Actual source code: mpisbaij.c
1: #define PETSCMAT_DLL
3: #include ../src/mat/impls/baij/mpi/mpibaij.h
4: #include mpisbaij.h
5: #include ../src/mat/impls/sbaij/seq/sbaij.h
6: #include petscblaslapack.h
8: EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ(Mat);
9: EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ_2comm(Mat);
10: EXTERN PetscErrorCode DisAssemble_MPISBAIJ(Mat);
11: EXTERN PetscErrorCode MatIncreaseOverlap_MPISBAIJ(Mat,PetscInt,IS[],PetscInt);
12: EXTERN PetscErrorCode MatGetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []);
13: EXTERN PetscErrorCode MatGetValues_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []);
14: EXTERN PetscErrorCode MatSetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt [],PetscInt,const PetscInt [],const PetscScalar [],InsertMode);
15: EXTERN PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode);
16: EXTERN PetscErrorCode MatSetValuesBlocked_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode);
17: EXTERN PetscErrorCode MatGetRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**);
18: EXTERN PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**);
19: EXTERN PetscErrorCode MatZeroRows_SeqSBAIJ(Mat,IS,PetscScalar*);
20: EXTERN PetscErrorCode MatZeroRows_SeqBAIJ(Mat,IS,PetscScalar *);
21: EXTERN PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat,Vec,PetscInt[]);
22: EXTERN PetscErrorCode MatSOR_MPISBAIJ(Mat,Vec,PetscReal,MatSORType,PetscReal,PetscInt,PetscInt,Vec);
27: PetscErrorCode MatStoreValues_MPISBAIJ(Mat mat)
28: {
29: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
33: MatStoreValues(aij->A);
34: MatStoreValues(aij->B);
35: return(0);
36: }
42: PetscErrorCode MatRetrieveValues_MPISBAIJ(Mat mat)
43: {
44: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
48: MatRetrieveValues(aij->A);
49: MatRetrieveValues(aij->B);
50: return(0);
51: }
55: #define CHUNKSIZE 10
57: #define MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv) \
58: { \
59: \
60: brow = row/bs; \
61: rp = aj + ai[brow]; ap = aa + bs2*ai[brow]; \
62: rmax = aimax[brow]; nrow = ailen[brow]; \
63: bcol = col/bs; \
64: ridx = row % bs; cidx = col % bs; \
65: low = 0; high = nrow; \
66: while (high-low > 3) { \
67: t = (low+high)/2; \
68: if (rp[t] > bcol) high = t; \
69: else low = t; \
70: } \
71: for (_i=low; _i<high; _i++) { \
72: if (rp[_i] > bcol) break; \
73: if (rp[_i] == bcol) { \
74: bap = ap + bs2*_i + bs*cidx + ridx; \
75: if (addv == ADD_VALUES) *bap += value; \
76: else *bap = value; \
77: goto a_noinsert; \
78: } \
79: } \
80: if (a->nonew == 1) goto a_noinsert; \
81: if (a->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \
82: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,aimax,a->nonew,MatScalar); \
83: N = nrow++ - 1; \
84: /* shift up all the later entries in this row */ \
85: for (ii=N; ii>=_i; ii--) { \
86: rp[ii+1] = rp[ii]; \
87: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
88: } \
89: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar)); } \
90: rp[_i] = bcol; \
91: ap[bs2*_i + bs*cidx + ridx] = value; \
92: a_noinsert:; \
93: ailen[brow] = nrow; \
94: }
96: #define MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv) \
97: { \
98: brow = row/bs; \
99: rp = bj + bi[brow]; ap = ba + bs2*bi[brow]; \
100: rmax = bimax[brow]; nrow = bilen[brow]; \
101: bcol = col/bs; \
102: ridx = row % bs; cidx = col % bs; \
103: low = 0; high = nrow; \
104: while (high-low > 3) { \
105: t = (low+high)/2; \
106: if (rp[t] > bcol) high = t; \
107: else low = t; \
108: } \
109: for (_i=low; _i<high; _i++) { \
110: if (rp[_i] > bcol) break; \
111: if (rp[_i] == bcol) { \
112: bap = ap + bs2*_i + bs*cidx + ridx; \
113: if (addv == ADD_VALUES) *bap += value; \
114: else *bap = value; \
115: goto b_noinsert; \
116: } \
117: } \
118: if (b->nonew == 1) goto b_noinsert; \
119: if (b->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \
120: MatSeqXAIJReallocateAIJ(B,b->mbs,bs2,nrow,brow,bcol,rmax,ba,bi,bj,rp,ap,bimax,b->nonew,MatScalar); \
121: N = nrow++ - 1; \
122: /* shift up all the later entries in this row */ \
123: for (ii=N; ii>=_i; ii--) { \
124: rp[ii+1] = rp[ii]; \
125: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
126: } \
127: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));} \
128: rp[_i] = bcol; \
129: ap[bs2*_i + bs*cidx + ridx] = value; \
130: b_noinsert:; \
131: bilen[brow] = nrow; \
132: }
134: /* Only add/insert a(i,j) with i<=j (blocks).
135: Any a(i,j) with i>j input by user is ingored.
136: */
139: PetscErrorCode MatSetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
140: {
141: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
142: MatScalar value;
143: PetscTruth roworiented = baij->roworiented;
145: PetscInt i,j,row,col;
146: PetscInt rstart_orig=mat->rmap->rstart;
147: PetscInt rend_orig=mat->rmap->rend,cstart_orig=mat->cmap->rstart;
148: PetscInt cend_orig=mat->cmap->rend,bs=mat->rmap->bs;
150: /* Some Variables required in the macro */
151: Mat A = baij->A;
152: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)(A)->data;
153: PetscInt *aimax=a->imax,*ai=a->i,*ailen=a->ilen,*aj=a->j;
154: MatScalar *aa=a->a;
156: Mat B = baij->B;
157: Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(B)->data;
158: PetscInt *bimax=b->imax,*bi=b->i,*bilen=b->ilen,*bj=b->j;
159: MatScalar *ba=b->a;
161: PetscInt *rp,ii,nrow,_i,rmax,N,brow,bcol;
162: PetscInt low,high,t,ridx,cidx,bs2=a->bs2;
163: MatScalar *ap,*bap;
165: /* for stash */
166: PetscInt n_loc, *in_loc = PETSC_NULL;
167: MatScalar *v_loc = PETSC_NULL;
171: if (!baij->donotstash){
172: if (n > baij->n_loc) {
173: PetscFree(baij->in_loc);
174: PetscFree(baij->v_loc);
175: PetscMalloc(n*sizeof(PetscInt),&baij->in_loc);
176: PetscMalloc(n*sizeof(MatScalar),&baij->v_loc);
177: baij->n_loc = n;
178: }
179: in_loc = baij->in_loc;
180: v_loc = baij->v_loc;
181: }
183: for (i=0; i<m; i++) {
184: if (im[i] < 0) continue;
185: #if defined(PETSC_USE_DEBUG)
186: if (im[i] >= mat->rmap->N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],mat->rmap->N-1);
187: #endif
188: if (im[i] >= rstart_orig && im[i] < rend_orig) { /* this processor entry */
189: row = im[i] - rstart_orig; /* local row index */
190: for (j=0; j<n; j++) {
191: if (im[i]/bs > in[j]/bs){
192: if (a->ignore_ltriangular){
193: continue; /* ignore lower triangular blocks */
194: } else {
195: SETERRQ(PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
196: }
197: }
198: if (in[j] >= cstart_orig && in[j] < cend_orig){ /* diag entry (A) */
199: col = in[j] - cstart_orig; /* local col index */
200: brow = row/bs; bcol = col/bs;
201: if (brow > bcol) continue; /* ignore lower triangular blocks of A */
202: if (roworiented) value = v[i*n+j]; else value = v[i+j*m];
203: MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv);
204: /* MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv); */
205: } else if (in[j] < 0) continue;
206: #if defined(PETSC_USE_DEBUG)
207: else if (in[j] >= mat->cmap->N) {SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[j],mat->cmap->N-1);}
208: #endif
209: else { /* off-diag entry (B) */
210: if (mat->was_assembled) {
211: if (!baij->colmap) {
212: CreateColmap_MPIBAIJ_Private(mat);
213: }
214: #if defined (PETSC_USE_CTABLE)
215: PetscTableFind(baij->colmap,in[j]/bs + 1,&col);
216: col = col - 1;
217: #else
218: col = baij->colmap[in[j]/bs] - 1;
219: #endif
220: if (col < 0 && !((Mat_SeqSBAIJ*)(baij->A->data))->nonew) {
221: DisAssemble_MPISBAIJ(mat);
222: col = in[j];
223: /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */
224: B = baij->B;
225: b = (Mat_SeqBAIJ*)(B)->data;
226: bimax=b->imax;bi=b->i;bilen=b->ilen;bj=b->j;
227: ba=b->a;
228: } else col += in[j]%bs;
229: } else col = in[j];
230: if (roworiented) value = v[i*n+j]; else value = v[i+j*m];
231: MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv);
232: /* MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv); */
233: }
234: }
235: } else { /* off processor entry */
236: if (!baij->donotstash) {
237: n_loc = 0;
238: for (j=0; j<n; j++){
239: if (im[i]/bs > in[j]/bs) continue; /* ignore lower triangular blocks */
240: in_loc[n_loc] = in[j];
241: if (roworiented) {
242: v_loc[n_loc] = v[i*n+j];
243: } else {
244: v_loc[n_loc] = v[j*m+i];
245: }
246: n_loc++;
247: }
248: MatStashValuesRow_Private(&mat->stash,im[i],n_loc,in_loc,v_loc,PETSC_FALSE);
249: }
250: }
251: }
252: return(0);
253: }
257: PetscErrorCode MatSetValuesBlocked_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const MatScalar v[],InsertMode addv)
258: {
259: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
260: const MatScalar *value;
261: MatScalar *barray=baij->barray;
262: PetscTruth roworiented = baij->roworiented,ignore_ltriangular = ((Mat_SeqSBAIJ*)baij->A->data)->ignore_ltriangular;
263: PetscErrorCode ierr;
264: PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs;
265: PetscInt rend=baij->rendbs,cstart=baij->rstartbs,stepval;
266: PetscInt cend=baij->rendbs,bs=mat->rmap->bs,bs2=baij->bs2;
269: if(!barray) {
270: PetscMalloc(bs2*sizeof(MatScalar),&barray);
271: baij->barray = barray;
272: }
274: if (roworiented) {
275: stepval = (n-1)*bs;
276: } else {
277: stepval = (m-1)*bs;
278: }
279: for (i=0; i<m; i++) {
280: if (im[i] < 0) continue;
281: #if defined(PETSC_USE_DEBUG)
282: if (im[i] >= baij->Mbs) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large, row %D max %D",im[i],baij->Mbs-1);
283: #endif
284: if (im[i] >= rstart && im[i] < rend) {
285: row = im[i] - rstart;
286: for (j=0; j<n; j++) {
287: if (im[i] > in[j]) {
288: if (ignore_ltriangular) continue; /* ignore lower triangular blocks */
289: else SETERRQ(PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
290: }
291: /* If NumCol = 1 then a copy is not required */
292: if ((roworiented) && (n == 1)) {
293: barray = (MatScalar*) v + i*bs2;
294: } else if((!roworiented) && (m == 1)) {
295: barray = (MatScalar*) v + j*bs2;
296: } else { /* Here a copy is required */
297: if (roworiented) {
298: value = v + i*(stepval+bs)*bs + j*bs;
299: } else {
300: value = v + j*(stepval+bs)*bs + i*bs;
301: }
302: for (ii=0; ii<bs; ii++,value+=stepval) {
303: for (jj=0; jj<bs; jj++) {
304: *barray++ = *value++;
305: }
306: }
307: barray -=bs2;
308: }
309:
310: if (in[j] >= cstart && in[j] < cend){
311: col = in[j] - cstart;
312: MatSetValuesBlocked_SeqSBAIJ(baij->A,1,&row,1,&col,barray,addv);
313: }
314: else if (in[j] < 0) continue;
315: #if defined(PETSC_USE_DEBUG)
316: else if (in[j] >= baij->Nbs) {SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large, col %D max %D",in[j],baij->Nbs-1);}
317: #endif
318: else {
319: if (mat->was_assembled) {
320: if (!baij->colmap) {
321: CreateColmap_MPIBAIJ_Private(mat);
322: }
324: #if defined(PETSC_USE_DEBUG)
325: #if defined (PETSC_USE_CTABLE)
326: { PetscInt data;
327: PetscTableFind(baij->colmap,in[j]+1,&data);
328: if ((data - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap");
329: }
330: #else
331: if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap");
332: #endif
333: #endif
334: #if defined (PETSC_USE_CTABLE)
335: PetscTableFind(baij->colmap,in[j]+1,&col);
336: col = (col - 1)/bs;
337: #else
338: col = (baij->colmap[in[j]] - 1)/bs;
339: #endif
340: if (col < 0 && !((Mat_SeqBAIJ*)(baij->A->data))->nonew) {
341: DisAssemble_MPISBAIJ(mat);
342: col = in[j];
343: }
344: }
345: else col = in[j];
346: MatSetValuesBlocked_SeqBAIJ(baij->B,1,&row,1,&col,barray,addv);
347: }
348: }
349: } else {
350: if (!baij->donotstash) {
351: if (roworiented) {
352: MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
353: } else {
354: MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
355: }
356: }
357: }
358: }
359: return(0);
360: }
364: PetscErrorCode MatGetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[])
365: {
366: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
368: PetscInt bs=mat->rmap->bs,i,j,bsrstart = mat->rmap->rstart,bsrend = mat->rmap->rend;
369: PetscInt bscstart = mat->cmap->rstart,bscend = mat->cmap->rend,row,col,data;
372: for (i=0; i<m; i++) {
373: if (idxm[i] < 0) continue; /* SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",idxm[i]); */
374: if (idxm[i] >= mat->rmap->N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",idxm[i],mat->rmap->N-1);
375: if (idxm[i] >= bsrstart && idxm[i] < bsrend) {
376: row = idxm[i] - bsrstart;
377: for (j=0; j<n; j++) {
378: if (idxn[j] < 0) continue; /* SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative column %D",idxn[j]); */
379: if (idxn[j] >= mat->cmap->N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",idxn[j],mat->cmap->N-1);
380: if (idxn[j] >= bscstart && idxn[j] < bscend){
381: col = idxn[j] - bscstart;
382: MatGetValues_SeqSBAIJ(baij->A,1,&row,1,&col,v+i*n+j);
383: } else {
384: if (!baij->colmap) {
385: CreateColmap_MPIBAIJ_Private(mat);
386: }
387: #if defined (PETSC_USE_CTABLE)
388: PetscTableFind(baij->colmap,idxn[j]/bs+1,&data);
389: data --;
390: #else
391: data = baij->colmap[idxn[j]/bs]-1;
392: #endif
393: if((data < 0) || (baij->garray[data/bs] != idxn[j]/bs)) *(v+i*n+j) = 0.0;
394: else {
395: col = data + idxn[j]%bs;
396: MatGetValues_SeqBAIJ(baij->B,1,&row,1,&col,v+i*n+j);
397: }
398: }
399: }
400: } else {
401: SETERRQ(PETSC_ERR_SUP,"Only local values currently supported");
402: }
403: }
404: return(0);
405: }
409: PetscErrorCode MatNorm_MPISBAIJ(Mat mat,NormType type,PetscReal *norm)
410: {
411: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
413: PetscReal sum[2],*lnorm2;
416: if (baij->size == 1) {
417: MatNorm(baij->A,type,norm);
418: } else {
419: if (type == NORM_FROBENIUS) {
420: PetscMalloc(2*sizeof(PetscReal),&lnorm2);
421: MatNorm(baij->A,type,lnorm2);
422: *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2++; /* squar power of norm(A) */
423: MatNorm(baij->B,type,lnorm2);
424: *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2--; /* squar power of norm(B) */
425: MPI_Allreduce(lnorm2,&sum,2,MPIU_REAL,MPI_SUM,((PetscObject)mat)->comm);
426: *norm = sqrt(sum[0] + 2*sum[1]);
427: PetscFree(lnorm2);
428: } else if (type == NORM_INFINITY || type == NORM_1) { /* max row/column sum */
429: Mat_SeqSBAIJ *amat=(Mat_SeqSBAIJ*)baij->A->data;
430: Mat_SeqBAIJ *bmat=(Mat_SeqBAIJ*)baij->B->data;
431: PetscReal *rsum,*rsum2,vabs;
432: PetscInt *jj,*garray=baij->garray,rstart=baij->rstartbs,nz;
433: PetscInt brow,bcol,col,bs=baij->A->rmap->bs,row,grow,gcol,mbs=amat->mbs;
434: MatScalar *v;
436: PetscMalloc2(mat->cmap->N,PetscReal,&rsum,mat->cmap->N,PetscReal,&rsum2);
437: PetscMemzero(rsum,mat->cmap->N*sizeof(PetscReal));
438: /* Amat */
439: v = amat->a; jj = amat->j;
440: for (brow=0; brow<mbs; brow++) {
441: grow = bs*(rstart + brow);
442: nz = amat->i[brow+1] - amat->i[brow];
443: for (bcol=0; bcol<nz; bcol++){
444: gcol = bs*(rstart + *jj); jj++;
445: for (col=0; col<bs; col++){
446: for (row=0; row<bs; row++){
447: vabs = PetscAbsScalar(*v); v++;
448: rsum[gcol+col] += vabs;
449: /* non-diagonal block */
450: if (bcol > 0 && vabs > 0.0) rsum[grow+row] += vabs;
451: }
452: }
453: }
454: }
455: /* Bmat */
456: v = bmat->a; jj = bmat->j;
457: for (brow=0; brow<mbs; brow++) {
458: grow = bs*(rstart + brow);
459: nz = bmat->i[brow+1] - bmat->i[brow];
460: for (bcol=0; bcol<nz; bcol++){
461: gcol = bs*garray[*jj]; jj++;
462: for (col=0; col<bs; col++){
463: for (row=0; row<bs; row++){
464: vabs = PetscAbsScalar(*v); v++;
465: rsum[gcol+col] += vabs;
466: rsum[grow+row] += vabs;
467: }
468: }
469: }
470: }
471: MPI_Allreduce(rsum,rsum2,mat->cmap->N,MPIU_REAL,MPI_SUM,((PetscObject)mat)->comm);
472: *norm = 0.0;
473: for (col=0; col<mat->cmap->N; col++) {
474: if (rsum2[col] > *norm) *norm = rsum2[col];
475: }
476: PetscFree2(rsum,rsum2);
477: } else {
478: SETERRQ(PETSC_ERR_SUP,"No support for this norm yet");
479: }
480: }
481: return(0);
482: }
486: PetscErrorCode MatAssemblyBegin_MPISBAIJ(Mat mat,MatAssemblyType mode)
487: {
488: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
490: PetscInt nstash,reallocs;
491: InsertMode addv;
494: if (baij->donotstash) {
495: return(0);
496: }
498: /* make sure all processors are either in INSERTMODE or ADDMODE */
499: MPI_Allreduce(&mat->insertmode,&addv,1,MPI_INT,MPI_BOR,((PetscObject)mat)->comm);
500: if (addv == (ADD_VALUES|INSERT_VALUES)) {
501: SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Some processors inserted others added");
502: }
503: mat->insertmode = addv; /* in case this processor had no cache */
505: MatStashScatterBegin_Private(mat,&mat->stash,mat->rmap->range);
506: MatStashScatterBegin_Private(mat,&mat->bstash,baij->rangebs);
507: MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);
508: PetscInfo2(mat,"Stash has %D entries,uses %D mallocs.\n",nstash,reallocs);
509: MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);
510: PetscInfo2(mat,"Block-Stash has %D entries, uses %D mallocs.\n",nstash,reallocs);
511: return(0);
512: }
516: PetscErrorCode MatAssemblyEnd_MPISBAIJ(Mat mat,MatAssemblyType mode)
517: {
518: Mat_MPISBAIJ *baij=(Mat_MPISBAIJ*)mat->data;
519: Mat_SeqSBAIJ *a=(Mat_SeqSBAIJ*)baij->A->data;
521: PetscInt i,j,rstart,ncols,flg,bs2=baij->bs2;
522: PetscInt *row,*col;
523: PetscTruth other_disassembled;
524: PetscMPIInt n;
525: PetscTruth r1,r2,r3;
526: MatScalar *val;
527: InsertMode addv = mat->insertmode;
529: /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */
532: if (!baij->donotstash) {
533: while (1) {
534: MatStashScatterGetMesg_Private(&mat->stash,&n,&row,&col,&val,&flg);
535: if (!flg) break;
537: for (i=0; i<n;) {
538: /* Now identify the consecutive vals belonging to the same row */
539: for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; }
540: if (j < n) ncols = j-i;
541: else ncols = n-i;
542: /* Now assemble all these values with a single function call */
543: MatSetValues_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i,addv);
544: i = j;
545: }
546: }
547: MatStashScatterEnd_Private(&mat->stash);
548: /* Now process the block-stash. Since the values are stashed column-oriented,
549: set the roworiented flag to column oriented, and after MatSetValues()
550: restore the original flags */
551: r1 = baij->roworiented;
552: r2 = a->roworiented;
553: r3 = ((Mat_SeqBAIJ*)baij->B->data)->roworiented;
554: baij->roworiented = PETSC_FALSE;
555: a->roworiented = PETSC_FALSE;
556: ((Mat_SeqBAIJ*)baij->B->data)->roworiented = PETSC_FALSE; /* b->roworinted */
557: while (1) {
558: MatStashScatterGetMesg_Private(&mat->bstash,&n,&row,&col,&val,&flg);
559: if (!flg) break;
560:
561: for (i=0; i<n;) {
562: /* Now identify the consecutive vals belonging to the same row */
563: for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; }
564: if (j < n) ncols = j-i;
565: else ncols = n-i;
566: MatSetValuesBlocked_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i*bs2,addv);
567: i = j;
568: }
569: }
570: MatStashScatterEnd_Private(&mat->bstash);
571: baij->roworiented = r1;
572: a->roworiented = r2;
573: ((Mat_SeqBAIJ*)baij->B->data)->roworiented = r3; /* b->roworinted */
574: }
576: MatAssemblyBegin(baij->A,mode);
577: MatAssemblyEnd(baij->A,mode);
579: /* determine if any processor has disassembled, if so we must
580: also disassemble ourselfs, in order that we may reassemble. */
581: /*
582: if nonzero structure of submatrix B cannot change then we know that
583: no processor disassembled thus we can skip this stuff
584: */
585: if (!((Mat_SeqBAIJ*)baij->B->data)->nonew) {
586: MPI_Allreduce(&mat->was_assembled,&other_disassembled,1,MPI_INT,MPI_PROD,((PetscObject)mat)->comm);
587: if (mat->was_assembled && !other_disassembled) {
588: DisAssemble_MPISBAIJ(mat);
589: }
590: }
592: if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) {
593: MatSetUpMultiply_MPISBAIJ(mat); /* setup Mvctx and sMvctx */
594: }
595: ((Mat_SeqBAIJ*)baij->B->data)->compressedrow.use = PETSC_TRUE; /* b->compressedrow.use */
596: MatAssemblyBegin(baij->B,mode);
597: MatAssemblyEnd(baij->B,mode);
598:
599: PetscFree2(baij->rowvalues,baij->rowindices);
600: baij->rowvalues = 0;
602: return(0);
603: }
608: static PetscErrorCode MatView_MPISBAIJ_ASCIIorDraworSocket(Mat mat,PetscViewer viewer)
609: {
610: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
611: PetscErrorCode ierr;
612: PetscInt bs = mat->rmap->bs;
613: PetscMPIInt size = baij->size,rank = baij->rank;
614: PetscTruth iascii,isdraw;
615: PetscViewer sviewer;
616: PetscViewerFormat format;
619: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);
620: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);
621: if (iascii) {
622: PetscViewerGetFormat(viewer,&format);
623: if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
624: MatInfo info;
625: MPI_Comm_rank(((PetscObject)mat)->comm,&rank);
626: MatGetInfo(mat,MAT_LOCAL,&info);
627: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] Local rows %D nz %D nz alloced %D bs %D mem %D\n",
628: rank,mat->rmap->N,(PetscInt)info.nz_used*bs,(PetscInt)info.nz_allocated*bs,
629: mat->rmap->bs,(PetscInt)info.memory);
630: MatGetInfo(baij->A,MAT_LOCAL,&info);
631: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] on-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);
632: MatGetInfo(baij->B,MAT_LOCAL,&info);
633: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] off-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);
634: PetscViewerFlush(viewer);
635: PetscViewerASCIIPrintf(viewer,"Information on VecScatter used in matrix-vector product: \n");
636: VecScatterView(baij->Mvctx,viewer);
637: return(0);
638: } else if (format == PETSC_VIEWER_ASCII_INFO) {
639: PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);
640: return(0);
641: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
642: return(0);
643: }
644: }
646: if (isdraw) {
647: PetscDraw draw;
648: PetscTruth isnull;
649: PetscViewerDrawGetDraw(viewer,0,&draw);
650: PetscDrawIsNull(draw,&isnull); if (isnull) return(0);
651: }
653: if (size == 1) {
654: PetscObjectSetName((PetscObject)baij->A,((PetscObject)mat)->name);
655: MatView(baij->A,viewer);
656: } else {
657: /* assemble the entire matrix onto first processor. */
658: Mat A;
659: Mat_SeqSBAIJ *Aloc;
660: Mat_SeqBAIJ *Bloc;
661: PetscInt M = mat->rmap->N,N = mat->cmap->N,*ai,*aj,col,i,j,k,*rvals,mbs = baij->mbs;
662: MatScalar *a;
664: /* Should this be the same type as mat? */
665: MatCreate(((PetscObject)mat)->comm,&A);
666: if (!rank) {
667: MatSetSizes(A,M,N,M,N);
668: } else {
669: MatSetSizes(A,0,0,M,N);
670: }
671: MatSetType(A,MATMPISBAIJ);
672: MatMPISBAIJSetPreallocation(A,mat->rmap->bs,0,PETSC_NULL,0,PETSC_NULL);
673: PetscLogObjectParent(mat,A);
675: /* copy over the A part */
676: Aloc = (Mat_SeqSBAIJ*)baij->A->data;
677: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
678: PetscMalloc(bs*sizeof(PetscInt),&rvals);
680: for (i=0; i<mbs; i++) {
681: rvals[0] = bs*(baij->rstartbs + i);
682: for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; }
683: for (j=ai[i]; j<ai[i+1]; j++) {
684: col = (baij->cstartbs+aj[j])*bs;
685: for (k=0; k<bs; k++) {
686: MatSetValues_MPISBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);
687: col++; a += bs;
688: }
689: }
690: }
691: /* copy over the B part */
692: Bloc = (Mat_SeqBAIJ*)baij->B->data;
693: ai = Bloc->i; aj = Bloc->j; a = Bloc->a;
694: for (i=0; i<mbs; i++) {
695:
696: rvals[0] = bs*(baij->rstartbs + i);
697: for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; }
698: for (j=ai[i]; j<ai[i+1]; j++) {
699: col = baij->garray[aj[j]]*bs;
700: for (k=0; k<bs; k++) {
701: MatSetValues_MPIBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);
702: col++; a += bs;
703: }
704: }
705: }
706: PetscFree(rvals);
707: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
708: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
709: /*
710: Everyone has to call to draw the matrix since the graphics waits are
711: synchronized across all processors that share the PetscDraw object
712: */
713: PetscViewerGetSingleton(viewer,&sviewer);
714: if (!rank) {
715: PetscObjectSetName((PetscObject)((Mat_MPISBAIJ*)(A->data))->A,((PetscObject)mat)->name);
716: MatView(((Mat_MPISBAIJ*)(A->data))->A,sviewer);
717: }
718: PetscViewerRestoreSingleton(viewer,&sviewer);
719: MatDestroy(A);
720: }
721: return(0);
722: }
726: PetscErrorCode MatView_MPISBAIJ(Mat mat,PetscViewer viewer)
727: {
729: PetscTruth iascii,isdraw,issocket,isbinary;
732: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);
733: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);
734: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_SOCKET,&issocket);
735: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_BINARY,&isbinary);
736: if (iascii || isdraw || issocket || isbinary) {
737: MatView_MPISBAIJ_ASCIIorDraworSocket(mat,viewer);
738: } else {
739: SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported by MPISBAIJ matrices",((PetscObject)viewer)->type_name);
740: }
741: return(0);
742: }
746: PetscErrorCode MatDestroy_MPISBAIJ(Mat mat)
747: {
748: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
752: #if defined(PETSC_USE_LOG)
753: PetscLogObjectState((PetscObject)mat,"Rows=%D,Cols=%D",mat->rmap->N,mat->cmap->N);
754: #endif
755: MatStashDestroy_Private(&mat->stash);
756: MatStashDestroy_Private(&mat->bstash);
757: MatDestroy(baij->A);
758: MatDestroy(baij->B);
759: #if defined (PETSC_USE_CTABLE)
760: if (baij->colmap) {PetscTableDestroy(baij->colmap);}
761: #else
762: PetscFree(baij->colmap);
763: #endif
764: PetscFree(baij->garray);
765: if (baij->lvec) {VecDestroy(baij->lvec);}
766: if (baij->Mvctx) {VecScatterDestroy(baij->Mvctx);}
767: if (baij->slvec0) {
768: VecDestroy(baij->slvec0);
769: VecDestroy(baij->slvec0b);
770: }
771: if (baij->slvec1) {
772: VecDestroy(baij->slvec1);
773: VecDestroy(baij->slvec1a);
774: VecDestroy(baij->slvec1b);
775: }
776: if (baij->sMvctx) {VecScatterDestroy(baij->sMvctx);}
777: PetscFree2(baij->rowvalues,baij->rowindices);
778: PetscFree(baij->barray);
779: PetscFree(baij->hd);
780: if (baij->diag) {VecDestroy(baij->diag);}
781: if (baij->bb1) {VecDestroy(baij->bb1);}
782: if (baij->xx1) {VecDestroy(baij->xx1);}
783: #if defined(PETSC_USE_SCALAR_MAT_SINGLE)
784: PetscFree(baij->setvaluescopy);
785: #endif
786: PetscFree(baij->in_loc);
787: PetscFree(baij->v_loc);
788: PetscFree(baij->rangebs);
789: PetscFree(baij);
791: PetscObjectChangeTypeName((PetscObject)mat,0);
792: PetscObjectComposeFunction((PetscObject)mat,"MatStoreValues_C","",PETSC_NULL);
793: PetscObjectComposeFunction((PetscObject)mat,"MatRetrieveValues_C","",PETSC_NULL);
794: PetscObjectComposeFunction((PetscObject)mat,"MatGetDiagonalBlock_C","",PETSC_NULL);
795: PetscObjectComposeFunction((PetscObject)mat,"MatMPISBAIJSetPreallocation_C","",PETSC_NULL);
796: return(0);
797: }
801: PetscErrorCode MatMult_MPISBAIJ_Hermitian(Mat A,Vec xx,Vec yy)
802: {
803: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
805: PetscInt nt,mbs=a->mbs,bs=A->rmap->bs;
806: PetscScalar *x,*from;
807:
809: VecGetLocalSize(xx,&nt);
810: if (nt != A->cmap->n) {
811: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
812: }
814: /* diagonal part */
815: (*a->A->ops->mult)(a->A,xx,a->slvec1a);
816: VecSet(a->slvec1b,0.0);
818: /* subdiagonal part */
819: (*a->B->ops->multhermitiantranspose)(a->B,xx,a->slvec0b);
821: /* copy x into the vec slvec0 */
822: VecGetArray(a->slvec0,&from);
823: VecGetArray(xx,&x);
825: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
826: VecRestoreArray(a->slvec0,&from);
827: VecRestoreArray(xx,&x);
828:
829: VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
830: VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
831: /* supperdiagonal part */
832: (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);
833: return(0);
834: }
838: PetscErrorCode MatMult_MPISBAIJ(Mat A,Vec xx,Vec yy)
839: {
840: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
842: PetscInt nt,mbs=a->mbs,bs=A->rmap->bs;
843: PetscScalar *x,*from;
844:
846: VecGetLocalSize(xx,&nt);
847: if (nt != A->cmap->n) {
848: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
849: }
851: /* diagonal part */
852: (*a->A->ops->mult)(a->A,xx,a->slvec1a);
853: VecSet(a->slvec1b,0.0);
855: /* subdiagonal part */
856: (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);
858: /* copy x into the vec slvec0 */
859: VecGetArray(a->slvec0,&from);
860: VecGetArray(xx,&x);
862: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
863: VecRestoreArray(a->slvec0,&from);
864: VecRestoreArray(xx,&x);
865:
866: VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
867: VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
868: /* supperdiagonal part */
869: (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);
870: return(0);
871: }
875: PetscErrorCode MatMult_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy)
876: {
877: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
879: PetscInt nt;
882: VecGetLocalSize(xx,&nt);
883: if (nt != A->cmap->n) {
884: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
885: }
886: VecGetLocalSize(yy,&nt);
887: if (nt != A->rmap->N) {
888: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible parition of A and yy");
889: }
891: VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
892: /* do diagonal part */
893: (*a->A->ops->mult)(a->A,xx,yy);
894: /* do supperdiagonal part */
895: VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
896: (*a->B->ops->multadd)(a->B,a->lvec,yy,yy);
897: /* do subdiagonal part */
898: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
899: VecScatterBegin(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
900: VecScatterEnd(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
902: return(0);
903: }
907: PetscErrorCode MatMultAdd_MPISBAIJ(Mat A,Vec xx,Vec yy,Vec zz)
908: {
909: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
911: PetscInt mbs=a->mbs,bs=A->rmap->bs;
912: PetscScalar *x,*from,zero=0.0;
913:
915: /*
916: PetscSynchronizedPrintf(((PetscObject)A)->comm," MatMultAdd is called ...\n");
917: PetscSynchronizedFlush(((PetscObject)A)->comm);
918: */
919: /* diagonal part */
920: (*a->A->ops->multadd)(a->A,xx,yy,a->slvec1a);
921: VecSet(a->slvec1b,zero);
923: /* subdiagonal part */
924: (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);
926: /* copy x into the vec slvec0 */
927: VecGetArray(a->slvec0,&from);
928: VecGetArray(xx,&x);
929: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
930: VecRestoreArray(a->slvec0,&from);
931:
932: VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
933: VecRestoreArray(xx,&x);
934: VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);
935:
936: /* supperdiagonal part */
937: (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,zz);
938:
939: return(0);
940: }
944: PetscErrorCode MatMultAdd_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy,Vec zz)
945: {
946: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
950: VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
951: /* do diagonal part */
952: (*a->A->ops->multadd)(a->A,xx,yy,zz);
953: /* do supperdiagonal part */
954: VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
955: (*a->B->ops->multadd)(a->B,a->lvec,zz,zz);
957: /* do subdiagonal part */
958: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
959: VecScatterBegin(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);
960: VecScatterEnd(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);
962: return(0);
963: }
965: /*
966: This only works correctly for square matrices where the subblock A->A is the
967: diagonal block
968: */
971: PetscErrorCode MatGetDiagonal_MPISBAIJ(Mat A,Vec v)
972: {
973: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
977: /* if (a->rmap->N != a->cmap->N) SETERRQ(PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block"); */
978: MatGetDiagonal(a->A,v);
979: return(0);
980: }
984: PetscErrorCode MatScale_MPISBAIJ(Mat A,PetscScalar aa)
985: {
986: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
990: MatScale(a->A,aa);
991: MatScale(a->B,aa);
992: return(0);
993: }
997: PetscErrorCode MatGetRow_MPISBAIJ(Mat matin,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
998: {
999: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
1000: PetscScalar *vworkA,*vworkB,**pvA,**pvB,*v_p;
1002: PetscInt bs = matin->rmap->bs,bs2 = mat->bs2,i,*cworkA,*cworkB,**pcA,**pcB;
1003: PetscInt nztot,nzA,nzB,lrow,brstart = matin->rmap->rstart,brend = matin->rmap->rend;
1004: PetscInt *cmap,*idx_p,cstart = mat->rstartbs;
1007: if (mat->getrowactive) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Already active");
1008: mat->getrowactive = PETSC_TRUE;
1010: if (!mat->rowvalues && (idx || v)) {
1011: /*
1012: allocate enough space to hold information from the longest row.
1013: */
1014: Mat_SeqSBAIJ *Aa = (Mat_SeqSBAIJ*)mat->A->data;
1015: Mat_SeqBAIJ *Ba = (Mat_SeqBAIJ*)mat->B->data;
1016: PetscInt max = 1,mbs = mat->mbs,tmp;
1017: for (i=0; i<mbs; i++) {
1018: tmp = Aa->i[i+1] - Aa->i[i] + Ba->i[i+1] - Ba->i[i]; /* row length */
1019: if (max < tmp) { max = tmp; }
1020: }
1021: PetscMalloc2(max*bs2,PetscScalar,&mat->rowvalues,max*bs2,PetscInt,&mat->rowindices);
1022: }
1023:
1024: if (row < brstart || row >= brend) SETERRQ(PETSC_ERR_SUP,"Only local rows")
1025: lrow = row - brstart; /* local row index */
1027: pvA = &vworkA; pcA = &cworkA; pvB = &vworkB; pcB = &cworkB;
1028: if (!v) {pvA = 0; pvB = 0;}
1029: if (!idx) {pcA = 0; if (!v) pcB = 0;}
1030: (*mat->A->ops->getrow)(mat->A,lrow,&nzA,pcA,pvA);
1031: (*mat->B->ops->getrow)(mat->B,lrow,&nzB,pcB,pvB);
1032: nztot = nzA + nzB;
1034: cmap = mat->garray;
1035: if (v || idx) {
1036: if (nztot) {
1037: /* Sort by increasing column numbers, assuming A and B already sorted */
1038: PetscInt imark = -1;
1039: if (v) {
1040: *v = v_p = mat->rowvalues;
1041: for (i=0; i<nzB; i++) {
1042: if (cmap[cworkB[i]/bs] < cstart) v_p[i] = vworkB[i];
1043: else break;
1044: }
1045: imark = i;
1046: for (i=0; i<nzA; i++) v_p[imark+i] = vworkA[i];
1047: for (i=imark; i<nzB; i++) v_p[nzA+i] = vworkB[i];
1048: }
1049: if (idx) {
1050: *idx = idx_p = mat->rowindices;
1051: if (imark > -1) {
1052: for (i=0; i<imark; i++) {
1053: idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs;
1054: }
1055: } else {
1056: for (i=0; i<nzB; i++) {
1057: if (cmap[cworkB[i]/bs] < cstart)
1058: idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ;
1059: else break;
1060: }
1061: imark = i;
1062: }
1063: for (i=0; i<nzA; i++) idx_p[imark+i] = cstart*bs + cworkA[i];
1064: for (i=imark; i<nzB; i++) idx_p[nzA+i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ;
1065: }
1066: } else {
1067: if (idx) *idx = 0;
1068: if (v) *v = 0;
1069: }
1070: }
1071: *nz = nztot;
1072: (*mat->A->ops->restorerow)(mat->A,lrow,&nzA,pcA,pvA);
1073: (*mat->B->ops->restorerow)(mat->B,lrow,&nzB,pcB,pvB);
1074: return(0);
1075: }
1079: PetscErrorCode MatRestoreRow_MPISBAIJ(Mat mat,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1080: {
1081: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
1084: if (!baij->getrowactive) {
1085: SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"MatGetRow() must be called first");
1086: }
1087: baij->getrowactive = PETSC_FALSE;
1088: return(0);
1089: }
1093: PetscErrorCode MatGetRowUpperTriangular_MPISBAIJ(Mat A)
1094: {
1095: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1096: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1099: aA->getrow_utriangular = PETSC_TRUE;
1100: return(0);
1101: }
1104: PetscErrorCode MatRestoreRowUpperTriangular_MPISBAIJ(Mat A)
1105: {
1106: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1107: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1110: aA->getrow_utriangular = PETSC_FALSE;
1111: return(0);
1112: }
1116: PetscErrorCode MatRealPart_MPISBAIJ(Mat A)
1117: {
1118: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1122: MatRealPart(a->A);
1123: MatRealPart(a->B);
1124: return(0);
1125: }
1129: PetscErrorCode MatImaginaryPart_MPISBAIJ(Mat A)
1130: {
1131: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1135: MatImaginaryPart(a->A);
1136: MatImaginaryPart(a->B);
1137: return(0);
1138: }
1142: PetscErrorCode MatZeroEntries_MPISBAIJ(Mat A)
1143: {
1144: Mat_MPISBAIJ *l = (Mat_MPISBAIJ*)A->data;
1148: MatZeroEntries(l->A);
1149: MatZeroEntries(l->B);
1150: return(0);
1151: }
1155: PetscErrorCode MatGetInfo_MPISBAIJ(Mat matin,MatInfoType flag,MatInfo *info)
1156: {
1157: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)matin->data;
1158: Mat A = a->A,B = a->B;
1160: PetscReal isend[5],irecv[5];
1163: info->block_size = (PetscReal)matin->rmap->bs;
1164: MatGetInfo(A,MAT_LOCAL,info);
1165: isend[0] = info->nz_used; isend[1] = info->nz_allocated; isend[2] = info->nz_unneeded;
1166: isend[3] = info->memory; isend[4] = info->mallocs;
1167: MatGetInfo(B,MAT_LOCAL,info);
1168: isend[0] += info->nz_used; isend[1] += info->nz_allocated; isend[2] += info->nz_unneeded;
1169: isend[3] += info->memory; isend[4] += info->mallocs;
1170: if (flag == MAT_LOCAL) {
1171: info->nz_used = isend[0];
1172: info->nz_allocated = isend[1];
1173: info->nz_unneeded = isend[2];
1174: info->memory = isend[3];
1175: info->mallocs = isend[4];
1176: } else if (flag == MAT_GLOBAL_MAX) {
1177: MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_MAX,((PetscObject)matin)->comm);
1178: info->nz_used = irecv[0];
1179: info->nz_allocated = irecv[1];
1180: info->nz_unneeded = irecv[2];
1181: info->memory = irecv[3];
1182: info->mallocs = irecv[4];
1183: } else if (flag == MAT_GLOBAL_SUM) {
1184: MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_SUM,((PetscObject)matin)->comm);
1185: info->nz_used = irecv[0];
1186: info->nz_allocated = irecv[1];
1187: info->nz_unneeded = irecv[2];
1188: info->memory = irecv[3];
1189: info->mallocs = irecv[4];
1190: } else {
1191: SETERRQ1(PETSC_ERR_ARG_WRONG,"Unknown MatInfoType argument %d",(int)flag);
1192: }
1193: info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */
1194: info->fill_ratio_needed = 0;
1195: info->factor_mallocs = 0;
1196: return(0);
1197: }
1201: PetscErrorCode MatSetOption_MPISBAIJ(Mat A,MatOption op,PetscTruth flg)
1202: {
1203: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1204: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1208: switch (op) {
1209: case MAT_NEW_NONZERO_LOCATIONS:
1210: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1211: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1212: case MAT_KEEP_NONZERO_PATTERN:
1213: case MAT_NEW_NONZERO_LOCATION_ERR:
1214: MatSetOption(a->A,op,flg);
1215: MatSetOption(a->B,op,flg);
1216: break;
1217: case MAT_ROW_ORIENTED:
1218: a->roworiented = flg;
1219: MatSetOption(a->A,op,flg);
1220: MatSetOption(a->B,op,flg);
1221: break;
1222: case MAT_NEW_DIAGONALS:
1223: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
1224: break;
1225: case MAT_IGNORE_OFF_PROC_ENTRIES:
1226: a->donotstash = flg;
1227: break;
1228: case MAT_USE_HASH_TABLE:
1229: a->ht_flag = flg;
1230: break;
1231: case MAT_HERMITIAN:
1232: if (!A->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Must call MatAssemblyEnd() first");
1233: MatSetOption(a->A,op,flg);
1234: A->ops->mult = MatMult_MPISBAIJ_Hermitian;
1235: break;
1236: case MAT_SYMMETRIC:
1237: MatSetOption(a->A,op,flg);
1238: break;
1239: case MAT_STRUCTURALLY_SYMMETRIC:
1240: MatSetOption(a->A,op,flg);
1241: break;
1242: case MAT_SYMMETRY_ETERNAL:
1243: if (!flg) SETERRQ(PETSC_ERR_SUP,"Matrix must be symmetric");
1244: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
1245: break;
1246: case MAT_IGNORE_LOWER_TRIANGULAR:
1247: aA->ignore_ltriangular = flg;
1248: break;
1249: case MAT_ERROR_LOWER_TRIANGULAR:
1250: aA->ignore_ltriangular = flg;
1251: break;
1252: case MAT_GETROW_UPPERTRIANGULAR:
1253: aA->getrow_utriangular = flg;
1254: break;
1255: default:
1256: SETERRQ1(PETSC_ERR_SUP,"unknown option %d",op);
1257: }
1258: return(0);
1259: }
1263: PetscErrorCode MatTranspose_MPISBAIJ(Mat A,MatReuse reuse,Mat *B)
1264: {
1267: if (MAT_INITIAL_MATRIX || *B != A) {
1268: MatDuplicate(A,MAT_COPY_VALUES,B);
1269: }
1270: return(0);
1271: }
1275: PetscErrorCode MatDiagonalScale_MPISBAIJ(Mat mat,Vec ll,Vec rr)
1276: {
1277: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
1278: Mat a=baij->A, b=baij->B;
1280: PetscInt nv,m,n;
1281: PetscTruth flg;
1284: if (ll != rr){
1285: VecEqual(ll,rr,&flg);
1286: if (!flg)
1287: SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"For symmetric format, left and right scaling vectors must be same\n");
1288: }
1289: if (!ll) return(0);
1291: MatGetLocalSize(mat,&m,&n);
1292: if (m != n) SETERRQ2(PETSC_ERR_ARG_SIZ,"For symmetric format, local size %d %d must be same",m,n);
1293:
1294: VecGetLocalSize(rr,&nv);
1295: if (nv!=n) SETERRQ(PETSC_ERR_ARG_SIZ,"Left and right vector non-conforming local size");
1297: VecScatterBegin(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);
1298:
1299: /* left diagonalscale the off-diagonal part */
1300: (*b->ops->diagonalscale)(b,ll,PETSC_NULL);
1301:
1302: /* scale the diagonal part */
1303: (*a->ops->diagonalscale)(a,ll,rr);
1305: /* right diagonalscale the off-diagonal part */
1306: VecScatterEnd(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);
1307: (*b->ops->diagonalscale)(b,PETSC_NULL,baij->lvec);
1308: return(0);
1309: }
1313: PetscErrorCode MatSetUnfactored_MPISBAIJ(Mat A)
1314: {
1315: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1319: MatSetUnfactored(a->A);
1320: return(0);
1321: }
1323: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat,MatDuplicateOption,Mat *);
1327: PetscErrorCode MatEqual_MPISBAIJ(Mat A,Mat B,PetscTruth *flag)
1328: {
1329: Mat_MPISBAIJ *matB = (Mat_MPISBAIJ*)B->data,*matA = (Mat_MPISBAIJ*)A->data;
1330: Mat a,b,c,d;
1331: PetscTruth flg;
1335: a = matA->A; b = matA->B;
1336: c = matB->A; d = matB->B;
1338: MatEqual(a,c,&flg);
1339: if (flg) {
1340: MatEqual(b,d,&flg);
1341: }
1342: MPI_Allreduce(&flg,flag,1,MPI_INT,MPI_LAND,((PetscObject)A)->comm);
1343: return(0);
1344: }
1348: PetscErrorCode MatCopy_MPISBAIJ(Mat A,Mat B,MatStructure str)
1349: {
1351: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1352: Mat_MPISBAIJ *b = (Mat_MPISBAIJ *)B->data;
1355: /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */
1356: if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) {
1357: MatGetRowUpperTriangular(A);
1358: MatCopy_Basic(A,B,str);
1359: MatRestoreRowUpperTriangular(A);
1360: } else {
1361: MatCopy(a->A,b->A,str);
1362: MatCopy(a->B,b->B,str);
1363: }
1364: return(0);
1365: }
1369: PetscErrorCode MatSetUpPreallocation_MPISBAIJ(Mat A)
1370: {
1374: MatMPISBAIJSetPreallocation(A,-PetscMax(A->rmap->bs,1),PETSC_DEFAULT,0,PETSC_DEFAULT,0);
1375: return(0);
1376: }
1380: PetscErrorCode MatAXPY_MPISBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
1381: {
1383: Mat_MPISBAIJ *xx=(Mat_MPISBAIJ *)X->data,*yy=(Mat_MPISBAIJ *)Y->data;
1384: PetscBLASInt bnz,one=1;
1385: Mat_SeqSBAIJ *xa,*ya;
1386: Mat_SeqBAIJ *xb,*yb;
1389: if (str == SAME_NONZERO_PATTERN) {
1390: PetscScalar alpha = a;
1391: xa = (Mat_SeqSBAIJ *)xx->A->data;
1392: ya = (Mat_SeqSBAIJ *)yy->A->data;
1393: bnz = PetscBLASIntCast(xa->nz);
1394: BLASaxpy_(&bnz,&alpha,xa->a,&one,ya->a,&one);
1395: xb = (Mat_SeqBAIJ *)xx->B->data;
1396: yb = (Mat_SeqBAIJ *)yy->B->data;
1397: bnz = PetscBLASIntCast(xb->nz);
1398: BLASaxpy_(&bnz,&alpha,xb->a,&one,yb->a,&one);
1399: } else {
1400: MatGetRowUpperTriangular(X);
1401: MatAXPY_Basic(Y,a,X,str);
1402: MatRestoreRowUpperTriangular(X);
1403: }
1404: return(0);
1405: }
1409: PetscErrorCode MatSetBlockSize_MPISBAIJ(Mat A,PetscInt bs)
1410: {
1411: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1412: PetscInt rbs,cbs;
1413: PetscErrorCode ierr;
1416: MatSetBlockSize(a->A,bs);
1417: MatSetBlockSize(a->B,bs);
1418: PetscLayoutGetBlockSize(A->rmap,&rbs);
1419: PetscLayoutGetBlockSize(A->cmap,&cbs);
1420: if (rbs != bs) SETERRQ2(PETSC_ERR_ARG_SIZ,"Attempt to set block size %d with SBAIJ %d",bs,rbs);
1421: if (cbs != bs) SETERRQ2(PETSC_ERR_ARG_SIZ,"Attempt to set block size %d with SBAIJ %d",bs,cbs);
1422: return(0);
1423: }
1427: PetscErrorCode MatGetSubMatrices_MPISBAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[])
1428: {
1430: PetscInt i;
1431: PetscTruth flg;
1434: for (i=0; i<n; i++) {
1435: ISEqual(irow[i],icol[i],&flg);
1436: if (!flg) {
1437: SETERRQ(PETSC_ERR_SUP,"Can only get symmetric submatrix for MPISBAIJ matrices");
1438: }
1439: }
1440: MatGetSubMatrices_MPIBAIJ(A,n,irow,icol,scall,B);
1441: return(0);
1442: }
1443:
1445: /* -------------------------------------------------------------------*/
1446: static struct _MatOps MatOps_Values = {
1447: MatSetValues_MPISBAIJ,
1448: MatGetRow_MPISBAIJ,
1449: MatRestoreRow_MPISBAIJ,
1450: MatMult_MPISBAIJ,
1451: /* 4*/ MatMultAdd_MPISBAIJ,
1452: MatMult_MPISBAIJ, /* transpose versions are same as non-transpose */
1453: MatMultAdd_MPISBAIJ,
1454: 0,
1455: 0,
1456: 0,
1457: /*10*/ 0,
1458: 0,
1459: 0,
1460: MatSOR_MPISBAIJ,
1461: MatTranspose_MPISBAIJ,
1462: /*15*/ MatGetInfo_MPISBAIJ,
1463: MatEqual_MPISBAIJ,
1464: MatGetDiagonal_MPISBAIJ,
1465: MatDiagonalScale_MPISBAIJ,
1466: MatNorm_MPISBAIJ,
1467: /*20*/ MatAssemblyBegin_MPISBAIJ,
1468: MatAssemblyEnd_MPISBAIJ,
1469: MatSetOption_MPISBAIJ,
1470: MatZeroEntries_MPISBAIJ,
1471: /*24*/ 0,
1472: 0,
1473: 0,
1474: 0,
1475: 0,
1476: /*29*/ MatSetUpPreallocation_MPISBAIJ,
1477: 0,
1478: 0,
1479: 0,
1480: 0,
1481: /*34*/ MatDuplicate_MPISBAIJ,
1482: 0,
1483: 0,
1484: 0,
1485: 0,
1486: /*39*/ MatAXPY_MPISBAIJ,
1487: MatGetSubMatrices_MPISBAIJ,
1488: MatIncreaseOverlap_MPISBAIJ,
1489: MatGetValues_MPISBAIJ,
1490: MatCopy_MPISBAIJ,
1491: /*44*/ 0,
1492: MatScale_MPISBAIJ,
1493: 0,
1494: 0,
1495: 0,
1496: /*49*/ MatSetBlockSize_MPISBAIJ,
1497: 0,
1498: 0,
1499: 0,
1500: 0,
1501: /*54*/ 0,
1502: 0,
1503: MatSetUnfactored_MPISBAIJ,
1504: 0,
1505: MatSetValuesBlocked_MPISBAIJ,
1506: /*59*/ 0,
1507: 0,
1508: 0,
1509: 0,
1510: 0,
1511: /*64*/ 0,
1512: 0,
1513: 0,
1514: 0,
1515: 0,
1516: /*69*/ MatGetRowMaxAbs_MPISBAIJ,
1517: 0,
1518: 0,
1519: 0,
1520: 0,
1521: /*74*/ 0,
1522: 0,
1523: 0,
1524: 0,
1525: 0,
1526: /*79*/ 0,
1527: 0,
1528: 0,
1529: 0,
1530: MatLoad_MPISBAIJ,
1531: /*84*/ 0,
1532: 0,
1533: 0,
1534: 0,
1535: 0,
1536: /*89*/ 0,
1537: 0,
1538: 0,
1539: 0,
1540: 0,
1541: /*94*/ 0,
1542: 0,
1543: 0,
1544: 0,
1545: 0,
1546: /*99*/ 0,
1547: 0,
1548: 0,
1549: 0,
1550: 0,
1551: /*104*/0,
1552: MatRealPart_MPISBAIJ,
1553: MatImaginaryPart_MPISBAIJ,
1554: MatGetRowUpperTriangular_MPISBAIJ,
1555: MatRestoreRowUpperTriangular_MPISBAIJ
1556: };
1562: PetscErrorCode MatGetDiagonalBlock_MPISBAIJ(Mat A,PetscTruth *iscopy,MatReuse reuse,Mat *a)
1563: {
1565: *a = ((Mat_MPISBAIJ *)A->data)->A;
1566: *iscopy = PETSC_FALSE;
1567: return(0);
1568: }
1574: PetscErrorCode MatMPISBAIJSetPreallocation_MPISBAIJ(Mat B,PetscInt bs,PetscInt d_nz,PetscInt *d_nnz,PetscInt o_nz,PetscInt *o_nnz)
1575: {
1576: Mat_MPISBAIJ *b;
1578: PetscInt i,mbs,Mbs,newbs = PetscAbs(bs);
1581: if (bs < 0){
1582: PetscOptionsBegin(((PetscObject)B)->comm,((PetscObject)B)->prefix,"Options for MPISBAIJ matrix","Mat");
1583: PetscOptionsInt("-mat_block_size","Set the blocksize used to store the matrix","MatMPIBAIJSetPreallocation",newbs,&newbs,PETSC_NULL);
1584: PetscOptionsEnd();
1585: bs = PetscAbs(bs);
1586: }
1587: if ((d_nnz || o_nnz) && newbs != bs) {
1588: SETERRQ(PETSC_ERR_ARG_WRONG,"Cannot change blocksize from command line if setting d_nnz or o_nnz");
1589: }
1590: bs = newbs;
1592: if (d_nz == PETSC_DECIDE || d_nz == PETSC_DEFAULT) d_nz = 3;
1593: if (o_nz == PETSC_DECIDE || o_nz == PETSC_DEFAULT) o_nz = 1;
1594: if (d_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"d_nz cannot be less than 0: value %D",d_nz);
1595: if (o_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"o_nz cannot be less than 0: value %D",o_nz);
1597: B->rmap->bs = B->cmap->bs = bs;
1598: PetscLayoutSetUp(B->rmap);
1599: PetscLayoutSetUp(B->cmap);
1601: if (d_nnz) {
1602: for (i=0; i<B->rmap->n/bs; i++) {
1603: if (d_nnz[i] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"d_nnz cannot be less than -1: local row %D value %D",i,d_nnz[i]);
1604: }
1605: }
1606: if (o_nnz) {
1607: for (i=0; i<B->rmap->n/bs; i++) {
1608: if (o_nnz[i] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"o_nnz cannot be less than -1: local row %D value %D",i,o_nnz[i]);
1609: }
1610: }
1612: b = (Mat_MPISBAIJ*)B->data;
1613: mbs = B->rmap->n/bs;
1614: Mbs = B->rmap->N/bs;
1615: if (mbs*bs != B->rmap->n) {
1616: SETERRQ2(PETSC_ERR_ARG_SIZ,"No of local rows %D must be divisible by blocksize %D",B->rmap->N,bs);
1617: }
1619: B->rmap->bs = bs;
1620: b->bs2 = bs*bs;
1621: b->mbs = mbs;
1622: b->nbs = mbs;
1623: b->Mbs = Mbs;
1624: b->Nbs = Mbs;
1626: for (i=0; i<=b->size; i++) {
1627: b->rangebs[i] = B->rmap->range[i]/bs;
1628: }
1629: b->rstartbs = B->rmap->rstart/bs;
1630: b->rendbs = B->rmap->rend/bs;
1631:
1632: b->cstartbs = B->cmap->rstart/bs;
1633: b->cendbs = B->cmap->rend/bs;
1635: if (!B->preallocated) {
1636: MatCreate(PETSC_COMM_SELF,&b->A);
1637: MatSetSizes(b->A,B->rmap->n,B->cmap->n,B->rmap->n,B->cmap->n);
1638: MatSetType(b->A,MATSEQSBAIJ);
1639: PetscLogObjectParent(B,b->A);
1640: MatCreate(PETSC_COMM_SELF,&b->B);
1641: MatSetSizes(b->B,B->rmap->n,B->cmap->N,B->rmap->n,B->cmap->N);
1642: MatSetType(b->B,MATSEQBAIJ);
1643: PetscLogObjectParent(B,b->B);
1644: MatStashCreate_Private(((PetscObject)B)->comm,bs,&B->bstash);
1645: }
1647: MatSeqSBAIJSetPreallocation(b->A,bs,d_nz,d_nnz);
1648: MatSeqBAIJSetPreallocation(b->B,bs,o_nz,o_nnz);
1649: B->preallocated = PETSC_TRUE;
1650: return(0);
1651: }
1655: #if defined(PETSC_HAVE_MUMPS)
1657: #endif
1658: #if defined(PETSC_HAVE_SPOOLES)
1660: #endif
1661: #if defined(PETSC_HAVE_PASTIX)
1663: #endif
1666: /*MC
1667: MATMPISBAIJ - MATMPISBAIJ = "mpisbaij" - A matrix type to be used for distributed symmetric sparse block matrices,
1668: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1670: Options Database Keys:
1671: . -mat_type mpisbaij - sets the matrix type to "mpisbaij" during a call to MatSetFromOptions()
1673: Level: beginner
1675: .seealso: MatCreateMPISBAIJ
1676: M*/
1681: PetscErrorCode MatCreate_MPISBAIJ(Mat B)
1682: {
1683: Mat_MPISBAIJ *b;
1685: PetscTruth flg;
1689: PetscNewLog(B,Mat_MPISBAIJ,&b);
1690: B->data = (void*)b;
1691: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
1693: B->ops->destroy = MatDestroy_MPISBAIJ;
1694: B->ops->view = MatView_MPISBAIJ;
1695: B->mapping = 0;
1696: B->assembled = PETSC_FALSE;
1698: B->insertmode = NOT_SET_VALUES;
1699: MPI_Comm_rank(((PetscObject)B)->comm,&b->rank);
1700: MPI_Comm_size(((PetscObject)B)->comm,&b->size);
1702: /* build local table of row and column ownerships */
1703: PetscMalloc((b->size+2)*sizeof(PetscInt),&b->rangebs);
1705: /* build cache for off array entries formed */
1706: MatStashCreate_Private(((PetscObject)B)->comm,1,&B->stash);
1707: b->donotstash = PETSC_FALSE;
1708: b->colmap = PETSC_NULL;
1709: b->garray = PETSC_NULL;
1710: b->roworiented = PETSC_TRUE;
1712: /* stuff used in block assembly */
1713: b->barray = 0;
1715: /* stuff used for matrix vector multiply */
1716: b->lvec = 0;
1717: b->Mvctx = 0;
1718: b->slvec0 = 0;
1719: b->slvec0b = 0;
1720: b->slvec1 = 0;
1721: b->slvec1a = 0;
1722: b->slvec1b = 0;
1723: b->sMvctx = 0;
1725: /* stuff for MatGetRow() */
1726: b->rowindices = 0;
1727: b->rowvalues = 0;
1728: b->getrowactive = PETSC_FALSE;
1730: /* hash table stuff */
1731: b->ht = 0;
1732: b->hd = 0;
1733: b->ht_size = 0;
1734: b->ht_flag = PETSC_FALSE;
1735: b->ht_fact = 0;
1736: b->ht_total_ct = 0;
1737: b->ht_insert_ct = 0;
1739: b->in_loc = 0;
1740: b->v_loc = 0;
1741: b->n_loc = 0;
1742: PetscOptionsBegin(((PetscObject)B)->comm,PETSC_NULL,"Options for loading MPISBAIJ matrix 1","Mat");
1743: PetscOptionsTruth("-mat_use_hash_table","Use hash table to save memory in constructing matrix","MatSetOption",PETSC_FALSE,&flg,PETSC_NULL);
1744: if (flg) {
1745: PetscReal fact = 1.39;
1746: MatSetOption(B,MAT_USE_HASH_TABLE,PETSC_TRUE);
1747: PetscOptionsReal("-mat_use_hash_table","Use hash table factor","MatMPIBAIJSetHashTableFactor",fact,&fact,PETSC_NULL);
1748: if (fact <= 1.0) fact = 1.39;
1749: MatMPIBAIJSetHashTableFactor(B,fact);
1750: PetscInfo1(B,"Hash table Factor used %5.2f\n",fact);
1751: }
1752: PetscOptionsEnd();
1754: #if defined(PETSC_HAVE_PASTIX)
1755: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetFactor_pastix_C",
1756: "MatGetFactor_mpisbaij_pastix",
1757: MatGetFactor_mpisbaij_pastix);
1758: #endif
1759: #if defined(PETSC_HAVE_MUMPS)
1760: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetFactor_mumps_C",
1761: "MatGetFactor_mpisbaij_mumps",
1762: MatGetFactor_mpisbaij_mumps);
1763: #endif
1764: #if defined(PETSC_HAVE_SPOOLES)
1765: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetFactor_spooles_C",
1766: "MatGetFactor_mpisbaij_spooles",
1767: MatGetFactor_mpisbaij_spooles);
1768: #endif
1769: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatStoreValues_C",
1770: "MatStoreValues_MPISBAIJ",
1771: MatStoreValues_MPISBAIJ);
1772: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatRetrieveValues_C",
1773: "MatRetrieveValues_MPISBAIJ",
1774: MatRetrieveValues_MPISBAIJ);
1775: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetDiagonalBlock_C",
1776: "MatGetDiagonalBlock_MPISBAIJ",
1777: MatGetDiagonalBlock_MPISBAIJ);
1778: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatMPISBAIJSetPreallocation_C",
1779: "MatMPISBAIJSetPreallocation_MPISBAIJ",
1780: MatMPISBAIJSetPreallocation_MPISBAIJ);
1781: B->symmetric = PETSC_TRUE;
1782: B->structurally_symmetric = PETSC_TRUE;
1783: B->symmetric_set = PETSC_TRUE;
1784: B->structurally_symmetric_set = PETSC_TRUE;
1785: PetscObjectChangeTypeName((PetscObject)B,MATMPISBAIJ);
1786: return(0);
1787: }
1790: /*MC
1791: MATSBAIJ - MATSBAIJ = "sbaij" - A matrix type to be used for symmetric block sparse matrices.
1793: This matrix type is identical to MATSEQSBAIJ when constructed with a single process communicator,
1794: and MATMPISBAIJ otherwise.
1796: Options Database Keys:
1797: . -mat_type sbaij - sets the matrix type to "sbaij" during a call to MatSetFromOptions()
1799: Level: beginner
1801: .seealso: MatCreateMPISBAIJ,MATSEQSBAIJ,MATMPISBAIJ
1802: M*/
1807: PetscErrorCode MatCreate_SBAIJ(Mat A)
1808: {
1810: PetscMPIInt size;
1813: MPI_Comm_size(((PetscObject)A)->comm,&size);
1814: if (size == 1) {
1815: MatSetType(A,MATSEQSBAIJ);
1816: } else {
1817: MatSetType(A,MATMPISBAIJ);
1818: }
1819: return(0);
1820: }
1825: /*@C
1826: MatMPISBAIJSetPreallocation - For good matrix assembly performance
1827: the user should preallocate the matrix storage by setting the parameters
1828: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
1829: performance can be increased by more than a factor of 50.
1831: Collective on Mat
1833: Input Parameters:
1834: + A - the matrix
1835: . bs - size of blockk
1836: . d_nz - number of block nonzeros per block row in diagonal portion of local
1837: submatrix (same for all local rows)
1838: . d_nnz - array containing the number of block nonzeros in the various block rows
1839: in the upper triangular and diagonal part of the in diagonal portion of the local
1840: (possibly different for each block row) or PETSC_NULL. You must leave room
1841: for the diagonal entry even if it is zero.
1842: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
1843: submatrix (same for all local rows).
1844: - o_nnz - array containing the number of nonzeros in the various block rows of the
1845: off-diagonal portion of the local submatrix (possibly different for
1846: each block row) or PETSC_NULL.
1849: Options Database Keys:
1850: . -mat_no_unroll - uses code that does not unroll the loops in the
1851: block calculations (much slower)
1852: . -mat_block_size - size of the blocks to use
1854: Notes:
1856: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
1857: than it must be used on all processors that share the object for that argument.
1859: If the *_nnz parameter is given then the *_nz parameter is ignored
1861: Storage Information:
1862: For a square global matrix we define each processor's diagonal portion
1863: to be its local rows and the corresponding columns (a square submatrix);
1864: each processor's off-diagonal portion encompasses the remainder of the
1865: local matrix (a rectangular submatrix).
1867: The user can specify preallocated storage for the diagonal part of
1868: the local submatrix with either d_nz or d_nnz (not both). Set
1869: d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic
1870: memory allocation. Likewise, specify preallocated storage for the
1871: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
1873: You can call MatGetInfo() to get information on how effective the preallocation was;
1874: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
1875: You can also run with the option -info and look for messages with the string
1876: malloc in them to see if additional memory allocation was needed.
1878: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
1879: the figure below we depict these three local rows and all columns (0-11).
1881: .vb
1882: 0 1 2 3 4 5 6 7 8 9 10 11
1883: -------------------
1884: row 3 | o o o d d d o o o o o o
1885: row 4 | o o o d d d o o o o o o
1886: row 5 | o o o d d d o o o o o o
1887: -------------------
1888: .ve
1889:
1890: Thus, any entries in the d locations are stored in the d (diagonal)
1891: submatrix, and any entries in the o locations are stored in the
1892: o (off-diagonal) submatrix. Note that the d matrix is stored in
1893: MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format.
1895: Now d_nz should indicate the number of block nonzeros per row in the upper triangular
1896: plus the diagonal part of the d matrix,
1897: and o_nz should indicate the number of block nonzeros per row in the o matrix.
1898: In general, for PDE problems in which most nonzeros are near the diagonal,
1899: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
1900: or you will get TERRIBLE performance; see the users' manual chapter on
1901: matrices.
1903: Level: intermediate
1905: .keywords: matrix, block, aij, compressed row, sparse, parallel
1907: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ()
1908: @*/
1909: PetscErrorCode MatMPISBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[])
1910: {
1911: PetscErrorCode ierr,(*f)(Mat,PetscInt,PetscInt,const PetscInt[],PetscInt,const PetscInt[]);
1914: PetscObjectQueryFunction((PetscObject)B,"MatMPISBAIJSetPreallocation_C",(void (**)(void))&f);
1915: if (f) {
1916: (*f)(B,bs,d_nz,d_nnz,o_nz,o_nnz);
1917: }
1918: return(0);
1919: }
1923: /*@C
1924: MatCreateMPISBAIJ - Creates a sparse parallel matrix in symmetric block AIJ format
1925: (block compressed row). For good matrix assembly performance
1926: the user should preallocate the matrix storage by setting the parameters
1927: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
1928: performance can be increased by more than a factor of 50.
1930: Collective on MPI_Comm
1932: Input Parameters:
1933: + comm - MPI communicator
1934: . bs - size of blockk
1935: . m - number of local rows (or PETSC_DECIDE to have calculated if M is given)
1936: This value should be the same as the local size used in creating the
1937: y vector for the matrix-vector product y = Ax.
1938: . n - number of local columns (or PETSC_DECIDE to have calculated if N is given)
1939: This value should be the same as the local size used in creating the
1940: x vector for the matrix-vector product y = Ax.
1941: . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given)
1942: . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given)
1943: . d_nz - number of block nonzeros per block row in diagonal portion of local
1944: submatrix (same for all local rows)
1945: . d_nnz - array containing the number of block nonzeros in the various block rows
1946: in the upper triangular portion of the in diagonal portion of the local
1947: (possibly different for each block block row) or PETSC_NULL.
1948: You must leave room for the diagonal entry even if it is zero.
1949: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
1950: submatrix (same for all local rows).
1951: - o_nnz - array containing the number of nonzeros in the various block rows of the
1952: off-diagonal portion of the local submatrix (possibly different for
1953: each block row) or PETSC_NULL.
1955: Output Parameter:
1956: . A - the matrix
1958: Options Database Keys:
1959: . -mat_no_unroll - uses code that does not unroll the loops in the
1960: block calculations (much slower)
1961: . -mat_block_size - size of the blocks to use
1962: . -mat_mpi - use the parallel matrix data structures even on one processor
1963: (defaults to using SeqBAIJ format on one processor)
1965: It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(),
1966: MatXXXXSetPreallocation() paradgm instead of this routine directly.
1967: [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation]
1969: Notes:
1970: The number of rows and columns must be divisible by blocksize.
1971: This matrix type does not support complex Hermitian operation.
1973: The user MUST specify either the local or global matrix dimensions
1974: (possibly both).
1976: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
1977: than it must be used on all processors that share the object for that argument.
1979: If the *_nnz parameter is given then the *_nz parameter is ignored
1981: Storage Information:
1982: For a square global matrix we define each processor's diagonal portion
1983: to be its local rows and the corresponding columns (a square submatrix);
1984: each processor's off-diagonal portion encompasses the remainder of the
1985: local matrix (a rectangular submatrix).
1987: The user can specify preallocated storage for the diagonal part of
1988: the local submatrix with either d_nz or d_nnz (not both). Set
1989: d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic
1990: memory allocation. Likewise, specify preallocated storage for the
1991: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
1993: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
1994: the figure below we depict these three local rows and all columns (0-11).
1996: .vb
1997: 0 1 2 3 4 5 6 7 8 9 10 11
1998: -------------------
1999: row 3 | o o o d d d o o o o o o
2000: row 4 | o o o d d d o o o o o o
2001: row 5 | o o o d d d o o o o o o
2002: -------------------
2003: .ve
2004:
2005: Thus, any entries in the d locations are stored in the d (diagonal)
2006: submatrix, and any entries in the o locations are stored in the
2007: o (off-diagonal) submatrix. Note that the d matrix is stored in
2008: MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format.
2010: Now d_nz should indicate the number of block nonzeros per row in the upper triangular
2011: plus the diagonal part of the d matrix,
2012: and o_nz should indicate the number of block nonzeros per row in the o matrix.
2013: In general, for PDE problems in which most nonzeros are near the diagonal,
2014: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
2015: or you will get TERRIBLE performance; see the users' manual chapter on
2016: matrices.
2018: Level: intermediate
2020: .keywords: matrix, block, aij, compressed row, sparse, parallel
2022: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ()
2023: @*/
2025: PetscErrorCode MatCreateMPISBAIJ(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt M,PetscInt N,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[],Mat *A)
2026: {
2028: PetscMPIInt size;
2031: MatCreate(comm,A);
2032: MatSetSizes(*A,m,n,M,N);
2033: MPI_Comm_size(comm,&size);
2034: if (size > 1) {
2035: MatSetType(*A,MATMPISBAIJ);
2036: MatMPISBAIJSetPreallocation(*A,bs,d_nz,d_nnz,o_nz,o_nnz);
2037: } else {
2038: MatSetType(*A,MATSEQSBAIJ);
2039: MatSeqSBAIJSetPreallocation(*A,bs,d_nz,d_nnz);
2040: }
2041: return(0);
2042: }
2047: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat matin,MatDuplicateOption cpvalues,Mat *newmat)
2048: {
2049: Mat mat;
2050: Mat_MPISBAIJ *a,*oldmat = (Mat_MPISBAIJ*)matin->data;
2052: PetscInt len=0,nt,bs=matin->rmap->bs,mbs=oldmat->mbs;
2053: PetscScalar *array;
2056: *newmat = 0;
2057: MatCreate(((PetscObject)matin)->comm,&mat);
2058: MatSetSizes(mat,matin->rmap->n,matin->cmap->n,matin->rmap->N,matin->cmap->N);
2059: MatSetType(mat,((PetscObject)matin)->type_name);
2060: PetscMemcpy(mat->ops,matin->ops,sizeof(struct _MatOps));
2061: PetscLayoutCopy(matin->rmap,&mat->rmap);
2062: PetscLayoutCopy(matin->cmap,&mat->cmap);
2063:
2064: mat->factor = matin->factor;
2065: mat->preallocated = PETSC_TRUE;
2066: mat->assembled = PETSC_TRUE;
2067: mat->insertmode = NOT_SET_VALUES;
2069: a = (Mat_MPISBAIJ*)mat->data;
2070: a->bs2 = oldmat->bs2;
2071: a->mbs = oldmat->mbs;
2072: a->nbs = oldmat->nbs;
2073: a->Mbs = oldmat->Mbs;
2074: a->Nbs = oldmat->Nbs;
2077: a->size = oldmat->size;
2078: a->rank = oldmat->rank;
2079: a->donotstash = oldmat->donotstash;
2080: a->roworiented = oldmat->roworiented;
2081: a->rowindices = 0;
2082: a->rowvalues = 0;
2083: a->getrowactive = PETSC_FALSE;
2084: a->barray = 0;
2085: a->rstartbs = oldmat->rstartbs;
2086: a->rendbs = oldmat->rendbs;
2087: a->cstartbs = oldmat->cstartbs;
2088: a->cendbs = oldmat->cendbs;
2090: /* hash table stuff */
2091: a->ht = 0;
2092: a->hd = 0;
2093: a->ht_size = 0;
2094: a->ht_flag = oldmat->ht_flag;
2095: a->ht_fact = oldmat->ht_fact;
2096: a->ht_total_ct = 0;
2097: a->ht_insert_ct = 0;
2098:
2099: PetscMemcpy(a->rangebs,oldmat->rangebs,(a->size+2)*sizeof(PetscInt));
2100: if (oldmat->colmap) {
2101: #if defined (PETSC_USE_CTABLE)
2102: PetscTableCreateCopy(oldmat->colmap,&a->colmap);
2103: #else
2104: PetscMalloc((a->Nbs)*sizeof(PetscInt),&a->colmap);
2105: PetscLogObjectMemory(mat,(a->Nbs)*sizeof(PetscInt));
2106: PetscMemcpy(a->colmap,oldmat->colmap,(a->Nbs)*sizeof(PetscInt));
2107: #endif
2108: } else a->colmap = 0;
2110: if (oldmat->garray && (len = ((Mat_SeqBAIJ*)(oldmat->B->data))->nbs)) {
2111: PetscMalloc(len*sizeof(PetscInt),&a->garray);
2112: PetscLogObjectMemory(mat,len*sizeof(PetscInt));
2113: PetscMemcpy(a->garray,oldmat->garray,len*sizeof(PetscInt));
2114: } else a->garray = 0;
2115:
2116: MatStashCreate_Private(((PetscObject)matin)->comm,matin->rmap->bs,&mat->bstash);
2117: VecDuplicate(oldmat->lvec,&a->lvec);
2118: PetscLogObjectParent(mat,a->lvec);
2119: VecScatterCopy(oldmat->Mvctx,&a->Mvctx);
2120: PetscLogObjectParent(mat,a->Mvctx);
2122: VecDuplicate(oldmat->slvec0,&a->slvec0);
2123: PetscLogObjectParent(mat,a->slvec0);
2124: VecDuplicate(oldmat->slvec1,&a->slvec1);
2125: PetscLogObjectParent(mat,a->slvec1);
2127: VecGetLocalSize(a->slvec1,&nt);
2128: VecGetArray(a->slvec1,&array);
2129: VecCreateSeqWithArray(PETSC_COMM_SELF,bs*mbs,array,&a->slvec1a);
2130: VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec1b);
2131: VecRestoreArray(a->slvec1,&array);
2132: VecGetArray(a->slvec0,&array);
2133: VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec0b);
2134: VecRestoreArray(a->slvec0,&array);
2135: PetscLogObjectParent(mat,a->slvec0);
2136: PetscLogObjectParent(mat,a->slvec1);
2137: PetscLogObjectParent(mat,a->slvec0b);
2138: PetscLogObjectParent(mat,a->slvec1a);
2139: PetscLogObjectParent(mat,a->slvec1b);
2141: /* VecScatterCopy(oldmat->sMvctx,&a->sMvctx); - not written yet, replaced by the lazy trick: */
2142: PetscObjectReference((PetscObject)oldmat->sMvctx);
2143: a->sMvctx = oldmat->sMvctx;
2144: PetscLogObjectParent(mat,a->sMvctx);
2146: MatDuplicate(oldmat->A,cpvalues,&a->A);
2147: PetscLogObjectParent(mat,a->A);
2148: MatDuplicate(oldmat->B,cpvalues,&a->B);
2149: PetscLogObjectParent(mat,a->B);
2150: PetscFListDuplicate(((PetscObject)matin)->qlist,&((PetscObject)mat)->qlist);
2151: *newmat = mat;
2152: return(0);
2153: }
2157: PetscErrorCode MatLoad_MPISBAIJ(PetscViewer viewer, const MatType type,Mat *newmat)
2158: {
2159: Mat A;
2161: PetscInt i,nz,j,rstart,rend;
2162: PetscScalar *vals,*buf;
2163: MPI_Comm comm = ((PetscObject)viewer)->comm;
2164: MPI_Status status;
2165: PetscMPIInt rank,size,tag = ((PetscObject)viewer)->tag,*sndcounts = 0,*browners,maxnz,*rowners,*locrowlens,mmbs;
2166: PetscInt header[4],*rowlengths = 0,M,N,m,*cols;
2167: PetscInt *procsnz = 0,jj,*mycols,*ibuf;
2168: PetscInt bs=1,Mbs,mbs,extra_rows;
2169: PetscInt *dlens,*odlens,*mask,*masked1,*masked2,rowcount,odcount;
2170: PetscInt dcount,kmax,k,nzcount,tmp;
2171: int fd;
2172:
2174: PetscOptionsBegin(comm,PETSC_NULL,"Options for loading MPISBAIJ matrix 2","Mat");
2175: PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,PETSC_NULL);
2176: PetscOptionsEnd();
2178: MPI_Comm_size(comm,&size);
2179: MPI_Comm_rank(comm,&rank);
2180: if (!rank) {
2181: PetscViewerBinaryGetDescriptor(viewer,&fd);
2182: PetscBinaryRead(fd,(char *)header,4,PETSC_INT);
2183: if (header[0] != MAT_FILE_COOKIE) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"not matrix object");
2184: if (header[3] < 0) {
2185: SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as MPISBAIJ");
2186: }
2187: }
2189: MPI_Bcast(header+1,3,MPIU_INT,0,comm);
2190: M = header[1]; N = header[2];
2192: if (M != N) SETERRQ(PETSC_ERR_SUP,"Can only do square matrices");
2194: /*
2195: This code adds extra rows to make sure the number of rows is
2196: divisible by the blocksize
2197: */
2198: Mbs = M/bs;
2199: extra_rows = bs - M + bs*(Mbs);
2200: if (extra_rows == bs) extra_rows = 0;
2201: else Mbs++;
2202: if (extra_rows &&!rank) {
2203: PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");
2204: }
2206: /* determine ownership of all rows */
2207: mbs = Mbs/size + ((Mbs % size) > rank);
2208: m = mbs*bs;
2209: PetscMalloc2(size+1,PetscMPIInt,&rowners,size+1,PetscMPIInt,&browners);
2210: mmbs = PetscMPIIntCast(mbs);
2211: MPI_Allgather(&mmbs,1,MPI_INT,rowners+1,1,MPI_INT,comm);
2212: rowners[0] = 0;
2213: for (i=2; i<=size; i++) rowners[i] += rowners[i-1];
2214: for (i=0; i<=size; i++) browners[i] = rowners[i]*bs;
2215: rstart = rowners[rank];
2216: rend = rowners[rank+1];
2217:
2218: /* distribute row lengths to all processors */
2219: PetscMalloc((rend-rstart)*bs*sizeof(PetscMPIInt),&locrowlens);
2220: if (!rank) {
2221: PetscMalloc((M+extra_rows)*sizeof(PetscInt),&rowlengths);
2222: PetscBinaryRead(fd,rowlengths,M,PETSC_INT);
2223: for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1;
2224: PetscMalloc(size*sizeof(PetscMPIInt),&sndcounts);
2225: for (i=0; i<size; i++) sndcounts[i] = browners[i+1] - browners[i];
2226: MPI_Scatterv(rowlengths,sndcounts,browners,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);
2227: PetscFree(sndcounts);
2228: } else {
2229: MPI_Scatterv(0,0,0,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);
2230: }
2231:
2232: if (!rank) { /* procs[0] */
2233: /* calculate the number of nonzeros on each processor */
2234: PetscMalloc(size*sizeof(PetscInt),&procsnz);
2235: PetscMemzero(procsnz,size*sizeof(PetscInt));
2236: for (i=0; i<size; i++) {
2237: for (j=rowners[i]*bs; j< rowners[i+1]*bs; j++) {
2238: procsnz[i] += rowlengths[j];
2239: }
2240: }
2241: PetscFree(rowlengths);
2242:
2243: /* determine max buffer needed and allocate it */
2244: maxnz = 0;
2245: for (i=0; i<size; i++) {
2246: maxnz = PetscMax(maxnz,procsnz[i]);
2247: }
2248: PetscMalloc(maxnz*sizeof(PetscInt),&cols);
2250: /* read in my part of the matrix column indices */
2251: nz = procsnz[0];
2252: PetscMalloc(nz*sizeof(PetscInt),&ibuf);
2253: mycols = ibuf;
2254: if (size == 1) nz -= extra_rows;
2255: PetscBinaryRead(fd,mycols,nz,PETSC_INT);
2256: if (size == 1) for (i=0; i< extra_rows; i++) { mycols[nz+i] = M+i; }
2258: /* read in every ones (except the last) and ship off */
2259: for (i=1; i<size-1; i++) {
2260: nz = procsnz[i];
2261: PetscBinaryRead(fd,cols,nz,PETSC_INT);
2262: MPI_Send(cols,nz,MPIU_INT,i,tag,comm);
2263: }
2264: /* read in the stuff for the last proc */
2265: if (size != 1) {
2266: nz = procsnz[size-1] - extra_rows; /* the extra rows are not on the disk */
2267: PetscBinaryRead(fd,cols,nz,PETSC_INT);
2268: for (i=0; i<extra_rows; i++) cols[nz+i] = M+i;
2269: MPI_Send(cols,nz+extra_rows,MPIU_INT,size-1,tag,comm);
2270: }
2271: PetscFree(cols);
2272: } else { /* procs[i], i>0 */
2273: /* determine buffer space needed for message */
2274: nz = 0;
2275: for (i=0; i<m; i++) {
2276: nz += locrowlens[i];
2277: }
2278: PetscMalloc(nz*sizeof(PetscInt),&ibuf);
2279: mycols = ibuf;
2280: /* receive message of column indices*/
2281: MPI_Recv(mycols,nz,MPIU_INT,0,tag,comm,&status);
2282: MPI_Get_count(&status,MPIU_INT,&maxnz);
2283: if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file");
2284: }
2286: /* loop over local rows, determining number of off diagonal entries */
2287: PetscMalloc2(rend-rstart,PetscInt,&dlens,rend-rstart,PetscInt,&odlens);
2288: PetscMalloc3(Mbs,PetscInt,&mask,Mbs,PetscInt,&masked1,Mbs,PetscInt,&masked2);
2289: PetscMemzero(mask,Mbs*sizeof(PetscInt));
2290: PetscMemzero(masked1,Mbs*sizeof(PetscInt));
2291: PetscMemzero(masked2,Mbs*sizeof(PetscInt));
2292: rowcount = 0;
2293: nzcount = 0;
2294: for (i=0; i<mbs; i++) {
2295: dcount = 0;
2296: odcount = 0;
2297: for (j=0; j<bs; j++) {
2298: kmax = locrowlens[rowcount];
2299: for (k=0; k<kmax; k++) {
2300: tmp = mycols[nzcount++]/bs; /* block col. index */
2301: if (!mask[tmp]) {
2302: mask[tmp] = 1;
2303: if (tmp < rstart || tmp >= rend) masked2[odcount++] = tmp; /* entry in off-diag portion */
2304: else masked1[dcount++] = tmp; /* entry in diag portion */
2305: }
2306: }
2307: rowcount++;
2308: }
2309:
2310: dlens[i] = dcount; /* d_nzz[i] */
2311: odlens[i] = odcount; /* o_nzz[i] */
2313: /* zero out the mask elements we set */
2314: for (j=0; j<dcount; j++) mask[masked1[j]] = 0;
2315: for (j=0; j<odcount; j++) mask[masked2[j]] = 0;
2316: }
2317:
2318: /* create our matrix */
2319: MatCreate(comm,&A);
2320: MatSetSizes(A,m,m,PETSC_DETERMINE,PETSC_DETERMINE);
2321: MatSetType(A,type);
2322: MatSetOption(A,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE);
2323: MatMPISBAIJSetPreallocation(A,bs,0,dlens,0,odlens);
2324:
2325: if (!rank) {
2326: PetscMalloc(maxnz*sizeof(PetscScalar),&buf);
2327: /* read in my part of the matrix numerical values */
2328: nz = procsnz[0];
2329: vals = buf;
2330: mycols = ibuf;
2331: if (size == 1) nz -= extra_rows;
2332: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2333: if (size == 1) for (i=0; i< extra_rows; i++) { vals[nz+i] = 1.0; }
2335: /* insert into matrix */
2336: jj = rstart*bs;
2337: for (i=0; i<m; i++) {
2338: MatSetValues(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
2339: mycols += locrowlens[i];
2340: vals += locrowlens[i];
2341: jj++;
2342: }
2344: /* read in other processors (except the last one) and ship out */
2345: for (i=1; i<size-1; i++) {
2346: nz = procsnz[i];
2347: vals = buf;
2348: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2349: MPI_Send(vals,nz,MPIU_SCALAR,i,((PetscObject)A)->tag,comm);
2350: }
2351: /* the last proc */
2352: if (size != 1){
2353: nz = procsnz[i] - extra_rows;
2354: vals = buf;
2355: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2356: for (i=0; i<extra_rows; i++) vals[nz+i] = 1.0;
2357: MPI_Send(vals,nz+extra_rows,MPIU_SCALAR,size-1,((PetscObject)A)->tag,comm);
2358: }
2359: PetscFree(procsnz);
2361: } else {
2362: /* receive numeric values */
2363: PetscMalloc(nz*sizeof(PetscScalar),&buf);
2365: /* receive message of values*/
2366: vals = buf;
2367: mycols = ibuf;
2368: MPI_Recv(vals,nz,MPIU_SCALAR,0,((PetscObject)A)->tag,comm,&status);
2369: MPI_Get_count(&status,MPIU_SCALAR,&maxnz);
2370: if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file");
2372: /* insert into matrix */
2373: jj = rstart*bs;
2374: for (i=0; i<m; i++) {
2375: MatSetValues_MPISBAIJ(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
2376: mycols += locrowlens[i];
2377: vals += locrowlens[i];
2378: jj++;
2379: }
2380: }
2382: PetscFree(locrowlens);
2383: PetscFree(buf);
2384: PetscFree(ibuf);
2385: PetscFree2(rowners,browners);
2386: PetscFree2(dlens,odlens);
2387: PetscFree3(mask,masked1,masked2);
2388: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
2389: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
2390: *newmat = A;
2391: return(0);
2392: }
2396: /*XXXXX@
2397: MatMPISBAIJSetHashTableFactor - Sets the factor required to compute the size of the HashTable.
2399: Input Parameters:
2400: . mat - the matrix
2401: . fact - factor
2403: Collective on Mat
2405: Level: advanced
2407: Notes:
2408: This can also be set by the command line option: -mat_use_hash_table fact
2410: .keywords: matrix, hashtable, factor, HT
2412: .seealso: MatSetOption()
2413: @XXXXX*/
2418: PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat A,Vec v,PetscInt idx[])
2419: {
2420: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
2421: Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(a->B)->data;
2422: PetscReal atmp;
2423: PetscReal *work,*svalues,*rvalues;
2425: PetscInt i,bs,mbs,*bi,*bj,brow,j,ncols,krow,kcol,col,row,Mbs,bcol;
2426: PetscMPIInt rank,size;
2427: PetscInt *rowners_bs,dest,count,source;
2428: PetscScalar *va;
2429: MatScalar *ba;
2430: MPI_Status stat;
2433: if (idx) SETERRQ(PETSC_ERR_SUP,"Send email to petsc-maint@mcs.anl.gov");
2434: MatGetRowMaxAbs(a->A,v,PETSC_NULL);
2435: VecGetArray(v,&va);
2437: MPI_Comm_size(((PetscObject)A)->comm,&size);
2438: MPI_Comm_rank(((PetscObject)A)->comm,&rank);
2440: bs = A->rmap->bs;
2441: mbs = a->mbs;
2442: Mbs = a->Mbs;
2443: ba = b->a;
2444: bi = b->i;
2445: bj = b->j;
2447: /* find ownerships */
2448: rowners_bs = A->rmap->range;
2450: /* each proc creates an array to be distributed */
2451: PetscMalloc(bs*Mbs*sizeof(PetscReal),&work);
2452: PetscMemzero(work,bs*Mbs*sizeof(PetscReal));
2454: /* row_max for B */
2455: if (rank != size-1){
2456: for (i=0; i<mbs; i++) {
2457: ncols = bi[1] - bi[0]; bi++;
2458: brow = bs*i;
2459: for (j=0; j<ncols; j++){
2460: bcol = bs*(*bj);
2461: for (kcol=0; kcol<bs; kcol++){
2462: col = bcol + kcol; /* local col index */
2463: col += rowners_bs[rank+1]; /* global col index */
2464: for (krow=0; krow<bs; krow++){
2465: atmp = PetscAbsScalar(*ba); ba++;
2466: row = brow + krow; /* local row index */
2467: if (PetscRealPart(va[row]) < atmp) va[row] = atmp;
2468: if (work[col] < atmp) work[col] = atmp;
2469: }
2470: }
2471: bj++;
2472: }
2473: }
2475: /* send values to its owners */
2476: for (dest=rank+1; dest<size; dest++){
2477: svalues = work + rowners_bs[dest];
2478: count = rowners_bs[dest+1]-rowners_bs[dest];
2479: MPI_Send(svalues,count,MPIU_REAL,dest,rank,((PetscObject)A)->comm);
2480: }
2481: }
2482:
2483: /* receive values */
2484: if (rank){
2485: rvalues = work;
2486: count = rowners_bs[rank+1]-rowners_bs[rank];
2487: for (source=0; source<rank; source++){
2488: MPI_Recv(rvalues,count,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,((PetscObject)A)->comm,&stat);
2489: /* process values */
2490: for (i=0; i<count; i++){
2491: if (PetscRealPart(va[i]) < rvalues[i]) va[i] = rvalues[i];
2492: }
2493: }
2494: }
2496: VecRestoreArray(v,&va);
2497: PetscFree(work);
2498: return(0);
2499: }
2503: PetscErrorCode MatSOR_MPISBAIJ(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
2504: {
2505: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
2507: PetscInt mbs=mat->mbs,bs=matin->rmap->bs;
2508: PetscScalar *x,*b,*ptr,*from;
2509: Vec bb1;
2510:
2512: if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits);
2513: if (bs > 1) SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented");
2515: if (flag == SOR_APPLY_UPPER) {
2516: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2517: return(0);
2518: }
2520: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){
2521: if ( flag & SOR_ZERO_INITIAL_GUESS ) {
2522: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,lits,xx);
2523: its--;
2524: }
2526: VecDuplicate(bb,&bb1);
2527: while (its--){
2528:
2529: /* lower triangular part: slvec0b = - B^T*xx */
2530: (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);
2531:
2532: /* copy xx into slvec0a */
2533: VecGetArray(mat->slvec0,&ptr);
2534: VecGetArray(xx,&x);
2535: PetscMemcpy(ptr,x,bs*mbs*sizeof(MatScalar));
2536: VecRestoreArray(mat->slvec0,&ptr);
2538: VecScale(mat->slvec0,-1.0);
2540: /* copy bb into slvec1a */
2541: VecGetArray(mat->slvec1,&ptr);
2542: VecGetArray(bb,&b);
2543: PetscMemcpy(ptr,b,bs*mbs*sizeof(MatScalar));
2544: VecRestoreArray(mat->slvec1,&ptr);
2546: /* set slvec1b = 0 */
2547: VecSet(mat->slvec1b,0.0);
2549: VecScatterBegin(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);
2550: VecRestoreArray(xx,&x);
2551: VecRestoreArray(bb,&b);
2552: VecScatterEnd(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);
2554: /* upper triangular part: bb1 = bb1 - B*x */
2555: (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,bb1);
2556:
2557: /* local diagonal sweep */
2558: (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);
2559: }
2560: VecDestroy(bb1);
2561: } else if ((flag & SOR_LOCAL_FORWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)){
2562: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2563: } else if ((flag & SOR_LOCAL_BACKWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)){
2564: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2565: } else if (flag & SOR_EISENSTAT) {
2566: Vec xx1;
2567: PetscTruth hasop;
2568: const PetscScalar *diag;
2569: PetscScalar *sl,scale = (omega - 2.0)/omega;
2570: PetscInt i,n;
2572: if (!mat->xx1) {
2573: VecDuplicate(bb,&mat->xx1);
2574: VecDuplicate(bb,&mat->bb1);
2575: }
2576: xx1 = mat->xx1;
2577: bb1 = mat->bb1;
2579: (*mat->A->ops->sor)(mat->A,bb,omega,(MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_BACKWARD_SWEEP),fshift,lits,1,xx);
2581: if (!mat->diag) {
2582: /* this is wrong for same matrix with new nonzero values */
2583: MatGetVecs(matin,&mat->diag,PETSC_NULL);
2584: MatGetDiagonal(matin,mat->diag);
2585: }
2586: MatHasOperation(matin,MATOP_MULT_DIAGONAL_BLOCK,&hasop);
2588: if (hasop) {
2589: MatMultDiagonalBlock(matin,xx,bb1);
2590: VecAYPX(mat->slvec1a,scale,bb);
2591: } else {
2592: /*
2593: These two lines are replaced by code that may be a bit faster for a good compiler
2594: VecPointwiseMult(mat->slvec1a,mat->diag,xx);
2595: VecAYPX(mat->slvec1a,scale,bb);
2596: */
2597: VecGetArray(mat->slvec1a,&sl);
2598: VecGetArray(mat->diag,(PetscScalar**)&diag);
2599: VecGetArray(bb,&b);
2600: VecGetArray(xx,&x);
2601: VecGetLocalSize(xx,&n);
2602: if (omega == 1.0) {
2603: for (i=0; i<n; i++) {
2604: sl[i] = b[i] - diag[i]*x[i];
2605: }
2606: PetscLogFlops(2.0*n);
2607: } else {
2608: for (i=0; i<n; i++) {
2609: sl[i] = b[i] + scale*diag[i]*x[i];
2610: }
2611: PetscLogFlops(3.0*n);
2612: }
2613: VecRestoreArray(mat->slvec1a,&sl);
2614: VecRestoreArray(mat->diag,(PetscScalar**)&diag);
2615: VecRestoreArray(bb,&b);
2616: VecRestoreArray(xx,&x);
2617: }
2619: /* multiply off-diagonal portion of matrix */
2620: VecSet(mat->slvec1b,0.0);
2621: (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);
2622: VecGetArray(mat->slvec0,&from);
2623: VecGetArray(xx,&x);
2624: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
2625: VecRestoreArray(mat->slvec0,&from);
2626: VecRestoreArray(xx,&x);
2627: VecScatterBegin(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);
2628: VecScatterEnd(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);
2629: (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,mat->slvec1a);
2631: /* local sweep */
2632: (*mat->A->ops->sor)(mat->A,mat->slvec1a,omega,(MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_FORWARD_SWEEP),fshift,lits,1,xx1);
2633: VecAXPY(xx,1.0,xx1);
2634: } else {
2635: SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format");
2636: }
2637: return(0);
2638: }
2642: PetscErrorCode MatSOR_MPISBAIJ_2comm(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
2643: {
2644: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
2646: Vec lvec1,bb1;
2647:
2649: if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits);
2650: if (matin->rmap->bs > 1) SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented");
2652: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){
2653: if ( flag & SOR_ZERO_INITIAL_GUESS ) {
2654: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,lits,xx);
2655: its--;
2656: }
2658: VecDuplicate(mat->lvec,&lvec1);
2659: VecDuplicate(bb,&bb1);
2660: while (its--){
2661: VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2662:
2663: /* lower diagonal part: bb1 = bb - B^T*xx */
2664: (*mat->B->ops->multtranspose)(mat->B,xx,lvec1);
2665: VecScale(lvec1,-1.0);
2667: VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2668: VecCopy(bb,bb1);
2669: VecScatterBegin(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);
2671: /* upper diagonal part: bb1 = bb1 - B*x */
2672: VecScale(mat->lvec,-1.0);
2673: (*mat->B->ops->multadd)(mat->B,mat->lvec,bb1,bb1);
2675: VecScatterEnd(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);
2676:
2677: /* diagonal sweep */
2678: (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);
2679: }
2680: VecDestroy(lvec1);
2681: VecDestroy(bb1);
2682: } else {
2683: SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format");
2684: }
2685: return(0);
2686: }