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));
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_FORCE_DIAGONAL_ENTRIES:
257: case MAT_IGNORE_OFF_PROC_ENTRIES:
258: case MAT_USE_HASH_TABLE:
259: case MAT_SORTED_FULL:
260: PetscCall(PetscInfo(A, "Option %s ignored\n", MatOptions[op]));
261: break;
262: case MAT_HERMITIAN:
263: #if defined(PETSC_USE_COMPLEX)
264: if (flg) { /* disable transpose ops */
265: PetscCheck(bs <= 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "No support for Hermitian with block size greater than 1");
266: A->ops->multtranspose = NULL;
267: A->ops->multtransposeadd = NULL;
268: A->symmetric = PETSC_BOOL3_FALSE;
269: }
270: #endif
271: break;
272: case MAT_SYMMETRIC:
273: case MAT_SPD:
274: #if defined(PETSC_USE_COMPLEX)
275: if (flg) { /* An hermitian and symmetric matrix has zero imaginary part (restore back transpose ops) */
276: A->ops->multtranspose = A->ops->mult;
277: A->ops->multtransposeadd = A->ops->multadd;
278: }
279: #endif
280: break;
281: /* These options are handled directly by MatSetOption() */
282: case MAT_STRUCTURALLY_SYMMETRIC:
283: case MAT_SYMMETRY_ETERNAL:
284: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
285: case MAT_STRUCTURE_ONLY:
286: case MAT_SPD_ETERNAL:
287: /* These options are handled directly by MatSetOption() */
288: break;
289: case MAT_IGNORE_LOWER_TRIANGULAR:
290: a->ignore_ltriangular = flg;
291: break;
292: case MAT_ERROR_LOWER_TRIANGULAR:
293: a->ignore_ltriangular = flg;
294: break;
295: case MAT_GETROW_UPPERTRIANGULAR:
296: a->getrow_utriangular = flg;
297: break;
298: case MAT_SUBMAT_SINGLEIS:
299: break;
300: default:
301: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "unknown option %d", op);
302: }
303: PetscFunctionReturn(PETSC_SUCCESS);
304: }
306: PetscErrorCode MatGetRow_SeqSBAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
307: {
308: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
310: PetscFunctionBegin;
311: 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()");
313: /* Get the upper triangular part of the row */
314: PetscCall(MatGetRow_SeqBAIJ_private(A, row, nz, idx, v, a->i, a->j, a->a));
315: PetscFunctionReturn(PETSC_SUCCESS);
316: }
318: PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
319: {
320: PetscFunctionBegin;
321: if (idx) PetscCall(PetscFree(*idx));
322: if (v) PetscCall(PetscFree(*v));
323: PetscFunctionReturn(PETSC_SUCCESS);
324: }
326: static PetscErrorCode MatGetRowUpperTriangular_SeqSBAIJ(Mat A)
327: {
328: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
330: PetscFunctionBegin;
331: a->getrow_utriangular = PETSC_TRUE;
332: PetscFunctionReturn(PETSC_SUCCESS);
333: }
335: static PetscErrorCode MatRestoreRowUpperTriangular_SeqSBAIJ(Mat A)
336: {
337: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
339: PetscFunctionBegin;
340: a->getrow_utriangular = PETSC_FALSE;
341: PetscFunctionReturn(PETSC_SUCCESS);
342: }
344: static PetscErrorCode MatTranspose_SeqSBAIJ(Mat A, MatReuse reuse, Mat *B)
345: {
346: PetscFunctionBegin;
347: if (reuse == MAT_REUSE_MATRIX) PetscCall(MatTransposeCheckNonzeroState_Private(A, *B));
348: if (reuse == MAT_INITIAL_MATRIX) {
349: PetscCall(MatDuplicate(A, MAT_COPY_VALUES, B));
350: } else if (reuse == MAT_REUSE_MATRIX) {
351: PetscCall(MatCopy(A, *B, SAME_NONZERO_PATTERN));
352: }
353: PetscFunctionReturn(PETSC_SUCCESS);
354: }
356: static PetscErrorCode MatView_SeqSBAIJ_ASCII(Mat A, PetscViewer viewer)
357: {
358: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
359: PetscInt i, j, bs = A->rmap->bs, k, l, bs2 = a->bs2;
360: PetscViewerFormat format;
361: PetscInt *diag;
362: const char *matname;
364: PetscFunctionBegin;
365: PetscCall(PetscViewerGetFormat(viewer, &format));
366: if (format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
367: PetscCall(PetscViewerASCIIPrintf(viewer, " block size is %" PetscInt_FMT "\n", bs));
368: } else if (format == PETSC_VIEWER_ASCII_MATLAB) {
369: Mat aij;
371: if (A->factortype && bs > 1) {
372: PetscCall(PetscPrintf(PETSC_COMM_SELF, "Warning: matrix is factored with bs>1. MatView() with PETSC_VIEWER_ASCII_MATLAB is not supported and ignored!\n"));
373: PetscFunctionReturn(PETSC_SUCCESS);
374: }
375: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &aij));
376: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
377: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)aij, matname));
378: PetscCall(MatView_SeqAIJ(aij, viewer));
379: PetscCall(MatDestroy(&aij));
380: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
381: Mat B;
383: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &B));
384: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
385: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)B, matname));
386: PetscCall(MatView_SeqAIJ(B, viewer));
387: PetscCall(MatDestroy(&B));
388: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
389: PetscFunctionReturn(PETSC_SUCCESS);
390: } else {
391: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
392: if (A->factortype) { /* for factored matrix */
393: PetscCheck(bs <= 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "matrix is factored with bs>1. Not implemented yet");
395: diag = a->diag;
396: for (i = 0; i < a->mbs; i++) { /* for row block i */
397: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
398: /* diagonal entry */
399: #if defined(PETSC_USE_COMPLEX)
400: if (PetscImaginaryPart(a->a[diag[i]]) > 0.0) {
401: 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]])));
402: } else if (PetscImaginaryPart(a->a[diag[i]]) < 0.0) {
403: 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]])));
404: } else {
405: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[diag[i]], (double)PetscRealPart(1.0 / a->a[diag[i]])));
406: }
407: #else
408: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[diag[i]], (double)(1.0 / a->a[diag[i]])));
409: #endif
410: /* off-diagonal entries */
411: for (k = a->i[i]; k < a->i[i + 1] - 1; k++) {
412: #if defined(PETSC_USE_COMPLEX)
413: if (PetscImaginaryPart(a->a[k]) > 0.0) {
414: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i) ", bs * a->j[k], (double)PetscRealPart(a->a[k]), (double)PetscImaginaryPart(a->a[k])));
415: } else if (PetscImaginaryPart(a->a[k]) < 0.0) {
416: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i) ", bs * a->j[k], (double)PetscRealPart(a->a[k]), -(double)PetscImaginaryPart(a->a[k])));
417: } else {
418: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k], (double)PetscRealPart(a->a[k])));
419: }
420: #else
421: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[k], (double)a->a[k]));
422: #endif
423: }
424: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
425: }
427: } else { /* for non-factored matrix */
428: for (i = 0; i < a->mbs; i++) { /* for row block i */
429: for (j = 0; j < bs; j++) { /* for row bs*i + j */
430: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i * bs + j));
431: for (k = a->i[i]; k < a->i[i + 1]; k++) { /* for column block */
432: for (l = 0; l < bs; l++) { /* for column */
433: #if defined(PETSC_USE_COMPLEX)
434: if (PetscImaginaryPart(a->a[bs2 * k + l * bs + j]) > 0.0) {
435: 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])));
436: } else if (PetscImaginaryPart(a->a[bs2 * k + l * bs + j]) < 0.0) {
437: 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])));
438: } else {
439: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k] + l, (double)PetscRealPart(a->a[bs2 * k + l * bs + j])));
440: }
441: #else
442: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", bs * a->j[k] + l, (double)a->a[bs2 * k + l * bs + j]));
443: #endif
444: }
445: }
446: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
447: }
448: }
449: }
450: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
451: }
452: PetscCall(PetscViewerFlush(viewer));
453: PetscFunctionReturn(PETSC_SUCCESS);
454: }
456: #include <petscdraw.h>
457: static PetscErrorCode MatView_SeqSBAIJ_Draw_Zoom(PetscDraw draw, void *Aa)
458: {
459: Mat A = (Mat)Aa;
460: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
461: PetscInt row, i, j, k, l, mbs = a->mbs, color, bs = A->rmap->bs, bs2 = a->bs2;
462: PetscReal xl, yl, xr, yr, x_l, x_r, y_l, y_r;
463: MatScalar *aa;
464: PetscViewer viewer;
466: PetscFunctionBegin;
467: PetscCall(PetscObjectQuery((PetscObject)A, "Zoomviewer", (PetscObject *)&viewer));
468: PetscCall(PetscDrawGetCoordinates(draw, &xl, &yl, &xr, &yr));
470: /* loop over matrix elements drawing boxes */
472: PetscDrawCollectiveBegin(draw);
473: PetscCall(PetscDrawString(draw, .3 * (xl + xr), .3 * (yl + yr), PETSC_DRAW_BLACK, "symmetric"));
474: /* Blue for negative, Cyan for zero and Red for positive */
475: color = PETSC_DRAW_BLUE;
476: for (i = 0, row = 0; i < mbs; i++, row += bs) {
477: for (j = a->i[i]; j < a->i[i + 1]; j++) {
478: y_l = A->rmap->N - row - 1.0;
479: y_r = y_l + 1.0;
480: x_l = a->j[j] * bs;
481: x_r = x_l + 1.0;
482: aa = a->a + j * bs2;
483: for (k = 0; k < bs; k++) {
484: for (l = 0; l < bs; l++) {
485: if (PetscRealPart(*aa++) >= 0.) continue;
486: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
487: }
488: }
489: }
490: }
491: color = PETSC_DRAW_CYAN;
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;
495: y_r = y_l + 1.0;
496: x_l = a->j[j] * bs;
497: x_r = x_l + 1.0;
498: aa = a->a + j * bs2;
499: for (k = 0; k < bs; k++) {
500: for (l = 0; l < bs; l++) {
501: if (PetscRealPart(*aa++) != 0.) continue;
502: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
503: }
504: }
505: }
506: }
507: color = PETSC_DRAW_RED;
508: for (i = 0, row = 0; i < mbs; i++, row += bs) {
509: for (j = a->i[i]; j < a->i[i + 1]; j++) {
510: y_l = A->rmap->N - row - 1.0;
511: y_r = y_l + 1.0;
512: x_l = a->j[j] * bs;
513: x_r = x_l + 1.0;
514: aa = a->a + j * bs2;
515: for (k = 0; k < bs; k++) {
516: for (l = 0; l < bs; l++) {
517: if (PetscRealPart(*aa++) <= 0.) continue;
518: PetscCall(PetscDrawRectangle(draw, x_l + k, y_l - l, x_r + k, y_r - l, color, color, color, color));
519: }
520: }
521: }
522: }
523: PetscDrawCollectiveEnd(draw);
524: PetscFunctionReturn(PETSC_SUCCESS);
525: }
527: static PetscErrorCode MatView_SeqSBAIJ_Draw(Mat A, PetscViewer viewer)
528: {
529: PetscReal xl, yl, xr, yr, w, h;
530: PetscDraw draw;
531: PetscBool isnull;
533: PetscFunctionBegin;
534: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
535: PetscCall(PetscDrawIsNull(draw, &isnull));
536: if (isnull) PetscFunctionReturn(PETSC_SUCCESS);
538: xr = A->rmap->N;
539: yr = A->rmap->N;
540: h = yr / 10.0;
541: w = xr / 10.0;
542: xr += w;
543: yr += h;
544: xl = -w;
545: yl = -h;
546: PetscCall(PetscDrawSetCoordinates(draw, xl, yl, xr, yr));
547: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", (PetscObject)viewer));
548: PetscCall(PetscDrawZoom(draw, MatView_SeqSBAIJ_Draw_Zoom, A));
549: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", NULL));
550: PetscCall(PetscDrawSave(draw));
551: PetscFunctionReturn(PETSC_SUCCESS);
552: }
554: /* Used for both MPIBAIJ and MPISBAIJ matrices */
555: #define MatView_SeqSBAIJ_Binary MatView_SeqBAIJ_Binary
557: PetscErrorCode MatView_SeqSBAIJ(Mat A, PetscViewer viewer)
558: {
559: PetscBool iascii, isbinary, isdraw;
561: PetscFunctionBegin;
562: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
563: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
564: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
565: if (iascii) {
566: PetscCall(MatView_SeqSBAIJ_ASCII(A, viewer));
567: } else if (isbinary) {
568: PetscCall(MatView_SeqSBAIJ_Binary(A, viewer));
569: } else if (isdraw) {
570: PetscCall(MatView_SeqSBAIJ_Draw(A, viewer));
571: } else {
572: Mat B;
573: const char *matname;
574: PetscCall(MatConvert(A, MATSEQAIJ, MAT_INITIAL_MATRIX, &B));
575: if (((PetscObject)A)->name) PetscCall(PetscObjectGetName((PetscObject)A, &matname));
576: if (((PetscObject)A)->name) PetscCall(PetscObjectSetName((PetscObject)B, matname));
577: PetscCall(MatView(B, viewer));
578: PetscCall(MatDestroy(&B));
579: }
580: PetscFunctionReturn(PETSC_SUCCESS);
581: }
583: PetscErrorCode MatGetValues_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], PetscScalar v[])
584: {
585: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
586: PetscInt *rp, k, low, high, t, row, nrow, i, col, l, *aj = a->j;
587: PetscInt *ai = a->i, *ailen = a->ilen;
588: PetscInt brow, bcol, ridx, cidx, bs = A->rmap->bs, bs2 = a->bs2;
589: MatScalar *ap, *aa = a->a;
591: PetscFunctionBegin;
592: for (k = 0; k < m; k++) { /* loop over rows */
593: row = im[k];
594: brow = row / bs;
595: if (row < 0) {
596: v += n;
597: continue;
598: } /* negative row */
599: 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);
600: rp = aj + ai[brow];
601: ap = aa + bs2 * ai[brow];
602: nrow = ailen[brow];
603: for (l = 0; l < n; l++) { /* loop over columns */
604: if (in[l] < 0) {
605: v++;
606: continue;
607: } /* negative column */
608: 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);
609: col = in[l];
610: bcol = col / bs;
611: cidx = col % bs;
612: ridx = row % bs;
613: high = nrow;
614: low = 0; /* assume unsorted */
615: while (high - low > 5) {
616: t = (low + high) / 2;
617: if (rp[t] > bcol) high = t;
618: else low = t;
619: }
620: for (i = low; i < high; i++) {
621: if (rp[i] > bcol) break;
622: if (rp[i] == bcol) {
623: *v++ = ap[bs2 * i + bs * cidx + ridx];
624: goto finished;
625: }
626: }
627: *v++ = 0.0;
628: finished:;
629: }
630: }
631: PetscFunctionReturn(PETSC_SUCCESS);
632: }
634: static PetscErrorCode MatPermute_SeqSBAIJ(Mat A, IS rowp, IS colp, Mat *B)
635: {
636: Mat C;
637: PetscBool flg = (PetscBool)(rowp == colp);
639: PetscFunctionBegin;
640: PetscCall(MatConvert(A, MATSEQBAIJ, MAT_INITIAL_MATRIX, &C));
641: PetscCall(MatPermute(C, rowp, colp, B));
642: PetscCall(MatDestroy(&C));
643: if (!flg) PetscCall(ISEqual(rowp, colp, &flg));
644: if (flg) PetscCall(MatConvert(*B, MATSEQSBAIJ, MAT_INPLACE_MATRIX, B));
645: PetscFunctionReturn(PETSC_SUCCESS);
646: }
648: PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
649: {
650: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
651: PetscInt *rp, k, low, high, t, ii, jj, row, nrow, i, col, l, rmax, N, lastcol = -1;
652: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
653: PetscInt *aj = a->j, nonew = a->nonew, bs2 = a->bs2, bs = A->rmap->bs, stepval;
654: PetscBool roworiented = a->roworiented;
655: const PetscScalar *value = v;
656: MatScalar *ap, *aa = a->a, *bap;
658: PetscFunctionBegin;
659: if (roworiented) stepval = (n - 1) * bs;
660: else stepval = (m - 1) * bs;
661: for (k = 0; k < m; k++) { /* loop over added rows */
662: row = im[k];
663: if (row < 0) continue;
664: 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);
665: rp = aj + ai[row];
666: ap = aa + bs2 * ai[row];
667: rmax = imax[row];
668: nrow = ailen[row];
669: low = 0;
670: high = nrow;
671: for (l = 0; l < n; l++) { /* loop over added columns */
672: if (in[l] < 0) continue;
673: col = in[l];
674: 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);
675: if (col < row) {
676: if (a->ignore_ltriangular) continue; /* ignore lower triangular block */
677: 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)");
678: }
679: if (roworiented) value = v + k * (stepval + bs) * bs + l * bs;
680: else value = v + l * (stepval + bs) * bs + k * bs;
682: if (col <= lastcol) low = 0;
683: else high = nrow;
685: lastcol = col;
686: while (high - low > 7) {
687: t = (low + high) / 2;
688: if (rp[t] > col) high = t;
689: else low = t;
690: }
691: for (i = low; i < high; i++) {
692: if (rp[i] > col) break;
693: if (rp[i] == col) {
694: bap = ap + bs2 * i;
695: if (roworiented) {
696: if (is == ADD_VALUES) {
697: for (ii = 0; ii < bs; ii++, value += stepval) {
698: for (jj = ii; jj < bs2; jj += bs) bap[jj] += *value++;
699: }
700: } else {
701: for (ii = 0; ii < bs; ii++, value += stepval) {
702: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
703: }
704: }
705: } else {
706: if (is == ADD_VALUES) {
707: for (ii = 0; ii < bs; ii++, value += stepval) {
708: for (jj = 0; jj < bs; jj++) *bap++ += *value++;
709: }
710: } else {
711: for (ii = 0; ii < bs; ii++, value += stepval) {
712: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
713: }
714: }
715: }
716: goto noinsert2;
717: }
718: }
719: if (nonew == 1) goto noinsert2;
720: 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);
721: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
722: N = nrow++ - 1;
723: high++;
724: /* shift up all the later entries in this row */
725: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
726: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
727: PetscCall(PetscArrayzero(ap + bs2 * i, bs2));
728: rp[i] = col;
729: bap = ap + bs2 * i;
730: if (roworiented) {
731: for (ii = 0; ii < bs; ii++, value += stepval) {
732: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
733: }
734: } else {
735: for (ii = 0; ii < bs; ii++, value += stepval) {
736: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
737: }
738: }
739: noinsert2:;
740: low = i;
741: }
742: ailen[row] = nrow;
743: }
744: PetscFunctionReturn(PETSC_SUCCESS);
745: }
747: static PetscErrorCode MatAssemblyEnd_SeqSBAIJ(Mat A, MatAssemblyType mode)
748: {
749: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
750: PetscInt fshift = 0, i, *ai = a->i, *aj = a->j, *imax = a->imax;
751: PetscInt m = A->rmap->N, *ip, N, *ailen = a->ilen;
752: PetscInt mbs = a->mbs, bs2 = a->bs2, rmax = 0;
753: MatScalar *aa = a->a, *ap;
755: PetscFunctionBegin;
756: if (mode == MAT_FLUSH_ASSEMBLY || (A->was_assembled && A->ass_nonzerostate == A->nonzerostate)) PetscFunctionReturn(PETSC_SUCCESS);
758: if (m) rmax = ailen[0];
759: for (i = 1; i < mbs; i++) {
760: /* move each row back by the amount of empty slots (fshift) before it*/
761: fshift += imax[i - 1] - ailen[i - 1];
762: rmax = PetscMax(rmax, ailen[i]);
763: if (fshift) {
764: ip = aj + ai[i];
765: ap = aa + bs2 * ai[i];
766: N = ailen[i];
767: PetscCall(PetscArraymove(ip - fshift, ip, N));
768: PetscCall(PetscArraymove(ap - bs2 * fshift, ap, bs2 * N));
769: }
770: ai[i] = ai[i - 1] + ailen[i - 1];
771: }
772: if (mbs) {
773: fshift += imax[mbs - 1] - ailen[mbs - 1];
774: ai[mbs] = ai[mbs - 1] + ailen[mbs - 1];
775: }
776: /* reset ilen and imax for each row */
777: for (i = 0; i < mbs; i++) ailen[i] = imax[i] = ai[i + 1] - ai[i];
778: a->nz = ai[mbs];
780: /* diagonals may have moved, reset it */
781: if (a->diag) PetscCall(PetscArraycpy(a->diag, ai, mbs));
782: 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);
784: 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));
785: PetscCall(PetscInfo(A, "Number of mallocs during MatSetValues is %" PetscInt_FMT "\n", a->reallocs));
786: PetscCall(PetscInfo(A, "Most nonzeros blocks in any row is %" PetscInt_FMT "\n", rmax));
788: A->info.mallocs += a->reallocs;
789: a->reallocs = 0;
790: A->info.nz_unneeded = (PetscReal)fshift * bs2;
791: a->idiagvalid = PETSC_FALSE;
792: a->rmax = rmax;
794: if (A->cmap->n < 65536 && A->cmap->bs == 1) {
795: if (a->jshort && a->free_jshort) {
796: /* when matrix data structure is changed, previous jshort must be replaced */
797: PetscCall(PetscFree(a->jshort));
798: }
799: PetscCall(PetscMalloc1(a->i[A->rmap->n], &a->jshort));
800: for (i = 0; i < a->i[A->rmap->n]; i++) a->jshort[i] = a->j[i];
801: A->ops->mult = MatMult_SeqSBAIJ_1_ushort;
802: A->ops->sor = MatSOR_SeqSBAIJ_ushort;
803: a->free_jshort = PETSC_TRUE;
804: }
805: PetscFunctionReturn(PETSC_SUCCESS);
806: }
808: /* Only add/insert a(i,j) with i<=j (blocks).
809: Any a(i,j) with i>j input by user is ignored.
810: */
812: PetscErrorCode MatSetValues_SeqSBAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
813: {
814: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
815: PetscInt *rp, k, low, high, t, ii, row, nrow, i, col, l, rmax, N, lastcol = -1;
816: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen, roworiented = a->roworiented;
817: PetscInt *aj = a->j, nonew = a->nonew, bs = A->rmap->bs, brow, bcol;
818: PetscInt ridx, cidx, bs2 = a->bs2;
819: MatScalar *ap, value, *aa = a->a, *bap;
821: PetscFunctionBegin;
822: for (k = 0; k < m; k++) { /* loop over added rows */
823: row = im[k]; /* row number */
824: brow = row / bs; /* block row number */
825: if (row < 0) continue;
826: 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);
827: rp = aj + ai[brow]; /*ptr to beginning of column value of the row block*/
828: ap = aa + bs2 * ai[brow]; /*ptr to beginning of element value of the row block*/
829: rmax = imax[brow]; /* maximum space allocated for this row */
830: nrow = ailen[brow]; /* actual length of this row */
831: low = 0;
832: high = nrow;
833: for (l = 0; l < n; l++) { /* loop over added columns */
834: if (in[l] < 0) continue;
835: 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);
836: col = in[l];
837: bcol = col / bs; /* block col number */
839: if (brow > bcol) {
840: if (a->ignore_ltriangular) continue; /* ignore lower triangular values */
841: 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)");
842: }
844: ridx = row % bs;
845: cidx = col % bs; /*row and col index inside the block */
846: if ((brow == bcol && ridx <= cidx) || (brow < bcol)) {
847: /* element value a(k,l) */
848: if (roworiented) value = v[l + k * n];
849: else value = v[k + l * m];
851: /* move pointer bap to a(k,l) quickly and add/insert value */
852: if (col <= lastcol) low = 0;
853: else high = nrow;
855: lastcol = col;
856: while (high - low > 7) {
857: t = (low + high) / 2;
858: if (rp[t] > bcol) high = t;
859: else low = t;
860: }
861: for (i = low; i < high; i++) {
862: if (rp[i] > bcol) break;
863: if (rp[i] == bcol) {
864: bap = ap + bs2 * i + bs * cidx + ridx;
865: if (is == ADD_VALUES) *bap += value;
866: else *bap = value;
867: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
868: if (brow == bcol && ridx < cidx) {
869: bap = ap + bs2 * i + bs * ridx + cidx;
870: if (is == ADD_VALUES) *bap += value;
871: else *bap = value;
872: }
873: goto noinsert1;
874: }
875: }
877: if (nonew == 1) goto noinsert1;
878: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", row, col);
879: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, brow, bcol, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
881: N = nrow++ - 1;
882: high++;
883: /* shift up all the later entries in this row */
884: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
885: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
886: PetscCall(PetscArrayzero(ap + bs2 * i, bs2));
887: rp[i] = bcol;
888: ap[bs2 * i + bs * cidx + ridx] = value;
889: /* for diag block, add/insert its symmetric element a(cidx,ridx) */
890: if (brow == bcol && ridx < cidx) ap[bs2 * i + bs * ridx + cidx] = value;
891: noinsert1:;
892: low = i;
893: }
894: } /* end of loop over added columns */
895: ailen[brow] = nrow;
896: } /* end of loop over added rows */
897: PetscFunctionReturn(PETSC_SUCCESS);
898: }
900: static PetscErrorCode MatICCFactor_SeqSBAIJ(Mat inA, IS row, const MatFactorInfo *info)
901: {
902: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)inA->data;
903: Mat outA;
904: PetscBool row_identity;
906: PetscFunctionBegin;
907: PetscCheck(info->levels == 0, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only levels=0 is supported for in-place icc");
908: PetscCall(ISIdentity(row, &row_identity));
909: PetscCheck(row_identity, PETSC_COMM_SELF, PETSC_ERR_SUP, "Matrix reordering is not supported");
910: 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()! */
912: outA = inA;
913: inA->factortype = MAT_FACTOR_ICC;
914: PetscCall(PetscFree(inA->solvertype));
915: PetscCall(PetscStrallocpy(MATSOLVERPETSC, &inA->solvertype));
917: PetscCall(MatMarkDiagonal_SeqSBAIJ(inA));
918: PetscCall(MatSeqSBAIJSetNumericFactorization_inplace(inA, row_identity));
920: PetscCall(PetscObjectReference((PetscObject)row));
921: PetscCall(ISDestroy(&a->row));
922: a->row = row;
923: PetscCall(PetscObjectReference((PetscObject)row));
924: PetscCall(ISDestroy(&a->col));
925: a->col = row;
927: /* Create the invert permutation so that it can be used in MatCholeskyFactorNumeric() */
928: if (a->icol) PetscCall(ISInvertPermutation(row, PETSC_DECIDE, &a->icol));
930: if (!a->solve_work) PetscCall(PetscMalloc1(inA->rmap->N + inA->rmap->bs, &a->solve_work));
932: PetscCall(MatCholeskyFactorNumeric(outA, inA, info));
933: PetscFunctionReturn(PETSC_SUCCESS);
934: }
936: static PetscErrorCode MatSeqSBAIJSetColumnIndices_SeqSBAIJ(Mat mat, PetscInt *indices)
937: {
938: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ *)mat->data;
939: PetscInt i, nz, n;
941: PetscFunctionBegin;
942: nz = baij->maxnz;
943: n = mat->cmap->n;
944: for (i = 0; i < nz; i++) baij->j[i] = indices[i];
946: baij->nz = nz;
947: for (i = 0; i < n; i++) baij->ilen[i] = baij->imax[i];
949: PetscCall(MatSetOption(mat, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
950: PetscFunctionReturn(PETSC_SUCCESS);
951: }
953: /*@
954: MatSeqSBAIJSetColumnIndices - Set the column indices for all the rows
955: in a `MATSEQSBAIJ` matrix.
957: Input Parameters:
958: + mat - the `MATSEQSBAIJ` matrix
959: - indices - the column indices
961: Level: advanced
963: Notes:
964: This can be called if you have precomputed the nonzero structure of the
965: matrix and want to provide it to the matrix object to improve the performance
966: of the `MatSetValues()` operation.
968: You MUST have set the correct numbers of nonzeros per row in the call to
969: `MatCreateSeqSBAIJ()`, and the columns indices MUST be sorted.
971: MUST be called before any calls to `MatSetValues()`
973: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreateSeqSBAIJ`
974: @*/
975: PetscErrorCode MatSeqSBAIJSetColumnIndices(Mat mat, PetscInt *indices)
976: {
977: PetscFunctionBegin;
979: PetscAssertPointer(indices, 2);
980: PetscUseMethod(mat, "MatSeqSBAIJSetColumnIndices_C", (Mat, PetscInt *), (mat, indices));
981: PetscFunctionReturn(PETSC_SUCCESS);
982: }
984: static PetscErrorCode MatCopy_SeqSBAIJ(Mat A, Mat B, MatStructure str)
985: {
986: PetscBool isbaij;
988: PetscFunctionBegin;
989: PetscCall(PetscObjectTypeCompareAny((PetscObject)B, &isbaij, MATSEQSBAIJ, MATMPISBAIJ, ""));
990: PetscCheck(isbaij, PetscObjectComm((PetscObject)B), PETSC_ERR_SUP, "Not for matrix type %s", ((PetscObject)B)->type_name);
991: /* If the two matrices have the same copy implementation and nonzero pattern, use fast copy. */
992: if (str == SAME_NONZERO_PATTERN && A->ops->copy == B->ops->copy) {
993: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
994: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
996: PetscCheck(a->i[a->mbs] == b->i[b->mbs], PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Number of nonzeros in two matrices are different");
997: PetscCheck(a->mbs == b->mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Number of rows in two matrices are different");
998: PetscCheck(a->bs2 == b->bs2, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Different block size");
999: PetscCall(PetscArraycpy(b->a, a->a, a->bs2 * a->i[a->mbs]));
1000: PetscCall(PetscObjectStateIncrease((PetscObject)B));
1001: } else {
1002: PetscCall(MatGetRowUpperTriangular(A));
1003: PetscCall(MatCopy_Basic(A, B, str));
1004: PetscCall(MatRestoreRowUpperTriangular(A));
1005: }
1006: PetscFunctionReturn(PETSC_SUCCESS);
1007: }
1009: static PetscErrorCode MatSeqSBAIJGetArray_SeqSBAIJ(Mat A, PetscScalar *array[])
1010: {
1011: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1013: PetscFunctionBegin;
1014: *array = a->a;
1015: PetscFunctionReturn(PETSC_SUCCESS);
1016: }
1018: static PetscErrorCode MatSeqSBAIJRestoreArray_SeqSBAIJ(Mat A, PetscScalar *array[])
1019: {
1020: PetscFunctionBegin;
1021: *array = NULL;
1022: PetscFunctionReturn(PETSC_SUCCESS);
1023: }
1025: PetscErrorCode MatAXPYGetPreallocation_SeqSBAIJ(Mat Y, Mat X, PetscInt *nnz)
1026: {
1027: PetscInt bs = Y->rmap->bs, mbs = Y->rmap->N / bs;
1028: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ *)X->data;
1029: Mat_SeqSBAIJ *y = (Mat_SeqSBAIJ *)Y->data;
1031: PetscFunctionBegin;
1032: /* Set the number of nonzeros in the new matrix */
1033: PetscCall(MatAXPYGetPreallocation_SeqX_private(mbs, x->i, x->j, y->i, y->j, nnz));
1034: PetscFunctionReturn(PETSC_SUCCESS);
1035: }
1037: static PetscErrorCode MatAXPY_SeqSBAIJ(Mat Y, PetscScalar a, Mat X, MatStructure str)
1038: {
1039: Mat_SeqSBAIJ *x = (Mat_SeqSBAIJ *)X->data, *y = (Mat_SeqSBAIJ *)Y->data;
1040: PetscInt bs = Y->rmap->bs, bs2 = bs * bs;
1041: PetscBLASInt one = 1;
1043: PetscFunctionBegin;
1044: if (str == UNKNOWN_NONZERO_PATTERN || (PetscDefined(USE_DEBUG) && str == SAME_NONZERO_PATTERN)) {
1045: PetscBool e = x->nz == y->nz && x->mbs == y->mbs ? PETSC_TRUE : PETSC_FALSE;
1046: if (e) {
1047: PetscCall(PetscArraycmp(x->i, y->i, x->mbs + 1, &e));
1048: if (e) {
1049: PetscCall(PetscArraycmp(x->j, y->j, x->i[x->mbs], &e));
1050: if (e) str = SAME_NONZERO_PATTERN;
1051: }
1052: }
1053: if (!e) PetscCheck(str != SAME_NONZERO_PATTERN, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MatStructure is not SAME_NONZERO_PATTERN");
1054: }
1055: if (str == SAME_NONZERO_PATTERN) {
1056: PetscScalar alpha = a;
1057: PetscBLASInt bnz;
1058: PetscCall(PetscBLASIntCast(x->nz * bs2, &bnz));
1059: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, x->a, &one, y->a, &one));
1060: PetscCall(PetscObjectStateIncrease((PetscObject)Y));
1061: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1062: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_TRUE));
1063: PetscCall(MatAXPY_Basic(Y, a, X, str));
1064: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_FALSE));
1065: } else {
1066: Mat B;
1067: PetscInt *nnz;
1068: PetscCheck(bs == X->rmap->bs, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Matrices must have same block size");
1069: PetscCall(MatGetRowUpperTriangular(X));
1070: PetscCall(MatGetRowUpperTriangular(Y));
1071: PetscCall(PetscMalloc1(Y->rmap->N, &nnz));
1072: PetscCall(MatCreate(PetscObjectComm((PetscObject)Y), &B));
1073: PetscCall(PetscObjectSetName((PetscObject)B, ((PetscObject)Y)->name));
1074: PetscCall(MatSetSizes(B, Y->rmap->n, Y->cmap->n, Y->rmap->N, Y->cmap->N));
1075: PetscCall(MatSetBlockSizesFromMats(B, Y, Y));
1076: PetscCall(MatSetType(B, ((PetscObject)Y)->type_name));
1077: PetscCall(MatAXPYGetPreallocation_SeqSBAIJ(Y, X, nnz));
1078: PetscCall(MatSeqSBAIJSetPreallocation(B, bs, 0, nnz));
1080: PetscCall(MatAXPY_BasicWithPreallocation(B, Y, a, X, str));
1082: PetscCall(MatHeaderMerge(Y, &B));
1083: PetscCall(PetscFree(nnz));
1084: PetscCall(MatRestoreRowUpperTriangular(X));
1085: PetscCall(MatRestoreRowUpperTriangular(Y));
1086: }
1087: PetscFunctionReturn(PETSC_SUCCESS);
1088: }
1090: static PetscErrorCode MatIsStructurallySymmetric_SeqSBAIJ(Mat A, PetscBool *flg)
1091: {
1092: PetscFunctionBegin;
1093: *flg = PETSC_TRUE;
1094: PetscFunctionReturn(PETSC_SUCCESS);
1095: }
1097: static PetscErrorCode MatConjugate_SeqSBAIJ(Mat A)
1098: {
1099: #if defined(PETSC_USE_COMPLEX)
1100: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1101: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1102: MatScalar *aa = a->a;
1104: PetscFunctionBegin;
1105: for (i = 0; i < nz; i++) aa[i] = PetscConj(aa[i]);
1106: #else
1107: PetscFunctionBegin;
1108: #endif
1109: PetscFunctionReturn(PETSC_SUCCESS);
1110: }
1112: static PetscErrorCode MatRealPart_SeqSBAIJ(Mat A)
1113: {
1114: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1115: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1116: MatScalar *aa = a->a;
1118: PetscFunctionBegin;
1119: for (i = 0; i < nz; i++) aa[i] = PetscRealPart(aa[i]);
1120: PetscFunctionReturn(PETSC_SUCCESS);
1121: }
1123: static PetscErrorCode MatImaginaryPart_SeqSBAIJ(Mat A)
1124: {
1125: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1126: PetscInt i, nz = a->bs2 * a->i[a->mbs];
1127: MatScalar *aa = a->a;
1129: PetscFunctionBegin;
1130: for (i = 0; i < nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1131: PetscFunctionReturn(PETSC_SUCCESS);
1132: }
1134: static PetscErrorCode MatZeroRowsColumns_SeqSBAIJ(Mat A, PetscInt is_n, const PetscInt is_idx[], PetscScalar diag, Vec x, Vec b)
1135: {
1136: Mat_SeqSBAIJ *baij = (Mat_SeqSBAIJ *)A->data;
1137: PetscInt i, j, k, count;
1138: PetscInt bs = A->rmap->bs, bs2 = baij->bs2, row, col;
1139: PetscScalar zero = 0.0;
1140: MatScalar *aa;
1141: const PetscScalar *xx;
1142: PetscScalar *bb;
1143: PetscBool *zeroed, vecs = PETSC_FALSE;
1145: PetscFunctionBegin;
1146: /* fix right-hand side if needed */
1147: if (x && b) {
1148: PetscCall(VecGetArrayRead(x, &xx));
1149: PetscCall(VecGetArray(b, &bb));
1150: vecs = PETSC_TRUE;
1151: }
1153: /* zero the columns */
1154: PetscCall(PetscCalloc1(A->rmap->n, &zeroed));
1155: for (i = 0; i < is_n; i++) {
1156: 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]);
1157: zeroed[is_idx[i]] = PETSC_TRUE;
1158: }
1159: if (vecs) {
1160: for (i = 0; i < A->rmap->N; i++) {
1161: row = i / bs;
1162: for (j = baij->i[row]; j < baij->i[row + 1]; j++) {
1163: for (k = 0; k < bs; k++) {
1164: col = bs * baij->j[j] + k;
1165: if (col <= i) continue;
1166: aa = ((MatScalar *)baij->a) + j * bs2 + (i % bs) + bs * k;
1167: if (!zeroed[i] && zeroed[col]) bb[i] -= aa[0] * xx[col];
1168: if (zeroed[i] && !zeroed[col]) bb[col] -= aa[0] * xx[i];
1169: }
1170: }
1171: }
1172: for (i = 0; i < is_n; i++) bb[is_idx[i]] = diag * xx[is_idx[i]];
1173: }
1175: for (i = 0; i < A->rmap->N; i++) {
1176: if (!zeroed[i]) {
1177: row = i / bs;
1178: for (j = baij->i[row]; j < baij->i[row + 1]; j++) {
1179: for (k = 0; k < bs; k++) {
1180: col = bs * baij->j[j] + k;
1181: if (zeroed[col]) {
1182: aa = ((MatScalar *)baij->a) + j * bs2 + (i % bs) + bs * k;
1183: aa[0] = 0.0;
1184: }
1185: }
1186: }
1187: }
1188: }
1189: PetscCall(PetscFree(zeroed));
1190: if (vecs) {
1191: PetscCall(VecRestoreArrayRead(x, &xx));
1192: PetscCall(VecRestoreArray(b, &bb));
1193: }
1195: /* zero the rows */
1196: for (i = 0; i < is_n; i++) {
1197: row = is_idx[i];
1198: count = (baij->i[row / bs + 1] - baij->i[row / bs]) * bs;
1199: aa = ((MatScalar *)baij->a) + baij->i[row / bs] * bs2 + (row % bs);
1200: for (k = 0; k < count; k++) {
1201: aa[0] = zero;
1202: aa += bs;
1203: }
1204: if (diag != 0.0) PetscUseTypeMethod(A, setvalues, 1, &row, 1, &row, &diag, INSERT_VALUES);
1205: }
1206: PetscCall(MatAssemblyEnd_SeqSBAIJ(A, MAT_FINAL_ASSEMBLY));
1207: PetscFunctionReturn(PETSC_SUCCESS);
1208: }
1210: static PetscErrorCode MatShift_SeqSBAIJ(Mat Y, PetscScalar a)
1211: {
1212: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)Y->data;
1214: PetscFunctionBegin;
1215: if (!Y->preallocated || !aij->nz) PetscCall(MatSeqSBAIJSetPreallocation(Y, Y->rmap->bs, 1, NULL));
1216: PetscCall(MatShift_Basic(Y, a));
1217: PetscFunctionReturn(PETSC_SUCCESS);
1218: }
1220: PetscErrorCode MatEliminateZeros_SeqSBAIJ(Mat A, PetscBool keep)
1221: {
1222: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
1223: PetscInt fshift = 0, fshift_prev = 0, i, *ai = a->i, *aj = a->j, *imax = a->imax, j, k;
1224: PetscInt m = A->rmap->N, *ailen = a->ilen;
1225: PetscInt mbs = a->mbs, bs2 = a->bs2, rmax = 0;
1226: MatScalar *aa = a->a, *ap;
1227: PetscBool zero;
1229: PetscFunctionBegin;
1230: PetscCheck(A->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot eliminate zeros for unassembled matrix");
1231: if (m) rmax = ailen[0];
1232: for (i = 1; i <= mbs; i++) {
1233: for (k = ai[i - 1]; k < ai[i]; k++) {
1234: zero = PETSC_TRUE;
1235: ap = aa + bs2 * k;
1236: for (j = 0; j < bs2 && zero; j++) {
1237: if (ap[j] != 0.0) zero = PETSC_FALSE;
1238: }
1239: if (zero && (aj[k] != i - 1 || !keep)) fshift++;
1240: else {
1241: if (zero && aj[k] == i - 1) PetscCall(PetscInfo(A, "Keep the diagonal block at row %" PetscInt_FMT "\n", i - 1));
1242: aj[k - fshift] = aj[k];
1243: PetscCall(PetscArraymove(ap - bs2 * fshift, ap, bs2));
1244: }
1245: }
1246: ai[i - 1] -= fshift_prev;
1247: fshift_prev = fshift;
1248: ailen[i - 1] = imax[i - 1] = ai[i] - fshift - ai[i - 1];
1249: a->nonzerorowcnt += ((ai[i] - fshift - ai[i - 1]) > 0);
1250: rmax = PetscMax(rmax, ailen[i - 1]);
1251: }
1252: if (fshift) {
1253: if (mbs) {
1254: ai[mbs] -= fshift;
1255: a->nz = ai[mbs];
1256: }
1257: 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));
1258: A->nonzerostate++;
1259: A->info.nz_unneeded += (PetscReal)fshift;
1260: a->rmax = rmax;
1261: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
1262: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
1263: }
1264: PetscFunctionReturn(PETSC_SUCCESS);
1265: }
1267: static struct _MatOps MatOps_Values = {MatSetValues_SeqSBAIJ,
1268: MatGetRow_SeqSBAIJ,
1269: MatRestoreRow_SeqSBAIJ,
1270: MatMult_SeqSBAIJ_N,
1271: /* 4*/ MatMultAdd_SeqSBAIJ_N,
1272: MatMult_SeqSBAIJ_N, /* transpose versions are same as non-transpose versions */
1273: MatMultAdd_SeqSBAIJ_N,
1274: NULL,
1275: NULL,
1276: NULL,
1277: /* 10*/ NULL,
1278: NULL,
1279: MatCholeskyFactor_SeqSBAIJ,
1280: MatSOR_SeqSBAIJ,
1281: MatTranspose_SeqSBAIJ,
1282: /* 15*/ MatGetInfo_SeqSBAIJ,
1283: MatEqual_SeqSBAIJ,
1284: MatGetDiagonal_SeqSBAIJ,
1285: MatDiagonalScale_SeqSBAIJ,
1286: MatNorm_SeqSBAIJ,
1287: /* 20*/ NULL,
1288: MatAssemblyEnd_SeqSBAIJ,
1289: MatSetOption_SeqSBAIJ,
1290: MatZeroEntries_SeqSBAIJ,
1291: /* 24*/ NULL,
1292: NULL,
1293: NULL,
1294: NULL,
1295: NULL,
1296: /* 29*/ MatSetUp_Seq_Hash,
1297: NULL,
1298: NULL,
1299: NULL,
1300: NULL,
1301: /* 34*/ MatDuplicate_SeqSBAIJ,
1302: NULL,
1303: NULL,
1304: NULL,
1305: MatICCFactor_SeqSBAIJ,
1306: /* 39*/ MatAXPY_SeqSBAIJ,
1307: MatCreateSubMatrices_SeqSBAIJ,
1308: MatIncreaseOverlap_SeqSBAIJ,
1309: MatGetValues_SeqSBAIJ,
1310: MatCopy_SeqSBAIJ,
1311: /* 44*/ NULL,
1312: MatScale_SeqSBAIJ,
1313: MatShift_SeqSBAIJ,
1314: NULL,
1315: MatZeroRowsColumns_SeqSBAIJ,
1316: /* 49*/ NULL,
1317: MatGetRowIJ_SeqSBAIJ,
1318: MatRestoreRowIJ_SeqSBAIJ,
1319: NULL,
1320: NULL,
1321: /* 54*/ NULL,
1322: NULL,
1323: NULL,
1324: MatPermute_SeqSBAIJ,
1325: MatSetValuesBlocked_SeqSBAIJ,
1326: /* 59*/ MatCreateSubMatrix_SeqSBAIJ,
1327: NULL,
1328: NULL,
1329: NULL,
1330: NULL,
1331: /* 64*/ NULL,
1332: NULL,
1333: NULL,
1334: NULL,
1335: NULL,
1336: /* 69*/ MatGetRowMaxAbs_SeqSBAIJ,
1337: NULL,
1338: MatConvert_MPISBAIJ_Basic,
1339: NULL,
1340: NULL,
1341: /* 74*/ NULL,
1342: NULL,
1343: NULL,
1344: NULL,
1345: NULL,
1346: /* 79*/ NULL,
1347: NULL,
1348: NULL,
1349: MatGetInertia_SeqSBAIJ,
1350: MatLoad_SeqSBAIJ,
1351: /* 84*/ NULL,
1352: NULL,
1353: MatIsStructurallySymmetric_SeqSBAIJ,
1354: NULL,
1355: NULL,
1356: /* 89*/ NULL,
1357: NULL,
1358: NULL,
1359: NULL,
1360: NULL,
1361: /* 94*/ NULL,
1362: NULL,
1363: NULL,
1364: NULL,
1365: NULL,
1366: /* 99*/ NULL,
1367: NULL,
1368: NULL,
1369: MatConjugate_SeqSBAIJ,
1370: NULL,
1371: /*104*/ NULL,
1372: MatRealPart_SeqSBAIJ,
1373: MatImaginaryPart_SeqSBAIJ,
1374: MatGetRowUpperTriangular_SeqSBAIJ,
1375: MatRestoreRowUpperTriangular_SeqSBAIJ,
1376: /*109*/ NULL,
1377: NULL,
1378: NULL,
1379: NULL,
1380: MatMissingDiagonal_SeqSBAIJ,
1381: /*114*/ NULL,
1382: NULL,
1383: NULL,
1384: NULL,
1385: NULL,
1386: /*119*/ NULL,
1387: NULL,
1388: NULL,
1389: NULL,
1390: NULL,
1391: /*124*/ NULL,
1392: NULL,
1393: NULL,
1394: NULL,
1395: NULL,
1396: /*129*/ NULL,
1397: NULL,
1398: NULL,
1399: NULL,
1400: NULL,
1401: /*134*/ NULL,
1402: NULL,
1403: NULL,
1404: NULL,
1405: NULL,
1406: /*139*/ MatSetBlockSizes_Default,
1407: NULL,
1408: NULL,
1409: NULL,
1410: NULL,
1411: /*144*/ MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ,
1412: NULL,
1413: NULL,
1414: NULL,
1415: NULL,
1416: NULL,
1417: /*150*/ NULL,
1418: MatEliminateZeros_SeqSBAIJ,
1419: NULL};
1421: static PetscErrorCode MatStoreValues_SeqSBAIJ(Mat mat)
1422: {
1423: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)mat->data;
1424: PetscInt nz = aij->i[mat->rmap->N] * mat->rmap->bs * aij->bs2;
1426: PetscFunctionBegin;
1427: PetscCheck(aij->nonew == 1, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1429: /* allocate space for values if not already there */
1430: if (!aij->saved_values) PetscCall(PetscMalloc1(nz + 1, &aij->saved_values));
1432: /* copy values over */
1433: PetscCall(PetscArraycpy(aij->saved_values, aij->a, nz));
1434: PetscFunctionReturn(PETSC_SUCCESS);
1435: }
1437: static PetscErrorCode MatRetrieveValues_SeqSBAIJ(Mat mat)
1438: {
1439: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)mat->data;
1440: PetscInt nz = aij->i[mat->rmap->N] * mat->rmap->bs * aij->bs2;
1442: PetscFunctionBegin;
1443: PetscCheck(aij->nonew == 1, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
1444: PetscCheck(aij->saved_values, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatStoreValues(A);first");
1446: /* copy values over */
1447: PetscCall(PetscArraycpy(aij->a, aij->saved_values, nz));
1448: PetscFunctionReturn(PETSC_SUCCESS);
1449: }
1451: static PetscErrorCode MatSeqSBAIJSetPreallocation_SeqSBAIJ(Mat B, PetscInt bs, PetscInt nz, const PetscInt nnz[])
1452: {
1453: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
1454: PetscInt i, mbs, nbs, bs2;
1455: PetscBool skipallocation = PETSC_FALSE, flg = PETSC_FALSE, realalloc = PETSC_FALSE;
1457: PetscFunctionBegin;
1458: if (B->hash_active) {
1459: PetscInt bs;
1460: B->ops[0] = b->cops;
1461: PetscCall(PetscHMapIJVDestroy(&b->ht));
1462: PetscCall(MatGetBlockSize(B, &bs));
1463: if (bs > 1) PetscCall(PetscHSetIJDestroy(&b->bht));
1464: PetscCall(PetscFree(b->dnz));
1465: PetscCall(PetscFree(b->bdnz));
1466: B->hash_active = PETSC_FALSE;
1467: }
1468: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
1470: PetscCall(MatSetBlockSize(B, PetscAbs(bs)));
1471: PetscCall(PetscLayoutSetUp(B->rmap));
1472: PetscCall(PetscLayoutSetUp(B->cmap));
1473: 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);
1474: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
1476: B->preallocated = PETSC_TRUE;
1478: mbs = B->rmap->N / bs;
1479: nbs = B->cmap->n / bs;
1480: bs2 = bs * bs;
1482: 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");
1484: if (nz == MAT_SKIP_ALLOCATION) {
1485: skipallocation = PETSC_TRUE;
1486: nz = 0;
1487: }
1489: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 3;
1490: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nz cannot be less than 0: value %" PetscInt_FMT, nz);
1491: if (nnz) {
1492: for (i = 0; i < mbs; i++) {
1493: 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]);
1494: 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);
1495: }
1496: }
1498: B->ops->mult = MatMult_SeqSBAIJ_N;
1499: B->ops->multadd = MatMultAdd_SeqSBAIJ_N;
1500: B->ops->multtranspose = MatMult_SeqSBAIJ_N;
1501: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_N;
1503: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_no_unroll", &flg, NULL));
1504: if (!flg) {
1505: switch (bs) {
1506: case 1:
1507: B->ops->mult = MatMult_SeqSBAIJ_1;
1508: B->ops->multadd = MatMultAdd_SeqSBAIJ_1;
1509: B->ops->multtranspose = MatMult_SeqSBAIJ_1;
1510: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_1;
1511: break;
1512: case 2:
1513: B->ops->mult = MatMult_SeqSBAIJ_2;
1514: B->ops->multadd = MatMultAdd_SeqSBAIJ_2;
1515: B->ops->multtranspose = MatMult_SeqSBAIJ_2;
1516: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_2;
1517: break;
1518: case 3:
1519: B->ops->mult = MatMult_SeqSBAIJ_3;
1520: B->ops->multadd = MatMultAdd_SeqSBAIJ_3;
1521: B->ops->multtranspose = MatMult_SeqSBAIJ_3;
1522: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_3;
1523: break;
1524: case 4:
1525: B->ops->mult = MatMult_SeqSBAIJ_4;
1526: B->ops->multadd = MatMultAdd_SeqSBAIJ_4;
1527: B->ops->multtranspose = MatMult_SeqSBAIJ_4;
1528: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_4;
1529: break;
1530: case 5:
1531: B->ops->mult = MatMult_SeqSBAIJ_5;
1532: B->ops->multadd = MatMultAdd_SeqSBAIJ_5;
1533: B->ops->multtranspose = MatMult_SeqSBAIJ_5;
1534: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_5;
1535: break;
1536: case 6:
1537: B->ops->mult = MatMult_SeqSBAIJ_6;
1538: B->ops->multadd = MatMultAdd_SeqSBAIJ_6;
1539: B->ops->multtranspose = MatMult_SeqSBAIJ_6;
1540: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_6;
1541: break;
1542: case 7:
1543: B->ops->mult = MatMult_SeqSBAIJ_7;
1544: B->ops->multadd = MatMultAdd_SeqSBAIJ_7;
1545: B->ops->multtranspose = MatMult_SeqSBAIJ_7;
1546: B->ops->multtransposeadd = MatMultAdd_SeqSBAIJ_7;
1547: break;
1548: }
1549: }
1551: b->mbs = mbs;
1552: b->nbs = nbs;
1553: if (!skipallocation) {
1554: if (!b->imax) {
1555: PetscCall(PetscMalloc2(mbs, &b->imax, mbs, &b->ilen));
1557: b->free_imax_ilen = PETSC_TRUE;
1558: }
1559: if (!nnz) {
1560: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
1561: else if (nz <= 0) nz = 1;
1562: nz = PetscMin(nbs, nz);
1563: for (i = 0; i < mbs; i++) b->imax[i] = nz;
1564: PetscCall(PetscIntMultError(nz, mbs, &nz));
1565: } else {
1566: PetscInt64 nz64 = 0;
1567: for (i = 0; i < mbs; i++) {
1568: b->imax[i] = nnz[i];
1569: nz64 += nnz[i];
1570: }
1571: PetscCall(PetscIntCast(nz64, &nz));
1572: }
1573: /* b->ilen will count nonzeros in each block row so far. */
1574: for (i = 0; i < mbs; i++) b->ilen[i] = 0;
1575: /* nz=(nz+mbs)/2; */ /* total diagonal and superdiagonal nonzero blocks */
1577: /* allocate the matrix space */
1578: PetscCall(MatSeqXAIJFreeAIJ(B, &b->a, &b->j, &b->i));
1579: PetscCall(PetscMalloc3(bs2 * nz, &b->a, nz, &b->j, B->rmap->N + 1, &b->i));
1580: PetscCall(PetscArrayzero(b->a, nz * bs2));
1581: PetscCall(PetscArrayzero(b->j, nz));
1583: b->singlemalloc = PETSC_TRUE;
1585: /* pointer to beginning of each row */
1586: b->i[0] = 0;
1587: for (i = 1; i < mbs + 1; i++) b->i[i] = b->i[i - 1] + b->imax[i - 1];
1589: b->free_a = PETSC_TRUE;
1590: b->free_ij = PETSC_TRUE;
1591: } else {
1592: b->free_a = PETSC_FALSE;
1593: b->free_ij = PETSC_FALSE;
1594: }
1596: b->bs2 = bs2;
1597: b->nz = 0;
1598: b->maxnz = nz;
1599: b->inew = NULL;
1600: b->jnew = NULL;
1601: b->anew = NULL;
1602: b->a2anew = NULL;
1603: b->permute = PETSC_FALSE;
1605: B->was_assembled = PETSC_FALSE;
1606: B->assembled = PETSC_FALSE;
1607: if (realalloc) PetscCall(MatSetOption(B, MAT_NEW_NONZERO_ALLOCATION_ERR, PETSC_TRUE));
1608: PetscFunctionReturn(PETSC_SUCCESS);
1609: }
1611: static PetscErrorCode MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ(Mat B, PetscInt bs, const PetscInt ii[], const PetscInt jj[], const PetscScalar V[])
1612: {
1613: PetscInt i, j, m, nz, anz, nz_max = 0, *nnz;
1614: PetscScalar *values = NULL;
1615: Mat_SeqSBAIJ *b = (Mat_SeqSBAIJ *)B->data;
1616: PetscBool roworiented = b->roworiented;
1617: PetscBool ilw = b->ignore_ltriangular;
1619: PetscFunctionBegin;
1620: PetscCheck(bs >= 1, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_OUTOFRANGE, "Invalid block size specified, must be positive but it is %" PetscInt_FMT, bs);
1621: PetscCall(PetscLayoutSetBlockSize(B->rmap, bs));
1622: PetscCall(PetscLayoutSetBlockSize(B->cmap, bs));
1623: PetscCall(PetscLayoutSetUp(B->rmap));
1624: PetscCall(PetscLayoutSetUp(B->cmap));
1625: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
1626: m = B->rmap->n / bs;
1628: PetscCheck(!ii[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "ii[0] must be 0 but it is %" PetscInt_FMT, ii[0]);
1629: PetscCall(PetscMalloc1(m + 1, &nnz));
1630: for (i = 0; i < m; i++) {
1631: nz = ii[i + 1] - ii[i];
1632: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row %" PetscInt_FMT " has a negative number of columns %" PetscInt_FMT, i, nz);
1633: PetscCheckSorted(nz, jj + ii[i]);
1634: anz = 0;
1635: for (j = 0; j < nz; j++) {
1636: /* count only values on the diagonal or above */
1637: if (jj[ii[i] + j] >= i) {
1638: anz = nz - j;
1639: break;
1640: }
1641: }
1642: nz_max = PetscMax(nz_max, nz);
1643: nnz[i] = anz;
1644: }
1645: PetscCall(MatSeqSBAIJSetPreallocation(B, bs, 0, nnz));
1646: PetscCall(PetscFree(nnz));
1648: values = (PetscScalar *)V;
1649: if (!values) PetscCall(PetscCalloc1(bs * bs * nz_max, &values));
1650: b->ignore_ltriangular = PETSC_TRUE;
1651: for (i = 0; i < m; i++) {
1652: PetscInt ncols = ii[i + 1] - ii[i];
1653: const PetscInt *icols = jj + ii[i];
1655: if (!roworiented || bs == 1) {
1656: const PetscScalar *svals = values + (V ? (bs * bs * ii[i]) : 0);
1657: PetscCall(MatSetValuesBlocked_SeqSBAIJ(B, 1, &i, ncols, icols, svals, INSERT_VALUES));
1658: } else {
1659: for (j = 0; j < ncols; j++) {
1660: const PetscScalar *svals = values + (V ? (bs * bs * (ii[i] + j)) : 0);
1661: PetscCall(MatSetValuesBlocked_SeqSBAIJ(B, 1, &i, 1, &icols[j], svals, INSERT_VALUES));
1662: }
1663: }
1664: }
1665: if (!V) PetscCall(PetscFree(values));
1666: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
1667: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
1668: PetscCall(MatSetOption(B, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
1669: b->ignore_ltriangular = ilw;
1670: PetscFunctionReturn(PETSC_SUCCESS);
1671: }
1673: /*
1674: This is used to set the numeric factorization for both Cholesky and ICC symbolic factorization
1675: */
1676: PetscErrorCode MatSeqSBAIJSetNumericFactorization_inplace(Mat B, PetscBool natural)
1677: {
1678: PetscBool flg = PETSC_FALSE;
1679: PetscInt bs = B->rmap->bs;
1681: PetscFunctionBegin;
1682: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_no_unroll", &flg, NULL));
1683: if (flg) bs = 8;
1685: if (!natural) {
1686: switch (bs) {
1687: case 1:
1688: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_inplace;
1689: break;
1690: case 2:
1691: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2;
1692: break;
1693: case 3:
1694: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3;
1695: break;
1696: case 4:
1697: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4;
1698: break;
1699: case 5:
1700: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5;
1701: break;
1702: case 6:
1703: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6;
1704: break;
1705: case 7:
1706: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7;
1707: break;
1708: default:
1709: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N;
1710: break;
1711: }
1712: } else {
1713: switch (bs) {
1714: case 1:
1715: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_1_NaturalOrdering_inplace;
1716: break;
1717: case 2:
1718: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_2_NaturalOrdering;
1719: break;
1720: case 3:
1721: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_3_NaturalOrdering;
1722: break;
1723: case 4:
1724: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_4_NaturalOrdering;
1725: break;
1726: case 5:
1727: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_5_NaturalOrdering;
1728: break;
1729: case 6:
1730: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_6_NaturalOrdering;
1731: break;
1732: case 7:
1733: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_7_NaturalOrdering;
1734: break;
1735: default:
1736: B->ops->choleskyfactornumeric = MatCholeskyFactorNumeric_SeqSBAIJ_N_NaturalOrdering;
1737: break;
1738: }
1739: }
1740: PetscFunctionReturn(PETSC_SUCCESS);
1741: }
1743: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqAIJ(Mat, MatType, MatReuse, Mat *);
1744: PETSC_INTERN PetscErrorCode MatConvert_SeqSBAIJ_SeqBAIJ(Mat, MatType, MatReuse, Mat *);
1745: static PetscErrorCode MatFactorGetSolverType_petsc(Mat A, MatSolverType *type)
1746: {
1747: PetscFunctionBegin;
1748: *type = MATSOLVERPETSC;
1749: PetscFunctionReturn(PETSC_SUCCESS);
1750: }
1752: PETSC_INTERN PetscErrorCode MatGetFactor_seqsbaij_petsc(Mat A, MatFactorType ftype, Mat *B)
1753: {
1754: PetscInt n = A->rmap->n;
1756: PetscFunctionBegin;
1757: #if defined(PETSC_USE_COMPLEX)
1758: if ((ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) && A->hermitian == PETSC_BOOL3_TRUE && A->symmetric != PETSC_BOOL3_TRUE) {
1759: PetscCall(PetscInfo(A, "Hermitian MAT_FACTOR_CHOLESKY or MAT_FACTOR_ICC are not supported. Use MAT_FACTOR_LU instead.\n"));
1760: *B = NULL;
1761: PetscFunctionReturn(PETSC_SUCCESS);
1762: }
1763: #endif
1765: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), B));
1766: PetscCall(MatSetSizes(*B, n, n, n, n));
1767: if (ftype == MAT_FACTOR_CHOLESKY || ftype == MAT_FACTOR_ICC) {
1768: PetscCall(MatSetType(*B, MATSEQSBAIJ));
1769: PetscCall(MatSeqSBAIJSetPreallocation(*B, A->rmap->bs, MAT_SKIP_ALLOCATION, NULL));
1771: (*B)->ops->choleskyfactorsymbolic = MatCholeskyFactorSymbolic_SeqSBAIJ;
1772: (*B)->ops->iccfactorsymbolic = MatICCFactorSymbolic_SeqSBAIJ;
1773: PetscCall(PetscStrallocpy(MATORDERINGNATURAL, (char **)&(*B)->preferredordering[MAT_FACTOR_CHOLESKY]));
1774: PetscCall(PetscStrallocpy(MATORDERINGNATURAL, (char **)&(*B)->preferredordering[MAT_FACTOR_ICC]));
1775: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Factor type not supported");
1777: (*B)->factortype = ftype;
1778: (*B)->canuseordering = PETSC_TRUE;
1779: PetscCall(PetscFree((*B)->solvertype));
1780: PetscCall(PetscStrallocpy(MATSOLVERPETSC, &(*B)->solvertype));
1781: PetscCall(PetscObjectComposeFunction((PetscObject)*B, "MatFactorGetSolverType_C", MatFactorGetSolverType_petsc));
1782: PetscFunctionReturn(PETSC_SUCCESS);
1783: }
1785: /*@C
1786: MatSeqSBAIJGetArray - gives access to the array where the numerical data for a `MATSEQSBAIJ` matrix is stored
1788: Not Collective
1790: Input Parameter:
1791: . A - a `MATSEQSBAIJ` matrix
1793: Output Parameter:
1794: . array - pointer to the data
1796: Level: intermediate
1798: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatSeqSBAIJRestoreArray()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`
1799: @*/
1800: PetscErrorCode MatSeqSBAIJGetArray(Mat A, PetscScalar **array)
1801: {
1802: PetscFunctionBegin;
1803: PetscUseMethod(A, "MatSeqSBAIJGetArray_C", (Mat, PetscScalar **), (A, array));
1804: PetscFunctionReturn(PETSC_SUCCESS);
1805: }
1807: /*@C
1808: MatSeqSBAIJRestoreArray - returns access to the array where the numerical data for a `MATSEQSBAIJ` matrix is stored obtained by `MatSeqSBAIJGetArray()`
1810: Not Collective
1812: Input Parameters:
1813: + A - a `MATSEQSBAIJ` matrix
1814: - array - pointer to the data
1816: Level: intermediate
1818: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatSeqSBAIJGetArray()`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArray()`
1819: @*/
1820: PetscErrorCode MatSeqSBAIJRestoreArray(Mat A, PetscScalar **array)
1821: {
1822: PetscFunctionBegin;
1823: PetscUseMethod(A, "MatSeqSBAIJRestoreArray_C", (Mat, PetscScalar **), (A, array));
1824: PetscFunctionReturn(PETSC_SUCCESS);
1825: }
1827: /*MC
1828: MATSEQSBAIJ - MATSEQSBAIJ = "seqsbaij" - A matrix type to be used for sequential symmetric block sparse matrices,
1829: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1831: For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you
1832: can call `MatSetOption`(`Mat`, `MAT_HERMITIAN`).
1834: Options Database Key:
1835: . -mat_type seqsbaij - sets the matrix type to "seqsbaij" during a call to `MatSetFromOptions()`
1837: Level: beginner
1839: Notes:
1840: By default if you insert values into the lower triangular part of the matrix they are simply ignored (since they are not
1841: stored and it is assumed they symmetric to the upper triangular). If you call `MatSetOption`(`Mat`,`MAT_IGNORE_LOWER_TRIANGULAR`,`PETSC_FALSE`) or use
1842: 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.
1844: The number of rows in the matrix must be less than or equal to the number of columns
1846: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreateSeqSBAIJ()`, `MatType`, `MATMPISBAIJ`
1847: M*/
1848: PETSC_EXTERN PetscErrorCode MatCreate_SeqSBAIJ(Mat B)
1849: {
1850: Mat_SeqSBAIJ *b;
1851: PetscMPIInt size;
1852: PetscBool no_unroll = PETSC_FALSE, no_inode = PETSC_FALSE;
1854: PetscFunctionBegin;
1855: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &size));
1856: PetscCheck(size <= 1, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Comm must be of size 1");
1858: PetscCall(PetscNew(&b));
1859: B->data = (void *)b;
1860: B->ops[0] = MatOps_Values;
1862: B->ops->destroy = MatDestroy_SeqSBAIJ;
1863: B->ops->view = MatView_SeqSBAIJ;
1864: b->row = NULL;
1865: b->icol = NULL;
1866: b->reallocs = 0;
1867: b->saved_values = NULL;
1868: b->inode.limit = 5;
1869: b->inode.max_limit = 5;
1871: b->roworiented = PETSC_TRUE;
1872: b->nonew = 0;
1873: b->diag = NULL;
1874: b->solve_work = NULL;
1875: b->mult_work = NULL;
1876: B->spptr = NULL;
1877: B->info.nz_unneeded = (PetscReal)b->maxnz * b->bs2;
1878: b->keepnonzeropattern = PETSC_FALSE;
1880: b->inew = NULL;
1881: b->jnew = NULL;
1882: b->anew = NULL;
1883: b->a2anew = NULL;
1884: b->permute = PETSC_FALSE;
1886: b->ignore_ltriangular = PETSC_TRUE;
1888: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_ignore_lower_triangular", &b->ignore_ltriangular, NULL));
1890: b->getrow_utriangular = PETSC_FALSE;
1892: PetscCall(PetscOptionsGetBool(((PetscObject)B)->options, ((PetscObject)B)->prefix, "-mat_getrow_uppertriangular", &b->getrow_utriangular, NULL));
1894: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJGetArray_C", MatSeqSBAIJGetArray_SeqSBAIJ));
1895: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJRestoreArray_C", MatSeqSBAIJRestoreArray_SeqSBAIJ));
1896: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatStoreValues_C", MatStoreValues_SeqSBAIJ));
1897: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatRetrieveValues_C", MatRetrieveValues_SeqSBAIJ));
1898: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetColumnIndices_C", MatSeqSBAIJSetColumnIndices_SeqSBAIJ));
1899: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_seqaij_C", MatConvert_SeqSBAIJ_SeqAIJ));
1900: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_seqbaij_C", MatConvert_SeqSBAIJ_SeqBAIJ));
1901: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetPreallocation_C", MatSeqSBAIJSetPreallocation_SeqSBAIJ));
1902: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqSBAIJSetPreallocationCSR_C", MatSeqSBAIJSetPreallocationCSR_SeqSBAIJ));
1903: #if defined(PETSC_HAVE_ELEMENTAL)
1904: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_elemental_C", MatConvert_SeqSBAIJ_Elemental));
1905: #endif
1906: #if defined(PETSC_HAVE_SCALAPACK)
1907: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqsbaij_scalapack_C", MatConvert_SBAIJ_ScaLAPACK));
1908: #endif
1910: B->symmetry_eternal = PETSC_TRUE;
1911: B->structural_symmetry_eternal = PETSC_TRUE;
1912: B->symmetric = PETSC_BOOL3_TRUE;
1913: B->structurally_symmetric = PETSC_BOOL3_TRUE;
1914: #if defined(PETSC_USE_COMPLEX)
1915: B->hermitian = PETSC_BOOL3_FALSE;
1916: #else
1917: B->hermitian = PETSC_BOOL3_TRUE;
1918: #endif
1920: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATSEQSBAIJ));
1922: PetscOptionsBegin(PetscObjectComm((PetscObject)B), ((PetscObject)B)->prefix, "Options for SEQSBAIJ matrix", "Mat");
1923: PetscCall(PetscOptionsBool("-mat_no_unroll", "Do not optimize for inodes (slower)", NULL, no_unroll, &no_unroll, NULL));
1924: if (no_unroll) PetscCall(PetscInfo(B, "Not using Inode routines due to -mat_no_unroll\n"));
1925: PetscCall(PetscOptionsBool("-mat_no_inode", "Do not optimize for inodes (slower)", NULL, no_inode, &no_inode, NULL));
1926: if (no_inode) PetscCall(PetscInfo(B, "Not using Inode routines due to -mat_no_inode\n"));
1927: PetscCall(PetscOptionsInt("-mat_inode_limit", "Do not use inodes larger then this value", NULL, b->inode.limit, &b->inode.limit, NULL));
1928: PetscOptionsEnd();
1929: b->inode.use = (PetscBool)(!(no_unroll || no_inode));
1930: if (b->inode.limit > b->inode.max_limit) b->inode.limit = b->inode.max_limit;
1931: PetscFunctionReturn(PETSC_SUCCESS);
1932: }
1934: /*@C
1935: MatSeqSBAIJSetPreallocation - Creates a sparse symmetric matrix in block AIJ (block
1936: compressed row) `MATSEQSBAIJ` format. For good matrix assembly performance the
1937: user should preallocate the matrix storage by setting the parameter `nz`
1938: (or the array `nnz`).
1940: Collective
1942: Input Parameters:
1943: + B - the symmetric matrix
1944: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
1945: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with `MatCreateVecs()`
1946: . nz - number of block nonzeros per block row (same for all rows)
1947: - nnz - array containing the number of block nonzeros in the upper triangular plus
1948: diagonal portion of each block (possibly different for each block row) or `NULL`
1950: Options Database Keys:
1951: + -mat_no_unroll - uses code that does not unroll the loops in the block calculations (much slower)
1952: - -mat_block_size - size of the blocks to use (only works if a negative bs is passed in
1954: Level: intermediate
1956: Notes:
1957: Specify the preallocated storage with either `nz` or `nnz` (not both).
1958: Set `nz` = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
1959: allocation. See [Sparse Matrices](sec_matsparse) for details.
1961: You can call `MatGetInfo()` to get information on how effective the preallocation was;
1962: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
1963: You can also run with the option `-info` and look for messages with the string
1964: malloc in them to see if additional memory allocation was needed.
1966: If the `nnz` parameter is given then the `nz` parameter is ignored
1968: .seealso: [](ch_matrices), `Mat`, [Sparse Matrices](sec_matsparse), `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatCreateSBAIJ()`
1969: @*/
1970: PetscErrorCode MatSeqSBAIJSetPreallocation(Mat B, PetscInt bs, PetscInt nz, const PetscInt nnz[])
1971: {
1972: PetscFunctionBegin;
1976: PetscTryMethod(B, "MatSeqSBAIJSetPreallocation_C", (Mat, PetscInt, PetscInt, const PetscInt[]), (B, bs, nz, nnz));
1977: PetscFunctionReturn(PETSC_SUCCESS);
1978: }
1980: /*@C
1981: MatSeqSBAIJSetPreallocationCSR - Creates a sparse parallel matrix in `MATSEQSBAIJ` format using the given nonzero structure and (optional) numerical values
1983: Input Parameters:
1984: + B - the matrix
1985: . bs - size of block, the blocks are ALWAYS square.
1986: . i - the indices into `j` for the start of each local row (indices start with zero)
1987: . j - the column indices for each local row (indices start with zero) these must be sorted for each row
1988: - v - optional values in the matrix, use `NULL` if not provided
1990: Level: advanced
1992: Notes:
1993: The `i`,`j`,`v` values are COPIED with this routine; to avoid the copy use `MatCreateSeqSBAIJWithArrays()`
1995: The order of the entries in values is specified by the `MatOption` `MAT_ROW_ORIENTED`. For example, C programs
1996: may want to use the default `MAT_ROW_ORIENTED` = `PETSC_TRUE` and use an array v[nnz][bs][bs] where the second index is
1997: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
1998: `MAT_ROW_ORIENTED` = `PETSC_FALSE` and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
1999: block column and the second index is over columns within a block.
2001: Any entries provided that lie below the diagonal are ignored
2003: Though this routine has Preallocation() in the name it also sets the exact nonzero locations of the matrix entries
2004: and usually the numerical values as well
2006: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqSBAIJ()`, `MatSetValuesBlocked()`, `MatSeqSBAIJSetPreallocation()`
2007: @*/
2008: PetscErrorCode MatSeqSBAIJSetPreallocationCSR(Mat B, PetscInt bs, const PetscInt i[], const PetscInt j[], const PetscScalar v[])
2009: {
2010: PetscFunctionBegin;
2014: PetscTryMethod(B, "MatSeqSBAIJSetPreallocationCSR_C", (Mat, PetscInt, const PetscInt[], const PetscInt[], const PetscScalar[]), (B, bs, i, j, v));
2015: PetscFunctionReturn(PETSC_SUCCESS);
2016: }
2018: /*@C
2019: MatCreateSeqSBAIJ - Creates a sparse symmetric matrix in (block
2020: compressed row) `MATSEQSBAIJ` format. For good matrix assembly performance the
2021: user should preallocate the matrix storage by setting the parameter `nz`
2022: (or the array `nnz`).
2024: Collective
2026: Input Parameters:
2027: + comm - MPI communicator, set to `PETSC_COMM_SELF`
2028: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
2029: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2030: . m - number of rows
2031: . n - number of columns
2032: . nz - number of block nonzeros per block row (same for all rows)
2033: - nnz - array containing the number of block nonzeros in the upper triangular plus
2034: diagonal portion of each block (possibly different for each block row) or `NULL`
2036: Output Parameter:
2037: . A - the symmetric matrix
2039: Options Database Keys:
2040: + -mat_no_unroll - uses code that does not unroll the loops in the block calculations (much slower)
2041: - -mat_block_size - size of the blocks to use
2043: Level: intermediate
2045: Notes:
2046: It is recommended that one use `MatCreateFromOptions()` or the `MatCreate()`, `MatSetType()` and/or `MatSetFromOptions()`,
2047: MatXXXXSetPreallocation() paradigm instead of this routine directly.
2048: [MatXXXXSetPreallocation() is, for example, `MatSeqAIJSetPreallocation()`]
2050: The number of rows and columns must be divisible by blocksize.
2051: This matrix type does not support complex Hermitian operation.
2053: Specify the preallocated storage with either `nz` or `nnz` (not both).
2054: Set `nz` = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
2055: allocation. See [Sparse Matrices](sec_matsparse) for details.
2057: If the `nnz` parameter is given then the `nz` parameter is ignored
2059: .seealso: [](ch_matrices), `Mat`, [Sparse Matrices](sec_matsparse), `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatCreateSBAIJ()`
2060: @*/
2061: PetscErrorCode MatCreateSeqSBAIJ(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt nz, const PetscInt nnz[], Mat *A)
2062: {
2063: PetscFunctionBegin;
2064: PetscCall(MatCreate(comm, A));
2065: PetscCall(MatSetSizes(*A, m, n, m, n));
2066: PetscCall(MatSetType(*A, MATSEQSBAIJ));
2067: PetscCall(MatSeqSBAIJSetPreallocation(*A, bs, nz, (PetscInt *)nnz));
2068: PetscFunctionReturn(PETSC_SUCCESS);
2069: }
2071: PetscErrorCode MatDuplicate_SeqSBAIJ(Mat A, MatDuplicateOption cpvalues, Mat *B)
2072: {
2073: Mat C;
2074: Mat_SeqSBAIJ *c, *a = (Mat_SeqSBAIJ *)A->data;
2075: PetscInt i, mbs = a->mbs, nz = a->nz, bs2 = a->bs2;
2077: PetscFunctionBegin;
2078: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Cannot duplicate unassembled matrix");
2079: PetscCheck(a->i[mbs] == nz, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Corrupt matrix");
2081: *B = NULL;
2082: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &C));
2083: PetscCall(MatSetSizes(C, A->rmap->N, A->cmap->n, A->rmap->N, A->cmap->n));
2084: PetscCall(MatSetBlockSizesFromMats(C, A, A));
2085: PetscCall(MatSetType(C, MATSEQSBAIJ));
2086: c = (Mat_SeqSBAIJ *)C->data;
2088: C->preallocated = PETSC_TRUE;
2089: C->factortype = A->factortype;
2090: c->row = NULL;
2091: c->icol = NULL;
2092: c->saved_values = NULL;
2093: c->keepnonzeropattern = a->keepnonzeropattern;
2094: C->assembled = PETSC_TRUE;
2096: PetscCall(PetscLayoutReference(A->rmap, &C->rmap));
2097: PetscCall(PetscLayoutReference(A->cmap, &C->cmap));
2098: c->bs2 = a->bs2;
2099: c->mbs = a->mbs;
2100: c->nbs = a->nbs;
2102: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2103: c->imax = a->imax;
2104: c->ilen = a->ilen;
2105: c->free_imax_ilen = PETSC_FALSE;
2106: } else {
2107: PetscCall(PetscMalloc2((mbs + 1), &c->imax, (mbs + 1), &c->ilen));
2108: for (i = 0; i < mbs; i++) {
2109: c->imax[i] = a->imax[i];
2110: c->ilen[i] = a->ilen[i];
2111: }
2112: c->free_imax_ilen = PETSC_TRUE;
2113: }
2115: /* allocate the matrix space */
2116: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2117: PetscCall(PetscMalloc1(bs2 * nz, &c->a));
2118: c->i = a->i;
2119: c->j = a->j;
2120: c->singlemalloc = PETSC_FALSE;
2121: c->free_a = PETSC_TRUE;
2122: c->free_ij = PETSC_FALSE;
2123: c->parent = A;
2124: PetscCall(PetscObjectReference((PetscObject)A));
2125: PetscCall(MatSetOption(A, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2126: PetscCall(MatSetOption(C, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2127: } else {
2128: PetscCall(PetscMalloc3(bs2 * nz, &c->a, nz, &c->j, mbs + 1, &c->i));
2129: PetscCall(PetscArraycpy(c->i, a->i, mbs + 1));
2130: c->singlemalloc = PETSC_TRUE;
2131: c->free_a = PETSC_TRUE;
2132: c->free_ij = PETSC_TRUE;
2133: }
2134: if (mbs > 0) {
2135: if (cpvalues != MAT_SHARE_NONZERO_PATTERN) PetscCall(PetscArraycpy(c->j, a->j, nz));
2136: if (cpvalues == MAT_COPY_VALUES) {
2137: PetscCall(PetscArraycpy(c->a, a->a, bs2 * nz));
2138: } else {
2139: PetscCall(PetscArrayzero(c->a, bs2 * nz));
2140: }
2141: if (a->jshort) {
2142: /* cannot share jshort, it is reallocated in MatAssemblyEnd_SeqSBAIJ() */
2143: /* if the parent matrix is reassembled, this child matrix will never notice */
2144: PetscCall(PetscMalloc1(nz, &c->jshort));
2145: PetscCall(PetscArraycpy(c->jshort, a->jshort, nz));
2147: c->free_jshort = PETSC_TRUE;
2148: }
2149: }
2151: c->roworiented = a->roworiented;
2152: c->nonew = a->nonew;
2154: if (a->diag) {
2155: if (cpvalues == MAT_SHARE_NONZERO_PATTERN) {
2156: c->diag = a->diag;
2157: c->free_diag = PETSC_FALSE;
2158: } else {
2159: PetscCall(PetscMalloc1(mbs, &c->diag));
2160: for (i = 0; i < mbs; i++) c->diag[i] = a->diag[i];
2161: c->free_diag = PETSC_TRUE;
2162: }
2163: }
2164: c->nz = a->nz;
2165: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
2166: c->solve_work = NULL;
2167: c->mult_work = NULL;
2169: *B = C;
2170: PetscCall(PetscFunctionListDuplicate(((PetscObject)A)->qlist, &((PetscObject)C)->qlist));
2171: PetscFunctionReturn(PETSC_SUCCESS);
2172: }
2174: /* Used for both SeqBAIJ and SeqSBAIJ matrices */
2175: #define MatLoad_SeqSBAIJ_Binary MatLoad_SeqBAIJ_Binary
2177: PetscErrorCode MatLoad_SeqSBAIJ(Mat mat, PetscViewer viewer)
2178: {
2179: PetscBool isbinary;
2181: PetscFunctionBegin;
2182: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
2183: 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);
2184: PetscCall(MatLoad_SeqSBAIJ_Binary(mat, viewer));
2185: PetscFunctionReturn(PETSC_SUCCESS);
2186: }
2188: /*@
2189: MatCreateSeqSBAIJWithArrays - Creates an sequential `MATSEQSBAIJ` matrix using matrix elements
2190: (upper triangular entries in CSR format) provided by the user.
2192: Collective
2194: Input Parameters:
2195: + comm - must be an MPI communicator of size 1
2196: . bs - size of block
2197: . m - number of rows
2198: . n - number of columns
2199: . 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
2200: . j - column indices
2201: - a - matrix values
2203: Output Parameter:
2204: . mat - the matrix
2206: Level: advanced
2208: Notes:
2209: The `i`, `j`, and `a` arrays are not copied by this routine, the user must free these arrays
2210: once the matrix is destroyed
2212: You cannot set new nonzero locations into this matrix, that will generate an error.
2214: The `i` and `j` indices are 0 based
2216: When block size is greater than 1 the matrix values must be stored using the `MATSBAIJ` storage format. For block size of 1
2217: it is the regular CSR format excluding the lower triangular elements.
2219: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MatCreate()`, `MatCreateSBAIJ()`, `MatCreateSeqSBAIJ()`
2220: @*/
2221: PetscErrorCode MatCreateSeqSBAIJWithArrays(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt i[], PetscInt j[], PetscScalar a[], Mat *mat)
2222: {
2223: PetscInt ii;
2224: Mat_SeqSBAIJ *sbaij;
2226: PetscFunctionBegin;
2227: PetscCheck(bs == 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "block size %" PetscInt_FMT " > 1 is not supported yet", bs);
2228: PetscCheck(m == 0 || i[0] == 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "i (row indices) must start with 0");
2230: PetscCall(MatCreate(comm, mat));
2231: PetscCall(MatSetSizes(*mat, m, n, m, n));
2232: PetscCall(MatSetType(*mat, MATSEQSBAIJ));
2233: PetscCall(MatSeqSBAIJSetPreallocation(*mat, bs, MAT_SKIP_ALLOCATION, NULL));
2234: sbaij = (Mat_SeqSBAIJ *)(*mat)->data;
2235: PetscCall(PetscMalloc2(m, &sbaij->imax, m, &sbaij->ilen));
2237: sbaij->i = i;
2238: sbaij->j = j;
2239: sbaij->a = a;
2241: sbaij->singlemalloc = PETSC_FALSE;
2242: sbaij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
2243: sbaij->free_a = PETSC_FALSE;
2244: sbaij->free_ij = PETSC_FALSE;
2245: sbaij->free_imax_ilen = PETSC_TRUE;
2247: for (ii = 0; ii < m; ii++) {
2248: sbaij->ilen[ii] = sbaij->imax[ii] = i[ii + 1] - i[ii];
2249: 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]);
2250: }
2251: if (PetscDefined(USE_DEBUG)) {
2252: for (ii = 0; ii < sbaij->i[m]; ii++) {
2253: PetscCheck(j[ii] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative column index at location = %" PetscInt_FMT " index = %" PetscInt_FMT, ii, j[ii]);
2254: 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]);
2255: }
2256: }
2258: PetscCall(MatAssemblyBegin(*mat, MAT_FINAL_ASSEMBLY));
2259: PetscCall(MatAssemblyEnd(*mat, MAT_FINAL_ASSEMBLY));
2260: PetscFunctionReturn(PETSC_SUCCESS);
2261: }
2263: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqSBAIJ(MPI_Comm comm, Mat inmat, PetscInt n, MatReuse scall, Mat *outmat)
2264: {
2265: PetscFunctionBegin;
2266: PetscCall(MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(comm, inmat, n, scall, outmat));
2267: PetscFunctionReturn(PETSC_SUCCESS);
2268: }