Actual source code: mpibaij.c
petsc-3.9.4 2018-09-11
2: #include <../src/mat/impls/baij/mpi/mpibaij.h>
4: #include <petscblaslapack.h>
5: #include <petscsf.h>
7: #if defined(PETSC_HAVE_HYPRE)
8: PETSC_INTERN PetscErrorCode MatConvert_AIJ_HYPRE(Mat,MatType,MatReuse,Mat*);
9: #endif
11: PetscErrorCode MatGetRowMaxAbs_MPIBAIJ(Mat A,Vec v,PetscInt idx[])
12: {
13: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
15: PetscInt i,*idxb = 0;
16: PetscScalar *va,*vb;
17: Vec vtmp;
20: MatGetRowMaxAbs(a->A,v,idx);
21: VecGetArray(v,&va);
22: if (idx) {
23: for (i=0; i<A->rmap->n; i++) {
24: if (PetscAbsScalar(va[i])) idx[i] += A->cmap->rstart;
25: }
26: }
28: VecCreateSeq(PETSC_COMM_SELF,A->rmap->n,&vtmp);
29: if (idx) {PetscMalloc1(A->rmap->n,&idxb);}
30: MatGetRowMaxAbs(a->B,vtmp,idxb);
31: VecGetArray(vtmp,&vb);
33: for (i=0; i<A->rmap->n; i++) {
34: if (PetscAbsScalar(va[i]) < PetscAbsScalar(vb[i])) {
35: va[i] = vb[i];
36: if (idx) idx[i] = A->cmap->bs*a->garray[idxb[i]/A->cmap->bs] + (idxb[i] % A->cmap->bs);
37: }
38: }
40: VecRestoreArray(v,&va);
41: VecRestoreArray(vtmp,&vb);
42: PetscFree(idxb);
43: VecDestroy(&vtmp);
44: return(0);
45: }
47: PetscErrorCode MatStoreValues_MPIBAIJ(Mat mat)
48: {
49: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ*)mat->data;
53: MatStoreValues(aij->A);
54: MatStoreValues(aij->B);
55: return(0);
56: }
58: PetscErrorCode MatRetrieveValues_MPIBAIJ(Mat mat)
59: {
60: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ*)mat->data;
64: MatRetrieveValues(aij->A);
65: MatRetrieveValues(aij->B);
66: return(0);
67: }
69: /*
70: Local utility routine that creates a mapping from the global column
71: number to the local number in the off-diagonal part of the local
72: storage of the matrix. This is done in a non scalable way since the
73: length of colmap equals the global matrix length.
74: */
75: PetscErrorCode MatCreateColmap_MPIBAIJ_Private(Mat mat)
76: {
77: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
78: Mat_SeqBAIJ *B = (Mat_SeqBAIJ*)baij->B->data;
80: PetscInt nbs = B->nbs,i,bs=mat->rmap->bs;
83: #if defined(PETSC_USE_CTABLE)
84: PetscTableCreate(baij->nbs,baij->Nbs+1,&baij->colmap);
85: for (i=0; i<nbs; i++) {
86: PetscTableAdd(baij->colmap,baij->garray[i]+1,i*bs+1,INSERT_VALUES);
87: }
88: #else
89: PetscMalloc1(baij->Nbs+1,&baij->colmap);
90: PetscLogObjectMemory((PetscObject)mat,baij->Nbs*sizeof(PetscInt));
91: PetscMemzero(baij->colmap,baij->Nbs*sizeof(PetscInt));
92: for (i=0; i<nbs; i++) baij->colmap[baij->garray[i]] = i*bs+1;
93: #endif
94: return(0);
95: }
97: #define MatSetValues_SeqBAIJ_A_Private(row,col,value,addv,orow,ocol) \
98: { \
99: \
100: brow = row/bs; \
101: rp = aj + ai[brow]; ap = aa + bs2*ai[brow]; \
102: rmax = aimax[brow]; nrow = ailen[brow]; \
103: bcol = col/bs; \
104: ridx = row % bs; cidx = col % bs; \
105: low = 0; high = nrow; \
106: while (high-low > 3) { \
107: t = (low+high)/2; \
108: if (rp[t] > bcol) high = t; \
109: else low = t; \
110: } \
111: for (_i=low; _i<high; _i++) { \
112: if (rp[_i] > bcol) break; \
113: if (rp[_i] == bcol) { \
114: bap = ap + bs2*_i + bs*cidx + ridx; \
115: if (addv == ADD_VALUES) *bap += value; \
116: else *bap = value; \
117: goto a_noinsert; \
118: } \
119: } \
120: if (a->nonew == 1) goto a_noinsert; \
121: if (a->nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero at global row/column (%D, %D) into matrix", orow, ocol); \
122: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,aimax,a->nonew,MatScalar); \
123: N = nrow++ - 1; \
124: /* shift up all the later entries in this row */ \
125: for (ii=N; ii>=_i; ii--) { \
126: rp[ii+1] = rp[ii]; \
127: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
128: } \
129: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar)); } \
130: rp[_i] = bcol; \
131: ap[bs2*_i + bs*cidx + ridx] = value; \
132: a_noinsert:; \
133: ailen[brow] = nrow; \
134: }
136: #define MatSetValues_SeqBAIJ_B_Private(row,col,value,addv,orow,ocol) \
137: { \
138: brow = row/bs; \
139: rp = bj + bi[brow]; ap = ba + bs2*bi[brow]; \
140: rmax = bimax[brow]; nrow = bilen[brow]; \
141: bcol = col/bs; \
142: ridx = row % bs; cidx = col % bs; \
143: low = 0; high = nrow; \
144: while (high-low > 3) { \
145: t = (low+high)/2; \
146: if (rp[t] > bcol) high = t; \
147: else low = t; \
148: } \
149: for (_i=low; _i<high; _i++) { \
150: if (rp[_i] > bcol) break; \
151: if (rp[_i] == bcol) { \
152: bap = ap + bs2*_i + bs*cidx + ridx; \
153: if (addv == ADD_VALUES) *bap += value; \
154: else *bap = value; \
155: goto b_noinsert; \
156: } \
157: } \
158: if (b->nonew == 1) goto b_noinsert; \
159: if (b->nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero at global row/column (%D, %D) into matrix", orow, ocol); \
160: MatSeqXAIJReallocateAIJ(B,b->mbs,bs2,nrow,brow,bcol,rmax,ba,bi,bj,rp,ap,bimax,b->nonew,MatScalar); \
161: N = nrow++ - 1; \
162: /* shift up all the later entries in this row */ \
163: for (ii=N; ii>=_i; ii--) { \
164: rp[ii+1] = rp[ii]; \
165: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar)); \
166: } \
167: if (N>=_i) { PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));} \
168: rp[_i] = bcol; \
169: ap[bs2*_i + bs*cidx + ridx] = value; \
170: b_noinsert:; \
171: bilen[brow] = nrow; \
172: }
174: PetscErrorCode MatSetValues_MPIBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
175: {
176: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
177: MatScalar value;
178: PetscBool roworiented = baij->roworiented;
180: PetscInt i,j,row,col;
181: PetscInt rstart_orig=mat->rmap->rstart;
182: PetscInt rend_orig =mat->rmap->rend,cstart_orig=mat->cmap->rstart;
183: PetscInt cend_orig =mat->cmap->rend,bs=mat->rmap->bs;
185: /* Some Variables required in the macro */
186: Mat A = baij->A;
187: Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)(A)->data;
188: PetscInt *aimax=a->imax,*ai=a->i,*ailen=a->ilen,*aj=a->j;
189: MatScalar *aa =a->a;
191: Mat B = baij->B;
192: Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(B)->data;
193: PetscInt *bimax=b->imax,*bi=b->i,*bilen=b->ilen,*bj=b->j;
194: MatScalar *ba =b->a;
196: PetscInt *rp,ii,nrow,_i,rmax,N,brow,bcol;
197: PetscInt low,high,t,ridx,cidx,bs2=a->bs2;
198: MatScalar *ap,*bap;
201: for (i=0; i<m; i++) {
202: if (im[i] < 0) continue;
203: #if defined(PETSC_USE_DEBUG)
204: if (im[i] >= mat->rmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],mat->rmap->N-1);
205: #endif
206: if (im[i] >= rstart_orig && im[i] < rend_orig) {
207: row = im[i] - rstart_orig;
208: for (j=0; j<n; j++) {
209: if (in[j] >= cstart_orig && in[j] < cend_orig) {
210: col = in[j] - cstart_orig;
211: if (roworiented) value = v[i*n+j];
212: else value = v[i+j*m];
213: MatSetValues_SeqBAIJ_A_Private(row,col,value,addv,im[i],in[j]);
214: /* MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv); */
215: } else if (in[j] < 0) continue;
216: #if defined(PETSC_USE_DEBUG)
217: else if (in[j] >= mat->cmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[j],mat->cmap->N-1);
218: #endif
219: else {
220: if (mat->was_assembled) {
221: if (!baij->colmap) {
222: MatCreateColmap_MPIBAIJ_Private(mat);
223: }
224: #if defined(PETSC_USE_CTABLE)
225: PetscTableFind(baij->colmap,in[j]/bs + 1,&col);
226: col = col - 1;
227: #else
228: col = baij->colmap[in[j]/bs] - 1;
229: #endif
230: if (col < 0 && !((Mat_SeqBAIJ*)(baij->B->data))->nonew) {
231: MatDisAssemble_MPIBAIJ(mat);
232: col = in[j];
233: /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */
234: B = baij->B;
235: b = (Mat_SeqBAIJ*)(B)->data;
236: bimax=b->imax;bi=b->i;bilen=b->ilen;bj=b->j;
237: ba =b->a;
238: } else if (col < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", im[i], in[j]);
239: else col += in[j]%bs;
240: } else col = in[j];
241: if (roworiented) value = v[i*n+j];
242: else value = v[i+j*m];
243: MatSetValues_SeqBAIJ_B_Private(row,col,value,addv,im[i],in[j]);
244: /* MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv); */
245: }
246: }
247: } else {
248: if (mat->nooffprocentries) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Setting off process row %D even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set",im[i]);
249: if (!baij->donotstash) {
250: mat->assembled = PETSC_FALSE;
251: if (roworiented) {
252: MatStashValuesRow_Private(&mat->stash,im[i],n,in,v+i*n,PETSC_FALSE);
253: } else {
254: MatStashValuesCol_Private(&mat->stash,im[i],n,in,v+i,m,PETSC_FALSE);
255: }
256: }
257: }
258: }
259: return(0);
260: }
262: PETSC_STATIC_INLINE PetscErrorCode MatSetValuesBlocked_SeqBAIJ_Inlined(Mat A,PetscInt row,PetscInt col,const PetscScalar v[],InsertMode is,PetscInt orow,PetscInt ocol)
263: {
264: Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data;
265: PetscInt *rp,low,high,t,ii,jj,nrow,i,rmax,N;
266: PetscInt *imax=a->imax,*ai=a->i,*ailen=a->ilen;
267: PetscErrorCode ierr;
268: PetscInt *aj =a->j,nonew=a->nonew,bs2=a->bs2,bs=A->rmap->bs;
269: PetscBool roworiented=a->roworiented;
270: const PetscScalar *value = v;
271: MatScalar *ap,*aa = a->a,*bap;
274: rp = aj + ai[row];
275: ap = aa + bs2*ai[row];
276: rmax = imax[row];
277: nrow = ailen[row];
278: value = v;
279: low = 0;
280: high = nrow;
281: while (high-low > 7) {
282: t = (low+high)/2;
283: if (rp[t] > col) high = t;
284: else low = t;
285: }
286: for (i=low; i<high; i++) {
287: if (rp[i] > col) break;
288: if (rp[i] == col) {
289: bap = ap + bs2*i;
290: if (roworiented) {
291: if (is == ADD_VALUES) {
292: for (ii=0; ii<bs; ii++) {
293: for (jj=ii; jj<bs2; jj+=bs) {
294: bap[jj] += *value++;
295: }
296: }
297: } else {
298: for (ii=0; ii<bs; ii++) {
299: for (jj=ii; jj<bs2; jj+=bs) {
300: bap[jj] = *value++;
301: }
302: }
303: }
304: } else {
305: if (is == ADD_VALUES) {
306: for (ii=0; ii<bs; ii++,value+=bs) {
307: for (jj=0; jj<bs; jj++) {
308: bap[jj] += value[jj];
309: }
310: bap += bs;
311: }
312: } else {
313: for (ii=0; ii<bs; ii++,value+=bs) {
314: for (jj=0; jj<bs; jj++) {
315: bap[jj] = value[jj];
316: }
317: bap += bs;
318: }
319: }
320: }
321: goto noinsert2;
322: }
323: }
324: if (nonew == 1) goto noinsert2;
325: if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new global block indexed nonzero block (%D, %D) in the matrix", orow, ocol);
326: MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar);
327: N = nrow++ - 1; high++;
328: /* shift up all the later entries in this row */
329: for (ii=N; ii>=i; ii--) {
330: rp[ii+1] = rp[ii];
331: PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));
332: }
333: if (N >= i) {
334: PetscMemzero(ap+bs2*i,bs2*sizeof(MatScalar));
335: }
336: rp[i] = col;
337: bap = ap + bs2*i;
338: if (roworiented) {
339: for (ii=0; ii<bs; ii++) {
340: for (jj=ii; jj<bs2; jj+=bs) {
341: bap[jj] = *value++;
342: }
343: }
344: } else {
345: for (ii=0; ii<bs; ii++) {
346: for (jj=0; jj<bs; jj++) {
347: *bap++ = *value++;
348: }
349: }
350: }
351: noinsert2:;
352: ailen[row] = nrow;
353: return(0);
354: }
356: /*
357: This routine should be optimized so that the block copy at ** Here a copy is required ** below is not needed
358: by passing additional stride information into the MatSetValuesBlocked_SeqBAIJ_Inlined() routine
359: */
360: PetscErrorCode MatSetValuesBlocked_MPIBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
361: {
362: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
363: const PetscScalar *value;
364: MatScalar *barray = baij->barray;
365: PetscBool roworiented = baij->roworiented;
366: PetscErrorCode ierr;
367: PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs;
368: PetscInt rend=baij->rendbs,cstart=baij->cstartbs,stepval;
369: PetscInt cend=baij->cendbs,bs=mat->rmap->bs,bs2=baij->bs2;
372: if (!barray) {
373: PetscMalloc1(bs2,&barray);
374: baij->barray = barray;
375: }
377: if (roworiented) stepval = (n-1)*bs;
378: else stepval = (m-1)*bs;
380: for (i=0; i<m; i++) {
381: if (im[i] < 0) continue;
382: #if defined(PETSC_USE_DEBUG)
383: if (im[i] >= baij->Mbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block indexed row too large %D max %D",im[i],baij->Mbs-1);
384: #endif
385: if (im[i] >= rstart && im[i] < rend) {
386: row = im[i] - rstart;
387: for (j=0; j<n; j++) {
388: /* If NumCol = 1 then a copy is not required */
389: if ((roworiented) && (n == 1)) {
390: barray = (MatScalar*)v + i*bs2;
391: } else if ((!roworiented) && (m == 1)) {
392: barray = (MatScalar*)v + j*bs2;
393: } else { /* Here a copy is required */
394: if (roworiented) {
395: value = v + (i*(stepval+bs) + j)*bs;
396: } else {
397: value = v + (j*(stepval+bs) + i)*bs;
398: }
399: for (ii=0; ii<bs; ii++,value+=bs+stepval) {
400: for (jj=0; jj<bs; jj++) barray[jj] = value[jj];
401: barray += bs;
402: }
403: barray -= bs2;
404: }
406: if (in[j] >= cstart && in[j] < cend) {
407: col = in[j] - cstart;
408: MatSetValuesBlocked_SeqBAIJ_Inlined(baij->A,row,col,barray,addv,im[i],in[j]);
409: } else if (in[j] < 0) continue;
410: #if defined(PETSC_USE_DEBUG)
411: else if (in[j] >= baij->Nbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Block indexed column too large %D max %D",in[j],baij->Nbs-1);
412: #endif
413: else {
414: if (mat->was_assembled) {
415: if (!baij->colmap) {
416: MatCreateColmap_MPIBAIJ_Private(mat);
417: }
419: #if defined(PETSC_USE_DEBUG)
420: #if defined(PETSC_USE_CTABLE)
421: { PetscInt data;
422: PetscTableFind(baij->colmap,in[j]+1,&data);
423: if ((data - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap");
424: }
425: #else
426: if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap");
427: #endif
428: #endif
429: #if defined(PETSC_USE_CTABLE)
430: PetscTableFind(baij->colmap,in[j]+1,&col);
431: col = (col - 1)/bs;
432: #else
433: col = (baij->colmap[in[j]] - 1)/bs;
434: #endif
435: if (col < 0 && !((Mat_SeqBAIJ*)(baij->B->data))->nonew) {
436: MatDisAssemble_MPIBAIJ(mat);
437: col = in[j];
438: } else if (col < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new blocked indexed nonzero block (%D, %D) into matrix",im[i],in[j]);
439: } else col = in[j];
440: MatSetValuesBlocked_SeqBAIJ_Inlined(baij->B,row,col,barray,addv,im[i],in[j]);
441: }
442: }
443: } else {
444: if (mat->nooffprocentries) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Setting off process block indexed row %D even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set",im[i]);
445: if (!baij->donotstash) {
446: if (roworiented) {
447: MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
448: } else {
449: MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
450: }
451: }
452: }
453: }
454: return(0);
455: }
457: #define HASH_KEY 0.6180339887
458: #define HASH(size,key,tmp) (tmp = (key)*HASH_KEY,(PetscInt)((size)*(tmp-(PetscInt)tmp)))
459: /* #define HASH(size,key) ((PetscInt)((size)*fmod(((key)*HASH_KEY),1))) */
460: /* #define HASH(size,key,tmp) ((PetscInt)((size)*fmod(((key)*HASH_KEY),1))) */
461: PetscErrorCode MatSetValues_MPIBAIJ_HT(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
462: {
463: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
464: PetscBool roworiented = baij->roworiented;
466: PetscInt i,j,row,col;
467: PetscInt rstart_orig=mat->rmap->rstart;
468: PetscInt rend_orig =mat->rmap->rend,Nbs=baij->Nbs;
469: PetscInt h1,key,size=baij->ht_size,bs=mat->rmap->bs,*HT=baij->ht,idx;
470: PetscReal tmp;
471: MatScalar **HD = baij->hd,value;
472: #if defined(PETSC_USE_DEBUG)
473: PetscInt total_ct=baij->ht_total_ct,insert_ct=baij->ht_insert_ct;
474: #endif
477: for (i=0; i<m; i++) {
478: #if defined(PETSC_USE_DEBUG)
479: if (im[i] < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row");
480: if (im[i] >= mat->rmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],mat->rmap->N-1);
481: #endif
482: row = im[i];
483: if (row >= rstart_orig && row < rend_orig) {
484: for (j=0; j<n; j++) {
485: col = in[j];
486: if (roworiented) value = v[i*n+j];
487: else value = v[i+j*m];
488: /* Look up PetscInto the Hash Table */
489: key = (row/bs)*Nbs+(col/bs)+1;
490: h1 = HASH(size,key,tmp);
493: idx = h1;
494: #if defined(PETSC_USE_DEBUG)
495: insert_ct++;
496: total_ct++;
497: if (HT[idx] != key) {
498: for (idx=h1; (idx<size) && (HT[idx]!=key); idx++,total_ct++) ;
499: if (idx == size) {
500: for (idx=0; (idx<h1) && (HT[idx]!=key); idx++,total_ct++) ;
501: if (idx == h1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"(%D,%D) has no entry in the hash table", row, col);
502: }
503: }
504: #else
505: if (HT[idx] != key) {
506: for (idx=h1; (idx<size) && (HT[idx]!=key); idx++) ;
507: if (idx == size) {
508: for (idx=0; (idx<h1) && (HT[idx]!=key); idx++) ;
509: if (idx == h1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"(%D,%D) has no entry in the hash table", row, col);
510: }
511: }
512: #endif
513: /* A HASH table entry is found, so insert the values at the correct address */
514: if (addv == ADD_VALUES) *(HD[idx]+ (col % bs)*bs + (row % bs)) += value;
515: else *(HD[idx]+ (col % bs)*bs + (row % bs)) = value;
516: }
517: } else if (!baij->donotstash) {
518: if (roworiented) {
519: MatStashValuesRow_Private(&mat->stash,im[i],n,in,v+i*n,PETSC_FALSE);
520: } else {
521: MatStashValuesCol_Private(&mat->stash,im[i],n,in,v+i,m,PETSC_FALSE);
522: }
523: }
524: }
525: #if defined(PETSC_USE_DEBUG)
526: baij->ht_total_ct += total_ct;
527: baij->ht_insert_ct += insert_ct;
528: #endif
529: return(0);
530: }
532: PetscErrorCode MatSetValuesBlocked_MPIBAIJ_HT(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv)
533: {
534: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
535: PetscBool roworiented = baij->roworiented;
536: PetscErrorCode ierr;
537: PetscInt i,j,ii,jj,row,col;
538: PetscInt rstart=baij->rstartbs;
539: PetscInt rend =mat->rmap->rend,stepval,bs=mat->rmap->bs,bs2=baij->bs2,nbs2=n*bs2;
540: PetscInt h1,key,size=baij->ht_size,idx,*HT=baij->ht,Nbs=baij->Nbs;
541: PetscReal tmp;
542: MatScalar **HD = baij->hd,*baij_a;
543: const PetscScalar *v_t,*value;
544: #if defined(PETSC_USE_DEBUG)
545: PetscInt total_ct=baij->ht_total_ct,insert_ct=baij->ht_insert_ct;
546: #endif
549: if (roworiented) stepval = (n-1)*bs;
550: else stepval = (m-1)*bs;
552: for (i=0; i<m; i++) {
553: #if defined(PETSC_USE_DEBUG)
554: if (im[i] < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",im[i]);
555: if (im[i] >= baij->Mbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],baij->Mbs-1);
556: #endif
557: row = im[i];
558: v_t = v + i*nbs2;
559: if (row >= rstart && row < rend) {
560: for (j=0; j<n; j++) {
561: col = in[j];
563: /* Look up into the Hash Table */
564: key = row*Nbs+col+1;
565: h1 = HASH(size,key,tmp);
567: idx = h1;
568: #if defined(PETSC_USE_DEBUG)
569: total_ct++;
570: insert_ct++;
571: if (HT[idx] != key) {
572: for (idx=h1; (idx<size) && (HT[idx]!=key); idx++,total_ct++) ;
573: if (idx == size) {
574: for (idx=0; (idx<h1) && (HT[idx]!=key); idx++,total_ct++) ;
575: if (idx == h1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"(%D,%D) has no entry in the hash table", row, col);
576: }
577: }
578: #else
579: if (HT[idx] != key) {
580: for (idx=h1; (idx<size) && (HT[idx]!=key); idx++) ;
581: if (idx == size) {
582: for (idx=0; (idx<h1) && (HT[idx]!=key); idx++) ;
583: if (idx == h1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"(%D,%D) has no entry in the hash table", row, col);
584: }
585: }
586: #endif
587: baij_a = HD[idx];
588: if (roworiented) {
589: /*value = v + i*(stepval+bs)*bs + j*bs;*/
590: /* value = v + (i*(stepval+bs)+j)*bs; */
591: value = v_t;
592: v_t += bs;
593: if (addv == ADD_VALUES) {
594: for (ii=0; ii<bs; ii++,value+=stepval) {
595: for (jj=ii; jj<bs2; jj+=bs) {
596: baij_a[jj] += *value++;
597: }
598: }
599: } else {
600: for (ii=0; ii<bs; ii++,value+=stepval) {
601: for (jj=ii; jj<bs2; jj+=bs) {
602: baij_a[jj] = *value++;
603: }
604: }
605: }
606: } else {
607: value = v + j*(stepval+bs)*bs + i*bs;
608: if (addv == ADD_VALUES) {
609: for (ii=0; ii<bs; ii++,value+=stepval,baij_a+=bs) {
610: for (jj=0; jj<bs; jj++) {
611: baij_a[jj] += *value++;
612: }
613: }
614: } else {
615: for (ii=0; ii<bs; ii++,value+=stepval,baij_a+=bs) {
616: for (jj=0; jj<bs; jj++) {
617: baij_a[jj] = *value++;
618: }
619: }
620: }
621: }
622: }
623: } else {
624: if (!baij->donotstash) {
625: if (roworiented) {
626: MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
627: } else {
628: MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
629: }
630: }
631: }
632: }
633: #if defined(PETSC_USE_DEBUG)
634: baij->ht_total_ct += total_ct;
635: baij->ht_insert_ct += insert_ct;
636: #endif
637: return(0);
638: }
640: PetscErrorCode MatGetValues_MPIBAIJ(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[])
641: {
642: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
644: PetscInt bs = mat->rmap->bs,i,j,bsrstart = mat->rmap->rstart,bsrend = mat->rmap->rend;
645: PetscInt bscstart = mat->cmap->rstart,bscend = mat->cmap->rend,row,col,data;
648: for (i=0; i<m; i++) {
649: if (idxm[i] < 0) continue; /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",idxm[i]);*/
650: if (idxm[i] >= mat->rmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",idxm[i],mat->rmap->N-1);
651: if (idxm[i] >= bsrstart && idxm[i] < bsrend) {
652: row = idxm[i] - bsrstart;
653: for (j=0; j<n; j++) {
654: if (idxn[j] < 0) continue; /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",idxn[j]); */
655: if (idxn[j] >= mat->cmap->N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",idxn[j],mat->cmap->N-1);
656: if (idxn[j] >= bscstart && idxn[j] < bscend) {
657: col = idxn[j] - bscstart;
658: MatGetValues_SeqBAIJ(baij->A,1,&row,1,&col,v+i*n+j);
659: } else {
660: if (!baij->colmap) {
661: MatCreateColmap_MPIBAIJ_Private(mat);
662: }
663: #if defined(PETSC_USE_CTABLE)
664: PetscTableFind(baij->colmap,idxn[j]/bs+1,&data);
665: data--;
666: #else
667: data = baij->colmap[idxn[j]/bs]-1;
668: #endif
669: if ((data < 0) || (baij->garray[data/bs] != idxn[j]/bs)) *(v+i*n+j) = 0.0;
670: else {
671: col = data + idxn[j]%bs;
672: MatGetValues_SeqBAIJ(baij->B,1,&row,1,&col,v+i*n+j);
673: }
674: }
675: }
676: } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only local values currently supported");
677: }
678: return(0);
679: }
681: PetscErrorCode MatNorm_MPIBAIJ(Mat mat,NormType type,PetscReal *nrm)
682: {
683: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
684: Mat_SeqBAIJ *amat = (Mat_SeqBAIJ*)baij->A->data,*bmat = (Mat_SeqBAIJ*)baij->B->data;
686: PetscInt i,j,bs2=baij->bs2,bs=baij->A->rmap->bs,nz,row,col;
687: PetscReal sum = 0.0;
688: MatScalar *v;
691: if (baij->size == 1) {
692: MatNorm(baij->A,type,nrm);
693: } else {
694: if (type == NORM_FROBENIUS) {
695: v = amat->a;
696: nz = amat->nz*bs2;
697: for (i=0; i<nz; i++) {
698: sum += PetscRealPart(PetscConj(*v)*(*v)); v++;
699: }
700: v = bmat->a;
701: nz = bmat->nz*bs2;
702: for (i=0; i<nz; i++) {
703: sum += PetscRealPart(PetscConj(*v)*(*v)); v++;
704: }
705: MPIU_Allreduce(&sum,nrm,1,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)mat));
706: *nrm = PetscSqrtReal(*nrm);
707: } else if (type == NORM_1) { /* max column sum */
708: PetscReal *tmp,*tmp2;
709: PetscInt *jj,*garray=baij->garray,cstart=baij->rstartbs;
710: PetscMalloc2(mat->cmap->N,&tmp,mat->cmap->N,&tmp2);
711: PetscMemzero(tmp,mat->cmap->N*sizeof(PetscReal));
712: v = amat->a; jj = amat->j;
713: for (i=0; i<amat->nz; i++) {
714: for (j=0; j<bs; j++) {
715: col = bs*(cstart + *jj) + j; /* column index */
716: for (row=0; row<bs; row++) {
717: tmp[col] += PetscAbsScalar(*v); v++;
718: }
719: }
720: jj++;
721: }
722: v = bmat->a; jj = bmat->j;
723: for (i=0; i<bmat->nz; i++) {
724: for (j=0; j<bs; j++) {
725: col = bs*garray[*jj] + j;
726: for (row=0; row<bs; row++) {
727: tmp[col] += PetscAbsScalar(*v); v++;
728: }
729: }
730: jj++;
731: }
732: MPIU_Allreduce(tmp,tmp2,mat->cmap->N,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)mat));
733: *nrm = 0.0;
734: for (j=0; j<mat->cmap->N; j++) {
735: if (tmp2[j] > *nrm) *nrm = tmp2[j];
736: }
737: PetscFree2(tmp,tmp2);
738: } else if (type == NORM_INFINITY) { /* max row sum */
739: PetscReal *sums;
740: PetscMalloc1(bs,&sums);
741: sum = 0.0;
742: for (j=0; j<amat->mbs; j++) {
743: for (row=0; row<bs; row++) sums[row] = 0.0;
744: v = amat->a + bs2*amat->i[j];
745: nz = amat->i[j+1]-amat->i[j];
746: for (i=0; i<nz; i++) {
747: for (col=0; col<bs; col++) {
748: for (row=0; row<bs; row++) {
749: sums[row] += PetscAbsScalar(*v); v++;
750: }
751: }
752: }
753: v = bmat->a + bs2*bmat->i[j];
754: nz = bmat->i[j+1]-bmat->i[j];
755: for (i=0; i<nz; i++) {
756: for (col=0; col<bs; col++) {
757: for (row=0; row<bs; row++) {
758: sums[row] += PetscAbsScalar(*v); v++;
759: }
760: }
761: }
762: for (row=0; row<bs; row++) {
763: if (sums[row] > sum) sum = sums[row];
764: }
765: }
766: MPIU_Allreduce(&sum,nrm,1,MPIU_REAL,MPIU_MAX,PetscObjectComm((PetscObject)mat));
767: PetscFree(sums);
768: } else SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"No support for this norm yet");
769: }
770: return(0);
771: }
773: /*
774: Creates the hash table, and sets the table
775: This table is created only once.
776: If new entried need to be added to the matrix
777: then the hash table has to be destroyed and
778: recreated.
779: */
780: PetscErrorCode MatCreateHashTable_MPIBAIJ_Private(Mat mat,PetscReal factor)
781: {
782: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
783: Mat A = baij->A,B=baij->B;
784: Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b=(Mat_SeqBAIJ*)B->data;
785: PetscInt i,j,k,nz=a->nz+b->nz,h1,*ai=a->i,*aj=a->j,*bi=b->i,*bj=b->j;
787: PetscInt ht_size,bs2=baij->bs2,rstart=baij->rstartbs;
788: PetscInt cstart=baij->cstartbs,*garray=baij->garray,row,col,Nbs=baij->Nbs;
789: PetscInt *HT,key;
790: MatScalar **HD;
791: PetscReal tmp;
792: #if defined(PETSC_USE_INFO)
793: PetscInt ct=0,max=0;
794: #endif
797: if (baij->ht) return(0);
799: baij->ht_size = (PetscInt)(factor*nz);
800: ht_size = baij->ht_size;
802: /* Allocate Memory for Hash Table */
803: PetscCalloc2(ht_size,&baij->hd,ht_size,&baij->ht);
804: HD = baij->hd;
805: HT = baij->ht;
807: /* Loop Over A */
808: for (i=0; i<a->mbs; i++) {
809: for (j=ai[i]; j<ai[i+1]; j++) {
810: row = i+rstart;
811: col = aj[j]+cstart;
813: key = row*Nbs + col + 1;
814: h1 = HASH(ht_size,key,tmp);
815: for (k=0; k<ht_size; k++) {
816: if (!HT[(h1+k)%ht_size]) {
817: HT[(h1+k)%ht_size] = key;
818: HD[(h1+k)%ht_size] = a->a + j*bs2;
819: break;
820: #if defined(PETSC_USE_INFO)
821: } else {
822: ct++;
823: #endif
824: }
825: }
826: #if defined(PETSC_USE_INFO)
827: if (k> max) max = k;
828: #endif
829: }
830: }
831: /* Loop Over B */
832: for (i=0; i<b->mbs; i++) {
833: for (j=bi[i]; j<bi[i+1]; j++) {
834: row = i+rstart;
835: col = garray[bj[j]];
836: key = row*Nbs + col + 1;
837: h1 = HASH(ht_size,key,tmp);
838: for (k=0; k<ht_size; k++) {
839: if (!HT[(h1+k)%ht_size]) {
840: HT[(h1+k)%ht_size] = key;
841: HD[(h1+k)%ht_size] = b->a + j*bs2;
842: break;
843: #if defined(PETSC_USE_INFO)
844: } else {
845: ct++;
846: #endif
847: }
848: }
849: #if defined(PETSC_USE_INFO)
850: if (k> max) max = k;
851: #endif
852: }
853: }
855: /* Print Summary */
856: #if defined(PETSC_USE_INFO)
857: for (i=0,j=0; i<ht_size; i++) {
858: if (HT[i]) j++;
859: }
860: PetscInfo2(mat,"Average Search = %5.2f,max search = %D\n",(!j)? 0.0:((PetscReal)(ct+j))/j,max);
861: #endif
862: return(0);
863: }
865: PetscErrorCode MatAssemblyBegin_MPIBAIJ(Mat mat,MatAssemblyType mode)
866: {
867: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
869: PetscInt nstash,reallocs;
872: if (baij->donotstash || mat->nooffprocentries) return(0);
874: MatStashScatterBegin_Private(mat,&mat->stash,mat->rmap->range);
875: MatStashScatterBegin_Private(mat,&mat->bstash,baij->rangebs);
876: MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);
877: PetscInfo2(mat,"Stash has %D entries,uses %D mallocs.\n",nstash,reallocs);
878: MatStashGetInfo_Private(&mat->bstash,&nstash,&reallocs);
879: PetscInfo2(mat,"Block-Stash has %D entries, uses %D mallocs.\n",nstash,reallocs);
880: return(0);
881: }
883: PetscErrorCode MatAssemblyEnd_MPIBAIJ(Mat mat,MatAssemblyType mode)
884: {
885: Mat_MPIBAIJ *baij=(Mat_MPIBAIJ*)mat->data;
886: Mat_SeqBAIJ *a =(Mat_SeqBAIJ*)baij->A->data;
888: PetscInt i,j,rstart,ncols,flg,bs2=baij->bs2;
889: PetscInt *row,*col;
890: PetscBool r1,r2,r3,other_disassembled;
891: MatScalar *val;
892: PetscMPIInt n;
895: /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */
896: if (!baij->donotstash && !mat->nooffprocentries) {
897: while (1) {
898: MatStashScatterGetMesg_Private(&mat->stash,&n,&row,&col,&val,&flg);
899: if (!flg) break;
901: for (i=0; i<n;) {
902: /* Now identify the consecutive vals belonging to the same row */
903: for (j=i,rstart=row[j]; j<n; j++) {
904: if (row[j] != rstart) break;
905: }
906: if (j < n) ncols = j-i;
907: else ncols = n-i;
908: /* Now assemble all these values with a single function call */
909: MatSetValues_MPIBAIJ(mat,1,row+i,ncols,col+i,val+i,mat->insertmode);
910: i = j;
911: }
912: }
913: MatStashScatterEnd_Private(&mat->stash);
914: /* Now process the block-stash. Since the values are stashed column-oriented,
915: set the roworiented flag to column oriented, and after MatSetValues()
916: restore the original flags */
917: r1 = baij->roworiented;
918: r2 = a->roworiented;
919: r3 = ((Mat_SeqBAIJ*)baij->B->data)->roworiented;
921: baij->roworiented = PETSC_FALSE;
922: a->roworiented = PETSC_FALSE;
924: (((Mat_SeqBAIJ*)baij->B->data))->roworiented = PETSC_FALSE; /* b->roworiented */
925: while (1) {
926: MatStashScatterGetMesg_Private(&mat->bstash,&n,&row,&col,&val,&flg);
927: if (!flg) break;
929: for (i=0; i<n;) {
930: /* Now identify the consecutive vals belonging to the same row */
931: for (j=i,rstart=row[j]; j<n; j++) {
932: if (row[j] != rstart) break;
933: }
934: if (j < n) ncols = j-i;
935: else ncols = n-i;
936: MatSetValuesBlocked_MPIBAIJ(mat,1,row+i,ncols,col+i,val+i*bs2,mat->insertmode);
937: i = j;
938: }
939: }
940: MatStashScatterEnd_Private(&mat->bstash);
942: baij->roworiented = r1;
943: a->roworiented = r2;
945: ((Mat_SeqBAIJ*)baij->B->data)->roworiented = r3; /* b->roworiented */
946: }
948: MatAssemblyBegin(baij->A,mode);
949: MatAssemblyEnd(baij->A,mode);
951: /* determine if any processor has disassembled, if so we must
952: also disassemble ourselfs, in order that we may reassemble. */
953: /*
954: if nonzero structure of submatrix B cannot change then we know that
955: no processor disassembled thus we can skip this stuff
956: */
957: if (!((Mat_SeqBAIJ*)baij->B->data)->nonew) {
958: MPIU_Allreduce(&mat->was_assembled,&other_disassembled,1,MPIU_BOOL,MPI_PROD,PetscObjectComm((PetscObject)mat));
959: if (mat->was_assembled && !other_disassembled) {
960: MatDisAssemble_MPIBAIJ(mat);
961: }
962: }
964: if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) {
965: MatSetUpMultiply_MPIBAIJ(mat);
966: }
967: MatAssemblyBegin(baij->B,mode);
968: MatAssemblyEnd(baij->B,mode);
970: #if defined(PETSC_USE_INFO)
971: if (baij->ht && mode== MAT_FINAL_ASSEMBLY) {
972: PetscInfo1(mat,"Average Hash Table Search in MatSetValues = %5.2f\n",(double)((PetscReal)baij->ht_total_ct)/baij->ht_insert_ct);
974: baij->ht_total_ct = 0;
975: baij->ht_insert_ct = 0;
976: }
977: #endif
978: if (baij->ht_flag && !baij->ht && mode == MAT_FINAL_ASSEMBLY) {
979: MatCreateHashTable_MPIBAIJ_Private(mat,baij->ht_fact);
981: mat->ops->setvalues = MatSetValues_MPIBAIJ_HT;
982: mat->ops->setvaluesblocked = MatSetValuesBlocked_MPIBAIJ_HT;
983: }
985: PetscFree2(baij->rowvalues,baij->rowindices);
987: baij->rowvalues = 0;
989: /* if no new nonzero locations are allowed in matrix then only set the matrix state the first time through */
990: if ((!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) || !((Mat_SeqBAIJ*)(baij->A->data))->nonew) {
991: PetscObjectState state = baij->A->nonzerostate + baij->B->nonzerostate;
992: MPIU_Allreduce(&state,&mat->nonzerostate,1,MPIU_INT64,MPI_SUM,PetscObjectComm((PetscObject)mat));
993: }
994: return(0);
995: }
997: extern PetscErrorCode MatView_SeqBAIJ(Mat,PetscViewer);
998: #include <petscdraw.h>
999: static PetscErrorCode MatView_MPIBAIJ_ASCIIorDraworSocket(Mat mat,PetscViewer viewer)
1000: {
1001: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
1002: PetscErrorCode ierr;
1003: PetscMPIInt rank = baij->rank;
1004: PetscInt bs = mat->rmap->bs;
1005: PetscBool iascii,isdraw;
1006: PetscViewer sviewer;
1007: PetscViewerFormat format;
1010: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
1011: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
1012: if (iascii) {
1013: PetscViewerGetFormat(viewer,&format);
1014: if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
1015: MatInfo info;
1016: MPI_Comm_rank(PetscObjectComm((PetscObject)mat),&rank);
1017: MatGetInfo(mat,MAT_LOCAL,&info);
1018: PetscViewerASCIIPushSynchronized(viewer);
1019: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] Local rows %D nz %D nz alloced %D bs %D mem %g\n",
1020: rank,mat->rmap->n,(PetscInt)info.nz_used,(PetscInt)info.nz_allocated,mat->rmap->bs,(double)info.memory);
1021: MatGetInfo(baij->A,MAT_LOCAL,&info);
1022: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] on-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used);
1023: MatGetInfo(baij->B,MAT_LOCAL,&info);
1024: PetscViewerASCIISynchronizedPrintf(viewer,"[%d] off-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used);
1025: PetscViewerFlush(viewer);
1026: PetscViewerASCIIPopSynchronized(viewer);
1027: PetscViewerASCIIPrintf(viewer,"Information on VecScatter used in matrix-vector product: \n");
1028: VecScatterView(baij->Mvctx,viewer);
1029: return(0);
1030: } else if (format == PETSC_VIEWER_ASCII_INFO) {
1031: PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);
1032: return(0);
1033: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
1034: return(0);
1035: }
1036: }
1038: if (isdraw) {
1039: PetscDraw draw;
1040: PetscBool isnull;
1041: PetscViewerDrawGetDraw(viewer,0,&draw);
1042: PetscDrawIsNull(draw,&isnull);
1043: if (isnull) return(0);
1044: }
1046: {
1047: /* assemble the entire matrix onto first processor. */
1048: Mat A;
1049: Mat_SeqBAIJ *Aloc;
1050: PetscInt M = mat->rmap->N,N = mat->cmap->N,*ai,*aj,col,i,j,k,*rvals,mbs = baij->mbs;
1051: MatScalar *a;
1052: const char *matname;
1054: /* Here we are creating a temporary matrix, so will assume MPIBAIJ is acceptable */
1055: /* Perhaps this should be the type of mat? */
1056: MatCreate(PetscObjectComm((PetscObject)mat),&A);
1057: if (!rank) {
1058: MatSetSizes(A,M,N,M,N);
1059: } else {
1060: MatSetSizes(A,0,0,M,N);
1061: }
1062: MatSetType(A,MATMPIBAIJ);
1063: MatMPIBAIJSetPreallocation(A,mat->rmap->bs,0,NULL,0,NULL);
1064: MatSetOption(A,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_FALSE);
1065: PetscLogObjectParent((PetscObject)mat,(PetscObject)A);
1067: /* copy over the A part */
1068: Aloc = (Mat_SeqBAIJ*)baij->A->data;
1069: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
1070: PetscMalloc1(bs,&rvals);
1072: for (i=0; i<mbs; i++) {
1073: rvals[0] = bs*(baij->rstartbs + i);
1074: for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1;
1075: for (j=ai[i]; j<ai[i+1]; j++) {
1076: col = (baij->cstartbs+aj[j])*bs;
1077: for (k=0; k<bs; k++) {
1078: MatSetValues_MPIBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);
1079: col++; a += bs;
1080: }
1081: }
1082: }
1083: /* copy over the B part */
1084: Aloc = (Mat_SeqBAIJ*)baij->B->data;
1085: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
1086: for (i=0; i<mbs; i++) {
1087: rvals[0] = bs*(baij->rstartbs + i);
1088: for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1;
1089: for (j=ai[i]; j<ai[i+1]; j++) {
1090: col = baij->garray[aj[j]]*bs;
1091: for (k=0; k<bs; k++) {
1092: MatSetValues_MPIBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);
1093: col++; a += bs;
1094: }
1095: }
1096: }
1097: PetscFree(rvals);
1098: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
1099: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
1100: /*
1101: Everyone has to call to draw the matrix since the graphics waits are
1102: synchronized across all processors that share the PetscDraw object
1103: */
1104: PetscViewerGetSubViewer(viewer,PETSC_COMM_SELF,&sviewer);
1105: PetscObjectGetName((PetscObject)mat,&matname);
1106: if (!rank) {
1107: PetscObjectSetName((PetscObject)((Mat_MPIBAIJ*)(A->data))->A,matname);
1108: MatView_SeqBAIJ(((Mat_MPIBAIJ*)(A->data))->A,sviewer);
1109: }
1110: PetscViewerRestoreSubViewer(viewer,PETSC_COMM_SELF,&sviewer);
1111: PetscViewerFlush(viewer);
1112: MatDestroy(&A);
1113: }
1114: return(0);
1115: }
1117: static PetscErrorCode MatView_MPIBAIJ_Binary(Mat mat,PetscViewer viewer)
1118: {
1119: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)mat->data;
1120: Mat_SeqBAIJ *A = (Mat_SeqBAIJ*)a->A->data;
1121: Mat_SeqBAIJ *B = (Mat_SeqBAIJ*)a->B->data;
1123: PetscInt i,*row_lens,*crow_lens,bs = mat->rmap->bs,j,k,bs2=a->bs2,header[4],nz,rlen;
1124: PetscInt *range=0,nzmax,*column_indices,cnt,col,*garray = a->garray,cstart = mat->cmap->rstart/bs,len,pcnt,l,ll;
1125: int fd;
1126: PetscScalar *column_values;
1127: FILE *file;
1128: PetscMPIInt rank,size,tag = ((PetscObject)viewer)->tag;
1129: PetscInt message_count,flowcontrolcount;
1132: MPI_Comm_rank(PetscObjectComm((PetscObject)mat),&rank);
1133: MPI_Comm_size(PetscObjectComm((PetscObject)mat),&size);
1134: nz = bs2*(A->nz + B->nz);
1135: rlen = mat->rmap->n;
1136: PetscViewerBinaryGetDescriptor(viewer,&fd);
1137: if (!rank) {
1138: header[0] = MAT_FILE_CLASSID;
1139: header[1] = mat->rmap->N;
1140: header[2] = mat->cmap->N;
1142: MPI_Reduce(&nz,&header[3],1,MPIU_INT,MPI_SUM,0,PetscObjectComm((PetscObject)mat));
1143: PetscBinaryWrite(fd,header,4,PETSC_INT,PETSC_TRUE);
1144: /* get largest number of rows any processor has */
1145: range = mat->rmap->range;
1146: for (i=1; i<size; i++) {
1147: rlen = PetscMax(rlen,range[i+1] - range[i]);
1148: }
1149: } else {
1150: MPI_Reduce(&nz,0,1,MPIU_INT,MPI_SUM,0,PetscObjectComm((PetscObject)mat));
1151: }
1153: PetscMalloc1(rlen/bs,&crow_lens);
1154: /* compute lengths of each row */
1155: for (i=0; i<a->mbs; i++) {
1156: crow_lens[i] = A->i[i+1] - A->i[i] + B->i[i+1] - B->i[i];
1157: }
1158: /* store the row lengths to the file */
1159: PetscViewerFlowControlStart(viewer,&message_count,&flowcontrolcount);
1160: if (!rank) {
1161: MPI_Status status;
1162: PetscMalloc1(rlen,&row_lens);
1163: rlen = (range[1] - range[0])/bs;
1164: for (i=0; i<rlen; i++) {
1165: for (j=0; j<bs; j++) {
1166: row_lens[i*bs+j] = bs*crow_lens[i];
1167: }
1168: }
1169: PetscBinaryWrite(fd,row_lens,bs*rlen,PETSC_INT,PETSC_TRUE);
1170: for (i=1; i<size; i++) {
1171: rlen = (range[i+1] - range[i])/bs;
1172: PetscViewerFlowControlStepMaster(viewer,i,&message_count,flowcontrolcount);
1173: MPI_Recv(crow_lens,rlen,MPIU_INT,i,tag,PetscObjectComm((PetscObject)mat),&status);
1174: for (k=0; k<rlen; k++) {
1175: for (j=0; j<bs; j++) {
1176: row_lens[k*bs+j] = bs*crow_lens[k];
1177: }
1178: }
1179: PetscBinaryWrite(fd,row_lens,bs*rlen,PETSC_INT,PETSC_TRUE);
1180: }
1181: PetscViewerFlowControlEndMaster(viewer,&message_count);
1182: PetscFree(row_lens);
1183: } else {
1184: PetscViewerFlowControlStepWorker(viewer,rank,&message_count);
1185: MPI_Send(crow_lens,mat->rmap->n/bs,MPIU_INT,0,tag,PetscObjectComm((PetscObject)mat));
1186: PetscViewerFlowControlEndWorker(viewer,&message_count);
1187: }
1188: PetscFree(crow_lens);
1190: /* load up the local column indices. Include for all rows not just one for each block row since process 0 does not have the
1191: information needed to make it for each row from a block row. This does require more communication but still not more than
1192: the communication needed for the nonzero values */
1193: nzmax = nz; /* space a largest processor needs */
1194: MPI_Reduce(&nz,&nzmax,1,MPIU_INT,MPI_MAX,0,PetscObjectComm((PetscObject)mat));
1195: PetscMalloc1(nzmax,&column_indices);
1196: cnt = 0;
1197: for (i=0; i<a->mbs; i++) {
1198: pcnt = cnt;
1199: for (j=B->i[i]; j<B->i[i+1]; j++) {
1200: if ((col = garray[B->j[j]]) > cstart) break;
1201: for (l=0; l<bs; l++) {
1202: column_indices[cnt++] = bs*col+l;
1203: }
1204: }
1205: for (k=A->i[i]; k<A->i[i+1]; k++) {
1206: for (l=0; l<bs; l++) {
1207: column_indices[cnt++] = bs*(A->j[k] + cstart)+l;
1208: }
1209: }
1210: for (; j<B->i[i+1]; j++) {
1211: for (l=0; l<bs; l++) {
1212: column_indices[cnt++] = bs*garray[B->j[j]]+l;
1213: }
1214: }
1215: len = cnt - pcnt;
1216: for (k=1; k<bs; k++) {
1217: PetscMemcpy(&column_indices[cnt],&column_indices[pcnt],len*sizeof(PetscInt));
1218: cnt += len;
1219: }
1220: }
1221: if (cnt != nz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_LIB,"Internal PETSc error: cnt = %D nz = %D",cnt,nz);
1223: /* store the columns to the file */
1224: PetscViewerFlowControlStart(viewer,&message_count,&flowcontrolcount);
1225: if (!rank) {
1226: MPI_Status status;
1227: PetscBinaryWrite(fd,column_indices,nz,PETSC_INT,PETSC_TRUE);
1228: for (i=1; i<size; i++) {
1229: PetscViewerFlowControlStepMaster(viewer,i,&message_count,flowcontrolcount);
1230: MPI_Recv(&cnt,1,MPIU_INT,i,tag,PetscObjectComm((PetscObject)mat),&status);
1231: MPI_Recv(column_indices,cnt,MPIU_INT,i,tag,PetscObjectComm((PetscObject)mat),&status);
1232: PetscBinaryWrite(fd,column_indices,cnt,PETSC_INT,PETSC_TRUE);
1233: }
1234: PetscViewerFlowControlEndMaster(viewer,&message_count);
1235: } else {
1236: PetscViewerFlowControlStepWorker(viewer,rank,&message_count);
1237: MPI_Send(&cnt,1,MPIU_INT,0,tag,PetscObjectComm((PetscObject)mat));
1238: MPI_Send(column_indices,cnt,MPIU_INT,0,tag,PetscObjectComm((PetscObject)mat));
1239: PetscViewerFlowControlEndWorker(viewer,&message_count);
1240: }
1241: PetscFree(column_indices);
1243: /* load up the numerical values */
1244: PetscMalloc1(nzmax,&column_values);
1245: cnt = 0;
1246: for (i=0; i<a->mbs; i++) {
1247: rlen = bs*(B->i[i+1] - B->i[i] + A->i[i+1] - A->i[i]);
1248: for (j=B->i[i]; j<B->i[i+1]; j++) {
1249: if (garray[B->j[j]] > cstart) break;
1250: for (l=0; l<bs; l++) {
1251: for (ll=0; ll<bs; ll++) {
1252: column_values[cnt + l*rlen + ll] = B->a[bs2*j+l+bs*ll];
1253: }
1254: }
1255: cnt += bs;
1256: }
1257: for (k=A->i[i]; k<A->i[i+1]; k++) {
1258: for (l=0; l<bs; l++) {
1259: for (ll=0; ll<bs; ll++) {
1260: column_values[cnt + l*rlen + ll] = A->a[bs2*k+l+bs*ll];
1261: }
1262: }
1263: cnt += bs;
1264: }
1265: for (; j<B->i[i+1]; j++) {
1266: for (l=0; l<bs; l++) {
1267: for (ll=0; ll<bs; ll++) {
1268: column_values[cnt + l*rlen + ll] = B->a[bs2*j+l+bs*ll];
1269: }
1270: }
1271: cnt += bs;
1272: }
1273: cnt += (bs-1)*rlen;
1274: }
1275: if (cnt != nz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Internal PETSc error: cnt = %D nz = %D",cnt,nz);
1277: /* store the column values to the file */
1278: PetscViewerFlowControlStart(viewer,&message_count,&flowcontrolcount);
1279: if (!rank) {
1280: MPI_Status status;
1281: PetscBinaryWrite(fd,column_values,nz,PETSC_SCALAR,PETSC_TRUE);
1282: for (i=1; i<size; i++) {
1283: PetscViewerFlowControlStepMaster(viewer,i,&message_count,flowcontrolcount);
1284: MPI_Recv(&cnt,1,MPIU_INT,i,tag,PetscObjectComm((PetscObject)mat),&status);
1285: MPI_Recv(column_values,cnt,MPIU_SCALAR,i,tag,PetscObjectComm((PetscObject)mat),&status);
1286: PetscBinaryWrite(fd,column_values,cnt,PETSC_SCALAR,PETSC_TRUE);
1287: }
1288: PetscViewerFlowControlEndMaster(viewer,&message_count);
1289: } else {
1290: PetscViewerFlowControlStepWorker(viewer,rank,&message_count);
1291: MPI_Send(&nz,1,MPIU_INT,0,tag,PetscObjectComm((PetscObject)mat));
1292: MPI_Send(column_values,nz,MPIU_SCALAR,0,tag,PetscObjectComm((PetscObject)mat));
1293: PetscViewerFlowControlEndWorker(viewer,&message_count);
1294: }
1295: PetscFree(column_values);
1297: PetscViewerBinaryGetInfoPointer(viewer,&file);
1298: if (file) {
1299: fprintf(file,"-matload_block_size %d\n",(int)mat->rmap->bs);
1300: }
1301: return(0);
1302: }
1304: PetscErrorCode MatView_MPIBAIJ(Mat mat,PetscViewer viewer)
1305: {
1307: PetscBool iascii,isdraw,issocket,isbinary;
1310: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
1311: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);
1312: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSOCKET,&issocket);
1313: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);
1314: if (iascii || isdraw || issocket) {
1315: MatView_MPIBAIJ_ASCIIorDraworSocket(mat,viewer);
1316: } else if (isbinary) {
1317: MatView_MPIBAIJ_Binary(mat,viewer);
1318: }
1319: return(0);
1320: }
1322: PetscErrorCode MatDestroy_MPIBAIJ(Mat mat)
1323: {
1324: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
1328: #if defined(PETSC_USE_LOG)
1329: PetscLogObjectState((PetscObject)mat,"Rows=%D,Cols=%D",mat->rmap->N,mat->cmap->N);
1330: #endif
1331: MatStashDestroy_Private(&mat->stash);
1332: MatStashDestroy_Private(&mat->bstash);
1333: MatDestroy(&baij->A);
1334: MatDestroy(&baij->B);
1335: #if defined(PETSC_USE_CTABLE)
1336: PetscTableDestroy(&baij->colmap);
1337: #else
1338: PetscFree(baij->colmap);
1339: #endif
1340: PetscFree(baij->garray);
1341: VecDestroy(&baij->lvec);
1342: VecScatterDestroy(&baij->Mvctx);
1343: PetscFree2(baij->rowvalues,baij->rowindices);
1344: PetscFree(baij->barray);
1345: PetscFree2(baij->hd,baij->ht);
1346: PetscFree(baij->rangebs);
1347: PetscFree(mat->data);
1349: PetscObjectChangeTypeName((PetscObject)mat,0);
1350: PetscObjectComposeFunction((PetscObject)mat,"MatStoreValues_C",NULL);
1351: PetscObjectComposeFunction((PetscObject)mat,"MatRetrieveValues_C",NULL);
1352: PetscObjectComposeFunction((PetscObject)mat,"MatMPIBAIJSetPreallocation_C",NULL);
1353: PetscObjectComposeFunction((PetscObject)mat,"MatMPIBAIJSetPreallocationCSR_C",NULL);
1354: PetscObjectComposeFunction((PetscObject)mat,"MatDiagonalScaleLocal_C",NULL);
1355: PetscObjectComposeFunction((PetscObject)mat,"MatSetHashTableFactor_C",NULL);
1356: PetscObjectComposeFunction((PetscObject)mat,"MatConvert_mpibaij_mpisbaij_C",NULL);
1357: PetscObjectComposeFunction((PetscObject)mat,"MatConvert_mpibaij_mpibstrm_C",NULL);
1358: #if defined(PETSC_HAVE_HYPRE)
1359: PetscObjectComposeFunction((PetscObject)mat,"MatConvert_mpibaij_hypre_C",NULL);
1360: #endif
1361: return(0);
1362: }
1364: PetscErrorCode MatMult_MPIBAIJ(Mat A,Vec xx,Vec yy)
1365: {
1366: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1368: PetscInt nt;
1371: VecGetLocalSize(xx,&nt);
1372: if (nt != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx");
1373: VecGetLocalSize(yy,&nt);
1374: if (nt != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible parition of A and yy");
1375: VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
1376: (*a->A->ops->mult)(a->A,xx,yy);
1377: VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
1378: (*a->B->ops->multadd)(a->B,a->lvec,yy,yy);
1379: return(0);
1380: }
1382: PetscErrorCode MatMultAdd_MPIBAIJ(Mat A,Vec xx,Vec yy,Vec zz)
1383: {
1384: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1388: VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
1389: (*a->A->ops->multadd)(a->A,xx,yy,zz);
1390: VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);
1391: (*a->B->ops->multadd)(a->B,a->lvec,zz,zz);
1392: return(0);
1393: }
1395: PetscErrorCode MatMultTranspose_MPIBAIJ(Mat A,Vec xx,Vec yy)
1396: {
1397: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1399: PetscBool merged;
1402: VecScatterGetMerged(a->Mvctx,&merged);
1403: /* do nondiagonal part */
1404: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
1405: if (!merged) {
1406: /* send it on its way */
1407: VecScatterBegin(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
1408: /* do local part */
1409: (*a->A->ops->multtranspose)(a->A,xx,yy);
1410: /* receive remote parts: note this assumes the values are not actually */
1411: /* inserted in yy until the next line */
1412: VecScatterEnd(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
1413: } else {
1414: /* do local part */
1415: (*a->A->ops->multtranspose)(a->A,xx,yy);
1416: /* send it on its way */
1417: VecScatterBegin(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
1418: /* values actually were received in the Begin() but we need to call this nop */
1419: VecScatterEnd(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);
1420: }
1421: return(0);
1422: }
1424: PetscErrorCode MatMultTransposeAdd_MPIBAIJ(Mat A,Vec xx,Vec yy,Vec zz)
1425: {
1426: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1430: /* do nondiagonal part */
1431: (*a->B->ops->multtranspose)(a->B,xx,a->lvec);
1432: /* send it on its way */
1433: VecScatterBegin(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);
1434: /* do local part */
1435: (*a->A->ops->multtransposeadd)(a->A,xx,yy,zz);
1436: /* receive remote parts: note this assumes the values are not actually */
1437: /* inserted in yy until the next line, which is true for my implementation*/
1438: /* but is not perhaps always true. */
1439: VecScatterEnd(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);
1440: return(0);
1441: }
1443: /*
1444: This only works correctly for square matrices where the subblock A->A is the
1445: diagonal block
1446: */
1447: PetscErrorCode MatGetDiagonal_MPIBAIJ(Mat A,Vec v)
1448: {
1449: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1453: if (A->rmap->N != A->cmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block");
1454: MatGetDiagonal(a->A,v);
1455: return(0);
1456: }
1458: PetscErrorCode MatScale_MPIBAIJ(Mat A,PetscScalar aa)
1459: {
1460: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1464: MatScale(a->A,aa);
1465: MatScale(a->B,aa);
1466: return(0);
1467: }
1469: PetscErrorCode MatGetRow_MPIBAIJ(Mat matin,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1470: {
1471: Mat_MPIBAIJ *mat = (Mat_MPIBAIJ*)matin->data;
1472: PetscScalar *vworkA,*vworkB,**pvA,**pvB,*v_p;
1474: PetscInt bs = matin->rmap->bs,bs2 = mat->bs2,i,*cworkA,*cworkB,**pcA,**pcB;
1475: PetscInt nztot,nzA,nzB,lrow,brstart = matin->rmap->rstart,brend = matin->rmap->rend;
1476: PetscInt *cmap,*idx_p,cstart = mat->cstartbs;
1479: if (row < brstart || row >= brend) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only local rows");
1480: if (mat->getrowactive) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Already active");
1481: mat->getrowactive = PETSC_TRUE;
1483: if (!mat->rowvalues && (idx || v)) {
1484: /*
1485: allocate enough space to hold information from the longest row.
1486: */
1487: Mat_SeqBAIJ *Aa = (Mat_SeqBAIJ*)mat->A->data,*Ba = (Mat_SeqBAIJ*)mat->B->data;
1488: PetscInt max = 1,mbs = mat->mbs,tmp;
1489: for (i=0; i<mbs; i++) {
1490: tmp = Aa->i[i+1] - Aa->i[i] + Ba->i[i+1] - Ba->i[i];
1491: if (max < tmp) max = tmp;
1492: }
1493: PetscMalloc2(max*bs2,&mat->rowvalues,max*bs2,&mat->rowindices);
1494: }
1495: lrow = row - brstart;
1497: pvA = &vworkA; pcA = &cworkA; pvB = &vworkB; pcB = &cworkB;
1498: if (!v) {pvA = 0; pvB = 0;}
1499: if (!idx) {pcA = 0; if (!v) pcB = 0;}
1500: (*mat->A->ops->getrow)(mat->A,lrow,&nzA,pcA,pvA);
1501: (*mat->B->ops->getrow)(mat->B,lrow,&nzB,pcB,pvB);
1502: nztot = nzA + nzB;
1504: cmap = mat->garray;
1505: if (v || idx) {
1506: if (nztot) {
1507: /* Sort by increasing column numbers, assuming A and B already sorted */
1508: PetscInt imark = -1;
1509: if (v) {
1510: *v = v_p = mat->rowvalues;
1511: for (i=0; i<nzB; i++) {
1512: if (cmap[cworkB[i]/bs] < cstart) v_p[i] = vworkB[i];
1513: else break;
1514: }
1515: imark = i;
1516: for (i=0; i<nzA; i++) v_p[imark+i] = vworkA[i];
1517: for (i=imark; i<nzB; i++) v_p[nzA+i] = vworkB[i];
1518: }
1519: if (idx) {
1520: *idx = idx_p = mat->rowindices;
1521: if (imark > -1) {
1522: for (i=0; i<imark; i++) {
1523: idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs;
1524: }
1525: } else {
1526: for (i=0; i<nzB; i++) {
1527: if (cmap[cworkB[i]/bs] < cstart) idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs;
1528: else break;
1529: }
1530: imark = i;
1531: }
1532: for (i=0; i<nzA; i++) idx_p[imark+i] = cstart*bs + cworkA[i];
1533: for (i=imark; i<nzB; i++) idx_p[nzA+i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ;
1534: }
1535: } else {
1536: if (idx) *idx = 0;
1537: if (v) *v = 0;
1538: }
1539: }
1540: *nz = nztot;
1541: (*mat->A->ops->restorerow)(mat->A,lrow,&nzA,pcA,pvA);
1542: (*mat->B->ops->restorerow)(mat->B,lrow,&nzB,pcB,pvB);
1543: return(0);
1544: }
1546: PetscErrorCode MatRestoreRow_MPIBAIJ(Mat mat,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1547: {
1548: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
1551: if (!baij->getrowactive) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"MatGetRow not called");
1552: baij->getrowactive = PETSC_FALSE;
1553: return(0);
1554: }
1556: PetscErrorCode MatZeroEntries_MPIBAIJ(Mat A)
1557: {
1558: Mat_MPIBAIJ *l = (Mat_MPIBAIJ*)A->data;
1562: MatZeroEntries(l->A);
1563: MatZeroEntries(l->B);
1564: return(0);
1565: }
1567: PetscErrorCode MatGetInfo_MPIBAIJ(Mat matin,MatInfoType flag,MatInfo *info)
1568: {
1569: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)matin->data;
1570: Mat A = a->A,B = a->B;
1572: PetscReal isend[5],irecv[5];
1575: info->block_size = (PetscReal)matin->rmap->bs;
1577: MatGetInfo(A,MAT_LOCAL,info);
1579: isend[0] = info->nz_used; isend[1] = info->nz_allocated; isend[2] = info->nz_unneeded;
1580: isend[3] = info->memory; isend[4] = info->mallocs;
1582: MatGetInfo(B,MAT_LOCAL,info);
1584: isend[0] += info->nz_used; isend[1] += info->nz_allocated; isend[2] += info->nz_unneeded;
1585: isend[3] += info->memory; isend[4] += info->mallocs;
1587: if (flag == MAT_LOCAL) {
1588: info->nz_used = isend[0];
1589: info->nz_allocated = isend[1];
1590: info->nz_unneeded = isend[2];
1591: info->memory = isend[3];
1592: info->mallocs = isend[4];
1593: } else if (flag == MAT_GLOBAL_MAX) {
1594: MPIU_Allreduce(isend,irecv,5,MPIU_REAL,MPIU_MAX,PetscObjectComm((PetscObject)matin));
1596: info->nz_used = irecv[0];
1597: info->nz_allocated = irecv[1];
1598: info->nz_unneeded = irecv[2];
1599: info->memory = irecv[3];
1600: info->mallocs = irecv[4];
1601: } else if (flag == MAT_GLOBAL_SUM) {
1602: MPIU_Allreduce(isend,irecv,5,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)matin));
1604: info->nz_used = irecv[0];
1605: info->nz_allocated = irecv[1];
1606: info->nz_unneeded = irecv[2];
1607: info->memory = irecv[3];
1608: info->mallocs = irecv[4];
1609: } else SETERRQ1(PetscObjectComm((PetscObject)matin),PETSC_ERR_ARG_WRONG,"Unknown MatInfoType argument %d",(int)flag);
1610: info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */
1611: info->fill_ratio_needed = 0;
1612: info->factor_mallocs = 0;
1613: return(0);
1614: }
1616: PetscErrorCode MatSetOption_MPIBAIJ(Mat A,MatOption op,PetscBool flg)
1617: {
1618: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1622: switch (op) {
1623: case MAT_NEW_NONZERO_LOCATIONS:
1624: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1625: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1626: case MAT_KEEP_NONZERO_PATTERN:
1627: case MAT_NEW_NONZERO_LOCATION_ERR:
1628: MatCheckPreallocated(A,1);
1629: MatSetOption(a->A,op,flg);
1630: MatSetOption(a->B,op,flg);
1631: break;
1632: case MAT_ROW_ORIENTED:
1633: MatCheckPreallocated(A,1);
1634: a->roworiented = flg;
1636: MatSetOption(a->A,op,flg);
1637: MatSetOption(a->B,op,flg);
1638: break;
1639: case MAT_NEW_DIAGONALS:
1640: PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);
1641: break;
1642: case MAT_IGNORE_OFF_PROC_ENTRIES:
1643: a->donotstash = flg;
1644: break;
1645: case MAT_USE_HASH_TABLE:
1646: a->ht_flag = flg;
1647: a->ht_fact = 1.39;
1648: break;
1649: case MAT_SYMMETRIC:
1650: case MAT_STRUCTURALLY_SYMMETRIC:
1651: case MAT_HERMITIAN:
1652: case MAT_SUBMAT_SINGLEIS:
1653: case MAT_SYMMETRY_ETERNAL:
1654: MatCheckPreallocated(A,1);
1655: MatSetOption(a->A,op,flg);
1656: break;
1657: default:
1658: SETERRQ1(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"unknown option %d",op);
1659: }
1660: return(0);
1661: }
1663: PetscErrorCode MatTranspose_MPIBAIJ(Mat A,MatReuse reuse,Mat *matout)
1664: {
1665: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)A->data;
1666: Mat_SeqBAIJ *Aloc;
1667: Mat B;
1669: PetscInt M =A->rmap->N,N=A->cmap->N,*ai,*aj,i,*rvals,j,k,col;
1670: PetscInt bs=A->rmap->bs,mbs=baij->mbs;
1671: MatScalar *a;
1674: if (reuse == MAT_INITIAL_MATRIX || reuse == MAT_INPLACE_MATRIX) {
1675: MatCreate(PetscObjectComm((PetscObject)A),&B);
1676: MatSetSizes(B,A->cmap->n,A->rmap->n,N,M);
1677: MatSetType(B,((PetscObject)A)->type_name);
1678: /* Do not know preallocation information, but must set block size */
1679: MatMPIBAIJSetPreallocation(B,A->rmap->bs,PETSC_DECIDE,NULL,PETSC_DECIDE,NULL);
1680: } else {
1681: B = *matout;
1682: }
1684: /* copy over the A part */
1685: Aloc = (Mat_SeqBAIJ*)baij->A->data;
1686: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
1687: PetscMalloc1(bs,&rvals);
1689: for (i=0; i<mbs; i++) {
1690: rvals[0] = bs*(baij->rstartbs + i);
1691: for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1;
1692: for (j=ai[i]; j<ai[i+1]; j++) {
1693: col = (baij->cstartbs+aj[j])*bs;
1694: for (k=0; k<bs; k++) {
1695: MatSetValues_MPIBAIJ(B,1,&col,bs,rvals,a,INSERT_VALUES);
1697: col++; a += bs;
1698: }
1699: }
1700: }
1701: /* copy over the B part */
1702: Aloc = (Mat_SeqBAIJ*)baij->B->data;
1703: ai = Aloc->i; aj = Aloc->j; a = Aloc->a;
1704: for (i=0; i<mbs; i++) {
1705: rvals[0] = bs*(baij->rstartbs + i);
1706: for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1;
1707: for (j=ai[i]; j<ai[i+1]; j++) {
1708: col = baij->garray[aj[j]]*bs;
1709: for (k=0; k<bs; k++) {
1710: MatSetValues_MPIBAIJ(B,1,&col,bs,rvals,a,INSERT_VALUES);
1711: col++;
1712: a += bs;
1713: }
1714: }
1715: }
1716: PetscFree(rvals);
1717: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
1718: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
1720: if (reuse == MAT_INITIAL_MATRIX || reuse == MAT_REUSE_MATRIX) *matout = B;
1721: else {
1722: MatHeaderMerge(A,&B);
1723: }
1724: return(0);
1725: }
1727: PetscErrorCode MatDiagonalScale_MPIBAIJ(Mat mat,Vec ll,Vec rr)
1728: {
1729: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
1730: Mat a = baij->A,b = baij->B;
1732: PetscInt s1,s2,s3;
1735: MatGetLocalSize(mat,&s2,&s3);
1736: if (rr) {
1737: VecGetLocalSize(rr,&s1);
1738: if (s1!=s3) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"right vector non-conforming local size");
1739: /* Overlap communication with computation. */
1740: VecScatterBegin(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);
1741: }
1742: if (ll) {
1743: VecGetLocalSize(ll,&s1);
1744: if (s1!=s2) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"left vector non-conforming local size");
1745: (*b->ops->diagonalscale)(b,ll,NULL);
1746: }
1747: /* scale the diagonal block */
1748: (*a->ops->diagonalscale)(a,ll,rr);
1750: if (rr) {
1751: /* Do a scatter end and then right scale the off-diagonal block */
1752: VecScatterEnd(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);
1753: (*b->ops->diagonalscale)(b,NULL,baij->lvec);
1754: }
1755: return(0);
1756: }
1758: PetscErrorCode MatZeroRows_MPIBAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
1759: {
1760: Mat_MPIBAIJ *l = (Mat_MPIBAIJ *) A->data;
1761: PetscInt *lrows;
1762: PetscInt r, len;
1766: /* get locally owned rows */
1767: MatZeroRowsMapLocal_Private(A,N,rows,&len,&lrows);
1768: /* fix right hand side if needed */
1769: if (x && b) {
1770: const PetscScalar *xx;
1771: PetscScalar *bb;
1773: VecGetArrayRead(x,&xx);
1774: VecGetArray(b,&bb);
1775: for (r = 0; r < len; ++r) bb[lrows[r]] = diag*xx[lrows[r]];
1776: VecRestoreArrayRead(x,&xx);
1777: VecRestoreArray(b,&bb);
1778: }
1780: /* actually zap the local rows */
1781: /*
1782: Zero the required rows. If the "diagonal block" of the matrix
1783: is square and the user wishes to set the diagonal we use separate
1784: code so that MatSetValues() is not called for each diagonal allocating
1785: new memory, thus calling lots of mallocs and slowing things down.
1787: */
1788: /* must zero l->B before l->A because the (diag) case below may put values into l->B*/
1789: MatZeroRows_SeqBAIJ(l->B,len,lrows,0.0,NULL,NULL);
1790: if (A->congruentlayouts == -1) { /* first time we compare rows and cols layouts */
1791: PetscBool cong;
1792: PetscLayoutCompare(A->rmap,A->cmap,&cong);
1793: if (cong) A->congruentlayouts = 1;
1794: else A->congruentlayouts = 0;
1795: }
1796: if ((diag != 0.0) && A->congruentlayouts) {
1797: MatZeroRows_SeqBAIJ(l->A,len,lrows,diag,NULL,NULL);
1798: } else if (diag != 0.0) {
1799: MatZeroRows_SeqBAIJ(l->A,len,lrows,0.0,0,0);
1800: if (((Mat_SeqBAIJ*)l->A->data)->nonew) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatZeroRows() on rectangular matrices cannot be used with the Mat options \n\
1801: MAT_NEW_NONZERO_LOCATIONS,MAT_NEW_NONZERO_LOCATION_ERR,MAT_NEW_NONZERO_ALLOCATION_ERR");
1802: for (r = 0; r < len; ++r) {
1803: const PetscInt row = lrows[r] + A->rmap->rstart;
1804: MatSetValues(A,1,&row,1,&row,&diag,INSERT_VALUES);
1805: }
1806: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
1807: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
1808: } else {
1809: MatZeroRows_SeqBAIJ(l->A,len,lrows,0.0,NULL,NULL);
1810: }
1811: PetscFree(lrows);
1813: /* only change matrix nonzero state if pattern was allowed to be changed */
1814: if (!((Mat_SeqBAIJ*)(l->A->data))->keepnonzeropattern) {
1815: PetscObjectState state = l->A->nonzerostate + l->B->nonzerostate;
1816: MPIU_Allreduce(&state,&A->nonzerostate,1,MPIU_INT64,MPI_SUM,PetscObjectComm((PetscObject)A));
1817: }
1818: return(0);
1819: }
1821: PetscErrorCode MatZeroRowsColumns_MPIBAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b)
1822: {
1823: Mat_MPIBAIJ *l = (Mat_MPIBAIJ*)A->data;
1824: PetscErrorCode ierr;
1825: PetscMPIInt n = A->rmap->n;
1826: PetscInt i,j,k,r,p = 0,len = 0,row,col,count;
1827: PetscInt *lrows,*owners = A->rmap->range;
1828: PetscSFNode *rrows;
1829: PetscSF sf;
1830: const PetscScalar *xx;
1831: PetscScalar *bb,*mask;
1832: Vec xmask,lmask;
1833: Mat_SeqBAIJ *baij = (Mat_SeqBAIJ*)l->B->data;
1834: PetscInt bs = A->rmap->bs, bs2 = baij->bs2;
1835: PetscScalar *aa;
1838: /* Create SF where leaves are input rows and roots are owned rows */
1839: PetscMalloc1(n, &lrows);
1840: for (r = 0; r < n; ++r) lrows[r] = -1;
1841: PetscMalloc1(N, &rrows);
1842: for (r = 0; r < N; ++r) {
1843: const PetscInt idx = rows[r];
1844: if (idx < 0 || A->rmap->N <= idx) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row %D out of range [0,%D)",idx,A->rmap->N);
1845: if (idx < owners[p] || owners[p+1] <= idx) { /* short-circuit the search if the last p owns this row too */
1846: PetscLayoutFindOwner(A->rmap,idx,&p);
1847: }
1848: rrows[r].rank = p;
1849: rrows[r].index = rows[r] - owners[p];
1850: }
1851: PetscSFCreate(PetscObjectComm((PetscObject) A), &sf);
1852: PetscSFSetGraph(sf, n, N, NULL, PETSC_OWN_POINTER, rrows, PETSC_OWN_POINTER);
1853: /* Collect flags for rows to be zeroed */
1854: PetscSFReduceBegin(sf, MPIU_INT, (PetscInt *) rows, lrows, MPI_LOR);
1855: PetscSFReduceEnd(sf, MPIU_INT, (PetscInt *) rows, lrows, MPI_LOR);
1856: PetscSFDestroy(&sf);
1857: /* Compress and put in row numbers */
1858: for (r = 0; r < n; ++r) if (lrows[r] >= 0) lrows[len++] = r;
1859: /* zero diagonal part of matrix */
1860: MatZeroRowsColumns(l->A,len,lrows,diag,x,b);
1861: /* handle off diagonal part of matrix */
1862: MatCreateVecs(A,&xmask,NULL);
1863: VecDuplicate(l->lvec,&lmask);
1864: VecGetArray(xmask,&bb);
1865: for (i=0; i<len; i++) bb[lrows[i]] = 1;
1866: VecRestoreArray(xmask,&bb);
1867: VecScatterBegin(l->Mvctx,xmask,lmask,ADD_VALUES,SCATTER_FORWARD);
1868: VecScatterEnd(l->Mvctx,xmask,lmask,ADD_VALUES,SCATTER_FORWARD);
1869: VecDestroy(&xmask);
1870: if (x) {
1871: VecScatterBegin(l->Mvctx,x,l->lvec,INSERT_VALUES,SCATTER_FORWARD);
1872: VecScatterEnd(l->Mvctx,x,l->lvec,INSERT_VALUES,SCATTER_FORWARD);
1873: VecGetArrayRead(l->lvec,&xx);
1874: VecGetArray(b,&bb);
1875: }
1876: VecGetArray(lmask,&mask);
1877: /* remove zeroed rows of off diagonal matrix */
1878: for (i = 0; i < len; ++i) {
1879: row = lrows[i];
1880: count = (baij->i[row/bs +1] - baij->i[row/bs])*bs;
1881: aa = ((MatScalar*)(baij->a)) + baij->i[row/bs]*bs2 + (row%bs);
1882: for (k = 0; k < count; ++k) {
1883: aa[0] = 0.0;
1884: aa += bs;
1885: }
1886: }
1887: /* loop over all elements of off process part of matrix zeroing removed columns*/
1888: for (i = 0; i < l->B->rmap->N; ++i) {
1889: row = i/bs;
1890: for (j = baij->i[row]; j < baij->i[row+1]; ++j) {
1891: for (k = 0; k < bs; ++k) {
1892: col = bs*baij->j[j] + k;
1893: if (PetscAbsScalar(mask[col])) {
1894: aa = ((MatScalar*)(baij->a)) + j*bs2 + (i%bs) + bs*k;
1895: if (x) bb[i] -= aa[0]*xx[col];
1896: aa[0] = 0.0;
1897: }
1898: }
1899: }
1900: }
1901: if (x) {
1902: VecRestoreArray(b,&bb);
1903: VecRestoreArrayRead(l->lvec,&xx);
1904: }
1905: VecRestoreArray(lmask,&mask);
1906: VecDestroy(&lmask);
1907: PetscFree(lrows);
1909: /* only change matrix nonzero state if pattern was allowed to be changed */
1910: if (!((Mat_SeqBAIJ*)(l->A->data))->keepnonzeropattern) {
1911: PetscObjectState state = l->A->nonzerostate + l->B->nonzerostate;
1912: MPIU_Allreduce(&state,&A->nonzerostate,1,MPIU_INT64,MPI_SUM,PetscObjectComm((PetscObject)A));
1913: }
1914: return(0);
1915: }
1917: PetscErrorCode MatSetUnfactored_MPIBAIJ(Mat A)
1918: {
1919: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1923: MatSetUnfactored(a->A);
1924: return(0);
1925: }
1927: static PetscErrorCode MatDuplicate_MPIBAIJ(Mat,MatDuplicateOption,Mat*);
1929: PetscErrorCode MatEqual_MPIBAIJ(Mat A,Mat B,PetscBool *flag)
1930: {
1931: Mat_MPIBAIJ *matB = (Mat_MPIBAIJ*)B->data,*matA = (Mat_MPIBAIJ*)A->data;
1932: Mat a,b,c,d;
1933: PetscBool flg;
1937: a = matA->A; b = matA->B;
1938: c = matB->A; d = matB->B;
1940: MatEqual(a,c,&flg);
1941: if (flg) {
1942: MatEqual(b,d,&flg);
1943: }
1944: MPIU_Allreduce(&flg,flag,1,MPIU_BOOL,MPI_LAND,PetscObjectComm((PetscObject)A));
1945: return(0);
1946: }
1948: PetscErrorCode MatCopy_MPIBAIJ(Mat A,Mat B,MatStructure str)
1949: {
1951: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
1952: Mat_MPIBAIJ *b = (Mat_MPIBAIJ*)B->data;
1955: /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */
1956: if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) {
1957: MatCopy_Basic(A,B,str);
1958: } else {
1959: MatCopy(a->A,b->A,str);
1960: MatCopy(a->B,b->B,str);
1961: }
1962: PetscObjectStateIncrease((PetscObject)B);
1963: return(0);
1964: }
1966: PetscErrorCode MatSetUp_MPIBAIJ(Mat A)
1967: {
1971: MatMPIBAIJSetPreallocation(A,A->rmap->bs,PETSC_DEFAULT,0,PETSC_DEFAULT,0);
1972: return(0);
1973: }
1975: PetscErrorCode MatAXPYGetPreallocation_MPIBAIJ(Mat Y,const PetscInt *yltog,Mat X,const PetscInt *xltog,PetscInt *nnz)
1976: {
1978: PetscInt bs = Y->rmap->bs,m = Y->rmap->N/bs;
1979: Mat_SeqBAIJ *x = (Mat_SeqBAIJ*)X->data;
1980: Mat_SeqBAIJ *y = (Mat_SeqBAIJ*)Y->data;
1983: MatAXPYGetPreallocation_MPIX_private(m,x->i,x->j,xltog,y->i,y->j,yltog,nnz);
1984: return(0);
1985: }
1987: PetscErrorCode MatAXPY_MPIBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
1988: {
1990: Mat_MPIBAIJ *xx=(Mat_MPIBAIJ*)X->data,*yy=(Mat_MPIBAIJ*)Y->data;
1991: PetscBLASInt bnz,one=1;
1992: Mat_SeqBAIJ *x,*y;
1993: PetscInt bs2 = Y->rmap->bs*Y->rmap->bs;
1996: if (str == SAME_NONZERO_PATTERN) {
1997: PetscScalar alpha = a;
1998: x = (Mat_SeqBAIJ*)xx->A->data;
1999: y = (Mat_SeqBAIJ*)yy->A->data;
2000: PetscBLASIntCast(x->nz*bs2,&bnz);
2001: PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one));
2002: x = (Mat_SeqBAIJ*)xx->B->data;
2003: y = (Mat_SeqBAIJ*)yy->B->data;
2004: PetscBLASIntCast(x->nz*bs2,&bnz);
2005: PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one));
2006: PetscObjectStateIncrease((PetscObject)Y);
2007: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
2008: MatAXPY_Basic(Y,a,X,str);
2009: } else {
2010: Mat B;
2011: PetscInt *nnz_d,*nnz_o,bs=Y->rmap->bs;
2012: PetscMalloc1(yy->A->rmap->N,&nnz_d);
2013: PetscMalloc1(yy->B->rmap->N,&nnz_o);
2014: MatCreate(PetscObjectComm((PetscObject)Y),&B);
2015: PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);
2016: MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);
2017: MatSetBlockSizesFromMats(B,Y,Y);
2018: MatSetType(B,MATMPIBAIJ);
2019: MatAXPYGetPreallocation_SeqBAIJ(yy->A,xx->A,nnz_d);
2020: MatAXPYGetPreallocation_MPIBAIJ(yy->B,yy->garray,xx->B,xx->garray,nnz_o);
2021: MatMPIBAIJSetPreallocation(B,bs,0,nnz_d,0,nnz_o);
2022: /* MatAXPY_BasicWithPreallocation() for BAIJ matrix is much slower than AIJ, even for bs=1 ! */
2023: MatAXPY_BasicWithPreallocation(B,Y,a,X,str);
2024: MatHeaderReplace(Y,&B);
2025: PetscFree(nnz_d);
2026: PetscFree(nnz_o);
2027: }
2028: return(0);
2029: }
2031: PetscErrorCode MatRealPart_MPIBAIJ(Mat A)
2032: {
2033: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
2037: MatRealPart(a->A);
2038: MatRealPart(a->B);
2039: return(0);
2040: }
2042: PetscErrorCode MatImaginaryPart_MPIBAIJ(Mat A)
2043: {
2044: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
2048: MatImaginaryPart(a->A);
2049: MatImaginaryPart(a->B);
2050: return(0);
2051: }
2053: PetscErrorCode MatCreateSubMatrix_MPIBAIJ(Mat mat,IS isrow,IS iscol,MatReuse call,Mat *newmat)
2054: {
2056: IS iscol_local;
2057: PetscInt csize;
2060: ISGetLocalSize(iscol,&csize);
2061: if (call == MAT_REUSE_MATRIX) {
2062: PetscObjectQuery((PetscObject)*newmat,"ISAllGather",(PetscObject*)&iscol_local);
2063: if (!iscol_local) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Submatrix passed in was not used before, cannot reuse");
2064: } else {
2065: ISAllGather(iscol,&iscol_local);
2066: }
2067: MatCreateSubMatrix_MPIBAIJ_Private(mat,isrow,iscol_local,csize,call,newmat);
2068: if (call == MAT_INITIAL_MATRIX) {
2069: PetscObjectCompose((PetscObject)*newmat,"ISAllGather",(PetscObject)iscol_local);
2070: ISDestroy(&iscol_local);
2071: }
2072: return(0);
2073: }
2075: /*
2076: Not great since it makes two copies of the submatrix, first an SeqBAIJ
2077: in local and then by concatenating the local matrices the end result.
2078: Writing it directly would be much like MatCreateSubMatrices_MPIBAIJ().
2079: This routine is used for BAIJ and SBAIJ matrices (unfortunate dependency).
2080: */
2081: PetscErrorCode MatCreateSubMatrix_MPIBAIJ_Private(Mat mat,IS isrow,IS iscol,PetscInt csize,MatReuse call,Mat *newmat)
2082: {
2084: PetscMPIInt rank,size;
2085: PetscInt i,m,n,rstart,row,rend,nz,*cwork,j,bs;
2086: PetscInt *ii,*jj,nlocal,*dlens,*olens,dlen,olen,jend,mglobal;
2087: Mat M,Mreuse;
2088: MatScalar *vwork,*aa;
2089: MPI_Comm comm;
2090: IS isrow_new, iscol_new;
2091: Mat_SeqBAIJ *aij;
2094: PetscObjectGetComm((PetscObject)mat,&comm);
2095: MPI_Comm_rank(comm,&rank);
2096: MPI_Comm_size(comm,&size);
2097: /* The compression and expansion should be avoided. Doesn't point
2098: out errors, might change the indices, hence buggey */
2099: ISCompressIndicesGeneral(mat->rmap->N,mat->rmap->n,mat->rmap->bs,1,&isrow,&isrow_new);
2100: ISCompressIndicesGeneral(mat->cmap->N,mat->cmap->n,mat->cmap->bs,1,&iscol,&iscol_new);
2102: if (call == MAT_REUSE_MATRIX) {
2103: PetscObjectQuery((PetscObject)*newmat,"SubMatrix",(PetscObject*)&Mreuse);
2104: if (!Mreuse) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Submatrix passed in was not used before, cannot reuse");
2105: MatCreateSubMatrices_MPIBAIJ_local(mat,1,&isrow_new,&iscol_new,MAT_REUSE_MATRIX,&Mreuse);
2106: } else {
2107: MatCreateSubMatrices_MPIBAIJ_local(mat,1,&isrow_new,&iscol_new,MAT_INITIAL_MATRIX,&Mreuse);
2108: }
2109: ISDestroy(&isrow_new);
2110: ISDestroy(&iscol_new);
2111: /*
2112: m - number of local rows
2113: n - number of columns (same on all processors)
2114: rstart - first row in new global matrix generated
2115: */
2116: MatGetBlockSize(mat,&bs);
2117: MatGetSize(Mreuse,&m,&n);
2118: m = m/bs;
2119: n = n/bs;
2121: if (call == MAT_INITIAL_MATRIX) {
2122: aij = (Mat_SeqBAIJ*)(Mreuse)->data;
2123: ii = aij->i;
2124: jj = aij->j;
2126: /*
2127: Determine the number of non-zeros in the diagonal and off-diagonal
2128: portions of the matrix in order to do correct preallocation
2129: */
2131: /* first get start and end of "diagonal" columns */
2132: if (csize == PETSC_DECIDE) {
2133: ISGetSize(isrow,&mglobal);
2134: if (mglobal == n*bs) { /* square matrix */
2135: nlocal = m;
2136: } else {
2137: nlocal = n/size + ((n % size) > rank);
2138: }
2139: } else {
2140: nlocal = csize/bs;
2141: }
2142: MPI_Scan(&nlocal,&rend,1,MPIU_INT,MPI_SUM,comm);
2143: rstart = rend - nlocal;
2144: if (rank == size - 1 && rend != n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Local column sizes %D do not add up to total number of columns %D",rend,n);
2146: /* next, compute all the lengths */
2147: PetscMalloc2(m+1,&dlens,m+1,&olens);
2148: for (i=0; i<m; i++) {
2149: jend = ii[i+1] - ii[i];
2150: olen = 0;
2151: dlen = 0;
2152: for (j=0; j<jend; j++) {
2153: if (*jj < rstart || *jj >= rend) olen++;
2154: else dlen++;
2155: jj++;
2156: }
2157: olens[i] = olen;
2158: dlens[i] = dlen;
2159: }
2160: MatCreate(comm,&M);
2161: MatSetSizes(M,bs*m,bs*nlocal,PETSC_DECIDE,bs*n);
2162: MatSetType(M,((PetscObject)mat)->type_name);
2163: MatMPIBAIJSetPreallocation(M,bs,0,dlens,0,olens);
2164: MatMPISBAIJSetPreallocation(M,bs,0,dlens,0,olens);
2165: PetscFree2(dlens,olens);
2166: } else {
2167: PetscInt ml,nl;
2169: M = *newmat;
2170: MatGetLocalSize(M,&ml,&nl);
2171: if (ml != m) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Previous matrix must be same size/layout as request");
2172: MatZeroEntries(M);
2173: /*
2174: The next two lines are needed so we may call MatSetValues_MPIAIJ() below directly,
2175: rather than the slower MatSetValues().
2176: */
2177: M->was_assembled = PETSC_TRUE;
2178: M->assembled = PETSC_FALSE;
2179: }
2180: MatSetOption(M,MAT_ROW_ORIENTED,PETSC_FALSE);
2181: MatGetOwnershipRange(M,&rstart,&rend);
2182: aij = (Mat_SeqBAIJ*)(Mreuse)->data;
2183: ii = aij->i;
2184: jj = aij->j;
2185: aa = aij->a;
2186: for (i=0; i<m; i++) {
2187: row = rstart/bs + i;
2188: nz = ii[i+1] - ii[i];
2189: cwork = jj; jj += nz;
2190: vwork = aa; aa += nz*bs*bs;
2191: MatSetValuesBlocked_MPIBAIJ(M,1,&row,nz,cwork,vwork,INSERT_VALUES);
2192: }
2194: MatAssemblyBegin(M,MAT_FINAL_ASSEMBLY);
2195: MatAssemblyEnd(M,MAT_FINAL_ASSEMBLY);
2196: *newmat = M;
2198: /* save submatrix used in processor for next request */
2199: if (call == MAT_INITIAL_MATRIX) {
2200: PetscObjectCompose((PetscObject)M,"SubMatrix",(PetscObject)Mreuse);
2201: PetscObjectDereference((PetscObject)Mreuse);
2202: }
2203: return(0);
2204: }
2206: PetscErrorCode MatPermute_MPIBAIJ(Mat A,IS rowp,IS colp,Mat *B)
2207: {
2208: MPI_Comm comm,pcomm;
2209: PetscInt clocal_size,nrows;
2210: const PetscInt *rows;
2211: PetscMPIInt size;
2212: IS crowp,lcolp;
2216: PetscObjectGetComm((PetscObject)A,&comm);
2217: /* make a collective version of 'rowp' */
2218: PetscObjectGetComm((PetscObject)rowp,&pcomm);
2219: if (pcomm==comm) {
2220: crowp = rowp;
2221: } else {
2222: ISGetSize(rowp,&nrows);
2223: ISGetIndices(rowp,&rows);
2224: ISCreateGeneral(comm,nrows,rows,PETSC_COPY_VALUES,&crowp);
2225: ISRestoreIndices(rowp,&rows);
2226: }
2227: ISSetPermutation(crowp);
2228: /* make a local version of 'colp' */
2229: PetscObjectGetComm((PetscObject)colp,&pcomm);
2230: MPI_Comm_size(pcomm,&size);
2231: if (size==1) {
2232: lcolp = colp;
2233: } else {
2234: ISAllGather(colp,&lcolp);
2235: }
2236: ISSetPermutation(lcolp);
2237: /* now we just get the submatrix */
2238: MatGetLocalSize(A,NULL,&clocal_size);
2239: MatCreateSubMatrix_MPIBAIJ_Private(A,crowp,lcolp,clocal_size,MAT_INITIAL_MATRIX,B);
2240: /* clean up */
2241: if (pcomm!=comm) {
2242: ISDestroy(&crowp);
2243: }
2244: if (size>1) {
2245: ISDestroy(&lcolp);
2246: }
2247: return(0);
2248: }
2250: PetscErrorCode MatGetGhosts_MPIBAIJ(Mat mat,PetscInt *nghosts,const PetscInt *ghosts[])
2251: {
2252: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*) mat->data;
2253: Mat_SeqBAIJ *B = (Mat_SeqBAIJ*)baij->B->data;
2256: if (nghosts) *nghosts = B->nbs;
2257: if (ghosts) *ghosts = baij->garray;
2258: return(0);
2259: }
2261: PetscErrorCode MatGetSeqNonzeroStructure_MPIBAIJ(Mat A,Mat *newmat)
2262: {
2263: Mat B;
2264: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
2265: Mat_SeqBAIJ *ad = (Mat_SeqBAIJ*)a->A->data,*bd = (Mat_SeqBAIJ*)a->B->data;
2266: Mat_SeqAIJ *b;
2268: PetscMPIInt size,rank,*recvcounts = 0,*displs = 0;
2269: PetscInt sendcount,i,*rstarts = A->rmap->range,n,cnt,j,bs = A->rmap->bs;
2270: PetscInt m,*garray = a->garray,*lens,*jsendbuf,*a_jsendbuf,*b_jsendbuf;
2273: MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);
2274: MPI_Comm_rank(PetscObjectComm((PetscObject)A),&rank);
2276: /* ----------------------------------------------------------------
2277: Tell every processor the number of nonzeros per row
2278: */
2279: PetscMalloc1(A->rmap->N/bs,&lens);
2280: for (i=A->rmap->rstart/bs; i<A->rmap->rend/bs; i++) {
2281: lens[i] = ad->i[i-A->rmap->rstart/bs+1] - ad->i[i-A->rmap->rstart/bs] + bd->i[i-A->rmap->rstart/bs+1] - bd->i[i-A->rmap->rstart/bs];
2282: }
2283: PetscMalloc1(2*size,&recvcounts);
2284: displs = recvcounts + size;
2285: for (i=0; i<size; i++) {
2286: recvcounts[i] = A->rmap->range[i+1]/bs - A->rmap->range[i]/bs;
2287: displs[i] = A->rmap->range[i]/bs;
2288: }
2289: #if defined(PETSC_HAVE_MPI_IN_PLACE)
2290: MPI_Allgatherv(MPI_IN_PLACE,0,MPI_DATATYPE_NULL,lens,recvcounts,displs,MPIU_INT,PetscObjectComm((PetscObject)A));
2291: #else
2292: sendcount = A->rmap->rend/bs - A->rmap->rstart/bs;
2293: MPI_Allgatherv(lens+A->rmap->rstart/bs,sendcount,MPIU_INT,lens,recvcounts,displs,MPIU_INT,PetscObjectComm((PetscObject)A));
2294: #endif
2295: /* ---------------------------------------------------------------
2296: Create the sequential matrix of the same type as the local block diagonal
2297: */
2298: MatCreate(PETSC_COMM_SELF,&B);
2299: MatSetSizes(B,A->rmap->N/bs,A->cmap->N/bs,PETSC_DETERMINE,PETSC_DETERMINE);
2300: MatSetType(B,MATSEQAIJ);
2301: MatSeqAIJSetPreallocation(B,0,lens);
2302: b = (Mat_SeqAIJ*)B->data;
2304: /*--------------------------------------------------------------------
2305: Copy my part of matrix column indices over
2306: */
2307: sendcount = ad->nz + bd->nz;
2308: jsendbuf = b->j + b->i[rstarts[rank]/bs];
2309: a_jsendbuf = ad->j;
2310: b_jsendbuf = bd->j;
2311: n = A->rmap->rend/bs - A->rmap->rstart/bs;
2312: cnt = 0;
2313: for (i=0; i<n; i++) {
2315: /* put in lower diagonal portion */
2316: m = bd->i[i+1] - bd->i[i];
2317: while (m > 0) {
2318: /* is it above diagonal (in bd (compressed) numbering) */
2319: if (garray[*b_jsendbuf] > A->rmap->rstart/bs + i) break;
2320: jsendbuf[cnt++] = garray[*b_jsendbuf++];
2321: m--;
2322: }
2324: /* put in diagonal portion */
2325: for (j=ad->i[i]; j<ad->i[i+1]; j++) {
2326: jsendbuf[cnt++] = A->rmap->rstart/bs + *a_jsendbuf++;
2327: }
2329: /* put in upper diagonal portion */
2330: while (m-- > 0) {
2331: jsendbuf[cnt++] = garray[*b_jsendbuf++];
2332: }
2333: }
2334: if (cnt != sendcount) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Corrupted PETSc matrix: nz given %D actual nz %D",sendcount,cnt);
2336: /*--------------------------------------------------------------------
2337: Gather all column indices to all processors
2338: */
2339: for (i=0; i<size; i++) {
2340: recvcounts[i] = 0;
2341: for (j=A->rmap->range[i]/bs; j<A->rmap->range[i+1]/bs; j++) {
2342: recvcounts[i] += lens[j];
2343: }
2344: }
2345: displs[0] = 0;
2346: for (i=1; i<size; i++) {
2347: displs[i] = displs[i-1] + recvcounts[i-1];
2348: }
2349: #if defined(PETSC_HAVE_MPI_IN_PLACE)
2350: MPI_Allgatherv(MPI_IN_PLACE,0,MPI_DATATYPE_NULL,b->j,recvcounts,displs,MPIU_INT,PetscObjectComm((PetscObject)A));
2351: #else
2352: MPI_Allgatherv(jsendbuf,sendcount,MPIU_INT,b->j,recvcounts,displs,MPIU_INT,PetscObjectComm((PetscObject)A));
2353: #endif
2354: /*--------------------------------------------------------------------
2355: Assemble the matrix into useable form (note numerical values not yet set)
2356: */
2357: /* set the b->ilen (length of each row) values */
2358: PetscMemcpy(b->ilen,lens,(A->rmap->N/bs)*sizeof(PetscInt));
2359: /* set the b->i indices */
2360: b->i[0] = 0;
2361: for (i=1; i<=A->rmap->N/bs; i++) {
2362: b->i[i] = b->i[i-1] + lens[i-1];
2363: }
2364: PetscFree(lens);
2365: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
2366: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
2367: PetscFree(recvcounts);
2369: if (A->symmetric) {
2370: MatSetOption(B,MAT_SYMMETRIC,PETSC_TRUE);
2371: } else if (A->hermitian) {
2372: MatSetOption(B,MAT_HERMITIAN,PETSC_TRUE);
2373: } else if (A->structurally_symmetric) {
2374: MatSetOption(B,MAT_STRUCTURALLY_SYMMETRIC,PETSC_TRUE);
2375: }
2376: *newmat = B;
2377: return(0);
2378: }
2380: PetscErrorCode MatSOR_MPIBAIJ(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
2381: {
2382: Mat_MPIBAIJ *mat = (Mat_MPIBAIJ*)matin->data;
2384: Vec bb1 = 0;
2387: if (flag == SOR_APPLY_UPPER) {
2388: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2389: return(0);
2390: }
2392: if (its > 1 || ~flag & SOR_ZERO_INITIAL_GUESS) {
2393: VecDuplicate(bb,&bb1);
2394: }
2396: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP) {
2397: if (flag & SOR_ZERO_INITIAL_GUESS) {
2398: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2399: its--;
2400: }
2402: while (its--) {
2403: VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2404: VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2406: /* update rhs: bb1 = bb - B*x */
2407: VecScale(mat->lvec,-1.0);
2408: (*mat->B->ops->multadd)(mat->B,mat->lvec,bb,bb1);
2410: /* local sweep */
2411: (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,1,xx);
2412: }
2413: } else if (flag & SOR_LOCAL_FORWARD_SWEEP) {
2414: if (flag & SOR_ZERO_INITIAL_GUESS) {
2415: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2416: its--;
2417: }
2418: while (its--) {
2419: VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2420: VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2422: /* update rhs: bb1 = bb - B*x */
2423: VecScale(mat->lvec,-1.0);
2424: (*mat->B->ops->multadd)(mat->B,mat->lvec,bb,bb1);
2426: /* local sweep */
2427: (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_FORWARD_SWEEP,fshift,lits,1,xx);
2428: }
2429: } else if (flag & SOR_LOCAL_BACKWARD_SWEEP) {
2430: if (flag & SOR_ZERO_INITIAL_GUESS) {
2431: (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);
2432: its--;
2433: }
2434: while (its--) {
2435: VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2436: VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);
2438: /* update rhs: bb1 = bb - B*x */
2439: VecScale(mat->lvec,-1.0);
2440: (*mat->B->ops->multadd)(mat->B,mat->lvec,bb,bb1);
2442: /* local sweep */
2443: (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_BACKWARD_SWEEP,fshift,lits,1,xx);
2444: }
2445: } else SETERRQ(PetscObjectComm((PetscObject)matin),PETSC_ERR_SUP,"Parallel version of SOR requested not supported");
2447: VecDestroy(&bb1);
2448: return(0);
2449: }
2451: PetscErrorCode MatGetColumnNorms_MPIBAIJ(Mat A,NormType type,PetscReal *norms)
2452: {
2454: Mat_MPIBAIJ *aij = (Mat_MPIBAIJ*)A->data;
2455: PetscInt N,i,*garray = aij->garray;
2456: PetscInt ib,jb,bs = A->rmap->bs;
2457: Mat_SeqBAIJ *a_aij = (Mat_SeqBAIJ*) aij->A->data;
2458: MatScalar *a_val = a_aij->a;
2459: Mat_SeqBAIJ *b_aij = (Mat_SeqBAIJ*) aij->B->data;
2460: MatScalar *b_val = b_aij->a;
2461: PetscReal *work;
2464: MatGetSize(A,NULL,&N);
2465: PetscCalloc1(N,&work);
2466: if (type == NORM_2) {
2467: for (i=a_aij->i[0]; i<a_aij->i[aij->A->rmap->n/bs]; i++) {
2468: for (jb=0; jb<bs; jb++) {
2469: for (ib=0; ib<bs; ib++) {
2470: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscAbsScalar(*a_val * *a_val);
2471: a_val++;
2472: }
2473: }
2474: }
2475: for (i=b_aij->i[0]; i<b_aij->i[aij->B->rmap->n/bs]; i++) {
2476: for (jb=0; jb<bs; jb++) {
2477: for (ib=0; ib<bs; ib++) {
2478: work[garray[b_aij->j[i]] * bs + jb] += PetscAbsScalar(*b_val * *b_val);
2479: b_val++;
2480: }
2481: }
2482: }
2483: } else if (type == NORM_1) {
2484: for (i=a_aij->i[0]; i<a_aij->i[aij->A->rmap->n/bs]; i++) {
2485: for (jb=0; jb<bs; jb++) {
2486: for (ib=0; ib<bs; ib++) {
2487: work[A->cmap->rstart + a_aij->j[i] * bs + jb] += PetscAbsScalar(*a_val);
2488: a_val++;
2489: }
2490: }
2491: }
2492: for (i=b_aij->i[0]; i<b_aij->i[aij->B->rmap->n/bs]; i++) {
2493: for (jb=0; jb<bs; jb++) {
2494: for (ib=0; ib<bs; ib++) {
2495: work[garray[b_aij->j[i]] * bs + jb] += PetscAbsScalar(*b_val);
2496: b_val++;
2497: }
2498: }
2499: }
2500: } else if (type == NORM_INFINITY) {
2501: for (i=a_aij->i[0]; i<a_aij->i[aij->A->rmap->n/bs]; i++) {
2502: for (jb=0; jb<bs; jb++) {
2503: for (ib=0; ib<bs; ib++) {
2504: int col = A->cmap->rstart + a_aij->j[i] * bs + jb;
2505: work[col] = PetscMax(PetscAbsScalar(*a_val), work[col]);
2506: a_val++;
2507: }
2508: }
2509: }
2510: for (i=b_aij->i[0]; i<b_aij->i[aij->B->rmap->n/bs]; i++) {
2511: for (jb=0; jb<bs; jb++) {
2512: for (ib=0; ib<bs; ib++) {
2513: int col = garray[b_aij->j[i]] * bs + jb;
2514: work[col] = PetscMax(PetscAbsScalar(*b_val), work[col]);
2515: b_val++;
2516: }
2517: }
2518: }
2519: } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONG,"Unknown NormType");
2520: if (type == NORM_INFINITY) {
2521: MPIU_Allreduce(work,norms,N,MPIU_REAL,MPIU_MAX,PetscObjectComm((PetscObject)A));
2522: } else {
2523: MPIU_Allreduce(work,norms,N,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)A));
2524: }
2525: PetscFree(work);
2526: if (type == NORM_2) {
2527: for (i=0; i<N; i++) norms[i] = PetscSqrtReal(norms[i]);
2528: }
2529: return(0);
2530: }
2532: PetscErrorCode MatInvertBlockDiagonal_MPIBAIJ(Mat A,const PetscScalar **values)
2533: {
2534: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*) A->data;
2538: MatInvertBlockDiagonal(a->A,values);
2539: A->factorerrortype = a->A->factorerrortype;
2540: A->factorerror_zeropivot_value = a->A->factorerror_zeropivot_value;
2541: A->factorerror_zeropivot_row = a->A->factorerror_zeropivot_row;
2542: return(0);
2543: }
2545: PetscErrorCode MatShift_MPIBAIJ(Mat Y,PetscScalar a)
2546: {
2548: Mat_MPIBAIJ *maij = (Mat_MPIBAIJ*)Y->data;
2549: Mat_SeqBAIJ *aij = (Mat_SeqBAIJ*)maij->A->data;
2552: if (!Y->preallocated) {
2553: MatMPIBAIJSetPreallocation(Y,Y->rmap->bs,1,NULL,0,NULL);
2554: } else if (!aij->nz) {
2555: PetscInt nonew = aij->nonew;
2556: MatSeqBAIJSetPreallocation(maij->A,Y->rmap->bs,1,NULL);
2557: aij->nonew = nonew;
2558: }
2559: MatShift_Basic(Y,a);
2560: return(0);
2561: }
2563: PetscErrorCode MatMissingDiagonal_MPIBAIJ(Mat A,PetscBool *missing,PetscInt *d)
2564: {
2565: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
2569: if (A->rmap->n != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only works for square matrices");
2570: MatMissingDiagonal(a->A,missing,d);
2571: if (d) {
2572: PetscInt rstart;
2573: MatGetOwnershipRange(A,&rstart,NULL);
2574: *d += rstart/A->rmap->bs;
2576: }
2577: return(0);
2578: }
2580: PetscErrorCode MatGetDiagonalBlock_MPIBAIJ(Mat A,Mat *a)
2581: {
2583: *a = ((Mat_MPIBAIJ*)A->data)->A;
2584: return(0);
2585: }
2587: /* -------------------------------------------------------------------*/
2588: static struct _MatOps MatOps_Values = {MatSetValues_MPIBAIJ,
2589: MatGetRow_MPIBAIJ,
2590: MatRestoreRow_MPIBAIJ,
2591: MatMult_MPIBAIJ,
2592: /* 4*/ MatMultAdd_MPIBAIJ,
2593: MatMultTranspose_MPIBAIJ,
2594: MatMultTransposeAdd_MPIBAIJ,
2595: 0,
2596: 0,
2597: 0,
2598: /*10*/ 0,
2599: 0,
2600: 0,
2601: MatSOR_MPIBAIJ,
2602: MatTranspose_MPIBAIJ,
2603: /*15*/ MatGetInfo_MPIBAIJ,
2604: MatEqual_MPIBAIJ,
2605: MatGetDiagonal_MPIBAIJ,
2606: MatDiagonalScale_MPIBAIJ,
2607: MatNorm_MPIBAIJ,
2608: /*20*/ MatAssemblyBegin_MPIBAIJ,
2609: MatAssemblyEnd_MPIBAIJ,
2610: MatSetOption_MPIBAIJ,
2611: MatZeroEntries_MPIBAIJ,
2612: /*24*/ MatZeroRows_MPIBAIJ,
2613: 0,
2614: 0,
2615: 0,
2616: 0,
2617: /*29*/ MatSetUp_MPIBAIJ,
2618: 0,
2619: 0,
2620: MatGetDiagonalBlock_MPIBAIJ,
2621: 0,
2622: /*34*/ MatDuplicate_MPIBAIJ,
2623: 0,
2624: 0,
2625: 0,
2626: 0,
2627: /*39*/ MatAXPY_MPIBAIJ,
2628: MatCreateSubMatrices_MPIBAIJ,
2629: MatIncreaseOverlap_MPIBAIJ,
2630: MatGetValues_MPIBAIJ,
2631: MatCopy_MPIBAIJ,
2632: /*44*/ 0,
2633: MatScale_MPIBAIJ,
2634: MatShift_MPIBAIJ,
2635: 0,
2636: MatZeroRowsColumns_MPIBAIJ,
2637: /*49*/ 0,
2638: 0,
2639: 0,
2640: 0,
2641: 0,
2642: /*54*/ MatFDColoringCreate_MPIXAIJ,
2643: 0,
2644: MatSetUnfactored_MPIBAIJ,
2645: MatPermute_MPIBAIJ,
2646: MatSetValuesBlocked_MPIBAIJ,
2647: /*59*/ MatCreateSubMatrix_MPIBAIJ,
2648: MatDestroy_MPIBAIJ,
2649: MatView_MPIBAIJ,
2650: 0,
2651: 0,
2652: /*64*/ 0,
2653: 0,
2654: 0,
2655: 0,
2656: 0,
2657: /*69*/ MatGetRowMaxAbs_MPIBAIJ,
2658: 0,
2659: 0,
2660: 0,
2661: 0,
2662: /*74*/ 0,
2663: MatFDColoringApply_BAIJ,
2664: 0,
2665: 0,
2666: 0,
2667: /*79*/ 0,
2668: 0,
2669: 0,
2670: 0,
2671: MatLoad_MPIBAIJ,
2672: /*84*/ 0,
2673: 0,
2674: 0,
2675: 0,
2676: 0,
2677: /*89*/ 0,
2678: 0,
2679: 0,
2680: 0,
2681: 0,
2682: /*94*/ 0,
2683: 0,
2684: 0,
2685: 0,
2686: 0,
2687: /*99*/ 0,
2688: 0,
2689: 0,
2690: 0,
2691: 0,
2692: /*104*/0,
2693: MatRealPart_MPIBAIJ,
2694: MatImaginaryPart_MPIBAIJ,
2695: 0,
2696: 0,
2697: /*109*/0,
2698: 0,
2699: 0,
2700: 0,
2701: MatMissingDiagonal_MPIBAIJ,
2702: /*114*/MatGetSeqNonzeroStructure_MPIBAIJ,
2703: 0,
2704: MatGetGhosts_MPIBAIJ,
2705: 0,
2706: 0,
2707: /*119*/0,
2708: 0,
2709: 0,
2710: 0,
2711: MatGetMultiProcBlock_MPIBAIJ,
2712: /*124*/0,
2713: MatGetColumnNorms_MPIBAIJ,
2714: MatInvertBlockDiagonal_MPIBAIJ,
2715: 0,
2716: 0,
2717: /*129*/ 0,
2718: 0,
2719: 0,
2720: 0,
2721: 0,
2722: /*134*/ 0,
2723: 0,
2724: 0,
2725: 0,
2726: 0,
2727: /*139*/ MatSetBlockSizes_Default,
2728: 0,
2729: 0,
2730: MatFDColoringSetUp_MPIXAIJ,
2731: 0,
2732: /*144*/MatCreateMPIMatConcatenateSeqMat_MPIBAIJ
2733: };
2736: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPISBAIJ(Mat, MatType,MatReuse,Mat*);
2738: PetscErrorCode MatMPIBAIJSetPreallocationCSR_MPIBAIJ(Mat B,PetscInt bs,const PetscInt ii[],const PetscInt jj[],const PetscScalar V[])
2739: {
2740: PetscInt m,rstart,cstart,cend;
2741: PetscInt i,j,dlen,olen,nz,nz_max=0,*d_nnz=0,*o_nnz=0;
2742: const PetscInt *JJ =0;
2743: PetscScalar *values=0;
2744: PetscBool roworiented = ((Mat_MPIBAIJ*)B->data)->roworiented;
2748: PetscLayoutSetBlockSize(B->rmap,bs);
2749: PetscLayoutSetBlockSize(B->cmap,bs);
2750: PetscLayoutSetUp(B->rmap);
2751: PetscLayoutSetUp(B->cmap);
2752: PetscLayoutGetBlockSize(B->rmap,&bs);
2753: m = B->rmap->n/bs;
2754: rstart = B->rmap->rstart/bs;
2755: cstart = B->cmap->rstart/bs;
2756: cend = B->cmap->rend/bs;
2758: if (ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"ii[0] must be 0 but it is %D",ii[0]);
2759: PetscMalloc2(m,&d_nnz,m,&o_nnz);
2760: for (i=0; i<m; i++) {
2761: nz = ii[i+1] - ii[i];
2762: if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Local row %D has a negative number of columns %D",i,nz);
2763: nz_max = PetscMax(nz_max,nz);
2764: dlen = 0;
2765: olen = 0;
2766: JJ = jj + ii[i];
2767: for (j=0; j<nz; j++) {
2768: if (*JJ < cstart || *JJ >= cend) olen++;
2769: else dlen++;
2770: JJ++;
2771: }
2772: d_nnz[i] = dlen;
2773: o_nnz[i] = olen;
2774: }
2775: MatMPIBAIJSetPreallocation(B,bs,0,d_nnz,0,o_nnz);
2776: PetscFree2(d_nnz,o_nnz);
2778: values = (PetscScalar*)V;
2779: if (!values) {
2780: PetscCalloc1(bs*bs*nz_max,&values);
2781: }
2782: for (i=0; i<m; i++) {
2783: PetscInt row = i + rstart;
2784: PetscInt ncols = ii[i+1] - ii[i];
2785: const PetscInt *icols = jj + ii[i];
2786: if (!roworiented) { /* block ordering matches the non-nested layout of MatSetValues so we can insert entire rows */
2787: const PetscScalar *svals = values + (V ? (bs*bs*ii[i]) : 0);
2788: MatSetValuesBlocked_MPIBAIJ(B,1,&row,ncols,icols,svals,INSERT_VALUES);
2789: } else { /* block ordering does not match so we can only insert one block at a time. */
2790: PetscInt j;
2791: for (j=0; j<ncols; j++) {
2792: const PetscScalar *svals = values + (V ? (bs*bs*(ii[i]+j)) : 0);
2793: MatSetValuesBlocked_MPIBAIJ(B,1,&row,1,&icols[j],svals,INSERT_VALUES);
2794: }
2795: }
2796: }
2798: if (!V) { PetscFree(values); }
2799: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
2800: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
2801: MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);
2802: return(0);
2803: }
2805: /*@C
2806: MatMPIBAIJSetPreallocationCSR - Allocates memory for a sparse parallel matrix in BAIJ format
2807: (the default parallel PETSc format).
2809: Collective on MPI_Comm
2811: Input Parameters:
2812: + B - the matrix
2813: . bs - the block size
2814: . i - the indices into j for the start of each local row (starts with zero)
2815: . j - the column indices for each local row (starts with zero) these must be sorted for each row
2816: - v - optional values in the matrix
2818: Level: developer
2820: Notes: The order of the entries in values is specified by the MatOption MAT_ROW_ORIENTED. For example, C programs
2821: may want to use the default MAT_ROW_ORIENTED=PETSC_TRUE and use an array v[nnz][bs][bs] where the second index is
2822: over rows within a block and the last index is over columns within a block row. Fortran programs will likely set
2823: MAT_ROW_ORIENTED=PETSC_FALSE and use a Fortran array v(bs,bs,nnz) in which the first index is over rows within a
2824: block column and the second index is over columns within a block.
2826: .keywords: matrix, aij, compressed row, sparse, parallel
2828: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatMPIBAIJSetPreallocation(), MatCreateAIJ(), MPIAIJ, MatCreateMPIBAIJWithArrays(), MPIBAIJ
2829: @*/
2830: PetscErrorCode MatMPIBAIJSetPreallocationCSR(Mat B,PetscInt bs,const PetscInt i[],const PetscInt j[], const PetscScalar v[])
2831: {
2838: PetscTryMethod(B,"MatMPIBAIJSetPreallocationCSR_C",(Mat,PetscInt,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,bs,i,j,v));
2839: return(0);
2840: }
2842: PetscErrorCode MatMPIBAIJSetPreallocation_MPIBAIJ(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt *d_nnz,PetscInt o_nz,const PetscInt *o_nnz)
2843: {
2844: Mat_MPIBAIJ *b;
2846: PetscInt i;
2849: MatSetBlockSize(B,PetscAbs(bs));
2850: PetscLayoutSetUp(B->rmap);
2851: PetscLayoutSetUp(B->cmap);
2852: PetscLayoutGetBlockSize(B->rmap,&bs);
2854: if (d_nnz) {
2855: for (i=0; i<B->rmap->n/bs; i++) {
2856: if (d_nnz[i] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"d_nnz cannot be less than -1: local row %D value %D",i,d_nnz[i]);
2857: }
2858: }
2859: if (o_nnz) {
2860: for (i=0; i<B->rmap->n/bs; i++) {
2861: if (o_nnz[i] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"o_nnz cannot be less than -1: local row %D value %D",i,o_nnz[i]);
2862: }
2863: }
2865: b = (Mat_MPIBAIJ*)B->data;
2866: b->bs2 = bs*bs;
2867: b->mbs = B->rmap->n/bs;
2868: b->nbs = B->cmap->n/bs;
2869: b->Mbs = B->rmap->N/bs;
2870: b->Nbs = B->cmap->N/bs;
2872: for (i=0; i<=b->size; i++) {
2873: b->rangebs[i] = B->rmap->range[i]/bs;
2874: }
2875: b->rstartbs = B->rmap->rstart/bs;
2876: b->rendbs = B->rmap->rend/bs;
2877: b->cstartbs = B->cmap->rstart/bs;
2878: b->cendbs = B->cmap->rend/bs;
2880: #if defined(PETSC_USE_CTABLE)
2881: PetscTableDestroy(&b->colmap);
2882: #else
2883: PetscFree(b->colmap);
2884: #endif
2885: PetscFree(b->garray);
2886: VecDestroy(&b->lvec);
2887: VecScatterDestroy(&b->Mvctx);
2889: /* Because the B will have been resized we simply destroy it and create a new one each time */
2890: MatDestroy(&b->B);
2891: MatCreate(PETSC_COMM_SELF,&b->B);
2892: MatSetSizes(b->B,B->rmap->n,B->cmap->N,B->rmap->n,B->cmap->N);
2893: MatSetType(b->B,MATSEQBAIJ);
2894: PetscLogObjectParent((PetscObject)B,(PetscObject)b->B);
2896: if (!B->preallocated) {
2897: MatCreate(PETSC_COMM_SELF,&b->A);
2898: MatSetSizes(b->A,B->rmap->n,B->cmap->n,B->rmap->n,B->cmap->n);
2899: MatSetType(b->A,MATSEQBAIJ);
2900: PetscLogObjectParent((PetscObject)B,(PetscObject)b->A);
2901: MatStashCreate_Private(PetscObjectComm((PetscObject)B),bs,&B->bstash);
2902: }
2904: MatSeqBAIJSetPreallocation(b->A,bs,d_nz,d_nnz);
2905: MatSeqBAIJSetPreallocation(b->B,bs,o_nz,o_nnz);
2906: B->preallocated = PETSC_TRUE;
2907: B->was_assembled = PETSC_FALSE;
2908: B->assembled = PETSC_FALSE;
2909: return(0);
2910: }
2912: extern PetscErrorCode MatDiagonalScaleLocal_MPIBAIJ(Mat,Vec);
2913: extern PetscErrorCode MatSetHashTableFactor_MPIBAIJ(Mat,PetscReal);
2915: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIAdj(Mat B, MatType newtype,MatReuse reuse,Mat *adj)
2916: {
2917: Mat_MPIBAIJ *b = (Mat_MPIBAIJ*)B->data;
2919: Mat_SeqBAIJ *d = (Mat_SeqBAIJ*) b->A->data,*o = (Mat_SeqBAIJ*) b->B->data;
2920: PetscInt M = B->rmap->n/B->rmap->bs,i,*ii,*jj,cnt,j,k,rstart = B->rmap->rstart/B->rmap->bs;
2921: const PetscInt *id = d->i, *jd = d->j, *io = o->i, *jo = o->j, *garray = b->garray;
2924: PetscMalloc1(M+1,&ii);
2925: ii[0] = 0;
2926: for (i=0; i<M; i++) {
2927: if ((id[i+1] - id[i]) < 0) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Indices wrong %D %D %D",i,id[i],id[i+1]);
2928: if ((io[i+1] - io[i]) < 0) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Indices wrong %D %D %D",i,io[i],io[i+1]);
2929: ii[i+1] = ii[i] + id[i+1] - id[i] + io[i+1] - io[i];
2930: /* remove one from count of matrix has diagonal */
2931: for (j=id[i]; j<id[i+1]; j++) {
2932: if (jd[j] == i) {ii[i+1]--;break;}
2933: }
2934: }
2935: PetscMalloc1(ii[M],&jj);
2936: cnt = 0;
2937: for (i=0; i<M; i++) {
2938: for (j=io[i]; j<io[i+1]; j++) {
2939: if (garray[jo[j]] > rstart) break;
2940: jj[cnt++] = garray[jo[j]];
2941: }
2942: for (k=id[i]; k<id[i+1]; k++) {
2943: if (jd[k] != i) {
2944: jj[cnt++] = rstart + jd[k];
2945: }
2946: }
2947: for (; j<io[i+1]; j++) {
2948: jj[cnt++] = garray[jo[j]];
2949: }
2950: }
2951: MatCreateMPIAdj(PetscObjectComm((PetscObject)B),M,B->cmap->N/B->rmap->bs,ii,jj,NULL,adj);
2952: return(0);
2953: }
2955: #include <../src/mat/impls/aij/mpi/mpiaij.h>
2957: PETSC_INTERN PetscErrorCode MatConvert_SeqBAIJ_SeqAIJ(Mat,MatType,MatReuse,Mat*);
2959: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIAIJ(Mat A,MatType newtype,MatReuse reuse,Mat *newmat)
2960: {
2962: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
2963: Mat B;
2964: Mat_MPIAIJ *b;
2967: if (!A->assembled) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"Matrix must be assembled");
2969: if (reuse == MAT_REUSE_MATRIX) {
2970: B = *newmat;
2971: } else {
2972: MatCreate(PetscObjectComm((PetscObject)A),&B);
2973: MatSetType(B,MATMPIAIJ);
2974: MatSetSizes(B,A->rmap->n,A->cmap->n,A->rmap->N,A->cmap->N);
2975: MatSetBlockSizes(B,A->rmap->bs,A->cmap->bs);
2976: MatSeqAIJSetPreallocation(B,0,NULL);
2977: MatMPIAIJSetPreallocation(B,0,NULL,0,NULL);
2978: }
2979: b = (Mat_MPIAIJ*) B->data;
2981: if (reuse == MAT_REUSE_MATRIX) {
2982: MatConvert_SeqBAIJ_SeqAIJ(a->A, MATSEQAIJ, MAT_REUSE_MATRIX, &b->A);
2983: MatConvert_SeqBAIJ_SeqAIJ(a->B, MATSEQAIJ, MAT_REUSE_MATRIX, &b->B);
2984: } else {
2985: MatDestroy(&b->A);
2986: MatDestroy(&b->B);
2987: MatDisAssemble_MPIBAIJ(A);
2988: MatConvert_SeqBAIJ_SeqAIJ(a->A, MATSEQAIJ, MAT_INITIAL_MATRIX, &b->A);
2989: MatConvert_SeqBAIJ_SeqAIJ(a->B, MATSEQAIJ, MAT_INITIAL_MATRIX, &b->B);
2990: MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);
2991: MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);
2992: }
2993: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
2994: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
2996: if (reuse == MAT_INPLACE_MATRIX) {
2997: MatHeaderReplace(A,&B);
2998: } else {
2999: *newmat = B;
3000: }
3001: return(0);
3002: }
3004: /*MC
3005: MATMPIBAIJ - MATMPIBAIJ = "mpibaij" - A matrix type to be used for distributed block sparse matrices.
3007: Options Database Keys:
3008: + -mat_type mpibaij - sets the matrix type to "mpibaij" during a call to MatSetFromOptions()
3009: . -mat_block_size <bs> - set the blocksize used to store the matrix
3010: - -mat_use_hash_table <fact>
3012: Level: beginner
3014: .seealso: MatCreateMPIBAIJ
3015: M*/
3017: PETSC_INTERN PetscErrorCode MatConvert_MPIBAIJ_MPIBSTRM(Mat,MatType,MatReuse,Mat*);
3019: PETSC_EXTERN PetscErrorCode MatCreate_MPIBAIJ(Mat B)
3020: {
3021: Mat_MPIBAIJ *b;
3023: PetscBool flg = PETSC_FALSE;
3026: PetscNewLog(B,&b);
3027: B->data = (void*)b;
3029: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
3030: B->assembled = PETSC_FALSE;
3032: B->insertmode = NOT_SET_VALUES;
3033: MPI_Comm_rank(PetscObjectComm((PetscObject)B),&b->rank);
3034: MPI_Comm_size(PetscObjectComm((PetscObject)B),&b->size);
3036: /* build local table of row and column ownerships */
3037: PetscMalloc1(b->size+1,&b->rangebs);
3039: /* build cache for off array entries formed */
3040: MatStashCreate_Private(PetscObjectComm((PetscObject)B),1,&B->stash);
3042: b->donotstash = PETSC_FALSE;
3043: b->colmap = NULL;
3044: b->garray = NULL;
3045: b->roworiented = PETSC_TRUE;
3047: /* stuff used in block assembly */
3048: b->barray = 0;
3050: /* stuff used for matrix vector multiply */
3051: b->lvec = 0;
3052: b->Mvctx = 0;
3054: /* stuff for MatGetRow() */
3055: b->rowindices = 0;
3056: b->rowvalues = 0;
3057: b->getrowactive = PETSC_FALSE;
3059: /* hash table stuff */
3060: b->ht = 0;
3061: b->hd = 0;
3062: b->ht_size = 0;
3063: b->ht_flag = PETSC_FALSE;
3064: b->ht_fact = 0;
3065: b->ht_total_ct = 0;
3066: b->ht_insert_ct = 0;
3068: /* stuff for MatCreateSubMatrices_MPIBAIJ_local() */
3069: b->ijonly = PETSC_FALSE;
3072: PetscObjectComposeFunction((PetscObject)B,"MatConvert_mpibaij_mpiadj_C",MatConvert_MPIBAIJ_MPIAdj);
3073: PetscObjectComposeFunction((PetscObject)B,"MatConvert_mpibaij_mpiaij_C",MatConvert_MPIBAIJ_MPIAIJ);
3074: PetscObjectComposeFunction((PetscObject)B,"MatConvert_mpibaij_mpisbaij_C",MatConvert_MPIBAIJ_MPISBAIJ);
3075: #if defined(PETSC_HAVE_HYPRE)
3076: PetscObjectComposeFunction((PetscObject)B,"MatConvert_mpibaij_hypre_C",MatConvert_AIJ_HYPRE);
3077: #endif
3078: PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_MPIBAIJ);
3079: PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_MPIBAIJ);
3080: PetscObjectComposeFunction((PetscObject)B,"MatMPIBAIJSetPreallocation_C",MatMPIBAIJSetPreallocation_MPIBAIJ);
3081: PetscObjectComposeFunction((PetscObject)B,"MatMPIBAIJSetPreallocationCSR_C",MatMPIBAIJSetPreallocationCSR_MPIBAIJ);
3082: PetscObjectComposeFunction((PetscObject)B,"MatDiagonalScaleLocal_C",MatDiagonalScaleLocal_MPIBAIJ);
3083: PetscObjectComposeFunction((PetscObject)B,"MatSetHashTableFactor_C",MatSetHashTableFactor_MPIBAIJ);
3084: PetscObjectChangeTypeName((PetscObject)B,MATMPIBAIJ);
3086: PetscOptionsBegin(PetscObjectComm((PetscObject)B),NULL,"Options for loading MPIBAIJ matrix 1","Mat");
3087: PetscOptionsName("-mat_use_hash_table","Use hash table to save time in constructing matrix","MatSetOption",&flg);
3088: if (flg) {
3089: PetscReal fact = 1.39;
3090: MatSetOption(B,MAT_USE_HASH_TABLE,PETSC_TRUE);
3091: PetscOptionsReal("-mat_use_hash_table","Use hash table factor","MatMPIBAIJSetHashTableFactor",fact,&fact,NULL);
3092: if (fact <= 1.0) fact = 1.39;
3093: MatMPIBAIJSetHashTableFactor(B,fact);
3094: PetscInfo1(B,"Hash table Factor used %5.2f\n",fact);
3095: }
3096: PetscOptionsEnd();
3097: return(0);
3098: }
3100: /*MC
3101: MATBAIJ - MATBAIJ = "baij" - A matrix type to be used for block sparse matrices.
3103: This matrix type is identical to MATSEQBAIJ when constructed with a single process communicator,
3104: and MATMPIBAIJ otherwise.
3106: Options Database Keys:
3107: . -mat_type baij - sets the matrix type to "baij" during a call to MatSetFromOptions()
3109: Level: beginner
3111: .seealso: MatCreateBAIJ(),MATSEQBAIJ,MATMPIBAIJ, MatMPIBAIJSetPreallocation(), MatMPIBAIJSetPreallocationCSR()
3112: M*/
3114: /*@C
3115: MatMPIBAIJSetPreallocation - Allocates memory for a sparse parallel matrix in block AIJ format
3116: (block compressed row). For good matrix assembly performance
3117: the user should preallocate the matrix storage by setting the parameters
3118: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
3119: performance can be increased by more than a factor of 50.
3121: Collective on Mat
3123: Input Parameters:
3124: + B - the matrix
3125: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
3126: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
3127: . d_nz - number of block nonzeros per block row in diagonal portion of local
3128: submatrix (same for all local rows)
3129: . d_nnz - array containing the number of block nonzeros in the various block rows
3130: of the in diagonal portion of the local (possibly different for each block
3131: row) or NULL. If you plan to factor the matrix you must leave room for the diagonal entry and
3132: set it even if it is zero.
3133: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
3134: submatrix (same for all local rows).
3135: - o_nnz - array containing the number of nonzeros in the various block rows of the
3136: off-diagonal portion of the local submatrix (possibly different for
3137: each block row) or NULL.
3139: If the *_nnz parameter is given then the *_nz parameter is ignored
3141: Options Database Keys:
3142: + -mat_block_size - size of the blocks to use
3143: - -mat_use_hash_table <fact>
3145: Notes:
3146: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
3147: than it must be used on all processors that share the object for that argument.
3149: Storage Information:
3150: For a square global matrix we define each processor's diagonal portion
3151: to be its local rows and the corresponding columns (a square submatrix);
3152: each processor's off-diagonal portion encompasses the remainder of the
3153: local matrix (a rectangular submatrix).
3155: The user can specify preallocated storage for the diagonal part of
3156: the local submatrix with either d_nz or d_nnz (not both). Set
3157: d_nz=PETSC_DEFAULT and d_nnz=NULL for PETSc to control dynamic
3158: memory allocation. Likewise, specify preallocated storage for the
3159: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
3161: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
3162: the figure below we depict these three local rows and all columns (0-11).
3164: .vb
3165: 0 1 2 3 4 5 6 7 8 9 10 11
3166: --------------------------
3167: row 3 |o o o d d d o o o o o o
3168: row 4 |o o o d d d o o o o o o
3169: row 5 |o o o d d d o o o o o o
3170: --------------------------
3171: .ve
3173: Thus, any entries in the d locations are stored in the d (diagonal)
3174: submatrix, and any entries in the o locations are stored in the
3175: o (off-diagonal) submatrix. Note that the d and the o submatrices are
3176: stored simply in the MATSEQBAIJ format for compressed row storage.
3178: Now d_nz should indicate the number of block nonzeros per row in the d matrix,
3179: and o_nz should indicate the number of block nonzeros per row in the o matrix.
3180: In general, for PDE problems in which most nonzeros are near the diagonal,
3181: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
3182: or you will get TERRIBLE performance; see the users' manual chapter on
3183: matrices.
3185: You can call MatGetInfo() to get information on how effective the preallocation was;
3186: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
3187: You can also run with the option -info and look for messages with the string
3188: malloc in them to see if additional memory allocation was needed.
3190: Level: intermediate
3192: .keywords: matrix, block, aij, compressed row, sparse, parallel
3194: .seealso: MatCreate(), MatCreateSeqBAIJ(), MatSetValues(), MatCreateBAIJ(), MatMPIBAIJSetPreallocationCSR(), PetscSplitOwnership()
3195: @*/
3196: PetscErrorCode MatMPIBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[])
3197: {
3204: PetscTryMethod(B,"MatMPIBAIJSetPreallocation_C",(Mat,PetscInt,PetscInt,const PetscInt[],PetscInt,const PetscInt[]),(B,bs,d_nz,d_nnz,o_nz,o_nnz));
3205: return(0);
3206: }
3208: /*@C
3209: MatCreateBAIJ - Creates a sparse parallel matrix in block AIJ format
3210: (block compressed row). For good matrix assembly performance
3211: the user should preallocate the matrix storage by setting the parameters
3212: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
3213: performance can be increased by more than a factor of 50.
3215: Collective on MPI_Comm
3217: Input Parameters:
3218: + comm - MPI communicator
3219: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
3220: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
3221: . m - number of local rows (or PETSC_DECIDE to have calculated if M is given)
3222: This value should be the same as the local size used in creating the
3223: y vector for the matrix-vector product y = Ax.
3224: . n - number of local columns (or PETSC_DECIDE to have calculated if N is given)
3225: This value should be the same as the local size used in creating the
3226: x vector for the matrix-vector product y = Ax.
3227: . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given)
3228: . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given)
3229: . d_nz - number of nonzero blocks per block row in diagonal portion of local
3230: submatrix (same for all local rows)
3231: . d_nnz - array containing the number of nonzero blocks in the various block rows
3232: of the in diagonal portion of the local (possibly different for each block
3233: row) or NULL. If you plan to factor the matrix you must leave room for the diagonal entry
3234: and set it even if it is zero.
3235: . o_nz - number of nonzero blocks per block row in the off-diagonal portion of local
3236: submatrix (same for all local rows).
3237: - o_nnz - array containing the number of nonzero blocks in the various block rows of the
3238: off-diagonal portion of the local submatrix (possibly different for
3239: each block row) or NULL.
3241: Output Parameter:
3242: . A - the matrix
3244: Options Database Keys:
3245: + -mat_block_size - size of the blocks to use
3246: - -mat_use_hash_table <fact>
3248: It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(),
3249: MatXXXXSetPreallocation() paradgm instead of this routine directly.
3250: [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation]
3252: Notes:
3253: If the *_nnz parameter is given then the *_nz parameter is ignored
3255: A nonzero block is any block that as 1 or more nonzeros in it
3257: The user MUST specify either the local or global matrix dimensions
3258: (possibly both).
3260: If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor
3261: than it must be used on all processors that share the object for that argument.
3263: Storage Information:
3264: For a square global matrix we define each processor's diagonal portion
3265: to be its local rows and the corresponding columns (a square submatrix);
3266: each processor's off-diagonal portion encompasses the remainder of the
3267: local matrix (a rectangular submatrix).
3269: The user can specify preallocated storage for the diagonal part of
3270: the local submatrix with either d_nz or d_nnz (not both). Set
3271: d_nz=PETSC_DEFAULT and d_nnz=NULL for PETSc to control dynamic
3272: memory allocation. Likewise, specify preallocated storage for the
3273: off-diagonal part of the local submatrix with o_nz or o_nnz (not both).
3275: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
3276: the figure below we depict these three local rows and all columns (0-11).
3278: .vb
3279: 0 1 2 3 4 5 6 7 8 9 10 11
3280: --------------------------
3281: row 3 |o o o d d d o o o o o o
3282: row 4 |o o o d d d o o o o o o
3283: row 5 |o o o d d d o o o o o o
3284: --------------------------
3285: .ve
3287: Thus, any entries in the d locations are stored in the d (diagonal)
3288: submatrix, and any entries in the o locations are stored in the
3289: o (off-diagonal) submatrix. Note that the d and the o submatrices are
3290: stored simply in the MATSEQBAIJ format for compressed row storage.
3292: Now d_nz should indicate the number of block nonzeros per row in the d matrix,
3293: and o_nz should indicate the number of block nonzeros per row in the o matrix.
3294: In general, for PDE problems in which most nonzeros are near the diagonal,
3295: one expects d_nz >> o_nz. For large problems you MUST preallocate memory
3296: or you will get TERRIBLE performance; see the users' manual chapter on
3297: matrices.
3299: Level: intermediate
3301: .keywords: matrix, block, aij, compressed row, sparse, parallel
3303: .seealso: MatCreate(), MatCreateSeqBAIJ(), MatSetValues(), MatCreateBAIJ(), MatMPIBAIJSetPreallocation(), MatMPIBAIJSetPreallocationCSR()
3304: @*/
3305: PetscErrorCode MatCreateBAIJ(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt M,PetscInt N,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[],Mat *A)
3306: {
3308: PetscMPIInt size;
3311: MatCreate(comm,A);
3312: MatSetSizes(*A,m,n,M,N);
3313: MPI_Comm_size(comm,&size);
3314: if (size > 1) {
3315: MatSetType(*A,MATMPIBAIJ);
3316: MatMPIBAIJSetPreallocation(*A,bs,d_nz,d_nnz,o_nz,o_nnz);
3317: } else {
3318: MatSetType(*A,MATSEQBAIJ);
3319: MatSeqBAIJSetPreallocation(*A,bs,d_nz,d_nnz);
3320: }
3321: return(0);
3322: }
3324: static PetscErrorCode MatDuplicate_MPIBAIJ(Mat matin,MatDuplicateOption cpvalues,Mat *newmat)
3325: {
3326: Mat mat;
3327: Mat_MPIBAIJ *a,*oldmat = (Mat_MPIBAIJ*)matin->data;
3329: PetscInt len=0;
3332: *newmat = 0;
3333: MatCreate(PetscObjectComm((PetscObject)matin),&mat);
3334: MatSetSizes(mat,matin->rmap->n,matin->cmap->n,matin->rmap->N,matin->cmap->N);
3335: MatSetType(mat,((PetscObject)matin)->type_name);
3336: PetscMemcpy(mat->ops,matin->ops,sizeof(struct _MatOps));
3338: mat->factortype = matin->factortype;
3339: mat->preallocated = PETSC_TRUE;
3340: mat->assembled = PETSC_TRUE;
3341: mat->insertmode = NOT_SET_VALUES;
3343: a = (Mat_MPIBAIJ*)mat->data;
3344: mat->rmap->bs = matin->rmap->bs;
3345: a->bs2 = oldmat->bs2;
3346: a->mbs = oldmat->mbs;
3347: a->nbs = oldmat->nbs;
3348: a->Mbs = oldmat->Mbs;
3349: a->Nbs = oldmat->Nbs;
3351: PetscLayoutReference(matin->rmap,&mat->rmap);
3352: PetscLayoutReference(matin->cmap,&mat->cmap);
3354: a->size = oldmat->size;
3355: a->rank = oldmat->rank;
3356: a->donotstash = oldmat->donotstash;
3357: a->roworiented = oldmat->roworiented;
3358: a->rowindices = 0;
3359: a->rowvalues = 0;
3360: a->getrowactive = PETSC_FALSE;
3361: a->barray = 0;
3362: a->rstartbs = oldmat->rstartbs;
3363: a->rendbs = oldmat->rendbs;
3364: a->cstartbs = oldmat->cstartbs;
3365: a->cendbs = oldmat->cendbs;
3367: /* hash table stuff */
3368: a->ht = 0;
3369: a->hd = 0;
3370: a->ht_size = 0;
3371: a->ht_flag = oldmat->ht_flag;
3372: a->ht_fact = oldmat->ht_fact;
3373: a->ht_total_ct = 0;
3374: a->ht_insert_ct = 0;
3376: PetscMemcpy(a->rangebs,oldmat->rangebs,(a->size+1)*sizeof(PetscInt));
3377: if (oldmat->colmap) {
3378: #if defined(PETSC_USE_CTABLE)
3379: PetscTableCreateCopy(oldmat->colmap,&a->colmap);
3380: #else
3381: PetscMalloc1(a->Nbs,&a->colmap);
3382: PetscLogObjectMemory((PetscObject)mat,(a->Nbs)*sizeof(PetscInt));
3383: PetscMemcpy(a->colmap,oldmat->colmap,(a->Nbs)*sizeof(PetscInt));
3384: #endif
3385: } else a->colmap = 0;
3387: if (oldmat->garray && (len = ((Mat_SeqBAIJ*)(oldmat->B->data))->nbs)) {
3388: PetscMalloc1(len,&a->garray);
3389: PetscLogObjectMemory((PetscObject)mat,len*sizeof(PetscInt));
3390: PetscMemcpy(a->garray,oldmat->garray,len*sizeof(PetscInt));
3391: } else a->garray = 0;
3393: MatStashCreate_Private(PetscObjectComm((PetscObject)matin),matin->rmap->bs,&mat->bstash);
3394: VecDuplicate(oldmat->lvec,&a->lvec);
3395: PetscLogObjectParent((PetscObject)mat,(PetscObject)a->lvec);
3396: VecScatterCopy(oldmat->Mvctx,&a->Mvctx);
3397: PetscLogObjectParent((PetscObject)mat,(PetscObject)a->Mvctx);
3399: MatDuplicate(oldmat->A,cpvalues,&a->A);
3400: PetscLogObjectParent((PetscObject)mat,(PetscObject)a->A);
3401: MatDuplicate(oldmat->B,cpvalues,&a->B);
3402: PetscLogObjectParent((PetscObject)mat,(PetscObject)a->B);
3403: PetscFunctionListDuplicate(((PetscObject)matin)->qlist,&((PetscObject)mat)->qlist);
3404: *newmat = mat;
3405: return(0);
3406: }
3408: PetscErrorCode MatLoad_MPIBAIJ(Mat newmat,PetscViewer viewer)
3409: {
3411: int fd;
3412: PetscInt i,nz,j,rstart,rend;
3413: PetscScalar *vals,*buf;
3414: MPI_Comm comm;
3415: MPI_Status status;
3416: PetscMPIInt rank,size,maxnz;
3417: PetscInt header[4],*rowlengths = 0,M,N,m,*rowners,*cols;
3418: PetscInt *locrowlens = NULL,*procsnz = NULL,*browners = NULL;
3419: PetscInt jj,*mycols,*ibuf,bs = newmat->rmap->bs,Mbs,mbs,extra_rows,mmax;
3420: PetscMPIInt tag = ((PetscObject)viewer)->tag;
3421: PetscInt *dlens = NULL,*odlens = NULL,*mask = NULL,*masked1 = NULL,*masked2 = NULL,rowcount,odcount;
3422: PetscInt dcount,kmax,k,nzcount,tmp,mend;
3425: /* force binary viewer to load .info file if it has not yet done so */
3426: PetscViewerSetUp(viewer);
3427: PetscObjectGetComm((PetscObject)viewer,&comm);
3428: PetscOptionsBegin(comm,NULL,"Options for loading MPIBAIJ matrix 2","Mat");
3429: PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,NULL);
3430: PetscOptionsEnd();
3431: if (bs < 0) bs = 1;
3433: MPI_Comm_size(comm,&size);
3434: MPI_Comm_rank(comm,&rank);
3435: PetscViewerBinaryGetDescriptor(viewer,&fd);
3436: if (!rank) {
3437: PetscBinaryRead(fd,(char*)header,4,PETSC_INT);
3438: if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not matrix object");
3439: if (header[3] < 0) SETERRQ(PetscObjectComm((PetscObject)newmat),PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format on disk, cannot load as MPIAIJ");
3440: }
3441: MPI_Bcast(header+1,3,MPIU_INT,0,comm);
3442: M = header[1]; N = header[2];
3444: /* If global sizes are set, check if they are consistent with that given in the file */
3445: if (newmat->rmap->N >= 0 && newmat->rmap->N != M) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Inconsistent # of rows:Matrix in file has (%D) and input matrix has (%D)",newmat->rmap->N,M);
3446: if (newmat->cmap->N >= 0 && newmat->cmap->N != N) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Inconsistent # of cols:Matrix in file has (%D) and input matrix has (%D)",newmat->cmap->N,N);
3448: if (M != N) SETERRQ(PetscObjectComm((PetscObject)viewer),PETSC_ERR_SUP,"Can only do square matrices");
3450: /*
3451: This code adds extra rows to make sure the number of rows is
3452: divisible by the blocksize
3453: */
3454: Mbs = M/bs;
3455: extra_rows = bs - M + bs*Mbs;
3456: if (extra_rows == bs) extra_rows = 0;
3457: else Mbs++;
3458: if (extra_rows && !rank) {
3459: PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");
3460: }
3462: /* determine ownership of all rows */
3463: if (newmat->rmap->n < 0) { /* PETSC_DECIDE */
3464: mbs = Mbs/size + ((Mbs % size) > rank);
3465: m = mbs*bs;
3466: } else { /* User set */
3467: m = newmat->rmap->n;
3468: mbs = m/bs;
3469: }
3470: PetscMalloc2(size+1,&rowners,size+1,&browners);
3471: MPI_Allgather(&mbs,1,MPIU_INT,rowners+1,1,MPIU_INT,comm);
3473: /* process 0 needs enough room for process with most rows */
3474: if (!rank) {
3475: mmax = rowners[1];
3476: for (i=2; i<=size; i++) {
3477: mmax = PetscMax(mmax,rowners[i]);
3478: }
3479: mmax*=bs;
3480: } else mmax = -1; /* unused, but compiler warns anyway */
3482: rowners[0] = 0;
3483: for (i=2; i<=size; i++) rowners[i] += rowners[i-1];
3484: for (i=0; i<=size; i++) browners[i] = rowners[i]*bs;
3485: rstart = rowners[rank];
3486: rend = rowners[rank+1];
3488: /* distribute row lengths to all processors */
3489: PetscMalloc1(m,&locrowlens);
3490: if (!rank) {
3491: mend = m;
3492: if (size == 1) mend = mend - extra_rows;
3493: PetscBinaryRead(fd,locrowlens,mend,PETSC_INT);
3494: for (j=mend; j<m; j++) locrowlens[j] = 1;
3495: PetscMalloc1(mmax,&rowlengths);
3496: PetscCalloc1(size,&procsnz);
3497: for (j=0; j<m; j++) {
3498: procsnz[0] += locrowlens[j];
3499: }
3500: for (i=1; i<size; i++) {
3501: mend = browners[i+1] - browners[i];
3502: if (i == size-1) mend = mend - extra_rows;
3503: PetscBinaryRead(fd,rowlengths,mend,PETSC_INT);
3504: for (j=mend; j<browners[i+1] - browners[i]; j++) rowlengths[j] = 1;
3505: /* calculate the number of nonzeros on each processor */
3506: for (j=0; j<browners[i+1]-browners[i]; j++) {
3507: procsnz[i] += rowlengths[j];
3508: }
3509: MPI_Send(rowlengths,browners[i+1]-browners[i],MPIU_INT,i,tag,comm);
3510: }
3511: PetscFree(rowlengths);
3512: } else {
3513: MPI_Recv(locrowlens,m,MPIU_INT,0,tag,comm,&status);
3514: }
3516: if (!rank) {
3517: /* determine max buffer needed and allocate it */
3518: maxnz = procsnz[0];
3519: for (i=1; i<size; i++) {
3520: maxnz = PetscMax(maxnz,procsnz[i]);
3521: }
3522: PetscMalloc1(maxnz,&cols);
3524: /* read in my part of the matrix column indices */
3525: nz = procsnz[0];
3526: PetscMalloc1(nz+1,&ibuf);
3527: mycols = ibuf;
3528: if (size == 1) nz -= extra_rows;
3529: PetscBinaryRead(fd,mycols,nz,PETSC_INT);
3530: if (size == 1) {
3531: for (i=0; i< extra_rows; i++) mycols[nz+i] = M+i;
3532: }
3534: /* read in every ones (except the last) and ship off */
3535: for (i=1; i<size-1; i++) {
3536: nz = procsnz[i];
3537: PetscBinaryRead(fd,cols,nz,PETSC_INT);
3538: MPI_Send(cols,nz,MPIU_INT,i,tag,comm);
3539: }
3540: /* read in the stuff for the last proc */
3541: if (size != 1) {
3542: nz = procsnz[size-1] - extra_rows; /* the extra rows are not on the disk */
3543: PetscBinaryRead(fd,cols,nz,PETSC_INT);
3544: for (i=0; i<extra_rows; i++) cols[nz+i] = M+i;
3545: MPI_Send(cols,nz+extra_rows,MPIU_INT,size-1,tag,comm);
3546: }
3547: PetscFree(cols);
3548: } else {
3549: /* determine buffer space needed for message */
3550: nz = 0;
3551: for (i=0; i<m; i++) {
3552: nz += locrowlens[i];
3553: }
3554: PetscMalloc1(nz+1,&ibuf);
3555: mycols = ibuf;
3556: /* receive message of column indices*/
3557: MPI_Recv(mycols,nz,MPIU_INT,0,tag,comm,&status);
3558: MPI_Get_count(&status,MPIU_INT,&maxnz);
3559: if (maxnz != nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file");
3560: }
3562: /* loop over local rows, determining number of off diagonal entries */
3563: PetscMalloc2(rend-rstart,&dlens,rend-rstart,&odlens);
3564: PetscCalloc3(Mbs,&mask,Mbs,&masked1,Mbs,&masked2);
3565: rowcount = 0; nzcount = 0;
3566: for (i=0; i<mbs; i++) {
3567: dcount = 0;
3568: odcount = 0;
3569: for (j=0; j<bs; j++) {
3570: kmax = locrowlens[rowcount];
3571: for (k=0; k<kmax; k++) {
3572: tmp = mycols[nzcount++]/bs;
3573: if (!mask[tmp]) {
3574: mask[tmp] = 1;
3575: if (tmp < rstart || tmp >= rend) masked2[odcount++] = tmp;
3576: else masked1[dcount++] = tmp;
3577: }
3578: }
3579: rowcount++;
3580: }
3582: dlens[i] = dcount;
3583: odlens[i] = odcount;
3585: /* zero out the mask elements we set */
3586: for (j=0; j<dcount; j++) mask[masked1[j]] = 0;
3587: for (j=0; j<odcount; j++) mask[masked2[j]] = 0;
3588: }
3590: MatSetSizes(newmat,m,m,M+extra_rows,N+extra_rows);
3591: MatMPIBAIJSetPreallocation(newmat,bs,0,dlens,0,odlens);
3593: if (!rank) {
3594: PetscMalloc1(maxnz+1,&buf);
3595: /* read in my part of the matrix numerical values */
3596: nz = procsnz[0];
3597: vals = buf;
3598: mycols = ibuf;
3599: if (size == 1) nz -= extra_rows;
3600: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
3601: if (size == 1) {
3602: for (i=0; i< extra_rows; i++) vals[nz+i] = 1.0;
3603: }
3605: /* insert into matrix */
3606: jj = rstart*bs;
3607: for (i=0; i<m; i++) {
3608: MatSetValues_MPIBAIJ(newmat,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
3609: mycols += locrowlens[i];
3610: vals += locrowlens[i];
3611: jj++;
3612: }
3613: /* read in other processors (except the last one) and ship out */
3614: for (i=1; i<size-1; i++) {
3615: nz = procsnz[i];
3616: vals = buf;
3617: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
3618: MPIULong_Send(vals,nz,MPIU_SCALAR,i,((PetscObject)newmat)->tag,comm);
3619: }
3620: /* the last proc */
3621: if (size != 1) {
3622: nz = procsnz[i] - extra_rows;
3623: vals = buf;
3624: PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);
3625: for (i=0; i<extra_rows; i++) vals[nz+i] = 1.0;
3626: MPIULong_Send(vals,nz+extra_rows,MPIU_SCALAR,size-1,((PetscObject)newmat)->tag,comm);
3627: }
3628: PetscFree(procsnz);
3629: } else {
3630: /* receive numeric values */
3631: PetscMalloc1(nz+1,&buf);
3633: /* receive message of values*/
3634: vals = buf;
3635: mycols = ibuf;
3636: MPIULong_Recv(vals,nz,MPIU_SCALAR,0,((PetscObject)newmat)->tag,comm);
3638: /* insert into matrix */
3639: jj = rstart*bs;
3640: for (i=0; i<m; i++) {
3641: MatSetValues_MPIBAIJ(newmat,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);
3642: mycols += locrowlens[i];
3643: vals += locrowlens[i];
3644: jj++;
3645: }
3646: }
3647: PetscFree(locrowlens);
3648: PetscFree(buf);
3649: PetscFree(ibuf);
3650: PetscFree2(rowners,browners);
3651: PetscFree2(dlens,odlens);
3652: PetscFree3(mask,masked1,masked2);
3653: MatAssemblyBegin(newmat,MAT_FINAL_ASSEMBLY);
3654: MatAssemblyEnd(newmat,MAT_FINAL_ASSEMBLY);
3655: return(0);
3656: }
3658: /*@
3659: MatMPIBAIJSetHashTableFactor - Sets the factor required to compute the size of the HashTable.
3661: Input Parameters:
3662: . mat - the matrix
3663: . fact - factor
3665: Not Collective, each process can use a different factor
3667: Level: advanced
3669: Notes:
3670: This can also be set by the command line option: -mat_use_hash_table <fact>
3672: .keywords: matrix, hashtable, factor, HT
3674: .seealso: MatSetOption()
3675: @*/
3676: PetscErrorCode MatMPIBAIJSetHashTableFactor(Mat mat,PetscReal fact)
3677: {
3681: PetscTryMethod(mat,"MatSetHashTableFactor_C",(Mat,PetscReal),(mat,fact));
3682: return(0);
3683: }
3685: PetscErrorCode MatSetHashTableFactor_MPIBAIJ(Mat mat,PetscReal fact)
3686: {
3687: Mat_MPIBAIJ *baij;
3690: baij = (Mat_MPIBAIJ*)mat->data;
3691: baij->ht_fact = fact;
3692: return(0);
3693: }
3695: PetscErrorCode MatMPIBAIJGetSeqBAIJ(Mat A,Mat *Ad,Mat *Ao,const PetscInt *colmap[])
3696: {
3697: Mat_MPIBAIJ *a = (Mat_MPIBAIJ*)A->data;
3698: PetscBool flg;
3702: PetscObjectTypeCompare((PetscObject)A,MATMPIBAIJ,&flg);
3703: if (!flg) SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"This function requires a MATMPIBAIJ matrix as input");
3704: if (Ad) *Ad = a->A;
3705: if (Ao) *Ao = a->B;
3706: if (colmap) *colmap = a->garray;
3707: return(0);
3708: }
3710: /*
3711: Special version for direct calls from Fortran (to eliminate two function call overheads
3712: */
3713: #if defined(PETSC_HAVE_FORTRAN_CAPS)
3714: #define matmpibaijsetvaluesblocked_ MATMPIBAIJSETVALUESBLOCKED
3715: #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE)
3716: #define matmpibaijsetvaluesblocked_ matmpibaijsetvaluesblocked
3717: #endif
3719: /*@C
3720: MatMPIBAIJSetValuesBlocked - Direct Fortran call to replace call to MatSetValuesBlocked()
3722: Collective on Mat
3724: Input Parameters:
3725: + mat - the matrix
3726: . min - number of input rows
3727: . im - input rows
3728: . nin - number of input columns
3729: . in - input columns
3730: . v - numerical values input
3731: - addvin - INSERT_VALUES or ADD_VALUES
3733: Notes: This has a complete copy of MatSetValuesBlocked_MPIBAIJ() which is terrible code un-reuse.
3735: Level: advanced
3737: .seealso: MatSetValuesBlocked()
3738: @*/
3739: PetscErrorCode matmpibaijsetvaluesblocked_(Mat *matin,PetscInt *min,const PetscInt im[],PetscInt *nin,const PetscInt in[],const MatScalar v[],InsertMode *addvin)
3740: {
3741: /* convert input arguments to C version */
3742: Mat mat = *matin;
3743: PetscInt m = *min, n = *nin;
3744: InsertMode addv = *addvin;
3746: Mat_MPIBAIJ *baij = (Mat_MPIBAIJ*)mat->data;
3747: const MatScalar *value;
3748: MatScalar *barray = baij->barray;
3749: PetscBool roworiented = baij->roworiented;
3750: PetscErrorCode ierr;
3751: PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs;
3752: PetscInt rend=baij->rendbs,cstart=baij->cstartbs,stepval;
3753: PetscInt cend=baij->cendbs,bs=mat->rmap->bs,bs2=baij->bs2;
3756: /* tasks normally handled by MatSetValuesBlocked() */
3757: if (mat->insertmode == NOT_SET_VALUES) mat->insertmode = addv;
3758: #if defined(PETSC_USE_DEBUG)
3759: else if (mat->insertmode != addv) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Cannot mix add values and insert values");
3760: if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
3761: #endif
3762: if (mat->assembled) {
3763: mat->was_assembled = PETSC_TRUE;
3764: mat->assembled = PETSC_FALSE;
3765: }
3766: PetscLogEventBegin(MAT_SetValues,mat,0,0,0);
3769: if (!barray) {
3770: PetscMalloc1(bs2,&barray);
3771: baij->barray = barray;
3772: }
3774: if (roworiented) stepval = (n-1)*bs;
3775: else stepval = (m-1)*bs;
3777: for (i=0; i<m; i++) {
3778: if (im[i] < 0) continue;
3779: #if defined(PETSC_USE_DEBUG)
3780: if (im[i] >= baij->Mbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large, row %D max %D",im[i],baij->Mbs-1);
3781: #endif
3782: if (im[i] >= rstart && im[i] < rend) {
3783: row = im[i] - rstart;
3784: for (j=0; j<n; j++) {
3785: /* If NumCol = 1 then a copy is not required */
3786: if ((roworiented) && (n == 1)) {
3787: barray = (MatScalar*)v + i*bs2;
3788: } else if ((!roworiented) && (m == 1)) {
3789: barray = (MatScalar*)v + j*bs2;
3790: } else { /* Here a copy is required */
3791: if (roworiented) {
3792: value = v + i*(stepval+bs)*bs + j*bs;
3793: } else {
3794: value = v + j*(stepval+bs)*bs + i*bs;
3795: }
3796: for (ii=0; ii<bs; ii++,value+=stepval) {
3797: for (jj=0; jj<bs; jj++) {
3798: *barray++ = *value++;
3799: }
3800: }
3801: barray -=bs2;
3802: }
3804: if (in[j] >= cstart && in[j] < cend) {
3805: col = in[j] - cstart;
3806: MatSetValuesBlocked_SeqBAIJ_Inlined(baij->A,row,col,barray,addv,im[i],in[j]);
3807: } else if (in[j] < 0) continue;
3808: #if defined(PETSC_USE_DEBUG)
3809: else if (in[j] >= baij->Nbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large, col %D max %D",in[j],baij->Nbs-1);
3810: #endif
3811: else {
3812: if (mat->was_assembled) {
3813: if (!baij->colmap) {
3814: MatCreateColmap_MPIBAIJ_Private(mat);
3815: }
3817: #if defined(PETSC_USE_DEBUG)
3818: #if defined(PETSC_USE_CTABLE)
3819: { PetscInt data;
3820: PetscTableFind(baij->colmap,in[j]+1,&data);
3821: if ((data - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap");
3822: }
3823: #else
3824: if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap");
3825: #endif
3826: #endif
3827: #if defined(PETSC_USE_CTABLE)
3828: PetscTableFind(baij->colmap,in[j]+1,&col);
3829: col = (col - 1)/bs;
3830: #else
3831: col = (baij->colmap[in[j]] - 1)/bs;
3832: #endif
3833: if (col < 0 && !((Mat_SeqBAIJ*)(baij->A->data))->nonew) {
3834: MatDisAssemble_MPIBAIJ(mat);
3835: col = in[j];
3836: }
3837: } else col = in[j];
3838: MatSetValuesBlocked_SeqBAIJ_Inlined(baij->B,row,col,barray,addv,im[i],in[j]);
3839: }
3840: }
3841: } else {
3842: if (!baij->donotstash) {
3843: if (roworiented) {
3844: MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
3845: } else {
3846: MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);
3847: }
3848: }
3849: }
3850: }
3852: /* task normally handled by MatSetValuesBlocked() */
3853: PetscLogEventEnd(MAT_SetValues,mat,0,0,0);
3854: return(0);
3855: }
3857: /*@
3858: MatCreateMPIBAIJWithArrays - creates a MPI BAIJ matrix using arrays that contain in standard
3859: CSR format the local rows.
3861: Collective on MPI_Comm
3863: Input Parameters:
3864: + comm - MPI communicator
3865: . bs - the block size, only a block size of 1 is supported
3866: . m - number of local rows (Cannot be PETSC_DECIDE)
3867: . n - This value should be the same as the local size used in creating the
3868: x vector for the matrix-vector product y = Ax. (or PETSC_DECIDE to have
3869: calculated if N is given) For square matrices n is almost always m.
3870: . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given)
3871: . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given)
3872: . i - row indices
3873: . j - column indices
3874: - a - matrix values
3876: Output Parameter:
3877: . mat - the matrix
3879: Level: intermediate
3881: Notes:
3882: The i, j, and a arrays ARE copied by this routine into the internal format used by PETSc;
3883: thus you CANNOT change the matrix entries by changing the values of a[] after you have
3884: called this routine. Use MatCreateMPIAIJWithSplitArrays() to avoid needing to copy the arrays.
3886: The order of the entries in values is the same as the block compressed sparse row storage format; that is, it is
3887: the same as a three dimensional array in Fortran values(bs,bs,nnz) that contains the first column of the first
3888: block, followed by the second column of the first block etc etc. That is, the blocks are contiguous in memory
3889: with column-major ordering within blocks.
3891: The i and j indices are 0 based, and i indices are indices corresponding to the local j array.
3893: .keywords: matrix, aij, compressed row, sparse, parallel
3895: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatMPIAIJSetPreallocation(), MatMPIAIJSetPreallocationCSR(),
3896: MPIAIJ, MatCreateAIJ(), MatCreateMPIAIJWithSplitArrays()
3897: @*/
3898: PetscErrorCode MatCreateMPIBAIJWithArrays(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt M,PetscInt N,const PetscInt i[],const PetscInt j[],const PetscScalar a[],Mat *mat)
3899: {
3903: if (i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0");
3904: if (m < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"local number of rows (m) cannot be PETSC_DECIDE, or negative");
3905: MatCreate(comm,mat);
3906: MatSetSizes(*mat,m,n,M,N);
3907: MatSetType(*mat,MATMPIBAIJ);
3908: MatSetBlockSize(*mat,bs);
3909: MatSetUp(*mat);
3910: MatSetOption(*mat,MAT_ROW_ORIENTED,PETSC_FALSE);
3911: MatMPIBAIJSetPreallocationCSR(*mat,bs,i,j,a);
3912: MatSetOption(*mat,MAT_ROW_ORIENTED,PETSC_TRUE);
3913: return(0);
3914: }
3916: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_MPIBAIJ(MPI_Comm comm,Mat inmat,PetscInt n,MatReuse scall,Mat *outmat)
3917: {
3919: PetscInt m,N,i,rstart,nnz,Ii,bs,cbs;
3920: PetscInt *indx;
3921: PetscScalar *values;
3924: MatGetSize(inmat,&m,&N);
3925: if (scall == MAT_INITIAL_MATRIX) { /* symbolic phase */
3926: Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)inmat->data;
3927: PetscInt *dnz,*onz,mbs,Nbs,nbs;
3928: PetscInt *bindx,rmax=a->rmax,j;
3929: PetscMPIInt rank,size;
3931: MatGetBlockSizes(inmat,&bs,&cbs);
3932: mbs = m/bs; Nbs = N/cbs;
3933: if (n == PETSC_DECIDE) {
3934: nbs = n;
3935: PetscSplitOwnership(comm,&nbs,&Nbs);
3936: n = nbs*cbs;
3937: } else {
3938: nbs = n/cbs;
3939: }
3941: PetscMalloc1(rmax,&bindx);
3942: MatPreallocateInitialize(comm,mbs,nbs,dnz,onz); /* inline function, output __end and __rstart are used below */
3944: MPI_Comm_rank(comm,&rank);
3945: MPI_Comm_rank(comm,&size);
3946: if (rank == size-1) {
3947: /* Check sum(nbs) = Nbs */
3948: if (__end != Nbs) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Sum of local block columns %D != global block columns %D",__end,Nbs);
3949: }
3951: rstart = __rstart; /* block rstart of *outmat; see inline function MatPreallocateInitialize */
3952: for (i=0; i<mbs; i++) {
3953: MatGetRow_SeqBAIJ(inmat,i*bs,&nnz,&indx,NULL); /* non-blocked nnz and indx */
3954: nnz = nnz/bs;
3955: for (j=0; j<nnz; j++) bindx[j] = indx[j*bs]/bs;
3956: MatPreallocateSet(i+rstart,nnz,bindx,dnz,onz);
3957: MatRestoreRow_SeqBAIJ(inmat,i*bs,&nnz,&indx,NULL);
3958: }
3959: PetscFree(bindx);
3961: MatCreate(comm,outmat);
3962: MatSetSizes(*outmat,m,n,PETSC_DETERMINE,PETSC_DETERMINE);
3963: MatSetBlockSizes(*outmat,bs,cbs);
3964: MatSetType(*outmat,MATBAIJ);
3965: MatSeqBAIJSetPreallocation(*outmat,bs,0,dnz);
3966: MatMPIBAIJSetPreallocation(*outmat,bs,0,dnz,0,onz);
3967: MatPreallocateFinalize(dnz,onz);
3968: }
3970: /* numeric phase */
3971: MatGetBlockSizes(inmat,&bs,&cbs);
3972: MatGetOwnershipRange(*outmat,&rstart,NULL);
3974: for (i=0; i<m; i++) {
3975: MatGetRow_SeqBAIJ(inmat,i,&nnz,&indx,&values);
3976: Ii = i + rstart;
3977: MatSetValues(*outmat,1,&Ii,nnz,indx,values,INSERT_VALUES);
3978: MatRestoreRow_SeqBAIJ(inmat,i,&nnz,&indx,&values);
3979: }
3980: MatAssemblyBegin(*outmat,MAT_FINAL_ASSEMBLY);
3981: MatAssemblyEnd(*outmat,MAT_FINAL_ASSEMBLY);
3982: return(0);
3983: }