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
7: EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ(Mat);
8: EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ_2comm(Mat);
9: EXTERN PetscErrorCode DisAssemble_MPISBAIJ(Mat);
10: EXTERN PetscErrorCode MatIncreaseOverlap_MPISBAIJ(Mat,PetscInt,IS[],PetscInt);
11: EXTERN PetscErrorCode MatGetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []);
12: EXTERN PetscErrorCode MatGetValues_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []);
13: EXTERN PetscErrorCode MatSetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt [],PetscInt,const PetscInt [],const PetscScalar [],InsertMode);
14: EXTERN PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode);
15: EXTERN PetscErrorCode MatSetValuesBlocked_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode);
16: EXTERN PetscErrorCode MatGetRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**);
17: EXTERN PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**);
18: EXTERN PetscErrorCode MatPrintHelp_SeqSBAIJ(Mat);
19: EXTERN PetscErrorCode MatZeroRows_SeqSBAIJ(Mat,IS,PetscScalar*);
20: EXTERN PetscErrorCode MatZeroRows_SeqBAIJ(Mat,IS,PetscScalar *);
21: EXTERN PetscErrorCode MatGetRowMax_MPISBAIJ(Mat,Vec);
22: EXTERN PetscErrorCode MatRelax_MPISBAIJ(Mat,Vec,PetscReal,MatSORType,PetscReal,PetscInt,PetscInt,Vec);
24: /* UGLY, ugly, ugly
25: When MatScalar == PetscScalar the function MatSetValuesBlocked_MPIBAIJ_MatScalar() does
26: not exist. Otherwise ..._MatScalar() takes matrix elements in single precision and
27: inserts them into the single precision data structure. The function MatSetValuesBlocked_MPIBAIJ()
28: converts the entries into single precision and then calls ..._MatScalar() to put them
29: into the single precision data structures.
30: */
31: #if defined(PETSC_USE_MAT_SINGLE)
32: EXTERN PetscErrorCode MatSetValuesBlocked_SeqSBAIJ_MatScalar(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const MatScalar[],InsertMode);
33: EXTERN PetscErrorCode MatSetValues_MPISBAIJ_MatScalar(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const MatScalar[],InsertMode);
34: EXTERN PetscErrorCode MatSetValuesBlocked_MPISBAIJ_MatScalar(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const MatScalar[],InsertMode);
35: EXTERN PetscErrorCode MatSetValues_MPISBAIJ_HT_MatScalar(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const MatScalar[],InsertMode);
36: EXTERN PetscErrorCode MatSetValuesBlocked_MPISBAIJ_HT_MatScalar(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const MatScalar[],InsertMode);
37: #else
38: #define MatSetValuesBlocked_SeqSBAIJ_MatScalar MatSetValuesBlocked_SeqSBAIJ
39: #define MatSetValues_MPISBAIJ_MatScalar MatSetValues_MPISBAIJ
40: #define MatSetValuesBlocked_MPISBAIJ_MatScalar MatSetValuesBlocked_MPISBAIJ
41: #define MatSetValues_MPISBAIJ_HT_MatScalar MatSetValues_MPISBAIJ_HT
42: #define MatSetValuesBlocked_MPISBAIJ_HT_MatScalar MatSetValuesBlocked_MPISBAIJ_HT
43: #endif
48: PetscErrorCode PETSCMAT_DLLEXPORT MatStoreValues_MPISBAIJ(Mat mat)
49: {
50: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
54: MatStoreValues(aij->A);
55: MatStoreValues(aij->B);
56: return(0);
57: }
63: PetscErrorCode PETSCMAT_DLLEXPORT MatRetrieveValues_MPISBAIJ(Mat mat)
64: {
65: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
69: MatRetrieveValues(aij->A);
70: MatRetrieveValues(aij->B);
71: return(0);
72: }
76: #define CHUNKSIZE 10
78: #define MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv) \
79: { \
80: \
81: brow = row/bs; \
82: rp = aj + ai[brow]; ap = aa + bs2*ai[brow]; \
83: rmax = aimax[brow]; nrow = ailen[brow]; \
84: bcol = col/bs; \
85: ridx = row % bs; cidx = col % bs; \
86: low = 0; high = nrow; \
87: while (high-low > 3) { \
88: t = (low+high)/2; \
89: if (rp[t] > bcol) high = t; \
90: else low = t; \
91: } \
92: for (_i=low; _i<high; _i++) { \
93: if (rp[_i] > bcol) break; \
94: if (rp[_i] == bcol) { \
95: bap = ap + bs2*_i + bs*cidx + ridx; \
96: if (addv == ADD_VALUES) *bap += value; \
97: else *bap = value; \
98: goto a_noinsert; \
99: } \
100: } \
101: if (a->nonew == 1) goto a_noinsert; \
102: if (a->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \
103: MatSeqXAIJReallocateAIJ(a,bs2,nrow,brow,bcol,rmax,aa,ai,aj,a->mbs,rp,ap,aimax,a->nonew); \
104: N = nrow++ - 1; \
105: /* shift up all the later entries in this row */ \
106: for (ii=N; ii>=_i; ii--) { \
107: rp[ii+1] = rp[ii]; \
108: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
109: } \
110: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar)); } \
111: rp[_i] = bcol; \
112: ap[bs2*_i + bs*cidx + ridx] = value; \
113: a_noinsert:; \
114: ailen[brow] = nrow; \
115: }
116: #ifndef MatSetValues_SeqBAIJ_B_Private
117: #define MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv) \
118: { \
119: brow = row/bs; \
120: rp = bj + bi[brow]; ap = ba + bs2*bi[brow]; \
121: rmax = bimax[brow]; nrow = bilen[brow]; \
122: bcol = col/bs; \
123: ridx = row % bs; cidx = col % bs; \
124: low = 0; high = nrow; \
125: while (high-low > 3) { \
126: t = (low+high)/2; \
127: if (rp[t] > bcol) high = t; \
128: else low = t; \
129: } \
130: for (_i=low; _i<high; _i++) { \
131: if (rp[_i] > bcol) break; \
132: if (rp[_i] == bcol) { \
133: bap = ap + bs2*_i + bs*cidx + ridx; \
134: if (addv == ADD_VALUES) *bap += value; \
135: else *bap = value; \
136: goto b_noinsert; \
137: } \
138: } \
139: if (b->nonew == 1) goto b_noinsert; \
140: if (b->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \
141: MatSeqXAIJReallocateAIJ(b,bs2,nrow,brow,bcol,rmax,ba,bi,bj,b->mbs,rp,ap,bimax,b->nonew); \
142: N = nrow++ - 1; \
143: /* shift up all the later entries in this row */ \
144: for (ii=N; ii>=_i; ii--) { \
145: rp[ii+1] = rp[ii]; \
146: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
147: } \
148: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));} \
149: rp[_i] = bcol; \
150: ap[bs2*_i + bs*cidx + ridx] = value; \
151: b_noinsert:; \
152: bilen[brow] = nrow; \
153: }
154: #endif
156: #if defined(PETSC_USE_MAT_SINGLE)
159: PetscErrorCode MatSetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
160: {
161: Mat_MPISBAIJ *b = (Mat_MPISBAIJ*)mat->data;
163: PetscInt i,N = m*n;
164: MatScalar *vsingle;
167: if (N > b->setvalueslen) {
168: PetscFree(b->setvaluescopy);
169: PetscMalloc(N*sizeof(MatScalar),&b->setvaluescopy);
170: b->setvalueslen = N;
171: }
172: vsingle = b->setvaluescopy;
174: for (i=0; i<N; i++) {
175: vsingle[i] = v[i];
176: }
177: MatSetValues_MPISBAIJ_MatScalar(mat,m,im,n,in,vsingle,addv);
178: return(0);
179: }
183: PetscErrorCode MatSetValuesBlocked_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
184: {
185: Mat_MPIBAIJ *b = (Mat_MPIBAIJ*)mat->data;
187: PetscInt i,N = m*n*b->bs2;
188: MatScalar *vsingle;
191: if (N > b->setvalueslen) {
192: PetscFree(b->setvaluescopy);
193: PetscMalloc(N*sizeof(MatScalar),&b->setvaluescopy);
194: b->setvalueslen = N;
195: }
196: vsingle = b->setvaluescopy;
197: for (i=0; i<N; i++) {
198: vsingle[i] = v[i];
199: }
200: MatSetValuesBlocked_MPISBAIJ_MatScalar(mat,m,im,n,in,vsingle,addv);
201: return(0);
202: }
203: #endif
205: /* Only add/insert a(i,j) with i<=j (blocks).
206: Any a(i,j) with i>j input by user is ingored.
207: */
210: PetscErrorCode MatSetValues_MPISBAIJ_MatScalar(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const MatScalar v[],InsertMode addv)
211: {
212: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
213: MatScalar value;
214: PetscTruth roworiented = baij->roworiented;
216: PetscInt i,j,row,col;
217: PetscInt rstart_orig=mat->rmap.rstart;
218: PetscInt rend_orig=mat->rmap.rend,cstart_orig=mat->cmap.rstart;
219: PetscInt cend_orig=mat->cmap.rend,bs=mat->rmap.bs;
221: /* Some Variables required in the macro */
222: Mat A = baij->A;
223: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)(A)->data;
224: PetscInt *aimax=a->imax,*ai=a->i,*ailen=a->ilen,*aj=a->j;
225: MatScalar *aa=a->a;
227: Mat B = baij->B;
228: Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(B)->data;
229: PetscInt *bimax=b->imax,*bi=b->i,*bilen=b->ilen,*bj=b->j;
230: MatScalar *ba=b->a;
232: PetscInt *rp,ii,nrow,_i,rmax,N,brow,bcol;
233: PetscInt low,high,t,ridx,cidx,bs2=a->bs2;
234: MatScalar *ap,*bap;
236: /* for stash */
237: PetscInt n_loc, *in_loc=0;
238: MatScalar *v_loc=0;
242: if(!baij->donotstash){
243: PetscMalloc(n*sizeof(PetscInt),&in_loc);
244: PetscMalloc(n*sizeof(MatScalar),&v_loc);
245: }
247: for (i=0; i<m; i++) {
248: if (im[i] < 0) continue;
249: #if defined(PETSC_USE_DEBUG)
250: if (im[i] >= mat->rmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],mat->rmap.N-1);
251: #endif
252: if (im[i] >= rstart_orig && im[i] < rend_orig) { /* this processor entry */
253: row = im[i] - rstart_orig; /* local row index */
254: for (j=0; j<n; j++) {
255: if (im[i]/bs > in[j]/bs){
256: if (a->ignore_ltriangular){
257: continue; /* ignore lower triangular blocks */
258: } else {
259: 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)");
260: }
261: }
262: if (in[j] >= cstart_orig && in[j] < cend_orig){ /* diag entry (A) */
263: col = in[j] - cstart_orig; /* local col index */
264: brow = row/bs; bcol = col/bs;
265: if (brow > bcol) continue; /* ignore lower triangular blocks of A */
266: if (roworiented) value = v[i*n+j]; else value = v[i+j*m];
267: MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv);
268: /* MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv); */
269: } else if (in[j] < 0) continue;
270: #if defined(PETSC_USE_DEBUG)
271: 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);}
272: #endif
273: else { /* off-diag entry (B) */
274: if (mat->was_assembled) {
275: if (!baij->colmap) {
276: CreateColmap_MPIBAIJ_Private(mat);
277: }
278: #if defined (PETSC_USE_CTABLE)
279: PetscTableFind(baij->colmap,in[j]/bs + 1,&col);
280: col = col - 1;
281: #else
282: col = baij->colmap[in[j]/bs] - 1;
283: #endif
284: if (col < 0 && !((Mat_SeqSBAIJ*)(baij->A->data))->nonew) {
285: DisAssemble_MPISBAIJ(mat);
286: col = in[j];
287: /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */
288: B = baij->B;
289: b = (Mat_SeqBAIJ*)(B)->data;
290: bimax=b->imax;bi=b->i;bilen=b->ilen;bj=b->j;
291: ba=b->a;
292: } else col += in[j]%bs;
293: } else col = in[j];
294: if (roworiented) value = v[i*n+j]; else value = v[i+j*m];
295: MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv);
296: /* MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv); */
297: }
298: }
299: } else { /* off processor entry */
300: if (!baij->donotstash) {
301: n_loc = 0;
302: for (j=0; j<n; j++){
303: if (im[i]/bs > in[j]/bs) continue; /* ignore lower triangular blocks */
304: in_loc[n_loc] = in[j];
305: if (roworiented) {
306: v_loc[n_loc] = v[i*n+j];
307: } else {
308: v_loc[n_loc] = v[j*m+i];
309: }
310: n_loc++;
311: }
312: MatStashValuesRow_Private(&mat->stash,im[i],n_loc,in_loc,v_loc);
313: }
314: }
315: }
317: if(!baij->donotstash){
318: PetscFree(in_loc);
319: PetscFree(v_loc);
320: }
321: return(0);
322: }
326: PetscErrorCode MatSetValuesBlocked_MPISBAIJ_MatScalar(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const MatScalar v[],InsertMode addv)
327: {
328: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
329: const MatScalar *value;
330: MatScalar *barray=baij->barray;
331: PetscTruth roworiented = baij->roworiented;
332: PetscErrorCode ierr;
333: PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs;
334: PetscInt rend=baij->rendbs,cstart=baij->rstartbs,stepval;
335: PetscInt cend=baij->rendbs,bs=mat->rmap.bs,bs2=baij->bs2;
338: if(!barray) {
339: PetscMalloc(bs2*sizeof(MatScalar),&barray);
340: baij->barray = barray;
341: }
343: if (roworiented) {
344: stepval = (n-1)*bs;
345: } else {
346: stepval = (m-1)*bs;
347: }
348: for (i=0; i<m; i++) {
349: if (im[i] < 0) continue;
350: #if defined(PETSC_USE_DEBUG)
351: if (im[i] >= baij->Mbs) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large, row %D max %D",im[i],baij->Mbs-1);
352: #endif
353: if (im[i] >= rstart && im[i] < rend) {
354: row = im[i] - rstart;
355: for (j=0; j<n; j++) {
356: /* If NumCol = 1 then a copy is not required */
357: if ((roworiented) && (n == 1)) {
358: barray = (MatScalar*) v + i*bs2;
359: } else if((!roworiented) && (m == 1)) {
360: barray = (MatScalar*) v + j*bs2;
361: } else { /* Here a copy is required */
362: if (roworiented) {
363: value = v + i*(stepval+bs)*bs + j*bs;
364: } else {
365: value = v + j*(stepval+bs)*bs + i*bs;
366: }
367: for (ii=0; ii<bs; ii++,value+=stepval) {
368: for (jj=0; jj<bs; jj++) {
369: *barray++ = *value++;
370: }
371: }
372: barray -=bs2;
373: }
374:
375: if (in[j] >= cstart && in[j] < cend){
376: col = in[j] - cstart;
377: MatSetValuesBlocked_SeqSBAIJ(baij->A,1,&row,1,&col,barray,addv);
378: }
379: else if (in[j] < 0) continue;
380: #if defined(PETSC_USE_DEBUG)
381: else if (in[j] >= baij->Nbs) {SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large, col %D max %D",in[j],baij->Nbs-1);}
382: #endif
383: else {
384: if (mat->was_assembled) {
385: if (!baij->colmap) {
386: CreateColmap_MPIBAIJ_Private(mat);
387: }
389: #if defined(PETSC_USE_DEBUG)
390: #if defined (PETSC_USE_CTABLE)
391: { PetscInt data;
392: PetscTableFind(baij->colmap,in[j]+1,&data);
393: if ((data - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap");
394: }
395: #else
396: if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap");
397: #endif
398: #endif
399: #if defined (PETSC_USE_CTABLE)
400: PetscTableFind(baij->colmap,in[j]+1,&col);
401: col = (col - 1)/bs;
402: #else
403: col = (baij->colmap[in[j]] - 1)/bs;
404: #endif
405: if (col < 0 && !((Mat_SeqBAIJ*)(baij->A->data))->nonew) {
406: DisAssemble_MPISBAIJ(mat);
407: col = in[j];
408: }
409: }
410: else col = in[j];
411: MatSetValuesBlocked_SeqBAIJ(baij->B,1,&row,1,&col,barray,addv);
412: }
413: }
414: } else {
415: if (!baij->donotstash) {
416: if (roworiented) {
417: MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
418: } else {
419: MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
420: }
421: }
422: }
423: }
424: return(0);
425: }
429: PetscErrorCode MatGetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[])
430: {
431: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
433: PetscInt bs=mat->rmap.bs,i,j,bsrstart = mat->rmap.rstart,bsrend = mat->rmap.rend;
434: PetscInt bscstart = mat->cmap.rstart,bscend = mat->cmap.rend,row,col,data;
437: for (i=0; i<m; i++) {
438: if (idxm[i] < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",idxm[i]);
439: if (idxm[i] >= mat->rmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",idxm[i],mat->rmap.N-1);
440: if (idxm[i] >= bsrstart && idxm[i] < bsrend) {
441: row = idxm[i] - bsrstart;
442: for (j=0; j<n; j++) {
443: if (idxn[j] < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative column %D",idxn[j]);
444: if (idxn[j] >= mat->cmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",idxn[j],mat->cmap.N-1);
445: if (idxn[j] >= bscstart && idxn[j] < bscend){
446: col = idxn[j] - bscstart;
447: MatGetValues_SeqSBAIJ(baij->A,1,&row,1,&col,v+i*n+j);
448: } else {
449: if (!baij->colmap) {
450: CreateColmap_MPIBAIJ_Private(mat);
451: }
452: #if defined (PETSC_USE_CTABLE)
453: PetscTableFind(baij->colmap,idxn[j]/bs+1,&data);
454: data --;
455: #else
456: data = baij->colmap[idxn[j]/bs]-1;
457: #endif
458: if((data < 0) || (baij->garray[data/bs] != idxn[j]/bs)) *(v+i*n+j) = 0.0;
459: else {
460: col = data + idxn[j]%bs;
461: MatGetValues_SeqBAIJ(baij->B,1,&row,1,&col,v+i*n+j);
462: }
463: }
464: }
465: } else {
466: SETERRQ(PETSC_ERR_SUP,"Only local values currently supported");
467: }
468: }
469: return(0);
470: }
474: PetscErrorCode MatNorm_MPISBAIJ(Mat mat,NormType type,PetscReal *norm)
475: {
476: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
478: PetscReal sum[2],*lnorm2;
481: if (baij->size == 1) {
482: MatNorm(baij->A,type,norm);
483: } else {
484: if (type == NORM_FROBENIUS) {
485: PetscMalloc(2*sizeof(PetscReal),&lnorm2);
486: MatNorm(baij->A,type,lnorm2);
487: *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2++; /* squar power of norm(A) */
488: MatNorm(baij->B,type,lnorm2);
489: *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2--; /* squar power of norm(B) */
490: MPI_Allreduce(lnorm2,&sum,2,MPIU_REAL,MPI_SUM,mat->comm);
491: *norm = sqrt(sum[0] + 2*sum[1]);
492: PetscFree(lnorm2);
493: } else if (type == NORM_INFINITY || type == NORM_1) { /* max row/column sum */
494: Mat_SeqSBAIJ *amat=(Mat_SeqSBAIJ*)baij->A->data;
495: Mat_SeqBAIJ *bmat=(Mat_SeqBAIJ*)baij->B->data;
496: PetscReal *rsum,*rsum2,vabs;
497: PetscInt *jj,*garray=baij->garray,rstart=baij->rstartbs,nz;
498: PetscInt brow,bcol,col,bs=baij->A->rmap.bs,row,grow,gcol,mbs=amat->mbs;
499: MatScalar *v;
501: PetscMalloc((2*mat->cmap.N+1)*sizeof(PetscReal),&rsum);
502: rsum2 = rsum + mat->cmap.N;
503: PetscMemzero(rsum,mat->cmap.N*sizeof(PetscReal));
504: /* Amat */
505: v = amat->a; jj = amat->j;
506: for (brow=0; brow<mbs; brow++) {
507: grow = bs*(rstart + brow);
508: nz = amat->i[brow+1] - amat->i[brow];
509: for (bcol=0; bcol<nz; bcol++){
510: gcol = bs*(rstart + *jj); jj++;
511: for (col=0; col<bs; col++){
512: for (row=0; row<bs; row++){
513: vabs = PetscAbsScalar(*v); v++;
514: rsum[gcol+col] += vabs;
515: /* non-diagonal block */
516: if (bcol > 0 && vabs > 0.0) rsum[grow+row] += vabs;
517: }
518: }
519: }
520: }
521: /* Bmat */
522: v = bmat->a; jj = bmat->j;
523: for (brow=0; brow<mbs; brow++) {
524: grow = bs*(rstart + brow);
525: nz = bmat->i[brow+1] - bmat->i[brow];
526: for (bcol=0; bcol<nz; bcol++){
527: gcol = bs*garray[*jj]; jj++;
528: for (col=0; col<bs; col++){
529: for (row=0; row<bs; row++){
530: vabs = PetscAbsScalar(*v); v++;
531: rsum[gcol+col] += vabs;
532: rsum[grow+row] += vabs;
533: }
534: }
535: }
536: }
537: MPI_Allreduce(rsum,rsum2,mat->cmap.N,MPIU_REAL,MPI_SUM,mat->comm);
538: *norm = 0.0;
539: for (col=0; col<mat->cmap.N; col++) {
540: if (rsum2[col] > *norm) *norm = rsum2[col];
541: }
542: PetscFree(rsum);
543: } else {
544: SETERRQ(PETSC_ERR_SUP,"No support for this norm yet");
545: }
546: }
547: return(0);
548: }
552: PetscErrorCode MatAssemblyBegin_MPISBAIJ(Mat mat,MatAssemblyType mode)
553: {
554: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
556: PetscInt nstash,reallocs;
557: InsertMode addv;
560: if (baij->donotstash) {
561: return(0);
562: }
564: /* make sure all processors are either in INSERTMODE or ADDMODE */
565: MPI_Allreduce(&mat->insertmode,&addv,1,MPI_INT,MPI_BOR,mat->comm);
566: if (addv == (ADD_VALUES|INSERT_VALUES)) {
567: SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Some processors inserted others added");
568: }
569: mat->insertmode = addv; /* in case this processor had no cache */
571: MatStashScatterBegin_Private(&mat->stash,mat->rmap.range);
572: MatStashScatterBegin_Private(&mat->bstash,baij->rangebs);
573: MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);
574: PetscInfo2(0,"Stash has %D entries,uses %D mallocs.\n",nstash,reallocs);
575: MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);
576: PetscInfo2(0,"Block-Stash has %D entries, uses %D mallocs.\n",nstash,reallocs);
577: return(0);
578: }
582: PetscErrorCode MatAssemblyEnd_MPISBAIJ(Mat mat,MatAssemblyType mode)
583: {
584: Mat_MPISBAIJ *baij=(Mat_MPISBAIJ*)mat->data;
585: Mat_SeqSBAIJ *a=(Mat_SeqSBAIJ*)baij->A->data;
587: PetscInt i,j,rstart,ncols,flg,bs2=baij->bs2;
588: PetscInt *row,*col,other_disassembled;
589: PetscMPIInt n;
590: PetscTruth r1,r2,r3;
591: MatScalar *val;
592: InsertMode addv = mat->insertmode;
594: /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */
597: if (!baij->donotstash) {
598: while (1) {
599: MatStashScatterGetMesg_Private(&mat->stash,&n,&row,&col,&val,&flg);
600: if (!flg) break;
602: for (i=0; i<n;) {
603: /* Now identify the consecutive vals belonging to the same row */
604: for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; }
605: if (j < n) ncols = j-i;
606: else ncols = n-i;
607: /* Now assemble all these values with a single function call */
608: MatSetValues_MPISBAIJ_MatScalar(mat,1,row+i,ncols,col+i,val+i,addv);
609: i = j;
610: }
611: }
612: MatStashScatterEnd_Private(&mat->stash);
613: /* Now process the block-stash. Since the values are stashed column-oriented,
614: set the roworiented flag to column oriented, and after MatSetValues()
615: restore the original flags */
616: r1 = baij->roworiented;
617: r2 = a->roworiented;
618: r3 = ((Mat_SeqBAIJ*)baij->B->data)->roworiented;
619: baij->roworiented = PETSC_FALSE;
620: a->roworiented = PETSC_FALSE;
621: ((Mat_SeqBAIJ*)baij->B->data)->roworiented = PETSC_FALSE; /* b->roworinted */
622: while (1) {
623: MatStashScatterGetMesg_Private(&mat->bstash,&n,&row,&col,&val,&flg);
624: if (!flg) break;
625:
626: for (i=0; i<n;) {
627: /* Now identify the consecutive vals belonging to the same row */
628: for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; }
629: if (j < n) ncols = j-i;
630: else ncols = n-i;
631: MatSetValuesBlocked_MPISBAIJ_MatScalar(mat,1,row+i,ncols,col+i,val+i*bs2,addv);
632: i = j;
633: }
634: }
635: MatStashScatterEnd_Private(&mat->bstash);
636: baij->roworiented = r1;
637: a->roworiented = r2;
638: ((Mat_SeqBAIJ*)baij->B->data)->roworiented = r3; /* b->roworinted */
639: }
641: MatAssemblyBegin(baij->A,mode);
642: MatAssemblyEnd(baij->A,mode);
644: /* determine if any processor has disassembled, if so we must
645: also disassemble ourselfs, in order that we may reassemble. */
646: /*
647: if nonzero structure of submatrix B cannot change then we know that
648: no processor disassembled thus we can skip this stuff
649: */
650: if (!((Mat_SeqBAIJ*)baij->B->data)->nonew) {
651: MPI_Allreduce(&mat->was_assembled,&other_disassembled,1,MPI_INT,MPI_PROD,mat->comm);
652: if (mat->was_assembled && !other_disassembled) {
653: DisAssemble_MPISBAIJ(mat);
654: }
655: }
657: if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) {
658: MatSetUpMultiply_MPISBAIJ(mat); /* setup Mvctx and sMvctx */
659: }
660: ((Mat_SeqBAIJ*)baij->B->data)->compressedrow.use = PETSC_TRUE; /* b->compressedrow.use */
661: MatAssemblyBegin(baij->B,mode);
662: MatAssemblyEnd(baij->B,mode);
663:
664: PetscFree(baij->rowvalues);
665: baij->rowvalues = 0;
667: return(0);
668: }
672: static PetscErrorCode MatView_MPISBAIJ_ASCIIorDraworSocket(Mat mat,PetscViewer viewer)
673: {
674: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
675: PetscErrorCode ierr;
676: PetscInt bs = mat->rmap.bs;
677: PetscMPIInt size = baij->size,rank = baij->rank;
678: PetscTruth iascii,isdraw;
679: PetscViewer sviewer;
680: PetscViewerFormat format;
683: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);
684: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);
685: if (iascii) {
686: PetscViewerGetFormat(viewer,&format);
687: if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
688: MatInfo info;
689: MPI_Comm_rank(mat->comm,&rank);
690: MatGetInfo(mat,MAT_LOCAL,&info);
691: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] Local rows %D nz %D nz alloced %D bs %D mem %D\n",
692: rank,mat->rmap.N,(PetscInt)info.nz_used*bs,(PetscInt)info.nz_allocated*bs,
693: mat->rmap.bs,(PetscInt)info.memory);
694: MatGetInfo(baij->A,MAT_LOCAL,&info);
695: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] on-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);
696: MatGetInfo(baij->B,MAT_LOCAL,&info);
697: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] off-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);
698: PetscViewerFlush(viewer);
699: VecScatterView(baij->Mvctx,viewer);
700: return(0);
701: } else if (format == PETSC_VIEWER_ASCII_INFO) {
702: PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);
703: return(0);
704: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
705: return(0);
706: }
707: }
709: if (isdraw) {
710: PetscDraw draw;
711: PetscTruth isnull;
712: PetscViewerDrawGetDraw(viewer,0,&draw);
713: PetscDrawIsNull(draw,&isnull); if (isnull) return(0);
714: }
716: if (size == 1) {
717: PetscObjectSetName((PetscObject)baij->A,mat->name);
718: MatView(baij->A,viewer);
719: } else {
720: /* assemble the entire matrix onto first processor. */
721: Mat A;
722: Mat_SeqSBAIJ *Aloc;
723: Mat_SeqBAIJ *Bloc;
724: PetscInt M = mat->rmap.N,N = mat->cmap.N,*ai,*aj,col,i,j,k,*rvals,mbs = baij->mbs;
725: MatScalar *a;
727: /* Should this be the same type as mat? */
728: MatCreate(mat->comm,&A);
729: if (!rank) {
730: MatSetSizes(A,M,N,M,N);
731: } else {
732: MatSetSizes(A,0,0,M,N);
733: }
734: MatSetType(A,MATMPISBAIJ);
735: MatMPISBAIJSetPreallocation(A,mat->rmap.bs,0,PETSC_NULL,0,PETSC_NULL);
736: PetscLogObjectParent(mat,A);
738: /* copy over the A part */
739: Aloc = (Mat_SeqSBAIJ*)baij->A->data;
740: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
741: PetscMalloc(bs*sizeof(PetscInt),&rvals);
743: for (i=0; i<mbs; i++) {
744: rvals[0] = mat->rmap.rstart + bs*i;
745: for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; }
746: for (j=ai[i]; j<ai[i+1]; j++) {
747: col = mat->cmap.rstart+aj[j]*bs;
748: for (k=0; k<bs; k++) {
749: MatSetValues_MPISBAIJ_MatScalar(A,bs,rvals,1,&col,a,INSERT_VALUES);
750: col++; a += bs;
751: }
752: }
753: }
754: /* copy over the B part */
755: Bloc = (Mat_SeqBAIJ*)baij->B->data;
756: ai = Bloc->i; aj = Bloc->j; a = Bloc->a;
757: for (i=0; i<mbs; i++) {
758: rvals[0] = mat->rmap.rstart + bs;
759: for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; }
760: for (j=ai[i]; j<ai[i+1]; j++) {
761: col = baij->garray[aj[j]]*bs;
762: for (k=0; k<bs; k++) {
763: MatSetValues_MPISBAIJ_MatScalar(A,bs,rvals,1,&col,a,INSERT_VALUES);
764: col++; a += bs;
765: }
766: }
767: }
768: PetscFree(rvals);
769: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
770: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
771: /*
772: Everyone has to call to draw the matrix since the graphics waits are
773: synchronized across all processors that share the PetscDraw object
774: */
775: PetscViewerGetSingleton(viewer,&sviewer);
776: if (!rank) {
777: PetscObjectSetName((PetscObject)((Mat_MPISBAIJ*)(A->data))->A,mat->name);
778: MatView(((Mat_MPISBAIJ*)(A->data))->A,sviewer);
779: }
780: PetscViewerRestoreSingleton(viewer,&sviewer);
781: MatDestroy(A);
782: }
783: return(0);
784: }
788: PetscErrorCode MatView_MPISBAIJ(Mat mat,PetscViewer viewer)
789: {
791: PetscTruth iascii,isdraw,issocket,isbinary;
794: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);
795: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);
796: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_SOCKET,&issocket);
797: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_BINARY,&isbinary);
798: if (iascii || isdraw || issocket || isbinary) {
799: MatView_MPISBAIJ_ASCIIorDraworSocket(mat,viewer);
800: } else {
801: SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported by MPISBAIJ matrices",((PetscObject)viewer)->type_name);
802: }
803: return(0);
804: }
808: PetscErrorCode MatDestroy_MPISBAIJ(Mat mat)
809: {
810: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
814: #if defined(PETSC_USE_LOG)
815: PetscLogObjectState((PetscObject)mat,"Rows=%D,Cols=%D",mat->rmap.N,mat->cmap.N);
816: #endif
817: MatStashDestroy_Private(&mat->stash);
818: MatStashDestroy_Private(&mat->bstash);
819: MatDestroy(baij->A);
820: MatDestroy(baij->B);
821: #if defined (PETSC_USE_CTABLE)
822: if (baij->colmap) {PetscTableDelete(baij->colmap);}
823: #else
824: PetscFree(baij->colmap);
825: #endif
826: PetscFree(baij->garray);
827: if (baij->lvec) {VecDestroy(baij->lvec);}
828: if (baij->Mvctx) {VecScatterDestroy(baij->Mvctx);}
829: if (baij->slvec0) {
830: VecDestroy(baij->slvec0);
831: VecDestroy(baij->slvec0b);
832: }
833: if (baij->slvec1) {
834: VecDestroy(baij->slvec1);
835: VecDestroy(baij->slvec1a);
836: VecDestroy(baij->slvec1b);
837: }
838: if (baij->sMvctx) {VecScatterDestroy(baij->sMvctx);}
839: PetscFree(baij->rowvalues);
840: PetscFree(baij->barray);
841: PetscFree(baij->hd);
842: #if defined(PETSC_USE_MAT_SINGLE)
843: PetscFree(baij->setvaluescopy);
844: #endif
845: PetscFree(baij->rangebs);
846: PetscFree(baij);
848: PetscObjectComposeFunction((PetscObject)mat,"MatStoreValues_C","",PETSC_NULL);
849: PetscObjectComposeFunction((PetscObject)mat,"MatRetrieveValues_C","",PETSC_NULL);
850: PetscObjectComposeFunction((PetscObject)mat,"MatGetDiagonalBlock_C","",PETSC_NULL);
851: PetscObjectComposeFunction((PetscObject)mat,"MatMPISBAIJSetPreallocation_C","",PETSC_NULL);
852: return(0);
853: }
857: PetscErrorCode MatMult_MPISBAIJ(Mat A,Vec xx,Vec yy)
858: {
859: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
861: PetscInt nt,mbs=a->mbs,bs=A->rmap.bs;
862: PetscScalar *x,*from,zero=0.0;
863:
865: VecGetLocalSize(xx,&nt);
866: if (nt != A->cmap.n) {
867: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
868: }
870: /* diagonal part */
871: (*a->A->ops->mult)(a->A,xx,a->slvec1a);
872: VecSet(a->slvec1b,zero);
874: /* subdiagonal part */
875: (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);
876: CHKMEMQ;
877: /* copy x into the vec slvec0 */
878: VecGetArray(a->slvec0,&from);
879: VecGetArray(xx,&x);
880: CHKMEMQ;
881: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
882: CHKMEMQ;
883: VecRestoreArray(a->slvec0,&from);
884:
885: CHKMEMQ;
886: VecScatterBegin(a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD,a->sMvctx);
887: CHKMEMQ;
888: VecRestoreArray(xx,&x);
889: CHKMEMQ;
890: VecScatterEnd(a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD,a->sMvctx);
891: CHKMEMQ;
892: /* supperdiagonal part */
893: (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);
894: CHKMEMQ;
895: return(0);
896: }
900: PetscErrorCode MatMult_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy)
901: {
902: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
904: PetscInt nt;
907: VecGetLocalSize(xx,&nt);
908: if (nt != A->cmap.n) {
909: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
910: }
911: VecGetLocalSize(yy,&nt);
912: if (nt != A->rmap.N) {
913: SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible parition of A and yy");
914: }
916: VecScatterBegin(xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD,a->Mvctx);
917: /* do diagonal part */
918: (*a->A->ops->mult)(a->A,xx,yy);
919: /* do supperdiagonal part */
920: VecScatterEnd(xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD,a->Mvctx);
921: (*a->B->ops->multadd)(a->B,a->lvec,yy,yy);
922: /* do subdiagonal part */
923: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
924: VecScatterBegin(a->lvec,yy,ADD_VALUES,SCATTER_REVERSE,a->Mvctx);
925: VecScatterEnd(a->lvec,yy,ADD_VALUES,SCATTER_REVERSE,a->Mvctx);
927: return(0);
928: }
932: PetscErrorCode MatMultAdd_MPISBAIJ(Mat A,Vec xx,Vec yy,Vec zz)
933: {
934: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
936: PetscInt mbs=a->mbs,bs=A->rmap.bs;
937: PetscScalar *x,*from,zero=0.0;
938:
940: /*
941: PetscSynchronizedPrintf(A->comm," MatMultAdd is called ...\n");
942: PetscSynchronizedFlush(A->comm);
943: */
944: /* diagonal part */
945: (*a->A->ops->multadd)(a->A,xx,yy,a->slvec1a);
946: VecSet(a->slvec1b,zero);
948: /* subdiagonal part */
949: (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);
951: /* copy x into the vec slvec0 */
952: VecGetArray(a->slvec0,&from);
953: VecGetArray(xx,&x);
954: PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));
955: VecRestoreArray(a->slvec0,&from);
956:
957: VecScatterBegin(a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD,a->sMvctx);
958: VecRestoreArray(xx,&x);
959: VecScatterEnd(a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD,a->sMvctx);
960:
961: /* supperdiagonal part */
962: (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,zz);
963:
964: return(0);
965: }
969: PetscErrorCode MatMultAdd_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy,Vec zz)
970: {
971: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
975: VecScatterBegin(xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD,a->Mvctx);
976: /* do diagonal part */
977: (*a->A->ops->multadd)(a->A,xx,yy,zz);
978: /* do supperdiagonal part */
979: VecScatterEnd(xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD,a->Mvctx);
980: (*a->B->ops->multadd)(a->B,a->lvec,zz,zz);
982: /* do subdiagonal part */
983: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
984: VecScatterBegin(a->lvec,zz,ADD_VALUES,SCATTER_REVERSE,a->Mvctx);
985: VecScatterEnd(a->lvec,zz,ADD_VALUES,SCATTER_REVERSE,a->Mvctx);
987: return(0);
988: }
990: /*
991: This only works correctly for square matrices where the subblock A->A is the
992: diagonal block
993: */
996: PetscErrorCode MatGetDiagonal_MPISBAIJ(Mat A,Vec v)
997: {
998: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1002: /* if (a->rmap.N != a->cmap.N) SETERRQ(PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block"); */
1003: MatGetDiagonal(a->A,v);
1004: return(0);
1005: }
1009: PetscErrorCode MatScale_MPISBAIJ(Mat A,PetscScalar aa)
1010: {
1011: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1015: MatScale(a->A,aa);
1016: MatScale(a->B,aa);
1017: return(0);
1018: }
1022: PetscErrorCode MatGetRow_MPISBAIJ(Mat matin,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1023: {
1024: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
1025: PetscScalar *vworkA,*vworkB,**pvA,**pvB,*v_p;
1027: PetscInt bs = matin->rmap.bs,bs2 = mat->bs2,i,*cworkA,*cworkB,**pcA,**pcB;
1028: PetscInt nztot,nzA,nzB,lrow,brstart = matin->rmap.rstart,brend = matin->rmap.rend;
1029: PetscInt *cmap,*idx_p,cstart = mat->rstartbs;
1032: if (mat->getrowactive == PETSC_TRUE) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Already active");
1033: mat->getrowactive = PETSC_TRUE;
1035: if (!mat->rowvalues && (idx || v)) {
1036: /*
1037: allocate enough space to hold information from the longest row.
1038: */
1039: Mat_SeqSBAIJ *Aa = (Mat_SeqSBAIJ*)mat->A->data;
1040: Mat_SeqBAIJ *Ba = (Mat_SeqBAIJ*)mat->B->data;
1041: PetscInt max = 1,mbs = mat->mbs,tmp;
1042: for (i=0; i<mbs; i++) {
1043: tmp = Aa->i[i+1] - Aa->i[i] + Ba->i[i+1] - Ba->i[i]; /* row length */
1044: if (max < tmp) { max = tmp; }
1045: }
1046: PetscMalloc(max*bs2*(sizeof(PetscInt)+sizeof(PetscScalar)),&mat->rowvalues);
1047: mat->rowindices = (PetscInt*)(mat->rowvalues + max*bs2);
1048: }
1049:
1050: if (row < brstart || row >= brend) SETERRQ(PETSC_ERR_SUP,"Only local rows")
1051: lrow = row - brstart; /* local row index */
1053: pvA = &vworkA; pcA = &cworkA; pvB = &vworkB; pcB = &cworkB;
1054: if (!v) {pvA = 0; pvB = 0;}
1055: if (!idx) {pcA = 0; if (!v) pcB = 0;}
1056: (*mat->A->ops->getrow)(mat->A,lrow,&nzA,pcA,pvA);
1057: (*mat->B->ops->getrow)(mat->B,lrow,&nzB,pcB,pvB);
1058: nztot = nzA + nzB;
1060: cmap = mat->garray;
1061: if (v || idx) {
1062: if (nztot) {
1063: /* Sort by increasing column numbers, assuming A and B already sorted */
1064: PetscInt imark = -1;
1065: if (v) {
1066: *v = v_p = mat->rowvalues;
1067: for (i=0; i<nzB; i++) {
1068: if (cmap[cworkB[i]/bs] < cstart) v_p[i] = vworkB[i];
1069: else break;
1070: }
1071: imark = i;
1072: for (i=0; i<nzA; i++) v_p[imark+i] = vworkA[i];
1073: for (i=imark; i<nzB; i++) v_p[nzA+i] = vworkB[i];
1074: }
1075: if (idx) {
1076: *idx = idx_p = mat->rowindices;
1077: if (imark > -1) {
1078: for (i=0; i<imark; i++) {
1079: idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs;
1080: }
1081: } else {
1082: for (i=0; i<nzB; i++) {
1083: if (cmap[cworkB[i]/bs] < cstart)
1084: idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ;
1085: else break;
1086: }
1087: imark = i;
1088: }
1089: for (i=0; i<nzA; i++) idx_p[imark+i] = cstart*bs + cworkA[i];
1090: for (i=imark; i<nzB; i++) idx_p[nzA+i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ;
1091: }
1092: } else {
1093: if (idx) *idx = 0;
1094: if (v) *v = 0;
1095: }
1096: }
1097: *nz = nztot;
1098: (*mat->A->ops->restorerow)(mat->A,lrow,&nzA,pcA,pvA);
1099: (*mat->B->ops->restorerow)(mat->B,lrow,&nzB,pcB,pvB);
1100: return(0);
1101: }
1105: PetscErrorCode MatRestoreRow_MPISBAIJ(Mat mat,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1106: {
1107: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
1110: if (!baij->getrowactive) {
1111: SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"MatGetRow() must be called first");
1112: }
1113: baij->getrowactive = PETSC_FALSE;
1114: return(0);
1115: }
1119: PetscErrorCode MatGetRowUpperTriangular_MPISBAIJ(Mat A)
1120: {
1121: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1122: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1125: aA->getrow_utriangular = PETSC_TRUE;
1126: return(0);
1127: }
1130: PetscErrorCode MatRestoreRowUpperTriangular_MPISBAIJ(Mat A)
1131: {
1132: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1133: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1136: aA->getrow_utriangular = PETSC_FALSE;
1137: return(0);
1138: }
1142: PetscErrorCode MatRealPart_MPISBAIJ(Mat A)
1143: {
1144: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1148: MatRealPart(a->A);
1149: MatRealPart(a->B);
1150: return(0);
1151: }
1155: PetscErrorCode MatImaginaryPart_MPISBAIJ(Mat A)
1156: {
1157: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1161: MatImaginaryPart(a->A);
1162: MatImaginaryPart(a->B);
1163: return(0);
1164: }
1168: PetscErrorCode MatZeroEntries_MPISBAIJ(Mat A)
1169: {
1170: Mat_MPISBAIJ *l = (Mat_MPISBAIJ*)A->data;
1174: MatZeroEntries(l->A);
1175: MatZeroEntries(l->B);
1176: return(0);
1177: }
1181: PetscErrorCode MatGetInfo_MPISBAIJ(Mat matin,MatInfoType flag,MatInfo *info)
1182: {
1183: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)matin->data;
1184: Mat A = a->A,B = a->B;
1186: PetscReal isend[5],irecv[5];
1189: info->block_size = (PetscReal)matin->rmap.bs;
1190: MatGetInfo(A,MAT_LOCAL,info);
1191: isend[0] = info->nz_used; isend[1] = info->nz_allocated; isend[2] = info->nz_unneeded;
1192: isend[3] = info->memory; isend[4] = info->mallocs;
1193: MatGetInfo(B,MAT_LOCAL,info);
1194: isend[0] += info->nz_used; isend[1] += info->nz_allocated; isend[2] += info->nz_unneeded;
1195: isend[3] += info->memory; isend[4] += info->mallocs;
1196: if (flag == MAT_LOCAL) {
1197: info->nz_used = isend[0];
1198: info->nz_allocated = isend[1];
1199: info->nz_unneeded = isend[2];
1200: info->memory = isend[3];
1201: info->mallocs = isend[4];
1202: } else if (flag == MAT_GLOBAL_MAX) {
1203: MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_MAX,matin->comm);
1204: info->nz_used = irecv[0];
1205: info->nz_allocated = irecv[1];
1206: info->nz_unneeded = irecv[2];
1207: info->memory = irecv[3];
1208: info->mallocs = irecv[4];
1209: } else if (flag == MAT_GLOBAL_SUM) {
1210: MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_SUM,matin->comm);
1211: info->nz_used = irecv[0];
1212: info->nz_allocated = irecv[1];
1213: info->nz_unneeded = irecv[2];
1214: info->memory = irecv[3];
1215: info->mallocs = irecv[4];
1216: } else {
1217: SETERRQ1(PETSC_ERR_ARG_WRONG,"Unknown MatInfoType argument %d",(int)flag);
1218: }
1219: info->rows_global = (PetscReal)A->rmap.N;
1220: info->columns_global = (PetscReal)A->cmap.N;
1221: info->rows_local = (PetscReal)A->rmap.N;
1222: info->columns_local = (PetscReal)A->cmap.N;
1223: info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */
1224: info->fill_ratio_needed = 0;
1225: info->factor_mallocs = 0;
1226: return(0);
1227: }
1231: PetscErrorCode MatSetOption_MPISBAIJ(Mat A,MatOption op)
1232: {
1233: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1234: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data;
1238: switch (op) {
1239: case MAT_NO_NEW_NONZERO_LOCATIONS:
1240: case MAT_YES_NEW_NONZERO_LOCATIONS:
1241: case MAT_COLUMNS_UNSORTED:
1242: case MAT_COLUMNS_SORTED:
1243: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1244: case MAT_KEEP_ZEROED_ROWS:
1245: case MAT_NEW_NONZERO_LOCATION_ERR:
1246: MatSetOption(a->A,op);
1247: MatSetOption(a->B,op);
1248: break;
1249: case MAT_ROW_ORIENTED:
1250: a->roworiented = PETSC_TRUE;
1251: MatSetOption(a->A,op);
1252: MatSetOption(a->B,op);
1253: break;
1254: case MAT_ROWS_SORTED:
1255: case MAT_ROWS_UNSORTED:
1256: case MAT_YES_NEW_DIAGONALS:
1257: PetscInfo(A,"Option ignored\n");
1258: break;
1259: case MAT_COLUMN_ORIENTED:
1260: a->roworiented = PETSC_FALSE;
1261: MatSetOption(a->A,op);
1262: MatSetOption(a->B,op);
1263: break;
1264: case MAT_IGNORE_OFF_PROC_ENTRIES:
1265: a->donotstash = PETSC_TRUE;
1266: break;
1267: case MAT_NO_NEW_DIAGONALS:
1268: SETERRQ(PETSC_ERR_SUP,"MAT_NO_NEW_DIAGONALS");
1269: case MAT_USE_HASH_TABLE:
1270: a->ht_flag = PETSC_TRUE;
1271: break;
1272: case MAT_NOT_SYMMETRIC:
1273: case MAT_NOT_STRUCTURALLY_SYMMETRIC:
1274: case MAT_HERMITIAN:
1275: SETERRQ(PETSC_ERR_SUP,"Matrix must be symmetric");
1276: case MAT_SYMMETRIC:
1277: case MAT_STRUCTURALLY_SYMMETRIC:
1278: case MAT_NOT_HERMITIAN:
1279: case MAT_SYMMETRY_ETERNAL:
1280: case MAT_NOT_SYMMETRY_ETERNAL:
1281: break;
1282: case MAT_IGNORE_LOWER_TRIANGULAR:
1283: aA->ignore_ltriangular = PETSC_TRUE;
1284: break;
1285: case MAT_ERROR_LOWER_TRIANGULAR:
1286: aA->ignore_ltriangular = PETSC_FALSE;
1287: break;
1288: case MAT_GETROW_UPPERTRIANGULAR:
1289: aA->getrow_utriangular = PETSC_TRUE;
1290: break;
1291: default:
1292: SETERRQ(PETSC_ERR_SUP,"unknown option");
1293: }
1294: return(0);
1295: }
1299: PetscErrorCode MatTranspose_MPISBAIJ(Mat A,Mat *B)
1300: {
1303: MatDuplicate(A,MAT_COPY_VALUES,B);
1304: return(0);
1305: }
1309: PetscErrorCode MatDiagonalScale_MPISBAIJ(Mat mat,Vec ll,Vec rr)
1310: {
1311: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data;
1312: Mat a=baij->A, b=baij->B;
1314: PetscInt nv,m,n;
1315: PetscTruth flg;
1318: if (ll != rr){
1319: VecEqual(ll,rr,&flg);
1320: if (!flg)
1321: SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"For symmetric format, left and right scaling vectors must be same\n");
1322: }
1323: if (!ll) return(0);
1325: MatGetLocalSize(mat,&m,&n);
1326: if (m != n) SETERRQ2(PETSC_ERR_ARG_SIZ,"For symmetric format, local size %d %d must be same",m,n);
1327:
1328: VecGetLocalSize(rr,&nv);
1329: if (nv!=n) SETERRQ(PETSC_ERR_ARG_SIZ,"Left and right vector non-conforming local size");
1331: VecScatterBegin(rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD,baij->Mvctx);
1332:
1333: /* left diagonalscale the off-diagonal part */
1334: (*b->ops->diagonalscale)(b,ll,PETSC_NULL);
1335:
1336: /* scale the diagonal part */
1337: (*a->ops->diagonalscale)(a,ll,rr);
1339: /* right diagonalscale the off-diagonal part */
1340: VecScatterEnd(rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD,baij->Mvctx);
1341: (*b->ops->diagonalscale)(b,PETSC_NULL,baij->lvec);
1342: return(0);
1343: }
1347: PetscErrorCode MatPrintHelp_MPISBAIJ(Mat A)
1348: {
1349: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1350: MPI_Comm comm = A->comm;
1351: static PetscTruth called = PETSC_FALSE;
1352: PetscErrorCode ierr;
1355: if (!a->rank) {
1356: MatPrintHelp_SeqSBAIJ(a->A);
1357: }
1358: if (called) {return(0);} else called = PETSC_TRUE;
1359: (*PetscHelpPrintf)(comm," Options for MATMPISBAIJ matrix format (the defaults):\n");
1360: (*PetscHelpPrintf)(comm," -mat_use_hash_table <factor>: Use hashtable for efficient matrix assembly\n");
1361: return(0);
1362: }
1366: PetscErrorCode MatSetUnfactored_MPISBAIJ(Mat A)
1367: {
1368: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
1372: MatSetUnfactored(a->A);
1373: return(0);
1374: }
1376: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat,MatDuplicateOption,Mat *);
1380: PetscErrorCode MatEqual_MPISBAIJ(Mat A,Mat B,PetscTruth *flag)
1381: {
1382: Mat_MPISBAIJ *matB = (Mat_MPISBAIJ*)B->data,*matA = (Mat_MPISBAIJ*)A->data;
1383: Mat a,b,c,d;
1384: PetscTruth flg;
1388: a = matA->A; b = matA->B;
1389: c = matB->A; d = matB->B;
1391: MatEqual(a,c,&flg);
1392: if (flg) {
1393: MatEqual(b,d,&flg);
1394: }
1395: MPI_Allreduce(&flg,flag,1,MPI_INT,MPI_LAND,A->comm);
1396: return(0);
1397: }
1401: PetscErrorCode MatCopy_MPISBAIJ(Mat A,Mat B,MatStructure str)
1402: {
1404: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1405: Mat_MPISBAIJ *b = (Mat_MPISBAIJ *)B->data;
1408: /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */
1409: if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) {
1410: MatGetRowUpperTriangular(A);
1411: MatCopy_Basic(A,B,str);
1412: MatRestoreRowUpperTriangular(A);
1413: } else {
1414: MatCopy(a->A,b->A,str);
1415: MatCopy(a->B,b->B,str);
1416: }
1417: return(0);
1418: }
1422: PetscErrorCode MatSetUpPreallocation_MPISBAIJ(Mat A)
1423: {
1427: MatMPISBAIJSetPreallocation(A,PetscMax(A->rmap.bs,1),PETSC_DEFAULT,0,PETSC_DEFAULT,0);
1428: return(0);
1429: }
1431: #include petscblaslapack.h
1434: PetscErrorCode MatAXPY_MPISBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
1435: {
1437: Mat_MPISBAIJ *xx=(Mat_MPISBAIJ *)X->data,*yy=(Mat_MPISBAIJ *)Y->data;
1438: PetscBLASInt bnz,one=1;
1439: Mat_SeqSBAIJ *xa,*ya;
1440: Mat_SeqBAIJ *xb,*yb;
1443: if (str == SAME_NONZERO_PATTERN) {
1444: PetscScalar alpha = a;
1445: xa = (Mat_SeqSBAIJ *)xx->A->data;
1446: ya = (Mat_SeqSBAIJ *)yy->A->data;
1447: bnz = (PetscBLASInt)xa->nz;
1448: BLASaxpy_(&bnz,&alpha,xa->a,&one,ya->a,&one);
1449: xb = (Mat_SeqBAIJ *)xx->B->data;
1450: yb = (Mat_SeqBAIJ *)yy->B->data;
1451: bnz = (PetscBLASInt)xb->nz;
1452: BLASaxpy_(&bnz,&alpha,xb->a,&one,yb->a,&one);
1453: } else {
1454: MatGetRowUpperTriangular(X);
1455: MatAXPY_Basic(Y,a,X,str);
1456: MatRestoreRowUpperTriangular(X);
1457: }
1458: return(0);
1459: }
1463: PetscErrorCode MatGetSubMatrices_MPISBAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[])
1464: {
1466: PetscInt i;
1467: PetscTruth flg;
1470: for (i=0; i<n; i++) {
1471: ISEqual(irow[i],icol[i],&flg);
1472: if (!flg) {
1473: SETERRQ(PETSC_ERR_SUP,"Can only get symmetric submatrix for MPISBAIJ matrices");
1474: }
1475: }
1476: MatGetSubMatrices_MPIBAIJ(A,n,irow,icol,scall,B);
1477: return(0);
1478: }
1479:
1481: /* -------------------------------------------------------------------*/
1482: static struct _MatOps MatOps_Values = {
1483: MatSetValues_MPISBAIJ,
1484: MatGetRow_MPISBAIJ,
1485: MatRestoreRow_MPISBAIJ,
1486: MatMult_MPISBAIJ,
1487: /* 4*/ MatMultAdd_MPISBAIJ,
1488: MatMult_MPISBAIJ, /* transpose versions are same as non-transpose */
1489: MatMultAdd_MPISBAIJ,
1490: 0,
1491: 0,
1492: 0,
1493: /*10*/ 0,
1494: 0,
1495: 0,
1496: MatRelax_MPISBAIJ,
1497: MatTranspose_MPISBAIJ,
1498: /*15*/ MatGetInfo_MPISBAIJ,
1499: MatEqual_MPISBAIJ,
1500: MatGetDiagonal_MPISBAIJ,
1501: MatDiagonalScale_MPISBAIJ,
1502: MatNorm_MPISBAIJ,
1503: /*20*/ MatAssemblyBegin_MPISBAIJ,
1504: MatAssemblyEnd_MPISBAIJ,
1505: 0,
1506: MatSetOption_MPISBAIJ,
1507: MatZeroEntries_MPISBAIJ,
1508: /*25*/ 0,
1509: 0,
1510: 0,
1511: 0,
1512: 0,
1513: /*30*/ MatSetUpPreallocation_MPISBAIJ,
1514: 0,
1515: 0,
1516: 0,
1517: 0,
1518: /*35*/ MatDuplicate_MPISBAIJ,
1519: 0,
1520: 0,
1521: 0,
1522: 0,
1523: /*40*/ MatAXPY_MPISBAIJ,
1524: MatGetSubMatrices_MPISBAIJ,
1525: MatIncreaseOverlap_MPISBAIJ,
1526: MatGetValues_MPISBAIJ,
1527: MatCopy_MPISBAIJ,
1528: /*45*/ MatPrintHelp_MPISBAIJ,
1529: MatScale_MPISBAIJ,
1530: 0,
1531: 0,
1532: 0,
1533: /*50*/ 0,
1534: 0,
1535: 0,
1536: 0,
1537: 0,
1538: /*55*/ 0,
1539: 0,
1540: MatSetUnfactored_MPISBAIJ,
1541: 0,
1542: MatSetValuesBlocked_MPISBAIJ,
1543: /*60*/ 0,
1544: 0,
1545: 0,
1546: 0,
1547: 0,
1548: /*65*/ 0,
1549: 0,
1550: 0,
1551: 0,
1552: 0,
1553: /*70*/ MatGetRowMax_MPISBAIJ,
1554: 0,
1555: 0,
1556: 0,
1557: 0,
1558: /*75*/ 0,
1559: 0,
1560: 0,
1561: 0,
1562: 0,
1563: /*80*/ 0,
1564: 0,
1565: 0,
1566: 0,
1567: MatLoad_MPISBAIJ,
1568: /*85*/ 0,
1569: 0,
1570: 0,
1571: 0,
1572: 0,
1573: /*90*/ 0,
1574: 0,
1575: 0,
1576: 0,
1577: 0,
1578: /*95*/ 0,
1579: 0,
1580: 0,
1581: 0,
1582: 0,
1583: /*100*/0,
1584: 0,
1585: 0,
1586: 0,
1587: 0,
1588: /*105*/0,
1589: MatRealPart_MPISBAIJ,
1590: MatImaginaryPart_MPISBAIJ,
1591: MatGetRowUpperTriangular_MPISBAIJ,
1592: MatRestoreRowUpperTriangular_MPISBAIJ
1593: };
1599: PetscErrorCode PETSCMAT_DLLEXPORT MatGetDiagonalBlock_MPISBAIJ(Mat A,PetscTruth *iscopy,MatReuse reuse,Mat *a)
1600: {
1602: *a = ((Mat_MPISBAIJ *)A->data)->A;
1603: *iscopy = PETSC_FALSE;
1604: return(0);
1605: }
1611: PetscErrorCode PETSCMAT_DLLEXPORT MatMPISBAIJSetPreallocation_MPISBAIJ(Mat B,PetscInt bs,PetscInt d_nz,PetscInt *d_nnz,PetscInt o_nz,PetscInt *o_nnz)
1612: {
1613: Mat_MPISBAIJ *b;
1615: PetscInt i,mbs,Mbs;
1618: PetscOptionsGetInt(B->prefix,"-mat_block_size",&bs,PETSC_NULL);
1620: if (bs < 1) SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"Invalid block size specified, must be positive");
1621: if (d_nz == PETSC_DECIDE || d_nz == PETSC_DEFAULT) d_nz = 3;
1622: if (o_nz == PETSC_DECIDE || o_nz == PETSC_DEFAULT) o_nz = 1;
1623: if (d_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"d_nz cannot be less than 0: value %D",d_nz);
1624: if (o_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"o_nz cannot be less than 0: value %D",o_nz);
1626: B->rmap.bs = B->cmap.bs = bs;
1627: PetscMapInitialize(B->comm,&B->rmap);
1628: PetscMapInitialize(B->comm,&B->cmap);
1630: if (d_nnz) {
1631: for (i=0; i<B->rmap.n/bs; i++) {
1632: 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]);
1633: }
1634: }
1635: if (o_nnz) {
1636: for (i=0; i<B->rmap.n/bs; i++) {
1637: 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]);
1638: }
1639: }
1640: B->preallocated = PETSC_TRUE;
1642: b = (Mat_MPISBAIJ*)B->data;
1643: mbs = B->rmap.n/bs;
1644: Mbs = B->rmap.N/bs;
1645: if (mbs*bs != B->rmap.n) {
1646: SETERRQ2(PETSC_ERR_ARG_SIZ,"No of local rows %D must be divisible by blocksize %D",B->rmap.N,bs);
1647: }
1649: B->rmap.bs = bs;
1650: b->bs2 = bs*bs;
1651: b->mbs = mbs;
1652: b->nbs = mbs;
1653: b->Mbs = Mbs;
1654: b->Nbs = Mbs;
1656: for (i=0; i<=b->size; i++) {
1657: b->rangebs[i] = B->rmap.range[i]/bs;
1658: }
1659: b->rstartbs = B->rmap.rstart/bs;
1660: b->rendbs = B->rmap.rend/bs;
1661:
1662: b->cstartbs = B->cmap.rstart/bs;
1663: b->cendbs = B->cmap.rend/bs;
1664:
1665: MatCreate(PETSC_COMM_SELF,&b->A);
1666: MatSetSizes(b->A,B->rmap.n,B->cmap.n,B->rmap.n,B->cmap.n);
1667: MatSetType(b->A,MATSEQSBAIJ);
1668: MatSeqSBAIJSetPreallocation(b->A,bs,d_nz,d_nnz);
1669: PetscLogObjectParent(B,b->A);
1671: MatCreate(PETSC_COMM_SELF,&b->B);
1672: MatSetSizes(b->B,B->rmap.n,B->cmap.N,B->rmap.n,B->cmap.N);
1673: MatSetType(b->B,MATSEQBAIJ);
1674: MatSeqBAIJSetPreallocation(b->B,bs,o_nz,o_nnz);
1675: PetscLogObjectParent(B,b->B);
1677: /* build cache for off array entries formed */
1678: MatStashCreate_Private(B->comm,bs,&B->bstash);
1680: return(0);
1681: }
1684: /*MC
1685: MATMPISBAIJ - MATMPISBAIJ = "mpisbaij" - A matrix type to be used for distributed symmetric sparse block matrices,
1686: based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored.
1688: Options Database Keys:
1689: . -mat_type mpisbaij - sets the matrix type to "mpisbaij" during a call to MatSetFromOptions()
1691: Level: beginner
1693: .seealso: MatCreateMPISBAIJ
1694: M*/
1699: PetscErrorCode PETSCMAT_DLLEXPORT MatCreate_MPISBAIJ(Mat B)
1700: {
1701: Mat_MPISBAIJ *b;
1703: PetscTruth flg;
1707: PetscNew(Mat_MPISBAIJ,&b);
1708: B->data = (void*)b;
1709: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
1711: B->ops->destroy = MatDestroy_MPISBAIJ;
1712: B->ops->view = MatView_MPISBAIJ;
1713: B->mapping = 0;
1714: B->factor = 0;
1715: B->assembled = PETSC_FALSE;
1717: B->insertmode = NOT_SET_VALUES;
1718: MPI_Comm_rank(B->comm,&b->rank);
1719: MPI_Comm_size(B->comm,&b->size);
1721: /* build local table of row and column ownerships */
1722: PetscMalloc((b->size+2)*sizeof(PetscInt),&b->rangebs);
1724: /* build cache for off array entries formed */
1725: MatStashCreate_Private(B->comm,1,&B->stash);
1726: b->donotstash = PETSC_FALSE;
1727: b->colmap = PETSC_NULL;
1728: b->garray = PETSC_NULL;
1729: b->roworiented = PETSC_TRUE;
1731: #if defined(PETSC_USE_MAT_SINGLE)
1732: /* stuff for MatSetValues_XXX in single precision */
1733: b->setvalueslen = 0;
1734: b->setvaluescopy = PETSC_NULL;
1735: #endif
1737: /* stuff used in block assembly */
1738: b->barray = 0;
1740: /* stuff used for matrix vector multiply */
1741: b->lvec = 0;
1742: b->Mvctx = 0;
1743: b->slvec0 = 0;
1744: b->slvec0b = 0;
1745: b->slvec1 = 0;
1746: b->slvec1a = 0;
1747: b->slvec1b = 0;
1748: b->sMvctx = 0;
1750: /* stuff for MatGetRow() */
1751: b->rowindices = 0;
1752: b->rowvalues = 0;
1753: b->getrowactive = PETSC_FALSE;
1755: /* hash table stuff */
1756: b->ht = 0;
1757: b->hd = 0;
1758: b->ht_size = 0;
1759: b->ht_flag = PETSC_FALSE;
1760: b->ht_fact = 0;
1761: b->ht_total_ct = 0;
1762: b->ht_insert_ct = 0;
1764: PetscOptionsHasName(B->prefix,"-mat_use_hash_table",&flg);
1765: if (flg) {
1766: PetscReal fact = 1.39;
1767: MatSetOption(B,MAT_USE_HASH_TABLE);
1768: PetscOptionsGetReal(B->prefix,"-mat_use_hash_table",&fact,PETSC_NULL);
1769: if (fact <= 1.0) fact = 1.39;
1770: MatMPIBAIJSetHashTableFactor(B,fact);
1771: PetscInfo1(0,"Hash table Factor used %5.2f\n",fact);
1772: }
1773: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatStoreValues_C",
1774: "MatStoreValues_MPISBAIJ",
1775: MatStoreValues_MPISBAIJ);
1776: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatRetrieveValues_C",
1777: "MatRetrieveValues_MPISBAIJ",
1778: MatRetrieveValues_MPISBAIJ);
1779: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetDiagonalBlock_C",
1780: "MatGetDiagonalBlock_MPISBAIJ",
1781: MatGetDiagonalBlock_MPISBAIJ);
1782: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatMPISBAIJSetPreallocation_C",
1783: "MatMPISBAIJSetPreallocation_MPISBAIJ",
1784: MatMPISBAIJSetPreallocation_MPISBAIJ);
1785: B->symmetric = PETSC_TRUE;
1786: B->structurally_symmetric = PETSC_TRUE;
1787: B->symmetric_set = PETSC_TRUE;
1788: B->structurally_symmetric_set = PETSC_TRUE;
1789: return(0);
1790: }
1793: /*MC
1794: MATSBAIJ - MATSBAIJ = "sbaij" - A matrix type to be used for symmetric block sparse matrices.
1796: This matrix type is identical to MATSEQSBAIJ when constructed with a single process communicator,
1797: and MATMPISBAIJ otherwise.
1799: Options Database Keys:
1800: . -mat_type sbaij - sets the matrix type to "sbaij" during a call to MatSetFromOptions()
1802: Level: beginner
1804: .seealso: MatCreateMPISBAIJ,MATSEQSBAIJ,MATMPISBAIJ
1805: M*/
1810: PetscErrorCode PETSCMAT_DLLEXPORT MatCreate_SBAIJ(Mat A)
1811: {
1813: PetscMPIInt size;
1816: PetscObjectChangeTypeName((PetscObject)A,MATSBAIJ);
1817: MPI_Comm_size(A->comm,&size);
1818: if (size == 1) {
1819: MatSetType(A,MATSEQSBAIJ);
1820: } else {
1821: MatSetType(A,MATMPISBAIJ);
1822: }
1823: return(0);
1824: }
1829: /*@C
1830: MatMPISBAIJSetPreallocation - For good matrix assembly performance
1831: the user should preallocate the matrix storage by setting the parameters
1832: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
1833: performance can be increased by more than a factor of 50.
1835: Collective on Mat
1837: Input Parameters:
1838: + A - the matrix
1839: . bs - size of blockk
1840: . d_nz - number of block nonzeros per block row in diagonal portion of local
1841: submatrix (same for all local rows)
1842: . d_nnz - array containing the number of block nonzeros in the various block rows
1843: in the upper triangular and diagonal part of the in diagonal portion of the local
1844: (possibly different for each block row) or PETSC_NULL. You must leave room
1845: for the diagonal entry even if it is zero.
1846: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
1847: submatrix (same for all local rows).
1848: - o_nnz - array containing the number of nonzeros in the various block rows of the
1849: off-diagonal portion of the local submatrix (possibly different for
1850: each block row) or PETSC_NULL.
1853: Options Database Keys:
1854: . -mat_no_unroll - uses code that does not unroll the loops in the
1855: block calculations (much slower)
1856: . -mat_block_size - size of the blocks to use
1858: Notes:
1860: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
1861: than it must be used on all processors that share the object for that argument.
1863: If the *_nnz parameter is given then the *_nz parameter is ignored
1865: Storage Information:
1866: For a square global matrix we define each processor's diagonal portion
1867: to be its local rows and the corresponding columns (a square submatrix);
1868: each processor's off-diagonal portion encompasses the remainder of the
1869: local matrix (a rectangular submatrix).
1871: The user can specify preallocated storage for the diagonal part of
1872: the local submatrix with either d_nz or d_nnz (not both). Set
1873: d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic
1874: memory allocation. Likewise, specify preallocated storage for the
1875: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
1877: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
1878: the figure below we depict these three local rows and all columns (0-11).
1880: .vb
1881: 0 1 2 3 4 5 6 7 8 9 10 11
1882: -------------------
1883: row 3 | o o o d d d o o o o o o
1884: row 4 | o o o d d d o o o o o o
1885: row 5 | o o o d d d o o o o o o
1886: -------------------
1887: .ve
1888:
1889: Thus, any entries in the d locations are stored in the d (diagonal)
1890: submatrix, and any entries in the o locations are stored in the
1891: o (off-diagonal) submatrix. Note that the d matrix is stored in
1892: MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format.
1894: Now d_nz should indicate the number of block nonzeros per row in the upper triangular
1895: plus the diagonal part of the d matrix,
1896: and o_nz should indicate the number of block nonzeros per row in the o matrix.
1897: In general, for PDE problems in which most nonzeros are near the diagonal,
1898: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
1899: or you will get TERRIBLE performance; see the users' manual chapter on
1900: matrices.
1902: Level: intermediate
1904: .keywords: matrix, block, aij, compressed row, sparse, parallel
1906: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ()
1907: @*/
1908: PetscErrorCode PETSCMAT_DLLEXPORT MatMPISBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[])
1909: {
1910: PetscErrorCode ierr,(*f)(Mat,PetscInt,PetscInt,const PetscInt[],PetscInt,const PetscInt[]);
1913: PetscObjectQueryFunction((PetscObject)B,"MatMPISBAIJSetPreallocation_C",(void (**)(void))&f);
1914: if (f) {
1915: (*f)(B,bs,d_nz,d_nnz,o_nz,o_nnz);
1916: }
1917: return(0);
1918: }
1922: /*@C
1923: MatCreateMPISBAIJ - Creates a sparse parallel matrix in symmetric block AIJ format
1924: (block compressed row). For good matrix assembly performance
1925: the user should preallocate the matrix storage by setting the parameters
1926: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
1927: performance can be increased by more than a factor of 50.
1929: Collective on MPI_Comm
1931: Input Parameters:
1932: + comm - MPI communicator
1933: . bs - size of blockk
1934: . m - number of local rows (or PETSC_DECIDE to have calculated if M is given)
1935: This value should be the same as the local size used in creating the
1936: y vector for the matrix-vector product y = Ax.
1937: . n - number of local columns (or PETSC_DECIDE to have calculated if N is given)
1938: This value should be the same as the local size used in creating the
1939: x vector for the matrix-vector product y = Ax.
1940: . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given)
1941: . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given)
1942: . d_nz - number of block nonzeros per block row in diagonal portion of local
1943: submatrix (same for all local rows)
1944: . d_nnz - array containing the number of block nonzeros in the various block rows
1945: in the upper triangular portion of the in diagonal portion of the local
1946: (possibly different for each block block row) or PETSC_NULL.
1947: You must leave room for the diagonal entry even if it is zero.
1948: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
1949: submatrix (same for all local rows).
1950: - o_nnz - array containing the number of nonzeros in the various block rows of the
1951: off-diagonal portion of the local submatrix (possibly different for
1952: each block row) or PETSC_NULL.
1954: Output Parameter:
1955: . A - the matrix
1957: Options Database Keys:
1958: . -mat_no_unroll - uses code that does not unroll the loops in the
1959: block calculations (much slower)
1960: . -mat_block_size - size of the blocks to use
1961: . -mat_mpi - use the parallel matrix data structures even on one processor
1962: (defaults to using SeqBAIJ format on one processor)
1964: Notes:
1965: The number of rows and columns must be divisible by blocksize.
1967: The user MUST specify either the local or global matrix dimensions
1968: (possibly both).
1970: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
1971: than it must be used on all processors that share the object for that argument.
1973: If the *_nnz parameter is given then the *_nz parameter is ignored
1975: Storage Information:
1976: For a square global matrix we define each processor's diagonal portion
1977: to be its local rows and the corresponding columns (a square submatrix);
1978: each processor's off-diagonal portion encompasses the remainder of the
1979: local matrix (a rectangular submatrix).
1981: The user can specify preallocated storage for the diagonal part of
1982: the local submatrix with either d_nz or d_nnz (not both). Set
1983: d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic
1984: memory allocation. Likewise, specify preallocated storage for the
1985: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
1987: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
1988: the figure below we depict these three local rows and all columns (0-11).
1990: .vb
1991: 0 1 2 3 4 5 6 7 8 9 10 11
1992: -------------------
1993: row 3 | o o o d d d o o o o o o
1994: row 4 | o o o d d d o o o o o o
1995: row 5 | o o o d d d o o o o o o
1996: -------------------
1997: .ve
1998:
1999: Thus, any entries in the d locations are stored in the d (diagonal)
2000: submatrix, and any entries in the o locations are stored in the
2001: o (off-diagonal) submatrix. Note that the d matrix is stored in
2002: MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format.
2004: Now d_nz should indicate the number of block nonzeros per row in the upper triangular
2005: plus the diagonal part of the d matrix,
2006: and o_nz should indicate the number of block nonzeros per row in the o matrix.
2007: In general, for PDE problems in which most nonzeros are near the diagonal,
2008: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
2009: or you will get TERRIBLE performance; see the users' manual chapter on
2010: matrices.
2012: Level: intermediate
2014: .keywords: matrix, block, aij, compressed row, sparse, parallel
2016: .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ()
2017: @*/
2019: PetscErrorCode PETSCMAT_DLLEXPORT 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)
2020: {
2022: PetscMPIInt size;
2025: MatCreate(comm,A);
2026: MatSetSizes(*A,m,n,M,N);
2027: MPI_Comm_size(comm,&size);
2028: if (size > 1) {
2029: MatSetType(*A,MATMPISBAIJ);
2030: MatMPISBAIJSetPreallocation(*A,bs,d_nz,d_nnz,o_nz,o_nnz);
2031: } else {
2032: MatSetType(*A,MATSEQSBAIJ);
2033: MatSeqSBAIJSetPreallocation(*A,bs,d_nz,d_nnz);
2034: }
2035: return(0);
2036: }
2041: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat matin,MatDuplicateOption cpvalues,Mat *newmat)
2042: {
2043: Mat mat;
2044: Mat_MPISBAIJ *a,*oldmat = (Mat_MPISBAIJ*)matin->data;
2046: PetscInt len=0,nt,bs=matin->rmap.bs,mbs=oldmat->mbs;
2047: PetscScalar *array;
2050: *newmat = 0;
2051: MatCreate(matin->comm,&mat);
2052: MatSetSizes(mat,matin->rmap.n,matin->cmap.n,matin->rmap.N,matin->cmap.N);
2053: MatSetType(mat,matin->type_name);
2054: PetscMemcpy(mat->ops,matin->ops,sizeof(struct _MatOps));
2055: PetscMapCopy(matin->comm,&matin->rmap,&mat->rmap);
2056: PetscMapCopy(matin->comm,&matin->cmap,&mat->cmap);
2057:
2058: mat->factor = matin->factor;
2059: mat->preallocated = PETSC_TRUE;
2060: mat->assembled = PETSC_TRUE;
2061: mat->insertmode = NOT_SET_VALUES;
2063: a = (Mat_MPISBAIJ*)mat->data;
2064: a->bs2 = oldmat->bs2;
2065: a->mbs = oldmat->mbs;
2066: a->nbs = oldmat->nbs;
2067: a->Mbs = oldmat->Mbs;
2068: a->Nbs = oldmat->Nbs;
2071: a->size = oldmat->size;
2072: a->rank = oldmat->rank;
2073: a->donotstash = oldmat->donotstash;
2074: a->roworiented = oldmat->roworiented;
2075: a->rowindices = 0;
2076: a->rowvalues = 0;
2077: a->getrowactive = PETSC_FALSE;
2078: a->barray = 0;
2079: a->rstartbs = oldmat->rstartbs;
2080: a->rendbs = oldmat->rendbs;
2081: a->cstartbs = oldmat->cstartbs;
2082: a->cendbs = oldmat->cendbs;
2084: /* hash table stuff */
2085: a->ht = 0;
2086: a->hd = 0;
2087: a->ht_size = 0;
2088: a->ht_flag = oldmat->ht_flag;
2089: a->ht_fact = oldmat->ht_fact;
2090: a->ht_total_ct = 0;
2091: a->ht_insert_ct = 0;
2092:
2093: PetscMemcpy(a->rangebs,oldmat->rangebs,(a->size+2)*sizeof(PetscInt));
2094: MatStashCreate_Private(matin->comm,1,&mat->stash);
2095: MatStashCreate_Private(matin->comm,matin->rmap.bs,&mat->bstash);
2096: if (oldmat->colmap) {
2097: #if defined (PETSC_USE_CTABLE)
2098: PetscTableCreateCopy(oldmat->colmap,&a->colmap);
2099: #else
2100: PetscMalloc((a->Nbs)*sizeof(PetscInt),&a->colmap);
2101: PetscLogObjectMemory(mat,(a->Nbs)*sizeof(PetscInt));
2102: PetscMemcpy(a->colmap,oldmat->colmap,(a->Nbs)*sizeof(PetscInt));
2103: #endif
2104: } else a->colmap = 0;
2106: if (oldmat->garray && (len = ((Mat_SeqBAIJ*)(oldmat->B->data))->nbs)) {
2107: PetscMalloc(len*sizeof(PetscInt),&a->garray);
2108: PetscLogObjectMemory(mat,len*sizeof(PetscInt));
2109: PetscMemcpy(a->garray,oldmat->garray,len*sizeof(PetscInt));
2110: } else a->garray = 0;
2111:
2112: VecDuplicate(oldmat->lvec,&a->lvec);
2113: PetscLogObjectParent(mat,a->lvec);
2114: VecScatterCopy(oldmat->Mvctx,&a->Mvctx);
2115: PetscLogObjectParent(mat,a->Mvctx);
2117: VecDuplicate(oldmat->slvec0,&a->slvec0);
2118: PetscLogObjectParent(mat,a->slvec0);
2119: VecDuplicate(oldmat->slvec1,&a->slvec1);
2120: PetscLogObjectParent(mat,a->slvec1);
2122: VecGetLocalSize(a->slvec1,&nt);
2123: VecGetArray(a->slvec1,&array);
2124: VecCreateSeqWithArray(PETSC_COMM_SELF,bs*mbs,array,&a->slvec1a);
2125: VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec1b);
2126: VecRestoreArray(a->slvec1,&array);
2127: VecGetArray(a->slvec0,&array);
2128: VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec0b);
2129: VecRestoreArray(a->slvec0,&array);
2130: PetscLogObjectParent(mat,a->slvec0);
2131: PetscLogObjectParent(mat,a->slvec1);
2132: PetscLogObjectParent(mat,a->slvec0b);
2133: PetscLogObjectParent(mat,a->slvec1a);
2134: PetscLogObjectParent(mat,a->slvec1b);
2136: /* VecScatterCopy(oldmat->sMvctx,&a->sMvctx); - not written yet, replaced by the lazy trick: */
2137: PetscObjectReference((PetscObject)oldmat->sMvctx);
2138: a->sMvctx = oldmat->sMvctx;
2139: PetscLogObjectParent(mat,a->sMvctx);
2141: MatDuplicate(oldmat->A,cpvalues,&a->A);
2142: PetscLogObjectParent(mat,a->A);
2143: MatDuplicate(oldmat->B,cpvalues,&a->B);
2144: PetscLogObjectParent(mat,a->B);
2145: PetscFListDuplicate(mat->qlist,&matin->qlist);
2146: *newmat = mat;
2147: return(0);
2148: }
2150: #include petscsys.h
2154: PetscErrorCode MatLoad_MPISBAIJ(PetscViewer viewer, MatType type,Mat *newmat)
2155: {
2156: Mat A;
2158: PetscInt i,nz,j,rstart,rend;
2159: PetscScalar *vals,*buf;
2160: MPI_Comm comm = ((PetscObject)viewer)->comm;
2161: MPI_Status status;
2162: PetscMPIInt rank,size,tag = ((PetscObject)viewer)->tag,*sndcounts = 0,*browners,maxnz,*rowners,*locrowlens;
2163: PetscInt header[4],*rowlengths = 0,M,N,m,*cols;
2164: PetscInt *procsnz = 0,jj,*mycols,*ibuf;
2165: PetscInt bs=1,Mbs,mbs,extra_rows;
2166: PetscInt *dlens,*odlens,*mask,*masked1,*masked2,rowcount,odcount;
2167: PetscInt dcount,kmax,k,nzcount,tmp;
2168: int fd;
2169:
2171: PetscOptionsGetInt(PETSC_NULL,"-matload_block_size",&bs,PETSC_NULL);
2173: MPI_Comm_size(comm,&size);
2174: MPI_Comm_rank(comm,&rank);
2175: if (!rank) {
2176: PetscViewerBinaryGetDescriptor(viewer,&fd);
2177: PetscBinaryRead(fd,(char *)header,4,PETSC_INT);
2178: if (header[0] != MAT_FILE_COOKIE) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"not matrix object");
2179: if (header[3] < 0) {
2180: SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as MPISBAIJ");
2181: }
2182: }
2184: MPI_Bcast(header+1,3,MPIU_INT,0,comm);
2185: M = header[1]; N = header[2];
2187: if (M != N) SETERRQ(PETSC_ERR_SUP,"Can only do square matrices");
2189: /*
2190: This code adds extra rows to make sure the number of rows is
2191: divisible by the blocksize
2192: */
2193: Mbs = M/bs;
2194: extra_rows = bs - M + bs*(Mbs);
2195: if (extra_rows == bs) extra_rows = 0;
2196: else Mbs++;
2197: if (extra_rows &&!rank) {
2198: PetscInfo(0,"Padding loaded matrix to match blocksize\n");
2199: }
2201: /* determine ownership of all rows */
2202: mbs = Mbs/size + ((Mbs % size) > rank);
2203: m = mbs*bs;
2204: PetscMalloc(2*(size+2)*sizeof(PetscMPIInt),&rowners);
2205: browners = rowners + size + 1;
2206: MPI_Allgather(&mbs,1,MPI_INT,rowners+1,1,MPI_INT,comm);
2207: rowners[0] = 0;
2208: for (i=2; i<=size; i++) rowners[i] += rowners[i-1];
2209: for (i=0; i<=size; i++) browners[i] = rowners[i]*bs;
2210: rstart = rowners[rank];
2211: rend = rowners[rank+1];
2212:
2213: /* distribute row lengths to all processors */
2214: PetscMalloc((rend-rstart)*bs*sizeof(PetscMPIInt),&locrowlens);
2215: if (!rank) {
2216: PetscMalloc((M+extra_rows)*sizeof(PetscInt),&rowlengths);
2217: PetscBinaryRead(fd,rowlengths,M,PETSC_INT);
2218: for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1;
2219: PetscMalloc(size*sizeof(PetscMPIInt),&sndcounts);
2220: for (i=0; i<size; i++) sndcounts[i] = browners[i+1] - browners[i];
2221: MPI_Scatterv(rowlengths,sndcounts,browners,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);
2222: PetscFree(sndcounts);
2223: } else {
2224: MPI_Scatterv(0,0,0,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);
2225: }
2226:
2227: if (!rank) { /* procs[0] */
2228: /* calculate the number of nonzeros on each processor */
2229: PetscMalloc(size*sizeof(PetscInt),&procsnz);
2230: PetscMemzero(procsnz,size*sizeof(PetscInt));
2231: for (i=0; i<size; i++) {
2232: for (j=rowners[i]*bs; j< rowners[i+1]*bs; j++) {
2233: procsnz[i] += rowlengths[j];
2234: }
2235: }
2236: PetscFree(rowlengths);
2237:
2238: /* determine max buffer needed and allocate it */
2239: maxnz = 0;
2240: for (i=0; i<size; i++) {
2241: maxnz = PetscMax(maxnz,procsnz[i]);
2242: }
2243: PetscMalloc(maxnz*sizeof(PetscInt),&cols);
2245: /* read in my part of the matrix column indices */
2246: nz = procsnz[0];
2247: PetscMalloc(nz*sizeof(PetscInt),&ibuf);
2248: mycols = ibuf;
2249: if (size == 1) nz -= extra_rows;
2250: PetscBinaryRead(fd,mycols,nz,PETSC_INT);
2251: if (size == 1) for (i=0; i< extra_rows; i++) { mycols[nz+i] = M+i; }
2253: /* read in every ones (except the last) and ship off */
2254: for (i=1; i<size-1; i++) {
2255: nz = procsnz[i];
2256: PetscBinaryRead(fd,cols,nz,PETSC_INT);
2257: MPI_Send(cols,nz,MPIU_INT,i,tag,comm);
2258: }
2259: /* read in the stuff for the last proc */
2260: if (size != 1) {
2261: nz = procsnz[size-1] - extra_rows; /* the extra rows are not on the disk */
2262: PetscBinaryRead(fd,cols,nz,PETSC_INT);
2263: for (i=0; i<extra_rows; i++) cols[nz+i] = M+i;
2264: MPI_Send(cols,nz+extra_rows,MPIU_INT,size-1,tag,comm);
2265: }
2266: PetscFree(cols);
2267: } else { /* procs[i], i>0 */
2268: /* determine buffer space needed for message */
2269: nz = 0;
2270: for (i=0; i<m; i++) {
2271: nz += locrowlens[i];
2272: }
2273: PetscMalloc(nz*sizeof(PetscInt),&ibuf);
2274: mycols = ibuf;
2275: /* receive message of column indices*/
2276: MPI_Recv(mycols,nz,MPIU_INT,0,tag,comm,&status);
2277: MPI_Get_count(&status,MPIU_INT,&maxnz);
2278: if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file");
2279: }
2281: /* loop over local rows, determining number of off diagonal entries */
2282: PetscMalloc(2*(rend-rstart+1)*sizeof(PetscInt),&dlens);
2283: odlens = dlens + (rend-rstart);
2284: PetscMalloc(3*Mbs*sizeof(PetscInt),&mask);
2285: PetscMemzero(mask,3*Mbs*sizeof(PetscInt));
2286: masked1 = mask + Mbs;
2287: masked2 = masked1 + Mbs;
2288: rowcount = 0; nzcount = 0;
2289: for (i=0; i<mbs; i++) {
2290: dcount = 0;
2291: odcount = 0;
2292: for (j=0; j<bs; j++) {
2293: kmax = locrowlens[rowcount];
2294: for (k=0; k<kmax; k++) {
2295: tmp = mycols[nzcount++]/bs; /* block col. index */
2296: if (!mask[tmp]) {
2297: mask[tmp] = 1;
2298: if (tmp < rstart || tmp >= rend) masked2[odcount++] = tmp; /* entry in off-diag portion */
2299: else masked1[dcount++] = tmp; /* entry in diag portion */
2300: }
2301: }
2302: rowcount++;
2303: }
2304:
2305: dlens[i] = dcount; /* d_nzz[i] */
2306: odlens[i] = odcount; /* o_nzz[i] */
2308: /* zero out the mask elements we set */
2309: for (j=0; j<dcount; j++) mask[masked1[j]] = 0;
2310: for (j=0; j<odcount; j++) mask[masked2[j]] = 0;
2311: }
2312:
2313: /* create our matrix */
2314: MatCreate(comm,&A);
2315: MatSetSizes(A,m,m,PETSC_DETERMINE,PETSC_DETERMINE);
2316: MatSetType(A,type);
2317: MatMPISBAIJSetPreallocation(A,bs,0,dlens,0,odlens);
2318: MatSetOption(A,MAT_COLUMNS_SORTED);
2319:
2320: if (!rank) {
2321: PetscMalloc(maxnz*sizeof(PetscScalar),&buf);
2322: /* read in my part of the matrix numerical values */
2323: nz = procsnz[0];
2324: vals = buf;
2325: mycols = ibuf;
2326: if (size == 1) nz -= extra_rows;
2327: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2328: if (size == 1) for (i=0; i< extra_rows; i++) { vals[nz+i] = 1.0; }
2330: /* insert into matrix */
2331: jj = rstart*bs;
2332: for (i=0; i<m; i++) {
2333: MatSetValues(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
2334: mycols += locrowlens[i];
2335: vals += locrowlens[i];
2336: jj++;
2337: }
2339: /* read in other processors (except the last one) and ship out */
2340: for (i=1; i<size-1; i++) {
2341: nz = procsnz[i];
2342: vals = buf;
2343: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2344: MPI_Send(vals,nz,MPIU_SCALAR,i,A->tag,comm);
2345: }
2346: /* the last proc */
2347: if (size != 1){
2348: nz = procsnz[i] - extra_rows;
2349: vals = buf;
2350: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
2351: for (i=0; i<extra_rows; i++) vals[nz+i] = 1.0;
2352: MPI_Send(vals,nz+extra_rows,MPIU_SCALAR,size-1,A->tag,comm);
2353: }
2354: PetscFree(procsnz);
2356: } else {
2357: /* receive numeric values */
2358: PetscMalloc(nz*sizeof(PetscScalar),&buf);
2360: /* receive message of values*/
2361: vals = buf;
2362: mycols = ibuf;
2363: MPI_Recv(vals,nz,MPIU_SCALAR,0,A->tag,comm,&status);
2364: MPI_Get_count(&status,MPIU_SCALAR,&maxnz);
2365: if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file");
2367: /* insert into matrix */
2368: jj = rstart*bs;
2369: for (i=0; i<m; i++) {
2370: MatSetValues_MPISBAIJ(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
2371: mycols += locrowlens[i];
2372: vals += locrowlens[i];
2373: jj++;
2374: }
2375: }
2377: PetscFree(locrowlens);
2378: PetscFree(buf);
2379: PetscFree(ibuf);
2380: PetscFree(rowners);
2381: PetscFree(dlens);
2382: PetscFree(mask);
2383: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
2384: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
2385: *newmat = A;
2386: return(0);
2387: }
2391: /*XXXXX@
2392: MatMPISBAIJSetHashTableFactor - Sets the factor required to compute the size of the HashTable.
2394: Input Parameters:
2395: . mat - the matrix
2396: . fact - factor
2398: Collective on Mat
2400: Level: advanced
2402: Notes:
2403: This can also be set by the command line option: -mat_use_hash_table fact
2405: .keywords: matrix, hashtable, factor, HT
2407: .seealso: MatSetOption()
2408: @XXXXX*/
2413: PetscErrorCode MatGetRowMax_MPISBAIJ(Mat A,Vec v)
2414: {
2415: Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data;
2416: Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(a->B)->data;
2417: PetscReal atmp;
2418: PetscReal *work,*svalues,*rvalues;
2420: PetscInt i,bs,mbs,*bi,*bj,brow,j,ncols,krow,kcol,col,row,Mbs,bcol;
2421: PetscMPIInt rank,size;
2422: PetscInt *rowners_bs,dest,count,source;
2423: PetscScalar *va;
2424: MatScalar *ba;
2425: MPI_Status stat;
2428: MatGetRowMax(a->A,v);
2429: VecGetArray(v,&va);
2431: MPI_Comm_size(A->comm,&size);
2432: MPI_Comm_rank(A->comm,&rank);
2434: bs = A->rmap.bs;
2435: mbs = a->mbs;
2436: Mbs = a->Mbs;
2437: ba = b->a;
2438: bi = b->i;
2439: bj = b->j;
2441: /* find ownerships */
2442: rowners_bs = A->rmap.range;
2444: /* each proc creates an array to be distributed */
2445: PetscMalloc(bs*Mbs*sizeof(PetscReal),&work);
2446: PetscMemzero(work,bs*Mbs*sizeof(PetscReal));
2448: /* row_max for B */
2449: if (rank != size-1){
2450: for (i=0; i<mbs; i++) {
2451: ncols = bi[1] - bi[0]; bi++;
2452: brow = bs*i;
2453: for (j=0; j<ncols; j++){
2454: bcol = bs*(*bj);
2455: for (kcol=0; kcol<bs; kcol++){
2456: col = bcol + kcol; /* local col index */
2457: col += rowners_bs[rank+1]; /* global col index */
2458: for (krow=0; krow<bs; krow++){
2459: atmp = PetscAbsScalar(*ba); ba++;
2460: row = brow + krow; /* local row index */
2461: if (PetscRealPart(va[row]) < atmp) va[row] = atmp;
2462: if (work[col] < atmp) work[col] = atmp;
2463: }
2464: }
2465: bj++;
2466: }
2467: }
2469: /* send values to its owners */
2470: for (dest=rank+1; dest<size; dest++){
2471: svalues = work + rowners_bs[dest];
2472: count = rowners_bs[dest+1]-rowners_bs[dest];
2473: MPI_Send(svalues,count,MPIU_REAL,dest,rank,A->comm);
2474: }
2475: }
2476:
2477: /* receive values */
2478: if (rank){
2479: rvalues = work;
2480: count = rowners_bs[rank+1]-rowners_bs[rank];
2481: for (source=0; source<rank; source++){
2482: MPI_Recv(rvalues,count,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,A->comm,&stat);
2483: /* process values */
2484: for (i=0; i<count; i++){
2485: if (PetscRealPart(va[i]) < rvalues[i]) va[i] = rvalues[i];
2486: }
2487: }
2488: }
2490: VecRestoreArray(v,&va);
2491: PetscFree(work);
2492: return(0);
2493: }
2497: PetscErrorCode MatRelax_MPISBAIJ(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
2498: {
2499: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
2501: PetscInt mbs=mat->mbs,bs=matin->rmap.bs;
2502: PetscScalar *x,*b,*ptr,zero=0.0;
2503: Vec bb1;
2504:
2506: if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits);
2507: if (bs > 1)
2508: SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented");
2510: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){
2511: if ( flag & SOR_ZERO_INITIAL_GUESS ) {
2512: (*mat->A->ops->relax)(mat->A,bb,omega,flag,fshift,lits,lits,xx);
2513: its--;
2514: }
2516: VecDuplicate(bb,&bb1);
2517: while (its--){
2518:
2519: /* lower triangular part: slvec0b = - B^T*xx */
2520: (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);
2521:
2522: /* copy xx into slvec0a */
2523: VecGetArray(mat->slvec0,&ptr);
2524: VecGetArray(xx,&x);
2525: PetscMemcpy(ptr,x,bs*mbs*sizeof(MatScalar));
2526: VecRestoreArray(mat->slvec0,&ptr);
2528: VecScale(mat->slvec0,-1.0);
2530: /* copy bb into slvec1a */
2531: VecGetArray(mat->slvec1,&ptr);
2532: VecGetArray(bb,&b);
2533: PetscMemcpy(ptr,b,bs*mbs*sizeof(MatScalar));
2534: VecRestoreArray(mat->slvec1,&ptr);
2536: /* set slvec1b = 0 */
2537: VecSet(mat->slvec1b,zero);
2539: VecScatterBegin(mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD,mat->sMvctx);
2540: VecRestoreArray(xx,&x);
2541: VecRestoreArray(bb,&b);
2542: VecScatterEnd(mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD,mat->sMvctx);
2544: /* upper triangular part: bb1 = bb1 - B*x */
2545: (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,bb1);
2546:
2547: /* local diagonal sweep */
2548: (*mat->A->ops->relax)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);
2549: }
2550: VecDestroy(bb1);
2551: } else {
2552: SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format");
2553: }
2554: return(0);
2555: }
2559: PetscErrorCode MatRelax_MPISBAIJ_2comm(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
2560: {
2561: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data;
2563: Vec lvec1,bb1;
2564:
2566: if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits);
2567: if (matin->rmap.bs > 1)
2568: SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented");
2570: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){
2571: if ( flag & SOR_ZERO_INITIAL_GUESS ) {
2572: (*mat->A->ops->relax)(mat->A,bb,omega,flag,fshift,lits,lits,xx);
2573: its--;
2574: }
2576: VecDuplicate(mat->lvec,&lvec1);
2577: VecDuplicate(bb,&bb1);
2578: while (its--){
2579: VecScatterBegin(xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD,mat->Mvctx);
2580:
2581: /* lower diagonal part: bb1 = bb - B^T*xx */
2582: (*mat->B->ops->multtranspose)(mat->B,xx,lvec1);
2583: VecScale(lvec1,-1.0);
2585: VecScatterEnd(xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD,mat->Mvctx);
2586: VecCopy(bb,bb1);
2587: VecScatterBegin(lvec1,bb1,ADD_VALUES,SCATTER_REVERSE,mat->Mvctx);
2589: /* upper diagonal part: bb1 = bb1 - B*x */
2590: VecScale(mat->lvec,-1.0);
2591: (*mat->B->ops->multadd)(mat->B,mat->lvec,bb1,bb1);
2593: VecScatterEnd(lvec1,bb1,ADD_VALUES,SCATTER_REVERSE,mat->Mvctx);
2594:
2595: /* diagonal sweep */
2596: (*mat->A->ops->relax)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);
2597: }
2598: VecDestroy(lvec1);
2599: VecDestroy(bb1);
2600: } else {
2601: SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format");
2602: }
2603: return(0);
2604: }