1102: PetscErrorCode MatZeroEntries_SeqAIJ(Mat A)
1103: {
1104: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1108: PetscArrayzero(a->a,a->i[A->rmap->n]);
1109: MatSeqAIJInvalidateDiagonal(A);
1110: return(0);
1111: }
1113: PetscErrorCode MatDestroy_SeqAIJ(Mat A)
1114: {
1115: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1119: #if defined(PETSC_USE_LOG)
1120: PetscLogObjectState((PetscObject)A,"Rows=%D, Cols=%D, NZ=%D",A->rmap->n,A->cmap->n,a->nz);
1121: #endif
1122: MatSeqXAIJFreeAIJ(A,&a->a,&a->j,&a->i);
1123: ISDestroy(&a->row);
1124: ISDestroy(&a->col);
1125: PetscFree(a->diag);
1126: PetscFree(a->ibdiag);
1127: PetscFree(a->imax);
1128: PetscFree(a->ilen);
1129: PetscFree(a->ipre);
1130: PetscFree3(a->idiag,a->mdiag,a->ssor_work);
1131: PetscFree(a->solve_work);
1132: ISDestroy(&a->icol);
1133: PetscFree(a->saved_values);
1134: ISColoringDestroy(&a->coloring);
1135: PetscFree2(a->compressedrow.i,a->compressedrow.rindex);
1136: PetscFree(a->matmult_abdense);
1138: MatDestroy_SeqAIJ_Inode(A);
1139: PetscFree(A->data);
1141: PetscObjectChangeTypeName((PetscObject)A,0);
1142: PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetColumnIndices_C",NULL);
1143: PetscObjectComposeFunction((PetscObject)A,"MatStoreValues_C",NULL);
1144: PetscObjectComposeFunction((PetscObject)A,"MatRetrieveValues_C",NULL);
1145: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqsbaij_C",NULL);
1146: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqbaij_C",NULL);
1147: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqaijperm_C",NULL);
1148: #if defined(PETSC_HAVE_ELEMENTAL)
1149: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_elemental_C",NULL);
1150: #endif
1151: #if defined(PETSC_HAVE_HYPRE)
1152: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_hypre_C",NULL);
1153: PetscObjectComposeFunction((PetscObject)A,"MatMatMatMult_transpose_seqaij_seqaij_C",NULL);
1154: #endif
1155: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqdense_C",NULL);
1156: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqsell_C",NULL);
1157: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_is_C",NULL);
1158: PetscObjectComposeFunction((PetscObject)A,"MatIsTranspose_C",NULL);
1159: PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetPreallocation_C",NULL);
1160: PetscObjectComposeFunction((PetscObject)A,"MatResetPreallocation_C",NULL);
1161: PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetPreallocationCSR_C",NULL);
1162: PetscObjectComposeFunction((PetscObject)A,"MatReorderForNonzeroDiagonal_C",NULL);
1163: PetscObjectComposeFunction((PetscObject)A,"MatPtAP_is_seqaij_C",NULL);
1164: return(0);
1165: }
1167: PetscErrorCode MatSetOption_SeqAIJ(Mat A,MatOption op,PetscBool flg)
1168: {
1169: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1173: switch (op) {
1174: case MAT_ROW_ORIENTED:
1175: a->roworiented = flg;
1176: break;
1177: case MAT_KEEP_NONZERO_PATTERN:
1178: a->keepnonzeropattern = flg;
1179: break;
1180: case MAT_NEW_NONZERO_LOCATIONS:
1181: a->nonew = (flg ? 0 : 1);
1182: break;
1183: case MAT_NEW_NONZERO_LOCATION_ERR:
1184: a->nonew = (flg ? -1 : 0);
1185: break;
1186: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1187: a->nonew = (flg ? -2 : 0);
1188: break;
1189: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1190: a->nounused = (flg ? -1 : 0);
1191: break;
1192: case MAT_IGNORE_ZERO_ENTRIES:
1193: a->ignorezeroentries = flg;
1194: break;
1195: case MAT_SPD:
1196: case MAT_SYMMETRIC:
1197: case MAT_STRUCTURALLY_SYMMETRIC:
1198: case MAT_HERMITIAN:
1199: case MAT_SYMMETRY_ETERNAL:
1200: case MAT_STRUCTURE_ONLY:
1201: /* These options are handled directly by MatSetOption() */
1202: break;
1203: case MAT_NEW_DIAGONALS:
1204: case MAT_IGNORE_OFF_PROC_ENTRIES:
1205: case MAT_USE_HASH_TABLE:
1206: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
1207: break;
1208: case MAT_USE_INODES:
1209: /* Not an error because MatSetOption_SeqAIJ_Inode handles this one */
1210: break;
1211: case MAT_SUBMAT_SINGLEIS:
1212: A->submat_singleis = flg;
1213: break;
1214: case MAT_SORTED_FULL:
1215: if (flg) A->ops->setvalues = MatSetValues_SeqAIJ_SortedFull;
1216: else A->ops->setvalues = MatSetValues_SeqAIJ;
1217: break;
1218: default:
1219: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"unknown option %d",op);
1220: }
1221: MatSetOption_SeqAIJ_Inode(A,op,flg);
1222: return(0);
1223: }
1225: PetscErrorCode MatGetDiagonal_SeqAIJ(Mat A,Vec v)
1226: {
1227: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1229: PetscInt i,j,n,*ai=a->i,*aj=a->j;
1230: PetscScalar *aa=a->a,*x;
1233: VecGetLocalSize(v,&n);
1234: if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
1236: if (A->factortype == MAT_FACTOR_ILU || A->factortype == MAT_FACTOR_LU) {
1237: PetscInt *diag=a->diag;
1238: VecGetArrayWrite(v,&x);
1239: for (i=0; i<n; i++) x[i] = 1.0/aa[diag[i]];
1240: VecRestoreArrayWrite(v,&x);
1241: return(0);
1242: }
1244: VecGetArrayWrite(v,&x);
1245: for (i=0; i<n; i++) {
1246: x[i] = 0.0;
1247: for (j=ai[i]; j<ai[i+1]; j++) {
1248: if (aj[j] == i) {
1249: x[i] = aa[j];
1250: break;
1251: }
1252: }
1253: }
1254: VecRestoreArrayWrite(v,&x);
1255: return(0);
1256: }
1258: #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h>
1259: PetscErrorCode MatMultTransposeAdd_SeqAIJ(Mat A,Vec xx,Vec zz,Vec yy)
1260: {
1261: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1262: PetscScalar *y;
1263: const PetscScalar *x;
1264: PetscErrorCode ierr;
1265: PetscInt m = A->rmap->n;
1266: #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
1267: const MatScalar *v;
1268: PetscScalar alpha;
1269: PetscInt n,i,j;
1270: const PetscInt *idx,*ii,*ridx=NULL;
1271: Mat_CompressedRow cprow = a->compressedrow;
1272: PetscBool usecprow = cprow.use;
1273: #endif
1276: if (zz != yy) {VecCopy(zz,yy);}
1277: VecGetArrayRead(xx,&x);
1278: VecGetArray(yy,&y);
1280: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
1281: fortranmulttransposeaddaij_(&m,x,a->i,a->j,a->a,y);
1282: #else
1283: if (usecprow) {
1284: m = cprow.nrows;
1285: ii = cprow.i;
1286: ridx = cprow.rindex;
1287: } else {
1288: ii = a->i;
1289: }
1290: for (i=0; i<m; i++) {
1291: idx = a->j + ii[i];
1292: v = a->a + ii[i];
1293: n = ii[i+1] - ii[i];
1294: if (usecprow) {
1295: alpha = x[ridx[i]];
1296: } else {
1297: alpha = x[i];
1298: }
1299: for (j=0; j<n; j++) y[idx[j]] += alpha*v[j];
1300: }
1301: #endif
1302: PetscLogFlops(2.0*a->nz);
1303: VecRestoreArrayRead(xx,&x);
1304: VecRestoreArray(yy,&y);
1305: return(0);
1306: }
1308: PetscErrorCode MatMultTranspose_SeqAIJ(Mat A,Vec xx,Vec yy)
1309: {
1313: VecSet(yy,0.0);
1314: MatMultTransposeAdd_SeqAIJ(A,xx,yy,yy);
1315: return(0);
1316: }
1318: #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h>
1320: PetscErrorCode MatMult_SeqAIJ(Mat A,Vec xx,Vec yy)
1321: {
1322: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1323: PetscScalar *y;
1324: const PetscScalar *x;
1325: const MatScalar *aa;
1326: PetscErrorCode ierr;
1327: PetscInt m=A->rmap->n;
1328: const PetscInt *aj,*ii,*ridx=NULL;
1329: PetscInt n,i;
1330: PetscScalar sum;
1331: PetscBool usecprow=a->compressedrow.use;
1333: #if defined(PETSC_HAVE_PRAGMA_DISJOINT)
1334: #pragma disjoint(*x,*y,*aa)
1335: #endif
1338: VecGetArrayRead(xx,&x);
1339: VecGetArray(yy,&y);
1340: ii = a->i;
1341: if (usecprow) { /* use compressed row format */
1342: PetscArrayzero(y,m);
1343: m = a->compressedrow.nrows;
1344: ii = a->compressedrow.i;
1345: ridx = a->compressedrow.rindex;
1346: for (i=0; i<m; i++) {
1347: n = ii[i+1] - ii[i];
1348: aj = a->j + ii[i];
1349: aa = a->a + ii[i];
1350: sum = 0.0;
1351: PetscSparseDensePlusDot(sum,x,aa,aj,n);
1352: /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */
1353: y[*ridx++] = sum;
1354: }
1355: } else { /* do not use compressed row format */
1356: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ)
1357: aj = a->j;
1358: aa = a->a;
1359: fortranmultaij_(&m,x,ii,aj,aa,y);
1360: #else
1361: for (i=0; i<m; i++) {
1362: n = ii[i+1] - ii[i];
1363: aj = a->j + ii[i];
1364: aa = a->a + ii[i];
1365: sum = 0.0;
1366: PetscSparseDensePlusDot(sum,x,aa,aj,n);
1367: y[i] = sum;
1368: }
1369: #endif
1370: }
1371: PetscLogFlops(2.0*a->nz - a->nonzerorowcnt);
1372: VecRestoreArrayRead(xx,&x);
1373: VecRestoreArray(yy,&y);
1374: return(0);
1375: }
1377: PetscErrorCode MatMultMax_SeqAIJ(Mat A,Vec xx,Vec yy)
1378: {
1379: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1380: PetscScalar *y;
1381: const PetscScalar *x;
1382: const MatScalar *aa;
1383: PetscErrorCode ierr;
1384: PetscInt m=A->rmap->n;
1385: const PetscInt *aj,*ii,*ridx=NULL;
1386: PetscInt n,i,nonzerorow=0;
1387: PetscScalar sum;
1388: PetscBool usecprow=a->compressedrow.use;
1390: #if defined(PETSC_HAVE_PRAGMA_DISJOINT)
1391: #pragma disjoint(*x,*y,*aa)
1392: #endif
1395: VecGetArrayRead(xx,&x);
1396: VecGetArray(yy,&y);
1397: if (usecprow) { /* use compressed row format */
1398: m = a->compressedrow.nrows;
1399: ii = a->compressedrow.i;
1400: ridx = a->compressedrow.rindex;
1401: for (i=0; i<m; i++) {
1402: n = ii[i+1] - ii[i];
1403: aj = a->j + ii[i];
1404: aa = a->a + ii[i];
1405: sum = 0.0;
1406: nonzerorow += (n>0);
1407: PetscSparseDenseMaxDot(sum,x,aa,aj,n);
1408: /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */
1409: y[*ridx++] = sum;
1410: }
1411: } else { /* do not use compressed row format */
1412: ii = a->i;
1413: for (i=0; i<m; i++) {
1414: n = ii[i+1] - ii[i];
1415: aj = a->j + ii[i];
1416: aa = a->a + ii[i];
1417: sum = 0.0;
1418: nonzerorow += (n>0);
1419: PetscSparseDenseMaxDot(sum,x,aa,aj,n);
1420: y[i] = sum;
1421: }
1422: }
1423: PetscLogFlops(2.0*a->nz - nonzerorow);
1424: VecRestoreArrayRead(xx,&x);
1425: VecRestoreArray(yy,&y);
1426: return(0);
1427: }
1429: PetscErrorCode MatMultAddMax_SeqAIJ(Mat A,Vec xx,Vec yy,Vec zz)
1430: {
1431: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1432: PetscScalar *y,*z;
1433: const PetscScalar *x;
1434: const MatScalar *aa;
1435: PetscErrorCode ierr;
1436: PetscInt m = A->rmap->n,*aj,*ii;
1437: PetscInt n,i,*ridx=NULL;
1438: PetscScalar sum;
1439: PetscBool usecprow=a->compressedrow.use;
1442: VecGetArrayRead(xx,&x);
1443: VecGetArrayPair(yy,zz,&y,&z);
1444: if (usecprow) { /* use compressed row format */
1445: if (zz != yy) {
1446: PetscArraycpy(z,y,m);
1447: }
1448: m = a->compressedrow.nrows;
1449: ii = a->compressedrow.i;
1450: ridx = a->compressedrow.rindex;
1451: for (i=0; i<m; i++) {
1452: n = ii[i+1] - ii[i];
1453: aj = a->j + ii[i];
1454: aa = a->a + ii[i];
1455: sum = y[*ridx];
1456: PetscSparseDenseMaxDot(sum,x,aa,aj,n);
1457: z[*ridx++] = sum;
1458: }
1459: } else { /* do not use compressed row format */
1460: ii = a->i;
1461: for (i=0; i<m; i++) {
1462: n = ii[i+1] - ii[i];
1463: aj = a->j + ii[i];
1464: aa = a->a + ii[i];
1465: sum = y[i];
1466: PetscSparseDenseMaxDot(sum,x,aa,aj,n);
1467: z[i] = sum;
1468: }
1469: }
1470: PetscLogFlops(2.0*a->nz);
1471: VecRestoreArrayRead(xx,&x);
1472: VecRestoreArrayPair(yy,zz,&y,&z);
1473: return(0);
1474: }
1476: #include <../src/mat/impls/aij/seq/ftn-kernels/fmultadd.h>
1477: PetscErrorCode MatMultAdd_SeqAIJ(Mat A,Vec xx,Vec yy,Vec zz)
1478: {
1479: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1480: PetscScalar *y,*z;
1481: const PetscScalar *x;
1482: const MatScalar *aa;
1483: PetscErrorCode ierr;
1484: const PetscInt *aj,*ii,*ridx=NULL;
1485: PetscInt m = A->rmap->n,n,i;
1486: PetscScalar sum;
1487: PetscBool usecprow=a->compressedrow.use;
1490: VecGetArrayRead(xx,&x);
1491: VecGetArrayPair(yy,zz,&y,&z);
1492: if (usecprow) { /* use compressed row format */
1493: if (zz != yy) {
1494: PetscArraycpy(z,y,m);
1495: }
1496: m = a->compressedrow.nrows;
1497: ii = a->compressedrow.i;
1498: ridx = a->compressedrow.rindex;
1499: for (i=0; i<m; i++) {
1500: n = ii[i+1] - ii[i];
1501: aj = a->j + ii[i];
1502: aa = a->a + ii[i];
1503: sum = y[*ridx];
1504: PetscSparseDensePlusDot(sum,x,aa,aj,n);
1505: z[*ridx++] = sum;
1506: }
1507: } else { /* do not use compressed row format */
1508: ii = a->i;
1509: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTADDAIJ)
1510: aj = a->j;
1511: aa = a->a;
1512: fortranmultaddaij_(&m,x,ii,aj,aa,y,z);
1513: #else
1514: for (i=0; i<m; i++) {
1515: n = ii[i+1] - ii[i];
1516: aj = a->j + ii[i];
1517: aa = a->a + ii[i];
1518: sum = y[i];
1519: PetscSparseDensePlusDot(sum,x,aa,aj,n);
1520: z[i] = sum;
1521: }
1522: #endif
1523: }
1524: PetscLogFlops(2.0*a->nz);
1525: VecRestoreArrayRead(xx,&x);
1526: VecRestoreArrayPair(yy,zz,&y,&z);
1527: return(0);
1528: }
1530: /*
1531: Adds diagonal pointers to sparse matrix structure.
1532: */
1533: PetscErrorCode MatMarkDiagonal_SeqAIJ(Mat A)
1534: {
1535: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1537: PetscInt i,j,m = A->rmap->n;
1540: if (!a->diag) {
1541: PetscMalloc1(m,&a->diag);
1542: PetscLogObjectMemory((PetscObject)A, m*sizeof(PetscInt));
1543: }
1544: for (i=0; i<A->rmap->n; i++) {
1545: a->diag[i] = a->i[i+1];
1546: for (j=a->i[i]; j<a->i[i+1]; j++) {
1547: if (a->j[j] == i) {
1548: a->diag[i] = j;
1549: break;
1550: }
1551: }
1552: }
1553: return(0);
1554: }
1556: PetscErrorCode MatShift_SeqAIJ(Mat A,PetscScalar v)
1557: {
1558: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1559: const PetscInt *diag = (const PetscInt*)a->diag;
1560: const PetscInt *ii = (const PetscInt*) a->i;
1561: PetscInt i,*mdiag = NULL;
1562: PetscErrorCode ierr;
1563: PetscInt cnt = 0; /* how many diagonals are missing */
1566: if (!A->preallocated || !a->nz) {
1567: MatSeqAIJSetPreallocation(A,1,NULL);
1568: MatShift_Basic(A,v);
1569: return(0);
1570: }
1572: if (a->diagonaldense) {
1573: cnt = 0;
1574: } else {
1575: PetscCalloc1(A->rmap->n,&mdiag);
1576: for (i=0; i<A->rmap->n; i++) {
1577: if (diag[i] >= ii[i+1]) {
1578: cnt++;
1579: mdiag[i] = 1;
1580: }
1581: }
1582: }
1583: if (!cnt) {
1584: MatShift_Basic(A,v);
1585: } else {
1586: PetscScalar *olda = a->a; /* preserve pointers to current matrix nonzeros structure and values */
1587: PetscInt *oldj = a->j, *oldi = a->i;
1588: PetscBool singlemalloc = a->singlemalloc,free_a = a->free_a,free_ij = a->free_ij;
1590: a->a = NULL;
1591: a->j = NULL;
1592: a->i = NULL;
1593: /* increase the values in imax for each row where a diagonal is being inserted then reallocate the matrix data structures */
1594: for (i=0; i<A->rmap->n; i++) {
1595: a->imax[i] += mdiag[i];
1596: a->imax[i] = PetscMin(a->imax[i],A->cmap->n);
1597: }
1598: MatSeqAIJSetPreallocation_SeqAIJ(A,0,a->imax);
1600: /* copy old values into new matrix data structure */
1601: for (i=0; i<A->rmap->n; i++) {
1602: MatSetValues(A,1,&i,a->imax[i] - mdiag[i],&oldj[oldi[i]],&olda[oldi[i]],ADD_VALUES);
1603: if (i < A->cmap->n) {
1604: MatSetValue(A,i,i,v,ADD_VALUES);
1605: }
1606: }
1607: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
1608: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
1609: if (singlemalloc) {
1610: PetscFree3(olda,oldj,oldi);
1611: } else {
1612: if (free_a) {PetscFree(olda);}
1613: if (free_ij) {PetscFree(oldj);}
1614: if (free_ij) {PetscFree(oldi);}
1615: }
1616: }
1617: PetscFree(mdiag);
1618: a->diagonaldense = PETSC_TRUE;
1619: return(0);
1620: }
1622: /*
1623: Checks for missing diagonals
1624: */
1625: PetscErrorCode MatMissingDiagonal_SeqAIJ(Mat A,PetscBool *missing,PetscInt *d)
1626: {
1627: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1628: PetscInt *diag,*ii = a->i,i;
1632: *missing = PETSC_FALSE;
1633: if (A->rmap->n > 0 && !ii) {
1634: *missing = PETSC_TRUE;
1635: if (d) *d = 0;
1636: PetscInfo(A,"Matrix has no entries therefore is missing diagonal\n");
1637: } else {
1638: PetscInt n;
1639: n = PetscMin(A->rmap->n, A->cmap->n);
1640: diag = a->diag;
1641: for (i=0; i<n; i++) {
1642: if (diag[i] >= ii[i+1]) {
1643: *missing = PETSC_TRUE;
1644: if (d) *d = i;
1645: PetscInfo1(A,"Matrix is missing diagonal number %D\n",i);
1646: break;
1647: }
1648: }
1649: }
1650: return(0);
1651: }
1653: #include <petscblaslapack.h>
1654: #include <petsc/private/kernels/blockinvert.h>
1656: /*
1657: Note that values is allocated externally by the PC and then passed into this routine
1658: */
1659: PetscErrorCode MatInvertVariableBlockDiagonal_SeqAIJ(Mat A,PetscInt nblocks,const PetscInt *bsizes,PetscScalar *diag)
1660: {
1661: PetscErrorCode ierr;
1662: PetscInt n = A->rmap->n, i, ncnt = 0, *indx,j,bsizemax = 0,*v_pivots;
1663: PetscBool allowzeropivot,zeropivotdetected=PETSC_FALSE;
1664: const PetscReal shift = 0.0;
1665: PetscInt ipvt[5];
1666: PetscScalar work[25],*v_work;
1669: allowzeropivot = PetscNot(A->erroriffailure);
1670: for (i=0; i<nblocks; i++) ncnt += bsizes[i];
1671: if (ncnt != n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Total blocksizes %D doesn't match number matrix rows %D",ncnt,n);
1672: for (i=0; i<nblocks; i++) {
1673: bsizemax = PetscMax(bsizemax,bsizes[i]);
1674: }
1675: PetscMalloc1(bsizemax,&indx);
1676: if (bsizemax > 7) {
1677: PetscMalloc2(bsizemax,&v_work,bsizemax,&v_pivots);
1678: }
1679: ncnt = 0;
1680: for (i=0; i<nblocks; i++) {
1681: for (j=0; j<bsizes[i]; j++) indx[j] = ncnt+j;
1682: MatGetValues(A,bsizes[i],indx,bsizes[i],indx,diag);
1683: switch (bsizes[i]) {
1684: case 1:
1685: *diag = 1.0/(*diag);
1686: break;
1687: case 2:
1688: PetscKernel_A_gets_inverse_A_2(diag,shift,allowzeropivot,&zeropivotdetected);
1689: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1690: PetscKernel_A_gets_transpose_A_2(diag);
1691: break;
1692: case 3:
1693: PetscKernel_A_gets_inverse_A_3(diag,shift,allowzeropivot,&zeropivotdetected);
1694: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1695: PetscKernel_A_gets_transpose_A_3(diag);
1696: break;
1697: case 4:
1698: PetscKernel_A_gets_inverse_A_4(diag,shift,allowzeropivot,&zeropivotdetected);
1699: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1700: PetscKernel_A_gets_transpose_A_4(diag);
1701: break;
1702: case 5:
1703: PetscKernel_A_gets_inverse_A_5(diag,ipvt,work,shift,allowzeropivot,&zeropivotdetected);
1704: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1705: PetscKernel_A_gets_transpose_A_5(diag);
1706: break;
1707: case 6:
1708: PetscKernel_A_gets_inverse_A_6(diag,shift,allowzeropivot,&zeropivotdetected);
1709: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1710: PetscKernel_A_gets_transpose_A_6(diag);
1711: break;
1712: case 7:
1713: PetscKernel_A_gets_inverse_A_7(diag,shift,allowzeropivot,&zeropivotdetected);
1714: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1715: PetscKernel_A_gets_transpose_A_7(diag);
1716: break;
1717: default:
1718: PetscKernel_A_gets_inverse_A(bsizes[i],diag,v_pivots,v_work,allowzeropivot,&zeropivotdetected);
1719: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1720: PetscKernel_A_gets_transpose_A_N(diag,bsizes[i]);
1721: }
1722: ncnt += bsizes[i];
1723: diag += bsizes[i]*bsizes[i];
1724: }
1725: if (bsizemax > 7) {
1726: PetscFree2(v_work,v_pivots);
1727: }
1728: PetscFree(indx);
1729: return(0);
1730: }
1732: /*
1733: Negative shift indicates do not generate an error if there is a zero diagonal, just invert it anyways
1734: */
1735: PetscErrorCode MatInvertDiagonal_SeqAIJ(Mat A,PetscScalar omega,PetscScalar fshift)
1736: {
1737: Mat_SeqAIJ *a = (Mat_SeqAIJ*) A->data;
1739: PetscInt i,*diag,m = A->rmap->n;
1740: MatScalar *v = a->a;
1741: PetscScalar *idiag,*mdiag;
1744: if (a->idiagvalid) return(0);
1745: MatMarkDiagonal_SeqAIJ(A);
1746: diag = a->diag;
1747: if (!a->idiag) {
1748: PetscMalloc3(m,&a->idiag,m,&a->mdiag,m,&a->ssor_work);
1749: PetscLogObjectMemory((PetscObject)A, 3*m*sizeof(PetscScalar));
1750: v = a->a;
1751: }
1752: mdiag = a->mdiag;
1753: idiag = a->idiag;
1755: if (omega == 1.0 && PetscRealPart(fshift) <= 0.0) {
1756: for (i=0; i<m; i++) {
1757: mdiag[i] = v[diag[i]];
1758: if (!PetscAbsScalar(mdiag[i])) { /* zero diagonal */
1759: if (PetscRealPart(fshift)) {
1760: PetscInfo1(A,"Zero diagonal on row %D\n",i);
1761: A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1762: A->factorerror_zeropivot_value = 0.0;
1763: A->factorerror_zeropivot_row = i;
1764: } else SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Zero diagonal on row %D",i);
1765: }
1766: idiag[i] = 1.0/v[diag[i]];
1767: }
1768: PetscLogFlops(m);
1769: } else {
1770: for (i=0; i<m; i++) {
1771: mdiag[i] = v[diag[i]];
1772: idiag[i] = omega/(fshift + v[diag[i]]);
1773: }
1774: PetscLogFlops(2.0*m);
1775: }
1776: a->idiagvalid = PETSC_TRUE;
1777: return(0);
1778: }
1780: #include <../src/mat/impls/aij/seq/ftn-kernels/frelax.h>
1781: PetscErrorCode MatSOR_SeqAIJ(Mat A,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
1782: {
1783: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1784: PetscScalar *x,d,sum,*t,scale;
1785: const MatScalar *v,*idiag=0,*mdiag;
1786: const PetscScalar *b, *bs,*xb, *ts;
1787: PetscErrorCode ierr;
1788: PetscInt n,m = A->rmap->n,i;
1789: const PetscInt *idx,*diag;
1792: its = its*lits;
1794: if (fshift != a->fshift || omega != a->omega) a->idiagvalid = PETSC_FALSE; /* must recompute idiag[] */
1795: if (!a->idiagvalid) {MatInvertDiagonal_SeqAIJ(A,omega,fshift);}
1796: a->fshift = fshift;
1797: a->omega = omega;
1799: diag = a->diag;
1800: t = a->ssor_work;
1801: idiag = a->idiag;
1802: mdiag = a->mdiag;
1804: VecGetArray(xx,&x);
1805: VecGetArrayRead(bb,&b);
1806: /* We count flops by assuming the upper triangular and lower triangular parts have the same number of nonzeros */
1807: if (flag == SOR_APPLY_UPPER) {
1808: /* apply (U + D/omega) to the vector */
1809: bs = b;
1810: for (i=0; i<m; i++) {
1811: d = fshift + mdiag[i];
1812: n = a->i[i+1] - diag[i] - 1;
1813: idx = a->j + diag[i] + 1;
1814: v = a->a + diag[i] + 1;
1815: sum = b[i]*d/omega;
1816: PetscSparseDensePlusDot(sum,bs,v,idx,n);
1817: x[i] = sum;
1818: }
1819: VecRestoreArray(xx,&x);
1820: VecRestoreArrayRead(bb,&b);
1821: PetscLogFlops(a->nz);
1822: return(0);
1823: }
1825: if (flag == SOR_APPLY_LOWER) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"SOR_APPLY_LOWER is not implemented");
1826: else if (flag & SOR_EISENSTAT) {
1827: /* Let A = L + U + D; where L is lower trianglar,
1828: U is upper triangular, E = D/omega; This routine applies
1830: (L + E)^{-1} A (U + E)^{-1}
1832: to a vector efficiently using Eisenstat's trick.
1833: */
1834: scale = (2.0/omega) - 1.0;
1836: /* x = (E + U)^{-1} b */
1837: for (i=m-1; i>=0; i--) {
1838: n = a->i[i+1] - diag[i] - 1;
1839: idx = a->j + diag[i] + 1;
1840: v = a->a + diag[i] + 1;
1841: sum = b[i];
1842: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1843: x[i] = sum*idiag[i];
1844: }
1846: /* t = b - (2*E - D)x */
1847: v = a->a;
1848: for (i=0; i<m; i++) t[i] = b[i] - scale*(v[*diag++])*x[i];
1850: /* t = (E + L)^{-1}t */
1851: ts = t;
1852: diag = a->diag;
1853: for (i=0; i<m; i++) {
1854: n = diag[i] - a->i[i];
1855: idx = a->j + a->i[i];
1856: v = a->a + a->i[i];
1857: sum = t[i];
1858: PetscSparseDenseMinusDot(sum,ts,v,idx,n);
1859: t[i] = sum*idiag[i];
1860: /* x = x + t */
1861: x[i] += t[i];
1862: }
1864: PetscLogFlops(6.0*m-1 + 2.0*a->nz);
1865: VecRestoreArray(xx,&x);
1866: VecRestoreArrayRead(bb,&b);
1867: return(0);
1868: }
1869: if (flag & SOR_ZERO_INITIAL_GUESS) {
1870: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) {
1871: for (i=0; i<m; i++) {
1872: n = diag[i] - a->i[i];
1873: idx = a->j + a->i[i];
1874: v = a->a + a->i[i];
1875: sum = b[i];
1876: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1877: t[i] = sum;
1878: x[i] = sum*idiag[i];
1879: }
1880: xb = t;
1881: PetscLogFlops(a->nz);
1882: } else xb = b;
1883: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) {
1884: for (i=m-1; i>=0; i--) {
1885: n = a->i[i+1] - diag[i] - 1;
1886: idx = a->j + diag[i] + 1;
1887: v = a->a + diag[i] + 1;
1888: sum = xb[i];
1889: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1890: if (xb == b) {
1891: x[i] = sum*idiag[i];
1892: } else {
1893: x[i] = (1-omega)*x[i] + sum*idiag[i]; /* omega in idiag */
1894: }
1895: }
1896: PetscLogFlops(a->nz); /* assumes 1/2 in upper */
1897: }
1898: its--;
1899: }
1900: while (its--) {
1901: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) {
1902: for (i=0; i<m; i++) {
1903: /* lower */
1904: n = diag[i] - a->i[i];
1905: idx = a->j + a->i[i];
1906: v = a->a + a->i[i];
1907: sum = b[i];
1908: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1909: t[i] = sum; /* save application of the lower-triangular part */
1910: /* upper */
1911: n = a->i[i+1] - diag[i] - 1;
1912: idx = a->j + diag[i] + 1;
1913: v = a->a + diag[i] + 1;
1914: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1915: x[i] = (1. - omega)*x[i] + sum*idiag[i]; /* omega in idiag */
1916: }
1917: xb = t;
1918: PetscLogFlops(2.0*a->nz);
1919: } else xb = b;
1920: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) {
1921: for (i=m-1; i>=0; i--) {
1922: sum = xb[i];
1923: if (xb == b) {
1924: /* whole matrix (no checkpointing available) */
1925: n = a->i[i+1] - a->i[i];
1926: idx = a->j + a->i[i];
1927: v = a->a + a->i[i];
1928: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1929: x[i] = (1. - omega)*x[i] + (sum + mdiag[i]*x[i])*idiag[i];
1930: } else { /* lower-triangular part has been saved, so only apply upper-triangular */
1931: n = a->i[i+1] - diag[i] - 1;
1932: idx = a->j + diag[i] + 1;
1933: v = a->a + diag[i] + 1;
1934: PetscSparseDenseMinusDot(sum,x,v,idx,n);
1935: x[i] = (1. - omega)*x[i] + sum*idiag[i]; /* omega in idiag */
1936: }
1937: }
1938: if (xb == b) {
1939: PetscLogFlops(2.0*a->nz);
1940: } else {
1941: PetscLogFlops(a->nz); /* assumes 1/2 in upper */
1942: }
1943: }
1944: }
1945: VecRestoreArray(xx,&x);
1946: VecRestoreArrayRead(bb,&b);
1947: return(0);
1948: }
1951: PetscErrorCode MatGetInfo_SeqAIJ(Mat A,MatInfoType flag,MatInfo *info)
1952: {
1953: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1956: info->block_size = 1.0;
1957: info->nz_allocated = (double)a->maxnz;
1958: info->nz_used = (double)a->nz;
1959: info->nz_unneeded = (double)(a->maxnz - a->nz);
1960: info->assemblies = (double)A->num_ass;
1961: info->mallocs = (double)A->info.mallocs;
1962: info->memory = ((PetscObject)A)->mem;
1963: if (A->factortype) {
1964: info->fill_ratio_given = A->info.fill_ratio_given;
1965: info->fill_ratio_needed = A->info.fill_ratio_needed;
1966: info->factor_mallocs = A->info.factor_mallocs;
1967: } else {
1968: info->fill_ratio_given = 0;
1969: info->fill_ratio_needed = 0;
1970: info->factor_mallocs = 0;
1971: }
1972: return(0);
1973: }
1975: PetscErrorCode MatZeroRows_SeqAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
1976: {
1977: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1978: PetscInt i,m = A->rmap->n - 1;
1979: PetscErrorCode ierr;
1980: const PetscScalar *xx;
1981: PetscScalar *bb;
1982: PetscInt d = 0;
1985: if (x && b) {
1986: VecGetArrayRead(x,&xx);
1987: VecGetArray(b,&bb);
1988: for (i=0; i<N; i++) {
1989: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
1990: if (rows[i] >= A->cmap->n) continue;
1991: bb[rows[i]] = diag*xx[rows[i]];
1992: }
1993: VecRestoreArrayRead(x,&xx);
1994: VecRestoreArray(b,&bb);
1995: }
1997: if (a->keepnonzeropattern) {
1998: for (i=0; i<N; i++) {
1999: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
2000: PetscArrayzero(&a->a[a->i[rows[i]]],a->ilen[rows[i]]);
2001: }
2002: if (diag != 0.0) {
2003: for (i=0; i<N; i++) {
2004: d = rows[i];
2005: if (rows[i] >= A->cmap->n) continue;
2006: if (a->diag[d] >= a->i[d+1]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry in the zeroed row %D",d);
2007: }
2008: for (i=0; i<N; i++) {
2009: if (rows[i] >= A->cmap->n) continue;
2010: a->a[a->diag[rows[i]]] = diag;
2011: }
2012: }
2013: } else {
2014: if (diag != 0.0) {
2015: for (i=0; i<N; i++) {
2016: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
2017: if (a->ilen[rows[i]] > 0) {
2018: if (rows[i] >= A->cmap->n) {
2019: a->ilen[rows[i]] = 0;
2020: } else {
2021: a->ilen[rows[i]] = 1;
2022: a->a[a->i[rows[i]]] = diag;
2023: a->j[a->i[rows[i]]] = rows[i];
2024: }
2025: } else if (rows[i] < A->cmap->n) { /* in case row was completely empty */
2026: MatSetValues_SeqAIJ(A,1,&rows[i],1,&rows[i],&diag,INSERT_VALUES);
2027: }
2028: }
2029: } else {
2030: for (i=0; i<N; i++) {
2031: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
2032: a->ilen[rows[i]] = 0;
2033: }
2034: }
2035: A->nonzerostate++;
2036: }
2037: (*A->ops->assemblyend)(A,MAT_FINAL_ASSEMBLY);
2038: return(0);
2039: }
2041: PetscErrorCode MatZeroRowsColumns_SeqAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
2042: {
2043: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2044: PetscInt i,j,m = A->rmap->n - 1,d = 0;
2045: PetscErrorCode ierr;
2046: PetscBool missing,*zeroed,vecs = PETSC_FALSE;
2047: const PetscScalar *xx;
2048: PetscScalar *bb;
2051: if (x && b) {
2052: VecGetArrayRead(x,&xx);
2053: VecGetArray(b,&bb);
2054: vecs = PETSC_TRUE;
2055: }
2056: PetscCalloc1(A->rmap->n,&zeroed);
2057: for (i=0; i<N; i++) {
2058: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
2059: PetscArrayzero(&a->a[a->i[rows[i]]],a->ilen[rows[i]]);
2061: zeroed[rows[i]] = PETSC_TRUE;
2062: }
2063: for (i=0; i<A->rmap->n; i++) {
2064: if (!zeroed[i]) {
2065: for (j=a->i[i]; j<a->i[i+1]; j++) {
2066: if (a->j[j] < A->rmap->n && zeroed[a->j[j]]) {
2067: if (vecs) bb[i] -= a->a[j]*xx[a->j[j]];
2068: a->a[j] = 0.0;
2069: }
2070: }
2071: } else if (vecs && i < A->cmap->N) bb[i] = diag*xx[i];
2072: }
2073: if (x && b) {
2074: VecRestoreArrayRead(x,&xx);
2075: VecRestoreArray(b,&bb);
2076: }
2077: PetscFree(zeroed);
2078: if (diag != 0.0) {
2079: MatMissingDiagonal_SeqAIJ(A,&missing,&d);
2080: if (missing) {
2081: for (i=0; i<N; i++) {
2082: if (rows[i] >= A->cmap->N) continue;
2083: if (a->nonew && rows[i] >= d) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry in row %D (%D)",d,rows[i]);
2084: MatSetValues_SeqAIJ(A,1,&rows[i],1,&rows[i],&diag,INSERT_VALUES);
2085: }
2086: } else {
2087: for (i=0; i<N; i++) {
2088: a->a[a->diag[rows[i]]] = diag;
2089: }
2090: }
2091: }
2092: (*A->ops->assemblyend)(A,MAT_FINAL_ASSEMBLY);
2093: return(0);
2094: }
2096: PetscErrorCode MatGetRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
2097: {
2098: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2099: PetscInt *itmp;
2102: if (row < 0 || row >= A->rmap->n) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row %D out of range",row);
2104: *nz = a->i[row+1] - a->i[row];
2105: if (v) *v = a->a + a->i[row];
2106: if (idx) {
2107: itmp = a->j + a->i[row];
2108: if (*nz) *idx = itmp;
2109: else *idx = 0;
2110: }
2111: return(0);
2112: }
2114: /* remove this function? */
2115: PetscErrorCode MatRestoreRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
2116: {
2118: return(0);
2119: }
2121: PetscErrorCode MatNorm_SeqAIJ(Mat A,NormType type,PetscReal *nrm)
2122: {
2123: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2124: MatScalar *v = a->a;
2125: PetscReal sum = 0.0;
2127: PetscInt i,j;
2130: if (type == NORM_FROBENIUS) {
2131: #if defined(PETSC_USE_REAL___FP16)
2132: PetscBLASInt one = 1,nz = a->nz;
2133: *nrm = BLASnrm2_(&nz,v,&one);
2134: #else
2135: for (i=0; i<a->nz; i++) {
2136: sum += PetscRealPart(PetscConj(*v)*(*v)); v++;
2137: }
2138: *nrm = PetscSqrtReal(sum);
2139: #endif
2140: PetscLogFlops(2*a->nz);
2141: } else if (type == NORM_1) {
2142: PetscReal *tmp;
2143: PetscInt *jj = a->j;
2144: PetscCalloc1(A->cmap->n+1,&tmp);
2145: *nrm = 0.0;
2146: for (j=0; j<a->nz; j++) {
2147: tmp[*jj++] += PetscAbsScalar(*v); v++;
2148: }
2149: for (j=0; j<A->cmap->n; j++) {
2150: if (tmp[j] > *nrm) *nrm = tmp[j];
2151: }
2152: PetscFree(tmp);
2153: PetscLogFlops(PetscMax(a->nz-1,0));
2154: } else if (type == NORM_INFINITY) {
2155: *nrm = 0.0;
2156: for (j=0; j<A->rmap->n; j++) {
2157: v = a->a + a->i[j];
2158: sum = 0.0;
2159: for (i=0; i<a->i[j+1]-a->i[j]; i++) {
2160: sum += PetscAbsScalar(*v); v++;
2161: }
2162: if (sum > *nrm) *nrm = sum;
2163: }
2164: PetscLogFlops(PetscMax(a->nz-1,0));
2165: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"No support for two norm");
2166: return(0);
2167: }
2169: /* Merged from MatGetSymbolicTranspose_SeqAIJ() - replace MatGetSymbolicTranspose_SeqAIJ()? */
2170: PetscErrorCode MatTransposeSymbolic_SeqAIJ(Mat A,Mat *B)
2171: {
2173: PetscInt i,j,anzj;
2174: Mat_SeqAIJ *a=(Mat_SeqAIJ*)A->data,*b;
2175: PetscInt an=A->cmap->N,am=A->rmap->N;
2176: PetscInt *ati,*atj,*atfill,*ai=a->i,*aj=a->j;
2179: /* Allocate space for symbolic transpose info and work array */
2180: PetscCalloc1(an+1,&ati);
2181: PetscMalloc1(ai[am],&atj);
2182: PetscMalloc1(an,&atfill);
2184: /* Walk through aj and count ## of non-zeros in each row of A^T. */
2185: /* Note: offset by 1 for fast conversion into csr format. */
2186: for (i=0;i<ai[am];i++) ati[aj[i]+1] += 1;
2187: /* Form ati for csr format of A^T. */
2188: for (i=0;i<an;i++) ati[i+1] += ati[i];
2190: /* Copy ati into atfill so we have locations of the next free space in atj */
2191: PetscArraycpy(atfill,ati,an);
2193: /* Walk through A row-wise and mark nonzero entries of A^T. */
2194: for (i=0;i<am;i++) {
2195: anzj = ai[i+1] - ai[i];
2196: for (j=0;j<anzj;j++) {
2197: atj[atfill[*aj]] = i;
2198: atfill[*aj++] += 1;
2199: }
2200: }
2202: /* Clean up temporary space and complete requests. */
2203: PetscFree(atfill);
2204: MatCreateSeqAIJWithArrays(PetscObjectComm((PetscObject)A),an,am,ati,atj,NULL,B);
2205: MatSetBlockSizes(*B,PetscAbs(A->cmap->bs),PetscAbs(A->rmap->bs));
2207: b = (Mat_SeqAIJ*)((*B)->data);
2208: b->free_a = PETSC_FALSE;
2209: b->free_ij = PETSC_TRUE;
2210: b->nonew = 0;
2211: return(0);
2212: }
2214: PetscErrorCode MatIsTranspose_SeqAIJ(Mat A,Mat B,PetscReal tol,PetscBool *f)
2215: {
2216: Mat_SeqAIJ *aij = (Mat_SeqAIJ*) A->data,*bij = (Mat_SeqAIJ*) B->data;
2217: PetscInt *adx,*bdx,*aii,*bii,*aptr,*bptr;
2218: MatScalar *va,*vb;
2220: PetscInt ma,na,mb,nb, i;
2223: MatGetSize(A,&ma,&na);
2224: MatGetSize(B,&mb,&nb);
2225: if (ma!=nb || na!=mb) {
2226: *f = PETSC_FALSE;
2227: return(0);
2228: }
2229: aii = aij->i; bii = bij->i;
2230: adx = aij->j; bdx = bij->j;
2231: va = aij->a; vb = bij->a;
2232: PetscMalloc1(ma,&aptr);
2233: PetscMalloc1(mb,&bptr);
2234: for (i=0; i<ma; i++) aptr[i] = aii[i];
2235: for (i=0; i<mb; i++) bptr[i] = bii[i];
2237: *f = PETSC_TRUE;
2238: for (i=0; i<ma; i++) {
2239: while (aptr[i]<aii[i+1]) {
2240: PetscInt idc,idr;
2241: PetscScalar vc,vr;
2242: /* column/row index/value */
2243: idc = adx[aptr[i]];
2244: idr = bdx[bptr[idc]];
2245: vc = va[aptr[i]];
2246: vr = vb[bptr[idc]];
2247: if (i!=idr || PetscAbsScalar(vc-vr) > tol) {
2248: *f = PETSC_FALSE;
2249: goto done;
2250: } else {
2251: aptr[i]++;
2252: if (B || i!=idc) bptr[idc]++;
2253: }
2254: }
2255: }
2256: done:
2257: PetscFree(aptr);
2258: PetscFree(bptr);
2259: return(0);
2260: }
2262: PetscErrorCode MatIsHermitianTranspose_SeqAIJ(Mat A,Mat B,PetscReal tol,PetscBool *f)
2263: {
2264: Mat_SeqAIJ *aij = (Mat_SeqAIJ*) A->data,*bij = (Mat_SeqAIJ*) B->data;
2265: PetscInt *adx,*bdx,*aii,*bii,*aptr,*bptr;
2266: MatScalar *va,*vb;
2268: PetscInt ma,na,mb,nb, i;
2271: MatGetSize(A,&ma,&na);
2272: MatGetSize(B,&mb,&nb);
2273: if (ma!=nb || na!=mb) {
2274: *f = PETSC_FALSE;
2275: return(0);
2276: }
2277: aii = aij->i; bii = bij->i;
2278: adx = aij->j; bdx = bij->j;
2279: va = aij->a; vb = bij->a;
2280: PetscMalloc1(ma,&aptr);
2281: PetscMalloc1(mb,&bptr);
2282: for (i=0; i<ma; i++) aptr[i] = aii[i];
2283: for (i=0; i<mb; i++) bptr[i] = bii[i];
2285: *f = PETSC_TRUE;
2286: for (i=0; i<ma; i++) {
2287: while (aptr[i]<aii[i+1]) {
2288: PetscInt idc,idr;
2289: PetscScalar vc,vr;
2290: /* column/row index/value */
2291: idc = adx[aptr[i]];
2292: idr = bdx[bptr[idc]];
2293: vc = va[aptr[i]];
2294: vr = vb[bptr[idc]];
2295: if (i!=idr || PetscAbsScalar(vc-PetscConj(vr)) > tol) {
2296: *f = PETSC_FALSE;
2297: goto done;
2298: } else {
2299: aptr[i]++;
2300: if (B || i!=idc) bptr[idc]++;
2301: }
2302: }
2303: }
2304: done:
2305: PetscFree(aptr);
2306: PetscFree(bptr);
2307: return(0);
2308: }
2310: PetscErrorCode MatIsSymmetric_SeqAIJ(Mat A,PetscReal tol,PetscBool *f)
2311: {
2315: MatIsTranspose_SeqAIJ(A,A,tol,f);
2316: return(0);
2317: }
2319: PetscErrorCode MatIsHermitian_SeqAIJ(Mat A,PetscReal tol,PetscBool *f)
2320: {
2324: MatIsHermitianTranspose_SeqAIJ(A,A,tol,f);
2325: return(0);
2326: }
2328: PetscErrorCode MatDiagonalScale_SeqAIJ(Mat A,Vec ll,Vec rr)
2329: {
2330: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2331: const PetscScalar *l,*r;
2332: PetscScalar x;
2333: MatScalar *v;
2334: PetscErrorCode ierr;
2335: PetscInt i,j,m = A->rmap->n,n = A->cmap->n,M,nz = a->nz;
2336: const PetscInt *jj;
2339: if (ll) {
2340: /* The local size is used so that VecMPI can be passed to this routine
2341: by MatDiagonalScale_MPIAIJ */
2342: VecGetLocalSize(ll,&m);
2343: if (m != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Left scaling vector wrong length");
2344: VecGetArrayRead(ll,&l);
2345: v = a->a;
2346: for (i=0; i<m; i++) {
2347: x = l[i];
2348: M = a->i[i+1] - a->i[i];
2349: for (j=0; j<M; j++) (*v++) *= x;
2350: }
2351: VecRestoreArrayRead(ll,&l);
2352: PetscLogFlops(nz);
2353: }
2354: if (rr) {
2355: VecGetLocalSize(rr,&n);
2356: if (n != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Right scaling vector wrong length");
2357: VecGetArrayRead(rr,&r);
2358: v = a->a; jj = a->j;
2359: for (i=0; i<nz; i++) (*v++) *= r[*jj++];
2360: VecRestoreArrayRead(rr,&r);
2361: PetscLogFlops(nz);
2362: }
2363: MatSeqAIJInvalidateDiagonal(A);
2364: return(0);
2365: }
2367: PetscErrorCode MatCreateSubMatrix_SeqAIJ(Mat A,IS isrow,IS iscol,PetscInt csize,MatReuse scall,Mat *B)
2368: {
2369: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data,*c;
2371: PetscInt *smap,i,k,kstart,kend,oldcols = A->cmap->n,*lens;
2372: PetscInt row,mat_i,*mat_j,tcol,first,step,*mat_ilen,sum,lensi;
2373: const PetscInt *irow,*icol;
2374: PetscInt nrows,ncols;
2375: PetscInt *starts,*j_new,*i_new,*aj = a->j,*ai = a->i,ii,*ailen = a->ilen;
2376: MatScalar *a_new,*mat_a;
2377: Mat C;
2378: PetscBool stride;
2382: ISGetIndices(isrow,&irow);
2383: ISGetLocalSize(isrow,&nrows);
2384: ISGetLocalSize(iscol,&ncols);
2386: PetscObjectTypeCompare((PetscObject)iscol,ISSTRIDE,&stride);
2387: if (stride) {
2388: ISStrideGetInfo(iscol,&first,&step);
2389: } else {
2390: first = 0;
2391: step = 0;
2392: }
2393: if (stride && step == 1) {
2394: /* special case of contiguous rows */
2395: PetscMalloc2(nrows,&lens,nrows,&starts);
2396: /* loop over new rows determining lens and starting points */
2397: for (i=0; i<nrows; i++) {
2398: kstart = ai[irow[i]];
2399: kend = kstart + ailen[irow[i]];
2400: starts[i] = kstart;
2401: for (k=kstart; k<kend; k++) {
2402: if (aj[k] >= first) {
2403: starts[i] = k;
2404: break;
2405: }
2406: }
2407: sum = 0;
2408: while (k < kend) {
2409: if (aj[k++] >= first+ncols) break;
2410: sum++;
2411: }
2412: lens[i] = sum;
2413: }
2414: /* create submatrix */
2415: if (scall == MAT_REUSE_MATRIX) {
2416: PetscInt n_cols,n_rows;
2417: MatGetSize(*B,&n_rows,&n_cols);
2418: if (n_rows != nrows || n_cols != ncols) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Reused submatrix wrong size");
2419: MatZeroEntries(*B);
2420: C = *B;
2421: } else {
2422: PetscInt rbs,cbs;
2423: MatCreate(PetscObjectComm((PetscObject)A),&C);
2424: MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);
2425: ISGetBlockSize(isrow,&rbs);
2426: ISGetBlockSize(iscol,&cbs);
2427: MatSetBlockSizes(C,rbs,cbs);
2428: MatSetType(C,((PetscObject)A)->type_name);
2429: MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);
2430: }
2431: c = (Mat_SeqAIJ*)C->data;
2433: /* loop over rows inserting into submatrix */
2434: a_new = c->a;
2435: j_new = c->j;
2436: i_new = c->i;
2438: for (i=0; i<nrows; i++) {
2439: ii = starts[i];
2440: lensi = lens[i];
2441: for (k=0; k<lensi; k++) {
2442: *j_new++ = aj[ii+k] - first;
2443: }
2444: PetscArraycpy(a_new,a->a + starts[i],lensi);
2445: a_new += lensi;
2446: i_new[i+1] = i_new[i] + lensi;
2447: c->ilen[i] = lensi;
2448: }
2449: PetscFree2(lens,starts);
2450: } else {
2451: ISGetIndices(iscol,&icol);
2452: PetscCalloc1(oldcols,&smap);
2453: PetscMalloc1(1+nrows,&lens);
2454: for (i=0; i<ncols; i++) {
2455: #if defined(PETSC_USE_DEBUG)
2456: if (icol[i] >= oldcols) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Requesting column beyond largest column icol[%D] %D <= A->cmap->n %D",i,icol[i],oldcols);
2457: #endif
2458: smap[icol[i]] = i+1;
2459: }
2461: /* determine lens of each row */
2462: for (i=0; i<nrows; i++) {
2463: kstart = ai[irow[i]];
2464: kend = kstart + a->ilen[irow[i]];
2465: lens[i] = 0;
2466: for (k=kstart; k<kend; k++) {
2467: if (smap[aj[k]]) {
2468: lens[i]++;
2469: }
2470: }
2471: }
2472: /* Create and fill new matrix */
2473: if (scall == MAT_REUSE_MATRIX) {
2474: PetscBool equal;
2476: c = (Mat_SeqAIJ*)((*B)->data);
2477: if ((*B)->rmap->n != nrows || (*B)->cmap->n != ncols) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong size");
2478: PetscArraycmp(c->ilen,lens,(*B)->rmap->n,&equal);
2479: if (!equal) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong no of nonzeros");
2480: PetscArrayzero(c->ilen,(*B)->rmap->n);
2481: C = *B;
2482: } else {
2483: PetscInt rbs,cbs;
2484: MatCreate(PetscObjectComm((PetscObject)A),&C);
2485: MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);
2486: ISGetBlockSize(isrow,&rbs);
2487: ISGetBlockSize(iscol,&cbs);
2488: MatSetBlockSizes(C,rbs,cbs);
2489: MatSetType(C,((PetscObject)A)->type_name);
2490: MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);
2491: }
2492: c = (Mat_SeqAIJ*)(C->data);
2493: for (i=0; i<nrows; i++) {
2494: row = irow[i];
2495: kstart = ai[row];
2496: kend = kstart + a->ilen[row];
2497: mat_i = c->i[i];
2498: mat_j = c->j + mat_i;
2499: mat_a = c->a + mat_i;
2500: mat_ilen = c->ilen + i;
2501: for (k=kstart; k<kend; k++) {
2502: if ((tcol=smap[a->j[k]])) {
2503: *mat_j++ = tcol - 1;
2504: *mat_a++ = a->a[k];
2505: (*mat_ilen)++;
2507: }
2508: }
2509: }
2510: /* Free work space */
2511: ISRestoreIndices(iscol,&icol);
2512: PetscFree(smap);
2513: PetscFree(lens);
2514: /* sort */
2515: for (i = 0; i < nrows; i++) {
2516: PetscInt ilen;
2518: mat_i = c->i[i];
2519: mat_j = c->j + mat_i;
2520: mat_a = c->a + mat_i;
2521: ilen = c->ilen[i];
2522: PetscSortIntWithScalarArray(ilen,mat_j,mat_a);
2523: }
2524: }
2525: MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);
2526: MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);
2528: ISRestoreIndices(isrow,&irow);
2529: *B = C;
2530: return(0);
2531: }
2533: PetscErrorCode MatGetMultiProcBlock_SeqAIJ(Mat mat,MPI_Comm subComm,MatReuse scall,Mat *subMat)
2534: {
2536: Mat B;
2539: if (scall == MAT_INITIAL_MATRIX) {
2540: MatCreate(subComm,&B);
2541: MatSetSizes(B,mat->rmap->n,mat->cmap->n,mat->rmap->n,mat->cmap->n);
2542: MatSetBlockSizesFromMats(B,mat,mat);
2543: MatSetType(B,MATSEQAIJ);
2544: MatDuplicateNoCreate_SeqAIJ(B,mat,MAT_COPY_VALUES,PETSC_TRUE);
2545: *subMat = B;
2546: } else {
2547: MatCopy_SeqAIJ(mat,*subMat,SAME_NONZERO_PATTERN);
2548: }
2549: return(0);
2550: }
2552: PetscErrorCode MatILUFactor_SeqAIJ(Mat inA,IS row,IS col,const MatFactorInfo *info)
2553: {
2554: Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data;
2556: Mat outA;
2557: PetscBool row_identity,col_identity;
2560: if (info->levels != 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only levels=0 supported for in-place ilu");
2562: ISIdentity(row,&row_identity);
2563: ISIdentity(col,&col_identity);
2565: outA = inA;
2566: outA->factortype = MAT_FACTOR_LU;
2567: PetscFree(inA->solvertype);
2568: PetscStrallocpy(MATSOLVERPETSC,&inA->solvertype);
2570: PetscObjectReference((PetscObject)row);
2571: ISDestroy(&a->row);
2573: a->row = row;
2575: PetscObjectReference((PetscObject)col);
2576: ISDestroy(&a->col);
2578: a->col = col;
2580: /* Create the inverse permutation so that it can be used in MatLUFactorNumeric() */
2581: ISDestroy(&a->icol);
2582: ISInvertPermutation(col,PETSC_DECIDE,&a->icol);
2583: PetscLogObjectParent((PetscObject)inA,(PetscObject)a->icol);
2585: if (!a->solve_work) { /* this matrix may have been factored before */
2586: PetscMalloc1(inA->rmap->n+1,&a->solve_work);
2587: PetscLogObjectMemory((PetscObject)inA, (inA->rmap->n+1)*sizeof(PetscScalar));
2588: }
2590: MatMarkDiagonal_SeqAIJ(inA);
2591: if (row_identity && col_identity) {
2592: MatLUFactorNumeric_SeqAIJ_inplace(outA,inA,info);
2593: } else {
2594: MatLUFactorNumeric_SeqAIJ_InplaceWithPerm(outA,inA,info);
2595: }
2596: return(0);
2597: }
2599: PetscErrorCode MatScale_SeqAIJ(Mat inA,PetscScalar alpha)
2600: {
2601: Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data;
2602: PetscScalar oalpha = alpha;
2604: PetscBLASInt one = 1,bnz;
2607: PetscBLASIntCast(a->nz,&bnz);
2608: PetscStackCallBLAS("BLASscal",BLASscal_(&bnz,&oalpha,a->a,&one));
2609: PetscLogFlops(a->nz);
2610: MatSeqAIJInvalidateDiagonal(inA);
2611: return(0);
2612: }
2614: PetscErrorCode MatDestroySubMatrix_Private(Mat_SubSppt *submatj)
2615: {
2617: PetscInt i;
2620: if (!submatj->id) { /* delete data that are linked only to submats[id=0] */
2621: PetscFree4(submatj->sbuf1,submatj->ptr,submatj->tmp,submatj->ctr);
2623: for (i=0; i<submatj->nrqr; ++i) {
2624: PetscFree(submatj->sbuf2[i]);
2625: }
2626: PetscFree3(submatj->sbuf2,submatj->req_size,submatj->req_source1);
2628: if (submatj->rbuf1) {
2629: PetscFree(submatj->rbuf1[0]);
2630: PetscFree(submatj->rbuf1);
2631: }
2633: for (i=0; i<submatj->nrqs; ++i) {
2634: PetscFree(submatj->rbuf3[i]);
2635: }
2636: PetscFree3(submatj->req_source2,submatj->rbuf2,submatj->rbuf3);
2637: PetscFree(submatj->pa);
2638: }
2640: #if defined(PETSC_USE_CTABLE)
2641: PetscTableDestroy((PetscTable*)&submatj->rmap);
2642: if (submatj->cmap_loc) {PetscFree(submatj->cmap_loc);}
2643: PetscFree(submatj->rmap_loc);
2644: #else
2645: PetscFree(submatj->rmap);
2646: #endif
2648: if (!submatj->allcolumns) {
2649: #if defined(PETSC_USE_CTABLE)
2650: PetscTableDestroy((PetscTable*)&submatj->cmap);
2651: #else
2652: PetscFree(submatj->cmap);
2653: #endif
2654: }
2655: PetscFree(submatj->row2proc);
2657: PetscFree(submatj);
2658: return(0);
2659: }
2661: PetscErrorCode MatDestroySubMatrix_SeqAIJ(Mat C)
2662: {
2664: Mat_SeqAIJ *c = (Mat_SeqAIJ*)C->data;
2665: Mat_SubSppt *submatj = c->submatis1;
2668: (*submatj->destroy)(C);
2669: MatDestroySubMatrix_Private(submatj);
2670: return(0);
2671: }
2673: PetscErrorCode MatDestroySubMatrices_SeqAIJ(PetscInt n,Mat *mat[])
2674: {
2676: PetscInt i;
2677: Mat C;
2678: Mat_SeqAIJ *c;
2679: Mat_SubSppt *submatj;
2682: for (i=0; i<n; i++) {
2683: C = (*mat)[i];
2684: c = (Mat_SeqAIJ*)C->data;
2685: submatj = c->submatis1;
2686: if (submatj) {
2687: if (--((PetscObject)C)->refct <= 0) {
2688: (*submatj->destroy)(C);
2689: MatDestroySubMatrix_Private(submatj);
2690: PetscFree(C->defaultvectype);
2691: PetscLayoutDestroy(&C->rmap);
2692: PetscLayoutDestroy(&C->cmap);
2693: PetscHeaderDestroy(&C);
2694: }
2695: } else {
2696: MatDestroy(&C);
2697: }
2698: }
2700: /* Destroy Dummy submatrices created for reuse */
2701: MatDestroySubMatrices_Dummy(n,mat);
2703: PetscFree(*mat);
2704: return(0);
2705: }
2707: PetscErrorCode MatCreateSubMatrices_SeqAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[])
2708: {
2710: PetscInt i;
2713: if (scall == MAT_INITIAL_MATRIX) {
2714: PetscCalloc1(n+1,B);
2715: }
2717: for (i=0; i<n; i++) {
2718: MatCreateSubMatrix_SeqAIJ(A,irow[i],icol[i],PETSC_DECIDE,scall,&(*B)[i]);
2719: }
2720: return(0);
2721: }
2723: PetscErrorCode MatIncreaseOverlap_SeqAIJ(Mat A,PetscInt is_max,IS is[],PetscInt ov)
2724: {
2725: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2727: PetscInt row,i,j,k,l,m,n,*nidx,isz,val;
2728: const PetscInt *idx;
2729: PetscInt start,end,*ai,*aj;
2730: PetscBT table;
2733: m = A->rmap->n;
2734: ai = a->i;
2735: aj = a->j;
2737: if (ov < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"illegal negative overlap value used");
2739: PetscMalloc1(m+1,&nidx);
2740: PetscBTCreate(m,&table);
2742: for (i=0; i<is_max; i++) {
2743: /* Initialize the two local arrays */
2744: isz = 0;
2745: PetscBTMemzero(m,table);
2747: /* Extract the indices, assume there can be duplicate entries */
2748: ISGetIndices(is[i],&idx);
2749: ISGetLocalSize(is[i],&n);
2751: /* Enter these into the temp arrays. I.e., mark table[row], enter row into new index */
2752: for (j=0; j<n; ++j) {
2753: if (!PetscBTLookupSet(table,idx[j])) nidx[isz++] = idx[j];
2754: }
2755: ISRestoreIndices(is[i],&idx);
2756: ISDestroy(&is[i]);
2758: k = 0;
2759: for (j=0; j<ov; j++) { /* for each overlap */
2760: n = isz;
2761: for (; k<n; k++) { /* do only those rows in nidx[k], which are not done yet */
2762: row = nidx[k];
2763: start = ai[row];
2764: end = ai[row+1];
2765: for (l = start; l<end; l++) {
2766: val = aj[l];
2767: if (!PetscBTLookupSet(table,val)) nidx[isz++] = val;
2768: }
2769: }
2770: }
2771: ISCreateGeneral(PETSC_COMM_SELF,isz,nidx,PETSC_COPY_VALUES,(is+i));
2772: }
2773: PetscBTDestroy(&table);
2774: PetscFree(nidx);
2775: return(0);
2776: }
2778: /* -------------------------------------------------------------- */
2779: PetscErrorCode MatPermute_SeqAIJ(Mat A,IS rowp,IS colp,Mat *B)
2780: {
2781: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2783: PetscInt i,nz = 0,m = A->rmap->n,n = A->cmap->n;
2784: const PetscInt *row,*col;
2785: PetscInt *cnew,j,*lens;
2786: IS icolp,irowp;
2787: PetscInt *cwork = NULL;
2788: PetscScalar *vwork = NULL;
2791: ISInvertPermutation(rowp,PETSC_DECIDE,&irowp);
2792: ISGetIndices(irowp,&row);
2793: ISInvertPermutation(colp,PETSC_DECIDE,&icolp);
2794: ISGetIndices(icolp,&col);
2796: /* determine lengths of permuted rows */
2797: PetscMalloc1(m+1,&lens);
2798: for (i=0; i<m; i++) lens[row[i]] = a->i[i+1] - a->i[i];
2799: MatCreate(PetscObjectComm((PetscObject)A),B);
2800: MatSetSizes(*B,m,n,m,n);
2801: MatSetBlockSizesFromMats(*B,A,A);
2802: MatSetType(*B,((PetscObject)A)->type_name);
2803: MatSeqAIJSetPreallocation_SeqAIJ(*B,0,lens);
2804: PetscFree(lens);
2806: PetscMalloc1(n,&cnew);
2807: for (i=0; i<m; i++) {
2808: MatGetRow_SeqAIJ(A,i,&nz,&cwork,&vwork);
2809: for (j=0; j<nz; j++) cnew[j] = col[cwork[j]];
2810: MatSetValues_SeqAIJ(*B,1,&row[i],nz,cnew,vwork,INSERT_VALUES);
2811: MatRestoreRow_SeqAIJ(A,i,&nz,&cwork,&vwork);
2812: }
2813: PetscFree(cnew);
2815: (*B)->assembled = PETSC_FALSE;
2817: MatAssemblyBegin(*B,MAT_FINAL_ASSEMBLY);
2818: MatAssemblyEnd(*B,MAT_FINAL_ASSEMBLY);
2819: ISRestoreIndices(irowp,&row);
2820: ISRestoreIndices(icolp,&col);
2821: ISDestroy(&irowp);
2822: ISDestroy(&icolp);
2823: return(0);
2824: }
2826: PetscErrorCode MatCopy_SeqAIJ(Mat A,Mat B,MatStructure str)
2827: {
2831: /* If the two matrices have the same copy implementation, use fast copy. */
2832: if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) {
2833: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2834: Mat_SeqAIJ *b = (Mat_SeqAIJ*)B->data;
2836: if (a->i[A->rmap->n] != b->i[B->rmap->n]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Number of nonzeros in two matrices are different");
2837: PetscArraycpy(b->a,a->a,a->i[A->rmap->n]);
2838: PetscObjectStateIncrease((PetscObject)B);
2839: } else {
2840: MatCopy_Basic(A,B,str);
2841: }
2842: return(0);
2843: }
2845: PetscErrorCode MatSetUp_SeqAIJ(Mat A)
2846: {
2850: MatSeqAIJSetPreallocation_SeqAIJ(A,PETSC_DEFAULT,0);
2851: return(0);
2852: }
2854: PetscErrorCode MatSeqAIJGetArray_SeqAIJ(Mat A,PetscScalar *array[])
2855: {
2856: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2859: *array = a->a;
2860: return(0);
2861: }
2863: PetscErrorCode MatSeqAIJRestoreArray_SeqAIJ(Mat A,PetscScalar *array[])
2864: {
2866: return(0);
2867: }
2869: /*
2870: Computes the number of nonzeros per row needed for preallocation when X and Y
2871: have different nonzero structure.
2872: */
2873: PetscErrorCode MatAXPYGetPreallocation_SeqX_private(PetscInt m,const PetscInt *xi,const PetscInt *xj,const PetscInt *yi,const PetscInt *yj,PetscInt *nnz)
2874: {
2875: PetscInt i,j,k,nzx,nzy;
2878: /* Set the number of nonzeros in the new matrix */
2879: for (i=0; i<m; i++) {
2880: const PetscInt *xjj = xj+xi[i],*yjj = yj+yi[i];
2881: nzx = xi[i+1] - xi[i];
2882: nzy = yi[i+1] - yi[i];
2883: nnz[i] = 0;
2884: for (j=0,k=0; j<nzx; j++) { /* Point in X */
2885: for (; k<nzy && yjj[k]<xjj[j]; k++) nnz[i]++; /* Catch up to X */
2886: if (k<nzy && yjj[k]==xjj[j]) k++; /* Skip duplicate */
2887: nnz[i]++;
2888: }
2889: for (; k<nzy; k++) nnz[i]++;
2890: }
2891: return(0);
2892: }
2894: PetscErrorCode MatAXPYGetPreallocation_SeqAIJ(Mat Y,Mat X,PetscInt *nnz)
2895: {
2896: PetscInt m = Y->rmap->N;
2897: Mat_SeqAIJ *x = (Mat_SeqAIJ*)X->data;
2898: Mat_SeqAIJ *y = (Mat_SeqAIJ*)Y->data;
2902: /* Set the number of nonzeros in the new matrix */
2903: MatAXPYGetPreallocation_SeqX_private(m,x->i,x->j,y->i,y->j,nnz);
2904: return(0);
2905: }
2907: PetscErrorCode MatAXPY_SeqAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
2908: {
2910: Mat_SeqAIJ *x = (Mat_SeqAIJ*)X->data,*y = (Mat_SeqAIJ*)Y->data;
2911: PetscBLASInt one=1,bnz;
2914: PetscBLASIntCast(x->nz,&bnz);
2915: if (str == SAME_NONZERO_PATTERN) {
2916: PetscScalar alpha = a;
2917: PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one));
2918: MatSeqAIJInvalidateDiagonal(Y);
2919: PetscObjectStateIncrease((PetscObject)Y);
2920: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
2921: MatAXPY_Basic(Y,a,X,str);
2922: } else {
2923: Mat B;
2924: PetscInt *nnz;
2925: PetscMalloc1(Y->rmap->N,&nnz);
2926: MatCreate(PetscObjectComm((PetscObject)Y),&B);
2927: PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);
2928: MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);
2929: MatSetBlockSizesFromMats(B,Y,Y);
2930: MatSetType(B,(MatType) ((PetscObject)Y)->type_name);
2931: MatAXPYGetPreallocation_SeqAIJ(Y,X,nnz);
2932: MatSeqAIJSetPreallocation(B,0,nnz);
2933: MatAXPY_BasicWithPreallocation(B,Y,a,X,str);
2934: MatHeaderReplace(Y,&B);
2935: PetscFree(nnz);
2936: }
2937: return(0);
2938: }
2940: PetscErrorCode MatConjugate_SeqAIJ(Mat mat)
2941: {
2942: #if defined(PETSC_USE_COMPLEX)
2943: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data;
2944: PetscInt i,nz;
2945: PetscScalar *a;
2948: nz = aij->nz;
2949: a = aij->a;
2950: for (i=0; i<nz; i++) a[i] = PetscConj(a[i]);
2951: #else
2953: #endif
2954: return(0);
2955: }
2957: PetscErrorCode MatGetRowMaxAbs_SeqAIJ(Mat A,Vec v,PetscInt idx[])
2958: {
2959: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2961: PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n;
2962: PetscReal atmp;
2963: PetscScalar *x;
2964: MatScalar *aa;
2967: if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2968: aa = a->a;
2969: ai = a->i;
2970: aj = a->j;
2972: VecSet(v,0.0);
2973: VecGetArray(v,&x);
2974: VecGetLocalSize(v,&n);
2975: if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
2976: for (i=0; i<m; i++) {
2977: ncols = ai[1] - ai[0]; ai++;
2978: x[i] = 0.0;
2979: for (j=0; j<ncols; j++) {
2980: atmp = PetscAbsScalar(*aa);
2981: if (PetscAbsScalar(x[i]) < atmp) {x[i] = atmp; if (idx) idx[i] = *aj;}
2982: aa++; aj++;
2983: }
2984: }
2985: VecRestoreArray(v,&x);
2986: return(0);
2987: }
2989: PetscErrorCode MatGetRowMax_SeqAIJ(Mat A,Vec v,PetscInt idx[])
2990: {
2991: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2993: PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n;
2994: PetscScalar *x;
2995: MatScalar *aa;
2998: if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2999: aa = a->a;
3000: ai = a->i;
3001: aj = a->j;
3003: VecSet(v,0.0);
3004: VecGetArray(v,&x);
3005: VecGetLocalSize(v,&n);
3006: if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
3007: for (i=0; i<m; i++) {
3008: ncols = ai[1] - ai[0]; ai++;
3009: if (ncols == A->cmap->n) { /* row is dense */
3010: x[i] = *aa; if (idx) idx[i] = 0;
3011: } else { /* row is sparse so already KNOW maximum is 0.0 or higher */
3012: x[i] = 0.0;
3013: if (idx) {
3014: idx[i] = 0; /* in case ncols is zero */
3015: for (j=0;j<ncols;j++) { /* find first implicit 0.0 in the row */
3016: if (aj[j] > j) {
3017: idx[i] = j;
3018: break;
3019: }
3020: }
3021: }
3022: }
3023: for (j=0; j<ncols; j++) {
3024: if (PetscRealPart(x[i]) < PetscRealPart(*aa)) {x[i] = *aa; if (idx) idx[i] = *aj;}
3025: aa++; aj++;
3026: }
3027: }
3028: VecRestoreArray(v,&x);
3029: return(0);
3030: }
3032: PetscErrorCode MatGetRowMinAbs_SeqAIJ(Mat A,Vec v,PetscInt idx[])
3033: {
3034: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3036: PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n;
3037: PetscReal atmp;
3038: PetscScalar *x;
3039: MatScalar *aa;
3042: if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3043: aa = a->a;
3044: ai = a->i;
3045: aj = a->j;
3047: VecSet(v,0.0);
3048: VecGetArray(v,&x);
3049: VecGetLocalSize(v,&n);
3050: if (n != A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector, %D vs. %D rows", A->rmap->n, n);
3051: for (i=0; i<m; i++) {
3052: ncols = ai[1] - ai[0]; ai++;
3053: if (ncols) {
3054: /* Get first nonzero */
3055: for (j = 0; j < ncols; j++) {
3056: atmp = PetscAbsScalar(aa[j]);
3057: if (atmp > 1.0e-12) {
3058: x[i] = atmp;
3059: if (idx) idx[i] = aj[j];
3060: break;
3061: }
3062: }
3063: if (j == ncols) {x[i] = PetscAbsScalar(*aa); if (idx) idx[i] = *aj;}
3064: } else {
3065: x[i] = 0.0; if (idx) idx[i] = 0;
3066: }
3067: for (j = 0; j < ncols; j++) {
3068: atmp = PetscAbsScalar(*aa);
3069: if (atmp > 1.0e-12 && PetscAbsScalar(x[i]) > atmp) {x[i] = atmp; if (idx) idx[i] = *aj;}
3070: aa++; aj++;
3071: }
3072: }
3073: VecRestoreArray(v,&x);
3074: return(0);
3075: }
3077: PetscErrorCode MatGetRowMin_SeqAIJ(Mat A,Vec v,PetscInt idx[])
3078: {
3079: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3080: PetscErrorCode ierr;
3081: PetscInt i,j,m = A->rmap->n,ncols,n;
3082: const PetscInt *ai,*aj;
3083: PetscScalar *x;
3084: const MatScalar *aa;
3087: if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3088: aa = a->a;
3089: ai = a->i;
3090: aj = a->j;
3092: VecSet(v,0.0);
3093: VecGetArray(v,&x);
3094: VecGetLocalSize(v,&n);
3095: if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
3096: for (i=0; i<m; i++) {
3097: ncols = ai[1] - ai[0]; ai++;
3098: if (ncols == A->cmap->n) { /* row is dense */
3099: x[i] = *aa; if (idx) idx[i] = 0;
3100: } else { /* row is sparse so already KNOW minimum is 0.0 or lower */
3101: x[i] = 0.0;
3102: if (idx) { /* find first implicit 0.0 in the row */
3103: idx[i] = 0; /* in case ncols is zero */
3104: for (j=0; j<ncols; j++) {
3105: if (aj[j] > j) {
3106: idx[i] = j;
3107: break;
3108: }
3109: }
3110: }
3111: }
3112: for (j=0; j<ncols; j++) {
3113: if (PetscRealPart(x[i]) > PetscRealPart(*aa)) {x[i] = *aa; if (idx) idx[i] = *aj;}
3114: aa++; aj++;
3115: }
3116: }
3117: VecRestoreArray(v,&x);
3118: return(0);
3119: }
3121: PetscErrorCode MatInvertBlockDiagonal_SeqAIJ(Mat A,const PetscScalar **values)
3122: {
3123: Mat_SeqAIJ *a = (Mat_SeqAIJ*) A->data;
3124: PetscErrorCode ierr;
3125: PetscInt i,bs = PetscAbs(A->rmap->bs),mbs = A->rmap->n/bs,ipvt[5],bs2 = bs*bs,*v_pivots,ij[7],*IJ,j;
3126: MatScalar *diag,work[25],*v_work;
3127: const PetscReal shift = 0.0;
3128: PetscBool allowzeropivot,zeropivotdetected=PETSC_FALSE;
3131: allowzeropivot = PetscNot(A->erroriffailure);
3132: if (a->ibdiagvalid) {
3133: if (values) *values = a->ibdiag;
3134: return(0);
3135: }
3136: MatMarkDiagonal_SeqAIJ(A);
3137: if (!a->ibdiag) {
3138: PetscMalloc1(bs2*mbs,&a->ibdiag);
3139: PetscLogObjectMemory((PetscObject)A,bs2*mbs*sizeof(PetscScalar));
3140: }
3141: diag = a->ibdiag;
3142: if (values) *values = a->ibdiag;
3143: /* factor and invert each block */
3144: switch (bs) {
3145: case 1:
3146: for (i=0; i<mbs; i++) {
3147: MatGetValues(A,1,&i,1,&i,diag+i);
3148: if (PetscAbsScalar(diag[i] + shift) < PETSC_MACHINE_EPSILON) {
3149: if (allowzeropivot) {
3150: A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3151: A->factorerror_zeropivot_value = PetscAbsScalar(diag[i]);
3152: A->factorerror_zeropivot_row = i;
3153: PetscInfo3(A,"Zero pivot, row %D pivot %g tolerance %g\n",i,(double)PetscAbsScalar(diag[i]),(double)PETSC_MACHINE_EPSILON);
3154: } else SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Zero pivot, row %D pivot %g tolerance %g",i,(double)PetscAbsScalar(diag[i]),(double)PETSC_MACHINE_EPSILON);
3155: }
3156: diag[i] = (PetscScalar)1.0 / (diag[i] + shift);
3157: }
3158: break;
3159: case 2:
3160: for (i=0; i<mbs; i++) {
3161: ij[0] = 2*i; ij[1] = 2*i + 1;
3162: MatGetValues(A,2,ij,2,ij,diag);
3163: PetscKernel_A_gets_inverse_A_2(diag,shift,allowzeropivot,&zeropivotdetected);
3164: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3165: PetscKernel_A_gets_transpose_A_2(diag);
3166: diag += 4;
3167: }
3168: break;
3169: case 3:
3170: for (i=0; i<mbs; i++) {
3171: ij[0] = 3*i; ij[1] = 3*i + 1; ij[2] = 3*i + 2;
3172: MatGetValues(A,3,ij,3,ij,diag);
3173: PetscKernel_A_gets_inverse_A_3(diag,shift,allowzeropivot,&zeropivotdetected);
3174: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3175: PetscKernel_A_gets_transpose_A_3(diag);
3176: diag += 9;
3177: }
3178: break;
3179: case 4:
3180: for (i=0; i<mbs; i++) {
3181: ij[0] = 4*i; ij[1] = 4*i + 1; ij[2] = 4*i + 2; ij[3] = 4*i + 3;
3182: MatGetValues(A,4,ij,4,ij,diag);
3183: PetscKernel_A_gets_inverse_A_4(diag,shift,allowzeropivot,&zeropivotdetected);
3184: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3185: PetscKernel_A_gets_transpose_A_4(diag);
3186: diag += 16;
3187: }
3188: break;
3189: case 5:
3190: for (i=0; i<mbs; i++) {
3191: ij[0] = 5*i; ij[1] = 5*i + 1; ij[2] = 5*i + 2; ij[3] = 5*i + 3; ij[4] = 5*i + 4;
3192: MatGetValues(A,5,ij,5,ij,diag);
3193: PetscKernel_A_gets_inverse_A_5(diag,ipvt,work,shift,allowzeropivot,&zeropivotdetected);
3194: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3195: PetscKernel_A_gets_transpose_A_5(diag);
3196: diag += 25;
3197: }
3198: break;
3199: case 6:
3200: for (i=0; i<mbs; i++) {
3201: ij[0] = 6*i; ij[1] = 6*i + 1; ij[2] = 6*i + 2; ij[3] = 6*i + 3; ij[4] = 6*i + 4; ij[5] = 6*i + 5;
3202: MatGetValues(A,6,ij,6,ij,diag);
3203: PetscKernel_A_gets_inverse_A_6(diag,shift,allowzeropivot,&zeropivotdetected);
3204: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3205: PetscKernel_A_gets_transpose_A_6(diag);
3206: diag += 36;
3207: }
3208: break;
3209: case 7:
3210: for (i=0; i<mbs; i++) {
3211: ij[0] = 7*i; ij[1] = 7*i + 1; ij[2] = 7*i + 2; ij[3] = 7*i + 3; ij[4] = 7*i + 4; ij[5] = 7*i + 5; ij[5] = 7*i + 6;
3212: MatGetValues(A,7,ij,7,ij,diag);
3213: PetscKernel_A_gets_inverse_A_7(diag,shift,allowzeropivot,&zeropivotdetected);
3214: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3215: PetscKernel_A_gets_transpose_A_7(diag);
3216: diag += 49;
3217: }
3218: break;
3219: default:
3220: PetscMalloc3(bs,&v_work,bs,&v_pivots,bs,&IJ);
3221: for (i=0; i<mbs; i++) {
3222: for (j=0; j<bs; j++) {
3223: IJ[j] = bs*i + j;
3224: }
3225: MatGetValues(A,bs,IJ,bs,IJ,diag);
3226: PetscKernel_A_gets_inverse_A(bs,diag,v_pivots,v_work,allowzeropivot,&zeropivotdetected);
3227: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3228: PetscKernel_A_gets_transpose_A_N(diag,bs);
3229: diag += bs2;
3230: }
3231: PetscFree3(v_work,v_pivots,IJ);
3232: }
3233: a->ibdiagvalid = PETSC_TRUE;
3234: return(0);
3235: }
3237: static PetscErrorCode MatSetRandom_SeqAIJ(Mat x,PetscRandom rctx)
3238: {
3240: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)x->data;
3241: PetscScalar a;
3242: PetscInt m,n,i,j,col;
3245: if (!x->assembled) {
3246: MatGetSize(x,&m,&n);
3247: for (i=0; i<m; i++) {
3248: for (j=0; j<aij->imax[i]; j++) {
3249: PetscRandomGetValue(rctx,&a);
3250: col = (PetscInt)(n*PetscRealPart(a));
3251: MatSetValues(x,1,&i,1,&col,&a,ADD_VALUES);
3252: }
3253: }
3254: } else {
3255: for (i=0; i<aij->nz; i++) {PetscRandomGetValue(rctx,aij->a+i);}
3256: }
3257: MatAssemblyBegin(x,MAT_FINAL_ASSEMBLY);
3258: MatAssemblyEnd(x,MAT_FINAL_ASSEMBLY);
3259: return(0);
3260: }
3262: /* Like MatSetRandom_SeqAIJ, but do not set values on columns in range of [low, high) */
3263: PetscErrorCode MatSetRandomSkipColumnRange_SeqAIJ_Private(Mat x,PetscInt low,PetscInt high,PetscRandom rctx)
3264: {
3266: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)x->data;
3267: PetscScalar a;
3268: PetscInt m,n,i,j,col,nskip;
3271: nskip = high - low;
3272: MatGetSize(x,&m,&n);
3273: n -= nskip; /* shrink number of columns where nonzeros can be set */
3274: for (i=0; i<m; i++) {
3275: for (j=0; j<aij->imax[i]; j++) {
3276: PetscRandomGetValue(rctx,&a);
3277: col = (PetscInt)(n*PetscRealPart(a));
3278: if (col >= low) col += nskip; /* shift col rightward to skip the hole */
3279: MatSetValues(x,1,&i,1,&col,&a,ADD_VALUES);
3280: }
3281: }
3282: MatAssemblyBegin(x,MAT_FINAL_ASSEMBLY);
3283: MatAssemblyEnd(x,MAT_FINAL_ASSEMBLY);
3284: return(0);
3285: }
3288: /* -------------------------------------------------------------------*/
3289: static struct _MatOps MatOps_Values = { MatSetValues_SeqAIJ,
3290: MatGetRow_SeqAIJ,
3291: MatRestoreRow_SeqAIJ,
3292: MatMult_SeqAIJ,
3293: /* 4*/ MatMultAdd_SeqAIJ,
3294: MatMultTranspose_SeqAIJ,
3295: MatMultTransposeAdd_SeqAIJ,
3296: 0,
3297: 0,
3298: 0,
3299: /* 10*/ 0,
3300: MatLUFactor_SeqAIJ,
3301: 0,
3302: MatSOR_SeqAIJ,
3303: MatTranspose_SeqAIJ,
3304: /*1 5*/ MatGetInfo_SeqAIJ,
3305: MatEqual_SeqAIJ,
3306: MatGetDiagonal_SeqAIJ,
3307: MatDiagonalScale_SeqAIJ,
3308: MatNorm_SeqAIJ,
3309: /* 20*/ 0,
3310: MatAssemblyEnd_SeqAIJ,
3311: MatSetOption_SeqAIJ,
3312: MatZeroEntries_SeqAIJ,
3313: /* 24*/ MatZeroRows_SeqAIJ,
3314: 0,
3315: 0,
3316: 0,
3317: 0,
3318: /* 29*/ MatSetUp_SeqAIJ,
3319: 0,
3320: 0,
3321: 0,
3322: 0,
3323: /* 34*/ MatDuplicate_SeqAIJ,
3324: 0,
3325: 0,
3326: MatILUFactor_SeqAIJ,
3327: 0,
3328: /* 39*/ MatAXPY_SeqAIJ,
3329: MatCreateSubMatrices_SeqAIJ,
3330: MatIncreaseOverlap_SeqAIJ,
3331: MatGetValues_SeqAIJ,
3332: MatCopy_SeqAIJ,
3333: /* 44*/ MatGetRowMax_SeqAIJ,
3334: MatScale_SeqAIJ,
3335: MatShift_SeqAIJ,
3336: MatDiagonalSet_SeqAIJ,
3337: MatZeroRowsColumns_SeqAIJ,
3338: /* 49*/ MatSetRandom_SeqAIJ,
3339: MatGetRowIJ_SeqAIJ,
3340: MatRestoreRowIJ_SeqAIJ,
3341: MatGetColumnIJ_SeqAIJ,
3342: MatRestoreColumnIJ_SeqAIJ,
3343: /* 54*/ MatFDColoringCreate_SeqXAIJ,
3344: 0,
3345: 0,
3346: MatPermute_SeqAIJ,
3347: 0,
3348: /* 59*/ 0,
3349: MatDestroy_SeqAIJ,
3350: MatView_SeqAIJ,
3351: 0,
3352: MatMatMatMult_SeqAIJ_SeqAIJ_SeqAIJ,
3353: /* 64*/ MatMatMatMultSymbolic_SeqAIJ_SeqAIJ_SeqAIJ,
3354: MatMatMatMultNumeric_SeqAIJ_SeqAIJ_SeqAIJ,
3355: 0,
3356: 0,
3357: 0,
3358: /* 69*/ MatGetRowMaxAbs_SeqAIJ,
3359: MatGetRowMinAbs_SeqAIJ,
3360: 0,
3361: 0,
3362: 0,
3363: /* 74*/ 0,
3364: MatFDColoringApply_AIJ,
3365: 0,
3366: 0,
3367: 0,
3368: /* 79*/ MatFindZeroDiagonals_SeqAIJ,
3369: 0,
3370: 0,
3371: 0,
3372: MatLoad_SeqAIJ,
3373: /* 84*/ MatIsSymmetric_SeqAIJ,
3374: MatIsHermitian_SeqAIJ,
3375: 0,
3376: 0,
3377: 0,
3378: /* 89*/ MatMatMult_SeqAIJ_SeqAIJ,
3379: MatMatMultSymbolic_SeqAIJ_SeqAIJ,
3380: MatMatMultNumeric_SeqAIJ_SeqAIJ,
3381: MatPtAP_SeqAIJ_SeqAIJ,
3382: MatPtAPSymbolic_SeqAIJ_SeqAIJ_SparseAxpy,
3383: /* 94*/ MatPtAPNumeric_SeqAIJ_SeqAIJ_SparseAxpy,
3384: MatMatTransposeMult_SeqAIJ_SeqAIJ,
3385: MatMatTransposeMultSymbolic_SeqAIJ_SeqAIJ,
3386: MatMatTransposeMultNumeric_SeqAIJ_SeqAIJ,
3387: 0,
3388: /* 99*/ 0,
3389: 0,
3390: 0,
3391: MatConjugate_SeqAIJ,
3392: 0,
3393: /*104*/ MatSetValuesRow_SeqAIJ,
3394: MatRealPart_SeqAIJ,
3395: MatImaginaryPart_SeqAIJ,
3396: 0,
3397: 0,
3398: /*109*/ MatMatSolve_SeqAIJ,
3399: 0,
3400: MatGetRowMin_SeqAIJ,
3401: 0,
3402: MatMissingDiagonal_SeqAIJ,
3403: /*114*/ 0,
3404: 0,
3405: 0,
3406: 0,
3407: 0,
3408: /*119*/ 0,
3409: 0,
3410: 0,
3411: 0,
3412: MatGetMultiProcBlock_SeqAIJ,
3413: /*124*/ MatFindNonzeroRows_SeqAIJ,
3414: MatGetColumnNorms_SeqAIJ,
3415: MatInvertBlockDiagonal_SeqAIJ,
3416: MatInvertVariableBlockDiagonal_SeqAIJ,
3417: 0,
3418: /*129*/ 0,
3419: MatTransposeMatMult_SeqAIJ_SeqAIJ,
3420: MatTransposeMatMultSymbolic_SeqAIJ_SeqAIJ,
3421: MatTransposeMatMultNumeric_SeqAIJ_SeqAIJ,
3422: MatTransposeColoringCreate_SeqAIJ,
3423: /*134*/ MatTransColoringApplySpToDen_SeqAIJ,
3424: MatTransColoringApplyDenToSp_SeqAIJ,
3425: MatRARt_SeqAIJ_SeqAIJ,
3426: MatRARtSymbolic_SeqAIJ_SeqAIJ,
3427: MatRARtNumeric_SeqAIJ_SeqAIJ,
3428: /*139*/0,
3429: 0,
3430: 0,
3431: MatFDColoringSetUp_SeqXAIJ,
3432: MatFindOffBlockDiagonalEntries_SeqAIJ,
3433: /*144*/MatCreateMPIMatConcatenateSeqMat_SeqAIJ,
3434: MatDestroySubMatrices_SeqAIJ
3435: };
3437: PetscErrorCode MatSeqAIJSetColumnIndices_SeqAIJ(Mat mat,PetscInt *indices)
3438: {
3439: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data;
3440: PetscInt i,nz,n;
3443: nz = aij->maxnz;
3444: n = mat->rmap->n;
3445: for (i=0; i<nz; i++) {
3446: aij->j[i] = indices[i];
3447: }
3448: aij->nz = nz;
3449: for (i=0; i<n; i++) {
3450: aij->ilen[i] = aij->imax[i];
3451: }
3452: return(0);
3453: }
3455: /*
3456: * When a sparse matrix has many zero columns, we should compact them out to save the space
3457: * This happens in MatPtAPSymbolic_MPIAIJ_MPIAIJ_scalable()
3458: * */
3459: PetscErrorCode MatSeqAIJCompactOutExtraColumns_SeqAIJ(Mat mat, ISLocalToGlobalMapping *mapping)
3460: {
3461: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data;
3462: PetscTable gid1_lid1;
3463: PetscTablePosition tpos;
3464: PetscInt gid,lid,i,j,ncols,ec;
3465: PetscInt *garray;
3466: PetscErrorCode ierr;
3471: /* use a table */
3472: PetscTableCreate(mat->rmap->n,mat->cmap->N+1,&gid1_lid1);
3473: ec = 0;
3474: for (i=0; i<mat->rmap->n; i++) {
3475: ncols = aij->i[i+1] - aij->i[i];
3476: for (j=0; j<ncols; j++) {
3477: PetscInt data,gid1 = aij->j[aij->i[i] + j] + 1;
3478: PetscTableFind(gid1_lid1,gid1,&data);
3479: if (!data) {
3480: /* one based table */
3481: PetscTableAdd(gid1_lid1,gid1,++ec,INSERT_VALUES);
3482: }
3483: }
3484: }
3485: /* form array of columns we need */
3486: PetscMalloc1(ec+1,&garray);
3487: PetscTableGetHeadPosition(gid1_lid1,&tpos);
3488: while (tpos) {
3489: PetscTableGetNext(gid1_lid1,&tpos,&gid,&lid);
3490: gid--;
3491: lid--;
3492: garray[lid] = gid;
3493: }
3494: PetscSortInt(ec,garray); /* sort, and rebuild */
3495: PetscTableRemoveAll(gid1_lid1);
3496: for (i=0; i<ec; i++) {
3497: PetscTableAdd(gid1_lid1,garray[i]+1,i+1,INSERT_VALUES);
3498: }
3499: /* compact out the extra columns in B */
3500: for (i=0; i<mat->rmap->n; i++) {
3501: ncols = aij->i[i+1] - aij->i[i];
3502: for (j=0; j<ncols; j++) {
3503: PetscInt gid1 = aij->j[aij->i[i] + j] + 1;
3504: PetscTableFind(gid1_lid1,gid1,&lid);
3505: lid--;
3506: aij->j[aij->i[i] + j] = lid;
3507: }
3508: }
3509: mat->cmap->n = mat->cmap->N = ec;
3510: mat->cmap->bs = 1;
3512: PetscTableDestroy(&gid1_lid1);
3513: PetscLayoutSetUp((mat->cmap));
3514: ISLocalToGlobalMappingCreate(PETSC_COMM_SELF,mat->cmap->bs,mat->cmap->n,garray,PETSC_OWN_POINTER,mapping);
3515: ISLocalToGlobalMappingSetType(*mapping,ISLOCALTOGLOBALMAPPINGHASH);
3516: return(0);
3517: }
3519: /*@
3520: MatSeqAIJSetColumnIndices - Set the column indices for all the rows
3521: in the matrix.
3523: Input Parameters:
3524: + mat - the SeqAIJ matrix
3525: - indices - the column indices
3527: Level: advanced
3529: Notes:
3530: This can be called if you have precomputed the nonzero structure of the
3531: matrix and want to provide it to the matrix object to improve the performance
3532: of the MatSetValues() operation.
3534: You MUST have set the correct numbers of nonzeros per row in the call to
3535: MatCreateSeqAIJ(), and the columns indices MUST be sorted.
3537: MUST be called before any calls to MatSetValues();
3539: The indices should start with zero, not one.
3541: @*/
3542: PetscErrorCode MatSeqAIJSetColumnIndices(Mat mat,PetscInt *indices)
3543: {
3549: PetscUseMethod(mat,"MatSeqAIJSetColumnIndices_C",(Mat,PetscInt*),(mat,indices));
3550: return(0);
3551: }
3553: /* ----------------------------------------------------------------------------------------*/
3555: PetscErrorCode MatStoreValues_SeqAIJ(Mat mat)
3556: {
3557: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data;
3559: size_t nz = aij->i[mat->rmap->n];
3562: if (!aij->nonew) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
3564: /* allocate space for values if not already there */
3565: if (!aij->saved_values) {
3566: PetscMalloc1(nz+1,&aij->saved_values);
3567: PetscLogObjectMemory((PetscObject)mat,(nz+1)*sizeof(PetscScalar));
3568: }
3570: /* copy values over */
3571: PetscArraycpy(aij->saved_values,aij->a,nz);
3572: return(0);
3573: }
3575: /*@
3576: MatStoreValues - Stashes a copy of the matrix values; this allows, for
3577: example, reuse of the linear part of a Jacobian, while recomputing the
3578: nonlinear portion.
3580: Collect on Mat
3582: Input Parameters:
3583: . mat - the matrix (currently only AIJ matrices support this option)
3585: Level: advanced
3587: Common Usage, with SNESSolve():
3588: $ Create Jacobian matrix
3589: $ Set linear terms into matrix
3590: $ Apply boundary conditions to matrix, at this time matrix must have
3591: $ final nonzero structure (i.e. setting the nonlinear terms and applying
3592: $ boundary conditions again will not change the nonzero structure
3593: $ MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);
3594: $ MatStoreValues(mat);
3595: $ Call SNESSetJacobian() with matrix
3596: $ In your Jacobian routine
3597: $ MatRetrieveValues(mat);
3598: $ Set nonlinear terms in matrix
3600: Common Usage without SNESSolve(), i.e. when you handle nonlinear solve yourself:
3601: $ // build linear portion of Jacobian
3602: $ MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);
3603: $ MatStoreValues(mat);
3604: $ loop over nonlinear iterations
3605: $ MatRetrieveValues(mat);
3606: $ // call MatSetValues(mat,...) to set nonliner portion of Jacobian
3607: $ // call MatAssemblyBegin/End() on matrix
3608: $ Solve linear system with Jacobian
3609: $ endloop
3611: Notes:
3612: Matrix must already be assemblied before calling this routine
3613: Must set the matrix option MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE); before
3614: calling this routine.
3616: When this is called multiple times it overwrites the previous set of stored values
3617: and does not allocated additional space.
3619: .seealso: MatRetrieveValues()
3621: @*/
3622: PetscErrorCode MatStoreValues(Mat mat)
3623: {
3628: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3629: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3630: PetscUseMethod(mat,"MatStoreValues_C",(Mat),(mat));
3631: return(0);
3632: }
3634: PetscErrorCode MatRetrieveValues_SeqAIJ(Mat mat)
3635: {
3636: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data;
3638: PetscInt nz = aij->i[mat->rmap->n];
3641: if (!aij->nonew) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
3642: if (!aij->saved_values) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatStoreValues(A);first");
3643: /* copy values over */
3644: PetscArraycpy(aij->a,aij->saved_values,nz);
3645: return(0);
3646: }
3648: /*@
3649: MatRetrieveValues - Retrieves the copy of the matrix values; this allows, for
3650: example, reuse of the linear part of a Jacobian, while recomputing the
3651: nonlinear portion.
3653: Collect on Mat
3655: Input Parameters:
3656: . mat - the matrix (currently only AIJ matrices support this option)
3658: Level: advanced
3660: .seealso: MatStoreValues()
3662: @*/
3663: PetscErrorCode MatRetrieveValues(Mat mat)
3664: {
3669: if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
3670: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3671: PetscUseMethod(mat,"MatRetrieveValues_C",(Mat),(mat));
3672: return(0);
3673: }
3676: /* --------------------------------------------------------------------------------*/
3677: /*@C
3678: MatCreateSeqAIJ - Creates a sparse matrix in AIJ (compressed row) format
3679: (the default parallel PETSc format). For good matrix assembly performance
3680: the user should preallocate the matrix storage by setting the parameter nz
3681: (or the array nnz). By setting these parameters accurately, performance
3682: during matrix assembly can be increased by more than a factor of 50.
3684: Collective
3686: Input Parameters:
3687: + comm - MPI communicator, set to PETSC_COMM_SELF
3688: . m - number of rows
3689: . n - number of columns
3690: . nz - number of nonzeros per row (same for all rows)
3691: - nnz - array containing the number of nonzeros in the various rows
3692: (possibly different for each row) or NULL
3694: Output Parameter:
3695: . A - the matrix
3697: It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(),
3698: MatXXXXSetPreallocation() paradigm instead of this routine directly.
3699: [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation]
3701: Notes:
3702: If nnz is given then nz is ignored
3704: The AIJ format (also called the Yale sparse matrix format or
3705: compressed row storage), is fully compatible with standard Fortran 77
3706: storage. That is, the stored row and column indices can begin at
3707: either one (as in Fortran) or zero. See the users' manual for details.
3709: Specify the preallocated storage with either nz or nnz (not both).
3710: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
3711: allocation. For large problems you MUST preallocate memory or you
3712: will get TERRIBLE performance, see the users' manual chapter on matrices.
3714: By default, this format uses inodes (identical nodes) when possible, to
3715: improve numerical efficiency of matrix-vector products and solves. We
3716: search for consecutive rows with the same nonzero structure, thereby
3717: reusing matrix information to achieve increased efficiency.
3719: Options Database Keys:
3720: + -mat_no_inode - Do not use inodes
3721: - -mat_inode_limit <limit> - Sets inode limit (max limit=5)
3723: Level: intermediate
3725: .seealso: MatCreate(), MatCreateAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays()
3727: @*/
3728: PetscErrorCode MatCreateSeqAIJ(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt nz,const PetscInt nnz[],Mat *A)
3729: {
3733: MatCreate(comm,A);
3734: MatSetSizes(*A,m,n,m,n);
3735: MatSetType(*A,MATSEQAIJ);
3736: MatSeqAIJSetPreallocation_SeqAIJ(*A,nz,nnz);
3737: return(0);
3738: }
3740: /*@C
3741: MatSeqAIJSetPreallocation - For good matrix assembly performance
3742: the user should preallocate the matrix storage by setting the parameter nz
3743: (or the array nnz). By setting these parameters accurately, performance
3744: during matrix assembly can be increased by more than a factor of 50.
3746: Collective
3748: Input Parameters:
3749: + B - The matrix
3750: . nz - number of nonzeros per row (same for all rows)
3751: - nnz - array containing the number of nonzeros in the various rows
3752: (possibly different for each row) or NULL
3754: Notes:
3755: If nnz is given then nz is ignored
3757: The AIJ format (also called the Yale sparse matrix format or
3758: compressed row storage), is fully compatible with standard Fortran 77
3759: storage. That is, the stored row and column indices can begin at
3760: either one (as in Fortran) or zero. See the users' manual for details.
3762: Specify the preallocated storage with either nz or nnz (not both).
3763: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
3764: allocation. For large problems you MUST preallocate memory or you
3765: will get TERRIBLE performance, see the users' manual chapter on matrices.
3767: You can call MatGetInfo() to get information on how effective the preallocation was;
3768: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
3769: You can also run with the option -info and look for messages with the string
3770: malloc in them to see if additional memory allocation was needed.
3772: Developers: Use nz of MAT_SKIP_ALLOCATION to not allocate any space for the matrix
3773: entries or columns indices
3775: By default, this format uses inodes (identical nodes) when possible, to
3776: improve numerical efficiency of matrix-vector products and solves. We
3777: search for consecutive rows with the same nonzero structure, thereby
3778: reusing matrix information to achieve increased efficiency.
3780: Options Database Keys:
3781: + -mat_no_inode - Do not use inodes
3782: - -mat_inode_limit <limit> - Sets inode limit (max limit=5)
3784: Level: intermediate
3786: .seealso: MatCreate(), MatCreateAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays(), MatGetInfo()
3788: @*/
3789: PetscErrorCode MatSeqAIJSetPreallocation(Mat B,PetscInt nz,const PetscInt nnz[])
3790: {
3796: PetscTryMethod(B,"MatSeqAIJSetPreallocation_C",(Mat,PetscInt,const PetscInt[]),(B,nz,nnz));
3797: return(0);
3798: }
3800: PetscErrorCode MatSeqAIJSetPreallocation_SeqAIJ(Mat B,PetscInt nz,const PetscInt *nnz)
3801: {
3802: Mat_SeqAIJ *b;
3803: PetscBool skipallocation = PETSC_FALSE,realalloc = PETSC_FALSE;
3805: PetscInt i;
3808: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
3809: if (nz == MAT_SKIP_ALLOCATION) {
3810: skipallocation = PETSC_TRUE;
3811: nz = 0;
3812: }
3813: PetscLayoutSetUp(B->rmap);
3814: PetscLayoutSetUp(B->cmap);
3816: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
3817: if (nz < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nz cannot be less than 0: value %D",nz);
3818: #if defined(PETSC_USE_DEBUG)
3819: if (nnz) {
3820: for (i=0; i<B->rmap->n; i++) {
3821: if (nnz[i] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be less than 0: local row %D value %D",i,nnz[i]);
3822: if (nnz[i] > B->cmap->n) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be greater than row length: local row %D value %d rowlength %D",i,nnz[i],B->cmap->n);
3823: }
3824: }
3825: #endif
3827: B->preallocated = PETSC_TRUE;
3829: b = (Mat_SeqAIJ*)B->data;
3831: if (!skipallocation) {
3832: if (!b->imax) {
3833: PetscMalloc1(B->rmap->n,&b->imax);
3834: PetscLogObjectMemory((PetscObject)B,B->rmap->n*sizeof(PetscInt));
3835: }
3836: if (!b->ilen) {
3837: /* b->ilen will count nonzeros in each row so far. */
3838: PetscCalloc1(B->rmap->n,&b->ilen);
3839: PetscLogObjectMemory((PetscObject)B,B->rmap->n*sizeof(PetscInt));
3840: } else {
3841: PetscMemzero(b->ilen,B->rmap->n*sizeof(PetscInt));
3842: }
3843: if (!b->ipre) {
3844: PetscMalloc1(B->rmap->n,&b->ipre);
3845: PetscLogObjectMemory((PetscObject)B,B->rmap->n*sizeof(PetscInt));
3846: }
3847: if (!nnz) {
3848: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 10;
3849: else if (nz < 0) nz = 1;
3850: nz = PetscMin(nz,B->cmap->n);
3851: for (i=0; i<B->rmap->n; i++) b->imax[i] = nz;
3852: nz = nz*B->rmap->n;
3853: } else {
3854: nz = 0;
3855: for (i=0; i<B->rmap->n; i++) {b->imax[i] = nnz[i]; nz += nnz[i];}
3856: }
3858: /* allocate the matrix space */
3859: /* FIXME: should B's old memory be unlogged? */
3860: MatSeqXAIJFreeAIJ(B,&b->a,&b->j,&b->i);
3861: if (B->structure_only) {
3862: PetscMalloc1(nz,&b->j);
3863: PetscMalloc1(B->rmap->n+1,&b->i);
3864: PetscLogObjectMemory((PetscObject)B,(B->rmap->n+1)*sizeof(PetscInt)+nz*sizeof(PetscInt));
3865: } else {
3866: PetscMalloc3(nz,&b->a,nz,&b->j,B->rmap->n+1,&b->i);
3867: PetscLogObjectMemory((PetscObject)B,(B->rmap->n+1)*sizeof(PetscInt)+nz*(sizeof(PetscScalar)+sizeof(PetscInt)));
3868: }
3869: b->i[0] = 0;
3870: for (i=1; i<B->rmap->n+1; i++) {
3871: b->i[i] = b->i[i-1] + b->imax[i-1];
3872: }
3873: if (B->structure_only) {
3874: b->singlemalloc = PETSC_FALSE;
3875: b->free_a = PETSC_FALSE;
3876: } else {
3877: b->singlemalloc = PETSC_TRUE;
3878: b->free_a = PETSC_TRUE;
3879: }
3880: b->free_ij = PETSC_TRUE;
3881: } else {
3882: b->free_a = PETSC_FALSE;
3883: b->free_ij = PETSC_FALSE;
3884: }
3886: if (b->ipre && nnz != b->ipre && b->imax) {
3887: /* reserve user-requested sparsity */
3888: PetscArraycpy(b->ipre,b->imax,B->rmap->n);
3889: }
3892: b->nz = 0;
3893: b->maxnz = nz;
3894: B->info.nz_unneeded = (double)b->maxnz;
3895: if (realalloc) {
3896: MatSetOption(B,MAT_NEW_NONZERO_ALLOCATION_ERR,PETSC_TRUE);
3897: }
3898: B->was_assembled = PETSC_FALSE;
3899: B->assembled = PETSC_FALSE;
3900: return(0);
3901: }
3904: PetscErrorCode MatResetPreallocation_SeqAIJ(Mat A)
3905: {
3906: Mat_SeqAIJ *a;
3907: PetscInt i;
3913: /* Check local size. If zero, then return */
3914: if (!A->rmap->n) return(0);
3916: a = (Mat_SeqAIJ*)A->data;
3917: /* if no saved info, we error out */
3918: if (!a->ipre) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_NULL,"No saved preallocation info \n");
3920: if (!a->i || !a->j || !a->a || !a->imax || !a->ilen) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_NULL,"Memory info is incomplete, and can not reset preallocation \n");
3922: PetscArraycpy(a->imax,a->ipre,A->rmap->n);
3923: PetscArrayzero(a->ilen,A->rmap->n);
3924: a->i[0] = 0;
3925: for (i=1; i<A->rmap->n+1; i++) {
3926: a->i[i] = a->i[i-1] + a->imax[i-1];
3927: }
3928: A->preallocated = PETSC_TRUE;
3929: a->nz = 0;
3930: a->maxnz = a->i[A->rmap->n];
3931: A->info.nz_unneeded = (double)a->maxnz;
3932: A->was_assembled = PETSC_FALSE;
3933: A->assembled = PETSC_FALSE;
3934: return(0);
3935: }
3937: /*@
3938: MatSeqAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in AIJ format.
3940: Input Parameters:
3941: + B - the matrix
3942: . i - the indices into j for the start of each row (starts with zero)
3943: . j - the column indices for each row (starts with zero) these must be sorted for each row
3944: - v - optional values in the matrix
3946: Level: developer
3948: The i,j,v values are COPIED with this routine; to avoid the copy use MatCreateSeqAIJWithArrays()
3950: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatSeqAIJSetPreallocation(), MatCreateSeqAIJ(), MATSEQAIJ
3951: @*/
3952: PetscErrorCode MatSeqAIJSetPreallocationCSR(Mat B,const PetscInt i[],const PetscInt j[],const PetscScalar v[])
3953: {
3959: PetscTryMethod(B,"MatSeqAIJSetPreallocationCSR_C",(Mat,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,i,j,v));
3960: return(0);
3961: }
3963: PetscErrorCode MatSeqAIJSetPreallocationCSR_SeqAIJ(Mat B,const PetscInt Ii[],const PetscInt J[],const PetscScalar v[])
3964: {
3965: PetscInt i;
3966: PetscInt m,n;
3967: PetscInt nz;
3968: PetscInt *nnz, nz_max = 0;
3972: if (Ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE, "Ii[0] must be 0 it is %D", Ii[0]);
3974: PetscLayoutSetUp(B->rmap);
3975: PetscLayoutSetUp(B->cmap);
3977: MatGetSize(B, &m, &n);
3978: PetscMalloc1(m+1, &nnz);
3979: for (i = 0; i < m; i++) {
3980: nz = Ii[i+1]- Ii[i];
3981: nz_max = PetscMax(nz_max, nz);
3982: if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE, "Local row %D has a negative number of columns %D", i, nnz);
3983: nnz[i] = nz;
3984: }
3985: MatSeqAIJSetPreallocation(B, 0, nnz);
3986: PetscFree(nnz);
3988: for (i = 0; i < m; i++) {
3989: MatSetValues_SeqAIJ(B, 1, &i, Ii[i+1] - Ii[i], J+Ii[i], v ? v + Ii[i] : NULL, INSERT_VALUES);
3990: }
3992: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
3993: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
3995: MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
3996: return(0);
3997: }
3999: #include <../src/mat/impls/dense/seq/dense.h>
4000: #include <petsc/private/kernels/petscaxpy.h>
4002: /*
4003: Computes (B'*A')' since computing B*A directly is untenable
4005: n p p
4006: ( ) ( ) ( )
4007: m ( A ) * n ( B ) = m ( C )
4008: ( ) ( ) ( )
4010: */
4011: PetscErrorCode MatMatMultNumeric_SeqDense_SeqAIJ(Mat A,Mat B,Mat C)
4012: {
4013: PetscErrorCode ierr;
4014: Mat_SeqDense *sub_a = (Mat_SeqDense*)A->data;
4015: Mat_SeqAIJ *sub_b = (Mat_SeqAIJ*)B->data;
4016: Mat_SeqDense *sub_c = (Mat_SeqDense*)C->data;
4017: PetscInt i,n,m,q,p;
4018: const PetscInt *ii,*idx;
4019: const PetscScalar *b,*a,*a_q;
4020: PetscScalar *c,*c_q;
4023: m = A->rmap->n;
4024: n = A->cmap->n;
4025: p = B->cmap->n;
4026: a = sub_a->v;
4027: b = sub_b->a;
4028: c = sub_c->v;
4029: PetscArrayzero(c,m*p);
4031: ii = sub_b->i;
4032: idx = sub_b->j;
4033: for (i=0; i<n; i++) {
4034: q = ii[i+1] - ii[i];
4035: while (q-->0) {
4036: c_q = c + m*(*idx);
4037: a_q = a + m*i;
4038: PetscKernelAXPY(c_q,*b,a_q,m);
4039: idx++;
4040: b++;
4041: }
4042: }
4043: return(0);
4044: }
4046: PetscErrorCode MatMatMultSymbolic_SeqDense_SeqAIJ(Mat A,Mat B,PetscReal fill,Mat *C)
4047: {
4049: PetscInt m=A->rmap->n,n=B->cmap->n;
4050: Mat Cmat;
4053: if (A->cmap->n != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"A->cmap->n %D != B->rmap->n %D\n",A->cmap->n,B->rmap->n);
4054: MatCreate(PetscObjectComm((PetscObject)A),&Cmat);
4055: MatSetSizes(Cmat,m,n,m,n);
4056: MatSetBlockSizesFromMats(Cmat,A,B);
4057: MatSetType(Cmat,MATSEQDENSE);
4058: MatSeqDenseSetPreallocation(Cmat,NULL);
4060: Cmat->ops->matmultnumeric = MatMatMultNumeric_SeqDense_SeqAIJ;
4062: *C = Cmat;
4063: return(0);
4064: }
4066: /* ----------------------------------------------------------------*/
4067: PETSC_INTERN PetscErrorCode MatMatMult_SeqDense_SeqAIJ(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C)
4068: {
4072: if (scall == MAT_INITIAL_MATRIX) {
4073: PetscLogEventBegin(MAT_MatMultSymbolic,A,B,0,0);
4074: MatMatMultSymbolic_SeqDense_SeqAIJ(A,B,fill,C);
4075: PetscLogEventEnd(MAT_MatMultSymbolic,A,B,0,0);
4076: }
4077: PetscLogEventBegin(MAT_MatMultNumeric,A,B,0,0);
4078: MatMatMultNumeric_SeqDense_SeqAIJ(A,B,*C);
4079: PetscLogEventEnd(MAT_MatMultNumeric,A,B,0,0);
4080: return(0);
4081: }
4084: /*MC
4085: MATSEQAIJ - MATSEQAIJ = "seqaij" - A matrix type to be used for sequential sparse matrices,
4086: based on compressed sparse row format.
4088: Options Database Keys:
4089: . -mat_type seqaij - sets the matrix type to "seqaij" during a call to MatSetFromOptions()
4091: Level: beginner
4093: .seealso: MatCreateSeqAIJ(), MatSetFromOptions(), MatSetType(), MatCreate(), MatType
4094: M*/
4096: /*MC
4097: MATAIJ - MATAIJ = "aij" - A matrix type to be used for sparse matrices.
4099: This matrix type is identical to MATSEQAIJ when constructed with a single process communicator,
4100: and MATMPIAIJ otherwise. As a result, for single process communicators,
4101: MatSeqAIJSetPreallocation is supported, and similarly MatMPIAIJSetPreallocation is supported
4102: for communicators controlling multiple processes. It is recommended that you call both of
4103: the above preallocation routines for simplicity.
4105: Options Database Keys:
4106: . -mat_type aij - sets the matrix type to "aij" during a call to MatSetFromOptions()
4108: Developer Notes:
4109: Subclasses include MATAIJCUSPARSE, MATAIJPERM, MATAIJSELL, MATAIJMKL, MATAIJCRL, and also automatically switches over to use inodes when
4110: enough exist.
4112: Level: beginner
4114: .seealso: MatCreateAIJ(), MatCreateSeqAIJ(), MATSEQAIJ,MATMPIAIJ
4115: M*/
4117: /*MC
4118: MATAIJCRL - MATAIJCRL = "aijcrl" - A matrix type to be used for sparse matrices.
4120: This matrix type is identical to MATSEQAIJCRL when constructed with a single process communicator,
4121: and MATMPIAIJCRL otherwise. As a result, for single process communicators,
4122: MatSeqAIJSetPreallocation() is supported, and similarly MatMPIAIJSetPreallocation() is supported
4123: for communicators controlling multiple processes. It is recommended that you call both of
4124: the above preallocation routines for simplicity.
4126: Options Database Keys:
4127: . -mat_type aijcrl - sets the matrix type to "aijcrl" during a call to MatSetFromOptions()
4129: Level: beginner
4131: .seealso: MatCreateMPIAIJCRL,MATSEQAIJCRL,MATMPIAIJCRL, MATSEQAIJCRL, MATMPIAIJCRL
4132: M*/
4134: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJCRL(Mat,MatType,MatReuse,Mat*);
4135: #if defined(PETSC_HAVE_ELEMENTAL)
4136: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_Elemental(Mat,MatType,MatReuse,Mat*);
4137: #endif
4138: #if defined(PETSC_HAVE_HYPRE)
4139: PETSC_INTERN PetscErrorCode MatConvert_AIJ_HYPRE(Mat A,MatType,MatReuse,Mat*);
4140: PETSC_INTERN PetscErrorCode MatMatMatMult_Transpose_AIJ_AIJ(Mat,Mat,Mat,MatReuse,PetscReal,Mat*);
4141: #endif
4142: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqDense(Mat,MatType,MatReuse,Mat*);
4144: PETSC_EXTERN PetscErrorCode MatConvert_SeqAIJ_SeqSELL(Mat,MatType,MatReuse,Mat*);
4145: PETSC_INTERN PetscErrorCode MatConvert_XAIJ_IS(Mat,MatType,MatReuse,Mat*);
4146: PETSC_INTERN PetscErrorCode MatPtAP_IS_XAIJ(Mat,Mat,MatReuse,PetscReal,Mat*);
4148: /*@C
4149: MatSeqAIJGetArray - gives access to the array where the data for a MATSEQAIJ matrix is stored
4151: Not Collective
4153: Input Parameter:
4154: . mat - a MATSEQAIJ matrix
4156: Output Parameter:
4157: . array - pointer to the data
4159: Level: intermediate
4161: .seealso: MatSeqAIJRestoreArray(), MatSeqAIJGetArrayF90()
4162: @*/
4163: PetscErrorCode MatSeqAIJGetArray(Mat A,PetscScalar **array)
4164: {
4168: PetscUseMethod(A,"MatSeqAIJGetArray_C",(Mat,PetscScalar**),(A,array));
4169: return(0);
4170: }
4172: /*@C
4173: MatSeqAIJGetMaxRowNonzeros - returns the maximum number of nonzeros in any row
4175: Not Collective
4177: Input Parameter:
4178: . mat - a MATSEQAIJ matrix
4180: Output Parameter:
4181: . nz - the maximum number of nonzeros in any row
4183: Level: intermediate
4185: .seealso: MatSeqAIJRestoreArray(), MatSeqAIJGetArrayF90()
4186: @*/
4187: PetscErrorCode MatSeqAIJGetMaxRowNonzeros(Mat A,PetscInt *nz)
4188: {
4189: Mat_SeqAIJ *aij = (Mat_SeqAIJ*)A->data;
4192: *nz = aij->rmax;
4193: return(0);
4194: }
4196: /*@C
4197: MatSeqAIJRestoreArray - returns access to the array where the data for a MATSEQAIJ matrix is stored obtained by MatSeqAIJGetArray()
4199: Not Collective
4201: Input Parameters:
4202: + mat - a MATSEQAIJ matrix
4203: - array - pointer to the data
4205: Level: intermediate
4207: .seealso: MatSeqAIJGetArray(), MatSeqAIJRestoreArrayF90()
4208: @*/
4209: PetscErrorCode MatSeqAIJRestoreArray(Mat A,PetscScalar **array)
4210: {
4214: PetscUseMethod(A,"MatSeqAIJRestoreArray_C",(Mat,PetscScalar**),(A,array));
4215: return(0);
4216: }
4218: #if defined(PETSC_HAVE_CUDA)
4219: PETSC_EXTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJCUSPARSE(Mat);
4220: #endif
4222: PETSC_EXTERN PetscErrorCode MatCreate_SeqAIJ(Mat B)
4223: {
4224: Mat_SeqAIJ *b;
4226: PetscMPIInt size;
4229: MPI_Comm_size(PetscObjectComm((PetscObject)B),&size);
4230: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Comm must be of size 1");
4232: PetscNewLog(B,&b);
4234: B->data = (void*)b;
4236: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
4237: if (B->sortedfull) B->ops->setvalues = MatSetValues_SeqAIJ_SortedFull;
4239: b->row = 0;
4240: b->col = 0;
4241: b->icol = 0;
4242: b->reallocs = 0;
4243: b->ignorezeroentries = PETSC_FALSE;
4244: b->roworiented = PETSC_TRUE;
4245: b->nonew = 0;
4246: b->diag = 0;
4247: b->solve_work = 0;
4248: B->spptr = 0;
4249: b->saved_values = 0;
4250: b->idiag = 0;
4251: b->mdiag = 0;
4252: b->ssor_work = 0;
4253: b->omega = 1.0;
4254: b->fshift = 0.0;
4255: b->idiagvalid = PETSC_FALSE;
4256: b->ibdiagvalid = PETSC_FALSE;
4257: b->keepnonzeropattern = PETSC_FALSE;
4259: PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);
4260: PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJGetArray_C",MatSeqAIJGetArray_SeqAIJ);
4261: PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJRestoreArray_C",MatSeqAIJRestoreArray_SeqAIJ);
4263: #if defined(PETSC_HAVE_MATLAB_ENGINE)
4264: PetscObjectComposeFunction((PetscObject)B,"PetscMatlabEnginePut_C",MatlabEnginePut_SeqAIJ);
4265: PetscObjectComposeFunction((PetscObject)B,"PetscMatlabEngineGet_C",MatlabEngineGet_SeqAIJ);
4266: #endif
4268: PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetColumnIndices_C",MatSeqAIJSetColumnIndices_SeqAIJ);
4269: PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_SeqAIJ);
4270: PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_SeqAIJ);
4271: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqsbaij_C",MatConvert_SeqAIJ_SeqSBAIJ);
4272: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqbaij_C",MatConvert_SeqAIJ_SeqBAIJ);
4273: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijperm_C",MatConvert_SeqAIJ_SeqAIJPERM);
4274: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijsell_C",MatConvert_SeqAIJ_SeqAIJSELL);
4275: #if defined(PETSC_HAVE_MKL_SPARSE)
4276: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijmkl_C",MatConvert_SeqAIJ_SeqAIJMKL);
4277: #endif
4278: #if defined(PETSC_HAVE_CUDA)
4279: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijcusparse_C",MatConvert_SeqAIJ_SeqAIJCUSPARSE);
4280: #endif
4281: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijcrl_C",MatConvert_SeqAIJ_SeqAIJCRL);
4282: #if defined(PETSC_HAVE_ELEMENTAL)
4283: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_elemental_C",MatConvert_SeqAIJ_Elemental);
4284: #endif
4285: #if defined(PETSC_HAVE_HYPRE)
4286: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_hypre_C",MatConvert_AIJ_HYPRE);
4287: PetscObjectComposeFunction((PetscObject)B,"MatMatMatMult_transpose_seqaij_seqaij_C",MatMatMatMult_Transpose_AIJ_AIJ);
4288: #endif
4289: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqdense_C",MatConvert_SeqAIJ_SeqDense);
4290: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqsell_C",MatConvert_SeqAIJ_SeqSELL);
4291: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_is_C",MatConvert_XAIJ_IS);
4292: PetscObjectComposeFunction((PetscObject)B,"MatIsTranspose_C",MatIsTranspose_SeqAIJ);
4293: PetscObjectComposeFunction((PetscObject)B,"MatIsHermitianTranspose_C",MatIsTranspose_SeqAIJ);
4294: PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetPreallocation_C",MatSeqAIJSetPreallocation_SeqAIJ);
4295: PetscObjectComposeFunction((PetscObject)B,"MatResetPreallocation_C",MatResetPreallocation_SeqAIJ);
4296: PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetPreallocationCSR_C",MatSeqAIJSetPreallocationCSR_SeqAIJ);
4297: PetscObjectComposeFunction((PetscObject)B,"MatReorderForNonzeroDiagonal_C",MatReorderForNonzeroDiagonal_SeqAIJ);
4298: PetscObjectComposeFunction((PetscObject)B,"MatMatMult_seqdense_seqaij_C",MatMatMult_SeqDense_SeqAIJ);
4299: PetscObjectComposeFunction((PetscObject)B,"MatMatMultSymbolic_seqdense_seqaij_C",MatMatMultSymbolic_SeqDense_SeqAIJ);
4300: PetscObjectComposeFunction((PetscObject)B,"MatMatMultNumeric_seqdense_seqaij_C",MatMatMultNumeric_SeqDense_SeqAIJ);
4301: PetscObjectComposeFunction((PetscObject)B,"MatPtAP_is_seqaij_C",MatPtAP_IS_XAIJ);
4302: MatCreate_SeqAIJ_Inode(B);
4303: PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);
4304: MatSeqAIJSetTypeFromOptions(B); /* this allows changing the matrix subtype to say MATSEQAIJPERM */
4305: return(0);
4306: }
4308: /*
4309: Given a matrix generated with MatGetFactor() duplicates all the information in A into B
4310: */
4311: PetscErrorCode MatDuplicateNoCreate_SeqAIJ(Mat C,Mat A,MatDuplicateOption cpvalues,PetscBool mallocmatspace)
4312: {
4313: Mat_SeqAIJ *c,*a = (Mat_SeqAIJ*)A->data;
4315: PetscInt m = A->rmap->n,i;
4318: c = (Mat_SeqAIJ*)C->data;
4320: C->factortype = A->factortype;
4321: c->row = 0;
4322: c->col = 0;
4323: c->icol = 0;
4324: c->reallocs = 0;
4326: C->assembled = PETSC_TRUE;
4328: PetscLayoutReference(A->rmap,&C->rmap);
4329: PetscLayoutReference(A->cmap,&C->cmap);
4331: PetscMalloc1(m,&c->imax);
4332: PetscMemcpy(c->imax,a->imax,m*sizeof(PetscInt));
4333: PetscMalloc1(m,&c->ilen);
4334: PetscMemcpy(c->ilen,a->ilen,m*sizeof(PetscInt));
4335: PetscLogObjectMemory((PetscObject)C, 2*m*sizeof(PetscInt));
4337: /* allocate the matrix space */
4338: if (mallocmatspace) {
4339: PetscMalloc3(a->i[m],&c->a,a->i[m],&c->j,m+1,&c->i);
4340: PetscLogObjectMemory((PetscObject)C, a->i[m]*(sizeof(PetscScalar)+sizeof(PetscInt))+(m+1)*sizeof(PetscInt));
4342: c->singlemalloc = PETSC_TRUE;
4344: PetscArraycpy(c->i,a->i,m+1);
4345: if (m > 0) {
4346: PetscArraycpy(c->j,a->j,a->i[m]);
4347: if (cpvalues == MAT_COPY_VALUES) {
4348: PetscArraycpy(c->a,a->a,a->i[m]);
4349: } else {
4350: PetscArrayzero(c->a,a->i[m]);
4351: }
4352: }
4353: }
4355: c->ignorezeroentries = a->ignorezeroentries;
4356: c->roworiented = a->roworiented;
4357: c->nonew = a->nonew;
4358: if (a->diag) {
4359: PetscMalloc1(m+1,&c->diag);
4360: PetscMemcpy(c->diag,a->diag,m*sizeof(PetscInt));
4361: PetscLogObjectMemory((PetscObject)C,(m+1)*sizeof(PetscInt));
4362: } else c->diag = NULL;
4364: c->solve_work = 0;
4365: c->saved_values = 0;
4366: c->idiag = 0;
4367: c->ssor_work = 0;
4368: c->keepnonzeropattern = a->keepnonzeropattern;
4369: c->free_a = PETSC_TRUE;
4370: c->free_ij = PETSC_TRUE;
4372: c->rmax = a->rmax;
4373: c->nz = a->nz;
4374: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
4375: C->preallocated = PETSC_TRUE;
4377: c->compressedrow.use = a->compressedrow.use;
4378: c->compressedrow.nrows = a->compressedrow.nrows;
4379: if (a->compressedrow.use) {
4380: i = a->compressedrow.nrows;
4381: PetscMalloc2(i+1,&c->compressedrow.i,i,&c->compressedrow.rindex);
4382: PetscArraycpy(c->compressedrow.i,a->compressedrow.i,i+1);
4383: PetscArraycpy(c->compressedrow.rindex,a->compressedrow.rindex,i);
4384: } else {
4385: c->compressedrow.use = PETSC_FALSE;
4386: c->compressedrow.i = NULL;
4387: c->compressedrow.rindex = NULL;
4388: }
4389: c->nonzerorowcnt = a->nonzerorowcnt;
4390: C->nonzerostate = A->nonzerostate;
4392: MatDuplicate_SeqAIJ_Inode(A,cpvalues,&C);
4393: PetscFunctionListDuplicate(((PetscObject)A)->qlist,&((PetscObject)C)->qlist);
4394: return(0);
4395: }
4397: PetscErrorCode MatDuplicate_SeqAIJ(Mat A,MatDuplicateOption cpvalues,Mat *B)
4398: {
4402: MatCreate(PetscObjectComm((PetscObject)A),B);
4403: MatSetSizes(*B,A->rmap->n,A->cmap->n,A->rmap->n,A->cmap->n);
4404: if (!(A->rmap->n % A->rmap->bs) && !(A->cmap->n % A->cmap->bs)) {
4405: MatSetBlockSizesFromMats(*B,A,A);
4406: }
4407: MatSetType(*B,((PetscObject)A)->type_name);
4408: MatDuplicateNoCreate_SeqAIJ(*B,A,cpvalues,PETSC_TRUE);
4409: return(0);
4410: }
4412: PetscErrorCode MatLoad_SeqAIJ(Mat newMat, PetscViewer viewer)
4413: {
4414: PetscBool isbinary, ishdf5;
4420: /* force binary viewer to load .info file if it has not yet done so */
4421: PetscViewerSetUp(viewer);
4422: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);
4423: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERHDF5, &ishdf5);
4424: if (isbinary) {
4425: MatLoad_SeqAIJ_Binary(newMat,viewer);
4426: } else if (ishdf5) {
4427: #if defined(PETSC_HAVE_HDF5)
4428: MatLoad_AIJ_HDF5(newMat,viewer);
4429: #else
4430: SETERRQ(PetscObjectComm((PetscObject)newMat),PETSC_ERR_SUP,"HDF5 not supported in this build.\nPlease reconfigure using --download-hdf5");
4431: #endif
4432: } else {
4433: SETERRQ2(PetscObjectComm((PetscObject)newMat),PETSC_ERR_SUP,"Viewer type %s not yet supported for reading %s matrices",((PetscObject)viewer)->type_name,((PetscObject)newMat)->type_name);
4434: }
4435: return(0);
4436: }
4438: PetscErrorCode MatLoad_SeqAIJ_Binary(Mat newMat, PetscViewer viewer)
4439: {
4440: Mat_SeqAIJ *a;
4442: PetscInt i,sum,nz,header[4],*rowlengths = 0,M,N,rows,cols;
4443: int fd;
4444: PetscMPIInt size;
4445: MPI_Comm comm;
4446: PetscInt bs = newMat->rmap->bs;
4449: PetscObjectGetComm((PetscObject)viewer,&comm);
4450: MPI_Comm_size(comm,&size);
4451: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"view must have one processor");
4453: PetscOptionsBegin(comm,NULL,"Options for loading SEQAIJ matrix","Mat");
4454: PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,NULL);
4455: PetscOptionsEnd();
4456: if (bs < 0) bs = 1;
4457: MatSetBlockSize(newMat,bs);
4459: PetscViewerBinaryGetDescriptor(viewer,&fd);
4460: PetscBinaryRead(fd,header,4,NULL,PETSC_INT);
4461: if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not matrix object in file");
4462: M = header[1]; N = header[2]; nz = header[3];
4464: if (nz < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format on disk,cannot load as SeqAIJ");
4466: /* read in row lengths */
4467: PetscMalloc1(M,&rowlengths);
4468: PetscBinaryRead(fd,rowlengths,M,NULL,PETSC_INT);
4470: /* check if sum of rowlengths is same as nz */
4471: for (i=0,sum=0; i< M; i++) sum +=rowlengths[i];
4472: if (sum != nz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_READ,"Inconsistant matrix data in file. no-nonzeros = %dD, sum-row-lengths = %D\n",nz,sum);
4474: /* set global size if not set already*/
4475: if (newMat->rmap->n < 0 && newMat->rmap->N < 0 && newMat->cmap->n < 0 && newMat->cmap->N < 0) {
4476: MatSetSizes(newMat,PETSC_DECIDE,PETSC_DECIDE,M,N);
4477: } else {
4478: /* if sizes and type are already set, check if the matrix global sizes are correct */
4479: MatGetSize(newMat,&rows,&cols);
4480: if (rows < 0 && cols < 0) { /* user might provide local size instead of global size */
4481: MatGetLocalSize(newMat,&rows,&cols);
4482: }
4483: if (M != rows || N != cols) SETERRQ4(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED, "Matrix in file of different length (%D, %D) than the input matrix (%D, %D)",M,N,rows,cols);
4484: }
4485: MatSeqAIJSetPreallocation_SeqAIJ(newMat,0,rowlengths);
4486: a = (Mat_SeqAIJ*)newMat->data;
4488: PetscBinaryRead(fd,a->j,nz,NULL,PETSC_INT);
4490: /* read in nonzero values */
4491: PetscBinaryRead(fd,a->a,nz,NULL,PETSC_SCALAR);
4493: /* set matrix "i" values */
4494: a->i[0] = 0;
4495: for (i=1; i<= M; i++) {
4496: a->i[i] = a->i[i-1] + rowlengths[i-1];
4497: a->ilen[i-1] = rowlengths[i-1];
4498: }
4499: PetscFree(rowlengths);
4501: MatAssemblyBegin(newMat,MAT_FINAL_ASSEMBLY);
4502: MatAssemblyEnd(newMat,MAT_FINAL_ASSEMBLY);
4503: return(0);
4504: }
4506: PetscErrorCode MatEqual_SeqAIJ(Mat A,Mat B,PetscBool * flg)
4507: {
4508: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data,*b = (Mat_SeqAIJ*)B->data;
4510: #if defined(PETSC_USE_COMPLEX)
4511: PetscInt k;
4512: #endif
4515: /* If the matrix dimensions are not equal,or no of nonzeros */
4516: if ((A->rmap->n != B->rmap->n) || (A->cmap->n != B->cmap->n) ||(a->nz != b->nz)) {
4517: *flg = PETSC_FALSE;
4518: return(0);
4519: }
4521: /* if the a->i are the same */
4522: PetscArraycmp(a->i,b->i,A->rmap->n+1,flg);
4523: if (!*flg) return(0);
4525: /* if a->j are the same */
4526: PetscArraycmp(a->j,b->j,a->nz,flg);
4527: if (!*flg) return(0);
4529: /* if a->a are the same */
4530: #if defined(PETSC_USE_COMPLEX)
4531: for (k=0; k<a->nz; k++) {
4532: if (PetscRealPart(a->a[k]) != PetscRealPart(b->a[k]) || PetscImaginaryPart(a->a[k]) != PetscImaginaryPart(b->a[k])) {
4533: *flg = PETSC_FALSE;
4534: return(0);
4535: }
4536: }
4537: #else
4538: PetscArraycmp(a->a,b->a,a->nz,flg);
4539: #endif
4540: return(0);
4541: }
4543: /*@
4544: MatCreateSeqAIJWithArrays - Creates an sequential AIJ matrix using matrix elements (in CSR format)
4545: provided by the user.
4547: Collective
4549: Input Parameters:
4550: + comm - must be an MPI communicator of size 1
4551: . m - number of rows
4552: . n - number of columns
4553: . i - row indices; that is i[0] = 0, i[row] = i[row-1] + number of elements in that row of the matrix
4554: . j - column indices
4555: - a - matrix values
4557: Output Parameter:
4558: . mat - the matrix
4560: Level: intermediate
4562: Notes:
4563: The i, j, and a arrays are not copied by this routine, the user must free these arrays
4564: once the matrix is destroyed and not before
4566: You cannot set new nonzero locations into this matrix, that will generate an error.
4568: The i and j indices are 0 based
4570: The format which is used for the sparse matrix input, is equivalent to a
4571: row-major ordering.. i.e for the following matrix, the input data expected is
4572: as shown
4574: $ 1 0 0
4575: $ 2 0 3
4576: $ 4 5 6
4577: $
4578: $ i = {0,1,3,6} [size = nrow+1 = 3+1]
4579: $ j = {0,0,2,0,1,2} [size = 6]; values must be sorted for each row
4580: $ v = {1,2,3,4,5,6} [size = 6]
4583: .seealso: MatCreate(), MatCreateAIJ(), MatCreateSeqAIJ(), MatCreateMPIAIJWithArrays(), MatMPIAIJSetPreallocationCSR()
4585: @*/
4586: PetscErrorCode MatCreateSeqAIJWithArrays(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt i[],PetscInt j[],PetscScalar a[],Mat *mat)
4587: {
4589: PetscInt ii;
4590: Mat_SeqAIJ *aij;
4591: #if defined(PETSC_USE_DEBUG)
4592: PetscInt jj;
4593: #endif
4596: if (m > 0 && i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0");
4597: MatCreate(comm,mat);
4598: MatSetSizes(*mat,m,n,m,n);
4599: /* MatSetBlockSizes(*mat,,); */
4600: MatSetType(*mat,MATSEQAIJ);
4601: MatSeqAIJSetPreallocation_SeqAIJ(*mat,MAT_SKIP_ALLOCATION,0);
4602: aij = (Mat_SeqAIJ*)(*mat)->data;
4603: PetscMalloc1(m,&aij->imax);
4604: PetscMalloc1(m,&aij->ilen);
4606: aij->i = i;
4607: aij->j = j;
4608: aij->a = a;
4609: aij->singlemalloc = PETSC_FALSE;
4610: aij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
4611: aij->free_a = PETSC_FALSE;
4612: aij->free_ij = PETSC_FALSE;
4614: for (ii=0; ii<m; ii++) {
4615: aij->ilen[ii] = aij->imax[ii] = i[ii+1] - i[ii];
4616: #if defined(PETSC_USE_DEBUG)
4617: if (i[ii+1] - i[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row length in i (row indices) row = %D length = %D",ii,i[ii+1] - i[ii]);
4618: for (jj=i[ii]+1; jj<i[ii+1]; jj++) {
4619: if (j[jj] < j[jj-1]) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column entry number %D (actual column %D) in row %D is not sorted",jj-i[ii],j[jj],ii);
4620: if (j[jj] == j[jj-1]) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column entry number %D (actual column %D) in row %D is identical to previous entry",jj-i[ii],j[jj],ii);
4621: }
4622: #endif
4623: }
4624: #if defined(PETSC_USE_DEBUG)
4625: for (ii=0; ii<aij->i[m]; ii++) {
4626: if (j[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column index at location = %D index = %D",ii,j[ii]);
4627: if (j[ii] > n - 1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column index to large at location = %D index = %D",ii,j[ii]);
4628: }
4629: #endif
4631: MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);
4632: MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);
4633: return(0);
4634: }
4635: /*@C
4636: MatCreateSeqAIJFromTriple - Creates an sequential AIJ matrix using matrix elements (in COO format)
4637: provided by the user.
4639: Collective
4641: Input Parameters:
4642: + comm - must be an MPI communicator of size 1
4643: . m - number of rows
4644: . n - number of columns
4645: . i - row indices
4646: . j - column indices
4647: . a - matrix values
4648: . nz - number of nonzeros
4649: - idx - 0 or 1 based
4651: Output Parameter:
4652: . mat - the matrix
4654: Level: intermediate
4656: Notes:
4657: The i and j indices are 0 based
4659: The format which is used for the sparse matrix input, is equivalent to a
4660: row-major ordering.. i.e for the following matrix, the input data expected is
4661: as shown:
4663: 1 0 0
4664: 2 0 3
4665: 4 5 6
4667: i = {0,1,1,2,2,2}
4668: j = {0,0,2,0,1,2}
4669: v = {1,2,3,4,5,6}
4672: .seealso: MatCreate(), MatCreateAIJ(), MatCreateSeqAIJ(), MatCreateSeqAIJWithArrays(), MatMPIAIJSetPreallocationCSR()
4674: @*/
4675: PetscErrorCode MatCreateSeqAIJFromTriple(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt i[],PetscInt j[],PetscScalar a[],Mat *mat,PetscInt nz,PetscBool idx)
4676: {
4678: PetscInt ii, *nnz, one = 1,row,col;
4682: PetscCalloc1(m,&nnz);
4683: for (ii = 0; ii < nz; ii++) {
4684: nnz[i[ii] - !!idx] += 1;
4685: }
4686: MatCreate(comm,mat);
4687: MatSetSizes(*mat,m,n,m,n);
4688: MatSetType(*mat,MATSEQAIJ);
4689: MatSeqAIJSetPreallocation_SeqAIJ(*mat,0,nnz);
4690: for (ii = 0; ii < nz; ii++) {
4691: if (idx) {
4692: row = i[ii] - 1;
4693: col = j[ii] - 1;
4694: } else {
4695: row = i[ii];
4696: col = j[ii];
4697: }
4698: MatSetValues(*mat,one,&row,one,&col,&a[ii],ADD_VALUES);
4699: }
4700: MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);
4701: MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);
4702: PetscFree(nnz);
4703: return(0);
4704: }
4706: PetscErrorCode MatSeqAIJInvalidateDiagonal(Mat A)
4707: {
4708: Mat_SeqAIJ *a=(Mat_SeqAIJ*)A->data;
4712: a->idiagvalid = PETSC_FALSE;
4713: a->ibdiagvalid = PETSC_FALSE;
4715: MatSeqAIJInvalidateDiagonal_Inode(A);
4716: return(0);
4717: }
4719: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqAIJ(MPI_Comm comm,Mat inmat,PetscInt n,MatReuse scall,Mat *outmat)
4720: {
4722: PetscMPIInt size;
4725: MPI_Comm_size(comm,&size);
4726: if (size == 1) {
4727: if (scall == MAT_INITIAL_MATRIX) {
4728: MatDuplicate(inmat,MAT_COPY_VALUES,outmat);
4729: } else {
4730: MatCopy(inmat,*outmat,SAME_NONZERO_PATTERN);
4731: }
4732: } else {
4733: MatCreateMPIMatConcatenateSeqMat_MPIAIJ(comm,inmat,n,scall,outmat);
4734: }
4735: return(0);
4736: }
4738: /*
4739: Permute A into C's *local* index space using rowemb,colemb.
4740: The embedding are supposed to be injections and the above implies that the range of rowemb is a subset
4741: of [0,m), colemb is in [0,n).
4742: If pattern == DIFFERENT_NONZERO_PATTERN, C is preallocated according to A.
4743: */
4744: PetscErrorCode MatSetSeqMat_SeqAIJ(Mat C,IS rowemb,IS colemb,MatStructure pattern,Mat B)
4745: {
4746: /* If making this function public, change the error returned in this function away from _PLIB. */
4748: Mat_SeqAIJ *Baij;
4749: PetscBool seqaij;
4750: PetscInt m,n,*nz,i,j,count;
4751: PetscScalar v;
4752: const PetscInt *rowindices,*colindices;
4755: if (!B) return(0);
4756: /* Check to make sure the target matrix (and embeddings) are compatible with C and each other. */
4757: PetscObjectBaseTypeCompare((PetscObject)B,MATSEQAIJ,&seqaij);
4758: if (!seqaij) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Input matrix is of wrong type");
4759: if (rowemb) {
4760: ISGetLocalSize(rowemb,&m);
4761: if (m != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Row IS of size %D is incompatible with matrix row size %D",m,B->rmap->n);
4762: } else {
4763: if (C->rmap->n != B->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Input matrix is row-incompatible with the target matrix");
4764: }
4765: if (colemb) {
4766: ISGetLocalSize(colemb,&n);
4767: if (n != B->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Diag col IS of size %D is incompatible with input matrix col size %D",n,B->cmap->n);
4768: } else {
4769: if (C->cmap->n != B->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Input matrix is col-incompatible with the target matrix");
4770: }
4772: Baij = (Mat_SeqAIJ*)(B->data);
4773: if (pattern == DIFFERENT_NONZERO_PATTERN) {
4774: PetscMalloc1(B->rmap->n,&nz);
4775: for (i=0; i<B->rmap->n; i++) {
4776: nz[i] = Baij->i[i+1] - Baij->i[i];
4777: }
4778: MatSeqAIJSetPreallocation(C,0,nz);
4779: PetscFree(nz);
4780: }
4781: if (pattern == SUBSET_NONZERO_PATTERN) {
4782: MatZeroEntries(C);
4783: }
4784: count = 0;
4785: rowindices = NULL;
4786: colindices = NULL;
4787: if (rowemb) {
4788: ISGetIndices(rowemb,&rowindices);
4789: }
4790: if (colemb) {
4791: ISGetIndices(colemb,&colindices);
4792: }
4793: for (i=0; i<B->rmap->n; i++) {
4794: PetscInt row;
4795: row = i;
4796: if (rowindices) row = rowindices[i];
4797: for (j=Baij->i[i]; j<Baij->i[i+1]; j++) {
4798: PetscInt col;
4799: col = Baij->j[count];
4800: if (colindices) col = colindices[col];
4801: v = Baij->a[count];
4802: MatSetValues(C,1,&row,1,&col,&v,INSERT_VALUES);
4803: ++count;
4804: }
4805: }
4806: /* FIXME: set C's nonzerostate correctly. */
4807: /* Assembly for C is necessary. */
4808: C->preallocated = PETSC_TRUE;
4809: C->assembled = PETSC_TRUE;
4810: C->was_assembled = PETSC_FALSE;
4811: return(0);
4812: }
4814: PetscFunctionList MatSeqAIJList = NULL;
4816: /*@C
4817: MatSeqAIJSetType - Converts a MATSEQAIJ matrix to a subtype
4819: Collective on Mat
4821: Input Parameters:
4822: + mat - the matrix object
4823: - matype - matrix type
4825: Options Database Key:
4826: . -mat_seqai_type <method> - for example seqaijcrl
4829: Level: intermediate
4831: .seealso: PCSetType(), VecSetType(), MatCreate(), MatType, Mat
4832: @*/
4833: PetscErrorCode MatSeqAIJSetType(Mat mat, MatType matype)
4834: {
4835: PetscErrorCode ierr,(*r)(Mat,MatType,MatReuse,Mat*);
4836: PetscBool sametype;
4840: PetscObjectTypeCompare((PetscObject)mat,matype,&sametype);
4841: if (sametype) return(0);
4843: PetscFunctionListFind(MatSeqAIJList,matype,&r);
4844: if (!r) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_UNKNOWN_TYPE,"Unknown Mat type given: %s",matype);
4845: (*r)(mat,matype,MAT_INPLACE_MATRIX,&mat);
4846: return(0);
4847: }
4850: /*@C
4851: MatSeqAIJRegister - - Adds a new sub-matrix type for sequential AIJ matrices
4853: Not Collective
4855: Input Parameters:
4856: + name - name of a new user-defined matrix type, for example MATSEQAIJCRL
4857: - function - routine to convert to subtype
4859: Notes:
4860: MatSeqAIJRegister() may be called multiple times to add several user-defined solvers.
4863: Then, your matrix can be chosen with the procedural interface at runtime via the option
4864: $ -mat_seqaij_type my_mat
4866: Level: advanced
4868: .seealso: MatSeqAIJRegisterAll()
4871: Level: advanced
4872: @*/
4873: PetscErrorCode MatSeqAIJRegister(const char sname[],PetscErrorCode (*function)(Mat,MatType,MatReuse,Mat *))
4874: {
4878: MatInitializePackage();
4879: PetscFunctionListAdd(&MatSeqAIJList,sname,function);
4880: return(0);
4881: }
4883: PetscBool MatSeqAIJRegisterAllCalled = PETSC_FALSE;
4885: /*@C
4886: MatSeqAIJRegisterAll - Registers all of the matrix subtypes of SeqAIJ
4888: Not Collective
4890: Level: advanced
4892: Developers Note: CUSP and CUSPARSE do not yet support the MatConvert_SeqAIJ..() paradigm and thus cannot be registered here
4894: .seealso: MatRegisterAll(), MatSeqAIJRegister()
4895: @*/
4896: PetscErrorCode MatSeqAIJRegisterAll(void)
4897: {
4901: if (MatSeqAIJRegisterAllCalled) return(0);
4902: MatSeqAIJRegisterAllCalled = PETSC_TRUE;
4904: MatSeqAIJRegister(MATSEQAIJCRL, MatConvert_SeqAIJ_SeqAIJCRL);
4905: MatSeqAIJRegister(MATSEQAIJPERM, MatConvert_SeqAIJ_SeqAIJPERM);
4906: MatSeqAIJRegister(MATSEQAIJSELL, MatConvert_SeqAIJ_SeqAIJSELL);
4907: #if defined(PETSC_HAVE_MKL_SPARSE)
4908: MatSeqAIJRegister(MATSEQAIJMKL, MatConvert_SeqAIJ_SeqAIJMKL);
4909: #endif
4910: #if defined(PETSC_HAVE_VIENNACL) && defined(PETSC_HAVE_VIENNACL_NO_CUDA)
4911: MatSeqAIJRegister(MATMPIAIJVIENNACL, MatConvert_SeqAIJ_SeqAIJViennaCL);
4912: #endif
4913: return(0);
4914: }
4916: /*
4917: Special version for direct calls from Fortran
4918: */
4919: #include <petsc/private/fortranimpl.h>
4920: #if defined(PETSC_HAVE_FORTRAN_CAPS)
4921: #define matsetvaluesseqaij_ MATSETVALUESSEQAIJ
4922: #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE)
4923: #define matsetvaluesseqaij_ matsetvaluesseqaij
4924: #endif
4926: /* Change these macros so can be used in void function */
4927: #undef CHKERRQ
4928: #define CHKERRQ(ierr) CHKERRABORT(PetscObjectComm((PetscObject)A),ierr)
4929: #undef SETERRQ2
4930: #define SETERRQ2(comm,ierr,b,c,d) CHKERRABORT(comm,ierr)
4931: #undef SETERRQ3
4932: #define SETERRQ3(comm,ierr,b,c,d,e) CHKERRABORT(comm,ierr)
4934: PETSC_EXTERN void PETSC_STDCALL matsetvaluesseqaij_(Mat *AA,PetscInt *mm,const PetscInt im[],PetscInt *nn,const PetscInt in[],const PetscScalar v[],InsertMode *isis, PetscErrorCode *_ierr)
4935: {
4936: Mat A = *AA;
4937: PetscInt m = *mm, n = *nn;
4938: InsertMode is = *isis;
4939: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
4940: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N;
4941: PetscInt *imax,*ai,*ailen;
4943: PetscInt *aj,nonew = a->nonew,lastcol = -1;
4944: MatScalar *ap,value,*aa;
4945: PetscBool ignorezeroentries = a->ignorezeroentries;
4946: PetscBool roworiented = a->roworiented;
4949: MatCheckPreallocated(A,1);
4950: imax = a->imax;
4951: ai = a->i;
4952: ailen = a->ilen;
4953: aj = a->j;
4954: aa = a->a;
4956: for (k=0; k<m; k++) { /* loop over added rows */
4957: row = im[k];
4958: if (row < 0) continue;
4959: #if defined(PETSC_USE_DEBUG)
4960: if (row >= A->rmap->n) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Row too large");
4961: #endif
4962: rp = aj + ai[row]; ap = aa + ai[row];
4963: rmax = imax[row]; nrow = ailen[row];
4964: low = 0;
4965: high = nrow;
4966: for (l=0; l<n; l++) { /* loop over added columns */
4967: if (in[l] < 0) continue;
4968: #if defined(PETSC_USE_DEBUG)
4969: if (in[l] >= A->cmap->n) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Column too large");
4970: #endif
4971: col = in[l];
4972: if (roworiented) value = v[l + k*n];
4973: else value = v[k + l*m];
4975: if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue;
4977: if (col <= lastcol) low = 0;
4978: else high = nrow;
4979: lastcol = col;
4980: while (high-low > 5) {
4981: t = (low+high)/2;
4982: if (rp[t] > col) high = t;
4983: else low = t;
4984: }
4985: for (i=low; i<high; i++) {
4986: if (rp[i] > col) break;
4987: if (rp[i] == col) {
4988: if (is == ADD_VALUES) ap[i] += value;
4989: else ap[i] = value;
4990: goto noinsert;
4991: }
4992: }
4993: if (value == 0.0 && ignorezeroentries) goto noinsert;
4994: if (nonew == 1) goto noinsert;
4995: if (nonew == -1) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero in the matrix");
4996: MatSeqXAIJReallocateAIJ(A,A->rmap->n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
4997: N = nrow++ - 1; a->nz++; high++;
4998: /* shift up all the later entries in this row */
4999: for (ii=N; ii>=i; ii--) {
5000: rp[ii+1] = rp[ii];
5001: ap[ii+1] = ap[ii];
5002: }
5003: rp[i] = col;
5004: ap[i] = value;
5005: A->nonzerostate++;
5006: noinsert:;
5007: low = i + 1;
5008: }
5009: ailen[row] = nrow;
5010: }
5011: PetscFunctionReturnVoid();
5012: }