Actual source code: sbaij.c
1: /*
2: Defines the basic matrix operations for the SBAIJ (compressed row)
3: matrix storage format.
4: */
5: #include <../src/mat/impls/baij/seq/baij.h>
6: #include <../src/mat/impls/sbaij/seq/sbaij.h>
7: #include <petscblaslapack.h>
9: #include <../src/mat/impls/sbaij/seq/relax.h>
10: #define USESHORT
11: #include <../src/mat/impls/sbaij/seq/relax.h>
13: /* defines MatSetValues_Seq_Hash(), MatAssemblyEnd_Seq_Hash(), MatSetUp_Seq_Hash() */
14: #define TYPE SBAIJ
15: #define TYPE_SBAIJ
16: #define TYPE_BS
17: #include "../src/mat/impls/aij/seq/seqhashmatsetvalues.h"
18: #undef TYPE_BS
19: #define TYPE_BS _BS
20: #define TYPE_BS_ON
21: #include "../src/mat/impls/aij/seq/seqhashmatsetvalues.h"
22: #undef TYPE_BS
23: #undef TYPE_SBAIJ
24: #include "../src/mat/impls/aij/seq/seqhashmat.h"
25: #undef TYPE
26: #undef TYPE_BS_ON
28: #if defined(PETSC_HAVE_ELEMENTAL)
29: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_Elemental(Mat, MatType, MatReuse, Mat *);
30: #endif
31: #if defined(PETSC_HAVE_SCALAPACK)
32: PETSC_INTERN PetscErrorCode MatConvert_SBAIJ_ScaLAPACK(Mat, MatType, MatReuse, Mat *);
33: #endif
34: PETSC_INTERN PetscErrorCode MatConvert_MPISBAIJ_Basic(Mat, MatType, MatReuse, Mat *);
36: /*
37: Checks for missing diagonals
38: */
39: static PetscErrorCode MatMissingDiagonal_SeqSBAIJ(Mat A, PetscBool *missing, PetscInt *dd)
40: {
41: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
42: PetscInt *diag, *ii = a->i, i;
44: PetscFunctionBegin;
45: PetscCall(MatMarkDiagonal_SeqSBAIJ(A));
46: *missing = PETSC_FALSE;
47: if (A->rmap->n > 0 && !ii) {
48: *missing = PETSC_TRUE;
49: if (dd) *dd = 0;
50: PetscCall(PetscInfo(A, "Matrix has no entries therefore is missing diagonal\n"));
51: } else {
52: diag = a->diag;
53: for (i = 0; i < a->mbs; i++) {
54: if (diag[i] >= ii[i + 1]) {
55: *missing = PETSC_TRUE;
56: if (dd) *dd = i;
57: break;
58: }
59: }
60: }
61: PetscFunctionReturn(PETSC_SUCCESS);
62: }
64: PetscErrorCode MatMarkDiagonal_SeqSBAIJ(Mat A)
65: {
66: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
67: PetscInt i, j;
69: PetscFunctionBegin;
70: if (!a->diag) {
71: PetscCall(PetscMalloc1(a->mbs, &a->diag));
72: a->free_diag = PETSC_TRUE;
73: }
74: for (i = 0; i < a->mbs; i++) {
75: a->diag[i] = a->i[i + 1];
76: for (j = a->i[i]; j < a->i[i + 1]; j++) {
77: if (a->j[j] == i) {
78: a->diag[i] = j;
79: break;
80: }
81: }
82: }
83: PetscFunctionReturn(PETSC_SUCCESS);
84: }
86: static PetscErrorCode MatGetRowIJ_SeqSBAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool blockcompressed, PetscInt *nn, const PetscInt *inia[], const PetscInt *inja[], PetscBool *done)
87: {
88: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
89: PetscInt i, j, n = a->mbs, nz = a->i[n], *tia, *tja, bs = A->rmap->bs, k, l, cnt;
90: PetscInt **ia = (PetscInt **)inia, **ja = (PetscInt **)inja;
92: PetscFunctionBegin;
93: *nn = n;
94: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
95: if (symmetric) {
96: PetscCall(MatToSymmetricIJ_SeqAIJ(n, a->i, a->j, PETSC_FALSE, 0, 0, &tia, &tja));
97: nz = tia[n];
98: } else {
99: tia = a->i;
100: tja = a->j;
101: }
103: if (!blockcompressed && bs > 1) {
104: (*nn) *= bs;
105: /* malloc & create the natural set of indices */
106: PetscCall(PetscMalloc1((n + 1) * bs, ia));
107: if (n) {
108: (*ia)[0] = oshift;
109: for (j = 1; j < bs; j++) (*ia)[j] = (tia[1] - tia[0]) * bs + (*ia)[j - 1];
110: }
112: for (i = 1; i < n; i++) {
113: (*ia)[i * bs] = (tia[i] - tia[i - 1]) * bs + (*ia)[i * bs - 1];
114: for (j = 1; j < bs; j++) (*ia)[i * bs + j] = (tia[i + 1] - tia[i]) * bs + (*ia)[i * bs + j - 1];
115: }
116: if (n) (*ia)[n * bs] = (tia[n] - tia[n - 1]) * bs + (*ia)[n * bs - 1];
118: if (inja) {
119: PetscCall(PetscMalloc1(nz * bs * bs, ja));
120: cnt = 0;
121: for (i = 0; i < n; i++) {
122: for (j = 0; j < bs; j++) {
123: for (k = tia[i]; k < tia[i + 1]; k++) {
124: for (l = 0; l < bs; l++) (*ja)[cnt++] = bs * tja[k] + l;
125: }
126: }
127: }
128: }
130: if (symmetric) { /* deallocate memory allocated in MatToSymmetricIJ_SeqAIJ() */
131: PetscCall(PetscFree(tia));
132: PetscCall(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: PetscCall(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: PetscCall(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: PetscFunctionReturn(PETSC_SUCCESS);
159: }
161: static PetscErrorCode MatRestoreRowIJ_SeqSBAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool blockcompressed, PetscInt *nn, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
162: {
163: PetscFunctionBegin;
164: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
165: if ((!blockcompressed && A->rmap->bs > 1) || (symmetric || oshift == 1)) {
166: PetscCall(PetscFree(*ia));
167: if (ja) PetscCall(PetscFree(*ja));
168: }
169: PetscFunctionReturn(PETSC_SUCCESS);
170: }
172: PetscErrorCode MatDestroy_SeqSBAIJ(Mat A)
173: {
174: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
176: PetscFunctionBegin;
177: if (A->hash_active) {
178: PetscInt bs;
179: A->ops[0] = a->cops;
180: PetscCall(PetscHMapIJVDestroy(&a->ht));
181: PetscCall(MatGetBlockSize(A, &bs));
182: if (bs > 1) PetscCall(PetscHSetIJDestroy(&a->bht));
183: PetscCall(PetscFree(a->dnz));
184: PetscCall(PetscFree(a->bdnz));
185: A->hash_active = PETSC_FALSE;
186: }
187: PetscCall(PetscLogObjectState((PetscObject)A, "Rows=%" PetscInt_FMT ", NZ=%" PetscInt_FMT, A->rmap->N, a->nz));
188: PetscCall(MatSeqXAIJFreeAIJ(A, &a->a, &a->j, &a->i));
189: if (a->free_diag) PetscCall(PetscFree(a->diag));
190: PetscCall(ISDestroy(&a->row));
191: PetscCall(ISDestroy(&a->col));
192: PetscCall(ISDestroy(&a->icol));
193: PetscCall(PetscFree(a->idiag));
194: PetscCall(PetscFree(a->inode.size_csr));
195: if (a->free_imax_ilen) PetscCall(PetscFree2(a->imax, a->ilen));
196: PetscCall(PetscFree(a->solve_work));
197: PetscCall(PetscFree(a->sor_work));
198: PetscCall(PetscFree(a->solves_work));
199: PetscCall(PetscFree(a->mult_work));
200: PetscCall(PetscFree(a->saved_values));
201: if (a->free_jshort) PetscCall(PetscFree(a->jshort));
202: PetscCall(PetscFree(a->inew));
203: PetscCall(MatDestroy(&a->parent));
204: PetscCall(PetscFree(A->data));
206: PetscCall(PetscObjectChangeTypeName((PetscObject)A, NULL));
207: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqSBAIJGetArray_C", NULL));
208: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqSBAIJRestoreArray_C", NULL));
209: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatStoreValues_C", NULL));
210: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatRetrieveValues_C", NULL));
211: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqSBAIJSetColumnIndices_C", NULL));
212: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqsbaij_seqaij_C", NULL));
213: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqsbaij_seqbaij_C", NULL));
214: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqSBAIJSetPreallocation_C", NULL));
215: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqSBAIJSetPreallocationCSR_C", NULL));
216: #if defined(PETSC_HAVE_ELEMENTAL)
217: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqsbaij_elemental_C", NULL));
218: #endif
219: #if defined(PETSC_HAVE_SCALAPACK)
220: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqsbaij_scalapack_C", NULL));
221: #endif
222: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatFactorGetSolverType_C", NULL));
223: PetscFunctionReturn(PETSC_SUCCESS);
224: }
226: static PetscErrorCode MatSetOption_SeqSBAIJ(Mat A, MatOption op, PetscBool flg)
227: {
228: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
229: #if defined(PETSC_USE_COMPLEX)
230: PetscInt bs;
231: #endif
233: PetscFunctionBegin;
234: #if defined(PETSC_USE_COMPLEX)
235: PetscCall(MatGetBlockSize(A, &bs));
236: #endif
237: switch (op) {
238: case MAT_ROW_ORIENTED:
239: a->roworiented = flg;
240: break;
241: case MAT_KEEP_NONZERO_PATTERN:
242: a->keepnonzeropattern = flg;
243: break;
244: case MAT_NEW_NONZERO_LOCATIONS:
245: a->nonew = (flg ? 0 : 1);
246: break;
247: case MAT_NEW_NONZERO_LOCATION_ERR:
248: a->nonew = (flg ? -1 : 0);
249: break;
250: case MAT_NEW_NONZERO_ALLOCATION_ERR:
251: a->nonew = (flg ? -2 : 0);
252: break;
253: case MAT_UNUSED_NONZERO_LOCATION_ERR:
254: a->nounused = (flg ? -1 : 0);
255: break;
256: case MAT_HERMITIAN:
257: #if defined(PETSC_USE_COMPLEX)
258: if (flg) { /* disable transpose ops */
259: PetscCheck(bs <= 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "No support for Hermitian with block size greater than 1");
260: A->ops->multtranspose = NULL;
261: A->ops->multtransposeadd = NULL;
262: A->symmetric = PETSC_BOOL3_FALSE;
263: }
264: #endif
265: break;
266: case MAT_SYMMETRIC:
267: case MAT_SPD:
268: #if defined(PETSC_USE_COMPLEX)
269: if (flg) { /* An hermitian and symmetric matrix has zero imaginary part (restore back transpose ops) */
270: A->ops->multtranspose = A->ops->mult;
271: A->ops->multtransposeadd = A->ops->multadd;
272: }
273: #endif
274: break;
275: case MAT_IGNORE_LOWER_TRIANGULAR:
276: a->ignore_ltriangular = flg;
277: break;
278: case MAT_ERROR_LOWER_TRIANGULAR:
279: a->ignore_ltriangular = flg;
280: break;
281: case MAT_GETROW_UPPERTRIANGULAR:
282: a->getrow_utriangular = flg;
283: break;
284: default:
285: break;
286: }
287: PetscFunctionReturn(PETSC_SUCCESS);
288: }
290: PetscErrorCode MatGetRow_SeqSBAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
291: {
292: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
294: PetscFunctionBegin;
295: PetscCheck(!A || a->getrow_utriangular, 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()");
297: /* Get the upper triangular part of the row */
298: PetscCall(MatGetRow_SeqBAIJ_private(A, row, nz, idx, v, a->i, a->j, a->a));
299: PetscFunctionReturn(PETSC_SUCCESS);
300: }
302: PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
303: {
304: PetscFunctionBegin;
305: if (idx) PetscCall(PetscFree(*idx));
306: if (v) PetscCall(PetscFree(*v));
307: PetscFunctionReturn(PETSC_SUCCESS);
308: }
310: static PetscErrorCode MatGetRowUpperTriangular_SeqSBAIJ(Mat A)
311: {
312: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
314: PetscFunctionBegin;
315: a->getrow_utriangular = PETSC_TRUE;
316: PetscFunctionReturn(PETSC_SUCCESS);
317: }
319: static PetscErrorCode MatRestoreRowUpperTriangular_SeqSBAIJ(Mat A)
320: {
321: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
323: PetscFunctionBegin;
324: a->getrow_utriangular = PETSC_FALSE;
325: PetscFunctionReturn(PETSC_SUCCESS);
326: }
328: static PetscErrorCode MatTranspose_SeqSBAIJ(Mat A, MatReuse reuse, Mat *B)
329: {
330: PetscFunctionBegin;
331: if (reuse == MAT_REUSE_MATRIX) PetscCall(MatTransposeCheckNonzeroState_Private(A, *B));
332: if (reuse == MAT_INITIAL_MATRIX) {
333: PetscCall(MatDuplicate(A, MAT_COPY_VALUES, B));
334: } else if (reuse == MAT_REUSE_MATRIX) {
335: PetscCall(MatCopy(A, *B, SAME_NONZERO_PATTERN));
336: }
337: PetscFunctionReturn(PETSC_SUCCESS);
338: }
340: static PetscErrorCode MatView_SeqSBAIJ_ASCII(Mat A, PetscViewer viewer)
341: {
342: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
343: PetscInt i, j, bs = A->rmap->bs, k, l, bs2 = a->bs2;
344: PetscViewerFormat format;
345: PetscInt *diag;
346: const char *matname;
348: PetscFunctionBegin;
349: PetscCall(PetscViewerGetFormat(viewer, &format));
350: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
351: PetscCall(PetscViewerASCIIPrintf(viewer, " block size is %" PetscInt_FMT "\n", bs));
352: } else if (format == PETSC_VIEWER_ASCII_MATLAB) {
353: Mat aij;
355: if (A->factortype && bs > 1) {
356: PetscCall(PetscPrintf(PETSC_COMM_SELF, "Warning: matrix is factored with bs>1. MatView() with PETSC_VIEWER_ASCII_MATLAB is not supported and ignored!\n"));
357: PetscFunctionReturn(PETSC_SUCCESS);
358: }
359: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &aij));
360: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
361: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)aij, matname));
362: PetscCall(MatView_SeqAIJ(aij, viewer));
363: PetscCall(MatDestroy(&aij));
364: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
365: Mat B;
367: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &B));
368: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
369: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)B, matname));
370: PetscCall(MatView_SeqAIJ(B, viewer));
371: PetscCall(MatDestroy(&B));
372: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
373: PetscFunctionReturn(PETSC_SUCCESS);
374: } else {
375: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
376: if (A->factortype) { /* for factored matrix */
377: PetscCheck(bs <= 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "matrix is factored with bs>1. Not implemented yet");
379: diag = a->diag;
380: for (i = 0; i < a->mbs; i++) { /* for row block i */
381: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
382: /* diagonal entry */
383: #if defined(PETSC_USE_COMPLEX)
384: if (PetscImaginaryPart(a->a[diag[i]]) > 0.0) {
385: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i) ", a->j[diag[i]], (double)PetscRealPart(1.0 / a->a[diag[i]]), (double)PetscImaginaryPart(1.0 / a->a[diag[i]])));
386: } else if (PetscImaginaryPart(a->a[diag[i]]) < 0.0) {
387: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i) ", a->j[diag[i]], (double)PetscRealPart(1.0 / a->a[diag[i]]), -(double)PetscImaginaryPart(1.0 / a->a[diag[i]])));
388: } else {
389: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[diag[i]], (double)PetscRealPart(1.0 / a->a[diag[i]])));
390: }
391: #else
392: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[diag[i]], (double)(1 / a->a[diag[i]])));
393: #endif
394: /* off-diagonal entries */
395: for (k = a->i[i]; k < a->i[i + 1] - 1; k++) {
396: #if defined(PETSC_USE_COMPLEX)
397: if (PetscImaginaryPart(a->a[k]) > 0.0) {
398: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i) ", bs * a->j[k], (double)PetscRealPart(a->a[k]), (double)PetscImaginaryPart(a->a[k])));
399: } else if (PetscImaginaryPart(a->a[k]) < 0.0) {
400: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i) ", bs * a->j[k], (double)PetscRealPart(a->a[k]), -(double)PetscImaginaryPart(a->a[k])));
401: } else {
402: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k], (double)PetscRealPart(a->a[k])));
403: }
404: #else
405: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[k], (double)a->a[k]));
406: #endif
407: }
408: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
409: }
411: } else { /* for non-factored matrix */
412: for (i = 0; i < a->mbs; i++) { /* for row block i */
413: for (j = 0; j < bs; j++) { /* for row bs*i + j */
414: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i * bs + j));
415: for (k = a->i[i]; k < a->i[i + 1]; k++) { /* for column block */
416: for (l = 0; l < bs; l++) { /* for column */
417: #if defined(PETSC_USE_COMPLEX)
418: if (PetscImaginaryPart(a->a[bs2 * k + l * bs + j]) > 0.0) {
419: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i) ", bs * a->j[k] + l, (double)PetscRealPart(a->a[bs2 * k + l * bs + j]), (double)PetscImaginaryPart(a->a[bs2 * k + l * bs + j])));
420: } else if (PetscImaginaryPart(a->a[bs2 * k + l * bs + j]) < 0.0) {
421: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i) ", bs * a->j[k] + l, (double)PetscRealPart(a->a[bs2 * k + l * bs + j]), -(double)PetscImaginaryPart(a->a[bs2 * k + l * bs + j])));
422: } else {
423: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k] + l, (double)PetscRealPart(a->a[bs2 * k + l * bs + j])));
424: }
425: #else
426: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k] + l, (double)a->a[bs2 * k + l * bs + j]));
427: #endif
428: }
429: }
430: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
431: }
432: }
433: }
434: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
435: }
436: PetscCall(PetscViewerFlush(viewer));
437: PetscFunctionReturn(PETSC_SUCCESS);
438: }
440: #include <petscdraw.h>
441: static PetscErrorCode MatView_SeqSBAIJ_Draw_Zoom(PetscDraw draw, void *Aa)
442: {
443: Mat A = (Mat)Aa;
444: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
445: PetscInt row, i, j, k, l, mbs = a->mbs, bs = A->rmap->bs, bs2 = a->bs2;
446: PetscReal xl, yl, xr, yr, x_l, x_r, y_l, y_r;
447: MatScalar *aa;
448: PetscViewer viewer;
449: int color;
451: PetscFunctionBegin;
452: PetscCall(PetscObjectQuery((PetscObject)A, "Zoomviewer", (PetscObject *)&viewer));
453: PetscCall(PetscDrawGetCoordinates(draw, &xl, &yl, &xr, &yr));
455: /* loop over matrix elements drawing boxes */
457: PetscDrawCollectiveBegin(draw);
458: PetscCall(PetscDrawString(draw, .3 * (xl + xr), .3 * (yl + yr), PETSC_DRAW_BLACK, "symmetric"));
459: /* Blue for negative, Cyan for zero and Red for positive */
460: color = PETSC_DRAW_BLUE;
461: for (i = 0, row = 0; i < mbs; i++, row += bs) {
462: for (j = a->i[i]; j < a->i[i + 1]; j++) {
463: y_l = A->rmap->N - row - 1.0;
464: y_r = y_l + 1.0;
465: x_l = a->j[j] * bs;
466: x_r = x_l + 1.0;
467: aa = a->a + j * bs2;
468: for (k = 0; k < bs; k++) {
469: for (l = 0; l < bs; l++) {
470: if (PetscRealPart(*aa++) >= 0.) continue;
471: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
472: }
473: }
474: }
475: }
476: color = PETSC_DRAW_CYAN;
477: for (i = 0, row = 0; i < mbs; i++, row += bs) {
478: for (j = a->i[i]; j < a->i[i + 1]; j++) {
479: y_l = A->rmap->N - row - 1.0;
480: y_r = y_l + 1.0;
481: x_l = a->j[j] * bs;
482: x_r = x_l + 1.0;
483: aa = a->a + j * bs2;
484: for (k = 0; k < bs; k++) {
485: for (l = 0; l < bs; l++) {
486: if (PetscRealPart(*aa++) != 0.) continue;
487: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
488: }
489: }
490: }
491: }
492: color = PETSC_DRAW_RED;
493: for (i = 0, row = 0; i < mbs; i++, row += bs) {
494: for (j = a->i[i]; j < a->i[i + 1]; j++) {
495: y_l = A->rmap->N - row - 1.0;
496: y_r = y_l + 1.0;
497: x_l = a->j[j] * bs;
498: x_r = x_l + 1.0;
499: aa = a->a + j * bs2;
500: for (k = 0; k < bs; k++) {
501: for (l = 0; l < bs; l++) {
502: if (PetscRealPart(*aa++) <= 0.) continue;
503: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
504: }
505: }
506: }
507: }
508: PetscDrawCollectiveEnd(draw);
509: PetscFunctionReturn(PETSC_SUCCESS);
510: }
512: static PetscErrorCode MatView_SeqSBAIJ_Draw(Mat A, PetscViewer viewer)
513: {
514: PetscReal xl, yl, xr, yr, w, h;
515: PetscDraw draw;
516: PetscBool isnull;
518: PetscFunctionBegin;
519: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
520: PetscCall(PetscDrawIsNull(draw, &isnull));
521: if (isnull) PetscFunctionReturn(PETSC_SUCCESS);
523: xr = A->rmap->N;
524: yr = A->rmap->N;
525: h = yr / 10.0;
526: w = xr / 10.0;
527: xr += w;
528: yr += h;
529: xl = -w;
530: yl = -h;
531: PetscCall(PetscDrawSetCoordinates(draw, xl, yl, xr, yr));
532: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", (PetscObject)viewer));
533: PetscCall(PetscDrawZoom(draw, MatView_SeqSBAIJ_Draw_Zoom, A));
534: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", NULL));
535: PetscCall(PetscDrawSave(draw));
536: PetscFunctionReturn(PETSC_SUCCESS);
537: }
539: /* Used for both MPIBAIJ and MPISBAIJ matrices */
540: #define MatView_SeqSBAIJ_Binary MatView_SeqBAIJ_Binary
542: PetscErrorCode MatView_SeqSBAIJ(Mat A, PetscViewer viewer)
543: {
544: PetscBool iascii, isbinary, isdraw;
546: PetscFunctionBegin;
547: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
548: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
549: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
550: if (iascii) {
551: PetscCall(MatView_SeqSBAIJ_ASCII(A, viewer));
552: } else if (isbinary) {
553: PetscCall(MatView_SeqSBAIJ_Binary(A, viewer));
554: } else if (isdraw) {
555: PetscCall(MatView_SeqSBAIJ_Draw(A, viewer));
556: } else {
557: Mat B;
558: const char *matname;
559: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &B));
560: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
561: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)B, matname));
562: PetscCall(MatView(B, viewer));
563: PetscCall(MatDestroy(&B));
564: }
565: PetscFunctionReturn(PETSC_SUCCESS);
566: }
568: PetscErrorCode MatGetValues_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], PetscScalar v[])
569: {
570: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
571: PetscInt *rp, k, low, high, t, row, nrow, i, col, l, *aj = a->j;
572: PetscInt *ai = a->i, *ailen = a->ilen;
573: PetscInt brow, bcol, ridx, cidx, bs = A->rmap->bs, bs2 = a->bs2;
574: MatScalar *ap, *aa = a->a;
576: PetscFunctionBegin;
577: for (k = 0; k < m; k++) { /* loop over rows */
578: row = im[k];
579: brow = row / bs;
580: if (row < 0) {
581: v += n;
582: continue;
583: } /* negative row */
584: PetscCheck(row < A->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, row, A->rmap->N - 1);
585: rp = aj + ai[brow];
586: ap = aa + bs2 * ai[brow];
587: nrow = ailen[brow];
588: for (l = 0; l < n; l++) { /* loop over columns */
589: if (in[l] < 0) {
590: v++;
591: continue;
592: } /* negative column */
593: PetscCheck(in[l] < A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[l], A->cmap->n - 1);
594: col = in[l];
595: bcol = col / bs;
596: cidx = col % bs;
597: ridx = row % bs;
598: high = nrow;
599: low = 0; /* assume unsorted */
600: while (high - low > 5) {
601: t = (low + high) / 2;
602: if (rp[t] > bcol) high = t;
603: else low = t;
604: }
605: for (i = low; i < high; i++) {
606: if (rp[i] > bcol) break;
607: if (rp[i] == bcol) {
608: *v++ = ap[bs2 * i + bs * cidx + ridx];
609: goto finished;
610: }
611: }
612: *v++ = 0.0;
613: finished:;
614: }
615: }
616: PetscFunctionReturn(PETSC_SUCCESS);
617: }
619: static PetscErrorCode MatPermute_SeqSBAIJ(Mat A, IS rowp, IS colp, Mat *B)
620: {
621: Mat C;
622: PetscBool flg = (PetscBool)(rowp == colp);
624: PetscFunctionBegin;
625: PetscCall(MatConvert(A, MATSEQBAIJ, MAT_INITIAL_MATRIX, &C));
626: PetscCall(MatPermute(C, rowp, colp, B));
627: PetscCall(MatDestroy(&C));
628: if (!flg) PetscCall(ISEqual(rowp, colp, &flg));
629: if (flg) PetscCall(MatConvert(*B, MATSEQSBAIJ, MAT_INPLACE_MATRIX, B));
630: PetscFunctionReturn(PETSC_SUCCESS);
631: }
633: PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
634: {
635: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
636: PetscInt *rp, k, low, high, t, ii, jj, row, nrow, i, col, l, rmax, N, lastcol = -1;
637: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
638: PetscInt *aj = a->j, nonew = a->nonew, bs2 = a->bs2, bs = A->rmap->bs, stepval;
639: PetscBool roworiented = a->roworiented;
640: const PetscScalar *value = v;
641: MatScalar *ap, *aa = a->a, *bap;
643: PetscFunctionBegin;
644: if (roworiented) stepval = (n - 1) * bs;
645: else stepval = (m - 1) * bs;
646: for (k = 0; k < m; k++) { /* loop over added rows */
647: row = im[k];
648: if (row < 0) continue;
649: PetscCheck(row < a->mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block index row too large %" PetscInt_FMT " max %" PetscInt_FMT, row, a->mbs - 1);
650: rp = aj + ai[row];
651: ap = aa + bs2 * ai[row];
652: rmax = imax[row];
653: nrow = ailen[row];
654: low = 0;
655: high = nrow;
656: for (l = 0; l < n; l++) { /* loop over added columns */
657: if (in[l] < 0) continue;
658: col = in[l];
659: PetscCheck(col < a->nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block index column too large %" PetscInt_FMT " max %" PetscInt_FMT, col, a->nbs - 1);
660: if (col < row) {
661: if (a->ignore_ltriangular) continue; /* ignore lower triangular block */
662: 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)");
663: }
664: if (roworiented) value = v + k * (stepval + bs) * bs + l * bs;
665: else value = v + l * (stepval + bs) * bs + k * bs;
667: if (col <= lastcol) low = 0;
668: else high = nrow;
670: lastcol = col;
671: while (high - low > 7) {
672: t = (low + high) / 2;
673: if (rp[t] > col) high = t;
674: else low = t;
675: }
676: for (i = low; i < high; i++) {
677: if (rp[i] > col) break;
678: if (rp[i] == col) {
679: bap = ap + bs2 * i;
680: if (roworiented) {
681: if (is == ADD_VALUES) {
682: for (ii = 0; ii < bs; ii++, value += stepval) {
683: for (jj = ii; jj < bs2; jj += bs) bap[jj] += *value++;
684: }
685: } else {
686: for (ii = 0; ii < bs; ii++, value += stepval) {
687: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
688: }
689: }
690: } else {
691: if (is == ADD_VALUES) {
692: for (ii = 0; ii < bs; ii++, value += stepval) {
693: for (jj = 0; jj < bs; jj++) *bap++ += *value++;
694: }
695: } else {
696: for (ii = 0; ii < bs; ii++, value += stepval) {
697: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
698: }
699: }
700: }
701: goto noinsert2;
702: }
703: }
704: if (nonew == 1) goto noinsert2;
705: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new block index nonzero block (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", row, col);
706: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
707: N = nrow++ - 1;
708: high++;
709: /* shift up all the later entries in this row */
710: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
711: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
712: PetscCall(PetscArrayzero(ap + bs2 * i, bs2));
713: rp[i] = col;
714: bap = ap + bs2 * i;
715: if (roworiented) {
716: for (ii = 0; ii < bs; ii++, value += stepval) {
717: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
718: }
719: } else {
720: for (ii = 0; ii < bs; ii++, value += stepval) {
721: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
722: }
723: }
724: noinsert2:;
725: low = i;
726: }
727: ailen[row] = nrow;
728: }
729: PetscFunctionReturn(PETSC_SUCCESS);
730: }
732: static PetscErrorCode MatAssemblyEnd_SeqSBAIJ(Mat A, MatAssemblyType mode)
733: {
734: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
735: PetscInt fshift = 0, i, *ai = a->i, *aj = a->j, *imax = a->imax;
736: PetscInt m = A->rmap->N, *ip, N, *ailen = a->ilen;
737: PetscInt mbs = a->mbs, bs2 = a->bs2, rmax = 0;
738: MatScalar *aa = a->a, *ap;
740: PetscFunctionBegin;
741: if (mode == MAT_FLUSH_ASSEMBLY || (A->was_assembled && A->ass_nonzerostate == A->nonzerostate)) PetscFunctionReturn(PETSC_SUCCESS);
743: if (m) rmax = ailen[0];
744: for (i = 1; i < mbs; i++) {
745: /* move each row back by the amount of empty slots (fshift) before it*/
746: fshift += imax[i - 1] - ailen[i - 1];
747: rmax = PetscMax(rmax, ailen[i]);
748: if (fshift) {
749: ip = aj + ai[i];
750: ap = aa + bs2 * ai[i];
751: N = ailen[i];
752: PetscCall(PetscArraymove(ip - fshift, ip, N));
753: PetscCall(PetscArraymove(ap - bs2 * fshift, ap, bs2 * N));
754: }
755: ai[i] = ai[i - 1] + ailen[i - 1];
756: }
757: if (mbs) {
758: fshift += imax[mbs - 1] - ailen[mbs - 1];
759: ai[mbs] = ai[mbs - 1] + ailen[mbs - 1];
760: }
761: /* reset ilen and imax for each row */
762: for (i = 0; i < mbs; i++) ailen[i] = imax[i] = ai[i + 1] - ai[i];
763: a->nz = ai[mbs];
765: /* diagonals may have moved, reset it */
766: if (a->diag) PetscCall(PetscArraycpy(a->diag, ai, mbs));
767: PetscCheck(!fshift || a->nounused != -1, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Unused space detected in matrix: %" PetscInt_FMT " X %" PetscInt_FMT " block size %" PetscInt_FMT ", %" PetscInt_FMT " unneeded", m, A->cmap->n, A->rmap->bs, fshift * bs2);
769: PetscCall(PetscInfo(A, "Matrix size: %" PetscInt_FMT " X %" PetscInt_FMT ", block size %" PetscInt_FMT "; storage space: %" PetscInt_FMT " unneeded, %" PetscInt_FMT " used\n", m, A->rmap->N, A->rmap->bs, fshift * bs2, a->nz * bs2));
770: PetscCall(PetscInfo(A, "Number of mallocs during MatSetValues is %" PetscInt_FMT "\n", a->reallocs));
771: PetscCall(PetscInfo(A, "Most nonzeros blocks in any row is %" PetscInt_FMT "\n", rmax));
773: A->info.mallocs += a->reallocs;
774: a->reallocs = 0;
775: A->info.nz_unneeded = (PetscReal)fshift * bs2;
776: a->idiagvalid = PETSC_FALSE;
777: a->rmax = rmax;
779: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
780: if (a->jshort && a->free_jshort) {
781: /* when matrix data structure is changed, previous jshort must be replaced */
782: PetscCall(PetscFree(a->jshort));
783: }
784: PetscCall(PetscMalloc1(a->i[A->rmap->n], &a->jshort));
785: for (i = 0; i < a->i[A->rmap->n]; i++) a->jshort[i] = (short)a->j[i];
786: A->ops->mult = MatMult_SeqSBAIJ_1_ushort;
787: A->ops->sor = MatSOR_SeqSBAIJ_ushort;
788: a->free_jshort = PETSC_TRUE;
789: }
790: PetscFunctionReturn(PETSC_SUCCESS);
791: }
793: /* Only add/insert a(i,j) with i<=j (blocks).
794: Any a(i,j) with i>j input by user is ignored.
795: */
797: PetscErrorCode MatSetValues_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
798: {
799: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
800: PetscInt *rp, k, low, high, t, ii, row, nrow, i, col, l, rmax, N, lastcol = -1;
801: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen, roworiented = a->roworiented;
802: PetscInt *aj = a->j, nonew = a->nonew, bs = A->rmap->bs, brow, bcol;
803: PetscInt ridx, cidx, bs2 = a->bs2;
804: MatScalar *ap, value, *aa = a->a, *bap;
806: PetscFunctionBegin;
807: for (k = 0; k < m; k++) { /* loop over added rows */
808: row = im[k]; /* row number */
809: brow = row / bs; /* block row number */
810: if (row < 0) continue;
811: PetscCheck(row < A->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, row, A->rmap->N - 1);
812: rp = aj + ai[brow]; /*ptr to beginning of column value of the row block*/
813: ap = aa + bs2 * ai[brow]; /*ptr to beginning of element value of the row block*/
814: rmax = imax[brow]; /* maximum space allocated for this row */
815: nrow = ailen[brow]; /* actual length of this row */
816: low = 0;
817: high = nrow;
818: for (l = 0; l < n; l++) { /* loop over added columns */
819: if (in[l] < 0) continue;
820: PetscCheck(in[l] < A->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[l], A->cmap->N - 1);
821: col = in[l];
822: bcol = col / bs; /* block col number */
824: if (brow > bcol) {
825: if (a->ignore_ltriangular) continue; /* ignore lower triangular values */
826: 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)");
827: }
829: ridx = row % bs;
830: cidx = col % bs; /*row and col index inside the block */
831: if ((brow == bcol && ridx <= cidx) || (brow < bcol)) {
832: /* element value a(k,l) */
833: if (roworiented) value = v[l + k * n];
834: else value = v[k + l * m];
836: /* move pointer bap to a(k,l) quickly and add/insert value */
837: if (col <= lastcol) low = 0;
838: else high = nrow;
840: lastcol = col;
841: while (high - low > 7) {
842: t = (low + high) / 2;
843: if (rp[t] > bcol) high = t;
844: else low = t;
845: }
846: for (i = low; i < high; i++) {
847: if (rp[i] > bcol) break;
848: if (rp[i] == bcol) {
849: bap = ap + bs2 * i + bs * cidx + ridx;
850: if (is == ADD_VALUES) *bap += value;
851: else *bap = value;
852: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
853: if (brow == bcol && ridx < cidx) {
854: bap = ap + bs2 * i + bs * ridx + cidx;
855: if (is == ADD_VALUES) *bap += value;
856: else *bap = value;
857: }
858: goto noinsert1;
859: }
860: }
862: if (nonew == 1) goto noinsert1;
863: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", row, col);
864: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, brow, bcol, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
866: N = nrow++ - 1;
867: high++;
868: /* shift up all the later entries in this row */
869: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
870: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
871: PetscCall(PetscArrayzero(ap + bs2 * i, bs2));
872: rp[i] = bcol;
873: ap[bs2 * i + bs * cidx + ridx] = value;
874: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
875: if (brow == bcol && ridx < cidx) ap[bs2 * i + bs * ridx + cidx] = value;
876: noinsert1:;
877: low = i;
878: }
879: } /* end of loop over added columns */
880: ailen[brow] = nrow;
881: } /* end of loop over added rows */
882: PetscFunctionReturn(PETSC_SUCCESS);
883: }
885: static PetscErrorCode MatICCFactor_SeqSBAIJ(Mat inA, IS row, const MatFactorInfo *info)
886: {
887: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)inA->data;
888: Mat outA;
889: PetscBool row_identity;
891: PetscFunctionBegin;
892: PetscCheck(info->levels == 0, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only levels=0 is supported for in-place icc");
893: PetscCall(ISIdentity(row, &row_identity));
894: PetscCheck(row_identity, PETSC_COMM_SELF, PETSC_ERR_SUP, "Matrix reordering is not supported");
895: PetscCheck(inA->rmap->bs == 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "Matrix block size %" PetscInt_FMT " is not supported", inA->rmap->bs); /* Need to replace MatCholeskyFactorSymbolic_SeqSBAIJ_MSR()! */
897: outA = inA;
898: inA->factortype = MAT_FACTOR_ICC;
899: PetscCall(PetscFree(inA->solvertype));
900: PetscCall(PetscStrallocpy(MATSOLVERPETSC, &inA->solvertype));
902: PetscCall(MatMarkDiagonal_SeqSBAIJ(inA));
903: PetscCall(MatSeqSBAIJSetNumericFactorization_inplace(inA, row_identity));
905: PetscCall(PetscObjectReference((PetscObject)row));
906: PetscCall(ISDestroy(&a->row));
907: a->row = row;
908: PetscCall(PetscObjectReference((PetscObject)row));
909: PetscCall(ISDestroy(&a->col));
910: a->col = row;
912: /* Create the invert permutation so that it can be used in MatCholeskyFactorNumeric() */
913: if (a->icol) PetscCall(ISInvertPermutation(row, PETSC_DECIDE, &a->icol));
915: if (!a->solve_work) PetscCall(PetscMalloc1(inA->rmap->N + inA->rmap->bs, &a->solve_work));
917: PetscCall(MatCholeskyFactorNumeric(outA, inA, info));
918: PetscFunctionReturn(PETSC_SUCCESS);
919: }
921: static PetscErrorCode MatSeqSBAIJSetColumnIndices_SeqSBAIJ(Mat mat, PetscInt *indices)
922: {
923: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ *)mat->data;
924: PetscInt i, nz, n;
926: PetscFunctionBegin;
927: nz = baij->maxnz;
928: n = mat->cmap->n;
929: for (i = 0; i < nz; i++) baij->j[i] = indices[i];
931: baij->nz = nz;
932: for (i = 0; i < n; i++) baij->ilen[i] = baij->imax[i];
934: PetscCall(MatSetOption(mat, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
935: PetscFunctionReturn(PETSC_SUCCESS);
936: }
938: /*@
939: MatSeqSBAIJSetColumnIndices - Set the column indices for all the rows
940: in a `MATSEQSBAIJ` matrix.
942: Input Parameters:
943: + mat - the `MATSEQSBAIJ` matrix
944: - indices - the column indices
946: Level: advanced
948: Notes:
949: This can be called if you have precomputed the nonzero structure of the
950: matrix and want to provide it to the matrix object to improve the performance
951: of the `MatSetValues()` operation.
953: You MUST have set the correct numbers of nonzeros per row in the call to
954: `MatCreateSeqSBAIJ()`, and the columns indices MUST be sorted.
956: MUST be called before any calls to `MatSetValues()`
958: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreateSeqSBAIJ`
959: @*/
960: PetscErrorCode MatSeqSBAIJSetColumnIndices(Mat mat, PetscInt *indices)
961: {
962: PetscFunctionBegin;
964: PetscAssertPointer(indices, 2);
965: PetscUseMethod(mat, "MatSeqSBAIJSetColumnIndices_C", (Mat, PetscInt *), (mat, indices));
966: PetscFunctionReturn(PETSC_SUCCESS);
967: }
969: static PetscErrorCode MatCopy_SeqSBAIJ(Mat A, Mat B, MatStructure str)
970: {
971: PetscBool isbaij;
973: PetscFunctionBegin;
974: PetscCall(PetscObjectTypeCompareAny((PetscObject)B, &isbaij, MATSEQSBAIJ, MATMPISBAIJ, ""));
975: PetscCheck(isbaij, PetscObjectComm((PetscObject)B), PETSC_ERR_SUP, "Not for matrix type %s", ((PetscObject)B)->type_name);
976: /* If the two matrices have the same copy implementation and nonzero pattern, use fast copy. */
977: if (str == SAME_NONZERO_PATTERN && A->ops->copy == B->ops->copy) {
978: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
979: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
981: PetscCheck(a->i[a->mbs] == b->i[b->mbs], PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Number of nonzeros in two matrices are different");
982: PetscCheck(a->mbs == b->mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Number of rows in two matrices are different");
983: PetscCheck(a->bs2 == b->bs2, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Different block size");
984: PetscCall(PetscArraycpy(b->a, a->a, a->bs2 * a->i[a->mbs]));
985: PetscCall(PetscObjectStateIncrease((PetscObject)B));
986: } else {
987: PetscCall(MatGetRowUpperTriangular(A));
988: PetscCall(MatCopy_Basic(A, B, str));
989: PetscCall(MatRestoreRowUpperTriangular(A));
990: }
991: PetscFunctionReturn(PETSC_SUCCESS);
992: }
994: static PetscErrorCode MatSeqSBAIJGetArray_SeqSBAIJ(Mat A, PetscScalar *array[])
995: {
996: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
998: PetscFunctionBegin;
999: *array = a->a;
1000: PetscFunctionReturn(PETSC_SUCCESS);
1001: }
1003: static PetscErrorCode MatSeqSBAIJRestoreArray_SeqSBAIJ(Mat A, PetscScalar *array[])
1004: {
1005: PetscFunctionBegin;
1006: *array = NULL;
1007: PetscFunctionReturn(PETSC_SUCCESS);
1008: }
1010: PetscErrorCode MatAXPYGetPreallocation_SeqSBAIJ(Mat Y, Mat X, PetscInt *nnz)
1011: {
1012: PetscInt bs = Y->rmap->bs, mbs = Y->rmap->N / bs;
1013: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ *)X->data;
1014: Mat_SeqSBAIJ *y = (Mat_SeqSBAIJ *)Y->data;
1016: PetscFunctionBegin;
1017: /* Set the number of nonzeros in the new matrix */
1018: PetscCall(MatAXPYGetPreallocation_SeqX_private(mbs, x->i, x->j, y->i, y->j, nnz));
1019: PetscFunctionReturn(PETSC_SUCCESS);
1020: }
1022: static PetscErrorCode MatAXPY_SeqSBAIJ(Mat Y, PetscScalar a, Mat X, MatStructure str)
1023: {
1024: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ *)X->data, *y = (Mat_SeqSBAIJ *)Y->data;
1025: PetscInt bs = Y->rmap->bs, bs2 = bs * bs;
1026: PetscBLASInt one = 1;
1028: PetscFunctionBegin;
1029: if (str == UNKNOWN_NONZERO_PATTERN || (PetscDefined(USE_DEBUG) && str == SAME_NONZERO_PATTERN)) {
1030: PetscBool e = x->nz == y->nz && x->mbs == y->mbs ? PETSC_TRUE : PETSC_FALSE;
1031: if (e) {
1032: PetscCall(PetscArraycmp(x->i, y->i, x->mbs + 1, &e));
1033: if (e) {
1034: PetscCall(PetscArraycmp(x->j, y->j, x->i[x->mbs], &e));
1035: if (e) str = SAME_NONZERO_PATTERN;
1036: }
1037: }
1038: if (!e) PetscCheck(str != SAME_NONZERO_PATTERN, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MatStructure is not SAME_NONZERO_PATTERN");
1039: }
1040: if (str == SAME_NONZERO_PATTERN) {
1041: PetscScalar alpha = a;
1042: PetscBLASInt bnz;
1043: PetscCall(PetscBLASIntCast(x->nz * bs2, &bnz));
1044: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, x->a, &one, y->a, &one));
1045: PetscCall(PetscObjectStateIncrease((PetscObject)Y));
1046: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1047: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_TRUE));
1048: PetscCall(MatAXPY_Basic(Y, a, X, str));
1049: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_FALSE));
1050: } else {
1051: Mat B;
1052: PetscInt *nnz;
1053: PetscCheck(bs == X->rmap->bs, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Matrices must have same block size");
1054: PetscCall(MatGetRowUpperTriangular(X));
1055: PetscCall(MatGetRowUpperTriangular(Y));
1056: PetscCall(PetscMalloc1(Y->rmap->N, &nnz));
1057: PetscCall(MatCreate(PetscObjectComm((PetscObject)Y), &B));
1058: PetscCall(PetscObjectSetName((PetscObject)B, ((PetscObject)Y)->name));
1059: PetscCall(MatSetSizes(B, Y->rmap->n, Y->cmap->n, Y->rmap->N, Y->cmap->N));
1060: PetscCall(MatSetBlockSizesFromMats(B, Y, Y));
1061: PetscCall(MatSetType(B, ((PetscObject)Y)->type_name));
1062: PetscCall(MatAXPYGetPreallocation_SeqSBAIJ(Y, X, nnz));
1063: PetscCall(MatSeqSBAIJSetPreallocation(B, bs, 0, nnz));
1065: PetscCall(MatAXPY_BasicWithPreallocation(B, Y, a, X, str));
1067: PetscCall(MatHeaderMerge(Y, &B));
1068: PetscCall(PetscFree(nnz));
1069: PetscCall(MatRestoreRowUpperTriangular(X));
1070: PetscCall(MatRestoreRowUpperTriangular(Y));
1071: }
1072: PetscFunctionReturn(PETSC_SUCCESS);
1073: }
1075: static PetscErrorCode MatIsStructurallySymmetric_SeqSBAIJ(Mat A, PetscBool *flg)
1076: {
1077: PetscFunctionBegin;
1078: *flg = PETSC_TRUE;
1079: PetscFunctionReturn(PETSC_SUCCESS);
1080: }
1082: static PetscErrorCode MatConjugate_SeqSBAIJ(Mat A)
1083: {
1084: #if defined(PETSC_USE_COMPLEX)
1085: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1086: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1087: MatScalar *aa = a->a;
1089: PetscFunctionBegin;
1090: for (i = 0; i < nz; i++) aa[i] = PetscConj(aa[i]);
1091: #else
1092: PetscFunctionBegin;
1093: #endif
1094: PetscFunctionReturn(PETSC_SUCCESS);
1095: }
1097: static PetscErrorCode MatRealPart_SeqSBAIJ(Mat A)
1098: {
1099: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1100: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1101: MatScalar *aa = a->a;
1103: PetscFunctionBegin;
1104: for (i = 0; i < nz; i++) aa[i] = PetscRealPart(aa[i]);
1105: PetscFunctionReturn(PETSC_SUCCESS);
1106: }
1108: static PetscErrorCode MatImaginaryPart_SeqSBAIJ(Mat A)
1109: {
1110: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1111: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1112: MatScalar *aa = a->a;
1114: PetscFunctionBegin;
1115: for (i = 0; i < nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1116: PetscFunctionReturn(PETSC_SUCCESS);
1117: }
1119: static PetscErrorCode MatZeroRowsColumns_SeqSBAIJ(Mat A, PetscInt is_n, const PetscInt is_idx[], PetscScalar diag, Vec x, Vec b)
1120: {
1121: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ *)A->data;
1122: PetscInt i, j, k, count;
1123: PetscInt bs = A->rmap->bs, bs2 = baij->bs2, row, col;
1124: PetscScalar zero = 0.0;
1125: MatScalar *aa;
1126: const PetscScalar *xx;
1127: PetscScalar *bb;
1128: PetscBool *zeroed, vecs = PETSC_FALSE;
1130: PetscFunctionBegin;
1131: /* fix right-hand side if needed */
1132: if (x && b) {
1133: PetscCall(VecGetArrayRead(x, &xx));
1134: PetscCall(VecGetArray(b, &bb));
1135: vecs = PETSC_TRUE;
1136: }
1138: /* zero the columns */
1139: PetscCall(PetscCalloc1(A->rmap->n, &zeroed));
1140: for (i = 0; i < is_n; i++) {
1141: PetscCheck(is_idx[i] >= 0 && is_idx[i] < A->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", is_idx[i]);
1142: zeroed[is_idx[i]] = PETSC_TRUE;
1143: }
1144: if (vecs) {
1145: for (i = 0; i < A->rmap->N; i++) {
1146: row = i / bs;
1147: for (j = baij->i[row]; j < baij->i[row + 1]; j++) {
1148: for (k = 0; k < bs; k++) {
1149: col = bs * baij->j[j] + k;
1150: if (col <= i) continue;
1151: aa = baij->a + j * bs2 + (i % bs) + bs * k;
1152: if (!zeroed[i] && zeroed[col]) bb[i] -= aa[0] * xx[col];
1153: if (zeroed[i] && !zeroed[col]) bb[col] -= aa[0] * xx[i];
1154: }
1155: }
1156: }
1157: for (i = 0; i < is_n; i++) bb[is_idx[i]] = diag * xx[is_idx[i]];
1158: }
1160: for (i = 0; i < A->rmap->N; i++) {
1161: if (!zeroed[i]) {
1162: row = i / bs;
1163: for (j = baij->i[row]; j < baij->i[row + 1]; j++) {
1164: for (k = 0; k < bs; k++) {
1165: col = bs * baij->j[j] + k;
1166: if (zeroed[col]) {
1167: aa = baij->a + j * bs2 + (i % bs) + bs * k;
1168: aa[0] = 0.0;
1169: }
1170: }
1171: }
1172: }
1173: }
1174: PetscCall(PetscFree(zeroed));
1175: if (vecs) {
1176: PetscCall(VecRestoreArrayRead(x, &xx));
1177: PetscCall(VecRestoreArray(b, &bb));
1178: }
1180: /* zero the rows */
1181: for (i = 0; i < is_n; i++) {
1182: row = is_idx[i];
1183: count = (baij->i[row / bs + 1] - baij->i[row / bs]) * bs;
1184: aa = baij->a + baij->i[row / bs] * bs2 + (row % bs);
1185: for (k = 0; k < count; k++) {
1186: aa[0] = zero;
1187: aa += bs;
1188: }
1189: if (diag != 0.0) PetscUseTypeMethod(A, setvalues, 1, &row, 1, &row, &diag, INSERT_VALUES);
1190: }
1191: PetscCall(MatAssemblyEnd_SeqSBAIJ(A, MAT_FINAL_ASSEMBLY));
1192: PetscFunctionReturn(PETSC_SUCCESS);
1193: }
1195: static PetscErrorCode MatShift_SeqSBAIJ(Mat Y, PetscScalar a)
1196: {
1197: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)Y->data;
1199: PetscFunctionBegin;
1200: if (!Y->preallocated || !aij->nz) PetscCall(MatSeqSBAIJSetPreallocation(Y, Y->rmap->bs, 1, NULL));
1201: PetscCall(MatShift_Basic(Y, a));
1202: PetscFunctionReturn(PETSC_SUCCESS);
1203: }
1205: PetscErrorCode MatEliminateZeros_SeqSBAIJ(Mat A, PetscBool keep)
1206: {
1207: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1208: PetscInt fshift = 0, fshift_prev = 0, i, *ai = a->i, *aj = a->j, *imax = a->imax, j, k;
1209: PetscInt m = A->rmap->N, *ailen = a->ilen;
1210: PetscInt mbs = a->mbs, bs2 = a->bs2, rmax = 0;
1211: MatScalar *aa = a->a, *ap;
1212: PetscBool zero;
1214: PetscFunctionBegin;
1215: PetscCheck(A->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot eliminate zeros for unassembled matrix");
1216: if (m) rmax = ailen[0];
1217: for (i = 1; i <= mbs; i++) {
1218: for (k = ai[i - 1]; k < ai[i]; k++) {
1219: zero = PETSC_TRUE;
1220: ap = aa + bs2 * k;
1221: for (j = 0; j < bs2 && zero; j++) {
1222: if (ap[j] != 0.0) zero = PETSC_FALSE;
1223: }
1224: if (zero && (aj[k] != i - 1 || !keep)) fshift++;
1225: else {
1226: if (zero && aj[k] == i - 1) PetscCall(PetscInfo(A, "Keep the diagonal block at row %" PetscInt_FMT "\n", i - 1));
1227: aj[k - fshift] = aj[k];
1228: PetscCall(PetscArraymove(ap - bs2 * fshift, ap, bs2));
1229: }
1230: }
1231: ai[i - 1] -= fshift_prev;
1232: fshift_prev = fshift;
1233: ailen[i - 1] = imax[i - 1] = ai[i] - fshift - ai[i - 1];
1234: a->nonzerorowcnt += ((ai[i] - fshift - ai[i - 1]) > 0);
1235: rmax = PetscMax(rmax, ailen[i - 1]);
1236: }
1237: if (fshift) {
1238: if (mbs) {
1239: ai[mbs] -= fshift;
1240: a->nz = ai[mbs];
1241: }
1242: PetscCall(PetscInfo(A, "Matrix size: %" PetscInt_FMT " X %" PetscInt_FMT "; zeros eliminated: %" PetscInt_FMT "; nonzeros left: %" PetscInt_FMT "\n", m, A->cmap->n, fshift, a->nz));
1243: A->nonzerostate++;
1244: A->info.nz_unneeded += (PetscReal)fshift;
1245: a->rmax = rmax;
1246: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
1247: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
1248: }
1249: PetscFunctionReturn(PETSC_SUCCESS);
1250: }
1252: static struct _MatOps MatOps_Values = {MatSetValues_SeqSBAIJ,
1253: MatGetRow_SeqSBAIJ,
1254: MatRestoreRow_SeqSBAIJ,
1255: MatMult_SeqSBAIJ_N,
1256: /* 4*/ MatMultAdd_SeqSBAIJ_N,
1257: MatMult_SeqSBAIJ_N, /* transpose versions are same as non-transpose versions */
1258: MatMultAdd_SeqSBAIJ_N,
1259: NULL,
1260: NULL,
1261: NULL,
1262: /* 10*/ NULL,
1263: NULL,
1264: MatCholeskyFactor_SeqSBAIJ,
1265: MatSOR_SeqSBAIJ,
1266: MatTranspose_SeqSBAIJ,
1267: /* 15*/ MatGetInfo_SeqSBAIJ,
1268: MatEqual_SeqSBAIJ,
1269: MatGetDiagonal_SeqSBAIJ,
1270: MatDiagonalScale_SeqSBAIJ,
1271: MatNorm_SeqSBAIJ,
1272: /* 20*/ NULL,
1273: MatAssemblyEnd_SeqSBAIJ,
1274: MatSetOption_SeqSBAIJ,
1275: MatZeroEntries_SeqSBAIJ,
1276: /* 24*/ NULL,
1277: NULL,
1278: NULL,
1279: NULL,
1280: NULL,
1281: /* 29*/ MatSetUp_Seq_Hash,
1282: NULL,
1283: NULL,
1284: NULL,
1285: NULL,
1286: /* 34*/ MatDuplicate_SeqSBAIJ,
1287: NULL,
1288: NULL,
1289: NULL,
1290: MatICCFactor_SeqSBAIJ,
1291: /* 39*/ MatAXPY_SeqSBAIJ,
1292: MatCreateSubMatrices_SeqSBAIJ,
1293: MatIncreaseOverlap_SeqSBAIJ,
1294: MatGetValues_SeqSBAIJ,
1295: MatCopy_SeqSBAIJ,
1296: /* 44*/ NULL,
1297: MatScale_SeqSBAIJ,
1298: MatShift_SeqSBAIJ,
1299: NULL,
1300: MatZeroRowsColumns_SeqSBAIJ,
1301: /* 49*/ NULL,
1302: MatGetRowIJ_SeqSBAIJ,
1303: MatRestoreRowIJ_SeqSBAIJ,
1304: NULL,
1305: NULL,
1306: /* 54*/ NULL,
1307: NULL,
1308: NULL,
1309: MatPermute_SeqSBAIJ,
1310: MatSetValuesBlocked_SeqSBAIJ,
1311: /* 59*/ MatCreateSubMatrix_SeqSBAIJ,
1312: NULL,
1313: NULL,
1314: NULL,
1315: NULL,
1316: /* 64*/ NULL,
1317: NULL,
1318: NULL,
1319: NULL,
1320: NULL,
1321: /* 69*/ MatGetRowMaxAbs_SeqSBAIJ,
1322: NULL,
1323: MatConvert_MPISBAIJ_Basic,
1324: NULL,
1325: NULL,
1326: /* 74*/ NULL,
1327: NULL,
1328: NULL,
1329: NULL,
1330: NULL,
1331: /* 79*/ NULL,
1332: NULL,
1333: NULL,
1334: MatGetInertia_SeqSBAIJ,
1335: MatLoad_SeqSBAIJ,
1336: /* 84*/ NULL,
1337: NULL,
1338: MatIsStructurallySymmetric_SeqSBAIJ,
1339: NULL,
1340: NULL,
1341: /* 89*/ NULL,
1342: NULL,
1343: NULL,
1344: NULL,
1345: NULL,
1346: /* 94*/ NULL,
1347: NULL,
1348: NULL,
1349: NULL,
1350: NULL,
1351: /* 99*/ NULL,
1352: NULL,
1353: NULL,
1354: MatConjugate_SeqSBAIJ,
1355: NULL,
1356: /*104*/ NULL,
1357: MatRealPart_SeqSBAIJ,
1358: MatImaginaryPart_SeqSBAIJ,
1359: MatGetRowUpperTriangular_SeqSBAIJ,
1360: MatRestoreRowUpperTriangular_SeqSBAIJ,
1361: /*109*/ NULL,
1362: NULL,
1363: NULL,
1364: NULL,
1365: MatMissingDiagonal_SeqSBAIJ,
1366: /*114*/ NULL,
1367: NULL,
1368: NULL,
1369: NULL,
1370: NULL,
1371: /*119*/ NULL,
1372: NULL,
1373: NULL,
1374: NULL,
1375: NULL,
1376: /*124*/ NULL,
1377: NULL,
1378: NULL,
1379: NULL,
1380: NULL,
1381: /*129*/ NULL,
1382: NULL,
1383: NULL,
1384: NULL,
1385: NULL,
1386: /*134*/ NULL,
1387: NULL,
1388: NULL,
1389: NULL,
1390: NULL,
1391: /*139*/ MatSetBlockSizes_Default,
1392: NULL,
1393: NULL,
1394: NULL,
1395: NULL,
1396: /*144*/ MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ,
1397: NULL,
1398: NULL,
1399: NULL,
1400: NULL,
1401: NULL,
1402: /*150*/ NULL,
1403: MatEliminateZeros_SeqSBAIJ,
1404: NULL,
1405: NULL,
1406: NULL,
1407: /*155*/ NULL,
1408: MatCopyHashToXAIJ_Seq_Hash};
1410: static PetscErrorCode MatStoreValues_SeqSBAIJ(Mat mat)
1411: {
1412: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)mat->data;
1413: PetscInt nz = aij->i[mat->rmap->N] * mat->rmap->bs * aij->bs2;
1415: PetscFunctionBegin;
1416: PetscCheck(aij->nonew == 1, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1418: /* allocate space for values if not already there */
1419: if (!aij->saved_values) PetscCall(PetscMalloc1(nz + 1, &aij->saved_values));
1421: /* copy values over */
1422: PetscCall(PetscArraycpy(aij->saved_values, aij->a, nz));
1423: PetscFunctionReturn(PETSC_SUCCESS);
1424: }
1426: static PetscErrorCode MatRetrieveValues_SeqSBAIJ(Mat mat)
1427: {
1428: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)mat->data;
1429: PetscInt nz = aij->i[mat->rmap->N] * mat->rmap->bs * aij->bs2;
1431: PetscFunctionBegin;
1432: PetscCheck(aij->nonew == 1, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1433: PetscCheck(aij->saved_values, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatStoreValues(A);first");
1435: /* copy values over */
1436: PetscCall(PetscArraycpy(aij->a, aij->saved_values, nz));
1437: PetscFunctionReturn(PETSC_SUCCESS);
1438: }
1440: static PetscErrorCode MatSeqSBAIJSetPreallocation_SeqSBAIJ(Mat B, PetscInt bs, PetscInt nz, const PetscInt nnz[])
1441: {
1442: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
1443: PetscInt i, mbs, nbs, bs2;
1444: PetscBool skipallocation = PETSC_FALSE, flg = PETSC_FALSE, realalloc = PETSC_FALSE;
1446: PetscFunctionBegin;
1447: if (B->hash_active) {
1448: PetscInt bs;
1449: B->ops[0] = b->cops;
1450: PetscCall(PetscHMapIJVDestroy(&b->ht));
1451: PetscCall(MatGetBlockSize(B, &bs));
1452: if (bs > 1) PetscCall(PetscHSetIJDestroy(&b->bht));
1453: PetscCall(PetscFree(b->dnz));
1454: PetscCall(PetscFree(b->bdnz));
1455: B->hash_active = PETSC_FALSE;
1456: }
1457: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
1459: PetscCall(MatSetBlockSize(B, bs));
1460: PetscCall(PetscLayoutSetUp(B->rmap));
1461: PetscCall(PetscLayoutSetUp(B->cmap));
1462: PetscCheck(B->rmap->N <= B->cmap->N, PETSC_COMM_SELF, PETSC_ERR_SUP, "SEQSBAIJ matrix cannot have more rows %" PetscInt_FMT " than columns %" PetscInt_FMT, B->rmap->N, B->cmap->N);
1463: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
1465: B->preallocated = PETSC_TRUE;
1467: mbs = B->rmap->N / bs;
1468: nbs = B->cmap->n / bs;
1469: bs2 = bs * bs;
1471: PetscCheck(mbs * bs == B->rmap->N && nbs * bs == B->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Number rows, cols must be divisible by blocksize");
1473: if (nz == MAT_SKIP_ALLOCATION) {
1474: skipallocation = PETSC_TRUE;
1475: nz = 0;
1476: }
1478: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 3;
1479: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nz cannot be less than 0: value %" PetscInt_FMT, nz);
1480: if (nnz) {
1481: for (i = 0; i < mbs; i++) {
1482: PetscCheck(nnz[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nnz cannot be less than 0: local row %" PetscInt_FMT " value %" PetscInt_FMT, i, nnz[i]);
1483: PetscCheck(nnz[i] <= nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nnz cannot be greater than block row length: local row %" PetscInt_FMT " value %" PetscInt_FMT " block rowlength %" PetscInt_FMT, i, nnz[i], nbs);
1484: }
1485: }
1487: B->ops->mult = MatMult_SeqSBAIJ_N;
1488: B->ops->multadd = MatMultAdd_SeqSBAIJ_N;
1489: B->ops->multtranspose = MatMult_SeqSBAIJ_N;
1490: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_N;
1492: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_no_unroll", &flg, NULL));
1493: if (!flg) {
1494: switch (bs) {
1495: case 1:
1496: B->ops->mult = MatMult_SeqSBAIJ_1;
1497: B->ops->multadd = MatMultAdd_SeqSBAIJ_1;
1498: B->ops->multtranspose = MatMult_SeqSBAIJ_1;
1499: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_1;
1500: break;
1501: case 2:
1502: B->ops->mult = MatMult_SeqSBAIJ_2;
1503: B->ops->multadd = MatMultAdd_SeqSBAIJ_2;
1504: B->ops->multtranspose = MatMult_SeqSBAIJ_2;
1505: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_2;
1506: break;
1507: case 3:
1508: B->ops->mult = MatMult_SeqSBAIJ_3;
1509: B->ops->multadd = MatMultAdd_SeqSBAIJ_3;
1510: B->ops->multtranspose = MatMult_SeqSBAIJ_3;
1511: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_3;
1512: break;
1513: case 4:
1514: B->ops->mult = MatMult_SeqSBAIJ_4;
1515: B->ops->multadd = MatMultAdd_SeqSBAIJ_4;
1516: B->ops->multtranspose = MatMult_SeqSBAIJ_4;
1517: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_4;
1518: break;
1519: case 5:
1520: B->ops->mult = MatMult_SeqSBAIJ_5;
1521: B->ops->multadd = MatMultAdd_SeqSBAIJ_5;
1522: B->ops->multtranspose = MatMult_SeqSBAIJ_5;
1523: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_5;
1524: break;
1525: case 6:
1526: B->ops->mult = MatMult_SeqSBAIJ_6;
1527: B->ops->multadd = MatMultAdd_SeqSBAIJ_6;
1528: B->ops->multtranspose = MatMult_SeqSBAIJ_6;
1529: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_6;
1530: break;
1531: case 7:
1532: B->ops->mult = MatMult_SeqSBAIJ_7;
1533: B->ops->multadd = MatMultAdd_SeqSBAIJ_7;
1534: B->ops->multtranspose = MatMult_SeqSBAIJ_7;
1535: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_7;
1536: break;
1537: }
1538: }
1540: b->mbs = mbs;
1541: b->nbs = nbs;
1542: if (!skipallocation) {
1543: if (!b->imax) {
1544: PetscCall(PetscMalloc2(mbs, &b->imax, mbs, &b->ilen));
1546: b->free_imax_ilen = PETSC_TRUE;
1547: }
1548: if (!nnz) {
1549: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
1550: else if (nz <= 0) nz = 1;
1551: nz = PetscMin(nbs, nz);
1552: for (i = 0; i < mbs; i++) b->imax[i] = nz;
1553: PetscCall(PetscIntMultError(nz, mbs, &nz));
1554: } else {
1555: PetscInt64 nz64 = 0;
1556: for (i = 0; i < mbs; i++) {
1557: b->imax[i] = nnz[i];
1558: nz64 += nnz[i];
1559: }
1560: PetscCall(PetscIntCast(nz64, &nz));
1561: }
1562: /* b->ilen will count nonzeros in each block row so far. */
1563: for (i = 0; i < mbs; i++) b->ilen[i] = 0;
1564: /* nz=(nz+mbs)/2; */ /* total diagonal and superdiagonal nonzero blocks */
1566: /* allocate the matrix space */
1567: PetscCall(MatSeqXAIJFreeAIJ(B, &b->a, &b->j, &b->i));
1568: PetscCall(PetscShmgetAllocateArray(bs2 * nz, sizeof(PetscScalar), (void **)&b->a));
1569: PetscCall(PetscShmgetAllocateArray(nz, sizeof(PetscInt), (void **)&b->j));
1570: PetscCall(PetscShmgetAllocateArray(B->rmap->n + 1, sizeof(PetscInt), (void **)&b->i));
1571: b->free_a = PETSC_TRUE;
1572: b->free_ij = PETSC_TRUE;
1573: PetscCall(PetscArrayzero(b->a, nz * bs2));
1574: PetscCall(PetscArrayzero(b->j, nz));
1575: b->free_a = PETSC_TRUE;
1576: b->free_ij = PETSC_TRUE;
1578: /* pointer to beginning of each row */
1579: b->i[0] = 0;
1580: for (i = 1; i < mbs + 1; i++) b->i[i] = b->i[i - 1] + b->imax[i - 1];
1582: } else {
1583: b->free_a = PETSC_FALSE;
1584: b->free_ij = PETSC_FALSE;
1585: }
1587: b->bs2 = bs2;
1588: b->nz = 0;
1589: b->maxnz = nz;
1590: b->inew = NULL;
1591: b->jnew = NULL;
1592: b->anew = NULL;
1593: b->a2anew = NULL;
1594: b->permute = PETSC_FALSE;
1596: B->was_assembled = PETSC_FALSE;
1597: B->assembled = PETSC_FALSE;
1598: if (realalloc) PetscCall(MatSetOption(B, MAT_NEW_NONZERO_ALLOCATION_ERR, PETSC_TRUE));
1599: PetscFunctionReturn(PETSC_SUCCESS);
1600: }
1602: static PetscErrorCode MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ(Mat B, PetscInt bs, const PetscInt ii[], const PetscInt jj[], const PetscScalar V[])
1603: {
1604: PetscInt i, j, m, nz, anz, nz_max = 0, *nnz;
1605: PetscScalar *values = NULL;
1606: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
1607: PetscBool roworiented = b->roworiented;
1608: PetscBool ilw = b->ignore_ltriangular;
1610: PetscFunctionBegin;
1611: PetscCheck(bs >= 1, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_OUTOFRANGE, "Invalid block size specified, must be positive but it is %" PetscInt_FMT, bs);
1612: PetscCall(PetscLayoutSetBlockSize(B->rmap, bs));
1613: PetscCall(PetscLayoutSetBlockSize(B->cmap, bs));
1614: PetscCall(PetscLayoutSetUp(B->rmap));
1615: PetscCall(PetscLayoutSetUp(B->cmap));
1616: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
1617: m = B->rmap->n / bs;
1619: PetscCheck(!ii[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "ii[0] must be 0 but it is %" PetscInt_FMT, ii[0]);
1620: PetscCall(PetscMalloc1(m + 1, &nnz));
1621: for (i = 0; i < m; i++) {
1622: nz = ii[i + 1] - ii[i];
1623: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row %" PetscInt_FMT " has a negative number of columns %" PetscInt_FMT, i, nz);
1624: PetscCheckSorted(nz, jj + ii[i]);
1625: anz = 0;
1626: for (j = 0; j < nz; j++) {
1627: /* count only values on the diagonal or above */
1628: if (jj[ii[i] + j] >= i) {
1629: anz = nz - j;
1630: break;
1631: }
1632: }
1633: nz_max = PetscMax(nz_max, nz);
1634: nnz[i] = anz;
1635: }
1636: PetscCall(MatSeqSBAIJSetPreallocation(B, bs, 0, nnz));
1637: PetscCall(PetscFree(nnz));
1639: values = (PetscScalar *)V;
1640: if (!values) PetscCall(PetscCalloc1(bs * bs * nz_max, &values));
1641: b->ignore_ltriangular = PETSC_TRUE;
1642: for (i = 0; i < m; i++) {
1643: PetscInt ncols = ii[i + 1] - ii[i];
1644: const PetscInt *icols = jj + ii[i];
1646: if (!roworiented || bs == 1) {
1647: const PetscScalar *svals = values + (V ? (bs * bs * ii[i]) : 0);
1648: PetscCall(MatSetValuesBlocked_SeqSBAIJ(B, 1, &i, ncols, icols, svals, INSERT_VALUES));
1649: } else {
1650: for (j = 0; j < ncols; j++) {
1651: const PetscScalar *svals = values + (V ? (bs * bs * (ii[i] + j)) : 0);
1652: PetscCall(MatSetValuesBlocked_SeqSBAIJ(B, 1, &i, 1, &icols[j], svals, INSERT_VALUES));
1653: }
1654: }
1655: }
1656: if (!V) PetscCall(PetscFree(values));
1657: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
1658: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
1659: PetscCall(MatSetOption(B, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
1660: b->ignore_ltriangular = ilw;
1661: PetscFunctionReturn(PETSC_SUCCESS);
1662: }
1664: /*
1665: This is used to set the numeric factorization for both Cholesky and ICC symbolic factorization
1666: */
1667: PetscErrorCode MatSeqSBAIJSetNumericFactorization_inplace(Mat B, PetscBool natural)
1668: {
1669: PetscBool flg = PETSC_FALSE;
1670: PetscInt bs = B->rmap->bs;
1672: PetscFunctionBegin;
1673: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_no_unroll", &flg, NULL));
1674: if (flg) bs = 8;
1676: if (!natural) {
1677: switch (bs) {
1678: case 1:
1679: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_inplace;
1680: break;
1681: case 2:
1682: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2;
1683: break;
1684: case 3:
1685: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3;
1686: break;
1687: case 4:
1688: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4;
1689: break;
1690: case 5:
1691: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5;
1692: break;
1693: case 6:
1694: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6;
1695: break;
1696: case 7:
1697: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7;
1698: break;
1699: default:
1700: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N;
1701: break;
1702: }
1703: } else {
1704: switch (bs) {
1705: case 1:
1706: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_NaturalOrdering_inplace;
1707: break;
1708: case 2:
1709: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2_NaturalOrdering;
1710: break;
1711: case 3:
1712: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3_NaturalOrdering;
1713: break;
1714: case 4:
1715: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4_NaturalOrdering;
1716: break;
1717: case 5:
1718: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5_NaturalOrdering;
1719: break;
1720: case 6:
1721: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6_NaturalOrdering;
1722: break;
1723: case 7:
1724: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7_NaturalOrdering;
1725: break;
1726: default:
1727: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N_NaturalOrdering;
1728: break;
1729: }
1730: }
1731: PetscFunctionReturn(PETSC_SUCCESS);
1732: }
1734: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqAIJ(Mat, MatType, MatReuse, Mat *);
1735: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqBAIJ(Mat, MatType, MatReuse, Mat *);
1736: static PetscErrorCode MatFactorGetSolverType_petsc(Mat A, MatSolverType *type)
1737: {
1738: PetscFunctionBegin;
1739: *type = MATSOLVERPETSC;
1740: PetscFunctionReturn(PETSC_SUCCESS);
1741: }
1743: PETSC_INTERN PetscErrorCode MatGetFactor_seqsbaij_petsc(Mat A, MatFactorType ftype, Mat *B)
1744: {
1745: PetscInt n = A->rmap->n;
1747: PetscFunctionBegin;
1748: #if defined(PETSC_USE_COMPLEX)
1749: if ((ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) && A->hermitian == PETSC_BOOL3_TRUE && A->symmetric != PETSC_BOOL3_TRUE) {
1750: PetscCall(PetscInfo(A, "Hermitian MAT_FACTOR_CHOLESKY or MAT_FACTOR_ICC are not supported. Use MAT_FACTOR_LU instead.\n"));
1751: *B = NULL;
1752: PetscFunctionReturn(PETSC_SUCCESS);
1753: }
1754: #endif
1756: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), B));
1757: PetscCall(MatSetSizes(*B, n, n, n, n));
1758: if (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) {
1759: PetscCall(MatSetType(*B, MATSEQSBAIJ));
1760: PetscCall(MatSeqSBAIJSetPreallocation(*B, A->rmap->bs, MAT_SKIP_ALLOCATION, NULL));
1762: (*B)->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SeqSBAIJ;
1763: (*B)->ops->iccfactorsymbolic = MatICCFactorSymbolic_SeqSBAIJ;
1764: PetscCall(PetscStrallocpy(MATORDERINGNATURAL, (char **)&(*B)->preferredordering[MAT_FACTOR_CHOLESKY]));
1765: PetscCall(PetscStrallocpy(MATORDERINGNATURAL, (char **)&(*B)->preferredordering[MAT_FACTOR_ICC]));
1766: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Factor type not supported");
1768: (*B)->factortype = ftype;
1769: (*B)->canuseordering = PETSC_TRUE;
1770: PetscCall(PetscFree((*B)->solvertype));
1771: PetscCall(PetscStrallocpy(MATSOLVERPETSC, &(*B)->solvertype));
1772: PetscCall(PetscObjectComposeFunction((PetscObject)*B, "MatFactorGetSolverType_C", MatFactorGetSolverType_petsc));
1773: PetscFunctionReturn(PETSC_SUCCESS);
1774: }
1776: /*@C
1777: MatSeqSBAIJGetArray - gives access to the array where the numerical data for a `MATSEQSBAIJ` matrix is stored
1779: Not Collective
1781: Input Parameter:
1782: . A - a `MATSEQSBAIJ` matrix
1784: Output Parameter:
1785: . array - pointer to the data
1787: Level: intermediate
1789: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatSeqSBAIJRestoreArray()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`
1790: @*/
1791: PetscErrorCode MatSeqSBAIJGetArray(Mat A, PetscScalar *array[])
1792: {
1793: PetscFunctionBegin;
1794: PetscUseMethod(A, "MatSeqSBAIJGetArray_C", (Mat, PetscScalar **), (A, array));
1795: PetscFunctionReturn(PETSC_SUCCESS);
1796: }
1798: /*@C
1799: MatSeqSBAIJRestoreArray - returns access to the array where the numerical data for a `MATSEQSBAIJ` matrix is stored obtained by `MatSeqSBAIJGetArray()`
1801: Not Collective
1803: Input Parameters:
1804: + A - a `MATSEQSBAIJ` matrix
1805: - array - pointer to the data
1807: Level: intermediate
1809: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatSeqSBAIJGetArray()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`
1810: @*/
1811: PetscErrorCode MatSeqSBAIJRestoreArray(Mat A, PetscScalar *array[])
1812: {
1813: PetscFunctionBegin;
1814: PetscUseMethod(A, "MatSeqSBAIJRestoreArray_C", (Mat, PetscScalar **), (A, array));
1815: PetscFunctionReturn(PETSC_SUCCESS);
1816: }
1818: /*MC
1819: MATSEQSBAIJ - MATSEQSBAIJ = "seqsbaij" - A matrix type to be used for sequential symmetric block sparse matrices,
1820: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1822: For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you
1823: can call `MatSetOption`(`Mat`, `MAT_HERMITIAN`).
1825: Options Database Key:
1826: . -mat_type seqsbaij - sets the matrix type to "seqsbaij" during a call to `MatSetFromOptions()`
1828: Level: beginner
1830: Notes:
1831: By default if you insert values into the lower triangular part of the matrix they are simply ignored (since they are not
1832: stored and it is assumed they symmetric to the upper triangular). If you call `MatSetOption`(`Mat`,`MAT_IGNORE_LOWER_TRIANGULAR`,`PETSC_FALSE`) or use
1833: 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.
1835: The number of rows in the matrix must be less than or equal to the number of columns
1837: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreateSeqSBAIJ()`, `MatType`, `MATMPISBAIJ`
1838: M*/
1839: PETSC_EXTERN PetscErrorCode MatCreate_SeqSBAIJ(Mat B)
1840: {
1841: Mat_SeqSBAIJ *b;
1842: PetscMPIInt size;
1843: PetscBool no_unroll = PETSC_FALSE, no_inode = PETSC_FALSE;
1845: PetscFunctionBegin;
1846: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &size));
1847: PetscCheck(size <= 1, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Comm must be of size 1");
1849: PetscCall(PetscNew(&b));
1850: B->data = (void *)b;
1851: B->ops[0] = MatOps_Values;
1853: B->ops->destroy = MatDestroy_SeqSBAIJ;
1854: B->ops->view = MatView_SeqSBAIJ;
1855: b->row = NULL;
1856: b->icol = NULL;
1857: b->reallocs = 0;
1858: b->saved_values = NULL;
1859: b->inode.limit = 5;
1860: b->inode.max_limit = 5;
1862: b->roworiented = PETSC_TRUE;
1863: b->nonew = 0;
1864: b->diag = NULL;
1865: b->solve_work = NULL;
1866: b->mult_work = NULL;
1867: B->spptr = NULL;
1868: B->info.nz_unneeded = (PetscReal)b->maxnz * b->bs2;
1869: b->keepnonzeropattern = PETSC_FALSE;
1871: b->inew = NULL;
1872: b->jnew = NULL;
1873: b->anew = NULL;
1874: b->a2anew = NULL;
1875: b->permute = PETSC_FALSE;
1877: b->ignore_ltriangular = PETSC_TRUE;
1879: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_ignore_lower_triangular", &b->ignore_ltriangular, NULL));
1881: b->getrow_utriangular = PETSC_FALSE;
1883: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_getrow_uppertriangular", &b->getrow_utriangular, NULL));
1885: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJGetArray_C", MatSeqSBAIJGetArray_SeqSBAIJ));
1886: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJRestoreArray_C", MatSeqSBAIJRestoreArray_SeqSBAIJ));
1887: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatStoreValues_C", MatStoreValues_SeqSBAIJ));
1888: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatRetrieveValues_C", MatRetrieveValues_SeqSBAIJ));
1889: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetColumnIndices_C", MatSeqSBAIJSetColumnIndices_SeqSBAIJ));
1890: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_seqaij_C", MatConvert_SeqSBAIJ_SeqAIJ));
1891: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_seqbaij_C", MatConvert_SeqSBAIJ_SeqBAIJ));
1892: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetPreallocation_C", MatSeqSBAIJSetPreallocation_SeqSBAIJ));
1893: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetPreallocationCSR_C", MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ));
1894: #if defined(PETSC_HAVE_ELEMENTAL)
1895: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_elemental_C", MatConvert_SeqSBAIJ_Elemental));
1896: #endif
1897: #if defined(PETSC_HAVE_SCALAPACK)
1898: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_scalapack_C", MatConvert_SBAIJ_ScaLAPACK));
1899: #endif
1901: B->symmetry_eternal = PETSC_TRUE;
1902: B->structural_symmetry_eternal = PETSC_TRUE;
1903: B->symmetric = PETSC_BOOL3_TRUE;
1904: B->structurally_symmetric = PETSC_BOOL3_TRUE;
1905: #if defined(PETSC_USE_COMPLEX)
1906: B->hermitian = PETSC_BOOL3_FALSE;
1907: #else
1908: B->hermitian = PETSC_BOOL3_TRUE;
1909: #endif
1911: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATSEQSBAIJ));
1913: PetscOptionsBegin(PetscObjectComm((PetscObject)B), ((PetscObject)B)->prefix, "Options for SEQSBAIJ matrix", "Mat");
1914: PetscCall(PetscOptionsBool("-mat_no_unroll", "Do not optimize for inodes (slower)", NULL, no_unroll, &no_unroll, NULL));
1915: if (no_unroll) PetscCall(PetscInfo(B, "Not using Inode routines due to -mat_no_unroll\n"));
1916: PetscCall(PetscOptionsBool("-mat_no_inode", "Do not optimize for inodes (slower)", NULL, no_inode, &no_inode, NULL));
1917: if (no_inode) PetscCall(PetscInfo(B, "Not using Inode routines due to -mat_no_inode\n"));
1918: PetscCall(PetscOptionsInt("-mat_inode_limit", "Do not use inodes larger then this value", NULL, b->inode.limit, &b->inode.limit, NULL));
1919: PetscOptionsEnd();
1920: b->inode.use = (PetscBool)(!(no_unroll || no_inode));
1921: if (b->inode.limit > b->inode.max_limit) b->inode.limit = b->inode.max_limit;
1922: PetscFunctionReturn(PETSC_SUCCESS);
1923: }
1925: /*@
1926: MatSeqSBAIJSetPreallocation - Creates a sparse symmetric matrix in block AIJ (block
1927: compressed row) `MATSEQSBAIJ` format. For good matrix assembly performance the
1928: user should preallocate the matrix storage by setting the parameter `nz`
1929: (or the array `nnz`).
1931: Collective
1933: Input Parameters:
1934: + B - the symmetric matrix
1935: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
1936: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with `MatCreateVecs()`
1937: . nz - number of block nonzeros per block row (same for all rows)
1938: - nnz - array containing the number of block nonzeros in the upper triangular plus
1939: diagonal portion of each block (possibly different for each block row) or `NULL`
1941: Options Database Keys:
1942: + -mat_no_unroll - uses code that does not unroll the loops in the block calculations (much slower)
1943: - -mat_block_size - size of the blocks to use (only works if a negative bs is passed in
1945: Level: intermediate
1947: Notes:
1948: Specify the preallocated storage with either `nz` or `nnz` (not both).
1949: Set `nz` = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
1950: allocation. See [Sparse Matrices](sec_matsparse) for details.
1952: You can call `MatGetInfo()` to get information on how effective the preallocation was;
1953: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
1954: You can also run with the option `-info` and look for messages with the string
1955: malloc in them to see if additional memory allocation was needed.
1957: If the `nnz` parameter is given then the `nz` parameter is ignored
1959: .seealso: [](ch_matrices), `Mat`, [Sparse Matrices](sec_matsparse), `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatCreateSBAIJ()`
1960: @*/
1961: PetscErrorCode MatSeqSBAIJSetPreallocation(Mat B, PetscInt bs, PetscInt nz, const PetscInt nnz[])
1962: {
1963: PetscFunctionBegin;
1967: PetscTryMethod(B, "MatSeqSBAIJSetPreallocation_C", (Mat, PetscInt, PetscInt, const PetscInt[]), (B, bs, nz, nnz));
1968: PetscFunctionReturn(PETSC_SUCCESS);
1969: }
1971: /*@C
1972: MatSeqSBAIJSetPreallocationCSR - Creates a sparse parallel matrix in `MATSEQSBAIJ` format using the given nonzero structure and (optional) numerical values
1974: Input Parameters:
1975: + B - the matrix
1976: . bs - size of block, the blocks are ALWAYS square.
1977: . i - the indices into `j` for the start of each local row (indices start with zero)
1978: . j - the column indices for each local row (indices start with zero) these must be sorted for each row
1979: - v - optional values in the matrix, use `NULL` if not provided
1981: Level: advanced
1983: Notes:
1984: The `i`,`j`,`v` values are COPIED with this routine; to avoid the copy use `MatCreateSeqSBAIJWithArrays()`
1986: The order of the entries in values is specified by the `MatOption` `MAT_ROW_ORIENTED`. For example, C programs
1987: may want to use the default `MAT_ROW_ORIENTED` = `PETSC_TRUE` and use an array v[nnz][bs][bs] where the second index is
1988: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
1989: `MAT_ROW_ORIENTED` = `PETSC_FALSE` and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
1990: block column and the second index is over columns within a block.
1992: Any entries provided that lie below the diagonal are ignored
1994: Though this routine has Preallocation() in the name it also sets the exact nonzero locations of the matrix entries
1995: and usually the numerical values as well
1997: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqSBAIJ()`, `MatSetValuesBlocked()`, `MatSeqSBAIJSetPreallocation()`
1998: @*/
1999: PetscErrorCode MatSeqSBAIJSetPreallocationCSR(Mat B, PetscInt bs, const PetscInt i[], const PetscInt j[], const PetscScalar v[])
2000: {
2001: PetscFunctionBegin;
2005: PetscTryMethod(B, "MatSeqSBAIJSetPreallocationCSR_C", (Mat, PetscInt, const PetscInt[], const PetscInt[], const PetscScalar[]), (B, bs, i, j, v));
2006: PetscFunctionReturn(PETSC_SUCCESS);
2007: }
2009: /*@
2010: MatCreateSeqSBAIJ - Creates a sparse symmetric matrix in (block
2011: compressed row) `MATSEQSBAIJ` format. For good matrix assembly performance the
2012: user should preallocate the matrix storage by setting the parameter `nz`
2013: (or the array `nnz`).
2015: Collective
2017: Input Parameters:
2018: + comm - MPI communicator, set to `PETSC_COMM_SELF`
2019: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
2020: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2021: . m - number of rows
2022: . n - number of columns
2023: . nz - number of block nonzeros per block row (same for all rows)
2024: - nnz - array containing the number of block nonzeros in the upper triangular plus
2025: diagonal portion of each block (possibly different for each block row) or `NULL`
2027: Output Parameter:
2028: . A - the symmetric matrix
2030: Options Database Keys:
2031: + -mat_no_unroll - uses code that does not unroll the loops in the block calculations (much slower)
2032: - -mat_block_size - size of the blocks to use
2034: Level: intermediate
2036: Notes:
2037: It is recommended that one use `MatCreateFromOptions()` or the `MatCreate()`, `MatSetType()` and/or `MatSetFromOptions()`,
2038: MatXXXXSetPreallocation() paradigm instead of this routine directly.
2039: [MatXXXXSetPreallocation() is, for example, `MatSeqAIJSetPreallocation()`]
2041: The number of rows and columns must be divisible by blocksize.
2042: This matrix type does not support complex Hermitian operation.
2044: Specify the preallocated storage with either `nz` or `nnz` (not both).
2045: Set `nz` = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
2046: allocation. See [Sparse Matrices](sec_matsparse) for details.
2048: If the `nnz` parameter is given then the `nz` parameter is ignored
2050: .seealso: [](ch_matrices), `Mat`, [Sparse Matrices](sec_matsparse), `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatCreateSBAIJ()`
2051: @*/
2052: PetscErrorCode MatCreateSeqSBAIJ(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt nz, const PetscInt nnz[], Mat *A)
2053: {
2054: PetscFunctionBegin;
2055: PetscCall(MatCreate(comm, A));
2056: PetscCall(MatSetSizes(*A, m, n, m, n));
2057: PetscCall(MatSetType(*A, MATSEQSBAIJ));
2058: PetscCall(MatSeqSBAIJSetPreallocation(*A, bs, nz, (PetscInt *)nnz));
2059: PetscFunctionReturn(PETSC_SUCCESS);
2060: }
2062: PetscErrorCode MatDuplicate_SeqSBAIJ(Mat A, MatDuplicateOption cpvalues, Mat *B)
2063: {
2064: Mat C;
2065: Mat_SeqSBAIJ *c, *a = (Mat_SeqSBAIJ *)A->data;
2066: PetscInt i, mbs = a->mbs, nz = a->nz, bs2 = a->bs2;
2068: PetscFunctionBegin;
2069: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Cannot duplicate unassembled matrix");
2070: PetscCheck(a->i[mbs] == nz, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Corrupt matrix");
2072: *B = NULL;
2073: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &C));
2074: PetscCall(MatSetSizes(C, A->rmap->N, A->cmap->n, A->rmap->N, A->cmap->n));
2075: PetscCall(MatSetBlockSizesFromMats(C, A, A));
2076: PetscCall(MatSetType(C, MATSEQSBAIJ));
2077: c = (Mat_SeqSBAIJ *)C->data;
2079: C->preallocated = PETSC_TRUE;
2080: C->factortype = A->factortype;
2081: c->row = NULL;
2082: c->icol = NULL;
2083: c->saved_values = NULL;
2084: c->keepnonzeropattern = a->keepnonzeropattern;
2085: C->assembled = PETSC_TRUE;
2087: PetscCall(PetscLayoutReference(A->rmap, &C->rmap));
2088: PetscCall(PetscLayoutReference(A->cmap, &C->cmap));
2089: c->bs2 = a->bs2;
2090: c->mbs = a->mbs;
2091: c->nbs = a->nbs;
2093: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2094: c->imax = a->imax;
2095: c->ilen = a->ilen;
2096: c->free_imax_ilen = PETSC_FALSE;
2097: } else {
2098: PetscCall(PetscMalloc2(mbs + 1, &c->imax, mbs + 1, &c->ilen));
2099: for (i = 0; i < mbs; i++) {
2100: c->imax[i] = a->imax[i];
2101: c->ilen[i] = a->ilen[i];
2102: }
2103: c->free_imax_ilen = PETSC_TRUE;
2104: }
2106: /* allocate the matrix space */
2107: PetscCall(PetscShmgetAllocateArray(bs2 * nz, sizeof(PetscScalar), (void **)&c->a));
2108: c->free_a = PETSC_TRUE;
2109: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2110: PetscCall(PetscArrayzero(c->a, bs2 * nz));
2111: c->i = a->i;
2112: c->j = a->j;
2113: c->free_ij = PETSC_FALSE;
2114: c->parent = A;
2115: PetscCall(PetscObjectReference((PetscObject)A));
2116: PetscCall(MatSetOption(A, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2117: PetscCall(MatSetOption(C, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2118: } else {
2119: PetscCall(PetscShmgetAllocateArray(nz, sizeof(PetscInt), (void **)&c->j));
2120: PetscCall(PetscShmgetAllocateArray(mbs + 1, sizeof(PetscInt), (void **)&c->i));
2121: PetscCall(PetscArraycpy(c->i, a->i, mbs + 1));
2122: c->free_ij = PETSC_TRUE;
2123: }
2124: if (mbs > 0) {
2125: if (cpvalues != MAT_SHARE_NONZERO_PATTERN) PetscCall(PetscArraycpy(c->j, a->j, nz));
2126: if (cpvalues == MAT_COPY_VALUES) {
2127: PetscCall(PetscArraycpy(c->a, a->a, bs2 * nz));
2128: } else {
2129: PetscCall(PetscArrayzero(c->a, bs2 * nz));
2130: }
2131: if (a->jshort) {
2132: /* cannot share jshort, it is reallocated in MatAssemblyEnd_SeqSBAIJ() */
2133: /* if the parent matrix is reassembled, this child matrix will never notice */
2134: PetscCall(PetscMalloc1(nz, &c->jshort));
2135: PetscCall(PetscArraycpy(c->jshort, a->jshort, nz));
2137: c->free_jshort = PETSC_TRUE;
2138: }
2139: }
2141: c->roworiented = a->roworiented;
2142: c->nonew = a->nonew;
2144: if (a->diag) {
2145: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2146: c->diag = a->diag;
2147: c->free_diag = PETSC_FALSE;
2148: } else {
2149: PetscCall(PetscMalloc1(mbs, &c->diag));
2150: for (i = 0; i < mbs; i++) c->diag[i] = a->diag[i];
2151: c->free_diag = PETSC_TRUE;
2152: }
2153: }
2154: c->nz = a->nz;
2155: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
2156: c->solve_work = NULL;
2157: c->mult_work = NULL;
2159: *B = C;
2160: PetscCall(PetscFunctionListDuplicate(((PetscObject)A)->qlist, &((PetscObject)C)->qlist));
2161: PetscFunctionReturn(PETSC_SUCCESS);
2162: }
2164: /* Used for both SeqBAIJ and SeqSBAIJ matrices */
2165: #define MatLoad_SeqSBAIJ_Binary MatLoad_SeqBAIJ_Binary
2167: PetscErrorCode MatLoad_SeqSBAIJ(Mat mat, PetscViewer viewer)
2168: {
2169: PetscBool isbinary;
2171: PetscFunctionBegin;
2172: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
2173: PetscCheck(isbinary, PetscObjectComm((PetscObject)viewer), PETSC_ERR_SUP, "Viewer type %s not yet supported for reading %s matrices", ((PetscObject)viewer)->type_name, ((PetscObject)mat)->type_name);
2174: PetscCall(MatLoad_SeqSBAIJ_Binary(mat, viewer));
2175: PetscFunctionReturn(PETSC_SUCCESS);
2176: }
2178: /*@
2179: MatCreateSeqSBAIJWithArrays - Creates an sequential `MATSEQSBAIJ` matrix using matrix elements
2180: (upper triangular entries in CSR format) provided by the user.
2182: Collective
2184: Input Parameters:
2185: + comm - must be an MPI communicator of size 1
2186: . bs - size of block
2187: . m - number of rows
2188: . n - number of columns
2189: . i - row indices; that is i[0] = 0, i[row] = i[row-1] + number of block elements in that row block row of the matrix
2190: . j - column indices
2191: - a - matrix values
2193: Output Parameter:
2194: . mat - the matrix
2196: Level: advanced
2198: Notes:
2199: The `i`, `j`, and `a` arrays are not copied by this routine, the user must free these arrays
2200: once the matrix is destroyed
2202: You cannot set new nonzero locations into this matrix, that will generate an error.
2204: The `i` and `j` indices are 0 based
2206: When block size is greater than 1 the matrix values must be stored using the `MATSBAIJ` storage format. For block size of 1
2207: it is the regular CSR format excluding the lower triangular elements.
2209: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSBAIJ()`, `MatCreateSeqSBAIJ()`
2210: @*/
2211: PetscErrorCode MatCreateSeqSBAIJWithArrays(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt i[], PetscInt j[], PetscScalar a[], Mat *mat)
2212: {
2213: PetscInt ii;
2214: Mat_SeqSBAIJ *sbaij;
2216: PetscFunctionBegin;
2217: PetscCheck(bs == 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "block size %" PetscInt_FMT " > 1 is not supported yet", bs);
2218: PetscCheck(m == 0 || i[0] == 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "i (row indices) must start with 0");
2220: PetscCall(MatCreate(comm, mat));
2221: PetscCall(MatSetSizes(*mat, m, n, m, n));
2222: PetscCall(MatSetType(*mat, MATSEQSBAIJ));
2223: PetscCall(MatSeqSBAIJSetPreallocation(*mat, bs, MAT_SKIP_ALLOCATION, NULL));
2224: sbaij = (Mat_SeqSBAIJ *)(*mat)->data;
2225: PetscCall(PetscMalloc2(m, &sbaij->imax, m, &sbaij->ilen));
2227: sbaij->i = i;
2228: sbaij->j = j;
2229: sbaij->a = a;
2231: sbaij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
2232: sbaij->free_a = PETSC_FALSE;
2233: sbaij->free_ij = PETSC_FALSE;
2234: sbaij->free_imax_ilen = PETSC_TRUE;
2236: for (ii = 0; ii < m; ii++) {
2237: sbaij->ilen[ii] = sbaij->imax[ii] = i[ii + 1] - i[ii];
2238: PetscCheck(i[ii + 1] >= i[ii], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative row length in i (row indices) row = %" PetscInt_FMT " length = %" PetscInt_FMT, ii, i[ii + 1] - i[ii]);
2239: }
2240: if (PetscDefined(USE_DEBUG)) {
2241: for (ii = 0; ii < sbaij->i[m]; ii++) {
2242: PetscCheck(j[ii] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative column index at location = %" PetscInt_FMT " index = %" PetscInt_FMT, ii, j[ii]);
2243: PetscCheck(j[ii] < n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column index too large at location = %" PetscInt_FMT " index = %" PetscInt_FMT, ii, j[ii]);
2244: }
2245: }
2247: PetscCall(MatAssemblyBegin(*mat, MAT_FINAL_ASSEMBLY));
2248: PetscCall(MatAssemblyEnd(*mat, MAT_FINAL_ASSEMBLY));
2249: PetscFunctionReturn(PETSC_SUCCESS);
2250: }
2252: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ(MPI_Comm comm, Mat inmat, PetscInt n, MatReuse scall, Mat *outmat)
2253: {
2254: PetscFunctionBegin;
2255: PetscCall(MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(comm, inmat, n, scall, outmat));
2256: PetscFunctionReturn(PETSC_SUCCESS);
2257: }