Actual source code: sbaij.c
petsc-3.7.1 2016-05-15
2: /*
3: Defines the basic matrix operations for the SBAIJ (compressed row)
4: matrix storage format.
5: */
6: #include <../src/mat/impls/baij/seq/baij.h> /*I "petscmat.h" I*/
7: #include <../src/mat/impls/sbaij/seq/sbaij.h>
8: #include <petscblaslapack.h>
10: #include <../src/mat/impls/sbaij/seq/relax.h>
11: #define USESHORT
12: #include <../src/mat/impls/sbaij/seq/relax.h>
14: extern PetscErrorCode MatSeqSBAIJSetNumericFactorization_inplace(Mat,PetscBool);
15: #if defined(PETSC_HAVE_ELEMENTAL)
16: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_Elemental(Mat,MatType,MatReuse,Mat*);
17: #endif
19: /*
20: Checks for missing diagonals
21: */
24: PetscErrorCode MatMissingDiagonal_SeqSBAIJ(Mat A,PetscBool *missing,PetscInt *dd)
25: {
26: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
28: PetscInt *diag,*ii = a->i,i;
31: MatMarkDiagonal_SeqSBAIJ(A);
32: *missing = PETSC_FALSE;
33: if (A->rmap->n > 0 && !ii) {
34: *missing = PETSC_TRUE;
35: if (dd) *dd = 0;
36: PetscInfo(A,"Matrix has no entries therefore is missing diagonal\n");
37: } else {
38: diag = a->diag;
39: for (i=0; i<a->mbs; i++) {
40: if (diag[i] >= ii[i+1]) {
41: *missing = PETSC_TRUE;
42: if (dd) *dd = i;
43: break;
44: }
45: }
46: }
47: return(0);
48: }
52: PetscErrorCode MatMarkDiagonal_SeqSBAIJ(Mat A)
53: {
54: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
56: PetscInt i,j;
59: if (!a->diag) {
60: PetscMalloc1(a->mbs,&a->diag);
61: PetscLogObjectMemory((PetscObject)A,a->mbs*sizeof(PetscInt));
62: a->free_diag = PETSC_TRUE;
63: }
64: for (i=0; i<a->mbs; i++) {
65: a->diag[i] = a->i[i+1];
66: for (j=a->i[i]; j<a->i[i+1]; j++) {
67: if (a->j[j] == i) {
68: a->diag[i] = j;
69: break;
70: }
71: }
72: }
73: return(0);
74: }
78: static PetscErrorCode MatGetRowIJ_SeqSBAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool blockcompressed,PetscInt *nn,const PetscInt *inia[],const PetscInt *inja[],PetscBool *done)
79: {
80: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
81: PetscInt i,j,n = a->mbs,nz = a->i[n],bs = A->rmap->bs;
82: PetscInt **ia = (PetscInt**)inia,**ja = (PetscInt**)inja;
86: *nn = n;
87: if (!ia) return(0);
88: if (!blockcompressed) {
89: /* malloc & create the natural set of indices */
90: PetscMalloc2((n+1)*bs,ia,nz*bs,ja);
91: for (i=0; i<n+1; i++) {
92: for (j=0; j<bs; j++) {
93: (*ia)[i*bs+j] = a->i[i]*bs+j+oshift;
94: }
95: }
96: for (i=0; i<nz; i++) {
97: for (j=0; j<bs; j++) {
98: (*ja)[i*bs+j] = a->j[i]*bs+j+oshift;
99: }
100: }
101: } else { /* blockcompressed */
102: if (oshift == 1) {
103: /* temporarily add 1 to i and j indices */
104: for (i=0; i<nz; i++) a->j[i]++;
105: for (i=0; i<n+1; i++) a->i[i]++;
106: }
107: *ia = a->i; *ja = a->j;
108: }
109: return(0);
110: }
114: static PetscErrorCode MatRestoreRowIJ_SeqSBAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool blockcompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
115: {
116: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
117: PetscInt i,n = a->mbs,nz = a->i[n];
121: if (!ia) return(0);
123: if (!blockcompressed) {
124: PetscFree2(*ia,*ja);
125: } else if (oshift == 1) { /* blockcompressed */
126: for (i=0; i<nz; i++) a->j[i]--;
127: for (i=0; i<n+1; i++) a->i[i]--;
128: }
129: return(0);
130: }
134: PetscErrorCode MatDestroy_SeqSBAIJ(Mat A)
135: {
136: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
140: #if defined(PETSC_USE_LOG)
141: PetscLogObjectState((PetscObject)A,"Rows=%D, NZ=%D",A->rmap->N,a->nz);
142: #endif
143: MatSeqXAIJFreeAIJ(A,&a->a,&a->j,&a->i);
144: if (a->free_diag) {PetscFree(a->diag);}
145: ISDestroy(&a->row);
146: ISDestroy(&a->col);
147: ISDestroy(&a->icol);
148: PetscFree(a->idiag);
149: PetscFree(a->inode.size);
150: if (a->free_imax_ilen) {PetscFree2(a->imax,a->ilen);}
151: PetscFree(a->solve_work);
152: PetscFree(a->sor_work);
153: PetscFree(a->solves_work);
154: PetscFree(a->mult_work);
155: PetscFree(a->saved_values);
156: if (a->free_jshort) {PetscFree(a->jshort);}
157: PetscFree(a->inew);
158: MatDestroy(&a->parent);
159: PetscFree(A->data);
161: PetscObjectChangeTypeName((PetscObject)A,0);
162: PetscObjectComposeFunction((PetscObject)A,"MatStoreValues_C",NULL);
163: PetscObjectComposeFunction((PetscObject)A,"MatRetrieveValues_C",NULL);
164: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetColumnIndices_C",NULL);
165: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqaij_C",NULL);
166: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqbaij_C",NULL);
167: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetPreallocation_C",NULL);
168: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetPreallocationCSR_C",NULL);
169: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqsbstrm_C",NULL);
170: #if defined(PETSC_HAVE_ELEMENTAL)
171: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_elemental_C",NULL);
172: #endif
173: return(0);
174: }
178: PetscErrorCode MatSetOption_SeqSBAIJ(Mat A,MatOption op,PetscBool flg)
179: {
180: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
184: switch (op) {
185: case MAT_ROW_ORIENTED:
186: a->roworiented = flg;
187: break;
188: case MAT_KEEP_NONZERO_PATTERN:
189: a->keepnonzeropattern = flg;
190: break;
191: case MAT_NEW_NONZERO_LOCATIONS:
192: a->nonew = (flg ? 0 : 1);
193: break;
194: case MAT_NEW_NONZERO_LOCATION_ERR:
195: a->nonew = (flg ? -1 : 0);
196: break;
197: case MAT_NEW_NONZERO_ALLOCATION_ERR:
198: a->nonew = (flg ? -2 : 0);
199: break;
200: case MAT_UNUSED_NONZERO_LOCATION_ERR:
201: a->nounused = (flg ? -1 : 0);
202: break;
203: case MAT_NEW_DIAGONALS:
204: case MAT_IGNORE_OFF_PROC_ENTRIES:
205: case MAT_USE_HASH_TABLE:
206: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
207: break;
208: case MAT_HERMITIAN:
209: if (!A->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Must call MatAssemblyEnd() first");
210: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
211: A->ops->mult = MatMult_SeqSBAIJ_1_Hermitian_ushort;
212: } else if (A->cmap->bs == 1) {
213: A->ops->mult = MatMult_SeqSBAIJ_1_Hermitian;
214: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"No support for Hermitian with block size greater than 1");
215: break;
216: case MAT_SPD:
217: /* These options are handled directly by MatSetOption() */
218: break;
219: case MAT_SYMMETRIC:
220: case MAT_STRUCTURALLY_SYMMETRIC:
221: case MAT_SYMMETRY_ETERNAL:
222: /* These options are handled directly by MatSetOption() */
223: break;
224: case MAT_IGNORE_LOWER_TRIANGULAR:
225: a->ignore_ltriangular = flg;
226: break;
227: case MAT_ERROR_LOWER_TRIANGULAR:
228: a->ignore_ltriangular = flg;
229: break;
230: case MAT_GETROW_UPPERTRIANGULAR:
231: a->getrow_utriangular = flg;
232: break;
233: default:
234: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"unknown option %d",op);
235: }
236: return(0);
237: }
241: PetscErrorCode MatGetRow_SeqSBAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
242: {
243: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
247: if (A && !a->getrow_utriangular) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatGetRow is not supported for SBAIJ matrix format. Getting the upper triangular part of row, run with -mat_getrow_uppertriangular, call MatSetOption(mat,MAT_GETROW_UPPERTRIANGULAR,PETSC_TRUE) or MatGetRowUpperTriangular()");
249: /* Get the upper triangular part of the row */
250: MatGetRow_SeqBAIJ_private(A,row,nz,idx,v,a->i,a->j,a->a);
251: return(0);
252: }
256: PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
257: {
261: if (idx) {PetscFree(*idx);}
262: if (v) {PetscFree(*v);}
263: return(0);
264: }
268: PetscErrorCode MatGetRowUpperTriangular_SeqSBAIJ(Mat A)
269: {
270: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
273: a->getrow_utriangular = PETSC_TRUE;
274: return(0);
275: }
278: PetscErrorCode MatRestoreRowUpperTriangular_SeqSBAIJ(Mat A)
279: {
280: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
283: a->getrow_utriangular = PETSC_FALSE;
284: return(0);
285: }
289: PetscErrorCode MatTranspose_SeqSBAIJ(Mat A,MatReuse reuse,Mat *B)
290: {
294: if (reuse == MAT_INITIAL_MATRIX || *B != A) {
295: MatDuplicate(A,MAT_COPY_VALUES,B);
296: }
297: return(0);
298: }
302: PetscErrorCode MatView_SeqSBAIJ_ASCII(Mat A,PetscViewer viewer)
303: {
304: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
305: PetscErrorCode ierr;
306: PetscInt i,j,bs = A->rmap->bs,k,l,bs2=a->bs2;
307: PetscViewerFormat format;
308: PetscInt *diag;
311: PetscViewerGetFormat(viewer,&format);
312: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
313: PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);
314: } else if (format == PETSC_VIEWER_ASCII_MATLAB) {
315: Mat aij;
316: const char *matname;
318: if (A->factortype && bs>1) {
319: PetscPrintf(PETSC_COMM_SELF,"Warning: matrix is factored with bs>1. MatView() with PETSC_VIEWER_ASCII_MATLAB is not supported and ignored!\n");
320: return(0);
321: }
322: MatConvert(A,MATSEQAIJ,MAT_INITIAL_MATRIX,&aij);
323: PetscObjectGetName((PetscObject)A,&matname);
324: PetscObjectSetName((PetscObject)aij,matname);
325: MatView(aij,viewer);
326: MatDestroy(&aij);
327: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
328: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
329: for (i=0; i<a->mbs; i++) {
330: for (j=0; j<bs; j++) {
331: PetscViewerASCIIPrintf(viewer,"row %D:",i*bs+j);
332: for (k=a->i[i]; k<a->i[i+1]; k++) {
333: for (l=0; l<bs; l++) {
334: #if defined(PETSC_USE_COMPLEX)
335: if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) > 0.0 && PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
336: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k]+l,
337: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
338: } else if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) < 0.0 && PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
339: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k]+l,
340: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),-(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
341: } else if (PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
342: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)PetscRealPart(a->a[bs2*k + l*bs + j]));
343: }
344: #else
345: if (a->a[bs2*k + l*bs + j] != 0.0) {
346: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)a->a[bs2*k + l*bs + j]);
347: }
348: #endif
349: }
350: }
351: PetscViewerASCIIPrintf(viewer,"\n");
352: }
353: }
354: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
355: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
356: return(0);
357: } else {
358: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
359: if (A->factortype) { /* for factored matrix */
360: if (bs>1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"matrix is factored with bs>1. Not implemented yet");
362: diag=a->diag;
363: for (i=0; i<a->mbs; i++) { /* for row block i */
364: PetscViewerASCIIPrintf(viewer,"row %D:",i);
365: /* diagonal entry */
366: #if defined(PETSC_USE_COMPLEX)
367: if (PetscImaginaryPart(a->a[diag[i]]) > 0.0) {
368: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",a->j[diag[i]],(double)PetscRealPart(1.0/a->a[diag[i]]),(double)PetscImaginaryPart(1.0/a->a[diag[i]]));
369: } else if (PetscImaginaryPart(a->a[diag[i]]) < 0.0) {
370: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",a->j[diag[i]],(double)PetscRealPart(1.0/a->a[diag[i]]),-(double)PetscImaginaryPart(1.0/a->a[diag[i]]));
371: } else {
372: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[diag[i]],(double)PetscRealPart(1.0/a->a[diag[i]]));
373: }
374: #else
375: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[diag[i]],(double)(1.0/a->a[diag[i]]));
376: #endif
377: /* off-diagonal entries */
378: for (k=a->i[i]; k<a->i[i+1]-1; k++) {
379: #if defined(PETSC_USE_COMPLEX)
380: if (PetscImaginaryPart(a->a[k]) > 0.0) {
381: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k],(double)PetscRealPart(a->a[k]),(double)PetscImaginaryPart(a->a[k]));
382: } else if (PetscImaginaryPart(a->a[k]) < 0.0) {
383: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k],(double)PetscRealPart(a->a[k]),-(double)PetscImaginaryPart(a->a[k]));
384: } else {
385: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k],(double)PetscRealPart(a->a[k]));
386: }
387: #else
388: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[k],(double)a->a[k]);
389: #endif
390: }
391: PetscViewerASCIIPrintf(viewer,"\n");
392: }
394: } else { /* for non-factored matrix */
395: for (i=0; i<a->mbs; i++) { /* for row block i */
396: for (j=0; j<bs; j++) { /* for row bs*i + j */
397: PetscViewerASCIIPrintf(viewer,"row %D:",i*bs+j);
398: for (k=a->i[i]; k<a->i[i+1]; k++) { /* for column block */
399: for (l=0; l<bs; l++) { /* for column */
400: #if defined(PETSC_USE_COMPLEX)
401: if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) > 0.0) {
402: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k]+l,
403: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
404: } else if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) < 0.0) {
405: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k]+l,
406: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),-(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
407: } else {
408: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)PetscRealPart(a->a[bs2*k + l*bs + j]));
409: }
410: #else
411: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)a->a[bs2*k + l*bs + j]);
412: #endif
413: }
414: }
415: PetscViewerASCIIPrintf(viewer,"\n");
416: }
417: }
418: }
419: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
420: }
421: PetscViewerFlush(viewer);
422: return(0);
423: }
425: #include <petscdraw.h>
428: static PetscErrorCode MatView_SeqSBAIJ_Draw_Zoom(PetscDraw draw,void *Aa)
429: {
430: Mat A = (Mat) Aa;
431: Mat_SeqSBAIJ *a=(Mat_SeqSBAIJ*)A->data;
433: PetscInt row,i,j,k,l,mbs=a->mbs,color,bs=A->rmap->bs,bs2=a->bs2;
434: PetscReal xl,yl,xr,yr,x_l,x_r,y_l,y_r;
435: MatScalar *aa;
436: PetscViewer viewer;
439: PetscObjectQuery((PetscObject)A,"Zoomviewer",(PetscObject*)&viewer);
440: PetscDrawGetCoordinates(draw,&xl,&yl,&xr,&yr);
442: /* loop over matrix elements drawing boxes */
444: PetscDrawCollectiveBegin(draw);
445: PetscDrawString(draw, .3*(xl+xr), .3*(yl+yr), PETSC_DRAW_BLACK, "symmetric");
446: /* Blue for negative, Cyan for zero and Red for positive */
447: color = PETSC_DRAW_BLUE;
448: for (i=0,row=0; i<mbs; i++,row+=bs) {
449: for (j=a->i[i]; j<a->i[i+1]; j++) {
450: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
451: x_l = a->j[j]*bs; x_r = x_l + 1.0;
452: aa = a->a + j*bs2;
453: for (k=0; k<bs; k++) {
454: for (l=0; l<bs; l++) {
455: if (PetscRealPart(*aa++) >= 0.) continue;
456: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
457: }
458: }
459: }
460: }
461: color = PETSC_DRAW_CYAN;
462: for (i=0,row=0; i<mbs; i++,row+=bs) {
463: for (j=a->i[i]; j<a->i[i+1]; j++) {
464: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
465: x_l = a->j[j]*bs; x_r = x_l + 1.0;
466: aa = a->a + j*bs2;
467: for (k=0; k<bs; k++) {
468: for (l=0; l<bs; l++) {
469: if (PetscRealPart(*aa++) != 0.) continue;
470: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
471: }
472: }
473: }
474: }
475: color = PETSC_DRAW_RED;
476: for (i=0,row=0; i<mbs; i++,row+=bs) {
477: for (j=a->i[i]; j<a->i[i+1]; j++) {
478: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
479: x_l = a->j[j]*bs; x_r = x_l + 1.0;
480: aa = a->a + j*bs2;
481: for (k=0; k<bs; k++) {
482: for (l=0; l<bs; l++) {
483: if (PetscRealPart(*aa++) <= 0.) continue;
484: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
485: }
486: }
487: }
488: }
489: PetscDrawCollectiveEnd(draw);
490: return(0);
491: }
495: static PetscErrorCode MatView_SeqSBAIJ_Draw(Mat A,PetscViewer viewer)
496: {
498: PetscReal xl,yl,xr,yr,w,h;
499: PetscDraw draw;
500: PetscBool isnull;
503: PetscViewerDrawGetDraw(viewer,0,&draw);
504: PetscDrawIsNull(draw,&isnull);
505: if (isnull) return(0);
507: xr = A->rmap->N; yr = A->rmap->N; h = yr/10.0; w = xr/10.0;
508: xr += w; yr += h; xl = -w; yl = -h;
509: PetscDrawSetCoordinates(draw,xl,yl,xr,yr);
510: PetscObjectCompose((PetscObject)A,"Zoomviewer",(PetscObject)viewer);
511: PetscDrawZoom(draw,MatView_SeqSBAIJ_Draw_Zoom,A);
512: PetscObjectCompose((PetscObject)A,"Zoomviewer",NULL);
513: PetscDrawSave(draw);
514: return(0);
515: }
519: PetscErrorCode MatView_SeqSBAIJ(Mat A,PetscViewer viewer)
520: {
522: PetscBool iascii,isdraw;
523: FILE *file = 0;
526: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
527: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
528: if (iascii) {
529: MatView_SeqSBAIJ_ASCII(A,viewer);
530: } else if (isdraw) {
531: MatView_SeqSBAIJ_Draw(A,viewer);
532: } else {
533: Mat B;
534: const char *matname;
535: MatConvert(A,MATSEQAIJ,MAT_INITIAL_MATRIX,&B);
536: PetscObjectGetName((PetscObject)A,&matname);
537: PetscObjectSetName((PetscObject)B,matname);
538: MatView(B,viewer);
539: MatDestroy(&B);
540: PetscViewerBinaryGetInfoPointer(viewer,&file);
541: if (file) {
542: fprintf(file,"-matload_block_size %d\n",(int)A->rmap->bs);
543: }
544: }
545: return(0);
546: }
551: PetscErrorCode MatGetValues_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],PetscScalar v[])
552: {
553: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
554: PetscInt *rp,k,low,high,t,row,nrow,i,col,l,*aj = a->j;
555: PetscInt *ai = a->i,*ailen = a->ilen;
556: PetscInt brow,bcol,ridx,cidx,bs=A->rmap->bs,bs2=a->bs2;
557: MatScalar *ap,*aa = a->a;
560: for (k=0; k<m; k++) { /* loop over rows */
561: row = im[k]; brow = row/bs;
562: if (row < 0) {v += n; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",row); */
563: 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);
564: rp = aj + ai[brow]; ap = aa + bs2*ai[brow];
565: nrow = ailen[brow];
566: for (l=0; l<n; l++) { /* loop over columns */
567: if (in[l] < 0) {v++; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",in[l]); */
568: 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);
569: col = in[l];
570: bcol = col/bs;
571: cidx = col%bs;
572: ridx = row%bs;
573: high = nrow;
574: low = 0; /* assume unsorted */
575: while (high-low > 5) {
576: t = (low+high)/2;
577: if (rp[t] > bcol) high = t;
578: else low = t;
579: }
580: for (i=low; i<high; i++) {
581: if (rp[i] > bcol) break;
582: if (rp[i] == bcol) {
583: *v++ = ap[bs2*i+bs*cidx+ridx];
584: goto finished;
585: }
586: }
587: *v++ = 0.0;
588: finished:;
589: }
590: }
591: return(0);
592: }
597: PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
598: {
599: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
600: PetscErrorCode ierr;
601: PetscInt *rp,k,low,high,t,ii,jj,row,nrow,i,col,l,rmax,N,lastcol = -1;
602: PetscInt *imax =a->imax,*ai=a->i,*ailen=a->ilen;
603: PetscInt *aj =a->j,nonew=a->nonew,bs2=a->bs2,bs=A->rmap->bs,stepval;
604: PetscBool roworiented=a->roworiented;
605: const PetscScalar *value = v;
606: MatScalar *ap,*aa = a->a,*bap;
609: if (roworiented) stepval = (n-1)*bs;
610: else stepval = (m-1)*bs;
612: for (k=0; k<m; k++) { /* loop over added rows */
613: row = im[k];
614: if (row < 0) continue;
615: #if defined(PETSC_USE_DEBUG)
616: if (row >= a->mbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block index row too large %D max %D",row,a->mbs-1);
617: #endif
618: rp = aj + ai[row];
619: ap = aa + bs2*ai[row];
620: rmax = imax[row];
621: nrow = ailen[row];
622: low = 0;
623: high = nrow;
624: for (l=0; l<n; l++) { /* loop over added columns */
625: if (in[l] < 0) continue;
626: col = in[l];
627: #if defined(PETSC_USE_DEBUG)
628: if (col >= a->nbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block index column too large %D max %D",col,a->nbs-1);
629: #endif
630: if (col < row) {
631: if (a->ignore_ltriangular) continue; /* ignore lower triangular block */
632: else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
633: }
634: if (roworiented) value = v + k*(stepval+bs)*bs + l*bs;
635: else value = v + l*(stepval+bs)*bs + k*bs;
637: if (col <= lastcol) low = 0;
638: else high = nrow;
640: lastcol = col;
641: while (high-low > 7) {
642: t = (low+high)/2;
643: if (rp[t] > col) high = t;
644: else low = t;
645: }
646: for (i=low; i<high; i++) {
647: if (rp[i] > col) break;
648: if (rp[i] == col) {
649: bap = ap + bs2*i;
650: if (roworiented) {
651: if (is == ADD_VALUES) {
652: for (ii=0; ii<bs; ii++,value+=stepval) {
653: for (jj=ii; jj<bs2; jj+=bs) {
654: bap[jj] += *value++;
655: }
656: }
657: } else {
658: for (ii=0; ii<bs; ii++,value+=stepval) {
659: for (jj=ii; jj<bs2; jj+=bs) {
660: bap[jj] = *value++;
661: }
662: }
663: }
664: } else {
665: if (is == ADD_VALUES) {
666: for (ii=0; ii<bs; ii++,value+=stepval) {
667: for (jj=0; jj<bs; jj++) {
668: *bap++ += *value++;
669: }
670: }
671: } else {
672: for (ii=0; ii<bs; ii++,value+=stepval) {
673: for (jj=0; jj<bs; jj++) {
674: *bap++ = *value++;
675: }
676: }
677: }
678: }
679: goto noinsert2;
680: }
681: }
682: if (nonew == 1) goto noinsert2;
683: if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new block index nonzero block (%D, %D) in the matrix", row, col);
684: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
685: N = nrow++ - 1; high++;
686: /* shift up all the later entries in this row */
687: for (ii=N; ii>=i; ii--) {
688: rp[ii+1] = rp[ii];
689: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));
690: }
691: if (N >= i) {
692: PetscMemzero(ap+bs2*i,bs2*sizeof(MatScalar));
693: }
694: rp[i] = col;
695: bap = ap + bs2*i;
696: if (roworiented) {
697: for (ii=0; ii<bs; ii++,value+=stepval) {
698: for (jj=ii; jj<bs2; jj+=bs) {
699: bap[jj] = *value++;
700: }
701: }
702: } else {
703: for (ii=0; ii<bs; ii++,value+=stepval) {
704: for (jj=0; jj<bs; jj++) {
705: *bap++ = *value++;
706: }
707: }
708: }
709: noinsert2:;
710: low = i;
711: }
712: ailen[row] = nrow;
713: }
714: return(0);
715: }
717: /*
718: This is not yet used
719: */
722: PetscErrorCode MatAssemblyEnd_SeqSBAIJ_SeqAIJ_Inode(Mat A)
723: {
724: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
726: const PetscInt *ai = a->i, *aj = a->j,*cols;
727: PetscInt i = 0,j,blk_size,m = A->rmap->n,node_count = 0,nzx,nzy,*ns,row,nz,cnt,cnt2,*counts;
728: PetscBool flag;
731: PetscMalloc1(m,&ns);
732: while (i < m) {
733: nzx = ai[i+1] - ai[i]; /* Number of nonzeros */
734: /* Limits the number of elements in a node to 'a->inode.limit' */
735: for (j=i+1,blk_size=1; j<m && blk_size <a->inode.limit; ++j,++blk_size) {
736: nzy = ai[j+1] - ai[j];
737: if (nzy != (nzx - j + i)) break;
738: PetscMemcmp(aj + ai[i] + j - i,aj + ai[j],nzy*sizeof(PetscInt),&flag);
739: if (!flag) break;
740: }
741: ns[node_count++] = blk_size;
743: i = j;
744: }
745: if (!a->inode.size && m && node_count > .9*m) {
746: PetscFree(ns);
747: PetscInfo2(A,"Found %D nodes out of %D rows. Not using Inode routines\n",node_count,m);
748: } else {
749: a->inode.node_count = node_count;
751: PetscMalloc1(node_count,&a->inode.size);
752: PetscLogObjectMemory((PetscObject)A,node_count*sizeof(PetscInt));
753: PetscMemcpy(a->inode.size,ns,node_count*sizeof(PetscInt));
754: PetscFree(ns);
755: PetscInfo3(A,"Found %D nodes of %D. Limit used: %D. Using Inode routines\n",node_count,m,a->inode.limit);
757: /* count collections of adjacent columns in each inode */
758: row = 0;
759: cnt = 0;
760: for (i=0; i<node_count; i++) {
761: cols = aj + ai[row] + a->inode.size[i];
762: nz = ai[row+1] - ai[row] - a->inode.size[i];
763: for (j=1; j<nz; j++) {
764: if (cols[j] != cols[j-1]+1) cnt++;
765: }
766: cnt++;
767: row += a->inode.size[i];
768: }
769: PetscMalloc1(2*cnt,&counts);
770: cnt = 0;
771: row = 0;
772: for (i=0; i<node_count; i++) {
773: cols = aj + ai[row] + a->inode.size[i];
774: counts[2*cnt] = cols[0];
775: nz = ai[row+1] - ai[row] - a->inode.size[i];
776: cnt2 = 1;
777: for (j=1; j<nz; j++) {
778: if (cols[j] != cols[j-1]+1) {
779: counts[2*(cnt++)+1] = cnt2;
780: counts[2*cnt] = cols[j];
781: cnt2 = 1;
782: } else cnt2++;
783: }
784: counts[2*(cnt++)+1] = cnt2;
785: row += a->inode.size[i];
786: }
787: PetscIntView(2*cnt,counts,0);
788: }
789: return(0);
790: }
794: PetscErrorCode MatAssemblyEnd_SeqSBAIJ(Mat A,MatAssemblyType mode)
795: {
796: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
798: PetscInt fshift = 0,i,j,*ai = a->i,*aj = a->j,*imax = a->imax;
799: PetscInt m = A->rmap->N,*ip,N,*ailen = a->ilen;
800: PetscInt mbs = a->mbs,bs2 = a->bs2,rmax = 0;
801: MatScalar *aa = a->a,*ap;
804: if (mode == MAT_FLUSH_ASSEMBLY) return(0);
806: if (m) rmax = ailen[0];
807: for (i=1; i<mbs; i++) {
808: /* move each row back by the amount of empty slots (fshift) before it*/
809: fshift += imax[i-1] - ailen[i-1];
810: rmax = PetscMax(rmax,ailen[i]);
811: if (fshift) {
812: ip = aj + ai[i]; ap = aa + bs2*ai[i];
813: N = ailen[i];
814: for (j=0; j<N; j++) {
815: ip[j-fshift] = ip[j];
816: PetscMemcpy(ap+(j-fshift)*bs2,ap+j*bs2,bs2*sizeof(MatScalar));
817: }
818: }
819: ai[i] = ai[i-1] + ailen[i-1];
820: }
821: if (mbs) {
822: fshift += imax[mbs-1] - ailen[mbs-1];
823: ai[mbs] = ai[mbs-1] + ailen[mbs-1];
824: }
825: /* reset ilen and imax for each row */
826: for (i=0; i<mbs; i++) {
827: ailen[i] = imax[i] = ai[i+1] - ai[i];
828: }
829: a->nz = ai[mbs];
831: /* diagonals may have moved, reset it */
832: if (a->diag) {
833: PetscMemcpy(a->diag,ai,mbs*sizeof(PetscInt));
834: }
835: if (fshift && a->nounused == -1) SETERRQ4(PETSC_COMM_SELF,PETSC_ERR_PLIB, "Unused space detected in matrix: %D X %D block size %D, %D unneeded", m, A->cmap->n, A->rmap->bs, fshift*bs2);
837: PetscInfo5(A,"Matrix size: %D X %D, block size %D; storage space: %D unneeded, %D used\n",m,A->rmap->N,A->rmap->bs,fshift*bs2,a->nz*bs2);
838: PetscInfo1(A,"Number of mallocs during MatSetValues is %D\n",a->reallocs);
839: PetscInfo1(A,"Most nonzeros blocks in any row is %D\n",rmax);
841: A->info.mallocs += a->reallocs;
842: a->reallocs = 0;
843: A->info.nz_unneeded = (PetscReal)fshift*bs2;
844: a->idiagvalid = PETSC_FALSE;
845: a->rmax = rmax;
847: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
848: if (a->jshort && a->free_jshort) {
849: /* when matrix data structure is changed, previous jshort must be replaced */
850: PetscFree(a->jshort);
851: }
852: PetscMalloc1(a->i[A->rmap->n],&a->jshort);
853: PetscLogObjectMemory((PetscObject)A,a->i[A->rmap->n]*sizeof(unsigned short));
854: for (i=0; i<a->i[A->rmap->n]; i++) a->jshort[i] = a->j[i];
855: A->ops->mult = MatMult_SeqSBAIJ_1_ushort;
856: A->ops->sor = MatSOR_SeqSBAIJ_ushort;
857: a->free_jshort = PETSC_TRUE;
858: }
859: return(0);
860: }
862: /*
863: This function returns an array of flags which indicate the locations of contiguous
864: blocks that should be zeroed. for eg: if bs = 3 and is = [0,1,2,3,5,6,7,8,9]
865: then the resulting sizes = [3,1,1,3,1] correspondig to sets [(0,1,2),(3),(5),(6,7,8),(9)]
866: Assume: sizes should be long enough to hold all the values.
867: */
870: PetscErrorCode MatZeroRows_SeqSBAIJ_Check_Blocks(PetscInt idx[],PetscInt n,PetscInt bs,PetscInt sizes[], PetscInt *bs_max)
871: {
872: PetscInt i,j,k,row;
873: PetscBool flg;
876: for (i=0,j=0; i<n; j++) {
877: row = idx[i];
878: if (row%bs!=0) { /* Not the begining of a block */
879: sizes[j] = 1;
880: i++;
881: } else if (i+bs > n) { /* Beginning of a block, but complete block doesn't exist (at idx end) */
882: sizes[j] = 1; /* Also makes sure atleast 'bs' values exist for next else */
883: i++;
884: } else { /* Begining of the block, so check if the complete block exists */
885: flg = PETSC_TRUE;
886: for (k=1; k<bs; k++) {
887: if (row+k != idx[i+k]) { /* break in the block */
888: flg = PETSC_FALSE;
889: break;
890: }
891: }
892: if (flg) { /* No break in the bs */
893: sizes[j] = bs;
894: i += bs;
895: } else {
896: sizes[j] = 1;
897: i++;
898: }
899: }
900: }
901: *bs_max = j;
902: return(0);
903: }
906: /* Only add/insert a(i,j) with i<=j (blocks).
907: Any a(i,j) with i>j input by user is ingored.
908: */
912: PetscErrorCode MatSetValues_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
913: {
914: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
916: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N,lastcol = -1;
917: PetscInt *imax=a->imax,*ai=a->i,*ailen=a->ilen,roworiented=a->roworiented;
918: PetscInt *aj =a->j,nonew=a->nonew,bs=A->rmap->bs,brow,bcol;
919: PetscInt ridx,cidx,bs2=a->bs2;
920: MatScalar *ap,value,*aa=a->a,*bap;
923: for (k=0; k<m; k++) { /* loop over added rows */
924: row = im[k]; /* row number */
925: brow = row/bs; /* block row number */
926: if (row < 0) continue;
927: #if defined(PETSC_USE_DEBUG)
928: 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);
929: #endif
930: rp = aj + ai[brow]; /*ptr to beginning of column value of the row block*/
931: ap = aa + bs2*ai[brow]; /*ptr to beginning of element value of the row block*/
932: rmax = imax[brow]; /* maximum space allocated for this row */
933: nrow = ailen[brow]; /* actual length of this row */
934: low = 0;
936: for (l=0; l<n; l++) { /* loop over added columns */
937: if (in[l] < 0) continue;
938: #if defined(PETSC_USE_DEBUG)
939: if (in[l] >= A->rmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->rmap->N-1);
940: #endif
941: col = in[l];
942: bcol = col/bs; /* block col number */
944: if (brow > bcol) {
945: if (a->ignore_ltriangular) continue; /* ignore lower triangular values */
946: else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
947: }
949: ridx = row % bs; cidx = col % bs; /*row and col index inside the block */
950: if ((brow==bcol && ridx<=cidx) || (brow<bcol)) {
951: /* element value a(k,l) */
952: if (roworiented) value = v[l + k*n];
953: else value = v[k + l*m];
955: /* move pointer bap to a(k,l) quickly and add/insert value */
956: if (col <= lastcol) low = 0;
957: high = nrow;
958: lastcol = col;
959: while (high-low > 7) {
960: t = (low+high)/2;
961: if (rp[t] > bcol) high = t;
962: else low = t;
963: }
964: for (i=low; i<high; i++) {
965: if (rp[i] > bcol) break;
966: if (rp[i] == bcol) {
967: bap = ap + bs2*i + bs*cidx + ridx;
968: if (is == ADD_VALUES) *bap += value;
969: else *bap = value;
970: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
971: if (brow == bcol && ridx < cidx) {
972: bap = ap + bs2*i + bs*ridx + cidx;
973: if (is == ADD_VALUES) *bap += value;
974: else *bap = value;
975: }
976: goto noinsert1;
977: }
978: }
980: if (nonew == 1) goto noinsert1;
981: if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) in the matrix", row, col);
982: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
984: N = nrow++ - 1; high++;
985: /* shift up all the later entries in this row */
986: for (ii=N; ii>=i; ii--) {
987: rp[ii+1] = rp[ii];
988: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));
989: }
990: if (N>=i) {
991: PetscMemzero(ap+bs2*i,bs2*sizeof(MatScalar));
992: }
993: rp[i] = bcol;
994: ap[bs2*i + bs*cidx + ridx] = value;
995: A->nonzerostate++;
996: noinsert1:;
997: low = i;
998: }
999: } /* end of loop over added columns */
1000: ailen[brow] = nrow;
1001: } /* end of loop over added rows */
1002: return(0);
1003: }
1007: PetscErrorCode MatICCFactor_SeqSBAIJ(Mat inA,IS row,const MatFactorInfo *info)
1008: {
1009: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)inA->data;
1010: Mat outA;
1012: PetscBool row_identity;
1015: if (info->levels != 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only levels=0 is supported for in-place icc");
1016: ISIdentity(row,&row_identity);
1017: if (!row_identity) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix reordering is not supported");
1018: if (inA->rmap->bs != 1) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix block size %D is not supported",inA->rmap->bs); /* Need to replace MatCholeskyFactorSymbolic_SeqSBAIJ_MSR()! */
1020: outA = inA;
1021: inA->factortype = MAT_FACTOR_ICC;
1022: PetscFree(inA->solvertype);
1023: PetscStrallocpy(MATSOLVERPETSC,&inA->solvertype);
1025: MatMarkDiagonal_SeqSBAIJ(inA);
1026: MatSeqSBAIJSetNumericFactorization_inplace(inA,row_identity);
1028: PetscObjectReference((PetscObject)row);
1029: ISDestroy(&a->row);
1030: a->row = row;
1031: PetscObjectReference((PetscObject)row);
1032: ISDestroy(&a->col);
1033: a->col = row;
1035: /* Create the invert permutation so that it can be used in MatCholeskyFactorNumeric() */
1036: if (a->icol) {ISInvertPermutation(row,PETSC_DECIDE, &a->icol);}
1037: PetscLogObjectParent((PetscObject)inA,(PetscObject)a->icol);
1039: if (!a->solve_work) {
1040: PetscMalloc1(inA->rmap->N+inA->rmap->bs,&a->solve_work);
1041: PetscLogObjectMemory((PetscObject)inA,(inA->rmap->N+inA->rmap->bs)*sizeof(PetscScalar));
1042: }
1044: MatCholeskyFactorNumeric(outA,inA,info);
1045: return(0);
1046: }
1050: PetscErrorCode MatSeqSBAIJSetColumnIndices_SeqSBAIJ(Mat mat,PetscInt *indices)
1051: {
1052: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ*)mat->data;
1053: PetscInt i,nz,n;
1057: nz = baij->maxnz;
1058: n = mat->cmap->n;
1059: for (i=0; i<nz; i++) baij->j[i] = indices[i];
1061: baij->nz = nz;
1062: for (i=0; i<n; i++) baij->ilen[i] = baij->imax[i];
1064: MatSetOption(mat,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
1065: return(0);
1066: }
1070: /*@
1071: MatSeqSBAIJSetColumnIndices - Set the column indices for all the rows
1072: in the matrix.
1074: Input Parameters:
1075: + mat - the SeqSBAIJ matrix
1076: - indices - the column indices
1078: Level: advanced
1080: Notes:
1081: This can be called if you have precomputed the nonzero structure of the
1082: matrix and want to provide it to the matrix object to improve the performance
1083: of the MatSetValues() operation.
1085: You MUST have set the correct numbers of nonzeros per row in the call to
1086: MatCreateSeqSBAIJ(), and the columns indices MUST be sorted.
1088: MUST be called before any calls to MatSetValues()
1090: .seealso: MatCreateSeqSBAIJ
1091: @*/
1092: PetscErrorCode MatSeqSBAIJSetColumnIndices(Mat mat,PetscInt *indices)
1093: {
1099: PetscUseMethod(mat,"MatSeqSBAIJSetColumnIndices_C",(Mat,PetscInt*),(mat,indices));
1100: return(0);
1101: }
1105: PetscErrorCode MatCopy_SeqSBAIJ(Mat A,Mat B,MatStructure str)
1106: {
1110: /* If the two matrices have the same copy implementation, use fast copy. */
1111: if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) {
1112: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1113: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ*)B->data;
1115: 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");
1116: PetscMemcpy(b->a,a->a,(a->i[A->rmap->N])*sizeof(PetscScalar));
1117: } else {
1118: MatGetRowUpperTriangular(A);
1119: MatCopy_Basic(A,B,str);
1120: MatRestoreRowUpperTriangular(A);
1121: }
1122: return(0);
1123: }
1127: PetscErrorCode MatSetUp_SeqSBAIJ(Mat A)
1128: {
1132: MatSeqSBAIJSetPreallocation_SeqSBAIJ(A,A->rmap->bs,PETSC_DEFAULT,0);
1133: return(0);
1134: }
1138: PetscErrorCode MatSeqSBAIJGetArray_SeqSBAIJ(Mat A,PetscScalar *array[])
1139: {
1140: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1143: *array = a->a;
1144: return(0);
1145: }
1149: PetscErrorCode MatSeqSBAIJRestoreArray_SeqSBAIJ(Mat A,PetscScalar *array[])
1150: {
1152: return(0);
1153: }
1157: PetscErrorCode MatAXPYGetPreallocation_SeqSBAIJ(Mat Y,Mat X,PetscInt *nnz)
1158: {
1159: PetscInt bs = Y->rmap->bs,mbs = Y->rmap->N/bs;
1160: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ*)X->data;
1161: Mat_SeqSBAIJ *y = (Mat_SeqSBAIJ*)Y->data;
1165: /* Set the number of nonzeros in the new matrix */
1166: MatAXPYGetPreallocation_SeqX_private(mbs,x->i,x->j,y->i,y->j,nnz);
1167: return(0);
1168: }
1172: PetscErrorCode MatAXPY_SeqSBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
1173: {
1174: Mat_SeqSBAIJ *x=(Mat_SeqSBAIJ*)X->data, *y=(Mat_SeqSBAIJ*)Y->data;
1176: PetscInt bs=Y->rmap->bs,bs2=bs*bs;
1177: PetscBLASInt one = 1;
1180: if (str == SAME_NONZERO_PATTERN) {
1181: PetscScalar alpha = a;
1182: PetscBLASInt bnz;
1183: PetscBLASIntCast(x->nz*bs2,&bnz);
1184: PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one));
1185: PetscObjectStateIncrease((PetscObject)Y);
1186: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1187: MatSetOption(X,MAT_GETROW_UPPERTRIANGULAR,PETSC_TRUE);
1188: MatAXPY_Basic(Y,a,X,str);
1189: MatSetOption(X,MAT_GETROW_UPPERTRIANGULAR,PETSC_FALSE);
1190: } else {
1191: Mat B;
1192: PetscInt *nnz;
1193: if (bs != X->rmap->bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Matrices must have same block size");
1194: MatGetRowUpperTriangular(X);
1195: MatGetRowUpperTriangular(Y);
1196: PetscMalloc1(Y->rmap->N,&nnz);
1197: MatCreate(PetscObjectComm((PetscObject)Y),&B);
1198: PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);
1199: MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);
1200: MatSetBlockSizesFromMats(B,Y,Y);
1201: MatSetType(B,(MatType) ((PetscObject)Y)->type_name);
1202: MatAXPYGetPreallocation_SeqSBAIJ(Y,X,nnz);
1203: MatSeqSBAIJSetPreallocation(B,bs,0,nnz);
1205: MatAXPY_BasicWithPreallocation(B,Y,a,X,str);
1207: MatHeaderReplace(Y,&B);
1208: PetscFree(nnz);
1209: MatRestoreRowUpperTriangular(X);
1210: MatRestoreRowUpperTriangular(Y);
1211: }
1212: return(0);
1213: }
1217: PetscErrorCode MatIsSymmetric_SeqSBAIJ(Mat A,PetscReal tol,PetscBool *flg)
1218: {
1220: *flg = PETSC_TRUE;
1221: return(0);
1222: }
1226: PetscErrorCode MatIsStructurallySymmetric_SeqSBAIJ(Mat A,PetscBool *flg)
1227: {
1229: *flg = PETSC_TRUE;
1230: return(0);
1231: }
1235: PetscErrorCode MatIsHermitian_SeqSBAIJ(Mat A,PetscReal tol,PetscBool *flg)
1236: {
1238: *flg = PETSC_FALSE;
1239: return(0);
1240: }
1244: PetscErrorCode MatRealPart_SeqSBAIJ(Mat A)
1245: {
1246: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1247: PetscInt i,nz = a->bs2*a->i[a->mbs];
1248: MatScalar *aa = a->a;
1251: for (i=0; i<nz; i++) aa[i] = PetscRealPart(aa[i]);
1252: return(0);
1253: }
1257: PetscErrorCode MatImaginaryPart_SeqSBAIJ(Mat A)
1258: {
1259: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1260: PetscInt i,nz = a->bs2*a->i[a->mbs];
1261: MatScalar *aa = a->a;
1264: for (i=0; i<nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1265: return(0);
1266: }
1270: PetscErrorCode MatZeroRowsColumns_SeqSBAIJ(Mat A,PetscInt is_n,const PetscInt is_idx[],PetscScalar diag,Vec x, Vec b)
1271: {
1272: Mat_SeqSBAIJ *baij=(Mat_SeqSBAIJ*)A->data;
1273: PetscErrorCode ierr;
1274: PetscInt i,j,k,count;
1275: PetscInt bs =A->rmap->bs,bs2=baij->bs2,row,col;
1276: PetscScalar zero = 0.0;
1277: MatScalar *aa;
1278: const PetscScalar *xx;
1279: PetscScalar *bb;
1280: PetscBool *zeroed,vecs = PETSC_FALSE;
1283: /* fix right hand side if needed */
1284: if (x && b) {
1285: VecGetArrayRead(x,&xx);
1286: VecGetArray(b,&bb);
1287: vecs = PETSC_TRUE;
1288: }
1290: /* zero the columns */
1291: PetscCalloc1(A->rmap->n,&zeroed);
1292: for (i=0; i<is_n; i++) {
1293: if (is_idx[i] < 0 || is_idx[i] >= A->rmap->N) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range",is_idx[i]);
1294: zeroed[is_idx[i]] = PETSC_TRUE;
1295: }
1296: if (vecs) {
1297: for (i=0; i<A->rmap->N; i++) {
1298: row = i/bs;
1299: for (j=baij->i[row]; j<baij->i[row+1]; j++) {
1300: for (k=0; k<bs; k++) {
1301: col = bs*baij->j[j] + k;
1302: if (col <= i) continue;
1303: aa = ((MatScalar*)(baij->a)) + j*bs2 + (i%bs) + bs*k;
1304: if (!zeroed[i] && zeroed[col]) bb[i] -= aa[0]*xx[col];
1305: if (zeroed[i] && !zeroed[col]) bb[col] -= aa[0]*xx[i];
1306: }
1307: }
1308: }
1309: for (i=0; i<is_n; i++) bb[is_idx[i]] = diag*xx[is_idx[i]];
1310: }
1312: for (i=0; i<A->rmap->N; i++) {
1313: if (!zeroed[i]) {
1314: row = i/bs;
1315: for (j=baij->i[row]; j<baij->i[row+1]; j++) {
1316: for (k=0; k<bs; k++) {
1317: col = bs*baij->j[j] + k;
1318: if (zeroed[col]) {
1319: aa = ((MatScalar*)(baij->a)) + j*bs2 + (i%bs) + bs*k;
1320: aa[0] = 0.0;
1321: }
1322: }
1323: }
1324: }
1325: }
1326: PetscFree(zeroed);
1327: if (vecs) {
1328: VecRestoreArrayRead(x,&xx);
1329: VecRestoreArray(b,&bb);
1330: }
1332: /* zero the rows */
1333: for (i=0; i<is_n; i++) {
1334: row = is_idx[i];
1335: count = (baij->i[row/bs +1] - baij->i[row/bs])*bs;
1336: aa = ((MatScalar*)(baij->a)) + baij->i[row/bs]*bs2 + (row%bs);
1337: for (k=0; k<count; k++) {
1338: aa[0] = zero;
1339: aa += bs;
1340: }
1341: if (diag != 0.0) {
1342: (*A->ops->setvalues)(A,1,&row,1,&row,&diag,INSERT_VALUES);
1343: }
1344: }
1345: MatAssemblyEnd_SeqSBAIJ(A,MAT_FINAL_ASSEMBLY);
1346: return(0);
1347: }
1351: PetscErrorCode MatShift_SeqSBAIJ(Mat Y,PetscScalar a)
1352: {
1354: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)Y->data;
1357: if (!Y->preallocated || !aij->nz) {
1358: MatSeqSBAIJSetPreallocation(Y,Y->rmap->bs,1,NULL);
1359: }
1360: MatShift_Basic(Y,a);
1361: return(0);
1362: }
1364: /* -------------------------------------------------------------------*/
1365: static struct _MatOps MatOps_Values = {MatSetValues_SeqSBAIJ,
1366: MatGetRow_SeqSBAIJ,
1367: MatRestoreRow_SeqSBAIJ,
1368: MatMult_SeqSBAIJ_N,
1369: /* 4*/ MatMultAdd_SeqSBAIJ_N,
1370: MatMult_SeqSBAIJ_N, /* transpose versions are same as non-transpose versions */
1371: MatMultAdd_SeqSBAIJ_N,
1372: 0,
1373: 0,
1374: 0,
1375: /* 10*/ 0,
1376: 0,
1377: MatCholeskyFactor_SeqSBAIJ,
1378: MatSOR_SeqSBAIJ,
1379: MatTranspose_SeqSBAIJ,
1380: /* 15*/ MatGetInfo_SeqSBAIJ,
1381: MatEqual_SeqSBAIJ,
1382: MatGetDiagonal_SeqSBAIJ,
1383: MatDiagonalScale_SeqSBAIJ,
1384: MatNorm_SeqSBAIJ,
1385: /* 20*/ 0,
1386: MatAssemblyEnd_SeqSBAIJ,
1387: MatSetOption_SeqSBAIJ,
1388: MatZeroEntries_SeqSBAIJ,
1389: /* 24*/ 0,
1390: 0,
1391: 0,
1392: 0,
1393: 0,
1394: /* 29*/ MatSetUp_SeqSBAIJ,
1395: 0,
1396: 0,
1397: 0,
1398: 0,
1399: /* 34*/ MatDuplicate_SeqSBAIJ,
1400: 0,
1401: 0,
1402: 0,
1403: MatICCFactor_SeqSBAIJ,
1404: /* 39*/ MatAXPY_SeqSBAIJ,
1405: MatGetSubMatrices_SeqSBAIJ,
1406: MatIncreaseOverlap_SeqSBAIJ,
1407: MatGetValues_SeqSBAIJ,
1408: MatCopy_SeqSBAIJ,
1409: /* 44*/ 0,
1410: MatScale_SeqSBAIJ,
1411: MatShift_SeqSBAIJ,
1412: 0,
1413: MatZeroRowsColumns_SeqSBAIJ,
1414: /* 49*/ 0,
1415: MatGetRowIJ_SeqSBAIJ,
1416: MatRestoreRowIJ_SeqSBAIJ,
1417: 0,
1418: 0,
1419: /* 54*/ 0,
1420: 0,
1421: 0,
1422: 0,
1423: MatSetValuesBlocked_SeqSBAIJ,
1424: /* 59*/ MatGetSubMatrix_SeqSBAIJ,
1425: 0,
1426: 0,
1427: 0,
1428: 0,
1429: /* 64*/ 0,
1430: 0,
1431: 0,
1432: 0,
1433: 0,
1434: /* 69*/ MatGetRowMaxAbs_SeqSBAIJ,
1435: 0,
1436: 0,
1437: 0,
1438: 0,
1439: /* 74*/ 0,
1440: 0,
1441: 0,
1442: 0,
1443: 0,
1444: /* 79*/ 0,
1445: 0,
1446: 0,
1447: MatGetInertia_SeqSBAIJ,
1448: MatLoad_SeqSBAIJ,
1449: /* 84*/ MatIsSymmetric_SeqSBAIJ,
1450: MatIsHermitian_SeqSBAIJ,
1451: MatIsStructurallySymmetric_SeqSBAIJ,
1452: 0,
1453: 0,
1454: /* 89*/ 0,
1455: 0,
1456: 0,
1457: 0,
1458: 0,
1459: /* 94*/ 0,
1460: 0,
1461: 0,
1462: 0,
1463: 0,
1464: /* 99*/ 0,
1465: 0,
1466: 0,
1467: 0,
1468: 0,
1469: /*104*/ 0,
1470: MatRealPart_SeqSBAIJ,
1471: MatImaginaryPart_SeqSBAIJ,
1472: MatGetRowUpperTriangular_SeqSBAIJ,
1473: MatRestoreRowUpperTriangular_SeqSBAIJ,
1474: /*109*/ 0,
1475: 0,
1476: 0,
1477: 0,
1478: MatMissingDiagonal_SeqSBAIJ,
1479: /*114*/ 0,
1480: 0,
1481: 0,
1482: 0,
1483: 0,
1484: /*119*/ 0,
1485: 0,
1486: 0,
1487: 0,
1488: 0,
1489: /*124*/ 0,
1490: 0,
1491: 0,
1492: 0,
1493: 0,
1494: /*129*/ 0,
1495: 0,
1496: 0,
1497: 0,
1498: 0,
1499: /*134*/ 0,
1500: 0,
1501: 0,
1502: 0,
1503: 0,
1504: /*139*/ 0,
1505: 0,
1506: 0,
1507: 0,
1508: 0,
1509: /*144*/MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ
1510: };
1514: PetscErrorCode MatStoreValues_SeqSBAIJ(Mat mat)
1515: {
1516: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)mat->data;
1517: PetscInt nz = aij->i[mat->rmap->N]*mat->rmap->bs*aij->bs2;
1521: if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1523: /* allocate space for values if not already there */
1524: if (!aij->saved_values) {
1525: PetscMalloc1(nz+1,&aij->saved_values);
1526: }
1528: /* copy values over */
1529: PetscMemcpy(aij->saved_values,aij->a,nz*sizeof(PetscScalar));
1530: return(0);
1531: }
1535: PetscErrorCode MatRetrieveValues_SeqSBAIJ(Mat mat)
1536: {
1537: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)mat->data;
1539: PetscInt nz = aij->i[mat->rmap->N]*mat->rmap->bs*aij->bs2;
1542: if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1543: if (!aij->saved_values) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatStoreValues(A);first");
1545: /* copy values over */
1546: PetscMemcpy(aij->a,aij->saved_values,nz*sizeof(PetscScalar));
1547: return(0);
1548: }
1552: PetscErrorCode MatSeqSBAIJSetPreallocation_SeqSBAIJ(Mat B,PetscInt bs,PetscInt nz,PetscInt *nnz)
1553: {
1554: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ*)B->data;
1556: PetscInt i,mbs,nbs,bs2;
1557: PetscBool skipallocation = PETSC_FALSE,flg = PETSC_FALSE,realalloc = PETSC_FALSE;
1560: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
1561: B->preallocated = PETSC_TRUE;
1563: MatSetBlockSize(B,PetscAbs(bs));
1564: PetscLayoutSetUp(B->rmap);
1565: PetscLayoutSetUp(B->cmap);
1566: PetscLayoutGetBlockSize(B->rmap,&bs);
1568: mbs = B->rmap->N/bs;
1569: nbs = B->cmap->n/bs;
1570: bs2 = bs*bs;
1572: if (mbs*bs != B->rmap->N || nbs*bs!=B->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Number rows, cols must be divisible by blocksize");
1574: if (nz == MAT_SKIP_ALLOCATION) {
1575: skipallocation = PETSC_TRUE;
1576: nz = 0;
1577: }
1579: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 3;
1580: if (nz < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nz cannot be less than 0: value %D",nz);
1581: if (nnz) {
1582: for (i=0; i<mbs; i++) {
1583: 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]);
1584: if (nnz[i] > nbs) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be greater than block row length: local row %D value %D block rowlength %D",i,nnz[i],nbs);
1585: }
1586: }
1588: B->ops->mult = MatMult_SeqSBAIJ_N;
1589: B->ops->multadd = MatMultAdd_SeqSBAIJ_N;
1590: B->ops->multtranspose = MatMult_SeqSBAIJ_N;
1591: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_N;
1593: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_no_unroll",&flg,NULL);
1594: if (!flg) {
1595: switch (bs) {
1596: case 1:
1597: B->ops->mult = MatMult_SeqSBAIJ_1;
1598: B->ops->multadd = MatMultAdd_SeqSBAIJ_1;
1599: B->ops->multtranspose = MatMult_SeqSBAIJ_1;
1600: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_1;
1601: break;
1602: case 2:
1603: B->ops->mult = MatMult_SeqSBAIJ_2;
1604: B->ops->multadd = MatMultAdd_SeqSBAIJ_2;
1605: B->ops->multtranspose = MatMult_SeqSBAIJ_2;
1606: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_2;
1607: break;
1608: case 3:
1609: B->ops->mult = MatMult_SeqSBAIJ_3;
1610: B->ops->multadd = MatMultAdd_SeqSBAIJ_3;
1611: B->ops->multtranspose = MatMult_SeqSBAIJ_3;
1612: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_3;
1613: break;
1614: case 4:
1615: B->ops->mult = MatMult_SeqSBAIJ_4;
1616: B->ops->multadd = MatMultAdd_SeqSBAIJ_4;
1617: B->ops->multtranspose = MatMult_SeqSBAIJ_4;
1618: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_4;
1619: break;
1620: case 5:
1621: B->ops->mult = MatMult_SeqSBAIJ_5;
1622: B->ops->multadd = MatMultAdd_SeqSBAIJ_5;
1623: B->ops->multtranspose = MatMult_SeqSBAIJ_5;
1624: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_5;
1625: break;
1626: case 6:
1627: B->ops->mult = MatMult_SeqSBAIJ_6;
1628: B->ops->multadd = MatMultAdd_SeqSBAIJ_6;
1629: B->ops->multtranspose = MatMult_SeqSBAIJ_6;
1630: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_6;
1631: break;
1632: case 7:
1633: B->ops->mult = MatMult_SeqSBAIJ_7;
1634: B->ops->multadd = MatMultAdd_SeqSBAIJ_7;
1635: B->ops->multtranspose = MatMult_SeqSBAIJ_7;
1636: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_7;
1637: break;
1638: }
1639: }
1641: b->mbs = mbs;
1642: b->nbs = nbs;
1643: if (!skipallocation) {
1644: if (!b->imax) {
1645: PetscMalloc2(mbs,&b->imax,mbs,&b->ilen);
1647: b->free_imax_ilen = PETSC_TRUE;
1649: PetscLogObjectMemory((PetscObject)B,2*mbs*sizeof(PetscInt));
1650: }
1651: if (!nnz) {
1652: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
1653: else if (nz <= 0) nz = 1;
1654: for (i=0; i<mbs; i++) b->imax[i] = nz;
1655: nz = nz*mbs; /* total nz */
1656: } else {
1657: nz = 0;
1658: for (i=0; i<mbs; i++) {b->imax[i] = nnz[i]; nz += nnz[i];}
1659: }
1660: /* b->ilen will count nonzeros in each block row so far. */
1661: for (i=0; i<mbs; i++) b->ilen[i] = 0;
1662: /* nz=(nz+mbs)/2; */ /* total diagonal and superdiagonal nonzero blocks */
1664: /* allocate the matrix space */
1665: MatSeqXAIJFreeAIJ(B,&b->a,&b->j,&b->i);
1666: PetscMalloc3(bs2*nz,&b->a,nz,&b->j,B->rmap->N+1,&b->i);
1667: PetscLogObjectMemory((PetscObject)B,(B->rmap->N+1)*sizeof(PetscInt)+nz*(bs2*sizeof(PetscScalar)+sizeof(PetscInt)));
1668: PetscMemzero(b->a,nz*bs2*sizeof(MatScalar));
1669: PetscMemzero(b->j,nz*sizeof(PetscInt));
1671: b->singlemalloc = PETSC_TRUE;
1673: /* pointer to beginning of each row */
1674: b->i[0] = 0;
1675: for (i=1; i<mbs+1; i++) b->i[i] = b->i[i-1] + b->imax[i-1];
1677: b->free_a = PETSC_TRUE;
1678: b->free_ij = PETSC_TRUE;
1679: } else {
1680: b->free_a = PETSC_FALSE;
1681: b->free_ij = PETSC_FALSE;
1682: }
1684: B->rmap->bs = bs;
1685: b->bs2 = bs2;
1686: b->nz = 0;
1687: b->maxnz = nz;
1689: b->inew = 0;
1690: b->jnew = 0;
1691: b->anew = 0;
1692: b->a2anew = 0;
1693: b->permute = PETSC_FALSE;
1694: if (realalloc) {MatSetOption(B,MAT_NEW_NONZERO_ALLOCATION_ERR,PETSC_TRUE);}
1695: return(0);
1696: }
1700: PetscErrorCode MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ(Mat B,PetscInt bs,const PetscInt ii[],const PetscInt jj[], const PetscScalar V[])
1701: {
1702: PetscInt i,j,m,nz,nz_max=0,*nnz;
1703: PetscScalar *values=0;
1704: PetscBool roworiented = ((Mat_SeqSBAIJ*)B->data)->roworiented;
1707: if (bs < 1) SETERRQ1(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_OUTOFRANGE,"Invalid block size specified, must be positive but it is %D",bs);
1708: PetscLayoutSetBlockSize(B->rmap,bs);
1709: PetscLayoutSetBlockSize(B->cmap,bs);
1710: PetscLayoutSetUp(B->rmap);
1711: PetscLayoutSetUp(B->cmap);
1712: PetscLayoutGetBlockSize(B->rmap,&bs);
1713: m = B->rmap->n/bs;
1715: if (ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"ii[0] must be 0 but it is %D",ii[0]);
1716: PetscMalloc1(m+1,&nnz);
1717: for (i=0; i<m; i++) {
1718: nz = ii[i+1] - ii[i];
1719: if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row %D has a negative number of columns %D",i,nz);
1720: nz_max = PetscMax(nz_max,nz);
1721: nnz[i] = nz;
1722: }
1723: MatSeqSBAIJSetPreallocation(B,bs,0,nnz);
1724: PetscFree(nnz);
1726: values = (PetscScalar*)V;
1727: if (!values) {
1728: PetscCalloc1(bs*bs*nz_max,&values);
1729: }
1730: for (i=0; i<m; i++) {
1731: PetscInt ncols = ii[i+1] - ii[i];
1732: const PetscInt *icols = jj + ii[i];
1733: if (!roworiented || bs == 1) {
1734: const PetscScalar *svals = values + (V ? (bs*bs*ii[i]) : 0);
1735: MatSetValuesBlocked_SeqSBAIJ(B,1,&i,ncols,icols,svals,INSERT_VALUES);
1736: } else {
1737: for (j=0; j<ncols; j++) {
1738: const PetscScalar *svals = values + (V ? (bs*bs*(ii[i]+j)) : 0);
1739: MatSetValuesBlocked_SeqSBAIJ(B,1,&i,1,&icols[j],svals,INSERT_VALUES);
1740: }
1741: }
1742: }
1743: if (!V) { PetscFree(values); }
1744: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
1745: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
1746: MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
1747: return(0);
1748: }
1750: /*
1751: This is used to set the numeric factorization for both Cholesky and ICC symbolic factorization
1752: */
1755: PetscErrorCode MatSeqSBAIJSetNumericFactorization_inplace(Mat B,PetscBool natural)
1756: {
1758: PetscBool flg = PETSC_FALSE;
1759: PetscInt bs = B->rmap->bs;
1762: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_no_unroll",&flg,NULL);
1763: if (flg) bs = 8;
1765: if (!natural) {
1766: switch (bs) {
1767: case 1:
1768: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_inplace;
1769: break;
1770: case 2:
1771: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2;
1772: break;
1773: case 3:
1774: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3;
1775: break;
1776: case 4:
1777: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4;
1778: break;
1779: case 5:
1780: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5;
1781: break;
1782: case 6:
1783: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6;
1784: break;
1785: case 7:
1786: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7;
1787: break;
1788: default:
1789: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N;
1790: break;
1791: }
1792: } else {
1793: switch (bs) {
1794: case 1:
1795: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_NaturalOrdering_inplace;
1796: break;
1797: case 2:
1798: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2_NaturalOrdering;
1799: break;
1800: case 3:
1801: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3_NaturalOrdering;
1802: break;
1803: case 4:
1804: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4_NaturalOrdering;
1805: break;
1806: case 5:
1807: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5_NaturalOrdering;
1808: break;
1809: case 6:
1810: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6_NaturalOrdering;
1811: break;
1812: case 7:
1813: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7_NaturalOrdering;
1814: break;
1815: default:
1816: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N_NaturalOrdering;
1817: break;
1818: }
1819: }
1820: return(0);
1821: }
1823: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqAIJ(Mat, MatType,MatReuse,Mat*);
1824: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqBAIJ(Mat, MatType,MatReuse,Mat*);
1828: PETSC_INTERN PetscErrorCode MatGetFactor_seqsbaij_petsc(Mat A,MatFactorType ftype,Mat *B)
1829: {
1830: PetscInt n = A->rmap->n;
1834: #if defined(PETSC_USE_COMPLEX)
1835: if (A->hermitian) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Hermitian Factor is not supported");
1836: #endif
1837: MatCreate(PetscObjectComm((PetscObject)A),B);
1838: MatSetSizes(*B,n,n,n,n);
1839: if (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) {
1840: MatSetType(*B,MATSEQSBAIJ);
1841: MatSeqSBAIJSetPreallocation(*B,A->rmap->bs,MAT_SKIP_ALLOCATION,NULL);
1843: (*B)->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SeqSBAIJ;
1844: (*B)->ops->iccfactorsymbolic = MatICCFactorSymbolic_SeqSBAIJ;
1845: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Factor type not supported");
1847: (*B)->factortype = ftype;
1848: PetscFree((*B)->solvertype);
1849: PetscStrallocpy(MATSOLVERPETSC,&(*B)->solvertype);
1850: return(0);
1851: }
1853: /*MC
1854: MATSEQSBAIJ - MATSEQSBAIJ = "seqsbaij" - A matrix type to be used for sequential symmetric block sparse matrices,
1855: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1857: For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you
1858: can call MatSetOption(Mat, MAT_HERMITIAN); after MatAssemblyEnd()
1860: Options Database Keys:
1861: . -mat_type seqsbaij - sets the matrix type to "seqsbaij" during a call to MatSetFromOptions()
1863: Notes: By default if you insert values into the lower triangular part of the matrix they are simply ignored (since they are not
1864: stored and it is assumed they symmetric to the upper triangular). If you call MatSetOption(Mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_FALSE) or use
1865: the options database -mat_ignore_lower_triangular false it will generate an error if you try to set a value in the lower triangular portion.
1868: Level: beginner
1870: .seealso: MatCreateSeqSBAIJ
1871: M*/
1873: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqSBSTRM(Mat, MatType,MatReuse,Mat*);
1877: PETSC_EXTERN PetscErrorCode MatCreate_SeqSBAIJ(Mat B)
1878: {
1879: Mat_SeqSBAIJ *b;
1881: PetscMPIInt size;
1882: PetscBool no_unroll = PETSC_FALSE,no_inode = PETSC_FALSE;
1885: MPI_Comm_size(PetscObjectComm((PetscObject)B),&size);
1886: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Comm must be of size 1");
1888: PetscNewLog(B,&b);
1889: B->data = (void*)b;
1890: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
1892: B->ops->destroy = MatDestroy_SeqSBAIJ;
1893: B->ops->view = MatView_SeqSBAIJ;
1894: b->row = 0;
1895: b->icol = 0;
1896: b->reallocs = 0;
1897: b->saved_values = 0;
1898: b->inode.limit = 5;
1899: b->inode.max_limit = 5;
1901: b->roworiented = PETSC_TRUE;
1902: b->nonew = 0;
1903: b->diag = 0;
1904: b->solve_work = 0;
1905: b->mult_work = 0;
1906: B->spptr = 0;
1907: B->info.nz_unneeded = (PetscReal)b->maxnz*b->bs2;
1908: b->keepnonzeropattern = PETSC_FALSE;
1910: b->inew = 0;
1911: b->jnew = 0;
1912: b->anew = 0;
1913: b->a2anew = 0;
1914: b->permute = PETSC_FALSE;
1916: b->ignore_ltriangular = PETSC_TRUE;
1918: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_ignore_lower_triangular",&b->ignore_ltriangular,NULL);
1920: b->getrow_utriangular = PETSC_FALSE;
1922: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_getrow_uppertriangular",&b->getrow_utriangular,NULL);
1924: PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_SeqSBAIJ);
1925: PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_SeqSBAIJ);
1926: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetColumnIndices_C",MatSeqSBAIJSetColumnIndices_SeqSBAIJ);
1927: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqaij_C",MatConvert_SeqSBAIJ_SeqAIJ);
1928: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqbaij_C",MatConvert_SeqSBAIJ_SeqBAIJ);
1929: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetPreallocation_C",MatSeqSBAIJSetPreallocation_SeqSBAIJ);
1930: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetPreallocationCSR_C",MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ);
1931: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqsbstrm_C",MatConvert_SeqSBAIJ_SeqSBSTRM);
1932: #if defined(PETSC_HAVE_ELEMENTAL)
1933: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_elemental_C",MatConvert_SeqSBAIJ_Elemental);
1934: #endif
1936: B->symmetric = PETSC_TRUE;
1937: B->structurally_symmetric = PETSC_TRUE;
1938: B->symmetric_set = PETSC_TRUE;
1939: B->structurally_symmetric_set = PETSC_TRUE;
1941: PetscObjectChangeTypeName((PetscObject)B,MATSEQSBAIJ);
1943: PetscOptionsBegin(PetscObjectComm((PetscObject)B),((PetscObject)B)->prefix,"Options for SEQSBAIJ matrix","Mat");
1944: PetscOptionsBool("-mat_no_unroll","Do not optimize for inodes (slower)",NULL,no_unroll,&no_unroll,NULL);
1945: if (no_unroll) {
1946: PetscInfo(B,"Not using Inode routines due to -mat_no_unroll\n");
1947: }
1948: PetscOptionsBool("-mat_no_inode","Do not optimize for inodes (slower)",NULL,no_inode,&no_inode,NULL);
1949: if (no_inode) {
1950: PetscInfo(B,"Not using Inode routines due to -mat_no_inode\n");
1951: }
1952: PetscOptionsInt("-mat_inode_limit","Do not use inodes larger then this value",NULL,b->inode.limit,&b->inode.limit,NULL);
1953: PetscOptionsEnd();
1954: b->inode.use = (PetscBool)(!(no_unroll || no_inode));
1955: if (b->inode.limit > b->inode.max_limit) b->inode.limit = b->inode.max_limit;
1956: return(0);
1957: }
1961: /*@C
1962: MatSeqSBAIJSetPreallocation - Creates a sparse symmetric matrix in block AIJ (block
1963: compressed row) format. For good matrix assembly performance the
1964: user should preallocate the matrix storage by setting the parameter nz
1965: (or the array nnz). By setting these parameters accurately, performance
1966: during matrix assembly can be increased by more than a factor of 50.
1968: Collective on Mat
1970: Input Parameters:
1971: + B - the symmetric matrix
1972: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
1973: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
1974: . nz - number of block nonzeros per block row (same for all rows)
1975: - nnz - array containing the number of block nonzeros in the upper triangular plus
1976: diagonal portion of each block (possibly different for each block row) or NULL
1978: Options Database Keys:
1979: . -mat_no_unroll - uses code that does not unroll the loops in the
1980: block calculations (much slower)
1981: . -mat_block_size - size of the blocks to use (only works if a negative bs is passed in
1983: Level: intermediate
1985: Notes:
1986: Specify the preallocated storage with either nz or nnz (not both).
1987: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
1988: allocation. See Users-Manual: ch_mat for details.
1990: You can call MatGetInfo() to get information on how effective the preallocation was;
1991: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
1992: You can also run with the option -info and look for messages with the string
1993: malloc in them to see if additional memory allocation was needed.
1995: If the nnz parameter is given then the nz parameter is ignored
1998: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatCreateSBAIJ()
1999: @*/
2000: PetscErrorCode MatSeqSBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt nz,const PetscInt nnz[])
2001: {
2008: PetscTryMethod(B,"MatSeqSBAIJSetPreallocation_C",(Mat,PetscInt,PetscInt,const PetscInt[]),(B,bs,nz,nnz));
2009: return(0);
2010: }
2012: #undef __FUNCT__
2014: /*@C
2015: MatSeqSBAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in symmetric block AIJ format.
2017: Input Parameters:
2018: + B - the matrix
2019: . bs - size of block, the blocks are ALWAYS square.
2020: . i - the indices into j for the start of each local row (starts with zero)
2021: . j - the column indices for each local row (starts with zero) these must be sorted for each row
2022: - v - optional values in the matrix
2024: Level: developer
2026: Notes:
2027: The order of the entries in values is specified by the MatOption MAT_ROW_ORIENTED. For example, C programs
2028: may want to use the default MAT_ROW_ORIENTED=PETSC_TRUE and use an array v[nnz][bs][bs] where the second index is
2029: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
2030: MAT_ROW_ORIENTED=PETSC_FALSE and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
2031: block column and the second index is over columns within a block.
2033: .keywords: matrix, block, aij, compressed row, sparse
2035: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValuesBlocked(), MatSeqSBAIJSetPreallocation(), MATSEQSBAIJ
2036: @*/
2037: PetscErrorCode MatSeqSBAIJSetPreallocationCSR(Mat B,PetscInt bs,const PetscInt i[],const PetscInt j[], const PetscScalar v[])
2038: {
2045: PetscTryMethod(B,"MatSeqSBAIJSetPreallocationCSR_C",(Mat,PetscInt,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,bs,i,j,v));
2046: return(0);
2047: }
2051: /*@C
2052: MatCreateSeqSBAIJ - Creates a sparse symmetric matrix in block AIJ (block
2053: compressed row) format. For good matrix assembly performance the
2054: user should preallocate the matrix storage by setting the parameter nz
2055: (or the array nnz). By setting these parameters accurately, performance
2056: during matrix assembly can be increased by more than a factor of 50.
2058: Collective on MPI_Comm
2060: Input Parameters:
2061: + comm - MPI communicator, set to PETSC_COMM_SELF
2062: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
2063: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2064: . m - number of rows, or number of columns
2065: . nz - number of block nonzeros per block row (same for all rows)
2066: - nnz - array containing the number of block nonzeros in the upper triangular plus
2067: diagonal portion of each block (possibly different for each block row) or NULL
2069: Output Parameter:
2070: . A - the symmetric matrix
2072: Options Database Keys:
2073: . -mat_no_unroll - uses code that does not unroll the loops in the
2074: block calculations (much slower)
2075: . -mat_block_size - size of the blocks to use
2077: Level: intermediate
2079: It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(),
2080: MatXXXXSetPreallocation() paradgm instead of this routine directly.
2081: [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation]
2083: Notes:
2084: The number of rows and columns must be divisible by blocksize.
2085: This matrix type does not support complex Hermitian operation.
2087: Specify the preallocated storage with either nz or nnz (not both).
2088: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
2089: allocation. See Users-Manual: ch_mat for details.
2091: If the nnz parameter is given then the nz parameter is ignored
2093: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatCreateSBAIJ()
2094: @*/
2095: PetscErrorCode MatCreateSeqSBAIJ(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt nz,const PetscInt nnz[],Mat *A)
2096: {
2100: MatCreate(comm,A);
2101: MatSetSizes(*A,m,n,m,n);
2102: MatSetType(*A,MATSEQSBAIJ);
2103: MatSeqSBAIJSetPreallocation_SeqSBAIJ(*A,bs,nz,(PetscInt*)nnz);
2104: return(0);
2105: }
2109: PetscErrorCode MatDuplicate_SeqSBAIJ(Mat A,MatDuplicateOption cpvalues,Mat *B)
2110: {
2111: Mat C;
2112: Mat_SeqSBAIJ *c,*a = (Mat_SeqSBAIJ*)A->data;
2114: PetscInt i,mbs = a->mbs,nz = a->nz,bs2 =a->bs2;
2117: if (a->i[mbs] != nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Corrupt matrix");
2119: *B = 0;
2120: MatCreate(PetscObjectComm((PetscObject)A),&C);
2121: MatSetSizes(C,A->rmap->N,A->cmap->n,A->rmap->N,A->cmap->n);
2122: MatSetType(C,MATSEQSBAIJ);
2123: PetscMemcpy(C->ops,A->ops,sizeof(struct _MatOps));
2124: c = (Mat_SeqSBAIJ*)C->data;
2126: C->preallocated = PETSC_TRUE;
2127: C->factortype = A->factortype;
2128: c->row = 0;
2129: c->icol = 0;
2130: c->saved_values = 0;
2131: c->keepnonzeropattern = a->keepnonzeropattern;
2132: C->assembled = PETSC_TRUE;
2134: PetscLayoutReference(A->rmap,&C->rmap);
2135: PetscLayoutReference(A->cmap,&C->cmap);
2136: c->bs2 = a->bs2;
2137: c->mbs = a->mbs;
2138: c->nbs = a->nbs;
2140: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2141: c->imax = a->imax;
2142: c->ilen = a->ilen;
2143: c->free_imax_ilen = PETSC_FALSE;
2144: } else {
2145: PetscMalloc2((mbs+1),&c->imax,(mbs+1),&c->ilen);
2146: PetscLogObjectMemory((PetscObject)C,2*(mbs+1)*sizeof(PetscInt));
2147: for (i=0; i<mbs; i++) {
2148: c->imax[i] = a->imax[i];
2149: c->ilen[i] = a->ilen[i];
2150: }
2151: c->free_imax_ilen = PETSC_TRUE;
2152: }
2154: /* allocate the matrix space */
2155: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2156: PetscMalloc1(bs2*nz,&c->a);
2157: PetscLogObjectMemory((PetscObject)C,nz*bs2*sizeof(MatScalar));
2158: c->i = a->i;
2159: c->j = a->j;
2160: c->singlemalloc = PETSC_FALSE;
2161: c->free_a = PETSC_TRUE;
2162: c->free_ij = PETSC_FALSE;
2163: c->parent = A;
2164: PetscObjectReference((PetscObject)A);
2165: MatSetOption(A,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
2166: MatSetOption(C,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
2167: } else {
2168: PetscMalloc3(bs2*nz,&c->a,nz,&c->j,mbs+1,&c->i);
2169: PetscMemcpy(c->i,a->i,(mbs+1)*sizeof(PetscInt));
2170: PetscLogObjectMemory((PetscObject)C,(mbs+1)*sizeof(PetscInt) + nz*(bs2*sizeof(MatScalar) + sizeof(PetscInt)));
2171: c->singlemalloc = PETSC_TRUE;
2172: c->free_a = PETSC_TRUE;
2173: c->free_ij = PETSC_TRUE;
2174: }
2175: if (mbs > 0) {
2176: if (cpvalues != MAT_SHARE_NONZERO_PATTERN) {
2177: PetscMemcpy(c->j,a->j,nz*sizeof(PetscInt));
2178: }
2179: if (cpvalues == MAT_COPY_VALUES) {
2180: PetscMemcpy(c->a,a->a,bs2*nz*sizeof(MatScalar));
2181: } else {
2182: PetscMemzero(c->a,bs2*nz*sizeof(MatScalar));
2183: }
2184: if (a->jshort) {
2185: /* cannot share jshort, it is reallocated in MatAssemblyEnd_SeqSBAIJ() */
2186: /* if the parent matrix is reassembled, this child matrix will never notice */
2187: PetscMalloc1(nz,&c->jshort);
2188: PetscLogObjectMemory((PetscObject)C,nz*sizeof(unsigned short));
2189: PetscMemcpy(c->jshort,a->jshort,nz*sizeof(unsigned short));
2191: c->free_jshort = PETSC_TRUE;
2192: }
2193: }
2195: c->roworiented = a->roworiented;
2196: c->nonew = a->nonew;
2198: if (a->diag) {
2199: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2200: c->diag = a->diag;
2201: c->free_diag = PETSC_FALSE;
2202: } else {
2203: PetscMalloc1(mbs,&c->diag);
2204: PetscLogObjectMemory((PetscObject)C,mbs*sizeof(PetscInt));
2205: for (i=0; i<mbs; i++) c->diag[i] = a->diag[i];
2206: c->free_diag = PETSC_TRUE;
2207: }
2208: }
2209: c->nz = a->nz;
2210: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
2211: c->solve_work = 0;
2212: c->mult_work = 0;
2214: *B = C;
2215: PetscFunctionListDuplicate(((PetscObject)A)->qlist,&((PetscObject)C)->qlist);
2216: return(0);
2217: }
2221: PetscErrorCode MatLoad_SeqSBAIJ(Mat newmat,PetscViewer viewer)
2222: {
2223: Mat_SeqSBAIJ *a;
2225: int fd;
2226: PetscMPIInt size;
2227: PetscInt i,nz,header[4],*rowlengths=0,M,N,bs = newmat->rmap->bs;
2228: PetscInt *mask,mbs,*jj,j,rowcount,nzcount,k,*s_browlengths,maskcount;
2229: PetscInt kmax,jcount,block,idx,point,nzcountb,extra_rows,rows,cols;
2230: PetscInt *masked,nmask,tmp,bs2,ishift;
2231: PetscScalar *aa;
2232: MPI_Comm comm;
2235: /* force binary viewer to load .info file if it has not yet done so */
2236: PetscViewerSetUp(viewer);
2237: PetscObjectGetComm((PetscObject)viewer,&comm);
2238: PetscOptionsGetInt(((PetscObject)newmat)->options,((PetscObject)newmat)->prefix,"-matload_block_size",&bs,NULL);
2239: if (bs < 0) bs = 1;
2240: bs2 = bs*bs;
2242: MPI_Comm_size(comm,&size);
2243: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"view must have one processor");
2244: PetscViewerBinaryGetDescriptor(viewer,&fd);
2245: PetscBinaryRead(fd,header,4,PETSC_INT);
2246: if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not Mat object");
2247: M = header[1]; N = header[2]; nz = header[3];
2249: if (header[3] < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as SeqSBAIJ");
2251: if (M != N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only do square matrices");
2253: /*
2254: This code adds extra rows to make sure the number of rows is
2255: divisible by the blocksize
2256: */
2257: mbs = M/bs;
2258: extra_rows = bs - M + bs*(mbs);
2259: if (extra_rows == bs) extra_rows = 0;
2260: else mbs++;
2261: if (extra_rows) {
2262: PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");
2263: }
2265: /* Set global sizes if not already set */
2266: if (newmat->rmap->n < 0 && newmat->rmap->N < 0 && newmat->cmap->n < 0 && newmat->cmap->N < 0) {
2267: MatSetSizes(newmat,PETSC_DECIDE,PETSC_DECIDE,M+extra_rows,N+extra_rows);
2268: } else { /* Check if the matrix global sizes are correct */
2269: MatGetSize(newmat,&rows,&cols);
2270: 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);
2271: }
2273: /* read in row lengths */
2274: PetscMalloc1(M+extra_rows,&rowlengths);
2275: PetscBinaryRead(fd,rowlengths,M,PETSC_INT);
2276: for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1;
2278: /* read in column indices */
2279: PetscMalloc1(nz+extra_rows,&jj);
2280: PetscBinaryRead(fd,jj,nz,PETSC_INT);
2281: for (i=0; i<extra_rows; i++) jj[nz+i] = M+i;
2283: /* loop over row lengths determining block row lengths */
2284: PetscCalloc1(mbs,&s_browlengths);
2285: PetscMalloc2(mbs,&mask,mbs,&masked);
2286: PetscMemzero(mask,mbs*sizeof(PetscInt));
2287: rowcount = 0;
2288: nzcount = 0;
2289: for (i=0; i<mbs; i++) {
2290: nmask = 0;
2291: for (j=0; j<bs; j++) {
2292: kmax = rowlengths[rowcount];
2293: for (k=0; k<kmax; k++) {
2294: tmp = jj[nzcount++]/bs; /* block col. index */
2295: if (!mask[tmp] && tmp >= i) {masked[nmask++] = tmp; mask[tmp] = 1;}
2296: }
2297: rowcount++;
2298: }
2299: s_browlengths[i] += nmask;
2301: /* zero out the mask elements we set */
2302: for (j=0; j<nmask; j++) mask[masked[j]] = 0;
2303: }
2305: /* Do preallocation */
2306: MatSeqSBAIJSetPreallocation_SeqSBAIJ(newmat,bs,0,s_browlengths);
2307: a = (Mat_SeqSBAIJ*)newmat->data;
2309: /* set matrix "i" values */
2310: a->i[0] = 0;
2311: for (i=1; i<= mbs; i++) {
2312: a->i[i] = a->i[i-1] + s_browlengths[i-1];
2313: a->ilen[i-1] = s_browlengths[i-1];
2314: }
2315: a->nz = a->i[mbs];
2317: /* read in nonzero values */
2318: PetscMalloc1(nz+extra_rows,&aa);
2319: PetscBinaryRead(fd,aa,nz,PETSC_SCALAR);
2320: for (i=0; i<extra_rows; i++) aa[nz+i] = 1.0;
2322: /* set "a" and "j" values into matrix */
2323: nzcount = 0; jcount = 0;
2324: for (i=0; i<mbs; i++) {
2325: nzcountb = nzcount;
2326: nmask = 0;
2327: for (j=0; j<bs; j++) {
2328: kmax = rowlengths[i*bs+j];
2329: for (k=0; k<kmax; k++) {
2330: tmp = jj[nzcount++]/bs; /* block col. index */
2331: if (!mask[tmp] && tmp >= i) { masked[nmask++] = tmp; mask[tmp] = 1;}
2332: }
2333: }
2334: /* sort the masked values */
2335: PetscSortInt(nmask,masked);
2337: /* set "j" values into matrix */
2338: maskcount = 1;
2339: for (j=0; j<nmask; j++) {
2340: a->j[jcount++] = masked[j];
2341: mask[masked[j]] = maskcount++;
2342: }
2344: /* set "a" values into matrix */
2345: ishift = bs2*a->i[i];
2346: for (j=0; j<bs; j++) {
2347: kmax = rowlengths[i*bs+j];
2348: for (k=0; k<kmax; k++) {
2349: tmp = jj[nzcountb]/bs; /* block col. index */
2350: if (tmp >= i) {
2351: block = mask[tmp] - 1;
2352: point = jj[nzcountb] - bs*tmp;
2353: idx = ishift + bs2*block + j + bs*point;
2354: a->a[idx] = aa[nzcountb];
2355: }
2356: nzcountb++;
2357: }
2358: }
2359: /* zero out the mask elements we set */
2360: for (j=0; j<nmask; j++) mask[masked[j]] = 0;
2361: }
2362: if (jcount != a->nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Bad binary matrix");
2364: PetscFree(rowlengths);
2365: PetscFree(s_browlengths);
2366: PetscFree(aa);
2367: PetscFree(jj);
2368: PetscFree2(mask,masked);
2370: MatAssemblyBegin(newmat,MAT_FINAL_ASSEMBLY);
2371: MatAssemblyEnd(newmat,MAT_FINAL_ASSEMBLY);
2372: return(0);
2373: }
2377: /*@
2378: MatCreateSeqSBAIJWithArrays - Creates an sequential SBAIJ matrix using matrix elements
2379: (upper triangular entries in CSR format) provided by the user.
2381: Collective on MPI_Comm
2383: Input Parameters:
2384: + comm - must be an MPI communicator of size 1
2385: . bs - size of block
2386: . m - number of rows
2387: . n - number of columns
2388: . i - row indices
2389: . j - column indices
2390: - a - matrix values
2392: Output Parameter:
2393: . mat - the matrix
2395: Level: advanced
2397: Notes:
2398: The i, j, and a arrays are not copied by this routine, the user must free these arrays
2399: once the matrix is destroyed
2401: You cannot set new nonzero locations into this matrix, that will generate an error.
2403: The i and j indices are 0 based
2405: When block size is greater than 1 the matrix values must be stored using the SBAIJ storage format (see the SBAIJ code to determine this). For block size of 1
2406: it is the regular CSR format excluding the lower triangular elements.
2408: .seealso: MatCreate(), MatCreateSBAIJ(), MatCreateSeqSBAIJ()
2410: @*/
2411: PetscErrorCode MatCreateSeqSBAIJWithArrays(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt *i,PetscInt *j,PetscScalar *a,Mat *mat)
2412: {
2414: PetscInt ii;
2415: Mat_SeqSBAIJ *sbaij;
2418: if (bs != 1) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"block size %D > 1 is not supported yet",bs);
2419: if (i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0");
2421: MatCreate(comm,mat);
2422: MatSetSizes(*mat,m,n,m,n);
2423: MatSetType(*mat,MATSEQSBAIJ);
2424: MatSeqSBAIJSetPreallocation_SeqSBAIJ(*mat,bs,MAT_SKIP_ALLOCATION,0);
2425: sbaij = (Mat_SeqSBAIJ*)(*mat)->data;
2426: PetscMalloc2(m,&sbaij->imax,m,&sbaij->ilen);
2427: PetscLogObjectMemory((PetscObject)*mat,2*m*sizeof(PetscInt));
2429: sbaij->i = i;
2430: sbaij->j = j;
2431: sbaij->a = a;
2433: sbaij->singlemalloc = PETSC_FALSE;
2434: sbaij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
2435: sbaij->free_a = PETSC_FALSE;
2436: sbaij->free_ij = PETSC_FALSE;
2438: for (ii=0; ii<m; ii++) {
2439: sbaij->ilen[ii] = sbaij->imax[ii] = i[ii+1] - i[ii];
2440: #if defined(PETSC_USE_DEBUG)
2441: 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]);
2442: #endif
2443: }
2444: #if defined(PETSC_USE_DEBUG)
2445: for (ii=0; ii<sbaij->i[m]; ii++) {
2446: if (j[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column index at location = %d index = %d",ii,j[ii]);
2447: 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]);
2448: }
2449: #endif
2451: MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);
2452: MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);
2453: return(0);
2454: }
2458: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ(MPI_Comm comm,Mat inmat,PetscInt n,MatReuse scall,Mat *outmat)
2459: {
2463: MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(comm,inmat,n,scall,outmat);
2464: return(0);
2465: }