Actual source code: sbaij.c
petsc-3.7.7 2017-09-25
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;
82: PetscInt i,j,n = a->mbs,nz = a->i[n],*tia,*tja,bs = A->rmap->bs,k,l,cnt;
83: PetscInt **ia = (PetscInt**)inia,**ja = (PetscInt**)inja;
86: *nn = n;
87: if (!ia) return(0);
88: if (symmetric) {
89: MatToSymmetricIJ_SeqAIJ(n,a->i,a->j,PETSC_FALSE,0,0,&tia,&tja);
90: nz = tia[n];
91: } else {
92: tia = a->i; tja = a->j;
93: }
95: if (!blockcompressed && bs > 1) {
96: (*nn) *= bs;
97: /* malloc & create the natural set of indices */
98: PetscMalloc1((n+1)*bs,ia);
99: if (n) {
100: (*ia)[0] = oshift;
101: for (j=1; j<bs; j++) {
102: (*ia)[j] = (tia[1]-tia[0])*bs+(*ia)[j-1];
103: }
104: }
106: for (i=1; i<n; i++) {
107: (*ia)[i*bs] = (tia[i]-tia[i-1])*bs + (*ia)[i*bs-1];
108: for (j=1; j<bs; j++) {
109: (*ia)[i*bs+j] = (tia[i+1]-tia[i])*bs + (*ia)[i*bs+j-1];
110: }
111: }
112: if (n) {
113: (*ia)[n*bs] = (tia[n]-tia[n-1])*bs + (*ia)[n*bs-1];
114: }
116: if (inja) {
117: PetscMalloc1(nz*bs*bs,ja);
118: cnt = 0;
119: for (i=0; i<n; i++) {
120: for (j=0; j<bs; j++) {
121: for (k=tia[i]; k<tia[i+1]; k++) {
122: for (l=0; l<bs; l++) {
123: (*ja)[cnt++] = bs*tja[k] + l;
124: }
125: }
126: }
127: }
128: }
130: if (symmetric) { /* deallocate memory allocated in MatToSymmetricIJ_SeqAIJ() */
131: PetscFree(tia);
132: PetscFree(tja);
133: }
134: } else if (oshift == 1) {
135: if (symmetric) {
136: nz = tia[A->rmap->n/bs];
137: /* add 1 to i and j indices */
138: for (i=0; i<A->rmap->n/bs+1; i++) tia[i] = tia[i] + 1;
139: *ia = tia;
140: if (ja) {
141: for (i=0; i<nz; i++) tja[i] = tja[i] + 1;
142: *ja = tja;
143: }
144: } else {
145: nz = a->i[A->rmap->n/bs];
146: /* malloc space and add 1 to i and j indices */
147: PetscMalloc1(A->rmap->n/bs+1,ia);
148: for (i=0; i<A->rmap->n/bs+1; i++) (*ia)[i] = a->i[i] + 1;
149: if (ja) {
150: PetscMalloc1(nz,ja);
151: for (i=0; i<nz; i++) (*ja)[i] = a->j[i] + 1;
152: }
153: }
154: } else {
155: *ia = tia;
156: if (ja) *ja = tja;
157: }
158: return(0);
159: }
163: static PetscErrorCode MatRestoreRowIJ_SeqSBAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool blockcompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done)
164: {
168: if (!ia) return(0);
169: if ((!blockcompressed && A->rmap->bs > 1) || (symmetric || oshift == 1)) {
170: PetscFree(*ia);
171: if (ja) {PetscFree(*ja);}
172: }
173: return(0);
174: }
178: PetscErrorCode MatDestroy_SeqSBAIJ(Mat A)
179: {
180: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
184: #if defined(PETSC_USE_LOG)
185: PetscLogObjectState((PetscObject)A,"Rows=%D, NZ=%D",A->rmap->N,a->nz);
186: #endif
187: MatSeqXAIJFreeAIJ(A,&a->a,&a->j,&a->i);
188: if (a->free_diag) {PetscFree(a->diag);}
189: ISDestroy(&a->row);
190: ISDestroy(&a->col);
191: ISDestroy(&a->icol);
192: PetscFree(a->idiag);
193: PetscFree(a->inode.size);
194: if (a->free_imax_ilen) {PetscFree2(a->imax,a->ilen);}
195: PetscFree(a->solve_work);
196: PetscFree(a->sor_work);
197: PetscFree(a->solves_work);
198: PetscFree(a->mult_work);
199: PetscFree(a->saved_values);
200: if (a->free_jshort) {PetscFree(a->jshort);}
201: PetscFree(a->inew);
202: MatDestroy(&a->parent);
203: PetscFree(A->data);
205: PetscObjectChangeTypeName((PetscObject)A,0);
206: PetscObjectComposeFunction((PetscObject)A,"MatStoreValues_C",NULL);
207: PetscObjectComposeFunction((PetscObject)A,"MatRetrieveValues_C",NULL);
208: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetColumnIndices_C",NULL);
209: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqaij_C",NULL);
210: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqbaij_C",NULL);
211: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetPreallocation_C",NULL);
212: PetscObjectComposeFunction((PetscObject)A,"MatSeqSBAIJSetPreallocationCSR_C",NULL);
213: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_seqsbstrm_C",NULL);
214: #if defined(PETSC_HAVE_ELEMENTAL)
215: PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqsbaij_elemental_C",NULL);
216: #endif
217: return(0);
218: }
222: PetscErrorCode MatSetOption_SeqSBAIJ(Mat A,MatOption op,PetscBool flg)
223: {
224: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
228: switch (op) {
229: case MAT_ROW_ORIENTED:
230: a->roworiented = flg;
231: break;
232: case MAT_KEEP_NONZERO_PATTERN:
233: a->keepnonzeropattern = flg;
234: break;
235: case MAT_NEW_NONZERO_LOCATIONS:
236: a->nonew = (flg ? 0 : 1);
237: break;
238: case MAT_NEW_NONZERO_LOCATION_ERR:
239: a->nonew = (flg ? -1 : 0);
240: break;
241: case MAT_NEW_NONZERO_ALLOCATION_ERR:
242: a->nonew = (flg ? -2 : 0);
243: break;
244: case MAT_UNUSED_NONZERO_LOCATION_ERR:
245: a->nounused = (flg ? -1 : 0);
246: break;
247: case MAT_NEW_DIAGONALS:
248: case MAT_IGNORE_OFF_PROC_ENTRIES:
249: case MAT_USE_HASH_TABLE:
250: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
251: break;
252: case MAT_HERMITIAN:
253: if (!A->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Must call MatAssemblyEnd() first");
254: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
255: A->ops->mult = MatMult_SeqSBAIJ_1_Hermitian_ushort;
256: } else if (A->cmap->bs == 1) {
257: A->ops->mult = MatMult_SeqSBAIJ_1_Hermitian;
258: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"No support for Hermitian with block size greater than 1");
259: break;
260: case MAT_SPD:
261: /* These options are handled directly by MatSetOption() */
262: break;
263: case MAT_SYMMETRIC:
264: case MAT_STRUCTURALLY_SYMMETRIC:
265: case MAT_SYMMETRY_ETERNAL:
266: /* These options are handled directly by MatSetOption() */
267: break;
268: case MAT_IGNORE_LOWER_TRIANGULAR:
269: a->ignore_ltriangular = flg;
270: break;
271: case MAT_ERROR_LOWER_TRIANGULAR:
272: a->ignore_ltriangular = flg;
273: break;
274: case MAT_GETROW_UPPERTRIANGULAR:
275: a->getrow_utriangular = flg;
276: break;
277: default:
278: SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"unknown option %d",op);
279: }
280: return(0);
281: }
285: PetscErrorCode MatGetRow_SeqSBAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
286: {
287: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
291: 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()");
293: /* Get the upper triangular part of the row */
294: MatGetRow_SeqBAIJ_private(A,row,nz,idx,v,a->i,a->j,a->a);
295: return(0);
296: }
300: PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
301: {
305: if (idx) {PetscFree(*idx);}
306: if (v) {PetscFree(*v);}
307: return(0);
308: }
312: PetscErrorCode MatGetRowUpperTriangular_SeqSBAIJ(Mat A)
313: {
314: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
317: a->getrow_utriangular = PETSC_TRUE;
318: return(0);
319: }
322: PetscErrorCode MatRestoreRowUpperTriangular_SeqSBAIJ(Mat A)
323: {
324: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
327: a->getrow_utriangular = PETSC_FALSE;
328: return(0);
329: }
333: PetscErrorCode MatTranspose_SeqSBAIJ(Mat A,MatReuse reuse,Mat *B)
334: {
338: if (reuse == MAT_INITIAL_MATRIX || *B != A) {
339: MatDuplicate(A,MAT_COPY_VALUES,B);
340: }
341: return(0);
342: }
346: PetscErrorCode MatView_SeqSBAIJ_ASCII(Mat A,PetscViewer viewer)
347: {
348: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
349: PetscErrorCode ierr;
350: PetscInt i,j,bs = A->rmap->bs,k,l,bs2=a->bs2;
351: PetscViewerFormat format;
352: PetscInt *diag;
355: PetscViewerGetFormat(viewer,&format);
356: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
357: PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);
358: } else if (format == PETSC_VIEWER_ASCII_MATLAB) {
359: Mat aij;
360: const char *matname;
362: if (A->factortype && bs>1) {
363: PetscPrintf(PETSC_COMM_SELF,"Warning: matrix is factored with bs>1. MatView() with PETSC_VIEWER_ASCII_MATLAB is not supported and ignored!\n");
364: return(0);
365: }
366: MatConvert(A,MATSEQAIJ,MAT_INITIAL_MATRIX,&aij);
367: PetscObjectGetName((PetscObject)A,&matname);
368: PetscObjectSetName((PetscObject)aij,matname);
369: MatView(aij,viewer);
370: MatDestroy(&aij);
371: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
372: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
373: for (i=0; i<a->mbs; i++) {
374: for (j=0; j<bs; j++) {
375: PetscViewerASCIIPrintf(viewer,"row %D:",i*bs+j);
376: for (k=a->i[i]; k<a->i[i+1]; k++) {
377: for (l=0; l<bs; l++) {
378: #if defined(PETSC_USE_COMPLEX)
379: if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) > 0.0 && PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
380: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k]+l,
381: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
382: } else if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) < 0.0 && PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
383: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k]+l,
384: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),-(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
385: } else if (PetscRealPart(a->a[bs2*k + l*bs + j]) != 0.0) {
386: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)PetscRealPart(a->a[bs2*k + l*bs + j]));
387: }
388: #else
389: if (a->a[bs2*k + l*bs + j] != 0.0) {
390: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)a->a[bs2*k + l*bs + j]);
391: }
392: #endif
393: }
394: }
395: PetscViewerASCIIPrintf(viewer,"\n");
396: }
397: }
398: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
399: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
400: return(0);
401: } else {
402: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
403: if (A->factortype) { /* for factored matrix */
404: if (bs>1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"matrix is factored with bs>1. Not implemented yet");
406: diag=a->diag;
407: for (i=0; i<a->mbs; i++) { /* for row block i */
408: PetscViewerASCIIPrintf(viewer,"row %D:",i);
409: /* diagonal entry */
410: #if defined(PETSC_USE_COMPLEX)
411: if (PetscImaginaryPart(a->a[diag[i]]) > 0.0) {
412: 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]]));
413: } else if (PetscImaginaryPart(a->a[diag[i]]) < 0.0) {
414: 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]]));
415: } else {
416: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[diag[i]],(double)PetscRealPart(1.0/a->a[diag[i]]));
417: }
418: #else
419: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[diag[i]],(double)(1.0/a->a[diag[i]]));
420: #endif
421: /* off-diagonal entries */
422: for (k=a->i[i]; k<a->i[i+1]-1; k++) {
423: #if defined(PETSC_USE_COMPLEX)
424: if (PetscImaginaryPart(a->a[k]) > 0.0) {
425: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k],(double)PetscRealPart(a->a[k]),(double)PetscImaginaryPart(a->a[k]));
426: } else if (PetscImaginaryPart(a->a[k]) < 0.0) {
427: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k],(double)PetscRealPart(a->a[k]),-(double)PetscImaginaryPart(a->a[k]));
428: } else {
429: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k],(double)PetscRealPart(a->a[k]));
430: }
431: #else
432: PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[k],(double)a->a[k]);
433: #endif
434: }
435: PetscViewerASCIIPrintf(viewer,"\n");
436: }
438: } else { /* for non-factored matrix */
439: for (i=0; i<a->mbs; i++) { /* for row block i */
440: for (j=0; j<bs; j++) { /* for row bs*i + j */
441: PetscViewerASCIIPrintf(viewer,"row %D:",i*bs+j);
442: for (k=a->i[i]; k<a->i[i+1]; k++) { /* for column block */
443: for (l=0; l<bs; l++) { /* for column */
444: #if defined(PETSC_USE_COMPLEX)
445: if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) > 0.0) {
446: PetscViewerASCIIPrintf(viewer," (%D, %g + %g i) ",bs*a->j[k]+l,
447: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
448: } else if (PetscImaginaryPart(a->a[bs2*k + l*bs + j]) < 0.0) {
449: PetscViewerASCIIPrintf(viewer," (%D, %g - %g i) ",bs*a->j[k]+l,
450: (double)PetscRealPart(a->a[bs2*k + l*bs + j]),-(double)PetscImaginaryPart(a->a[bs2*k + l*bs + j]));
451: } else {
452: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)PetscRealPart(a->a[bs2*k + l*bs + j]));
453: }
454: #else
455: PetscViewerASCIIPrintf(viewer," (%D, %g) ",bs*a->j[k]+l,(double)a->a[bs2*k + l*bs + j]);
456: #endif
457: }
458: }
459: PetscViewerASCIIPrintf(viewer,"\n");
460: }
461: }
462: }
463: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
464: }
465: PetscViewerFlush(viewer);
466: return(0);
467: }
469: #include <petscdraw.h>
472: static PetscErrorCode MatView_SeqSBAIJ_Draw_Zoom(PetscDraw draw,void *Aa)
473: {
474: Mat A = (Mat) Aa;
475: Mat_SeqSBAIJ *a=(Mat_SeqSBAIJ*)A->data;
477: PetscInt row,i,j,k,l,mbs=a->mbs,color,bs=A->rmap->bs,bs2=a->bs2;
478: PetscReal xl,yl,xr,yr,x_l,x_r,y_l,y_r;
479: MatScalar *aa;
480: PetscViewer viewer;
483: PetscObjectQuery((PetscObject)A,"Zoomviewer",(PetscObject*)&viewer);
484: PetscDrawGetCoordinates(draw,&xl,&yl,&xr,&yr);
486: /* loop over matrix elements drawing boxes */
488: PetscDrawCollectiveBegin(draw);
489: PetscDrawString(draw, .3*(xl+xr), .3*(yl+yr), PETSC_DRAW_BLACK, "symmetric");
490: /* Blue for negative, Cyan for zero and Red for positive */
491: color = PETSC_DRAW_BLUE;
492: for (i=0,row=0; i<mbs; i++,row+=bs) {
493: for (j=a->i[i]; j<a->i[i+1]; j++) {
494: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
495: x_l = a->j[j]*bs; x_r = x_l + 1.0;
496: aa = a->a + j*bs2;
497: for (k=0; k<bs; k++) {
498: for (l=0; l<bs; l++) {
499: if (PetscRealPart(*aa++) >= 0.) continue;
500: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
501: }
502: }
503: }
504: }
505: color = PETSC_DRAW_CYAN;
506: for (i=0,row=0; i<mbs; i++,row+=bs) {
507: for (j=a->i[i]; j<a->i[i+1]; j++) {
508: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
509: x_l = a->j[j]*bs; x_r = x_l + 1.0;
510: aa = a->a + j*bs2;
511: for (k=0; k<bs; k++) {
512: for (l=0; l<bs; l++) {
513: if (PetscRealPart(*aa++) != 0.) continue;
514: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
515: }
516: }
517: }
518: }
519: color = PETSC_DRAW_RED;
520: for (i=0,row=0; i<mbs; i++,row+=bs) {
521: for (j=a->i[i]; j<a->i[i+1]; j++) {
522: y_l = A->rmap->N - row - 1.0; y_r = y_l + 1.0;
523: x_l = a->j[j]*bs; x_r = x_l + 1.0;
524: aa = a->a + j*bs2;
525: for (k=0; k<bs; k++) {
526: for (l=0; l<bs; l++) {
527: if (PetscRealPart(*aa++) <= 0.) continue;
528: PetscDrawRectangle(draw,x_l+k,y_l-l,x_r+k,y_r-l,color,color,color,color);
529: }
530: }
531: }
532: }
533: PetscDrawCollectiveEnd(draw);
534: return(0);
535: }
539: static PetscErrorCode MatView_SeqSBAIJ_Draw(Mat A,PetscViewer viewer)
540: {
542: PetscReal xl,yl,xr,yr,w,h;
543: PetscDraw draw;
544: PetscBool isnull;
547: PetscViewerDrawGetDraw(viewer,0,&draw);
548: PetscDrawIsNull(draw,&isnull);
549: if (isnull) return(0);
551: xr = A->rmap->N; yr = A->rmap->N; h = yr/10.0; w = xr/10.0;
552: xr += w; yr += h; xl = -w; yl = -h;
553: PetscDrawSetCoordinates(draw,xl,yl,xr,yr);
554: PetscObjectCompose((PetscObject)A,"Zoomviewer",(PetscObject)viewer);
555: PetscDrawZoom(draw,MatView_SeqSBAIJ_Draw_Zoom,A);
556: PetscObjectCompose((PetscObject)A,"Zoomviewer",NULL);
557: PetscDrawSave(draw);
558: return(0);
559: }
563: PetscErrorCode MatView_SeqSBAIJ(Mat A,PetscViewer viewer)
564: {
566: PetscBool iascii,isdraw;
567: FILE *file = 0;
570: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
571: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
572: if (iascii) {
573: MatView_SeqSBAIJ_ASCII(A,viewer);
574: } else if (isdraw) {
575: MatView_SeqSBAIJ_Draw(A,viewer);
576: } else {
577: Mat B;
578: const char *matname;
579: MatConvert(A,MATSEQAIJ,MAT_INITIAL_MATRIX,&B);
580: PetscObjectGetName((PetscObject)A,&matname);
581: PetscObjectSetName((PetscObject)B,matname);
582: MatView(B,viewer);
583: MatDestroy(&B);
584: PetscViewerBinaryGetInfoPointer(viewer,&file);
585: if (file) {
586: fprintf(file,"-matload_block_size %d\n",(int)A->rmap->bs);
587: }
588: }
589: return(0);
590: }
595: PetscErrorCode MatGetValues_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],PetscScalar v[])
596: {
597: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
598: PetscInt *rp,k,low,high,t,row,nrow,i,col,l,*aj = a->j;
599: PetscInt *ai = a->i,*ailen = a->ilen;
600: PetscInt brow,bcol,ridx,cidx,bs=A->rmap->bs,bs2=a->bs2;
601: MatScalar *ap,*aa = a->a;
604: for (k=0; k<m; k++) { /* loop over rows */
605: row = im[k]; brow = row/bs;
606: if (row < 0) {v += n; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",row); */
607: 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);
608: rp = aj + ai[brow]; ap = aa + bs2*ai[brow];
609: nrow = ailen[brow];
610: for (l=0; l<n; l++) { /* loop over columns */
611: if (in[l] < 0) {v++; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",in[l]); */
612: 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);
613: col = in[l];
614: bcol = col/bs;
615: cidx = col%bs;
616: ridx = row%bs;
617: high = nrow;
618: low = 0; /* assume unsorted */
619: while (high-low > 5) {
620: t = (low+high)/2;
621: if (rp[t] > bcol) high = t;
622: else low = t;
623: }
624: for (i=low; i<high; i++) {
625: if (rp[i] > bcol) break;
626: if (rp[i] == bcol) {
627: *v++ = ap[bs2*i+bs*cidx+ridx];
628: goto finished;
629: }
630: }
631: *v++ = 0.0;
632: finished:;
633: }
634: }
635: return(0);
636: }
641: PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
642: {
643: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
644: PetscErrorCode ierr;
645: PetscInt *rp,k,low,high,t,ii,jj,row,nrow,i,col,l,rmax,N,lastcol = -1;
646: PetscInt *imax =a->imax,*ai=a->i,*ailen=a->ilen;
647: PetscInt *aj =a->j,nonew=a->nonew,bs2=a->bs2,bs=A->rmap->bs,stepval;
648: PetscBool roworiented=a->roworiented;
649: const PetscScalar *value = v;
650: MatScalar *ap,*aa = a->a,*bap;
653: if (roworiented) stepval = (n-1)*bs;
654: else stepval = (m-1)*bs;
656: for (k=0; k<m; k++) { /* loop over added rows */
657: row = im[k];
658: if (row < 0) continue;
659: #if defined(PETSC_USE_DEBUG)
660: if (row >= a->mbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block index row too large %D max %D",row,a->mbs-1);
661: #endif
662: rp = aj + ai[row];
663: ap = aa + bs2*ai[row];
664: rmax = imax[row];
665: nrow = ailen[row];
666: low = 0;
667: high = nrow;
668: for (l=0; l<n; l++) { /* loop over added columns */
669: if (in[l] < 0) continue;
670: col = in[l];
671: #if defined(PETSC_USE_DEBUG)
672: if (col >= a->nbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block index column too large %D max %D",col,a->nbs-1);
673: #endif
674: if (col < row) {
675: if (a->ignore_ltriangular) continue; /* ignore lower triangular block */
676: 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)");
677: }
678: if (roworiented) value = v + k*(stepval+bs)*bs + l*bs;
679: else value = v + l*(stepval+bs)*bs + k*bs;
681: if (col <= lastcol) low = 0;
682: else high = nrow;
684: lastcol = col;
685: while (high-low > 7) {
686: t = (low+high)/2;
687: if (rp[t] > col) high = t;
688: else low = t;
689: }
690: for (i=low; i<high; i++) {
691: if (rp[i] > col) break;
692: if (rp[i] == col) {
693: bap = ap + bs2*i;
694: if (roworiented) {
695: if (is == ADD_VALUES) {
696: for (ii=0; ii<bs; ii++,value+=stepval) {
697: for (jj=ii; jj<bs2; jj+=bs) {
698: bap[jj] += *value++;
699: }
700: }
701: } else {
702: for (ii=0; ii<bs; ii++,value+=stepval) {
703: for (jj=ii; jj<bs2; jj+=bs) {
704: bap[jj] = *value++;
705: }
706: }
707: }
708: } else {
709: if (is == ADD_VALUES) {
710: for (ii=0; ii<bs; ii++,value+=stepval) {
711: for (jj=0; jj<bs; jj++) {
712: *bap++ += *value++;
713: }
714: }
715: } else {
716: for (ii=0; ii<bs; ii++,value+=stepval) {
717: for (jj=0; jj<bs; jj++) {
718: *bap++ = *value++;
719: }
720: }
721: }
722: }
723: goto noinsert2;
724: }
725: }
726: if (nonew == 1) goto noinsert2;
727: 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);
728: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
729: N = nrow++ - 1; high++;
730: /* shift up all the later entries in this row */
731: for (ii=N; ii>=i; ii--) {
732: rp[ii+1] = rp[ii];
733: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));
734: }
735: if (N >= i) {
736: PetscMemzero(ap+bs2*i,bs2*sizeof(MatScalar));
737: }
738: rp[i] = col;
739: bap = ap + bs2*i;
740: if (roworiented) {
741: for (ii=0; ii<bs; ii++,value+=stepval) {
742: for (jj=ii; jj<bs2; jj+=bs) {
743: bap[jj] = *value++;
744: }
745: }
746: } else {
747: for (ii=0; ii<bs; ii++,value+=stepval) {
748: for (jj=0; jj<bs; jj++) {
749: *bap++ = *value++;
750: }
751: }
752: }
753: noinsert2:;
754: low = i;
755: }
756: ailen[row] = nrow;
757: }
758: return(0);
759: }
761: /*
762: This is not yet used
763: */
766: PetscErrorCode MatAssemblyEnd_SeqSBAIJ_SeqAIJ_Inode(Mat A)
767: {
768: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
770: const PetscInt *ai = a->i, *aj = a->j,*cols;
771: PetscInt i = 0,j,blk_size,m = A->rmap->n,node_count = 0,nzx,nzy,*ns,row,nz,cnt,cnt2,*counts;
772: PetscBool flag;
775: PetscMalloc1(m,&ns);
776: while (i < m) {
777: nzx = ai[i+1] - ai[i]; /* Number of nonzeros */
778: /* Limits the number of elements in a node to 'a->inode.limit' */
779: for (j=i+1,blk_size=1; j<m && blk_size <a->inode.limit; ++j,++blk_size) {
780: nzy = ai[j+1] - ai[j];
781: if (nzy != (nzx - j + i)) break;
782: PetscMemcmp(aj + ai[i] + j - i,aj + ai[j],nzy*sizeof(PetscInt),&flag);
783: if (!flag) break;
784: }
785: ns[node_count++] = blk_size;
787: i = j;
788: }
789: if (!a->inode.size && m && node_count > .9*m) {
790: PetscFree(ns);
791: PetscInfo2(A,"Found %D nodes out of %D rows. Not using Inode routines\n",node_count,m);
792: } else {
793: a->inode.node_count = node_count;
795: PetscMalloc1(node_count,&a->inode.size);
796: PetscLogObjectMemory((PetscObject)A,node_count*sizeof(PetscInt));
797: PetscMemcpy(a->inode.size,ns,node_count*sizeof(PetscInt));
798: PetscFree(ns);
799: PetscInfo3(A,"Found %D nodes of %D. Limit used: %D. Using Inode routines\n",node_count,m,a->inode.limit);
801: /* count collections of adjacent columns in each inode */
802: row = 0;
803: cnt = 0;
804: for (i=0; i<node_count; i++) {
805: cols = aj + ai[row] + a->inode.size[i];
806: nz = ai[row+1] - ai[row] - a->inode.size[i];
807: for (j=1; j<nz; j++) {
808: if (cols[j] != cols[j-1]+1) cnt++;
809: }
810: cnt++;
811: row += a->inode.size[i];
812: }
813: PetscMalloc1(2*cnt,&counts);
814: cnt = 0;
815: row = 0;
816: for (i=0; i<node_count; i++) {
817: cols = aj + ai[row] + a->inode.size[i];
818: counts[2*cnt] = cols[0];
819: nz = ai[row+1] - ai[row] - a->inode.size[i];
820: cnt2 = 1;
821: for (j=1; j<nz; j++) {
822: if (cols[j] != cols[j-1]+1) {
823: counts[2*(cnt++)+1] = cnt2;
824: counts[2*cnt] = cols[j];
825: cnt2 = 1;
826: } else cnt2++;
827: }
828: counts[2*(cnt++)+1] = cnt2;
829: row += a->inode.size[i];
830: }
831: PetscIntView(2*cnt,counts,0);
832: }
833: return(0);
834: }
838: PetscErrorCode MatAssemblyEnd_SeqSBAIJ(Mat A,MatAssemblyType mode)
839: {
840: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
842: PetscInt fshift = 0,i,j,*ai = a->i,*aj = a->j,*imax = a->imax;
843: PetscInt m = A->rmap->N,*ip,N,*ailen = a->ilen;
844: PetscInt mbs = a->mbs,bs2 = a->bs2,rmax = 0;
845: MatScalar *aa = a->a,*ap;
848: if (mode == MAT_FLUSH_ASSEMBLY) return(0);
850: if (m) rmax = ailen[0];
851: for (i=1; i<mbs; i++) {
852: /* move each row back by the amount of empty slots (fshift) before it*/
853: fshift += imax[i-1] - ailen[i-1];
854: rmax = PetscMax(rmax,ailen[i]);
855: if (fshift) {
856: ip = aj + ai[i]; ap = aa + bs2*ai[i];
857: N = ailen[i];
858: for (j=0; j<N; j++) {
859: ip[j-fshift] = ip[j];
860: PetscMemcpy(ap+(j-fshift)*bs2,ap+j*bs2,bs2*sizeof(MatScalar));
861: }
862: }
863: ai[i] = ai[i-1] + ailen[i-1];
864: }
865: if (mbs) {
866: fshift += imax[mbs-1] - ailen[mbs-1];
867: ai[mbs] = ai[mbs-1] + ailen[mbs-1];
868: }
869: /* reset ilen and imax for each row */
870: for (i=0; i<mbs; i++) {
871: ailen[i] = imax[i] = ai[i+1] - ai[i];
872: }
873: a->nz = ai[mbs];
875: /* diagonals may have moved, reset it */
876: if (a->diag) {
877: PetscMemcpy(a->diag,ai,mbs*sizeof(PetscInt));
878: }
879: 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);
881: 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);
882: PetscInfo1(A,"Number of mallocs during MatSetValues is %D\n",a->reallocs);
883: PetscInfo1(A,"Most nonzeros blocks in any row is %D\n",rmax);
885: A->info.mallocs += a->reallocs;
886: a->reallocs = 0;
887: A->info.nz_unneeded = (PetscReal)fshift*bs2;
888: a->idiagvalid = PETSC_FALSE;
889: a->rmax = rmax;
891: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
892: if (a->jshort && a->free_jshort) {
893: /* when matrix data structure is changed, previous jshort must be replaced */
894: PetscFree(a->jshort);
895: }
896: PetscMalloc1(a->i[A->rmap->n],&a->jshort);
897: PetscLogObjectMemory((PetscObject)A,a->i[A->rmap->n]*sizeof(unsigned short));
898: for (i=0; i<a->i[A->rmap->n]; i++) a->jshort[i] = a->j[i];
899: A->ops->mult = MatMult_SeqSBAIJ_1_ushort;
900: A->ops->sor = MatSOR_SeqSBAIJ_ushort;
901: a->free_jshort = PETSC_TRUE;
902: }
903: return(0);
904: }
906: /*
907: This function returns an array of flags which indicate the locations of contiguous
908: blocks that should be zeroed. for eg: if bs = 3 and is = [0,1,2,3,5,6,7,8,9]
909: then the resulting sizes = [3,1,1,3,1] correspondig to sets [(0,1,2),(3),(5),(6,7,8),(9)]
910: Assume: sizes should be long enough to hold all the values.
911: */
914: PetscErrorCode MatZeroRows_SeqSBAIJ_Check_Blocks(PetscInt idx[],PetscInt n,PetscInt bs,PetscInt sizes[], PetscInt *bs_max)
915: {
916: PetscInt i,j,k,row;
917: PetscBool flg;
920: for (i=0,j=0; i<n; j++) {
921: row = idx[i];
922: if (row%bs!=0) { /* Not the begining of a block */
923: sizes[j] = 1;
924: i++;
925: } else if (i+bs > n) { /* Beginning of a block, but complete block doesn't exist (at idx end) */
926: sizes[j] = 1; /* Also makes sure atleast 'bs' values exist for next else */
927: i++;
928: } else { /* Begining of the block, so check if the complete block exists */
929: flg = PETSC_TRUE;
930: for (k=1; k<bs; k++) {
931: if (row+k != idx[i+k]) { /* break in the block */
932: flg = PETSC_FALSE;
933: break;
934: }
935: }
936: if (flg) { /* No break in the bs */
937: sizes[j] = bs;
938: i += bs;
939: } else {
940: sizes[j] = 1;
941: i++;
942: }
943: }
944: }
945: *bs_max = j;
946: return(0);
947: }
950: /* Only add/insert a(i,j) with i<=j (blocks).
951: Any a(i,j) with i>j input by user is ingored.
952: */
956: PetscErrorCode MatSetValues_SeqSBAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
957: {
958: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
960: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N,lastcol = -1;
961: PetscInt *imax=a->imax,*ai=a->i,*ailen=a->ilen,roworiented=a->roworiented;
962: PetscInt *aj =a->j,nonew=a->nonew,bs=A->rmap->bs,brow,bcol;
963: PetscInt ridx,cidx,bs2=a->bs2;
964: MatScalar *ap,value,*aa=a->a,*bap;
967: for (k=0; k<m; k++) { /* loop over added rows */
968: row = im[k]; /* row number */
969: brow = row/bs; /* block row number */
970: if (row < 0) continue;
971: #if defined(PETSC_USE_DEBUG)
972: 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);
973: #endif
974: rp = aj + ai[brow]; /*ptr to beginning of column value of the row block*/
975: ap = aa + bs2*ai[brow]; /*ptr to beginning of element value of the row block*/
976: rmax = imax[brow]; /* maximum space allocated for this row */
977: nrow = ailen[brow]; /* actual length of this row */
978: low = 0;
980: for (l=0; l<n; l++) { /* loop over added columns */
981: if (in[l] < 0) continue;
982: #if defined(PETSC_USE_DEBUG)
983: 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);
984: #endif
985: col = in[l];
986: bcol = col/bs; /* block col number */
988: if (brow > bcol) {
989: if (a->ignore_ltriangular) continue; /* ignore lower triangular values */
990: 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)");
991: }
993: ridx = row % bs; cidx = col % bs; /*row and col index inside the block */
994: if ((brow==bcol && ridx<=cidx) || (brow<bcol)) {
995: /* element value a(k,l) */
996: if (roworiented) value = v[l + k*n];
997: else value = v[k + l*m];
999: /* move pointer bap to a(k,l) quickly and add/insert value */
1000: if (col <= lastcol) low = 0;
1001: high = nrow;
1002: lastcol = col;
1003: while (high-low > 7) {
1004: t = (low+high)/2;
1005: if (rp[t] > bcol) high = t;
1006: else low = t;
1007: }
1008: for (i=low; i<high; i++) {
1009: if (rp[i] > bcol) break;
1010: if (rp[i] == bcol) {
1011: bap = ap + bs2*i + bs*cidx + ridx;
1012: if (is == ADD_VALUES) *bap += value;
1013: else *bap = value;
1014: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
1015: if (brow == bcol && ridx < cidx) {
1016: bap = ap + bs2*i + bs*ridx + cidx;
1017: if (is == ADD_VALUES) *bap += value;
1018: else *bap = value;
1019: }
1020: goto noinsert1;
1021: }
1022: }
1024: if (nonew == 1) goto noinsert1;
1025: if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) in the matrix", row, col);
1026: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
1028: N = nrow++ - 1; high++;
1029: /* shift up all the later entries in this row */
1030: for (ii=N; ii>=i; ii--) {
1031: rp[ii+1] = rp[ii];
1032: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));
1033: }
1034: if (N>=i) {
1035: PetscMemzero(ap+bs2*i,bs2*sizeof(MatScalar));
1036: }
1037: rp[i] = bcol;
1038: ap[bs2*i + bs*cidx + ridx] = value;
1039: A->nonzerostate++;
1040: noinsert1:;
1041: low = i;
1042: }
1043: } /* end of loop over added columns */
1044: ailen[brow] = nrow;
1045: } /* end of loop over added rows */
1046: return(0);
1047: }
1051: PetscErrorCode MatICCFactor_SeqSBAIJ(Mat inA,IS row,const MatFactorInfo *info)
1052: {
1053: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)inA->data;
1054: Mat outA;
1056: PetscBool row_identity;
1059: if (info->levels != 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only levels=0 is supported for in-place icc");
1060: ISIdentity(row,&row_identity);
1061: if (!row_identity) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix reordering is not supported");
1062: 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()! */
1064: outA = inA;
1065: inA->factortype = MAT_FACTOR_ICC;
1066: PetscFree(inA->solvertype);
1067: PetscStrallocpy(MATSOLVERPETSC,&inA->solvertype);
1069: MatMarkDiagonal_SeqSBAIJ(inA);
1070: MatSeqSBAIJSetNumericFactorization_inplace(inA,row_identity);
1072: PetscObjectReference((PetscObject)row);
1073: ISDestroy(&a->row);
1074: a->row = row;
1075: PetscObjectReference((PetscObject)row);
1076: ISDestroy(&a->col);
1077: a->col = row;
1079: /* Create the invert permutation so that it can be used in MatCholeskyFactorNumeric() */
1080: if (a->icol) {ISInvertPermutation(row,PETSC_DECIDE, &a->icol);}
1081: PetscLogObjectParent((PetscObject)inA,(PetscObject)a->icol);
1083: if (!a->solve_work) {
1084: PetscMalloc1(inA->rmap->N+inA->rmap->bs,&a->solve_work);
1085: PetscLogObjectMemory((PetscObject)inA,(inA->rmap->N+inA->rmap->bs)*sizeof(PetscScalar));
1086: }
1088: MatCholeskyFactorNumeric(outA,inA,info);
1089: return(0);
1090: }
1094: PetscErrorCode MatSeqSBAIJSetColumnIndices_SeqSBAIJ(Mat mat,PetscInt *indices)
1095: {
1096: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ*)mat->data;
1097: PetscInt i,nz,n;
1101: nz = baij->maxnz;
1102: n = mat->cmap->n;
1103: for (i=0; i<nz; i++) baij->j[i] = indices[i];
1105: baij->nz = nz;
1106: for (i=0; i<n; i++) baij->ilen[i] = baij->imax[i];
1108: MatSetOption(mat,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
1109: return(0);
1110: }
1114: /*@
1115: MatSeqSBAIJSetColumnIndices - Set the column indices for all the rows
1116: in the matrix.
1118: Input Parameters:
1119: + mat - the SeqSBAIJ matrix
1120: - indices - the column indices
1122: Level: advanced
1124: Notes:
1125: This can be called if you have precomputed the nonzero structure of the
1126: matrix and want to provide it to the matrix object to improve the performance
1127: of the MatSetValues() operation.
1129: You MUST have set the correct numbers of nonzeros per row in the call to
1130: MatCreateSeqSBAIJ(), and the columns indices MUST be sorted.
1132: MUST be called before any calls to MatSetValues()
1134: .seealso: MatCreateSeqSBAIJ
1135: @*/
1136: PetscErrorCode MatSeqSBAIJSetColumnIndices(Mat mat,PetscInt *indices)
1137: {
1143: PetscUseMethod(mat,"MatSeqSBAIJSetColumnIndices_C",(Mat,PetscInt*),(mat,indices));
1144: return(0);
1145: }
1149: PetscErrorCode MatCopy_SeqSBAIJ(Mat A,Mat B,MatStructure str)
1150: {
1154: /* If the two matrices have the same copy implementation, use fast copy. */
1155: if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) {
1156: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1157: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ*)B->data;
1159: 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");
1160: PetscMemcpy(b->a,a->a,(a->i[A->rmap->N])*sizeof(PetscScalar));
1161: } else {
1162: MatGetRowUpperTriangular(A);
1163: MatCopy_Basic(A,B,str);
1164: MatRestoreRowUpperTriangular(A);
1165: }
1166: return(0);
1167: }
1171: PetscErrorCode MatSetUp_SeqSBAIJ(Mat A)
1172: {
1176: MatSeqSBAIJSetPreallocation_SeqSBAIJ(A,A->rmap->bs,PETSC_DEFAULT,0);
1177: return(0);
1178: }
1182: PetscErrorCode MatSeqSBAIJGetArray_SeqSBAIJ(Mat A,PetscScalar *array[])
1183: {
1184: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1187: *array = a->a;
1188: return(0);
1189: }
1193: PetscErrorCode MatSeqSBAIJRestoreArray_SeqSBAIJ(Mat A,PetscScalar *array[])
1194: {
1196: return(0);
1197: }
1201: PetscErrorCode MatAXPYGetPreallocation_SeqSBAIJ(Mat Y,Mat X,PetscInt *nnz)
1202: {
1203: PetscInt bs = Y->rmap->bs,mbs = Y->rmap->N/bs;
1204: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ*)X->data;
1205: Mat_SeqSBAIJ *y = (Mat_SeqSBAIJ*)Y->data;
1209: /* Set the number of nonzeros in the new matrix */
1210: MatAXPYGetPreallocation_SeqX_private(mbs,x->i,x->j,y->i,y->j,nnz);
1211: return(0);
1212: }
1216: PetscErrorCode MatAXPY_SeqSBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
1217: {
1218: Mat_SeqSBAIJ *x=(Mat_SeqSBAIJ*)X->data, *y=(Mat_SeqSBAIJ*)Y->data;
1220: PetscInt bs=Y->rmap->bs,bs2=bs*bs;
1221: PetscBLASInt one = 1;
1224: if (str == SAME_NONZERO_PATTERN) {
1225: PetscScalar alpha = a;
1226: PetscBLASInt bnz;
1227: PetscBLASIntCast(x->nz*bs2,&bnz);
1228: PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one));
1229: PetscObjectStateIncrease((PetscObject)Y);
1230: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1231: MatSetOption(X,MAT_GETROW_UPPERTRIANGULAR,PETSC_TRUE);
1232: MatAXPY_Basic(Y,a,X,str);
1233: MatSetOption(X,MAT_GETROW_UPPERTRIANGULAR,PETSC_FALSE);
1234: } else {
1235: Mat B;
1236: PetscInt *nnz;
1237: if (bs != X->rmap->bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Matrices must have same block size");
1238: MatGetRowUpperTriangular(X);
1239: MatGetRowUpperTriangular(Y);
1240: PetscMalloc1(Y->rmap->N,&nnz);
1241: MatCreate(PetscObjectComm((PetscObject)Y),&B);
1242: PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);
1243: MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);
1244: MatSetBlockSizesFromMats(B,Y,Y);
1245: MatSetType(B,(MatType) ((PetscObject)Y)->type_name);
1246: MatAXPYGetPreallocation_SeqSBAIJ(Y,X,nnz);
1247: MatSeqSBAIJSetPreallocation(B,bs,0,nnz);
1249: MatAXPY_BasicWithPreallocation(B,Y,a,X,str);
1251: MatHeaderReplace(Y,&B);
1252: PetscFree(nnz);
1253: MatRestoreRowUpperTriangular(X);
1254: MatRestoreRowUpperTriangular(Y);
1255: }
1256: return(0);
1257: }
1261: PetscErrorCode MatIsSymmetric_SeqSBAIJ(Mat A,PetscReal tol,PetscBool *flg)
1262: {
1264: *flg = PETSC_TRUE;
1265: return(0);
1266: }
1270: PetscErrorCode MatIsStructurallySymmetric_SeqSBAIJ(Mat A,PetscBool *flg)
1271: {
1273: *flg = PETSC_TRUE;
1274: return(0);
1275: }
1279: PetscErrorCode MatIsHermitian_SeqSBAIJ(Mat A,PetscReal tol,PetscBool *flg)
1280: {
1282: *flg = PETSC_FALSE;
1283: return(0);
1284: }
1288: PetscErrorCode MatRealPart_SeqSBAIJ(Mat A)
1289: {
1290: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1291: PetscInt i,nz = a->bs2*a->i[a->mbs];
1292: MatScalar *aa = a->a;
1295: for (i=0; i<nz; i++) aa[i] = PetscRealPart(aa[i]);
1296: return(0);
1297: }
1301: PetscErrorCode MatImaginaryPart_SeqSBAIJ(Mat A)
1302: {
1303: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1304: PetscInt i,nz = a->bs2*a->i[a->mbs];
1305: MatScalar *aa = a->a;
1308: for (i=0; i<nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1309: return(0);
1310: }
1314: PetscErrorCode MatZeroRowsColumns_SeqSBAIJ(Mat A,PetscInt is_n,const PetscInt is_idx[],PetscScalar diag,Vec x, Vec b)
1315: {
1316: Mat_SeqSBAIJ *baij=(Mat_SeqSBAIJ*)A->data;
1317: PetscErrorCode ierr;
1318: PetscInt i,j,k,count;
1319: PetscInt bs =A->rmap->bs,bs2=baij->bs2,row,col;
1320: PetscScalar zero = 0.0;
1321: MatScalar *aa;
1322: const PetscScalar *xx;
1323: PetscScalar *bb;
1324: PetscBool *zeroed,vecs = PETSC_FALSE;
1327: /* fix right hand side if needed */
1328: if (x && b) {
1329: VecGetArrayRead(x,&xx);
1330: VecGetArray(b,&bb);
1331: vecs = PETSC_TRUE;
1332: }
1334: /* zero the columns */
1335: PetscCalloc1(A->rmap->n,&zeroed);
1336: for (i=0; i<is_n; i++) {
1337: 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]);
1338: zeroed[is_idx[i]] = PETSC_TRUE;
1339: }
1340: if (vecs) {
1341: for (i=0; i<A->rmap->N; i++) {
1342: row = i/bs;
1343: for (j=baij->i[row]; j<baij->i[row+1]; j++) {
1344: for (k=0; k<bs; k++) {
1345: col = bs*baij->j[j] + k;
1346: if (col <= i) continue;
1347: aa = ((MatScalar*)(baij->a)) + j*bs2 + (i%bs) + bs*k;
1348: if (!zeroed[i] && zeroed[col]) bb[i] -= aa[0]*xx[col];
1349: if (zeroed[i] && !zeroed[col]) bb[col] -= aa[0]*xx[i];
1350: }
1351: }
1352: }
1353: for (i=0; i<is_n; i++) bb[is_idx[i]] = diag*xx[is_idx[i]];
1354: }
1356: for (i=0; i<A->rmap->N; i++) {
1357: if (!zeroed[i]) {
1358: row = i/bs;
1359: for (j=baij->i[row]; j<baij->i[row+1]; j++) {
1360: for (k=0; k<bs; k++) {
1361: col = bs*baij->j[j] + k;
1362: if (zeroed[col]) {
1363: aa = ((MatScalar*)(baij->a)) + j*bs2 + (i%bs) + bs*k;
1364: aa[0] = 0.0;
1365: }
1366: }
1367: }
1368: }
1369: }
1370: PetscFree(zeroed);
1371: if (vecs) {
1372: VecRestoreArrayRead(x,&xx);
1373: VecRestoreArray(b,&bb);
1374: }
1376: /* zero the rows */
1377: for (i=0; i<is_n; i++) {
1378: row = is_idx[i];
1379: count = (baij->i[row/bs +1] - baij->i[row/bs])*bs;
1380: aa = ((MatScalar*)(baij->a)) + baij->i[row/bs]*bs2 + (row%bs);
1381: for (k=0; k<count; k++) {
1382: aa[0] = zero;
1383: aa += bs;
1384: }
1385: if (diag != 0.0) {
1386: (*A->ops->setvalues)(A,1,&row,1,&row,&diag,INSERT_VALUES);
1387: }
1388: }
1389: MatAssemblyEnd_SeqSBAIJ(A,MAT_FINAL_ASSEMBLY);
1390: return(0);
1391: }
1395: PetscErrorCode MatShift_SeqSBAIJ(Mat Y,PetscScalar a)
1396: {
1398: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)Y->data;
1401: if (!Y->preallocated || !aij->nz) {
1402: MatSeqSBAIJSetPreallocation(Y,Y->rmap->bs,1,NULL);
1403: }
1404: MatShift_Basic(Y,a);
1405: return(0);
1406: }
1408: /* -------------------------------------------------------------------*/
1409: static struct _MatOps MatOps_Values = {MatSetValues_SeqSBAIJ,
1410: MatGetRow_SeqSBAIJ,
1411: MatRestoreRow_SeqSBAIJ,
1412: MatMult_SeqSBAIJ_N,
1413: /* 4*/ MatMultAdd_SeqSBAIJ_N,
1414: MatMult_SeqSBAIJ_N, /* transpose versions are same as non-transpose versions */
1415: MatMultAdd_SeqSBAIJ_N,
1416: 0,
1417: 0,
1418: 0,
1419: /* 10*/ 0,
1420: 0,
1421: MatCholeskyFactor_SeqSBAIJ,
1422: MatSOR_SeqSBAIJ,
1423: MatTranspose_SeqSBAIJ,
1424: /* 15*/ MatGetInfo_SeqSBAIJ,
1425: MatEqual_SeqSBAIJ,
1426: MatGetDiagonal_SeqSBAIJ,
1427: MatDiagonalScale_SeqSBAIJ,
1428: MatNorm_SeqSBAIJ,
1429: /* 20*/ 0,
1430: MatAssemblyEnd_SeqSBAIJ,
1431: MatSetOption_SeqSBAIJ,
1432: MatZeroEntries_SeqSBAIJ,
1433: /* 24*/ 0,
1434: 0,
1435: 0,
1436: 0,
1437: 0,
1438: /* 29*/ MatSetUp_SeqSBAIJ,
1439: 0,
1440: 0,
1441: 0,
1442: 0,
1443: /* 34*/ MatDuplicate_SeqSBAIJ,
1444: 0,
1445: 0,
1446: 0,
1447: MatICCFactor_SeqSBAIJ,
1448: /* 39*/ MatAXPY_SeqSBAIJ,
1449: MatGetSubMatrices_SeqSBAIJ,
1450: MatIncreaseOverlap_SeqSBAIJ,
1451: MatGetValues_SeqSBAIJ,
1452: MatCopy_SeqSBAIJ,
1453: /* 44*/ 0,
1454: MatScale_SeqSBAIJ,
1455: MatShift_SeqSBAIJ,
1456: 0,
1457: MatZeroRowsColumns_SeqSBAIJ,
1458: /* 49*/ 0,
1459: MatGetRowIJ_SeqSBAIJ,
1460: MatRestoreRowIJ_SeqSBAIJ,
1461: 0,
1462: 0,
1463: /* 54*/ 0,
1464: 0,
1465: 0,
1466: 0,
1467: MatSetValuesBlocked_SeqSBAIJ,
1468: /* 59*/ MatGetSubMatrix_SeqSBAIJ,
1469: 0,
1470: 0,
1471: 0,
1472: 0,
1473: /* 64*/ 0,
1474: 0,
1475: 0,
1476: 0,
1477: 0,
1478: /* 69*/ MatGetRowMaxAbs_SeqSBAIJ,
1479: 0,
1480: 0,
1481: 0,
1482: 0,
1483: /* 74*/ 0,
1484: 0,
1485: 0,
1486: 0,
1487: 0,
1488: /* 79*/ 0,
1489: 0,
1490: 0,
1491: MatGetInertia_SeqSBAIJ,
1492: MatLoad_SeqSBAIJ,
1493: /* 84*/ MatIsSymmetric_SeqSBAIJ,
1494: MatIsHermitian_SeqSBAIJ,
1495: MatIsStructurallySymmetric_SeqSBAIJ,
1496: 0,
1497: 0,
1498: /* 89*/ 0,
1499: 0,
1500: 0,
1501: 0,
1502: 0,
1503: /* 94*/ 0,
1504: 0,
1505: 0,
1506: 0,
1507: 0,
1508: /* 99*/ 0,
1509: 0,
1510: 0,
1511: 0,
1512: 0,
1513: /*104*/ 0,
1514: MatRealPart_SeqSBAIJ,
1515: MatImaginaryPart_SeqSBAIJ,
1516: MatGetRowUpperTriangular_SeqSBAIJ,
1517: MatRestoreRowUpperTriangular_SeqSBAIJ,
1518: /*109*/ 0,
1519: 0,
1520: 0,
1521: 0,
1522: MatMissingDiagonal_SeqSBAIJ,
1523: /*114*/ 0,
1524: 0,
1525: 0,
1526: 0,
1527: 0,
1528: /*119*/ 0,
1529: 0,
1530: 0,
1531: 0,
1532: 0,
1533: /*124*/ 0,
1534: 0,
1535: 0,
1536: 0,
1537: 0,
1538: /*129*/ 0,
1539: 0,
1540: 0,
1541: 0,
1542: 0,
1543: /*134*/ 0,
1544: 0,
1545: 0,
1546: 0,
1547: 0,
1548: /*139*/ 0,
1549: 0,
1550: 0,
1551: 0,
1552: 0,
1553: /*144*/MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ
1554: };
1558: PetscErrorCode MatStoreValues_SeqSBAIJ(Mat mat)
1559: {
1560: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)mat->data;
1561: PetscInt nz = aij->i[mat->rmap->N]*mat->rmap->bs*aij->bs2;
1565: if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1567: /* allocate space for values if not already there */
1568: if (!aij->saved_values) {
1569: PetscMalloc1(nz+1,&aij->saved_values);
1570: }
1572: /* copy values over */
1573: PetscMemcpy(aij->saved_values,aij->a,nz*sizeof(PetscScalar));
1574: return(0);
1575: }
1579: PetscErrorCode MatRetrieveValues_SeqSBAIJ(Mat mat)
1580: {
1581: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ*)mat->data;
1583: PetscInt nz = aij->i[mat->rmap->N]*mat->rmap->bs*aij->bs2;
1586: if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1587: if (!aij->saved_values) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatStoreValues(A);first");
1589: /* copy values over */
1590: PetscMemcpy(aij->a,aij->saved_values,nz*sizeof(PetscScalar));
1591: return(0);
1592: }
1596: PetscErrorCode MatSeqSBAIJSetPreallocation_SeqSBAIJ(Mat B,PetscInt bs,PetscInt nz,PetscInt *nnz)
1597: {
1598: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ*)B->data;
1600: PetscInt i,mbs,nbs,bs2;
1601: PetscBool skipallocation = PETSC_FALSE,flg = PETSC_FALSE,realalloc = PETSC_FALSE;
1604: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
1605: B->preallocated = PETSC_TRUE;
1607: MatSetBlockSize(B,PetscAbs(bs));
1608: PetscLayoutSetUp(B->rmap);
1609: PetscLayoutSetUp(B->cmap);
1610: PetscLayoutGetBlockSize(B->rmap,&bs);
1612: mbs = B->rmap->N/bs;
1613: nbs = B->cmap->n/bs;
1614: bs2 = bs*bs;
1616: 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");
1618: if (nz == MAT_SKIP_ALLOCATION) {
1619: skipallocation = PETSC_TRUE;
1620: nz = 0;
1621: }
1623: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 3;
1624: if (nz < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nz cannot be less than 0: value %D",nz);
1625: if (nnz) {
1626: for (i=0; i<mbs; i++) {
1627: 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]);
1628: 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);
1629: }
1630: }
1632: B->ops->mult = MatMult_SeqSBAIJ_N;
1633: B->ops->multadd = MatMultAdd_SeqSBAIJ_N;
1634: B->ops->multtranspose = MatMult_SeqSBAIJ_N;
1635: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_N;
1637: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_no_unroll",&flg,NULL);
1638: if (!flg) {
1639: switch (bs) {
1640: case 1:
1641: B->ops->mult = MatMult_SeqSBAIJ_1;
1642: B->ops->multadd = MatMultAdd_SeqSBAIJ_1;
1643: B->ops->multtranspose = MatMult_SeqSBAIJ_1;
1644: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_1;
1645: break;
1646: case 2:
1647: B->ops->mult = MatMult_SeqSBAIJ_2;
1648: B->ops->multadd = MatMultAdd_SeqSBAIJ_2;
1649: B->ops->multtranspose = MatMult_SeqSBAIJ_2;
1650: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_2;
1651: break;
1652: case 3:
1653: B->ops->mult = MatMult_SeqSBAIJ_3;
1654: B->ops->multadd = MatMultAdd_SeqSBAIJ_3;
1655: B->ops->multtranspose = MatMult_SeqSBAIJ_3;
1656: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_3;
1657: break;
1658: case 4:
1659: B->ops->mult = MatMult_SeqSBAIJ_4;
1660: B->ops->multadd = MatMultAdd_SeqSBAIJ_4;
1661: B->ops->multtranspose = MatMult_SeqSBAIJ_4;
1662: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_4;
1663: break;
1664: case 5:
1665: B->ops->mult = MatMult_SeqSBAIJ_5;
1666: B->ops->multadd = MatMultAdd_SeqSBAIJ_5;
1667: B->ops->multtranspose = MatMult_SeqSBAIJ_5;
1668: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_5;
1669: break;
1670: case 6:
1671: B->ops->mult = MatMult_SeqSBAIJ_6;
1672: B->ops->multadd = MatMultAdd_SeqSBAIJ_6;
1673: B->ops->multtranspose = MatMult_SeqSBAIJ_6;
1674: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_6;
1675: break;
1676: case 7:
1677: B->ops->mult = MatMult_SeqSBAIJ_7;
1678: B->ops->multadd = MatMultAdd_SeqSBAIJ_7;
1679: B->ops->multtranspose = MatMult_SeqSBAIJ_7;
1680: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_7;
1681: break;
1682: }
1683: }
1685: b->mbs = mbs;
1686: b->nbs = nbs;
1687: if (!skipallocation) {
1688: if (!b->imax) {
1689: PetscMalloc2(mbs,&b->imax,mbs,&b->ilen);
1691: b->free_imax_ilen = PETSC_TRUE;
1693: PetscLogObjectMemory((PetscObject)B,2*mbs*sizeof(PetscInt));
1694: }
1695: if (!nnz) {
1696: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
1697: else if (nz <= 0) nz = 1;
1698: for (i=0; i<mbs; i++) b->imax[i] = nz;
1699: nz = nz*mbs; /* total nz */
1700: } else {
1701: nz = 0;
1702: for (i=0; i<mbs; i++) {b->imax[i] = nnz[i]; nz += nnz[i];}
1703: }
1704: /* b->ilen will count nonzeros in each block row so far. */
1705: for (i=0; i<mbs; i++) b->ilen[i] = 0;
1706: /* nz=(nz+mbs)/2; */ /* total diagonal and superdiagonal nonzero blocks */
1708: /* allocate the matrix space */
1709: MatSeqXAIJFreeAIJ(B,&b->a,&b->j,&b->i);
1710: PetscMalloc3(bs2*nz,&b->a,nz,&b->j,B->rmap->N+1,&b->i);
1711: PetscLogObjectMemory((PetscObject)B,(B->rmap->N+1)*sizeof(PetscInt)+nz*(bs2*sizeof(PetscScalar)+sizeof(PetscInt)));
1712: PetscMemzero(b->a,nz*bs2*sizeof(MatScalar));
1713: PetscMemzero(b->j,nz*sizeof(PetscInt));
1715: b->singlemalloc = PETSC_TRUE;
1717: /* pointer to beginning of each row */
1718: b->i[0] = 0;
1719: for (i=1; i<mbs+1; i++) b->i[i] = b->i[i-1] + b->imax[i-1];
1721: b->free_a = PETSC_TRUE;
1722: b->free_ij = PETSC_TRUE;
1723: } else {
1724: b->free_a = PETSC_FALSE;
1725: b->free_ij = PETSC_FALSE;
1726: }
1728: B->rmap->bs = bs;
1729: b->bs2 = bs2;
1730: b->nz = 0;
1731: b->maxnz = nz;
1733: b->inew = 0;
1734: b->jnew = 0;
1735: b->anew = 0;
1736: b->a2anew = 0;
1737: b->permute = PETSC_FALSE;
1738: if (realalloc) {MatSetOption(B,MAT_NEW_NONZERO_ALLOCATION_ERR,PETSC_TRUE);}
1739: return(0);
1740: }
1744: PetscErrorCode MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ(Mat B,PetscInt bs,const PetscInt ii[],const PetscInt jj[], const PetscScalar V[])
1745: {
1746: PetscInt i,j,m,nz,nz_max=0,*nnz;
1747: PetscScalar *values=0;
1748: PetscBool roworiented = ((Mat_SeqSBAIJ*)B->data)->roworiented;
1751: if (bs < 1) SETERRQ1(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_OUTOFRANGE,"Invalid block size specified, must be positive but it is %D",bs);
1752: PetscLayoutSetBlockSize(B->rmap,bs);
1753: PetscLayoutSetBlockSize(B->cmap,bs);
1754: PetscLayoutSetUp(B->rmap);
1755: PetscLayoutSetUp(B->cmap);
1756: PetscLayoutGetBlockSize(B->rmap,&bs);
1757: m = B->rmap->n/bs;
1759: if (ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"ii[0] must be 0 but it is %D",ii[0]);
1760: PetscMalloc1(m+1,&nnz);
1761: for (i=0; i<m; i++) {
1762: nz = ii[i+1] - ii[i];
1763: if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row %D has a negative number of columns %D",i,nz);
1764: nz_max = PetscMax(nz_max,nz);
1765: nnz[i] = nz;
1766: }
1767: MatSeqSBAIJSetPreallocation(B,bs,0,nnz);
1768: PetscFree(nnz);
1770: values = (PetscScalar*)V;
1771: if (!values) {
1772: PetscCalloc1(bs*bs*nz_max,&values);
1773: }
1774: for (i=0; i<m; i++) {
1775: PetscInt ncols = ii[i+1] - ii[i];
1776: const PetscInt *icols = jj + ii[i];
1777: if (!roworiented || bs == 1) {
1778: const PetscScalar *svals = values + (V ? (bs*bs*ii[i]) : 0);
1779: MatSetValuesBlocked_SeqSBAIJ(B,1,&i,ncols,icols,svals,INSERT_VALUES);
1780: } else {
1781: for (j=0; j<ncols; j++) {
1782: const PetscScalar *svals = values + (V ? (bs*bs*(ii[i]+j)) : 0);
1783: MatSetValuesBlocked_SeqSBAIJ(B,1,&i,1,&icols[j],svals,INSERT_VALUES);
1784: }
1785: }
1786: }
1787: if (!V) { PetscFree(values); }
1788: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
1789: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
1790: MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
1791: return(0);
1792: }
1794: /*
1795: This is used to set the numeric factorization for both Cholesky and ICC symbolic factorization
1796: */
1799: PetscErrorCode MatSeqSBAIJSetNumericFactorization_inplace(Mat B,PetscBool natural)
1800: {
1802: PetscBool flg = PETSC_FALSE;
1803: PetscInt bs = B->rmap->bs;
1806: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_no_unroll",&flg,NULL);
1807: if (flg) bs = 8;
1809: if (!natural) {
1810: switch (bs) {
1811: case 1:
1812: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_inplace;
1813: break;
1814: case 2:
1815: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2;
1816: break;
1817: case 3:
1818: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3;
1819: break;
1820: case 4:
1821: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4;
1822: break;
1823: case 5:
1824: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5;
1825: break;
1826: case 6:
1827: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6;
1828: break;
1829: case 7:
1830: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7;
1831: break;
1832: default:
1833: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N;
1834: break;
1835: }
1836: } else {
1837: switch (bs) {
1838: case 1:
1839: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_NaturalOrdering_inplace;
1840: break;
1841: case 2:
1842: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2_NaturalOrdering;
1843: break;
1844: case 3:
1845: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3_NaturalOrdering;
1846: break;
1847: case 4:
1848: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4_NaturalOrdering;
1849: break;
1850: case 5:
1851: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5_NaturalOrdering;
1852: break;
1853: case 6:
1854: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6_NaturalOrdering;
1855: break;
1856: case 7:
1857: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7_NaturalOrdering;
1858: break;
1859: default:
1860: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N_NaturalOrdering;
1861: break;
1862: }
1863: }
1864: return(0);
1865: }
1867: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqAIJ(Mat, MatType,MatReuse,Mat*);
1868: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqBAIJ(Mat, MatType,MatReuse,Mat*);
1872: PETSC_INTERN PetscErrorCode MatGetFactor_seqsbaij_petsc(Mat A,MatFactorType ftype,Mat *B)
1873: {
1874: PetscInt n = A->rmap->n;
1878: #if defined(PETSC_USE_COMPLEX)
1879: if (A->hermitian) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Hermitian Factor is not supported");
1880: #endif
1881: MatCreate(PetscObjectComm((PetscObject)A),B);
1882: MatSetSizes(*B,n,n,n,n);
1883: if (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) {
1884: MatSetType(*B,MATSEQSBAIJ);
1885: MatSeqSBAIJSetPreallocation(*B,A->rmap->bs,MAT_SKIP_ALLOCATION,NULL);
1887: (*B)->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SeqSBAIJ;
1888: (*B)->ops->iccfactorsymbolic = MatICCFactorSymbolic_SeqSBAIJ;
1889: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Factor type not supported");
1891: (*B)->factortype = ftype;
1892: PetscFree((*B)->solvertype);
1893: PetscStrallocpy(MATSOLVERPETSC,&(*B)->solvertype);
1894: return(0);
1895: }
1897: /*MC
1898: MATSEQSBAIJ - MATSEQSBAIJ = "seqsbaij" - A matrix type to be used for sequential symmetric block sparse matrices,
1899: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1901: For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you
1902: can call MatSetOption(Mat, MAT_HERMITIAN); after MatAssemblyEnd()
1904: Options Database Keys:
1905: . -mat_type seqsbaij - sets the matrix type to "seqsbaij" during a call to MatSetFromOptions()
1907: Notes: By default if you insert values into the lower triangular part of the matrix they are simply ignored (since they are not
1908: stored and it is assumed they symmetric to the upper triangular). If you call MatSetOption(Mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_FALSE) or use
1909: 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.
1912: Level: beginner
1914: .seealso: MatCreateSeqSBAIJ
1915: M*/
1917: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqSBSTRM(Mat, MatType,MatReuse,Mat*);
1921: PETSC_EXTERN PetscErrorCode MatCreate_SeqSBAIJ(Mat B)
1922: {
1923: Mat_SeqSBAIJ *b;
1925: PetscMPIInt size;
1926: PetscBool no_unroll = PETSC_FALSE,no_inode = PETSC_FALSE;
1929: MPI_Comm_size(PetscObjectComm((PetscObject)B),&size);
1930: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Comm must be of size 1");
1932: PetscNewLog(B,&b);
1933: B->data = (void*)b;
1934: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
1936: B->ops->destroy = MatDestroy_SeqSBAIJ;
1937: B->ops->view = MatView_SeqSBAIJ;
1938: b->row = 0;
1939: b->icol = 0;
1940: b->reallocs = 0;
1941: b->saved_values = 0;
1942: b->inode.limit = 5;
1943: b->inode.max_limit = 5;
1945: b->roworiented = PETSC_TRUE;
1946: b->nonew = 0;
1947: b->diag = 0;
1948: b->solve_work = 0;
1949: b->mult_work = 0;
1950: B->spptr = 0;
1951: B->info.nz_unneeded = (PetscReal)b->maxnz*b->bs2;
1952: b->keepnonzeropattern = PETSC_FALSE;
1954: b->inew = 0;
1955: b->jnew = 0;
1956: b->anew = 0;
1957: b->a2anew = 0;
1958: b->permute = PETSC_FALSE;
1960: b->ignore_ltriangular = PETSC_TRUE;
1962: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_ignore_lower_triangular",&b->ignore_ltriangular,NULL);
1964: b->getrow_utriangular = PETSC_FALSE;
1966: PetscOptionsGetBool(((PetscObject)B)->options,((PetscObject)B)->prefix,"-mat_getrow_uppertriangular",&b->getrow_utriangular,NULL);
1968: PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_SeqSBAIJ);
1969: PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_SeqSBAIJ);
1970: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetColumnIndices_C",MatSeqSBAIJSetColumnIndices_SeqSBAIJ);
1971: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqaij_C",MatConvert_SeqSBAIJ_SeqAIJ);
1972: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqbaij_C",MatConvert_SeqSBAIJ_SeqBAIJ);
1973: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetPreallocation_C",MatSeqSBAIJSetPreallocation_SeqSBAIJ);
1974: PetscObjectComposeFunction((PetscObject)B,"MatSeqSBAIJSetPreallocationCSR_C",MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ);
1975: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_seqsbstrm_C",MatConvert_SeqSBAIJ_SeqSBSTRM);
1976: #if defined(PETSC_HAVE_ELEMENTAL)
1977: PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqsbaij_elemental_C",MatConvert_SeqSBAIJ_Elemental);
1978: #endif
1980: B->symmetric = PETSC_TRUE;
1981: B->structurally_symmetric = PETSC_TRUE;
1982: B->symmetric_set = PETSC_TRUE;
1983: B->structurally_symmetric_set = PETSC_TRUE;
1985: PetscObjectChangeTypeName((PetscObject)B,MATSEQSBAIJ);
1987: PetscOptionsBegin(PetscObjectComm((PetscObject)B),((PetscObject)B)->prefix,"Options for SEQSBAIJ matrix","Mat");
1988: PetscOptionsBool("-mat_no_unroll","Do not optimize for inodes (slower)",NULL,no_unroll,&no_unroll,NULL);
1989: if (no_unroll) {
1990: PetscInfo(B,"Not using Inode routines due to -mat_no_unroll\n");
1991: }
1992: PetscOptionsBool("-mat_no_inode","Do not optimize for inodes (slower)",NULL,no_inode,&no_inode,NULL);
1993: if (no_inode) {
1994: PetscInfo(B,"Not using Inode routines due to -mat_no_inode\n");
1995: }
1996: PetscOptionsInt("-mat_inode_limit","Do not use inodes larger then this value",NULL,b->inode.limit,&b->inode.limit,NULL);
1997: PetscOptionsEnd();
1998: b->inode.use = (PetscBool)(!(no_unroll || no_inode));
1999: if (b->inode.limit > b->inode.max_limit) b->inode.limit = b->inode.max_limit;
2000: return(0);
2001: }
2005: /*@C
2006: MatSeqSBAIJSetPreallocation - Creates a sparse symmetric matrix in block AIJ (block
2007: compressed row) format. For good matrix assembly performance the
2008: user should preallocate the matrix storage by setting the parameter nz
2009: (or the array nnz). By setting these parameters accurately, performance
2010: during matrix assembly can be increased by more than a factor of 50.
2012: Collective on Mat
2014: Input Parameters:
2015: + B - the symmetric matrix
2016: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
2017: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2018: . nz - number of block nonzeros per block row (same for all rows)
2019: - nnz - array containing the number of block nonzeros in the upper triangular plus
2020: diagonal portion of each block (possibly different for each block row) or NULL
2022: Options Database Keys:
2023: . -mat_no_unroll - uses code that does not unroll the loops in the
2024: block calculations (much slower)
2025: . -mat_block_size - size of the blocks to use (only works if a negative bs is passed in
2027: Level: intermediate
2029: Notes:
2030: Specify the preallocated storage with either nz or nnz (not both).
2031: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
2032: allocation. See Users-Manual: ch_mat for details.
2034: You can call MatGetInfo() to get information on how effective the preallocation was;
2035: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
2036: You can also run with the option -info and look for messages with the string
2037: malloc in them to see if additional memory allocation was needed.
2039: If the nnz parameter is given then the nz parameter is ignored
2042: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatCreateSBAIJ()
2043: @*/
2044: PetscErrorCode MatSeqSBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt nz,const PetscInt nnz[])
2045: {
2052: PetscTryMethod(B,"MatSeqSBAIJSetPreallocation_C",(Mat,PetscInt,PetscInt,const PetscInt[]),(B,bs,nz,nnz));
2053: return(0);
2054: }
2056: #undef __FUNCT__
2058: /*@C
2059: MatSeqSBAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in symmetric block AIJ format.
2061: Input Parameters:
2062: + B - the matrix
2063: . bs - size of block, the blocks are ALWAYS square.
2064: . i - the indices into j for the start of each local row (starts with zero)
2065: . j - the column indices for each local row (starts with zero) these must be sorted for each row
2066: - v - optional values in the matrix
2068: Level: developer
2070: Notes:
2071: The order of the entries in values is specified by the MatOption MAT_ROW_ORIENTED. For example, C programs
2072: may want to use the default MAT_ROW_ORIENTED=PETSC_TRUE and use an array v[nnz][bs][bs] where the second index is
2073: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
2074: MAT_ROW_ORIENTED=PETSC_FALSE and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
2075: block column and the second index is over columns within a block.
2077: .keywords: matrix, block, aij, compressed row, sparse
2079: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValuesBlocked(), MatSeqSBAIJSetPreallocation(), MATSEQSBAIJ
2080: @*/
2081: PetscErrorCode MatSeqSBAIJSetPreallocationCSR(Mat B,PetscInt bs,const PetscInt i[],const PetscInt j[], const PetscScalar v[])
2082: {
2089: PetscTryMethod(B,"MatSeqSBAIJSetPreallocationCSR_C",(Mat,PetscInt,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,bs,i,j,v));
2090: return(0);
2091: }
2095: /*@C
2096: MatCreateSeqSBAIJ - Creates a sparse symmetric matrix in block AIJ (block
2097: compressed row) format. For good matrix assembly performance the
2098: user should preallocate the matrix storage by setting the parameter nz
2099: (or the array nnz). By setting these parameters accurately, performance
2100: during matrix assembly can be increased by more than a factor of 50.
2102: Collective on MPI_Comm
2104: Input Parameters:
2105: + comm - MPI communicator, set to PETSC_COMM_SELF
2106: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
2107: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2108: . m - number of rows, or number of columns
2109: . nz - number of block nonzeros per block row (same for all rows)
2110: - nnz - array containing the number of block nonzeros in the upper triangular plus
2111: diagonal portion of each block (possibly different for each block row) or NULL
2113: Output Parameter:
2114: . A - the symmetric matrix
2116: Options Database Keys:
2117: . -mat_no_unroll - uses code that does not unroll the loops in the
2118: block calculations (much slower)
2119: . -mat_block_size - size of the blocks to use
2121: Level: intermediate
2123: It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(),
2124: MatXXXXSetPreallocation() paradgm instead of this routine directly.
2125: [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation]
2127: Notes:
2128: The number of rows and columns must be divisible by blocksize.
2129: This matrix type does not support complex Hermitian operation.
2131: Specify the preallocated storage with either nz or nnz (not both).
2132: Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory
2133: allocation. See Users-Manual: ch_mat for details.
2135: If the nnz parameter is given then the nz parameter is ignored
2137: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatCreateSBAIJ()
2138: @*/
2139: PetscErrorCode MatCreateSeqSBAIJ(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt nz,const PetscInt nnz[],Mat *A)
2140: {
2144: MatCreate(comm,A);
2145: MatSetSizes(*A,m,n,m,n);
2146: MatSetType(*A,MATSEQSBAIJ);
2147: MatSeqSBAIJSetPreallocation_SeqSBAIJ(*A,bs,nz,(PetscInt*)nnz);
2148: return(0);
2149: }
2153: PetscErrorCode MatDuplicate_SeqSBAIJ(Mat A,MatDuplicateOption cpvalues,Mat *B)
2154: {
2155: Mat C;
2156: Mat_SeqSBAIJ *c,*a = (Mat_SeqSBAIJ*)A->data;
2158: PetscInt i,mbs = a->mbs,nz = a->nz,bs2 =a->bs2;
2161: if (a->i[mbs] != nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Corrupt matrix");
2163: *B = 0;
2164: MatCreate(PetscObjectComm((PetscObject)A),&C);
2165: MatSetSizes(C,A->rmap->N,A->cmap->n,A->rmap->N,A->cmap->n);
2166: MatSetType(C,MATSEQSBAIJ);
2167: PetscMemcpy(C->ops,A->ops,sizeof(struct _MatOps));
2168: c = (Mat_SeqSBAIJ*)C->data;
2170: C->preallocated = PETSC_TRUE;
2171: C->factortype = A->factortype;
2172: c->row = 0;
2173: c->icol = 0;
2174: c->saved_values = 0;
2175: c->keepnonzeropattern = a->keepnonzeropattern;
2176: C->assembled = PETSC_TRUE;
2178: PetscLayoutReference(A->rmap,&C->rmap);
2179: PetscLayoutReference(A->cmap,&C->cmap);
2180: c->bs2 = a->bs2;
2181: c->mbs = a->mbs;
2182: c->nbs = a->nbs;
2184: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2185: c->imax = a->imax;
2186: c->ilen = a->ilen;
2187: c->free_imax_ilen = PETSC_FALSE;
2188: } else {
2189: PetscMalloc2((mbs+1),&c->imax,(mbs+1),&c->ilen);
2190: PetscLogObjectMemory((PetscObject)C,2*(mbs+1)*sizeof(PetscInt));
2191: for (i=0; i<mbs; i++) {
2192: c->imax[i] = a->imax[i];
2193: c->ilen[i] = a->ilen[i];
2194: }
2195: c->free_imax_ilen = PETSC_TRUE;
2196: }
2198: /* allocate the matrix space */
2199: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2200: PetscMalloc1(bs2*nz,&c->a);
2201: PetscLogObjectMemory((PetscObject)C,nz*bs2*sizeof(MatScalar));
2202: c->i = a->i;
2203: c->j = a->j;
2204: c->singlemalloc = PETSC_FALSE;
2205: c->free_a = PETSC_TRUE;
2206: c->free_ij = PETSC_FALSE;
2207: c->parent = A;
2208: PetscObjectReference((PetscObject)A);
2209: MatSetOption(A,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
2210: MatSetOption(C,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
2211: } else {
2212: PetscMalloc3(bs2*nz,&c->a,nz,&c->j,mbs+1,&c->i);
2213: PetscMemcpy(c->i,a->i,(mbs+1)*sizeof(PetscInt));
2214: PetscLogObjectMemory((PetscObject)C,(mbs+1)*sizeof(PetscInt) + nz*(bs2*sizeof(MatScalar) + sizeof(PetscInt)));
2215: c->singlemalloc = PETSC_TRUE;
2216: c->free_a = PETSC_TRUE;
2217: c->free_ij = PETSC_TRUE;
2218: }
2219: if (mbs > 0) {
2220: if (cpvalues != MAT_SHARE_NONZERO_PATTERN) {
2221: PetscMemcpy(c->j,a->j,nz*sizeof(PetscInt));
2222: }
2223: if (cpvalues == MAT_COPY_VALUES) {
2224: PetscMemcpy(c->a,a->a,bs2*nz*sizeof(MatScalar));
2225: } else {
2226: PetscMemzero(c->a,bs2*nz*sizeof(MatScalar));
2227: }
2228: if (a->jshort) {
2229: /* cannot share jshort, it is reallocated in MatAssemblyEnd_SeqSBAIJ() */
2230: /* if the parent matrix is reassembled, this child matrix will never notice */
2231: PetscMalloc1(nz,&c->jshort);
2232: PetscLogObjectMemory((PetscObject)C,nz*sizeof(unsigned short));
2233: PetscMemcpy(c->jshort,a->jshort,nz*sizeof(unsigned short));
2235: c->free_jshort = PETSC_TRUE;
2236: }
2237: }
2239: c->roworiented = a->roworiented;
2240: c->nonew = a->nonew;
2242: if (a->diag) {
2243: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2244: c->diag = a->diag;
2245: c->free_diag = PETSC_FALSE;
2246: } else {
2247: PetscMalloc1(mbs,&c->diag);
2248: PetscLogObjectMemory((PetscObject)C,mbs*sizeof(PetscInt));
2249: for (i=0; i<mbs; i++) c->diag[i] = a->diag[i];
2250: c->free_diag = PETSC_TRUE;
2251: }
2252: }
2253: c->nz = a->nz;
2254: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
2255: c->solve_work = 0;
2256: c->mult_work = 0;
2258: *B = C;
2259: PetscFunctionListDuplicate(((PetscObject)A)->qlist,&((PetscObject)C)->qlist);
2260: return(0);
2261: }
2265: PetscErrorCode MatLoad_SeqSBAIJ(Mat newmat,PetscViewer viewer)
2266: {
2267: Mat_SeqSBAIJ *a;
2269: int fd;
2270: PetscMPIInt size;
2271: PetscInt i,nz,header[4],*rowlengths=0,M,N,bs = newmat->rmap->bs;
2272: PetscInt *mask,mbs,*jj,j,rowcount,nzcount,k,*s_browlengths,maskcount;
2273: PetscInt kmax,jcount,block,idx,point,nzcountb,extra_rows,rows,cols;
2274: PetscInt *masked,nmask,tmp,bs2,ishift;
2275: PetscScalar *aa;
2276: MPI_Comm comm;
2279: /* force binary viewer to load .info file if it has not yet done so */
2280: PetscViewerSetUp(viewer);
2281: PetscObjectGetComm((PetscObject)viewer,&comm);
2282: PetscOptionsGetInt(((PetscObject)newmat)->options,((PetscObject)newmat)->prefix,"-matload_block_size",&bs,NULL);
2283: if (bs < 0) bs = 1;
2284: bs2 = bs*bs;
2286: MPI_Comm_size(comm,&size);
2287: if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"view must have one processor");
2288: PetscViewerBinaryGetDescriptor(viewer,&fd);
2289: PetscBinaryRead(fd,header,4,PETSC_INT);
2290: if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not Mat object");
2291: M = header[1]; N = header[2]; nz = header[3];
2293: if (header[3] < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as SeqSBAIJ");
2295: if (M != N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only do square matrices");
2297: /*
2298: This code adds extra rows to make sure the number of rows is
2299: divisible by the blocksize
2300: */
2301: mbs = M/bs;
2302: extra_rows = bs - M + bs*(mbs);
2303: if (extra_rows == bs) extra_rows = 0;
2304: else mbs++;
2305: if (extra_rows) {
2306: PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");
2307: }
2309: /* Set global sizes if not already set */
2310: if (newmat->rmap->n < 0 && newmat->rmap->N < 0 && newmat->cmap->n < 0 && newmat->cmap->N < 0) {
2311: MatSetSizes(newmat,PETSC_DECIDE,PETSC_DECIDE,M+extra_rows,N+extra_rows);
2312: } else { /* Check if the matrix global sizes are correct */
2313: MatGetSize(newmat,&rows,&cols);
2314: 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);
2315: }
2317: /* read in row lengths */
2318: PetscMalloc1(M+extra_rows,&rowlengths);
2319: PetscBinaryRead(fd,rowlengths,M,PETSC_INT);
2320: for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1;
2322: /* read in column indices */
2323: PetscMalloc1(nz+extra_rows,&jj);
2324: PetscBinaryRead(fd,jj,nz,PETSC_INT);
2325: for (i=0; i<extra_rows; i++) jj[nz+i] = M+i;
2327: /* loop over row lengths determining block row lengths */
2328: PetscCalloc1(mbs,&s_browlengths);
2329: PetscMalloc2(mbs,&mask,mbs,&masked);
2330: PetscMemzero(mask,mbs*sizeof(PetscInt));
2331: rowcount = 0;
2332: nzcount = 0;
2333: for (i=0; i<mbs; i++) {
2334: nmask = 0;
2335: for (j=0; j<bs; j++) {
2336: kmax = rowlengths[rowcount];
2337: for (k=0; k<kmax; k++) {
2338: tmp = jj[nzcount++]/bs; /* block col. index */
2339: if (!mask[tmp] && tmp >= i) {masked[nmask++] = tmp; mask[tmp] = 1;}
2340: }
2341: rowcount++;
2342: }
2343: s_browlengths[i] += nmask;
2345: /* zero out the mask elements we set */
2346: for (j=0; j<nmask; j++) mask[masked[j]] = 0;
2347: }
2349: /* Do preallocation */
2350: MatSeqSBAIJSetPreallocation_SeqSBAIJ(newmat,bs,0,s_browlengths);
2351: a = (Mat_SeqSBAIJ*)newmat->data;
2353: /* set matrix "i" values */
2354: a->i[0] = 0;
2355: for (i=1; i<= mbs; i++) {
2356: a->i[i] = a->i[i-1] + s_browlengths[i-1];
2357: a->ilen[i-1] = s_browlengths[i-1];
2358: }
2359: a->nz = a->i[mbs];
2361: /* read in nonzero values */
2362: PetscMalloc1(nz+extra_rows,&aa);
2363: PetscBinaryRead(fd,aa,nz,PETSC_SCALAR);
2364: for (i=0; i<extra_rows; i++) aa[nz+i] = 1.0;
2366: /* set "a" and "j" values into matrix */
2367: nzcount = 0; jcount = 0;
2368: for (i=0; i<mbs; i++) {
2369: nzcountb = nzcount;
2370: nmask = 0;
2371: for (j=0; j<bs; j++) {
2372: kmax = rowlengths[i*bs+j];
2373: for (k=0; k<kmax; k++) {
2374: tmp = jj[nzcount++]/bs; /* block col. index */
2375: if (!mask[tmp] && tmp >= i) { masked[nmask++] = tmp; mask[tmp] = 1;}
2376: }
2377: }
2378: /* sort the masked values */
2379: PetscSortInt(nmask,masked);
2381: /* set "j" values into matrix */
2382: maskcount = 1;
2383: for (j=0; j<nmask; j++) {
2384: a->j[jcount++] = masked[j];
2385: mask[masked[j]] = maskcount++;
2386: }
2388: /* set "a" values into matrix */
2389: ishift = bs2*a->i[i];
2390: for (j=0; j<bs; j++) {
2391: kmax = rowlengths[i*bs+j];
2392: for (k=0; k<kmax; k++) {
2393: tmp = jj[nzcountb]/bs; /* block col. index */
2394: if (tmp >= i) {
2395: block = mask[tmp] - 1;
2396: point = jj[nzcountb] - bs*tmp;
2397: idx = ishift + bs2*block + j + bs*point;
2398: a->a[idx] = aa[nzcountb];
2399: }
2400: nzcountb++;
2401: }
2402: }
2403: /* zero out the mask elements we set */
2404: for (j=0; j<nmask; j++) mask[masked[j]] = 0;
2405: }
2406: if (jcount != a->nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Bad binary matrix");
2408: PetscFree(rowlengths);
2409: PetscFree(s_browlengths);
2410: PetscFree(aa);
2411: PetscFree(jj);
2412: PetscFree2(mask,masked);
2414: MatAssemblyBegin(newmat,MAT_FINAL_ASSEMBLY);
2415: MatAssemblyEnd(newmat,MAT_FINAL_ASSEMBLY);
2416: return(0);
2417: }
2421: /*@
2422: MatCreateSeqSBAIJWithArrays - Creates an sequential SBAIJ matrix using matrix elements
2423: (upper triangular entries in CSR format) provided by the user.
2425: Collective on MPI_Comm
2427: Input Parameters:
2428: + comm - must be an MPI communicator of size 1
2429: . bs - size of block
2430: . m - number of rows
2431: . n - number of columns
2432: . i - row indices
2433: . j - column indices
2434: - a - matrix values
2436: Output Parameter:
2437: . mat - the matrix
2439: Level: advanced
2441: Notes:
2442: The i, j, and a arrays are not copied by this routine, the user must free these arrays
2443: once the matrix is destroyed
2445: You cannot set new nonzero locations into this matrix, that will generate an error.
2447: The i and j indices are 0 based
2449: 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
2450: it is the regular CSR format excluding the lower triangular elements.
2452: .seealso: MatCreate(), MatCreateSBAIJ(), MatCreateSeqSBAIJ()
2454: @*/
2455: PetscErrorCode MatCreateSeqSBAIJWithArrays(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt *i,PetscInt *j,PetscScalar *a,Mat *mat)
2456: {
2458: PetscInt ii;
2459: Mat_SeqSBAIJ *sbaij;
2462: if (bs != 1) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"block size %D > 1 is not supported yet",bs);
2463: if (m > 0 && i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0");
2465: MatCreate(comm,mat);
2466: MatSetSizes(*mat,m,n,m,n);
2467: MatSetType(*mat,MATSEQSBAIJ);
2468: MatSeqSBAIJSetPreallocation_SeqSBAIJ(*mat,bs,MAT_SKIP_ALLOCATION,0);
2469: sbaij = (Mat_SeqSBAIJ*)(*mat)->data;
2470: PetscMalloc2(m,&sbaij->imax,m,&sbaij->ilen);
2471: PetscLogObjectMemory((PetscObject)*mat,2*m*sizeof(PetscInt));
2473: sbaij->i = i;
2474: sbaij->j = j;
2475: sbaij->a = a;
2477: sbaij->singlemalloc = PETSC_FALSE;
2478: sbaij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
2479: sbaij->free_a = PETSC_FALSE;
2480: sbaij->free_ij = PETSC_FALSE;
2482: for (ii=0; ii<m; ii++) {
2483: sbaij->ilen[ii] = sbaij->imax[ii] = i[ii+1] - i[ii];
2484: #if defined(PETSC_USE_DEBUG)
2485: 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]);
2486: #endif
2487: }
2488: #if defined(PETSC_USE_DEBUG)
2489: for (ii=0; ii<sbaij->i[m]; ii++) {
2490: if (j[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column index at location = %d index = %d",ii,j[ii]);
2491: 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]);
2492: }
2493: #endif
2495: MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);
2496: MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);
2497: return(0);
2498: }
2502: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ(MPI_Comm comm,Mat inmat,PetscInt n,MatReuse scall,Mat *outmat)
2503: {
2507: MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(comm,inmat,n,scall,outmat);
2508: return(0);
2509: }