Actual source code: mpibaij.c
1: #include <../src/mat/impls/baij/mpi/mpibaij.h>
3: #include <petsc/private/hashseti.h>
4: #include <petscblaslapack.h>
5: #include <petscsf.h>
7: static PetscErrorCode MatDestroy_MPIBAIJ(Mat mat)
8: {
9: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
11: PetscFunctionBegin;
12: PetscCall(PetscLogObjectState((PetscObject)mat, "Rows=%" PetscInt_FMT ",Cols=%" PetscInt_FMT, mat->rmap->N, mat->cmap->N));
13: PetscCall(MatStashDestroy_Private(&mat->stash));
14: PetscCall(MatStashDestroy_Private(&mat->bstash));
15: PetscCall(MatDestroy(&baij->A));
16: PetscCall(MatDestroy(&baij->B));
17: #if defined(PETSC_USE_CTABLE)
18: PetscCall(PetscHMapIDestroy(&baij->colmap));
19: #else
20: PetscCall(PetscFree(baij->colmap));
21: #endif
22: PetscCall(PetscFree(baij->garray));
23: PetscCall(VecDestroy(&baij->lvec));
24: PetscCall(VecScatterDestroy(&baij->Mvctx));
25: PetscCall(PetscFree2(baij->rowvalues, baij->rowindices));
26: PetscCall(PetscFree(baij->barray));
27: PetscCall(PetscFree2(baij->hd, baij->ht));
28: PetscCall(PetscFree(baij->rangebs));
29: PetscCall(PetscFree(mat->data));
31: PetscCall(PetscObjectChangeTypeName((PetscObject)mat, NULL));
32: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatStoreValues_C", NULL));
33: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatRetrieveValues_C", NULL));
34: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatMPIBAIJSetPreallocation_C", NULL));
35: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatMPIBAIJSetPreallocationCSR_C", NULL));
36: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatDiagonalScaleLocal_C", NULL));
37: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatSetHashTableFactor_C", NULL));
38: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpibaij_mpisbaij_C", NULL));
39: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpibaij_mpiadj_C", NULL));
40: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpibaij_mpiaij_C", NULL));
41: #if defined(PETSC_HAVE_HYPRE)
42: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpibaij_hypre_C", NULL));
43: #endif
44: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpibaij_is_C", NULL));
45: PetscFunctionReturn(PETSC_SUCCESS);
46: }
48: /* defines MatSetValues_MPI_Hash(), MatAssemblyBegin_MPI_Hash(), and MatAssemblyEnd_MPI_Hash() */
49: #define TYPE BAIJ
50: #include "../src/mat/impls/aij/mpi/mpihashmat.h"
51: #undef TYPE
53: #if defined(PETSC_HAVE_HYPRE)
54: PETSC_INTERN PetscErrorCode MatConvert_AIJ_HYPRE(Mat, MatType, MatReuse, Mat *);
55: #endif
57: static PetscErrorCode MatGetRowMaxAbs_MPIBAIJ(Mat A, Vec v, PetscInt idx[])
58: {
59: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
60: PetscInt i, *idxb = NULL, m = A->rmap->n, bs = A->cmap->bs;
61: PetscScalar *va, *vv;
62: Vec vB, vA;
63: const PetscScalar *vb;
65: PetscFunctionBegin;
66: PetscCall(VecCreateSeq(PETSC_COMM_SELF, m, &vA));
67: PetscCall(MatGetRowMaxAbs(a->A, vA, idx));
69: PetscCall(VecGetArrayWrite(vA, &va));
70: if (idx) {
71: for (i = 0; i < m; i++) {
72: if (PetscAbsScalar(va[i])) idx[i] += A->cmap->rstart;
73: }
74: }
76: PetscCall(VecCreateSeq(PETSC_COMM_SELF, m, &vB));
77: PetscCall(PetscMalloc1(m, &idxb));
78: PetscCall(MatGetRowMaxAbs(a->B, vB, idxb));
80: PetscCall(VecGetArrayWrite(v, &vv));
81: PetscCall(VecGetArrayRead(vB, &vb));
82: for (i = 0; i < m; i++) {
83: if (PetscAbsScalar(va[i]) < PetscAbsScalar(vb[i])) {
84: vv[i] = vb[i];
85: if (idx) idx[i] = bs * a->garray[idxb[i] / bs] + (idxb[i] % bs);
86: } else {
87: vv[i] = va[i];
88: if (idx && PetscAbsScalar(va[i]) == PetscAbsScalar(vb[i]) && idxb[i] != -1 && idx[i] > bs * a->garray[idxb[i] / bs] + (idxb[i] % bs)) idx[i] = bs * a->garray[idxb[i] / bs] + (idxb[i] % bs);
89: }
90: }
91: PetscCall(VecRestoreArrayWrite(vA, &vv));
92: PetscCall(VecRestoreArrayWrite(vA, &va));
93: PetscCall(VecRestoreArrayRead(vB, &vb));
94: PetscCall(PetscFree(idxb));
95: PetscCall(VecDestroy(&vA));
96: PetscCall(VecDestroy(&vB));
97: PetscFunctionReturn(PETSC_SUCCESS);
98: }
100: static PetscErrorCode MatStoreValues_MPIBAIJ(Mat mat)
101: {
102: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ *)mat->data;
104: PetscFunctionBegin;
105: PetscCall(MatStoreValues(aij->A));
106: PetscCall(MatStoreValues(aij->B));
107: PetscFunctionReturn(PETSC_SUCCESS);
108: }
110: static PetscErrorCode MatRetrieveValues_MPIBAIJ(Mat mat)
111: {
112: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ *)mat->data;
114: PetscFunctionBegin;
115: PetscCall(MatRetrieveValues(aij->A));
116: PetscCall(MatRetrieveValues(aij->B));
117: PetscFunctionReturn(PETSC_SUCCESS);
118: }
120: /*
121: Local utility routine that creates a mapping from the global column
122: number to the local number in the off-diagonal part of the local
123: storage of the matrix. This is done in a non scalable way since the
124: length of colmap equals the global matrix length.
125: */
126: PetscErrorCode MatCreateColmap_MPIBAIJ_Private(Mat mat)
127: {
128: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
129: Mat_SeqBAIJ *B = (Mat_SeqBAIJ *)baij->B->data;
130: PetscInt nbs = B->nbs, i, bs = mat->rmap->bs;
132: PetscFunctionBegin;
133: #if defined(PETSC_USE_CTABLE)
134: PetscCall(PetscHMapICreateWithSize(baij->nbs, &baij->colmap));
135: for (i = 0; i < nbs; i++) PetscCall(PetscHMapISet(baij->colmap, baij->garray[i] + 1, i * bs + 1));
136: #else
137: PetscCall(PetscCalloc1(baij->Nbs + 1, &baij->colmap));
138: for (i = 0; i < nbs; i++) baij->colmap[baij->garray[i]] = i * bs + 1;
139: #endif
140: PetscFunctionReturn(PETSC_SUCCESS);
141: }
143: #define MatSetValues_SeqBAIJ_A_Private(row, col, value, addv, orow, ocol) \
144: do { \
145: brow = row / bs; \
146: rp = aj + ai[brow]; \
147: ap = aa + bs2 * ai[brow]; \
148: rmax = aimax[brow]; \
149: nrow = ailen[brow]; \
150: bcol = col / bs; \
151: ridx = row % bs; \
152: cidx = col % bs; \
153: low = 0; \
154: high = nrow; \
155: while (high - low > 3) { \
156: t = (low + high) / 2; \
157: if (rp[t] > bcol) high = t; \
158: else low = t; \
159: } \
160: for (_i = low; _i < high; _i++) { \
161: if (rp[_i] > bcol) break; \
162: if (rp[_i] == bcol) { \
163: bap = ap + bs2 * _i + bs * cidx + ridx; \
164: if (addv == ADD_VALUES) *bap += value; \
165: else *bap = value; \
166: goto a_noinsert; \
167: } \
168: } \
169: if (a->nonew == 1) goto a_noinsert; \
170: PetscCheck(a->nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero at global row/column (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", orow, ocol); \
171: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, brow, bcol, rmax, aa, ai, aj, rp, ap, aimax, a->nonew, MatScalar); \
172: N = nrow++ - 1; \
173: /* shift up all the later entries in this row */ \
174: PetscCall(PetscArraymove(rp + _i + 1, rp + _i, N - _i + 1)); \
175: PetscCall(PetscArraymove(ap + bs2 * (_i + 1), ap + bs2 * _i, bs2 * (N - _i + 1))); \
176: PetscCall(PetscArrayzero(ap + bs2 * _i, bs2)); \
177: rp[_i] = bcol; \
178: ap[bs2 * _i + bs * cidx + ridx] = value; \
179: a_noinsert:; \
180: ailen[brow] = nrow; \
181: } while (0)
183: #define MatSetValues_SeqBAIJ_B_Private(row, col, value, addv, orow, ocol) \
184: do { \
185: brow = row / bs; \
186: rp = bj + bi[brow]; \
187: ap = ba + bs2 * bi[brow]; \
188: rmax = bimax[brow]; \
189: nrow = bilen[brow]; \
190: bcol = col / bs; \
191: ridx = row % bs; \
192: cidx = col % bs; \
193: low = 0; \
194: high = nrow; \
195: while (high - low > 3) { \
196: t = (low + high) / 2; \
197: if (rp[t] > bcol) high = t; \
198: else low = t; \
199: } \
200: for (_i = low; _i < high; _i++) { \
201: if (rp[_i] > bcol) break; \
202: if (rp[_i] == bcol) { \
203: bap = ap + bs2 * _i + bs * cidx + ridx; \
204: if (addv == ADD_VALUES) *bap += value; \
205: else *bap = value; \
206: goto b_noinsert; \
207: } \
208: } \
209: if (b->nonew == 1) goto b_noinsert; \
210: PetscCheck(b->nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero at global row/column (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", orow, ocol); \
211: MatSeqXAIJReallocateAIJ(B, b->mbs, bs2, nrow, brow, bcol, rmax, ba, bi, bj, rp, ap, bimax, b->nonew, MatScalar); \
212: N = nrow++ - 1; \
213: /* shift up all the later entries in this row */ \
214: PetscCall(PetscArraymove(rp + _i + 1, rp + _i, N - _i + 1)); \
215: PetscCall(PetscArraymove(ap + bs2 * (_i + 1), ap + bs2 * _i, bs2 * (N - _i + 1))); \
216: PetscCall(PetscArrayzero(ap + bs2 * _i, bs2)); \
217: rp[_i] = bcol; \
218: ap[bs2 * _i + bs * cidx + ridx] = value; \
219: b_noinsert:; \
220: bilen[brow] = nrow; \
221: } while (0)
223: PetscErrorCode MatSetValues_MPIBAIJ(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode addv)
224: {
225: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
226: MatScalar value;
227: PetscBool roworiented = baij->roworiented;
228: PetscInt i, j, row, col;
229: PetscInt rstart_orig = mat->rmap->rstart;
230: PetscInt rend_orig = mat->rmap->rend, cstart_orig = mat->cmap->rstart;
231: PetscInt cend_orig = mat->cmap->rend, bs = mat->rmap->bs;
233: /* Some Variables required in the macro */
234: Mat A = baij->A;
235: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)(A)->data;
236: PetscInt *aimax = a->imax, *ai = a->i, *ailen = a->ilen, *aj = a->j;
237: MatScalar *aa = a->a;
239: Mat B = baij->B;
240: Mat_SeqBAIJ *b = (Mat_SeqBAIJ *)(B)->data;
241: PetscInt *bimax = b->imax, *bi = b->i, *bilen = b->ilen, *bj = b->j;
242: MatScalar *ba = b->a;
244: PetscInt *rp, ii, nrow, _i, rmax, N, brow, bcol;
245: PetscInt low, high, t, ridx, cidx, bs2 = a->bs2;
246: MatScalar *ap, *bap;
248: PetscFunctionBegin;
249: for (i = 0; i < m; i++) {
250: if (im[i] < 0) continue;
251: PetscCheck(im[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, im[i], mat->rmap->N - 1);
252: if (im[i] >= rstart_orig && im[i] < rend_orig) {
253: row = im[i] - rstart_orig;
254: for (j = 0; j < n; j++) {
255: if (in[j] >= cstart_orig && in[j] < cend_orig) {
256: col = in[j] - cstart_orig;
257: if (roworiented) value = v[i * n + j];
258: else value = v[i + j * m];
259: MatSetValues_SeqBAIJ_A_Private(row, col, value, addv, im[i], in[j]);
260: } else if (in[j] < 0) {
261: continue;
262: } else {
263: PetscCheck(in[j] < mat->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[j], mat->cmap->N - 1);
264: if (mat->was_assembled) {
265: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
266: #if defined(PETSC_USE_CTABLE)
267: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] / bs + 1, 0, &col));
268: col = col - 1;
269: #else
270: col = baij->colmap[in[j] / bs] - 1;
271: #endif
272: if (col < 0 && !((Mat_SeqBAIJ *)(baij->B->data))->nonew) {
273: PetscCall(MatDisAssemble_MPIBAIJ(mat));
274: col = in[j];
275: /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */
276: B = baij->B;
277: b = (Mat_SeqBAIJ *)(B)->data;
278: bimax = b->imax;
279: bi = b->i;
280: bilen = b->ilen;
281: bj = b->j;
282: ba = b->a;
283: } else {
284: PetscCheck(col >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", im[i], in[j]);
285: col += in[j] % bs;
286: }
287: } else col = in[j];
288: if (roworiented) value = v[i * n + j];
289: else value = v[i + j * m];
290: MatSetValues_SeqBAIJ_B_Private(row, col, value, addv, im[i], in[j]);
291: /* PetscCall(MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv)); */
292: }
293: }
294: } else {
295: PetscCheck(!mat->nooffprocentries, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Setting off process row %" PetscInt_FMT " even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set", im[i]);
296: if (!baij->donotstash) {
297: mat->assembled = PETSC_FALSE;
298: if (roworiented) {
299: PetscCall(MatStashValuesRow_Private(&mat->stash, im[i], n, in, v + i * n, PETSC_FALSE));
300: } else {
301: PetscCall(MatStashValuesCol_Private(&mat->stash, im[i], n, in, v + i, m, PETSC_FALSE));
302: }
303: }
304: }
305: }
306: PetscFunctionReturn(PETSC_SUCCESS);
307: }
309: static inline PetscErrorCode MatSetValuesBlocked_SeqBAIJ_Inlined(Mat A, PetscInt row, PetscInt col, const PetscScalar v[], InsertMode is, PetscInt orow, PetscInt ocol)
310: {
311: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)A->data;
312: PetscInt *rp, low, high, t, ii, jj, nrow, i, rmax, N;
313: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
314: PetscInt *aj = a->j, nonew = a->nonew, bs2 = a->bs2, bs = A->rmap->bs;
315: PetscBool roworiented = a->roworiented;
316: const PetscScalar *value = v;
317: MatScalar *ap, *aa = a->a, *bap;
319: PetscFunctionBegin;
320: rp = aj + ai[row];
321: ap = aa + bs2 * ai[row];
322: rmax = imax[row];
323: nrow = ailen[row];
324: value = v;
325: low = 0;
326: high = nrow;
327: while (high - low > 7) {
328: t = (low + high) / 2;
329: if (rp[t] > col) high = t;
330: else low = t;
331: }
332: for (i = low; i < high; i++) {
333: if (rp[i] > col) break;
334: if (rp[i] == col) {
335: bap = ap + bs2 * i;
336: if (roworiented) {
337: if (is == ADD_VALUES) {
338: for (ii = 0; ii < bs; ii++) {
339: for (jj = ii; jj < bs2; jj += bs) bap[jj] += *value++;
340: }
341: } else {
342: for (ii = 0; ii < bs; ii++) {
343: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
344: }
345: }
346: } else {
347: if (is == ADD_VALUES) {
348: for (ii = 0; ii < bs; ii++, value += bs) {
349: for (jj = 0; jj < bs; jj++) bap[jj] += value[jj];
350: bap += bs;
351: }
352: } else {
353: for (ii = 0; ii < bs; ii++, value += bs) {
354: for (jj = 0; jj < bs; jj++) bap[jj] = value[jj];
355: bap += bs;
356: }
357: }
358: }
359: goto noinsert2;
360: }
361: }
362: if (nonew == 1) goto noinsert2;
363: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new global block indexed nonzero block (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", orow, ocol);
364: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
365: N = nrow++ - 1;
366: high++;
367: /* shift up all the later entries in this row */
368: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
369: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
370: rp[i] = col;
371: bap = ap + bs2 * i;
372: if (roworiented) {
373: for (ii = 0; ii < bs; ii++) {
374: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
375: }
376: } else {
377: for (ii = 0; ii < bs; ii++) {
378: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
379: }
380: }
381: noinsert2:;
382: ailen[row] = nrow;
383: PetscFunctionReturn(PETSC_SUCCESS);
384: }
386: /*
387: This routine should be optimized so that the block copy at ** Here a copy is required ** below is not needed
388: by passing additional stride information into the MatSetValuesBlocked_SeqBAIJ_Inlined() routine
389: */
390: static PetscErrorCode MatSetValuesBlocked_MPIBAIJ(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode addv)
391: {
392: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
393: const PetscScalar *value;
394: MatScalar *barray = baij->barray;
395: PetscBool roworiented = baij->roworiented;
396: PetscInt i, j, ii, jj, row, col, rstart = baij->rstartbs;
397: PetscInt rend = baij->rendbs, cstart = baij->cstartbs, stepval;
398: PetscInt cend = baij->cendbs, bs = mat->rmap->bs, bs2 = baij->bs2;
400: PetscFunctionBegin;
401: if (!barray) {
402: PetscCall(PetscMalloc1(bs2, &barray));
403: baij->barray = barray;
404: }
406: if (roworiented) stepval = (n - 1) * bs;
407: else stepval = (m - 1) * bs;
409: for (i = 0; i < m; i++) {
410: if (im[i] < 0) continue;
411: PetscCheck(im[i] < baij->Mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block indexed row too large %" PetscInt_FMT " max %" PetscInt_FMT, im[i], baij->Mbs - 1);
412: if (im[i] >= rstart && im[i] < rend) {
413: row = im[i] - rstart;
414: for (j = 0; j < n; j++) {
415: /* If NumCol = 1 then a copy is not required */
416: if ((roworiented) && (n == 1)) {
417: barray = (MatScalar *)v + i * bs2;
418: } else if ((!roworiented) && (m == 1)) {
419: barray = (MatScalar *)v + j * bs2;
420: } else { /* Here a copy is required */
421: if (roworiented) {
422: value = v + (i * (stepval + bs) + j) * bs;
423: } else {
424: value = v + (j * (stepval + bs) + i) * bs;
425: }
426: for (ii = 0; ii < bs; ii++, value += bs + stepval) {
427: for (jj = 0; jj < bs; jj++) barray[jj] = value[jj];
428: barray += bs;
429: }
430: barray -= bs2;
431: }
433: if (in[j] >= cstart && in[j] < cend) {
434: col = in[j] - cstart;
435: PetscCall(MatSetValuesBlocked_SeqBAIJ_Inlined(baij->A, row, col, barray, addv, im[i], in[j]));
436: } else if (in[j] < 0) {
437: continue;
438: } else {
439: PetscCheck(in[j] < baij->Nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block indexed column too large %" PetscInt_FMT " max %" PetscInt_FMT, in[j], baij->Nbs - 1);
440: if (mat->was_assembled) {
441: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
443: #if defined(PETSC_USE_DEBUG)
444: #if defined(PETSC_USE_CTABLE)
445: {
446: PetscInt data;
447: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] + 1, 0, &data));
448: PetscCheck((data - 1) % bs == 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Incorrect colmap");
449: }
450: #else
451: PetscCheck((baij->colmap[in[j]] - 1) % bs == 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Incorrect colmap");
452: #endif
453: #endif
454: #if defined(PETSC_USE_CTABLE)
455: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] + 1, 0, &col));
456: col = (col - 1) / bs;
457: #else
458: col = (baij->colmap[in[j]] - 1) / bs;
459: #endif
460: if (col < 0 && !((Mat_SeqBAIJ *)(baij->B->data))->nonew) {
461: PetscCall(MatDisAssemble_MPIBAIJ(mat));
462: col = in[j];
463: } else PetscCheck(col >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new blocked indexed nonzero block (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", im[i], in[j]);
464: } else col = in[j];
465: PetscCall(MatSetValuesBlocked_SeqBAIJ_Inlined(baij->B, row, col, barray, addv, im[i], in[j]));
466: }
467: }
468: } else {
469: PetscCheck(!mat->nooffprocentries, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Setting off process block indexed row %" PetscInt_FMT " even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set", im[i]);
470: if (!baij->donotstash) {
471: if (roworiented) {
472: PetscCall(MatStashValuesRowBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
473: } else {
474: PetscCall(MatStashValuesColBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
475: }
476: }
477: }
478: }
479: PetscFunctionReturn(PETSC_SUCCESS);
480: }
482: #define HASH_KEY 0.6180339887
483: #define HASH(size, key, tmp) (tmp = (key)*HASH_KEY, (PetscInt)((size) * (tmp - (PetscInt)tmp)))
484: /* #define HASH(size,key) ((PetscInt)((size)*fmod(((key)*HASH_KEY),1))) */
485: /* #define HASH(size,key,tmp) ((PetscInt)((size)*fmod(((key)*HASH_KEY),1))) */
486: static PetscErrorCode MatSetValues_MPIBAIJ_HT(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode addv)
487: {
488: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
489: PetscBool roworiented = baij->roworiented;
490: PetscInt i, j, row, col;
491: PetscInt rstart_orig = mat->rmap->rstart;
492: PetscInt rend_orig = mat->rmap->rend, Nbs = baij->Nbs;
493: PetscInt h1, key, size = baij->ht_size, bs = mat->rmap->bs, *HT = baij->ht, idx;
494: PetscReal tmp;
495: MatScalar **HD = baij->hd, value;
496: PetscInt total_ct = baij->ht_total_ct, insert_ct = baij->ht_insert_ct;
498: PetscFunctionBegin;
499: for (i = 0; i < m; i++) {
500: if (PetscDefined(USE_DEBUG)) {
501: PetscCheck(im[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative row");
502: PetscCheck(im[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, im[i], mat->rmap->N - 1);
503: }
504: row = im[i];
505: if (row >= rstart_orig && row < rend_orig) {
506: for (j = 0; j < n; j++) {
507: col = in[j];
508: if (roworiented) value = v[i * n + j];
509: else value = v[i + j * m];
510: /* Look up PetscInto the Hash Table */
511: key = (row / bs) * Nbs + (col / bs) + 1;
512: h1 = HASH(size, key, tmp);
514: idx = h1;
515: if (PetscDefined(USE_DEBUG)) {
516: insert_ct++;
517: total_ct++;
518: if (HT[idx] != key) {
519: for (idx = h1; (idx < size) && (HT[idx] != key); idx++, total_ct++)
520: ;
521: if (idx == size) {
522: for (idx = 0; (idx < h1) && (HT[idx] != key); idx++, total_ct++)
523: ;
524: PetscCheck(idx != h1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "(%" PetscInt_FMT ",%" PetscInt_FMT ") has no entry in the hash table", row, col);
525: }
526: }
527: } else if (HT[idx] != key) {
528: for (idx = h1; (idx < size) && (HT[idx] != key); idx++)
529: ;
530: if (idx == size) {
531: for (idx = 0; (idx < h1) && (HT[idx] != key); idx++)
532: ;
533: PetscCheck(idx != h1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "(%" PetscInt_FMT ",%" PetscInt_FMT ") has no entry in the hash table", row, col);
534: }
535: }
536: /* A HASH table entry is found, so insert the values at the correct address */
537: if (addv == ADD_VALUES) *(HD[idx] + (col % bs) * bs + (row % bs)) += value;
538: else *(HD[idx] + (col % bs) * bs + (row % bs)) = value;
539: }
540: } else if (!baij->donotstash) {
541: if (roworiented) {
542: PetscCall(MatStashValuesRow_Private(&mat->stash, im[i], n, in, v + i * n, PETSC_FALSE));
543: } else {
544: PetscCall(MatStashValuesCol_Private(&mat->stash, im[i], n, in, v + i, m, PETSC_FALSE));
545: }
546: }
547: }
548: if (PetscDefined(USE_DEBUG)) {
549: baij->ht_total_ct += total_ct;
550: baij->ht_insert_ct += insert_ct;
551: }
552: PetscFunctionReturn(PETSC_SUCCESS);
553: }
555: static PetscErrorCode MatSetValuesBlocked_MPIBAIJ_HT(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode addv)
556: {
557: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
558: PetscBool roworiented = baij->roworiented;
559: PetscInt i, j, ii, jj, row, col;
560: PetscInt rstart = baij->rstartbs;
561: PetscInt rend = mat->rmap->rend, stepval, bs = mat->rmap->bs, bs2 = baij->bs2, nbs2 = n * bs2;
562: PetscInt h1, key, size = baij->ht_size, idx, *HT = baij->ht, Nbs = baij->Nbs;
563: PetscReal tmp;
564: MatScalar **HD = baij->hd, *baij_a;
565: const PetscScalar *v_t, *value;
566: PetscInt total_ct = baij->ht_total_ct, insert_ct = baij->ht_insert_ct;
568: PetscFunctionBegin;
569: if (roworiented) stepval = (n - 1) * bs;
570: else stepval = (m - 1) * bs;
572: for (i = 0; i < m; i++) {
573: if (PetscDefined(USE_DEBUG)) {
574: PetscCheck(im[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative row: %" PetscInt_FMT, im[i]);
575: PetscCheck(im[i] < baij->Mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, im[i], baij->Mbs - 1);
576: }
577: row = im[i];
578: v_t = v + i * nbs2;
579: if (row >= rstart && row < rend) {
580: for (j = 0; j < n; j++) {
581: col = in[j];
583: /* Look up into the Hash Table */
584: key = row * Nbs + col + 1;
585: h1 = HASH(size, key, tmp);
587: idx = h1;
588: if (PetscDefined(USE_DEBUG)) {
589: total_ct++;
590: insert_ct++;
591: if (HT[idx] != key) {
592: for (idx = h1; (idx < size) && (HT[idx] != key); idx++, total_ct++)
593: ;
594: if (idx == size) {
595: for (idx = 0; (idx < h1) && (HT[idx] != key); idx++, total_ct++)
596: ;
597: PetscCheck(idx != h1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "(%" PetscInt_FMT ",%" PetscInt_FMT ") has no entry in the hash table", row, col);
598: }
599: }
600: } else if (HT[idx] != key) {
601: for (idx = h1; (idx < size) && (HT[idx] != key); idx++)
602: ;
603: if (idx == size) {
604: for (idx = 0; (idx < h1) && (HT[idx] != key); idx++)
605: ;
606: PetscCheck(idx != h1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "(%" PetscInt_FMT ",%" PetscInt_FMT ") has no entry in the hash table", row, col);
607: }
608: }
609: baij_a = HD[idx];
610: if (roworiented) {
611: /*value = v + i*(stepval+bs)*bs + j*bs;*/
612: /* value = v + (i*(stepval+bs)+j)*bs; */
613: value = v_t;
614: v_t += bs;
615: if (addv == ADD_VALUES) {
616: for (ii = 0; ii < bs; ii++, value += stepval) {
617: for (jj = ii; jj < bs2; jj += bs) baij_a[jj] += *value++;
618: }
619: } else {
620: for (ii = 0; ii < bs; ii++, value += stepval) {
621: for (jj = ii; jj < bs2; jj += bs) baij_a[jj] = *value++;
622: }
623: }
624: } else {
625: value = v + j * (stepval + bs) * bs + i * bs;
626: if (addv == ADD_VALUES) {
627: for (ii = 0; ii < bs; ii++, value += stepval, baij_a += bs) {
628: for (jj = 0; jj < bs; jj++) baij_a[jj] += *value++;
629: }
630: } else {
631: for (ii = 0; ii < bs; ii++, value += stepval, baij_a += bs) {
632: for (jj = 0; jj < bs; jj++) baij_a[jj] = *value++;
633: }
634: }
635: }
636: }
637: } else {
638: if (!baij->donotstash) {
639: if (roworiented) {
640: PetscCall(MatStashValuesRowBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
641: } else {
642: PetscCall(MatStashValuesColBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
643: }
644: }
645: }
646: }
647: if (PetscDefined(USE_DEBUG)) {
648: baij->ht_total_ct += total_ct;
649: baij->ht_insert_ct += insert_ct;
650: }
651: PetscFunctionReturn(PETSC_SUCCESS);
652: }
654: static PetscErrorCode MatGetValues_MPIBAIJ(Mat mat, PetscInt m, const PetscInt idxm[], PetscInt n, const PetscInt idxn[], PetscScalar v[])
655: {
656: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
657: PetscInt bs = mat->rmap->bs, i, j, bsrstart = mat->rmap->rstart, bsrend = mat->rmap->rend;
658: PetscInt bscstart = mat->cmap->rstart, bscend = mat->cmap->rend, row, col, data;
660: PetscFunctionBegin;
661: for (i = 0; i < m; i++) {
662: if (idxm[i] < 0) continue; /* negative row */
663: PetscCheck(idxm[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, idxm[i], mat->rmap->N - 1);
664: if (idxm[i] >= bsrstart && idxm[i] < bsrend) {
665: row = idxm[i] - bsrstart;
666: for (j = 0; j < n; j++) {
667: if (idxn[j] < 0) continue; /* negative column */
668: PetscCheck(idxn[j] < mat->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, idxn[j], mat->cmap->N - 1);
669: if (idxn[j] >= bscstart && idxn[j] < bscend) {
670: col = idxn[j] - bscstart;
671: PetscCall(MatGetValues_SeqBAIJ(baij->A, 1, &row, 1, &col, v + i * n + j));
672: } else {
673: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
674: #if defined(PETSC_USE_CTABLE)
675: PetscCall(PetscHMapIGetWithDefault(baij->colmap, idxn[j] / bs + 1, 0, &data));
676: data--;
677: #else
678: data = baij->colmap[idxn[j] / bs] - 1;
679: #endif
680: if ((data < 0) || (baij->garray[data / bs] != idxn[j] / bs)) *(v + i * n + j) = 0.0;
681: else {
682: col = data + idxn[j] % bs;
683: PetscCall(MatGetValues_SeqBAIJ(baij->B, 1, &row, 1, &col, v + i * n + j));
684: }
685: }
686: }
687: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Only local values currently supported");
688: }
689: PetscFunctionReturn(PETSC_SUCCESS);
690: }
692: static PetscErrorCode MatNorm_MPIBAIJ(Mat mat, NormType type, PetscReal *nrm)
693: {
694: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
695: Mat_SeqBAIJ *amat = (Mat_SeqBAIJ *)baij->A->data, *bmat = (Mat_SeqBAIJ *)baij->B->data;
696: PetscInt i, j, bs2 = baij->bs2, bs = baij->A->rmap->bs, nz, row, col;
697: PetscReal sum = 0.0;
698: MatScalar *v;
700: PetscFunctionBegin;
701: if (baij->size == 1) {
702: PetscCall(MatNorm(baij->A, type, nrm));
703: } else {
704: if (type == NORM_FROBENIUS) {
705: v = amat->a;
706: nz = amat->nz * bs2;
707: for (i = 0; i < nz; i++) {
708: sum += PetscRealPart(PetscConj(*v) * (*v));
709: v++;
710: }
711: v = bmat->a;
712: nz = bmat->nz * bs2;
713: for (i = 0; i < nz; i++) {
714: sum += PetscRealPart(PetscConj(*v) * (*v));
715: v++;
716: }
717: PetscCall(MPIU_Allreduce(&sum, nrm, 1, MPIU_REAL, MPIU_SUM, PetscObjectComm((PetscObject)mat)));
718: *nrm = PetscSqrtReal(*nrm);
719: } else if (type == NORM_1) { /* max column sum */
720: PetscReal *tmp, *tmp2;
721: PetscInt *jj, *garray = baij->garray, cstart = baij->rstartbs;
722: PetscCall(PetscCalloc1(mat->cmap->N, &tmp));
723: PetscCall(PetscMalloc1(mat->cmap->N, &tmp2));
724: v = amat->a;
725: jj = amat->j;
726: for (i = 0; i < amat->nz; i++) {
727: for (j = 0; j < bs; j++) {
728: col = bs * (cstart + *jj) + j; /* column index */
729: for (row = 0; row < bs; row++) {
730: tmp[col] += PetscAbsScalar(*v);
731: v++;
732: }
733: }
734: jj++;
735: }
736: v = bmat->a;
737: jj = bmat->j;
738: for (i = 0; i < bmat->nz; i++) {
739: for (j = 0; j < bs; j++) {
740: col = bs * garray[*jj] + j;
741: for (row = 0; row < bs; row++) {
742: tmp[col] += PetscAbsScalar(*v);
743: v++;
744: }
745: }
746: jj++;
747: }
748: PetscCall(MPIU_Allreduce(tmp, tmp2, mat->cmap->N, MPIU_REAL, MPIU_SUM, PetscObjectComm((PetscObject)mat)));
749: *nrm = 0.0;
750: for (j = 0; j < mat->cmap->N; j++) {
751: if (tmp2[j] > *nrm) *nrm = tmp2[j];
752: }
753: PetscCall(PetscFree(tmp));
754: PetscCall(PetscFree(tmp2));
755: } else if (type == NORM_INFINITY) { /* max row sum */
756: PetscReal *sums;
757: PetscCall(PetscMalloc1(bs, &sums));
758: sum = 0.0;
759: for (j = 0; j < amat->mbs; j++) {
760: for (row = 0; row < bs; row++) sums[row] = 0.0;
761: v = amat->a + bs2 * amat->i[j];
762: nz = amat->i[j + 1] - amat->i[j];
763: for (i = 0; i < nz; i++) {
764: for (col = 0; col < bs; col++) {
765: for (row = 0; row < bs; row++) {
766: sums[row] += PetscAbsScalar(*v);
767: v++;
768: }
769: }
770: }
771: v = bmat->a + bs2 * bmat->i[j];
772: nz = bmat->i[j + 1] - bmat->i[j];
773: for (i = 0; i < nz; i++) {
774: for (col = 0; col < bs; col++) {
775: for (row = 0; row < bs; row++) {
776: sums[row] += PetscAbsScalar(*v);
777: v++;
778: }
779: }
780: }
781: for (row = 0; row < bs; row++) {
782: if (sums[row] > sum) sum = sums[row];
783: }
784: }
785: PetscCall(MPIU_Allreduce(&sum, nrm, 1, MPIU_REAL, MPIU_MAX, PetscObjectComm((PetscObject)mat)));
786: PetscCall(PetscFree(sums));
787: } else SETERRQ(PetscObjectComm((PetscObject)mat), PETSC_ERR_SUP, "No support for this norm yet");
788: }
789: PetscFunctionReturn(PETSC_SUCCESS);
790: }
792: /*
793: Creates the hash table, and sets the table
794: This table is created only once.
795: If new entried need to be added to the matrix
796: then the hash table has to be destroyed and
797: recreated.
798: */
799: static PetscErrorCode MatCreateHashTable_MPIBAIJ_Private(Mat mat, PetscReal factor)
800: {
801: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
802: Mat A = baij->A, B = baij->B;
803: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)A->data, *b = (Mat_SeqBAIJ *)B->data;
804: PetscInt i, j, k, nz = a->nz + b->nz, h1, *ai = a->i, *aj = a->j, *bi = b->i, *bj = b->j;
805: PetscInt ht_size, bs2 = baij->bs2, rstart = baij->rstartbs;
806: PetscInt cstart = baij->cstartbs, *garray = baij->garray, row, col, Nbs = baij->Nbs;
807: PetscInt *HT, key;
808: MatScalar **HD;
809: PetscReal tmp;
810: #if defined(PETSC_USE_INFO)
811: PetscInt ct = 0, max = 0;
812: #endif
814: PetscFunctionBegin;
815: if (baij->ht) PetscFunctionReturn(PETSC_SUCCESS);
817: baij->ht_size = (PetscInt)(factor * nz);
818: ht_size = baij->ht_size;
820: /* Allocate Memory for Hash Table */
821: PetscCall(PetscCalloc2(ht_size, &baij->hd, ht_size, &baij->ht));
822: HD = baij->hd;
823: HT = baij->ht;
825: /* Loop Over A */
826: for (i = 0; i < a->mbs; i++) {
827: for (j = ai[i]; j < ai[i + 1]; j++) {
828: row = i + rstart;
829: col = aj[j] + cstart;
831: key = row * Nbs + col + 1;
832: h1 = HASH(ht_size, key, tmp);
833: for (k = 0; k < ht_size; k++) {
834: if (!HT[(h1 + k) % ht_size]) {
835: HT[(h1 + k) % ht_size] = key;
836: HD[(h1 + k) % ht_size] = a->a + j * bs2;
837: break;
838: #if defined(PETSC_USE_INFO)
839: } else {
840: ct++;
841: #endif
842: }
843: }
844: #if defined(PETSC_USE_INFO)
845: if (k > max) max = k;
846: #endif
847: }
848: }
849: /* Loop Over B */
850: for (i = 0; i < b->mbs; i++) {
851: for (j = bi[i]; j < bi[i + 1]; j++) {
852: row = i + rstart;
853: col = garray[bj[j]];
854: key = row * Nbs + col + 1;
855: h1 = HASH(ht_size, key, tmp);
856: for (k = 0; k < ht_size; k++) {
857: if (!HT[(h1 + k) % ht_size]) {
858: HT[(h1 + k) % ht_size] = key;
859: HD[(h1 + k) % ht_size] = b->a + j * bs2;
860: break;
861: #if defined(PETSC_USE_INFO)
862: } else {
863: ct++;
864: #endif
865: }
866: }
867: #if defined(PETSC_USE_INFO)
868: if (k > max) max = k;
869: #endif
870: }
871: }
873: /* Print Summary */
874: #if defined(PETSC_USE_INFO)
875: for (i = 0, j = 0; i < ht_size; i++) {
876: if (HT[i]) j++;
877: }
878: PetscCall(PetscInfo(mat, "Average Search = %5.2g,max search = %" PetscInt_FMT "\n", (!j) ? (double)0.0 : (double)(((PetscReal)(ct + j)) / (double)j), max));
879: #endif
880: PetscFunctionReturn(PETSC_SUCCESS);
881: }
883: static PetscErrorCode MatAssemblyBegin_MPIBAIJ(Mat mat, MatAssemblyType mode)
884: {
885: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
886: PetscInt nstash, reallocs;
888: PetscFunctionBegin;
889: if (baij->donotstash || mat->nooffprocentries) PetscFunctionReturn(PETSC_SUCCESS);
891: PetscCall(MatStashScatterBegin_Private(mat, &mat->stash, mat->rmap->range));
892: PetscCall(MatStashScatterBegin_Private(mat, &mat->bstash, baij->rangebs));
893: PetscCall(MatStashGetInfo_Private(&mat->stash, &nstash, &reallocs));
894: PetscCall(PetscInfo(mat, "Stash has %" PetscInt_FMT " entries,uses %" PetscInt_FMT " mallocs.\n", nstash, reallocs));
895: PetscCall(MatStashGetInfo_Private(&mat->bstash, &nstash, &reallocs));
896: PetscCall(PetscInfo(mat, "Block-Stash has %" PetscInt_FMT " entries, uses %" PetscInt_FMT " mallocs.\n", nstash, reallocs));
897: PetscFunctionReturn(PETSC_SUCCESS);
898: }
900: static PetscErrorCode MatAssemblyEnd_MPIBAIJ(Mat mat, MatAssemblyType mode)
901: {
902: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
903: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)baij->A->data;
904: PetscInt i, j, rstart, ncols, flg, bs2 = baij->bs2;
905: PetscInt *row, *col;
906: PetscBool r1, r2, r3, other_disassembled;
907: MatScalar *val;
908: PetscMPIInt n;
910: PetscFunctionBegin;
911: /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */
912: if (!baij->donotstash && !mat->nooffprocentries) {
913: while (1) {
914: PetscCall(MatStashScatterGetMesg_Private(&mat->stash, &n, &row, &col, &val, &flg));
915: if (!flg) break;
917: for (i = 0; i < n;) {
918: /* Now identify the consecutive vals belonging to the same row */
919: for (j = i, rstart = row[j]; j < n; j++) {
920: if (row[j] != rstart) break;
921: }
922: if (j < n) ncols = j - i;
923: else ncols = n - i;
924: /* Now assemble all these values with a single function call */
925: PetscCall(MatSetValues_MPIBAIJ(mat, 1, row + i, ncols, col + i, val + i, mat->insertmode));
926: i = j;
927: }
928: }
929: PetscCall(MatStashScatterEnd_Private(&mat->stash));
930: /* Now process the block-stash. Since the values are stashed column-oriented,
931: set the roworiented flag to column oriented, and after MatSetValues()
932: restore the original flags */
933: r1 = baij->roworiented;
934: r2 = a->roworiented;
935: r3 = ((Mat_SeqBAIJ *)baij->B->data)->roworiented;
937: baij->roworiented = PETSC_FALSE;
938: a->roworiented = PETSC_FALSE;
940: (((Mat_SeqBAIJ *)baij->B->data))->roworiented = PETSC_FALSE; /* b->roworiented */
941: while (1) {
942: PetscCall(MatStashScatterGetMesg_Private(&mat->bstash, &n, &row, &col, &val, &flg));
943: if (!flg) break;
945: for (i = 0; i < n;) {
946: /* Now identify the consecutive vals belonging to the same row */
947: for (j = i, rstart = row[j]; j < n; j++) {
948: if (row[j] != rstart) break;
949: }
950: if (j < n) ncols = j - i;
951: else ncols = n - i;
952: PetscCall(MatSetValuesBlocked_MPIBAIJ(mat, 1, row + i, ncols, col + i, val + i * bs2, mat->insertmode));
953: i = j;
954: }
955: }
956: PetscCall(MatStashScatterEnd_Private(&mat->bstash));
958: baij->roworiented = r1;
959: a->roworiented = r2;
961: ((Mat_SeqBAIJ *)baij->B->data)->roworiented = r3; /* b->roworiented */
962: }
964: PetscCall(MatAssemblyBegin(baij->A, mode));
965: PetscCall(MatAssemblyEnd(baij->A, mode));
967: /* determine if any processor has disassembled, if so we must
968: also disassemble ourselves, in order that we may reassemble. */
969: /*
970: if nonzero structure of submatrix B cannot change then we know that
971: no processor disassembled thus we can skip this stuff
972: */
973: if (!((Mat_SeqBAIJ *)baij->B->data)->nonew) {
974: PetscCall(MPIU_Allreduce(&mat->was_assembled, &other_disassembled, 1, MPIU_BOOL, MPI_LAND, PetscObjectComm((PetscObject)mat)));
975: if (mat->was_assembled && !other_disassembled) PetscCall(MatDisAssemble_MPIBAIJ(mat));
976: }
978: if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) PetscCall(MatSetUpMultiply_MPIBAIJ(mat));
979: PetscCall(MatAssemblyBegin(baij->B, mode));
980: PetscCall(MatAssemblyEnd(baij->B, mode));
982: #if defined(PETSC_USE_INFO)
983: if (baij->ht && mode == MAT_FINAL_ASSEMBLY) {
984: PetscCall(PetscInfo(mat, "Average Hash Table Search in MatSetValues = %5.2f\n", (double)((PetscReal)baij->ht_total_ct) / baij->ht_insert_ct));
986: baij->ht_total_ct = 0;
987: baij->ht_insert_ct = 0;
988: }
989: #endif
990: if (baij->ht_flag && !baij->ht && mode == MAT_FINAL_ASSEMBLY) {
991: PetscCall(MatCreateHashTable_MPIBAIJ_Private(mat, baij->ht_fact));
993: mat->ops->setvalues = MatSetValues_MPIBAIJ_HT;
994: mat->ops->setvaluesblocked = MatSetValuesBlocked_MPIBAIJ_HT;
995: }
997: PetscCall(PetscFree2(baij->rowvalues, baij->rowindices));
999: baij->rowvalues = NULL;
1001: /* if no new nonzero locations are allowed in matrix then only set the matrix state the first time through */
1002: if ((!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) || !((Mat_SeqBAIJ *)(baij->A->data))->nonew) {
1003: PetscObjectState state = baij->A->nonzerostate + baij->B->nonzerostate;
1004: PetscCall(MPIU_Allreduce(&state, &mat->nonzerostate, 1, MPIU_INT64, MPI_SUM, PetscObjectComm((PetscObject)mat)));
1005: }
1006: PetscFunctionReturn(PETSC_SUCCESS);
1007: }
1009: extern PetscErrorCode MatView_SeqBAIJ(Mat, PetscViewer);
1010: #include <petscdraw.h>
1011: static PetscErrorCode MatView_MPIBAIJ_ASCIIorDraworSocket(Mat mat, PetscViewer viewer)
1012: {
1013: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
1014: PetscMPIInt rank = baij->rank;
1015: PetscInt bs = mat->rmap->bs;
1016: PetscBool iascii, isdraw;
1017: PetscViewer sviewer;
1018: PetscViewerFormat format;
1020: PetscFunctionBegin;
1021: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
1022: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
1023: if (iascii) {
1024: PetscCall(PetscViewerGetFormat(viewer, &format));
1025: if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
1026: MatInfo info;
1027: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)mat), &rank));
1028: PetscCall(MatGetInfo(mat, MAT_LOCAL, &info));
1029: PetscCall(PetscViewerASCIIPushSynchronized(viewer));
1030: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] Local rows %" PetscInt_FMT " nz %" PetscInt_FMT " nz alloced %" PetscInt_FMT " bs %" PetscInt_FMT " mem %g\n", rank, mat->rmap->n, (PetscInt)info.nz_used, (PetscInt)info.nz_allocated,
1031: mat->rmap->bs, (double)info.memory));
1032: PetscCall(MatGetInfo(baij->A, MAT_LOCAL, &info));
1033: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] on-diagonal part: nz %" PetscInt_FMT " \n", rank, (PetscInt)info.nz_used));
1034: PetscCall(MatGetInfo(baij->B, MAT_LOCAL, &info));
1035: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] off-diagonal part: nz %" PetscInt_FMT " \n", rank, (PetscInt)info.nz_used));
1036: PetscCall(PetscViewerFlush(viewer));
1037: PetscCall(PetscViewerASCIIPopSynchronized(viewer));
1038: PetscCall(PetscViewerASCIIPrintf(viewer, "Information on VecScatter used in matrix-vector product: \n"));
1039: PetscCall(VecScatterView(baij->Mvctx, viewer));
1040: PetscFunctionReturn(PETSC_SUCCESS);
1041: } else if (format == PETSC_VIEWER_ASCII_INFO) {
1042: PetscCall(PetscViewerASCIIPrintf(viewer, " block size is %" PetscInt_FMT "\n", bs));
1043: PetscFunctionReturn(PETSC_SUCCESS);
1044: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
1045: PetscFunctionReturn(PETSC_SUCCESS);
1046: }
1047: }
1049: if (isdraw) {
1050: PetscDraw draw;
1051: PetscBool isnull;
1052: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
1053: PetscCall(PetscDrawIsNull(draw, &isnull));
1054: if (isnull) PetscFunctionReturn(PETSC_SUCCESS);
1055: }
1057: {
1058: /* assemble the entire matrix onto first processor. */
1059: Mat A;
1060: Mat_SeqBAIJ *Aloc;
1061: PetscInt M = mat->rmap->N, N = mat->cmap->N, *ai, *aj, col, i, j, k, *rvals, mbs = baij->mbs;
1062: MatScalar *a;
1063: const char *matname;
1065: /* Here we are creating a temporary matrix, so will assume MPIBAIJ is acceptable */
1066: /* Perhaps this should be the type of mat? */
1067: PetscCall(MatCreate(PetscObjectComm((PetscObject)mat), &A));
1068: if (rank == 0) {
1069: PetscCall(MatSetSizes(A, M, N, M, N));
1070: } else {
1071: PetscCall(MatSetSizes(A, 0, 0, M, N));
1072: }
1073: PetscCall(MatSetType(A, MATMPIBAIJ));
1074: PetscCall(MatMPIBAIJSetPreallocation(A, mat->rmap->bs, 0, NULL, 0, NULL));
1075: PetscCall(MatSetOption(A, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_FALSE));
1077: /* copy over the A part */
1078: Aloc = (Mat_SeqBAIJ *)baij->A->data;
1079: ai = Aloc->i;
1080: aj = Aloc->j;
1081: a = Aloc->a;
1082: PetscCall(PetscMalloc1(bs, &rvals));
1084: for (i = 0; i < mbs; i++) {
1085: rvals[0] = bs * (baij->rstartbs + i);
1086: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
1087: for (j = ai[i]; j < ai[i + 1]; j++) {
1088: col = (baij->cstartbs + aj[j]) * bs;
1089: for (k = 0; k < bs; k++) {
1090: PetscCall(MatSetValues_MPIBAIJ(A, bs, rvals, 1, &col, a, INSERT_VALUES));
1091: col++;
1092: a += bs;
1093: }
1094: }
1095: }
1096: /* copy over the B part */
1097: Aloc = (Mat_SeqBAIJ *)baij->B->data;
1098: ai = Aloc->i;
1099: aj = Aloc->j;
1100: a = Aloc->a;
1101: for (i = 0; i < mbs; i++) {
1102: rvals[0] = bs * (baij->rstartbs + i);
1103: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
1104: for (j = ai[i]; j < ai[i + 1]; j++) {
1105: col = baij->garray[aj[j]] * bs;
1106: for (k = 0; k < bs; k++) {
1107: PetscCall(MatSetValues_MPIBAIJ(A, bs, rvals, 1, &col, a, INSERT_VALUES));
1108: col++;
1109: a += bs;
1110: }
1111: }
1112: }
1113: PetscCall(PetscFree(rvals));
1114: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
1115: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
1116: /*
1117: Everyone has to call to draw the matrix since the graphics waits are
1118: synchronized across all processors that share the PetscDraw object
1119: */
1120: PetscCall(PetscViewerGetSubViewer(viewer, PETSC_COMM_SELF, &sviewer));
1121: if (((PetscObject)mat)->name) PetscCall(PetscObjectGetName((PetscObject)mat, &matname));
1122: if (rank == 0) {
1123: if (((PetscObject)mat)->name) PetscCall(PetscObjectSetName((PetscObject)((Mat_MPIBAIJ *)(A->data))->A, matname));
1124: PetscCall(MatView_SeqBAIJ(((Mat_MPIBAIJ *)(A->data))->A, sviewer));
1125: }
1126: PetscCall(PetscViewerRestoreSubViewer(viewer, PETSC_COMM_SELF, &sviewer));
1127: PetscCall(PetscViewerFlush(viewer));
1128: PetscCall(MatDestroy(&A));
1129: }
1130: PetscFunctionReturn(PETSC_SUCCESS);
1131: }
1133: /* Used for both MPIBAIJ and MPISBAIJ matrices */
1134: PetscErrorCode MatView_MPIBAIJ_Binary(Mat mat, PetscViewer viewer)
1135: {
1136: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ *)mat->data;
1137: Mat_SeqBAIJ *A = (Mat_SeqBAIJ *)aij->A->data;
1138: Mat_SeqBAIJ *B = (Mat_SeqBAIJ *)aij->B->data;
1139: const PetscInt *garray = aij->garray;
1140: PetscInt header[4], M, N, m, rs, cs, bs, cnt, i, j, ja, jb, k, l;
1141: PetscInt64 nz, hnz;
1142: PetscInt *rowlens, *colidxs;
1143: PetscScalar *matvals;
1144: PetscMPIInt rank;
1146: PetscFunctionBegin;
1147: PetscCall(PetscViewerSetUp(viewer));
1149: M = mat->rmap->N;
1150: N = mat->cmap->N;
1151: m = mat->rmap->n;
1152: rs = mat->rmap->rstart;
1153: cs = mat->cmap->rstart;
1154: bs = mat->rmap->bs;
1155: nz = bs * bs * (A->nz + B->nz);
1157: /* write matrix header */
1158: header[0] = MAT_FILE_CLASSID;
1159: header[1] = M;
1160: header[2] = N;
1161: PetscCallMPI(MPI_Reduce(&nz, &hnz, 1, MPIU_INT64, MPI_SUM, 0, PetscObjectComm((PetscObject)mat)));
1162: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)mat), &rank));
1163: if (rank == 0) PetscCall(PetscIntCast(hnz, &header[3]));
1164: PetscCall(PetscViewerBinaryWrite(viewer, header, 4, PETSC_INT));
1166: /* fill in and store row lengths */
1167: PetscCall(PetscMalloc1(m, &rowlens));
1168: for (cnt = 0, i = 0; i < A->mbs; i++)
1169: for (j = 0; j < bs; j++) rowlens[cnt++] = bs * (A->i[i + 1] - A->i[i] + B->i[i + 1] - B->i[i]);
1170: PetscCall(PetscViewerBinaryWriteAll(viewer, rowlens, m, rs, M, PETSC_INT));
1171: PetscCall(PetscFree(rowlens));
1173: /* fill in and store column indices */
1174: PetscCall(PetscMalloc1(nz, &colidxs));
1175: for (cnt = 0, i = 0; i < A->mbs; i++) {
1176: for (k = 0; k < bs; k++) {
1177: for (jb = B->i[i]; jb < B->i[i + 1]; jb++) {
1178: if (garray[B->j[jb]] > cs / bs) break;
1179: for (l = 0; l < bs; l++) colidxs[cnt++] = bs * garray[B->j[jb]] + l;
1180: }
1181: for (ja = A->i[i]; ja < A->i[i + 1]; ja++)
1182: for (l = 0; l < bs; l++) colidxs[cnt++] = bs * A->j[ja] + l + cs;
1183: for (; jb < B->i[i + 1]; jb++)
1184: for (l = 0; l < bs; l++) colidxs[cnt++] = bs * garray[B->j[jb]] + l;
1185: }
1186: }
1187: PetscCheck(cnt == nz, PETSC_COMM_SELF, PETSC_ERR_LIB, "Internal PETSc error: cnt = %" PetscInt_FMT " nz = %" PetscInt64_FMT, cnt, nz);
1188: PetscCall(PetscViewerBinaryWriteAll(viewer, colidxs, nz, PETSC_DECIDE, PETSC_DECIDE, PETSC_INT));
1189: PetscCall(PetscFree(colidxs));
1191: /* fill in and store nonzero values */
1192: PetscCall(PetscMalloc1(nz, &matvals));
1193: for (cnt = 0, i = 0; i < A->mbs; i++) {
1194: for (k = 0; k < bs; k++) {
1195: for (jb = B->i[i]; jb < B->i[i + 1]; jb++) {
1196: if (garray[B->j[jb]] > cs / bs) break;
1197: for (l = 0; l < bs; l++) matvals[cnt++] = B->a[bs * (bs * jb + l) + k];
1198: }
1199: for (ja = A->i[i]; ja < A->i[i + 1]; ja++)
1200: for (l = 0; l < bs; l++) matvals[cnt++] = A->a[bs * (bs * ja + l) + k];
1201: for (; jb < B->i[i + 1]; jb++)
1202: for (l = 0; l < bs; l++) matvals[cnt++] = B->a[bs * (bs * jb + l) + k];
1203: }
1204: }
1205: PetscCall(PetscViewerBinaryWriteAll(viewer, matvals, nz, PETSC_DECIDE, PETSC_DECIDE, PETSC_SCALAR));
1206: PetscCall(PetscFree(matvals));
1208: /* write block size option to the viewer's .info file */
1209: PetscCall(MatView_Binary_BlockSizes(mat, viewer));
1210: PetscFunctionReturn(PETSC_SUCCESS);
1211: }
1213: PetscErrorCode MatView_MPIBAIJ(Mat mat, PetscViewer viewer)
1214: {
1215: PetscBool iascii, isdraw, issocket, isbinary;
1217: PetscFunctionBegin;
1218: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
1219: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
1220: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSOCKET, &issocket));
1221: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
1222: if (iascii || isdraw || issocket) {
1223: PetscCall(MatView_MPIBAIJ_ASCIIorDraworSocket(mat, viewer));
1224: } else if (isbinary) PetscCall(MatView_MPIBAIJ_Binary(mat, viewer));
1225: PetscFunctionReturn(PETSC_SUCCESS);
1226: }
1228: static PetscErrorCode MatMult_MPIBAIJ(Mat A, Vec xx, Vec yy)
1229: {
1230: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1231: PetscInt nt;
1233: PetscFunctionBegin;
1234: PetscCall(VecGetLocalSize(xx, &nt));
1235: PetscCheck(nt == A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible partition of A and xx");
1236: PetscCall(VecGetLocalSize(yy, &nt));
1237: PetscCheck(nt == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Incompatible partition of A and yy");
1238: PetscCall(VecScatterBegin(a->Mvctx, xx, a->lvec, INSERT_VALUES, SCATTER_FORWARD));
1239: PetscCall((*a->A->ops->mult)(a->A, xx, yy));
1240: PetscCall(VecScatterEnd(a->Mvctx, xx, a->lvec, INSERT_VALUES, SCATTER_FORWARD));
1241: PetscCall((*a->B->ops->multadd)(a->B, a->lvec, yy, yy));
1242: PetscFunctionReturn(PETSC_SUCCESS);
1243: }
1245: static PetscErrorCode MatMultAdd_MPIBAIJ(Mat A, Vec xx, Vec yy, Vec zz)
1246: {
1247: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1249: PetscFunctionBegin;
1250: PetscCall(VecScatterBegin(a->Mvctx, xx, a->lvec, INSERT_VALUES, SCATTER_FORWARD));
1251: PetscCall((*a->A->ops->multadd)(a->A, xx, yy, zz));
1252: PetscCall(VecScatterEnd(a->Mvctx, xx, a->lvec, INSERT_VALUES, SCATTER_FORWARD));
1253: PetscCall((*a->B->ops->multadd)(a->B, a->lvec, zz, zz));
1254: PetscFunctionReturn(PETSC_SUCCESS);
1255: }
1257: static PetscErrorCode MatMultTranspose_MPIBAIJ(Mat A, Vec xx, Vec yy)
1258: {
1259: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1261: PetscFunctionBegin;
1262: /* do nondiagonal part */
1263: PetscCall((*a->B->ops->multtranspose)(a->B, xx, a->lvec));
1264: /* do local part */
1265: PetscCall((*a->A->ops->multtranspose)(a->A, xx, yy));
1266: /* add partial results together */
1267: PetscCall(VecScatterBegin(a->Mvctx, a->lvec, yy, ADD_VALUES, SCATTER_REVERSE));
1268: PetscCall(VecScatterEnd(a->Mvctx, a->lvec, yy, ADD_VALUES, SCATTER_REVERSE));
1269: PetscFunctionReturn(PETSC_SUCCESS);
1270: }
1272: static PetscErrorCode MatMultTransposeAdd_MPIBAIJ(Mat A, Vec xx, Vec yy, Vec zz)
1273: {
1274: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1276: PetscFunctionBegin;
1277: /* do nondiagonal part */
1278: PetscCall((*a->B->ops->multtranspose)(a->B, xx, a->lvec));
1279: /* do local part */
1280: PetscCall((*a->A->ops->multtransposeadd)(a->A, xx, yy, zz));
1281: /* add partial results together */
1282: PetscCall(VecScatterBegin(a->Mvctx, a->lvec, zz, ADD_VALUES, SCATTER_REVERSE));
1283: PetscCall(VecScatterEnd(a->Mvctx, a->lvec, zz, ADD_VALUES, SCATTER_REVERSE));
1284: PetscFunctionReturn(PETSC_SUCCESS);
1285: }
1287: /*
1288: This only works correctly for square matrices where the subblock A->A is the
1289: diagonal block
1290: */
1291: static PetscErrorCode MatGetDiagonal_MPIBAIJ(Mat A, Vec v)
1292: {
1293: PetscFunctionBegin;
1294: PetscCheck(A->rmap->N == A->cmap->N, PETSC_COMM_SELF, PETSC_ERR_SUP, "Supports only square matrix where A->A is diag block");
1295: PetscCall(MatGetDiagonal(((Mat_MPIBAIJ *)A->data)->A, v));
1296: PetscFunctionReturn(PETSC_SUCCESS);
1297: }
1299: static PetscErrorCode MatScale_MPIBAIJ(Mat A, PetscScalar aa)
1300: {
1301: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1303: PetscFunctionBegin;
1304: PetscCall(MatScale(a->A, aa));
1305: PetscCall(MatScale(a->B, aa));
1306: PetscFunctionReturn(PETSC_SUCCESS);
1307: }
1309: PetscErrorCode MatGetRow_MPIBAIJ(Mat matin, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
1310: {
1311: Mat_MPIBAIJ *mat = (Mat_MPIBAIJ *)matin->data;
1312: PetscScalar *vworkA, *vworkB, **pvA, **pvB, *v_p;
1313: PetscInt bs = matin->rmap->bs, bs2 = mat->bs2, i, *cworkA, *cworkB, **pcA, **pcB;
1314: PetscInt nztot, nzA, nzB, lrow, brstart = matin->rmap->rstart, brend = matin->rmap->rend;
1315: PetscInt *cmap, *idx_p, cstart = mat->cstartbs;
1317: PetscFunctionBegin;
1318: PetscCheck(row >= brstart && row < brend, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only local rows");
1319: PetscCheck(!mat->getrowactive, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Already active");
1320: mat->getrowactive = PETSC_TRUE;
1322: if (!mat->rowvalues && (idx || v)) {
1323: /*
1324: allocate enough space to hold information from the longest row.
1325: */
1326: Mat_SeqBAIJ *Aa = (Mat_SeqBAIJ *)mat->A->data, *Ba = (Mat_SeqBAIJ *)mat->B->data;
1327: PetscInt max = 1, mbs = mat->mbs, tmp;
1328: for (i = 0; i < mbs; i++) {
1329: tmp = Aa->i[i + 1] - Aa->i[i] + Ba->i[i + 1] - Ba->i[i];
1330: if (max < tmp) max = tmp;
1331: }
1332: PetscCall(PetscMalloc2(max * bs2, &mat->rowvalues, max * bs2, &mat->rowindices));
1333: }
1334: lrow = row - brstart;
1336: pvA = &vworkA;
1337: pcA = &cworkA;
1338: pvB = &vworkB;
1339: pcB = &cworkB;
1340: if (!v) {
1341: pvA = NULL;
1342: pvB = NULL;
1343: }
1344: if (!idx) {
1345: pcA = NULL;
1346: if (!v) pcB = NULL;
1347: }
1348: PetscCall((*mat->A->ops->getrow)(mat->A, lrow, &nzA, pcA, pvA));
1349: PetscCall((*mat->B->ops->getrow)(mat->B, lrow, &nzB, pcB, pvB));
1350: nztot = nzA + nzB;
1352: cmap = mat->garray;
1353: if (v || idx) {
1354: if (nztot) {
1355: /* Sort by increasing column numbers, assuming A and B already sorted */
1356: PetscInt imark = -1;
1357: if (v) {
1358: *v = v_p = mat->rowvalues;
1359: for (i = 0; i < nzB; i++) {
1360: if (cmap[cworkB[i] / bs] < cstart) v_p[i] = vworkB[i];
1361: else break;
1362: }
1363: imark = i;
1364: for (i = 0; i < nzA; i++) v_p[imark + i] = vworkA[i];
1365: for (i = imark; i < nzB; i++) v_p[nzA + i] = vworkB[i];
1366: }
1367: if (idx) {
1368: *idx = idx_p = mat->rowindices;
1369: if (imark > -1) {
1370: for (i = 0; i < imark; i++) idx_p[i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1371: } else {
1372: for (i = 0; i < nzB; i++) {
1373: if (cmap[cworkB[i] / bs] < cstart) idx_p[i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1374: else break;
1375: }
1376: imark = i;
1377: }
1378: for (i = 0; i < nzA; i++) idx_p[imark + i] = cstart * bs + cworkA[i];
1379: for (i = imark; i < nzB; i++) idx_p[nzA + i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1380: }
1381: } else {
1382: if (idx) *idx = NULL;
1383: if (v) *v = NULL;
1384: }
1385: }
1386: *nz = nztot;
1387: PetscCall((*mat->A->ops->restorerow)(mat->A, lrow, &nzA, pcA, pvA));
1388: PetscCall((*mat->B->ops->restorerow)(mat->B, lrow, &nzB, pcB, pvB));
1389: PetscFunctionReturn(PETSC_SUCCESS);
1390: }
1392: PetscErrorCode MatRestoreRow_MPIBAIJ(Mat mat, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
1393: {
1394: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
1396: PetscFunctionBegin;
1397: PetscCheck(baij->getrowactive, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "MatGetRow not called");
1398: baij->getrowactive = PETSC_FALSE;
1399: PetscFunctionReturn(PETSC_SUCCESS);
1400: }
1402: static PetscErrorCode MatZeroEntries_MPIBAIJ(Mat A)
1403: {
1404: Mat_MPIBAIJ *l = (Mat_MPIBAIJ *)A->data;
1406: PetscFunctionBegin;
1407: PetscCall(MatZeroEntries(l->A));
1408: PetscCall(MatZeroEntries(l->B));
1409: PetscFunctionReturn(PETSC_SUCCESS);
1410: }
1412: static PetscErrorCode MatGetInfo_MPIBAIJ(Mat matin, MatInfoType flag, MatInfo *info)
1413: {
1414: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)matin->data;
1415: Mat A = a->A, B = a->B;
1416: PetscLogDouble isend[5], irecv[5];
1418: PetscFunctionBegin;
1419: info->block_size = (PetscReal)matin->rmap->bs;
1421: PetscCall(MatGetInfo(A, MAT_LOCAL, info));
1423: isend[0] = info->nz_used;
1424: isend[1] = info->nz_allocated;
1425: isend[2] = info->nz_unneeded;
1426: isend[3] = info->memory;
1427: isend[4] = info->mallocs;
1429: PetscCall(MatGetInfo(B, MAT_LOCAL, info));
1431: isend[0] += info->nz_used;
1432: isend[1] += info->nz_allocated;
1433: isend[2] += info->nz_unneeded;
1434: isend[3] += info->memory;
1435: isend[4] += info->mallocs;
1437: if (flag == MAT_LOCAL) {
1438: info->nz_used = isend[0];
1439: info->nz_allocated = isend[1];
1440: info->nz_unneeded = isend[2];
1441: info->memory = isend[3];
1442: info->mallocs = isend[4];
1443: } else if (flag == MAT_GLOBAL_MAX) {
1444: PetscCall(MPIU_Allreduce(isend, irecv, 5, MPIU_PETSCLOGDOUBLE, MPI_MAX, PetscObjectComm((PetscObject)matin)));
1446: info->nz_used = irecv[0];
1447: info->nz_allocated = irecv[1];
1448: info->nz_unneeded = irecv[2];
1449: info->memory = irecv[3];
1450: info->mallocs = irecv[4];
1451: } else if (flag == MAT_GLOBAL_SUM) {
1452: PetscCall(MPIU_Allreduce(isend, irecv, 5, MPIU_PETSCLOGDOUBLE, MPI_SUM, PetscObjectComm((PetscObject)matin)));
1454: info->nz_used = irecv[0];
1455: info->nz_allocated = irecv[1];
1456: info->nz_unneeded = irecv[2];
1457: info->memory = irecv[3];
1458: info->mallocs = irecv[4];
1459: } else SETERRQ(PetscObjectComm((PetscObject)matin), PETSC_ERR_ARG_WRONG, "Unknown MatInfoType argument %d", (int)flag);
1460: info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */
1461: info->fill_ratio_needed = 0;
1462: info->factor_mallocs = 0;
1463: PetscFunctionReturn(PETSC_SUCCESS);
1464: }
1466: static PetscErrorCode MatSetOption_MPIBAIJ(Mat A, MatOption op, PetscBool flg)
1467: {
1468: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1470: PetscFunctionBegin;
1471: switch (op) {
1472: case MAT_NEW_NONZERO_LOCATIONS:
1473: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1474: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1475: case MAT_KEEP_NONZERO_PATTERN:
1476: case MAT_NEW_NONZERO_LOCATION_ERR:
1477: MatCheckPreallocated(A, 1);
1478: PetscCall(MatSetOption(a->A, op, flg));
1479: PetscCall(MatSetOption(a->B, op, flg));
1480: break;
1481: case MAT_ROW_ORIENTED:
1482: MatCheckPreallocated(A, 1);
1483: a->roworiented = flg;
1485: PetscCall(MatSetOption(a->A, op, flg));
1486: PetscCall(MatSetOption(a->B, op, flg));
1487: break;
1488: case MAT_FORCE_DIAGONAL_ENTRIES:
1489: case MAT_SORTED_FULL:
1490: PetscCall(PetscInfo(A, "Option %s ignored\n", MatOptions[op]));
1491: break;
1492: case MAT_IGNORE_OFF_PROC_ENTRIES:
1493: a->donotstash = flg;
1494: break;
1495: case MAT_USE_HASH_TABLE:
1496: a->ht_flag = flg;
1497: a->ht_fact = 1.39;
1498: break;
1499: case MAT_SYMMETRIC:
1500: case MAT_STRUCTURALLY_SYMMETRIC:
1501: case MAT_HERMITIAN:
1502: case MAT_SUBMAT_SINGLEIS:
1503: case MAT_SYMMETRY_ETERNAL:
1504: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
1505: case MAT_SPD_ETERNAL:
1506: /* if the diagonal matrix is square it inherits some of the properties above */
1507: break;
1508: default:
1509: SETERRQ(PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "unknown option %d", op);
1510: }
1511: PetscFunctionReturn(PETSC_SUCCESS);
1512: }
1514: static PetscErrorCode MatTranspose_MPIBAIJ(Mat A, MatReuse reuse, Mat *matout)
1515: {
1516: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)A->data;
1517: Mat_SeqBAIJ *Aloc;
1518: Mat B;
1519: PetscInt M = A->rmap->N, N = A->cmap->N, *ai, *aj, i, *rvals, j, k, col;
1520: PetscInt bs = A->rmap->bs, mbs = baij->mbs;
1521: MatScalar *a;
1523: PetscFunctionBegin;
1524: if (reuse == MAT_REUSE_MATRIX) PetscCall(MatTransposeCheckNonzeroState_Private(A, *matout));
1525: if (reuse == MAT_INITIAL_MATRIX || reuse == MAT_INPLACE_MATRIX) {
1526: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &B));
1527: PetscCall(MatSetSizes(B, A->cmap->n, A->rmap->n, N, M));
1528: PetscCall(MatSetType(B, ((PetscObject)A)->type_name));
1529: /* Do not know preallocation information, but must set block size */
1530: PetscCall(MatMPIBAIJSetPreallocation(B, A->rmap->bs, PETSC_DECIDE, NULL, PETSC_DECIDE, NULL));
1531: } else {
1532: B = *matout;
1533: }
1535: /* copy over the A part */
1536: Aloc = (Mat_SeqBAIJ *)baij->A->data;
1537: ai = Aloc->i;
1538: aj = Aloc->j;
1539: a = Aloc->a;
1540: PetscCall(PetscMalloc1(bs, &rvals));
1542: for (i = 0; i < mbs; i++) {
1543: rvals[0] = bs * (baij->rstartbs + i);
1544: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
1545: for (j = ai[i]; j < ai[i + 1]; j++) {
1546: col = (baij->cstartbs + aj[j]) * bs;
1547: for (k = 0; k < bs; k++) {
1548: PetscCall(MatSetValues_MPIBAIJ(B, 1, &col, bs, rvals, a, INSERT_VALUES));
1550: col++;
1551: a += bs;
1552: }
1553: }
1554: }
1555: /* copy over the B part */
1556: Aloc = (Mat_SeqBAIJ *)baij->B->data;
1557: ai = Aloc->i;
1558: aj = Aloc->j;
1559: a = Aloc->a;
1560: for (i = 0; i < mbs; i++) {
1561: rvals[0] = bs * (baij->rstartbs + i);
1562: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
1563: for (j = ai[i]; j < ai[i + 1]; j++) {
1564: col = baij->garray[aj[j]] * bs;
1565: for (k = 0; k < bs; k++) {
1566: PetscCall(MatSetValues_MPIBAIJ(B, 1, &col, bs, rvals, a, INSERT_VALUES));
1567: col++;
1568: a += bs;
1569: }
1570: }
1571: }
1572: PetscCall(PetscFree(rvals));
1573: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
1574: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
1576: if (reuse == MAT_INITIAL_MATRIX || reuse == MAT_REUSE_MATRIX) *matout = B;
1577: else PetscCall(MatHeaderMerge(A, &B));
1578: PetscFunctionReturn(PETSC_SUCCESS);
1579: }
1581: static PetscErrorCode MatDiagonalScale_MPIBAIJ(Mat mat, Vec ll, Vec rr)
1582: {
1583: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
1584: Mat a = baij->A, b = baij->B;
1585: PetscInt s1, s2, s3;
1587: PetscFunctionBegin;
1588: PetscCall(MatGetLocalSize(mat, &s2, &s3));
1589: if (rr) {
1590: PetscCall(VecGetLocalSize(rr, &s1));
1591: PetscCheck(s1 == s3, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "right vector non-conforming local size");
1592: /* Overlap communication with computation. */
1593: PetscCall(VecScatterBegin(baij->Mvctx, rr, baij->lvec, INSERT_VALUES, SCATTER_FORWARD));
1594: }
1595: if (ll) {
1596: PetscCall(VecGetLocalSize(ll, &s1));
1597: PetscCheck(s1 == s2, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "left vector non-conforming local size");
1598: PetscUseTypeMethod(b, diagonalscale, ll, NULL);
1599: }
1600: /* scale the diagonal block */
1601: PetscUseTypeMethod(a, diagonalscale, ll, rr);
1603: if (rr) {
1604: /* Do a scatter end and then right scale the off-diagonal block */
1605: PetscCall(VecScatterEnd(baij->Mvctx, rr, baij->lvec, INSERT_VALUES, SCATTER_FORWARD));
1606: PetscUseTypeMethod(b, diagonalscale, NULL, baij->lvec);
1607: }
1608: PetscFunctionReturn(PETSC_SUCCESS);
1609: }
1611: static PetscErrorCode MatZeroRows_MPIBAIJ(Mat A, PetscInt N, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
1612: {
1613: Mat_MPIBAIJ *l = (Mat_MPIBAIJ *)A->data;
1614: PetscInt *lrows;
1615: PetscInt r, len;
1616: PetscBool cong;
1618: PetscFunctionBegin;
1619: /* get locally owned rows */
1620: PetscCall(MatZeroRowsMapLocal_Private(A, N, rows, &len, &lrows));
1621: /* fix right hand side if needed */
1622: if (x && b) {
1623: const PetscScalar *xx;
1624: PetscScalar *bb;
1626: PetscCall(VecGetArrayRead(x, &xx));
1627: PetscCall(VecGetArray(b, &bb));
1628: for (r = 0; r < len; ++r) bb[lrows[r]] = diag * xx[lrows[r]];
1629: PetscCall(VecRestoreArrayRead(x, &xx));
1630: PetscCall(VecRestoreArray(b, &bb));
1631: }
1633: /* actually zap the local rows */
1634: /*
1635: Zero the required rows. If the "diagonal block" of the matrix
1636: is square and the user wishes to set the diagonal we use separate
1637: code so that MatSetValues() is not called for each diagonal allocating
1638: new memory, thus calling lots of mallocs and slowing things down.
1640: */
1641: /* must zero l->B before l->A because the (diag) case below may put values into l->B*/
1642: PetscCall(MatZeroRows_SeqBAIJ(l->B, len, lrows, 0.0, NULL, NULL));
1643: PetscCall(MatHasCongruentLayouts(A, &cong));
1644: if ((diag != 0.0) && cong) {
1645: PetscCall(MatZeroRows_SeqBAIJ(l->A, len, lrows, diag, NULL, NULL));
1646: } else if (diag != 0.0) {
1647: PetscCall(MatZeroRows_SeqBAIJ(l->A, len, lrows, 0.0, NULL, NULL));
1648: PetscCheck(!((Mat_SeqBAIJ *)l->A->data)->nonew, PETSC_COMM_SELF, PETSC_ERR_SUP, "MatZeroRows() on rectangular matrices cannot be used with the Mat options \n\
1649: MAT_NEW_NONZERO_LOCATIONS,MAT_NEW_NONZERO_LOCATION_ERR,MAT_NEW_NONZERO_ALLOCATION_ERR");
1650: for (r = 0; r < len; ++r) {
1651: const PetscInt row = lrows[r] + A->rmap->rstart;
1652: PetscCall(MatSetValues(A, 1, &row, 1, &row, &diag, INSERT_VALUES));
1653: }
1654: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
1655: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
1656: } else {
1657: PetscCall(MatZeroRows_SeqBAIJ(l->A, len, lrows, 0.0, NULL, NULL));
1658: }
1659: PetscCall(PetscFree(lrows));
1661: /* only change matrix nonzero state if pattern was allowed to be changed */
1662: if (!((Mat_SeqBAIJ *)(l->A->data))->keepnonzeropattern) {
1663: PetscObjectState state = l->A->nonzerostate + l->B->nonzerostate;
1664: PetscCall(MPIU_Allreduce(&state, &A->nonzerostate, 1, MPIU_INT64, MPI_SUM, PetscObjectComm((PetscObject)A)));
1665: }
1666: PetscFunctionReturn(PETSC_SUCCESS);
1667: }
1669: static PetscErrorCode MatZeroRowsColumns_MPIBAIJ(Mat A, PetscInt N, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
1670: {
1671: Mat_MPIBAIJ *l = (Mat_MPIBAIJ *)A->data;
1672: PetscMPIInt n = A->rmap->n, p = 0;
1673: PetscInt i, j, k, r, len = 0, row, col, count;
1674: PetscInt *lrows, *owners = A->rmap->range;
1675: PetscSFNode *rrows;
1676: PetscSF sf;
1677: const PetscScalar *xx;
1678: PetscScalar *bb, *mask;
1679: Vec xmask, lmask;
1680: Mat_SeqBAIJ *baij = (Mat_SeqBAIJ *)l->B->data;
1681: PetscInt bs = A->rmap->bs, bs2 = baij->bs2;
1682: PetscScalar *aa;
1684: PetscFunctionBegin;
1685: /* Create SF where leaves are input rows and roots are owned rows */
1686: PetscCall(PetscMalloc1(n, &lrows));
1687: for (r = 0; r < n; ++r) lrows[r] = -1;
1688: PetscCall(PetscMalloc1(N, &rrows));
1689: for (r = 0; r < N; ++r) {
1690: const PetscInt idx = rows[r];
1691: PetscCheck(idx >= 0 && A->rmap->N > idx, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row %" PetscInt_FMT " out of range [0,%" PetscInt_FMT ")", idx, A->rmap->N);
1692: if (idx < owners[p] || owners[p + 1] <= idx) { /* short-circuit the search if the last p owns this row too */
1693: PetscCall(PetscLayoutFindOwner(A->rmap, idx, &p));
1694: }
1695: rrows[r].rank = p;
1696: rrows[r].index = rows[r] - owners[p];
1697: }
1698: PetscCall(PetscSFCreate(PetscObjectComm((PetscObject)A), &sf));
1699: PetscCall(PetscSFSetGraph(sf, n, N, NULL, PETSC_OWN_POINTER, rrows, PETSC_OWN_POINTER));
1700: /* Collect flags for rows to be zeroed */
1701: PetscCall(PetscSFReduceBegin(sf, MPIU_INT, (PetscInt *)rows, lrows, MPI_LOR));
1702: PetscCall(PetscSFReduceEnd(sf, MPIU_INT, (PetscInt *)rows, lrows, MPI_LOR));
1703: PetscCall(PetscSFDestroy(&sf));
1704: /* Compress and put in row numbers */
1705: for (r = 0; r < n; ++r)
1706: if (lrows[r] >= 0) lrows[len++] = r;
1707: /* zero diagonal part of matrix */
1708: PetscCall(MatZeroRowsColumns(l->A, len, lrows, diag, x, b));
1709: /* handle off-diagonal part of matrix */
1710: PetscCall(MatCreateVecs(A, &xmask, NULL));
1711: PetscCall(VecDuplicate(l->lvec, &lmask));
1712: PetscCall(VecGetArray(xmask, &bb));
1713: for (i = 0; i < len; i++) bb[lrows[i]] = 1;
1714: PetscCall(VecRestoreArray(xmask, &bb));
1715: PetscCall(VecScatterBegin(l->Mvctx, xmask, lmask, ADD_VALUES, SCATTER_FORWARD));
1716: PetscCall(VecScatterEnd(l->Mvctx, xmask, lmask, ADD_VALUES, SCATTER_FORWARD));
1717: PetscCall(VecDestroy(&xmask));
1718: if (x) {
1719: PetscCall(VecScatterBegin(l->Mvctx, x, l->lvec, INSERT_VALUES, SCATTER_FORWARD));
1720: PetscCall(VecScatterEnd(l->Mvctx, x, l->lvec, INSERT_VALUES, SCATTER_FORWARD));
1721: PetscCall(VecGetArrayRead(l->lvec, &xx));
1722: PetscCall(VecGetArray(b, &bb));
1723: }
1724: PetscCall(VecGetArray(lmask, &mask));
1725: /* remove zeroed rows of off-diagonal matrix */
1726: for (i = 0; i < len; ++i) {
1727: row = lrows[i];
1728: count = (baij->i[row / bs + 1] - baij->i[row / bs]) * bs;
1729: aa = ((MatScalar *)(baij->a)) + baij->i[row / bs] * bs2 + (row % bs);
1730: for (k = 0; k < count; ++k) {
1731: aa[0] = 0.0;
1732: aa += bs;
1733: }
1734: }
1735: /* loop over all elements of off process part of matrix zeroing removed columns*/
1736: for (i = 0; i < l->B->rmap->N; ++i) {
1737: row = i / bs;
1738: for (j = baij->i[row]; j < baij->i[row + 1]; ++j) {
1739: for (k = 0; k < bs; ++k) {
1740: col = bs * baij->j[j] + k;
1741: if (PetscAbsScalar(mask[col])) {
1742: aa = ((MatScalar *)(baij->a)) + j * bs2 + (i % bs) + bs * k;
1743: if (x) bb[i] -= aa[0] * xx[col];
1744: aa[0] = 0.0;
1745: }
1746: }
1747: }
1748: }
1749: if (x) {
1750: PetscCall(VecRestoreArray(b, &bb));
1751: PetscCall(VecRestoreArrayRead(l->lvec, &xx));
1752: }
1753: PetscCall(VecRestoreArray(lmask, &mask));
1754: PetscCall(VecDestroy(&lmask));
1755: PetscCall(PetscFree(lrows));
1757: /* only change matrix nonzero state if pattern was allowed to be changed */
1758: if (!((Mat_SeqBAIJ *)(l->A->data))->keepnonzeropattern) {
1759: PetscObjectState state = l->A->nonzerostate + l->B->nonzerostate;
1760: PetscCall(MPIU_Allreduce(&state, &A->nonzerostate, 1, MPIU_INT64, MPI_SUM, PetscObjectComm((PetscObject)A)));
1761: }
1762: PetscFunctionReturn(PETSC_SUCCESS);
1763: }
1765: static PetscErrorCode MatSetUnfactored_MPIBAIJ(Mat A)
1766: {
1767: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1769: PetscFunctionBegin;
1770: PetscCall(MatSetUnfactored(a->A));
1771: PetscFunctionReturn(PETSC_SUCCESS);
1772: }
1774: static PetscErrorCode MatDuplicate_MPIBAIJ(Mat, MatDuplicateOption, Mat *);
1776: static PetscErrorCode MatEqual_MPIBAIJ(Mat A, Mat B, PetscBool *flag)
1777: {
1778: Mat_MPIBAIJ *matB = (Mat_MPIBAIJ *)B->data, *matA = (Mat_MPIBAIJ *)A->data;
1779: Mat a, b, c, d;
1780: PetscBool flg;
1782: PetscFunctionBegin;
1783: a = matA->A;
1784: b = matA->B;
1785: c = matB->A;
1786: d = matB->B;
1788: PetscCall(MatEqual(a, c, &flg));
1789: if (flg) PetscCall(MatEqual(b, d, &flg));
1790: PetscCall(MPIU_Allreduce(&flg, flag, 1, MPIU_BOOL, MPI_LAND, PetscObjectComm((PetscObject)A)));
1791: PetscFunctionReturn(PETSC_SUCCESS);
1792: }
1794: static PetscErrorCode MatCopy_MPIBAIJ(Mat A, Mat B, MatStructure str)
1795: {
1796: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1797: Mat_MPIBAIJ *b = (Mat_MPIBAIJ *)B->data;
1799: PetscFunctionBegin;
1800: /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */
1801: if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) {
1802: PetscCall(MatCopy_Basic(A, B, str));
1803: } else {
1804: PetscCall(MatCopy(a->A, b->A, str));
1805: PetscCall(MatCopy(a->B, b->B, str));
1806: }
1807: PetscCall(PetscObjectStateIncrease((PetscObject)B));
1808: PetscFunctionReturn(PETSC_SUCCESS);
1809: }
1811: PetscErrorCode MatAXPYGetPreallocation_MPIBAIJ(Mat Y, const PetscInt *yltog, Mat X, const PetscInt *xltog, PetscInt *nnz)
1812: {
1813: PetscInt bs = Y->rmap->bs, m = Y->rmap->N / bs;
1814: Mat_SeqBAIJ *x = (Mat_SeqBAIJ *)X->data;
1815: Mat_SeqBAIJ *y = (Mat_SeqBAIJ *)Y->data;
1817: PetscFunctionBegin;
1818: PetscCall(MatAXPYGetPreallocation_MPIX_private(m, x->i, x->j, xltog, y->i, y->j, yltog, nnz));
1819: PetscFunctionReturn(PETSC_SUCCESS);
1820: }
1822: static PetscErrorCode MatAXPY_MPIBAIJ(Mat Y, PetscScalar a, Mat X, MatStructure str)
1823: {
1824: Mat_MPIBAIJ *xx = (Mat_MPIBAIJ *)X->data, *yy = (Mat_MPIBAIJ *)Y->data;
1825: PetscBLASInt bnz, one = 1;
1826: Mat_SeqBAIJ *x, *y;
1827: PetscInt bs2 = Y->rmap->bs * Y->rmap->bs;
1829: PetscFunctionBegin;
1830: if (str == SAME_NONZERO_PATTERN) {
1831: PetscScalar alpha = a;
1832: x = (Mat_SeqBAIJ *)xx->A->data;
1833: y = (Mat_SeqBAIJ *)yy->A->data;
1834: PetscCall(PetscBLASIntCast(x->nz * bs2, &bnz));
1835: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, x->a, &one, y->a, &one));
1836: x = (Mat_SeqBAIJ *)xx->B->data;
1837: y = (Mat_SeqBAIJ *)yy->B->data;
1838: PetscCall(PetscBLASIntCast(x->nz * bs2, &bnz));
1839: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, x->a, &one, y->a, &one));
1840: PetscCall(PetscObjectStateIncrease((PetscObject)Y));
1841: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1842: PetscCall(MatAXPY_Basic(Y, a, X, str));
1843: } else {
1844: Mat B;
1845: PetscInt *nnz_d, *nnz_o, bs = Y->rmap->bs;
1846: PetscCall(PetscMalloc1(yy->A->rmap->N, &nnz_d));
1847: PetscCall(PetscMalloc1(yy->B->rmap->N, &nnz_o));
1848: PetscCall(MatCreate(PetscObjectComm((PetscObject)Y), &B));
1849: PetscCall(PetscObjectSetName((PetscObject)B, ((PetscObject)Y)->name));
1850: PetscCall(MatSetSizes(B, Y->rmap->n, Y->cmap->n, Y->rmap->N, Y->cmap->N));
1851: PetscCall(MatSetBlockSizesFromMats(B, Y, Y));
1852: PetscCall(MatSetType(B, MATMPIBAIJ));
1853: PetscCall(MatAXPYGetPreallocation_SeqBAIJ(yy->A, xx->A, nnz_d));
1854: PetscCall(MatAXPYGetPreallocation_MPIBAIJ(yy->B, yy->garray, xx->B, xx->garray, nnz_o));
1855: PetscCall(MatMPIBAIJSetPreallocation(B, bs, 0, nnz_d, 0, nnz_o));
1856: /* MatAXPY_BasicWithPreallocation() for BAIJ matrix is much slower than AIJ, even for bs=1 ! */
1857: PetscCall(MatAXPY_BasicWithPreallocation(B, Y, a, X, str));
1858: PetscCall(MatHeaderMerge(Y, &B));
1859: PetscCall(PetscFree(nnz_d));
1860: PetscCall(PetscFree(nnz_o));
1861: }
1862: PetscFunctionReturn(PETSC_SUCCESS);
1863: }
1865: static PetscErrorCode MatConjugate_MPIBAIJ(Mat mat)
1866: {
1867: PetscFunctionBegin;
1868: if (PetscDefined(USE_COMPLEX)) {
1869: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)mat->data;
1871: PetscCall(MatConjugate_SeqBAIJ(a->A));
1872: PetscCall(MatConjugate_SeqBAIJ(a->B));
1873: }
1874: PetscFunctionReturn(PETSC_SUCCESS);
1875: }
1877: static PetscErrorCode MatRealPart_MPIBAIJ(Mat A)
1878: {
1879: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1881: PetscFunctionBegin;
1882: PetscCall(MatRealPart(a->A));
1883: PetscCall(MatRealPart(a->B));
1884: PetscFunctionReturn(PETSC_SUCCESS);
1885: }
1887: static PetscErrorCode MatImaginaryPart_MPIBAIJ(Mat A)
1888: {
1889: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
1891: PetscFunctionBegin;
1892: PetscCall(MatImaginaryPart(a->A));
1893: PetscCall(MatImaginaryPart(a->B));
1894: PetscFunctionReturn(PETSC_SUCCESS);
1895: }
1897: static PetscErrorCode MatCreateSubMatrix_MPIBAIJ(Mat mat, IS isrow, IS iscol, MatReuse call, Mat *newmat)
1898: {
1899: IS iscol_local;
1900: PetscInt csize;
1902: PetscFunctionBegin;
1903: PetscCall(ISGetLocalSize(iscol, &csize));
1904: if (call == MAT_REUSE_MATRIX) {
1905: PetscCall(PetscObjectQuery((PetscObject)*newmat, "ISAllGather", (PetscObject *)&iscol_local));
1906: PetscCheck(iscol_local, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Submatrix passed in was not used before, cannot reuse");
1907: } else {
1908: PetscCall(ISAllGather(iscol, &iscol_local));
1909: }
1910: PetscCall(MatCreateSubMatrix_MPIBAIJ_Private(mat, isrow, iscol_local, csize, call, newmat, PETSC_FALSE));
1911: if (call == MAT_INITIAL_MATRIX) {
1912: PetscCall(PetscObjectCompose((PetscObject)*newmat, "ISAllGather", (PetscObject)iscol_local));
1913: PetscCall(ISDestroy(&iscol_local));
1914: }
1915: PetscFunctionReturn(PETSC_SUCCESS);
1916: }
1918: /*
1919: Not great since it makes two copies of the submatrix, first an SeqBAIJ
1920: in local and then by concatenating the local matrices the end result.
1921: Writing it directly would be much like MatCreateSubMatrices_MPIBAIJ().
1922: This routine is used for BAIJ and SBAIJ matrices (unfortunate dependency).
1923: */
1924: PetscErrorCode MatCreateSubMatrix_MPIBAIJ_Private(Mat mat, IS isrow, IS iscol, PetscInt csize, MatReuse call, Mat *newmat, PetscBool sym)
1925: {
1926: PetscMPIInt rank, size;
1927: PetscInt i, m, n, rstart, row, rend, nz, *cwork, j, bs;
1928: PetscInt *ii, *jj, nlocal, *dlens, *olens, dlen, olen, jend, mglobal;
1929: Mat M, Mreuse;
1930: MatScalar *vwork, *aa;
1931: MPI_Comm comm;
1932: IS isrow_new, iscol_new;
1933: Mat_SeqBAIJ *aij;
1935: PetscFunctionBegin;
1936: PetscCall(PetscObjectGetComm((PetscObject)mat, &comm));
1937: PetscCallMPI(MPI_Comm_rank(comm, &rank));
1938: PetscCallMPI(MPI_Comm_size(comm, &size));
1939: /* The compression and expansion should be avoided. Doesn't point
1940: out errors, might change the indices, hence buggey */
1941: PetscCall(ISCompressIndicesGeneral(mat->rmap->N, mat->rmap->n, mat->rmap->bs, 1, &isrow, &isrow_new));
1942: if (isrow == iscol) {
1943: iscol_new = isrow_new;
1944: PetscCall(PetscObjectReference((PetscObject)iscol_new));
1945: } else PetscCall(ISCompressIndicesGeneral(mat->cmap->N, mat->cmap->n, mat->cmap->bs, 1, &iscol, &iscol_new));
1947: if (call == MAT_REUSE_MATRIX) {
1948: PetscCall(PetscObjectQuery((PetscObject)*newmat, "SubMatrix", (PetscObject *)&Mreuse));
1949: PetscCheck(Mreuse, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Submatrix passed in was not used before, cannot reuse");
1950: PetscCall(MatCreateSubMatrices_MPIBAIJ_local(mat, 1, &isrow_new, &iscol_new, MAT_REUSE_MATRIX, &Mreuse, sym));
1951: } else {
1952: PetscCall(MatCreateSubMatrices_MPIBAIJ_local(mat, 1, &isrow_new, &iscol_new, MAT_INITIAL_MATRIX, &Mreuse, sym));
1953: }
1954: PetscCall(ISDestroy(&isrow_new));
1955: PetscCall(ISDestroy(&iscol_new));
1956: /*
1957: m - number of local rows
1958: n - number of columns (same on all processors)
1959: rstart - first row in new global matrix generated
1960: */
1961: PetscCall(MatGetBlockSize(mat, &bs));
1962: PetscCall(MatGetSize(Mreuse, &m, &n));
1963: m = m / bs;
1964: n = n / bs;
1966: if (call == MAT_INITIAL_MATRIX) {
1967: aij = (Mat_SeqBAIJ *)(Mreuse)->data;
1968: ii = aij->i;
1969: jj = aij->j;
1971: /*
1972: Determine the number of non-zeros in the diagonal and off-diagonal
1973: portions of the matrix in order to do correct preallocation
1974: */
1976: /* first get start and end of "diagonal" columns */
1977: if (csize == PETSC_DECIDE) {
1978: PetscCall(ISGetSize(isrow, &mglobal));
1979: if (mglobal == n * bs) { /* square matrix */
1980: nlocal = m;
1981: } else {
1982: nlocal = n / size + ((n % size) > rank);
1983: }
1984: } else {
1985: nlocal = csize / bs;
1986: }
1987: PetscCallMPI(MPI_Scan(&nlocal, &rend, 1, MPIU_INT, MPI_SUM, comm));
1988: rstart = rend - nlocal;
1989: PetscCheck(rank != size - 1 || rend == n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Local column sizes %" PetscInt_FMT " do not add up to total number of columns %" PetscInt_FMT, rend, n);
1991: /* next, compute all the lengths */
1992: PetscCall(PetscMalloc2(m + 1, &dlens, m + 1, &olens));
1993: for (i = 0; i < m; i++) {
1994: jend = ii[i + 1] - ii[i];
1995: olen = 0;
1996: dlen = 0;
1997: for (j = 0; j < jend; j++) {
1998: if (*jj < rstart || *jj >= rend) olen++;
1999: else dlen++;
2000: jj++;
2001: }
2002: olens[i] = olen;
2003: dlens[i] = dlen;
2004: }
2005: PetscCall(MatCreate(comm, &M));
2006: PetscCall(MatSetSizes(M, bs * m, bs * nlocal, PETSC_DECIDE, bs * n));
2007: PetscCall(MatSetType(M, sym ? ((PetscObject)mat)->type_name : MATMPIBAIJ));
2008: PetscCall(MatMPIBAIJSetPreallocation(M, bs, 0, dlens, 0, olens));
2009: PetscCall(MatMPISBAIJSetPreallocation(M, bs, 0, dlens, 0, olens));
2010: PetscCall(PetscFree2(dlens, olens));
2011: } else {
2012: PetscInt ml, nl;
2014: M = *newmat;
2015: PetscCall(MatGetLocalSize(M, &ml, &nl));
2016: PetscCheck(ml == m, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Previous matrix must be same size/layout as request");
2017: PetscCall(MatZeroEntries(M));
2018: /*
2019: The next two lines are needed so we may call MatSetValues_MPIAIJ() below directly,
2020: rather than the slower MatSetValues().
2021: */
2022: M->was_assembled = PETSC_TRUE;
2023: M->assembled = PETSC_FALSE;
2024: }
2025: PetscCall(MatSetOption(M, MAT_ROW_ORIENTED, PETSC_FALSE));
2026: PetscCall(MatGetOwnershipRange(M, &rstart, &rend));
2027: aij = (Mat_SeqBAIJ *)(Mreuse)->data;
2028: ii = aij->i;
2029: jj = aij->j;
2030: aa = aij->a;
2031: for (i = 0; i < m; i++) {
2032: row = rstart / bs + i;
2033: nz = ii[i + 1] - ii[i];
2034: cwork = jj;
2035: jj += nz;
2036: vwork = aa;
2037: aa += nz * bs * bs;
2038: PetscCall(MatSetValuesBlocked_MPIBAIJ(M, 1, &row, nz, cwork, vwork, INSERT_VALUES));
2039: }
2041: PetscCall(MatAssemblyBegin(M, MAT_FINAL_ASSEMBLY));
2042: PetscCall(MatAssemblyEnd(M, MAT_FINAL_ASSEMBLY));
2043: *newmat = M;
2045: /* save submatrix used in processor for next request */
2046: if (call == MAT_INITIAL_MATRIX) {
2047: PetscCall(PetscObjectCompose((PetscObject)M, "SubMatrix", (PetscObject)Mreuse));
2048: PetscCall(PetscObjectDereference((PetscObject)Mreuse));
2049: }
2050: PetscFunctionReturn(PETSC_SUCCESS);
2051: }
2053: static PetscErrorCode MatPermute_MPIBAIJ(Mat A, IS rowp, IS colp, Mat *B)
2054: {
2055: MPI_Comm comm, pcomm;
2056: PetscInt clocal_size, nrows;
2057: const PetscInt *rows;
2058: PetscMPIInt size;
2059: IS crowp, lcolp;
2061: PetscFunctionBegin;
2062: PetscCall(PetscObjectGetComm((PetscObject)A, &comm));
2063: /* make a collective version of 'rowp' */
2064: PetscCall(PetscObjectGetComm((PetscObject)rowp, &pcomm));
2065: if (pcomm == comm) {
2066: crowp = rowp;
2067: } else {
2068: PetscCall(ISGetSize(rowp, &nrows));
2069: PetscCall(ISGetIndices(rowp, &rows));
2070: PetscCall(ISCreateGeneral(comm, nrows, rows, PETSC_COPY_VALUES, &crowp));
2071: PetscCall(ISRestoreIndices(rowp, &rows));
2072: }
2073: PetscCall(ISSetPermutation(crowp));
2074: /* make a local version of 'colp' */
2075: PetscCall(PetscObjectGetComm((PetscObject)colp, &pcomm));
2076: PetscCallMPI(MPI_Comm_size(pcomm, &size));
2077: if (size == 1) {
2078: lcolp = colp;
2079: } else {
2080: PetscCall(ISAllGather(colp, &lcolp));
2081: }
2082: PetscCall(ISSetPermutation(lcolp));
2083: /* now we just get the submatrix */
2084: PetscCall(MatGetLocalSize(A, NULL, &clocal_size));
2085: PetscCall(MatCreateSubMatrix_MPIBAIJ_Private(A, crowp, lcolp, clocal_size, MAT_INITIAL_MATRIX, B, PETSC_FALSE));
2086: /* clean up */
2087: if (pcomm != comm) PetscCall(ISDestroy(&crowp));
2088: if (size > 1) PetscCall(ISDestroy(&lcolp));
2089: PetscFunctionReturn(PETSC_SUCCESS);
2090: }
2092: static PetscErrorCode MatGetGhosts_MPIBAIJ(Mat mat, PetscInt *nghosts, const PetscInt *ghosts[])
2093: {
2094: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
2095: Mat_SeqBAIJ *B = (Mat_SeqBAIJ *)baij->B->data;
2097: PetscFunctionBegin;
2098: if (nghosts) *nghosts = B->nbs;
2099: if (ghosts) *ghosts = baij->garray;
2100: PetscFunctionReturn(PETSC_SUCCESS);
2101: }
2103: static PetscErrorCode MatGetSeqNonzeroStructure_MPIBAIJ(Mat A, Mat *newmat)
2104: {
2105: Mat B;
2106: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
2107: Mat_SeqBAIJ *ad = (Mat_SeqBAIJ *)a->A->data, *bd = (Mat_SeqBAIJ *)a->B->data;
2108: Mat_SeqAIJ *b;
2109: PetscMPIInt size, rank, *recvcounts = NULL, *displs = NULL;
2110: PetscInt sendcount, i, *rstarts = A->rmap->range, n, cnt, j, bs = A->rmap->bs;
2111: PetscInt m, *garray = a->garray, *lens, *jsendbuf, *a_jsendbuf, *b_jsendbuf;
2113: PetscFunctionBegin;
2114: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)A), &size));
2115: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)A), &rank));
2117: /* Tell every processor the number of nonzeros per row */
2118: PetscCall(PetscMalloc1(A->rmap->N / bs, &lens));
2119: for (i = A->rmap->rstart / bs; i < A->rmap->rend / bs; i++) lens[i] = ad->i[i - A->rmap->rstart / bs + 1] - ad->i[i - A->rmap->rstart / bs] + bd->i[i - A->rmap->rstart / bs + 1] - bd->i[i - A->rmap->rstart / bs];
2120: PetscCall(PetscMalloc1(2 * size, &recvcounts));
2121: displs = recvcounts + size;
2122: for (i = 0; i < size; i++) {
2123: recvcounts[i] = A->rmap->range[i + 1] / bs - A->rmap->range[i] / bs;
2124: displs[i] = A->rmap->range[i] / bs;
2125: }
2126: PetscCallMPI(MPI_Allgatherv(MPI_IN_PLACE, 0, MPI_DATATYPE_NULL, lens, recvcounts, displs, MPIU_INT, PetscObjectComm((PetscObject)A)));
2127: /* Create the sequential matrix of the same type as the local block diagonal */
2128: PetscCall(MatCreate(PETSC_COMM_SELF, &B));
2129: PetscCall(MatSetSizes(B, A->rmap->N / bs, A->cmap->N / bs, PETSC_DETERMINE, PETSC_DETERMINE));
2130: PetscCall(MatSetType(B, MATSEQAIJ));
2131: PetscCall(MatSeqAIJSetPreallocation(B, 0, lens));
2132: b = (Mat_SeqAIJ *)B->data;
2134: /* Copy my part of matrix column indices over */
2135: sendcount = ad->nz + bd->nz;
2136: jsendbuf = b->j + b->i[rstarts[rank] / bs];
2137: a_jsendbuf = ad->j;
2138: b_jsendbuf = bd->j;
2139: n = A->rmap->rend / bs - A->rmap->rstart / bs;
2140: cnt = 0;
2141: for (i = 0; i < n; i++) {
2142: /* put in lower diagonal portion */
2143: m = bd->i[i + 1] - bd->i[i];
2144: while (m > 0) {
2145: /* is it above diagonal (in bd (compressed) numbering) */
2146: if (garray[*b_jsendbuf] > A->rmap->rstart / bs + i) break;
2147: jsendbuf[cnt++] = garray[*b_jsendbuf++];
2148: m--;
2149: }
2151: /* put in diagonal portion */
2152: for (j = ad->i[i]; j < ad->i[i + 1]; j++) jsendbuf[cnt++] = A->rmap->rstart / bs + *a_jsendbuf++;
2154: /* put in upper diagonal portion */
2155: while (m-- > 0) jsendbuf[cnt++] = garray[*b_jsendbuf++];
2156: }
2157: PetscCheck(cnt == sendcount, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Corrupted PETSc matrix: nz given %" PetscInt_FMT " actual nz %" PetscInt_FMT, sendcount, cnt);
2159: /* Gather all column indices to all processors */
2160: for (i = 0; i < size; i++) {
2161: recvcounts[i] = 0;
2162: for (j = A->rmap->range[i] / bs; j < A->rmap->range[i + 1] / bs; j++) recvcounts[i] += lens[j];
2163: }
2164: displs[0] = 0;
2165: for (i = 1; i < size; i++) displs[i] = displs[i - 1] + recvcounts[i - 1];
2166: PetscCallMPI(MPI_Allgatherv(MPI_IN_PLACE, 0, MPI_DATATYPE_NULL, b->j, recvcounts, displs, MPIU_INT, PetscObjectComm((PetscObject)A)));
2167: /* Assemble the matrix into useable form (note numerical values not yet set) */
2168: /* set the b->ilen (length of each row) values */
2169: PetscCall(PetscArraycpy(b->ilen, lens, A->rmap->N / bs));
2170: /* set the b->i indices */
2171: b->i[0] = 0;
2172: for (i = 1; i <= A->rmap->N / bs; i++) b->i[i] = b->i[i - 1] + lens[i - 1];
2173: PetscCall(PetscFree(lens));
2174: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
2175: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
2176: PetscCall(PetscFree(recvcounts));
2178: PetscCall(MatPropagateSymmetryOptions(A, B));
2179: *newmat = B;
2180: PetscFunctionReturn(PETSC_SUCCESS);
2181: }
2183: static PetscErrorCode MatSOR_MPIBAIJ(Mat matin, Vec bb, PetscReal omega, MatSORType flag, PetscReal fshift, PetscInt its, PetscInt lits, Vec xx)
2184: {
2185: Mat_MPIBAIJ *mat = (Mat_MPIBAIJ *)matin->data;
2186: Vec bb1 = NULL;
2188: PetscFunctionBegin;
2189: if (flag == SOR_APPLY_UPPER) {
2190: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2191: PetscFunctionReturn(PETSC_SUCCESS);
2192: }
2194: if (its > 1 || ~flag & SOR_ZERO_INITIAL_GUESS) PetscCall(VecDuplicate(bb, &bb1));
2196: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP) {
2197: if (flag & SOR_ZERO_INITIAL_GUESS) {
2198: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2199: its--;
2200: }
2202: while (its--) {
2203: PetscCall(VecScatterBegin(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2204: PetscCall(VecScatterEnd(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2206: /* update rhs: bb1 = bb - B*x */
2207: PetscCall(VecScale(mat->lvec, -1.0));
2208: PetscCall((*mat->B->ops->multadd)(mat->B, mat->lvec, bb, bb1));
2210: /* local sweep */
2211: PetscCall((*mat->A->ops->sor)(mat->A, bb1, omega, SOR_SYMMETRIC_SWEEP, fshift, lits, 1, xx));
2212: }
2213: } else if (flag & SOR_LOCAL_FORWARD_SWEEP) {
2214: if (flag & SOR_ZERO_INITIAL_GUESS) {
2215: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2216: its--;
2217: }
2218: while (its--) {
2219: PetscCall(VecScatterBegin(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2220: PetscCall(VecScatterEnd(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2222: /* update rhs: bb1 = bb - B*x */
2223: PetscCall(VecScale(mat->lvec, -1.0));
2224: PetscCall((*mat->B->ops->multadd)(mat->B, mat->lvec, bb, bb1));
2226: /* local sweep */
2227: PetscCall((*mat->A->ops->sor)(mat->A, bb1, omega, SOR_FORWARD_SWEEP, fshift, lits, 1, xx));
2228: }
2229: } else if (flag & SOR_LOCAL_BACKWARD_SWEEP) {
2230: if (flag & SOR_ZERO_INITIAL_GUESS) {
2231: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2232: its--;
2233: }
2234: while (its--) {
2235: PetscCall(VecScatterBegin(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2236: PetscCall(VecScatterEnd(mat->Mvctx, xx, mat->lvec, INSERT_VALUES, SCATTER_FORWARD));
2238: /* update rhs: bb1 = bb - B*x */
2239: PetscCall(VecScale(mat->lvec, -1.0));
2240: PetscCall((*mat->B->ops->multadd)(mat->B, mat->lvec, bb, bb1));
2242: /* local sweep */
2243: PetscCall((*mat->A->ops->sor)(mat->A, bb1, omega, SOR_BACKWARD_SWEEP, fshift, lits, 1, xx));
2244: }
2245: } else SETERRQ(PetscObjectComm((PetscObject)matin), PETSC_ERR_SUP, "Parallel version of SOR requested not supported");
2247: PetscCall(VecDestroy(&bb1));
2248: PetscFunctionReturn(PETSC_SUCCESS);
2249: }
2251: static PetscErrorCode MatGetColumnReductions_MPIBAIJ(Mat A, PetscInt type, PetscReal *reductions)
2252: {
2253: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ *)A->data;
2254: PetscInt m, N, i, *garray = aij->garray;
2255: PetscInt ib, jb, bs = A->rmap->bs;
2256: Mat_SeqBAIJ *a_aij = (Mat_SeqBAIJ *)aij->A->data;
2257: MatScalar *a_val = a_aij->a;
2258: Mat_SeqBAIJ *b_aij = (Mat_SeqBAIJ *)aij->B->data;
2259: MatScalar *b_val = b_aij->a;
2260: PetscReal *work;
2262: PetscFunctionBegin;
2263: PetscCall(MatGetSize(A, &m, &N));
2264: PetscCall(PetscCalloc1(N, &work));
2265: if (type == NORM_2) {
2266: for (i = a_aij->i[0]; i < a_aij->i[aij->A->rmap->n / bs]; i++) {
2267: for (jb = 0; jb < bs; jb++) {
2268: for (ib = 0; ib < bs; ib++) {
2269: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscAbsScalar(*a_val * *a_val);
2270: a_val++;
2271: }
2272: }
2273: }
2274: for (i = b_aij->i[0]; i < b_aij->i[aij->B->rmap->n / bs]; i++) {
2275: for (jb = 0; jb < bs; jb++) {
2276: for (ib = 0; ib < bs; ib++) {
2277: work[garray[b_aij->j[i]] * bs + jb] += PetscAbsScalar(*b_val * *b_val);
2278: b_val++;
2279: }
2280: }
2281: }
2282: } else if (type == NORM_1) {
2283: for (i = a_aij->i[0]; i < a_aij->i[aij->A->rmap->n / bs]; i++) {
2284: for (jb = 0; jb < bs; jb++) {
2285: for (ib = 0; ib < bs; ib++) {
2286: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscAbsScalar(*a_val);
2287: a_val++;
2288: }
2289: }
2290: }
2291: for (i = b_aij->i[0]; i < b_aij->i[aij->B->rmap->n / bs]; i++) {
2292: for (jb = 0; jb < bs; jb++) {
2293: for (ib = 0; ib < bs; ib++) {
2294: work[garray[b_aij->j[i]] * bs + jb] += PetscAbsScalar(*b_val);
2295: b_val++;
2296: }
2297: }
2298: }
2299: } else if (type == NORM_INFINITY) {
2300: for (i = a_aij->i[0]; i < a_aij->i[aij->A->rmap->n / bs]; i++) {
2301: for (jb = 0; jb < bs; jb++) {
2302: for (ib = 0; ib < bs; ib++) {
2303: int col = A->cmap->rstart + a_aij->j[i] * bs + jb;
2304: work[col] = PetscMax(PetscAbsScalar(*a_val), work[col]);
2305: a_val++;
2306: }
2307: }
2308: }
2309: for (i = b_aij->i[0]; i < b_aij->i[aij->B->rmap->n / bs]; i++) {
2310: for (jb = 0; jb < bs; jb++) {
2311: for (ib = 0; ib < bs; ib++) {
2312: int col = garray[b_aij->j[i]] * bs + jb;
2313: work[col] = PetscMax(PetscAbsScalar(*b_val), work[col]);
2314: b_val++;
2315: }
2316: }
2317: }
2318: } else if (type == REDUCTION_SUM_REALPART || type == REDUCTION_MEAN_REALPART) {
2319: for (i = a_aij->i[0]; i < a_aij->i[aij->A->rmap->n / bs]; i++) {
2320: for (jb = 0; jb < bs; jb++) {
2321: for (ib = 0; ib < bs; ib++) {
2322: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscRealPart(*a_val);
2323: a_val++;
2324: }
2325: }
2326: }
2327: for (i = b_aij->i[0]; i < b_aij->i[aij->B->rmap->n / bs]; i++) {
2328: for (jb = 0; jb < bs; jb++) {
2329: for (ib = 0; ib < bs; ib++) {
2330: work[garray[b_aij->j[i]] * bs + jb] += PetscRealPart(*b_val);
2331: b_val++;
2332: }
2333: }
2334: }
2335: } else if (type == REDUCTION_SUM_IMAGINARYPART || type == REDUCTION_MEAN_IMAGINARYPART) {
2336: for (i = a_aij->i[0]; i < a_aij->i[aij->A->rmap->n / bs]; i++) {
2337: for (jb = 0; jb < bs; jb++) {
2338: for (ib = 0; ib < bs; ib++) {
2339: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscImaginaryPart(*a_val);
2340: a_val++;
2341: }
2342: }
2343: }
2344: for (i = b_aij->i[0]; i < b_aij->i[aij->B->rmap->n / bs]; i++) {
2345: for (jb = 0; jb < bs; jb++) {
2346: for (ib = 0; ib < bs; ib++) {
2347: work[garray[b_aij->j[i]] * bs + jb] += PetscImaginaryPart(*b_val);
2348: b_val++;
2349: }
2350: }
2351: }
2352: } else SETERRQ(PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONG, "Unknown reduction type");
2353: if (type == NORM_INFINITY) {
2354: PetscCall(MPIU_Allreduce(work, reductions, N, MPIU_REAL, MPIU_MAX, PetscObjectComm((PetscObject)A)));
2355: } else {
2356: PetscCall(MPIU_Allreduce(work, reductions, N, MPIU_REAL, MPIU_SUM, PetscObjectComm((PetscObject)A)));
2357: }
2358: PetscCall(PetscFree(work));
2359: if (type == NORM_2) {
2360: for (i = 0; i < N; i++) reductions[i] = PetscSqrtReal(reductions[i]);
2361: } else if (type == REDUCTION_MEAN_REALPART || type == REDUCTION_MEAN_IMAGINARYPART) {
2362: for (i = 0; i < N; i++) reductions[i] /= m;
2363: }
2364: PetscFunctionReturn(PETSC_SUCCESS);
2365: }
2367: static PetscErrorCode MatInvertBlockDiagonal_MPIBAIJ(Mat A, const PetscScalar **values)
2368: {
2369: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
2371: PetscFunctionBegin;
2372: PetscCall(MatInvertBlockDiagonal(a->A, values));
2373: A->factorerrortype = a->A->factorerrortype;
2374: A->factorerror_zeropivot_value = a->A->factorerror_zeropivot_value;
2375: A->factorerror_zeropivot_row = a->A->factorerror_zeropivot_row;
2376: PetscFunctionReturn(PETSC_SUCCESS);
2377: }
2379: static PetscErrorCode MatShift_MPIBAIJ(Mat Y, PetscScalar a)
2380: {
2381: Mat_MPIBAIJ *maij = (Mat_MPIBAIJ *)Y->data;
2382: Mat_SeqBAIJ *aij = (Mat_SeqBAIJ *)maij->A->data;
2384: PetscFunctionBegin;
2385: if (!Y->preallocated) {
2386: PetscCall(MatMPIBAIJSetPreallocation(Y, Y->rmap->bs, 1, NULL, 0, NULL));
2387: } else if (!aij->nz) {
2388: PetscInt nonew = aij->nonew;
2389: PetscCall(MatSeqBAIJSetPreallocation(maij->A, Y->rmap->bs, 1, NULL));
2390: aij->nonew = nonew;
2391: }
2392: PetscCall(MatShift_Basic(Y, a));
2393: PetscFunctionReturn(PETSC_SUCCESS);
2394: }
2396: static PetscErrorCode MatMissingDiagonal_MPIBAIJ(Mat A, PetscBool *missing, PetscInt *d)
2397: {
2398: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
2400: PetscFunctionBegin;
2401: PetscCheck(A->rmap->n == A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only works for square matrices");
2402: PetscCall(MatMissingDiagonal(a->A, missing, d));
2403: if (d) {
2404: PetscInt rstart;
2405: PetscCall(MatGetOwnershipRange(A, &rstart, NULL));
2406: *d += rstart / A->rmap->bs;
2407: }
2408: PetscFunctionReturn(PETSC_SUCCESS);
2409: }
2411: static PetscErrorCode MatGetDiagonalBlock_MPIBAIJ(Mat A, Mat *a)
2412: {
2413: PetscFunctionBegin;
2414: *a = ((Mat_MPIBAIJ *)A->data)->A;
2415: PetscFunctionReturn(PETSC_SUCCESS);
2416: }
2418: static PetscErrorCode MatEliminateZeros_MPIBAIJ(Mat A, PetscBool keep)
2419: {
2420: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
2422: PetscFunctionBegin;
2423: PetscCall(MatEliminateZeros_SeqBAIJ(a->A, keep)); // possibly keep zero diagonal coefficients
2424: PetscCall(MatEliminateZeros_SeqBAIJ(a->B, PETSC_FALSE)); // never keep zero diagonal coefficients
2425: PetscFunctionReturn(PETSC_SUCCESS);
2426: }
2428: static struct _MatOps MatOps_Values = {MatSetValues_MPIBAIJ,
2429: MatGetRow_MPIBAIJ,
2430: MatRestoreRow_MPIBAIJ,
2431: MatMult_MPIBAIJ,
2432: /* 4*/ MatMultAdd_MPIBAIJ,
2433: MatMultTranspose_MPIBAIJ,
2434: MatMultTransposeAdd_MPIBAIJ,
2435: NULL,
2436: NULL,
2437: NULL,
2438: /*10*/ NULL,
2439: NULL,
2440: NULL,
2441: MatSOR_MPIBAIJ,
2442: MatTranspose_MPIBAIJ,
2443: /*15*/ MatGetInfo_MPIBAIJ,
2444: MatEqual_MPIBAIJ,
2445: MatGetDiagonal_MPIBAIJ,
2446: MatDiagonalScale_MPIBAIJ,
2447: MatNorm_MPIBAIJ,
2448: /*20*/ MatAssemblyBegin_MPIBAIJ,
2449: MatAssemblyEnd_MPIBAIJ,
2450: MatSetOption_MPIBAIJ,
2451: MatZeroEntries_MPIBAIJ,
2452: /*24*/ MatZeroRows_MPIBAIJ,
2453: NULL,
2454: NULL,
2455: NULL,
2456: NULL,
2457: /*29*/ MatSetUp_MPI_Hash,
2458: NULL,
2459: NULL,
2460: MatGetDiagonalBlock_MPIBAIJ,
2461: NULL,
2462: /*34*/ MatDuplicate_MPIBAIJ,
2463: NULL,
2464: NULL,
2465: NULL,
2466: NULL,
2467: /*39*/ MatAXPY_MPIBAIJ,
2468: MatCreateSubMatrices_MPIBAIJ,
2469: MatIncreaseOverlap_MPIBAIJ,
2470: MatGetValues_MPIBAIJ,
2471: MatCopy_MPIBAIJ,
2472: /*44*/ NULL,
2473: MatScale_MPIBAIJ,
2474: MatShift_MPIBAIJ,
2475: NULL,
2476: MatZeroRowsColumns_MPIBAIJ,
2477: /*49*/ NULL,
2478: NULL,
2479: NULL,
2480: NULL,
2481: NULL,
2482: /*54*/ MatFDColoringCreate_MPIXAIJ,
2483: NULL,
2484: MatSetUnfactored_MPIBAIJ,
2485: MatPermute_MPIBAIJ,
2486: MatSetValuesBlocked_MPIBAIJ,
2487: /*59*/ MatCreateSubMatrix_MPIBAIJ,
2488: MatDestroy_MPIBAIJ,
2489: MatView_MPIBAIJ,
2490: NULL,
2491: NULL,
2492: /*64*/ NULL,
2493: NULL,
2494: NULL,
2495: NULL,
2496: NULL,
2497: /*69*/ MatGetRowMaxAbs_MPIBAIJ,
2498: NULL,
2499: NULL,
2500: NULL,
2501: NULL,
2502: /*74*/ NULL,
2503: MatFDColoringApply_BAIJ,
2504: NULL,
2505: NULL,
2506: NULL,
2507: /*79*/ NULL,
2508: NULL,
2509: NULL,
2510: NULL,
2511: MatLoad_MPIBAIJ,
2512: /*84*/ NULL,
2513: NULL,
2514: NULL,
2515: NULL,
2516: NULL,
2517: /*89*/ NULL,
2518: NULL,
2519: NULL,
2520: NULL,
2521: NULL,
2522: /*94*/ NULL,
2523: NULL,
2524: NULL,
2525: NULL,
2526: NULL,
2527: /*99*/ NULL,
2528: NULL,
2529: NULL,
2530: MatConjugate_MPIBAIJ,
2531: NULL,
2532: /*104*/ NULL,
2533: MatRealPart_MPIBAIJ,
2534: MatImaginaryPart_MPIBAIJ,
2535: NULL,
2536: NULL,
2537: /*109*/ NULL,
2538: NULL,
2539: NULL,
2540: NULL,
2541: MatMissingDiagonal_MPIBAIJ,
2542: /*114*/ MatGetSeqNonzeroStructure_MPIBAIJ,
2543: NULL,
2544: MatGetGhosts_MPIBAIJ,
2545: NULL,
2546: NULL,
2547: /*119*/ NULL,
2548: NULL,
2549: NULL,
2550: NULL,
2551: MatGetMultiProcBlock_MPIBAIJ,
2552: /*124*/ NULL,
2553: MatGetColumnReductions_MPIBAIJ,
2554: MatInvertBlockDiagonal_MPIBAIJ,
2555: NULL,
2556: NULL,
2557: /*129*/ NULL,
2558: NULL,
2559: NULL,
2560: NULL,
2561: NULL,
2562: /*134*/ NULL,
2563: NULL,
2564: NULL,
2565: NULL,
2566: NULL,
2567: /*139*/ MatSetBlockSizes_Default,
2568: NULL,
2569: NULL,
2570: MatFDColoringSetUp_MPIXAIJ,
2571: NULL,
2572: /*144*/ MatCreateMPIMatConcatenateSeqMat_MPIBAIJ,
2573: NULL,
2574: NULL,
2575: NULL,
2576: NULL,
2577: NULL,
2578: /*150*/ NULL,
2579: MatEliminateZeros_MPIBAIJ};
2581: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPISBAIJ(Mat, MatType, MatReuse, Mat *);
2582: PETSC_INTERN PetscErrorCode MatConvert_XAIJ_IS(Mat, MatType, MatReuse, Mat *);
2584: static PetscErrorCode MatMPIBAIJSetPreallocationCSR_MPIBAIJ(Mat B, PetscInt bs, const PetscInt ii[], const PetscInt jj[], const PetscScalar V[])
2585: {
2586: PetscInt m, rstart, cstart, cend;
2587: PetscInt i, j, dlen, olen, nz, nz_max = 0, *d_nnz = NULL, *o_nnz = NULL;
2588: const PetscInt *JJ = NULL;
2589: PetscScalar *values = NULL;
2590: PetscBool roworiented = ((Mat_MPIBAIJ *)B->data)->roworiented;
2591: PetscBool nooffprocentries;
2593: PetscFunctionBegin;
2594: PetscCall(PetscLayoutSetBlockSize(B->rmap, bs));
2595: PetscCall(PetscLayoutSetBlockSize(B->cmap, bs));
2596: PetscCall(PetscLayoutSetUp(B->rmap));
2597: PetscCall(PetscLayoutSetUp(B->cmap));
2598: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
2599: m = B->rmap->n / bs;
2600: rstart = B->rmap->rstart / bs;
2601: cstart = B->cmap->rstart / bs;
2602: cend = B->cmap->rend / bs;
2604: PetscCheck(!ii[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "ii[0] must be 0 but it is %" PetscInt_FMT, ii[0]);
2605: PetscCall(PetscMalloc2(m, &d_nnz, m, &o_nnz));
2606: for (i = 0; i < m; i++) {
2607: nz = ii[i + 1] - ii[i];
2608: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Local row %" PetscInt_FMT " has a negative number of columns %" PetscInt_FMT, i, nz);
2609: nz_max = PetscMax(nz_max, nz);
2610: dlen = 0;
2611: olen = 0;
2612: JJ = jj + ii[i];
2613: for (j = 0; j < nz; j++) {
2614: if (*JJ < cstart || *JJ >= cend) olen++;
2615: else dlen++;
2616: JJ++;
2617: }
2618: d_nnz[i] = dlen;
2619: o_nnz[i] = olen;
2620: }
2621: PetscCall(MatMPIBAIJSetPreallocation(B, bs, 0, d_nnz, 0, o_nnz));
2622: PetscCall(PetscFree2(d_nnz, o_nnz));
2624: values = (PetscScalar *)V;
2625: if (!values) PetscCall(PetscCalloc1(bs * bs * nz_max, &values));
2626: for (i = 0; i < m; i++) {
2627: PetscInt row = i + rstart;
2628: PetscInt ncols = ii[i + 1] - ii[i];
2629: const PetscInt *icols = jj + ii[i];
2630: if (bs == 1 || !roworiented) { /* block ordering matches the non-nested layout of MatSetValues so we can insert entire rows */
2631: const PetscScalar *svals = values + (V ? (bs * bs * ii[i]) : 0);
2632: PetscCall(MatSetValuesBlocked_MPIBAIJ(B, 1, &row, ncols, icols, svals, INSERT_VALUES));
2633: } else { /* block ordering does not match so we can only insert one block at a time. */
2634: PetscInt j;
2635: for (j = 0; j < ncols; j++) {
2636: const PetscScalar *svals = values + (V ? (bs * bs * (ii[i] + j)) : 0);
2637: PetscCall(MatSetValuesBlocked_MPIBAIJ(B, 1, &row, 1, &icols[j], svals, INSERT_VALUES));
2638: }
2639: }
2640: }
2642: if (!V) PetscCall(PetscFree(values));
2643: nooffprocentries = B->nooffprocentries;
2644: B->nooffprocentries = PETSC_TRUE;
2645: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
2646: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
2647: B->nooffprocentries = nooffprocentries;
2649: PetscCall(MatSetOption(B, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2650: PetscFunctionReturn(PETSC_SUCCESS);
2651: }
2653: /*@C
2654: MatMPIBAIJSetPreallocationCSR - Creates a sparse parallel matrix in `MATBAIJ` format using the given nonzero structure and (optional) numerical values
2656: Collective
2658: Input Parameters:
2659: + B - the matrix
2660: . bs - the block size
2661: . i - the indices into `j` for the start of each local row (starts with zero)
2662: . j - the column indices for each local row (starts with zero) these must be sorted for each row
2663: - v - optional values in the matrix
2665: Level: advanced
2667: Notes:
2668: The order of the entries in values is specified by the `MatOption` `MAT_ROW_ORIENTED`. For example, C programs
2669: may want to use the default `MAT_ROW_ORIENTED` with value `PETSC_TRUE` and use an array v[nnz][bs][bs] where the second index is
2670: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
2671: `MAT_ROW_ORIENTED` with value `PETSC_FALSE` and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
2672: block column and the second index is over columns within a block.
2674: Though this routine has Preallocation() in the name it also sets the exact nonzero locations of the matrix entries and usually the numerical values as well
2676: .seealso: `Mat`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatMPIBAIJSetPreallocation()`, `MatCreateAIJ()`, `MATMPIAIJ`, `MatCreateMPIBAIJWithArrays()`, `MATMPIBAIJ`
2677: @*/
2678: PetscErrorCode MatMPIBAIJSetPreallocationCSR(Mat B, PetscInt bs, const PetscInt i[], const PetscInt j[], const PetscScalar v[])
2679: {
2680: PetscFunctionBegin;
2684: PetscTryMethod(B, "MatMPIBAIJSetPreallocationCSR_C", (Mat, PetscInt, const PetscInt[], const PetscInt[], const PetscScalar[]), (B, bs, i, j, v));
2685: PetscFunctionReturn(PETSC_SUCCESS);
2686: }
2688: PetscErrorCode MatMPIBAIJSetPreallocation_MPIBAIJ(Mat B, PetscInt bs, PetscInt d_nz, const PetscInt *d_nnz, PetscInt o_nz, const PetscInt *o_nnz)
2689: {
2690: Mat_MPIBAIJ *b = (Mat_MPIBAIJ *)B->data;
2691: PetscInt i;
2692: PetscMPIInt size;
2694: PetscFunctionBegin;
2695: if (B->hash_active) {
2696: B->ops[0] = b->cops;
2697: B->hash_active = PETSC_FALSE;
2698: }
2699: if (!B->preallocated) PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)B), bs, &B->bstash));
2700: PetscCall(MatSetBlockSize(B, PetscAbs(bs)));
2701: PetscCall(PetscLayoutSetUp(B->rmap));
2702: PetscCall(PetscLayoutSetUp(B->cmap));
2703: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
2705: if (d_nnz) {
2706: for (i = 0; i < B->rmap->n / bs; i++) PetscCheck(d_nnz[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "d_nnz cannot be less than -1: local row %" PetscInt_FMT " value %" PetscInt_FMT, i, d_nnz[i]);
2707: }
2708: if (o_nnz) {
2709: for (i = 0; i < B->rmap->n / bs; i++) PetscCheck(o_nnz[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "o_nnz cannot be less than -1: local row %" PetscInt_FMT " value %" PetscInt_FMT, i, o_nnz[i]);
2710: }
2712: b->bs2 = bs * bs;
2713: b->mbs = B->rmap->n / bs;
2714: b->nbs = B->cmap->n / bs;
2715: b->Mbs = B->rmap->N / bs;
2716: b->Nbs = B->cmap->N / bs;
2718: for (i = 0; i <= b->size; i++) b->rangebs[i] = B->rmap->range[i] / bs;
2719: b->rstartbs = B->rmap->rstart / bs;
2720: b->rendbs = B->rmap->rend / bs;
2721: b->cstartbs = B->cmap->rstart / bs;
2722: b->cendbs = B->cmap->rend / bs;
2724: #if defined(PETSC_USE_CTABLE)
2725: PetscCall(PetscHMapIDestroy(&b->colmap));
2726: #else
2727: PetscCall(PetscFree(b->colmap));
2728: #endif
2729: PetscCall(PetscFree(b->garray));
2730: PetscCall(VecDestroy(&b->lvec));
2731: PetscCall(VecScatterDestroy(&b->Mvctx));
2733: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &size));
2734: PetscCall(MatDestroy(&b->B));
2735: PetscCall(MatCreate(PETSC_COMM_SELF, &b->B));
2736: PetscCall(MatSetSizes(b->B, B->rmap->n, size > 1 ? B->cmap->N : 0, B->rmap->n, size > 1 ? B->cmap->N : 0));
2737: PetscCall(MatSetType(b->B, MATSEQBAIJ));
2739: PetscCall(MatDestroy(&b->A));
2740: PetscCall(MatCreate(PETSC_COMM_SELF, &b->A));
2741: PetscCall(MatSetSizes(b->A, B->rmap->n, B->cmap->n, B->rmap->n, B->cmap->n));
2742: PetscCall(MatSetType(b->A, MATSEQBAIJ));
2744: PetscCall(MatSeqBAIJSetPreallocation(b->A, bs, d_nz, d_nnz));
2745: PetscCall(MatSeqBAIJSetPreallocation(b->B, bs, o_nz, o_nnz));
2746: B->preallocated = PETSC_TRUE;
2747: B->was_assembled = PETSC_FALSE;
2748: B->assembled = PETSC_FALSE;
2749: PetscFunctionReturn(PETSC_SUCCESS);
2750: }
2752: extern PetscErrorCode MatDiagonalScaleLocal_MPIBAIJ(Mat, Vec);
2753: extern PetscErrorCode MatSetHashTableFactor_MPIBAIJ(Mat, PetscReal);
2755: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIAdj(Mat B, MatType newtype, MatReuse reuse, Mat *adj)
2756: {
2757: Mat_MPIBAIJ *b = (Mat_MPIBAIJ *)B->data;
2758: Mat_SeqBAIJ *d = (Mat_SeqBAIJ *)b->A->data, *o = (Mat_SeqBAIJ *)b->B->data;
2759: PetscInt M = B->rmap->n / B->rmap->bs, i, *ii, *jj, cnt, j, k, rstart = B->rmap->rstart / B->rmap->bs;
2760: const PetscInt *id = d->i, *jd = d->j, *io = o->i, *jo = o->j, *garray = b->garray;
2762: PetscFunctionBegin;
2763: PetscCall(PetscMalloc1(M + 1, &ii));
2764: ii[0] = 0;
2765: for (i = 0; i < M; i++) {
2766: PetscCheck((id[i + 1] - id[i]) >= 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Indices wrong %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT, i, id[i], id[i + 1]);
2767: PetscCheck((io[i + 1] - io[i]) >= 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Indices wrong %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT, i, io[i], io[i + 1]);
2768: ii[i + 1] = ii[i] + id[i + 1] - id[i] + io[i + 1] - io[i];
2769: /* remove one from count of matrix has diagonal */
2770: for (j = id[i]; j < id[i + 1]; j++) {
2771: if (jd[j] == i) {
2772: ii[i + 1]--;
2773: break;
2774: }
2775: }
2776: }
2777: PetscCall(PetscMalloc1(ii[M], &jj));
2778: cnt = 0;
2779: for (i = 0; i < M; i++) {
2780: for (j = io[i]; j < io[i + 1]; j++) {
2781: if (garray[jo[j]] > rstart) break;
2782: jj[cnt++] = garray[jo[j]];
2783: }
2784: for (k = id[i]; k < id[i + 1]; k++) {
2785: if (jd[k] != i) jj[cnt++] = rstart + jd[k];
2786: }
2787: for (; j < io[i + 1]; j++) jj[cnt++] = garray[jo[j]];
2788: }
2789: PetscCall(MatCreateMPIAdj(PetscObjectComm((PetscObject)B), M, B->cmap->N / B->rmap->bs, ii, jj, NULL, adj));
2790: PetscFunctionReturn(PETSC_SUCCESS);
2791: }
2793: #include <../src/mat/impls/aij/mpi/mpiaij.h>
2795: PETSC_INTERN PetscErrorCode MatConvert_SeqBAIJ_SeqAIJ(Mat, MatType, MatReuse, Mat *);
2797: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIAIJ(Mat A, MatType newtype, MatReuse reuse, Mat *newmat)
2798: {
2799: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
2800: Mat_MPIAIJ *b;
2801: Mat B;
2803: PetscFunctionBegin;
2804: PetscCheck(A->assembled, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "Matrix must be assembled");
2806: if (reuse == MAT_REUSE_MATRIX) {
2807: B = *newmat;
2808: } else {
2809: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &B));
2810: PetscCall(MatSetType(B, MATMPIAIJ));
2811: PetscCall(MatSetSizes(B, A->rmap->n, A->cmap->n, A->rmap->N, A->cmap->N));
2812: PetscCall(MatSetBlockSizes(B, A->rmap->bs, A->cmap->bs));
2813: PetscCall(MatSeqAIJSetPreallocation(B, 0, NULL));
2814: PetscCall(MatMPIAIJSetPreallocation(B, 0, NULL, 0, NULL));
2815: }
2816: b = (Mat_MPIAIJ *)B->data;
2818: if (reuse == MAT_REUSE_MATRIX) {
2819: PetscCall(MatConvert_SeqBAIJ_SeqAIJ(a->A, MATSEQAIJ, MAT_REUSE_MATRIX, &b->A));
2820: PetscCall(MatConvert_SeqBAIJ_SeqAIJ(a->B, MATSEQAIJ, MAT_REUSE_MATRIX, &b->B));
2821: } else {
2822: PetscInt *garray = a->garray;
2823: Mat_SeqAIJ *bB;
2824: PetscInt bs, nnz;
2825: PetscCall(MatDestroy(&b->A));
2826: PetscCall(MatDestroy(&b->B));
2827: /* just clear out the data structure */
2828: PetscCall(MatDisAssemble_MPIAIJ(B));
2829: PetscCall(MatConvert_SeqBAIJ_SeqAIJ(a->A, MATSEQAIJ, MAT_INITIAL_MATRIX, &b->A));
2830: PetscCall(MatConvert_SeqBAIJ_SeqAIJ(a->B, MATSEQAIJ, MAT_INITIAL_MATRIX, &b->B));
2832: /* Global numbering for b->B columns */
2833: bB = (Mat_SeqAIJ *)b->B->data;
2834: bs = A->rmap->bs;
2835: nnz = bB->i[A->rmap->n];
2836: for (PetscInt k = 0; k < nnz; k++) {
2837: PetscInt bj = bB->j[k] / bs;
2838: PetscInt br = bB->j[k] % bs;
2839: bB->j[k] = garray[bj] * bs + br;
2840: }
2841: }
2842: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
2843: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
2845: if (reuse == MAT_INPLACE_MATRIX) {
2846: PetscCall(MatHeaderReplace(A, &B));
2847: } else {
2848: *newmat = B;
2849: }
2850: PetscFunctionReturn(PETSC_SUCCESS);
2851: }
2853: /*MC
2854: MATMPIBAIJ - MATMPIBAIJ = "mpibaij" - A matrix type to be used for distributed block sparse matrices.
2856: Options Database Keys:
2857: + -mat_type mpibaij - sets the matrix type to `MATMPIBAIJ` during a call to `MatSetFromOptions()`
2858: . -mat_block_size <bs> - set the blocksize used to store the matrix
2859: . -mat_baij_mult_version version - indicate the version of the matrix-vector product to use (0 often indicates using BLAS)
2860: - -mat_use_hash_table <fact> - set hash table factor
2862: Level: beginner
2864: Note:
2865: `MatSetOption(A, MAT_STRUCTURE_ONLY, PETSC_TRUE)` may be called for this matrix type. In this no
2866: space is allocated for the nonzero entries and any entries passed with `MatSetValues()` are ignored
2868: .seealso: `Mat`, `MATBAIJ`, `MATSEQBAIJ`, `MatCreateBAIJ`
2869: M*/
2871: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIBSTRM(Mat, MatType, MatReuse, Mat *);
2873: PETSC_EXTERN PetscErrorCode MatCreate_MPIBAIJ(Mat B)
2874: {
2875: Mat_MPIBAIJ *b;
2876: PetscBool flg = PETSC_FALSE;
2878: PetscFunctionBegin;
2879: PetscCall(PetscNew(&b));
2880: B->data = (void *)b;
2881: B->ops[0] = MatOps_Values;
2882: B->assembled = PETSC_FALSE;
2884: B->insertmode = NOT_SET_VALUES;
2885: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)B), &b->rank));
2886: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &b->size));
2888: /* build local table of row and column ownerships */
2889: PetscCall(PetscMalloc1(b->size + 1, &b->rangebs));
2891: /* build cache for off array entries formed */
2892: PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)B), 1, &B->stash));
2894: b->donotstash = PETSC_FALSE;
2895: b->colmap = NULL;
2896: b->garray = NULL;
2897: b->roworiented = PETSC_TRUE;
2899: /* stuff used in block assembly */
2900: b->barray = NULL;
2902: /* stuff used for matrix vector multiply */
2903: b->lvec = NULL;
2904: b->Mvctx = NULL;
2906: /* stuff for MatGetRow() */
2907: b->rowindices = NULL;
2908: b->rowvalues = NULL;
2909: b->getrowactive = PETSC_FALSE;
2911: /* hash table stuff */
2912: b->ht = NULL;
2913: b->hd = NULL;
2914: b->ht_size = 0;
2915: b->ht_flag = PETSC_FALSE;
2916: b->ht_fact = 0;
2917: b->ht_total_ct = 0;
2918: b->ht_insert_ct = 0;
2920: /* stuff for MatCreateSubMatrices_MPIBAIJ_local() */
2921: b->ijonly = PETSC_FALSE;
2923: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpibaij_mpiadj_C", MatConvert_MPIBAIJ_MPIAdj));
2924: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpibaij_mpiaij_C", MatConvert_MPIBAIJ_MPIAIJ));
2925: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpibaij_mpisbaij_C", MatConvert_MPIBAIJ_MPISBAIJ));
2926: #if defined(PETSC_HAVE_HYPRE)
2927: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpibaij_hypre_C", MatConvert_AIJ_HYPRE));
2928: #endif
2929: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatStoreValues_C", MatStoreValues_MPIBAIJ));
2930: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatRetrieveValues_C", MatRetrieveValues_MPIBAIJ));
2931: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatMPIBAIJSetPreallocation_C", MatMPIBAIJSetPreallocation_MPIBAIJ));
2932: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatMPIBAIJSetPreallocationCSR_C", MatMPIBAIJSetPreallocationCSR_MPIBAIJ));
2933: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatDiagonalScaleLocal_C", MatDiagonalScaleLocal_MPIBAIJ));
2934: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSetHashTableFactor_C", MatSetHashTableFactor_MPIBAIJ));
2935: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpibaij_is_C", MatConvert_XAIJ_IS));
2936: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATMPIBAIJ));
2938: PetscOptionsBegin(PetscObjectComm((PetscObject)B), NULL, "Options for loading MPIBAIJ matrix 1", "Mat");
2939: PetscCall(PetscOptionsName("-mat_use_hash_table", "Use hash table to save time in constructing matrix", "MatSetOption", &flg));
2940: if (flg) {
2941: PetscReal fact = 1.39;
2942: PetscCall(MatSetOption(B, MAT_USE_HASH_TABLE, PETSC_TRUE));
2943: PetscCall(PetscOptionsReal("-mat_use_hash_table", "Use hash table factor", "MatMPIBAIJSetHashTableFactor", fact, &fact, NULL));
2944: if (fact <= 1.0) fact = 1.39;
2945: PetscCall(MatMPIBAIJSetHashTableFactor(B, fact));
2946: PetscCall(PetscInfo(B, "Hash table Factor used %5.2g\n", (double)fact));
2947: }
2948: PetscOptionsEnd();
2949: PetscFunctionReturn(PETSC_SUCCESS);
2950: }
2952: // PetscClangLinter pragma disable: -fdoc-section-header-unknown
2953: /*MC
2954: MATBAIJ - MATBAIJ = "baij" - A matrix type to be used for block sparse matrices.
2956: This matrix type is identical to `MATSEQBAIJ` when constructed with a single process communicator,
2957: and `MATMPIBAIJ` otherwise.
2959: Options Database Keys:
2960: . -mat_type baij - sets the matrix type to `MATBAIJ` during a call to `MatSetFromOptions()`
2962: Level: beginner
2964: .seealso: `Mat`, `MatCreateBAIJ()`, `MATSEQBAIJ`, `MATMPIBAIJ`, `MatMPIBAIJSetPreallocation()`, `MatMPIBAIJSetPreallocationCSR()`
2965: M*/
2967: /*@C
2968: MatMPIBAIJSetPreallocation - Allocates memory for a sparse parallel matrix in `MATMPIBAIJ` format
2969: (block compressed row).
2971: Collective
2973: Input Parameters:
2974: + B - the matrix
2975: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
2976: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with `MatCreateVecs()`
2977: . d_nz - number of block nonzeros per block row in diagonal portion of local
2978: submatrix (same for all local rows)
2979: . d_nnz - array containing the number of block nonzeros in the various block rows
2980: of the in diagonal portion of the local (possibly different for each block
2981: row) or `NULL`. If you plan to factor the matrix you must leave room for the diagonal entry and
2982: set it even if it is zero.
2983: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
2984: submatrix (same for all local rows).
2985: - o_nnz - array containing the number of nonzeros in the various block rows of the
2986: off-diagonal portion of the local submatrix (possibly different for
2987: each block row) or `NULL`.
2989: If the *_nnz parameter is given then the *_nz parameter is ignored
2991: Options Database Keys:
2992: + -mat_block_size - size of the blocks to use
2993: - -mat_use_hash_table <fact> - set hash table factor
2995: Level: intermediate
2997: Notes:
2998: For good matrix assembly performance
2999: the user should preallocate the matrix storage by setting the parameters
3000: `d_nz` (or `d_nnz`) and `o_nz` (or `o_nnz`). By setting these parameters accurately,
3001: performance can be increased by more than a factor of 50.
3003: If `PETSC_DECIDE` or `PETSC_DETERMINE` is used for a particular argument on one processor
3004: than it must be used on all processors that share the object for that argument.
3006: Storage Information:
3007: For a square global matrix we define each processor's diagonal portion
3008: to be its local rows and the corresponding columns (a square submatrix);
3009: each processor's off-diagonal portion encompasses the remainder of the
3010: local matrix (a rectangular submatrix).
3012: The user can specify preallocated storage for the diagonal part of
3013: the local submatrix with either `d_nz` or `d_nnz` (not both). Set
3014: `d_nz` = `PETSC_DEFAULT` and `d_nnz` = `NULL` for PETSc to control dynamic
3015: memory allocation. Likewise, specify preallocated storage for the
3016: off-diagonal part of the local submatrix with `o_nz` or `o_nnz` (not both).
3018: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
3019: the figure below we depict these three local rows and all columns (0-11).
3021: .vb
3022: 0 1 2 3 4 5 6 7 8 9 10 11
3023: --------------------------
3024: row 3 |o o o d d d o o o o o o
3025: row 4 |o o o d d d o o o o o o
3026: row 5 |o o o d d d o o o o o o
3027: --------------------------
3028: .ve
3030: Thus, any entries in the d locations are stored in the d (diagonal)
3031: submatrix, and any entries in the o locations are stored in the
3032: o (off-diagonal) submatrix. Note that the d and the o submatrices are
3033: stored simply in the `MATSEQBAIJ` format for compressed row storage.
3035: Now `d_nz` should indicate the number of block nonzeros per row in the d matrix,
3036: and `o_nz` should indicate the number of block nonzeros per row in the o matrix.
3037: In general, for PDE problems in which most nonzeros are near the diagonal,
3038: one expects `d_nz` >> `o_nz`.
3040: You can call `MatGetInfo()` to get information on how effective the preallocation was;
3041: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
3042: You can also run with the option `-info` and look for messages with the string
3043: malloc in them to see if additional memory allocation was needed.
3045: .seealso: `Mat`, `MATMPIBAIJ`, `MatCreate()`, `MatCreateSeqBAIJ()`, `MatSetValues()`, `MatCreateBAIJ()`, `MatMPIBAIJSetPreallocationCSR()`, `PetscSplitOwnership()`
3046: @*/
3047: PetscErrorCode MatMPIBAIJSetPreallocation(Mat B, PetscInt bs, PetscInt d_nz, const PetscInt d_nnz[], PetscInt o_nz, const PetscInt o_nnz[])
3048: {
3049: PetscFunctionBegin;
3053: PetscTryMethod(B, "MatMPIBAIJSetPreallocation_C", (Mat, PetscInt, PetscInt, const PetscInt[], PetscInt, const PetscInt[]), (B, bs, d_nz, d_nnz, o_nz, o_nnz));
3054: PetscFunctionReturn(PETSC_SUCCESS);
3055: }
3057: // PetscClangLinter pragma disable: -fdoc-section-header-unknown
3058: /*@C
3059: MatCreateBAIJ - Creates a sparse parallel matrix in `MATBAIJ` format
3060: (block compressed row).
3062: Collective
3064: Input Parameters:
3065: + comm - MPI communicator
3066: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
3067: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with `MatCreateVecs()`
3068: . m - number of local rows (or `PETSC_DECIDE` to have calculated if M is given)
3069: This value should be the same as the local size used in creating the
3070: y vector for the matrix-vector product y = Ax.
3071: . n - number of local columns (or `PETSC_DECIDE` to have calculated if N is given)
3072: This value should be the same as the local size used in creating the
3073: x vector for the matrix-vector product y = Ax.
3074: . M - number of global rows (or `PETSC_DETERMINE` to have calculated if m is given)
3075: . N - number of global columns (or `PETSC_DETERMINE` to have calculated if n is given)
3076: . d_nz - number of nonzero blocks per block row in diagonal portion of local
3077: submatrix (same for all local rows)
3078: . d_nnz - array containing the number of nonzero blocks in the various block rows
3079: of the in diagonal portion of the local (possibly different for each block
3080: row) or NULL. If you plan to factor the matrix you must leave room for the diagonal entry
3081: and set it even if it is zero.
3082: . o_nz - number of nonzero blocks per block row in the off-diagonal portion of local
3083: submatrix (same for all local rows).
3084: - o_nnz - array containing the number of nonzero blocks in the various block rows of the
3085: off-diagonal portion of the local submatrix (possibly different for
3086: each block row) or NULL.
3088: Output Parameter:
3089: . A - the matrix
3091: Options Database Keys:
3092: + -mat_block_size - size of the blocks to use
3093: - -mat_use_hash_table <fact> - set hash table factor
3095: Level: intermediate
3097: Notes:
3098: It is recommended that one use `MatCreateFromOptions()` or the `MatCreate()`, `MatSetType()` and/or `MatSetFromOptions()`,
3099: MatXXXXSetPreallocation() paradigm instead of this routine directly.
3100: [MatXXXXSetPreallocation() is, for example, `MatSeqBAIJSetPreallocation()`]
3102: For good matrix assembly performance
3103: the user should preallocate the matrix storage by setting the parameters
3104: `d_nz` (or `d_nnz`) and `o_nz` (or `o_nnz`). By setting these parameters accurately,
3105: performance can be increased by more than a factor of 50.
3107: If the *_nnz parameter is given then the *_nz parameter is ignored
3109: A nonzero block is any block that as 1 or more nonzeros in it
3111: The user MUST specify either the local or global matrix dimensions
3112: (possibly both).
3114: If `PETSC_DECIDE` or `PETSC_DETERMINE` is used for a particular argument on one processor
3115: than it must be used on all processors that share the object for that argument.
3117: Storage Information:
3118: For a square global matrix we define each processor's diagonal portion
3119: to be its local rows and the corresponding columns (a square submatrix);
3120: each processor's off-diagonal portion encompasses the remainder of the
3121: local matrix (a rectangular submatrix).
3123: The user can specify preallocated storage for the diagonal part of
3124: the local submatrix with either d_nz or d_nnz (not both). Set
3125: `d_nz` = `PETSC_DEFAULT` and `d_nnz` = `NULL` for PETSc to control dynamic
3126: memory allocation. Likewise, specify preallocated storage for the
3127: off-diagonal part of the local submatrix with `o_nz` or `o_nnz` (not both).
3129: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
3130: the figure below we depict these three local rows and all columns (0-11).
3132: .vb
3133: 0 1 2 3 4 5 6 7 8 9 10 11
3134: --------------------------
3135: row 3 |o o o d d d o o o o o o
3136: row 4 |o o o d d d o o o o o o
3137: row 5 |o o o d d d o o o o o o
3138: --------------------------
3139: .ve
3141: Thus, any entries in the d locations are stored in the d (diagonal)
3142: submatrix, and any entries in the o locations are stored in the
3143: o (off-diagonal) submatrix. Note that the d and the o submatrices are
3144: stored simply in the `MATSEQBAIJ` format for compressed row storage.
3146: Now `d_nz` should indicate the number of block nonzeros per row in the d matrix,
3147: and `o_nz` should indicate the number of block nonzeros per row in the o matrix.
3148: In general, for PDE problems in which most nonzeros are near the diagonal,
3149: one expects `d_nz` >> `o_nz`.
3151: .seealso: `Mat`, `MatCreate()`, `MatCreateSeqBAIJ()`, `MatSetValues()`, `MatMPIBAIJSetPreallocation()`, `MatMPIBAIJSetPreallocationCSR()`
3152: @*/
3153: PetscErrorCode MatCreateBAIJ(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt M, PetscInt N, PetscInt d_nz, const PetscInt d_nnz[], PetscInt o_nz, const PetscInt o_nnz[], Mat *A)
3154: {
3155: PetscMPIInt size;
3157: PetscFunctionBegin;
3158: PetscCall(MatCreate(comm, A));
3159: PetscCall(MatSetSizes(*A, m, n, M, N));
3160: PetscCallMPI(MPI_Comm_size(comm, &size));
3161: if (size > 1) {
3162: PetscCall(MatSetType(*A, MATMPIBAIJ));
3163: PetscCall(MatMPIBAIJSetPreallocation(*A, bs, d_nz, d_nnz, o_nz, o_nnz));
3164: } else {
3165: PetscCall(MatSetType(*A, MATSEQBAIJ));
3166: PetscCall(MatSeqBAIJSetPreallocation(*A, bs, d_nz, d_nnz));
3167: }
3168: PetscFunctionReturn(PETSC_SUCCESS);
3169: }
3171: static PetscErrorCode MatDuplicate_MPIBAIJ(Mat matin, MatDuplicateOption cpvalues, Mat *newmat)
3172: {
3173: Mat mat;
3174: Mat_MPIBAIJ *a, *oldmat = (Mat_MPIBAIJ *)matin->data;
3175: PetscInt len = 0;
3177: PetscFunctionBegin;
3178: *newmat = NULL;
3179: PetscCall(MatCreate(PetscObjectComm((PetscObject)matin), &mat));
3180: PetscCall(MatSetSizes(mat, matin->rmap->n, matin->cmap->n, matin->rmap->N, matin->cmap->N));
3181: PetscCall(MatSetType(mat, ((PetscObject)matin)->type_name));
3183: PetscCall(PetscLayoutReference(matin->rmap, &mat->rmap));
3184: PetscCall(PetscLayoutReference(matin->cmap, &mat->cmap));
3185: if (matin->hash_active) {
3186: PetscCall(MatSetUp(mat));
3187: } else {
3188: mat->factortype = matin->factortype;
3189: mat->preallocated = PETSC_TRUE;
3190: mat->assembled = PETSC_TRUE;
3191: mat->insertmode = NOT_SET_VALUES;
3193: a = (Mat_MPIBAIJ *)mat->data;
3194: mat->rmap->bs = matin->rmap->bs;
3195: a->bs2 = oldmat->bs2;
3196: a->mbs = oldmat->mbs;
3197: a->nbs = oldmat->nbs;
3198: a->Mbs = oldmat->Mbs;
3199: a->Nbs = oldmat->Nbs;
3201: a->size = oldmat->size;
3202: a->rank = oldmat->rank;
3203: a->donotstash = oldmat->donotstash;
3204: a->roworiented = oldmat->roworiented;
3205: a->rowindices = NULL;
3206: a->rowvalues = NULL;
3207: a->getrowactive = PETSC_FALSE;
3208: a->barray = NULL;
3209: a->rstartbs = oldmat->rstartbs;
3210: a->rendbs = oldmat->rendbs;
3211: a->cstartbs = oldmat->cstartbs;
3212: a->cendbs = oldmat->cendbs;
3214: /* hash table stuff */
3215: a->ht = NULL;
3216: a->hd = NULL;
3217: a->ht_size = 0;
3218: a->ht_flag = oldmat->ht_flag;
3219: a->ht_fact = oldmat->ht_fact;
3220: a->ht_total_ct = 0;
3221: a->ht_insert_ct = 0;
3223: PetscCall(PetscArraycpy(a->rangebs, oldmat->rangebs, a->size + 1));
3224: if (oldmat->colmap) {
3225: #if defined(PETSC_USE_CTABLE)
3226: PetscCall(PetscHMapIDuplicate(oldmat->colmap, &a->colmap));
3227: #else
3228: PetscCall(PetscMalloc1(a->Nbs, &a->colmap));
3229: PetscCall(PetscArraycpy(a->colmap, oldmat->colmap, a->Nbs));
3230: #endif
3231: } else a->colmap = NULL;
3233: if (oldmat->garray && (len = ((Mat_SeqBAIJ *)(oldmat->B->data))->nbs)) {
3234: PetscCall(PetscMalloc1(len, &a->garray));
3235: PetscCall(PetscArraycpy(a->garray, oldmat->garray, len));
3236: } else a->garray = NULL;
3238: PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)matin), matin->rmap->bs, &mat->bstash));
3239: PetscCall(VecDuplicate(oldmat->lvec, &a->lvec));
3240: PetscCall(VecScatterCopy(oldmat->Mvctx, &a->Mvctx));
3242: PetscCall(MatDuplicate(oldmat->A, cpvalues, &a->A));
3243: PetscCall(MatDuplicate(oldmat->B, cpvalues, &a->B));
3244: }
3245: PetscCall(PetscFunctionListDuplicate(((PetscObject)matin)->qlist, &((PetscObject)mat)->qlist));
3246: *newmat = mat;
3247: PetscFunctionReturn(PETSC_SUCCESS);
3248: }
3250: /* Used for both MPIBAIJ and MPISBAIJ matrices */
3251: PetscErrorCode MatLoad_MPIBAIJ_Binary(Mat mat, PetscViewer viewer)
3252: {
3253: PetscInt header[4], M, N, nz, bs, m, n, mbs, nbs, rows, cols, sum, i, j, k;
3254: PetscInt *rowidxs, *colidxs, rs, cs, ce;
3255: PetscScalar *matvals;
3257: PetscFunctionBegin;
3258: PetscCall(PetscViewerSetUp(viewer));
3260: /* read in matrix header */
3261: PetscCall(PetscViewerBinaryRead(viewer, header, 4, NULL, PETSC_INT));
3262: PetscCheck(header[0] == MAT_FILE_CLASSID, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Not a matrix object in file");
3263: M = header[1];
3264: N = header[2];
3265: nz = header[3];
3266: PetscCheck(M >= 0, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Matrix row size (%" PetscInt_FMT ") in file is negative", M);
3267: PetscCheck(N >= 0, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Matrix column size (%" PetscInt_FMT ") in file is negative", N);
3268: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Matrix stored in special format on disk, cannot load as MPIBAIJ");
3270: /* set block sizes from the viewer's .info file */
3271: PetscCall(MatLoad_Binary_BlockSizes(mat, viewer));
3272: /* set local sizes if not set already */
3273: if (mat->rmap->n < 0 && M == N) mat->rmap->n = mat->cmap->n;
3274: if (mat->cmap->n < 0 && M == N) mat->cmap->n = mat->rmap->n;
3275: /* set global sizes if not set already */
3276: if (mat->rmap->N < 0) mat->rmap->N = M;
3277: if (mat->cmap->N < 0) mat->cmap->N = N;
3278: PetscCall(PetscLayoutSetUp(mat->rmap));
3279: PetscCall(PetscLayoutSetUp(mat->cmap));
3281: /* check if the matrix sizes are correct */
3282: PetscCall(MatGetSize(mat, &rows, &cols));
3283: PetscCheck(M == rows && N == cols, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Matrix in file of different sizes (%" PetscInt_FMT ", %" PetscInt_FMT ") than the input matrix (%" PetscInt_FMT ", %" PetscInt_FMT ")", M, N, rows, cols);
3284: PetscCall(MatGetBlockSize(mat, &bs));
3285: PetscCall(MatGetLocalSize(mat, &m, &n));
3286: PetscCall(PetscLayoutGetRange(mat->rmap, &rs, NULL));
3287: PetscCall(PetscLayoutGetRange(mat->cmap, &cs, &ce));
3288: mbs = m / bs;
3289: nbs = n / bs;
3291: /* read in row lengths and build row indices */
3292: PetscCall(PetscMalloc1(m + 1, &rowidxs));
3293: PetscCall(PetscViewerBinaryReadAll(viewer, rowidxs + 1, m, PETSC_DECIDE, M, PETSC_INT));
3294: rowidxs[0] = 0;
3295: for (i = 0; i < m; i++) rowidxs[i + 1] += rowidxs[i];
3296: PetscCall(MPIU_Allreduce(&rowidxs[m], &sum, 1, MPIU_INT, MPI_SUM, PetscObjectComm((PetscObject)viewer)));
3297: PetscCheck(sum == nz, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Inconsistent matrix data in file: nonzeros = %" PetscInt_FMT ", sum-row-lengths = %" PetscInt_FMT, nz, sum);
3299: /* read in column indices and matrix values */
3300: PetscCall(PetscMalloc2(rowidxs[m], &colidxs, rowidxs[m], &matvals));
3301: PetscCall(PetscViewerBinaryReadAll(viewer, colidxs, rowidxs[m], PETSC_DETERMINE, PETSC_DETERMINE, PETSC_INT));
3302: PetscCall(PetscViewerBinaryReadAll(viewer, matvals, rowidxs[m], PETSC_DETERMINE, PETSC_DETERMINE, PETSC_SCALAR));
3304: { /* preallocate matrix storage */
3305: PetscBT bt; /* helper bit set to count diagonal nonzeros */
3306: PetscHSetI ht; /* helper hash set to count off-diagonal nonzeros */
3307: PetscBool sbaij, done;
3308: PetscInt *d_nnz, *o_nnz;
3310: PetscCall(PetscBTCreate(nbs, &bt));
3311: PetscCall(PetscHSetICreate(&ht));
3312: PetscCall(PetscCalloc2(mbs, &d_nnz, mbs, &o_nnz));
3313: PetscCall(PetscObjectTypeCompare((PetscObject)mat, MATMPISBAIJ, &sbaij));
3314: for (i = 0; i < mbs; i++) {
3315: PetscCall(PetscBTMemzero(nbs, bt));
3316: PetscCall(PetscHSetIClear(ht));
3317: for (k = 0; k < bs; k++) {
3318: PetscInt row = bs * i + k;
3319: for (j = rowidxs[row]; j < rowidxs[row + 1]; j++) {
3320: PetscInt col = colidxs[j];
3321: if (!sbaij || col >= row) {
3322: if (col >= cs && col < ce) {
3323: if (!PetscBTLookupSet(bt, (col - cs) / bs)) d_nnz[i]++;
3324: } else {
3325: PetscCall(PetscHSetIQueryAdd(ht, col / bs, &done));
3326: if (done) o_nnz[i]++;
3327: }
3328: }
3329: }
3330: }
3331: }
3332: PetscCall(PetscBTDestroy(&bt));
3333: PetscCall(PetscHSetIDestroy(&ht));
3334: PetscCall(MatMPIBAIJSetPreallocation(mat, bs, 0, d_nnz, 0, o_nnz));
3335: PetscCall(MatMPISBAIJSetPreallocation(mat, bs, 0, d_nnz, 0, o_nnz));
3336: PetscCall(PetscFree2(d_nnz, o_nnz));
3337: }
3339: /* store matrix values */
3340: for (i = 0; i < m; i++) {
3341: PetscInt row = rs + i, s = rowidxs[i], e = rowidxs[i + 1];
3342: PetscCall((*mat->ops->setvalues)(mat, 1, &row, e - s, colidxs + s, matvals + s, INSERT_VALUES));
3343: }
3345: PetscCall(PetscFree(rowidxs));
3346: PetscCall(PetscFree2(colidxs, matvals));
3347: PetscCall(MatAssemblyBegin(mat, MAT_FINAL_ASSEMBLY));
3348: PetscCall(MatAssemblyEnd(mat, MAT_FINAL_ASSEMBLY));
3349: PetscFunctionReturn(PETSC_SUCCESS);
3350: }
3352: PetscErrorCode MatLoad_MPIBAIJ(Mat mat, PetscViewer viewer)
3353: {
3354: PetscBool isbinary;
3356: PetscFunctionBegin;
3357: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
3358: 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);
3359: PetscCall(MatLoad_MPIBAIJ_Binary(mat, viewer));
3360: PetscFunctionReturn(PETSC_SUCCESS);
3361: }
3363: /*@
3364: MatMPIBAIJSetHashTableFactor - Sets the factor required to compute the size of the matrices hash table
3366: Input Parameters:
3367: + mat - the matrix
3368: - fact - factor
3370: Options Database Key:
3371: . -mat_use_hash_table <fact> - provide the factor
3373: Level: advanced
3375: .seealso: `Mat`, `MATMPIBAIJ`, `MatSetOption()`
3376: @*/
3377: PetscErrorCode MatMPIBAIJSetHashTableFactor(Mat mat, PetscReal fact)
3378: {
3379: PetscFunctionBegin;
3380: PetscTryMethod(mat, "MatSetHashTableFactor_C", (Mat, PetscReal), (mat, fact));
3381: PetscFunctionReturn(PETSC_SUCCESS);
3382: }
3384: PetscErrorCode MatSetHashTableFactor_MPIBAIJ(Mat mat, PetscReal fact)
3385: {
3386: Mat_MPIBAIJ *baij;
3388: PetscFunctionBegin;
3389: baij = (Mat_MPIBAIJ *)mat->data;
3390: baij->ht_fact = fact;
3391: PetscFunctionReturn(PETSC_SUCCESS);
3392: }
3394: PetscErrorCode MatMPIBAIJGetSeqBAIJ(Mat A, Mat *Ad, Mat *Ao, const PetscInt *colmap[])
3395: {
3396: Mat_MPIBAIJ *a = (Mat_MPIBAIJ *)A->data;
3397: PetscBool flg;
3399: PetscFunctionBegin;
3400: PetscCall(PetscObjectTypeCompare((PetscObject)A, MATMPIBAIJ, &flg));
3401: PetscCheck(flg, PetscObjectComm((PetscObject)A), PETSC_ERR_SUP, "This function requires a MATMPIBAIJ matrix as input");
3402: if (Ad) *Ad = a->A;
3403: if (Ao) *Ao = a->B;
3404: if (colmap) *colmap = a->garray;
3405: PetscFunctionReturn(PETSC_SUCCESS);
3406: }
3408: /*
3409: Special version for direct calls from Fortran (to eliminate two function call overheads
3410: */
3411: #if defined(PETSC_HAVE_FORTRAN_CAPS)
3412: #define matmpibaijsetvaluesblocked_ MATMPIBAIJSETVALUESBLOCKED
3413: #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE)
3414: #define matmpibaijsetvaluesblocked_ matmpibaijsetvaluesblocked
3415: #endif
3417: // PetscClangLinter pragma disable: -fdoc-synopsis-matching-symbol-name
3418: /*@C
3419: MatMPIBAIJSetValuesBlocked - Direct Fortran call to replace call to `MatSetValuesBlocked()`
3421: Collective
3423: Input Parameters:
3424: + matin - the matrix
3425: . min - number of input rows
3426: . im - input rows
3427: . nin - number of input columns
3428: . in - input columns
3429: . v - numerical values input
3430: - addvin - `INSERT_VALUES` or `ADD_VALUES`
3432: Level: advanced
3434: Developer Notes:
3435: This has a complete copy of `MatSetValuesBlocked_MPIBAIJ()` which is terrible code un-reuse.
3437: .seealso: `Mat`, `MatSetValuesBlocked()`
3438: @*/
3439: PETSC_EXTERN PetscErrorCode matmpibaijsetvaluesblocked_(Mat *matin, PetscInt *min, const PetscInt im[], PetscInt *nin, const PetscInt in[], const MatScalar v[], InsertMode *addvin)
3440: {
3441: /* convert input arguments to C version */
3442: Mat mat = *matin;
3443: PetscInt m = *min, n = *nin;
3444: InsertMode addv = *addvin;
3446: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ *)mat->data;
3447: const MatScalar *value;
3448: MatScalar *barray = baij->barray;
3449: PetscBool roworiented = baij->roworiented;
3450: PetscInt i, j, ii, jj, row, col, rstart = baij->rstartbs;
3451: PetscInt rend = baij->rendbs, cstart = baij->cstartbs, stepval;
3452: PetscInt cend = baij->cendbs, bs = mat->rmap->bs, bs2 = baij->bs2;
3454: PetscFunctionBegin;
3455: /* tasks normally handled by MatSetValuesBlocked() */
3456: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
3457: else PetscCheck(mat->insertmode == addv, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot mix add values and insert values");
3458: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3459: if (mat->assembled) {
3460: mat->was_assembled = PETSC_TRUE;
3461: mat->assembled = PETSC_FALSE;
3462: }
3463: PetscCall(PetscLogEventBegin(MAT_SetValues, mat, 0, 0, 0));
3465: if (!barray) {
3466: PetscCall(PetscMalloc1(bs2, &barray));
3467: baij->barray = barray;
3468: }
3470: if (roworiented) stepval = (n - 1) * bs;
3471: else stepval = (m - 1) * bs;
3473: for (i = 0; i < m; i++) {
3474: if (im[i] < 0) continue;
3475: PetscCheck(im[i] < baij->Mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large, row %" PetscInt_FMT " max %" PetscInt_FMT, im[i], baij->Mbs - 1);
3476: if (im[i] >= rstart && im[i] < rend) {
3477: row = im[i] - rstart;
3478: for (j = 0; j < n; j++) {
3479: /* If NumCol = 1 then a copy is not required */
3480: if ((roworiented) && (n == 1)) {
3481: barray = (MatScalar *)v + i * bs2;
3482: } else if ((!roworiented) && (m == 1)) {
3483: barray = (MatScalar *)v + j * bs2;
3484: } else { /* Here a copy is required */
3485: if (roworiented) {
3486: value = v + i * (stepval + bs) * bs + j * bs;
3487: } else {
3488: value = v + j * (stepval + bs) * bs + i * bs;
3489: }
3490: for (ii = 0; ii < bs; ii++, value += stepval) {
3491: for (jj = 0; jj < bs; jj++) *barray++ = *value++;
3492: }
3493: barray -= bs2;
3494: }
3496: if (in[j] >= cstart && in[j] < cend) {
3497: col = in[j] - cstart;
3498: PetscCall(MatSetValuesBlocked_SeqBAIJ_Inlined(baij->A, row, col, barray, addv, im[i], in[j]));
3499: } else if (in[j] < 0) {
3500: continue;
3501: } else {
3502: PetscCheck(in[j] < baij->Nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large, col %" PetscInt_FMT " max %" PetscInt_FMT, in[j], baij->Nbs - 1);
3503: if (mat->was_assembled) {
3504: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
3506: #if defined(PETSC_USE_DEBUG)
3507: #if defined(PETSC_USE_CTABLE)
3508: {
3509: PetscInt data;
3510: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] + 1, 0, &data));
3511: PetscCheck((data - 1) % bs == 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Incorrect colmap");
3512: }
3513: #else
3514: PetscCheck((baij->colmap[in[j]] - 1) % bs == 0, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Incorrect colmap");
3515: #endif
3516: #endif
3517: #if defined(PETSC_USE_CTABLE)
3518: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] + 1, 0, &col));
3519: col = (col - 1) / bs;
3520: #else
3521: col = (baij->colmap[in[j]] - 1) / bs;
3522: #endif
3523: if (col < 0 && !((Mat_SeqBAIJ *)(baij->A->data))->nonew) {
3524: PetscCall(MatDisAssemble_MPIBAIJ(mat));
3525: col = in[j];
3526: }
3527: } else col = in[j];
3528: PetscCall(MatSetValuesBlocked_SeqBAIJ_Inlined(baij->B, row, col, barray, addv, im[i], in[j]));
3529: }
3530: }
3531: } else {
3532: if (!baij->donotstash) {
3533: if (roworiented) {
3534: PetscCall(MatStashValuesRowBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
3535: } else {
3536: PetscCall(MatStashValuesColBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
3537: }
3538: }
3539: }
3540: }
3542: /* task normally handled by MatSetValuesBlocked() */
3543: PetscCall(PetscLogEventEnd(MAT_SetValues, mat, 0, 0, 0));
3544: PetscFunctionReturn(PETSC_SUCCESS);
3545: }
3547: /*@
3548: MatCreateMPIBAIJWithArrays - creates a `MATMPIBAIJ` matrix using arrays that contain in standard block
3549: CSR format the local rows.
3551: Collective
3553: Input Parameters:
3554: + comm - MPI communicator
3555: . bs - the block size, only a block size of 1 is supported
3556: . m - number of local rows (Cannot be `PETSC_DECIDE`)
3557: . n - This value should be the same as the local size used in creating the
3558: x vector for the matrix-vector product y = Ax. (or `PETSC_DECIDE` to have
3559: calculated if N is given) For square matrices n is almost always m.
3560: . M - number of global rows (or `PETSC_DETERMINE` to have calculated if m is given)
3561: . N - number of global columns (or `PETSC_DETERMINE` to have calculated if n is given)
3562: . i - row indices; that is i[0] = 0, i[row] = i[row-1] + number of block elements in that rowth block row of the matrix
3563: . j - column indices
3564: - a - matrix values
3566: Output Parameter:
3567: . mat - the matrix
3569: Level: intermediate
3571: Notes:
3572: The `i`, `j`, and `a` arrays ARE copied by this routine into the internal format used by PETSc;
3573: thus you CANNOT change the matrix entries by changing the values of a[] after you have
3574: called this routine. Use `MatCreateMPIAIJWithSplitArrays()` to avoid needing to copy the arrays.
3576: The order of the entries in values is the same as the block compressed sparse row storage format; that is, it is
3577: the same as a three dimensional array in Fortran values(bs,bs,nnz) that contains the first column of the first
3578: block, followed by the second column of the first block etc etc. That is, the blocks are contiguous in memory
3579: with column-major ordering within blocks.
3581: The `i` and `j` indices are 0 based, and i indices are indices corresponding to the local `j` array.
3583: .seealso: `Mat`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatMPIAIJSetPreallocation()`, `MatMPIAIJSetPreallocationCSR()`,
3584: `MATMPIAIJ`, `MatCreateAIJ()`, `MatCreateMPIAIJWithSplitArrays()`
3585: @*/
3586: PetscErrorCode MatCreateMPIBAIJWithArrays(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt M, PetscInt N, const PetscInt i[], const PetscInt j[], const PetscScalar a[], Mat *mat)
3587: {
3588: PetscFunctionBegin;
3589: PetscCheck(!i[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "i (row indices) must start with 0");
3590: PetscCheck(m >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "local number of rows (m) cannot be PETSC_DECIDE, or negative");
3591: PetscCall(MatCreate(comm, mat));
3592: PetscCall(MatSetSizes(*mat, m, n, M, N));
3593: PetscCall(MatSetType(*mat, MATMPIBAIJ));
3594: PetscCall(MatSetBlockSize(*mat, bs));
3595: PetscCall(MatSetUp(*mat));
3596: PetscCall(MatSetOption(*mat, MAT_ROW_ORIENTED, PETSC_FALSE));
3597: PetscCall(MatMPIBAIJSetPreallocationCSR(*mat, bs, i, j, a));
3598: PetscCall(MatSetOption(*mat, MAT_ROW_ORIENTED, PETSC_TRUE));
3599: PetscFunctionReturn(PETSC_SUCCESS);
3600: }
3602: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_MPIBAIJ(MPI_Comm comm, Mat inmat, PetscInt n, MatReuse scall, Mat *outmat)
3603: {
3604: PetscInt m, N, i, rstart, nnz, Ii, bs, cbs;
3605: PetscInt *indx;
3606: PetscScalar *values;
3608: PetscFunctionBegin;
3609: PetscCall(MatGetSize(inmat, &m, &N));
3610: if (scall == MAT_INITIAL_MATRIX) { /* symbolic phase */
3611: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)inmat->data;
3612: PetscInt *dnz, *onz, mbs, Nbs, nbs;
3613: PetscInt *bindx, rmax = a->rmax, j;
3614: PetscMPIInt rank, size;
3616: PetscCall(MatGetBlockSizes(inmat, &bs, &cbs));
3617: mbs = m / bs;
3618: Nbs = N / cbs;
3619: if (n == PETSC_DECIDE) PetscCall(PetscSplitOwnershipBlock(comm, cbs, &n, &N));
3620: nbs = n / cbs;
3622: PetscCall(PetscMalloc1(rmax, &bindx));
3623: MatPreallocateBegin(comm, mbs, nbs, dnz, onz); /* inline function, output __end and __rstart are used below */
3625: PetscCallMPI(MPI_Comm_rank(comm, &rank));
3626: PetscCallMPI(MPI_Comm_rank(comm, &size));
3627: if (rank == size - 1) {
3628: /* Check sum(nbs) = Nbs */
3629: PetscCheck(__end == Nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Sum of local block columns %" PetscInt_FMT " != global block columns %" PetscInt_FMT, __end, Nbs);
3630: }
3632: rstart = __rstart; /* block rstart of *outmat; see inline function MatPreallocateBegin */
3633: for (i = 0; i < mbs; i++) {
3634: PetscCall(MatGetRow_SeqBAIJ(inmat, i * bs, &nnz, &indx, NULL)); /* non-blocked nnz and indx */
3635: nnz = nnz / bs;
3636: for (j = 0; j < nnz; j++) bindx[j] = indx[j * bs] / bs;
3637: PetscCall(MatPreallocateSet(i + rstart, nnz, bindx, dnz, onz));
3638: PetscCall(MatRestoreRow_SeqBAIJ(inmat, i * bs, &nnz, &indx, NULL));
3639: }
3640: PetscCall(PetscFree(bindx));
3642: PetscCall(MatCreate(comm, outmat));
3643: PetscCall(MatSetSizes(*outmat, m, n, PETSC_DETERMINE, PETSC_DETERMINE));
3644: PetscCall(MatSetBlockSizes(*outmat, bs, cbs));
3645: PetscCall(MatSetType(*outmat, MATBAIJ));
3646: PetscCall(MatSeqBAIJSetPreallocation(*outmat, bs, 0, dnz));
3647: PetscCall(MatMPIBAIJSetPreallocation(*outmat, bs, 0, dnz, 0, onz));
3648: MatPreallocateEnd(dnz, onz);
3649: PetscCall(MatSetOption(*outmat, MAT_NO_OFF_PROC_ENTRIES, PETSC_TRUE));
3650: }
3652: /* numeric phase */
3653: PetscCall(MatGetBlockSizes(inmat, &bs, &cbs));
3654: PetscCall(MatGetOwnershipRange(*outmat, &rstart, NULL));
3656: for (i = 0; i < m; i++) {
3657: PetscCall(MatGetRow_SeqBAIJ(inmat, i, &nnz, &indx, &values));
3658: Ii = i + rstart;
3659: PetscCall(MatSetValues(*outmat, 1, &Ii, nnz, indx, values, INSERT_VALUES));
3660: PetscCall(MatRestoreRow_SeqBAIJ(inmat, i, &nnz, &indx, &values));
3661: }
3662: PetscCall(MatAssemblyBegin(*outmat, MAT_FINAL_ASSEMBLY));
3663: PetscCall(MatAssemblyEnd(*outmat, MAT_FINAL_ASSEMBLY));
3664: PetscFunctionReturn(PETSC_SUCCESS);
3665: }