Actual source code: aij.c
1: /*
2: Defines the basic matrix operations for the AIJ (compressed row)
3: matrix storage format.
4: */
6: #include <../src/mat/impls/aij/seq/aij.h>
7: #include <petscblaslapack.h>
8: #include <petscbt.h>
9: #include <petsc/private/kernels/blocktranspose.h>
11: /* defines MatSetValues_Seq_Hash(), MatAssemblyEnd_Seq_Hash(), MatSetUp_Seq_Hash() */
12: #define TYPE AIJ
13: #define TYPE_BS
14: #include "../src/mat/impls/aij/seq/seqhashmatsetvalues.h"
15: #include "../src/mat/impls/aij/seq/seqhashmat.h"
16: #undef TYPE
17: #undef TYPE_BS
19: static PetscErrorCode MatSeqAIJSetTypeFromOptions(Mat A)
20: {
21: PetscBool flg;
22: char type[256];
24: PetscFunctionBegin;
25: PetscObjectOptionsBegin((PetscObject)A);
26: PetscCall(PetscOptionsFList("-mat_seqaij_type", "Matrix SeqAIJ type", "MatSeqAIJSetType", MatSeqAIJList, "seqaij", type, 256, &flg));
27: if (flg) PetscCall(MatSeqAIJSetType(A, type));
28: PetscOptionsEnd();
29: PetscFunctionReturn(PETSC_SUCCESS);
30: }
32: static PetscErrorCode MatGetColumnReductions_SeqAIJ(Mat A, PetscInt type, PetscReal *reductions)
33: {
34: PetscInt i, m, n;
35: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
37: PetscFunctionBegin;
38: PetscCall(MatGetSize(A, &m, &n));
39: PetscCall(PetscArrayzero(reductions, n));
40: if (type == NORM_2) {
41: for (i = 0; i < aij->i[m]; i++) reductions[aij->j[i]] += PetscAbsScalar(aij->a[i] * aij->a[i]);
42: } else if (type == NORM_1) {
43: for (i = 0; i < aij->i[m]; i++) reductions[aij->j[i]] += PetscAbsScalar(aij->a[i]);
44: } else if (type == NORM_INFINITY) {
45: for (i = 0; i < aij->i[m]; i++) reductions[aij->j[i]] = PetscMax(PetscAbsScalar(aij->a[i]), reductions[aij->j[i]]);
46: } else if (type == REDUCTION_SUM_REALPART || type == REDUCTION_MEAN_REALPART) {
47: for (i = 0; i < aij->i[m]; i++) reductions[aij->j[i]] += PetscRealPart(aij->a[i]);
48: } else if (type == REDUCTION_SUM_IMAGINARYPART || type == REDUCTION_MEAN_IMAGINARYPART) {
49: for (i = 0; i < aij->i[m]; i++) reductions[aij->j[i]] += PetscImaginaryPart(aij->a[i]);
50: } else SETERRQ(PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONG, "Unknown reduction type");
52: if (type == NORM_2) {
53: for (i = 0; i < n; i++) reductions[i] = PetscSqrtReal(reductions[i]);
54: } else if (type == REDUCTION_MEAN_REALPART || type == REDUCTION_MEAN_IMAGINARYPART) {
55: for (i = 0; i < n; i++) reductions[i] /= m;
56: }
57: PetscFunctionReturn(PETSC_SUCCESS);
58: }
60: static PetscErrorCode MatFindOffBlockDiagonalEntries_SeqAIJ(Mat A, IS *is)
61: {
62: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
63: PetscInt i, m = A->rmap->n, cnt = 0, bs = A->rmap->bs;
64: const PetscInt *jj = a->j, *ii = a->i;
65: PetscInt *rows;
67: PetscFunctionBegin;
68: for (i = 0; i < m; i++) {
69: if ((ii[i] != ii[i + 1]) && ((jj[ii[i]] < bs * (i / bs)) || (jj[ii[i + 1] - 1] > bs * ((i + bs) / bs) - 1))) cnt++;
70: }
71: PetscCall(PetscMalloc1(cnt, &rows));
72: cnt = 0;
73: for (i = 0; i < m; i++) {
74: if ((ii[i] != ii[i + 1]) && ((jj[ii[i]] < bs * (i / bs)) || (jj[ii[i + 1] - 1] > bs * ((i + bs) / bs) - 1))) {
75: rows[cnt] = i;
76: cnt++;
77: }
78: }
79: PetscCall(ISCreateGeneral(PETSC_COMM_SELF, cnt, rows, PETSC_OWN_POINTER, is));
80: PetscFunctionReturn(PETSC_SUCCESS);
81: }
83: PetscErrorCode MatFindZeroDiagonals_SeqAIJ_Private(Mat A, PetscInt *nrows, PetscInt **zrows)
84: {
85: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
86: const MatScalar *aa;
87: PetscInt i, m = A->rmap->n, cnt = 0;
88: const PetscInt *ii = a->i, *jj = a->j, *diag;
89: PetscInt *rows;
91: PetscFunctionBegin;
92: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
93: PetscCall(MatMarkDiagonal_SeqAIJ(A));
94: diag = a->diag;
95: for (i = 0; i < m; i++) {
96: if ((diag[i] >= ii[i + 1]) || (jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) cnt++;
97: }
98: PetscCall(PetscMalloc1(cnt, &rows));
99: cnt = 0;
100: for (i = 0; i < m; i++) {
101: if ((diag[i] >= ii[i + 1]) || (jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) rows[cnt++] = i;
102: }
103: *nrows = cnt;
104: *zrows = rows;
105: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
106: PetscFunctionReturn(PETSC_SUCCESS);
107: }
109: static PetscErrorCode MatFindZeroDiagonals_SeqAIJ(Mat A, IS *zrows)
110: {
111: PetscInt nrows, *rows;
113: PetscFunctionBegin;
114: *zrows = NULL;
115: PetscCall(MatFindZeroDiagonals_SeqAIJ_Private(A, &nrows, &rows));
116: PetscCall(ISCreateGeneral(PetscObjectComm((PetscObject)A), nrows, rows, PETSC_OWN_POINTER, zrows));
117: PetscFunctionReturn(PETSC_SUCCESS);
118: }
120: static PetscErrorCode MatFindNonzeroRows_SeqAIJ(Mat A, IS *keptrows)
121: {
122: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
123: const MatScalar *aa;
124: PetscInt m = A->rmap->n, cnt = 0;
125: const PetscInt *ii;
126: PetscInt n, i, j, *rows;
128: PetscFunctionBegin;
129: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
130: *keptrows = NULL;
131: ii = a->i;
132: for (i = 0; i < m; i++) {
133: n = ii[i + 1] - ii[i];
134: if (!n) {
135: cnt++;
136: goto ok1;
137: }
138: for (j = ii[i]; j < ii[i + 1]; j++) {
139: if (aa[j] != 0.0) goto ok1;
140: }
141: cnt++;
142: ok1:;
143: }
144: if (!cnt) {
145: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
146: PetscFunctionReturn(PETSC_SUCCESS);
147: }
148: PetscCall(PetscMalloc1(A->rmap->n - cnt, &rows));
149: cnt = 0;
150: for (i = 0; i < m; i++) {
151: n = ii[i + 1] - ii[i];
152: if (!n) continue;
153: for (j = ii[i]; j < ii[i + 1]; j++) {
154: if (aa[j] != 0.0) {
155: rows[cnt++] = i;
156: break;
157: }
158: }
159: }
160: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
161: PetscCall(ISCreateGeneral(PETSC_COMM_SELF, cnt, rows, PETSC_OWN_POINTER, keptrows));
162: PetscFunctionReturn(PETSC_SUCCESS);
163: }
165: PetscErrorCode MatDiagonalSet_SeqAIJ(Mat Y, Vec D, InsertMode is)
166: {
167: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)Y->data;
168: PetscInt i, m = Y->rmap->n;
169: const PetscInt *diag;
170: MatScalar *aa;
171: const PetscScalar *v;
172: PetscBool missing;
174: PetscFunctionBegin;
175: if (Y->assembled) {
176: PetscCall(MatMissingDiagonal_SeqAIJ(Y, &missing, NULL));
177: if (!missing) {
178: diag = aij->diag;
179: PetscCall(VecGetArrayRead(D, &v));
180: PetscCall(MatSeqAIJGetArray(Y, &aa));
181: if (is == INSERT_VALUES) {
182: for (i = 0; i < m; i++) aa[diag[i]] = v[i];
183: } else {
184: for (i = 0; i < m; i++) aa[diag[i]] += v[i];
185: }
186: PetscCall(MatSeqAIJRestoreArray(Y, &aa));
187: PetscCall(VecRestoreArrayRead(D, &v));
188: PetscFunctionReturn(PETSC_SUCCESS);
189: }
190: PetscCall(MatSeqAIJInvalidateDiagonal(Y));
191: }
192: PetscCall(MatDiagonalSet_Default(Y, D, is));
193: PetscFunctionReturn(PETSC_SUCCESS);
194: }
196: PetscErrorCode MatGetRowIJ_SeqAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *m, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
197: {
198: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
199: PetscInt i, ishift;
201: PetscFunctionBegin;
202: if (m) *m = A->rmap->n;
203: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
204: ishift = 0;
205: if (symmetric && A->structurally_symmetric != PETSC_BOOL3_TRUE) {
206: PetscCall(MatToSymmetricIJ_SeqAIJ(A->rmap->n, a->i, a->j, PETSC_TRUE, ishift, oshift, (PetscInt **)ia, (PetscInt **)ja));
207: } else if (oshift == 1) {
208: PetscInt *tia;
209: PetscInt nz = a->i[A->rmap->n];
210: /* malloc space and add 1 to i and j indices */
211: PetscCall(PetscMalloc1(A->rmap->n + 1, &tia));
212: for (i = 0; i < A->rmap->n + 1; i++) tia[i] = a->i[i] + 1;
213: *ia = tia;
214: if (ja) {
215: PetscInt *tja;
216: PetscCall(PetscMalloc1(nz + 1, &tja));
217: for (i = 0; i < nz; i++) tja[i] = a->j[i] + 1;
218: *ja = tja;
219: }
220: } else {
221: *ia = a->i;
222: if (ja) *ja = a->j;
223: }
224: PetscFunctionReturn(PETSC_SUCCESS);
225: }
227: PetscErrorCode MatRestoreRowIJ_SeqAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
228: {
229: PetscFunctionBegin;
230: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
231: if ((symmetric && A->structurally_symmetric != PETSC_BOOL3_TRUE) || oshift == 1) {
232: PetscCall(PetscFree(*ia));
233: if (ja) PetscCall(PetscFree(*ja));
234: }
235: PetscFunctionReturn(PETSC_SUCCESS);
236: }
238: PetscErrorCode MatGetColumnIJ_SeqAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *nn, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
239: {
240: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
241: PetscInt i, *collengths, *cia, *cja, n = A->cmap->n, m = A->rmap->n;
242: PetscInt nz = a->i[m], row, *jj, mr, col;
244: PetscFunctionBegin;
245: *nn = n;
246: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
247: if (symmetric) {
248: PetscCall(MatToSymmetricIJ_SeqAIJ(A->rmap->n, a->i, a->j, PETSC_TRUE, 0, oshift, (PetscInt **)ia, (PetscInt **)ja));
249: } else {
250: PetscCall(PetscCalloc1(n, &collengths));
251: PetscCall(PetscMalloc1(n + 1, &cia));
252: PetscCall(PetscMalloc1(nz, &cja));
253: jj = a->j;
254: for (i = 0; i < nz; i++) collengths[jj[i]]++;
255: cia[0] = oshift;
256: for (i = 0; i < n; i++) cia[i + 1] = cia[i] + collengths[i];
257: PetscCall(PetscArrayzero(collengths, n));
258: jj = a->j;
259: for (row = 0; row < m; row++) {
260: mr = a->i[row + 1] - a->i[row];
261: for (i = 0; i < mr; i++) {
262: col = *jj++;
264: cja[cia[col] + collengths[col]++ - oshift] = row + oshift;
265: }
266: }
267: PetscCall(PetscFree(collengths));
268: *ia = cia;
269: *ja = cja;
270: }
271: PetscFunctionReturn(PETSC_SUCCESS);
272: }
274: PetscErrorCode MatRestoreColumnIJ_SeqAIJ(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscBool *done)
275: {
276: PetscFunctionBegin;
277: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
279: PetscCall(PetscFree(*ia));
280: PetscCall(PetscFree(*ja));
281: PetscFunctionReturn(PETSC_SUCCESS);
282: }
284: /*
285: MatGetColumnIJ_SeqAIJ_Color() and MatRestoreColumnIJ_SeqAIJ_Color() are customized from
286: MatGetColumnIJ_SeqAIJ() and MatRestoreColumnIJ_SeqAIJ() by adding an output
287: spidx[], index of a->a, to be used in MatTransposeColoringCreate_SeqAIJ() and MatFDColoringCreate_SeqXAIJ()
288: */
289: PetscErrorCode MatGetColumnIJ_SeqAIJ_Color(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *nn, const PetscInt *ia[], const PetscInt *ja[], PetscInt *spidx[], PetscBool *done)
290: {
291: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
292: PetscInt i, *collengths, *cia, *cja, n = A->cmap->n, m = A->rmap->n;
293: PetscInt nz = a->i[m], row, mr, col, tmp;
294: PetscInt *cspidx;
295: const PetscInt *jj;
297: PetscFunctionBegin;
298: *nn = n;
299: if (!ia) PetscFunctionReturn(PETSC_SUCCESS);
301: PetscCall(PetscCalloc1(n, &collengths));
302: PetscCall(PetscMalloc1(n + 1, &cia));
303: PetscCall(PetscMalloc1(nz, &cja));
304: PetscCall(PetscMalloc1(nz, &cspidx));
305: jj = a->j;
306: for (i = 0; i < nz; i++) collengths[jj[i]]++;
307: cia[0] = oshift;
308: for (i = 0; i < n; i++) cia[i + 1] = cia[i] + collengths[i];
309: PetscCall(PetscArrayzero(collengths, n));
310: jj = a->j;
311: for (row = 0; row < m; row++) {
312: mr = a->i[row + 1] - a->i[row];
313: for (i = 0; i < mr; i++) {
314: col = *jj++;
315: tmp = cia[col] + collengths[col]++ - oshift;
316: cspidx[tmp] = a->i[row] + i; /* index of a->j */
317: cja[tmp] = row + oshift;
318: }
319: }
320: PetscCall(PetscFree(collengths));
321: *ia = cia;
322: *ja = cja;
323: *spidx = cspidx;
324: PetscFunctionReturn(PETSC_SUCCESS);
325: }
327: PetscErrorCode MatRestoreColumnIJ_SeqAIJ_Color(Mat A, PetscInt oshift, PetscBool symmetric, PetscBool inodecompressed, PetscInt *n, const PetscInt *ia[], const PetscInt *ja[], PetscInt *spidx[], PetscBool *done)
328: {
329: PetscFunctionBegin;
330: PetscCall(MatRestoreColumnIJ_SeqAIJ(A, oshift, symmetric, inodecompressed, n, ia, ja, done));
331: PetscCall(PetscFree(*spidx));
332: PetscFunctionReturn(PETSC_SUCCESS);
333: }
335: static PetscErrorCode MatSetValuesRow_SeqAIJ(Mat A, PetscInt row, const PetscScalar v[])
336: {
337: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
338: PetscInt *ai = a->i;
339: PetscScalar *aa;
341: PetscFunctionBegin;
342: PetscCall(MatSeqAIJGetArray(A, &aa));
343: PetscCall(PetscArraycpy(aa + ai[row], v, ai[row + 1] - ai[row]));
344: PetscCall(MatSeqAIJRestoreArray(A, &aa));
345: PetscFunctionReturn(PETSC_SUCCESS);
346: }
348: /*
349: MatSeqAIJSetValuesLocalFast - An optimized version of MatSetValuesLocal() for SeqAIJ matrices with several assumptions
351: - a single row of values is set with each call
352: - no row or column indices are negative or (in error) larger than the number of rows or columns
353: - the values are always added to the matrix, not set
354: - no new locations are introduced in the nonzero structure of the matrix
356: This does NOT assume the global column indices are sorted
358: */
360: #include <petsc/private/isimpl.h>
361: PetscErrorCode MatSeqAIJSetValuesLocalFast(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
362: {
363: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
364: PetscInt low, high, t, row, nrow, i, col, l;
365: const PetscInt *rp, *ai = a->i, *ailen = a->ilen, *aj = a->j;
366: PetscInt lastcol = -1;
367: MatScalar *ap, value, *aa;
368: const PetscInt *ridx = A->rmap->mapping->indices, *cidx = A->cmap->mapping->indices;
370: PetscFunctionBegin;
371: PetscCall(MatSeqAIJGetArray(A, &aa));
372: row = ridx[im[0]];
373: rp = aj + ai[row];
374: ap = aa + ai[row];
375: nrow = ailen[row];
376: low = 0;
377: high = nrow;
378: for (l = 0; l < n; l++) { /* loop over added columns */
379: col = cidx[in[l]];
380: value = v[l];
382: if (col <= lastcol) low = 0;
383: else high = nrow;
384: lastcol = col;
385: while (high - low > 5) {
386: t = (low + high) / 2;
387: if (rp[t] > col) high = t;
388: else low = t;
389: }
390: for (i = low; i < high; i++) {
391: if (rp[i] == col) {
392: ap[i] += value;
393: low = i + 1;
394: break;
395: }
396: }
397: }
398: PetscCall(MatSeqAIJRestoreArray(A, &aa));
399: return PETSC_SUCCESS;
400: }
402: PetscErrorCode MatSetValues_SeqAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
403: {
404: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
405: PetscInt *rp, k, low, high, t, ii, row, nrow, i, col, l, rmax, N;
406: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
407: PetscInt *aj = a->j, nonew = a->nonew, lastcol = -1;
408: MatScalar *ap = NULL, value = 0.0, *aa;
409: PetscBool ignorezeroentries = a->ignorezeroentries;
410: PetscBool roworiented = a->roworiented;
412: PetscFunctionBegin;
413: PetscCall(MatSeqAIJGetArray(A, &aa));
414: for (k = 0; k < m; k++) { /* loop over added rows */
415: row = im[k];
416: if (row < 0) continue;
417: PetscCheck(row < A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, row, A->rmap->n - 1);
418: rp = PetscSafePointerPlusOffset(aj, ai[row]);
419: if (!A->structure_only) ap = PetscSafePointerPlusOffset(aa, ai[row]);
420: rmax = imax[row];
421: nrow = ailen[row];
422: low = 0;
423: high = nrow;
424: for (l = 0; l < n; l++) { /* loop over added columns */
425: if (in[l] < 0) continue;
426: PetscCheck(in[l] < A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[l], A->cmap->n - 1);
427: col = in[l];
428: if (v && !A->structure_only) value = roworiented ? v[l + k * n] : v[k + l * m];
429: if (!A->structure_only && value == 0.0 && ignorezeroentries && is == ADD_VALUES && row != col) continue;
431: if (col <= lastcol) low = 0;
432: else high = nrow;
433: lastcol = col;
434: while (high - low > 5) {
435: t = (low + high) / 2;
436: if (rp[t] > col) high = t;
437: else low = t;
438: }
439: for (i = low; i < high; i++) {
440: if (rp[i] > col) break;
441: if (rp[i] == col) {
442: if (!A->structure_only) {
443: if (is == ADD_VALUES) {
444: ap[i] += value;
445: (void)PetscLogFlops(1.0);
446: } else ap[i] = value;
447: }
448: low = i + 1;
449: goto noinsert;
450: }
451: }
452: if (value == 0.0 && ignorezeroentries && row != col) goto noinsert;
453: if (nonew == 1) goto noinsert;
454: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero at (%" PetscInt_FMT ",%" PetscInt_FMT ") in the matrix", row, col);
455: if (A->structure_only) {
456: MatSeqXAIJReallocateAIJ_structure_only(A, A->rmap->n, 1, nrow, row, col, rmax, ai, aj, rp, imax, nonew, MatScalar);
457: } else {
458: MatSeqXAIJReallocateAIJ(A, A->rmap->n, 1, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
459: }
460: N = nrow++ - 1;
461: a->nz++;
462: high++;
463: /* shift up all the later entries in this row */
464: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
465: rp[i] = col;
466: if (!A->structure_only) {
467: PetscCall(PetscArraymove(ap + i + 1, ap + i, N - i + 1));
468: ap[i] = value;
469: }
470: low = i + 1;
471: noinsert:;
472: }
473: ailen[row] = nrow;
474: }
475: PetscCall(MatSeqAIJRestoreArray(A, &aa));
476: PetscFunctionReturn(PETSC_SUCCESS);
477: }
479: static PetscErrorCode MatSetValues_SeqAIJ_SortedFullNoPreallocation(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
480: {
481: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
482: PetscInt *rp, k, row;
483: PetscInt *ai = a->i;
484: PetscInt *aj = a->j;
485: MatScalar *aa, *ap;
487: PetscFunctionBegin;
488: PetscCheck(!A->was_assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot call on assembled matrix.");
489: PetscCheck(m * n + a->nz <= a->maxnz, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Number of entries in matrix will be larger than maximum nonzeros allocated for %" PetscInt_FMT " in MatSeqAIJSetTotalPreallocation()", a->maxnz);
491: PetscCall(MatSeqAIJGetArray(A, &aa));
492: for (k = 0; k < m; k++) { /* loop over added rows */
493: row = im[k];
494: rp = aj + ai[row];
495: ap = PetscSafePointerPlusOffset(aa, ai[row]);
497: PetscCall(PetscMemcpy(rp, in, n * sizeof(PetscInt)));
498: if (!A->structure_only) {
499: if (v) {
500: PetscCall(PetscMemcpy(ap, v, n * sizeof(PetscScalar)));
501: v += n;
502: } else {
503: PetscCall(PetscMemzero(ap, n * sizeof(PetscScalar)));
504: }
505: }
506: a->ilen[row] = n;
507: a->imax[row] = n;
508: a->i[row + 1] = a->i[row] + n;
509: a->nz += n;
510: }
511: PetscCall(MatSeqAIJRestoreArray(A, &aa));
512: PetscFunctionReturn(PETSC_SUCCESS);
513: }
515: /*@
516: MatSeqAIJSetTotalPreallocation - Sets an upper bound on the total number of expected nonzeros in the matrix.
518: Input Parameters:
519: + A - the `MATSEQAIJ` matrix
520: - nztotal - bound on the number of nonzeros
522: Level: advanced
524: Notes:
525: This can be called if you will be provided the matrix row by row (from row zero) with sorted column indices for each row.
526: Simply call `MatSetValues()` after this call to provide the matrix entries in the usual manner. This matrix may be used
527: as always with multiple matrix assemblies.
529: .seealso: [](ch_matrices), `Mat`, `MatSetOption()`, `MAT_SORTED_FULL`, `MatSetValues()`, `MatSeqAIJSetPreallocation()`
530: @*/
531: PetscErrorCode MatSeqAIJSetTotalPreallocation(Mat A, PetscInt nztotal)
532: {
533: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
535: PetscFunctionBegin;
536: PetscCall(PetscLayoutSetUp(A->rmap));
537: PetscCall(PetscLayoutSetUp(A->cmap));
538: a->maxnz = nztotal;
539: if (!a->imax) { PetscCall(PetscMalloc1(A->rmap->n, &a->imax)); }
540: if (!a->ilen) {
541: PetscCall(PetscMalloc1(A->rmap->n, &a->ilen));
542: } else {
543: PetscCall(PetscMemzero(a->ilen, A->rmap->n * sizeof(PetscInt)));
544: }
546: /* allocate the matrix space */
547: PetscCall(PetscShmgetAllocateArray(A->rmap->n + 1, sizeof(PetscInt), (void **)&a->i));
548: PetscCall(PetscShmgetAllocateArray(nztotal, sizeof(PetscInt), (void **)&a->j));
549: a->free_ij = PETSC_TRUE;
550: if (A->structure_only) {
551: a->free_a = PETSC_FALSE;
552: } else {
553: PetscCall(PetscShmgetAllocateArray(nztotal, sizeof(PetscScalar), (void **)&a->a));
554: a->free_a = PETSC_TRUE;
555: }
556: a->i[0] = 0;
557: A->ops->setvalues = MatSetValues_SeqAIJ_SortedFullNoPreallocation;
558: A->preallocated = PETSC_TRUE;
559: PetscFunctionReturn(PETSC_SUCCESS);
560: }
562: static PetscErrorCode MatSetValues_SeqAIJ_SortedFull(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode is)
563: {
564: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
565: PetscInt *rp, k, row;
566: PetscInt *ai = a->i, *ailen = a->ilen;
567: PetscInt *aj = a->j;
568: MatScalar *aa, *ap;
570: PetscFunctionBegin;
571: PetscCall(MatSeqAIJGetArray(A, &aa));
572: for (k = 0; k < m; k++) { /* loop over added rows */
573: row = im[k];
574: PetscCheck(n <= a->imax[row], PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Preallocation for row %" PetscInt_FMT " does not match number of columns provided", n);
575: rp = aj + ai[row];
576: ap = aa + ai[row];
577: if (!A->was_assembled) PetscCall(PetscMemcpy(rp, in, n * sizeof(PetscInt)));
578: if (!A->structure_only) {
579: if (v) {
580: PetscCall(PetscMemcpy(ap, v, n * sizeof(PetscScalar)));
581: v += n;
582: } else {
583: PetscCall(PetscMemzero(ap, n * sizeof(PetscScalar)));
584: }
585: }
586: ailen[row] = n;
587: a->nz += n;
588: }
589: PetscCall(MatSeqAIJRestoreArray(A, &aa));
590: PetscFunctionReturn(PETSC_SUCCESS);
591: }
593: static PetscErrorCode MatGetValues_SeqAIJ(Mat A, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], PetscScalar v[])
594: {
595: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
596: PetscInt *rp, k, low, high, t, row, nrow, i, col, l, *aj = a->j;
597: PetscInt *ai = a->i, *ailen = a->ilen;
598: const MatScalar *ap, *aa;
600: PetscFunctionBegin;
601: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
602: for (k = 0; k < m; k++) { /* loop over rows */
603: row = im[k];
604: if (row < 0) {
605: v += n;
606: continue;
607: } /* negative row */
608: PetscCheck(row < A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, row, A->rmap->n - 1);
609: rp = PetscSafePointerPlusOffset(aj, ai[row]);
610: ap = PetscSafePointerPlusOffset(aa, ai[row]);
611: nrow = ailen[row];
612: for (l = 0; l < n; l++) { /* loop over columns */
613: if (in[l] < 0) {
614: v++;
615: continue;
616: } /* negative column */
617: PetscCheck(in[l] < A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[l], A->cmap->n - 1);
618: col = in[l];
619: high = nrow;
620: low = 0; /* assume unsorted */
621: while (high - low > 5) {
622: t = (low + high) / 2;
623: if (rp[t] > col) high = t;
624: else low = t;
625: }
626: for (i = low; i < high; i++) {
627: if (rp[i] > col) break;
628: if (rp[i] == col) {
629: *v++ = ap[i];
630: goto finished;
631: }
632: }
633: *v++ = 0.0;
634: finished:;
635: }
636: }
637: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
638: PetscFunctionReturn(PETSC_SUCCESS);
639: }
641: static PetscErrorCode MatView_SeqAIJ_Binary(Mat mat, PetscViewer viewer)
642: {
643: Mat_SeqAIJ *A = (Mat_SeqAIJ *)mat->data;
644: const PetscScalar *av;
645: PetscInt header[4], M, N, m, nz, i;
646: PetscInt *rowlens;
648: PetscFunctionBegin;
649: PetscCall(PetscViewerSetUp(viewer));
651: M = mat->rmap->N;
652: N = mat->cmap->N;
653: m = mat->rmap->n;
654: nz = A->nz;
656: /* write matrix header */
657: header[0] = MAT_FILE_CLASSID;
658: header[1] = M;
659: header[2] = N;
660: header[3] = nz;
661: PetscCall(PetscViewerBinaryWrite(viewer, header, 4, PETSC_INT));
663: /* fill in and store row lengths */
664: PetscCall(PetscMalloc1(m, &rowlens));
665: for (i = 0; i < m; i++) rowlens[i] = A->i[i + 1] - A->i[i];
666: PetscCall(PetscViewerBinaryWrite(viewer, rowlens, m, PETSC_INT));
667: PetscCall(PetscFree(rowlens));
668: /* store column indices */
669: PetscCall(PetscViewerBinaryWrite(viewer, A->j, nz, PETSC_INT));
670: /* store nonzero values */
671: PetscCall(MatSeqAIJGetArrayRead(mat, &av));
672: PetscCall(PetscViewerBinaryWrite(viewer, av, nz, PETSC_SCALAR));
673: PetscCall(MatSeqAIJRestoreArrayRead(mat, &av));
675: /* write block size option to the viewer's .info file */
676: PetscCall(MatView_Binary_BlockSizes(mat, viewer));
677: PetscFunctionReturn(PETSC_SUCCESS);
678: }
680: static PetscErrorCode MatView_SeqAIJ_ASCII_structonly(Mat A, PetscViewer viewer)
681: {
682: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
683: PetscInt i, k, m = A->rmap->N;
685: PetscFunctionBegin;
686: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
687: for (i = 0; i < m; i++) {
688: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
689: for (k = a->i[i]; k < a->i[i + 1]; k++) PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ") ", a->j[k]));
690: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
691: }
692: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
693: PetscFunctionReturn(PETSC_SUCCESS);
694: }
696: extern PetscErrorCode MatSeqAIJFactorInfo_Matlab(Mat, PetscViewer);
698: static PetscErrorCode MatView_SeqAIJ_ASCII(Mat A, PetscViewer viewer)
699: {
700: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
701: const PetscScalar *av;
702: PetscInt i, j, m = A->rmap->n;
703: const char *name;
704: PetscViewerFormat format;
706: PetscFunctionBegin;
707: if (A->structure_only) {
708: PetscCall(MatView_SeqAIJ_ASCII_structonly(A, viewer));
709: PetscFunctionReturn(PETSC_SUCCESS);
710: }
712: PetscCall(PetscViewerGetFormat(viewer, &format));
713: // By petsc's rule, even PETSC_VIEWER_ASCII_INFO_DETAIL doesn't print matrix entries
714: if (format == PETSC_VIEWER_ASCII_FACTOR_INFO || format == PETSC_VIEWER_ASCII_INFO || format == PETSC_VIEWER_ASCII_INFO_DETAIL) PetscFunctionReturn(PETSC_SUCCESS);
716: /* trigger copy to CPU if needed */
717: PetscCall(MatSeqAIJGetArrayRead(A, &av));
718: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
719: if (format == PETSC_VIEWER_ASCII_MATLAB) {
720: PetscInt nofinalvalue = 0;
721: if (m && ((a->i[m] == a->i[m - 1]) || (a->j[a->nz - 1] != A->cmap->n - 1))) {
722: /* Need a dummy value to ensure the dimension of the matrix. */
723: nofinalvalue = 1;
724: }
725: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
726: PetscCall(PetscViewerASCIIPrintf(viewer, "%% Size = %" PetscInt_FMT " %" PetscInt_FMT " \n", m, A->cmap->n));
727: PetscCall(PetscViewerASCIIPrintf(viewer, "%% Nonzeros = %" PetscInt_FMT " \n", a->nz));
728: #if defined(PETSC_USE_COMPLEX)
729: PetscCall(PetscViewerASCIIPrintf(viewer, "zzz = zeros(%" PetscInt_FMT ",4);\n", a->nz + nofinalvalue));
730: #else
731: PetscCall(PetscViewerASCIIPrintf(viewer, "zzz = zeros(%" PetscInt_FMT ",3);\n", a->nz + nofinalvalue));
732: #endif
733: PetscCall(PetscViewerASCIIPrintf(viewer, "zzz = [\n"));
735: for (i = 0; i < m; i++) {
736: for (j = a->i[i]; j < a->i[i + 1]; j++) {
737: #if defined(PETSC_USE_COMPLEX)
738: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %18.16e %18.16e\n", i + 1, a->j[j] + 1, (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
739: #else
740: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %18.16e\n", i + 1, a->j[j] + 1, (double)a->a[j]));
741: #endif
742: }
743: }
744: if (nofinalvalue) {
745: #if defined(PETSC_USE_COMPLEX)
746: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %18.16e %18.16e\n", m, A->cmap->n, 0., 0.));
747: #else
748: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %18.16e\n", m, A->cmap->n, 0.0));
749: #endif
750: }
751: PetscCall(PetscObjectGetName((PetscObject)A, &name));
752: PetscCall(PetscViewerASCIIPrintf(viewer, "];\n %s = spconvert(zzz);\n", name));
753: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
754: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
755: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
756: for (i = 0; i < m; i++) {
757: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
758: for (j = a->i[i]; j < a->i[i + 1]; j++) {
759: #if defined(PETSC_USE_COMPLEX)
760: if (PetscImaginaryPart(a->a[j]) > 0.0 && PetscRealPart(a->a[j]) != 0.0) {
761: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
762: } else if (PetscImaginaryPart(a->a[j]) < 0.0 && PetscRealPart(a->a[j]) != 0.0) {
763: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)-PetscImaginaryPart(a->a[j])));
764: } else if (PetscRealPart(a->a[j]) != 0.0) {
765: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)PetscRealPart(a->a[j])));
766: }
767: #else
768: if (a->a[j] != 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)a->a[j]));
769: #endif
770: }
771: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
772: }
773: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
774: } else if (format == PETSC_VIEWER_ASCII_SYMMODU) {
775: PetscInt nzd = 0, fshift = 1, *sptr;
776: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
777: PetscCall(PetscMalloc1(m + 1, &sptr));
778: for (i = 0; i < m; i++) {
779: sptr[i] = nzd + 1;
780: for (j = a->i[i]; j < a->i[i + 1]; j++) {
781: if (a->j[j] >= i) {
782: #if defined(PETSC_USE_COMPLEX)
783: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) nzd++;
784: #else
785: if (a->a[j] != 0.0) nzd++;
786: #endif
787: }
788: }
789: }
790: sptr[m] = nzd + 1;
791: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT "\n\n", m, nzd));
792: for (i = 0; i < m + 1; i += 6) {
793: if (i + 4 < m) {
794: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT "\n", sptr[i], sptr[i + 1], sptr[i + 2], sptr[i + 3], sptr[i + 4], sptr[i + 5]));
795: } else if (i + 3 < m) {
796: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT "\n", sptr[i], sptr[i + 1], sptr[i + 2], sptr[i + 3], sptr[i + 4]));
797: } else if (i + 2 < m) {
798: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT "\n", sptr[i], sptr[i + 1], sptr[i + 2], sptr[i + 3]));
799: } else if (i + 1 < m) {
800: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT "\n", sptr[i], sptr[i + 1], sptr[i + 2]));
801: } else if (i < m) {
802: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " %" PetscInt_FMT "\n", sptr[i], sptr[i + 1]));
803: } else {
804: PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT "\n", sptr[i]));
805: }
806: }
807: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
808: PetscCall(PetscFree(sptr));
809: for (i = 0; i < m; i++) {
810: for (j = a->i[i]; j < a->i[i + 1]; j++) {
811: if (a->j[j] >= i) PetscCall(PetscViewerASCIIPrintf(viewer, " %" PetscInt_FMT " ", a->j[j] + fshift));
812: }
813: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
814: }
815: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
816: for (i = 0; i < m; i++) {
817: for (j = a->i[i]; j < a->i[i + 1]; j++) {
818: if (a->j[j] >= i) {
819: #if defined(PETSC_USE_COMPLEX)
820: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " %18.16e %18.16e ", (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
821: #else
822: if (a->a[j] != 0.0) PetscCall(PetscViewerASCIIPrintf(viewer, " %18.16e ", (double)a->a[j]));
823: #endif
824: }
825: }
826: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
827: }
828: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
829: } else if (format == PETSC_VIEWER_ASCII_DENSE) {
830: PetscInt cnt = 0, jcnt;
831: PetscScalar value;
832: #if defined(PETSC_USE_COMPLEX)
833: PetscBool realonly = PETSC_TRUE;
835: for (i = 0; i < a->i[m]; i++) {
836: if (PetscImaginaryPart(a->a[i]) != 0.0) {
837: realonly = PETSC_FALSE;
838: break;
839: }
840: }
841: #endif
843: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
844: for (i = 0; i < m; i++) {
845: jcnt = 0;
846: for (j = 0; j < A->cmap->n; j++) {
847: if (jcnt < a->i[i + 1] - a->i[i] && j == a->j[cnt]) {
848: value = a->a[cnt++];
849: jcnt++;
850: } else {
851: value = 0.0;
852: }
853: #if defined(PETSC_USE_COMPLEX)
854: if (realonly) {
855: PetscCall(PetscViewerASCIIPrintf(viewer, " %7.5e ", (double)PetscRealPart(value)));
856: } else {
857: PetscCall(PetscViewerASCIIPrintf(viewer, " %7.5e+%7.5e i ", (double)PetscRealPart(value), (double)PetscImaginaryPart(value)));
858: }
859: #else
860: PetscCall(PetscViewerASCIIPrintf(viewer, " %7.5e ", (double)value));
861: #endif
862: }
863: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
864: }
865: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
866: } else if (format == PETSC_VIEWER_ASCII_MATRIXMARKET) {
867: PetscInt fshift = 1;
868: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
869: #if defined(PETSC_USE_COMPLEX)
870: PetscCall(PetscViewerASCIIPrintf(viewer, "%%%%MatrixMarket matrix coordinate complex general\n"));
871: #else
872: PetscCall(PetscViewerASCIIPrintf(viewer, "%%%%MatrixMarket matrix coordinate real general\n"));
873: #endif
874: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %" PetscInt_FMT "\n", m, A->cmap->n, a->nz));
875: for (i = 0; i < m; i++) {
876: for (j = a->i[i]; j < a->i[i + 1]; j++) {
877: #if defined(PETSC_USE_COMPLEX)
878: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %g %g\n", i + fshift, a->j[j] + fshift, (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
879: #else
880: PetscCall(PetscViewerASCIIPrintf(viewer, "%" PetscInt_FMT " %" PetscInt_FMT " %g\n", i + fshift, a->j[j] + fshift, (double)a->a[j]));
881: #endif
882: }
883: }
884: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
885: } else {
886: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_FALSE));
887: if (A->factortype) {
888: for (i = 0; i < m; i++) {
889: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
890: /* L part */
891: for (j = a->i[i]; j < a->i[i + 1]; j++) {
892: #if defined(PETSC_USE_COMPLEX)
893: if (PetscImaginaryPart(a->a[j]) > 0.0) {
894: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
895: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
896: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)(-PetscImaginaryPart(a->a[j]))));
897: } else {
898: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)PetscRealPart(a->a[j])));
899: }
900: #else
901: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)a->a[j]));
902: #endif
903: }
904: /* diagonal */
905: j = a->diag[i];
906: #if defined(PETSC_USE_COMPLEX)
907: if (PetscImaginaryPart(a->a[j]) > 0.0) {
908: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i)", a->j[j], (double)PetscRealPart(1 / a->a[j]), (double)PetscImaginaryPart(1 / a->a[j])));
909: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
910: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i)", a->j[j], (double)PetscRealPart(1 / a->a[j]), (double)(-PetscImaginaryPart(1 / a->a[j]))));
911: } else {
912: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)PetscRealPart(1 / a->a[j])));
913: }
914: #else
915: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)(1 / a->a[j])));
916: #endif
918: /* U part */
919: for (j = a->diag[i + 1] + 1; j < a->diag[i]; j++) {
920: #if defined(PETSC_USE_COMPLEX)
921: if (PetscImaginaryPart(a->a[j]) > 0.0) {
922: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
923: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
924: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)(-PetscImaginaryPart(a->a[j]))));
925: } else {
926: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)PetscRealPart(a->a[j])));
927: }
928: #else
929: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)a->a[j]));
930: #endif
931: }
932: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
933: }
934: } else {
935: for (i = 0; i < m; i++) {
936: PetscCall(PetscViewerASCIIPrintf(viewer, "row %" PetscInt_FMT ":", i));
937: for (j = a->i[i]; j < a->i[i + 1]; j++) {
938: #if defined(PETSC_USE_COMPLEX)
939: if (PetscImaginaryPart(a->a[j]) > 0.0) {
940: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g + %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)PetscImaginaryPart(a->a[j])));
941: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
942: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g - %g i)", a->j[j], (double)PetscRealPart(a->a[j]), (double)-PetscImaginaryPart(a->a[j])));
943: } else {
944: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)PetscRealPart(a->a[j])));
945: }
946: #else
947: PetscCall(PetscViewerASCIIPrintf(viewer, " (%" PetscInt_FMT ", %g) ", a->j[j], (double)a->a[j]));
948: #endif
949: }
950: PetscCall(PetscViewerASCIIPrintf(viewer, "\n"));
951: }
952: }
953: PetscCall(PetscViewerASCIIUseTabs(viewer, PETSC_TRUE));
954: }
955: PetscCall(PetscViewerFlush(viewer));
956: PetscFunctionReturn(PETSC_SUCCESS);
957: }
959: #include <petscdraw.h>
960: static PetscErrorCode MatView_SeqAIJ_Draw_Zoom(PetscDraw draw, void *Aa)
961: {
962: Mat A = (Mat)Aa;
963: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
964: PetscInt i, j, m = A->rmap->n;
965: int color;
966: PetscReal xl, yl, xr, yr, x_l, x_r, y_l, y_r;
967: PetscViewer viewer;
968: PetscViewerFormat format;
969: const PetscScalar *aa;
971: PetscFunctionBegin;
972: PetscCall(PetscObjectQuery((PetscObject)A, "Zoomviewer", (PetscObject *)&viewer));
973: PetscCall(PetscViewerGetFormat(viewer, &format));
974: PetscCall(PetscDrawGetCoordinates(draw, &xl, &yl, &xr, &yr));
976: /* loop over matrix elements drawing boxes */
977: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
978: if (format != PETSC_VIEWER_DRAW_CONTOUR) {
979: PetscDrawCollectiveBegin(draw);
980: /* Blue for negative, Cyan for zero and Red for positive */
981: color = PETSC_DRAW_BLUE;
982: for (i = 0; i < m; i++) {
983: y_l = m - i - 1.0;
984: y_r = y_l + 1.0;
985: for (j = a->i[i]; j < a->i[i + 1]; j++) {
986: x_l = a->j[j];
987: x_r = x_l + 1.0;
988: if (PetscRealPart(aa[j]) >= 0.) continue;
989: PetscCall(PetscDrawRectangle(draw, x_l, y_l, x_r, y_r, color, color, color, color));
990: }
991: }
992: color = PETSC_DRAW_CYAN;
993: for (i = 0; i < m; i++) {
994: y_l = m - i - 1.0;
995: y_r = y_l + 1.0;
996: for (j = a->i[i]; j < a->i[i + 1]; j++) {
997: x_l = a->j[j];
998: x_r = x_l + 1.0;
999: if (aa[j] != 0.) continue;
1000: PetscCall(PetscDrawRectangle(draw, x_l, y_l, x_r, y_r, color, color, color, color));
1001: }
1002: }
1003: color = PETSC_DRAW_RED;
1004: for (i = 0; i < m; i++) {
1005: y_l = m - i - 1.0;
1006: y_r = y_l + 1.0;
1007: for (j = a->i[i]; j < a->i[i + 1]; j++) {
1008: x_l = a->j[j];
1009: x_r = x_l + 1.0;
1010: if (PetscRealPart(aa[j]) <= 0.) continue;
1011: PetscCall(PetscDrawRectangle(draw, x_l, y_l, x_r, y_r, color, color, color, color));
1012: }
1013: }
1014: PetscDrawCollectiveEnd(draw);
1015: } else {
1016: /* use contour shading to indicate magnitude of values */
1017: /* first determine max of all nonzero values */
1018: PetscReal minv = 0.0, maxv = 0.0;
1019: PetscInt nz = a->nz, count = 0;
1020: PetscDraw popup;
1022: for (i = 0; i < nz; i++) {
1023: if (PetscAbsScalar(aa[i]) > maxv) maxv = PetscAbsScalar(aa[i]);
1024: }
1025: if (minv >= maxv) maxv = minv + PETSC_SMALL;
1026: PetscCall(PetscDrawGetPopup(draw, &popup));
1027: PetscCall(PetscDrawScalePopup(popup, minv, maxv));
1029: PetscDrawCollectiveBegin(draw);
1030: for (i = 0; i < m; i++) {
1031: y_l = m - i - 1.0;
1032: y_r = y_l + 1.0;
1033: for (j = a->i[i]; j < a->i[i + 1]; j++) {
1034: x_l = a->j[j];
1035: x_r = x_l + 1.0;
1036: color = PetscDrawRealToColor(PetscAbsScalar(aa[count]), minv, maxv);
1037: PetscCall(PetscDrawRectangle(draw, x_l, y_l, x_r, y_r, color, color, color, color));
1038: count++;
1039: }
1040: }
1041: PetscDrawCollectiveEnd(draw);
1042: }
1043: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
1044: PetscFunctionReturn(PETSC_SUCCESS);
1045: }
1047: #include <petscdraw.h>
1048: static PetscErrorCode MatView_SeqAIJ_Draw(Mat A, PetscViewer viewer)
1049: {
1050: PetscDraw draw;
1051: PetscReal xr, yr, xl, yl, h, w;
1052: PetscBool isnull;
1054: PetscFunctionBegin;
1055: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
1056: PetscCall(PetscDrawIsNull(draw, &isnull));
1057: if (isnull) PetscFunctionReturn(PETSC_SUCCESS);
1059: xr = A->cmap->n;
1060: yr = A->rmap->n;
1061: h = yr / 10.0;
1062: w = xr / 10.0;
1063: xr += w;
1064: yr += h;
1065: xl = -w;
1066: yl = -h;
1067: PetscCall(PetscDrawSetCoordinates(draw, xl, yl, xr, yr));
1068: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", (PetscObject)viewer));
1069: PetscCall(PetscDrawZoom(draw, MatView_SeqAIJ_Draw_Zoom, A));
1070: PetscCall(PetscObjectCompose((PetscObject)A, "Zoomviewer", NULL));
1071: PetscCall(PetscDrawSave(draw));
1072: PetscFunctionReturn(PETSC_SUCCESS);
1073: }
1075: PetscErrorCode MatView_SeqAIJ(Mat A, PetscViewer viewer)
1076: {
1077: PetscBool iascii, isbinary, isdraw;
1079: PetscFunctionBegin;
1080: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
1081: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
1082: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
1083: if (iascii) PetscCall(MatView_SeqAIJ_ASCII(A, viewer));
1084: else if (isbinary) PetscCall(MatView_SeqAIJ_Binary(A, viewer));
1085: else if (isdraw) PetscCall(MatView_SeqAIJ_Draw(A, viewer));
1086: PetscCall(MatView_SeqAIJ_Inode(A, viewer));
1087: PetscFunctionReturn(PETSC_SUCCESS);
1088: }
1090: PetscErrorCode MatAssemblyEnd_SeqAIJ(Mat A, MatAssemblyType mode)
1091: {
1092: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1093: PetscInt fshift = 0, i, *ai = a->i, *aj = a->j, *imax = a->imax;
1094: PetscInt m = A->rmap->n, *ip, N, *ailen = a->ilen, rmax = 0, n;
1095: MatScalar *aa = a->a, *ap;
1096: PetscReal ratio = 0.6;
1098: PetscFunctionBegin;
1099: if (mode == MAT_FLUSH_ASSEMBLY) PetscFunctionReturn(PETSC_SUCCESS);
1100: PetscCall(MatSeqAIJInvalidateDiagonal(A));
1101: if (A->was_assembled && A->ass_nonzerostate == A->nonzerostate) {
1102: /* we need to respect users asking to use or not the inodes routine in between matrix assemblies */
1103: PetscCall(MatAssemblyEnd_SeqAIJ_Inode(A, mode));
1104: PetscFunctionReturn(PETSC_SUCCESS);
1105: }
1107: if (m) rmax = ailen[0]; /* determine row with most nonzeros */
1108: for (i = 1; i < m; i++) {
1109: /* move each row back by the amount of empty slots (fshift) before it*/
1110: fshift += imax[i - 1] - ailen[i - 1];
1111: rmax = PetscMax(rmax, ailen[i]);
1112: if (fshift) {
1113: ip = aj + ai[i];
1114: ap = aa + ai[i];
1115: N = ailen[i];
1116: PetscCall(PetscArraymove(ip - fshift, ip, N));
1117: if (!A->structure_only) PetscCall(PetscArraymove(ap - fshift, ap, N));
1118: }
1119: ai[i] = ai[i - 1] + ailen[i - 1];
1120: }
1121: if (m) {
1122: fshift += imax[m - 1] - ailen[m - 1];
1123: ai[m] = ai[m - 1] + ailen[m - 1];
1124: }
1125: /* reset ilen and imax for each row */
1126: a->nonzerorowcnt = 0;
1127: if (A->structure_only) {
1128: PetscCall(PetscFree(a->imax));
1129: PetscCall(PetscFree(a->ilen));
1130: } else { /* !A->structure_only */
1131: for (i = 0; i < m; i++) {
1132: ailen[i] = imax[i] = ai[i + 1] - ai[i];
1133: a->nonzerorowcnt += ((ai[i + 1] - ai[i]) > 0);
1134: }
1135: }
1136: a->nz = ai[m];
1137: PetscCheck(!fshift || a->nounused != -1, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Unused space detected in matrix: %" PetscInt_FMT " X %" PetscInt_FMT ", %" PetscInt_FMT " unneeded", m, A->cmap->n, fshift);
1138: PetscCall(MatMarkDiagonal_SeqAIJ(A)); // since diagonal info is used a lot, it is helpful to set them up at the end of assembly
1139: a->diagonaldense = PETSC_TRUE;
1140: n = PetscMin(A->rmap->n, A->cmap->n);
1141: for (i = 0; i < n; i++) {
1142: if (a->diag[i] >= ai[i + 1]) {
1143: a->diagonaldense = PETSC_FALSE;
1144: break;
1145: }
1146: }
1147: PetscCall(PetscInfo(A, "Matrix size: %" PetscInt_FMT " X %" PetscInt_FMT "; storage space: %" PetscInt_FMT " unneeded,%" PetscInt_FMT " used\n", m, A->cmap->n, fshift, a->nz));
1148: PetscCall(PetscInfo(A, "Number of mallocs during MatSetValues() is %" PetscInt_FMT "\n", a->reallocs));
1149: PetscCall(PetscInfo(A, "Maximum nonzeros in any row is %" PetscInt_FMT "\n", rmax));
1151: A->info.mallocs += a->reallocs;
1152: a->reallocs = 0;
1153: A->info.nz_unneeded = (PetscReal)fshift;
1154: a->rmax = rmax;
1156: if (!A->structure_only) PetscCall(MatCheckCompressedRow(A, a->nonzerorowcnt, &a->compressedrow, a->i, m, ratio));
1157: PetscCall(MatAssemblyEnd_SeqAIJ_Inode(A, mode));
1158: PetscFunctionReturn(PETSC_SUCCESS);
1159: }
1161: static PetscErrorCode MatRealPart_SeqAIJ(Mat A)
1162: {
1163: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1164: PetscInt i, nz = a->nz;
1165: MatScalar *aa;
1167: PetscFunctionBegin;
1168: PetscCall(MatSeqAIJGetArray(A, &aa));
1169: for (i = 0; i < nz; i++) aa[i] = PetscRealPart(aa[i]);
1170: PetscCall(MatSeqAIJRestoreArray(A, &aa));
1171: PetscCall(MatSeqAIJInvalidateDiagonal(A));
1172: PetscFunctionReturn(PETSC_SUCCESS);
1173: }
1175: static PetscErrorCode MatImaginaryPart_SeqAIJ(Mat A)
1176: {
1177: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1178: PetscInt i, nz = a->nz;
1179: MatScalar *aa;
1181: PetscFunctionBegin;
1182: PetscCall(MatSeqAIJGetArray(A, &aa));
1183: for (i = 0; i < nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
1184: PetscCall(MatSeqAIJRestoreArray(A, &aa));
1185: PetscCall(MatSeqAIJInvalidateDiagonal(A));
1186: PetscFunctionReturn(PETSC_SUCCESS);
1187: }
1189: PetscErrorCode MatZeroEntries_SeqAIJ(Mat A)
1190: {
1191: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1192: MatScalar *aa;
1194: PetscFunctionBegin;
1195: PetscCall(MatSeqAIJGetArrayWrite(A, &aa));
1196: PetscCall(PetscArrayzero(aa, a->i[A->rmap->n]));
1197: PetscCall(MatSeqAIJRestoreArrayWrite(A, &aa));
1198: PetscCall(MatSeqAIJInvalidateDiagonal(A));
1199: PetscFunctionReturn(PETSC_SUCCESS);
1200: }
1202: static PetscErrorCode MatReset_SeqAIJ(Mat A)
1203: {
1204: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1206: PetscFunctionBegin;
1207: if (A->hash_active) {
1208: A->ops[0] = a->cops;
1209: PetscCall(PetscHMapIJVDestroy(&a->ht));
1210: PetscCall(PetscFree(a->dnz));
1211: A->hash_active = PETSC_FALSE;
1212: }
1214: PetscCall(PetscLogObjectState((PetscObject)A, "Rows=%" PetscInt_FMT ", Cols=%" PetscInt_FMT ", NZ=%" PetscInt_FMT, A->rmap->n, A->cmap->n, a->nz));
1215: PetscCall(MatSeqXAIJFreeAIJ(A, &a->a, &a->j, &a->i));
1216: PetscCall(ISDestroy(&a->row));
1217: PetscCall(ISDestroy(&a->col));
1218: PetscCall(PetscFree(a->diag));
1219: PetscCall(PetscFree(a->ibdiag));
1220: PetscCall(PetscFree(a->imax));
1221: PetscCall(PetscFree(a->ilen));
1222: PetscCall(PetscFree(a->ipre));
1223: PetscCall(PetscFree3(a->idiag, a->mdiag, a->ssor_work));
1224: PetscCall(PetscFree(a->solve_work));
1225: PetscCall(ISDestroy(&a->icol));
1226: PetscCall(PetscFree(a->saved_values));
1227: PetscCall(PetscFree2(a->compressedrow.i, a->compressedrow.rindex));
1228: PetscCall(MatDestroy_SeqAIJ_Inode(A));
1229: PetscFunctionReturn(PETSC_SUCCESS);
1230: }
1232: static PetscErrorCode MatResetHash_SeqAIJ(Mat A)
1233: {
1234: PetscFunctionBegin;
1235: PetscCall(MatReset_SeqAIJ(A));
1236: PetscCall(MatCreate_SeqAIJ_Inode(A));
1237: PetscCall(MatSetUp_Seq_Hash(A));
1238: A->nonzerostate++;
1239: PetscFunctionReturn(PETSC_SUCCESS);
1240: }
1242: PetscErrorCode MatDestroy_SeqAIJ(Mat A)
1243: {
1244: PetscFunctionBegin;
1245: PetscCall(MatReset_SeqAIJ(A));
1246: PetscCall(PetscFree(A->data));
1248: /* MatMatMultNumeric_SeqAIJ_SeqAIJ_Sorted may allocate this.
1249: That function is so heavily used (sometimes in an hidden way through multnumeric function pointers)
1250: that is hard to properly add this data to the MatProduct data. We free it here to avoid
1251: users reusing the matrix object with different data to incur in obscure segmentation faults
1252: due to different matrix sizes */
1253: PetscCall(PetscObjectCompose((PetscObject)A, "__PETSc__ab_dense", NULL));
1255: PetscCall(PetscObjectChangeTypeName((PetscObject)A, NULL));
1256: PetscCall(PetscObjectComposeFunction((PetscObject)A, "PetscMatlabEnginePut_C", NULL));
1257: PetscCall(PetscObjectComposeFunction((PetscObject)A, "PetscMatlabEngineGet_C", NULL));
1258: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqAIJSetColumnIndices_C", NULL));
1259: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatStoreValues_C", NULL));
1260: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatRetrieveValues_C", NULL));
1261: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqsbaij_C", NULL));
1262: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqbaij_C", NULL));
1263: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijperm_C", NULL));
1264: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijsell_C", NULL));
1265: #if defined(PETSC_HAVE_MKL_SPARSE)
1266: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijmkl_C", NULL));
1267: #endif
1268: #if defined(PETSC_HAVE_CUDA)
1269: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijcusparse_C", NULL));
1270: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaijcusparse_seqaij_C", NULL));
1271: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaij_seqaijcusparse_C", NULL));
1272: #endif
1273: #if defined(PETSC_HAVE_HIP)
1274: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijhipsparse_C", NULL));
1275: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaijhipsparse_seqaij_C", NULL));
1276: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaij_seqaijhipsparse_C", NULL));
1277: #endif
1278: #if defined(PETSC_HAVE_KOKKOS_KERNELS)
1279: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijkokkos_C", NULL));
1280: #endif
1281: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijcrl_C", NULL));
1282: #if defined(PETSC_HAVE_ELEMENTAL)
1283: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_elemental_C", NULL));
1284: #endif
1285: #if defined(PETSC_HAVE_SCALAPACK)
1286: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_scalapack_C", NULL));
1287: #endif
1288: #if defined(PETSC_HAVE_HYPRE)
1289: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_hypre_C", NULL));
1290: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_transpose_seqaij_seqaij_C", NULL));
1291: #endif
1292: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqdense_C", NULL));
1293: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqsell_C", NULL));
1294: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_is_C", NULL));
1295: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatIsTranspose_C", NULL));
1296: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatIsHermitianTranspose_C", NULL));
1297: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqAIJSetPreallocation_C", NULL));
1298: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatResetPreallocation_C", NULL));
1299: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatResetHash_C", NULL));
1300: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqAIJSetPreallocationCSR_C", NULL));
1301: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatReorderForNonzeroDiagonal_C", NULL));
1302: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_is_seqaij_C", NULL));
1303: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqdense_seqaij_C", NULL));
1304: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaij_seqaij_C", NULL));
1305: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSeqAIJKron_C", NULL));
1306: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSetPreallocationCOO_C", NULL));
1307: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatSetValuesCOO_C", NULL));
1308: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatFactorGetSolverType_C", NULL));
1309: /* these calls do not belong here: the subclasses Duplicate/Destroy are wrong */
1310: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaijsell_seqaij_C", NULL));
1311: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaijperm_seqaij_C", NULL));
1312: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatConvert_seqaij_seqaijviennacl_C", NULL));
1313: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaijviennacl_seqdense_C", NULL));
1314: PetscCall(PetscObjectComposeFunction((PetscObject)A, "MatProductSetFromOptions_seqaijviennacl_seqaij_C", NULL));
1315: PetscFunctionReturn(PETSC_SUCCESS);
1316: }
1318: PetscErrorCode MatSetOption_SeqAIJ(Mat A, MatOption op, PetscBool flg)
1319: {
1320: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1322: PetscFunctionBegin;
1323: switch (op) {
1324: case MAT_ROW_ORIENTED:
1325: a->roworiented = flg;
1326: break;
1327: case MAT_KEEP_NONZERO_PATTERN:
1328: a->keepnonzeropattern = flg;
1329: break;
1330: case MAT_NEW_NONZERO_LOCATIONS:
1331: a->nonew = (flg ? 0 : 1);
1332: break;
1333: case MAT_NEW_NONZERO_LOCATION_ERR:
1334: a->nonew = (flg ? -1 : 0);
1335: break;
1336: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1337: a->nonew = (flg ? -2 : 0);
1338: break;
1339: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1340: a->nounused = (flg ? -1 : 0);
1341: break;
1342: case MAT_IGNORE_ZERO_ENTRIES:
1343: a->ignorezeroentries = flg;
1344: break;
1345: case MAT_SPD:
1346: case MAT_SYMMETRIC:
1347: case MAT_STRUCTURALLY_SYMMETRIC:
1348: case MAT_HERMITIAN:
1349: case MAT_SYMMETRY_ETERNAL:
1350: case MAT_STRUCTURE_ONLY:
1351: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
1352: case MAT_SPD_ETERNAL:
1353: /* if the diagonal matrix is square it inherits some of the properties above */
1354: break;
1355: case MAT_FORCE_DIAGONAL_ENTRIES:
1356: case MAT_IGNORE_OFF_PROC_ENTRIES:
1357: case MAT_USE_HASH_TABLE:
1358: PetscCall(PetscInfo(A, "Option %s ignored\n", MatOptions[op]));
1359: break;
1360: case MAT_USE_INODES:
1361: PetscCall(MatSetOption_SeqAIJ_Inode(A, MAT_USE_INODES, flg));
1362: break;
1363: case MAT_SUBMAT_SINGLEIS:
1364: A->submat_singleis = flg;
1365: break;
1366: case MAT_SORTED_FULL:
1367: if (flg) A->ops->setvalues = MatSetValues_SeqAIJ_SortedFull;
1368: else A->ops->setvalues = MatSetValues_SeqAIJ;
1369: break;
1370: case MAT_FORM_EXPLICIT_TRANSPOSE:
1371: A->form_explicit_transpose = flg;
1372: break;
1373: default:
1374: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "unknown option %d", op);
1375: }
1376: PetscFunctionReturn(PETSC_SUCCESS);
1377: }
1379: static PetscErrorCode MatGetDiagonal_SeqAIJ(Mat A, Vec v)
1380: {
1381: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1382: PetscInt i, j, n, *ai = a->i, *aj = a->j;
1383: PetscScalar *x;
1384: const PetscScalar *aa;
1386: PetscFunctionBegin;
1387: PetscCall(VecGetLocalSize(v, &n));
1388: PetscCheck(n == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector");
1389: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
1390: if (A->factortype == MAT_FACTOR_ILU || A->factortype == MAT_FACTOR_LU) {
1391: PetscInt *diag = a->diag;
1392: PetscCall(VecGetArrayWrite(v, &x));
1393: for (i = 0; i < n; i++) x[i] = 1.0 / aa[diag[i]];
1394: PetscCall(VecRestoreArrayWrite(v, &x));
1395: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
1396: PetscFunctionReturn(PETSC_SUCCESS);
1397: }
1399: PetscCall(VecGetArrayWrite(v, &x));
1400: for (i = 0; i < n; i++) {
1401: x[i] = 0.0;
1402: for (j = ai[i]; j < ai[i + 1]; j++) {
1403: if (aj[j] == i) {
1404: x[i] = aa[j];
1405: break;
1406: }
1407: }
1408: }
1409: PetscCall(VecRestoreArrayWrite(v, &x));
1410: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
1411: PetscFunctionReturn(PETSC_SUCCESS);
1412: }
1414: #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h>
1415: PetscErrorCode MatMultTransposeAdd_SeqAIJ(Mat A, Vec xx, Vec zz, Vec yy)
1416: {
1417: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1418: const MatScalar *aa;
1419: PetscScalar *y;
1420: const PetscScalar *x;
1421: PetscInt m = A->rmap->n;
1422: #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
1423: const MatScalar *v;
1424: PetscScalar alpha;
1425: PetscInt n, i, j;
1426: const PetscInt *idx, *ii, *ridx = NULL;
1427: Mat_CompressedRow cprow = a->compressedrow;
1428: PetscBool usecprow = cprow.use;
1429: #endif
1431: PetscFunctionBegin;
1432: if (zz != yy) PetscCall(VecCopy(zz, yy));
1433: PetscCall(VecGetArrayRead(xx, &x));
1434: PetscCall(VecGetArray(yy, &y));
1435: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
1437: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
1438: fortranmulttransposeaddaij_(&m, x, a->i, a->j, aa, y);
1439: #else
1440: if (usecprow) {
1441: m = cprow.nrows;
1442: ii = cprow.i;
1443: ridx = cprow.rindex;
1444: } else {
1445: ii = a->i;
1446: }
1447: for (i = 0; i < m; i++) {
1448: idx = a->j + ii[i];
1449: v = aa + ii[i];
1450: n = ii[i + 1] - ii[i];
1451: if (usecprow) {
1452: alpha = x[ridx[i]];
1453: } else {
1454: alpha = x[i];
1455: }
1456: for (j = 0; j < n; j++) y[idx[j]] += alpha * v[j];
1457: }
1458: #endif
1459: PetscCall(PetscLogFlops(2.0 * a->nz));
1460: PetscCall(VecRestoreArrayRead(xx, &x));
1461: PetscCall(VecRestoreArray(yy, &y));
1462: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
1463: PetscFunctionReturn(PETSC_SUCCESS);
1464: }
1466: PetscErrorCode MatMultTranspose_SeqAIJ(Mat A, Vec xx, Vec yy)
1467: {
1468: PetscFunctionBegin;
1469: PetscCall(VecSet(yy, 0.0));
1470: PetscCall(MatMultTransposeAdd_SeqAIJ(A, xx, yy, yy));
1471: PetscFunctionReturn(PETSC_SUCCESS);
1472: }
1474: #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h>
1476: PetscErrorCode MatMult_SeqAIJ(Mat A, Vec xx, Vec yy)
1477: {
1478: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1479: PetscScalar *y;
1480: const PetscScalar *x;
1481: const MatScalar *a_a;
1482: PetscInt m = A->rmap->n;
1483: const PetscInt *ii, *ridx = NULL;
1484: PetscBool usecprow = a->compressedrow.use;
1486: #if defined(PETSC_HAVE_PRAGMA_DISJOINT)
1487: #pragma disjoint(*x, *y, *aa)
1488: #endif
1490: PetscFunctionBegin;
1491: if (a->inode.use && a->inode.checked) {
1492: PetscCall(MatMult_SeqAIJ_Inode(A, xx, yy));
1493: PetscFunctionReturn(PETSC_SUCCESS);
1494: }
1495: PetscCall(MatSeqAIJGetArrayRead(A, &a_a));
1496: PetscCall(VecGetArrayRead(xx, &x));
1497: PetscCall(VecGetArray(yy, &y));
1498: ii = a->i;
1499: if (usecprow) { /* use compressed row format */
1500: PetscCall(PetscArrayzero(y, m));
1501: m = a->compressedrow.nrows;
1502: ii = a->compressedrow.i;
1503: ridx = a->compressedrow.rindex;
1504: PetscPragmaUseOMPKernels(parallel for)
1505: for (PetscInt i = 0; i < m; i++) {
1506: PetscInt n = ii[i + 1] - ii[i];
1507: const PetscInt *aj = a->j + ii[i];
1508: const PetscScalar *aa = a_a + ii[i];
1509: PetscScalar sum = 0.0;
1510: PetscSparseDensePlusDot(sum, x, aa, aj, n);
1511: /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */
1512: y[*ridx++] = sum;
1513: }
1514: } else { /* do not use compressed row format */
1515: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ)
1516: fortranmultaij_(&m, x, ii, a->j, a_a, y);
1517: #else
1518: PetscPragmaUseOMPKernels(parallel for)
1519: for (PetscInt i = 0; i < m; i++) {
1520: PetscInt n = ii[i + 1] - ii[i];
1521: const PetscInt *aj = a->j + ii[i];
1522: const PetscScalar *aa = a_a + ii[i];
1523: PetscScalar sum = 0.0;
1524: PetscSparseDensePlusDot(sum, x, aa, aj, n);
1525: y[i] = sum;
1526: }
1527: #endif
1528: }
1529: PetscCall(PetscLogFlops(2.0 * a->nz - a->nonzerorowcnt));
1530: PetscCall(VecRestoreArrayRead(xx, &x));
1531: PetscCall(VecRestoreArray(yy, &y));
1532: PetscCall(MatSeqAIJRestoreArrayRead(A, &a_a));
1533: PetscFunctionReturn(PETSC_SUCCESS);
1534: }
1536: // HACK!!!!! Used by src/mat/tests/ex170.c
1537: PETSC_EXTERN PetscErrorCode MatMultMax_SeqAIJ(Mat A, Vec xx, Vec yy)
1538: {
1539: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1540: PetscScalar *y;
1541: const PetscScalar *x;
1542: const MatScalar *aa, *a_a;
1543: PetscInt m = A->rmap->n;
1544: const PetscInt *aj, *ii, *ridx = NULL;
1545: PetscInt n, i, nonzerorow = 0;
1546: PetscScalar sum;
1547: PetscBool usecprow = a->compressedrow.use;
1549: #if defined(PETSC_HAVE_PRAGMA_DISJOINT)
1550: #pragma disjoint(*x, *y, *aa)
1551: #endif
1553: PetscFunctionBegin;
1554: PetscCall(MatSeqAIJGetArrayRead(A, &a_a));
1555: PetscCall(VecGetArrayRead(xx, &x));
1556: PetscCall(VecGetArray(yy, &y));
1557: if (usecprow) { /* use compressed row format */
1558: m = a->compressedrow.nrows;
1559: ii = a->compressedrow.i;
1560: ridx = a->compressedrow.rindex;
1561: for (i = 0; i < m; i++) {
1562: n = ii[i + 1] - ii[i];
1563: aj = a->j + ii[i];
1564: aa = a_a + ii[i];
1565: sum = 0.0;
1566: nonzerorow += (n > 0);
1567: PetscSparseDenseMaxDot(sum, x, aa, aj, n);
1568: /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */
1569: y[*ridx++] = sum;
1570: }
1571: } else { /* do not use compressed row format */
1572: ii = a->i;
1573: for (i = 0; i < m; i++) {
1574: n = ii[i + 1] - ii[i];
1575: aj = a->j + ii[i];
1576: aa = a_a + ii[i];
1577: sum = 0.0;
1578: nonzerorow += (n > 0);
1579: PetscSparseDenseMaxDot(sum, x, aa, aj, n);
1580: y[i] = sum;
1581: }
1582: }
1583: PetscCall(PetscLogFlops(2.0 * a->nz - nonzerorow));
1584: PetscCall(VecRestoreArrayRead(xx, &x));
1585: PetscCall(VecRestoreArray(yy, &y));
1586: PetscCall(MatSeqAIJRestoreArrayRead(A, &a_a));
1587: PetscFunctionReturn(PETSC_SUCCESS);
1588: }
1590: // HACK!!!!! Used by src/mat/tests/ex170.c
1591: PETSC_EXTERN PetscErrorCode MatMultAddMax_SeqAIJ(Mat A, Vec xx, Vec yy, Vec zz)
1592: {
1593: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1594: PetscScalar *y, *z;
1595: const PetscScalar *x;
1596: const MatScalar *aa, *a_a;
1597: PetscInt m = A->rmap->n, *aj, *ii;
1598: PetscInt n, i, *ridx = NULL;
1599: PetscScalar sum;
1600: PetscBool usecprow = a->compressedrow.use;
1602: PetscFunctionBegin;
1603: PetscCall(MatSeqAIJGetArrayRead(A, &a_a));
1604: PetscCall(VecGetArrayRead(xx, &x));
1605: PetscCall(VecGetArrayPair(yy, zz, &y, &z));
1606: if (usecprow) { /* use compressed row format */
1607: if (zz != yy) PetscCall(PetscArraycpy(z, y, m));
1608: m = a->compressedrow.nrows;
1609: ii = a->compressedrow.i;
1610: ridx = a->compressedrow.rindex;
1611: for (i = 0; i < m; i++) {
1612: n = ii[i + 1] - ii[i];
1613: aj = a->j + ii[i];
1614: aa = a_a + ii[i];
1615: sum = y[*ridx];
1616: PetscSparseDenseMaxDot(sum, x, aa, aj, n);
1617: z[*ridx++] = sum;
1618: }
1619: } else { /* do not use compressed row format */
1620: ii = a->i;
1621: for (i = 0; i < m; i++) {
1622: n = ii[i + 1] - ii[i];
1623: aj = a->j + ii[i];
1624: aa = a_a + ii[i];
1625: sum = y[i];
1626: PetscSparseDenseMaxDot(sum, x, aa, aj, n);
1627: z[i] = sum;
1628: }
1629: }
1630: PetscCall(PetscLogFlops(2.0 * a->nz));
1631: PetscCall(VecRestoreArrayRead(xx, &x));
1632: PetscCall(VecRestoreArrayPair(yy, zz, &y, &z));
1633: PetscCall(MatSeqAIJRestoreArrayRead(A, &a_a));
1634: PetscFunctionReturn(PETSC_SUCCESS);
1635: }
1637: #include <../src/mat/impls/aij/seq/ftn-kernels/fmultadd.h>
1638: PetscErrorCode MatMultAdd_SeqAIJ(Mat A, Vec xx, Vec yy, Vec zz)
1639: {
1640: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1641: PetscScalar *y, *z;
1642: const PetscScalar *x;
1643: const MatScalar *a_a;
1644: const PetscInt *ii, *ridx = NULL;
1645: PetscInt m = A->rmap->n;
1646: PetscBool usecprow = a->compressedrow.use;
1648: PetscFunctionBegin;
1649: if (a->inode.use && a->inode.checked) {
1650: PetscCall(MatMultAdd_SeqAIJ_Inode(A, xx, yy, zz));
1651: PetscFunctionReturn(PETSC_SUCCESS);
1652: }
1653: PetscCall(MatSeqAIJGetArrayRead(A, &a_a));
1654: PetscCall(VecGetArrayRead(xx, &x));
1655: PetscCall(VecGetArrayPair(yy, zz, &y, &z));
1656: if (usecprow) { /* use compressed row format */
1657: if (zz != yy) PetscCall(PetscArraycpy(z, y, m));
1658: m = a->compressedrow.nrows;
1659: ii = a->compressedrow.i;
1660: ridx = a->compressedrow.rindex;
1661: for (PetscInt i = 0; i < m; i++) {
1662: PetscInt n = ii[i + 1] - ii[i];
1663: const PetscInt *aj = a->j + ii[i];
1664: const PetscScalar *aa = a_a + ii[i];
1665: PetscScalar sum = y[*ridx];
1666: PetscSparseDensePlusDot(sum, x, aa, aj, n);
1667: z[*ridx++] = sum;
1668: }
1669: } else { /* do not use compressed row format */
1670: ii = a->i;
1671: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTADDAIJ)
1672: fortranmultaddaij_(&m, x, ii, a->j, a_a, y, z);
1673: #else
1674: PetscPragmaUseOMPKernels(parallel for)
1675: for (PetscInt i = 0; i < m; i++) {
1676: PetscInt n = ii[i + 1] - ii[i];
1677: const PetscInt *aj = a->j + ii[i];
1678: const PetscScalar *aa = a_a + ii[i];
1679: PetscScalar sum = y[i];
1680: PetscSparseDensePlusDot(sum, x, aa, aj, n);
1681: z[i] = sum;
1682: }
1683: #endif
1684: }
1685: PetscCall(PetscLogFlops(2.0 * a->nz));
1686: PetscCall(VecRestoreArrayRead(xx, &x));
1687: PetscCall(VecRestoreArrayPair(yy, zz, &y, &z));
1688: PetscCall(MatSeqAIJRestoreArrayRead(A, &a_a));
1689: PetscFunctionReturn(PETSC_SUCCESS);
1690: }
1692: /*
1693: Adds diagonal pointers to sparse matrix structure.
1694: */
1695: PetscErrorCode MatMarkDiagonal_SeqAIJ(Mat A)
1696: {
1697: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1698: PetscInt i, j, m = A->rmap->n;
1699: PetscBool alreadySet = PETSC_TRUE;
1701: PetscFunctionBegin;
1702: if (!a->diag) {
1703: PetscCall(PetscMalloc1(m, &a->diag));
1704: alreadySet = PETSC_FALSE;
1705: }
1706: for (i = 0; i < A->rmap->n; i++) {
1707: /* If A's diagonal is already correctly set, this fast track enables cheap and repeated MatMarkDiagonal_SeqAIJ() calls */
1708: if (alreadySet) {
1709: PetscInt pos = a->diag[i];
1710: if (pos >= a->i[i] && pos < a->i[i + 1] && a->j[pos] == i) continue;
1711: }
1713: a->diag[i] = a->i[i + 1];
1714: for (j = a->i[i]; j < a->i[i + 1]; j++) {
1715: if (a->j[j] == i) {
1716: a->diag[i] = j;
1717: break;
1718: }
1719: }
1720: }
1721: PetscFunctionReturn(PETSC_SUCCESS);
1722: }
1724: static PetscErrorCode MatShift_SeqAIJ(Mat A, PetscScalar v)
1725: {
1726: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1727: const PetscInt *diag = (const PetscInt *)a->diag;
1728: const PetscInt *ii = (const PetscInt *)a->i;
1729: PetscInt i, *mdiag = NULL;
1730: PetscInt cnt = 0; /* how many diagonals are missing */
1732: PetscFunctionBegin;
1733: if (!A->preallocated || !a->nz) {
1734: PetscCall(MatSeqAIJSetPreallocation(A, 1, NULL));
1735: PetscCall(MatShift_Basic(A, v));
1736: PetscFunctionReturn(PETSC_SUCCESS);
1737: }
1739: if (a->diagonaldense) {
1740: cnt = 0;
1741: } else {
1742: PetscCall(PetscCalloc1(A->rmap->n, &mdiag));
1743: for (i = 0; i < A->rmap->n; i++) {
1744: if (i < A->cmap->n && diag[i] >= ii[i + 1]) { /* 'out of range' rows never have diagonals */
1745: cnt++;
1746: mdiag[i] = 1;
1747: }
1748: }
1749: }
1750: if (!cnt) {
1751: PetscCall(MatShift_Basic(A, v));
1752: } else {
1753: PetscScalar *olda = a->a; /* preserve pointers to current matrix nonzeros structure and values */
1754: PetscInt *oldj = a->j, *oldi = a->i;
1755: PetscBool free_a = a->free_a, free_ij = a->free_ij;
1756: const PetscScalar *Aa;
1758: PetscCall(MatSeqAIJGetArrayRead(A, &Aa)); // sync the host
1759: PetscCall(MatSeqAIJRestoreArrayRead(A, &Aa));
1761: a->a = NULL;
1762: a->j = NULL;
1763: a->i = NULL;
1764: /* increase the values in imax for each row where a diagonal is being inserted then reallocate the matrix data structures */
1765: for (i = 0; i < PetscMin(A->rmap->n, A->cmap->n); i++) a->imax[i] += mdiag[i];
1766: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(A, 0, a->imax));
1768: /* copy old values into new matrix data structure */
1769: for (i = 0; i < A->rmap->n; i++) {
1770: PetscCall(MatSetValues(A, 1, &i, a->imax[i] - mdiag[i], &oldj[oldi[i]], &olda[oldi[i]], ADD_VALUES));
1771: if (i < A->cmap->n) PetscCall(MatSetValue(A, i, i, v, ADD_VALUES));
1772: }
1773: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
1774: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
1775: if (free_a) PetscCall(PetscShmgetDeallocateArray((void **)&olda));
1776: if (free_ij) PetscCall(PetscShmgetDeallocateArray((void **)&oldj));
1777: if (free_ij) PetscCall(PetscShmgetDeallocateArray((void **)&oldi));
1778: }
1779: PetscCall(PetscFree(mdiag));
1780: a->diagonaldense = PETSC_TRUE;
1781: PetscFunctionReturn(PETSC_SUCCESS);
1782: }
1784: /*
1785: Checks for missing diagonals
1786: */
1787: PetscErrorCode MatMissingDiagonal_SeqAIJ(Mat A, PetscBool *missing, PetscInt *d)
1788: {
1789: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1790: PetscInt *diag, *ii = a->i, i;
1792: PetscFunctionBegin;
1793: *missing = PETSC_FALSE;
1794: if (A->rmap->n > 0 && !ii) {
1795: *missing = PETSC_TRUE;
1796: if (d) *d = 0;
1797: PetscCall(PetscInfo(A, "Matrix has no entries therefore is missing diagonal\n"));
1798: } else {
1799: PetscInt n;
1800: n = PetscMin(A->rmap->n, A->cmap->n);
1801: diag = a->diag;
1802: for (i = 0; i < n; i++) {
1803: if (diag[i] >= ii[i + 1]) {
1804: *missing = PETSC_TRUE;
1805: if (d) *d = i;
1806: PetscCall(PetscInfo(A, "Matrix is missing diagonal number %" PetscInt_FMT "\n", i));
1807: break;
1808: }
1809: }
1810: }
1811: PetscFunctionReturn(PETSC_SUCCESS);
1812: }
1814: #include <petscblaslapack.h>
1815: #include <petsc/private/kernels/blockinvert.h>
1817: /*
1818: Note that values is allocated externally by the PC and then passed into this routine
1819: */
1820: static PetscErrorCode MatInvertVariableBlockDiagonal_SeqAIJ(Mat A, PetscInt nblocks, const PetscInt *bsizes, PetscScalar *diag)
1821: {
1822: PetscInt n = A->rmap->n, i, ncnt = 0, *indx, j, bsizemax = 0, *v_pivots;
1823: PetscBool allowzeropivot, zeropivotdetected = PETSC_FALSE;
1824: const PetscReal shift = 0.0;
1825: PetscInt ipvt[5];
1826: PetscCount flops = 0;
1827: PetscScalar work[25], *v_work;
1829: PetscFunctionBegin;
1830: allowzeropivot = PetscNot(A->erroriffailure);
1831: for (i = 0; i < nblocks; i++) ncnt += bsizes[i];
1832: PetscCheck(ncnt == n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Total blocksizes %" PetscInt_FMT " doesn't match number matrix rows %" PetscInt_FMT, ncnt, n);
1833: for (i = 0; i < nblocks; i++) bsizemax = PetscMax(bsizemax, bsizes[i]);
1834: PetscCall(PetscMalloc1(bsizemax, &indx));
1835: if (bsizemax > 7) PetscCall(PetscMalloc2(bsizemax, &v_work, bsizemax, &v_pivots));
1836: ncnt = 0;
1837: for (i = 0; i < nblocks; i++) {
1838: for (j = 0; j < bsizes[i]; j++) indx[j] = ncnt + j;
1839: PetscCall(MatGetValues(A, bsizes[i], indx, bsizes[i], indx, diag));
1840: switch (bsizes[i]) {
1841: case 1:
1842: *diag = 1.0 / (*diag);
1843: break;
1844: case 2:
1845: PetscCall(PetscKernel_A_gets_inverse_A_2(diag, shift, allowzeropivot, &zeropivotdetected));
1846: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1847: PetscCall(PetscKernel_A_gets_transpose_A_2(diag));
1848: break;
1849: case 3:
1850: PetscCall(PetscKernel_A_gets_inverse_A_3(diag, shift, allowzeropivot, &zeropivotdetected));
1851: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1852: PetscCall(PetscKernel_A_gets_transpose_A_3(diag));
1853: break;
1854: case 4:
1855: PetscCall(PetscKernel_A_gets_inverse_A_4(diag, shift, allowzeropivot, &zeropivotdetected));
1856: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1857: PetscCall(PetscKernel_A_gets_transpose_A_4(diag));
1858: break;
1859: case 5:
1860: PetscCall(PetscKernel_A_gets_inverse_A_5(diag, ipvt, work, shift, allowzeropivot, &zeropivotdetected));
1861: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1862: PetscCall(PetscKernel_A_gets_transpose_A_5(diag));
1863: break;
1864: case 6:
1865: PetscCall(PetscKernel_A_gets_inverse_A_6(diag, shift, allowzeropivot, &zeropivotdetected));
1866: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1867: PetscCall(PetscKernel_A_gets_transpose_A_6(diag));
1868: break;
1869: case 7:
1870: PetscCall(PetscKernel_A_gets_inverse_A_7(diag, shift, allowzeropivot, &zeropivotdetected));
1871: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1872: PetscCall(PetscKernel_A_gets_transpose_A_7(diag));
1873: break;
1874: default:
1875: PetscCall(PetscKernel_A_gets_inverse_A(bsizes[i], diag, v_pivots, v_work, allowzeropivot, &zeropivotdetected));
1876: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1877: PetscCall(PetscKernel_A_gets_transpose_A_N(diag, bsizes[i]));
1878: }
1879: ncnt += bsizes[i];
1880: diag += bsizes[i] * bsizes[i];
1881: flops += 2 * PetscPowInt64(bsizes[i], 3) / 3;
1882: }
1883: PetscCall(PetscLogFlops(flops));
1884: if (bsizemax > 7) PetscCall(PetscFree2(v_work, v_pivots));
1885: PetscCall(PetscFree(indx));
1886: PetscFunctionReturn(PETSC_SUCCESS);
1887: }
1889: /*
1890: Negative shift indicates do not generate an error if there is a zero diagonal, just invert it anyways
1891: */
1892: static PetscErrorCode MatInvertDiagonal_SeqAIJ(Mat A, PetscScalar omega, PetscScalar fshift)
1893: {
1894: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1895: PetscInt i, *diag, m = A->rmap->n;
1896: const MatScalar *v;
1897: PetscScalar *idiag, *mdiag;
1899: PetscFunctionBegin;
1900: if (a->idiagvalid) PetscFunctionReturn(PETSC_SUCCESS);
1901: PetscCall(MatMarkDiagonal_SeqAIJ(A));
1902: diag = a->diag;
1903: if (!a->idiag) { PetscCall(PetscMalloc3(m, &a->idiag, m, &a->mdiag, m, &a->ssor_work)); }
1905: mdiag = a->mdiag;
1906: idiag = a->idiag;
1907: PetscCall(MatSeqAIJGetArrayRead(A, &v));
1908: if (omega == 1.0 && PetscRealPart(fshift) <= 0.0) {
1909: for (i = 0; i < m; i++) {
1910: mdiag[i] = v[diag[i]];
1911: if (!PetscAbsScalar(mdiag[i])) { /* zero diagonal */
1912: if (PetscRealPart(fshift)) {
1913: PetscCall(PetscInfo(A, "Zero diagonal on row %" PetscInt_FMT "\n", i));
1914: A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
1915: A->factorerror_zeropivot_value = 0.0;
1916: A->factorerror_zeropivot_row = i;
1917: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Zero diagonal on row %" PetscInt_FMT, i);
1918: }
1919: idiag[i] = 1.0 / v[diag[i]];
1920: }
1921: PetscCall(PetscLogFlops(m));
1922: } else {
1923: for (i = 0; i < m; i++) {
1924: mdiag[i] = v[diag[i]];
1925: idiag[i] = omega / (fshift + v[diag[i]]);
1926: }
1927: PetscCall(PetscLogFlops(2.0 * m));
1928: }
1929: a->idiagvalid = PETSC_TRUE;
1930: PetscCall(MatSeqAIJRestoreArrayRead(A, &v));
1931: PetscFunctionReturn(PETSC_SUCCESS);
1932: }
1934: PetscErrorCode MatSOR_SeqAIJ(Mat A, Vec bb, PetscReal omega, MatSORType flag, PetscReal fshift, PetscInt its, PetscInt lits, Vec xx)
1935: {
1936: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
1937: PetscScalar *x, d, sum, *t, scale;
1938: const MatScalar *v, *idiag = NULL, *mdiag, *aa;
1939: const PetscScalar *b, *bs, *xb, *ts;
1940: PetscInt n, m = A->rmap->n, i;
1941: const PetscInt *idx, *diag;
1943: PetscFunctionBegin;
1944: if (a->inode.use && a->inode.checked && omega == 1.0 && fshift == 0.0) {
1945: PetscCall(MatSOR_SeqAIJ_Inode(A, bb, omega, flag, fshift, its, lits, xx));
1946: PetscFunctionReturn(PETSC_SUCCESS);
1947: }
1948: its = its * lits;
1950: if (fshift != a->fshift || omega != a->omega) a->idiagvalid = PETSC_FALSE; /* must recompute idiag[] */
1951: if (!a->idiagvalid) PetscCall(MatInvertDiagonal_SeqAIJ(A, omega, fshift));
1952: a->fshift = fshift;
1953: a->omega = omega;
1955: diag = a->diag;
1956: t = a->ssor_work;
1957: idiag = a->idiag;
1958: mdiag = a->mdiag;
1960: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
1961: PetscCall(VecGetArray(xx, &x));
1962: PetscCall(VecGetArrayRead(bb, &b));
1963: /* We count flops by assuming the upper triangular and lower triangular parts have the same number of nonzeros */
1964: if (flag == SOR_APPLY_UPPER) {
1965: /* apply (U + D/omega) to the vector */
1966: bs = b;
1967: for (i = 0; i < m; i++) {
1968: d = fshift + mdiag[i];
1969: n = a->i[i + 1] - diag[i] - 1;
1970: idx = a->j + diag[i] + 1;
1971: v = aa + diag[i] + 1;
1972: sum = b[i] * d / omega;
1973: PetscSparseDensePlusDot(sum, bs, v, idx, n);
1974: x[i] = sum;
1975: }
1976: PetscCall(VecRestoreArray(xx, &x));
1977: PetscCall(VecRestoreArrayRead(bb, &b));
1978: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
1979: PetscCall(PetscLogFlops(a->nz));
1980: PetscFunctionReturn(PETSC_SUCCESS);
1981: }
1983: PetscCheck(flag != SOR_APPLY_LOWER, PETSC_COMM_SELF, PETSC_ERR_SUP, "SOR_APPLY_LOWER is not implemented");
1984: if (flag & SOR_EISENSTAT) {
1985: /* Let A = L + U + D; where L is lower triangular,
1986: U is upper triangular, E = D/omega; This routine applies
1988: (L + E)^{-1} A (U + E)^{-1}
1990: to a vector efficiently using Eisenstat's trick.
1991: */
1992: scale = (2.0 / omega) - 1.0;
1994: /* x = (E + U)^{-1} b */
1995: for (i = m - 1; i >= 0; i--) {
1996: n = a->i[i + 1] - diag[i] - 1;
1997: idx = a->j + diag[i] + 1;
1998: v = aa + diag[i] + 1;
1999: sum = b[i];
2000: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2001: x[i] = sum * idiag[i];
2002: }
2004: /* t = b - (2*E - D)x */
2005: v = aa;
2006: for (i = 0; i < m; i++) t[i] = b[i] - scale * (v[*diag++]) * x[i];
2008: /* t = (E + L)^{-1}t */
2009: ts = t;
2010: diag = a->diag;
2011: for (i = 0; i < m; i++) {
2012: n = diag[i] - a->i[i];
2013: idx = a->j + a->i[i];
2014: v = aa + a->i[i];
2015: sum = t[i];
2016: PetscSparseDenseMinusDot(sum, ts, v, idx, n);
2017: t[i] = sum * idiag[i];
2018: /* x = x + t */
2019: x[i] += t[i];
2020: }
2022: PetscCall(PetscLogFlops(6.0 * m - 1 + 2.0 * a->nz));
2023: PetscCall(VecRestoreArray(xx, &x));
2024: PetscCall(VecRestoreArrayRead(bb, &b));
2025: PetscFunctionReturn(PETSC_SUCCESS);
2026: }
2027: if (flag & SOR_ZERO_INITIAL_GUESS) {
2028: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) {
2029: for (i = 0; i < m; i++) {
2030: n = diag[i] - a->i[i];
2031: idx = a->j + a->i[i];
2032: v = aa + a->i[i];
2033: sum = b[i];
2034: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2035: t[i] = sum;
2036: x[i] = sum * idiag[i];
2037: }
2038: xb = t;
2039: PetscCall(PetscLogFlops(a->nz));
2040: } else xb = b;
2041: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) {
2042: for (i = m - 1; i >= 0; i--) {
2043: n = a->i[i + 1] - diag[i] - 1;
2044: idx = a->j + diag[i] + 1;
2045: v = aa + diag[i] + 1;
2046: sum = xb[i];
2047: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2048: if (xb == b) {
2049: x[i] = sum * idiag[i];
2050: } else {
2051: x[i] = (1 - omega) * x[i] + sum * idiag[i]; /* omega in idiag */
2052: }
2053: }
2054: PetscCall(PetscLogFlops(a->nz)); /* assumes 1/2 in upper */
2055: }
2056: its--;
2057: }
2058: while (its--) {
2059: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) {
2060: for (i = 0; i < m; i++) {
2061: /* lower */
2062: n = diag[i] - a->i[i];
2063: idx = a->j + a->i[i];
2064: v = aa + a->i[i];
2065: sum = b[i];
2066: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2067: t[i] = sum; /* save application of the lower-triangular part */
2068: /* upper */
2069: n = a->i[i + 1] - diag[i] - 1;
2070: idx = a->j + diag[i] + 1;
2071: v = aa + diag[i] + 1;
2072: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2073: x[i] = (1. - omega) * x[i] + sum * idiag[i]; /* omega in idiag */
2074: }
2075: xb = t;
2076: PetscCall(PetscLogFlops(2.0 * a->nz));
2077: } else xb = b;
2078: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) {
2079: for (i = m - 1; i >= 0; i--) {
2080: sum = xb[i];
2081: if (xb == b) {
2082: /* whole matrix (no checkpointing available) */
2083: n = a->i[i + 1] - a->i[i];
2084: idx = a->j + a->i[i];
2085: v = aa + a->i[i];
2086: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2087: x[i] = (1. - omega) * x[i] + (sum + mdiag[i] * x[i]) * idiag[i];
2088: } else { /* lower-triangular part has been saved, so only apply upper-triangular */
2089: n = a->i[i + 1] - diag[i] - 1;
2090: idx = a->j + diag[i] + 1;
2091: v = aa + diag[i] + 1;
2092: PetscSparseDenseMinusDot(sum, x, v, idx, n);
2093: x[i] = (1. - omega) * x[i] + sum * idiag[i]; /* omega in idiag */
2094: }
2095: }
2096: if (xb == b) {
2097: PetscCall(PetscLogFlops(2.0 * a->nz));
2098: } else {
2099: PetscCall(PetscLogFlops(a->nz)); /* assumes 1/2 in upper */
2100: }
2101: }
2102: }
2103: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
2104: PetscCall(VecRestoreArray(xx, &x));
2105: PetscCall(VecRestoreArrayRead(bb, &b));
2106: PetscFunctionReturn(PETSC_SUCCESS);
2107: }
2109: static PetscErrorCode MatGetInfo_SeqAIJ(Mat A, MatInfoType flag, MatInfo *info)
2110: {
2111: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2113: PetscFunctionBegin;
2114: info->block_size = 1.0;
2115: info->nz_allocated = a->maxnz;
2116: info->nz_used = a->nz;
2117: info->nz_unneeded = (a->maxnz - a->nz);
2118: info->assemblies = A->num_ass;
2119: info->mallocs = A->info.mallocs;
2120: info->memory = 0; /* REVIEW ME */
2121: if (A->factortype) {
2122: info->fill_ratio_given = A->info.fill_ratio_given;
2123: info->fill_ratio_needed = A->info.fill_ratio_needed;
2124: info->factor_mallocs = A->info.factor_mallocs;
2125: } else {
2126: info->fill_ratio_given = 0;
2127: info->fill_ratio_needed = 0;
2128: info->factor_mallocs = 0;
2129: }
2130: PetscFunctionReturn(PETSC_SUCCESS);
2131: }
2133: static PetscErrorCode MatZeroRows_SeqAIJ(Mat A, PetscInt N, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
2134: {
2135: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2136: PetscInt i, m = A->rmap->n - 1;
2137: const PetscScalar *xx;
2138: PetscScalar *bb, *aa;
2139: PetscInt d = 0;
2141: PetscFunctionBegin;
2142: if (x && b) {
2143: PetscCall(VecGetArrayRead(x, &xx));
2144: PetscCall(VecGetArray(b, &bb));
2145: for (i = 0; i < N; i++) {
2146: PetscCheck(rows[i] >= 0 && rows[i] <= m, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", rows[i]);
2147: if (rows[i] >= A->cmap->n) continue;
2148: bb[rows[i]] = diag * xx[rows[i]];
2149: }
2150: PetscCall(VecRestoreArrayRead(x, &xx));
2151: PetscCall(VecRestoreArray(b, &bb));
2152: }
2154: PetscCall(MatSeqAIJGetArray(A, &aa));
2155: if (a->keepnonzeropattern) {
2156: for (i = 0; i < N; i++) {
2157: PetscCheck(rows[i] >= 0 && rows[i] <= m, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", rows[i]);
2158: PetscCall(PetscArrayzero(&aa[a->i[rows[i]]], a->ilen[rows[i]]));
2159: }
2160: if (diag != 0.0) {
2161: for (i = 0; i < N; i++) {
2162: d = rows[i];
2163: if (rows[i] >= A->cmap->n) continue;
2164: PetscCheck(a->diag[d] < a->i[d + 1], PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Matrix is missing diagonal entry in the zeroed row %" PetscInt_FMT, d);
2165: }
2166: for (i = 0; i < N; i++) {
2167: if (rows[i] >= A->cmap->n) continue;
2168: aa[a->diag[rows[i]]] = diag;
2169: }
2170: }
2171: } else {
2172: if (diag != 0.0) {
2173: for (i = 0; i < N; i++) {
2174: PetscCheck(rows[i] >= 0 && rows[i] <= m, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", rows[i]);
2175: if (a->ilen[rows[i]] > 0) {
2176: if (rows[i] >= A->cmap->n) {
2177: a->ilen[rows[i]] = 0;
2178: } else {
2179: a->ilen[rows[i]] = 1;
2180: aa[a->i[rows[i]]] = diag;
2181: a->j[a->i[rows[i]]] = rows[i];
2182: }
2183: } else if (rows[i] < A->cmap->n) { /* in case row was completely empty */
2184: PetscCall(MatSetValues_SeqAIJ(A, 1, &rows[i], 1, &rows[i], &diag, INSERT_VALUES));
2185: }
2186: }
2187: } else {
2188: for (i = 0; i < N; i++) {
2189: PetscCheck(rows[i] >= 0 && rows[i] <= m, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", rows[i]);
2190: a->ilen[rows[i]] = 0;
2191: }
2192: }
2193: A->nonzerostate++;
2194: }
2195: PetscCall(MatSeqAIJRestoreArray(A, &aa));
2196: PetscUseTypeMethod(A, assemblyend, MAT_FINAL_ASSEMBLY);
2197: PetscFunctionReturn(PETSC_SUCCESS);
2198: }
2200: static PetscErrorCode MatZeroRowsColumns_SeqAIJ(Mat A, PetscInt N, const PetscInt rows[], PetscScalar diag, Vec x, Vec b)
2201: {
2202: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2203: PetscInt i, j, m = A->rmap->n - 1, d = 0;
2204: PetscBool missing, *zeroed, vecs = PETSC_FALSE;
2205: const PetscScalar *xx;
2206: PetscScalar *bb, *aa;
2208: PetscFunctionBegin;
2209: if (!N) PetscFunctionReturn(PETSC_SUCCESS);
2210: PetscCall(MatSeqAIJGetArray(A, &aa));
2211: if (x && b) {
2212: PetscCall(VecGetArrayRead(x, &xx));
2213: PetscCall(VecGetArray(b, &bb));
2214: vecs = PETSC_TRUE;
2215: }
2216: PetscCall(PetscCalloc1(A->rmap->n, &zeroed));
2217: for (i = 0; i < N; i++) {
2218: PetscCheck(rows[i] >= 0 && rows[i] <= m, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "row %" PetscInt_FMT " out of range", rows[i]);
2219: PetscCall(PetscArrayzero(PetscSafePointerPlusOffset(aa, a->i[rows[i]]), a->ilen[rows[i]]));
2221: zeroed[rows[i]] = PETSC_TRUE;
2222: }
2223: for (i = 0; i < A->rmap->n; i++) {
2224: if (!zeroed[i]) {
2225: for (j = a->i[i]; j < a->i[i + 1]; j++) {
2226: if (a->j[j] < A->rmap->n && zeroed[a->j[j]]) {
2227: if (vecs) bb[i] -= aa[j] * xx[a->j[j]];
2228: aa[j] = 0.0;
2229: }
2230: }
2231: } else if (vecs && i < A->cmap->N) bb[i] = diag * xx[i];
2232: }
2233: if (x && b) {
2234: PetscCall(VecRestoreArrayRead(x, &xx));
2235: PetscCall(VecRestoreArray(b, &bb));
2236: }
2237: PetscCall(PetscFree(zeroed));
2238: if (diag != 0.0) {
2239: PetscCall(MatMissingDiagonal_SeqAIJ(A, &missing, &d));
2240: if (missing) {
2241: for (i = 0; i < N; i++) {
2242: if (rows[i] >= A->cmap->N) continue;
2243: PetscCheck(!a->nonew || rows[i] < d, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Matrix is missing diagonal entry in row %" PetscInt_FMT " (%" PetscInt_FMT ")", d, rows[i]);
2244: PetscCall(MatSetValues_SeqAIJ(A, 1, &rows[i], 1, &rows[i], &diag, INSERT_VALUES));
2245: }
2246: } else {
2247: for (i = 0; i < N; i++) aa[a->diag[rows[i]]] = diag;
2248: }
2249: }
2250: PetscCall(MatSeqAIJRestoreArray(A, &aa));
2251: PetscUseTypeMethod(A, assemblyend, MAT_FINAL_ASSEMBLY);
2252: PetscFunctionReturn(PETSC_SUCCESS);
2253: }
2255: PetscErrorCode MatGetRow_SeqAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
2256: {
2257: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2258: const PetscScalar *aa;
2260: PetscFunctionBegin;
2261: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
2262: *nz = a->i[row + 1] - a->i[row];
2263: if (v) *v = PetscSafePointerPlusOffset((PetscScalar *)aa, a->i[row]);
2264: if (idx) {
2265: if (*nz && a->j) *idx = a->j + a->i[row];
2266: else *idx = NULL;
2267: }
2268: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
2269: PetscFunctionReturn(PETSC_SUCCESS);
2270: }
2272: PetscErrorCode MatRestoreRow_SeqAIJ(Mat A, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
2273: {
2274: PetscFunctionBegin;
2275: PetscFunctionReturn(PETSC_SUCCESS);
2276: }
2278: static PetscErrorCode MatNorm_SeqAIJ(Mat A, NormType type, PetscReal *nrm)
2279: {
2280: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2281: const MatScalar *v;
2282: PetscReal sum = 0.0;
2283: PetscInt i, j;
2285: PetscFunctionBegin;
2286: PetscCall(MatSeqAIJGetArrayRead(A, &v));
2287: if (type == NORM_FROBENIUS) {
2288: #if defined(PETSC_USE_REAL___FP16)
2289: PetscBLASInt one = 1, nz = a->nz;
2290: PetscCallBLAS("BLASnrm2", *nrm = BLASnrm2_(&nz, v, &one));
2291: #else
2292: for (i = 0; i < a->nz; i++) {
2293: sum += PetscRealPart(PetscConj(*v) * (*v));
2294: v++;
2295: }
2296: *nrm = PetscSqrtReal(sum);
2297: #endif
2298: PetscCall(PetscLogFlops(2.0 * a->nz));
2299: } else if (type == NORM_1) {
2300: PetscReal *tmp;
2301: PetscInt *jj = a->j;
2302: PetscCall(PetscCalloc1(A->cmap->n + 1, &tmp));
2303: *nrm = 0.0;
2304: for (j = 0; j < a->nz; j++) {
2305: tmp[*jj++] += PetscAbsScalar(*v);
2306: v++;
2307: }
2308: for (j = 0; j < A->cmap->n; j++) {
2309: if (tmp[j] > *nrm) *nrm = tmp[j];
2310: }
2311: PetscCall(PetscFree(tmp));
2312: PetscCall(PetscLogFlops(PetscMax(a->nz - 1, 0)));
2313: } else if (type == NORM_INFINITY) {
2314: *nrm = 0.0;
2315: for (j = 0; j < A->rmap->n; j++) {
2316: const PetscScalar *v2 = PetscSafePointerPlusOffset(v, a->i[j]);
2317: sum = 0.0;
2318: for (i = 0; i < a->i[j + 1] - a->i[j]; i++) {
2319: sum += PetscAbsScalar(*v2);
2320: v2++;
2321: }
2322: if (sum > *nrm) *nrm = sum;
2323: }
2324: PetscCall(PetscLogFlops(PetscMax(a->nz - 1, 0)));
2325: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "No support for two norm");
2326: PetscCall(MatSeqAIJRestoreArrayRead(A, &v));
2327: PetscFunctionReturn(PETSC_SUCCESS);
2328: }
2330: static PetscErrorCode MatIsTranspose_SeqAIJ(Mat A, Mat B, PetscReal tol, PetscBool *f)
2331: {
2332: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data, *bij = (Mat_SeqAIJ *)B->data;
2333: PetscInt *adx, *bdx, *aii, *bii, *aptr, *bptr;
2334: const MatScalar *va, *vb;
2335: PetscInt ma, na, mb, nb, i;
2337: PetscFunctionBegin;
2338: PetscCall(MatGetSize(A, &ma, &na));
2339: PetscCall(MatGetSize(B, &mb, &nb));
2340: if (ma != nb || na != mb) {
2341: *f = PETSC_FALSE;
2342: PetscFunctionReturn(PETSC_SUCCESS);
2343: }
2344: PetscCall(MatSeqAIJGetArrayRead(A, &va));
2345: PetscCall(MatSeqAIJGetArrayRead(B, &vb));
2346: aii = aij->i;
2347: bii = bij->i;
2348: adx = aij->j;
2349: bdx = bij->j;
2350: PetscCall(PetscMalloc1(ma, &aptr));
2351: PetscCall(PetscMalloc1(mb, &bptr));
2352: for (i = 0; i < ma; i++) aptr[i] = aii[i];
2353: for (i = 0; i < mb; i++) bptr[i] = bii[i];
2355: *f = PETSC_TRUE;
2356: for (i = 0; i < ma; i++) {
2357: while (aptr[i] < aii[i + 1]) {
2358: PetscInt idc, idr;
2359: PetscScalar vc, vr;
2360: /* column/row index/value */
2361: idc = adx[aptr[i]];
2362: idr = bdx[bptr[idc]];
2363: vc = va[aptr[i]];
2364: vr = vb[bptr[idc]];
2365: if (i != idr || PetscAbsScalar(vc - vr) > tol) {
2366: *f = PETSC_FALSE;
2367: goto done;
2368: } else {
2369: aptr[i]++;
2370: if (B || i != idc) bptr[idc]++;
2371: }
2372: }
2373: }
2374: done:
2375: PetscCall(PetscFree(aptr));
2376: PetscCall(PetscFree(bptr));
2377: PetscCall(MatSeqAIJRestoreArrayRead(A, &va));
2378: PetscCall(MatSeqAIJRestoreArrayRead(B, &vb));
2379: PetscFunctionReturn(PETSC_SUCCESS);
2380: }
2382: static PetscErrorCode MatIsHermitianTranspose_SeqAIJ(Mat A, Mat B, PetscReal tol, PetscBool *f)
2383: {
2384: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data, *bij = (Mat_SeqAIJ *)B->data;
2385: PetscInt *adx, *bdx, *aii, *bii, *aptr, *bptr;
2386: MatScalar *va, *vb;
2387: PetscInt ma, na, mb, nb, i;
2389: PetscFunctionBegin;
2390: PetscCall(MatGetSize(A, &ma, &na));
2391: PetscCall(MatGetSize(B, &mb, &nb));
2392: if (ma != nb || na != mb) {
2393: *f = PETSC_FALSE;
2394: PetscFunctionReturn(PETSC_SUCCESS);
2395: }
2396: aii = aij->i;
2397: bii = bij->i;
2398: adx = aij->j;
2399: bdx = bij->j;
2400: va = aij->a;
2401: vb = bij->a;
2402: PetscCall(PetscMalloc1(ma, &aptr));
2403: PetscCall(PetscMalloc1(mb, &bptr));
2404: for (i = 0; i < ma; i++) aptr[i] = aii[i];
2405: for (i = 0; i < mb; i++) bptr[i] = bii[i];
2407: *f = PETSC_TRUE;
2408: for (i = 0; i < ma; i++) {
2409: while (aptr[i] < aii[i + 1]) {
2410: PetscInt idc, idr;
2411: PetscScalar vc, vr;
2412: /* column/row index/value */
2413: idc = adx[aptr[i]];
2414: idr = bdx[bptr[idc]];
2415: vc = va[aptr[i]];
2416: vr = vb[bptr[idc]];
2417: if (i != idr || PetscAbsScalar(vc - PetscConj(vr)) > tol) {
2418: *f = PETSC_FALSE;
2419: goto done;
2420: } else {
2421: aptr[i]++;
2422: if (B || i != idc) bptr[idc]++;
2423: }
2424: }
2425: }
2426: done:
2427: PetscCall(PetscFree(aptr));
2428: PetscCall(PetscFree(bptr));
2429: PetscFunctionReturn(PETSC_SUCCESS);
2430: }
2432: PetscErrorCode MatDiagonalScale_SeqAIJ(Mat A, Vec ll, Vec rr)
2433: {
2434: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2435: const PetscScalar *l, *r;
2436: PetscScalar x;
2437: MatScalar *v;
2438: PetscInt i, j, m = A->rmap->n, n = A->cmap->n, M, nz = a->nz;
2439: const PetscInt *jj;
2441: PetscFunctionBegin;
2442: if (ll) {
2443: /* The local size is used so that VecMPI can be passed to this routine
2444: by MatDiagonalScale_MPIAIJ */
2445: PetscCall(VecGetLocalSize(ll, &m));
2446: PetscCheck(m == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Left scaling vector wrong length");
2447: PetscCall(VecGetArrayRead(ll, &l));
2448: PetscCall(MatSeqAIJGetArray(A, &v));
2449: for (i = 0; i < m; i++) {
2450: x = l[i];
2451: M = a->i[i + 1] - a->i[i];
2452: for (j = 0; j < M; j++) (*v++) *= x;
2453: }
2454: PetscCall(VecRestoreArrayRead(ll, &l));
2455: PetscCall(PetscLogFlops(nz));
2456: PetscCall(MatSeqAIJRestoreArray(A, &v));
2457: }
2458: if (rr) {
2459: PetscCall(VecGetLocalSize(rr, &n));
2460: PetscCheck(n == A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Right scaling vector wrong length");
2461: PetscCall(VecGetArrayRead(rr, &r));
2462: PetscCall(MatSeqAIJGetArray(A, &v));
2463: jj = a->j;
2464: for (i = 0; i < nz; i++) (*v++) *= r[*jj++];
2465: PetscCall(MatSeqAIJRestoreArray(A, &v));
2466: PetscCall(VecRestoreArrayRead(rr, &r));
2467: PetscCall(PetscLogFlops(nz));
2468: }
2469: PetscCall(MatSeqAIJInvalidateDiagonal(A));
2470: PetscFunctionReturn(PETSC_SUCCESS);
2471: }
2473: PetscErrorCode MatCreateSubMatrix_SeqAIJ(Mat A, IS isrow, IS iscol, PetscInt csize, MatReuse scall, Mat *B)
2474: {
2475: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data, *c;
2476: PetscInt *smap, i, k, kstart, kend, oldcols = A->cmap->n, *lens;
2477: PetscInt row, mat_i, *mat_j, tcol, first, step, *mat_ilen, sum, lensi;
2478: const PetscInt *irow, *icol;
2479: const PetscScalar *aa;
2480: PetscInt nrows, ncols;
2481: PetscInt *starts, *j_new, *i_new, *aj = a->j, *ai = a->i, ii, *ailen = a->ilen;
2482: MatScalar *a_new, *mat_a, *c_a;
2483: Mat C;
2484: PetscBool stride;
2486: PetscFunctionBegin;
2487: PetscCall(ISGetIndices(isrow, &irow));
2488: PetscCall(ISGetLocalSize(isrow, &nrows));
2489: PetscCall(ISGetLocalSize(iscol, &ncols));
2491: PetscCall(PetscObjectTypeCompare((PetscObject)iscol, ISSTRIDE, &stride));
2492: if (stride) {
2493: PetscCall(ISStrideGetInfo(iscol, &first, &step));
2494: } else {
2495: first = 0;
2496: step = 0;
2497: }
2498: if (stride && step == 1) {
2499: /* special case of contiguous rows */
2500: PetscCall(PetscMalloc2(nrows, &lens, nrows, &starts));
2501: /* loop over new rows determining lens and starting points */
2502: for (i = 0; i < nrows; i++) {
2503: kstart = ai[irow[i]];
2504: kend = kstart + ailen[irow[i]];
2505: starts[i] = kstart;
2506: for (k = kstart; k < kend; k++) {
2507: if (aj[k] >= first) {
2508: starts[i] = k;
2509: break;
2510: }
2511: }
2512: sum = 0;
2513: while (k < kend) {
2514: if (aj[k++] >= first + ncols) break;
2515: sum++;
2516: }
2517: lens[i] = sum;
2518: }
2519: /* create submatrix */
2520: if (scall == MAT_REUSE_MATRIX) {
2521: PetscInt n_cols, n_rows;
2522: PetscCall(MatGetSize(*B, &n_rows, &n_cols));
2523: PetscCheck(n_rows == nrows && n_cols == ncols, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Reused submatrix wrong size");
2524: PetscCall(MatZeroEntries(*B));
2525: C = *B;
2526: } else {
2527: PetscInt rbs, cbs;
2528: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &C));
2529: PetscCall(MatSetSizes(C, nrows, ncols, PETSC_DETERMINE, PETSC_DETERMINE));
2530: PetscCall(ISGetBlockSize(isrow, &rbs));
2531: PetscCall(ISGetBlockSize(iscol, &cbs));
2532: PetscCall(MatSetBlockSizes(C, rbs, cbs));
2533: PetscCall(MatSetType(C, ((PetscObject)A)->type_name));
2534: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(C, 0, lens));
2535: }
2536: c = (Mat_SeqAIJ *)C->data;
2538: /* loop over rows inserting into submatrix */
2539: PetscCall(MatSeqAIJGetArrayWrite(C, &a_new)); // Not 'a_new = c->a-new', since that raw usage ignores offload state of C
2540: j_new = c->j;
2541: i_new = c->i;
2542: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
2543: for (i = 0; i < nrows; i++) {
2544: ii = starts[i];
2545: lensi = lens[i];
2546: if (lensi) {
2547: for (k = 0; k < lensi; k++) *j_new++ = aj[ii + k] - first;
2548: PetscCall(PetscArraycpy(a_new, aa + starts[i], lensi));
2549: a_new += lensi;
2550: }
2551: i_new[i + 1] = i_new[i] + lensi;
2552: c->ilen[i] = lensi;
2553: }
2554: PetscCall(MatSeqAIJRestoreArrayWrite(C, &a_new)); // Set C's offload state properly
2555: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
2556: PetscCall(PetscFree2(lens, starts));
2557: } else {
2558: PetscCall(ISGetIndices(iscol, &icol));
2559: PetscCall(PetscCalloc1(oldcols, &smap));
2560: PetscCall(PetscMalloc1(1 + nrows, &lens));
2561: for (i = 0; i < ncols; i++) {
2562: PetscCheck(icol[i] < oldcols, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Requesting column beyond largest column icol[%" PetscInt_FMT "] %" PetscInt_FMT " >= A->cmap->n %" PetscInt_FMT, i, icol[i], oldcols);
2563: smap[icol[i]] = i + 1;
2564: }
2566: /* determine lens of each row */
2567: for (i = 0; i < nrows; i++) {
2568: kstart = ai[irow[i]];
2569: kend = kstart + a->ilen[irow[i]];
2570: lens[i] = 0;
2571: for (k = kstart; k < kend; k++) {
2572: if (smap[aj[k]]) lens[i]++;
2573: }
2574: }
2575: /* Create and fill new matrix */
2576: if (scall == MAT_REUSE_MATRIX) {
2577: PetscBool equal;
2579: c = (Mat_SeqAIJ *)((*B)->data);
2580: PetscCheck((*B)->rmap->n == nrows && (*B)->cmap->n == ncols, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Cannot reuse matrix. wrong size");
2581: PetscCall(PetscArraycmp(c->ilen, lens, (*B)->rmap->n, &equal));
2582: PetscCheck(equal, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Cannot reuse matrix. wrong number of nonzeros");
2583: PetscCall(PetscArrayzero(c->ilen, (*B)->rmap->n));
2584: C = *B;
2585: } else {
2586: PetscInt rbs, cbs;
2587: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &C));
2588: PetscCall(MatSetSizes(C, nrows, ncols, PETSC_DETERMINE, PETSC_DETERMINE));
2589: PetscCall(ISGetBlockSize(isrow, &rbs));
2590: PetscCall(ISGetBlockSize(iscol, &cbs));
2591: if (rbs > 1 || cbs > 1) PetscCall(MatSetBlockSizes(C, rbs, cbs));
2592: PetscCall(MatSetType(C, ((PetscObject)A)->type_name));
2593: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(C, 0, lens));
2594: }
2595: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
2597: c = (Mat_SeqAIJ *)C->data;
2598: PetscCall(MatSeqAIJGetArrayWrite(C, &c_a)); // Not 'c->a', since that raw usage ignores offload state of C
2599: for (i = 0; i < nrows; i++) {
2600: row = irow[i];
2601: kstart = ai[row];
2602: kend = kstart + a->ilen[row];
2603: mat_i = c->i[i];
2604: mat_j = PetscSafePointerPlusOffset(c->j, mat_i);
2605: mat_a = PetscSafePointerPlusOffset(c_a, mat_i);
2606: mat_ilen = c->ilen + i;
2607: for (k = kstart; k < kend; k++) {
2608: if ((tcol = smap[a->j[k]])) {
2609: *mat_j++ = tcol - 1;
2610: *mat_a++ = aa[k];
2611: (*mat_ilen)++;
2612: }
2613: }
2614: }
2615: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
2616: /* Free work space */
2617: PetscCall(ISRestoreIndices(iscol, &icol));
2618: PetscCall(PetscFree(smap));
2619: PetscCall(PetscFree(lens));
2620: /* sort */
2621: for (i = 0; i < nrows; i++) {
2622: PetscInt ilen;
2624: mat_i = c->i[i];
2625: mat_j = PetscSafePointerPlusOffset(c->j, mat_i);
2626: mat_a = PetscSafePointerPlusOffset(c_a, mat_i);
2627: ilen = c->ilen[i];
2628: PetscCall(PetscSortIntWithScalarArray(ilen, mat_j, mat_a));
2629: }
2630: PetscCall(MatSeqAIJRestoreArrayWrite(C, &c_a));
2631: }
2632: #if defined(PETSC_HAVE_DEVICE)
2633: PetscCall(MatBindToCPU(C, A->boundtocpu));
2634: #endif
2635: PetscCall(MatAssemblyBegin(C, MAT_FINAL_ASSEMBLY));
2636: PetscCall(MatAssemblyEnd(C, MAT_FINAL_ASSEMBLY));
2638: PetscCall(ISRestoreIndices(isrow, &irow));
2639: *B = C;
2640: PetscFunctionReturn(PETSC_SUCCESS);
2641: }
2643: static PetscErrorCode MatGetMultiProcBlock_SeqAIJ(Mat mat, MPI_Comm subComm, MatReuse scall, Mat *subMat)
2644: {
2645: Mat B;
2647: PetscFunctionBegin;
2648: if (scall == MAT_INITIAL_MATRIX) {
2649: PetscCall(MatCreate(subComm, &B));
2650: PetscCall(MatSetSizes(B, mat->rmap->n, mat->cmap->n, mat->rmap->n, mat->cmap->n));
2651: PetscCall(MatSetBlockSizesFromMats(B, mat, mat));
2652: PetscCall(MatSetType(B, MATSEQAIJ));
2653: PetscCall(MatDuplicateNoCreate_SeqAIJ(B, mat, MAT_COPY_VALUES, PETSC_TRUE));
2654: *subMat = B;
2655: } else {
2656: PetscCall(MatCopy_SeqAIJ(mat, *subMat, SAME_NONZERO_PATTERN));
2657: }
2658: PetscFunctionReturn(PETSC_SUCCESS);
2659: }
2661: static PetscErrorCode MatILUFactor_SeqAIJ(Mat inA, IS row, IS col, const MatFactorInfo *info)
2662: {
2663: Mat_SeqAIJ *a = (Mat_SeqAIJ *)inA->data;
2664: Mat outA;
2665: PetscBool row_identity, col_identity;
2667: PetscFunctionBegin;
2668: PetscCheck(info->levels == 0, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only levels=0 supported for in-place ilu");
2670: PetscCall(ISIdentity(row, &row_identity));
2671: PetscCall(ISIdentity(col, &col_identity));
2673: outA = inA;
2674: outA->factortype = MAT_FACTOR_LU;
2675: PetscCall(PetscFree(inA->solvertype));
2676: PetscCall(PetscStrallocpy(MATSOLVERPETSC, &inA->solvertype));
2678: PetscCall(PetscObjectReference((PetscObject)row));
2679: PetscCall(ISDestroy(&a->row));
2681: a->row = row;
2683: PetscCall(PetscObjectReference((PetscObject)col));
2684: PetscCall(ISDestroy(&a->col));
2686: a->col = col;
2688: /* Create the inverse permutation so that it can be used in MatLUFactorNumeric() */
2689: PetscCall(ISDestroy(&a->icol));
2690: PetscCall(ISInvertPermutation(col, PETSC_DECIDE, &a->icol));
2692: if (!a->solve_work) { /* this matrix may have been factored before */
2693: PetscCall(PetscMalloc1(inA->rmap->n + 1, &a->solve_work));
2694: }
2696: PetscCall(MatMarkDiagonal_SeqAIJ(inA));
2697: if (row_identity && col_identity) {
2698: PetscCall(MatLUFactorNumeric_SeqAIJ_inplace(outA, inA, info));
2699: } else {
2700: PetscCall(MatLUFactorNumeric_SeqAIJ_InplaceWithPerm(outA, inA, info));
2701: }
2702: PetscFunctionReturn(PETSC_SUCCESS);
2703: }
2705: PetscErrorCode MatScale_SeqAIJ(Mat inA, PetscScalar alpha)
2706: {
2707: Mat_SeqAIJ *a = (Mat_SeqAIJ *)inA->data;
2708: PetscScalar *v;
2709: PetscBLASInt one = 1, bnz;
2711: PetscFunctionBegin;
2712: PetscCall(MatSeqAIJGetArray(inA, &v));
2713: PetscCall(PetscBLASIntCast(a->nz, &bnz));
2714: PetscCallBLAS("BLASscal", BLASscal_(&bnz, &alpha, v, &one));
2715: PetscCall(PetscLogFlops(a->nz));
2716: PetscCall(MatSeqAIJRestoreArray(inA, &v));
2717: PetscCall(MatSeqAIJInvalidateDiagonal(inA));
2718: PetscFunctionReturn(PETSC_SUCCESS);
2719: }
2721: PetscErrorCode MatDestroySubMatrix_Private(Mat_SubSppt *submatj)
2722: {
2723: PetscInt i;
2725: PetscFunctionBegin;
2726: if (!submatj->id) { /* delete data that are linked only to submats[id=0] */
2727: PetscCall(PetscFree4(submatj->sbuf1, submatj->ptr, submatj->tmp, submatj->ctr));
2729: for (i = 0; i < submatj->nrqr; ++i) PetscCall(PetscFree(submatj->sbuf2[i]));
2730: PetscCall(PetscFree3(submatj->sbuf2, submatj->req_size, submatj->req_source1));
2732: if (submatj->rbuf1) {
2733: PetscCall(PetscFree(submatj->rbuf1[0]));
2734: PetscCall(PetscFree(submatj->rbuf1));
2735: }
2737: for (i = 0; i < submatj->nrqs; ++i) PetscCall(PetscFree(submatj->rbuf3[i]));
2738: PetscCall(PetscFree3(submatj->req_source2, submatj->rbuf2, submatj->rbuf3));
2739: PetscCall(PetscFree(submatj->pa));
2740: }
2742: #if defined(PETSC_USE_CTABLE)
2743: PetscCall(PetscHMapIDestroy(&submatj->rmap));
2744: if (submatj->cmap_loc) PetscCall(PetscFree(submatj->cmap_loc));
2745: PetscCall(PetscFree(submatj->rmap_loc));
2746: #else
2747: PetscCall(PetscFree(submatj->rmap));
2748: #endif
2750: if (!submatj->allcolumns) {
2751: #if defined(PETSC_USE_CTABLE)
2752: PetscCall(PetscHMapIDestroy(&submatj->cmap));
2753: #else
2754: PetscCall(PetscFree(submatj->cmap));
2755: #endif
2756: }
2757: PetscCall(PetscFree(submatj->row2proc));
2759: PetscCall(PetscFree(submatj));
2760: PetscFunctionReturn(PETSC_SUCCESS);
2761: }
2763: PetscErrorCode MatDestroySubMatrix_SeqAIJ(Mat C)
2764: {
2765: Mat_SeqAIJ *c = (Mat_SeqAIJ *)C->data;
2766: Mat_SubSppt *submatj = c->submatis1;
2768: PetscFunctionBegin;
2769: PetscCall((*submatj->destroy)(C));
2770: PetscCall(MatDestroySubMatrix_Private(submatj));
2771: PetscFunctionReturn(PETSC_SUCCESS);
2772: }
2774: /* Note this has code duplication with MatDestroySubMatrices_SeqBAIJ() */
2775: static PetscErrorCode MatDestroySubMatrices_SeqAIJ(PetscInt n, Mat *mat[])
2776: {
2777: PetscInt i;
2778: Mat C;
2779: Mat_SeqAIJ *c;
2780: Mat_SubSppt *submatj;
2782: PetscFunctionBegin;
2783: for (i = 0; i < n; i++) {
2784: C = (*mat)[i];
2785: c = (Mat_SeqAIJ *)C->data;
2786: submatj = c->submatis1;
2787: if (submatj) {
2788: if (--((PetscObject)C)->refct <= 0) {
2789: PetscCall(PetscFree(C->factorprefix));
2790: PetscCall((*submatj->destroy)(C));
2791: PetscCall(MatDestroySubMatrix_Private(submatj));
2792: PetscCall(PetscFree(C->defaultvectype));
2793: PetscCall(PetscFree(C->defaultrandtype));
2794: PetscCall(PetscLayoutDestroy(&C->rmap));
2795: PetscCall(PetscLayoutDestroy(&C->cmap));
2796: PetscCall(PetscHeaderDestroy(&C));
2797: }
2798: } else {
2799: PetscCall(MatDestroy(&C));
2800: }
2801: }
2803: /* Destroy Dummy submatrices created for reuse */
2804: PetscCall(MatDestroySubMatrices_Dummy(n, mat));
2806: PetscCall(PetscFree(*mat));
2807: PetscFunctionReturn(PETSC_SUCCESS);
2808: }
2810: static PetscErrorCode MatCreateSubMatrices_SeqAIJ(Mat A, PetscInt n, const IS irow[], const IS icol[], MatReuse scall, Mat *B[])
2811: {
2812: PetscInt i;
2814: PetscFunctionBegin;
2815: if (scall == MAT_INITIAL_MATRIX) PetscCall(PetscCalloc1(n + 1, B));
2817: for (i = 0; i < n; i++) PetscCall(MatCreateSubMatrix_SeqAIJ(A, irow[i], icol[i], PETSC_DECIDE, scall, &(*B)[i]));
2818: PetscFunctionReturn(PETSC_SUCCESS);
2819: }
2821: static PetscErrorCode MatIncreaseOverlap_SeqAIJ(Mat A, PetscInt is_max, IS is[], PetscInt ov)
2822: {
2823: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2824: PetscInt row, i, j, k, l, ll, m, n, *nidx, isz, val;
2825: const PetscInt *idx;
2826: PetscInt start, end, *ai, *aj, bs = (A->rmap->bs > 0 && A->rmap->bs == A->cmap->bs) ? A->rmap->bs : 1;
2827: PetscBT table;
2829: PetscFunctionBegin;
2830: m = A->rmap->n / bs;
2831: ai = a->i;
2832: aj = a->j;
2834: PetscCheck(ov >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "illegal negative overlap value used");
2836: PetscCall(PetscMalloc1(m + 1, &nidx));
2837: PetscCall(PetscBTCreate(m, &table));
2839: for (i = 0; i < is_max; i++) {
2840: /* Initialize the two local arrays */
2841: isz = 0;
2842: PetscCall(PetscBTMemzero(m, table));
2844: /* Extract the indices, assume there can be duplicate entries */
2845: PetscCall(ISGetIndices(is[i], &idx));
2846: PetscCall(ISGetLocalSize(is[i], &n));
2848: if (bs > 1) {
2849: /* Enter these into the temp arrays. I.e., mark table[row], enter row into new index */
2850: for (j = 0; j < n; ++j) {
2851: if (!PetscBTLookupSet(table, idx[j] / bs)) nidx[isz++] = idx[j] / bs;
2852: }
2853: PetscCall(ISRestoreIndices(is[i], &idx));
2854: PetscCall(ISDestroy(&is[i]));
2856: k = 0;
2857: for (j = 0; j < ov; j++) { /* for each overlap */
2858: n = isz;
2859: for (; k < n; k++) { /* do only those rows in nidx[k], which are not done yet */
2860: for (ll = 0; ll < bs; ll++) {
2861: row = bs * nidx[k] + ll;
2862: start = ai[row];
2863: end = ai[row + 1];
2864: for (l = start; l < end; l++) {
2865: val = aj[l] / bs;
2866: if (!PetscBTLookupSet(table, val)) nidx[isz++] = val;
2867: }
2868: }
2869: }
2870: }
2871: PetscCall(ISCreateBlock(PETSC_COMM_SELF, bs, isz, nidx, PETSC_COPY_VALUES, is + i));
2872: } else {
2873: /* Enter these into the temp arrays. I.e., mark table[row], enter row into new index */
2874: for (j = 0; j < n; ++j) {
2875: if (!PetscBTLookupSet(table, idx[j])) nidx[isz++] = idx[j];
2876: }
2877: PetscCall(ISRestoreIndices(is[i], &idx));
2878: PetscCall(ISDestroy(&is[i]));
2880: k = 0;
2881: for (j = 0; j < ov; j++) { /* for each overlap */
2882: n = isz;
2883: for (; k < n; k++) { /* do only those rows in nidx[k], which are not done yet */
2884: row = nidx[k];
2885: start = ai[row];
2886: end = ai[row + 1];
2887: for (l = start; l < end; l++) {
2888: val = aj[l];
2889: if (!PetscBTLookupSet(table, val)) nidx[isz++] = val;
2890: }
2891: }
2892: }
2893: PetscCall(ISCreateGeneral(PETSC_COMM_SELF, isz, nidx, PETSC_COPY_VALUES, is + i));
2894: }
2895: }
2896: PetscCall(PetscBTDestroy(&table));
2897: PetscCall(PetscFree(nidx));
2898: PetscFunctionReturn(PETSC_SUCCESS);
2899: }
2901: static PetscErrorCode MatPermute_SeqAIJ(Mat A, IS rowp, IS colp, Mat *B)
2902: {
2903: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2904: PetscInt i, nz = 0, m = A->rmap->n, n = A->cmap->n;
2905: const PetscInt *row, *col;
2906: PetscInt *cnew, j, *lens;
2907: IS icolp, irowp;
2908: PetscInt *cwork = NULL;
2909: PetscScalar *vwork = NULL;
2911: PetscFunctionBegin;
2912: PetscCall(ISInvertPermutation(rowp, PETSC_DECIDE, &irowp));
2913: PetscCall(ISGetIndices(irowp, &row));
2914: PetscCall(ISInvertPermutation(colp, PETSC_DECIDE, &icolp));
2915: PetscCall(ISGetIndices(icolp, &col));
2917: /* determine lengths of permuted rows */
2918: PetscCall(PetscMalloc1(m + 1, &lens));
2919: for (i = 0; i < m; i++) lens[row[i]] = a->i[i + 1] - a->i[i];
2920: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), B));
2921: PetscCall(MatSetSizes(*B, m, n, m, n));
2922: PetscCall(MatSetBlockSizesFromMats(*B, A, A));
2923: PetscCall(MatSetType(*B, ((PetscObject)A)->type_name));
2924: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(*B, 0, lens));
2925: PetscCall(PetscFree(lens));
2927: PetscCall(PetscMalloc1(n, &cnew));
2928: for (i = 0; i < m; i++) {
2929: PetscCall(MatGetRow_SeqAIJ(A, i, &nz, &cwork, &vwork));
2930: for (j = 0; j < nz; j++) cnew[j] = col[cwork[j]];
2931: PetscCall(MatSetValues_SeqAIJ(*B, 1, &row[i], nz, cnew, vwork, INSERT_VALUES));
2932: PetscCall(MatRestoreRow_SeqAIJ(A, i, &nz, &cwork, &vwork));
2933: }
2934: PetscCall(PetscFree(cnew));
2936: (*B)->assembled = PETSC_FALSE;
2938: #if defined(PETSC_HAVE_DEVICE)
2939: PetscCall(MatBindToCPU(*B, A->boundtocpu));
2940: #endif
2941: PetscCall(MatAssemblyBegin(*B, MAT_FINAL_ASSEMBLY));
2942: PetscCall(MatAssemblyEnd(*B, MAT_FINAL_ASSEMBLY));
2943: PetscCall(ISRestoreIndices(irowp, &row));
2944: PetscCall(ISRestoreIndices(icolp, &col));
2945: PetscCall(ISDestroy(&irowp));
2946: PetscCall(ISDestroy(&icolp));
2947: if (rowp == colp) PetscCall(MatPropagateSymmetryOptions(A, *B));
2948: PetscFunctionReturn(PETSC_SUCCESS);
2949: }
2951: PetscErrorCode MatCopy_SeqAIJ(Mat A, Mat B, MatStructure str)
2952: {
2953: PetscFunctionBegin;
2954: /* If the two matrices have the same copy implementation, use fast copy. */
2955: if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) {
2956: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2957: Mat_SeqAIJ *b = (Mat_SeqAIJ *)B->data;
2958: const PetscScalar *aa;
2960: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
2961: PetscCheck(a->i[A->rmap->n] == b->i[B->rmap->n], PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Number of nonzeros in two matrices are different %" PetscInt_FMT " != %" PetscInt_FMT, a->i[A->rmap->n], b->i[B->rmap->n]);
2962: PetscCall(PetscArraycpy(b->a, aa, a->i[A->rmap->n]));
2963: PetscCall(PetscObjectStateIncrease((PetscObject)B));
2964: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
2965: } else {
2966: PetscCall(MatCopy_Basic(A, B, str));
2967: }
2968: PetscFunctionReturn(PETSC_SUCCESS);
2969: }
2971: PETSC_INTERN PetscErrorCode MatSeqAIJGetArray_SeqAIJ(Mat A, PetscScalar *array[])
2972: {
2973: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
2975: PetscFunctionBegin;
2976: *array = a->a;
2977: PetscFunctionReturn(PETSC_SUCCESS);
2978: }
2980: PETSC_INTERN PetscErrorCode MatSeqAIJRestoreArray_SeqAIJ(Mat A, PetscScalar *array[])
2981: {
2982: PetscFunctionBegin;
2983: *array = NULL;
2984: PetscFunctionReturn(PETSC_SUCCESS);
2985: }
2987: /*
2988: Computes the number of nonzeros per row needed for preallocation when X and Y
2989: have different nonzero structure.
2990: */
2991: PetscErrorCode MatAXPYGetPreallocation_SeqX_private(PetscInt m, const PetscInt *xi, const PetscInt *xj, const PetscInt *yi, const PetscInt *yj, PetscInt *nnz)
2992: {
2993: PetscInt i, j, k, nzx, nzy;
2995: PetscFunctionBegin;
2996: /* Set the number of nonzeros in the new matrix */
2997: for (i = 0; i < m; i++) {
2998: const PetscInt *xjj = PetscSafePointerPlusOffset(xj, xi[i]), *yjj = PetscSafePointerPlusOffset(yj, yi[i]);
2999: nzx = xi[i + 1] - xi[i];
3000: nzy = yi[i + 1] - yi[i];
3001: nnz[i] = 0;
3002: for (j = 0, k = 0; j < nzx; j++) { /* Point in X */
3003: for (; k < nzy && yjj[k] < xjj[j]; k++) nnz[i]++; /* Catch up to X */
3004: if (k < nzy && yjj[k] == xjj[j]) k++; /* Skip duplicate */
3005: nnz[i]++;
3006: }
3007: for (; k < nzy; k++) nnz[i]++;
3008: }
3009: PetscFunctionReturn(PETSC_SUCCESS);
3010: }
3012: PetscErrorCode MatAXPYGetPreallocation_SeqAIJ(Mat Y, Mat X, PetscInt *nnz)
3013: {
3014: PetscInt m = Y->rmap->N;
3015: Mat_SeqAIJ *x = (Mat_SeqAIJ *)X->data;
3016: Mat_SeqAIJ *y = (Mat_SeqAIJ *)Y->data;
3018: PetscFunctionBegin;
3019: /* Set the number of nonzeros in the new matrix */
3020: PetscCall(MatAXPYGetPreallocation_SeqX_private(m, x->i, x->j, y->i, y->j, nnz));
3021: PetscFunctionReturn(PETSC_SUCCESS);
3022: }
3024: PetscErrorCode MatAXPY_SeqAIJ(Mat Y, PetscScalar a, Mat X, MatStructure str)
3025: {
3026: Mat_SeqAIJ *x = (Mat_SeqAIJ *)X->data, *y = (Mat_SeqAIJ *)Y->data;
3028: PetscFunctionBegin;
3029: if (str == UNKNOWN_NONZERO_PATTERN || (PetscDefined(USE_DEBUG) && str == SAME_NONZERO_PATTERN)) {
3030: PetscBool e = x->nz == y->nz ? PETSC_TRUE : PETSC_FALSE;
3031: if (e) {
3032: PetscCall(PetscArraycmp(x->i, y->i, Y->rmap->n + 1, &e));
3033: if (e) {
3034: PetscCall(PetscArraycmp(x->j, y->j, y->nz, &e));
3035: if (e) str = SAME_NONZERO_PATTERN;
3036: }
3037: }
3038: if (!e) PetscCheck(str != SAME_NONZERO_PATTERN, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "MatStructure is not SAME_NONZERO_PATTERN");
3039: }
3040: if (str == SAME_NONZERO_PATTERN) {
3041: const PetscScalar *xa;
3042: PetscScalar *ya, alpha = a;
3043: PetscBLASInt one = 1, bnz;
3045: PetscCall(PetscBLASIntCast(x->nz, &bnz));
3046: PetscCall(MatSeqAIJGetArray(Y, &ya));
3047: PetscCall(MatSeqAIJGetArrayRead(X, &xa));
3048: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, xa, &one, ya, &one));
3049: PetscCall(MatSeqAIJRestoreArrayRead(X, &xa));
3050: PetscCall(MatSeqAIJRestoreArray(Y, &ya));
3051: PetscCall(PetscLogFlops(2.0 * bnz));
3052: PetscCall(MatSeqAIJInvalidateDiagonal(Y));
3053: PetscCall(PetscObjectStateIncrease((PetscObject)Y));
3054: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
3055: PetscCall(MatAXPY_Basic(Y, a, X, str));
3056: } else {
3057: Mat B;
3058: PetscInt *nnz;
3059: PetscCall(PetscMalloc1(Y->rmap->N, &nnz));
3060: PetscCall(MatCreate(PetscObjectComm((PetscObject)Y), &B));
3061: PetscCall(PetscObjectSetName((PetscObject)B, ((PetscObject)Y)->name));
3062: PetscCall(MatSetLayouts(B, Y->rmap, Y->cmap));
3063: PetscCall(MatSetType(B, ((PetscObject)Y)->type_name));
3064: PetscCall(MatAXPYGetPreallocation_SeqAIJ(Y, X, nnz));
3065: PetscCall(MatSeqAIJSetPreallocation(B, 0, nnz));
3066: PetscCall(MatAXPY_BasicWithPreallocation(B, Y, a, X, str));
3067: PetscCall(MatHeaderMerge(Y, &B));
3068: PetscCall(MatSeqAIJCheckInode(Y));
3069: PetscCall(PetscFree(nnz));
3070: }
3071: PetscFunctionReturn(PETSC_SUCCESS);
3072: }
3074: PETSC_INTERN PetscErrorCode MatConjugate_SeqAIJ(Mat mat)
3075: {
3076: #if defined(PETSC_USE_COMPLEX)
3077: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
3078: PetscInt i, nz;
3079: PetscScalar *a;
3081: PetscFunctionBegin;
3082: nz = aij->nz;
3083: PetscCall(MatSeqAIJGetArray(mat, &a));
3084: for (i = 0; i < nz; i++) a[i] = PetscConj(a[i]);
3085: PetscCall(MatSeqAIJRestoreArray(mat, &a));
3086: #else
3087: PetscFunctionBegin;
3088: #endif
3089: PetscFunctionReturn(PETSC_SUCCESS);
3090: }
3092: static PetscErrorCode MatGetRowMaxAbs_SeqAIJ(Mat A, Vec v, PetscInt idx[])
3093: {
3094: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3095: PetscInt i, j, m = A->rmap->n, *ai, *aj, ncols, n;
3096: PetscReal atmp;
3097: PetscScalar *x;
3098: const MatScalar *aa, *av;
3100: PetscFunctionBegin;
3101: PetscCheck(!A->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3102: PetscCall(MatSeqAIJGetArrayRead(A, &av));
3103: aa = av;
3104: ai = a->i;
3105: aj = a->j;
3107: PetscCall(VecSet(v, 0.0));
3108: PetscCall(VecGetArrayWrite(v, &x));
3109: PetscCall(VecGetLocalSize(v, &n));
3110: PetscCheck(n == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector");
3111: for (i = 0; i < m; i++) {
3112: ncols = ai[1] - ai[0];
3113: ai++;
3114: for (j = 0; j < ncols; j++) {
3115: atmp = PetscAbsScalar(*aa);
3116: if (PetscAbsScalar(x[i]) < atmp) {
3117: x[i] = atmp;
3118: if (idx) idx[i] = *aj;
3119: }
3120: aa++;
3121: aj++;
3122: }
3123: }
3124: PetscCall(VecRestoreArrayWrite(v, &x));
3125: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
3126: PetscFunctionReturn(PETSC_SUCCESS);
3127: }
3129: static PetscErrorCode MatGetRowSumAbs_SeqAIJ(Mat A, Vec v)
3130: {
3131: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3132: PetscInt i, j, m = A->rmap->n, *ai, ncols, n;
3133: PetscScalar *x;
3134: const MatScalar *aa, *av;
3136: PetscFunctionBegin;
3137: PetscCheck(!A->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3138: PetscCall(MatSeqAIJGetArrayRead(A, &av));
3139: aa = av;
3140: ai = a->i;
3142: PetscCall(VecSet(v, 0.0));
3143: PetscCall(VecGetArrayWrite(v, &x));
3144: PetscCall(VecGetLocalSize(v, &n));
3145: PetscCheck(n == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector");
3146: for (i = 0; i < m; i++) {
3147: ncols = ai[1] - ai[0];
3148: ai++;
3149: for (j = 0; j < ncols; j++) {
3150: x[i] += PetscAbsScalar(*aa);
3151: aa++;
3152: }
3153: }
3154: PetscCall(VecRestoreArrayWrite(v, &x));
3155: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
3156: PetscFunctionReturn(PETSC_SUCCESS);
3157: }
3159: static PetscErrorCode MatGetRowMax_SeqAIJ(Mat A, Vec v, PetscInt idx[])
3160: {
3161: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3162: PetscInt i, j, m = A->rmap->n, *ai, *aj, ncols, n;
3163: PetscScalar *x;
3164: const MatScalar *aa, *av;
3166: PetscFunctionBegin;
3167: PetscCheck(!A->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3168: PetscCall(MatSeqAIJGetArrayRead(A, &av));
3169: aa = av;
3170: ai = a->i;
3171: aj = a->j;
3173: PetscCall(VecSet(v, 0.0));
3174: PetscCall(VecGetArrayWrite(v, &x));
3175: PetscCall(VecGetLocalSize(v, &n));
3176: PetscCheck(n == A->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector");
3177: for (i = 0; i < m; i++) {
3178: ncols = ai[1] - ai[0];
3179: ai++;
3180: if (ncols == A->cmap->n) { /* row is dense */
3181: x[i] = *aa;
3182: if (idx) idx[i] = 0;
3183: } else { /* row is sparse so already KNOW maximum is 0.0 or higher */
3184: x[i] = 0.0;
3185: if (idx) {
3186: for (j = 0; j < ncols; j++) { /* find first implicit 0.0 in the row */
3187: if (aj[j] > j) {
3188: idx[i] = j;
3189: break;
3190: }
3191: }
3192: /* in case first implicit 0.0 in the row occurs at ncols-th column */
3193: if (j == ncols && j < A->cmap->n) idx[i] = j;
3194: }
3195: }
3196: for (j = 0; j < ncols; j++) {
3197: if (PetscRealPart(x[i]) < PetscRealPart(*aa)) {
3198: x[i] = *aa;
3199: if (idx) idx[i] = *aj;
3200: }
3201: aa++;
3202: aj++;
3203: }
3204: }
3205: PetscCall(VecRestoreArrayWrite(v, &x));
3206: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
3207: PetscFunctionReturn(PETSC_SUCCESS);
3208: }
3210: static PetscErrorCode MatGetRowMinAbs_SeqAIJ(Mat A, Vec v, PetscInt idx[])
3211: {
3212: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3213: PetscInt i, j, m = A->rmap->n, *ai, *aj, ncols, n;
3214: PetscScalar *x;
3215: const MatScalar *aa, *av;
3217: PetscFunctionBegin;
3218: PetscCall(MatSeqAIJGetArrayRead(A, &av));
3219: aa = av;
3220: ai = a->i;
3221: aj = a->j;
3223: PetscCall(VecSet(v, 0.0));
3224: PetscCall(VecGetArrayWrite(v, &x));
3225: PetscCall(VecGetLocalSize(v, &n));
3226: PetscCheck(n == m, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector, %" PetscInt_FMT " vs. %" PetscInt_FMT " rows", m, n);
3227: for (i = 0; i < m; i++) {
3228: ncols = ai[1] - ai[0];
3229: ai++;
3230: if (ncols == A->cmap->n) { /* row is dense */
3231: x[i] = *aa;
3232: if (idx) idx[i] = 0;
3233: } else { /* row is sparse so already KNOW minimum is 0.0 or higher */
3234: x[i] = 0.0;
3235: if (idx) { /* find first implicit 0.0 in the row */
3236: for (j = 0; j < ncols; j++) {
3237: if (aj[j] > j) {
3238: idx[i] = j;
3239: break;
3240: }
3241: }
3242: /* in case first implicit 0.0 in the row occurs at ncols-th column */
3243: if (j == ncols && j < A->cmap->n) idx[i] = j;
3244: }
3245: }
3246: for (j = 0; j < ncols; j++) {
3247: if (PetscAbsScalar(x[i]) > PetscAbsScalar(*aa)) {
3248: x[i] = *aa;
3249: if (idx) idx[i] = *aj;
3250: }
3251: aa++;
3252: aj++;
3253: }
3254: }
3255: PetscCall(VecRestoreArrayWrite(v, &x));
3256: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
3257: PetscFunctionReturn(PETSC_SUCCESS);
3258: }
3260: static PetscErrorCode MatGetRowMin_SeqAIJ(Mat A, Vec v, PetscInt idx[])
3261: {
3262: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3263: PetscInt i, j, m = A->rmap->n, ncols, n;
3264: const PetscInt *ai, *aj;
3265: PetscScalar *x;
3266: const MatScalar *aa, *av;
3268: PetscFunctionBegin;
3269: PetscCheck(!A->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3270: PetscCall(MatSeqAIJGetArrayRead(A, &av));
3271: aa = av;
3272: ai = a->i;
3273: aj = a->j;
3275: PetscCall(VecSet(v, 0.0));
3276: PetscCall(VecGetArrayWrite(v, &x));
3277: PetscCall(VecGetLocalSize(v, &n));
3278: PetscCheck(n == m, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Nonconforming matrix and vector");
3279: for (i = 0; i < m; i++) {
3280: ncols = ai[1] - ai[0];
3281: ai++;
3282: if (ncols == A->cmap->n) { /* row is dense */
3283: x[i] = *aa;
3284: if (idx) idx[i] = 0;
3285: } else { /* row is sparse so already KNOW minimum is 0.0 or lower */
3286: x[i] = 0.0;
3287: if (idx) { /* find first implicit 0.0 in the row */
3288: for (j = 0; j < ncols; j++) {
3289: if (aj[j] > j) {
3290: idx[i] = j;
3291: break;
3292: }
3293: }
3294: /* in case first implicit 0.0 in the row occurs at ncols-th column */
3295: if (j == ncols && j < A->cmap->n) idx[i] = j;
3296: }
3297: }
3298: for (j = 0; j < ncols; j++) {
3299: if (PetscRealPart(x[i]) > PetscRealPart(*aa)) {
3300: x[i] = *aa;
3301: if (idx) idx[i] = *aj;
3302: }
3303: aa++;
3304: aj++;
3305: }
3306: }
3307: PetscCall(VecRestoreArrayWrite(v, &x));
3308: PetscCall(MatSeqAIJRestoreArrayRead(A, &av));
3309: PetscFunctionReturn(PETSC_SUCCESS);
3310: }
3312: static PetscErrorCode MatInvertBlockDiagonal_SeqAIJ(Mat A, const PetscScalar **values)
3313: {
3314: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
3315: PetscInt i, bs = PetscAbs(A->rmap->bs), mbs = A->rmap->n / bs, ipvt[5], bs2 = bs * bs, *v_pivots, ij[7], *IJ, j;
3316: MatScalar *diag, work[25], *v_work;
3317: const PetscReal shift = 0.0;
3318: PetscBool allowzeropivot, zeropivotdetected = PETSC_FALSE;
3320: PetscFunctionBegin;
3321: allowzeropivot = PetscNot(A->erroriffailure);
3322: if (a->ibdiagvalid) {
3323: if (values) *values = a->ibdiag;
3324: PetscFunctionReturn(PETSC_SUCCESS);
3325: }
3326: PetscCall(MatMarkDiagonal_SeqAIJ(A));
3327: if (!a->ibdiag) { PetscCall(PetscMalloc1(bs2 * mbs, &a->ibdiag)); }
3328: diag = a->ibdiag;
3329: if (values) *values = a->ibdiag;
3330: /* factor and invert each block */
3331: switch (bs) {
3332: case 1:
3333: for (i = 0; i < mbs; i++) {
3334: PetscCall(MatGetValues(A, 1, &i, 1, &i, diag + i));
3335: if (PetscAbsScalar(diag[i] + shift) < PETSC_MACHINE_EPSILON) {
3336: if (allowzeropivot) {
3337: A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3338: A->factorerror_zeropivot_value = PetscAbsScalar(diag[i]);
3339: A->factorerror_zeropivot_row = i;
3340: PetscCall(PetscInfo(A, "Zero pivot, row %" PetscInt_FMT " pivot %g tolerance %g\n", i, (double)PetscAbsScalar(diag[i]), (double)PETSC_MACHINE_EPSILON));
3341: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_MAT_LU_ZRPVT, "Zero pivot, row %" PetscInt_FMT " pivot %g tolerance %g", i, (double)PetscAbsScalar(diag[i]), (double)PETSC_MACHINE_EPSILON);
3342: }
3343: diag[i] = (PetscScalar)1.0 / (diag[i] + shift);
3344: }
3345: break;
3346: case 2:
3347: for (i = 0; i < mbs; i++) {
3348: ij[0] = 2 * i;
3349: ij[1] = 2 * i + 1;
3350: PetscCall(MatGetValues(A, 2, ij, 2, ij, diag));
3351: PetscCall(PetscKernel_A_gets_inverse_A_2(diag, shift, allowzeropivot, &zeropivotdetected));
3352: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3353: PetscCall(PetscKernel_A_gets_transpose_A_2(diag));
3354: diag += 4;
3355: }
3356: break;
3357: case 3:
3358: for (i = 0; i < mbs; i++) {
3359: ij[0] = 3 * i;
3360: ij[1] = 3 * i + 1;
3361: ij[2] = 3 * i + 2;
3362: PetscCall(MatGetValues(A, 3, ij, 3, ij, diag));
3363: PetscCall(PetscKernel_A_gets_inverse_A_3(diag, shift, allowzeropivot, &zeropivotdetected));
3364: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3365: PetscCall(PetscKernel_A_gets_transpose_A_3(diag));
3366: diag += 9;
3367: }
3368: break;
3369: case 4:
3370: for (i = 0; i < mbs; i++) {
3371: ij[0] = 4 * i;
3372: ij[1] = 4 * i + 1;
3373: ij[2] = 4 * i + 2;
3374: ij[3] = 4 * i + 3;
3375: PetscCall(MatGetValues(A, 4, ij, 4, ij, diag));
3376: PetscCall(PetscKernel_A_gets_inverse_A_4(diag, shift, allowzeropivot, &zeropivotdetected));
3377: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3378: PetscCall(PetscKernel_A_gets_transpose_A_4(diag));
3379: diag += 16;
3380: }
3381: break;
3382: case 5:
3383: for (i = 0; i < mbs; i++) {
3384: ij[0] = 5 * i;
3385: ij[1] = 5 * i + 1;
3386: ij[2] = 5 * i + 2;
3387: ij[3] = 5 * i + 3;
3388: ij[4] = 5 * i + 4;
3389: PetscCall(MatGetValues(A, 5, ij, 5, ij, diag));
3390: PetscCall(PetscKernel_A_gets_inverse_A_5(diag, ipvt, work, shift, allowzeropivot, &zeropivotdetected));
3391: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3392: PetscCall(PetscKernel_A_gets_transpose_A_5(diag));
3393: diag += 25;
3394: }
3395: break;
3396: case 6:
3397: for (i = 0; i < mbs; i++) {
3398: ij[0] = 6 * i;
3399: ij[1] = 6 * i + 1;
3400: ij[2] = 6 * i + 2;
3401: ij[3] = 6 * i + 3;
3402: ij[4] = 6 * i + 4;
3403: ij[5] = 6 * i + 5;
3404: PetscCall(MatGetValues(A, 6, ij, 6, ij, diag));
3405: PetscCall(PetscKernel_A_gets_inverse_A_6(diag, shift, allowzeropivot, &zeropivotdetected));
3406: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3407: PetscCall(PetscKernel_A_gets_transpose_A_6(diag));
3408: diag += 36;
3409: }
3410: break;
3411: case 7:
3412: for (i = 0; i < mbs; i++) {
3413: ij[0] = 7 * i;
3414: ij[1] = 7 * i + 1;
3415: ij[2] = 7 * i + 2;
3416: ij[3] = 7 * i + 3;
3417: ij[4] = 7 * i + 4;
3418: ij[5] = 7 * i + 5;
3419: ij[6] = 7 * i + 6;
3420: PetscCall(MatGetValues(A, 7, ij, 7, ij, diag));
3421: PetscCall(PetscKernel_A_gets_inverse_A_7(diag, shift, allowzeropivot, &zeropivotdetected));
3422: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3423: PetscCall(PetscKernel_A_gets_transpose_A_7(diag));
3424: diag += 49;
3425: }
3426: break;
3427: default:
3428: PetscCall(PetscMalloc3(bs, &v_work, bs, &v_pivots, bs, &IJ));
3429: for (i = 0; i < mbs; i++) {
3430: for (j = 0; j < bs; j++) IJ[j] = bs * i + j;
3431: PetscCall(MatGetValues(A, bs, IJ, bs, IJ, diag));
3432: PetscCall(PetscKernel_A_gets_inverse_A(bs, diag, v_pivots, v_work, allowzeropivot, &zeropivotdetected));
3433: if (zeropivotdetected) A->factorerrortype = MAT_FACTOR_NUMERIC_ZEROPIVOT;
3434: PetscCall(PetscKernel_A_gets_transpose_A_N(diag, bs));
3435: diag += bs2;
3436: }
3437: PetscCall(PetscFree3(v_work, v_pivots, IJ));
3438: }
3439: a->ibdiagvalid = PETSC_TRUE;
3440: PetscFunctionReturn(PETSC_SUCCESS);
3441: }
3443: static PetscErrorCode MatSetRandom_SeqAIJ(Mat x, PetscRandom rctx)
3444: {
3445: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)x->data;
3446: PetscScalar a, *aa;
3447: PetscInt m, n, i, j, col;
3449: PetscFunctionBegin;
3450: if (!x->assembled) {
3451: PetscCall(MatGetSize(x, &m, &n));
3452: for (i = 0; i < m; i++) {
3453: for (j = 0; j < aij->imax[i]; j++) {
3454: PetscCall(PetscRandomGetValue(rctx, &a));
3455: col = (PetscInt)(n * PetscRealPart(a));
3456: PetscCall(MatSetValues(x, 1, &i, 1, &col, &a, ADD_VALUES));
3457: }
3458: }
3459: } else {
3460: PetscCall(MatSeqAIJGetArrayWrite(x, &aa));
3461: for (i = 0; i < aij->nz; i++) PetscCall(PetscRandomGetValue(rctx, aa + i));
3462: PetscCall(MatSeqAIJRestoreArrayWrite(x, &aa));
3463: }
3464: PetscCall(MatAssemblyBegin(x, MAT_FINAL_ASSEMBLY));
3465: PetscCall(MatAssemblyEnd(x, MAT_FINAL_ASSEMBLY));
3466: PetscFunctionReturn(PETSC_SUCCESS);
3467: }
3469: /* Like MatSetRandom_SeqAIJ, but do not set values on columns in range of [low, high) */
3470: PetscErrorCode MatSetRandomSkipColumnRange_SeqAIJ_Private(Mat x, PetscInt low, PetscInt high, PetscRandom rctx)
3471: {
3472: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)x->data;
3473: PetscScalar a;
3474: PetscInt m, n, i, j, col, nskip;
3476: PetscFunctionBegin;
3477: nskip = high - low;
3478: PetscCall(MatGetSize(x, &m, &n));
3479: n -= nskip; /* shrink number of columns where nonzeros can be set */
3480: for (i = 0; i < m; i++) {
3481: for (j = 0; j < aij->imax[i]; j++) {
3482: PetscCall(PetscRandomGetValue(rctx, &a));
3483: col = (PetscInt)(n * PetscRealPart(a));
3484: if (col >= low) col += nskip; /* shift col rightward to skip the hole */
3485: PetscCall(MatSetValues(x, 1, &i, 1, &col, &a, ADD_VALUES));
3486: }
3487: }
3488: PetscCall(MatAssemblyBegin(x, MAT_FINAL_ASSEMBLY));
3489: PetscCall(MatAssemblyEnd(x, MAT_FINAL_ASSEMBLY));
3490: PetscFunctionReturn(PETSC_SUCCESS);
3491: }
3493: static struct _MatOps MatOps_Values = {MatSetValues_SeqAIJ,
3494: MatGetRow_SeqAIJ,
3495: MatRestoreRow_SeqAIJ,
3496: MatMult_SeqAIJ,
3497: /* 4*/ MatMultAdd_SeqAIJ,
3498: MatMultTranspose_SeqAIJ,
3499: MatMultTransposeAdd_SeqAIJ,
3500: NULL,
3501: NULL,
3502: NULL,
3503: /* 10*/ NULL,
3504: MatLUFactor_SeqAIJ,
3505: NULL,
3506: MatSOR_SeqAIJ,
3507: MatTranspose_SeqAIJ,
3508: /*1 5*/ MatGetInfo_SeqAIJ,
3509: MatEqual_SeqAIJ,
3510: MatGetDiagonal_SeqAIJ,
3511: MatDiagonalScale_SeqAIJ,
3512: MatNorm_SeqAIJ,
3513: /* 20*/ NULL,
3514: MatAssemblyEnd_SeqAIJ,
3515: MatSetOption_SeqAIJ,
3516: MatZeroEntries_SeqAIJ,
3517: /* 24*/ MatZeroRows_SeqAIJ,
3518: NULL,
3519: NULL,
3520: NULL,
3521: NULL,
3522: /* 29*/ MatSetUp_Seq_Hash,
3523: NULL,
3524: NULL,
3525: NULL,
3526: NULL,
3527: /* 34*/ MatDuplicate_SeqAIJ,
3528: NULL,
3529: NULL,
3530: MatILUFactor_SeqAIJ,
3531: NULL,
3532: /* 39*/ MatAXPY_SeqAIJ,
3533: MatCreateSubMatrices_SeqAIJ,
3534: MatIncreaseOverlap_SeqAIJ,
3535: MatGetValues_SeqAIJ,
3536: MatCopy_SeqAIJ,
3537: /* 44*/ MatGetRowMax_SeqAIJ,
3538: MatScale_SeqAIJ,
3539: MatShift_SeqAIJ,
3540: MatDiagonalSet_SeqAIJ,
3541: MatZeroRowsColumns_SeqAIJ,
3542: /* 49*/ MatSetRandom_SeqAIJ,
3543: MatGetRowIJ_SeqAIJ,
3544: MatRestoreRowIJ_SeqAIJ,
3545: MatGetColumnIJ_SeqAIJ,
3546: MatRestoreColumnIJ_SeqAIJ,
3547: /* 54*/ MatFDColoringCreate_SeqXAIJ,
3548: NULL,
3549: NULL,
3550: MatPermute_SeqAIJ,
3551: NULL,
3552: /* 59*/ NULL,
3553: MatDestroy_SeqAIJ,
3554: MatView_SeqAIJ,
3555: NULL,
3556: NULL,
3557: /* 64*/ NULL,
3558: MatMatMatMultNumeric_SeqAIJ_SeqAIJ_SeqAIJ,
3559: NULL,
3560: NULL,
3561: NULL,
3562: /* 69*/ MatGetRowMaxAbs_SeqAIJ,
3563: MatGetRowMinAbs_SeqAIJ,
3564: NULL,
3565: NULL,
3566: NULL,
3567: /* 74*/ NULL,
3568: MatFDColoringApply_AIJ,
3569: NULL,
3570: NULL,
3571: NULL,
3572: /* 79*/ MatFindZeroDiagonals_SeqAIJ,
3573: NULL,
3574: NULL,
3575: NULL,
3576: MatLoad_SeqAIJ,
3577: /* 84*/ NULL,
3578: NULL,
3579: NULL,
3580: NULL,
3581: NULL,
3582: /* 89*/ NULL,
3583: NULL,
3584: MatMatMultNumeric_SeqAIJ_SeqAIJ,
3585: NULL,
3586: NULL,
3587: /* 94*/ MatPtAPNumeric_SeqAIJ_SeqAIJ_SparseAxpy,
3588: NULL,
3589: NULL,
3590: MatMatTransposeMultNumeric_SeqAIJ_SeqAIJ,
3591: NULL,
3592: /* 99*/ MatProductSetFromOptions_SeqAIJ,
3593: NULL,
3594: NULL,
3595: MatConjugate_SeqAIJ,
3596: NULL,
3597: /*104*/ MatSetValuesRow_SeqAIJ,
3598: MatRealPart_SeqAIJ,
3599: MatImaginaryPart_SeqAIJ,
3600: NULL,
3601: NULL,
3602: /*109*/ MatMatSolve_SeqAIJ,
3603: NULL,
3604: MatGetRowMin_SeqAIJ,
3605: NULL,
3606: MatMissingDiagonal_SeqAIJ,
3607: /*114*/ NULL,
3608: NULL,
3609: NULL,
3610: NULL,
3611: NULL,
3612: /*119*/ NULL,
3613: NULL,
3614: NULL,
3615: NULL,
3616: MatGetMultiProcBlock_SeqAIJ,
3617: /*124*/ MatFindNonzeroRows_SeqAIJ,
3618: MatGetColumnReductions_SeqAIJ,
3619: MatInvertBlockDiagonal_SeqAIJ,
3620: MatInvertVariableBlockDiagonal_SeqAIJ,
3621: NULL,
3622: /*129*/ NULL,
3623: NULL,
3624: NULL,
3625: MatTransposeMatMultNumeric_SeqAIJ_SeqAIJ,
3626: MatTransposeColoringCreate_SeqAIJ,
3627: /*134*/ MatTransColoringApplySpToDen_SeqAIJ,
3628: MatTransColoringApplyDenToSp_SeqAIJ,
3629: NULL,
3630: NULL,
3631: MatRARtNumeric_SeqAIJ_SeqAIJ,
3632: /*139*/ NULL,
3633: NULL,
3634: NULL,
3635: MatFDColoringSetUp_SeqXAIJ,
3636: MatFindOffBlockDiagonalEntries_SeqAIJ,
3637: MatCreateMPIMatConcatenateSeqMat_SeqAIJ,
3638: /*145*/ MatDestroySubMatrices_SeqAIJ,
3639: NULL,
3640: NULL,
3641: MatCreateGraph_Simple_AIJ,
3642: NULL,
3643: /*150*/ MatTransposeSymbolic_SeqAIJ,
3644: MatEliminateZeros_SeqAIJ,
3645: MatGetRowSumAbs_SeqAIJ,
3646: NULL,
3647: NULL,
3648: /*155*/ NULL,
3649: MatCopyHashToXAIJ_Seq_Hash};
3651: static PetscErrorCode MatSeqAIJSetColumnIndices_SeqAIJ(Mat mat, PetscInt *indices)
3652: {
3653: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
3654: PetscInt i, nz, n;
3656: PetscFunctionBegin;
3657: nz = aij->maxnz;
3658: n = mat->rmap->n;
3659: for (i = 0; i < nz; i++) aij->j[i] = indices[i];
3660: aij->nz = nz;
3661: for (i = 0; i < n; i++) aij->ilen[i] = aij->imax[i];
3662: PetscFunctionReturn(PETSC_SUCCESS);
3663: }
3665: /*
3666: * Given a sparse matrix with global column indices, compact it by using a local column space.
3667: * The result matrix helps saving memory in other algorithms, such as MatPtAPSymbolic_MPIAIJ_MPIAIJ_scalable()
3668: */
3669: PetscErrorCode MatSeqAIJCompactOutExtraColumns_SeqAIJ(Mat mat, ISLocalToGlobalMapping *mapping)
3670: {
3671: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
3672: PetscHMapI gid1_lid1;
3673: PetscHashIter tpos;
3674: PetscInt gid, lid, i, ec, nz = aij->nz;
3675: PetscInt *garray, *jj = aij->j;
3677: PetscFunctionBegin;
3679: PetscAssertPointer(mapping, 2);
3680: /* use a table */
3681: PetscCall(PetscHMapICreateWithSize(mat->rmap->n, &gid1_lid1));
3682: ec = 0;
3683: for (i = 0; i < nz; i++) {
3684: PetscInt data, gid1 = jj[i] + 1;
3685: PetscCall(PetscHMapIGetWithDefault(gid1_lid1, gid1, 0, &data));
3686: if (!data) {
3687: /* one based table */
3688: PetscCall(PetscHMapISet(gid1_lid1, gid1, ++ec));
3689: }
3690: }
3691: /* form array of columns we need */
3692: PetscCall(PetscMalloc1(ec, &garray));
3693: PetscHashIterBegin(gid1_lid1, tpos);
3694: while (!PetscHashIterAtEnd(gid1_lid1, tpos)) {
3695: PetscHashIterGetKey(gid1_lid1, tpos, gid);
3696: PetscHashIterGetVal(gid1_lid1, tpos, lid);
3697: PetscHashIterNext(gid1_lid1, tpos);
3698: gid--;
3699: lid--;
3700: garray[lid] = gid;
3701: }
3702: PetscCall(PetscSortInt(ec, garray)); /* sort, and rebuild */
3703: PetscCall(PetscHMapIClear(gid1_lid1));
3704: for (i = 0; i < ec; i++) PetscCall(PetscHMapISet(gid1_lid1, garray[i] + 1, i + 1));
3705: /* compact out the extra columns in B */
3706: for (i = 0; i < nz; i++) {
3707: PetscInt gid1 = jj[i] + 1;
3708: PetscCall(PetscHMapIGetWithDefault(gid1_lid1, gid1, 0, &lid));
3709: lid--;
3710: jj[i] = lid;
3711: }
3712: PetscCall(PetscLayoutDestroy(&mat->cmap));
3713: PetscCall(PetscHMapIDestroy(&gid1_lid1));
3714: PetscCall(PetscLayoutCreateFromSizes(PetscObjectComm((PetscObject)mat), ec, ec, 1, &mat->cmap));
3715: PetscCall(ISLocalToGlobalMappingCreate(PETSC_COMM_SELF, mat->cmap->bs, mat->cmap->n, garray, PETSC_OWN_POINTER, mapping));
3716: PetscCall(ISLocalToGlobalMappingSetType(*mapping, ISLOCALTOGLOBALMAPPINGHASH));
3717: PetscFunctionReturn(PETSC_SUCCESS);
3718: }
3720: /*@
3721: MatSeqAIJSetColumnIndices - Set the column indices for all the rows
3722: in the matrix.
3724: Input Parameters:
3725: + mat - the `MATSEQAIJ` matrix
3726: - indices - the column indices
3728: Level: advanced
3730: Notes:
3731: This can be called if you have precomputed the nonzero structure of the
3732: matrix and want to provide it to the matrix object to improve the performance
3733: of the `MatSetValues()` operation.
3735: You MUST have set the correct numbers of nonzeros per row in the call to
3736: `MatCreateSeqAIJ()`, and the columns indices MUST be sorted.
3738: MUST be called before any calls to `MatSetValues()`
3740: The indices should start with zero, not one.
3742: .seealso: [](ch_matrices), `Mat`, `MATSEQAIJ`
3743: @*/
3744: PetscErrorCode MatSeqAIJSetColumnIndices(Mat mat, PetscInt *indices)
3745: {
3746: PetscFunctionBegin;
3748: PetscAssertPointer(indices, 2);
3749: PetscUseMethod(mat, "MatSeqAIJSetColumnIndices_C", (Mat, PetscInt *), (mat, indices));
3750: PetscFunctionReturn(PETSC_SUCCESS);
3751: }
3753: static PetscErrorCode MatStoreValues_SeqAIJ(Mat mat)
3754: {
3755: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
3756: size_t nz = aij->i[mat->rmap->n];
3758: PetscFunctionBegin;
3759: PetscCheck(aij->nonew, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
3761: /* allocate space for values if not already there */
3762: if (!aij->saved_values) { PetscCall(PetscMalloc1(nz + 1, &aij->saved_values)); }
3764: /* copy values over */
3765: PetscCall(PetscArraycpy(aij->saved_values, aij->a, nz));
3766: PetscFunctionReturn(PETSC_SUCCESS);
3767: }
3769: /*@
3770: MatStoreValues - Stashes a copy of the matrix values; this allows reusing of the linear part of a Jacobian, while recomputing only the
3771: nonlinear portion.
3773: Logically Collect
3775: Input Parameter:
3776: . mat - the matrix (currently only `MATAIJ` matrices support this option)
3778: Level: advanced
3780: Example Usage:
3781: .vb
3782: Using SNES
3783: Create Jacobian matrix
3784: Set linear terms into matrix
3785: Apply boundary conditions to matrix, at this time matrix must have
3786: final nonzero structure (i.e. setting the nonlinear terms and applying
3787: boundary conditions again will not change the nonzero structure
3788: MatSetOption(mat, MAT_NEW_NONZERO_LOCATIONS, PETSC_FALSE);
3789: MatStoreValues(mat);
3790: Call SNESSetJacobian() with matrix
3791: In your Jacobian routine
3792: MatRetrieveValues(mat);
3793: Set nonlinear terms in matrix
3795: Without `SNESSolve()`, i.e. when you handle nonlinear solve yourself:
3796: // build linear portion of Jacobian
3797: MatSetOption(mat, MAT_NEW_NONZERO_LOCATIONS, PETSC_FALSE);
3798: MatStoreValues(mat);
3799: loop over nonlinear iterations
3800: MatRetrieveValues(mat);
3801: // call MatSetValues(mat,...) to set nonliner portion of Jacobian
3802: // call MatAssemblyBegin/End() on matrix
3803: Solve linear system with Jacobian
3804: endloop
3805: .ve
3807: Notes:
3808: Matrix must already be assembled before calling this routine
3809: Must set the matrix option `MatSetOption`(mat,`MAT_NEW_NONZERO_LOCATIONS`,`PETSC_FALSE`); before
3810: calling this routine.
3812: When this is called multiple times it overwrites the previous set of stored values
3813: and does not allocated additional space.
3815: .seealso: [](ch_matrices), `Mat`, `MatRetrieveValues()`
3816: @*/
3817: PetscErrorCode MatStoreValues(Mat mat)
3818: {
3819: PetscFunctionBegin;
3821: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3822: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3823: PetscUseMethod(mat, "MatStoreValues_C", (Mat), (mat));
3824: PetscFunctionReturn(PETSC_SUCCESS);
3825: }
3827: static PetscErrorCode MatRetrieveValues_SeqAIJ(Mat mat)
3828: {
3829: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
3830: PetscInt nz = aij->i[mat->rmap->n];
3832: PetscFunctionBegin;
3833: PetscCheck(aij->nonew, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first");
3834: PetscCheck(aij->saved_values, PETSC_COMM_SELF, PETSC_ERR_ORDER, "Must call MatStoreValues(A);first");
3835: /* copy values over */
3836: PetscCall(PetscArraycpy(aij->a, aij->saved_values, nz));
3837: PetscFunctionReturn(PETSC_SUCCESS);
3838: }
3840: /*@
3841: MatRetrieveValues - Retrieves the copy of the matrix values that was stored with `MatStoreValues()`
3843: Logically Collect
3845: Input Parameter:
3846: . mat - the matrix (currently only `MATAIJ` matrices support this option)
3848: Level: advanced
3850: .seealso: [](ch_matrices), `Mat`, `MatStoreValues()`
3851: @*/
3852: PetscErrorCode MatRetrieveValues(Mat mat)
3853: {
3854: PetscFunctionBegin;
3856: PetscCheck(mat->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
3857: PetscCheck(!mat->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
3858: PetscUseMethod(mat, "MatRetrieveValues_C", (Mat), (mat));
3859: PetscFunctionReturn(PETSC_SUCCESS);
3860: }
3862: /*@
3863: MatCreateSeqAIJ - Creates a sparse matrix in `MATSEQAIJ` (compressed row) format
3864: (the default parallel PETSc format). For good matrix assembly performance
3865: the user should preallocate the matrix storage by setting the parameter `nz`
3866: (or the array `nnz`).
3868: Collective
3870: Input Parameters:
3871: + comm - MPI communicator, set to `PETSC_COMM_SELF`
3872: . m - number of rows
3873: . n - number of columns
3874: . nz - number of nonzeros per row (same for all rows)
3875: - nnz - array containing the number of nonzeros in the various rows
3876: (possibly different for each row) or NULL
3878: Output Parameter:
3879: . A - the matrix
3881: Options Database Keys:
3882: + -mat_no_inode - Do not use inodes
3883: - -mat_inode_limit <limit> - Sets inode limit (max limit=5)
3885: Level: intermediate
3887: Notes:
3888: It is recommend to use `MatCreateFromOptions()` instead of this routine
3890: If `nnz` is given then `nz` is ignored
3892: The `MATSEQAIJ` format, also called
3893: compressed row storage, is fully compatible with standard Fortran
3894: storage. That is, the stored row and column indices can begin at
3895: either one (as in Fortran) or zero.
3897: Specify the preallocated storage with either `nz` or `nnz` (not both).
3898: Set `nz` = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
3899: allocation.
3901: By default, this format uses inodes (identical nodes) when possible, to
3902: improve numerical efficiency of matrix-vector products and solves. We
3903: search for consecutive rows with the same nonzero structure, thereby
3904: reusing matrix information to achieve increased efficiency.
3906: .seealso: [](ch_matrices), `Mat`, [Sparse Matrix Creation](sec_matsparse), `MatCreate()`, `MatCreateAIJ()`, `MatSetValues()`, `MatSeqAIJSetColumnIndices()`, `MatCreateSeqAIJWithArrays()`
3907: @*/
3908: PetscErrorCode MatCreateSeqAIJ(MPI_Comm comm, PetscInt m, PetscInt n, PetscInt nz, const PetscInt nnz[], Mat *A)
3909: {
3910: PetscFunctionBegin;
3911: PetscCall(MatCreate(comm, A));
3912: PetscCall(MatSetSizes(*A, m, n, m, n));
3913: PetscCall(MatSetType(*A, MATSEQAIJ));
3914: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(*A, nz, nnz));
3915: PetscFunctionReturn(PETSC_SUCCESS);
3916: }
3918: /*@
3919: MatSeqAIJSetPreallocation - For good matrix assembly performance
3920: the user should preallocate the matrix storage by setting the parameter nz
3921: (or the array nnz). By setting these parameters accurately, performance
3922: during matrix assembly can be increased by more than a factor of 50.
3924: Collective
3926: Input Parameters:
3927: + B - The matrix
3928: . nz - number of nonzeros per row (same for all rows)
3929: - nnz - array containing the number of nonzeros in the various rows
3930: (possibly different for each row) or NULL
3932: Options Database Keys:
3933: + -mat_no_inode - Do not use inodes
3934: - -mat_inode_limit <limit> - Sets inode limit (max limit=5)
3936: Level: intermediate
3938: Notes:
3939: If `nnz` is given then `nz` is ignored
3941: The `MATSEQAIJ` format also called
3942: compressed row storage, is fully compatible with standard Fortran
3943: storage. That is, the stored row and column indices can begin at
3944: either one (as in Fortran) or zero. See the users' manual for details.
3946: Specify the preallocated storage with either `nz` or `nnz` (not both).
3947: Set nz = `PETSC_DEFAULT` and `nnz` = `NULL` for PETSc to control dynamic memory
3948: allocation.
3950: You can call `MatGetInfo()` to get information on how effective the preallocation was;
3951: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
3952: You can also run with the option -info and look for messages with the string
3953: malloc in them to see if additional memory allocation was needed.
3955: Developer Notes:
3956: Use nz of `MAT_SKIP_ALLOCATION` to not allocate any space for the matrix
3957: entries or columns indices
3959: By default, this format uses inodes (identical nodes) when possible, to
3960: improve numerical efficiency of matrix-vector products and solves. We
3961: search for consecutive rows with the same nonzero structure, thereby
3962: reusing matrix information to achieve increased efficiency.
3964: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatCreateAIJ()`, `MatSetValues()`, `MatSeqAIJSetColumnIndices()`, `MatCreateSeqAIJWithArrays()`, `MatGetInfo()`,
3965: `MatSeqAIJSetTotalPreallocation()`
3966: @*/
3967: PetscErrorCode MatSeqAIJSetPreallocation(Mat B, PetscInt nz, const PetscInt nnz[])
3968: {
3969: PetscFunctionBegin;
3972: PetscTryMethod(B, "MatSeqAIJSetPreallocation_C", (Mat, PetscInt, const PetscInt[]), (B, nz, nnz));
3973: PetscFunctionReturn(PETSC_SUCCESS);
3974: }
3976: PetscErrorCode MatSeqAIJSetPreallocation_SeqAIJ(Mat B, PetscInt nz, const PetscInt *nnz)
3977: {
3978: Mat_SeqAIJ *b = (Mat_SeqAIJ *)B->data;
3979: PetscBool skipallocation = PETSC_FALSE, realalloc = PETSC_FALSE;
3980: PetscInt i;
3982: PetscFunctionBegin;
3983: if (B->hash_active) {
3984: B->ops[0] = b->cops;
3985: PetscCall(PetscHMapIJVDestroy(&b->ht));
3986: PetscCall(PetscFree(b->dnz));
3987: B->hash_active = PETSC_FALSE;
3988: }
3989: if (nz >= 0 || nnz) realalloc = PETSC_TRUE;
3990: if (nz == MAT_SKIP_ALLOCATION) {
3991: skipallocation = PETSC_TRUE;
3992: nz = 0;
3993: }
3994: PetscCall(PetscLayoutSetUp(B->rmap));
3995: PetscCall(PetscLayoutSetUp(B->cmap));
3997: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
3998: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nz cannot be less than 0: value %" PetscInt_FMT, nz);
3999: if (nnz) {
4000: for (i = 0; i < B->rmap->n; i++) {
4001: PetscCheck(nnz[i] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nnz cannot be less than 0: local row %" PetscInt_FMT " value %" PetscInt_FMT, i, nnz[i]);
4002: PetscCheck(nnz[i] <= B->cmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "nnz cannot be greater than row length: local row %" PetscInt_FMT " value %" PetscInt_FMT " rowlength %" PetscInt_FMT, i, nnz[i], B->cmap->n);
4003: }
4004: }
4006: B->preallocated = PETSC_TRUE;
4007: if (!skipallocation) {
4008: if (!b->imax) { PetscCall(PetscMalloc1(B->rmap->n, &b->imax)); }
4009: if (!b->ilen) {
4010: /* b->ilen will count nonzeros in each row so far. */
4011: PetscCall(PetscCalloc1(B->rmap->n, &b->ilen));
4012: } else {
4013: PetscCall(PetscMemzero(b->ilen, B->rmap->n * sizeof(PetscInt)));
4014: }
4015: if (!b->ipre) PetscCall(PetscMalloc1(B->rmap->n, &b->ipre));
4016: if (!nnz) {
4017: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 10;
4018: else if (nz < 0) nz = 1;
4019: nz = PetscMin(nz, B->cmap->n);
4020: for (i = 0; i < B->rmap->n; i++) b->imax[i] = nz;
4021: PetscCall(PetscIntMultError(nz, B->rmap->n, &nz));
4022: } else {
4023: PetscInt64 nz64 = 0;
4024: for (i = 0; i < B->rmap->n; i++) {
4025: b->imax[i] = nnz[i];
4026: nz64 += nnz[i];
4027: }
4028: PetscCall(PetscIntCast(nz64, &nz));
4029: }
4031: /* allocate the matrix space */
4032: PetscCall(MatSeqXAIJFreeAIJ(B, &b->a, &b->j, &b->i));
4033: PetscCall(PetscShmgetAllocateArray(nz, sizeof(PetscInt), (void **)&b->j));
4034: PetscCall(PetscShmgetAllocateArray(B->rmap->n + 1, sizeof(PetscInt), (void **)&b->i));
4035: b->free_ij = PETSC_TRUE;
4036: if (B->structure_only) {
4037: b->free_a = PETSC_FALSE;
4038: } else {
4039: PetscCall(PetscShmgetAllocateArray(nz, sizeof(PetscScalar), (void **)&b->a));
4040: b->free_a = PETSC_TRUE;
4041: }
4042: b->i[0] = 0;
4043: for (i = 1; i < B->rmap->n + 1; i++) b->i[i] = b->i[i - 1] + b->imax[i - 1];
4044: } else {
4045: b->free_a = PETSC_FALSE;
4046: b->free_ij = PETSC_FALSE;
4047: }
4049: if (b->ipre && nnz != b->ipre && b->imax) {
4050: /* reserve user-requested sparsity */
4051: PetscCall(PetscArraycpy(b->ipre, b->imax, B->rmap->n));
4052: }
4054: b->nz = 0;
4055: b->maxnz = nz;
4056: B->info.nz_unneeded = (double)b->maxnz;
4057: if (realalloc) PetscCall(MatSetOption(B, MAT_NEW_NONZERO_ALLOCATION_ERR, PETSC_TRUE));
4058: B->was_assembled = PETSC_FALSE;
4059: B->assembled = PETSC_FALSE;
4060: /* We simply deem preallocation has changed nonzero state. Updating the state
4061: will give clients (like AIJKokkos) a chance to know something has happened.
4062: */
4063: B->nonzerostate++;
4064: PetscFunctionReturn(PETSC_SUCCESS);
4065: }
4067: static PetscErrorCode MatResetPreallocation_SeqAIJ(Mat A)
4068: {
4069: Mat_SeqAIJ *a;
4070: PetscInt i;
4071: PetscBool skipreset;
4073: PetscFunctionBegin;
4076: /* Check local size. If zero, then return */
4077: if (!A->rmap->n) PetscFunctionReturn(PETSC_SUCCESS);
4079: a = (Mat_SeqAIJ *)A->data;
4080: /* if no saved info, we error out */
4081: PetscCheck(a->ipre, PETSC_COMM_SELF, PETSC_ERR_ARG_NULL, "No saved preallocation info ");
4083: PetscCheck(a->i && a->imax && a->ilen, PETSC_COMM_SELF, PETSC_ERR_ARG_NULL, "Memory info is incomplete, and can not reset preallocation ");
4085: PetscCall(PetscArraycmp(a->ipre, a->ilen, A->rmap->n, &skipreset));
4086: if (!skipreset) {
4087: PetscCall(PetscArraycpy(a->imax, a->ipre, A->rmap->n));
4088: PetscCall(PetscArrayzero(a->ilen, A->rmap->n));
4089: a->i[0] = 0;
4090: for (i = 1; i < A->rmap->n + 1; i++) a->i[i] = a->i[i - 1] + a->imax[i - 1];
4091: A->preallocated = PETSC_TRUE;
4092: a->nz = 0;
4093: a->maxnz = a->i[A->rmap->n];
4094: A->info.nz_unneeded = (double)a->maxnz;
4095: A->was_assembled = PETSC_FALSE;
4096: A->assembled = PETSC_FALSE;
4097: }
4098: PetscFunctionReturn(PETSC_SUCCESS);
4099: }
4101: /*@
4102: MatSeqAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in `MATSEQAIJ` format.
4104: Input Parameters:
4105: + B - the matrix
4106: . i - the indices into `j` for the start of each row (indices start with zero)
4107: . j - the column indices for each row (indices start with zero) these must be sorted for each row
4108: - v - optional values in the matrix, use `NULL` if not provided
4110: Level: developer
4112: Notes:
4113: The `i`,`j`,`v` values are COPIED with this routine; to avoid the copy use `MatCreateSeqAIJWithArrays()`
4115: This routine may be called multiple times with different nonzero patterns (or the same nonzero pattern). The nonzero
4116: structure will be the union of all the previous nonzero structures.
4118: Developer Notes:
4119: An optimization could be added to the implementation where it checks if the `i`, and `j` are identical to the current `i` and `j` and
4120: then just copies the `v` values directly with `PetscMemcpy()`.
4122: This routine could also take a `PetscCopyMode` argument to allow sharing the values instead of always copying them.
4124: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatSeqAIJSetPreallocation()`, `MATSEQAIJ`, `MatResetPreallocation()`
4125: @*/
4126: PetscErrorCode MatSeqAIJSetPreallocationCSR(Mat B, const PetscInt i[], const PetscInt j[], const PetscScalar v[])
4127: {
4128: PetscFunctionBegin;
4131: PetscTryMethod(B, "MatSeqAIJSetPreallocationCSR_C", (Mat, const PetscInt[], const PetscInt[], const PetscScalar[]), (B, i, j, v));
4132: PetscFunctionReturn(PETSC_SUCCESS);
4133: }
4135: static PetscErrorCode MatSeqAIJSetPreallocationCSR_SeqAIJ(Mat B, const PetscInt Ii[], const PetscInt J[], const PetscScalar v[])
4136: {
4137: PetscInt i;
4138: PetscInt m, n;
4139: PetscInt nz;
4140: PetscInt *nnz;
4142: PetscFunctionBegin;
4143: PetscCheck(Ii[0] == 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Ii[0] must be 0 it is %" PetscInt_FMT, Ii[0]);
4145: PetscCall(PetscLayoutSetUp(B->rmap));
4146: PetscCall(PetscLayoutSetUp(B->cmap));
4148: PetscCall(MatGetSize(B, &m, &n));
4149: PetscCall(PetscMalloc1(m + 1, &nnz));
4150: for (i = 0; i < m; i++) {
4151: nz = Ii[i + 1] - Ii[i];
4152: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Local row %" PetscInt_FMT " has a negative number of columns %" PetscInt_FMT, i, nz);
4153: nnz[i] = nz;
4154: }
4155: PetscCall(MatSeqAIJSetPreallocation(B, 0, nnz));
4156: PetscCall(PetscFree(nnz));
4158: for (i = 0; i < m; i++) PetscCall(MatSetValues_SeqAIJ(B, 1, &i, Ii[i + 1] - Ii[i], J + Ii[i], PetscSafePointerPlusOffset(v, Ii[i]), INSERT_VALUES));
4160: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
4161: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
4163: PetscCall(MatSetOption(B, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
4164: PetscFunctionReturn(PETSC_SUCCESS);
4165: }
4167: /*@
4168: MatSeqAIJKron - Computes `C`, the Kronecker product of `A` and `B`.
4170: Input Parameters:
4171: + A - left-hand side matrix
4172: . B - right-hand side matrix
4173: - reuse - either `MAT_INITIAL_MATRIX` or `MAT_REUSE_MATRIX`
4175: Output Parameter:
4176: . C - Kronecker product of `A` and `B`
4178: Level: intermediate
4180: Note:
4181: `MAT_REUSE_MATRIX` can only be used when the nonzero structure of the product matrix has not changed from that last call to `MatSeqAIJKron()`.
4183: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqAIJ()`, `MATSEQAIJ`, `MATKAIJ`, `MatReuse`
4184: @*/
4185: PetscErrorCode MatSeqAIJKron(Mat A, Mat B, MatReuse reuse, Mat *C)
4186: {
4187: PetscFunctionBegin;
4192: PetscAssertPointer(C, 4);
4193: if (reuse == MAT_REUSE_MATRIX) {
4196: }
4197: PetscTryMethod(A, "MatSeqAIJKron_C", (Mat, Mat, MatReuse, Mat *), (A, B, reuse, C));
4198: PetscFunctionReturn(PETSC_SUCCESS);
4199: }
4201: static PetscErrorCode MatSeqAIJKron_SeqAIJ(Mat A, Mat B, MatReuse reuse, Mat *C)
4202: {
4203: Mat newmat;
4204: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
4205: Mat_SeqAIJ *b = (Mat_SeqAIJ *)B->data;
4206: PetscScalar *v;
4207: const PetscScalar *aa, *ba;
4208: PetscInt *i, *j, m, n, p, q, nnz = 0, am = A->rmap->n, bm = B->rmap->n, an = A->cmap->n, bn = B->cmap->n;
4209: PetscBool flg;
4211: PetscFunctionBegin;
4212: PetscCheck(!A->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4213: PetscCheck(A->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4214: PetscCheck(!B->factortype, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for factored matrix");
4215: PetscCheck(B->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Not for unassembled matrix");
4216: PetscCall(PetscObjectTypeCompare((PetscObject)B, MATSEQAIJ, &flg));
4217: PetscCheck(flg, PETSC_COMM_SELF, PETSC_ERR_SUP, "MatType %s", ((PetscObject)B)->type_name);
4218: PetscCheck(reuse == MAT_INITIAL_MATRIX || reuse == MAT_REUSE_MATRIX, PETSC_COMM_SELF, PETSC_ERR_SUP, "MatReuse %d", (int)reuse);
4219: if (reuse == MAT_INITIAL_MATRIX) {
4220: PetscCall(PetscMalloc2(am * bm + 1, &i, a->i[am] * b->i[bm], &j));
4221: PetscCall(MatCreate(PETSC_COMM_SELF, &newmat));
4222: PetscCall(MatSetSizes(newmat, am * bm, an * bn, am * bm, an * bn));
4223: PetscCall(MatSetType(newmat, MATAIJ));
4224: i[0] = 0;
4225: for (m = 0; m < am; ++m) {
4226: for (p = 0; p < bm; ++p) {
4227: i[m * bm + p + 1] = i[m * bm + p] + (a->i[m + 1] - a->i[m]) * (b->i[p + 1] - b->i[p]);
4228: for (n = a->i[m]; n < a->i[m + 1]; ++n) {
4229: for (q = b->i[p]; q < b->i[p + 1]; ++q) j[nnz++] = a->j[n] * bn + b->j[q];
4230: }
4231: }
4232: }
4233: PetscCall(MatSeqAIJSetPreallocationCSR(newmat, i, j, NULL));
4234: *C = newmat;
4235: PetscCall(PetscFree2(i, j));
4236: nnz = 0;
4237: }
4238: PetscCall(MatSeqAIJGetArray(*C, &v));
4239: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
4240: PetscCall(MatSeqAIJGetArrayRead(B, &ba));
4241: for (m = 0; m < am; ++m) {
4242: for (p = 0; p < bm; ++p) {
4243: for (n = a->i[m]; n < a->i[m + 1]; ++n) {
4244: for (q = b->i[p]; q < b->i[p + 1]; ++q) v[nnz++] = aa[n] * ba[q];
4245: }
4246: }
4247: }
4248: PetscCall(MatSeqAIJRestoreArray(*C, &v));
4249: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
4250: PetscCall(MatSeqAIJRestoreArrayRead(B, &ba));
4251: PetscFunctionReturn(PETSC_SUCCESS);
4252: }
4254: #include <../src/mat/impls/dense/seq/dense.h>
4255: #include <petsc/private/kernels/petscaxpy.h>
4257: /*
4258: Computes (B'*A')' since computing B*A directly is untenable
4260: n p p
4261: [ ] [ ] [ ]
4262: m [ A ] * n [ B ] = m [ C ]
4263: [ ] [ ] [ ]
4265: */
4266: PetscErrorCode MatMatMultNumeric_SeqDense_SeqAIJ(Mat A, Mat B, Mat C)
4267: {
4268: Mat_SeqDense *sub_a = (Mat_SeqDense *)A->data;
4269: Mat_SeqAIJ *sub_b = (Mat_SeqAIJ *)B->data;
4270: Mat_SeqDense *sub_c = (Mat_SeqDense *)C->data;
4271: PetscInt i, j, n, m, q, p;
4272: const PetscInt *ii, *idx;
4273: const PetscScalar *b, *a, *a_q;
4274: PetscScalar *c, *c_q;
4275: PetscInt clda = sub_c->lda;
4276: PetscInt alda = sub_a->lda;
4278: PetscFunctionBegin;
4279: m = A->rmap->n;
4280: n = A->cmap->n;
4281: p = B->cmap->n;
4282: a = sub_a->v;
4283: b = sub_b->a;
4284: c = sub_c->v;
4285: if (clda == m) {
4286: PetscCall(PetscArrayzero(c, m * p));
4287: } else {
4288: for (j = 0; j < p; j++)
4289: for (i = 0; i < m; i++) c[j * clda + i] = 0.0;
4290: }
4291: ii = sub_b->i;
4292: idx = sub_b->j;
4293: for (i = 0; i < n; i++) {
4294: q = ii[i + 1] - ii[i];
4295: while (q-- > 0) {
4296: c_q = c + clda * (*idx);
4297: a_q = a + alda * i;
4298: PetscKernelAXPY(c_q, *b, a_q, m);
4299: idx++;
4300: b++;
4301: }
4302: }
4303: PetscFunctionReturn(PETSC_SUCCESS);
4304: }
4306: PetscErrorCode MatMatMultSymbolic_SeqDense_SeqAIJ(Mat A, Mat B, PetscReal fill, Mat C)
4307: {
4308: PetscInt m = A->rmap->n, n = B->cmap->n;
4309: PetscBool cisdense;
4311: PetscFunctionBegin;
4312: PetscCheck(A->cmap->n == B->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "A->cmap->n %" PetscInt_FMT " != B->rmap->n %" PetscInt_FMT, A->cmap->n, B->rmap->n);
4313: PetscCall(MatSetSizes(C, m, n, m, n));
4314: PetscCall(MatSetBlockSizesFromMats(C, A, B));
4315: PetscCall(PetscObjectTypeCompareAny((PetscObject)C, &cisdense, MATSEQDENSE, MATSEQDENSECUDA, MATSEQDENSEHIP, ""));
4316: if (!cisdense) PetscCall(MatSetType(C, MATDENSE));
4317: PetscCall(MatSetUp(C));
4319: C->ops->matmultnumeric = MatMatMultNumeric_SeqDense_SeqAIJ;
4320: PetscFunctionReturn(PETSC_SUCCESS);
4321: }
4323: /*MC
4324: MATSEQAIJ - MATSEQAIJ = "seqaij" - A matrix type to be used for sequential sparse matrices,
4325: based on compressed sparse row format.
4327: Options Database Key:
4328: . -mat_type seqaij - sets the matrix type to "seqaij" during a call to MatSetFromOptions()
4330: Level: beginner
4332: Notes:
4333: `MatSetValues()` may be called for this matrix type with a `NULL` argument for the numerical values,
4334: in this case the values associated with the rows and columns one passes in are set to zero
4335: in the matrix
4337: `MatSetOptions`(,`MAT_STRUCTURE_ONLY`,`PETSC_TRUE`) may be called for this matrix type. In this no
4338: space is allocated for the nonzero entries and any entries passed with `MatSetValues()` are ignored
4340: Developer Note:
4341: It would be nice if all matrix formats supported passing `NULL` in for the numerical values
4343: .seealso: [](ch_matrices), `Mat`, `MatCreateSeqAIJ()`, `MatSetFromOptions()`, `MatSetType()`, `MatCreate()`, `MatType`, `MATSELL`, `MATSEQSELL`, `MATMPISELL`
4344: M*/
4346: /*MC
4347: MATAIJ - MATAIJ = "aij" - A matrix type to be used for sparse matrices.
4349: This matrix type is identical to `MATSEQAIJ` when constructed with a single process communicator,
4350: and `MATMPIAIJ` otherwise. As a result, for single process communicators,
4351: `MatSeqAIJSetPreallocation()` is supported, and similarly `MatMPIAIJSetPreallocation()` is supported
4352: for communicators controlling multiple processes. It is recommended that you call both of
4353: the above preallocation routines for simplicity.
4355: Options Database Key:
4356: . -mat_type aij - sets the matrix type to "aij" during a call to `MatSetFromOptions()`
4358: Level: beginner
4360: Note:
4361: Subclasses include `MATAIJCUSPARSE`, `MATAIJPERM`, `MATAIJSELL`, `MATAIJMKL`, `MATAIJCRL`, and also automatically switches over to use inodes when
4362: enough exist.
4364: .seealso: [](ch_matrices), `Mat`, `MatCreateAIJ()`, `MatCreateSeqAIJ()`, `MATSEQAIJ`, `MATMPIAIJ`, `MATSELL`, `MATSEQSELL`, `MATMPISELL`
4365: M*/
4367: /*MC
4368: MATAIJCRL - MATAIJCRL = "aijcrl" - A matrix type to be used for sparse matrices.
4370: Options Database Key:
4371: . -mat_type aijcrl - sets the matrix type to "aijcrl" during a call to `MatSetFromOptions()`
4373: Level: beginner
4375: Note:
4376: This matrix type is identical to `MATSEQAIJCRL` when constructed with a single process communicator,
4377: and `MATMPIAIJCRL` otherwise. As a result, for single process communicators,
4378: `MatSeqAIJSetPreallocation()` is supported, and similarly `MatMPIAIJSetPreallocation()` is supported
4379: for communicators controlling multiple processes. It is recommended that you call both of
4380: the above preallocation routines for simplicity.
4382: .seealso: [](ch_matrices), `Mat`, `MatCreateMPIAIJCRL`, `MATSEQAIJCRL`, `MATMPIAIJCRL`, `MATSEQAIJCRL`, `MATMPIAIJCRL`
4383: M*/
4385: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJCRL(Mat, MatType, MatReuse, Mat *);
4386: #if defined(PETSC_HAVE_ELEMENTAL)
4387: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_Elemental(Mat, MatType, MatReuse, Mat *);
4388: #endif
4389: #if defined(PETSC_HAVE_SCALAPACK)
4390: PETSC_INTERN PetscErrorCode MatConvert_AIJ_ScaLAPACK(Mat, MatType, MatReuse, Mat *);
4391: #endif
4392: #if defined(PETSC_HAVE_HYPRE)
4393: PETSC_INTERN PetscErrorCode MatConvert_AIJ_HYPRE(Mat A, MatType, MatReuse, Mat *);
4394: #endif
4396: PETSC_EXTERN PetscErrorCode MatConvert_SeqAIJ_SeqSELL(Mat, MatType, MatReuse, Mat *);
4397: PETSC_INTERN PetscErrorCode MatConvert_XAIJ_IS(Mat, MatType, MatReuse, Mat *);
4398: PETSC_INTERN PetscErrorCode MatProductSetFromOptions_IS_XAIJ(Mat);
4400: /*@C
4401: MatSeqAIJGetArray - gives read/write access to the array where the data for a `MATSEQAIJ` matrix is stored
4403: Not Collective
4405: Input Parameter:
4406: . A - a `MATSEQAIJ` matrix
4408: Output Parameter:
4409: . array - pointer to the data
4411: Level: intermediate
4413: Fortran Notes:
4414: `MatSeqAIJGetArray()` Fortran binding is deprecated (since PETSc 3.19), use `MatSeqAIJGetArrayF90()`
4416: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJRestoreArray()`, `MatSeqAIJGetArrayF90()`
4417: @*/
4418: PetscErrorCode MatSeqAIJGetArray(Mat A, PetscScalar *array[])
4419: {
4420: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4422: PetscFunctionBegin;
4423: if (aij->ops->getarray) {
4424: PetscCall((*aij->ops->getarray)(A, array));
4425: } else {
4426: *array = aij->a;
4427: }
4428: PetscFunctionReturn(PETSC_SUCCESS);
4429: }
4431: /*@C
4432: MatSeqAIJRestoreArray - returns access to the array where the data for a `MATSEQAIJ` matrix is stored obtained by `MatSeqAIJGetArray()`
4434: Not Collective
4436: Input Parameters:
4437: + A - a `MATSEQAIJ` matrix
4438: - array - pointer to the data
4440: Level: intermediate
4442: Fortran Notes:
4443: `MatSeqAIJRestoreArray()` Fortran binding is deprecated (since PETSc 3.19), use `MatSeqAIJRestoreArrayF90()`
4445: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArrayF90()`
4446: @*/
4447: PetscErrorCode MatSeqAIJRestoreArray(Mat A, PetscScalar *array[])
4448: {
4449: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4451: PetscFunctionBegin;
4452: if (aij->ops->restorearray) {
4453: PetscCall((*aij->ops->restorearray)(A, array));
4454: } else {
4455: *array = NULL;
4456: }
4457: PetscCall(MatSeqAIJInvalidateDiagonal(A));
4458: PetscCall(PetscObjectStateIncrease((PetscObject)A));
4459: PetscFunctionReturn(PETSC_SUCCESS);
4460: }
4462: /*@C
4463: MatSeqAIJGetArrayRead - gives read-only access to the array where the data for a `MATSEQAIJ` matrix is stored
4465: Not Collective; No Fortran Support
4467: Input Parameter:
4468: . A - a `MATSEQAIJ` matrix
4470: Output Parameter:
4471: . array - pointer to the data
4473: Level: intermediate
4475: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArrayRead()`
4476: @*/
4477: PetscErrorCode MatSeqAIJGetArrayRead(Mat A, const PetscScalar *array[])
4478: {
4479: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4481: PetscFunctionBegin;
4482: if (aij->ops->getarrayread) {
4483: PetscCall((*aij->ops->getarrayread)(A, array));
4484: } else {
4485: *array = aij->a;
4486: }
4487: PetscFunctionReturn(PETSC_SUCCESS);
4488: }
4490: /*@C
4491: MatSeqAIJRestoreArrayRead - restore the read-only access array obtained from `MatSeqAIJGetArrayRead()`
4493: Not Collective; No Fortran Support
4495: Input Parameter:
4496: . A - a `MATSEQAIJ` matrix
4498: Output Parameter:
4499: . array - pointer to the data
4501: Level: intermediate
4503: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJGetArrayRead()`
4504: @*/
4505: PetscErrorCode MatSeqAIJRestoreArrayRead(Mat A, const PetscScalar *array[])
4506: {
4507: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4509: PetscFunctionBegin;
4510: if (aij->ops->restorearrayread) {
4511: PetscCall((*aij->ops->restorearrayread)(A, array));
4512: } else {
4513: *array = NULL;
4514: }
4515: PetscFunctionReturn(PETSC_SUCCESS);
4516: }
4518: /*@C
4519: MatSeqAIJGetArrayWrite - gives write-only access to the array where the data for a `MATSEQAIJ` matrix is stored
4521: Not Collective; No Fortran Support
4523: Input Parameter:
4524: . A - a `MATSEQAIJ` matrix
4526: Output Parameter:
4527: . array - pointer to the data
4529: Level: intermediate
4531: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJRestoreArrayRead()`
4532: @*/
4533: PetscErrorCode MatSeqAIJGetArrayWrite(Mat A, PetscScalar *array[])
4534: {
4535: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4537: PetscFunctionBegin;
4538: if (aij->ops->getarraywrite) {
4539: PetscCall((*aij->ops->getarraywrite)(A, array));
4540: } else {
4541: *array = aij->a;
4542: }
4543: PetscCall(MatSeqAIJInvalidateDiagonal(A));
4544: PetscCall(PetscObjectStateIncrease((PetscObject)A));
4545: PetscFunctionReturn(PETSC_SUCCESS);
4546: }
4548: /*@C
4549: MatSeqAIJRestoreArrayWrite - restore the read-only access array obtained from MatSeqAIJGetArrayRead
4551: Not Collective; No Fortran Support
4553: Input Parameter:
4554: . A - a MATSEQAIJ matrix
4556: Output Parameter:
4557: . array - pointer to the data
4559: Level: intermediate
4561: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJGetArrayRead()`
4562: @*/
4563: PetscErrorCode MatSeqAIJRestoreArrayWrite(Mat A, PetscScalar *array[])
4564: {
4565: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4567: PetscFunctionBegin;
4568: if (aij->ops->restorearraywrite) {
4569: PetscCall((*aij->ops->restorearraywrite)(A, array));
4570: } else {
4571: *array = NULL;
4572: }
4573: PetscFunctionReturn(PETSC_SUCCESS);
4574: }
4576: /*@C
4577: MatSeqAIJGetCSRAndMemType - Get the CSR arrays and the memory type of the `MATSEQAIJ` matrix
4579: Not Collective; No Fortran Support
4581: Input Parameter:
4582: . mat - a matrix of type `MATSEQAIJ` or its subclasses
4584: Output Parameters:
4585: + i - row map array of the matrix
4586: . j - column index array of the matrix
4587: . a - data array of the matrix
4588: - mtype - memory type of the arrays
4590: Level: developer
4592: Notes:
4593: Any of the output parameters can be `NULL`, in which case the corresponding value is not returned.
4594: If mat is a device matrix, the arrays are on the device. Otherwise, they are on the host.
4596: One can call this routine on a preallocated but not assembled matrix to just get the memory of the CSR underneath the matrix.
4597: If the matrix is assembled, the data array `a` is guaranteed to have the latest values of the matrix.
4599: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJGetArray()`, `MatSeqAIJGetArrayRead()`
4600: @*/
4601: PetscErrorCode MatSeqAIJGetCSRAndMemType(Mat mat, const PetscInt *i[], const PetscInt *j[], PetscScalar *a[], PetscMemType *mtype)
4602: {
4603: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
4605: PetscFunctionBegin;
4606: PetscCheck(mat->preallocated, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "matrix is not preallocated");
4607: if (aij->ops->getcsrandmemtype) {
4608: PetscCall((*aij->ops->getcsrandmemtype)(mat, i, j, a, mtype));
4609: } else {
4610: if (i) *i = aij->i;
4611: if (j) *j = aij->j;
4612: if (a) *a = aij->a;
4613: if (mtype) *mtype = PETSC_MEMTYPE_HOST;
4614: }
4615: PetscFunctionReturn(PETSC_SUCCESS);
4616: }
4618: /*@
4619: MatSeqAIJGetMaxRowNonzeros - returns the maximum number of nonzeros in any row
4621: Not Collective
4623: Input Parameter:
4624: . A - a `MATSEQAIJ` matrix
4626: Output Parameter:
4627: . nz - the maximum number of nonzeros in any row
4629: Level: intermediate
4631: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJRestoreArray()`, `MatSeqAIJGetArrayF90()`
4632: @*/
4633: PetscErrorCode MatSeqAIJGetMaxRowNonzeros(Mat A, PetscInt *nz)
4634: {
4635: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)A->data;
4637: PetscFunctionBegin;
4638: *nz = aij->rmax;
4639: PetscFunctionReturn(PETSC_SUCCESS);
4640: }
4642: static PetscErrorCode MatCOOStructDestroy_SeqAIJ(void **data)
4643: {
4644: MatCOOStruct_SeqAIJ *coo = (MatCOOStruct_SeqAIJ *)*data;
4646: PetscFunctionBegin;
4647: PetscCall(PetscFree(coo->perm));
4648: PetscCall(PetscFree(coo->jmap));
4649: PetscCall(PetscFree(coo));
4650: PetscFunctionReturn(PETSC_SUCCESS);
4651: }
4653: PetscErrorCode MatSetPreallocationCOO_SeqAIJ(Mat mat, PetscCount coo_n, PetscInt coo_i[], PetscInt coo_j[])
4654: {
4655: MPI_Comm comm;
4656: PetscInt *i, *j;
4657: PetscInt M, N, row, iprev;
4658: PetscCount k, p, q, nneg, nnz, start, end; /* Index the coo array, so use PetscCount as their type */
4659: PetscInt *Ai; /* Change to PetscCount once we use it for row pointers */
4660: PetscInt *Aj;
4661: PetscScalar *Aa;
4662: Mat_SeqAIJ *seqaij = (Mat_SeqAIJ *)mat->data;
4663: MatType rtype;
4664: PetscCount *perm, *jmap;
4665: MatCOOStruct_SeqAIJ *coo;
4666: PetscBool isorted;
4667: PetscBool hypre;
4668: const char *name;
4670: PetscFunctionBegin;
4671: PetscCall(PetscObjectGetComm((PetscObject)mat, &comm));
4672: PetscCall(MatGetSize(mat, &M, &N));
4673: i = coo_i;
4674: j = coo_j;
4675: PetscCall(PetscMalloc1(coo_n, &perm));
4677: /* Ignore entries with negative row or col indices; at the same time, check if i[] is already sorted (e.g., MatConvert_AlJ_HYPRE results in this case) */
4678: isorted = PETSC_TRUE;
4679: iprev = PETSC_INT_MIN;
4680: for (k = 0; k < coo_n; k++) {
4681: if (j[k] < 0) i[k] = -1;
4682: if (isorted) {
4683: if (i[k] < iprev) isorted = PETSC_FALSE;
4684: else iprev = i[k];
4685: }
4686: perm[k] = k;
4687: }
4689: /* Sort by row if not already */
4690: if (!isorted) PetscCall(PetscSortIntWithIntCountArrayPair(coo_n, i, j, perm));
4692: /* Advance k to the first row with a non-negative index */
4693: for (k = 0; k < coo_n; k++)
4694: if (i[k] >= 0) break;
4695: nneg = k;
4696: PetscCall(PetscMalloc1(coo_n - nneg + 1, &jmap)); /* +1 to make a CSR-like data structure. jmap[i] originally is the number of repeats for i-th nonzero */
4697: nnz = 0; /* Total number of unique nonzeros to be counted */
4698: jmap++; /* Inc jmap by 1 for convenience */
4700: PetscCall(PetscShmgetAllocateArray(M + 1, sizeof(PetscInt), (void **)&Ai)); /* CSR of A */
4701: PetscCall(PetscArrayzero(Ai, M + 1));
4702: PetscCall(PetscShmgetAllocateArray(coo_n - nneg, sizeof(PetscInt), (void **)&Aj)); /* We have at most coo_n-nneg unique nonzeros */
4704: PetscCall(PetscObjectGetName((PetscObject)mat, &name));
4705: PetscCall(PetscStrcmp("_internal_COO_mat_for_hypre", name, &hypre));
4707: /* In each row, sort by column, then unique column indices to get row length */
4708: Ai++; /* Inc by 1 for convenience */
4709: q = 0; /* q-th unique nonzero, with q starting from 0 */
4710: while (k < coo_n) {
4711: PetscBool strictly_sorted; // this row is strictly sorted?
4712: PetscInt jprev;
4714: /* get [start,end) indices for this row; also check if cols in this row are strictly sorted */
4715: row = i[k];
4716: start = k;
4717: jprev = PETSC_INT_MIN;
4718: strictly_sorted = PETSC_TRUE;
4719: while (k < coo_n && i[k] == row) {
4720: if (strictly_sorted) {
4721: if (j[k] <= jprev) strictly_sorted = PETSC_FALSE;
4722: else jprev = j[k];
4723: }
4724: k++;
4725: }
4726: end = k;
4728: /* hack for HYPRE: swap min column to diag so that diagonal values will go first */
4729: if (hypre) {
4730: PetscInt minj = PETSC_INT_MAX;
4731: PetscBool hasdiag = PETSC_FALSE;
4733: if (strictly_sorted) { // fast path to swap the first and the diag
4734: PetscCount tmp;
4735: for (p = start; p < end; p++) {
4736: if (j[p] == row && p != start) {
4737: j[p] = j[start];
4738: j[start] = row;
4739: tmp = perm[start];
4740: perm[start] = perm[p];
4741: perm[p] = tmp;
4742: break;
4743: }
4744: }
4745: } else {
4746: for (p = start; p < end; p++) {
4747: hasdiag = (PetscBool)(hasdiag || (j[p] == row));
4748: minj = PetscMin(minj, j[p]);
4749: }
4751: if (hasdiag) {
4752: for (p = start; p < end; p++) {
4753: if (j[p] == minj) j[p] = row;
4754: else if (j[p] == row) j[p] = minj;
4755: }
4756: }
4757: }
4758: }
4759: // sort by columns in a row
4760: if (!strictly_sorted) PetscCall(PetscSortIntWithCountArray(end - start, j + start, perm + start));
4762: if (strictly_sorted) { // fast path to set Aj[], jmap[], Ai[], nnz, q
4763: for (p = start; p < end; p++, q++) {
4764: Aj[q] = j[p];
4765: jmap[q] = 1;
4766: }
4767: PetscCall(PetscIntCast(end - start, Ai + row));
4768: nnz += Ai[row]; // q is already advanced
4769: } else {
4770: /* Find number of unique col entries in this row */
4771: Aj[q] = j[start]; /* Log the first nonzero in this row */
4772: jmap[q] = 1; /* Number of repeats of this nonzero entry */
4773: Ai[row] = 1;
4774: nnz++;
4776: for (p = start + 1; p < end; p++) { /* Scan remaining nonzero in this row */
4777: if (j[p] != j[p - 1]) { /* Meet a new nonzero */
4778: q++;
4779: jmap[q] = 1;
4780: Aj[q] = j[p];
4781: Ai[row]++;
4782: nnz++;
4783: } else {
4784: jmap[q]++;
4785: }
4786: }
4787: q++; /* Move to next row and thus next unique nonzero */
4788: }
4789: }
4791: Ai--; /* Back to the beginning of Ai[] */
4792: for (k = 0; k < M; k++) Ai[k + 1] += Ai[k];
4793: jmap--; // Back to the beginning of jmap[]
4794: jmap[0] = 0;
4795: for (k = 0; k < nnz; k++) jmap[k + 1] += jmap[k];
4797: if (nnz < coo_n - nneg) { /* Reallocate with actual number of unique nonzeros */
4798: PetscCount *jmap_new;
4799: PetscInt *Aj_new;
4801: PetscCall(PetscMalloc1(nnz + 1, &jmap_new));
4802: PetscCall(PetscArraycpy(jmap_new, jmap, nnz + 1));
4803: PetscCall(PetscFree(jmap));
4804: jmap = jmap_new;
4806: PetscCall(PetscShmgetAllocateArray(nnz, sizeof(PetscInt), (void **)&Aj_new));
4807: PetscCall(PetscArraycpy(Aj_new, Aj, nnz));
4808: PetscCall(PetscShmgetDeallocateArray((void **)&Aj));
4809: Aj = Aj_new;
4810: }
4812: if (nneg) { /* Discard heading entries with negative indices in perm[], as we'll access it from index 0 in MatSetValuesCOO */
4813: PetscCount *perm_new;
4815: PetscCall(PetscMalloc1(coo_n - nneg, &perm_new));
4816: PetscCall(PetscArraycpy(perm_new, perm + nneg, coo_n - nneg));
4817: PetscCall(PetscFree(perm));
4818: perm = perm_new;
4819: }
4821: PetscCall(MatGetRootType_Private(mat, &rtype));
4822: PetscCall(PetscShmgetAllocateArray(nnz, sizeof(PetscScalar), (void **)&Aa));
4823: PetscCall(PetscArrayzero(Aa, nnz));
4824: PetscCall(MatSetSeqAIJWithArrays_private(PETSC_COMM_SELF, M, N, Ai, Aj, Aa, rtype, mat));
4826: seqaij->free_a = seqaij->free_ij = PETSC_TRUE; /* Let newmat own Ai, Aj and Aa */
4828: // Put the COO struct in a container and then attach that to the matrix
4829: PetscCall(PetscMalloc1(1, &coo));
4830: PetscCall(PetscIntCast(nnz, &coo->nz));
4831: coo->n = coo_n;
4832: coo->Atot = coo_n - nneg; // Annz is seqaij->nz, so no need to record that again
4833: coo->jmap = jmap; // of length nnz+1
4834: coo->perm = perm;
4835: PetscCall(PetscObjectContainerCompose((PetscObject)mat, "__PETSc_MatCOOStruct_Host", coo, MatCOOStructDestroy_SeqAIJ));
4836: PetscFunctionReturn(PETSC_SUCCESS);
4837: }
4839: static PetscErrorCode MatSetValuesCOO_SeqAIJ(Mat A, const PetscScalar v[], InsertMode imode)
4840: {
4841: Mat_SeqAIJ *aseq = (Mat_SeqAIJ *)A->data;
4842: PetscCount i, j, Annz = aseq->nz;
4843: PetscCount *perm, *jmap;
4844: PetscScalar *Aa;
4845: PetscContainer container;
4846: MatCOOStruct_SeqAIJ *coo;
4848: PetscFunctionBegin;
4849: PetscCall(PetscObjectQuery((PetscObject)A, "__PETSc_MatCOOStruct_Host", (PetscObject *)&container));
4850: PetscCheck(container, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Not found MatCOOStruct on this matrix");
4851: PetscCall(PetscContainerGetPointer(container, (void **)&coo));
4852: perm = coo->perm;
4853: jmap = coo->jmap;
4854: PetscCall(MatSeqAIJGetArray(A, &Aa));
4855: for (i = 0; i < Annz; i++) {
4856: PetscScalar sum = 0.0;
4857: for (j = jmap[i]; j < jmap[i + 1]; j++) sum += v[perm[j]];
4858: Aa[i] = (imode == INSERT_VALUES ? 0.0 : Aa[i]) + sum;
4859: }
4860: PetscCall(MatSeqAIJRestoreArray(A, &Aa));
4861: PetscFunctionReturn(PETSC_SUCCESS);
4862: }
4864: #if defined(PETSC_HAVE_CUDA)
4865: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJCUSPARSE(Mat, MatType, MatReuse, Mat *);
4866: #endif
4867: #if defined(PETSC_HAVE_HIP)
4868: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJHIPSPARSE(Mat, MatType, MatReuse, Mat *);
4869: #endif
4870: #if defined(PETSC_HAVE_KOKKOS_KERNELS)
4871: PETSC_INTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJKokkos(Mat, MatType, MatReuse, Mat *);
4872: #endif
4874: PETSC_EXTERN PetscErrorCode MatCreate_SeqAIJ(Mat B)
4875: {
4876: Mat_SeqAIJ *b;
4877: PetscMPIInt size;
4879: PetscFunctionBegin;
4880: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &size));
4881: PetscCheck(size <= 1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Comm must be of size 1");
4883: PetscCall(PetscNew(&b));
4885: B->data = (void *)b;
4886: B->ops[0] = MatOps_Values;
4887: if (B->sortedfull) B->ops->setvalues = MatSetValues_SeqAIJ_SortedFull;
4889: b->row = NULL;
4890: b->col = NULL;
4891: b->icol = NULL;
4892: b->reallocs = 0;
4893: b->ignorezeroentries = PETSC_FALSE;
4894: b->roworiented = PETSC_TRUE;
4895: b->nonew = 0;
4896: b->diag = NULL;
4897: b->solve_work = NULL;
4898: B->spptr = NULL;
4899: b->saved_values = NULL;
4900: b->idiag = NULL;
4901: b->mdiag = NULL;
4902: b->ssor_work = NULL;
4903: b->omega = 1.0;
4904: b->fshift = 0.0;
4905: b->idiagvalid = PETSC_FALSE;
4906: b->ibdiagvalid = PETSC_FALSE;
4907: b->keepnonzeropattern = PETSC_FALSE;
4909: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATSEQAIJ));
4910: #if defined(PETSC_HAVE_MATLAB)
4911: PetscCall(PetscObjectComposeFunction((PetscObject)B, "PetscMatlabEnginePut_C", MatlabEnginePut_SeqAIJ));
4912: PetscCall(PetscObjectComposeFunction((PetscObject)B, "PetscMatlabEngineGet_C", MatlabEngineGet_SeqAIJ));
4913: #endif
4914: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqAIJSetColumnIndices_C", MatSeqAIJSetColumnIndices_SeqAIJ));
4915: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatStoreValues_C", MatStoreValues_SeqAIJ));
4916: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatRetrieveValues_C", MatRetrieveValues_SeqAIJ));
4917: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqsbaij_C", MatConvert_SeqAIJ_SeqSBAIJ));
4918: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqbaij_C", MatConvert_SeqAIJ_SeqBAIJ));
4919: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijperm_C", MatConvert_SeqAIJ_SeqAIJPERM));
4920: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijsell_C", MatConvert_SeqAIJ_SeqAIJSELL));
4921: #if defined(PETSC_HAVE_MKL_SPARSE)
4922: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijmkl_C", MatConvert_SeqAIJ_SeqAIJMKL));
4923: #endif
4924: #if defined(PETSC_HAVE_CUDA)
4925: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijcusparse_C", MatConvert_SeqAIJ_SeqAIJCUSPARSE));
4926: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqaijcusparse_seqaij_C", MatProductSetFromOptions_SeqAIJ));
4927: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqaij_seqaijcusparse_C", MatProductSetFromOptions_SeqAIJ));
4928: #endif
4929: #if defined(PETSC_HAVE_HIP)
4930: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijhipsparse_C", MatConvert_SeqAIJ_SeqAIJHIPSPARSE));
4931: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqaijhipsparse_seqaij_C", MatProductSetFromOptions_SeqAIJ));
4932: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqaij_seqaijhipsparse_C", MatProductSetFromOptions_SeqAIJ));
4933: #endif
4934: #if defined(PETSC_HAVE_KOKKOS_KERNELS)
4935: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijkokkos_C", MatConvert_SeqAIJ_SeqAIJKokkos));
4936: #endif
4937: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqaijcrl_C", MatConvert_SeqAIJ_SeqAIJCRL));
4938: #if defined(PETSC_HAVE_ELEMENTAL)
4939: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_elemental_C", MatConvert_SeqAIJ_Elemental));
4940: #endif
4941: #if defined(PETSC_HAVE_SCALAPACK)
4942: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_scalapack_C", MatConvert_AIJ_ScaLAPACK));
4943: #endif
4944: #if defined(PETSC_HAVE_HYPRE)
4945: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_hypre_C", MatConvert_AIJ_HYPRE));
4946: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_transpose_seqaij_seqaij_C", MatProductSetFromOptions_Transpose_AIJ_AIJ));
4947: #endif
4948: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqdense_C", MatConvert_SeqAIJ_SeqDense));
4949: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_seqsell_C", MatConvert_SeqAIJ_SeqSELL));
4950: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_seqaij_is_C", MatConvert_XAIJ_IS));
4951: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatIsTranspose_C", MatIsTranspose_SeqAIJ));
4952: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatIsHermitianTranspose_C", MatIsHermitianTranspose_SeqAIJ));
4953: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqAIJSetPreallocation_C", MatSeqAIJSetPreallocation_SeqAIJ));
4954: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatResetPreallocation_C", MatResetPreallocation_SeqAIJ));
4955: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatResetHash_C", MatResetHash_SeqAIJ));
4956: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqAIJSetPreallocationCSR_C", MatSeqAIJSetPreallocationCSR_SeqAIJ));
4957: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatReorderForNonzeroDiagonal_C", MatReorderForNonzeroDiagonal_SeqAIJ));
4958: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_is_seqaij_C", MatProductSetFromOptions_IS_XAIJ));
4959: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqdense_seqaij_C", MatProductSetFromOptions_SeqDense_SeqAIJ));
4960: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatProductSetFromOptions_seqaij_seqaij_C", MatProductSetFromOptions_SeqAIJ));
4961: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSeqAIJKron_C", MatSeqAIJKron_SeqAIJ));
4962: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSetPreallocationCOO_C", MatSetPreallocationCOO_SeqAIJ));
4963: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatSetValuesCOO_C", MatSetValuesCOO_SeqAIJ));
4964: PetscCall(MatCreate_SeqAIJ_Inode(B));
4965: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATSEQAIJ));
4966: PetscCall(MatSeqAIJSetTypeFromOptions(B)); /* this allows changing the matrix subtype to say MATSEQAIJPERM */
4967: PetscFunctionReturn(PETSC_SUCCESS);
4968: }
4970: /*
4971: Given a matrix generated with MatGetFactor() duplicates all the information in A into C
4972: */
4973: PetscErrorCode MatDuplicateNoCreate_SeqAIJ(Mat C, Mat A, MatDuplicateOption cpvalues, PetscBool mallocmatspace)
4974: {
4975: Mat_SeqAIJ *c = (Mat_SeqAIJ *)C->data, *a = (Mat_SeqAIJ *)A->data;
4976: PetscInt m = A->rmap->n, i;
4978: PetscFunctionBegin;
4979: PetscCheck(A->assembled || cpvalues == MAT_DO_NOT_COPY_VALUES, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot duplicate unassembled matrix");
4981: C->factortype = A->factortype;
4982: c->row = NULL;
4983: c->col = NULL;
4984: c->icol = NULL;
4985: c->reallocs = 0;
4986: c->diagonaldense = a->diagonaldense;
4988: C->assembled = A->assembled;
4990: if (A->preallocated) {
4991: PetscCall(PetscLayoutReference(A->rmap, &C->rmap));
4992: PetscCall(PetscLayoutReference(A->cmap, &C->cmap));
4994: if (!A->hash_active) {
4995: PetscCall(PetscMalloc1(m, &c->imax));
4996: PetscCall(PetscMemcpy(c->imax, a->imax, m * sizeof(PetscInt)));
4997: PetscCall(PetscMalloc1(m, &c->ilen));
4998: PetscCall(PetscMemcpy(c->ilen, a->ilen, m * sizeof(PetscInt)));
5000: /* allocate the matrix space */
5001: if (mallocmatspace) {
5002: PetscCall(PetscShmgetAllocateArray(a->i[m], sizeof(PetscScalar), (void **)&c->a));
5003: PetscCall(PetscShmgetAllocateArray(a->i[m], sizeof(PetscInt), (void **)&c->j));
5004: PetscCall(PetscShmgetAllocateArray(m + 1, sizeof(PetscInt), (void **)&c->i));
5005: PetscCall(PetscArraycpy(c->i, a->i, m + 1));
5006: c->free_a = PETSC_TRUE;
5007: c->free_ij = PETSC_TRUE;
5008: if (m > 0) {
5009: PetscCall(PetscArraycpy(c->j, a->j, a->i[m]));
5010: if (cpvalues == MAT_COPY_VALUES) {
5011: const PetscScalar *aa;
5013: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
5014: PetscCall(PetscArraycpy(c->a, aa, a->i[m]));
5015: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
5016: } else {
5017: PetscCall(PetscArrayzero(c->a, a->i[m]));
5018: }
5019: }
5020: }
5021: C->preallocated = PETSC_TRUE;
5022: } else {
5023: PetscCheck(mallocmatspace, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_WRONGSTATE, "Cannot malloc matrix memory from a non-preallocated matrix");
5024: PetscCall(MatSetUp(C));
5025: }
5027: c->ignorezeroentries = a->ignorezeroentries;
5028: c->roworiented = a->roworiented;
5029: c->nonew = a->nonew;
5030: if (a->diag) {
5031: PetscCall(PetscMalloc1(m + 1, &c->diag));
5032: PetscCall(PetscMemcpy(c->diag, a->diag, m * sizeof(PetscInt)));
5033: } else c->diag = NULL;
5035: c->solve_work = NULL;
5036: c->saved_values = NULL;
5037: c->idiag = NULL;
5038: c->ssor_work = NULL;
5039: c->keepnonzeropattern = a->keepnonzeropattern;
5041: c->rmax = a->rmax;
5042: c->nz = a->nz;
5043: c->maxnz = a->nz; /* Since we allocate exactly the right amount */
5045: c->compressedrow.use = a->compressedrow.use;
5046: c->compressedrow.nrows = a->compressedrow.nrows;
5047: if (a->compressedrow.use) {
5048: i = a->compressedrow.nrows;
5049: PetscCall(PetscMalloc2(i + 1, &c->compressedrow.i, i, &c->compressedrow.rindex));
5050: PetscCall(PetscArraycpy(c->compressedrow.i, a->compressedrow.i, i + 1));
5051: PetscCall(PetscArraycpy(c->compressedrow.rindex, a->compressedrow.rindex, i));
5052: } else {
5053: c->compressedrow.use = PETSC_FALSE;
5054: c->compressedrow.i = NULL;
5055: c->compressedrow.rindex = NULL;
5056: }
5057: c->nonzerorowcnt = a->nonzerorowcnt;
5058: C->nonzerostate = A->nonzerostate;
5060: PetscCall(MatDuplicate_SeqAIJ_Inode(A, cpvalues, &C));
5061: }
5062: PetscCall(PetscFunctionListDuplicate(((PetscObject)A)->qlist, &((PetscObject)C)->qlist));
5063: PetscFunctionReturn(PETSC_SUCCESS);
5064: }
5066: PetscErrorCode MatDuplicate_SeqAIJ(Mat A, MatDuplicateOption cpvalues, Mat *B)
5067: {
5068: PetscFunctionBegin;
5069: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), B));
5070: PetscCall(MatSetSizes(*B, A->rmap->n, A->cmap->n, A->rmap->n, A->cmap->n));
5071: if (!(A->rmap->n % A->rmap->bs) && !(A->cmap->n % A->cmap->bs)) PetscCall(MatSetBlockSizesFromMats(*B, A, A));
5072: PetscCall(MatSetType(*B, ((PetscObject)A)->type_name));
5073: PetscCall(MatDuplicateNoCreate_SeqAIJ(*B, A, cpvalues, PETSC_TRUE));
5074: PetscFunctionReturn(PETSC_SUCCESS);
5075: }
5077: PetscErrorCode MatLoad_SeqAIJ(Mat newMat, PetscViewer viewer)
5078: {
5079: PetscBool isbinary, ishdf5;
5081: PetscFunctionBegin;
5084: /* force binary viewer to load .info file if it has not yet done so */
5085: PetscCall(PetscViewerSetUp(viewer));
5086: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
5087: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERHDF5, &ishdf5));
5088: if (isbinary) {
5089: PetscCall(MatLoad_SeqAIJ_Binary(newMat, viewer));
5090: } else if (ishdf5) {
5091: #if defined(PETSC_HAVE_HDF5)
5092: PetscCall(MatLoad_AIJ_HDF5(newMat, viewer));
5093: #else
5094: SETERRQ(PetscObjectComm((PetscObject)newMat), PETSC_ERR_SUP, "HDF5 not supported in this build.\nPlease reconfigure using --download-hdf5");
5095: #endif
5096: } else {
5097: SETERRQ(PetscObjectComm((PetscObject)newMat), PETSC_ERR_SUP, "Viewer type %s not yet supported for reading %s matrices", ((PetscObject)viewer)->type_name, ((PetscObject)newMat)->type_name);
5098: }
5099: PetscFunctionReturn(PETSC_SUCCESS);
5100: }
5102: PetscErrorCode MatLoad_SeqAIJ_Binary(Mat mat, PetscViewer viewer)
5103: {
5104: Mat_SeqAIJ *a = (Mat_SeqAIJ *)mat->data;
5105: PetscInt header[4], *rowlens, M, N, nz, sum, rows, cols, i;
5107: PetscFunctionBegin;
5108: PetscCall(PetscViewerSetUp(viewer));
5110: /* read in matrix header */
5111: PetscCall(PetscViewerBinaryRead(viewer, header, 4, NULL, PETSC_INT));
5112: PetscCheck(header[0] == MAT_FILE_CLASSID, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Not a matrix object in file");
5113: M = header[1];
5114: N = header[2];
5115: nz = header[3];
5116: PetscCheck(M >= 0, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Matrix row size (%" PetscInt_FMT ") in file is negative", M);
5117: PetscCheck(N >= 0, PetscObjectComm((PetscObject)viewer), PETSC_ERR_FILE_UNEXPECTED, "Matrix column size (%" PetscInt_FMT ") in file is negative", N);
5118: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Matrix stored in special format on disk, cannot load as SeqAIJ");
5120: /* set block sizes from the viewer's .info file */
5121: PetscCall(MatLoad_Binary_BlockSizes(mat, viewer));
5122: /* set local and global sizes if not set already */
5123: if (mat->rmap->n < 0) mat->rmap->n = M;
5124: if (mat->cmap->n < 0) mat->cmap->n = N;
5125: if (mat->rmap->N < 0) mat->rmap->N = M;
5126: if (mat->cmap->N < 0) mat->cmap->N = N;
5127: PetscCall(PetscLayoutSetUp(mat->rmap));
5128: PetscCall(PetscLayoutSetUp(mat->cmap));
5130: /* check if the matrix sizes are correct */
5131: PetscCall(MatGetSize(mat, &rows, &cols));
5132: 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);
5134: /* read in row lengths */
5135: PetscCall(PetscMalloc1(M, &rowlens));
5136: PetscCall(PetscViewerBinaryRead(viewer, rowlens, M, NULL, PETSC_INT));
5137: /* check if sum(rowlens) is same as nz */
5138: sum = 0;
5139: for (i = 0; i < M; i++) sum += rowlens[i];
5140: PetscCheck(sum == nz, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Inconsistent matrix data in file: nonzeros = %" PetscInt_FMT ", sum-row-lengths = %" PetscInt_FMT, nz, sum);
5141: /* preallocate and check sizes */
5142: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(mat, 0, rowlens));
5143: PetscCall(MatGetSize(mat, &rows, &cols));
5144: PetscCheck(M == rows && N == cols, PETSC_COMM_SELF, PETSC_ERR_FILE_UNEXPECTED, "Matrix in file of different length (%" PetscInt_FMT ", %" PetscInt_FMT ") than the input matrix (%" PetscInt_FMT ", %" PetscInt_FMT ")", M, N, rows, cols);
5145: /* store row lengths */
5146: PetscCall(PetscArraycpy(a->ilen, rowlens, M));
5147: PetscCall(PetscFree(rowlens));
5149: /* fill in "i" row pointers */
5150: a->i[0] = 0;
5151: for (i = 0; i < M; i++) a->i[i + 1] = a->i[i] + a->ilen[i];
5152: /* read in "j" column indices */
5153: PetscCall(PetscViewerBinaryRead(viewer, a->j, nz, NULL, PETSC_INT));
5154: /* read in "a" nonzero values */
5155: PetscCall(PetscViewerBinaryRead(viewer, a->a, nz, NULL, PETSC_SCALAR));
5157: PetscCall(MatAssemblyBegin(mat, MAT_FINAL_ASSEMBLY));
5158: PetscCall(MatAssemblyEnd(mat, MAT_FINAL_ASSEMBLY));
5159: PetscFunctionReturn(PETSC_SUCCESS);
5160: }
5162: PetscErrorCode MatEqual_SeqAIJ(Mat A, Mat B, PetscBool *flg)
5163: {
5164: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data, *b = (Mat_SeqAIJ *)B->data;
5165: const PetscScalar *aa, *ba;
5166: #if defined(PETSC_USE_COMPLEX)
5167: PetscInt k;
5168: #endif
5170: PetscFunctionBegin;
5171: /* If the matrix dimensions are not equal,or no of nonzeros */
5172: if ((A->rmap->n != B->rmap->n) || (A->cmap->n != B->cmap->n) || (a->nz != b->nz)) {
5173: *flg = PETSC_FALSE;
5174: PetscFunctionReturn(PETSC_SUCCESS);
5175: }
5177: /* if the a->i are the same */
5178: PetscCall(PetscArraycmp(a->i, b->i, A->rmap->n + 1, flg));
5179: if (!*flg) PetscFunctionReturn(PETSC_SUCCESS);
5181: /* if a->j are the same */
5182: PetscCall(PetscArraycmp(a->j, b->j, a->nz, flg));
5183: if (!*flg) PetscFunctionReturn(PETSC_SUCCESS);
5185: PetscCall(MatSeqAIJGetArrayRead(A, &aa));
5186: PetscCall(MatSeqAIJGetArrayRead(B, &ba));
5187: /* if a->a are the same */
5188: #if defined(PETSC_USE_COMPLEX)
5189: for (k = 0; k < a->nz; k++) {
5190: if (PetscRealPart(aa[k]) != PetscRealPart(ba[k]) || PetscImaginaryPart(aa[k]) != PetscImaginaryPart(ba[k])) {
5191: *flg = PETSC_FALSE;
5192: PetscFunctionReturn(PETSC_SUCCESS);
5193: }
5194: }
5195: #else
5196: PetscCall(PetscArraycmp(aa, ba, a->nz, flg));
5197: #endif
5198: PetscCall(MatSeqAIJRestoreArrayRead(A, &aa));
5199: PetscCall(MatSeqAIJRestoreArrayRead(B, &ba));
5200: PetscFunctionReturn(PETSC_SUCCESS);
5201: }
5203: /*@
5204: MatCreateSeqAIJWithArrays - Creates an sequential `MATSEQAIJ` matrix using matrix elements (in CSR format)
5205: provided by the user.
5207: Collective
5209: Input Parameters:
5210: + comm - must be an MPI communicator of size 1
5211: . m - number of rows
5212: . n - number of columns
5213: . i - row indices; that is i[0] = 0, i[row] = i[row-1] + number of elements in that row of the matrix
5214: . j - column indices
5215: - a - matrix values
5217: Output Parameter:
5218: . mat - the matrix
5220: Level: intermediate
5222: Notes:
5223: The `i`, `j`, and `a` arrays are not copied by this routine, the user must free these arrays
5224: once the matrix is destroyed and not before
5226: You cannot set new nonzero locations into this matrix, that will generate an error.
5228: The `i` and `j` indices are 0 based
5230: The format which is used for the sparse matrix input, is equivalent to a
5231: row-major ordering.. i.e for the following matrix, the input data expected is
5232: as shown
5233: .vb
5234: 1 0 0
5235: 2 0 3
5236: 4 5 6
5238: i = {0,1,3,6} [size = nrow+1 = 3+1]
5239: j = {0,0,2,0,1,2} [size = 6]; values must be sorted for each row
5240: v = {1,2,3,4,5,6} [size = 6]
5241: .ve
5243: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatCreateAIJ()`, `MatCreateSeqAIJ()`, `MatCreateMPIAIJWithArrays()`, `MatMPIAIJSetPreallocationCSR()`
5244: @*/
5245: PetscErrorCode MatCreateSeqAIJWithArrays(MPI_Comm comm, PetscInt m, PetscInt n, PetscInt i[], PetscInt j[], PetscScalar a[], Mat *mat)
5246: {
5247: PetscInt ii;
5248: Mat_SeqAIJ *aij;
5249: PetscInt jj;
5251: PetscFunctionBegin;
5252: PetscCheck(m <= 0 || i[0] == 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "i (row indices) must start with 0");
5253: PetscCall(MatCreate(comm, mat));
5254: PetscCall(MatSetSizes(*mat, m, n, m, n));
5255: /* PetscCall(MatSetBlockSizes(*mat,,)); */
5256: PetscCall(MatSetType(*mat, MATSEQAIJ));
5257: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(*mat, MAT_SKIP_ALLOCATION, NULL));
5258: aij = (Mat_SeqAIJ *)(*mat)->data;
5259: PetscCall(PetscMalloc1(m, &aij->imax));
5260: PetscCall(PetscMalloc1(m, &aij->ilen));
5262: aij->i = i;
5263: aij->j = j;
5264: aij->a = a;
5265: aij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
5266: aij->free_a = PETSC_FALSE;
5267: aij->free_ij = PETSC_FALSE;
5269: for (ii = 0, aij->nonzerorowcnt = 0, aij->rmax = 0; ii < m; ii++) {
5270: aij->ilen[ii] = aij->imax[ii] = i[ii + 1] - i[ii];
5271: if (PetscDefined(USE_DEBUG)) {
5272: PetscCheck(i[ii + 1] - i[ii] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative row length in i (row indices) row = %" PetscInt_FMT " length = %" PetscInt_FMT, ii, i[ii + 1] - i[ii]);
5273: for (jj = i[ii] + 1; jj < i[ii + 1]; jj++) {
5274: PetscCheck(j[jj] >= j[jj - 1], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column entry number %" PetscInt_FMT " (actual column %" PetscInt_FMT ") in row %" PetscInt_FMT " is not sorted", jj - i[ii], j[jj], ii);
5275: PetscCheck(j[jj] != j[jj - 1], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column entry number %" PetscInt_FMT " (actual column %" PetscInt_FMT ") in row %" PetscInt_FMT " is identical to previous entry", jj - i[ii], j[jj], ii);
5276: }
5277: }
5278: }
5279: if (PetscDefined(USE_DEBUG)) {
5280: for (ii = 0; ii < aij->i[m]; ii++) {
5281: PetscCheck(j[ii] >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Negative column index at location = %" PetscInt_FMT " index = %" PetscInt_FMT, ii, j[ii]);
5282: PetscCheck(j[ii] <= n - 1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column index to large at location = %" PetscInt_FMT " index = %" PetscInt_FMT " last column = %" PetscInt_FMT, ii, j[ii], n - 1);
5283: }
5284: }
5286: PetscCall(MatAssemblyBegin(*mat, MAT_FINAL_ASSEMBLY));
5287: PetscCall(MatAssemblyEnd(*mat, MAT_FINAL_ASSEMBLY));
5288: PetscFunctionReturn(PETSC_SUCCESS);
5289: }
5291: /*@
5292: MatCreateSeqAIJFromTriple - Creates an sequential `MATSEQAIJ` matrix using matrix elements (in COO format)
5293: provided by the user.
5295: Collective
5297: Input Parameters:
5298: + comm - must be an MPI communicator of size 1
5299: . m - number of rows
5300: . n - number of columns
5301: . i - row indices
5302: . j - column indices
5303: . a - matrix values
5304: . nz - number of nonzeros
5305: - idx - if the `i` and `j` indices start with 1 use `PETSC_TRUE` otherwise use `PETSC_FALSE`
5307: Output Parameter:
5308: . mat - the matrix
5310: Level: intermediate
5312: Example:
5313: For the following matrix, the input data expected is as shown (using 0 based indexing)
5314: .vb
5315: 1 0 0
5316: 2 0 3
5317: 4 5 6
5319: i = {0,1,1,2,2,2}
5320: j = {0,0,2,0,1,2}
5321: v = {1,2,3,4,5,6}
5322: .ve
5324: Note:
5325: Instead of using this function, users should also consider `MatSetPreallocationCOO()` and `MatSetValuesCOO()`, which allow repeated or remote entries,
5326: and are particularly useful in iterative applications.
5328: .seealso: [](ch_matrices), `Mat`, `MatCreate()`, `MatCreateAIJ()`, `MatCreateSeqAIJ()`, `MatCreateSeqAIJWithArrays()`, `MatMPIAIJSetPreallocationCSR()`, `MatSetValuesCOO()`, `MatSetPreallocationCOO()`
5329: @*/
5330: PetscErrorCode MatCreateSeqAIJFromTriple(MPI_Comm comm, PetscInt m, PetscInt n, PetscInt i[], PetscInt j[], PetscScalar a[], Mat *mat, PetscInt nz, PetscBool idx)
5331: {
5332: PetscInt ii, *nnz, one = 1, row, col;
5334: PetscFunctionBegin;
5335: PetscCall(PetscCalloc1(m, &nnz));
5336: for (ii = 0; ii < nz; ii++) nnz[i[ii] - !!idx] += 1;
5337: PetscCall(MatCreate(comm, mat));
5338: PetscCall(MatSetSizes(*mat, m, n, m, n));
5339: PetscCall(MatSetType(*mat, MATSEQAIJ));
5340: PetscCall(MatSeqAIJSetPreallocation_SeqAIJ(*mat, 0, nnz));
5341: for (ii = 0; ii < nz; ii++) {
5342: if (idx) {
5343: row = i[ii] - 1;
5344: col = j[ii] - 1;
5345: } else {
5346: row = i[ii];
5347: col = j[ii];
5348: }
5349: PetscCall(MatSetValues(*mat, one, &row, one, &col, &a[ii], ADD_VALUES));
5350: }
5351: PetscCall(MatAssemblyBegin(*mat, MAT_FINAL_ASSEMBLY));
5352: PetscCall(MatAssemblyEnd(*mat, MAT_FINAL_ASSEMBLY));
5353: PetscCall(PetscFree(nnz));
5354: PetscFunctionReturn(PETSC_SUCCESS);
5355: }
5357: PetscErrorCode MatSeqAIJInvalidateDiagonal(Mat A)
5358: {
5359: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
5361: PetscFunctionBegin;
5362: a->idiagvalid = PETSC_FALSE;
5363: a->ibdiagvalid = PETSC_FALSE;
5365: PetscCall(MatSeqAIJInvalidateDiagonal_Inode(A));
5366: PetscFunctionReturn(PETSC_SUCCESS);
5367: }
5369: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_SeqAIJ(MPI_Comm comm, Mat inmat, PetscInt n, MatReuse scall, Mat *outmat)
5370: {
5371: PetscFunctionBegin;
5372: PetscCall(MatCreateMPIMatConcatenateSeqMat_MPIAIJ(comm, inmat, n, scall, outmat));
5373: PetscFunctionReturn(PETSC_SUCCESS);
5374: }
5376: /*
5377: Permute A into C's *local* index space using rowemb,colemb.
5378: The embedding are supposed to be injections and the above implies that the range of rowemb is a subset
5379: of [0,m), colemb is in [0,n).
5380: If pattern == DIFFERENT_NONZERO_PATTERN, C is preallocated according to A.
5381: */
5382: PetscErrorCode MatSetSeqMat_SeqAIJ(Mat C, IS rowemb, IS colemb, MatStructure pattern, Mat B)
5383: {
5384: /* If making this function public, change the error returned in this function away from _PLIB. */
5385: Mat_SeqAIJ *Baij;
5386: PetscBool seqaij;
5387: PetscInt m, n, *nz, i, j, count;
5388: PetscScalar v;
5389: const PetscInt *rowindices, *colindices;
5391: PetscFunctionBegin;
5392: if (!B) PetscFunctionReturn(PETSC_SUCCESS);
5393: /* Check to make sure the target matrix (and embeddings) are compatible with C and each other. */
5394: PetscCall(PetscObjectBaseTypeCompare((PetscObject)B, MATSEQAIJ, &seqaij));
5395: PetscCheck(seqaij, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Input matrix is of wrong type");
5396: if (rowemb) {
5397: PetscCall(ISGetLocalSize(rowemb, &m));
5398: PetscCheck(m == B->rmap->n, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Row IS of size %" PetscInt_FMT " is incompatible with matrix row size %" PetscInt_FMT, m, B->rmap->n);
5399: } else {
5400: PetscCheck(C->rmap->n == B->rmap->n, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Input matrix is row-incompatible with the target matrix");
5401: }
5402: if (colemb) {
5403: PetscCall(ISGetLocalSize(colemb, &n));
5404: PetscCheck(n == B->cmap->n, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Diag col IS of size %" PetscInt_FMT " is incompatible with input matrix col size %" PetscInt_FMT, n, B->cmap->n);
5405: } else {
5406: PetscCheck(C->cmap->n == B->cmap->n, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Input matrix is col-incompatible with the target matrix");
5407: }
5409: Baij = (Mat_SeqAIJ *)B->data;
5410: if (pattern == DIFFERENT_NONZERO_PATTERN) {
5411: PetscCall(PetscMalloc1(B->rmap->n, &nz));
5412: for (i = 0; i < B->rmap->n; i++) nz[i] = Baij->i[i + 1] - Baij->i[i];
5413: PetscCall(MatSeqAIJSetPreallocation(C, 0, nz));
5414: PetscCall(PetscFree(nz));
5415: }
5416: if (pattern == SUBSET_NONZERO_PATTERN) PetscCall(MatZeroEntries(C));
5417: count = 0;
5418: rowindices = NULL;
5419: colindices = NULL;
5420: if (rowemb) PetscCall(ISGetIndices(rowemb, &rowindices));
5421: if (colemb) PetscCall(ISGetIndices(colemb, &colindices));
5422: for (i = 0; i < B->rmap->n; i++) {
5423: PetscInt row;
5424: row = i;
5425: if (rowindices) row = rowindices[i];
5426: for (j = Baij->i[i]; j < Baij->i[i + 1]; j++) {
5427: PetscInt col;
5428: col = Baij->j[count];
5429: if (colindices) col = colindices[col];
5430: v = Baij->a[count];
5431: PetscCall(MatSetValues(C, 1, &row, 1, &col, &v, INSERT_VALUES));
5432: ++count;
5433: }
5434: }
5435: /* FIXME: set C's nonzerostate correctly. */
5436: /* Assembly for C is necessary. */
5437: C->preallocated = PETSC_TRUE;
5438: C->assembled = PETSC_TRUE;
5439: C->was_assembled = PETSC_FALSE;
5440: PetscFunctionReturn(PETSC_SUCCESS);
5441: }
5443: PetscErrorCode MatEliminateZeros_SeqAIJ(Mat A, PetscBool keep)
5444: {
5445: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
5446: MatScalar *aa = a->a;
5447: PetscInt m = A->rmap->n, fshift = 0, fshift_prev = 0, i, k;
5448: PetscInt *ailen = a->ilen, *imax = a->imax, *ai = a->i, *aj = a->j, rmax = 0;
5450: PetscFunctionBegin;
5451: PetscCheck(A->assembled, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Cannot eliminate zeros for unassembled matrix");
5452: if (m) rmax = ailen[0]; /* determine row with most nonzeros */
5453: for (i = 1; i <= m; i++) {
5454: /* move each nonzero entry back by the amount of zero slots (fshift) before it*/
5455: for (k = ai[i - 1]; k < ai[i]; k++) {
5456: if (aa[k] == 0 && (aj[k] != i - 1 || !keep)) fshift++;
5457: else {
5458: if (aa[k] == 0 && aj[k] == i - 1) PetscCall(PetscInfo(A, "Keep the diagonal zero at row %" PetscInt_FMT "\n", i - 1));
5459: aa[k - fshift] = aa[k];
5460: aj[k - fshift] = aj[k];
5461: }
5462: }
5463: ai[i - 1] -= fshift_prev; // safe to update ai[i-1] now since it will not be used in the next iteration
5464: fshift_prev = fshift;
5465: /* reset ilen and imax for each row */
5466: ailen[i - 1] = imax[i - 1] = ai[i] - fshift - ai[i - 1];
5467: a->nonzerorowcnt += ((ai[i] - fshift - ai[i - 1]) > 0);
5468: rmax = PetscMax(rmax, ailen[i - 1]);
5469: }
5470: if (fshift) {
5471: if (m) {
5472: ai[m] -= fshift;
5473: a->nz = ai[m];
5474: }
5475: PetscCall(PetscInfo(A, "Matrix size: %" PetscInt_FMT " X %" PetscInt_FMT "; zeros eliminated: %" PetscInt_FMT "; nonzeros left: %" PetscInt_FMT "\n", m, A->cmap->n, fshift, a->nz));
5476: A->nonzerostate++;
5477: A->info.nz_unneeded += (PetscReal)fshift;
5478: a->rmax = rmax;
5479: if (a->inode.use && a->inode.checked) PetscCall(MatSeqAIJCheckInode(A));
5480: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
5481: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
5482: }
5483: PetscFunctionReturn(PETSC_SUCCESS);
5484: }
5486: PetscFunctionList MatSeqAIJList = NULL;
5488: /*@
5489: MatSeqAIJSetType - Converts a `MATSEQAIJ` matrix to a subtype
5491: Collective
5493: Input Parameters:
5494: + mat - the matrix object
5495: - matype - matrix type
5497: Options Database Key:
5498: . -mat_seqaij_type <method> - for example seqaijcrl
5500: Level: intermediate
5502: .seealso: [](ch_matrices), `Mat`, `PCSetType()`, `VecSetType()`, `MatCreate()`, `MatType`
5503: @*/
5504: PetscErrorCode MatSeqAIJSetType(Mat mat, MatType matype)
5505: {
5506: PetscBool sametype;
5507: PetscErrorCode (*r)(Mat, MatType, MatReuse, Mat *);
5509: PetscFunctionBegin;
5511: PetscCall(PetscObjectTypeCompare((PetscObject)mat, matype, &sametype));
5512: if (sametype) PetscFunctionReturn(PETSC_SUCCESS);
5514: PetscCall(PetscFunctionListFind(MatSeqAIJList, matype, &r));
5515: PetscCheck(r, PetscObjectComm((PetscObject)mat), PETSC_ERR_ARG_UNKNOWN_TYPE, "Unknown Mat type given: %s", matype);
5516: PetscCall((*r)(mat, matype, MAT_INPLACE_MATRIX, &mat));
5517: PetscFunctionReturn(PETSC_SUCCESS);
5518: }
5520: /*@C
5521: MatSeqAIJRegister - - Adds a new sub-matrix type for sequential `MATSEQAIJ` matrices
5523: Not Collective, No Fortran Support
5525: Input Parameters:
5526: + sname - name of a new user-defined matrix type, for example `MATSEQAIJCRL`
5527: - function - routine to convert to subtype
5529: Level: advanced
5531: Notes:
5532: `MatSeqAIJRegister()` may be called multiple times to add several user-defined solvers.
5534: Then, your matrix can be chosen with the procedural interface at runtime via the option
5535: $ -mat_seqaij_type my_mat
5537: .seealso: [](ch_matrices), `Mat`, `MatSeqAIJRegisterAll()`
5538: @*/
5539: PetscErrorCode MatSeqAIJRegister(const char sname[], PetscErrorCode (*function)(Mat, MatType, MatReuse, Mat *))
5540: {
5541: PetscFunctionBegin;
5542: PetscCall(MatInitializePackage());
5543: PetscCall(PetscFunctionListAdd(&MatSeqAIJList, sname, function));
5544: PetscFunctionReturn(PETSC_SUCCESS);
5545: }
5547: PetscBool MatSeqAIJRegisterAllCalled = PETSC_FALSE;
5549: /*@C
5550: MatSeqAIJRegisterAll - Registers all of the matrix subtypes of `MATSSEQAIJ`
5552: Not Collective
5554: Level: advanced
5556: Note:
5557: This registers the versions of `MATSEQAIJ` for GPUs
5559: .seealso: [](ch_matrices), `Mat`, `MatRegisterAll()`, `MatSeqAIJRegister()`
5560: @*/
5561: PetscErrorCode MatSeqAIJRegisterAll(void)
5562: {
5563: PetscFunctionBegin;
5564: if (MatSeqAIJRegisterAllCalled) PetscFunctionReturn(PETSC_SUCCESS);
5565: MatSeqAIJRegisterAllCalled = PETSC_TRUE;
5567: PetscCall(MatSeqAIJRegister(MATSEQAIJCRL, MatConvert_SeqAIJ_SeqAIJCRL));
5568: PetscCall(MatSeqAIJRegister(MATSEQAIJPERM, MatConvert_SeqAIJ_SeqAIJPERM));
5569: PetscCall(MatSeqAIJRegister(MATSEQAIJSELL, MatConvert_SeqAIJ_SeqAIJSELL));
5570: #if defined(PETSC_HAVE_MKL_SPARSE)
5571: PetscCall(MatSeqAIJRegister(MATSEQAIJMKL, MatConvert_SeqAIJ_SeqAIJMKL));
5572: #endif
5573: #if defined(PETSC_HAVE_CUDA)
5574: PetscCall(MatSeqAIJRegister(MATSEQAIJCUSPARSE, MatConvert_SeqAIJ_SeqAIJCUSPARSE));
5575: #endif
5576: #if defined(PETSC_HAVE_HIP)
5577: PetscCall(MatSeqAIJRegister(MATSEQAIJHIPSPARSE, MatConvert_SeqAIJ_SeqAIJHIPSPARSE));
5578: #endif
5579: #if defined(PETSC_HAVE_KOKKOS_KERNELS)
5580: PetscCall(MatSeqAIJRegister(MATSEQAIJKOKKOS, MatConvert_SeqAIJ_SeqAIJKokkos));
5581: #endif
5582: #if defined(PETSC_HAVE_VIENNACL) && defined(PETSC_HAVE_VIENNACL_NO_CUDA)
5583: PetscCall(MatSeqAIJRegister(MATMPIAIJVIENNACL, MatConvert_SeqAIJ_SeqAIJViennaCL));
5584: #endif
5585: PetscFunctionReturn(PETSC_SUCCESS);
5586: }
5588: /*
5589: Special version for direct calls from Fortran
5590: */
5591: #if defined(PETSC_HAVE_FORTRAN_CAPS)
5592: #define matsetvaluesseqaij_ MATSETVALUESSEQAIJ
5593: #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE)
5594: #define matsetvaluesseqaij_ matsetvaluesseqaij
5595: #endif
5597: /* Change these macros so can be used in void function */
5599: /* Change these macros so can be used in void function */
5600: /* Identical to PetscCallVoid, except it assigns to *_ierr */
5601: #undef PetscCall
5602: #define PetscCall(...) \
5603: do { \
5604: PetscErrorCode ierr_msv_mpiaij = __VA_ARGS__; \
5605: if (PetscUnlikely(ierr_msv_mpiaij)) { \
5606: *_ierr = PetscError(PETSC_COMM_SELF, __LINE__, PETSC_FUNCTION_NAME, __FILE__, ierr_msv_mpiaij, PETSC_ERROR_REPEAT, " "); \
5607: return; \
5608: } \
5609: } while (0)
5611: #undef SETERRQ
5612: #define SETERRQ(comm, ierr, ...) \
5613: do { \
5614: *_ierr = PetscError(comm, __LINE__, PETSC_FUNCTION_NAME, __FILE__, ierr, PETSC_ERROR_INITIAL, __VA_ARGS__); \
5615: return; \
5616: } while (0)
5618: PETSC_EXTERN void matsetvaluesseqaij_(Mat *AA, PetscInt *mm, const PetscInt im[], PetscInt *nn, const PetscInt in[], const PetscScalar v[], InsertMode *isis, PetscErrorCode *_ierr)
5619: {
5620: Mat A = *AA;
5621: PetscInt m = *mm, n = *nn;
5622: InsertMode is = *isis;
5623: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data;
5624: PetscInt *rp, k, low, high, t, ii, row, nrow, i, col, l, rmax, N;
5625: PetscInt *imax, *ai, *ailen;
5626: PetscInt *aj, nonew = a->nonew, lastcol = -1;
5627: MatScalar *ap, value, *aa;
5628: PetscBool ignorezeroentries = a->ignorezeroentries;
5629: PetscBool roworiented = a->roworiented;
5631: PetscFunctionBegin;
5632: MatCheckPreallocated(A, 1);
5633: imax = a->imax;
5634: ai = a->i;
5635: ailen = a->ilen;
5636: aj = a->j;
5637: aa = a->a;
5639: for (k = 0; k < m; k++) { /* loop over added rows */
5640: row = im[k];
5641: if (row < 0) continue;
5642: PetscCheck(row < A->rmap->n, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_OUTOFRANGE, "Row too large");
5643: rp = aj + ai[row];
5644: ap = aa + ai[row];
5645: rmax = imax[row];
5646: nrow = ailen[row];
5647: low = 0;
5648: high = nrow;
5649: for (l = 0; l < n; l++) { /* loop over added columns */
5650: if (in[l] < 0) continue;
5651: PetscCheck(in[l] < A->cmap->n, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_OUTOFRANGE, "Column too large");
5652: col = in[l];
5653: if (roworiented) value = v[l + k * n];
5654: else value = v[k + l * m];
5656: if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue;
5658: if (col <= lastcol) low = 0;
5659: else high = nrow;
5660: lastcol = col;
5661: while (high - low > 5) {
5662: t = (low + high) / 2;
5663: if (rp[t] > col) high = t;
5664: else low = t;
5665: }
5666: for (i = low; i < high; i++) {
5667: if (rp[i] > col) break;
5668: if (rp[i] == col) {
5669: if (is == ADD_VALUES) ap[i] += value;
5670: else ap[i] = value;
5671: goto noinsert;
5672: }
5673: }
5674: if (value == 0.0 && ignorezeroentries) goto noinsert;
5675: if (nonew == 1) goto noinsert;
5676: PetscCheck(nonew != -1, PetscObjectComm((PetscObject)A), PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero in the matrix");
5677: MatSeqXAIJReallocateAIJ(A, A->rmap->n, 1, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
5678: N = nrow++ - 1;
5679: a->nz++;
5680: high++;
5681: /* shift up all the later entries in this row */
5682: for (ii = N; ii >= i; ii--) {
5683: rp[ii + 1] = rp[ii];
5684: ap[ii + 1] = ap[ii];
5685: }
5686: rp[i] = col;
5687: ap[i] = value;
5688: noinsert:;
5689: low = i + 1;
5690: }
5691: ailen[row] = nrow;
5692: }
5693: PetscFunctionReturnVoid();
5694: }
5695: /* Undefining these here since they were redefined from their original definition above! No
5696: * other PETSc functions should be defined past this point, as it is impossible to recover the
5697: * original definitions */
5698: #undef PetscCall
5699: #undef SETERRQ