1 2 #include <../src/mat/impls/baij/mpi/mpibaij.h> /*I "petscmat.h" I*/ 3 #include <../src/mat/impls/sbaij/mpi/mpisbaij.h> 4 #include <../src/mat/impls/sbaij/seq/sbaij.h> 5 #include <petscblaslapack.h> 6 7 extern PetscErrorCode MatSetUpMultiply_MPISBAIJ(Mat); 8 extern PetscErrorCode MatSetUpMultiply_MPISBAIJ_2comm(Mat); 9 extern PetscErrorCode MatDisAssemble_MPISBAIJ(Mat); 10 extern PetscErrorCode MatIncreaseOverlap_MPISBAIJ(Mat,PetscInt,IS[],PetscInt); 11 extern PetscErrorCode MatGetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []); 12 extern PetscErrorCode MatGetValues_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []); 13 extern PetscErrorCode MatSetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt [],PetscInt,const PetscInt [],const PetscScalar [],InsertMode); 14 extern PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 15 extern PetscErrorCode MatSetValuesBlocked_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 16 extern PetscErrorCode MatGetRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**); 17 extern PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**); 18 extern PetscErrorCode MatZeroRows_SeqSBAIJ(Mat,IS,PetscScalar*,Vec,Vec); 19 extern PetscErrorCode MatZeroRows_SeqBAIJ(Mat,IS,PetscScalar*,Vec,Vec); 20 extern PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat,Vec,PetscInt[]); 21 extern PetscErrorCode MatSOR_MPISBAIJ(Mat,Vec,PetscReal,MatSORType,PetscReal,PetscInt,PetscInt,Vec); 22 23 #undef __FUNCT__ 24 #define __FUNCT__ "MatStoreValues_MPISBAIJ" 25 PetscErrorCode MatStoreValues_MPISBAIJ(Mat mat) 26 { 27 Mat_MPISBAIJ *aij = (Mat_MPISBAIJ*)mat->data; 28 PetscErrorCode ierr; 29 30 PetscFunctionBegin; 31 ierr = MatStoreValues(aij->A);CHKERRQ(ierr); 32 ierr = MatStoreValues(aij->B);CHKERRQ(ierr); 33 PetscFunctionReturn(0); 34 } 35 36 #undef __FUNCT__ 37 #define __FUNCT__ "MatRetrieveValues_MPISBAIJ" 38 PetscErrorCode MatRetrieveValues_MPISBAIJ(Mat mat) 39 { 40 Mat_MPISBAIJ *aij = (Mat_MPISBAIJ*)mat->data; 41 PetscErrorCode ierr; 42 43 PetscFunctionBegin; 44 ierr = MatRetrieveValues(aij->A);CHKERRQ(ierr); 45 ierr = MatRetrieveValues(aij->B);CHKERRQ(ierr); 46 PetscFunctionReturn(0); 47 } 48 49 #define MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv) \ 50 { \ 51 \ 52 brow = row/bs; \ 53 rp = aj + ai[brow]; ap = aa + bs2*ai[brow]; \ 54 rmax = aimax[brow]; nrow = ailen[brow]; \ 55 bcol = col/bs; \ 56 ridx = row % bs; cidx = col % bs; \ 57 low = 0; high = nrow; \ 58 while (high-low > 3) { \ 59 t = (low+high)/2; \ 60 if (rp[t] > bcol) high = t; \ 61 else low = t; \ 62 } \ 63 for (_i=low; _i<high; _i++) { \ 64 if (rp[_i] > bcol) break; \ 65 if (rp[_i] == bcol) { \ 66 bap = ap + bs2*_i + bs*cidx + ridx; \ 67 if (addv == ADD_VALUES) *bap += value; \ 68 else *bap = value; \ 69 goto a_noinsert; \ 70 } \ 71 } \ 72 if (a->nonew == 1) goto a_noinsert; \ 73 if (a->nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \ 74 MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,aimax,a->nonew,MatScalar); \ 75 N = nrow++ - 1; \ 76 /* shift up all the later entries in this row */ \ 77 for (ii=N; ii>=_i; ii--) { \ 78 rp[ii+1] = rp[ii]; \ 79 ierr = PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));CHKERRQ(ierr); \ 80 } \ 81 if (N>=_i) { ierr = PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));CHKERRQ(ierr); } \ 82 rp[_i] = bcol; \ 83 ap[bs2*_i + bs*cidx + ridx] = value; \ 84 A->nonzerostate++;\ 85 a_noinsert:; \ 86 ailen[brow] = nrow; \ 87 } 88 89 #define MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv) \ 90 { \ 91 brow = row/bs; \ 92 rp = bj + bi[brow]; ap = ba + bs2*bi[brow]; \ 93 rmax = bimax[brow]; nrow = bilen[brow]; \ 94 bcol = col/bs; \ 95 ridx = row % bs; cidx = col % bs; \ 96 low = 0; high = nrow; \ 97 while (high-low > 3) { \ 98 t = (low+high)/2; \ 99 if (rp[t] > bcol) high = t; \ 100 else low = t; \ 101 } \ 102 for (_i=low; _i<high; _i++) { \ 103 if (rp[_i] > bcol) break; \ 104 if (rp[_i] == bcol) { \ 105 bap = ap + bs2*_i + bs*cidx + ridx; \ 106 if (addv == ADD_VALUES) *bap += value; \ 107 else *bap = value; \ 108 goto b_noinsert; \ 109 } \ 110 } \ 111 if (b->nonew == 1) goto b_noinsert; \ 112 if (b->nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \ 113 MatSeqXAIJReallocateAIJ(B,b->mbs,bs2,nrow,brow,bcol,rmax,ba,bi,bj,rp,ap,bimax,b->nonew,MatScalar); \ 114 N = nrow++ - 1; \ 115 /* shift up all the later entries in this row */ \ 116 for (ii=N; ii>=_i; ii--) { \ 117 rp[ii+1] = rp[ii]; \ 118 ierr = PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));CHKERRQ(ierr); \ 119 } \ 120 if (N>=_i) { ierr = PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));CHKERRQ(ierr);} \ 121 rp[_i] = bcol; \ 122 ap[bs2*_i + bs*cidx + ridx] = value; \ 123 B->nonzerostate++;\ 124 b_noinsert:; \ 125 bilen[brow] = nrow; \ 126 } 127 128 /* Only add/insert a(i,j) with i<=j (blocks). 129 Any a(i,j) with i>j input by user is ingored. 130 */ 131 #undef __FUNCT__ 132 #define __FUNCT__ "MatSetValues_MPISBAIJ" 133 PetscErrorCode MatSetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv) 134 { 135 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 136 MatScalar value; 137 PetscBool roworiented = baij->roworiented; 138 PetscErrorCode ierr; 139 PetscInt i,j,row,col; 140 PetscInt rstart_orig=mat->rmap->rstart; 141 PetscInt rend_orig =mat->rmap->rend,cstart_orig=mat->cmap->rstart; 142 PetscInt cend_orig =mat->cmap->rend,bs=mat->rmap->bs; 143 144 /* Some Variables required in the macro */ 145 Mat A = baij->A; 146 Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)(A)->data; 147 PetscInt *aimax=a->imax,*ai=a->i,*ailen=a->ilen,*aj=a->j; 148 MatScalar *aa =a->a; 149 150 Mat B = baij->B; 151 Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(B)->data; 152 PetscInt *bimax=b->imax,*bi=b->i,*bilen=b->ilen,*bj=b->j; 153 MatScalar *ba =b->a; 154 155 PetscInt *rp,ii,nrow,_i,rmax,N,brow,bcol; 156 PetscInt low,high,t,ridx,cidx,bs2=a->bs2; 157 MatScalar *ap,*bap; 158 159 /* for stash */ 160 PetscInt n_loc, *in_loc = NULL; 161 MatScalar *v_loc = NULL; 162 163 PetscFunctionBegin; 164 if (!baij->donotstash) { 165 if (n > baij->n_loc) { 166 ierr = PetscFree(baij->in_loc);CHKERRQ(ierr); 167 ierr = PetscFree(baij->v_loc);CHKERRQ(ierr); 168 ierr = PetscMalloc1(n,&baij->in_loc);CHKERRQ(ierr); 169 ierr = PetscMalloc1(n,&baij->v_loc);CHKERRQ(ierr); 170 171 baij->n_loc = n; 172 } 173 in_loc = baij->in_loc; 174 v_loc = baij->v_loc; 175 } 176 177 for (i=0; i<m; i++) { 178 if (im[i] < 0) continue; 179 #if defined(PETSC_USE_DEBUG) 180 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); 181 #endif 182 if (im[i] >= rstart_orig && im[i] < rend_orig) { /* this processor entry */ 183 row = im[i] - rstart_orig; /* local row index */ 184 for (j=0; j<n; j++) { 185 if (im[i]/bs > in[j]/bs) { 186 if (a->ignore_ltriangular) { 187 continue; /* ignore lower triangular blocks */ 188 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)"); 189 } 190 if (in[j] >= cstart_orig && in[j] < cend_orig) { /* diag entry (A) */ 191 col = in[j] - cstart_orig; /* local col index */ 192 brow = row/bs; bcol = col/bs; 193 if (brow > bcol) continue; /* ignore lower triangular blocks of A */ 194 if (roworiented) value = v[i*n+j]; 195 else value = v[i+j*m]; 196 MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv); 197 /* ierr = MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv);CHKERRQ(ierr); */ 198 } else if (in[j] < 0) continue; 199 #if defined(PETSC_USE_DEBUG) 200 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); 201 #endif 202 else { /* off-diag entry (B) */ 203 if (mat->was_assembled) { 204 if (!baij->colmap) { 205 ierr = MatCreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 206 } 207 #if defined(PETSC_USE_CTABLE) 208 ierr = PetscTableFind(baij->colmap,in[j]/bs + 1,&col);CHKERRQ(ierr); 209 col = col - 1; 210 #else 211 col = baij->colmap[in[j]/bs] - 1; 212 #endif 213 if (col < 0 && !((Mat_SeqSBAIJ*)(baij->A->data))->nonew) { 214 ierr = MatDisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 215 col = in[j]; 216 /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */ 217 B = baij->B; 218 b = (Mat_SeqBAIJ*)(B)->data; 219 bimax= b->imax;bi=b->i;bilen=b->ilen;bj=b->j; 220 ba = b->a; 221 } else col += in[j]%bs; 222 } else col = in[j]; 223 if (roworiented) value = v[i*n+j]; 224 else value = v[i+j*m]; 225 MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv); 226 /* ierr = MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv);CHKERRQ(ierr); */ 227 } 228 } 229 } else { /* off processor entry */ 230 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]); 231 if (!baij->donotstash) { 232 mat->assembled = PETSC_FALSE; 233 n_loc = 0; 234 for (j=0; j<n; j++) { 235 if (im[i]/bs > in[j]/bs) continue; /* ignore lower triangular blocks */ 236 in_loc[n_loc] = in[j]; 237 if (roworiented) { 238 v_loc[n_loc] = v[i*n+j]; 239 } else { 240 v_loc[n_loc] = v[j*m+i]; 241 } 242 n_loc++; 243 } 244 ierr = MatStashValuesRow_Private(&mat->stash,im[i],n_loc,in_loc,v_loc,PETSC_FALSE);CHKERRQ(ierr); 245 } 246 } 247 } 248 PetscFunctionReturn(0); 249 } 250 251 #undef __FUNCT__ 252 #define __FUNCT__ "MatSetValuesBlocked_MPISBAIJ" 253 PetscErrorCode MatSetValuesBlocked_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const MatScalar v[],InsertMode addv) 254 { 255 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 256 const MatScalar *value; 257 MatScalar *barray =baij->barray; 258 PetscBool roworiented = baij->roworiented,ignore_ltriangular = ((Mat_SeqSBAIJ*)baij->A->data)->ignore_ltriangular; 259 PetscErrorCode ierr; 260 PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs; 261 PetscInt rend=baij->rendbs,cstart=baij->rstartbs,stepval; 262 PetscInt cend=baij->rendbs,bs=mat->rmap->bs,bs2=baij->bs2; 263 264 PetscFunctionBegin; 265 if (!barray) { 266 ierr = PetscMalloc1(bs2,&barray);CHKERRQ(ierr); 267 baij->barray = barray; 268 } 269 270 if (roworiented) { 271 stepval = (n-1)*bs; 272 } else { 273 stepval = (m-1)*bs; 274 } 275 for (i=0; i<m; i++) { 276 if (im[i] < 0) continue; 277 #if defined(PETSC_USE_DEBUG) 278 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); 279 #endif 280 if (im[i] >= rstart && im[i] < rend) { 281 row = im[i] - rstart; 282 for (j=0; j<n; j++) { 283 if (im[i] > in[j]) { 284 if (ignore_ltriangular) continue; /* ignore lower triangular blocks */ 285 else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)"); 286 } 287 /* If NumCol = 1 then a copy is not required */ 288 if ((roworiented) && (n == 1)) { 289 barray = (MatScalar*) v + i*bs2; 290 } else if ((!roworiented) && (m == 1)) { 291 barray = (MatScalar*) v + j*bs2; 292 } else { /* Here a copy is required */ 293 if (roworiented) { 294 value = v + i*(stepval+bs)*bs + j*bs; 295 } else { 296 value = v + j*(stepval+bs)*bs + i*bs; 297 } 298 for (ii=0; ii<bs; ii++,value+=stepval) { 299 for (jj=0; jj<bs; jj++) { 300 *barray++ = *value++; 301 } 302 } 303 barray -=bs2; 304 } 305 306 if (in[j] >= cstart && in[j] < cend) { 307 col = in[j] - cstart; 308 ierr = MatSetValuesBlocked_SeqSBAIJ(baij->A,1,&row,1,&col,barray,addv);CHKERRQ(ierr); 309 } else if (in[j] < 0) continue; 310 #if defined(PETSC_USE_DEBUG) 311 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); 312 #endif 313 else { 314 if (mat->was_assembled) { 315 if (!baij->colmap) { 316 ierr = MatCreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 317 } 318 319 #if defined(PETSC_USE_DEBUG) 320 #if defined(PETSC_USE_CTABLE) 321 { PetscInt data; 322 ierr = PetscTableFind(baij->colmap,in[j]+1,&data);CHKERRQ(ierr); 323 if ((data - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap"); 324 } 325 #else 326 if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Incorrect colmap"); 327 #endif 328 #endif 329 #if defined(PETSC_USE_CTABLE) 330 ierr = PetscTableFind(baij->colmap,in[j]+1,&col);CHKERRQ(ierr); 331 col = (col - 1)/bs; 332 #else 333 col = (baij->colmap[in[j]] - 1)/bs; 334 #endif 335 if (col < 0 && !((Mat_SeqBAIJ*)(baij->A->data))->nonew) { 336 ierr = MatDisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 337 col = in[j]; 338 } 339 } else col = in[j]; 340 ierr = MatSetValuesBlocked_SeqBAIJ(baij->B,1,&row,1,&col,barray,addv);CHKERRQ(ierr); 341 } 342 } 343 } else { 344 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]); 345 if (!baij->donotstash) { 346 if (roworiented) { 347 ierr = MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);CHKERRQ(ierr); 348 } else { 349 ierr = MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);CHKERRQ(ierr); 350 } 351 } 352 } 353 } 354 PetscFunctionReturn(0); 355 } 356 357 #undef __FUNCT__ 358 #define __FUNCT__ "MatGetValues_MPISBAIJ" 359 PetscErrorCode MatGetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[]) 360 { 361 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 362 PetscErrorCode ierr; 363 PetscInt bs = mat->rmap->bs,i,j,bsrstart = mat->rmap->rstart,bsrend = mat->rmap->rend; 364 PetscInt bscstart = mat->cmap->rstart,bscend = mat->cmap->rend,row,col,data; 365 366 PetscFunctionBegin; 367 for (i=0; i<m; i++) { 368 if (idxm[i] < 0) continue; /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",idxm[i]); */ 369 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); 370 if (idxm[i] >= bsrstart && idxm[i] < bsrend) { 371 row = idxm[i] - bsrstart; 372 for (j=0; j<n; j++) { 373 if (idxn[j] < 0) continue; /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column %D",idxn[j]); */ 374 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); 375 if (idxn[j] >= bscstart && idxn[j] < bscend) { 376 col = idxn[j] - bscstart; 377 ierr = MatGetValues_SeqSBAIJ(baij->A,1,&row,1,&col,v+i*n+j);CHKERRQ(ierr); 378 } else { 379 if (!baij->colmap) { 380 ierr = MatCreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 381 } 382 #if defined(PETSC_USE_CTABLE) 383 ierr = PetscTableFind(baij->colmap,idxn[j]/bs+1,&data);CHKERRQ(ierr); 384 data--; 385 #else 386 data = baij->colmap[idxn[j]/bs]-1; 387 #endif 388 if ((data < 0) || (baij->garray[data/bs] != idxn[j]/bs)) *(v+i*n+j) = 0.0; 389 else { 390 col = data + idxn[j]%bs; 391 ierr = MatGetValues_SeqBAIJ(baij->B,1,&row,1,&col,v+i*n+j);CHKERRQ(ierr); 392 } 393 } 394 } 395 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only local values currently supported"); 396 } 397 PetscFunctionReturn(0); 398 } 399 400 #undef __FUNCT__ 401 #define __FUNCT__ "MatNorm_MPISBAIJ" 402 PetscErrorCode MatNorm_MPISBAIJ(Mat mat,NormType type,PetscReal *norm) 403 { 404 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 405 PetscErrorCode ierr; 406 PetscReal sum[2],*lnorm2; 407 408 PetscFunctionBegin; 409 if (baij->size == 1) { 410 ierr = MatNorm(baij->A,type,norm);CHKERRQ(ierr); 411 } else { 412 if (type == NORM_FROBENIUS) { 413 ierr = PetscMalloc1(2,&lnorm2);CHKERRQ(ierr); 414 ierr = MatNorm(baij->A,type,lnorm2);CHKERRQ(ierr); 415 *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2++; /* squar power of norm(A) */ 416 ierr = MatNorm(baij->B,type,lnorm2);CHKERRQ(ierr); 417 *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2--; /* squar power of norm(B) */ 418 ierr = MPI_Allreduce(lnorm2,sum,2,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)mat));CHKERRQ(ierr); 419 *norm = PetscSqrtReal(sum[0] + 2*sum[1]); 420 ierr = PetscFree(lnorm2);CHKERRQ(ierr); 421 } else if (type == NORM_INFINITY || type == NORM_1) { /* max row/column sum */ 422 Mat_SeqSBAIJ *amat=(Mat_SeqSBAIJ*)baij->A->data; 423 Mat_SeqBAIJ *bmat=(Mat_SeqBAIJ*)baij->B->data; 424 PetscReal *rsum,*rsum2,vabs; 425 PetscInt *jj,*garray=baij->garray,rstart=baij->rstartbs,nz; 426 PetscInt brow,bcol,col,bs=baij->A->rmap->bs,row,grow,gcol,mbs=amat->mbs; 427 MatScalar *v; 428 429 ierr = PetscMalloc2(mat->cmap->N,&rsum,mat->cmap->N,&rsum2);CHKERRQ(ierr); 430 ierr = PetscMemzero(rsum,mat->cmap->N*sizeof(PetscReal));CHKERRQ(ierr); 431 /* Amat */ 432 v = amat->a; jj = amat->j; 433 for (brow=0; brow<mbs; brow++) { 434 grow = bs*(rstart + brow); 435 nz = amat->i[brow+1] - amat->i[brow]; 436 for (bcol=0; bcol<nz; bcol++) { 437 gcol = bs*(rstart + *jj); jj++; 438 for (col=0; col<bs; col++) { 439 for (row=0; row<bs; row++) { 440 vabs = PetscAbsScalar(*v); v++; 441 rsum[gcol+col] += vabs; 442 /* non-diagonal block */ 443 if (bcol > 0 && vabs > 0.0) rsum[grow+row] += vabs; 444 } 445 } 446 } 447 } 448 /* Bmat */ 449 v = bmat->a; jj = bmat->j; 450 for (brow=0; brow<mbs; brow++) { 451 grow = bs*(rstart + brow); 452 nz = bmat->i[brow+1] - bmat->i[brow]; 453 for (bcol=0; bcol<nz; bcol++) { 454 gcol = bs*garray[*jj]; jj++; 455 for (col=0; col<bs; col++) { 456 for (row=0; row<bs; row++) { 457 vabs = PetscAbsScalar(*v); v++; 458 rsum[gcol+col] += vabs; 459 rsum[grow+row] += vabs; 460 } 461 } 462 } 463 } 464 ierr = MPI_Allreduce(rsum,rsum2,mat->cmap->N,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)mat));CHKERRQ(ierr); 465 *norm = 0.0; 466 for (col=0; col<mat->cmap->N; col++) { 467 if (rsum2[col] > *norm) *norm = rsum2[col]; 468 } 469 ierr = PetscFree2(rsum,rsum2);CHKERRQ(ierr); 470 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"No support for this norm yet"); 471 } 472 PetscFunctionReturn(0); 473 } 474 475 #undef __FUNCT__ 476 #define __FUNCT__ "MatAssemblyBegin_MPISBAIJ" 477 PetscErrorCode MatAssemblyBegin_MPISBAIJ(Mat mat,MatAssemblyType mode) 478 { 479 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 480 PetscErrorCode ierr; 481 PetscInt nstash,reallocs; 482 InsertMode addv; 483 484 PetscFunctionBegin; 485 if (baij->donotstash || mat->nooffprocentries) PetscFunctionReturn(0); 486 487 /* make sure all processors are either in INSERTMODE or ADDMODE */ 488 ierr = MPI_Allreduce((PetscEnum*)&mat->insertmode,(PetscEnum*)&addv,1,MPIU_ENUM,MPI_BOR,PetscObjectComm((PetscObject)mat));CHKERRQ(ierr); 489 if (addv == (ADD_VALUES|INSERT_VALUES)) SETERRQ(PetscObjectComm((PetscObject)mat),PETSC_ERR_ARG_WRONGSTATE,"Some processors inserted others added"); 490 mat->insertmode = addv; /* in case this processor had no cache */ 491 492 ierr = MatStashScatterBegin_Private(mat,&mat->stash,mat->rmap->range);CHKERRQ(ierr); 493 ierr = MatStashScatterBegin_Private(mat,&mat->bstash,baij->rangebs);CHKERRQ(ierr); 494 ierr = MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);CHKERRQ(ierr); 495 ierr = PetscInfo2(mat,"Stash has %D entries,uses %D mallocs.\n",nstash,reallocs);CHKERRQ(ierr); 496 ierr = MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);CHKERRQ(ierr); 497 ierr = PetscInfo2(mat,"Block-Stash has %D entries, uses %D mallocs.\n",nstash,reallocs);CHKERRQ(ierr); 498 PetscFunctionReturn(0); 499 } 500 501 #undef __FUNCT__ 502 #define __FUNCT__ "MatAssemblyEnd_MPISBAIJ" 503 PetscErrorCode MatAssemblyEnd_MPISBAIJ(Mat mat,MatAssemblyType mode) 504 { 505 Mat_MPISBAIJ *baij=(Mat_MPISBAIJ*)mat->data; 506 Mat_SeqSBAIJ *a =(Mat_SeqSBAIJ*)baij->A->data; 507 PetscErrorCode ierr; 508 PetscInt i,j,rstart,ncols,flg,bs2=baij->bs2; 509 PetscInt *row,*col; 510 PetscBool other_disassembled; 511 PetscMPIInt n; 512 PetscBool r1,r2,r3; 513 MatScalar *val; 514 InsertMode addv = mat->insertmode; 515 516 /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */ 517 PetscFunctionBegin; 518 if (!baij->donotstash && !mat->nooffprocentries) { 519 while (1) { 520 ierr = MatStashScatterGetMesg_Private(&mat->stash,&n,&row,&col,&val,&flg);CHKERRQ(ierr); 521 if (!flg) break; 522 523 for (i=0; i<n;) { 524 /* Now identify the consecutive vals belonging to the same row */ 525 for (j=i,rstart=row[j]; j<n; j++) { 526 if (row[j] != rstart) break; 527 } 528 if (j < n) ncols = j-i; 529 else ncols = n-i; 530 /* Now assemble all these values with a single function call */ 531 ierr = MatSetValues_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i,addv);CHKERRQ(ierr); 532 i = j; 533 } 534 } 535 ierr = MatStashScatterEnd_Private(&mat->stash);CHKERRQ(ierr); 536 /* Now process the block-stash. Since the values are stashed column-oriented, 537 set the roworiented flag to column oriented, and after MatSetValues() 538 restore the original flags */ 539 r1 = baij->roworiented; 540 r2 = a->roworiented; 541 r3 = ((Mat_SeqBAIJ*)baij->B->data)->roworiented; 542 543 baij->roworiented = PETSC_FALSE; 544 a->roworiented = PETSC_FALSE; 545 546 ((Mat_SeqBAIJ*)baij->B->data)->roworiented = PETSC_FALSE; /* b->roworinted */ 547 while (1) { 548 ierr = MatStashScatterGetMesg_Private(&mat->bstash,&n,&row,&col,&val,&flg);CHKERRQ(ierr); 549 if (!flg) break; 550 551 for (i=0; i<n;) { 552 /* Now identify the consecutive vals belonging to the same row */ 553 for (j=i,rstart=row[j]; j<n; j++) { 554 if (row[j] != rstart) break; 555 } 556 if (j < n) ncols = j-i; 557 else ncols = n-i; 558 ierr = MatSetValuesBlocked_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i*bs2,addv);CHKERRQ(ierr); 559 i = j; 560 } 561 } 562 ierr = MatStashScatterEnd_Private(&mat->bstash);CHKERRQ(ierr); 563 564 baij->roworiented = r1; 565 a->roworiented = r2; 566 567 ((Mat_SeqBAIJ*)baij->B->data)->roworiented = r3; /* b->roworinted */ 568 } 569 570 ierr = MatAssemblyBegin(baij->A,mode);CHKERRQ(ierr); 571 ierr = MatAssemblyEnd(baij->A,mode);CHKERRQ(ierr); 572 573 /* determine if any processor has disassembled, if so we must 574 also disassemble ourselfs, in order that we may reassemble. */ 575 /* 576 if nonzero structure of submatrix B cannot change then we know that 577 no processor disassembled thus we can skip this stuff 578 */ 579 if (!((Mat_SeqBAIJ*)baij->B->data)->nonew) { 580 ierr = MPI_Allreduce(&mat->was_assembled,&other_disassembled,1,MPIU_BOOL,MPI_PROD,PetscObjectComm((PetscObject)mat));CHKERRQ(ierr); 581 if (mat->was_assembled && !other_disassembled) { 582 ierr = MatDisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 583 } 584 } 585 586 if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) { 587 ierr = MatSetUpMultiply_MPISBAIJ(mat);CHKERRQ(ierr); /* setup Mvctx and sMvctx */ 588 } 589 ierr = MatAssemblyBegin(baij->B,mode);CHKERRQ(ierr); 590 ierr = MatAssemblyEnd(baij->B,mode);CHKERRQ(ierr); 591 592 ierr = PetscFree2(baij->rowvalues,baij->rowindices);CHKERRQ(ierr); 593 594 baij->rowvalues = 0; 595 596 /* if no new nonzero locations are allowed in matrix then only set the matrix state the first time through */ 597 if ((!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) || !((Mat_SeqBAIJ*)(baij->A->data))->nonew) { 598 PetscObjectState state = baij->A->nonzerostate + baij->B->nonzerostate; 599 ierr = MPI_Allreduce(&state,&mat->nonzerostate,1,MPIU_INT64,MPI_SUM,PetscObjectComm((PetscObject)mat));CHKERRQ(ierr); 600 } 601 PetscFunctionReturn(0); 602 } 603 604 extern PetscErrorCode MatView_SeqSBAIJ_ASCII(Mat,PetscViewer); 605 extern PetscErrorCode MatSetValues_MPIBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 606 #include <petscdraw.h> 607 #undef __FUNCT__ 608 #define __FUNCT__ "MatView_MPISBAIJ_ASCIIorDraworSocket" 609 static PetscErrorCode MatView_MPISBAIJ_ASCIIorDraworSocket(Mat mat,PetscViewer viewer) 610 { 611 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 612 PetscErrorCode ierr; 613 PetscInt bs = mat->rmap->bs; 614 PetscMPIInt rank = baij->rank; 615 PetscBool iascii,isdraw; 616 PetscViewer sviewer; 617 PetscViewerFormat format; 618 619 PetscFunctionBegin; 620 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 621 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);CHKERRQ(ierr); 622 if (iascii) { 623 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 624 if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) { 625 MatInfo info; 626 ierr = MPI_Comm_rank(PetscObjectComm((PetscObject)mat),&rank);CHKERRQ(ierr); 627 ierr = MatGetInfo(mat,MAT_LOCAL,&info);CHKERRQ(ierr); 628 ierr = PetscViewerASCIISynchronizedAllow(viewer,PETSC_TRUE);CHKERRQ(ierr); 629 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] Local rows %D nz %D nz alloced %D bs %D mem %D\n",rank,mat->rmap->n,(PetscInt)info.nz_used,(PetscInt)info.nz_allocated,mat->rmap->bs,(PetscInt)info.memory);CHKERRQ(ierr); 630 ierr = MatGetInfo(baij->A,MAT_LOCAL,&info);CHKERRQ(ierr); 631 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] on-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used);CHKERRQ(ierr); 632 ierr = MatGetInfo(baij->B,MAT_LOCAL,&info);CHKERRQ(ierr); 633 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] off-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used);CHKERRQ(ierr); 634 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 635 ierr = PetscViewerASCIISynchronizedAllow(viewer,PETSC_FALSE);CHKERRQ(ierr); 636 ierr = PetscViewerASCIIPrintf(viewer,"Information on VecScatter used in matrix-vector product: \n");CHKERRQ(ierr); 637 ierr = VecScatterView(baij->Mvctx,viewer);CHKERRQ(ierr); 638 PetscFunctionReturn(0); 639 } else if (format == PETSC_VIEWER_ASCII_INFO) { 640 ierr = PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);CHKERRQ(ierr); 641 PetscFunctionReturn(0); 642 } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) { 643 PetscFunctionReturn(0); 644 } 645 } 646 647 if (isdraw) { 648 PetscDraw draw; 649 PetscBool isnull; 650 ierr = PetscViewerDrawGetDraw(viewer,0,&draw);CHKERRQ(ierr); 651 ierr = PetscDrawIsNull(draw,&isnull);CHKERRQ(ierr); if (isnull) PetscFunctionReturn(0); 652 } 653 654 { 655 /* assemble the entire matrix onto first processor. */ 656 Mat A; 657 Mat_SeqSBAIJ *Aloc; 658 Mat_SeqBAIJ *Bloc; 659 PetscInt M = mat->rmap->N,N = mat->cmap->N,*ai,*aj,col,i,j,k,*rvals,mbs = baij->mbs; 660 MatScalar *a; 661 662 /* Should this be the same type as mat? */ 663 ierr = MatCreate(PetscObjectComm((PetscObject)mat),&A);CHKERRQ(ierr); 664 if (!rank) { 665 ierr = MatSetSizes(A,M,N,M,N);CHKERRQ(ierr); 666 } else { 667 ierr = MatSetSizes(A,0,0,M,N);CHKERRQ(ierr); 668 } 669 ierr = MatSetType(A,MATMPISBAIJ);CHKERRQ(ierr); 670 ierr = MatMPISBAIJSetPreallocation(A,mat->rmap->bs,0,NULL,0,NULL);CHKERRQ(ierr); 671 ierr = MatSetOption(A,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_FALSE);CHKERRQ(ierr); 672 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)A);CHKERRQ(ierr); 673 674 /* copy over the A part */ 675 Aloc = (Mat_SeqSBAIJ*)baij->A->data; 676 ai = Aloc->i; aj = Aloc->j; a = Aloc->a; 677 ierr = PetscMalloc1(bs,&rvals);CHKERRQ(ierr); 678 679 for (i=0; i<mbs; i++) { 680 rvals[0] = bs*(baij->rstartbs + i); 681 for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1; 682 for (j=ai[i]; j<ai[i+1]; j++) { 683 col = (baij->cstartbs+aj[j])*bs; 684 for (k=0; k<bs; k++) { 685 ierr = MatSetValues_MPISBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);CHKERRQ(ierr); 686 col++; 687 a += bs; 688 } 689 } 690 } 691 /* copy over the B part */ 692 Bloc = (Mat_SeqBAIJ*)baij->B->data; 693 ai = Bloc->i; aj = Bloc->j; a = Bloc->a; 694 for (i=0; i<mbs; i++) { 695 696 rvals[0] = bs*(baij->rstartbs + i); 697 for (j=1; j<bs; j++) rvals[j] = rvals[j-1] + 1; 698 for (j=ai[i]; j<ai[i+1]; j++) { 699 col = baij->garray[aj[j]]*bs; 700 for (k=0; k<bs; k++) { 701 ierr = MatSetValues_MPIBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);CHKERRQ(ierr); 702 col++; 703 a += bs; 704 } 705 } 706 } 707 ierr = PetscFree(rvals);CHKERRQ(ierr); 708 ierr = MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 709 ierr = MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 710 /* 711 Everyone has to call to draw the matrix since the graphics waits are 712 synchronized across all processors that share the PetscDraw object 713 */ 714 ierr = PetscViewerGetSingleton(viewer,&sviewer);CHKERRQ(ierr); 715 if (!rank) { 716 ierr = MatView_SeqSBAIJ_ASCII(((Mat_MPISBAIJ*)(A->data))->A,sviewer);CHKERRQ(ierr); 717 } 718 ierr = PetscViewerRestoreSingleton(viewer,&sviewer);CHKERRQ(ierr); 719 ierr = MatDestroy(&A);CHKERRQ(ierr); 720 } 721 PetscFunctionReturn(0); 722 } 723 724 #undef __FUNCT__ 725 #define __FUNCT__ "MatView_MPISBAIJ" 726 PetscErrorCode MatView_MPISBAIJ(Mat mat,PetscViewer viewer) 727 { 728 PetscErrorCode ierr; 729 PetscBool iascii,isdraw,issocket,isbinary; 730 731 PetscFunctionBegin; 732 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 733 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);CHKERRQ(ierr); 734 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSOCKET,&issocket);CHKERRQ(ierr); 735 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);CHKERRQ(ierr); 736 if (iascii || isdraw || issocket || isbinary) { 737 ierr = MatView_MPISBAIJ_ASCIIorDraworSocket(mat,viewer);CHKERRQ(ierr); 738 } 739 PetscFunctionReturn(0); 740 } 741 742 #undef __FUNCT__ 743 #define __FUNCT__ "MatDestroy_MPISBAIJ" 744 PetscErrorCode MatDestroy_MPISBAIJ(Mat mat) 745 { 746 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 747 PetscErrorCode ierr; 748 749 PetscFunctionBegin; 750 #if defined(PETSC_USE_LOG) 751 PetscLogObjectState((PetscObject)mat,"Rows=%D,Cols=%D",mat->rmap->N,mat->cmap->N); 752 #endif 753 ierr = MatStashDestroy_Private(&mat->stash);CHKERRQ(ierr); 754 ierr = MatStashDestroy_Private(&mat->bstash);CHKERRQ(ierr); 755 ierr = MatDestroy(&baij->A);CHKERRQ(ierr); 756 ierr = MatDestroy(&baij->B);CHKERRQ(ierr); 757 #if defined(PETSC_USE_CTABLE) 758 ierr = PetscTableDestroy(&baij->colmap);CHKERRQ(ierr); 759 #else 760 ierr = PetscFree(baij->colmap);CHKERRQ(ierr); 761 #endif 762 ierr = PetscFree(baij->garray);CHKERRQ(ierr); 763 ierr = VecDestroy(&baij->lvec);CHKERRQ(ierr); 764 ierr = VecScatterDestroy(&baij->Mvctx);CHKERRQ(ierr); 765 ierr = VecDestroy(&baij->slvec0);CHKERRQ(ierr); 766 ierr = VecDestroy(&baij->slvec0b);CHKERRQ(ierr); 767 ierr = VecDestroy(&baij->slvec1);CHKERRQ(ierr); 768 ierr = VecDestroy(&baij->slvec1a);CHKERRQ(ierr); 769 ierr = VecDestroy(&baij->slvec1b);CHKERRQ(ierr); 770 ierr = VecScatterDestroy(&baij->sMvctx);CHKERRQ(ierr); 771 ierr = PetscFree2(baij->rowvalues,baij->rowindices);CHKERRQ(ierr); 772 ierr = PetscFree(baij->barray);CHKERRQ(ierr); 773 ierr = PetscFree(baij->hd);CHKERRQ(ierr); 774 ierr = VecDestroy(&baij->diag);CHKERRQ(ierr); 775 ierr = VecDestroy(&baij->bb1);CHKERRQ(ierr); 776 ierr = VecDestroy(&baij->xx1);CHKERRQ(ierr); 777 #if defined(PETSC_USE_REAL_MAT_SINGLE) 778 ierr = PetscFree(baij->setvaluescopy);CHKERRQ(ierr); 779 #endif 780 ierr = PetscFree(baij->in_loc);CHKERRQ(ierr); 781 ierr = PetscFree(baij->v_loc);CHKERRQ(ierr); 782 ierr = PetscFree(baij->rangebs);CHKERRQ(ierr); 783 ierr = PetscFree(mat->data);CHKERRQ(ierr); 784 785 ierr = PetscObjectChangeTypeName((PetscObject)mat,0);CHKERRQ(ierr); 786 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatStoreValues_C",NULL);CHKERRQ(ierr); 787 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatRetrieveValues_C",NULL);CHKERRQ(ierr); 788 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatGetDiagonalBlock_C",NULL);CHKERRQ(ierr); 789 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatMPISBAIJSetPreallocation_C",NULL);CHKERRQ(ierr); 790 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatConvert_mpisbaij_mpisbstrm_C",NULL);CHKERRQ(ierr); 791 PetscFunctionReturn(0); 792 } 793 794 #undef __FUNCT__ 795 #define __FUNCT__ "MatMult_MPISBAIJ_Hermitian" 796 PetscErrorCode MatMult_MPISBAIJ_Hermitian(Mat A,Vec xx,Vec yy) 797 { 798 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 799 PetscErrorCode ierr; 800 PetscInt nt,mbs=a->mbs,bs=A->rmap->bs; 801 PetscScalar *x,*from; 802 803 PetscFunctionBegin; 804 ierr = VecGetLocalSize(xx,&nt);CHKERRQ(ierr); 805 if (nt != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx"); 806 807 /* diagonal part */ 808 ierr = (*a->A->ops->mult)(a->A,xx,a->slvec1a);CHKERRQ(ierr); 809 ierr = VecSet(a->slvec1b,0.0);CHKERRQ(ierr); 810 811 /* subdiagonal part */ 812 ierr = (*a->B->ops->multhermitiantranspose)(a->B,xx,a->slvec0b);CHKERRQ(ierr); 813 814 /* copy x into the vec slvec0 */ 815 ierr = VecGetArray(a->slvec0,&from);CHKERRQ(ierr); 816 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 817 818 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 819 ierr = VecRestoreArray(a->slvec0,&from);CHKERRQ(ierr); 820 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 821 822 ierr = VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 823 ierr = VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 824 /* supperdiagonal part */ 825 ierr = (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);CHKERRQ(ierr); 826 PetscFunctionReturn(0); 827 } 828 829 #undef __FUNCT__ 830 #define __FUNCT__ "MatMult_MPISBAIJ" 831 PetscErrorCode MatMult_MPISBAIJ(Mat A,Vec xx,Vec yy) 832 { 833 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 834 PetscErrorCode ierr; 835 PetscInt nt,mbs=a->mbs,bs=A->rmap->bs; 836 PetscScalar *x,*from; 837 838 PetscFunctionBegin; 839 ierr = VecGetLocalSize(xx,&nt);CHKERRQ(ierr); 840 if (nt != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx"); 841 842 /* diagonal part */ 843 ierr = (*a->A->ops->mult)(a->A,xx,a->slvec1a);CHKERRQ(ierr); 844 ierr = VecSet(a->slvec1b,0.0);CHKERRQ(ierr); 845 846 /* subdiagonal part */ 847 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);CHKERRQ(ierr); 848 849 /* copy x into the vec slvec0 */ 850 ierr = VecGetArray(a->slvec0,&from);CHKERRQ(ierr); 851 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 852 853 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 854 ierr = VecRestoreArray(a->slvec0,&from);CHKERRQ(ierr); 855 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 856 857 ierr = VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 858 ierr = VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 859 /* supperdiagonal part */ 860 ierr = (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);CHKERRQ(ierr); 861 PetscFunctionReturn(0); 862 } 863 864 #undef __FUNCT__ 865 #define __FUNCT__ "MatMult_MPISBAIJ_2comm" 866 PetscErrorCode MatMult_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy) 867 { 868 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 869 PetscErrorCode ierr; 870 PetscInt nt; 871 872 PetscFunctionBegin; 873 ierr = VecGetLocalSize(xx,&nt);CHKERRQ(ierr); 874 if (nt != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx"); 875 876 ierr = VecGetLocalSize(yy,&nt);CHKERRQ(ierr); 877 if (nt != A->rmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Incompatible parition of A and yy"); 878 879 ierr = VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 880 /* do diagonal part */ 881 ierr = (*a->A->ops->mult)(a->A,xx,yy);CHKERRQ(ierr); 882 /* do supperdiagonal part */ 883 ierr = VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 884 ierr = (*a->B->ops->multadd)(a->B,a->lvec,yy,yy);CHKERRQ(ierr); 885 /* do subdiagonal part */ 886 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->lvec);CHKERRQ(ierr); 887 ierr = VecScatterBegin(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 888 ierr = VecScatterEnd(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 889 PetscFunctionReturn(0); 890 } 891 892 #undef __FUNCT__ 893 #define __FUNCT__ "MatMultAdd_MPISBAIJ" 894 PetscErrorCode MatMultAdd_MPISBAIJ(Mat A,Vec xx,Vec yy,Vec zz) 895 { 896 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 897 PetscErrorCode ierr; 898 PetscInt mbs=a->mbs,bs=A->rmap->bs; 899 PetscScalar *x,*from,zero=0.0; 900 901 PetscFunctionBegin; 902 /* 903 PetscSynchronizedPrintf(PetscObjectComm((PetscObject)A)," MatMultAdd is called ...\n"); 904 PetscSynchronizedFlush(PetscObjectComm((PetscObject)A),PETSC_STDOUT); 905 */ 906 /* diagonal part */ 907 ierr = (*a->A->ops->multadd)(a->A,xx,yy,a->slvec1a);CHKERRQ(ierr); 908 ierr = VecSet(a->slvec1b,zero);CHKERRQ(ierr); 909 910 /* subdiagonal part */ 911 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);CHKERRQ(ierr); 912 913 /* copy x into the vec slvec0 */ 914 ierr = VecGetArray(a->slvec0,&from);CHKERRQ(ierr); 915 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 916 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 917 ierr = VecRestoreArray(a->slvec0,&from);CHKERRQ(ierr); 918 919 ierr = VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 920 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 921 ierr = VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 922 923 /* supperdiagonal part */ 924 ierr = (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,zz);CHKERRQ(ierr); 925 PetscFunctionReturn(0); 926 } 927 928 #undef __FUNCT__ 929 #define __FUNCT__ "MatMultAdd_MPISBAIJ_2comm" 930 PetscErrorCode MatMultAdd_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy,Vec zz) 931 { 932 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 933 PetscErrorCode ierr; 934 935 PetscFunctionBegin; 936 ierr = VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 937 /* do diagonal part */ 938 ierr = (*a->A->ops->multadd)(a->A,xx,yy,zz);CHKERRQ(ierr); 939 /* do supperdiagonal part */ 940 ierr = VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 941 ierr = (*a->B->ops->multadd)(a->B,a->lvec,zz,zz);CHKERRQ(ierr); 942 943 /* do subdiagonal part */ 944 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->lvec);CHKERRQ(ierr); 945 ierr = VecScatterBegin(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 946 ierr = VecScatterEnd(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 947 PetscFunctionReturn(0); 948 } 949 950 /* 951 This only works correctly for square matrices where the subblock A->A is the 952 diagonal block 953 */ 954 #undef __FUNCT__ 955 #define __FUNCT__ "MatGetDiagonal_MPISBAIJ" 956 PetscErrorCode MatGetDiagonal_MPISBAIJ(Mat A,Vec v) 957 { 958 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 959 PetscErrorCode ierr; 960 961 PetscFunctionBegin; 962 /* if (a->rmap->N != a->cmap->N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block"); */ 963 ierr = MatGetDiagonal(a->A,v);CHKERRQ(ierr); 964 PetscFunctionReturn(0); 965 } 966 967 #undef __FUNCT__ 968 #define __FUNCT__ "MatScale_MPISBAIJ" 969 PetscErrorCode MatScale_MPISBAIJ(Mat A,PetscScalar aa) 970 { 971 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 972 PetscErrorCode ierr; 973 974 PetscFunctionBegin; 975 ierr = MatScale(a->A,aa);CHKERRQ(ierr); 976 ierr = MatScale(a->B,aa);CHKERRQ(ierr); 977 PetscFunctionReturn(0); 978 } 979 980 #undef __FUNCT__ 981 #define __FUNCT__ "MatGetRow_MPISBAIJ" 982 PetscErrorCode MatGetRow_MPISBAIJ(Mat matin,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 983 { 984 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 985 PetscScalar *vworkA,*vworkB,**pvA,**pvB,*v_p; 986 PetscErrorCode ierr; 987 PetscInt bs = matin->rmap->bs,bs2 = mat->bs2,i,*cworkA,*cworkB,**pcA,**pcB; 988 PetscInt nztot,nzA,nzB,lrow,brstart = matin->rmap->rstart,brend = matin->rmap->rend; 989 PetscInt *cmap,*idx_p,cstart = mat->rstartbs; 990 991 PetscFunctionBegin; 992 if (mat->getrowactive) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Already active"); 993 mat->getrowactive = PETSC_TRUE; 994 995 if (!mat->rowvalues && (idx || v)) { 996 /* 997 allocate enough space to hold information from the longest row. 998 */ 999 Mat_SeqSBAIJ *Aa = (Mat_SeqSBAIJ*)mat->A->data; 1000 Mat_SeqBAIJ *Ba = (Mat_SeqBAIJ*)mat->B->data; 1001 PetscInt max = 1,mbs = mat->mbs,tmp; 1002 for (i=0; i<mbs; i++) { 1003 tmp = Aa->i[i+1] - Aa->i[i] + Ba->i[i+1] - Ba->i[i]; /* row length */ 1004 if (max < tmp) max = tmp; 1005 } 1006 ierr = PetscMalloc2(max*bs2,&mat->rowvalues,max*bs2,&mat->rowindices);CHKERRQ(ierr); 1007 } 1008 1009 if (row < brstart || row >= brend) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only local rows"); 1010 lrow = row - brstart; /* local row index */ 1011 1012 pvA = &vworkA; pcA = &cworkA; pvB = &vworkB; pcB = &cworkB; 1013 if (!v) {pvA = 0; pvB = 0;} 1014 if (!idx) {pcA = 0; if (!v) pcB = 0;} 1015 ierr = (*mat->A->ops->getrow)(mat->A,lrow,&nzA,pcA,pvA);CHKERRQ(ierr); 1016 ierr = (*mat->B->ops->getrow)(mat->B,lrow,&nzB,pcB,pvB);CHKERRQ(ierr); 1017 nztot = nzA + nzB; 1018 1019 cmap = mat->garray; 1020 if (v || idx) { 1021 if (nztot) { 1022 /* Sort by increasing column numbers, assuming A and B already sorted */ 1023 PetscInt imark = -1; 1024 if (v) { 1025 *v = v_p = mat->rowvalues; 1026 for (i=0; i<nzB; i++) { 1027 if (cmap[cworkB[i]/bs] < cstart) v_p[i] = vworkB[i]; 1028 else break; 1029 } 1030 imark = i; 1031 for (i=0; i<nzA; i++) v_p[imark+i] = vworkA[i]; 1032 for (i=imark; i<nzB; i++) v_p[nzA+i] = vworkB[i]; 1033 } 1034 if (idx) { 1035 *idx = idx_p = mat->rowindices; 1036 if (imark > -1) { 1037 for (i=0; i<imark; i++) { 1038 idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs; 1039 } 1040 } else { 1041 for (i=0; i<nzB; i++) { 1042 if (cmap[cworkB[i]/bs] < cstart) idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs; 1043 else break; 1044 } 1045 imark = i; 1046 } 1047 for (i=0; i<nzA; i++) idx_p[imark+i] = cstart*bs + cworkA[i]; 1048 for (i=imark; i<nzB; i++) idx_p[nzA+i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ; 1049 } 1050 } else { 1051 if (idx) *idx = 0; 1052 if (v) *v = 0; 1053 } 1054 } 1055 *nz = nztot; 1056 ierr = (*mat->A->ops->restorerow)(mat->A,lrow,&nzA,pcA,pvA);CHKERRQ(ierr); 1057 ierr = (*mat->B->ops->restorerow)(mat->B,lrow,&nzB,pcB,pvB);CHKERRQ(ierr); 1058 PetscFunctionReturn(0); 1059 } 1060 1061 #undef __FUNCT__ 1062 #define __FUNCT__ "MatRestoreRow_MPISBAIJ" 1063 PetscErrorCode MatRestoreRow_MPISBAIJ(Mat mat,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 1064 { 1065 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 1066 1067 PetscFunctionBegin; 1068 if (!baij->getrowactive) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"MatGetRow() must be called first"); 1069 baij->getrowactive = PETSC_FALSE; 1070 PetscFunctionReturn(0); 1071 } 1072 1073 #undef __FUNCT__ 1074 #define __FUNCT__ "MatGetRowUpperTriangular_MPISBAIJ" 1075 PetscErrorCode MatGetRowUpperTriangular_MPISBAIJ(Mat A) 1076 { 1077 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1078 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1079 1080 PetscFunctionBegin; 1081 aA->getrow_utriangular = PETSC_TRUE; 1082 PetscFunctionReturn(0); 1083 } 1084 #undef __FUNCT__ 1085 #define __FUNCT__ "MatRestoreRowUpperTriangular_MPISBAIJ" 1086 PetscErrorCode MatRestoreRowUpperTriangular_MPISBAIJ(Mat A) 1087 { 1088 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1089 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1090 1091 PetscFunctionBegin; 1092 aA->getrow_utriangular = PETSC_FALSE; 1093 PetscFunctionReturn(0); 1094 } 1095 1096 #undef __FUNCT__ 1097 #define __FUNCT__ "MatRealPart_MPISBAIJ" 1098 PetscErrorCode MatRealPart_MPISBAIJ(Mat A) 1099 { 1100 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1101 PetscErrorCode ierr; 1102 1103 PetscFunctionBegin; 1104 ierr = MatRealPart(a->A);CHKERRQ(ierr); 1105 ierr = MatRealPart(a->B);CHKERRQ(ierr); 1106 PetscFunctionReturn(0); 1107 } 1108 1109 #undef __FUNCT__ 1110 #define __FUNCT__ "MatImaginaryPart_MPISBAIJ" 1111 PetscErrorCode MatImaginaryPart_MPISBAIJ(Mat A) 1112 { 1113 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1114 PetscErrorCode ierr; 1115 1116 PetscFunctionBegin; 1117 ierr = MatImaginaryPart(a->A);CHKERRQ(ierr); 1118 ierr = MatImaginaryPart(a->B);CHKERRQ(ierr); 1119 PetscFunctionReturn(0); 1120 } 1121 1122 #undef __FUNCT__ 1123 #define __FUNCT__ "MatZeroEntries_MPISBAIJ" 1124 PetscErrorCode MatZeroEntries_MPISBAIJ(Mat A) 1125 { 1126 Mat_MPISBAIJ *l = (Mat_MPISBAIJ*)A->data; 1127 PetscErrorCode ierr; 1128 1129 PetscFunctionBegin; 1130 ierr = MatZeroEntries(l->A);CHKERRQ(ierr); 1131 ierr = MatZeroEntries(l->B);CHKERRQ(ierr); 1132 PetscFunctionReturn(0); 1133 } 1134 1135 #undef __FUNCT__ 1136 #define __FUNCT__ "MatGetInfo_MPISBAIJ" 1137 PetscErrorCode MatGetInfo_MPISBAIJ(Mat matin,MatInfoType flag,MatInfo *info) 1138 { 1139 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)matin->data; 1140 Mat A = a->A,B = a->B; 1141 PetscErrorCode ierr; 1142 PetscReal isend[5],irecv[5]; 1143 1144 PetscFunctionBegin; 1145 info->block_size = (PetscReal)matin->rmap->bs; 1146 1147 ierr = MatGetInfo(A,MAT_LOCAL,info);CHKERRQ(ierr); 1148 1149 isend[0] = info->nz_used; isend[1] = info->nz_allocated; isend[2] = info->nz_unneeded; 1150 isend[3] = info->memory; isend[4] = info->mallocs; 1151 1152 ierr = MatGetInfo(B,MAT_LOCAL,info);CHKERRQ(ierr); 1153 1154 isend[0] += info->nz_used; isend[1] += info->nz_allocated; isend[2] += info->nz_unneeded; 1155 isend[3] += info->memory; isend[4] += info->mallocs; 1156 if (flag == MAT_LOCAL) { 1157 info->nz_used = isend[0]; 1158 info->nz_allocated = isend[1]; 1159 info->nz_unneeded = isend[2]; 1160 info->memory = isend[3]; 1161 info->mallocs = isend[4]; 1162 } else if (flag == MAT_GLOBAL_MAX) { 1163 ierr = MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPIU_MAX,PetscObjectComm((PetscObject)matin));CHKERRQ(ierr); 1164 1165 info->nz_used = irecv[0]; 1166 info->nz_allocated = irecv[1]; 1167 info->nz_unneeded = irecv[2]; 1168 info->memory = irecv[3]; 1169 info->mallocs = irecv[4]; 1170 } else if (flag == MAT_GLOBAL_SUM) { 1171 ierr = MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)matin));CHKERRQ(ierr); 1172 1173 info->nz_used = irecv[0]; 1174 info->nz_allocated = irecv[1]; 1175 info->nz_unneeded = irecv[2]; 1176 info->memory = irecv[3]; 1177 info->mallocs = irecv[4]; 1178 } else SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown MatInfoType argument %d",(int)flag); 1179 info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */ 1180 info->fill_ratio_needed = 0; 1181 info->factor_mallocs = 0; 1182 PetscFunctionReturn(0); 1183 } 1184 1185 #undef __FUNCT__ 1186 #define __FUNCT__ "MatSetOption_MPISBAIJ" 1187 PetscErrorCode MatSetOption_MPISBAIJ(Mat A,MatOption op,PetscBool flg) 1188 { 1189 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1190 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1191 PetscErrorCode ierr; 1192 1193 PetscFunctionBegin; 1194 switch (op) { 1195 case MAT_NEW_NONZERO_LOCATIONS: 1196 case MAT_NEW_NONZERO_ALLOCATION_ERR: 1197 case MAT_UNUSED_NONZERO_LOCATION_ERR: 1198 case MAT_KEEP_NONZERO_PATTERN: 1199 case MAT_NEW_NONZERO_LOCATION_ERR: 1200 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1201 ierr = MatSetOption(a->B,op,flg);CHKERRQ(ierr); 1202 break; 1203 case MAT_ROW_ORIENTED: 1204 a->roworiented = flg; 1205 1206 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1207 ierr = MatSetOption(a->B,op,flg);CHKERRQ(ierr); 1208 break; 1209 case MAT_NEW_DIAGONALS: 1210 ierr = PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);CHKERRQ(ierr); 1211 break; 1212 case MAT_IGNORE_OFF_PROC_ENTRIES: 1213 a->donotstash = flg; 1214 break; 1215 case MAT_USE_HASH_TABLE: 1216 a->ht_flag = flg; 1217 break; 1218 case MAT_HERMITIAN: 1219 if (!A->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Must call MatAssemblyEnd() first"); 1220 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1221 1222 A->ops->mult = MatMult_MPISBAIJ_Hermitian; 1223 break; 1224 case MAT_SPD: 1225 A->spd_set = PETSC_TRUE; 1226 A->spd = flg; 1227 if (flg) { 1228 A->symmetric = PETSC_TRUE; 1229 A->structurally_symmetric = PETSC_TRUE; 1230 A->symmetric_set = PETSC_TRUE; 1231 A->structurally_symmetric_set = PETSC_TRUE; 1232 } 1233 break; 1234 case MAT_SYMMETRIC: 1235 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1236 break; 1237 case MAT_STRUCTURALLY_SYMMETRIC: 1238 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1239 break; 1240 case MAT_SYMMETRY_ETERNAL: 1241 if (!flg) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Matrix must be symmetric"); 1242 ierr = PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);CHKERRQ(ierr); 1243 break; 1244 case MAT_IGNORE_LOWER_TRIANGULAR: 1245 aA->ignore_ltriangular = flg; 1246 break; 1247 case MAT_ERROR_LOWER_TRIANGULAR: 1248 aA->ignore_ltriangular = flg; 1249 break; 1250 case MAT_GETROW_UPPERTRIANGULAR: 1251 aA->getrow_utriangular = flg; 1252 break; 1253 default: 1254 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"unknown option %d",op); 1255 } 1256 PetscFunctionReturn(0); 1257 } 1258 1259 #undef __FUNCT__ 1260 #define __FUNCT__ "MatTranspose_MPISBAIJ" 1261 PetscErrorCode MatTranspose_MPISBAIJ(Mat A,MatReuse reuse,Mat *B) 1262 { 1263 PetscErrorCode ierr; 1264 1265 PetscFunctionBegin; 1266 if (MAT_INITIAL_MATRIX || *B != A) { 1267 ierr = MatDuplicate(A,MAT_COPY_VALUES,B);CHKERRQ(ierr); 1268 } 1269 PetscFunctionReturn(0); 1270 } 1271 1272 #undef __FUNCT__ 1273 #define __FUNCT__ "MatDiagonalScale_MPISBAIJ" 1274 PetscErrorCode MatDiagonalScale_MPISBAIJ(Mat mat,Vec ll,Vec rr) 1275 { 1276 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 1277 Mat a = baij->A, b=baij->B; 1278 PetscErrorCode ierr; 1279 PetscInt nv,m,n; 1280 PetscBool flg; 1281 1282 PetscFunctionBegin; 1283 if (ll != rr) { 1284 ierr = VecEqual(ll,rr,&flg);CHKERRQ(ierr); 1285 if (!flg) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"For symmetric format, left and right scaling vectors must be same\n"); 1286 } 1287 if (!ll) PetscFunctionReturn(0); 1288 1289 ierr = MatGetLocalSize(mat,&m,&n);CHKERRQ(ierr); 1290 if (m != n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"For symmetric format, local size %d %d must be same",m,n); 1291 1292 ierr = VecGetLocalSize(rr,&nv);CHKERRQ(ierr); 1293 if (nv!=n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Left and right vector non-conforming local size"); 1294 1295 ierr = VecScatterBegin(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1296 1297 /* left diagonalscale the off-diagonal part */ 1298 ierr = (*b->ops->diagonalscale)(b,ll,NULL);CHKERRQ(ierr); 1299 1300 /* scale the diagonal part */ 1301 ierr = (*a->ops->diagonalscale)(a,ll,rr);CHKERRQ(ierr); 1302 1303 /* right diagonalscale the off-diagonal part */ 1304 ierr = VecScatterEnd(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1305 ierr = (*b->ops->diagonalscale)(b,NULL,baij->lvec);CHKERRQ(ierr); 1306 PetscFunctionReturn(0); 1307 } 1308 1309 #undef __FUNCT__ 1310 #define __FUNCT__ "MatSetUnfactored_MPISBAIJ" 1311 PetscErrorCode MatSetUnfactored_MPISBAIJ(Mat A) 1312 { 1313 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1314 PetscErrorCode ierr; 1315 1316 PetscFunctionBegin; 1317 ierr = MatSetUnfactored(a->A);CHKERRQ(ierr); 1318 PetscFunctionReturn(0); 1319 } 1320 1321 static PetscErrorCode MatDuplicate_MPISBAIJ(Mat,MatDuplicateOption,Mat*); 1322 1323 #undef __FUNCT__ 1324 #define __FUNCT__ "MatEqual_MPISBAIJ" 1325 PetscErrorCode MatEqual_MPISBAIJ(Mat A,Mat B,PetscBool *flag) 1326 { 1327 Mat_MPISBAIJ *matB = (Mat_MPISBAIJ*)B->data,*matA = (Mat_MPISBAIJ*)A->data; 1328 Mat a,b,c,d; 1329 PetscBool flg; 1330 PetscErrorCode ierr; 1331 1332 PetscFunctionBegin; 1333 a = matA->A; b = matA->B; 1334 c = matB->A; d = matB->B; 1335 1336 ierr = MatEqual(a,c,&flg);CHKERRQ(ierr); 1337 if (flg) { 1338 ierr = MatEqual(b,d,&flg);CHKERRQ(ierr); 1339 } 1340 ierr = MPI_Allreduce(&flg,flag,1,MPIU_BOOL,MPI_LAND,PetscObjectComm((PetscObject)A));CHKERRQ(ierr); 1341 PetscFunctionReturn(0); 1342 } 1343 1344 #undef __FUNCT__ 1345 #define __FUNCT__ "MatCopy_MPISBAIJ" 1346 PetscErrorCode MatCopy_MPISBAIJ(Mat A,Mat B,MatStructure str) 1347 { 1348 PetscErrorCode ierr; 1349 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1350 Mat_MPISBAIJ *b = (Mat_MPISBAIJ*)B->data; 1351 1352 PetscFunctionBegin; 1353 /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */ 1354 if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) { 1355 ierr = MatGetRowUpperTriangular(A);CHKERRQ(ierr); 1356 ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr); 1357 ierr = MatRestoreRowUpperTriangular(A);CHKERRQ(ierr); 1358 } else { 1359 ierr = MatCopy(a->A,b->A,str);CHKERRQ(ierr); 1360 ierr = MatCopy(a->B,b->B,str);CHKERRQ(ierr); 1361 } 1362 PetscFunctionReturn(0); 1363 } 1364 1365 #undef __FUNCT__ 1366 #define __FUNCT__ "MatSetUp_MPISBAIJ" 1367 PetscErrorCode MatSetUp_MPISBAIJ(Mat A) 1368 { 1369 PetscErrorCode ierr; 1370 1371 PetscFunctionBegin; 1372 ierr = MatMPISBAIJSetPreallocation(A,A->rmap->bs,PETSC_DEFAULT,0,PETSC_DEFAULT,0);CHKERRQ(ierr); 1373 PetscFunctionReturn(0); 1374 } 1375 1376 #undef __FUNCT__ 1377 #define __FUNCT__ "MatAXPY_MPISBAIJ" 1378 PetscErrorCode MatAXPY_MPISBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str) 1379 { 1380 PetscErrorCode ierr; 1381 Mat_MPISBAIJ *xx=(Mat_MPISBAIJ*)X->data,*yy=(Mat_MPISBAIJ*)Y->data; 1382 PetscBLASInt bnz,one=1; 1383 Mat_SeqSBAIJ *xa,*ya; 1384 Mat_SeqBAIJ *xb,*yb; 1385 1386 PetscFunctionBegin; 1387 if (str == SAME_NONZERO_PATTERN) { 1388 PetscScalar alpha = a; 1389 xa = (Mat_SeqSBAIJ*)xx->A->data; 1390 ya = (Mat_SeqSBAIJ*)yy->A->data; 1391 ierr = PetscBLASIntCast(xa->nz,&bnz);CHKERRQ(ierr); 1392 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,xa->a,&one,ya->a,&one)); 1393 xb = (Mat_SeqBAIJ*)xx->B->data; 1394 yb = (Mat_SeqBAIJ*)yy->B->data; 1395 ierr = PetscBLASIntCast(xb->nz,&bnz);CHKERRQ(ierr); 1396 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,xb->a,&one,yb->a,&one)); 1397 ierr = PetscObjectStateIncrease((PetscObject)Y);CHKERRQ(ierr); 1398 } else { 1399 Mat B; 1400 PetscInt *nnz_d,*nnz_o,bs=Y->rmap->bs; 1401 if (bs != X->rmap->bs) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Matrices must have same block size"); 1402 ierr = MatGetRowUpperTriangular(X);CHKERRQ(ierr); 1403 ierr = MatGetRowUpperTriangular(Y);CHKERRQ(ierr); 1404 ierr = PetscMalloc1(yy->A->rmap->N,&nnz_d);CHKERRQ(ierr); 1405 ierr = PetscMalloc1(yy->B->rmap->N,&nnz_o);CHKERRQ(ierr); 1406 ierr = MatCreate(PetscObjectComm((PetscObject)Y),&B);CHKERRQ(ierr); 1407 ierr = PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);CHKERRQ(ierr); 1408 ierr = MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);CHKERRQ(ierr); 1409 ierr = MatSetBlockSizesFromMats(B,Y,Y);CHKERRQ(ierr); 1410 ierr = MatSetType(B,MATMPISBAIJ);CHKERRQ(ierr); 1411 ierr = MatAXPYGetPreallocation_SeqSBAIJ(yy->A,xx->A,nnz_d);CHKERRQ(ierr); 1412 ierr = MatAXPYGetPreallocation_MPIBAIJ(yy->B,yy->garray,xx->B,xx->garray,nnz_o);CHKERRQ(ierr); 1413 ierr = MatMPISBAIJSetPreallocation(B,bs,0,nnz_d,0,nnz_o);CHKERRQ(ierr); 1414 ierr = MatAXPY_BasicWithPreallocation(B,Y,a,X,str);CHKERRQ(ierr); 1415 ierr = MatHeaderReplace(Y,B);CHKERRQ(ierr); 1416 ierr = PetscFree(nnz_d);CHKERRQ(ierr); 1417 ierr = PetscFree(nnz_o);CHKERRQ(ierr); 1418 ierr = MatRestoreRowUpperTriangular(X);CHKERRQ(ierr); 1419 ierr = MatRestoreRowUpperTriangular(Y);CHKERRQ(ierr); 1420 } 1421 PetscFunctionReturn(0); 1422 } 1423 1424 #undef __FUNCT__ 1425 #define __FUNCT__ "MatGetSubMatrices_MPISBAIJ" 1426 PetscErrorCode MatGetSubMatrices_MPISBAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[]) 1427 { 1428 PetscErrorCode ierr; 1429 PetscInt i; 1430 PetscBool flg; 1431 1432 PetscFunctionBegin; 1433 ierr = MatGetSubMatrices_MPIBAIJ(A,n,irow,icol,scall,B);CHKERRQ(ierr); 1434 for (i=0; i<n; i++) { 1435 ierr = ISEqual(irow[i],icol[i],&flg);CHKERRQ(ierr); 1436 if (!flg) { /* *B[i] is non-symmetric, set flag */ 1437 ierr = MatSetOption(*B[i],MAT_SYMMETRIC,PETSC_FALSE);CHKERRQ(ierr); 1438 } 1439 } 1440 PetscFunctionReturn(0); 1441 } 1442 1443 /* -------------------------------------------------------------------*/ 1444 static struct _MatOps MatOps_Values = {MatSetValues_MPISBAIJ, 1445 MatGetRow_MPISBAIJ, 1446 MatRestoreRow_MPISBAIJ, 1447 MatMult_MPISBAIJ, 1448 /* 4*/ MatMultAdd_MPISBAIJ, 1449 MatMult_MPISBAIJ, /* transpose versions are same as non-transpose */ 1450 MatMultAdd_MPISBAIJ, 1451 0, 1452 0, 1453 0, 1454 /* 10*/ 0, 1455 0, 1456 0, 1457 MatSOR_MPISBAIJ, 1458 MatTranspose_MPISBAIJ, 1459 /* 15*/ MatGetInfo_MPISBAIJ, 1460 MatEqual_MPISBAIJ, 1461 MatGetDiagonal_MPISBAIJ, 1462 MatDiagonalScale_MPISBAIJ, 1463 MatNorm_MPISBAIJ, 1464 /* 20*/ MatAssemblyBegin_MPISBAIJ, 1465 MatAssemblyEnd_MPISBAIJ, 1466 MatSetOption_MPISBAIJ, 1467 MatZeroEntries_MPISBAIJ, 1468 /* 24*/ 0, 1469 0, 1470 0, 1471 0, 1472 0, 1473 /* 29*/ MatSetUp_MPISBAIJ, 1474 0, 1475 0, 1476 0, 1477 0, 1478 /* 34*/ MatDuplicate_MPISBAIJ, 1479 0, 1480 0, 1481 0, 1482 0, 1483 /* 39*/ MatAXPY_MPISBAIJ, 1484 MatGetSubMatrices_MPISBAIJ, 1485 MatIncreaseOverlap_MPISBAIJ, 1486 MatGetValues_MPISBAIJ, 1487 MatCopy_MPISBAIJ, 1488 /* 44*/ 0, 1489 MatScale_MPISBAIJ, 1490 0, 1491 0, 1492 0, 1493 /* 49*/ 0, 1494 0, 1495 0, 1496 0, 1497 0, 1498 /* 54*/ 0, 1499 0, 1500 MatSetUnfactored_MPISBAIJ, 1501 0, 1502 MatSetValuesBlocked_MPISBAIJ, 1503 /* 59*/ 0, 1504 0, 1505 0, 1506 0, 1507 0, 1508 /* 64*/ 0, 1509 0, 1510 0, 1511 0, 1512 0, 1513 /* 69*/ MatGetRowMaxAbs_MPISBAIJ, 1514 0, 1515 0, 1516 0, 1517 0, 1518 /* 74*/ 0, 1519 0, 1520 0, 1521 0, 1522 0, 1523 /* 79*/ 0, 1524 0, 1525 0, 1526 0, 1527 MatLoad_MPISBAIJ, 1528 /* 84*/ 0, 1529 0, 1530 0, 1531 0, 1532 0, 1533 /* 89*/ 0, 1534 0, 1535 0, 1536 0, 1537 0, 1538 /* 94*/ 0, 1539 0, 1540 0, 1541 0, 1542 0, 1543 /* 99*/ 0, 1544 0, 1545 0, 1546 0, 1547 0, 1548 /*104*/ 0, 1549 MatRealPart_MPISBAIJ, 1550 MatImaginaryPart_MPISBAIJ, 1551 MatGetRowUpperTriangular_MPISBAIJ, 1552 MatRestoreRowUpperTriangular_MPISBAIJ, 1553 /*109*/ 0, 1554 0, 1555 0, 1556 0, 1557 0, 1558 /*114*/ 0, 1559 0, 1560 0, 1561 0, 1562 0, 1563 /*119*/ 0, 1564 0, 1565 0, 1566 0, 1567 0, 1568 /*124*/ 0, 1569 0, 1570 0, 1571 0, 1572 0, 1573 /*129*/ 0, 1574 0, 1575 0, 1576 0, 1577 0, 1578 /*134*/ 0, 1579 0, 1580 0, 1581 0, 1582 0, 1583 /*139*/ 0, 1584 0, 1585 0, 1586 0, 1587 0, 1588 /*144*/MatCreateMPIMatConcatenateSeqMat_MPISBAIJ 1589 }; 1590 1591 #undef __FUNCT__ 1592 #define __FUNCT__ "MatGetDiagonalBlock_MPISBAIJ" 1593 PetscErrorCode MatGetDiagonalBlock_MPISBAIJ(Mat A,Mat *a) 1594 { 1595 PetscFunctionBegin; 1596 *a = ((Mat_MPISBAIJ*)A->data)->A; 1597 PetscFunctionReturn(0); 1598 } 1599 1600 #undef __FUNCT__ 1601 #define __FUNCT__ "MatMPISBAIJSetPreallocation_MPISBAIJ" 1602 PetscErrorCode MatMPISBAIJSetPreallocation_MPISBAIJ(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt *d_nnz,PetscInt o_nz,const PetscInt *o_nnz) 1603 { 1604 Mat_MPISBAIJ *b; 1605 PetscErrorCode ierr; 1606 PetscInt i,mbs,Mbs; 1607 1608 PetscFunctionBegin; 1609 ierr = MatSetBlockSize(B,PetscAbs(bs));CHKERRQ(ierr); 1610 ierr = PetscLayoutSetUp(B->rmap);CHKERRQ(ierr); 1611 ierr = PetscLayoutSetUp(B->cmap);CHKERRQ(ierr); 1612 ierr = PetscLayoutGetBlockSize(B->rmap,&bs);CHKERRQ(ierr); 1613 1614 b = (Mat_MPISBAIJ*)B->data; 1615 mbs = B->rmap->n/bs; 1616 Mbs = B->rmap->N/bs; 1617 if (mbs*bs != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"No of local rows %D must be divisible by blocksize %D",B->rmap->N,bs); 1618 1619 B->rmap->bs = bs; 1620 b->bs2 = bs*bs; 1621 b->mbs = mbs; 1622 b->Mbs = Mbs; 1623 b->nbs = B->cmap->n/bs; 1624 b->Nbs = B->cmap->N/bs; 1625 1626 for (i=0; i<=b->size; i++) { 1627 b->rangebs[i] = B->rmap->range[i]/bs; 1628 } 1629 b->rstartbs = B->rmap->rstart/bs; 1630 b->rendbs = B->rmap->rend/bs; 1631 1632 b->cstartbs = B->cmap->rstart/bs; 1633 b->cendbs = B->cmap->rend/bs; 1634 1635 if (!B->preallocated) { 1636 ierr = MatCreate(PETSC_COMM_SELF,&b->A);CHKERRQ(ierr); 1637 ierr = MatSetSizes(b->A,B->rmap->n,B->cmap->n,B->rmap->n,B->cmap->n);CHKERRQ(ierr); 1638 ierr = MatSetType(b->A,MATSEQSBAIJ);CHKERRQ(ierr); 1639 ierr = PetscLogObjectParent((PetscObject)B,(PetscObject)b->A);CHKERRQ(ierr); 1640 ierr = MatCreate(PETSC_COMM_SELF,&b->B);CHKERRQ(ierr); 1641 ierr = MatSetSizes(b->B,B->rmap->n,B->cmap->N,B->rmap->n,B->cmap->N);CHKERRQ(ierr); 1642 ierr = MatSetType(b->B,MATSEQBAIJ);CHKERRQ(ierr); 1643 ierr = PetscLogObjectParent((PetscObject)B,(PetscObject)b->B);CHKERRQ(ierr); 1644 ierr = MatStashCreate_Private(PetscObjectComm((PetscObject)B),bs,&B->bstash);CHKERRQ(ierr); 1645 } 1646 1647 ierr = MatSeqSBAIJSetPreallocation(b->A,bs,d_nz,d_nnz);CHKERRQ(ierr); 1648 ierr = MatSeqBAIJSetPreallocation(b->B,bs,o_nz,o_nnz);CHKERRQ(ierr); 1649 1650 B->preallocated = PETSC_TRUE; 1651 PetscFunctionReturn(0); 1652 } 1653 1654 #undef __FUNCT__ 1655 #define __FUNCT__ "MatMPISBAIJSetPreallocationCSR_MPISBAIJ" 1656 PetscErrorCode MatMPISBAIJSetPreallocationCSR_MPISBAIJ(Mat B,PetscInt bs,const PetscInt ii[],const PetscInt jj[],const PetscScalar V[]) 1657 { 1658 PetscInt m,rstart,cstart,cend; 1659 PetscInt i,j,d,nz,nz_max=0,*d_nnz=0,*o_nnz=0; 1660 const PetscInt *JJ =0; 1661 PetscScalar *values=0; 1662 PetscErrorCode ierr; 1663 1664 PetscFunctionBegin; 1665 if (bs < 1) SETERRQ1(PetscObjectComm((PetscObject)B),PETSC_ERR_ARG_OUTOFRANGE,"Invalid block size specified, must be positive but it is %D",bs); 1666 ierr = PetscLayoutSetBlockSize(B->rmap,bs);CHKERRQ(ierr); 1667 ierr = PetscLayoutSetBlockSize(B->cmap,bs);CHKERRQ(ierr); 1668 ierr = PetscLayoutSetUp(B->rmap);CHKERRQ(ierr); 1669 ierr = PetscLayoutSetUp(B->cmap);CHKERRQ(ierr); 1670 ierr = PetscLayoutGetBlockSize(B->rmap,&bs);CHKERRQ(ierr); 1671 m = B->rmap->n/bs; 1672 rstart = B->rmap->rstart/bs; 1673 cstart = B->cmap->rstart/bs; 1674 cend = B->cmap->rend/bs; 1675 1676 if (ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"ii[0] must be 0 but it is %D",ii[0]); 1677 ierr = PetscMalloc2(m,&d_nnz,m,&o_nnz);CHKERRQ(ierr); 1678 for (i=0; i<m; i++) { 1679 nz = ii[i+1] - ii[i]; 1680 if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Local row %D has a negative number of columns %D",i,nz); 1681 nz_max = PetscMax(nz_max,nz); 1682 JJ = jj + ii[i]; 1683 for (j=0; j<nz; j++) { 1684 if (*JJ >= cstart) break; 1685 JJ++; 1686 } 1687 d = 0; 1688 for (; j<nz; j++) { 1689 if (*JJ++ >= cend) break; 1690 d++; 1691 } 1692 d_nnz[i] = d; 1693 o_nnz[i] = nz - d; 1694 } 1695 ierr = MatMPISBAIJSetPreallocation(B,bs,0,d_nnz,0,o_nnz);CHKERRQ(ierr); 1696 ierr = PetscFree2(d_nnz,o_nnz);CHKERRQ(ierr); 1697 1698 values = (PetscScalar*)V; 1699 if (!values) { 1700 ierr = PetscMalloc1(bs*bs*nz_max,&values);CHKERRQ(ierr); 1701 ierr = PetscMemzero(values,bs*bs*nz_max*sizeof(PetscScalar));CHKERRQ(ierr); 1702 } 1703 for (i=0; i<m; i++) { 1704 PetscInt row = i + rstart; 1705 PetscInt ncols = ii[i+1] - ii[i]; 1706 const PetscInt *icols = jj + ii[i]; 1707 const PetscScalar *svals = values + (V ? (bs*bs*ii[i]) : 0); 1708 ierr = MatSetValuesBlocked_MPISBAIJ(B,1,&row,ncols,icols,svals,INSERT_VALUES);CHKERRQ(ierr); 1709 } 1710 1711 if (!V) { ierr = PetscFree(values);CHKERRQ(ierr); } 1712 ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1713 ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1714 ierr = MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);CHKERRQ(ierr); 1715 PetscFunctionReturn(0); 1716 } 1717 1718 #if defined(PETSC_HAVE_MUMPS) 1719 PETSC_EXTERN PetscErrorCode MatGetFactor_sbaij_mumps(Mat,MatFactorType,Mat*); 1720 #endif 1721 #if defined(PETSC_HAVE_PASTIX) 1722 PETSC_EXTERN PetscErrorCode MatGetFactor_mpisbaij_pastix(Mat,MatFactorType,Mat*); 1723 #endif 1724 1725 /*MC 1726 MATMPISBAIJ - MATMPISBAIJ = "mpisbaij" - A matrix type to be used for distributed symmetric sparse block matrices, 1727 based on block compressed sparse row format. Only the upper triangular portion of the "diagonal" portion of 1728 the matrix is stored. 1729 1730 For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you 1731 can call MatSetOption(Mat, MAT_HERMITIAN); 1732 1733 Options Database Keys: 1734 . -mat_type mpisbaij - sets the matrix type to "mpisbaij" during a call to MatSetFromOptions() 1735 1736 Level: beginner 1737 1738 .seealso: MatCreateMPISBAIJ 1739 M*/ 1740 1741 PETSC_EXTERN PetscErrorCode MatConvert_MPISBAIJ_MPISBSTRM(Mat,MatType,MatReuse,Mat*); 1742 1743 #undef __FUNCT__ 1744 #define __FUNCT__ "MatCreate_MPISBAIJ" 1745 PETSC_EXTERN PetscErrorCode MatCreate_MPISBAIJ(Mat B) 1746 { 1747 Mat_MPISBAIJ *b; 1748 PetscErrorCode ierr; 1749 PetscBool flg; 1750 1751 PetscFunctionBegin; 1752 ierr = PetscNewLog(B,&b);CHKERRQ(ierr); 1753 B->data = (void*)b; 1754 ierr = PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));CHKERRQ(ierr); 1755 1756 B->ops->destroy = MatDestroy_MPISBAIJ; 1757 B->ops->view = MatView_MPISBAIJ; 1758 B->assembled = PETSC_FALSE; 1759 B->insertmode = NOT_SET_VALUES; 1760 1761 ierr = MPI_Comm_rank(PetscObjectComm((PetscObject)B),&b->rank);CHKERRQ(ierr); 1762 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)B),&b->size);CHKERRQ(ierr); 1763 1764 /* build local table of row and column ownerships */ 1765 ierr = PetscMalloc1((b->size+2),&b->rangebs);CHKERRQ(ierr); 1766 1767 /* build cache for off array entries formed */ 1768 ierr = MatStashCreate_Private(PetscObjectComm((PetscObject)B),1,&B->stash);CHKERRQ(ierr); 1769 1770 b->donotstash = PETSC_FALSE; 1771 b->colmap = NULL; 1772 b->garray = NULL; 1773 b->roworiented = PETSC_TRUE; 1774 1775 /* stuff used in block assembly */ 1776 b->barray = 0; 1777 1778 /* stuff used for matrix vector multiply */ 1779 b->lvec = 0; 1780 b->Mvctx = 0; 1781 b->slvec0 = 0; 1782 b->slvec0b = 0; 1783 b->slvec1 = 0; 1784 b->slvec1a = 0; 1785 b->slvec1b = 0; 1786 b->sMvctx = 0; 1787 1788 /* stuff for MatGetRow() */ 1789 b->rowindices = 0; 1790 b->rowvalues = 0; 1791 b->getrowactive = PETSC_FALSE; 1792 1793 /* hash table stuff */ 1794 b->ht = 0; 1795 b->hd = 0; 1796 b->ht_size = 0; 1797 b->ht_flag = PETSC_FALSE; 1798 b->ht_fact = 0; 1799 b->ht_total_ct = 0; 1800 b->ht_insert_ct = 0; 1801 1802 /* stuff for MatGetSubMatrices_MPIBAIJ_local() */ 1803 b->ijonly = PETSC_FALSE; 1804 1805 b->in_loc = 0; 1806 b->v_loc = 0; 1807 b->n_loc = 0; 1808 ierr = PetscOptionsBegin(PetscObjectComm((PetscObject)B),NULL,"Options for loading MPISBAIJ matrix 1","Mat");CHKERRQ(ierr); 1809 ierr = PetscOptionsBool("-mat_use_hash_table","Use hash table to save memory in constructing matrix","MatSetOption",PETSC_FALSE,&flg,NULL);CHKERRQ(ierr); 1810 if (flg) { 1811 PetscReal fact = 1.39; 1812 ierr = MatSetOption(B,MAT_USE_HASH_TABLE,PETSC_TRUE);CHKERRQ(ierr); 1813 ierr = PetscOptionsReal("-mat_use_hash_table","Use hash table factor","MatMPIBAIJSetHashTableFactor",fact,&fact,NULL);CHKERRQ(ierr); 1814 if (fact <= 1.0) fact = 1.39; 1815 ierr = MatMPIBAIJSetHashTableFactor(B,fact);CHKERRQ(ierr); 1816 ierr = PetscInfo1(B,"Hash table Factor used %5.2f\n",fact);CHKERRQ(ierr); 1817 } 1818 ierr = PetscOptionsEnd();CHKERRQ(ierr); 1819 1820 #if defined(PETSC_HAVE_PASTIX) 1821 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_pastix_C",MatGetFactor_mpisbaij_pastix);CHKERRQ(ierr); 1822 #endif 1823 #if defined(PETSC_HAVE_MUMPS) 1824 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_mumps_C",MatGetFactor_sbaij_mumps);CHKERRQ(ierr); 1825 #endif 1826 ierr = PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_MPISBAIJ);CHKERRQ(ierr); 1827 ierr = PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_MPISBAIJ);CHKERRQ(ierr); 1828 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetDiagonalBlock_C",MatGetDiagonalBlock_MPISBAIJ);CHKERRQ(ierr); 1829 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMPISBAIJSetPreallocation_C",MatMPISBAIJSetPreallocation_MPISBAIJ);CHKERRQ(ierr); 1830 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMPISBAIJSetPreallocationCSR_C",MatMPISBAIJSetPreallocationCSR_MPISBAIJ);CHKERRQ(ierr); 1831 ierr = PetscObjectComposeFunction((PetscObject)B,"MatConvert_mpisbaij_mpisbstrm_C",MatConvert_MPISBAIJ_MPISBSTRM);CHKERRQ(ierr); 1832 1833 B->symmetric = PETSC_TRUE; 1834 B->structurally_symmetric = PETSC_TRUE; 1835 B->symmetric_set = PETSC_TRUE; 1836 B->structurally_symmetric_set = PETSC_TRUE; 1837 1838 ierr = PetscObjectChangeTypeName((PetscObject)B,MATMPISBAIJ);CHKERRQ(ierr); 1839 PetscFunctionReturn(0); 1840 } 1841 1842 /*MC 1843 MATSBAIJ - MATSBAIJ = "sbaij" - A matrix type to be used for symmetric block sparse matrices. 1844 1845 This matrix type is identical to MATSEQSBAIJ when constructed with a single process communicator, 1846 and MATMPISBAIJ otherwise. 1847 1848 Options Database Keys: 1849 . -mat_type sbaij - sets the matrix type to "sbaij" during a call to MatSetFromOptions() 1850 1851 Level: beginner 1852 1853 .seealso: MatCreateMPISBAIJ,MATSEQSBAIJ,MATMPISBAIJ 1854 M*/ 1855 1856 #undef __FUNCT__ 1857 #define __FUNCT__ "MatMPISBAIJSetPreallocation" 1858 /*@C 1859 MatMPISBAIJSetPreallocation - For good matrix assembly performance 1860 the user should preallocate the matrix storage by setting the parameters 1861 d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately, 1862 performance can be increased by more than a factor of 50. 1863 1864 Collective on Mat 1865 1866 Input Parameters: 1867 + B - the matrix 1868 . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row 1869 blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs() 1870 . d_nz - number of block nonzeros per block row in diagonal portion of local 1871 submatrix (same for all local rows) 1872 . d_nnz - array containing the number of block nonzeros in the various block rows 1873 in the upper triangular and diagonal part of the in diagonal portion of the local 1874 (possibly different for each block row) or NULL. If you plan to factor the matrix you must leave room 1875 for the diagonal entry and set a value even if it is zero. 1876 . o_nz - number of block nonzeros per block row in the off-diagonal portion of local 1877 submatrix (same for all local rows). 1878 - o_nnz - array containing the number of nonzeros in the various block rows of the 1879 off-diagonal portion of the local submatrix that is right of the diagonal 1880 (possibly different for each block row) or NULL. 1881 1882 1883 Options Database Keys: 1884 . -mat_no_unroll - uses code that does not unroll the loops in the 1885 block calculations (much slower) 1886 . -mat_block_size - size of the blocks to use 1887 1888 Notes: 1889 1890 If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor 1891 than it must be used on all processors that share the object for that argument. 1892 1893 If the *_nnz parameter is given then the *_nz parameter is ignored 1894 1895 Storage Information: 1896 For a square global matrix we define each processor's diagonal portion 1897 to be its local rows and the corresponding columns (a square submatrix); 1898 each processor's off-diagonal portion encompasses the remainder of the 1899 local matrix (a rectangular submatrix). 1900 1901 The user can specify preallocated storage for the diagonal part of 1902 the local submatrix with either d_nz or d_nnz (not both). Set 1903 d_nz=PETSC_DEFAULT and d_nnz=NULL for PETSc to control dynamic 1904 memory allocation. Likewise, specify preallocated storage for the 1905 off-diagonal part of the local submatrix with o_nz or o_nnz (not both). 1906 1907 You can call MatGetInfo() to get information on how effective the preallocation was; 1908 for example the fields mallocs,nz_allocated,nz_used,nz_unneeded; 1909 You can also run with the option -info and look for messages with the string 1910 malloc in them to see if additional memory allocation was needed. 1911 1912 Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In 1913 the figure below we depict these three local rows and all columns (0-11). 1914 1915 .vb 1916 0 1 2 3 4 5 6 7 8 9 10 11 1917 -------------------------- 1918 row 3 |. . . d d d o o o o o o 1919 row 4 |. . . d d d o o o o o o 1920 row 5 |. . . d d d o o o o o o 1921 -------------------------- 1922 .ve 1923 1924 Thus, any entries in the d locations are stored in the d (diagonal) 1925 submatrix, and any entries in the o locations are stored in the 1926 o (off-diagonal) submatrix. Note that the d matrix is stored in 1927 MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format. 1928 1929 Now d_nz should indicate the number of block nonzeros per row in the upper triangular 1930 plus the diagonal part of the d matrix, 1931 and o_nz should indicate the number of block nonzeros per row in the o matrix 1932 1933 In general, for PDE problems in which most nonzeros are near the diagonal, 1934 one expects d_nz >> o_nz. For large problems you MUST preallocate memory 1935 or you will get TERRIBLE performance; see the users' manual chapter on 1936 matrices. 1937 1938 Level: intermediate 1939 1940 .keywords: matrix, block, aij, compressed row, sparse, parallel 1941 1942 .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateBAIJ(), PetscSplitOwnership() 1943 @*/ 1944 PetscErrorCode MatMPISBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[]) 1945 { 1946 PetscErrorCode ierr; 1947 1948 PetscFunctionBegin; 1949 PetscValidHeaderSpecific(B,MAT_CLASSID,1); 1950 PetscValidType(B,1); 1951 PetscValidLogicalCollectiveInt(B,bs,2); 1952 ierr = PetscTryMethod(B,"MatMPISBAIJSetPreallocation_C",(Mat,PetscInt,PetscInt,const PetscInt[],PetscInt,const PetscInt[]),(B,bs,d_nz,d_nnz,o_nz,o_nnz));CHKERRQ(ierr); 1953 PetscFunctionReturn(0); 1954 } 1955 1956 #undef __FUNCT__ 1957 #define __FUNCT__ "MatCreateSBAIJ" 1958 /*@C 1959 MatCreateSBAIJ - Creates a sparse parallel matrix in symmetric block AIJ format 1960 (block compressed row). For good matrix assembly performance 1961 the user should preallocate the matrix storage by setting the parameters 1962 d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately, 1963 performance can be increased by more than a factor of 50. 1964 1965 Collective on MPI_Comm 1966 1967 Input Parameters: 1968 + comm - MPI communicator 1969 . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row 1970 blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs() 1971 . m - number of local rows (or PETSC_DECIDE to have calculated if M is given) 1972 This value should be the same as the local size used in creating the 1973 y vector for the matrix-vector product y = Ax. 1974 . n - number of local columns (or PETSC_DECIDE to have calculated if N is given) 1975 This value should be the same as the local size used in creating the 1976 x vector for the matrix-vector product y = Ax. 1977 . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given) 1978 . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given) 1979 . d_nz - number of block nonzeros per block row in diagonal portion of local 1980 submatrix (same for all local rows) 1981 . d_nnz - array containing the number of block nonzeros in the various block rows 1982 in the upper triangular portion of the in diagonal portion of the local 1983 (possibly different for each block block row) or NULL. 1984 If you plan to factor the matrix you must leave room for the diagonal entry and 1985 set its value even if it is zero. 1986 . o_nz - number of block nonzeros per block row in the off-diagonal portion of local 1987 submatrix (same for all local rows). 1988 - o_nnz - array containing the number of nonzeros in the various block rows of the 1989 off-diagonal portion of the local submatrix (possibly different for 1990 each block row) or NULL. 1991 1992 Output Parameter: 1993 . A - the matrix 1994 1995 Options Database Keys: 1996 . -mat_no_unroll - uses code that does not unroll the loops in the 1997 block calculations (much slower) 1998 . -mat_block_size - size of the blocks to use 1999 . -mat_mpi - use the parallel matrix data structures even on one processor 2000 (defaults to using SeqBAIJ format on one processor) 2001 2002 It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(), 2003 MatXXXXSetPreallocation() paradgm instead of this routine directly. 2004 [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation] 2005 2006 Notes: 2007 The number of rows and columns must be divisible by blocksize. 2008 This matrix type does not support complex Hermitian operation. 2009 2010 The user MUST specify either the local or global matrix dimensions 2011 (possibly both). 2012 2013 If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor 2014 than it must be used on all processors that share the object for that argument. 2015 2016 If the *_nnz parameter is given then the *_nz parameter is ignored 2017 2018 Storage Information: 2019 For a square global matrix we define each processor's diagonal portion 2020 to be its local rows and the corresponding columns (a square submatrix); 2021 each processor's off-diagonal portion encompasses the remainder of the 2022 local matrix (a rectangular submatrix). 2023 2024 The user can specify preallocated storage for the diagonal part of 2025 the local submatrix with either d_nz or d_nnz (not both). Set 2026 d_nz=PETSC_DEFAULT and d_nnz=NULL for PETSc to control dynamic 2027 memory allocation. Likewise, specify preallocated storage for the 2028 off-diagonal part of the local submatrix with o_nz or o_nnz (not both). 2029 2030 Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In 2031 the figure below we depict these three local rows and all columns (0-11). 2032 2033 .vb 2034 0 1 2 3 4 5 6 7 8 9 10 11 2035 -------------------------- 2036 row 3 |. . . d d d o o o o o o 2037 row 4 |. . . d d d o o o o o o 2038 row 5 |. . . d d d o o o o o o 2039 -------------------------- 2040 .ve 2041 2042 Thus, any entries in the d locations are stored in the d (diagonal) 2043 submatrix, and any entries in the o locations are stored in the 2044 o (off-diagonal) submatrix. Note that the d matrix is stored in 2045 MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format. 2046 2047 Now d_nz should indicate the number of block nonzeros per row in the upper triangular 2048 plus the diagonal part of the d matrix, 2049 and o_nz should indicate the number of block nonzeros per row in the o matrix. 2050 In general, for PDE problems in which most nonzeros are near the diagonal, 2051 one expects d_nz >> o_nz. For large problems you MUST preallocate memory 2052 or you will get TERRIBLE performance; see the users' manual chapter on 2053 matrices. 2054 2055 Level: intermediate 2056 2057 .keywords: matrix, block, aij, compressed row, sparse, parallel 2058 2059 .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateBAIJ() 2060 @*/ 2061 2062 PetscErrorCode MatCreateSBAIJ(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) 2063 { 2064 PetscErrorCode ierr; 2065 PetscMPIInt size; 2066 2067 PetscFunctionBegin; 2068 ierr = MatCreate(comm,A);CHKERRQ(ierr); 2069 ierr = MatSetSizes(*A,m,n,M,N);CHKERRQ(ierr); 2070 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 2071 if (size > 1) { 2072 ierr = MatSetType(*A,MATMPISBAIJ);CHKERRQ(ierr); 2073 ierr = MatMPISBAIJSetPreallocation(*A,bs,d_nz,d_nnz,o_nz,o_nnz);CHKERRQ(ierr); 2074 } else { 2075 ierr = MatSetType(*A,MATSEQSBAIJ);CHKERRQ(ierr); 2076 ierr = MatSeqSBAIJSetPreallocation(*A,bs,d_nz,d_nnz);CHKERRQ(ierr); 2077 } 2078 PetscFunctionReturn(0); 2079 } 2080 2081 2082 #undef __FUNCT__ 2083 #define __FUNCT__ "MatDuplicate_MPISBAIJ" 2084 static PetscErrorCode MatDuplicate_MPISBAIJ(Mat matin,MatDuplicateOption cpvalues,Mat *newmat) 2085 { 2086 Mat mat; 2087 Mat_MPISBAIJ *a,*oldmat = (Mat_MPISBAIJ*)matin->data; 2088 PetscErrorCode ierr; 2089 PetscInt len=0,nt,bs=matin->rmap->bs,mbs=oldmat->mbs; 2090 PetscScalar *array; 2091 2092 PetscFunctionBegin; 2093 *newmat = 0; 2094 2095 ierr = MatCreate(PetscObjectComm((PetscObject)matin),&mat);CHKERRQ(ierr); 2096 ierr = MatSetSizes(mat,matin->rmap->n,matin->cmap->n,matin->rmap->N,matin->cmap->N);CHKERRQ(ierr); 2097 ierr = MatSetType(mat,((PetscObject)matin)->type_name);CHKERRQ(ierr); 2098 ierr = PetscMemcpy(mat->ops,matin->ops,sizeof(struct _MatOps));CHKERRQ(ierr); 2099 ierr = PetscLayoutReference(matin->rmap,&mat->rmap);CHKERRQ(ierr); 2100 ierr = PetscLayoutReference(matin->cmap,&mat->cmap);CHKERRQ(ierr); 2101 2102 mat->factortype = matin->factortype; 2103 mat->preallocated = PETSC_TRUE; 2104 mat->assembled = PETSC_TRUE; 2105 mat->insertmode = NOT_SET_VALUES; 2106 2107 a = (Mat_MPISBAIJ*)mat->data; 2108 a->bs2 = oldmat->bs2; 2109 a->mbs = oldmat->mbs; 2110 a->nbs = oldmat->nbs; 2111 a->Mbs = oldmat->Mbs; 2112 a->Nbs = oldmat->Nbs; 2113 2114 2115 a->size = oldmat->size; 2116 a->rank = oldmat->rank; 2117 a->donotstash = oldmat->donotstash; 2118 a->roworiented = oldmat->roworiented; 2119 a->rowindices = 0; 2120 a->rowvalues = 0; 2121 a->getrowactive = PETSC_FALSE; 2122 a->barray = 0; 2123 a->rstartbs = oldmat->rstartbs; 2124 a->rendbs = oldmat->rendbs; 2125 a->cstartbs = oldmat->cstartbs; 2126 a->cendbs = oldmat->cendbs; 2127 2128 /* hash table stuff */ 2129 a->ht = 0; 2130 a->hd = 0; 2131 a->ht_size = 0; 2132 a->ht_flag = oldmat->ht_flag; 2133 a->ht_fact = oldmat->ht_fact; 2134 a->ht_total_ct = 0; 2135 a->ht_insert_ct = 0; 2136 2137 ierr = PetscMemcpy(a->rangebs,oldmat->rangebs,(a->size+2)*sizeof(PetscInt));CHKERRQ(ierr); 2138 if (oldmat->colmap) { 2139 #if defined(PETSC_USE_CTABLE) 2140 ierr = PetscTableCreateCopy(oldmat->colmap,&a->colmap);CHKERRQ(ierr); 2141 #else 2142 ierr = PetscMalloc1((a->Nbs),&a->colmap);CHKERRQ(ierr); 2143 ierr = PetscLogObjectMemory((PetscObject)mat,(a->Nbs)*sizeof(PetscInt));CHKERRQ(ierr); 2144 ierr = PetscMemcpy(a->colmap,oldmat->colmap,(a->Nbs)*sizeof(PetscInt));CHKERRQ(ierr); 2145 #endif 2146 } else a->colmap = 0; 2147 2148 if (oldmat->garray && (len = ((Mat_SeqBAIJ*)(oldmat->B->data))->nbs)) { 2149 ierr = PetscMalloc1(len,&a->garray);CHKERRQ(ierr); 2150 ierr = PetscLogObjectMemory((PetscObject)mat,len*sizeof(PetscInt));CHKERRQ(ierr); 2151 ierr = PetscMemcpy(a->garray,oldmat->garray,len*sizeof(PetscInt));CHKERRQ(ierr); 2152 } else a->garray = 0; 2153 2154 ierr = MatStashCreate_Private(PetscObjectComm((PetscObject)matin),matin->rmap->bs,&mat->bstash);CHKERRQ(ierr); 2155 ierr = VecDuplicate(oldmat->lvec,&a->lvec);CHKERRQ(ierr); 2156 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->lvec);CHKERRQ(ierr); 2157 ierr = VecScatterCopy(oldmat->Mvctx,&a->Mvctx);CHKERRQ(ierr); 2158 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->Mvctx);CHKERRQ(ierr); 2159 2160 ierr = VecDuplicate(oldmat->slvec0,&a->slvec0);CHKERRQ(ierr); 2161 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec0);CHKERRQ(ierr); 2162 ierr = VecDuplicate(oldmat->slvec1,&a->slvec1);CHKERRQ(ierr); 2163 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec1);CHKERRQ(ierr); 2164 2165 ierr = VecGetLocalSize(a->slvec1,&nt);CHKERRQ(ierr); 2166 ierr = VecGetArray(a->slvec1,&array);CHKERRQ(ierr); 2167 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,1,bs*mbs,array,&a->slvec1a);CHKERRQ(ierr); 2168 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,1,nt-bs*mbs,array+bs*mbs,&a->slvec1b);CHKERRQ(ierr); 2169 ierr = VecRestoreArray(a->slvec1,&array);CHKERRQ(ierr); 2170 ierr = VecGetArray(a->slvec0,&array);CHKERRQ(ierr); 2171 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,1,nt-bs*mbs,array+bs*mbs,&a->slvec0b);CHKERRQ(ierr); 2172 ierr = VecRestoreArray(a->slvec0,&array);CHKERRQ(ierr); 2173 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec0);CHKERRQ(ierr); 2174 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec1);CHKERRQ(ierr); 2175 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec0b);CHKERRQ(ierr); 2176 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec1a);CHKERRQ(ierr); 2177 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->slvec1b);CHKERRQ(ierr); 2178 2179 /* ierr = VecScatterCopy(oldmat->sMvctx,&a->sMvctx); - not written yet, replaced by the lazy trick: */ 2180 ierr = PetscObjectReference((PetscObject)oldmat->sMvctx);CHKERRQ(ierr); 2181 a->sMvctx = oldmat->sMvctx; 2182 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->sMvctx);CHKERRQ(ierr); 2183 2184 ierr = MatDuplicate(oldmat->A,cpvalues,&a->A);CHKERRQ(ierr); 2185 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->A);CHKERRQ(ierr); 2186 ierr = MatDuplicate(oldmat->B,cpvalues,&a->B);CHKERRQ(ierr); 2187 ierr = PetscLogObjectParent((PetscObject)mat,(PetscObject)a->B);CHKERRQ(ierr); 2188 ierr = PetscFunctionListDuplicate(((PetscObject)matin)->qlist,&((PetscObject)mat)->qlist);CHKERRQ(ierr); 2189 *newmat = mat; 2190 PetscFunctionReturn(0); 2191 } 2192 2193 #undef __FUNCT__ 2194 #define __FUNCT__ "MatLoad_MPISBAIJ" 2195 PetscErrorCode MatLoad_MPISBAIJ(Mat newmat,PetscViewer viewer) 2196 { 2197 PetscErrorCode ierr; 2198 PetscInt i,nz,j,rstart,rend; 2199 PetscScalar *vals,*buf; 2200 MPI_Comm comm; 2201 MPI_Status status; 2202 PetscMPIInt rank,size,tag = ((PetscObject)viewer)->tag,*sndcounts = 0,*browners,maxnz,*rowners,mmbs; 2203 PetscInt header[4],*rowlengths = 0,M,N,m,*cols,*locrowlens; 2204 PetscInt *procsnz = 0,jj,*mycols,*ibuf; 2205 PetscInt bs =1,Mbs,mbs,extra_rows; 2206 PetscInt *dlens,*odlens,*mask,*masked1,*masked2,rowcount,odcount; 2207 PetscInt dcount,kmax,k,nzcount,tmp,sizesset=1,grows,gcols; 2208 int fd; 2209 2210 PetscFunctionBegin; 2211 ierr = PetscObjectGetComm((PetscObject)viewer,&comm);CHKERRQ(ierr); 2212 ierr = PetscOptionsBegin(comm,NULL,"Options for loading MPISBAIJ matrix 2","Mat");CHKERRQ(ierr); 2213 ierr = PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,NULL);CHKERRQ(ierr); 2214 ierr = PetscOptionsEnd();CHKERRQ(ierr); 2215 2216 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 2217 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 2218 if (!rank) { 2219 ierr = PetscViewerBinaryGetDescriptor(viewer,&fd);CHKERRQ(ierr); 2220 ierr = PetscBinaryRead(fd,(char*)header,4,PETSC_INT);CHKERRQ(ierr); 2221 if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not matrix object"); 2222 if (header[3] < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as MPISBAIJ"); 2223 } 2224 2225 if (newmat->rmap->n < 0 && newmat->rmap->N < 0 && newmat->cmap->n < 0 && newmat->cmap->N < 0) sizesset = 0; 2226 2227 ierr = MPI_Bcast(header+1,3,MPIU_INT,0,comm);CHKERRQ(ierr); 2228 M = header[1]; 2229 N = header[2]; 2230 2231 /* If global rows/cols are set to PETSC_DECIDE, set it to the sizes given in the file */ 2232 if (sizesset && newmat->rmap->N < 0) newmat->rmap->N = M; 2233 if (sizesset && newmat->cmap->N < 0) newmat->cmap->N = N; 2234 2235 /* If global sizes are set, check if they are consistent with that given in the file */ 2236 if (sizesset) { 2237 ierr = MatGetSize(newmat,&grows,&gcols);CHKERRQ(ierr); 2238 } 2239 if (sizesset && newmat->rmap->N != grows) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED, "Inconsistent # of rows:Matrix in file has (%d) and input matrix has (%d)",M,grows); 2240 if (sizesset && newmat->cmap->N != gcols) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED, "Inconsistent # of cols:Matrix in file has (%d) and input matrix has (%d)",N,gcols); 2241 2242 if (M != N) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Can only do square matrices"); 2243 2244 /* 2245 This code adds extra rows to make sure the number of rows is 2246 divisible by the blocksize 2247 */ 2248 Mbs = M/bs; 2249 extra_rows = bs - M + bs*(Mbs); 2250 if (extra_rows == bs) extra_rows = 0; 2251 else Mbs++; 2252 if (extra_rows &&!rank) { 2253 ierr = PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");CHKERRQ(ierr); 2254 } 2255 2256 /* determine ownership of all rows */ 2257 if (newmat->rmap->n < 0) { /* PETSC_DECIDE */ 2258 mbs = Mbs/size + ((Mbs % size) > rank); 2259 m = mbs*bs; 2260 } else { /* User Set */ 2261 m = newmat->rmap->n; 2262 mbs = m/bs; 2263 } 2264 ierr = PetscMalloc2(size+1,&rowners,size+1,&browners);CHKERRQ(ierr); 2265 ierr = PetscMPIIntCast(mbs,&mmbs);CHKERRQ(ierr); 2266 ierr = MPI_Allgather(&mmbs,1,MPI_INT,rowners+1,1,MPI_INT,comm);CHKERRQ(ierr); 2267 rowners[0] = 0; 2268 for (i=2; i<=size; i++) rowners[i] += rowners[i-1]; 2269 for (i=0; i<=size; i++) browners[i] = rowners[i]*bs; 2270 rstart = rowners[rank]; 2271 rend = rowners[rank+1]; 2272 2273 /* distribute row lengths to all processors */ 2274 ierr = PetscMalloc1((rend-rstart)*bs,&locrowlens);CHKERRQ(ierr); 2275 if (!rank) { 2276 ierr = PetscMalloc1((M+extra_rows),&rowlengths);CHKERRQ(ierr); 2277 ierr = PetscBinaryRead(fd,rowlengths,M,PETSC_INT);CHKERRQ(ierr); 2278 for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1; 2279 ierr = PetscMalloc1(size,&sndcounts);CHKERRQ(ierr); 2280 for (i=0; i<size; i++) sndcounts[i] = browners[i+1] - browners[i]; 2281 ierr = MPI_Scatterv(rowlengths,sndcounts,browners,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);CHKERRQ(ierr); 2282 ierr = PetscFree(sndcounts);CHKERRQ(ierr); 2283 } else { 2284 ierr = MPI_Scatterv(0,0,0,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);CHKERRQ(ierr); 2285 } 2286 2287 if (!rank) { /* procs[0] */ 2288 /* calculate the number of nonzeros on each processor */ 2289 ierr = PetscMalloc1(size,&procsnz);CHKERRQ(ierr); 2290 ierr = PetscMemzero(procsnz,size*sizeof(PetscInt));CHKERRQ(ierr); 2291 for (i=0; i<size; i++) { 2292 for (j=rowners[i]*bs; j< rowners[i+1]*bs; j++) { 2293 procsnz[i] += rowlengths[j]; 2294 } 2295 } 2296 ierr = PetscFree(rowlengths);CHKERRQ(ierr); 2297 2298 /* determine max buffer needed and allocate it */ 2299 maxnz = 0; 2300 for (i=0; i<size; i++) { 2301 maxnz = PetscMax(maxnz,procsnz[i]); 2302 } 2303 ierr = PetscMalloc1(maxnz,&cols);CHKERRQ(ierr); 2304 2305 /* read in my part of the matrix column indices */ 2306 nz = procsnz[0]; 2307 ierr = PetscMalloc1(nz,&ibuf);CHKERRQ(ierr); 2308 mycols = ibuf; 2309 if (size == 1) nz -= extra_rows; 2310 ierr = PetscBinaryRead(fd,mycols,nz,PETSC_INT);CHKERRQ(ierr); 2311 if (size == 1) { 2312 for (i=0; i< extra_rows; i++) mycols[nz+i] = M+i; 2313 } 2314 2315 /* read in every ones (except the last) and ship off */ 2316 for (i=1; i<size-1; i++) { 2317 nz = procsnz[i]; 2318 ierr = PetscBinaryRead(fd,cols,nz,PETSC_INT);CHKERRQ(ierr); 2319 ierr = MPI_Send(cols,nz,MPIU_INT,i,tag,comm);CHKERRQ(ierr); 2320 } 2321 /* read in the stuff for the last proc */ 2322 if (size != 1) { 2323 nz = procsnz[size-1] - extra_rows; /* the extra rows are not on the disk */ 2324 ierr = PetscBinaryRead(fd,cols,nz,PETSC_INT);CHKERRQ(ierr); 2325 for (i=0; i<extra_rows; i++) cols[nz+i] = M+i; 2326 ierr = MPI_Send(cols,nz+extra_rows,MPIU_INT,size-1,tag,comm);CHKERRQ(ierr); 2327 } 2328 ierr = PetscFree(cols);CHKERRQ(ierr); 2329 } else { /* procs[i], i>0 */ 2330 /* determine buffer space needed for message */ 2331 nz = 0; 2332 for (i=0; i<m; i++) nz += locrowlens[i]; 2333 ierr = PetscMalloc1(nz,&ibuf);CHKERRQ(ierr); 2334 mycols = ibuf; 2335 /* receive message of column indices*/ 2336 ierr = MPI_Recv(mycols,nz,MPIU_INT,0,tag,comm,&status);CHKERRQ(ierr); 2337 ierr = MPI_Get_count(&status,MPIU_INT,&maxnz);CHKERRQ(ierr); 2338 if (maxnz != nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file"); 2339 } 2340 2341 /* loop over local rows, determining number of off diagonal entries */ 2342 ierr = PetscMalloc2(rend-rstart,&dlens,rend-rstart,&odlens);CHKERRQ(ierr); 2343 ierr = PetscMalloc3(Mbs,&mask,Mbs,&masked1,Mbs,&masked2);CHKERRQ(ierr); 2344 ierr = PetscMemzero(mask,Mbs*sizeof(PetscInt));CHKERRQ(ierr); 2345 ierr = PetscMemzero(masked1,Mbs*sizeof(PetscInt));CHKERRQ(ierr); 2346 ierr = PetscMemzero(masked2,Mbs*sizeof(PetscInt));CHKERRQ(ierr); 2347 rowcount = 0; 2348 nzcount = 0; 2349 for (i=0; i<mbs; i++) { 2350 dcount = 0; 2351 odcount = 0; 2352 for (j=0; j<bs; j++) { 2353 kmax = locrowlens[rowcount]; 2354 for (k=0; k<kmax; k++) { 2355 tmp = mycols[nzcount++]/bs; /* block col. index */ 2356 if (!mask[tmp]) { 2357 mask[tmp] = 1; 2358 if (tmp < rstart || tmp >= rend) masked2[odcount++] = tmp; /* entry in off-diag portion */ 2359 else masked1[dcount++] = tmp; /* entry in diag portion */ 2360 } 2361 } 2362 rowcount++; 2363 } 2364 2365 dlens[i] = dcount; /* d_nzz[i] */ 2366 odlens[i] = odcount; /* o_nzz[i] */ 2367 2368 /* zero out the mask elements we set */ 2369 for (j=0; j<dcount; j++) mask[masked1[j]] = 0; 2370 for (j=0; j<odcount; j++) mask[masked2[j]] = 0; 2371 } 2372 if (!sizesset) { 2373 ierr = MatSetSizes(newmat,m,m,M+extra_rows,N+extra_rows);CHKERRQ(ierr); 2374 } 2375 ierr = MatMPISBAIJSetPreallocation(newmat,bs,0,dlens,0,odlens);CHKERRQ(ierr); 2376 ierr = MatSetOption(newmat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE);CHKERRQ(ierr); 2377 2378 if (!rank) { 2379 ierr = PetscMalloc1(maxnz,&buf);CHKERRQ(ierr); 2380 /* read in my part of the matrix numerical values */ 2381 nz = procsnz[0]; 2382 vals = buf; 2383 mycols = ibuf; 2384 if (size == 1) nz -= extra_rows; 2385 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2386 if (size == 1) { 2387 for (i=0; i< extra_rows; i++) vals[nz+i] = 1.0; 2388 } 2389 2390 /* insert into matrix */ 2391 jj = rstart*bs; 2392 for (i=0; i<m; i++) { 2393 ierr = MatSetValues(newmat,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);CHKERRQ(ierr); 2394 mycols += locrowlens[i]; 2395 vals += locrowlens[i]; 2396 jj++; 2397 } 2398 2399 /* read in other processors (except the last one) and ship out */ 2400 for (i=1; i<size-1; i++) { 2401 nz = procsnz[i]; 2402 vals = buf; 2403 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2404 ierr = MPI_Send(vals,nz,MPIU_SCALAR,i,((PetscObject)newmat)->tag,comm);CHKERRQ(ierr); 2405 } 2406 /* the last proc */ 2407 if (size != 1) { 2408 nz = procsnz[i] - extra_rows; 2409 vals = buf; 2410 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2411 for (i=0; i<extra_rows; i++) vals[nz+i] = 1.0; 2412 ierr = MPI_Send(vals,nz+extra_rows,MPIU_SCALAR,size-1,((PetscObject)newmat)->tag,comm);CHKERRQ(ierr); 2413 } 2414 ierr = PetscFree(procsnz);CHKERRQ(ierr); 2415 2416 } else { 2417 /* receive numeric values */ 2418 ierr = PetscMalloc1(nz,&buf);CHKERRQ(ierr); 2419 2420 /* receive message of values*/ 2421 vals = buf; 2422 mycols = ibuf; 2423 ierr = MPI_Recv(vals,nz,MPIU_SCALAR,0,((PetscObject)newmat)->tag,comm,&status);CHKERRQ(ierr); 2424 ierr = MPI_Get_count(&status,MPIU_SCALAR,&maxnz);CHKERRQ(ierr); 2425 if (maxnz != nz) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file"); 2426 2427 /* insert into matrix */ 2428 jj = rstart*bs; 2429 for (i=0; i<m; i++) { 2430 ierr = MatSetValues_MPISBAIJ(newmat,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);CHKERRQ(ierr); 2431 mycols += locrowlens[i]; 2432 vals += locrowlens[i]; 2433 jj++; 2434 } 2435 } 2436 2437 ierr = PetscFree(locrowlens);CHKERRQ(ierr); 2438 ierr = PetscFree(buf);CHKERRQ(ierr); 2439 ierr = PetscFree(ibuf);CHKERRQ(ierr); 2440 ierr = PetscFree2(rowners,browners);CHKERRQ(ierr); 2441 ierr = PetscFree2(dlens,odlens);CHKERRQ(ierr); 2442 ierr = PetscFree3(mask,masked1,masked2);CHKERRQ(ierr); 2443 ierr = MatAssemblyBegin(newmat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2444 ierr = MatAssemblyEnd(newmat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2445 PetscFunctionReturn(0); 2446 } 2447 2448 #undef __FUNCT__ 2449 #define __FUNCT__ "MatMPISBAIJSetHashTableFactor" 2450 /*XXXXX@ 2451 MatMPISBAIJSetHashTableFactor - Sets the factor required to compute the size of the HashTable. 2452 2453 Input Parameters: 2454 . mat - the matrix 2455 . fact - factor 2456 2457 Not Collective on Mat, each process can have a different hash factor 2458 2459 Level: advanced 2460 2461 Notes: 2462 This can also be set by the command line option: -mat_use_hash_table fact 2463 2464 .keywords: matrix, hashtable, factor, HT 2465 2466 .seealso: MatSetOption() 2467 @XXXXX*/ 2468 2469 2470 #undef __FUNCT__ 2471 #define __FUNCT__ "MatGetRowMaxAbs_MPISBAIJ" 2472 PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat A,Vec v,PetscInt idx[]) 2473 { 2474 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 2475 Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(a->B)->data; 2476 PetscReal atmp; 2477 PetscReal *work,*svalues,*rvalues; 2478 PetscErrorCode ierr; 2479 PetscInt i,bs,mbs,*bi,*bj,brow,j,ncols,krow,kcol,col,row,Mbs,bcol; 2480 PetscMPIInt rank,size; 2481 PetscInt *rowners_bs,dest,count,source; 2482 PetscScalar *va; 2483 MatScalar *ba; 2484 MPI_Status stat; 2485 2486 PetscFunctionBegin; 2487 if (idx) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Send email to petsc-maint@mcs.anl.gov"); 2488 ierr = MatGetRowMaxAbs(a->A,v,NULL);CHKERRQ(ierr); 2489 ierr = VecGetArray(v,&va);CHKERRQ(ierr); 2490 2491 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)A),&size);CHKERRQ(ierr); 2492 ierr = MPI_Comm_rank(PetscObjectComm((PetscObject)A),&rank);CHKERRQ(ierr); 2493 2494 bs = A->rmap->bs; 2495 mbs = a->mbs; 2496 Mbs = a->Mbs; 2497 ba = b->a; 2498 bi = b->i; 2499 bj = b->j; 2500 2501 /* find ownerships */ 2502 rowners_bs = A->rmap->range; 2503 2504 /* each proc creates an array to be distributed */ 2505 ierr = PetscMalloc1(bs*Mbs,&work);CHKERRQ(ierr); 2506 ierr = PetscMemzero(work,bs*Mbs*sizeof(PetscReal));CHKERRQ(ierr); 2507 2508 /* row_max for B */ 2509 if (rank != size-1) { 2510 for (i=0; i<mbs; i++) { 2511 ncols = bi[1] - bi[0]; bi++; 2512 brow = bs*i; 2513 for (j=0; j<ncols; j++) { 2514 bcol = bs*(*bj); 2515 for (kcol=0; kcol<bs; kcol++) { 2516 col = bcol + kcol; /* local col index */ 2517 col += rowners_bs[rank+1]; /* global col index */ 2518 for (krow=0; krow<bs; krow++) { 2519 atmp = PetscAbsScalar(*ba); ba++; 2520 row = brow + krow; /* local row index */ 2521 if (PetscRealPart(va[row]) < atmp) va[row] = atmp; 2522 if (work[col] < atmp) work[col] = atmp; 2523 } 2524 } 2525 bj++; 2526 } 2527 } 2528 2529 /* send values to its owners */ 2530 for (dest=rank+1; dest<size; dest++) { 2531 svalues = work + rowners_bs[dest]; 2532 count = rowners_bs[dest+1]-rowners_bs[dest]; 2533 ierr = MPI_Send(svalues,count,MPIU_REAL,dest,rank,PetscObjectComm((PetscObject)A));CHKERRQ(ierr); 2534 } 2535 } 2536 2537 /* receive values */ 2538 if (rank) { 2539 rvalues = work; 2540 count = rowners_bs[rank+1]-rowners_bs[rank]; 2541 for (source=0; source<rank; source++) { 2542 ierr = MPI_Recv(rvalues,count,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PetscObjectComm((PetscObject)A),&stat);CHKERRQ(ierr); 2543 /* process values */ 2544 for (i=0; i<count; i++) { 2545 if (PetscRealPart(va[i]) < rvalues[i]) va[i] = rvalues[i]; 2546 } 2547 } 2548 } 2549 2550 ierr = VecRestoreArray(v,&va);CHKERRQ(ierr); 2551 ierr = PetscFree(work);CHKERRQ(ierr); 2552 PetscFunctionReturn(0); 2553 } 2554 2555 #undef __FUNCT__ 2556 #define __FUNCT__ "MatSOR_MPISBAIJ" 2557 PetscErrorCode MatSOR_MPISBAIJ(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx) 2558 { 2559 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 2560 PetscErrorCode ierr; 2561 PetscInt mbs=mat->mbs,bs=matin->rmap->bs; 2562 PetscScalar *x,*ptr,*from; 2563 Vec bb1; 2564 const PetscScalar *b; 2565 2566 PetscFunctionBegin; 2567 if (its <= 0 || lits <= 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits); 2568 if (bs > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented"); 2569 2570 if (flag == SOR_APPLY_UPPER) { 2571 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);CHKERRQ(ierr); 2572 PetscFunctionReturn(0); 2573 } 2574 2575 if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP) { 2576 if (flag & SOR_ZERO_INITIAL_GUESS) { 2577 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,lits,xx);CHKERRQ(ierr); 2578 its--; 2579 } 2580 2581 ierr = VecDuplicate(bb,&bb1);CHKERRQ(ierr); 2582 while (its--) { 2583 2584 /* lower triangular part: slvec0b = - B^T*xx */ 2585 ierr = (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);CHKERRQ(ierr); 2586 2587 /* copy xx into slvec0a */ 2588 ierr = VecGetArray(mat->slvec0,&ptr);CHKERRQ(ierr); 2589 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 2590 ierr = PetscMemcpy(ptr,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 2591 ierr = VecRestoreArray(mat->slvec0,&ptr);CHKERRQ(ierr); 2592 2593 ierr = VecScale(mat->slvec0,-1.0);CHKERRQ(ierr); 2594 2595 /* copy bb into slvec1a */ 2596 ierr = VecGetArray(mat->slvec1,&ptr);CHKERRQ(ierr); 2597 ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr); 2598 ierr = PetscMemcpy(ptr,b,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 2599 ierr = VecRestoreArray(mat->slvec1,&ptr);CHKERRQ(ierr); 2600 2601 /* set slvec1b = 0 */ 2602 ierr = VecSet(mat->slvec1b,0.0);CHKERRQ(ierr); 2603 2604 ierr = VecScatterBegin(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2605 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 2606 ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr); 2607 ierr = VecScatterEnd(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2608 2609 /* upper triangular part: bb1 = bb1 - B*x */ 2610 ierr = (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,bb1);CHKERRQ(ierr); 2611 2612 /* local diagonal sweep */ 2613 ierr = (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);CHKERRQ(ierr); 2614 } 2615 ierr = VecDestroy(&bb1);CHKERRQ(ierr); 2616 } else if ((flag & SOR_LOCAL_FORWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)) { 2617 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);CHKERRQ(ierr); 2618 } else if ((flag & SOR_LOCAL_BACKWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)) { 2619 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,1,xx);CHKERRQ(ierr); 2620 } else if (flag & SOR_EISENSTAT) { 2621 Vec xx1; 2622 PetscBool hasop; 2623 const PetscScalar *diag; 2624 PetscScalar *sl,scale = (omega - 2.0)/omega; 2625 PetscInt i,n; 2626 2627 if (!mat->xx1) { 2628 ierr = VecDuplicate(bb,&mat->xx1);CHKERRQ(ierr); 2629 ierr = VecDuplicate(bb,&mat->bb1);CHKERRQ(ierr); 2630 } 2631 xx1 = mat->xx1; 2632 bb1 = mat->bb1; 2633 2634 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,(MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_BACKWARD_SWEEP),fshift,lits,1,xx);CHKERRQ(ierr); 2635 2636 if (!mat->diag) { 2637 /* this is wrong for same matrix with new nonzero values */ 2638 ierr = MatCreateVecs(matin,&mat->diag,NULL);CHKERRQ(ierr); 2639 ierr = MatGetDiagonal(matin,mat->diag);CHKERRQ(ierr); 2640 } 2641 ierr = MatHasOperation(matin,MATOP_MULT_DIAGONAL_BLOCK,&hasop);CHKERRQ(ierr); 2642 2643 if (hasop) { 2644 ierr = MatMultDiagonalBlock(matin,xx,bb1);CHKERRQ(ierr); 2645 ierr = VecAYPX(mat->slvec1a,scale,bb);CHKERRQ(ierr); 2646 } else { 2647 /* 2648 These two lines are replaced by code that may be a bit faster for a good compiler 2649 ierr = VecPointwiseMult(mat->slvec1a,mat->diag,xx);CHKERRQ(ierr); 2650 ierr = VecAYPX(mat->slvec1a,scale,bb);CHKERRQ(ierr); 2651 */ 2652 ierr = VecGetArray(mat->slvec1a,&sl);CHKERRQ(ierr); 2653 ierr = VecGetArrayRead(mat->diag,&diag);CHKERRQ(ierr); 2654 ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr); 2655 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 2656 ierr = VecGetLocalSize(xx,&n);CHKERRQ(ierr); 2657 if (omega == 1.0) { 2658 for (i=0; i<n; i++) sl[i] = b[i] - diag[i]*x[i]; 2659 ierr = PetscLogFlops(2.0*n);CHKERRQ(ierr); 2660 } else { 2661 for (i=0; i<n; i++) sl[i] = b[i] + scale*diag[i]*x[i]; 2662 ierr = PetscLogFlops(3.0*n);CHKERRQ(ierr); 2663 } 2664 ierr = VecRestoreArray(mat->slvec1a,&sl);CHKERRQ(ierr); 2665 ierr = VecRestoreArrayRead(mat->diag,&diag);CHKERRQ(ierr); 2666 ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr); 2667 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 2668 } 2669 2670 /* multiply off-diagonal portion of matrix */ 2671 ierr = VecSet(mat->slvec1b,0.0);CHKERRQ(ierr); 2672 ierr = (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);CHKERRQ(ierr); 2673 ierr = VecGetArray(mat->slvec0,&from);CHKERRQ(ierr); 2674 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 2675 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 2676 ierr = VecRestoreArray(mat->slvec0,&from);CHKERRQ(ierr); 2677 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 2678 ierr = VecScatterBegin(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2679 ierr = VecScatterEnd(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2680 ierr = (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,mat->slvec1a);CHKERRQ(ierr); 2681 2682 /* local sweep */ 2683 ierr = (*mat->A->ops->sor)(mat->A,mat->slvec1a,omega,(MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_FORWARD_SWEEP),fshift,lits,1,xx1);CHKERRQ(ierr); 2684 ierr = VecAXPY(xx,1.0,xx1);CHKERRQ(ierr); 2685 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format"); 2686 PetscFunctionReturn(0); 2687 } 2688 2689 #undef __FUNCT__ 2690 #define __FUNCT__ "MatSOR_MPISBAIJ_2comm" 2691 PetscErrorCode MatSOR_MPISBAIJ_2comm(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx) 2692 { 2693 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 2694 PetscErrorCode ierr; 2695 Vec lvec1,bb1; 2696 2697 PetscFunctionBegin; 2698 if (its <= 0 || lits <= 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits); 2699 if (matin->rmap->bs > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented"); 2700 2701 if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP) { 2702 if (flag & SOR_ZERO_INITIAL_GUESS) { 2703 ierr = (*mat->A->ops->sor)(mat->A,bb,omega,flag,fshift,lits,lits,xx);CHKERRQ(ierr); 2704 its--; 2705 } 2706 2707 ierr = VecDuplicate(mat->lvec,&lvec1);CHKERRQ(ierr); 2708 ierr = VecDuplicate(bb,&bb1);CHKERRQ(ierr); 2709 while (its--) { 2710 ierr = VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2711 2712 /* lower diagonal part: bb1 = bb - B^T*xx */ 2713 ierr = (*mat->B->ops->multtranspose)(mat->B,xx,lvec1);CHKERRQ(ierr); 2714 ierr = VecScale(lvec1,-1.0);CHKERRQ(ierr); 2715 2716 ierr = VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2717 ierr = VecCopy(bb,bb1);CHKERRQ(ierr); 2718 ierr = VecScatterBegin(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 2719 2720 /* upper diagonal part: bb1 = bb1 - B*x */ 2721 ierr = VecScale(mat->lvec,-1.0);CHKERRQ(ierr); 2722 ierr = (*mat->B->ops->multadd)(mat->B,mat->lvec,bb1,bb1);CHKERRQ(ierr); 2723 2724 ierr = VecScatterEnd(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 2725 2726 /* diagonal sweep */ 2727 ierr = (*mat->A->ops->sor)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);CHKERRQ(ierr); 2728 } 2729 ierr = VecDestroy(&lvec1);CHKERRQ(ierr); 2730 ierr = VecDestroy(&bb1);CHKERRQ(ierr); 2731 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format"); 2732 PetscFunctionReturn(0); 2733 } 2734 2735 #undef __FUNCT__ 2736 #define __FUNCT__ "MatCreateMPISBAIJWithArrays" 2737 /*@ 2738 MatCreateMPISBAIJWithArrays - creates a MPI SBAIJ matrix using arrays that contain in standard 2739 CSR format the local rows. 2740 2741 Collective on MPI_Comm 2742 2743 Input Parameters: 2744 + comm - MPI communicator 2745 . bs - the block size, only a block size of 1 is supported 2746 . m - number of local rows (Cannot be PETSC_DECIDE) 2747 . n - This value should be the same as the local size used in creating the 2748 x vector for the matrix-vector product y = Ax. (or PETSC_DECIDE to have 2749 calculated if N is given) For square matrices n is almost always m. 2750 . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given) 2751 . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given) 2752 . i - row indices 2753 . j - column indices 2754 - a - matrix values 2755 2756 Output Parameter: 2757 . mat - the matrix 2758 2759 Level: intermediate 2760 2761 Notes: 2762 The i, j, and a arrays ARE copied by this routine into the internal format used by PETSc; 2763 thus you CANNOT change the matrix entries by changing the values of a[] after you have 2764 called this routine. Use MatCreateMPIAIJWithSplitArrays() to avoid needing to copy the arrays. 2765 2766 The i and j indices are 0 based, and i indices are indices corresponding to the local j array. 2767 2768 .keywords: matrix, aij, compressed row, sparse, parallel 2769 2770 .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatMPIAIJSetPreallocation(), MatMPIAIJSetPreallocationCSR(), 2771 MPIAIJ, MatCreateAIJ(), MatCreateMPIAIJWithSplitArrays() 2772 @*/ 2773 PetscErrorCode MatCreateMPISBAIJWithArrays(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt M,PetscInt N,const PetscInt i[],const PetscInt j[],const PetscScalar a[],Mat *mat) 2774 { 2775 PetscErrorCode ierr; 2776 2777 2778 PetscFunctionBegin; 2779 if (i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0"); 2780 if (m < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"local number of rows (m) cannot be PETSC_DECIDE, or negative"); 2781 ierr = MatCreate(comm,mat);CHKERRQ(ierr); 2782 ierr = MatSetSizes(*mat,m,n,M,N);CHKERRQ(ierr); 2783 ierr = MatSetType(*mat,MATMPISBAIJ);CHKERRQ(ierr); 2784 ierr = MatMPISBAIJSetPreallocationCSR(*mat,bs,i,j,a);CHKERRQ(ierr); 2785 PetscFunctionReturn(0); 2786 } 2787 2788 2789 #undef __FUNCT__ 2790 #define __FUNCT__ "MatMPISBAIJSetPreallocationCSR" 2791 /*@C 2792 MatMPISBAIJSetPreallocationCSR - Allocates memory for a sparse parallel matrix in BAIJ format 2793 (the default parallel PETSc format). 2794 2795 Collective on MPI_Comm 2796 2797 Input Parameters: 2798 + B - the matrix 2799 . bs - the block size 2800 . i - the indices into j for the start of each local row (starts with zero) 2801 . j - the column indices for each local row (starts with zero) these must be sorted for each row 2802 - v - optional values in the matrix 2803 2804 Level: developer 2805 2806 .keywords: matrix, aij, compressed row, sparse, parallel 2807 2808 .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatMPIBAIJSetPreallocation(), MatCreateAIJ(), MPIAIJ 2809 @*/ 2810 PetscErrorCode MatMPISBAIJSetPreallocationCSR(Mat B,PetscInt bs,const PetscInt i[],const PetscInt j[], const PetscScalar v[]) 2811 { 2812 PetscErrorCode ierr; 2813 2814 PetscFunctionBegin; 2815 ierr = PetscTryMethod(B,"MatMPISBAIJSetPreallocationCSR_C",(Mat,PetscInt,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,bs,i,j,v));CHKERRQ(ierr); 2816 PetscFunctionReturn(0); 2817 } 2818 2819 #undef __FUNCT__ 2820 #define __FUNCT__ "MatCreateMPIMatConcatenateSeqMat_MPISBAIJ" 2821 PetscErrorCode MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(MPI_Comm comm,Mat inmat,PetscInt n,MatReuse scall,Mat *outmat) 2822 { 2823 PetscErrorCode ierr; 2824 PetscInt m,N,i,rstart,nnz,Ii,bs,cbs; 2825 PetscInt *indx; 2826 PetscScalar *values; 2827 2828 PetscFunctionBegin; 2829 ierr = MatGetSize(inmat,&m,&N);CHKERRQ(ierr); 2830 if (scall == MAT_INITIAL_MATRIX) { /* symbolic phase */ 2831 Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)inmat->data; 2832 PetscInt *dnz,*onz,sum,bs,cbs,mbs,Nbs; 2833 PetscInt *bindx,rmax=a->rmax,j; 2834 2835 ierr = MatGetBlockSizes(inmat,&bs,&cbs);CHKERRQ(ierr); 2836 mbs = m/bs; Nbs = N/cbs; 2837 if (n == PETSC_DECIDE) { 2838 ierr = PetscSplitOwnership(comm,&n,&Nbs);CHKERRQ(ierr); 2839 } 2840 /* Check sum(n) = Nbs */ 2841 ierr = MPI_Allreduce(&n,&sum,1,MPIU_INT,MPI_SUM,comm);CHKERRQ(ierr); 2842 if (sum != Nbs) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Sum of local columns != global columns %d",Nbs); 2843 2844 ierr = MPI_Scan(&mbs, &rstart,1,MPIU_INT,MPI_SUM,comm);CHKERRQ(ierr); 2845 rstart -= mbs; 2846 2847 ierr = PetscMalloc1(rmax,&bindx);CHKERRQ(ierr); 2848 ierr = MatPreallocateInitialize(comm,mbs,n,dnz,onz);CHKERRQ(ierr); 2849 ierr = MatSetOption(inmat,MAT_GETROW_UPPERTRIANGULAR,PETSC_TRUE);CHKERRQ(ierr); 2850 for (i=0; i<mbs; i++) { 2851 ierr = MatGetRow_SeqSBAIJ(inmat,i*bs,&nnz,&indx,NULL);CHKERRQ(ierr); /* non-blocked nnz and indx */ 2852 nnz = nnz/bs; 2853 for (j=0; j<nnz; j++) bindx[j] = indx[j*bs]/bs; 2854 ierr = MatPreallocateSet(i+rstart,nnz,bindx,dnz,onz);CHKERRQ(ierr); 2855 ierr = MatRestoreRow_SeqSBAIJ(inmat,i*bs,&nnz,&indx,NULL);CHKERRQ(ierr); 2856 } 2857 ierr = MatSetOption(inmat,MAT_GETROW_UPPERTRIANGULAR,PETSC_FALSE);CHKERRQ(ierr); 2858 ierr = PetscFree(bindx);CHKERRQ(ierr); 2859 2860 ierr = MatCreate(comm,outmat);CHKERRQ(ierr); 2861 ierr = MatSetSizes(*outmat,m,n*bs,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 2862 ierr = MatSetBlockSizes(*outmat,bs,cbs);CHKERRQ(ierr); 2863 ierr = MatSetType(*outmat,MATMPISBAIJ);CHKERRQ(ierr); 2864 ierr = MatMPISBAIJSetPreallocation(*outmat,bs,0,dnz,0,onz);CHKERRQ(ierr); 2865 ierr = MatPreallocateFinalize(dnz,onz);CHKERRQ(ierr); 2866 } 2867 2868 /* numeric phase */ 2869 ierr = MatGetBlockSizes(inmat,&bs,&cbs);CHKERRQ(ierr); 2870 ierr = MatGetOwnershipRange(*outmat,&rstart,NULL);CHKERRQ(ierr); 2871 2872 ierr = MatSetOption(inmat,MAT_GETROW_UPPERTRIANGULAR,PETSC_TRUE);CHKERRQ(ierr); 2873 for (i=0; i<m; i++) { 2874 ierr = MatGetRow_SeqSBAIJ(inmat,i,&nnz,&indx,&values);CHKERRQ(ierr); 2875 Ii = i + rstart; 2876 ierr = MatSetValues(*outmat,1,&Ii,nnz,indx,values,INSERT_VALUES);CHKERRQ(ierr); 2877 ierr = MatRestoreRow_SeqSBAIJ(inmat,i,&nnz,&indx,&values);CHKERRQ(ierr); 2878 } 2879 ierr = MatSetOption(inmat,MAT_GETROW_UPPERTRIANGULAR,PETSC_FALSE);CHKERRQ(ierr); 2880 ierr = MatAssemblyBegin(*outmat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2881 ierr = MatAssemblyEnd(*outmat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2882 PetscFunctionReturn(0); 2883 } 2884