1 2 /* 3 Defines the basic matrix operations for the AIJ (compressed row) 4 matrix storage format. 5 */ 6 7 8 #include <../src/mat/impls/aij/seq/aij.h> /*I "petscmat.h" I*/ 9 #include <petscblaslapack.h> 10 #include <petscbt.h> 11 #include <petsc-private/kernels/blocktranspose.h> 12 #if defined(PETSC_THREADCOMM_ACTIVE) 13 #include <petscthreadcomm.h> 14 #endif 15 16 #undef __FUNCT__ 17 #define __FUNCT__ "MatGetColumnNorms_SeqAIJ" 18 PetscErrorCode MatGetColumnNorms_SeqAIJ(Mat A,NormType type,PetscReal *norms) 19 { 20 PetscErrorCode ierr; 21 PetscInt i,m,n; 22 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)A->data; 23 24 PetscFunctionBegin; 25 ierr = MatGetSize(A,&m,&n);CHKERRQ(ierr); 26 ierr = PetscMemzero(norms,n*sizeof(PetscReal));CHKERRQ(ierr); 27 if (type == NORM_2) { 28 for (i=0; i<aij->i[m]; i++) { 29 norms[aij->j[i]] += PetscAbsScalar(aij->a[i]*aij->a[i]); 30 } 31 } else if (type == NORM_1) { 32 for (i=0; i<aij->i[m]; i++) { 33 norms[aij->j[i]] += PetscAbsScalar(aij->a[i]); 34 } 35 } else if (type == NORM_INFINITY) { 36 for (i=0; i<aij->i[m]; i++) { 37 norms[aij->j[i]] = PetscMax(PetscAbsScalar(aij->a[i]),norms[aij->j[i]]); 38 } 39 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONG,"Unknown NormType"); 40 41 if (type == NORM_2) { 42 for (i=0; i<n; i++) norms[i] = PetscSqrtReal(norms[i]); 43 } 44 PetscFunctionReturn(0); 45 } 46 47 #undef __FUNCT__ 48 #define __FUNCT__ "MatFindZeroDiagonals_SeqAIJ_Private" 49 PetscErrorCode MatFindZeroDiagonals_SeqAIJ_Private(Mat A,PetscInt *nrows,PetscInt **zrows) 50 { 51 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 52 const MatScalar *aa = a->a; 53 PetscInt i,m=A->rmap->n,cnt = 0; 54 const PetscInt *jj = a->j,*diag; 55 PetscInt *rows; 56 PetscErrorCode ierr; 57 58 PetscFunctionBegin; 59 ierr = MatMarkDiagonal_SeqAIJ(A);CHKERRQ(ierr); 60 diag = a->diag; 61 for (i=0; i<m; i++) { 62 if ((jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) { 63 cnt++; 64 } 65 } 66 ierr = PetscMalloc(cnt*sizeof(PetscInt),&rows);CHKERRQ(ierr); 67 cnt = 0; 68 for (i=0; i<m; i++) { 69 if ((jj[diag[i]] != i) || (aa[diag[i]] == 0.0)) { 70 rows[cnt++] = i; 71 } 72 } 73 *nrows = cnt; 74 *zrows = rows; 75 PetscFunctionReturn(0); 76 } 77 78 #undef __FUNCT__ 79 #define __FUNCT__ "MatFindZeroDiagonals_SeqAIJ" 80 PetscErrorCode MatFindZeroDiagonals_SeqAIJ(Mat A,IS *zrows) 81 { 82 PetscInt nrows,*rows; 83 PetscErrorCode ierr; 84 85 PetscFunctionBegin; 86 *zrows = NULL; 87 ierr = MatFindZeroDiagonals_SeqAIJ_Private(A,&nrows,&rows);CHKERRQ(ierr); 88 ierr = ISCreateGeneral(PetscObjectComm((PetscObject)A),nrows,rows,PETSC_OWN_POINTER,zrows);CHKERRQ(ierr); 89 PetscFunctionReturn(0); 90 } 91 92 #undef __FUNCT__ 93 #define __FUNCT__ "MatFindNonzeroRows_SeqAIJ" 94 PetscErrorCode MatFindNonzeroRows_SeqAIJ(Mat A,IS *keptrows) 95 { 96 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 97 const MatScalar *aa; 98 PetscInt m=A->rmap->n,cnt = 0; 99 const PetscInt *ii; 100 PetscInt n,i,j,*rows; 101 PetscErrorCode ierr; 102 103 PetscFunctionBegin; 104 *keptrows = 0; 105 ii = a->i; 106 for (i=0; i<m; i++) { 107 n = ii[i+1] - ii[i]; 108 if (!n) { 109 cnt++; 110 goto ok1; 111 } 112 aa = a->a + ii[i]; 113 for (j=0; j<n; j++) { 114 if (aa[j] != 0.0) goto ok1; 115 } 116 cnt++; 117 ok1:; 118 } 119 if (!cnt) PetscFunctionReturn(0); 120 ierr = PetscMalloc((A->rmap->n-cnt)*sizeof(PetscInt),&rows);CHKERRQ(ierr); 121 cnt = 0; 122 for (i=0; i<m; i++) { 123 n = ii[i+1] - ii[i]; 124 if (!n) continue; 125 aa = a->a + ii[i]; 126 for (j=0; j<n; j++) { 127 if (aa[j] != 0.0) { 128 rows[cnt++] = i; 129 break; 130 } 131 } 132 } 133 ierr = ISCreateGeneral(PETSC_COMM_SELF,cnt,rows,PETSC_OWN_POINTER,keptrows);CHKERRQ(ierr); 134 PetscFunctionReturn(0); 135 } 136 137 #undef __FUNCT__ 138 #define __FUNCT__ "MatDiagonalSet_SeqAIJ" 139 PetscErrorCode MatDiagonalSet_SeqAIJ(Mat Y,Vec D,InsertMode is) 140 { 141 PetscErrorCode ierr; 142 Mat_SeqAIJ *aij = (Mat_SeqAIJ*) Y->data; 143 PetscInt i,*diag, m = Y->rmap->n; 144 MatScalar *aa = aij->a; 145 PetscScalar *v; 146 PetscBool missing; 147 148 PetscFunctionBegin; 149 if (Y->assembled) { 150 ierr = MatMissingDiagonal_SeqAIJ(Y,&missing,NULL);CHKERRQ(ierr); 151 if (!missing) { 152 diag = aij->diag; 153 ierr = VecGetArray(D,&v);CHKERRQ(ierr); 154 if (is == INSERT_VALUES) { 155 for (i=0; i<m; i++) { 156 aa[diag[i]] = v[i]; 157 } 158 } else { 159 for (i=0; i<m; i++) { 160 aa[diag[i]] += v[i]; 161 } 162 } 163 ierr = VecRestoreArray(D,&v);CHKERRQ(ierr); 164 PetscFunctionReturn(0); 165 } 166 ierr = MatSeqAIJInvalidateDiagonal(Y);CHKERRQ(ierr); 167 } 168 ierr = MatDiagonalSet_Default(Y,D,is);CHKERRQ(ierr); 169 PetscFunctionReturn(0); 170 } 171 172 #undef __FUNCT__ 173 #define __FUNCT__ "MatGetRowIJ_SeqAIJ" 174 PetscErrorCode MatGetRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *m,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 175 { 176 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 177 PetscErrorCode ierr; 178 PetscInt i,ishift; 179 180 PetscFunctionBegin; 181 *m = A->rmap->n; 182 if (!ia) PetscFunctionReturn(0); 183 ishift = 0; 184 if (symmetric && !A->structurally_symmetric) { 185 ierr = MatToSymmetricIJ_SeqAIJ(A->rmap->n,a->i,a->j,ishift,oshift,(PetscInt**)ia,(PetscInt**)ja);CHKERRQ(ierr); 186 } else if (oshift == 1) { 187 PetscInt *tia; 188 PetscInt nz = a->i[A->rmap->n]; 189 /* malloc space and add 1 to i and j indices */ 190 ierr = PetscMalloc((A->rmap->n+1)*sizeof(PetscInt),&tia);CHKERRQ(ierr); 191 for (i=0; i<A->rmap->n+1; i++) tia[i] = a->i[i] + 1; 192 *ia = tia; 193 if (ja) { 194 PetscInt *tja; 195 ierr = PetscMalloc((nz+1)*sizeof(PetscInt),&tja);CHKERRQ(ierr); 196 for (i=0; i<nz; i++) tja[i] = a->j[i] + 1; 197 *ja = tja; 198 } 199 } else { 200 *ia = a->i; 201 if (ja) *ja = a->j; 202 } 203 PetscFunctionReturn(0); 204 } 205 206 #undef __FUNCT__ 207 #define __FUNCT__ "MatRestoreRowIJ_SeqAIJ" 208 PetscErrorCode MatRestoreRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 209 { 210 PetscErrorCode ierr; 211 212 PetscFunctionBegin; 213 if (!ia) PetscFunctionReturn(0); 214 if ((symmetric && !A->structurally_symmetric) || oshift == 1) { 215 ierr = PetscFree(*ia);CHKERRQ(ierr); 216 if (ja) {ierr = PetscFree(*ja);CHKERRQ(ierr);} 217 } 218 PetscFunctionReturn(0); 219 } 220 221 #undef __FUNCT__ 222 #define __FUNCT__ "MatGetColumnIJ_SeqAIJ" 223 PetscErrorCode MatGetColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 224 { 225 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 226 PetscErrorCode ierr; 227 PetscInt i,*collengths,*cia,*cja,n = A->cmap->n,m = A->rmap->n; 228 PetscInt nz = a->i[m],row,*jj,mr,col; 229 230 PetscFunctionBegin; 231 *nn = n; 232 if (!ia) PetscFunctionReturn(0); 233 if (symmetric) { 234 ierr = MatToSymmetricIJ_SeqAIJ(A->rmap->n,a->i,a->j,0,oshift,(PetscInt**)ia,(PetscInt**)ja);CHKERRQ(ierr); 235 } else { 236 ierr = PetscMalloc((n+1)*sizeof(PetscInt),&collengths);CHKERRQ(ierr); 237 ierr = PetscMemzero(collengths,n*sizeof(PetscInt));CHKERRQ(ierr); 238 ierr = PetscMalloc((n+1)*sizeof(PetscInt),&cia);CHKERRQ(ierr); 239 ierr = PetscMalloc((nz+1)*sizeof(PetscInt),&cja);CHKERRQ(ierr); 240 jj = a->j; 241 for (i=0; i<nz; i++) { 242 collengths[jj[i]]++; 243 } 244 cia[0] = oshift; 245 for (i=0; i<n; i++) { 246 cia[i+1] = cia[i] + collengths[i]; 247 } 248 ierr = PetscMemzero(collengths,n*sizeof(PetscInt));CHKERRQ(ierr); 249 jj = a->j; 250 for (row=0; row<m; row++) { 251 mr = a->i[row+1] - a->i[row]; 252 for (i=0; i<mr; i++) { 253 col = *jj++; 254 255 cja[cia[col] + collengths[col]++ - oshift] = row + oshift; 256 } 257 } 258 ierr = PetscFree(collengths);CHKERRQ(ierr); 259 *ia = cia; *ja = cja; 260 } 261 PetscFunctionReturn(0); 262 } 263 264 #undef __FUNCT__ 265 #define __FUNCT__ "MatRestoreColumnIJ_SeqAIJ" 266 PetscErrorCode MatRestoreColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscBool *done) 267 { 268 PetscErrorCode ierr; 269 270 PetscFunctionBegin; 271 if (!ia) PetscFunctionReturn(0); 272 273 ierr = PetscFree(*ia);CHKERRQ(ierr); 274 ierr = PetscFree(*ja);CHKERRQ(ierr); 275 PetscFunctionReturn(0); 276 } 277 278 /* 279 MatGetColumnIJ_SeqAIJ_Color() and MatRestoreColumnIJ_SeqAIJ_Color() are customized from 280 MatGetColumnIJ_SeqAIJ() and MatRestoreColumnIJ_SeqAIJ() by adding an output 281 spidx[], index of a->a, to be used in MatTransposeColoringCreate_SeqAIJ() and MatFDColoringCreate_SeqAIJ() 282 */ 283 #undef __FUNCT__ 284 #define __FUNCT__ "MatGetColumnIJ_SeqAIJ_Color" 285 PetscErrorCode MatGetColumnIJ_SeqAIJ_Color(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *nn,const PetscInt *ia[],const PetscInt *ja[],PetscInt *spidx[],PetscBool *done) 286 { 287 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 288 PetscErrorCode ierr; 289 PetscInt i,*collengths,*cia,*cja,n = A->cmap->n,m = A->rmap->n; 290 PetscInt nz = a->i[m],row,*jj,mr,col; 291 PetscInt *cspidx; 292 293 PetscFunctionBegin; 294 *nn = n; 295 if (!ia) PetscFunctionReturn(0); 296 if (symmetric) { 297 SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_SUP,"MatGetColumnIJ_SeqAIJ_Color() not supported for the case symmetric"); 298 ierr = MatToSymmetricIJ_SeqAIJ(A->rmap->n,a->i,a->j,0,oshift,(PetscInt**)ia,(PetscInt**)ja);CHKERRQ(ierr); 299 } else { 300 ierr = PetscMalloc((n+1)*sizeof(PetscInt),&collengths);CHKERRQ(ierr); 301 ierr = PetscMemzero(collengths,n*sizeof(PetscInt));CHKERRQ(ierr); 302 ierr = PetscMalloc((n+1)*sizeof(PetscInt),&cia);CHKERRQ(ierr); 303 ierr = PetscMalloc((nz+1)*sizeof(PetscInt),&cja);CHKERRQ(ierr); 304 ierr = PetscMalloc((nz+1)*sizeof(PetscInt),&cspidx);CHKERRQ(ierr); 305 jj = a->j; 306 for (i=0; i<nz; i++) { 307 collengths[jj[i]]++; 308 } 309 cia[0] = oshift; 310 for (i=0; i<n; i++) { 311 cia[i+1] = cia[i] + collengths[i]; 312 } 313 ierr = PetscMemzero(collengths,n*sizeof(PetscInt));CHKERRQ(ierr); 314 jj = a->j; 315 for (row=0; row<m; row++) { 316 mr = a->i[row+1] - a->i[row]; 317 for (i=0; i<mr; i++) { 318 col = *jj++; 319 320 cspidx[cia[col] + collengths[col] - oshift] = a->i[row] + i; /* index of a->j */ 321 cja[cia[col] + collengths[col]++ - oshift] = row + oshift; 322 } 323 } 324 ierr = PetscFree(collengths);CHKERRQ(ierr); 325 *ia = cia; *ja = cja; 326 *spidx = cspidx; 327 } 328 PetscFunctionReturn(0); 329 } 330 331 #undef __FUNCT__ 332 #define __FUNCT__ "MatRestoreColumnIJ_SeqAIJ_Color" 333 PetscErrorCode MatRestoreColumnIJ_SeqAIJ_Color(Mat A,PetscInt oshift,PetscBool symmetric,PetscBool inodecompressed,PetscInt *n,const PetscInt *ia[],const PetscInt *ja[],PetscInt *spidx[],PetscBool *done) 334 { 335 PetscErrorCode ierr; 336 337 PetscFunctionBegin; 338 if (!ia) PetscFunctionReturn(0); 339 340 ierr = PetscFree(*ia);CHKERRQ(ierr); 341 ierr = PetscFree(*ja);CHKERRQ(ierr); 342 ierr = PetscFree(*spidx);CHKERRQ(ierr); 343 PetscFunctionReturn(0); 344 } 345 346 #undef __FUNCT__ 347 #define __FUNCT__ "MatSetValuesRow_SeqAIJ" 348 PetscErrorCode MatSetValuesRow_SeqAIJ(Mat A,PetscInt row,const PetscScalar v[]) 349 { 350 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 351 PetscInt *ai = a->i; 352 PetscErrorCode ierr; 353 354 PetscFunctionBegin; 355 ierr = PetscMemcpy(a->a+ai[row],v,(ai[row+1]-ai[row])*sizeof(PetscScalar));CHKERRQ(ierr); 356 PetscFunctionReturn(0); 357 } 358 359 #undef __FUNCT__ 360 #define __FUNCT__ "MatSetValues_SeqAIJ" 361 PetscErrorCode MatSetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is) 362 { 363 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 364 PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N; 365 PetscInt *imax = a->imax,*ai = a->i,*ailen = a->ilen; 366 PetscErrorCode ierr; 367 PetscInt *aj = a->j,nonew = a->nonew,lastcol = -1; 368 MatScalar *ap,value,*aa = a->a; 369 PetscBool ignorezeroentries = a->ignorezeroentries; 370 PetscBool roworiented = a->roworiented; 371 372 PetscFunctionBegin; 373 if (v) PetscValidScalarPointer(v,6); 374 for (k=0; k<m; k++) { /* loop over added rows */ 375 row = im[k]; 376 if (row < 0) continue; 377 #if defined(PETSC_USE_DEBUG) 378 if (row >= A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap->n-1); 379 #endif 380 rp = aj + ai[row]; ap = aa + ai[row]; 381 rmax = imax[row]; nrow = ailen[row]; 382 low = 0; 383 high = nrow; 384 for (l=0; l<n; l++) { /* loop over added columns */ 385 if (in[l] < 0) continue; 386 #if defined(PETSC_USE_DEBUG) 387 if (in[l] >= A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap->n-1); 388 #endif 389 col = in[l]; 390 if (v) { 391 if (roworiented) { 392 value = v[l + k*n]; 393 } else { 394 value = v[k + l*m]; 395 } 396 } else { 397 value = 0.; 398 } 399 if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue; 400 401 if (col <= lastcol) low = 0; 402 else high = nrow; 403 lastcol = col; 404 while (high-low > 5) { 405 t = (low+high)/2; 406 if (rp[t] > col) high = t; 407 else low = t; 408 } 409 for (i=low; i<high; i++) { 410 if (rp[i] > col) break; 411 if (rp[i] == col) { 412 if (is == ADD_VALUES) ap[i] += value; 413 else ap[i] = value; 414 low = i + 1; 415 goto noinsert; 416 } 417 } 418 if (value == 0.0 && ignorezeroentries) goto noinsert; 419 if (nonew == 1) goto noinsert; 420 if (nonew == -1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero at (%D,%D) in the matrix",row,col); 421 MatSeqXAIJReallocateAIJ(A,A->rmap->n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar); 422 N = nrow++ - 1; a->nz++; high++; 423 /* shift up all the later entries in this row */ 424 for (ii=N; ii>=i; ii--) { 425 rp[ii+1] = rp[ii]; 426 ap[ii+1] = ap[ii]; 427 } 428 rp[i] = col; 429 ap[i] = value; 430 low = i + 1; 431 noinsert:; 432 } 433 ailen[row] = nrow; 434 } 435 A->same_nonzero = PETSC_FALSE; 436 PetscFunctionReturn(0); 437 } 438 439 440 #undef __FUNCT__ 441 #define __FUNCT__ "MatGetValues_SeqAIJ" 442 PetscErrorCode MatGetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],PetscScalar v[]) 443 { 444 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 445 PetscInt *rp,k,low,high,t,row,nrow,i,col,l,*aj = a->j; 446 PetscInt *ai = a->i,*ailen = a->ilen; 447 MatScalar *ap,*aa = a->a; 448 449 PetscFunctionBegin; 450 for (k=0; k<m; k++) { /* loop over rows */ 451 row = im[k]; 452 if (row < 0) {v += n; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",row); */ 453 if (row >= A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap->n-1); 454 rp = aj + ai[row]; ap = aa + ai[row]; 455 nrow = ailen[row]; 456 for (l=0; l<n; l++) { /* loop over columns */ 457 if (in[l] < 0) {v++; continue;} /* SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",in[l]); */ 458 if (in[l] >= A->cmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap->n-1); 459 col = in[l]; 460 high = nrow; low = 0; /* assume unsorted */ 461 while (high-low > 5) { 462 t = (low+high)/2; 463 if (rp[t] > col) high = t; 464 else low = t; 465 } 466 for (i=low; i<high; i++) { 467 if (rp[i] > col) break; 468 if (rp[i] == col) { 469 *v++ = ap[i]; 470 goto finished; 471 } 472 } 473 *v++ = 0.0; 474 finished:; 475 } 476 } 477 PetscFunctionReturn(0); 478 } 479 480 481 #undef __FUNCT__ 482 #define __FUNCT__ "MatView_SeqAIJ_Binary" 483 PetscErrorCode MatView_SeqAIJ_Binary(Mat A,PetscViewer viewer) 484 { 485 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 486 PetscErrorCode ierr; 487 PetscInt i,*col_lens; 488 int fd; 489 FILE *file; 490 491 PetscFunctionBegin; 492 ierr = PetscViewerBinaryGetDescriptor(viewer,&fd);CHKERRQ(ierr); 493 ierr = PetscMalloc((4+A->rmap->n)*sizeof(PetscInt),&col_lens);CHKERRQ(ierr); 494 495 col_lens[0] = MAT_FILE_CLASSID; 496 col_lens[1] = A->rmap->n; 497 col_lens[2] = A->cmap->n; 498 col_lens[3] = a->nz; 499 500 /* store lengths of each row and write (including header) to file */ 501 for (i=0; i<A->rmap->n; i++) { 502 col_lens[4+i] = a->i[i+1] - a->i[i]; 503 } 504 ierr = PetscBinaryWrite(fd,col_lens,4+A->rmap->n,PETSC_INT,PETSC_TRUE);CHKERRQ(ierr); 505 ierr = PetscFree(col_lens);CHKERRQ(ierr); 506 507 /* store column indices (zero start index) */ 508 ierr = PetscBinaryWrite(fd,a->j,a->nz,PETSC_INT,PETSC_FALSE);CHKERRQ(ierr); 509 510 /* store nonzero values */ 511 ierr = PetscBinaryWrite(fd,a->a,a->nz,PETSC_SCALAR,PETSC_FALSE);CHKERRQ(ierr); 512 513 ierr = PetscViewerBinaryGetInfoPointer(viewer,&file);CHKERRQ(ierr); 514 if (file) { 515 fprintf(file,"-matload_block_size %d\n",(int)A->rmap->bs); 516 } 517 PetscFunctionReturn(0); 518 } 519 520 extern PetscErrorCode MatSeqAIJFactorInfo_Matlab(Mat,PetscViewer); 521 522 #undef __FUNCT__ 523 #define __FUNCT__ "MatView_SeqAIJ_ASCII" 524 PetscErrorCode MatView_SeqAIJ_ASCII(Mat A,PetscViewer viewer) 525 { 526 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 527 PetscErrorCode ierr; 528 PetscInt i,j,m = A->rmap->n,shift=0; 529 const char *name; 530 PetscViewerFormat format; 531 532 PetscFunctionBegin; 533 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 534 if (format == PETSC_VIEWER_ASCII_MATLAB) { 535 PetscInt nofinalvalue = 0; 536 if (m && ((a->i[m] == a->i[m-1]) || (a->j[a->nz-1] != A->cmap->n-!shift))) { 537 nofinalvalue = 1; 538 } 539 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 540 ierr = PetscViewerASCIIPrintf(viewer,"%% Size = %D %D \n",m,A->cmap->n);CHKERRQ(ierr); 541 ierr = PetscViewerASCIIPrintf(viewer,"%% Nonzeros = %D \n",a->nz);CHKERRQ(ierr); 542 ierr = PetscViewerASCIIPrintf(viewer,"zzz = zeros(%D,3);\n",a->nz+nofinalvalue);CHKERRQ(ierr); 543 ierr = PetscViewerASCIIPrintf(viewer,"zzz = [\n");CHKERRQ(ierr); 544 545 for (i=0; i<m; i++) { 546 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 547 #if defined(PETSC_USE_COMPLEX) 548 ierr = PetscViewerASCIIPrintf(viewer,"%D %D %18.16e + %18.16ei \n",i+1,a->j[j]+!shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 549 #else 550 ierr = PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",i+1,a->j[j]+!shift,a->a[j]);CHKERRQ(ierr); 551 #endif 552 } 553 } 554 if (nofinalvalue) { 555 ierr = PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",m,A->cmap->n,0.0);CHKERRQ(ierr); 556 } 557 ierr = PetscObjectGetName((PetscObject)A,&name);CHKERRQ(ierr); 558 ierr = PetscViewerASCIIPrintf(viewer,"];\n %s = spconvert(zzz);\n",name);CHKERRQ(ierr); 559 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 560 } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO || format == PETSC_VIEWER_ASCII_INFO) { 561 PetscFunctionReturn(0); 562 } else if (format == PETSC_VIEWER_ASCII_COMMON) { 563 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 564 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)A,viewer);CHKERRQ(ierr); 565 for (i=0; i<m; i++) { 566 ierr = PetscViewerASCIIPrintf(viewer,"row %D:",i);CHKERRQ(ierr); 567 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 568 #if defined(PETSC_USE_COMPLEX) 569 if (PetscImaginaryPart(a->a[j]) > 0.0 && PetscRealPart(a->a[j]) != 0.0) { 570 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 571 } else if (PetscImaginaryPart(a->a[j]) < 0.0 && PetscRealPart(a->a[j]) != 0.0) { 572 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 573 } else if (PetscRealPart(a->a[j]) != 0.0) { 574 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,PetscRealPart(a->a[j]));CHKERRQ(ierr); 575 } 576 #else 577 if (a->a[j] != 0.0) {ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,(double)a->a[j]);CHKERRQ(ierr);} 578 #endif 579 } 580 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 581 } 582 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 583 } else if (format == PETSC_VIEWER_ASCII_SYMMODU) { 584 PetscInt nzd=0,fshift=1,*sptr; 585 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 586 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)A,viewer);CHKERRQ(ierr); 587 ierr = PetscMalloc((m+1)*sizeof(PetscInt),&sptr);CHKERRQ(ierr); 588 for (i=0; i<m; i++) { 589 sptr[i] = nzd+1; 590 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 591 if (a->j[j] >= i) { 592 #if defined(PETSC_USE_COMPLEX) 593 if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) nzd++; 594 #else 595 if (a->a[j] != 0.0) nzd++; 596 #endif 597 } 598 } 599 } 600 sptr[m] = nzd+1; 601 ierr = PetscViewerASCIIPrintf(viewer," %D %D\n\n",m,nzd);CHKERRQ(ierr); 602 for (i=0; i<m+1; i+=6) { 603 if (i+4<m) { 604 ierr = PetscViewerASCIIPrintf(viewer," %D %D %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3],sptr[i+4],sptr[i+5]);CHKERRQ(ierr); 605 } else if (i+3<m) { 606 ierr = PetscViewerASCIIPrintf(viewer," %D %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3],sptr[i+4]);CHKERRQ(ierr); 607 } else if (i+2<m) { 608 ierr = PetscViewerASCIIPrintf(viewer," %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3]);CHKERRQ(ierr); 609 } else if (i+1<m) { 610 ierr = PetscViewerASCIIPrintf(viewer," %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2]);CHKERRQ(ierr); 611 } else if (i<m) { 612 ierr = PetscViewerASCIIPrintf(viewer," %D %D\n",sptr[i],sptr[i+1]);CHKERRQ(ierr); 613 } else { 614 ierr = PetscViewerASCIIPrintf(viewer," %D\n",sptr[i]);CHKERRQ(ierr); 615 } 616 } 617 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 618 ierr = PetscFree(sptr);CHKERRQ(ierr); 619 for (i=0; i<m; i++) { 620 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 621 if (a->j[j] >= i) {ierr = PetscViewerASCIIPrintf(viewer," %D ",a->j[j]+fshift);CHKERRQ(ierr);} 622 } 623 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 624 } 625 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 626 for (i=0; i<m; i++) { 627 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 628 if (a->j[j] >= i) { 629 #if defined(PETSC_USE_COMPLEX) 630 if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) { 631 ierr = PetscViewerASCIIPrintf(viewer," %18.16e %18.16e ",PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 632 } 633 #else 634 if (a->a[j] != 0.0) {ierr = PetscViewerASCIIPrintf(viewer," %18.16e ",a->a[j]);CHKERRQ(ierr);} 635 #endif 636 } 637 } 638 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 639 } 640 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 641 } else if (format == PETSC_VIEWER_ASCII_DENSE) { 642 PetscInt cnt = 0,jcnt; 643 PetscScalar value; 644 645 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 646 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)A,viewer);CHKERRQ(ierr); 647 for (i=0; i<m; i++) { 648 jcnt = 0; 649 for (j=0; j<A->cmap->n; j++) { 650 if (jcnt < a->i[i+1]-a->i[i] && j == a->j[cnt]) { 651 value = a->a[cnt++]; 652 jcnt++; 653 } else { 654 value = 0.0; 655 } 656 #if defined(PETSC_USE_COMPLEX) 657 ierr = PetscViewerASCIIPrintf(viewer," %7.5e+%7.5e i ",PetscRealPart(value),PetscImaginaryPart(value));CHKERRQ(ierr); 658 #else 659 ierr = PetscViewerASCIIPrintf(viewer," %7.5e ",value);CHKERRQ(ierr); 660 #endif 661 } 662 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 663 } 664 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 665 } else if (format == PETSC_VIEWER_ASCII_MATRIXMARKET) { 666 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 667 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)A,viewer);CHKERRQ(ierr); 668 #if defined(PETSC_USE_COMPLEX) 669 ierr = PetscViewerASCIIPrintf(viewer,"%%matrix complex general\n");CHKERRQ(ierr); 670 #else 671 ierr = PetscViewerASCIIPrintf(viewer,"%%matrix real general\n");CHKERRQ(ierr); 672 #endif 673 ierr = PetscViewerASCIIPrintf(viewer,"%D %D %D\n", m, A->cmap->n, a->nz);CHKERRQ(ierr); 674 for (i=0; i<m; i++) { 675 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 676 #if defined(PETSC_USE_COMPLEX) 677 if (PetscImaginaryPart(a->a[j]) > 0.0) { 678 ierr = PetscViewerASCIIPrintf(viewer,"%D %D, %g %g\n", i+shift,a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 679 } else if (PetscImaginaryPart(a->a[j]) < 0.0) { 680 ierr = PetscViewerASCIIPrintf(viewer,"%D %D, %g -%g\n", i+shift,a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 681 } else { 682 ierr = PetscViewerASCIIPrintf(viewer,"%D %D, %g\n", i+shift,a->j[j]+shift,PetscRealPart(a->a[j]));CHKERRQ(ierr); 683 } 684 #else 685 ierr = PetscViewerASCIIPrintf(viewer,"%D %D %g\n", i+shift, a->j[j]+shift, (double)a->a[j]);CHKERRQ(ierr); 686 #endif 687 } 688 } 689 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 690 } else { 691 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);CHKERRQ(ierr); 692 ierr = PetscObjectPrintClassNamePrefixType((PetscObject)A,viewer);CHKERRQ(ierr); 693 if (A->factortype) { 694 for (i=0; i<m; i++) { 695 ierr = PetscViewerASCIIPrintf(viewer,"row %D:",i);CHKERRQ(ierr); 696 /* L part */ 697 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 698 #if defined(PETSC_USE_COMPLEX) 699 if (PetscImaginaryPart(a->a[j]) > 0.0) { 700 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 701 } else if (PetscImaginaryPart(a->a[j]) < 0.0) { 702 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 703 } else { 704 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,PetscRealPart(a->a[j]));CHKERRQ(ierr); 705 } 706 #else 707 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,(double)a->a[j]);CHKERRQ(ierr); 708 #endif 709 } 710 /* diagonal */ 711 j = a->diag[i]; 712 #if defined(PETSC_USE_COMPLEX) 713 if (PetscImaginaryPart(a->a[j]) > 0.0) { 714 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j]+shift,PetscRealPart(1.0/a->a[j]),PetscImaginaryPart(1.0/a->a[j]));CHKERRQ(ierr); 715 } else if (PetscImaginaryPart(a->a[j]) < 0.0) { 716 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j]+shift,PetscRealPart(1.0/a->a[j]),-PetscImaginaryPart(1.0/a->a[j]));CHKERRQ(ierr); 717 } else { 718 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,PetscRealPart(1.0/a->a[j]));CHKERRQ(ierr); 719 } 720 #else 721 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,(double)1.0/a->a[j]);CHKERRQ(ierr); 722 #endif 723 724 /* U part */ 725 for (j=a->diag[i+1]+1+shift; j<a->diag[i]+shift; j++) { 726 #if defined(PETSC_USE_COMPLEX) 727 if (PetscImaginaryPart(a->a[j]) > 0.0) { 728 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 729 } else if (PetscImaginaryPart(a->a[j]) < 0.0) { 730 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 731 } else { 732 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,PetscRealPart(a->a[j]));CHKERRQ(ierr); 733 } 734 #else 735 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,(double)a->a[j]);CHKERRQ(ierr); 736 #endif 737 } 738 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 739 } 740 } else { 741 for (i=0; i<m; i++) { 742 ierr = PetscViewerASCIIPrintf(viewer,"row %D:",i);CHKERRQ(ierr); 743 for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) { 744 #if defined(PETSC_USE_COMPLEX) 745 if (PetscImaginaryPart(a->a[j]) > 0.0) { 746 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g + %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 747 } else if (PetscImaginaryPart(a->a[j]) < 0.0) { 748 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g - %g i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));CHKERRQ(ierr); 749 } else { 750 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,PetscRealPart(a->a[j]));CHKERRQ(ierr); 751 } 752 #else 753 ierr = PetscViewerASCIIPrintf(viewer," (%D, %g) ",a->j[j]+shift,(double)a->a[j]);CHKERRQ(ierr); 754 #endif 755 } 756 ierr = PetscViewerASCIIPrintf(viewer,"\n");CHKERRQ(ierr); 757 } 758 } 759 ierr = PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);CHKERRQ(ierr); 760 } 761 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 762 PetscFunctionReturn(0); 763 } 764 765 #include <petscdraw.h> 766 #undef __FUNCT__ 767 #define __FUNCT__ "MatView_SeqAIJ_Draw_Zoom" 768 PetscErrorCode MatView_SeqAIJ_Draw_Zoom(PetscDraw draw,void *Aa) 769 { 770 Mat A = (Mat) Aa; 771 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 772 PetscErrorCode ierr; 773 PetscInt i,j,m = A->rmap->n,color; 774 PetscReal xl,yl,xr,yr,x_l,x_r,y_l,y_r,maxv = 0.0; 775 PetscViewer viewer; 776 PetscViewerFormat format; 777 778 PetscFunctionBegin; 779 ierr = PetscObjectQuery((PetscObject)A,"Zoomviewer",(PetscObject*)&viewer);CHKERRQ(ierr); 780 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 781 782 ierr = PetscDrawGetCoordinates(draw,&xl,&yl,&xr,&yr);CHKERRQ(ierr); 783 /* loop over matrix elements drawing boxes */ 784 785 if (format != PETSC_VIEWER_DRAW_CONTOUR) { 786 /* Blue for negative, Cyan for zero and Red for positive */ 787 color = PETSC_DRAW_BLUE; 788 for (i=0; i<m; i++) { 789 y_l = m - i - 1.0; y_r = y_l + 1.0; 790 for (j=a->i[i]; j<a->i[i+1]; j++) { 791 x_l = a->j[j]; x_r = x_l + 1.0; 792 if (PetscRealPart(a->a[j]) >= 0.) continue; 793 ierr = PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);CHKERRQ(ierr); 794 } 795 } 796 color = PETSC_DRAW_CYAN; 797 for (i=0; i<m; i++) { 798 y_l = m - i - 1.0; y_r = y_l + 1.0; 799 for (j=a->i[i]; j<a->i[i+1]; j++) { 800 x_l = a->j[j]; x_r = x_l + 1.0; 801 if (a->a[j] != 0.) continue; 802 ierr = PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);CHKERRQ(ierr); 803 } 804 } 805 color = PETSC_DRAW_RED; 806 for (i=0; i<m; i++) { 807 y_l = m - i - 1.0; y_r = y_l + 1.0; 808 for (j=a->i[i]; j<a->i[i+1]; j++) { 809 x_l = a->j[j]; x_r = x_l + 1.0; 810 if (PetscRealPart(a->a[j]) <= 0.) continue; 811 ierr = PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);CHKERRQ(ierr); 812 } 813 } 814 } else { 815 /* use contour shading to indicate magnitude of values */ 816 /* first determine max of all nonzero values */ 817 PetscInt nz = a->nz,count; 818 PetscDraw popup; 819 PetscReal scale; 820 821 for (i=0; i<nz; i++) { 822 if (PetscAbsScalar(a->a[i]) > maxv) maxv = PetscAbsScalar(a->a[i]); 823 } 824 scale = (245.0 - PETSC_DRAW_BASIC_COLORS)/maxv; 825 ierr = PetscDrawGetPopup(draw,&popup);CHKERRQ(ierr); 826 if (popup) { 827 ierr = PetscDrawScalePopup(popup,0.0,maxv);CHKERRQ(ierr); 828 } 829 count = 0; 830 for (i=0; i<m; i++) { 831 y_l = m - i - 1.0; y_r = y_l + 1.0; 832 for (j=a->i[i]; j<a->i[i+1]; j++) { 833 x_l = a->j[j]; x_r = x_l + 1.0; 834 color = PETSC_DRAW_BASIC_COLORS + (PetscInt)(scale*PetscAbsScalar(a->a[count])); 835 ierr = PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);CHKERRQ(ierr); 836 count++; 837 } 838 } 839 } 840 PetscFunctionReturn(0); 841 } 842 843 #include <petscdraw.h> 844 #undef __FUNCT__ 845 #define __FUNCT__ "MatView_SeqAIJ_Draw" 846 PetscErrorCode MatView_SeqAIJ_Draw(Mat A,PetscViewer viewer) 847 { 848 PetscErrorCode ierr; 849 PetscDraw draw; 850 PetscReal xr,yr,xl,yl,h,w; 851 PetscBool isnull; 852 853 PetscFunctionBegin; 854 ierr = PetscViewerDrawGetDraw(viewer,0,&draw);CHKERRQ(ierr); 855 ierr = PetscDrawIsNull(draw,&isnull);CHKERRQ(ierr); 856 if (isnull) PetscFunctionReturn(0); 857 858 ierr = PetscObjectCompose((PetscObject)A,"Zoomviewer",(PetscObject)viewer);CHKERRQ(ierr); 859 xr = A->cmap->n; yr = A->rmap->n; h = yr/10.0; w = xr/10.0; 860 xr += w; yr += h; xl = -w; yl = -h; 861 ierr = PetscDrawSetCoordinates(draw,xl,yl,xr,yr);CHKERRQ(ierr); 862 ierr = PetscDrawZoom(draw,MatView_SeqAIJ_Draw_Zoom,A);CHKERRQ(ierr); 863 ierr = PetscObjectCompose((PetscObject)A,"Zoomviewer",NULL);CHKERRQ(ierr); 864 PetscFunctionReturn(0); 865 } 866 867 #undef __FUNCT__ 868 #define __FUNCT__ "MatView_SeqAIJ" 869 PetscErrorCode MatView_SeqAIJ(Mat A,PetscViewer viewer) 870 { 871 PetscErrorCode ierr; 872 PetscBool iascii,isbinary,isdraw; 873 874 PetscFunctionBegin; 875 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);CHKERRQ(ierr); 876 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERBINARY,&isbinary);CHKERRQ(ierr); 877 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERDRAW,&isdraw);CHKERRQ(ierr); 878 if (iascii) { 879 ierr = MatView_SeqAIJ_ASCII(A,viewer);CHKERRQ(ierr); 880 } else if (isbinary) { 881 ierr = MatView_SeqAIJ_Binary(A,viewer);CHKERRQ(ierr); 882 } else if (isdraw) { 883 ierr = MatView_SeqAIJ_Draw(A,viewer);CHKERRQ(ierr); 884 } 885 ierr = MatView_SeqAIJ_Inode(A,viewer);CHKERRQ(ierr); 886 PetscFunctionReturn(0); 887 } 888 889 #undef __FUNCT__ 890 #define __FUNCT__ "MatAssemblyEnd_SeqAIJ" 891 PetscErrorCode MatAssemblyEnd_SeqAIJ(Mat A,MatAssemblyType mode) 892 { 893 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 894 PetscErrorCode ierr; 895 PetscInt fshift = 0,i,j,*ai = a->i,*aj = a->j,*imax = a->imax; 896 PetscInt m = A->rmap->n,*ip,N,*ailen = a->ilen,rmax = 0; 897 MatScalar *aa = a->a,*ap; 898 PetscReal ratio = 0.6; 899 900 PetscFunctionBegin; 901 if (mode == MAT_FLUSH_ASSEMBLY) PetscFunctionReturn(0); 902 903 if (m) rmax = ailen[0]; /* determine row with most nonzeros */ 904 for (i=1; i<m; i++) { 905 /* move each row back by the amount of empty slots (fshift) before it*/ 906 fshift += imax[i-1] - ailen[i-1]; 907 rmax = PetscMax(rmax,ailen[i]); 908 if (fshift) { 909 ip = aj + ai[i]; 910 ap = aa + ai[i]; 911 N = ailen[i]; 912 for (j=0; j<N; j++) { 913 ip[j-fshift] = ip[j]; 914 ap[j-fshift] = ap[j]; 915 } 916 } 917 ai[i] = ai[i-1] + ailen[i-1]; 918 } 919 if (m) { 920 fshift += imax[m-1] - ailen[m-1]; 921 ai[m] = ai[m-1] + ailen[m-1]; 922 } 923 /* reset ilen and imax for each row */ 924 for (i=0; i<m; i++) { 925 ailen[i] = imax[i] = ai[i+1] - ai[i]; 926 } 927 a->nz = ai[m]; 928 if (fshift && a->nounused == -1) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_PLIB, "Unused space detected in matrix: %D X %D, %D unneeded", m, A->cmap->n, fshift); 929 930 ierr = MatMarkDiagonal_SeqAIJ(A);CHKERRQ(ierr); 931 ierr = PetscInfo4(A,"Matrix size: %D X %D; storage space: %D unneeded,%D used\n",m,A->cmap->n,fshift,a->nz);CHKERRQ(ierr); 932 ierr = PetscInfo1(A,"Number of mallocs during MatSetValues() is %D\n",a->reallocs);CHKERRQ(ierr); 933 ierr = PetscInfo1(A,"Maximum nonzeros in any row is %D\n",rmax);CHKERRQ(ierr); 934 935 A->info.mallocs += a->reallocs; 936 a->reallocs = 0; 937 A->info.nz_unneeded = (double)fshift; 938 a->rmax = rmax; 939 940 ierr = MatCheckCompressedRow(A,&a->compressedrow,a->i,m,ratio);CHKERRQ(ierr); 941 942 A->same_nonzero = PETSC_TRUE; 943 944 ierr = MatAssemblyEnd_SeqAIJ_Inode(A,mode);CHKERRQ(ierr); 945 946 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 947 PetscFunctionReturn(0); 948 } 949 950 #undef __FUNCT__ 951 #define __FUNCT__ "MatRealPart_SeqAIJ" 952 PetscErrorCode MatRealPart_SeqAIJ(Mat A) 953 { 954 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 955 PetscInt i,nz = a->nz; 956 MatScalar *aa = a->a; 957 PetscErrorCode ierr; 958 959 PetscFunctionBegin; 960 for (i=0; i<nz; i++) aa[i] = PetscRealPart(aa[i]); 961 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 962 PetscFunctionReturn(0); 963 } 964 965 #undef __FUNCT__ 966 #define __FUNCT__ "MatImaginaryPart_SeqAIJ" 967 PetscErrorCode MatImaginaryPart_SeqAIJ(Mat A) 968 { 969 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 970 PetscInt i,nz = a->nz; 971 MatScalar *aa = a->a; 972 PetscErrorCode ierr; 973 974 PetscFunctionBegin; 975 for (i=0; i<nz; i++) aa[i] = PetscImaginaryPart(aa[i]); 976 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 977 PetscFunctionReturn(0); 978 } 979 980 #if defined(PETSC_THREADCOMM_ACTIVE) 981 PetscErrorCode MatZeroEntries_SeqAIJ_Kernel(PetscInt thread_id,Mat A) 982 { 983 PetscErrorCode ierr; 984 PetscInt *trstarts=A->rmap->trstarts; 985 PetscInt n,start,end; 986 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 987 988 start = trstarts[thread_id]; 989 end = trstarts[thread_id+1]; 990 n = a->i[end] - a->i[start]; 991 ierr = PetscMemzero(a->a+a->i[start],n*sizeof(PetscScalar));CHKERRQ(ierr); 992 return 0; 993 } 994 995 #undef __FUNCT__ 996 #define __FUNCT__ "MatZeroEntries_SeqAIJ" 997 PetscErrorCode MatZeroEntries_SeqAIJ(Mat A) 998 { 999 PetscErrorCode ierr; 1000 1001 PetscFunctionBegin; 1002 ierr = PetscThreadCommRunKernel(PetscObjectComm((PetscObject)A),(PetscThreadKernel)MatZeroEntries_SeqAIJ_Kernel,1,A);CHKERRQ(ierr); 1003 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 1004 PetscFunctionReturn(0); 1005 } 1006 #else 1007 #undef __FUNCT__ 1008 #define __FUNCT__ "MatZeroEntries_SeqAIJ" 1009 PetscErrorCode MatZeroEntries_SeqAIJ(Mat A) 1010 { 1011 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1012 PetscErrorCode ierr; 1013 1014 PetscFunctionBegin; 1015 ierr = PetscMemzero(a->a,(a->i[A->rmap->n])*sizeof(PetscScalar));CHKERRQ(ierr); 1016 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 1017 PetscFunctionReturn(0); 1018 } 1019 #endif 1020 1021 #undef __FUNCT__ 1022 #define __FUNCT__ "MatDestroy_SeqAIJ" 1023 PetscErrorCode MatDestroy_SeqAIJ(Mat A) 1024 { 1025 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1026 PetscErrorCode ierr; 1027 1028 PetscFunctionBegin; 1029 #if defined(PETSC_USE_LOG) 1030 PetscLogObjectState((PetscObject)A,"Rows=%D, Cols=%D, NZ=%D",A->rmap->n,A->cmap->n,a->nz); 1031 #endif 1032 ierr = MatSeqXAIJFreeAIJ(A,&a->a,&a->j,&a->i);CHKERRQ(ierr); 1033 ierr = ISDestroy(&a->row);CHKERRQ(ierr); 1034 ierr = ISDestroy(&a->col);CHKERRQ(ierr); 1035 ierr = PetscFree(a->diag);CHKERRQ(ierr); 1036 ierr = PetscFree(a->ibdiag);CHKERRQ(ierr); 1037 ierr = PetscFree2(a->imax,a->ilen);CHKERRQ(ierr); 1038 ierr = PetscFree3(a->idiag,a->mdiag,a->ssor_work);CHKERRQ(ierr); 1039 ierr = PetscFree(a->solve_work);CHKERRQ(ierr); 1040 ierr = ISDestroy(&a->icol);CHKERRQ(ierr); 1041 ierr = PetscFree(a->saved_values);CHKERRQ(ierr); 1042 ierr = ISColoringDestroy(&a->coloring);CHKERRQ(ierr); 1043 ierr = PetscFree(a->xtoy);CHKERRQ(ierr); 1044 ierr = MatDestroy(&a->XtoY);CHKERRQ(ierr); 1045 ierr = PetscFree2(a->compressedrow.i,a->compressedrow.rindex);CHKERRQ(ierr); 1046 ierr = PetscFree(a->matmult_abdense);CHKERRQ(ierr); 1047 1048 ierr = MatDestroy_SeqAIJ_Inode(A);CHKERRQ(ierr); 1049 ierr = PetscFree(A->data);CHKERRQ(ierr); 1050 1051 ierr = PetscObjectChangeTypeName((PetscObject)A,0);CHKERRQ(ierr); 1052 ierr = PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetColumnIndices_C",NULL);CHKERRQ(ierr); 1053 ierr = PetscObjectComposeFunction((PetscObject)A,"MatStoreValues_C",NULL);CHKERRQ(ierr); 1054 ierr = PetscObjectComposeFunction((PetscObject)A,"MatRetrieveValues_C",NULL);CHKERRQ(ierr); 1055 ierr = PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqsbaij_C",NULL);CHKERRQ(ierr); 1056 ierr = PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqbaij_C",NULL);CHKERRQ(ierr); 1057 ierr = PetscObjectComposeFunction((PetscObject)A,"MatConvert_seqaij_seqaijperm_C",NULL);CHKERRQ(ierr); 1058 ierr = PetscObjectComposeFunction((PetscObject)A,"MatIsTranspose_C",NULL);CHKERRQ(ierr); 1059 ierr = PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetPreallocation_C",NULL);CHKERRQ(ierr); 1060 ierr = PetscObjectComposeFunction((PetscObject)A,"MatSeqAIJSetPreallocationCSR_C",NULL);CHKERRQ(ierr); 1061 ierr = PetscObjectComposeFunction((PetscObject)A,"MatReorderForNonzeroDiagonal_C",NULL);CHKERRQ(ierr); 1062 PetscFunctionReturn(0); 1063 } 1064 1065 #undef __FUNCT__ 1066 #define __FUNCT__ "MatSetOption_SeqAIJ" 1067 PetscErrorCode MatSetOption_SeqAIJ(Mat A,MatOption op,PetscBool flg) 1068 { 1069 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1070 PetscErrorCode ierr; 1071 1072 PetscFunctionBegin; 1073 switch (op) { 1074 case MAT_ROW_ORIENTED: 1075 a->roworiented = flg; 1076 break; 1077 case MAT_KEEP_NONZERO_PATTERN: 1078 a->keepnonzeropattern = flg; 1079 break; 1080 case MAT_NEW_NONZERO_LOCATIONS: 1081 a->nonew = (flg ? 0 : 1); 1082 break; 1083 case MAT_NEW_NONZERO_LOCATION_ERR: 1084 a->nonew = (flg ? -1 : 0); 1085 break; 1086 case MAT_NEW_NONZERO_ALLOCATION_ERR: 1087 a->nonew = (flg ? -2 : 0); 1088 break; 1089 case MAT_UNUSED_NONZERO_LOCATION_ERR: 1090 a->nounused = (flg ? -1 : 0); 1091 break; 1092 case MAT_IGNORE_ZERO_ENTRIES: 1093 a->ignorezeroentries = flg; 1094 break; 1095 case MAT_CHECK_COMPRESSED_ROW: 1096 a->compressedrow.check = flg; 1097 break; 1098 case MAT_SPD: 1099 case MAT_SYMMETRIC: 1100 case MAT_STRUCTURALLY_SYMMETRIC: 1101 case MAT_HERMITIAN: 1102 case MAT_SYMMETRY_ETERNAL: 1103 /* These options are handled directly by MatSetOption() */ 1104 break; 1105 case MAT_NEW_DIAGONALS: 1106 case MAT_IGNORE_OFF_PROC_ENTRIES: 1107 case MAT_USE_HASH_TABLE: 1108 ierr = PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);CHKERRQ(ierr); 1109 break; 1110 case MAT_USE_INODES: 1111 /* Not an error because MatSetOption_SeqAIJ_Inode handles this one */ 1112 break; 1113 default: 1114 SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"unknown option %d",op); 1115 } 1116 ierr = MatSetOption_SeqAIJ_Inode(A,op,flg);CHKERRQ(ierr); 1117 PetscFunctionReturn(0); 1118 } 1119 1120 #undef __FUNCT__ 1121 #define __FUNCT__ "MatGetDiagonal_SeqAIJ" 1122 PetscErrorCode MatGetDiagonal_SeqAIJ(Mat A,Vec v) 1123 { 1124 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1125 PetscErrorCode ierr; 1126 PetscInt i,j,n,*ai=a->i,*aj=a->j,nz; 1127 PetscScalar *aa=a->a,*x,zero=0.0; 1128 1129 PetscFunctionBegin; 1130 ierr = VecGetLocalSize(v,&n);CHKERRQ(ierr); 1131 if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector"); 1132 1133 if (A->factortype == MAT_FACTOR_ILU || A->factortype == MAT_FACTOR_LU) { 1134 PetscInt *diag=a->diag; 1135 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 1136 for (i=0; i<n; i++) x[i] = 1.0/aa[diag[i]]; 1137 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 1138 PetscFunctionReturn(0); 1139 } 1140 1141 ierr = VecSet(v,zero);CHKERRQ(ierr); 1142 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 1143 for (i=0; i<n; i++) { 1144 nz = ai[i+1] - ai[i]; 1145 if (!nz) x[i] = 0.0; 1146 for (j=ai[i]; j<ai[i+1]; j++) { 1147 if (aj[j] == i) { 1148 x[i] = aa[j]; 1149 break; 1150 } 1151 } 1152 } 1153 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 1154 PetscFunctionReturn(0); 1155 } 1156 1157 #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h> 1158 #undef __FUNCT__ 1159 #define __FUNCT__ "MatMultTransposeAdd_SeqAIJ" 1160 PetscErrorCode MatMultTransposeAdd_SeqAIJ(Mat A,Vec xx,Vec zz,Vec yy) 1161 { 1162 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1163 PetscScalar *x,*y; 1164 PetscErrorCode ierr; 1165 PetscInt m = A->rmap->n; 1166 #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ) 1167 MatScalar *v; 1168 PetscScalar alpha; 1169 PetscInt n,i,j,*idx,*ii,*ridx=NULL; 1170 Mat_CompressedRow cprow = a->compressedrow; 1171 PetscBool usecprow = cprow.use; 1172 #endif 1173 1174 PetscFunctionBegin; 1175 if (zz != yy) {ierr = VecCopy(zz,yy);CHKERRQ(ierr);} 1176 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 1177 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1178 1179 #if defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ) 1180 fortranmulttransposeaddaij_(&m,x,a->i,a->j,a->a,y); 1181 #else 1182 if (usecprow) { 1183 m = cprow.nrows; 1184 ii = cprow.i; 1185 ridx = cprow.rindex; 1186 } else { 1187 ii = a->i; 1188 } 1189 for (i=0; i<m; i++) { 1190 idx = a->j + ii[i]; 1191 v = a->a + ii[i]; 1192 n = ii[i+1] - ii[i]; 1193 if (usecprow) { 1194 alpha = x[ridx[i]]; 1195 } else { 1196 alpha = x[i]; 1197 } 1198 for (j=0; j<n; j++) y[idx[j]] += alpha*v[j]; 1199 } 1200 #endif 1201 ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr); 1202 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 1203 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1204 PetscFunctionReturn(0); 1205 } 1206 1207 #undef __FUNCT__ 1208 #define __FUNCT__ "MatMultTranspose_SeqAIJ" 1209 PetscErrorCode MatMultTranspose_SeqAIJ(Mat A,Vec xx,Vec yy) 1210 { 1211 PetscErrorCode ierr; 1212 1213 PetscFunctionBegin; 1214 ierr = VecSet(yy,0.0);CHKERRQ(ierr); 1215 ierr = MatMultTransposeAdd_SeqAIJ(A,xx,yy,yy);CHKERRQ(ierr); 1216 PetscFunctionReturn(0); 1217 } 1218 1219 #include <../src/mat/impls/aij/seq/ftn-kernels/fmult.h> 1220 #if defined(PETSC_THREADCOMM_ACTIVE) 1221 PetscErrorCode MatMult_SeqAIJ_Kernel(PetscInt thread_id,Mat A,Vec xx,Vec yy) 1222 { 1223 PetscErrorCode ierr; 1224 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1225 PetscScalar *y; 1226 const PetscScalar *x; 1227 const MatScalar *aa; 1228 PetscInt *trstarts=A->rmap->trstarts; 1229 PetscInt n,start,end,i; 1230 const PetscInt *aj,*ai; 1231 PetscScalar sum; 1232 1233 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1234 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1235 start = trstarts[thread_id]; 1236 end = trstarts[thread_id+1]; 1237 aj = a->j; 1238 aa = a->a; 1239 ai = a->i; 1240 for (i=start; i<end; i++) { 1241 n = ai[i+1] - ai[i]; 1242 aj = a->j + ai[i]; 1243 aa = a->a + ai[i]; 1244 sum = 0.0; 1245 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1246 y[i] = sum; 1247 } 1248 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1249 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1250 return 0; 1251 } 1252 1253 #undef __FUNCT__ 1254 #define __FUNCT__ "MatMult_SeqAIJ" 1255 PetscErrorCode MatMult_SeqAIJ(Mat A,Vec xx,Vec yy) 1256 { 1257 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1258 PetscScalar *y; 1259 const PetscScalar *x; 1260 const MatScalar *aa; 1261 PetscErrorCode ierr; 1262 PetscInt m=A->rmap->n; 1263 const PetscInt *aj,*ii,*ridx=NULL; 1264 PetscInt n,i,nonzerorow=0; 1265 PetscScalar sum; 1266 PetscBool usecprow=a->compressedrow.use; 1267 1268 #if defined(PETSC_HAVE_PRAGMA_DISJOINT) 1269 #pragma disjoint(*x,*y,*aa) 1270 #endif 1271 1272 PetscFunctionBegin; 1273 aj = a->j; 1274 aa = a->a; 1275 ii = a->i; 1276 if (usecprow) { /* use compressed row format */ 1277 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1278 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1279 m = a->compressedrow.nrows; 1280 ii = a->compressedrow.i; 1281 ridx = a->compressedrow.rindex; 1282 for (i=0; i<m; i++) { 1283 n = ii[i+1] - ii[i]; 1284 aj = a->j + ii[i]; 1285 aa = a->a + ii[i]; 1286 sum = 0.0; 1287 nonzerorow += (n>0); 1288 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1289 /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */ 1290 y[*ridx++] = sum; 1291 } 1292 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1293 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1294 } else { /* do not use compressed row format */ 1295 #if defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ) 1296 fortranmultaij_(&m,x,ii,aj,aa,y); 1297 #else 1298 ierr = PetscThreadCommRunKernel(PetscObjectComm((PetscObject)A),(PetscThreadKernel)MatMult_SeqAIJ_Kernel,3,A,xx,yy);CHKERRQ(ierr); 1299 #endif 1300 } 1301 ierr = PetscLogFlops(2.0*a->nz - nonzerorow);CHKERRQ(ierr); 1302 PetscFunctionReturn(0); 1303 } 1304 #else 1305 #undef __FUNCT__ 1306 #define __FUNCT__ "MatMult_SeqAIJ" 1307 PetscErrorCode MatMult_SeqAIJ(Mat A,Vec xx,Vec yy) 1308 { 1309 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1310 PetscScalar *y; 1311 const PetscScalar *x; 1312 const MatScalar *aa; 1313 PetscErrorCode ierr; 1314 PetscInt m=A->rmap->n; 1315 const PetscInt *aj,*ii,*ridx=NULL; 1316 PetscInt n,i,nonzerorow=0; 1317 PetscScalar sum; 1318 PetscBool usecprow=a->compressedrow.use; 1319 1320 #if defined(PETSC_HAVE_PRAGMA_DISJOINT) 1321 #pragma disjoint(*x,*y,*aa) 1322 #endif 1323 1324 PetscFunctionBegin; 1325 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1326 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1327 aj = a->j; 1328 aa = a->a; 1329 ii = a->i; 1330 if (usecprow) { /* use compressed row format */ 1331 m = a->compressedrow.nrows; 1332 ii = a->compressedrow.i; 1333 ridx = a->compressedrow.rindex; 1334 for (i=0; i<m; i++) { 1335 n = ii[i+1] - ii[i]; 1336 aj = a->j + ii[i]; 1337 aa = a->a + ii[i]; 1338 sum = 0.0; 1339 nonzerorow += (n>0); 1340 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1341 /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */ 1342 y[*ridx++] = sum; 1343 } 1344 } else { /* do not use compressed row format */ 1345 #if defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ) 1346 fortranmultaij_(&m,x,ii,aj,aa,y); 1347 #else 1348 #if defined(PETSC_THREADCOMM_ACTIVE) 1349 ierr = PetscThreadCommRunKernel(PetscObjectComm((PetscObject)A),(PetscThreadKernel)MatMult_SeqAIJ_Kernel,3,A,xx,yy);CHKERRQ(ierr); 1350 #else 1351 for (i=0; i<m; i++) { 1352 n = ii[i+1] - ii[i]; 1353 aj = a->j + ii[i]; 1354 aa = a->a + ii[i]; 1355 sum = 0.0; 1356 nonzerorow += (n>0); 1357 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1358 y[i] = sum; 1359 } 1360 #endif 1361 #endif 1362 } 1363 ierr = PetscLogFlops(2.0*a->nz - nonzerorow);CHKERRQ(ierr); 1364 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1365 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1366 PetscFunctionReturn(0); 1367 } 1368 #endif 1369 1370 #undef __FUNCT__ 1371 #define __FUNCT__ "MatMultMax_SeqAIJ" 1372 PetscErrorCode MatMultMax_SeqAIJ(Mat A,Vec xx,Vec yy) 1373 { 1374 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1375 PetscScalar *y; 1376 const PetscScalar *x; 1377 const MatScalar *aa; 1378 PetscErrorCode ierr; 1379 PetscInt m=A->rmap->n; 1380 const PetscInt *aj,*ii,*ridx=NULL; 1381 PetscInt n,i,nonzerorow=0; 1382 PetscScalar sum; 1383 PetscBool usecprow=a->compressedrow.use; 1384 1385 #if defined(PETSC_HAVE_PRAGMA_DISJOINT) 1386 #pragma disjoint(*x,*y,*aa) 1387 #endif 1388 1389 PetscFunctionBegin; 1390 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1391 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1392 aj = a->j; 1393 aa = a->a; 1394 ii = a->i; 1395 if (usecprow) { /* use compressed row format */ 1396 m = a->compressedrow.nrows; 1397 ii = a->compressedrow.i; 1398 ridx = a->compressedrow.rindex; 1399 for (i=0; i<m; i++) { 1400 n = ii[i+1] - ii[i]; 1401 aj = a->j + ii[i]; 1402 aa = a->a + ii[i]; 1403 sum = 0.0; 1404 nonzerorow += (n>0); 1405 PetscSparseDenseMaxDot(sum,x,aa,aj,n); 1406 /* for (j=0; j<n; j++) sum += (*aa++)*x[*aj++]; */ 1407 y[*ridx++] = sum; 1408 } 1409 } else { /* do not use compressed row format */ 1410 for (i=0; i<m; i++) { 1411 n = ii[i+1] - ii[i]; 1412 aj = a->j + ii[i]; 1413 aa = a->a + ii[i]; 1414 sum = 0.0; 1415 nonzerorow += (n>0); 1416 PetscSparseDenseMaxDot(sum,x,aa,aj,n); 1417 y[i] = sum; 1418 } 1419 } 1420 ierr = PetscLogFlops(2.0*a->nz - nonzerorow);CHKERRQ(ierr); 1421 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1422 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1423 PetscFunctionReturn(0); 1424 } 1425 1426 #undef __FUNCT__ 1427 #define __FUNCT__ "MatMultAddMax_SeqAIJ" 1428 PetscErrorCode MatMultAddMax_SeqAIJ(Mat A,Vec xx,Vec yy,Vec zz) 1429 { 1430 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1431 PetscScalar *y,*z; 1432 const PetscScalar *x; 1433 const MatScalar *aa; 1434 PetscErrorCode ierr; 1435 PetscInt m = A->rmap->n,*aj,*ii; 1436 PetscInt n,i,*ridx=NULL; 1437 PetscScalar sum; 1438 PetscBool usecprow=a->compressedrow.use; 1439 1440 PetscFunctionBegin; 1441 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1442 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1443 if (zz != yy) { 1444 ierr = VecGetArray(zz,&z);CHKERRQ(ierr); 1445 } else { 1446 z = y; 1447 } 1448 1449 aj = a->j; 1450 aa = a->a; 1451 ii = a->i; 1452 if (usecprow) { /* use compressed row format */ 1453 if (zz != yy) { 1454 ierr = PetscMemcpy(z,y,m*sizeof(PetscScalar));CHKERRQ(ierr); 1455 } 1456 m = a->compressedrow.nrows; 1457 ii = a->compressedrow.i; 1458 ridx = a->compressedrow.rindex; 1459 for (i=0; i<m; i++) { 1460 n = ii[i+1] - ii[i]; 1461 aj = a->j + ii[i]; 1462 aa = a->a + ii[i]; 1463 sum = y[*ridx]; 1464 PetscSparseDenseMaxDot(sum,x,aa,aj,n); 1465 z[*ridx++] = sum; 1466 } 1467 } else { /* do not use compressed row format */ 1468 for (i=0; i<m; i++) { 1469 n = ii[i+1] - ii[i]; 1470 aj = a->j + ii[i]; 1471 aa = a->a + ii[i]; 1472 sum = y[i]; 1473 PetscSparseDenseMaxDot(sum,x,aa,aj,n); 1474 z[i] = sum; 1475 } 1476 } 1477 ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr); 1478 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1479 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1480 if (zz != yy) { 1481 ierr = VecRestoreArray(zz,&z);CHKERRQ(ierr); 1482 } 1483 PetscFunctionReturn(0); 1484 } 1485 1486 #include <../src/mat/impls/aij/seq/ftn-kernels/fmultadd.h> 1487 #undef __FUNCT__ 1488 #define __FUNCT__ "MatMultAdd_SeqAIJ" 1489 PetscErrorCode MatMultAdd_SeqAIJ(Mat A,Vec xx,Vec yy,Vec zz) 1490 { 1491 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1492 PetscScalar *y,*z; 1493 const PetscScalar *x; 1494 const MatScalar *aa; 1495 PetscErrorCode ierr; 1496 PetscInt m = A->rmap->n,*aj,*ii; 1497 PetscInt n,i,*ridx=NULL; 1498 PetscScalar sum; 1499 PetscBool usecprow=a->compressedrow.use; 1500 1501 PetscFunctionBegin; 1502 ierr = VecGetArrayRead(xx,&x);CHKERRQ(ierr); 1503 ierr = VecGetArray(yy,&y);CHKERRQ(ierr); 1504 if (zz != yy) { 1505 ierr = VecGetArray(zz,&z);CHKERRQ(ierr); 1506 } else { 1507 z = y; 1508 } 1509 1510 aj = a->j; 1511 aa = a->a; 1512 ii = a->i; 1513 if (usecprow) { /* use compressed row format */ 1514 if (zz != yy) { 1515 ierr = PetscMemcpy(z,y,m*sizeof(PetscScalar));CHKERRQ(ierr); 1516 } 1517 m = a->compressedrow.nrows; 1518 ii = a->compressedrow.i; 1519 ridx = a->compressedrow.rindex; 1520 for (i=0; i<m; i++) { 1521 n = ii[i+1] - ii[i]; 1522 aj = a->j + ii[i]; 1523 aa = a->a + ii[i]; 1524 sum = y[*ridx]; 1525 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1526 z[*ridx++] = sum; 1527 } 1528 } else { /* do not use compressed row format */ 1529 #if defined(PETSC_USE_FORTRAN_KERNEL_MULTADDAIJ) 1530 fortranmultaddaij_(&m,x,ii,aj,aa,y,z); 1531 #else 1532 for (i=0; i<m; i++) { 1533 n = ii[i+1] - ii[i]; 1534 aj = a->j + ii[i]; 1535 aa = a->a + ii[i]; 1536 sum = y[i]; 1537 PetscSparseDensePlusDot(sum,x,aa,aj,n); 1538 z[i] = sum; 1539 } 1540 #endif 1541 } 1542 ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr); 1543 ierr = VecRestoreArrayRead(xx,&x);CHKERRQ(ierr); 1544 ierr = VecRestoreArray(yy,&y);CHKERRQ(ierr); 1545 if (zz != yy) { 1546 ierr = VecRestoreArray(zz,&z);CHKERRQ(ierr); 1547 } 1548 #if defined(PETSC_HAVE_CUSP) 1549 /* 1550 ierr = VecView(xx,0);CHKERRQ(ierr); 1551 ierr = VecView(zz,0);CHKERRQ(ierr); 1552 ierr = MatView(A,0);CHKERRQ(ierr); 1553 */ 1554 #endif 1555 PetscFunctionReturn(0); 1556 } 1557 1558 /* 1559 Adds diagonal pointers to sparse matrix structure. 1560 */ 1561 #undef __FUNCT__ 1562 #define __FUNCT__ "MatMarkDiagonal_SeqAIJ" 1563 PetscErrorCode MatMarkDiagonal_SeqAIJ(Mat A) 1564 { 1565 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1566 PetscErrorCode ierr; 1567 PetscInt i,j,m = A->rmap->n; 1568 1569 PetscFunctionBegin; 1570 if (!a->diag) { 1571 ierr = PetscMalloc(m*sizeof(PetscInt),&a->diag);CHKERRQ(ierr); 1572 ierr = PetscLogObjectMemory((PetscObject)A, m*sizeof(PetscInt));CHKERRQ(ierr); 1573 } 1574 for (i=0; i<A->rmap->n; i++) { 1575 a->diag[i] = a->i[i+1]; 1576 for (j=a->i[i]; j<a->i[i+1]; j++) { 1577 if (a->j[j] == i) { 1578 a->diag[i] = j; 1579 break; 1580 } 1581 } 1582 } 1583 PetscFunctionReturn(0); 1584 } 1585 1586 /* 1587 Checks for missing diagonals 1588 */ 1589 #undef __FUNCT__ 1590 #define __FUNCT__ "MatMissingDiagonal_SeqAIJ" 1591 PetscErrorCode MatMissingDiagonal_SeqAIJ(Mat A,PetscBool *missing,PetscInt *d) 1592 { 1593 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1594 PetscInt *diag,*jj = a->j,i; 1595 1596 PetscFunctionBegin; 1597 *missing = PETSC_FALSE; 1598 if (A->rmap->n > 0 && !jj) { 1599 *missing = PETSC_TRUE; 1600 if (d) *d = 0; 1601 PetscInfo(A,"Matrix has no entries therefore is missing diagonal"); 1602 } else { 1603 diag = a->diag; 1604 for (i=0; i<A->rmap->n; i++) { 1605 if (jj[diag[i]] != i) { 1606 *missing = PETSC_TRUE; 1607 if (d) *d = i; 1608 PetscInfo1(A,"Matrix is missing diagonal number %D",i); 1609 break; 1610 } 1611 } 1612 } 1613 PetscFunctionReturn(0); 1614 } 1615 1616 #undef __FUNCT__ 1617 #define __FUNCT__ "MatInvertDiagonal_SeqAIJ" 1618 PetscErrorCode MatInvertDiagonal_SeqAIJ(Mat A,PetscScalar omega,PetscScalar fshift) 1619 { 1620 Mat_SeqAIJ *a = (Mat_SeqAIJ*) A->data; 1621 PetscErrorCode ierr; 1622 PetscInt i,*diag,m = A->rmap->n; 1623 MatScalar *v = a->a; 1624 PetscScalar *idiag,*mdiag; 1625 1626 PetscFunctionBegin; 1627 if (a->idiagvalid) PetscFunctionReturn(0); 1628 ierr = MatMarkDiagonal_SeqAIJ(A);CHKERRQ(ierr); 1629 diag = a->diag; 1630 if (!a->idiag) { 1631 ierr = PetscMalloc3(m,PetscScalar,&a->idiag,m,PetscScalar,&a->mdiag,m,PetscScalar,&a->ssor_work);CHKERRQ(ierr); 1632 ierr = PetscLogObjectMemory((PetscObject)A, 3*m*sizeof(PetscScalar));CHKERRQ(ierr); 1633 v = a->a; 1634 } 1635 mdiag = a->mdiag; 1636 idiag = a->idiag; 1637 1638 if (omega == 1.0 && !PetscAbsScalar(fshift)) { 1639 for (i=0; i<m; i++) { 1640 mdiag[i] = v[diag[i]]; 1641 if (!PetscAbsScalar(mdiag[i])) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Zero diagonal on row %D",i); 1642 idiag[i] = 1.0/v[diag[i]]; 1643 } 1644 ierr = PetscLogFlops(m);CHKERRQ(ierr); 1645 } else { 1646 for (i=0; i<m; i++) { 1647 mdiag[i] = v[diag[i]]; 1648 idiag[i] = omega/(fshift + v[diag[i]]); 1649 } 1650 ierr = PetscLogFlops(2.0*m);CHKERRQ(ierr); 1651 } 1652 a->idiagvalid = PETSC_TRUE; 1653 PetscFunctionReturn(0); 1654 } 1655 1656 #include <../src/mat/impls/aij/seq/ftn-kernels/frelax.h> 1657 #undef __FUNCT__ 1658 #define __FUNCT__ "MatSOR_SeqAIJ" 1659 PetscErrorCode MatSOR_SeqAIJ(Mat A,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx) 1660 { 1661 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1662 PetscScalar *x,d,sum,*t,scale; 1663 const MatScalar *v = a->a,*idiag=0,*mdiag; 1664 const PetscScalar *b, *bs,*xb, *ts; 1665 PetscErrorCode ierr; 1666 PetscInt n = A->cmap->n,m = A->rmap->n,i; 1667 const PetscInt *idx,*diag; 1668 1669 PetscFunctionBegin; 1670 its = its*lits; 1671 1672 if (fshift != a->fshift || omega != a->omega) a->idiagvalid = PETSC_FALSE; /* must recompute idiag[] */ 1673 if (!a->idiagvalid) {ierr = MatInvertDiagonal_SeqAIJ(A,omega,fshift);CHKERRQ(ierr);} 1674 a->fshift = fshift; 1675 a->omega = omega; 1676 1677 diag = a->diag; 1678 t = a->ssor_work; 1679 idiag = a->idiag; 1680 mdiag = a->mdiag; 1681 1682 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 1683 ierr = VecGetArrayRead(bb,&b);CHKERRQ(ierr); 1684 /* We count flops by assuming the upper triangular and lower triangular parts have the same number of nonzeros */ 1685 if (flag == SOR_APPLY_UPPER) { 1686 /* apply (U + D/omega) to the vector */ 1687 bs = b; 1688 for (i=0; i<m; i++) { 1689 d = fshift + mdiag[i]; 1690 n = a->i[i+1] - diag[i] - 1; 1691 idx = a->j + diag[i] + 1; 1692 v = a->a + diag[i] + 1; 1693 sum = b[i]*d/omega; 1694 PetscSparseDensePlusDot(sum,bs,v,idx,n); 1695 x[i] = sum; 1696 } 1697 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 1698 ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr); 1699 ierr = PetscLogFlops(a->nz);CHKERRQ(ierr); 1700 PetscFunctionReturn(0); 1701 } 1702 1703 if (flag == SOR_APPLY_LOWER) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"SOR_APPLY_LOWER is not implemented"); 1704 else if (flag & SOR_EISENSTAT) { 1705 /* Let A = L + U + D; where L is lower trianglar, 1706 U is upper triangular, E = D/omega; This routine applies 1707 1708 (L + E)^{-1} A (U + E)^{-1} 1709 1710 to a vector efficiently using Eisenstat's trick. 1711 */ 1712 scale = (2.0/omega) - 1.0; 1713 1714 /* x = (E + U)^{-1} b */ 1715 for (i=m-1; i>=0; i--) { 1716 n = a->i[i+1] - diag[i] - 1; 1717 idx = a->j + diag[i] + 1; 1718 v = a->a + diag[i] + 1; 1719 sum = b[i]; 1720 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1721 x[i] = sum*idiag[i]; 1722 } 1723 1724 /* t = b - (2*E - D)x */ 1725 v = a->a; 1726 for (i=0; i<m; i++) t[i] = b[i] - scale*(v[*diag++])*x[i]; 1727 1728 /* t = (E + L)^{-1}t */ 1729 ts = t; 1730 diag = a->diag; 1731 for (i=0; i<m; i++) { 1732 n = diag[i] - a->i[i]; 1733 idx = a->j + a->i[i]; 1734 v = a->a + a->i[i]; 1735 sum = t[i]; 1736 PetscSparseDenseMinusDot(sum,ts,v,idx,n); 1737 t[i] = sum*idiag[i]; 1738 /* x = x + t */ 1739 x[i] += t[i]; 1740 } 1741 1742 ierr = PetscLogFlops(6.0*m-1 + 2.0*a->nz);CHKERRQ(ierr); 1743 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 1744 ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr); 1745 PetscFunctionReturn(0); 1746 } 1747 if (flag & SOR_ZERO_INITIAL_GUESS) { 1748 if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) { 1749 for (i=0; i<m; i++) { 1750 n = diag[i] - a->i[i]; 1751 idx = a->j + a->i[i]; 1752 v = a->a + a->i[i]; 1753 sum = b[i]; 1754 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1755 t[i] = sum; 1756 x[i] = sum*idiag[i]; 1757 } 1758 xb = t; 1759 ierr = PetscLogFlops(a->nz);CHKERRQ(ierr); 1760 } else xb = b; 1761 if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) { 1762 for (i=m-1; i>=0; i--) { 1763 n = a->i[i+1] - diag[i] - 1; 1764 idx = a->j + diag[i] + 1; 1765 v = a->a + diag[i] + 1; 1766 sum = xb[i]; 1767 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1768 if (xb == b) { 1769 x[i] = sum*idiag[i]; 1770 } else { 1771 x[i] = (1-omega)*x[i] + sum*idiag[i]; /* omega in idiag */ 1772 } 1773 } 1774 ierr = PetscLogFlops(a->nz);CHKERRQ(ierr); /* assumes 1/2 in upper */ 1775 } 1776 its--; 1777 } 1778 while (its--) { 1779 if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) { 1780 for (i=0; i<m; i++) { 1781 /* lower */ 1782 n = diag[i] - a->i[i]; 1783 idx = a->j + a->i[i]; 1784 v = a->a + a->i[i]; 1785 sum = b[i]; 1786 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1787 t[i] = sum; /* save application of the lower-triangular part */ 1788 /* upper */ 1789 n = a->i[i+1] - diag[i] - 1; 1790 idx = a->j + diag[i] + 1; 1791 v = a->a + diag[i] + 1; 1792 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1793 x[i] = (1. - omega)*x[i] + sum*idiag[i]; /* omega in idiag */ 1794 } 1795 xb = t; 1796 ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr); 1797 } else xb = b; 1798 if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP) { 1799 for (i=m-1; i>=0; i--) { 1800 sum = xb[i]; 1801 if (xb == b) { 1802 /* whole matrix (no checkpointing available) */ 1803 n = a->i[i+1] - a->i[i]; 1804 idx = a->j + a->i[i]; 1805 v = a->a + a->i[i]; 1806 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1807 x[i] = (1. - omega)*x[i] + (sum + mdiag[i]*x[i])*idiag[i]; 1808 } else { /* lower-triangular part has been saved, so only apply upper-triangular */ 1809 n = a->i[i+1] - diag[i] - 1; 1810 idx = a->j + diag[i] + 1; 1811 v = a->a + diag[i] + 1; 1812 PetscSparseDenseMinusDot(sum,x,v,idx,n); 1813 x[i] = (1. - omega)*x[i] + sum*idiag[i]; /* omega in idiag */ 1814 } 1815 } 1816 if (xb == b) { 1817 ierr = PetscLogFlops(2.0*a->nz);CHKERRQ(ierr); 1818 } else { 1819 ierr = PetscLogFlops(a->nz);CHKERRQ(ierr); /* assumes 1/2 in upper */ 1820 } 1821 } 1822 } 1823 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 1824 ierr = VecRestoreArrayRead(bb,&b);CHKERRQ(ierr); 1825 PetscFunctionReturn(0); 1826 } 1827 1828 1829 #undef __FUNCT__ 1830 #define __FUNCT__ "MatGetInfo_SeqAIJ" 1831 PetscErrorCode MatGetInfo_SeqAIJ(Mat A,MatInfoType flag,MatInfo *info) 1832 { 1833 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1834 1835 PetscFunctionBegin; 1836 info->block_size = 1.0; 1837 info->nz_allocated = (double)a->maxnz; 1838 info->nz_used = (double)a->nz; 1839 info->nz_unneeded = (double)(a->maxnz - a->nz); 1840 info->assemblies = (double)A->num_ass; 1841 info->mallocs = (double)A->info.mallocs; 1842 info->memory = ((PetscObject)A)->mem; 1843 if (A->factortype) { 1844 info->fill_ratio_given = A->info.fill_ratio_given; 1845 info->fill_ratio_needed = A->info.fill_ratio_needed; 1846 info->factor_mallocs = A->info.factor_mallocs; 1847 } else { 1848 info->fill_ratio_given = 0; 1849 info->fill_ratio_needed = 0; 1850 info->factor_mallocs = 0; 1851 } 1852 PetscFunctionReturn(0); 1853 } 1854 1855 #undef __FUNCT__ 1856 #define __FUNCT__ "MatZeroRows_SeqAIJ" 1857 PetscErrorCode MatZeroRows_SeqAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 1858 { 1859 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1860 PetscInt i,m = A->rmap->n - 1,d = 0; 1861 PetscErrorCode ierr; 1862 const PetscScalar *xx; 1863 PetscScalar *bb; 1864 PetscBool missing; 1865 1866 PetscFunctionBegin; 1867 if (x && b) { 1868 ierr = VecGetArrayRead(x,&xx);CHKERRQ(ierr); 1869 ierr = VecGetArray(b,&bb);CHKERRQ(ierr); 1870 for (i=0; i<N; i++) { 1871 if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]); 1872 bb[rows[i]] = diag*xx[rows[i]]; 1873 } 1874 ierr = VecRestoreArrayRead(x,&xx);CHKERRQ(ierr); 1875 ierr = VecRestoreArray(b,&bb);CHKERRQ(ierr); 1876 } 1877 1878 if (a->keepnonzeropattern) { 1879 for (i=0; i<N; i++) { 1880 if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]); 1881 ierr = PetscMemzero(&a->a[a->i[rows[i]]],a->ilen[rows[i]]*sizeof(PetscScalar));CHKERRQ(ierr); 1882 } 1883 if (diag != 0.0) { 1884 ierr = MatMissingDiagonal_SeqAIJ(A,&missing,&d);CHKERRQ(ierr); 1885 if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry in row %D",d); 1886 for (i=0; i<N; i++) { 1887 a->a[a->diag[rows[i]]] = diag; 1888 } 1889 } 1890 A->same_nonzero = PETSC_TRUE; 1891 } else { 1892 if (diag != 0.0) { 1893 for (i=0; i<N; i++) { 1894 if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]); 1895 if (a->ilen[rows[i]] > 0) { 1896 a->ilen[rows[i]] = 1; 1897 a->a[a->i[rows[i]]] = diag; 1898 a->j[a->i[rows[i]]] = rows[i]; 1899 } else { /* in case row was completely empty */ 1900 ierr = MatSetValues_SeqAIJ(A,1,&rows[i],1,&rows[i],&diag,INSERT_VALUES);CHKERRQ(ierr); 1901 } 1902 } 1903 } else { 1904 for (i=0; i<N; i++) { 1905 if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]); 1906 a->ilen[rows[i]] = 0; 1907 } 1908 } 1909 A->same_nonzero = PETSC_FALSE; 1910 } 1911 ierr = MatAssemblyEnd_SeqAIJ(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1912 PetscFunctionReturn(0); 1913 } 1914 1915 #undef __FUNCT__ 1916 #define __FUNCT__ "MatZeroRowsColumns_SeqAIJ" 1917 PetscErrorCode MatZeroRowsColumns_SeqAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag,Vec x,Vec b) 1918 { 1919 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1920 PetscInt i,j,m = A->rmap->n - 1,d = 0; 1921 PetscErrorCode ierr; 1922 PetscBool missing,*zeroed,vecs = PETSC_FALSE; 1923 const PetscScalar *xx; 1924 PetscScalar *bb; 1925 1926 PetscFunctionBegin; 1927 if (x && b) { 1928 ierr = VecGetArrayRead(x,&xx);CHKERRQ(ierr); 1929 ierr = VecGetArray(b,&bb);CHKERRQ(ierr); 1930 vecs = PETSC_TRUE; 1931 } 1932 ierr = PetscMalloc(A->rmap->n*sizeof(PetscBool),&zeroed);CHKERRQ(ierr); 1933 ierr = PetscMemzero(zeroed,A->rmap->n*sizeof(PetscBool));CHKERRQ(ierr); 1934 for (i=0; i<N; i++) { 1935 if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]); 1936 ierr = PetscMemzero(&a->a[a->i[rows[i]]],a->ilen[rows[i]]*sizeof(PetscScalar));CHKERRQ(ierr); 1937 1938 zeroed[rows[i]] = PETSC_TRUE; 1939 } 1940 for (i=0; i<A->rmap->n; i++) { 1941 if (!zeroed[i]) { 1942 for (j=a->i[i]; j<a->i[i+1]; j++) { 1943 if (zeroed[a->j[j]]) { 1944 if (vecs) bb[i] -= a->a[j]*xx[a->j[j]]; 1945 a->a[j] = 0.0; 1946 } 1947 } 1948 } else if (vecs) bb[i] = diag*xx[i]; 1949 } 1950 if (x && b) { 1951 ierr = VecRestoreArrayRead(x,&xx);CHKERRQ(ierr); 1952 ierr = VecRestoreArray(b,&bb);CHKERRQ(ierr); 1953 } 1954 ierr = PetscFree(zeroed);CHKERRQ(ierr); 1955 if (diag != 0.0) { 1956 ierr = MatMissingDiagonal_SeqAIJ(A,&missing,&d);CHKERRQ(ierr); 1957 if (missing) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry in row %D",d); 1958 for (i=0; i<N; i++) { 1959 a->a[a->diag[rows[i]]] = diag; 1960 } 1961 } 1962 A->same_nonzero = PETSC_TRUE; 1963 ierr = MatAssemblyEnd_SeqAIJ(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1964 PetscFunctionReturn(0); 1965 } 1966 1967 #undef __FUNCT__ 1968 #define __FUNCT__ "MatGetRow_SeqAIJ" 1969 PetscErrorCode MatGetRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 1970 { 1971 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 1972 PetscInt *itmp; 1973 1974 PetscFunctionBegin; 1975 if (row < 0 || row >= A->rmap->n) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Row %D out of range",row); 1976 1977 *nz = a->i[row+1] - a->i[row]; 1978 if (v) *v = a->a + a->i[row]; 1979 if (idx) { 1980 itmp = a->j + a->i[row]; 1981 if (*nz) *idx = itmp; 1982 else *idx = 0; 1983 } 1984 PetscFunctionReturn(0); 1985 } 1986 1987 /* remove this function? */ 1988 #undef __FUNCT__ 1989 #define __FUNCT__ "MatRestoreRow_SeqAIJ" 1990 PetscErrorCode MatRestoreRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 1991 { 1992 PetscFunctionBegin; 1993 PetscFunctionReturn(0); 1994 } 1995 1996 #undef __FUNCT__ 1997 #define __FUNCT__ "MatNorm_SeqAIJ" 1998 PetscErrorCode MatNorm_SeqAIJ(Mat A,NormType type,PetscReal *nrm) 1999 { 2000 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2001 MatScalar *v = a->a; 2002 PetscReal sum = 0.0; 2003 PetscErrorCode ierr; 2004 PetscInt i,j; 2005 2006 PetscFunctionBegin; 2007 if (type == NORM_FROBENIUS) { 2008 for (i=0; i<a->nz; i++) { 2009 sum += PetscRealPart(PetscConj(*v)*(*v)); v++; 2010 } 2011 *nrm = PetscSqrtReal(sum); 2012 } else if (type == NORM_1) { 2013 PetscReal *tmp; 2014 PetscInt *jj = a->j; 2015 ierr = PetscMalloc((A->cmap->n+1)*sizeof(PetscReal),&tmp);CHKERRQ(ierr); 2016 ierr = PetscMemzero(tmp,A->cmap->n*sizeof(PetscReal));CHKERRQ(ierr); 2017 *nrm = 0.0; 2018 for (j=0; j<a->nz; j++) { 2019 tmp[*jj++] += PetscAbsScalar(*v); v++; 2020 } 2021 for (j=0; j<A->cmap->n; j++) { 2022 if (tmp[j] > *nrm) *nrm = tmp[j]; 2023 } 2024 ierr = PetscFree(tmp);CHKERRQ(ierr); 2025 } else if (type == NORM_INFINITY) { 2026 *nrm = 0.0; 2027 for (j=0; j<A->rmap->n; j++) { 2028 v = a->a + a->i[j]; 2029 sum = 0.0; 2030 for (i=0; i<a->i[j+1]-a->i[j]; i++) { 2031 sum += PetscAbsScalar(*v); v++; 2032 } 2033 if (sum > *nrm) *nrm = sum; 2034 } 2035 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"No support for two norm"); 2036 PetscFunctionReturn(0); 2037 } 2038 2039 /* Merged from MatGetSymbolicTranspose_SeqAIJ() - replace MatGetSymbolicTranspose_SeqAIJ()? */ 2040 #undef __FUNCT__ 2041 #define __FUNCT__ "MatTransposeSymbolic_SeqAIJ" 2042 PetscErrorCode MatTransposeSymbolic_SeqAIJ(Mat A,Mat *B) 2043 { 2044 PetscErrorCode ierr; 2045 PetscInt i,j,anzj; 2046 Mat_SeqAIJ *a=(Mat_SeqAIJ*)A->data,*b; 2047 PetscInt an=A->cmap->N,am=A->rmap->N; 2048 PetscInt *ati,*atj,*atfill,*ai=a->i,*aj=a->j; 2049 2050 PetscFunctionBegin; 2051 /* Allocate space for symbolic transpose info and work array */ 2052 ierr = PetscMalloc((an+1)*sizeof(PetscInt),&ati);CHKERRQ(ierr); 2053 ierr = PetscMalloc(ai[am]*sizeof(PetscInt),&atj);CHKERRQ(ierr); 2054 ierr = PetscMalloc(an*sizeof(PetscInt),&atfill);CHKERRQ(ierr); 2055 ierr = PetscMemzero(ati,(an+1)*sizeof(PetscInt));CHKERRQ(ierr); 2056 2057 /* Walk through aj and count ## of non-zeros in each row of A^T. */ 2058 /* Note: offset by 1 for fast conversion into csr format. */ 2059 for (i=0;i<ai[am];i++) ati[aj[i]+1] += 1; 2060 /* Form ati for csr format of A^T. */ 2061 for (i=0;i<an;i++) ati[i+1] += ati[i]; 2062 2063 /* Copy ati into atfill so we have locations of the next free space in atj */ 2064 ierr = PetscMemcpy(atfill,ati,an*sizeof(PetscInt));CHKERRQ(ierr); 2065 2066 /* Walk through A row-wise and mark nonzero entries of A^T. */ 2067 for (i=0;i<am;i++) { 2068 anzj = ai[i+1] - ai[i]; 2069 for (j=0;j<anzj;j++) { 2070 atj[atfill[*aj]] = i; 2071 atfill[*aj++] += 1; 2072 } 2073 } 2074 2075 /* Clean up temporary space and complete requests. */ 2076 ierr = PetscFree(atfill);CHKERRQ(ierr); 2077 ierr = MatCreateSeqAIJWithArrays(PetscObjectComm((PetscObject)A),an,am,ati,atj,NULL,B);CHKERRQ(ierr); 2078 2079 (*B)->rmap->bs = A->cmap->bs; 2080 (*B)->cmap->bs = A->rmap->bs; 2081 2082 b = (Mat_SeqAIJ*)((*B)->data); 2083 b->free_a = PETSC_FALSE; 2084 b->free_ij = PETSC_TRUE; 2085 b->nonew = 0; 2086 PetscFunctionReturn(0); 2087 } 2088 2089 #undef __FUNCT__ 2090 #define __FUNCT__ "MatTranspose_SeqAIJ" 2091 PetscErrorCode MatTranspose_SeqAIJ(Mat A,MatReuse reuse,Mat *B) 2092 { 2093 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2094 Mat C; 2095 PetscErrorCode ierr; 2096 PetscInt i,*aj = a->j,*ai = a->i,m = A->rmap->n,len,*col; 2097 MatScalar *array = a->a; 2098 2099 PetscFunctionBegin; 2100 if (reuse == MAT_REUSE_MATRIX && A == *B && m != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Square matrix only for in-place"); 2101 2102 if (reuse == MAT_INITIAL_MATRIX || *B == A) { 2103 ierr = PetscMalloc((1+A->cmap->n)*sizeof(PetscInt),&col);CHKERRQ(ierr); 2104 ierr = PetscMemzero(col,(1+A->cmap->n)*sizeof(PetscInt));CHKERRQ(ierr); 2105 2106 for (i=0; i<ai[m]; i++) col[aj[i]] += 1; 2107 ierr = MatCreate(PetscObjectComm((PetscObject)A),&C);CHKERRQ(ierr); 2108 ierr = MatSetSizes(C,A->cmap->n,m,A->cmap->n,m);CHKERRQ(ierr); 2109 ierr = MatSetBlockSizes(C,A->cmap->bs,A->rmap->bs);CHKERRQ(ierr); 2110 ierr = MatSetType(C,((PetscObject)A)->type_name);CHKERRQ(ierr); 2111 ierr = MatSeqAIJSetPreallocation_SeqAIJ(C,0,col);CHKERRQ(ierr); 2112 ierr = PetscFree(col);CHKERRQ(ierr); 2113 } else { 2114 C = *B; 2115 } 2116 2117 for (i=0; i<m; i++) { 2118 len = ai[i+1]-ai[i]; 2119 ierr = MatSetValues_SeqAIJ(C,len,aj,1,&i,array,INSERT_VALUES);CHKERRQ(ierr); 2120 array += len; 2121 aj += len; 2122 } 2123 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2124 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2125 2126 if (reuse == MAT_INITIAL_MATRIX || *B != A) { 2127 *B = C; 2128 } else { 2129 ierr = MatHeaderMerge(A,C);CHKERRQ(ierr); 2130 } 2131 PetscFunctionReturn(0); 2132 } 2133 2134 #undef __FUNCT__ 2135 #define __FUNCT__ "MatIsTranspose_SeqAIJ" 2136 PetscErrorCode MatIsTranspose_SeqAIJ(Mat A,Mat B,PetscReal tol,PetscBool *f) 2137 { 2138 Mat_SeqAIJ *aij = (Mat_SeqAIJ*) A->data,*bij = (Mat_SeqAIJ*) A->data; 2139 PetscInt *adx,*bdx,*aii,*bii,*aptr,*bptr; 2140 MatScalar *va,*vb; 2141 PetscErrorCode ierr; 2142 PetscInt ma,na,mb,nb, i; 2143 2144 PetscFunctionBegin; 2145 bij = (Mat_SeqAIJ*) B->data; 2146 2147 ierr = MatGetSize(A,&ma,&na);CHKERRQ(ierr); 2148 ierr = MatGetSize(B,&mb,&nb);CHKERRQ(ierr); 2149 if (ma!=nb || na!=mb) { 2150 *f = PETSC_FALSE; 2151 PetscFunctionReturn(0); 2152 } 2153 aii = aij->i; bii = bij->i; 2154 adx = aij->j; bdx = bij->j; 2155 va = aij->a; vb = bij->a; 2156 ierr = PetscMalloc(ma*sizeof(PetscInt),&aptr);CHKERRQ(ierr); 2157 ierr = PetscMalloc(mb*sizeof(PetscInt),&bptr);CHKERRQ(ierr); 2158 for (i=0; i<ma; i++) aptr[i] = aii[i]; 2159 for (i=0; i<mb; i++) bptr[i] = bii[i]; 2160 2161 *f = PETSC_TRUE; 2162 for (i=0; i<ma; i++) { 2163 while (aptr[i]<aii[i+1]) { 2164 PetscInt idc,idr; 2165 PetscScalar vc,vr; 2166 /* column/row index/value */ 2167 idc = adx[aptr[i]]; 2168 idr = bdx[bptr[idc]]; 2169 vc = va[aptr[i]]; 2170 vr = vb[bptr[idc]]; 2171 if (i!=idr || PetscAbsScalar(vc-vr) > tol) { 2172 *f = PETSC_FALSE; 2173 goto done; 2174 } else { 2175 aptr[i]++; 2176 if (B || i!=idc) bptr[idc]++; 2177 } 2178 } 2179 } 2180 done: 2181 ierr = PetscFree(aptr);CHKERRQ(ierr); 2182 ierr = PetscFree(bptr);CHKERRQ(ierr); 2183 PetscFunctionReturn(0); 2184 } 2185 2186 #undef __FUNCT__ 2187 #define __FUNCT__ "MatIsHermitianTranspose_SeqAIJ" 2188 PetscErrorCode MatIsHermitianTranspose_SeqAIJ(Mat A,Mat B,PetscReal tol,PetscBool *f) 2189 { 2190 Mat_SeqAIJ *aij = (Mat_SeqAIJ*) A->data,*bij = (Mat_SeqAIJ*) A->data; 2191 PetscInt *adx,*bdx,*aii,*bii,*aptr,*bptr; 2192 MatScalar *va,*vb; 2193 PetscErrorCode ierr; 2194 PetscInt ma,na,mb,nb, i; 2195 2196 PetscFunctionBegin; 2197 bij = (Mat_SeqAIJ*) B->data; 2198 2199 ierr = MatGetSize(A,&ma,&na);CHKERRQ(ierr); 2200 ierr = MatGetSize(B,&mb,&nb);CHKERRQ(ierr); 2201 if (ma!=nb || na!=mb) { 2202 *f = PETSC_FALSE; 2203 PetscFunctionReturn(0); 2204 } 2205 aii = aij->i; bii = bij->i; 2206 adx = aij->j; bdx = bij->j; 2207 va = aij->a; vb = bij->a; 2208 ierr = PetscMalloc(ma*sizeof(PetscInt),&aptr);CHKERRQ(ierr); 2209 ierr = PetscMalloc(mb*sizeof(PetscInt),&bptr);CHKERRQ(ierr); 2210 for (i=0; i<ma; i++) aptr[i] = aii[i]; 2211 for (i=0; i<mb; i++) bptr[i] = bii[i]; 2212 2213 *f = PETSC_TRUE; 2214 for (i=0; i<ma; i++) { 2215 while (aptr[i]<aii[i+1]) { 2216 PetscInt idc,idr; 2217 PetscScalar vc,vr; 2218 /* column/row index/value */ 2219 idc = adx[aptr[i]]; 2220 idr = bdx[bptr[idc]]; 2221 vc = va[aptr[i]]; 2222 vr = vb[bptr[idc]]; 2223 if (i!=idr || PetscAbsScalar(vc-PetscConj(vr)) > tol) { 2224 *f = PETSC_FALSE; 2225 goto done; 2226 } else { 2227 aptr[i]++; 2228 if (B || i!=idc) bptr[idc]++; 2229 } 2230 } 2231 } 2232 done: 2233 ierr = PetscFree(aptr);CHKERRQ(ierr); 2234 ierr = PetscFree(bptr);CHKERRQ(ierr); 2235 PetscFunctionReturn(0); 2236 } 2237 2238 #undef __FUNCT__ 2239 #define __FUNCT__ "MatIsSymmetric_SeqAIJ" 2240 PetscErrorCode MatIsSymmetric_SeqAIJ(Mat A,PetscReal tol,PetscBool *f) 2241 { 2242 PetscErrorCode ierr; 2243 2244 PetscFunctionBegin; 2245 ierr = MatIsTranspose_SeqAIJ(A,A,tol,f);CHKERRQ(ierr); 2246 PetscFunctionReturn(0); 2247 } 2248 2249 #undef __FUNCT__ 2250 #define __FUNCT__ "MatIsHermitian_SeqAIJ" 2251 PetscErrorCode MatIsHermitian_SeqAIJ(Mat A,PetscReal tol,PetscBool *f) 2252 { 2253 PetscErrorCode ierr; 2254 2255 PetscFunctionBegin; 2256 ierr = MatIsHermitianTranspose_SeqAIJ(A,A,tol,f);CHKERRQ(ierr); 2257 PetscFunctionReturn(0); 2258 } 2259 2260 #undef __FUNCT__ 2261 #define __FUNCT__ "MatDiagonalScale_SeqAIJ" 2262 PetscErrorCode MatDiagonalScale_SeqAIJ(Mat A,Vec ll,Vec rr) 2263 { 2264 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2265 PetscScalar *l,*r,x; 2266 MatScalar *v; 2267 PetscErrorCode ierr; 2268 PetscInt i,j,m = A->rmap->n,n = A->cmap->n,M,nz = a->nz,*jj; 2269 2270 PetscFunctionBegin; 2271 if (ll) { 2272 /* The local size is used so that VecMPI can be passed to this routine 2273 by MatDiagonalScale_MPIAIJ */ 2274 ierr = VecGetLocalSize(ll,&m);CHKERRQ(ierr); 2275 if (m != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Left scaling vector wrong length"); 2276 ierr = VecGetArray(ll,&l);CHKERRQ(ierr); 2277 v = a->a; 2278 for (i=0; i<m; i++) { 2279 x = l[i]; 2280 M = a->i[i+1] - a->i[i]; 2281 for (j=0; j<M; j++) (*v++) *= x; 2282 } 2283 ierr = VecRestoreArray(ll,&l);CHKERRQ(ierr); 2284 ierr = PetscLogFlops(nz);CHKERRQ(ierr); 2285 } 2286 if (rr) { 2287 ierr = VecGetLocalSize(rr,&n);CHKERRQ(ierr); 2288 if (n != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Right scaling vector wrong length"); 2289 ierr = VecGetArray(rr,&r);CHKERRQ(ierr); 2290 v = a->a; jj = a->j; 2291 for (i=0; i<nz; i++) (*v++) *= r[*jj++]; 2292 ierr = VecRestoreArray(rr,&r);CHKERRQ(ierr); 2293 ierr = PetscLogFlops(nz);CHKERRQ(ierr); 2294 } 2295 ierr = MatSeqAIJInvalidateDiagonal(A);CHKERRQ(ierr); 2296 PetscFunctionReturn(0); 2297 } 2298 2299 #undef __FUNCT__ 2300 #define __FUNCT__ "MatGetSubMatrix_SeqAIJ" 2301 PetscErrorCode MatGetSubMatrix_SeqAIJ(Mat A,IS isrow,IS iscol,PetscInt csize,MatReuse scall,Mat *B) 2302 { 2303 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data,*c; 2304 PetscErrorCode ierr; 2305 PetscInt *smap,i,k,kstart,kend,oldcols = A->cmap->n,*lens; 2306 PetscInt row,mat_i,*mat_j,tcol,first,step,*mat_ilen,sum,lensi; 2307 const PetscInt *irow,*icol; 2308 PetscInt nrows,ncols; 2309 PetscInt *starts,*j_new,*i_new,*aj = a->j,*ai = a->i,ii,*ailen = a->ilen; 2310 MatScalar *a_new,*mat_a; 2311 Mat C; 2312 PetscBool stride,sorted; 2313 2314 PetscFunctionBegin; 2315 ierr = ISSorted(isrow,&sorted);CHKERRQ(ierr); 2316 if (!sorted) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"ISrow is not sorted"); 2317 ierr = ISSorted(iscol,&sorted);CHKERRQ(ierr); 2318 if (!sorted) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"IScol is not sorted"); 2319 2320 ierr = ISGetIndices(isrow,&irow);CHKERRQ(ierr); 2321 ierr = ISGetLocalSize(isrow,&nrows);CHKERRQ(ierr); 2322 ierr = ISGetLocalSize(iscol,&ncols);CHKERRQ(ierr); 2323 2324 ierr = ISStrideGetInfo(iscol,&first,&step);CHKERRQ(ierr); 2325 ierr = PetscObjectTypeCompare((PetscObject)iscol,ISSTRIDE,&stride);CHKERRQ(ierr); 2326 if (stride && step == 1) { 2327 /* special case of contiguous rows */ 2328 ierr = PetscMalloc2(nrows,PetscInt,&lens,nrows,PetscInt,&starts);CHKERRQ(ierr); 2329 /* loop over new rows determining lens and starting points */ 2330 for (i=0; i<nrows; i++) { 2331 kstart = ai[irow[i]]; 2332 kend = kstart + ailen[irow[i]]; 2333 for (k=kstart; k<kend; k++) { 2334 if (aj[k] >= first) { 2335 starts[i] = k; 2336 break; 2337 } 2338 } 2339 sum = 0; 2340 while (k < kend) { 2341 if (aj[k++] >= first+ncols) break; 2342 sum++; 2343 } 2344 lens[i] = sum; 2345 } 2346 /* create submatrix */ 2347 if (scall == MAT_REUSE_MATRIX) { 2348 PetscInt n_cols,n_rows; 2349 ierr = MatGetSize(*B,&n_rows,&n_cols);CHKERRQ(ierr); 2350 if (n_rows != nrows || n_cols != ncols) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Reused submatrix wrong size"); 2351 ierr = MatZeroEntries(*B);CHKERRQ(ierr); 2352 C = *B; 2353 } else { 2354 PetscInt rbs,cbs; 2355 ierr = MatCreate(PetscObjectComm((PetscObject)A),&C);CHKERRQ(ierr); 2356 ierr = MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 2357 ierr = ISGetBlockSize(isrow,&rbs);CHKERRQ(ierr); 2358 ierr = ISGetBlockSize(iscol,&cbs);CHKERRQ(ierr); 2359 ierr = MatSetBlockSizes(C,rbs,cbs);CHKERRQ(ierr); 2360 ierr = MatSetType(C,((PetscObject)A)->type_name);CHKERRQ(ierr); 2361 ierr = MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);CHKERRQ(ierr); 2362 } 2363 c = (Mat_SeqAIJ*)C->data; 2364 2365 /* loop over rows inserting into submatrix */ 2366 a_new = c->a; 2367 j_new = c->j; 2368 i_new = c->i; 2369 2370 for (i=0; i<nrows; i++) { 2371 ii = starts[i]; 2372 lensi = lens[i]; 2373 for (k=0; k<lensi; k++) { 2374 *j_new++ = aj[ii+k] - first; 2375 } 2376 ierr = PetscMemcpy(a_new,a->a + starts[i],lensi*sizeof(PetscScalar));CHKERRQ(ierr); 2377 a_new += lensi; 2378 i_new[i+1] = i_new[i] + lensi; 2379 c->ilen[i] = lensi; 2380 } 2381 ierr = PetscFree2(lens,starts);CHKERRQ(ierr); 2382 } else { 2383 ierr = ISGetIndices(iscol,&icol);CHKERRQ(ierr); 2384 ierr = PetscMalloc(oldcols*sizeof(PetscInt),&smap);CHKERRQ(ierr); 2385 ierr = PetscMemzero(smap,oldcols*sizeof(PetscInt));CHKERRQ(ierr); 2386 ierr = PetscMalloc((1+nrows)*sizeof(PetscInt),&lens);CHKERRQ(ierr); 2387 for (i=0; i<ncols; i++) { 2388 #if defined(PETSC_USE_DEBUG) 2389 if (icol[i] >= oldcols) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Requesting column beyond largest column icol[%D] %D <= A->cmap->n %D",i,icol[i],oldcols); 2390 #endif 2391 smap[icol[i]] = i+1; 2392 } 2393 2394 /* determine lens of each row */ 2395 for (i=0; i<nrows; i++) { 2396 kstart = ai[irow[i]]; 2397 kend = kstart + a->ilen[irow[i]]; 2398 lens[i] = 0; 2399 for (k=kstart; k<kend; k++) { 2400 if (smap[aj[k]]) { 2401 lens[i]++; 2402 } 2403 } 2404 } 2405 /* Create and fill new matrix */ 2406 if (scall == MAT_REUSE_MATRIX) { 2407 PetscBool equal; 2408 2409 c = (Mat_SeqAIJ*)((*B)->data); 2410 if ((*B)->rmap->n != nrows || (*B)->cmap->n != ncols) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong size"); 2411 ierr = PetscMemcmp(c->ilen,lens,(*B)->rmap->n*sizeof(PetscInt),&equal);CHKERRQ(ierr); 2412 if (!equal) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong no of nonzeros"); 2413 ierr = PetscMemzero(c->ilen,(*B)->rmap->n*sizeof(PetscInt));CHKERRQ(ierr); 2414 C = *B; 2415 } else { 2416 PetscInt rbs,cbs; 2417 ierr = MatCreate(PetscObjectComm((PetscObject)A),&C);CHKERRQ(ierr); 2418 ierr = MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 2419 ierr = ISGetBlockSize(isrow,&rbs);CHKERRQ(ierr); 2420 ierr = ISGetBlockSize(iscol,&cbs);CHKERRQ(ierr); 2421 ierr = MatSetBlockSizes(C,rbs,cbs);CHKERRQ(ierr); 2422 ierr = MatSetType(C,((PetscObject)A)->type_name);CHKERRQ(ierr); 2423 ierr = MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);CHKERRQ(ierr); 2424 } 2425 c = (Mat_SeqAIJ*)(C->data); 2426 for (i=0; i<nrows; i++) { 2427 row = irow[i]; 2428 kstart = ai[row]; 2429 kend = kstart + a->ilen[row]; 2430 mat_i = c->i[i]; 2431 mat_j = c->j + mat_i; 2432 mat_a = c->a + mat_i; 2433 mat_ilen = c->ilen + i; 2434 for (k=kstart; k<kend; k++) { 2435 if ((tcol=smap[a->j[k]])) { 2436 *mat_j++ = tcol - 1; 2437 *mat_a++ = a->a[k]; 2438 (*mat_ilen)++; 2439 2440 } 2441 } 2442 } 2443 /* Free work space */ 2444 ierr = ISRestoreIndices(iscol,&icol);CHKERRQ(ierr); 2445 ierr = PetscFree(smap);CHKERRQ(ierr); 2446 ierr = PetscFree(lens);CHKERRQ(ierr); 2447 } 2448 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2449 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2450 2451 ierr = ISRestoreIndices(isrow,&irow);CHKERRQ(ierr); 2452 *B = C; 2453 PetscFunctionReturn(0); 2454 } 2455 2456 #undef __FUNCT__ 2457 #define __FUNCT__ "MatGetMultiProcBlock_SeqAIJ" 2458 PetscErrorCode MatGetMultiProcBlock_SeqAIJ(Mat mat,MPI_Comm subComm,MatReuse scall,Mat *subMat) 2459 { 2460 PetscErrorCode ierr; 2461 Mat B; 2462 2463 PetscFunctionBegin; 2464 if (scall == MAT_INITIAL_MATRIX) { 2465 ierr = MatCreate(subComm,&B);CHKERRQ(ierr); 2466 ierr = MatSetSizes(B,mat->rmap->n,mat->cmap->n,mat->rmap->n,mat->cmap->n);CHKERRQ(ierr); 2467 ierr = MatSetBlockSizes(B,mat->rmap->bs,mat->cmap->bs);CHKERRQ(ierr); 2468 ierr = MatSetType(B,MATSEQAIJ);CHKERRQ(ierr); 2469 ierr = MatDuplicateNoCreate_SeqAIJ(B,mat,MAT_COPY_VALUES,PETSC_TRUE);CHKERRQ(ierr); 2470 *subMat = B; 2471 } else { 2472 ierr = MatCopy_SeqAIJ(mat,*subMat,SAME_NONZERO_PATTERN);CHKERRQ(ierr); 2473 } 2474 PetscFunctionReturn(0); 2475 } 2476 2477 #undef __FUNCT__ 2478 #define __FUNCT__ "MatILUFactor_SeqAIJ" 2479 PetscErrorCode MatILUFactor_SeqAIJ(Mat inA,IS row,IS col,const MatFactorInfo *info) 2480 { 2481 Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data; 2482 PetscErrorCode ierr; 2483 Mat outA; 2484 PetscBool row_identity,col_identity; 2485 2486 PetscFunctionBegin; 2487 if (info->levels != 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Only levels=0 supported for in-place ilu"); 2488 2489 ierr = ISIdentity(row,&row_identity);CHKERRQ(ierr); 2490 ierr = ISIdentity(col,&col_identity);CHKERRQ(ierr); 2491 2492 outA = inA; 2493 outA->factortype = MAT_FACTOR_LU; 2494 2495 ierr = PetscObjectReference((PetscObject)row);CHKERRQ(ierr); 2496 ierr = ISDestroy(&a->row);CHKERRQ(ierr); 2497 2498 a->row = row; 2499 2500 ierr = PetscObjectReference((PetscObject)col);CHKERRQ(ierr); 2501 ierr = ISDestroy(&a->col);CHKERRQ(ierr); 2502 2503 a->col = col; 2504 2505 /* Create the inverse permutation so that it can be used in MatLUFactorNumeric() */ 2506 ierr = ISDestroy(&a->icol);CHKERRQ(ierr); 2507 ierr = ISInvertPermutation(col,PETSC_DECIDE,&a->icol);CHKERRQ(ierr); 2508 ierr = PetscLogObjectParent((PetscObject)inA,(PetscObject)a->icol);CHKERRQ(ierr); 2509 2510 if (!a->solve_work) { /* this matrix may have been factored before */ 2511 ierr = PetscMalloc((inA->rmap->n+1)*sizeof(PetscScalar),&a->solve_work);CHKERRQ(ierr); 2512 ierr = PetscLogObjectMemory((PetscObject)inA, (inA->rmap->n+1)*sizeof(PetscScalar));CHKERRQ(ierr); 2513 } 2514 2515 ierr = MatMarkDiagonal_SeqAIJ(inA);CHKERRQ(ierr); 2516 if (row_identity && col_identity) { 2517 ierr = MatLUFactorNumeric_SeqAIJ_inplace(outA,inA,info);CHKERRQ(ierr); 2518 } else { 2519 ierr = MatLUFactorNumeric_SeqAIJ_InplaceWithPerm(outA,inA,info);CHKERRQ(ierr); 2520 } 2521 PetscFunctionReturn(0); 2522 } 2523 2524 #undef __FUNCT__ 2525 #define __FUNCT__ "MatScale_SeqAIJ" 2526 PetscErrorCode MatScale_SeqAIJ(Mat inA,PetscScalar alpha) 2527 { 2528 Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data; 2529 PetscScalar oalpha = alpha; 2530 PetscErrorCode ierr; 2531 PetscBLASInt one = 1,bnz; 2532 2533 PetscFunctionBegin; 2534 ierr = PetscBLASIntCast(a->nz,&bnz);CHKERRQ(ierr); 2535 PetscStackCallBLAS("BLASscal",BLASscal_(&bnz,&oalpha,a->a,&one)); 2536 ierr = PetscLogFlops(a->nz);CHKERRQ(ierr); 2537 ierr = MatSeqAIJInvalidateDiagonal(inA);CHKERRQ(ierr); 2538 PetscFunctionReturn(0); 2539 } 2540 2541 #undef __FUNCT__ 2542 #define __FUNCT__ "MatGetSubMatrices_SeqAIJ" 2543 PetscErrorCode MatGetSubMatrices_SeqAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[]) 2544 { 2545 PetscErrorCode ierr; 2546 PetscInt i; 2547 2548 PetscFunctionBegin; 2549 if (scall == MAT_INITIAL_MATRIX) { 2550 ierr = PetscMalloc((n+1)*sizeof(Mat),B);CHKERRQ(ierr); 2551 } 2552 2553 for (i=0; i<n; i++) { 2554 ierr = MatGetSubMatrix_SeqAIJ(A,irow[i],icol[i],PETSC_DECIDE,scall,&(*B)[i]);CHKERRQ(ierr); 2555 } 2556 PetscFunctionReturn(0); 2557 } 2558 2559 #undef __FUNCT__ 2560 #define __FUNCT__ "MatIncreaseOverlap_SeqAIJ" 2561 PetscErrorCode MatIncreaseOverlap_SeqAIJ(Mat A,PetscInt is_max,IS is[],PetscInt ov) 2562 { 2563 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2564 PetscErrorCode ierr; 2565 PetscInt row,i,j,k,l,m,n,*nidx,isz,val; 2566 const PetscInt *idx; 2567 PetscInt start,end,*ai,*aj; 2568 PetscBT table; 2569 2570 PetscFunctionBegin; 2571 m = A->rmap->n; 2572 ai = a->i; 2573 aj = a->j; 2574 2575 if (ov < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"illegal negative overlap value used"); 2576 2577 ierr = PetscMalloc((m+1)*sizeof(PetscInt),&nidx);CHKERRQ(ierr); 2578 ierr = PetscBTCreate(m,&table);CHKERRQ(ierr); 2579 2580 for (i=0; i<is_max; i++) { 2581 /* Initialize the two local arrays */ 2582 isz = 0; 2583 ierr = PetscBTMemzero(m,table);CHKERRQ(ierr); 2584 2585 /* Extract the indices, assume there can be duplicate entries */ 2586 ierr = ISGetIndices(is[i],&idx);CHKERRQ(ierr); 2587 ierr = ISGetLocalSize(is[i],&n);CHKERRQ(ierr); 2588 2589 /* Enter these into the temp arrays. I.e., mark table[row], enter row into new index */ 2590 for (j=0; j<n; ++j) { 2591 if (!PetscBTLookupSet(table,idx[j])) nidx[isz++] = idx[j]; 2592 } 2593 ierr = ISRestoreIndices(is[i],&idx);CHKERRQ(ierr); 2594 ierr = ISDestroy(&is[i]);CHKERRQ(ierr); 2595 2596 k = 0; 2597 for (j=0; j<ov; j++) { /* for each overlap */ 2598 n = isz; 2599 for (; k<n; k++) { /* do only those rows in nidx[k], which are not done yet */ 2600 row = nidx[k]; 2601 start = ai[row]; 2602 end = ai[row+1]; 2603 for (l = start; l<end; l++) { 2604 val = aj[l]; 2605 if (!PetscBTLookupSet(table,val)) nidx[isz++] = val; 2606 } 2607 } 2608 } 2609 ierr = ISCreateGeneral(PETSC_COMM_SELF,isz,nidx,PETSC_COPY_VALUES,(is+i));CHKERRQ(ierr); 2610 } 2611 ierr = PetscBTDestroy(&table);CHKERRQ(ierr); 2612 ierr = PetscFree(nidx);CHKERRQ(ierr); 2613 PetscFunctionReturn(0); 2614 } 2615 2616 /* -------------------------------------------------------------- */ 2617 #undef __FUNCT__ 2618 #define __FUNCT__ "MatPermute_SeqAIJ" 2619 PetscErrorCode MatPermute_SeqAIJ(Mat A,IS rowp,IS colp,Mat *B) 2620 { 2621 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2622 PetscErrorCode ierr; 2623 PetscInt i,nz = 0,m = A->rmap->n,n = A->cmap->n; 2624 const PetscInt *row,*col; 2625 PetscInt *cnew,j,*lens; 2626 IS icolp,irowp; 2627 PetscInt *cwork = NULL; 2628 PetscScalar *vwork = NULL; 2629 2630 PetscFunctionBegin; 2631 ierr = ISInvertPermutation(rowp,PETSC_DECIDE,&irowp);CHKERRQ(ierr); 2632 ierr = ISGetIndices(irowp,&row);CHKERRQ(ierr); 2633 ierr = ISInvertPermutation(colp,PETSC_DECIDE,&icolp);CHKERRQ(ierr); 2634 ierr = ISGetIndices(icolp,&col);CHKERRQ(ierr); 2635 2636 /* determine lengths of permuted rows */ 2637 ierr = PetscMalloc((m+1)*sizeof(PetscInt),&lens);CHKERRQ(ierr); 2638 for (i=0; i<m; i++) lens[row[i]] = a->i[i+1] - a->i[i]; 2639 ierr = MatCreate(PetscObjectComm((PetscObject)A),B);CHKERRQ(ierr); 2640 ierr = MatSetSizes(*B,m,n,m,n);CHKERRQ(ierr); 2641 ierr = MatSetBlockSizes(*B,A->rmap->bs,A->cmap->bs);CHKERRQ(ierr); 2642 ierr = MatSetType(*B,((PetscObject)A)->type_name);CHKERRQ(ierr); 2643 ierr = MatSeqAIJSetPreallocation_SeqAIJ(*B,0,lens);CHKERRQ(ierr); 2644 ierr = PetscFree(lens);CHKERRQ(ierr); 2645 2646 ierr = PetscMalloc(n*sizeof(PetscInt),&cnew);CHKERRQ(ierr); 2647 for (i=0; i<m; i++) { 2648 ierr = MatGetRow_SeqAIJ(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 2649 for (j=0; j<nz; j++) cnew[j] = col[cwork[j]]; 2650 ierr = MatSetValues_SeqAIJ(*B,1,&row[i],nz,cnew,vwork,INSERT_VALUES);CHKERRQ(ierr); 2651 ierr = MatRestoreRow_SeqAIJ(A,i,&nz,&cwork,&vwork);CHKERRQ(ierr); 2652 } 2653 ierr = PetscFree(cnew);CHKERRQ(ierr); 2654 2655 (*B)->assembled = PETSC_FALSE; 2656 2657 ierr = MatAssemblyBegin(*B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2658 ierr = MatAssemblyEnd(*B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2659 ierr = ISRestoreIndices(irowp,&row);CHKERRQ(ierr); 2660 ierr = ISRestoreIndices(icolp,&col);CHKERRQ(ierr); 2661 ierr = ISDestroy(&irowp);CHKERRQ(ierr); 2662 ierr = ISDestroy(&icolp);CHKERRQ(ierr); 2663 PetscFunctionReturn(0); 2664 } 2665 2666 #undef __FUNCT__ 2667 #define __FUNCT__ "MatCopy_SeqAIJ" 2668 PetscErrorCode MatCopy_SeqAIJ(Mat A,Mat B,MatStructure str) 2669 { 2670 PetscErrorCode ierr; 2671 2672 PetscFunctionBegin; 2673 /* If the two matrices have the same copy implementation, use fast copy. */ 2674 if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) { 2675 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2676 Mat_SeqAIJ *b = (Mat_SeqAIJ*)B->data; 2677 2678 if (a->i[A->rmap->n] != b->i[B->rmap->n]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_INCOMP,"Number of nonzeros in two matrices are different"); 2679 ierr = PetscMemcpy(b->a,a->a,(a->i[A->rmap->n])*sizeof(PetscScalar));CHKERRQ(ierr); 2680 } else { 2681 ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr); 2682 } 2683 PetscFunctionReturn(0); 2684 } 2685 2686 #undef __FUNCT__ 2687 #define __FUNCT__ "MatSetUp_SeqAIJ" 2688 PetscErrorCode MatSetUp_SeqAIJ(Mat A) 2689 { 2690 PetscErrorCode ierr; 2691 2692 PetscFunctionBegin; 2693 ierr = MatSeqAIJSetPreallocation_SeqAIJ(A,PETSC_DEFAULT,0);CHKERRQ(ierr); 2694 PetscFunctionReturn(0); 2695 } 2696 2697 #undef __FUNCT__ 2698 #define __FUNCT__ "MatSeqAIJGetArray_SeqAIJ" 2699 PetscErrorCode MatSeqAIJGetArray_SeqAIJ(Mat A,PetscScalar *array[]) 2700 { 2701 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 2702 2703 PetscFunctionBegin; 2704 *array = a->a; 2705 PetscFunctionReturn(0); 2706 } 2707 2708 #undef __FUNCT__ 2709 #define __FUNCT__ "MatSeqAIJRestoreArray_SeqAIJ" 2710 PetscErrorCode MatSeqAIJRestoreArray_SeqAIJ(Mat A,PetscScalar *array[]) 2711 { 2712 PetscFunctionBegin; 2713 PetscFunctionReturn(0); 2714 } 2715 2716 /* Optimize MatFDColoringApply_AIJ() by using array den2sp to skip calling MatSetValues() */ 2717 /* #define JACOBIANCOLOROPT */ 2718 #if defined(JACOBIANCOLOROPT) 2719 #include <petsctime.h> 2720 #endif 2721 #undef __FUNCT__ 2722 #define __FUNCT__ "MatFDColoringApply_SeqAIJ" 2723 PetscErrorCode MatFDColoringApply_SeqAIJ(Mat J,MatFDColoring coloring,Vec x1,MatStructure *flag,void *sctx) 2724 { 2725 PetscErrorCode (*f)(void*,Vec,Vec,void*) = (PetscErrorCode (*)(void*,Vec,Vec,void*))coloring->f; 2726 PetscErrorCode ierr; 2727 PetscInt k,l,row,col,N; 2728 PetscScalar dx,*y,*xx,*w3_array; 2729 PetscScalar *vscale_array; 2730 PetscReal epsilon=coloring->error_rel,umin = coloring->umin,unorm; 2731 Vec w1=coloring->w1,w2=coloring->w2,w3; 2732 void *fctx=coloring->fctx; 2733 PetscBool flg=PETSC_FALSE; 2734 Mat_SeqAIJ *csp=(Mat_SeqAIJ*)J->data; 2735 PetscScalar *ca=csp->a; 2736 const PetscInt ncolors=coloring->ncolors,*ncolumns=coloring->ncolumns,*nrows=coloring->nrows; 2737 PetscInt **columns=coloring->columns,ncolumns_k,nrows_k,nz,spidx; 2738 #if defined(JACOBIANCOLOROPT) 2739 PetscLogDouble t0,t1,t_init=0.0,t_setvals=0.0,t_func=0.0,t_dx=0.0,t_kl=0.0,t00,t11; 2740 #endif 2741 2742 PetscFunctionBegin; 2743 #if defined(JACOBIANCOLOROPT) 2744 ierr = PetscTime(&t0);CHKERRQ(ierr); 2745 #endif 2746 ierr = MatSetUnfactored(J);CHKERRQ(ierr); 2747 ierr = PetscOptionsGetBool(NULL,"-mat_fd_coloring_dont_rezero",&flg,NULL);CHKERRQ(ierr); 2748 if (flg) { 2749 ierr = PetscInfo(coloring,"Not calling MatZeroEntries()\n");CHKERRQ(ierr); 2750 } else { 2751 PetscBool assembled; 2752 ierr = MatAssembled(J,&assembled);CHKERRQ(ierr); 2753 if (assembled) { 2754 ierr = MatZeroEntries(J);CHKERRQ(ierr); 2755 } 2756 } 2757 2758 if (!coloring->vscale) { 2759 ierr = VecDuplicate(x1,&coloring->vscale);CHKERRQ(ierr); 2760 } 2761 2762 if (coloring->htype[0] == 'w') { /* tacky test; need to make systematic if we add other approaches to computing h*/ 2763 ierr = VecNorm(x1,NORM_2,&unorm);CHKERRQ(ierr); 2764 } 2765 #if defined(JACOBIANCOLOROPT) 2766 ierr = PetscTime(&t1);CHKERRQ(ierr); 2767 t_init += t1 - t0; 2768 #endif 2769 2770 /* Set w1 = F(x1) */ 2771 #if defined(JACOBIANCOLOROPT) 2772 ierr = PetscTime(&t0);CHKERRQ(ierr); 2773 #endif 2774 if (!coloring->fset) { 2775 ierr = PetscLogEventBegin(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2776 ierr = (*f)(sctx,x1,w1,fctx);CHKERRQ(ierr); 2777 ierr = PetscLogEventEnd(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2778 } else { 2779 coloring->fset = PETSC_FALSE; 2780 } 2781 #if defined(JACOBIANCOLOROPT) 2782 ierr = PetscTime(&t1);CHKERRQ(ierr); 2783 t_setvals += t1 - t0; 2784 #endif 2785 2786 #if defined(JACOBIANCOLOROPT) 2787 ierr = PetscTime(&t0);CHKERRQ(ierr); 2788 #endif 2789 if (!coloring->w3) { 2790 ierr = VecDuplicate(x1,&coloring->w3);CHKERRQ(ierr); 2791 ierr = PetscLogObjectParent((PetscObject)coloring,(PetscObject)coloring->w3);CHKERRQ(ierr); 2792 } 2793 w3 = coloring->w3; 2794 2795 /* Compute scale factors: vscale = 1./dx = 1./(epsilon*xx) */ 2796 ierr = VecGetArray(x1,&xx);CHKERRQ(ierr); 2797 ierr = VecGetArray(coloring->vscale,&vscale_array);CHKERRQ(ierr); 2798 ierr = VecGetSize(x1,&N);CHKERRQ(ierr); 2799 for (col=0; col<N; col++) { 2800 if (coloring->htype[0] == 'w') { 2801 dx = 1.0 + unorm; 2802 } else { 2803 dx = xx[col]; 2804 } 2805 if (dx == (PetscScalar)0.0) dx = 1.0; 2806 if (PetscAbsScalar(dx) < umin && PetscRealPart(dx) >= 0.0) dx = umin; 2807 else if (PetscRealPart(dx) < 0.0 && PetscAbsScalar(dx) < umin) dx = -umin; 2808 dx *= epsilon; 2809 vscale_array[col] = (PetscScalar)dx; 2810 } 2811 #if defined(JACOBIANCOLOROPT) 2812 ierr = PetscTime(&t1);CHKERRQ(ierr); 2813 t_init += t1 - t0; 2814 #endif 2815 2816 nz = 0; 2817 for (k=0; k<ncolors; k++) { /* loop over colors */ 2818 #if defined(JACOBIANCOLOROPT) 2819 ierr = PetscTime(&t0);CHKERRQ(ierr); 2820 #endif 2821 coloring->currentcolor = k; 2822 2823 /* 2824 Loop over each column associated with color 2825 adding the perturbation to the vector w3 = x1 + dx. 2826 */ 2827 ierr = VecCopy(x1,w3);CHKERRQ(ierr); 2828 ierr = VecGetArray(w3,&w3_array);CHKERRQ(ierr); 2829 ncolumns_k = ncolumns[k]; 2830 for (l=0; l<ncolumns_k; l++) { /* loop over columns */ 2831 col = columns[k][l]; 2832 w3_array[col] += vscale_array[col]; 2833 } 2834 ierr = VecRestoreArray(w3,&w3_array);CHKERRQ(ierr); 2835 #if defined(JACOBIANCOLOROPT) 2836 ierr = PetscTime(&t1);CHKERRQ(ierr); 2837 t_dx += t1 - t0; 2838 #endif 2839 2840 /* 2841 Evaluate function at w3 = x1 + dx (here dx is a vector of perturbations) 2842 w2 = F(x1 + dx) - F(x1) 2843 */ 2844 #if defined(JACOBIANCOLOROPT) 2845 ierr = PetscTime(&t0);CHKERRQ(ierr); 2846 #endif 2847 ierr = PetscLogEventBegin(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2848 ierr = (*f)(sctx,w3,w2,fctx);CHKERRQ(ierr); 2849 ierr = PetscLogEventEnd(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2850 #if defined(JACOBIANCOLOROPT) 2851 ierr = PetscTime(&t1);CHKERRQ(ierr); 2852 t_func += t1 - t0; 2853 #endif 2854 2855 #if defined(JACOBIANCOLOROPT) 2856 ierr = PetscTime(&t0);CHKERRQ(ierr); 2857 #endif 2858 ierr = VecAXPY(w2,-1.0,w1);CHKERRQ(ierr); 2859 2860 /* 2861 Loop over rows of vector, putting w2/dx into Jacobian matrix 2862 */ 2863 ierr = VecGetArray(w2,&y);CHKERRQ(ierr); 2864 nrows_k = nrows[k]; 2865 for (l=0; l<nrows_k; l++) { /* loop over rows */ 2866 #if defined(JACOBIANCOLOROPT) 2867 ierr = PetscTime(&t00);CHKERRQ(ierr); 2868 #endif 2869 row = coloring->rowcolden2sp3[nz++]; 2870 col = coloring->rowcolden2sp3[nz++]; 2871 spidx = coloring->rowcolden2sp3[nz++]; 2872 #if defined(JACOBIANCOLOROPT) 2873 ierr = PetscTime(&t11);CHKERRQ(ierr); 2874 t_kl += t11 - t00; 2875 #endif 2876 ca[spidx] = y[row]/vscale_array[col]; 2877 } 2878 ierr = VecRestoreArray(w2,&y);CHKERRQ(ierr); 2879 2880 #if defined(JACOBIANCOLOROPT) 2881 ierr = PetscTime(&t1);CHKERRQ(ierr); 2882 t_setvals += t1 - t0; 2883 #endif 2884 } /* endof for each color */ 2885 #if defined(JACOBIANCOLOROPT) 2886 printf(" FDColorApply time: init %g + func %g + setvalues %g + dx %g= %g\n",t_init,t_func,t_setvals,t_dx,t_init+t_func+t_setvals+t_dx); 2887 printf(" FDColorApply time: kl %g\n",t_kl); 2888 #endif 2889 ierr = VecRestoreArray(coloring->vscale,&vscale_array);CHKERRQ(ierr); 2890 ierr = VecRestoreArray(x1,&xx);CHKERRQ(ierr); 2891 2892 coloring->currentcolor = -1; 2893 PetscFunctionReturn(0); 2894 } 2895 /* --------------------------------------------------------*/ 2896 2897 #undef __FUNCT__ 2898 #define __FUNCT__ "MatFDColoringApply_SeqAIJ_old" 2899 PetscErrorCode MatFDColoringApply_SeqAIJ_old(Mat J,MatFDColoring coloring,Vec x1,MatStructure *flag,void *sctx) 2900 { 2901 PetscErrorCode (*f)(void*,Vec,Vec,void*) = (PetscErrorCode (*)(void*,Vec,Vec,void*))coloring->f; 2902 PetscErrorCode ierr; 2903 PetscInt k,N,start,end,l,row,col,srow,**vscaleforrow; 2904 PetscScalar dx,*y,*xx,*w3_array; 2905 PetscScalar *vscale_array; 2906 PetscReal epsilon = coloring->error_rel,umin = coloring->umin; 2907 Vec w1,w2,w3; 2908 void *fctx = coloring->fctx; 2909 PetscBool flg = PETSC_FALSE; 2910 2911 PetscFunctionBegin; 2912 printf("MatFDColoringApply_SeqAIJ ...\n"); 2913 if (!coloring->w1) { 2914 ierr = VecDuplicate(x1,&coloring->w1);CHKERRQ(ierr); 2915 ierr = PetscLogObjectParent((PetscObject)coloring,(PetscObject)coloring->w1);CHKERRQ(ierr); 2916 ierr = VecDuplicate(x1,&coloring->w2);CHKERRQ(ierr); 2917 ierr = PetscLogObjectParent((PetscObject)coloring,(PetscObject)coloring->w2);CHKERRQ(ierr); 2918 ierr = VecDuplicate(x1,&coloring->w3);CHKERRQ(ierr); 2919 ierr = PetscLogObjectParent((PetscObject)coloring,(PetscObject)coloring->w3);CHKERRQ(ierr); 2920 } 2921 w1 = coloring->w1; w2 = coloring->w2; w3 = coloring->w3; 2922 2923 ierr = MatSetUnfactored(J);CHKERRQ(ierr); 2924 ierr = PetscOptionsGetBool(((PetscObject)coloring)->prefix,"-mat_fd_coloring_dont_rezero",&flg,NULL);CHKERRQ(ierr); 2925 if (flg) { 2926 ierr = PetscInfo(coloring,"Not calling MatZeroEntries()\n");CHKERRQ(ierr); 2927 } else { 2928 PetscBool assembled; 2929 ierr = MatAssembled(J,&assembled);CHKERRQ(ierr); 2930 if (assembled) { 2931 ierr = MatZeroEntries(J);CHKERRQ(ierr); 2932 } 2933 } 2934 2935 ierr = VecGetOwnershipRange(x1,&start,&end);CHKERRQ(ierr); 2936 ierr = VecGetSize(x1,&N);CHKERRQ(ierr); 2937 2938 if (!coloring->fset) { 2939 ierr = PetscLogEventBegin(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2940 ierr = (*f)(sctx,x1,w1,fctx);CHKERRQ(ierr); 2941 ierr = PetscLogEventEnd(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 2942 } else { 2943 coloring->fset = PETSC_FALSE; 2944 } 2945 2946 /* 2947 Compute all the scale factors and share with other processors 2948 */ 2949 ierr = VecGetArray(x1,&xx);CHKERRQ(ierr);xx = xx - start; 2950 ierr = VecGetArray(coloring->vscale,&vscale_array);CHKERRQ(ierr);vscale_array = vscale_array - start; 2951 for (k=0; k<coloring->ncolors; k++) { 2952 /* 2953 Loop over each column associated with color adding the 2954 perturbation to the vector w3. 2955 */ 2956 for (l=0; l<coloring->ncolumns[k]; l++) { 2957 col = coloring->columns[k][l]; /* column of the matrix we are probing for */ 2958 dx = xx[col]; 2959 if (dx == 0.0) dx = 1.0; 2960 if (PetscAbsScalar(dx) < umin && PetscRealPart(dx) >= 0.0) dx = umin; 2961 else if (PetscRealPart(dx) < 0.0 && PetscAbsScalar(dx) < umin) dx = -umin; 2962 dx *= epsilon; 2963 vscale_array[col] = 1.0/dx; 2964 } 2965 } 2966 vscale_array = vscale_array + start; 2967 2968 ierr = VecRestoreArray(coloring->vscale,&vscale_array);CHKERRQ(ierr); 2969 ierr = VecGhostUpdateBegin(coloring->vscale,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2970 ierr = VecGhostUpdateEnd(coloring->vscale,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2971 2972 /* ierr = VecView(coloring->vscale,PETSC_VIEWER_STDOUT_WORLD); 2973 ierr = VecView(x1,PETSC_VIEWER_STDOUT_WORLD);*/ 2974 2975 if (coloring->vscaleforrow) vscaleforrow = coloring->vscaleforrow; 2976 else vscaleforrow = coloring->columnsforrow; 2977 2978 ierr = VecGetArray(coloring->vscale,&vscale_array);CHKERRQ(ierr); 2979 /* 2980 Loop over each color 2981 */ 2982 for (k=0; k<coloring->ncolors; k++) { 2983 coloring->currentcolor = k; 2984 2985 ierr = VecCopy(x1,w3);CHKERRQ(ierr); 2986 ierr = VecGetArray(w3,&w3_array);CHKERRQ(ierr);w3_array = w3_array - start; 2987 /* 2988 Loop over each column associated with color adding the 2989 perturbation to the vector w3. 2990 */ 2991 for (l=0; l<coloring->ncolumns[k]; l++) { 2992 col = coloring->columns[k][l]; /* column of the matrix we are probing for */ 2993 dx = xx[col]; 2994 if (dx == 0.0) dx = 1.0; 2995 if (PetscAbsScalar(dx) < umin && PetscRealPart(dx) >= 0.0) dx = umin; 2996 else if (PetscRealPart(dx) < 0.0 && PetscAbsScalar(dx) < umin) dx = -umin; 2997 dx *= epsilon; 2998 if (!PetscAbsScalar(dx)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Computed 0 differencing parameter"); 2999 w3_array[col] += dx; 3000 } 3001 w3_array = w3_array + start; ierr = VecRestoreArray(w3,&w3_array);CHKERRQ(ierr); 3002 3003 /* 3004 Evaluate function at x1 + dx (here dx is a vector of perturbations) 3005 */ 3006 3007 ierr = PetscLogEventBegin(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 3008 ierr = (*f)(sctx,w3,w2,fctx);CHKERRQ(ierr); 3009 ierr = PetscLogEventEnd(MAT_FDColoringFunction,0,0,0,0);CHKERRQ(ierr); 3010 ierr = VecAXPY(w2,-1.0,w1);CHKERRQ(ierr); 3011 3012 /* 3013 Loop over rows of vector, putting results into Jacobian matrix 3014 */ 3015 ierr = VecGetArray(w2,&y);CHKERRQ(ierr); 3016 for (l=0; l<coloring->nrows[k]; l++) { 3017 row = coloring->rows[k][l]; 3018 col = coloring->columnsforrow[k][l]; 3019 y[row] *= vscale_array[vscaleforrow[k][l]]; 3020 srow = row + start; 3021 ierr = MatSetValues_SeqAIJ(J,1,&srow,1,&col,y+row,INSERT_VALUES);CHKERRQ(ierr); 3022 } 3023 ierr = VecRestoreArray(w2,&y);CHKERRQ(ierr); 3024 } 3025 coloring->currentcolor = k; 3026 3027 ierr = VecRestoreArray(coloring->vscale,&vscale_array);CHKERRQ(ierr); 3028 xx = xx + start; 3029 ierr = VecRestoreArray(x1,&xx);CHKERRQ(ierr); 3030 ierr = MatAssemblyBegin(J,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3031 ierr = MatAssemblyEnd(J,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3032 PetscFunctionReturn(0); 3033 } 3034 3035 /* 3036 Computes the number of nonzeros per row needed for preallocation when X and Y 3037 have different nonzero structure. 3038 */ 3039 #undef __FUNCT__ 3040 #define __FUNCT__ "MatAXPYGetPreallocation_SeqAIJ" 3041 PetscErrorCode MatAXPYGetPreallocation_SeqAIJ(Mat Y,Mat X,PetscInt *nnz) 3042 { 3043 PetscInt i,m=Y->rmap->N; 3044 Mat_SeqAIJ *x = (Mat_SeqAIJ*)X->data; 3045 Mat_SeqAIJ *y = (Mat_SeqAIJ*)Y->data; 3046 const PetscInt *xi = x->i,*yi = y->i; 3047 3048 PetscFunctionBegin; 3049 /* Set the number of nonzeros in the new matrix */ 3050 for (i=0; i<m; i++) { 3051 PetscInt j,k,nzx = xi[i+1] - xi[i],nzy = yi[i+1] - yi[i]; 3052 const PetscInt *xj = x->j+xi[i],*yj = y->j+yi[i]; 3053 nnz[i] = 0; 3054 for (j=0,k=0; j<nzx; j++) { /* Point in X */ 3055 for (; k<nzy && yj[k]<xj[j]; k++) nnz[i]++; /* Catch up to X */ 3056 if (k<nzy && yj[k]==xj[j]) k++; /* Skip duplicate */ 3057 nnz[i]++; 3058 } 3059 for (; k<nzy; k++) nnz[i]++; 3060 } 3061 PetscFunctionReturn(0); 3062 } 3063 3064 #undef __FUNCT__ 3065 #define __FUNCT__ "MatAXPY_SeqAIJ" 3066 PetscErrorCode MatAXPY_SeqAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str) 3067 { 3068 PetscErrorCode ierr; 3069 PetscInt i; 3070 Mat_SeqAIJ *x = (Mat_SeqAIJ*)X->data,*y = (Mat_SeqAIJ*)Y->data; 3071 PetscBLASInt one=1,bnz; 3072 3073 PetscFunctionBegin; 3074 ierr = PetscBLASIntCast(x->nz,&bnz);CHKERRQ(ierr); 3075 if (str == SAME_NONZERO_PATTERN) { 3076 PetscScalar alpha = a; 3077 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one)); 3078 ierr = MatSeqAIJInvalidateDiagonal(Y);CHKERRQ(ierr); 3079 } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */ 3080 if (y->xtoy && y->XtoY != X) { 3081 ierr = PetscFree(y->xtoy);CHKERRQ(ierr); 3082 ierr = MatDestroy(&y->XtoY);CHKERRQ(ierr); 3083 } 3084 if (!y->xtoy) { /* get xtoy */ 3085 ierr = MatAXPYGetxtoy_Private(X->rmap->n,x->i,x->j,NULL, y->i,y->j,NULL, &y->xtoy);CHKERRQ(ierr); 3086 y->XtoY = X; 3087 ierr = PetscObjectReference((PetscObject)X);CHKERRQ(ierr); 3088 } 3089 for (i=0; i<x->nz; i++) y->a[y->xtoy[i]] += a*(x->a[i]); 3090 ierr = PetscInfo3(Y,"ratio of nnz(X)/nnz(Y): %d/%d = %g\n",x->nz,y->nz,(double)(PetscReal)(x->nz)/(y->nz+1));CHKERRQ(ierr); 3091 } else { 3092 Mat B; 3093 PetscInt *nnz; 3094 ierr = PetscMalloc(Y->rmap->N*sizeof(PetscInt),&nnz);CHKERRQ(ierr); 3095 ierr = MatCreate(PetscObjectComm((PetscObject)Y),&B);CHKERRQ(ierr); 3096 ierr = PetscObjectSetName((PetscObject)B,((PetscObject)Y)->name);CHKERRQ(ierr); 3097 ierr = MatSetSizes(B,Y->rmap->n,Y->cmap->n,Y->rmap->N,Y->cmap->N);CHKERRQ(ierr); 3098 ierr = MatSetBlockSizes(B,Y->rmap->bs,Y->cmap->bs);CHKERRQ(ierr); 3099 ierr = MatSetType(B,(MatType) ((PetscObject)Y)->type_name);CHKERRQ(ierr); 3100 ierr = MatAXPYGetPreallocation_SeqAIJ(Y,X,nnz);CHKERRQ(ierr); 3101 ierr = MatSeqAIJSetPreallocation(B,0,nnz);CHKERRQ(ierr); 3102 ierr = MatAXPY_BasicWithPreallocation(B,Y,a,X,str);CHKERRQ(ierr); 3103 ierr = MatHeaderReplace(Y,B);CHKERRQ(ierr); 3104 ierr = PetscFree(nnz);CHKERRQ(ierr); 3105 } 3106 PetscFunctionReturn(0); 3107 } 3108 3109 #undef __FUNCT__ 3110 #define __FUNCT__ "MatConjugate_SeqAIJ" 3111 PetscErrorCode MatConjugate_SeqAIJ(Mat mat) 3112 { 3113 #if defined(PETSC_USE_COMPLEX) 3114 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data; 3115 PetscInt i,nz; 3116 PetscScalar *a; 3117 3118 PetscFunctionBegin; 3119 nz = aij->nz; 3120 a = aij->a; 3121 for (i=0; i<nz; i++) a[i] = PetscConj(a[i]); 3122 #else 3123 PetscFunctionBegin; 3124 #endif 3125 PetscFunctionReturn(0); 3126 } 3127 3128 #undef __FUNCT__ 3129 #define __FUNCT__ "MatGetRowMaxAbs_SeqAIJ" 3130 PetscErrorCode MatGetRowMaxAbs_SeqAIJ(Mat A,Vec v,PetscInt idx[]) 3131 { 3132 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 3133 PetscErrorCode ierr; 3134 PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n; 3135 PetscReal atmp; 3136 PetscScalar *x; 3137 MatScalar *aa; 3138 3139 PetscFunctionBegin; 3140 if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3141 aa = a->a; 3142 ai = a->i; 3143 aj = a->j; 3144 3145 ierr = VecSet(v,0.0);CHKERRQ(ierr); 3146 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 3147 ierr = VecGetLocalSize(v,&n);CHKERRQ(ierr); 3148 if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector"); 3149 for (i=0; i<m; i++) { 3150 ncols = ai[1] - ai[0]; ai++; 3151 x[i] = 0.0; 3152 for (j=0; j<ncols; j++) { 3153 atmp = PetscAbsScalar(*aa); 3154 if (PetscAbsScalar(x[i]) < atmp) {x[i] = atmp; if (idx) idx[i] = *aj;} 3155 aa++; aj++; 3156 } 3157 } 3158 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 3159 PetscFunctionReturn(0); 3160 } 3161 3162 #undef __FUNCT__ 3163 #define __FUNCT__ "MatGetRowMax_SeqAIJ" 3164 PetscErrorCode MatGetRowMax_SeqAIJ(Mat A,Vec v,PetscInt idx[]) 3165 { 3166 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 3167 PetscErrorCode ierr; 3168 PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n; 3169 PetscScalar *x; 3170 MatScalar *aa; 3171 3172 PetscFunctionBegin; 3173 if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3174 aa = a->a; 3175 ai = a->i; 3176 aj = a->j; 3177 3178 ierr = VecSet(v,0.0);CHKERRQ(ierr); 3179 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 3180 ierr = VecGetLocalSize(v,&n);CHKERRQ(ierr); 3181 if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector"); 3182 for (i=0; i<m; i++) { 3183 ncols = ai[1] - ai[0]; ai++; 3184 if (ncols == A->cmap->n) { /* row is dense */ 3185 x[i] = *aa; if (idx) idx[i] = 0; 3186 } else { /* row is sparse so already KNOW maximum is 0.0 or higher */ 3187 x[i] = 0.0; 3188 if (idx) { 3189 idx[i] = 0; /* in case ncols is zero */ 3190 for (j=0;j<ncols;j++) { /* find first implicit 0.0 in the row */ 3191 if (aj[j] > j) { 3192 idx[i] = j; 3193 break; 3194 } 3195 } 3196 } 3197 } 3198 for (j=0; j<ncols; j++) { 3199 if (PetscRealPart(x[i]) < PetscRealPart(*aa)) {x[i] = *aa; if (idx) idx[i] = *aj;} 3200 aa++; aj++; 3201 } 3202 } 3203 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 3204 PetscFunctionReturn(0); 3205 } 3206 3207 #undef __FUNCT__ 3208 #define __FUNCT__ "MatGetRowMinAbs_SeqAIJ" 3209 PetscErrorCode MatGetRowMinAbs_SeqAIJ(Mat A,Vec v,PetscInt idx[]) 3210 { 3211 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 3212 PetscErrorCode ierr; 3213 PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n; 3214 PetscReal atmp; 3215 PetscScalar *x; 3216 MatScalar *aa; 3217 3218 PetscFunctionBegin; 3219 if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3220 aa = a->a; 3221 ai = a->i; 3222 aj = a->j; 3223 3224 ierr = VecSet(v,0.0);CHKERRQ(ierr); 3225 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 3226 ierr = VecGetLocalSize(v,&n);CHKERRQ(ierr); 3227 if (n != A->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector, %d vs. %d rows", A->rmap->n, n); 3228 for (i=0; i<m; i++) { 3229 ncols = ai[1] - ai[0]; ai++; 3230 if (ncols) { 3231 /* Get first nonzero */ 3232 for (j = 0; j < ncols; j++) { 3233 atmp = PetscAbsScalar(aa[j]); 3234 if (atmp > 1.0e-12) { 3235 x[i] = atmp; 3236 if (idx) idx[i] = aj[j]; 3237 break; 3238 } 3239 } 3240 if (j == ncols) {x[i] = PetscAbsScalar(*aa); if (idx) idx[i] = *aj;} 3241 } else { 3242 x[i] = 0.0; if (idx) idx[i] = 0; 3243 } 3244 for (j = 0; j < ncols; j++) { 3245 atmp = PetscAbsScalar(*aa); 3246 if (atmp > 1.0e-12 && PetscAbsScalar(x[i]) > atmp) {x[i] = atmp; if (idx) idx[i] = *aj;} 3247 aa++; aj++; 3248 } 3249 } 3250 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 3251 PetscFunctionReturn(0); 3252 } 3253 3254 #undef __FUNCT__ 3255 #define __FUNCT__ "MatGetRowMin_SeqAIJ" 3256 PetscErrorCode MatGetRowMin_SeqAIJ(Mat A,Vec v,PetscInt idx[]) 3257 { 3258 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 3259 PetscErrorCode ierr; 3260 PetscInt i,j,m = A->rmap->n,*ai,*aj,ncols,n; 3261 PetscScalar *x; 3262 MatScalar *aa; 3263 3264 PetscFunctionBegin; 3265 if (A->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3266 aa = a->a; 3267 ai = a->i; 3268 aj = a->j; 3269 3270 ierr = VecSet(v,0.0);CHKERRQ(ierr); 3271 ierr = VecGetArray(v,&x);CHKERRQ(ierr); 3272 ierr = VecGetLocalSize(v,&n);CHKERRQ(ierr); 3273 if (n != A->rmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector"); 3274 for (i=0; i<m; i++) { 3275 ncols = ai[1] - ai[0]; ai++; 3276 if (ncols == A->cmap->n) { /* row is dense */ 3277 x[i] = *aa; if (idx) idx[i] = 0; 3278 } else { /* row is sparse so already KNOW minimum is 0.0 or lower */ 3279 x[i] = 0.0; 3280 if (idx) { /* find first implicit 0.0 in the row */ 3281 idx[i] = 0; /* in case ncols is zero */ 3282 for (j=0; j<ncols; j++) { 3283 if (aj[j] > j) { 3284 idx[i] = j; 3285 break; 3286 } 3287 } 3288 } 3289 } 3290 for (j=0; j<ncols; j++) { 3291 if (PetscRealPart(x[i]) > PetscRealPart(*aa)) {x[i] = *aa; if (idx) idx[i] = *aj;} 3292 aa++; aj++; 3293 } 3294 } 3295 ierr = VecRestoreArray(v,&x);CHKERRQ(ierr); 3296 PetscFunctionReturn(0); 3297 } 3298 3299 #include <petscblaslapack.h> 3300 #include <petsc-private/kernels/blockinvert.h> 3301 3302 #undef __FUNCT__ 3303 #define __FUNCT__ "MatInvertBlockDiagonal_SeqAIJ" 3304 PetscErrorCode MatInvertBlockDiagonal_SeqAIJ(Mat A,const PetscScalar **values) 3305 { 3306 Mat_SeqAIJ *a = (Mat_SeqAIJ*) A->data; 3307 PetscErrorCode ierr; 3308 PetscInt i,bs = A->rmap->bs,mbs = A->rmap->n/A->rmap->bs,ipvt[5],bs2 = bs*bs,*v_pivots,ij[7],*IJ,j; 3309 MatScalar *diag,work[25],*v_work; 3310 PetscReal shift = 0.0; 3311 3312 PetscFunctionBegin; 3313 if (a->ibdiagvalid) { 3314 if (values) *values = a->ibdiag; 3315 PetscFunctionReturn(0); 3316 } 3317 ierr = MatMarkDiagonal_SeqAIJ(A);CHKERRQ(ierr); 3318 if (!a->ibdiag) { 3319 ierr = PetscMalloc(bs2*mbs*sizeof(PetscScalar),&a->ibdiag);CHKERRQ(ierr); 3320 ierr = PetscLogObjectMemory((PetscObject)A,bs2*mbs*sizeof(PetscScalar));CHKERRQ(ierr); 3321 } 3322 diag = a->ibdiag; 3323 if (values) *values = a->ibdiag; 3324 /* factor and invert each block */ 3325 switch (bs) { 3326 case 1: 3327 for (i=0; i<mbs; i++) { 3328 ierr = MatGetValues(A,1,&i,1,&i,diag+i);CHKERRQ(ierr); 3329 diag[i] = (PetscScalar)1.0 / (diag[i] + shift); 3330 } 3331 break; 3332 case 2: 3333 for (i=0; i<mbs; i++) { 3334 ij[0] = 2*i; ij[1] = 2*i + 1; 3335 ierr = MatGetValues(A,2,ij,2,ij,diag);CHKERRQ(ierr); 3336 ierr = PetscKernel_A_gets_inverse_A_2(diag,shift);CHKERRQ(ierr); 3337 ierr = PetscKernel_A_gets_transpose_A_2(diag);CHKERRQ(ierr); 3338 diag += 4; 3339 } 3340 break; 3341 case 3: 3342 for (i=0; i<mbs; i++) { 3343 ij[0] = 3*i; ij[1] = 3*i + 1; ij[2] = 3*i + 2; 3344 ierr = MatGetValues(A,3,ij,3,ij,diag);CHKERRQ(ierr); 3345 ierr = PetscKernel_A_gets_inverse_A_3(diag,shift);CHKERRQ(ierr); 3346 ierr = PetscKernel_A_gets_transpose_A_3(diag);CHKERRQ(ierr); 3347 diag += 9; 3348 } 3349 break; 3350 case 4: 3351 for (i=0; i<mbs; i++) { 3352 ij[0] = 4*i; ij[1] = 4*i + 1; ij[2] = 4*i + 2; ij[3] = 4*i + 3; 3353 ierr = MatGetValues(A,4,ij,4,ij,diag);CHKERRQ(ierr); 3354 ierr = PetscKernel_A_gets_inverse_A_4(diag,shift);CHKERRQ(ierr); 3355 ierr = PetscKernel_A_gets_transpose_A_4(diag);CHKERRQ(ierr); 3356 diag += 16; 3357 } 3358 break; 3359 case 5: 3360 for (i=0; i<mbs; i++) { 3361 ij[0] = 5*i; ij[1] = 5*i + 1; ij[2] = 5*i + 2; ij[3] = 5*i + 3; ij[4] = 5*i + 4; 3362 ierr = MatGetValues(A,5,ij,5,ij,diag);CHKERRQ(ierr); 3363 ierr = PetscKernel_A_gets_inverse_A_5(diag,ipvt,work,shift);CHKERRQ(ierr); 3364 ierr = PetscKernel_A_gets_transpose_A_5(diag);CHKERRQ(ierr); 3365 diag += 25; 3366 } 3367 break; 3368 case 6: 3369 for (i=0; i<mbs; i++) { 3370 ij[0] = 6*i; ij[1] = 6*i + 1; ij[2] = 6*i + 2; ij[3] = 6*i + 3; ij[4] = 6*i + 4; ij[5] = 6*i + 5; 3371 ierr = MatGetValues(A,6,ij,6,ij,diag);CHKERRQ(ierr); 3372 ierr = PetscKernel_A_gets_inverse_A_6(diag,shift);CHKERRQ(ierr); 3373 ierr = PetscKernel_A_gets_transpose_A_6(diag);CHKERRQ(ierr); 3374 diag += 36; 3375 } 3376 break; 3377 case 7: 3378 for (i=0; i<mbs; i++) { 3379 ij[0] = 7*i; ij[1] = 7*i + 1; ij[2] = 7*i + 2; ij[3] = 7*i + 3; ij[4] = 7*i + 4; ij[5] = 7*i + 5; ij[5] = 7*i + 6; 3380 ierr = MatGetValues(A,7,ij,7,ij,diag);CHKERRQ(ierr); 3381 ierr = PetscKernel_A_gets_inverse_A_7(diag,shift);CHKERRQ(ierr); 3382 ierr = PetscKernel_A_gets_transpose_A_7(diag);CHKERRQ(ierr); 3383 diag += 49; 3384 } 3385 break; 3386 default: 3387 ierr = PetscMalloc3(bs,MatScalar,&v_work,bs,PetscInt,&v_pivots,bs,PetscInt,&IJ);CHKERRQ(ierr); 3388 for (i=0; i<mbs; i++) { 3389 for (j=0; j<bs; j++) { 3390 IJ[j] = bs*i + j; 3391 } 3392 ierr = MatGetValues(A,bs,IJ,bs,IJ,diag);CHKERRQ(ierr); 3393 ierr = PetscKernel_A_gets_inverse_A(bs,diag,v_pivots,v_work);CHKERRQ(ierr); 3394 ierr = PetscKernel_A_gets_transpose_A_N(diag,bs);CHKERRQ(ierr); 3395 diag += bs2; 3396 } 3397 ierr = PetscFree3(v_work,v_pivots,IJ);CHKERRQ(ierr); 3398 } 3399 a->ibdiagvalid = PETSC_TRUE; 3400 PetscFunctionReturn(0); 3401 } 3402 3403 #undef __FUNCT__ 3404 #define __FUNCT__ "MatSetRandom_SeqAIJ" 3405 static PetscErrorCode MatSetRandom_SeqAIJ(Mat x,PetscRandom rctx) 3406 { 3407 PetscErrorCode ierr; 3408 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)x->data; 3409 PetscScalar a; 3410 PetscInt m,n,i,j,col; 3411 3412 PetscFunctionBegin; 3413 if (!x->assembled) { 3414 ierr = MatGetSize(x,&m,&n);CHKERRQ(ierr); 3415 for (i=0; i<m; i++) { 3416 for (j=0; j<aij->imax[i]; j++) { 3417 ierr = PetscRandomGetValue(rctx,&a);CHKERRQ(ierr); 3418 col = (PetscInt)(n*PetscRealPart(a)); 3419 ierr = MatSetValues(x,1,&i,1,&col,&a,ADD_VALUES);CHKERRQ(ierr); 3420 } 3421 } 3422 } else SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Not yet coded"); 3423 ierr = MatAssemblyBegin(x,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3424 ierr = MatAssemblyEnd(x,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 3425 PetscFunctionReturn(0); 3426 } 3427 3428 extern PetscErrorCode MatFDColoringApply_AIJ(Mat,MatFDColoring,Vec,MatStructure*,void*); 3429 /* -------------------------------------------------------------------*/ 3430 static struct _MatOps MatOps_Values = { MatSetValues_SeqAIJ, 3431 MatGetRow_SeqAIJ, 3432 MatRestoreRow_SeqAIJ, 3433 MatMult_SeqAIJ, 3434 /* 4*/ MatMultAdd_SeqAIJ, 3435 MatMultTranspose_SeqAIJ, 3436 MatMultTransposeAdd_SeqAIJ, 3437 0, 3438 0, 3439 0, 3440 /* 10*/ 0, 3441 MatLUFactor_SeqAIJ, 3442 0, 3443 MatSOR_SeqAIJ, 3444 MatTranspose_SeqAIJ, 3445 /*1 5*/ MatGetInfo_SeqAIJ, 3446 MatEqual_SeqAIJ, 3447 MatGetDiagonal_SeqAIJ, 3448 MatDiagonalScale_SeqAIJ, 3449 MatNorm_SeqAIJ, 3450 /* 20*/ 0, 3451 MatAssemblyEnd_SeqAIJ, 3452 MatSetOption_SeqAIJ, 3453 MatZeroEntries_SeqAIJ, 3454 /* 24*/ MatZeroRows_SeqAIJ, 3455 0, 3456 0, 3457 0, 3458 0, 3459 /* 29*/ MatSetUp_SeqAIJ, 3460 0, 3461 0, 3462 0, 3463 0, 3464 /* 34*/ MatDuplicate_SeqAIJ, 3465 0, 3466 0, 3467 MatILUFactor_SeqAIJ, 3468 0, 3469 /* 39*/ MatAXPY_SeqAIJ, 3470 MatGetSubMatrices_SeqAIJ, 3471 MatIncreaseOverlap_SeqAIJ, 3472 MatGetValues_SeqAIJ, 3473 MatCopy_SeqAIJ, 3474 /* 44*/ MatGetRowMax_SeqAIJ, 3475 MatScale_SeqAIJ, 3476 0, 3477 MatDiagonalSet_SeqAIJ, 3478 MatZeroRowsColumns_SeqAIJ, 3479 /* 49*/ MatSetRandom_SeqAIJ, 3480 MatGetRowIJ_SeqAIJ, 3481 MatRestoreRowIJ_SeqAIJ, 3482 MatGetColumnIJ_SeqAIJ, 3483 MatRestoreColumnIJ_SeqAIJ, 3484 /* 54*/ MatFDColoringCreate_SeqAIJ, 3485 0, 3486 0, 3487 MatPermute_SeqAIJ, 3488 0, 3489 /* 59*/ 0, 3490 MatDestroy_SeqAIJ, 3491 MatView_SeqAIJ, 3492 0, 3493 MatMatMatMult_SeqAIJ_SeqAIJ_SeqAIJ, 3494 /* 64*/ MatMatMatMultSymbolic_SeqAIJ_SeqAIJ_SeqAIJ, 3495 MatMatMatMultNumeric_SeqAIJ_SeqAIJ_SeqAIJ, 3496 0, 3497 0, 3498 0, 3499 /* 69*/ MatGetRowMaxAbs_SeqAIJ, 3500 MatGetRowMinAbs_SeqAIJ, 3501 0, 3502 MatSetColoring_SeqAIJ, 3503 0, 3504 /* 74*/ MatSetValuesAdifor_SeqAIJ, 3505 MatFDColoringApply_AIJ, 3506 0, 3507 0, 3508 0, 3509 /* 79*/ MatFindZeroDiagonals_SeqAIJ, 3510 0, 3511 0, 3512 0, 3513 MatLoad_SeqAIJ, 3514 /* 84*/ MatIsSymmetric_SeqAIJ, 3515 MatIsHermitian_SeqAIJ, 3516 0, 3517 0, 3518 0, 3519 /* 89*/ MatMatMult_SeqAIJ_SeqAIJ, 3520 MatMatMultSymbolic_SeqAIJ_SeqAIJ, 3521 MatMatMultNumeric_SeqAIJ_SeqAIJ, 3522 MatPtAP_SeqAIJ_SeqAIJ, 3523 MatPtAPSymbolic_SeqAIJ_SeqAIJ_DenseAxpy, 3524 /* 94*/ MatPtAPNumeric_SeqAIJ_SeqAIJ, 3525 MatMatTransposeMult_SeqAIJ_SeqAIJ, 3526 MatMatTransposeMultSymbolic_SeqAIJ_SeqAIJ, 3527 MatMatTransposeMultNumeric_SeqAIJ_SeqAIJ, 3528 0, 3529 /* 99*/ 0, 3530 0, 3531 0, 3532 MatConjugate_SeqAIJ, 3533 0, 3534 /*104*/ MatSetValuesRow_SeqAIJ, 3535 MatRealPart_SeqAIJ, 3536 MatImaginaryPart_SeqAIJ, 3537 0, 3538 0, 3539 /*109*/ MatMatSolve_SeqAIJ, 3540 0, 3541 MatGetRowMin_SeqAIJ, 3542 0, 3543 MatMissingDiagonal_SeqAIJ, 3544 /*114*/ 0, 3545 0, 3546 0, 3547 0, 3548 0, 3549 /*119*/ 0, 3550 0, 3551 0, 3552 0, 3553 MatGetMultiProcBlock_SeqAIJ, 3554 /*124*/ MatFindNonzeroRows_SeqAIJ, 3555 MatGetColumnNorms_SeqAIJ, 3556 MatInvertBlockDiagonal_SeqAIJ, 3557 0, 3558 0, 3559 /*129*/ 0, 3560 MatTransposeMatMult_SeqAIJ_SeqAIJ, 3561 MatTransposeMatMultSymbolic_SeqAIJ_SeqAIJ, 3562 MatTransposeMatMultNumeric_SeqAIJ_SeqAIJ, 3563 MatTransposeColoringCreate_SeqAIJ, 3564 /*134*/ MatTransColoringApplySpToDen_SeqAIJ, 3565 MatTransColoringApplyDenToSp_SeqAIJ, 3566 MatRARt_SeqAIJ_SeqAIJ, 3567 MatRARtSymbolic_SeqAIJ_SeqAIJ, 3568 MatRARtNumeric_SeqAIJ_SeqAIJ, 3569 /*139*/0, 3570 0 3571 }; 3572 3573 #undef __FUNCT__ 3574 #define __FUNCT__ "MatSeqAIJSetColumnIndices_SeqAIJ" 3575 PetscErrorCode MatSeqAIJSetColumnIndices_SeqAIJ(Mat mat,PetscInt *indices) 3576 { 3577 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data; 3578 PetscInt i,nz,n; 3579 3580 PetscFunctionBegin; 3581 nz = aij->maxnz; 3582 n = mat->rmap->n; 3583 for (i=0; i<nz; i++) { 3584 aij->j[i] = indices[i]; 3585 } 3586 aij->nz = nz; 3587 for (i=0; i<n; i++) { 3588 aij->ilen[i] = aij->imax[i]; 3589 } 3590 PetscFunctionReturn(0); 3591 } 3592 3593 #undef __FUNCT__ 3594 #define __FUNCT__ "MatSeqAIJSetColumnIndices" 3595 /*@ 3596 MatSeqAIJSetColumnIndices - Set the column indices for all the rows 3597 in the matrix. 3598 3599 Input Parameters: 3600 + mat - the SeqAIJ matrix 3601 - indices - the column indices 3602 3603 Level: advanced 3604 3605 Notes: 3606 This can be called if you have precomputed the nonzero structure of the 3607 matrix and want to provide it to the matrix object to improve the performance 3608 of the MatSetValues() operation. 3609 3610 You MUST have set the correct numbers of nonzeros per row in the call to 3611 MatCreateSeqAIJ(), and the columns indices MUST be sorted. 3612 3613 MUST be called before any calls to MatSetValues(); 3614 3615 The indices should start with zero, not one. 3616 3617 @*/ 3618 PetscErrorCode MatSeqAIJSetColumnIndices(Mat mat,PetscInt *indices) 3619 { 3620 PetscErrorCode ierr; 3621 3622 PetscFunctionBegin; 3623 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3624 PetscValidPointer(indices,2); 3625 ierr = PetscUseMethod(mat,"MatSeqAIJSetColumnIndices_C",(Mat,PetscInt*),(mat,indices));CHKERRQ(ierr); 3626 PetscFunctionReturn(0); 3627 } 3628 3629 /* ----------------------------------------------------------------------------------------*/ 3630 3631 #undef __FUNCT__ 3632 #define __FUNCT__ "MatStoreValues_SeqAIJ" 3633 PetscErrorCode MatStoreValues_SeqAIJ(Mat mat) 3634 { 3635 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data; 3636 PetscErrorCode ierr; 3637 size_t nz = aij->i[mat->rmap->n]; 3638 3639 PetscFunctionBegin; 3640 if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first"); 3641 3642 /* allocate space for values if not already there */ 3643 if (!aij->saved_values) { 3644 ierr = PetscMalloc((nz+1)*sizeof(PetscScalar),&aij->saved_values);CHKERRQ(ierr); 3645 ierr = PetscLogObjectMemory((PetscObject)mat,(nz+1)*sizeof(PetscScalar));CHKERRQ(ierr); 3646 } 3647 3648 /* copy values over */ 3649 ierr = PetscMemcpy(aij->saved_values,aij->a,nz*sizeof(PetscScalar));CHKERRQ(ierr); 3650 PetscFunctionReturn(0); 3651 } 3652 3653 #undef __FUNCT__ 3654 #define __FUNCT__ "MatStoreValues" 3655 /*@ 3656 MatStoreValues - Stashes a copy of the matrix values; this allows, for 3657 example, reuse of the linear part of a Jacobian, while recomputing the 3658 nonlinear portion. 3659 3660 Collect on Mat 3661 3662 Input Parameters: 3663 . mat - the matrix (currently only AIJ matrices support this option) 3664 3665 Level: advanced 3666 3667 Common Usage, with SNESSolve(): 3668 $ Create Jacobian matrix 3669 $ Set linear terms into matrix 3670 $ Apply boundary conditions to matrix, at this time matrix must have 3671 $ final nonzero structure (i.e. setting the nonlinear terms and applying 3672 $ boundary conditions again will not change the nonzero structure 3673 $ ierr = MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE); 3674 $ ierr = MatStoreValues(mat); 3675 $ Call SNESSetJacobian() with matrix 3676 $ In your Jacobian routine 3677 $ ierr = MatRetrieveValues(mat); 3678 $ Set nonlinear terms in matrix 3679 3680 Common Usage without SNESSolve(), i.e. when you handle nonlinear solve yourself: 3681 $ // build linear portion of Jacobian 3682 $ ierr = MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE); 3683 $ ierr = MatStoreValues(mat); 3684 $ loop over nonlinear iterations 3685 $ ierr = MatRetrieveValues(mat); 3686 $ // call MatSetValues(mat,...) to set nonliner portion of Jacobian 3687 $ // call MatAssemblyBegin/End() on matrix 3688 $ Solve linear system with Jacobian 3689 $ endloop 3690 3691 Notes: 3692 Matrix must already be assemblied before calling this routine 3693 Must set the matrix option MatSetOption(mat,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE); before 3694 calling this routine. 3695 3696 When this is called multiple times it overwrites the previous set of stored values 3697 and does not allocated additional space. 3698 3699 .seealso: MatRetrieveValues() 3700 3701 @*/ 3702 PetscErrorCode MatStoreValues(Mat mat) 3703 { 3704 PetscErrorCode ierr; 3705 3706 PetscFunctionBegin; 3707 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3708 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3709 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3710 ierr = PetscUseMethod(mat,"MatStoreValues_C",(Mat),(mat));CHKERRQ(ierr); 3711 PetscFunctionReturn(0); 3712 } 3713 3714 #undef __FUNCT__ 3715 #define __FUNCT__ "MatRetrieveValues_SeqAIJ" 3716 PetscErrorCode MatRetrieveValues_SeqAIJ(Mat mat) 3717 { 3718 Mat_SeqAIJ *aij = (Mat_SeqAIJ*)mat->data; 3719 PetscErrorCode ierr; 3720 PetscInt nz = aij->i[mat->rmap->n]; 3721 3722 PetscFunctionBegin; 3723 if (aij->nonew != 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NEW_NONZERO_LOCATIONS,PETSC_FALSE);first"); 3724 if (!aij->saved_values) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Must call MatStoreValues(A);first"); 3725 /* copy values over */ 3726 ierr = PetscMemcpy(aij->a,aij->saved_values,nz*sizeof(PetscScalar));CHKERRQ(ierr); 3727 PetscFunctionReturn(0); 3728 } 3729 3730 #undef __FUNCT__ 3731 #define __FUNCT__ "MatRetrieveValues" 3732 /*@ 3733 MatRetrieveValues - Retrieves the copy of the matrix values; this allows, for 3734 example, reuse of the linear part of a Jacobian, while recomputing the 3735 nonlinear portion. 3736 3737 Collect on Mat 3738 3739 Input Parameters: 3740 . mat - the matrix (currently on AIJ matrices support this option) 3741 3742 Level: advanced 3743 3744 .seealso: MatStoreValues() 3745 3746 @*/ 3747 PetscErrorCode MatRetrieveValues(Mat mat) 3748 { 3749 PetscErrorCode ierr; 3750 3751 PetscFunctionBegin; 3752 PetscValidHeaderSpecific(mat,MAT_CLASSID,1); 3753 if (!mat->assembled) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix"); 3754 if (mat->factortype) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix"); 3755 ierr = PetscUseMethod(mat,"MatRetrieveValues_C",(Mat),(mat));CHKERRQ(ierr); 3756 PetscFunctionReturn(0); 3757 } 3758 3759 3760 /* --------------------------------------------------------------------------------*/ 3761 #undef __FUNCT__ 3762 #define __FUNCT__ "MatCreateSeqAIJ" 3763 /*@C 3764 MatCreateSeqAIJ - Creates a sparse matrix in AIJ (compressed row) format 3765 (the default parallel PETSc format). For good matrix assembly performance 3766 the user should preallocate the matrix storage by setting the parameter nz 3767 (or the array nnz). By setting these parameters accurately, performance 3768 during matrix assembly can be increased by more than a factor of 50. 3769 3770 Collective on MPI_Comm 3771 3772 Input Parameters: 3773 + comm - MPI communicator, set to PETSC_COMM_SELF 3774 . m - number of rows 3775 . n - number of columns 3776 . nz - number of nonzeros per row (same for all rows) 3777 - nnz - array containing the number of nonzeros in the various rows 3778 (possibly different for each row) or NULL 3779 3780 Output Parameter: 3781 . A - the matrix 3782 3783 It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(), 3784 MatXXXXSetPreallocation() paradgm instead of this routine directly. 3785 [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation] 3786 3787 Notes: 3788 If nnz is given then nz is ignored 3789 3790 The AIJ format (also called the Yale sparse matrix format or 3791 compressed row storage), is fully compatible with standard Fortran 77 3792 storage. That is, the stored row and column indices can begin at 3793 either one (as in Fortran) or zero. See the users' manual for details. 3794 3795 Specify the preallocated storage with either nz or nnz (not both). 3796 Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory 3797 allocation. For large problems you MUST preallocate memory or you 3798 will get TERRIBLE performance, see the users' manual chapter on matrices. 3799 3800 By default, this format uses inodes (identical nodes) when possible, to 3801 improve numerical efficiency of matrix-vector products and solves. We 3802 search for consecutive rows with the same nonzero structure, thereby 3803 reusing matrix information to achieve increased efficiency. 3804 3805 Options Database Keys: 3806 + -mat_no_inode - Do not use inodes 3807 - -mat_inode_limit <limit> - Sets inode limit (max limit=5) 3808 3809 Level: intermediate 3810 3811 .seealso: MatCreate(), MatCreateAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays() 3812 3813 @*/ 3814 PetscErrorCode MatCreateSeqAIJ(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt nz,const PetscInt nnz[],Mat *A) 3815 { 3816 PetscErrorCode ierr; 3817 3818 PetscFunctionBegin; 3819 ierr = MatCreate(comm,A);CHKERRQ(ierr); 3820 ierr = MatSetSizes(*A,m,n,m,n);CHKERRQ(ierr); 3821 ierr = MatSetType(*A,MATSEQAIJ);CHKERRQ(ierr); 3822 ierr = MatSeqAIJSetPreallocation_SeqAIJ(*A,nz,nnz);CHKERRQ(ierr); 3823 PetscFunctionReturn(0); 3824 } 3825 3826 #undef __FUNCT__ 3827 #define __FUNCT__ "MatSeqAIJSetPreallocation" 3828 /*@C 3829 MatSeqAIJSetPreallocation - For good matrix assembly performance 3830 the user should preallocate the matrix storage by setting the parameter nz 3831 (or the array nnz). By setting these parameters accurately, performance 3832 during matrix assembly can be increased by more than a factor of 50. 3833 3834 Collective on MPI_Comm 3835 3836 Input Parameters: 3837 + B - The matrix-free 3838 . nz - number of nonzeros per row (same for all rows) 3839 - nnz - array containing the number of nonzeros in the various rows 3840 (possibly different for each row) or NULL 3841 3842 Notes: 3843 If nnz is given then nz is ignored 3844 3845 The AIJ format (also called the Yale sparse matrix format or 3846 compressed row storage), is fully compatible with standard Fortran 77 3847 storage. That is, the stored row and column indices can begin at 3848 either one (as in Fortran) or zero. See the users' manual for details. 3849 3850 Specify the preallocated storage with either nz or nnz (not both). 3851 Set nz=PETSC_DEFAULT and nnz=NULL for PETSc to control dynamic memory 3852 allocation. For large problems you MUST preallocate memory or you 3853 will get TERRIBLE performance, see the users' manual chapter on matrices. 3854 3855 You can call MatGetInfo() to get information on how effective the preallocation was; 3856 for example the fields mallocs,nz_allocated,nz_used,nz_unneeded; 3857 You can also run with the option -info and look for messages with the string 3858 malloc in them to see if additional memory allocation was needed. 3859 3860 Developers: Use nz of MAT_SKIP_ALLOCATION to not allocate any space for the matrix 3861 entries or columns indices 3862 3863 By default, this format uses inodes (identical nodes) when possible, to 3864 improve numerical efficiency of matrix-vector products and solves. We 3865 search for consecutive rows with the same nonzero structure, thereby 3866 reusing matrix information to achieve increased efficiency. 3867 3868 Options Database Keys: 3869 + -mat_no_inode - Do not use inodes 3870 . -mat_inode_limit <limit> - Sets inode limit (max limit=5) 3871 - -mat_aij_oneindex - Internally use indexing starting at 1 3872 rather than 0. Note that when calling MatSetValues(), 3873 the user still MUST index entries starting at 0! 3874 3875 Level: intermediate 3876 3877 .seealso: MatCreate(), MatCreateAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays(), MatGetInfo() 3878 3879 @*/ 3880 PetscErrorCode MatSeqAIJSetPreallocation(Mat B,PetscInt nz,const PetscInt nnz[]) 3881 { 3882 PetscErrorCode ierr; 3883 3884 PetscFunctionBegin; 3885 PetscValidHeaderSpecific(B,MAT_CLASSID,1); 3886 PetscValidType(B,1); 3887 ierr = PetscTryMethod(B,"MatSeqAIJSetPreallocation_C",(Mat,PetscInt,const PetscInt[]),(B,nz,nnz));CHKERRQ(ierr); 3888 PetscFunctionReturn(0); 3889 } 3890 3891 #undef __FUNCT__ 3892 #define __FUNCT__ "MatSeqAIJSetPreallocation_SeqAIJ" 3893 PetscErrorCode MatSeqAIJSetPreallocation_SeqAIJ(Mat B,PetscInt nz,const PetscInt *nnz) 3894 { 3895 Mat_SeqAIJ *b; 3896 PetscBool skipallocation = PETSC_FALSE,realalloc = PETSC_FALSE; 3897 PetscErrorCode ierr; 3898 PetscInt i; 3899 3900 PetscFunctionBegin; 3901 if (nz >= 0 || nnz) realalloc = PETSC_TRUE; 3902 if (nz == MAT_SKIP_ALLOCATION) { 3903 skipallocation = PETSC_TRUE; 3904 nz = 0; 3905 } 3906 3907 ierr = PetscLayoutSetUp(B->rmap);CHKERRQ(ierr); 3908 ierr = PetscLayoutSetUp(B->cmap);CHKERRQ(ierr); 3909 3910 if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5; 3911 if (nz < 0) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nz cannot be less than 0: value %d",nz); 3912 if (nnz) { 3913 for (i=0; i<B->rmap->n; i++) { 3914 if (nnz[i] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be less than 0: local row %d value %d",i,nnz[i]); 3915 if (nnz[i] > B->cmap->n) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be greater than row length: local row %d value %d rowlength %d",i,nnz[i],B->cmap->n); 3916 } 3917 } 3918 3919 B->preallocated = PETSC_TRUE; 3920 3921 b = (Mat_SeqAIJ*)B->data; 3922 3923 if (!skipallocation) { 3924 if (!b->imax) { 3925 ierr = PetscMalloc2(B->rmap->n,PetscInt,&b->imax,B->rmap->n,PetscInt,&b->ilen);CHKERRQ(ierr); 3926 ierr = PetscLogObjectMemory((PetscObject)B,2*B->rmap->n*sizeof(PetscInt));CHKERRQ(ierr); 3927 } 3928 if (!nnz) { 3929 if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 10; 3930 else if (nz < 0) nz = 1; 3931 for (i=0; i<B->rmap->n; i++) b->imax[i] = nz; 3932 nz = nz*B->rmap->n; 3933 } else { 3934 nz = 0; 3935 for (i=0; i<B->rmap->n; i++) {b->imax[i] = nnz[i]; nz += nnz[i];} 3936 } 3937 /* b->ilen will count nonzeros in each row so far. */ 3938 for (i=0; i<B->rmap->n; i++) b->ilen[i] = 0; 3939 3940 /* allocate the matrix space */ 3941 ierr = MatSeqXAIJFreeAIJ(B,&b->a,&b->j,&b->i);CHKERRQ(ierr); 3942 ierr = PetscMalloc3(nz,PetscScalar,&b->a,nz,PetscInt,&b->j,B->rmap->n+1,PetscInt,&b->i);CHKERRQ(ierr); 3943 ierr = PetscLogObjectMemory((PetscObject)B,(B->rmap->n+1)*sizeof(PetscInt)+nz*(sizeof(PetscScalar)+sizeof(PetscInt)));CHKERRQ(ierr); 3944 b->i[0] = 0; 3945 for (i=1; i<B->rmap->n+1; i++) { 3946 b->i[i] = b->i[i-1] + b->imax[i-1]; 3947 } 3948 b->singlemalloc = PETSC_TRUE; 3949 b->free_a = PETSC_TRUE; 3950 b->free_ij = PETSC_TRUE; 3951 #if defined(PETSC_THREADCOMM_ACTIVE) 3952 ierr = MatZeroEntries_SeqAIJ(B);CHKERRQ(ierr); 3953 #endif 3954 } else { 3955 b->free_a = PETSC_FALSE; 3956 b->free_ij = PETSC_FALSE; 3957 } 3958 3959 b->nz = 0; 3960 b->maxnz = nz; 3961 B->info.nz_unneeded = (double)b->maxnz; 3962 if (realalloc) { 3963 ierr = MatSetOption(B,MAT_NEW_NONZERO_ALLOCATION_ERR,PETSC_TRUE);CHKERRQ(ierr); 3964 } 3965 PetscFunctionReturn(0); 3966 } 3967 3968 #undef __FUNCT__ 3969 #define __FUNCT__ "MatSeqAIJSetPreallocationCSR" 3970 /*@ 3971 MatSeqAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in AIJ format. 3972 3973 Input Parameters: 3974 + B - the matrix 3975 . i - the indices into j for the start of each row (starts with zero) 3976 . j - the column indices for each row (starts with zero) these must be sorted for each row 3977 - v - optional values in the matrix 3978 3979 Level: developer 3980 3981 The i,j,v values are COPIED with this routine; to avoid the copy use MatCreateSeqAIJWithArrays() 3982 3983 .keywords: matrix, aij, compressed row, sparse, sequential 3984 3985 .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatSeqAIJSetPreallocation(), MatCreateSeqAIJ(), SeqAIJ 3986 @*/ 3987 PetscErrorCode MatSeqAIJSetPreallocationCSR(Mat B,const PetscInt i[],const PetscInt j[],const PetscScalar v[]) 3988 { 3989 PetscErrorCode ierr; 3990 3991 PetscFunctionBegin; 3992 PetscValidHeaderSpecific(B,MAT_CLASSID,1); 3993 PetscValidType(B,1); 3994 ierr = PetscTryMethod(B,"MatSeqAIJSetPreallocationCSR_C",(Mat,const PetscInt[],const PetscInt[],const PetscScalar[]),(B,i,j,v));CHKERRQ(ierr); 3995 PetscFunctionReturn(0); 3996 } 3997 3998 #undef __FUNCT__ 3999 #define __FUNCT__ "MatSeqAIJSetPreallocationCSR_SeqAIJ" 4000 PetscErrorCode MatSeqAIJSetPreallocationCSR_SeqAIJ(Mat B,const PetscInt Ii[],const PetscInt J[],const PetscScalar v[]) 4001 { 4002 PetscInt i; 4003 PetscInt m,n; 4004 PetscInt nz; 4005 PetscInt *nnz, nz_max = 0; 4006 PetscScalar *values; 4007 PetscErrorCode ierr; 4008 4009 PetscFunctionBegin; 4010 if (Ii[0]) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE, "Ii[0] must be 0 it is %D", Ii[0]); 4011 4012 ierr = PetscLayoutSetUp(B->rmap);CHKERRQ(ierr); 4013 ierr = PetscLayoutSetUp(B->cmap);CHKERRQ(ierr); 4014 4015 ierr = MatGetSize(B, &m, &n);CHKERRQ(ierr); 4016 ierr = PetscMalloc((m+1) * sizeof(PetscInt), &nnz);CHKERRQ(ierr); 4017 for (i = 0; i < m; i++) { 4018 nz = Ii[i+1]- Ii[i]; 4019 nz_max = PetscMax(nz_max, nz); 4020 if (nz < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE, "Local row %D has a negative number of columns %D", i, nnz); 4021 nnz[i] = nz; 4022 } 4023 ierr = MatSeqAIJSetPreallocation(B, 0, nnz);CHKERRQ(ierr); 4024 ierr = PetscFree(nnz);CHKERRQ(ierr); 4025 4026 if (v) { 4027 values = (PetscScalar*) v; 4028 } else { 4029 ierr = PetscMalloc(nz_max*sizeof(PetscScalar), &values);CHKERRQ(ierr); 4030 ierr = PetscMemzero(values, nz_max*sizeof(PetscScalar));CHKERRQ(ierr); 4031 } 4032 4033 for (i = 0; i < m; i++) { 4034 nz = Ii[i+1] - Ii[i]; 4035 ierr = MatSetValues_SeqAIJ(B, 1, &i, nz, J+Ii[i], values + (v ? Ii[i] : 0), INSERT_VALUES);CHKERRQ(ierr); 4036 } 4037 4038 ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4039 ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4040 4041 if (!v) { 4042 ierr = PetscFree(values);CHKERRQ(ierr); 4043 } 4044 ierr = MatSetOption(B,MAT_NEW_NONZERO_LOCATION_ERR,PETSC_TRUE);CHKERRQ(ierr); 4045 PetscFunctionReturn(0); 4046 } 4047 4048 #include <../src/mat/impls/dense/seq/dense.h> 4049 #include <petsc-private/kernels/petscaxpy.h> 4050 4051 #undef __FUNCT__ 4052 #define __FUNCT__ "MatMatMultNumeric_SeqDense_SeqAIJ" 4053 /* 4054 Computes (B'*A')' since computing B*A directly is untenable 4055 4056 n p p 4057 ( ) ( ) ( ) 4058 m ( A ) * n ( B ) = m ( C ) 4059 ( ) ( ) ( ) 4060 4061 */ 4062 PetscErrorCode MatMatMultNumeric_SeqDense_SeqAIJ(Mat A,Mat B,Mat C) 4063 { 4064 PetscErrorCode ierr; 4065 Mat_SeqDense *sub_a = (Mat_SeqDense*)A->data; 4066 Mat_SeqAIJ *sub_b = (Mat_SeqAIJ*)B->data; 4067 Mat_SeqDense *sub_c = (Mat_SeqDense*)C->data; 4068 PetscInt i,n,m,q,p; 4069 const PetscInt *ii,*idx; 4070 const PetscScalar *b,*a,*a_q; 4071 PetscScalar *c,*c_q; 4072 4073 PetscFunctionBegin; 4074 m = A->rmap->n; 4075 n = A->cmap->n; 4076 p = B->cmap->n; 4077 a = sub_a->v; 4078 b = sub_b->a; 4079 c = sub_c->v; 4080 ierr = PetscMemzero(c,m*p*sizeof(PetscScalar));CHKERRQ(ierr); 4081 4082 ii = sub_b->i; 4083 idx = sub_b->j; 4084 for (i=0; i<n; i++) { 4085 q = ii[i+1] - ii[i]; 4086 while (q-->0) { 4087 c_q = c + m*(*idx); 4088 a_q = a + m*i; 4089 PetscKernelAXPY(c_q,*b,a_q,m); 4090 idx++; 4091 b++; 4092 } 4093 } 4094 PetscFunctionReturn(0); 4095 } 4096 4097 #undef __FUNCT__ 4098 #define __FUNCT__ "MatMatMultSymbolic_SeqDense_SeqAIJ" 4099 PetscErrorCode MatMatMultSymbolic_SeqDense_SeqAIJ(Mat A,Mat B,PetscReal fill,Mat *C) 4100 { 4101 PetscErrorCode ierr; 4102 PetscInt m=A->rmap->n,n=B->cmap->n; 4103 Mat Cmat; 4104 4105 PetscFunctionBegin; 4106 if (A->cmap->n != B->rmap->n) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"A->cmap->n %d != B->rmap->n %d\n",A->cmap->n,B->rmap->n); 4107 ierr = MatCreate(PetscObjectComm((PetscObject)A),&Cmat);CHKERRQ(ierr); 4108 ierr = MatSetSizes(Cmat,m,n,m,n);CHKERRQ(ierr); 4109 ierr = MatSetBlockSizes(Cmat,A->rmap->bs,B->cmap->bs);CHKERRQ(ierr); 4110 ierr = MatSetType(Cmat,MATSEQDENSE);CHKERRQ(ierr); 4111 ierr = MatSeqDenseSetPreallocation(Cmat,NULL);CHKERRQ(ierr); 4112 4113 Cmat->ops->matmultnumeric = MatMatMultNumeric_SeqDense_SeqAIJ; 4114 4115 *C = Cmat; 4116 PetscFunctionReturn(0); 4117 } 4118 4119 /* ----------------------------------------------------------------*/ 4120 #undef __FUNCT__ 4121 #define __FUNCT__ "MatMatMult_SeqDense_SeqAIJ" 4122 PetscErrorCode MatMatMult_SeqDense_SeqAIJ(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C) 4123 { 4124 PetscErrorCode ierr; 4125 4126 PetscFunctionBegin; 4127 if (scall == MAT_INITIAL_MATRIX) { 4128 ierr = PetscLogEventBegin(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr); 4129 ierr = MatMatMultSymbolic_SeqDense_SeqAIJ(A,B,fill,C);CHKERRQ(ierr); 4130 ierr = PetscLogEventEnd(MAT_MatMultSymbolic,A,B,0,0);CHKERRQ(ierr); 4131 } 4132 ierr = PetscLogEventBegin(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr); 4133 ierr = MatMatMultNumeric_SeqDense_SeqAIJ(A,B,*C);CHKERRQ(ierr); 4134 ierr = PetscLogEventEnd(MAT_MatMultNumeric,A,B,0,0);CHKERRQ(ierr); 4135 PetscFunctionReturn(0); 4136 } 4137 4138 4139 /*MC 4140 MATSEQAIJ - MATSEQAIJ = "seqaij" - A matrix type to be used for sequential sparse matrices, 4141 based on compressed sparse row format. 4142 4143 Options Database Keys: 4144 . -mat_type seqaij - sets the matrix type to "seqaij" during a call to MatSetFromOptions() 4145 4146 Level: beginner 4147 4148 .seealso: MatCreateSeqAIJ(), MatSetFromOptions(), MatSetType(), MatCreate(), MatType 4149 M*/ 4150 4151 /*MC 4152 MATAIJ - MATAIJ = "aij" - A matrix type to be used for sparse matrices. 4153 4154 This matrix type is identical to MATSEQAIJ when constructed with a single process communicator, 4155 and MATMPIAIJ otherwise. As a result, for single process communicators, 4156 MatSeqAIJSetPreallocation is supported, and similarly MatMPIAIJSetPreallocation is supported 4157 for communicators controlling multiple processes. It is recommended that you call both of 4158 the above preallocation routines for simplicity. 4159 4160 Options Database Keys: 4161 . -mat_type aij - sets the matrix type to "aij" during a call to MatSetFromOptions() 4162 4163 Developer Notes: Subclasses include MATAIJCUSP, MATAIJPERM, MATAIJCRL, and also automatically switches over to use inodes when 4164 enough exist. 4165 4166 Level: beginner 4167 4168 .seealso: MatCreateAIJ(), MatCreateSeqAIJ(), MATSEQAIJ,MATMPIAIJ 4169 M*/ 4170 4171 /*MC 4172 MATAIJCRL - MATAIJCRL = "aijcrl" - A matrix type to be used for sparse matrices. 4173 4174 This matrix type is identical to MATSEQAIJCRL when constructed with a single process communicator, 4175 and MATMPIAIJCRL otherwise. As a result, for single process communicators, 4176 MatSeqAIJSetPreallocation() is supported, and similarly MatMPIAIJSetPreallocation() is supported 4177 for communicators controlling multiple processes. It is recommended that you call both of 4178 the above preallocation routines for simplicity. 4179 4180 Options Database Keys: 4181 . -mat_type aijcrl - sets the matrix type to "aijcrl" during a call to MatSetFromOptions() 4182 4183 Level: beginner 4184 4185 .seealso: MatCreateMPIAIJCRL,MATSEQAIJCRL,MATMPIAIJCRL, MATSEQAIJCRL, MATMPIAIJCRL 4186 M*/ 4187 4188 #if defined(PETSC_HAVE_PASTIX) 4189 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_pastix(Mat,MatFactorType,Mat*); 4190 #endif 4191 #if defined(PETSC_HAVE_ESSL) && !defined(PETSC_USE_COMPLEX) && !defined(PETSC_USE_REAL_SINGLE) && !defined(PETSC_USE_REAL___FLOAT128) 4192 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_essl(Mat,MatFactorType,Mat*); 4193 #endif 4194 PETSC_EXTERN PetscErrorCode MatConvert_SeqAIJ_SeqAIJCRL(Mat,MatType,MatReuse,Mat*); 4195 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_petsc(Mat,MatFactorType,Mat*); 4196 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_bas(Mat,MatFactorType,Mat*); 4197 extern PetscErrorCode MatGetFactorAvailable_seqaij_petsc(Mat,MatFactorType,PetscBool*); 4198 #if defined(PETSC_HAVE_MUMPS) 4199 PETSC_EXTERN PetscErrorCode MatGetFactor_aij_mumps(Mat,MatFactorType,Mat*); 4200 #endif 4201 #if defined(PETSC_HAVE_SUPERLU) 4202 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_superlu(Mat,MatFactorType,Mat*); 4203 #endif 4204 #if defined(PETSC_HAVE_SUPERLU_DIST) 4205 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_superlu_dist(Mat,MatFactorType,Mat*); 4206 #endif 4207 #if defined(PETSC_HAVE_UMFPACK) 4208 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_umfpack(Mat,MatFactorType,Mat*); 4209 #endif 4210 #if defined(PETSC_HAVE_CHOLMOD) 4211 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_cholmod(Mat,MatFactorType,Mat*); 4212 #endif 4213 #if defined(PETSC_HAVE_LUSOL) 4214 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_lusol(Mat,MatFactorType,Mat*); 4215 #endif 4216 #if defined(PETSC_HAVE_MATLAB_ENGINE) 4217 PETSC_EXTERN PetscErrorCode MatGetFactor_seqaij_matlab(Mat,MatFactorType,Mat*); 4218 extern PetscErrorCode MatlabEnginePut_SeqAIJ(PetscObject,void*); 4219 extern PetscErrorCode MatlabEngineGet_SeqAIJ(PetscObject,void*); 4220 #endif 4221 #if defined(PETSC_HAVE_CLIQUE) 4222 PETSC_EXTERN PetscErrorCode MatGetFactor_aij_clique(Mat,MatFactorType,Mat*); 4223 #endif 4224 4225 4226 #undef __FUNCT__ 4227 #define __FUNCT__ "MatSeqAIJGetArray" 4228 /*@C 4229 MatSeqAIJGetArray - gives access to the array where the data for a SeqSeqAIJ matrix is stored 4230 4231 Not Collective 4232 4233 Input Parameter: 4234 . mat - a MATSEQDENSE matrix 4235 4236 Output Parameter: 4237 . array - pointer to the data 4238 4239 Level: intermediate 4240 4241 .seealso: MatSeqAIJRestoreArray(), MatSeqAIJGetArrayF90() 4242 @*/ 4243 PetscErrorCode MatSeqAIJGetArray(Mat A,PetscScalar **array) 4244 { 4245 PetscErrorCode ierr; 4246 4247 PetscFunctionBegin; 4248 ierr = PetscUseMethod(A,"MatSeqAIJGetArray_C",(Mat,PetscScalar**),(A,array));CHKERRQ(ierr); 4249 PetscFunctionReturn(0); 4250 } 4251 4252 #undef __FUNCT__ 4253 #define __FUNCT__ "MatSeqAIJRestoreArray" 4254 /*@C 4255 MatSeqAIJRestoreArray - returns access to the array where the data for a SeqSeqAIJ matrix is stored obtained by MatSeqAIJGetArray() 4256 4257 Not Collective 4258 4259 Input Parameters: 4260 . mat - a MATSEQDENSE matrix 4261 . array - pointer to the data 4262 4263 Level: intermediate 4264 4265 .seealso: MatSeqAIJGetArray(), MatSeqAIJRestoreArrayF90() 4266 @*/ 4267 PetscErrorCode MatSeqAIJRestoreArray(Mat A,PetscScalar **array) 4268 { 4269 PetscErrorCode ierr; 4270 4271 PetscFunctionBegin; 4272 ierr = PetscUseMethod(A,"MatSeqAIJRestoreArray_C",(Mat,PetscScalar**),(A,array));CHKERRQ(ierr); 4273 PetscFunctionReturn(0); 4274 } 4275 4276 #undef __FUNCT__ 4277 #define __FUNCT__ "MatCreate_SeqAIJ" 4278 PETSC_EXTERN PetscErrorCode MatCreate_SeqAIJ(Mat B) 4279 { 4280 Mat_SeqAIJ *b; 4281 PetscErrorCode ierr; 4282 PetscMPIInt size; 4283 4284 PetscFunctionBegin; 4285 ierr = MPI_Comm_size(PetscObjectComm((PetscObject)B),&size);CHKERRQ(ierr); 4286 if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Comm must be of size 1"); 4287 4288 ierr = PetscNewLog(B,Mat_SeqAIJ,&b);CHKERRQ(ierr); 4289 4290 B->data = (void*)b; 4291 4292 ierr = PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));CHKERRQ(ierr); 4293 4294 b->row = 0; 4295 b->col = 0; 4296 b->icol = 0; 4297 b->reallocs = 0; 4298 b->ignorezeroentries = PETSC_FALSE; 4299 b->roworiented = PETSC_TRUE; 4300 b->nonew = 0; 4301 b->diag = 0; 4302 b->solve_work = 0; 4303 B->spptr = 0; 4304 b->saved_values = 0; 4305 b->idiag = 0; 4306 b->mdiag = 0; 4307 b->ssor_work = 0; 4308 b->omega = 1.0; 4309 b->fshift = 0.0; 4310 b->idiagvalid = PETSC_FALSE; 4311 b->ibdiagvalid = PETSC_FALSE; 4312 b->keepnonzeropattern = PETSC_FALSE; 4313 b->xtoy = 0; 4314 b->XtoY = 0; 4315 B->same_nonzero = PETSC_FALSE; 4316 4317 ierr = PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);CHKERRQ(ierr); 4318 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJGetArray_C",MatSeqAIJGetArray_SeqAIJ);CHKERRQ(ierr); 4319 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJRestoreArray_C",MatSeqAIJRestoreArray_SeqAIJ);CHKERRQ(ierr); 4320 4321 #if defined(PETSC_HAVE_MATLAB_ENGINE) 4322 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_matlab_C",MatGetFactor_seqaij_matlab);CHKERRQ(ierr); 4323 ierr = PetscObjectComposeFunction((PetscObject)B,"PetscMatlabEnginePut_C",MatlabEnginePut_SeqAIJ);CHKERRQ(ierr); 4324 ierr = PetscObjectComposeFunction((PetscObject)B,"PetscMatlabEngineGet_C",MatlabEngineGet_SeqAIJ);CHKERRQ(ierr); 4325 #endif 4326 #if defined(PETSC_HAVE_PASTIX) 4327 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_pastix_C",MatGetFactor_seqaij_pastix);CHKERRQ(ierr); 4328 #endif 4329 #if defined(PETSC_HAVE_ESSL) && !defined(PETSC_USE_COMPLEX) && !defined(PETSC_USE_REAL_SINGLE) && !defined(PETSC_USE_REAL___FLOAT128) 4330 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_essl_C",MatGetFactor_seqaij_essl);CHKERRQ(ierr); 4331 #endif 4332 #if defined(PETSC_HAVE_SUPERLU) 4333 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_superlu_C",MatGetFactor_seqaij_superlu);CHKERRQ(ierr); 4334 #endif 4335 #if defined(PETSC_HAVE_SUPERLU_DIST) 4336 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_superlu_dist_C",MatGetFactor_seqaij_superlu_dist);CHKERRQ(ierr); 4337 #endif 4338 #if defined(PETSC_HAVE_MUMPS) 4339 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_mumps_C",MatGetFactor_aij_mumps);CHKERRQ(ierr); 4340 #endif 4341 #if defined(PETSC_HAVE_UMFPACK) 4342 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_umfpack_C",MatGetFactor_seqaij_umfpack);CHKERRQ(ierr); 4343 #endif 4344 #if defined(PETSC_HAVE_CHOLMOD) 4345 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_cholmod_C",MatGetFactor_seqaij_cholmod);CHKERRQ(ierr); 4346 #endif 4347 #if defined(PETSC_HAVE_LUSOL) 4348 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_lusol_C",MatGetFactor_seqaij_lusol);CHKERRQ(ierr); 4349 #endif 4350 #if defined(PETSC_HAVE_CLIQUE) 4351 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_clique_C",MatGetFactor_aij_clique);CHKERRQ(ierr); 4352 #endif 4353 4354 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_petsc_C",MatGetFactor_seqaij_petsc);CHKERRQ(ierr); 4355 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactorAvailable_petsc_C",MatGetFactorAvailable_seqaij_petsc);CHKERRQ(ierr); 4356 ierr = PetscObjectComposeFunction((PetscObject)B,"MatGetFactor_bas_C",MatGetFactor_seqaij_bas);CHKERRQ(ierr); 4357 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetColumnIndices_C",MatSeqAIJSetColumnIndices_SeqAIJ);CHKERRQ(ierr); 4358 ierr = PetscObjectComposeFunction((PetscObject)B,"MatStoreValues_C",MatStoreValues_SeqAIJ);CHKERRQ(ierr); 4359 ierr = PetscObjectComposeFunction((PetscObject)B,"MatRetrieveValues_C",MatRetrieveValues_SeqAIJ);CHKERRQ(ierr); 4360 ierr = PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqsbaij_C",MatConvert_SeqAIJ_SeqSBAIJ);CHKERRQ(ierr); 4361 ierr = PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqbaij_C",MatConvert_SeqAIJ_SeqBAIJ);CHKERRQ(ierr); 4362 ierr = PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijperm_C",MatConvert_SeqAIJ_SeqAIJPERM);CHKERRQ(ierr); 4363 ierr = PetscObjectComposeFunction((PetscObject)B,"MatConvert_seqaij_seqaijcrl_C",MatConvert_SeqAIJ_SeqAIJCRL);CHKERRQ(ierr); 4364 ierr = PetscObjectComposeFunction((PetscObject)B,"MatIsTranspose_C",MatIsTranspose_SeqAIJ);CHKERRQ(ierr); 4365 ierr = PetscObjectComposeFunction((PetscObject)B,"MatIsHermitianTranspose_C",MatIsTranspose_SeqAIJ);CHKERRQ(ierr); 4366 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetPreallocation_C",MatSeqAIJSetPreallocation_SeqAIJ);CHKERRQ(ierr); 4367 ierr = PetscObjectComposeFunction((PetscObject)B,"MatSeqAIJSetPreallocationCSR_C",MatSeqAIJSetPreallocationCSR_SeqAIJ);CHKERRQ(ierr); 4368 ierr = PetscObjectComposeFunction((PetscObject)B,"MatReorderForNonzeroDiagonal_C",MatReorderForNonzeroDiagonal_SeqAIJ);CHKERRQ(ierr); 4369 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMatMult_seqdense_seqaij_C",MatMatMult_SeqDense_SeqAIJ);CHKERRQ(ierr); 4370 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMatMultSymbolic_seqdense_seqaij_C",MatMatMultSymbolic_SeqDense_SeqAIJ);CHKERRQ(ierr); 4371 ierr = PetscObjectComposeFunction((PetscObject)B,"MatMatMultNumeric_seqdense_seqaij_C",MatMatMultNumeric_SeqDense_SeqAIJ);CHKERRQ(ierr); 4372 ierr = MatCreate_SeqAIJ_Inode(B);CHKERRQ(ierr); 4373 ierr = PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);CHKERRQ(ierr); 4374 PetscFunctionReturn(0); 4375 } 4376 4377 #undef __FUNCT__ 4378 #define __FUNCT__ "MatDuplicateNoCreate_SeqAIJ" 4379 /* 4380 Given a matrix generated with MatGetFactor() duplicates all the information in A into B 4381 */ 4382 PetscErrorCode MatDuplicateNoCreate_SeqAIJ(Mat C,Mat A,MatDuplicateOption cpvalues,PetscBool mallocmatspace) 4383 { 4384 Mat_SeqAIJ *c,*a = (Mat_SeqAIJ*)A->data; 4385 PetscErrorCode ierr; 4386 PetscInt i,m = A->rmap->n; 4387 4388 PetscFunctionBegin; 4389 c = (Mat_SeqAIJ*)C->data; 4390 4391 C->factortype = A->factortype; 4392 c->row = 0; 4393 c->col = 0; 4394 c->icol = 0; 4395 c->reallocs = 0; 4396 4397 C->assembled = PETSC_TRUE; 4398 4399 ierr = PetscLayoutReference(A->rmap,&C->rmap);CHKERRQ(ierr); 4400 ierr = PetscLayoutReference(A->cmap,&C->cmap);CHKERRQ(ierr); 4401 4402 ierr = PetscMalloc2(m,PetscInt,&c->imax,m,PetscInt,&c->ilen);CHKERRQ(ierr); 4403 ierr = PetscLogObjectMemory((PetscObject)C, 2*m*sizeof(PetscInt));CHKERRQ(ierr); 4404 for (i=0; i<m; i++) { 4405 c->imax[i] = a->imax[i]; 4406 c->ilen[i] = a->ilen[i]; 4407 } 4408 4409 /* allocate the matrix space */ 4410 if (mallocmatspace) { 4411 ierr = PetscMalloc3(a->i[m],PetscScalar,&c->a,a->i[m],PetscInt,&c->j,m+1,PetscInt,&c->i);CHKERRQ(ierr); 4412 ierr = PetscLogObjectMemory((PetscObject)C, a->i[m]*(sizeof(PetscScalar)+sizeof(PetscInt))+(m+1)*sizeof(PetscInt));CHKERRQ(ierr); 4413 4414 c->singlemalloc = PETSC_TRUE; 4415 4416 ierr = PetscMemcpy(c->i,a->i,(m+1)*sizeof(PetscInt));CHKERRQ(ierr); 4417 if (m > 0) { 4418 ierr = PetscMemcpy(c->j,a->j,(a->i[m])*sizeof(PetscInt));CHKERRQ(ierr); 4419 if (cpvalues == MAT_COPY_VALUES) { 4420 ierr = PetscMemcpy(c->a,a->a,(a->i[m])*sizeof(PetscScalar));CHKERRQ(ierr); 4421 } else { 4422 ierr = PetscMemzero(c->a,(a->i[m])*sizeof(PetscScalar));CHKERRQ(ierr); 4423 } 4424 } 4425 } 4426 4427 c->ignorezeroentries = a->ignorezeroentries; 4428 c->roworiented = a->roworiented; 4429 c->nonew = a->nonew; 4430 if (a->diag) { 4431 ierr = PetscMalloc((m+1)*sizeof(PetscInt),&c->diag);CHKERRQ(ierr); 4432 ierr = PetscLogObjectMemory((PetscObject)C,(m+1)*sizeof(PetscInt));CHKERRQ(ierr); 4433 for (i=0; i<m; i++) { 4434 c->diag[i] = a->diag[i]; 4435 } 4436 } else c->diag = 0; 4437 4438 c->solve_work = 0; 4439 c->saved_values = 0; 4440 c->idiag = 0; 4441 c->ssor_work = 0; 4442 c->keepnonzeropattern = a->keepnonzeropattern; 4443 c->free_a = PETSC_TRUE; 4444 c->free_ij = PETSC_TRUE; 4445 c->xtoy = 0; 4446 c->XtoY = 0; 4447 4448 c->rmax = a->rmax; 4449 c->nz = a->nz; 4450 c->maxnz = a->nz; /* Since we allocate exactly the right amount */ 4451 C->preallocated = PETSC_TRUE; 4452 4453 c->compressedrow.use = a->compressedrow.use; 4454 c->compressedrow.nrows = a->compressedrow.nrows; 4455 c->compressedrow.check = a->compressedrow.check; 4456 if (a->compressedrow.use) { 4457 i = a->compressedrow.nrows; 4458 ierr = PetscMalloc2(i+1,PetscInt,&c->compressedrow.i,i,PetscInt,&c->compressedrow.rindex);CHKERRQ(ierr); 4459 ierr = PetscMemcpy(c->compressedrow.i,a->compressedrow.i,(i+1)*sizeof(PetscInt));CHKERRQ(ierr); 4460 ierr = PetscMemcpy(c->compressedrow.rindex,a->compressedrow.rindex,i*sizeof(PetscInt));CHKERRQ(ierr); 4461 } else { 4462 c->compressedrow.use = PETSC_FALSE; 4463 c->compressedrow.i = NULL; 4464 c->compressedrow.rindex = NULL; 4465 } 4466 C->same_nonzero = A->same_nonzero; 4467 4468 ierr = MatDuplicate_SeqAIJ_Inode(A,cpvalues,&C);CHKERRQ(ierr); 4469 ierr = PetscFunctionListDuplicate(((PetscObject)A)->qlist,&((PetscObject)C)->qlist);CHKERRQ(ierr); 4470 PetscFunctionReturn(0); 4471 } 4472 4473 #undef __FUNCT__ 4474 #define __FUNCT__ "MatDuplicate_SeqAIJ" 4475 PetscErrorCode MatDuplicate_SeqAIJ(Mat A,MatDuplicateOption cpvalues,Mat *B) 4476 { 4477 PetscErrorCode ierr; 4478 4479 PetscFunctionBegin; 4480 ierr = MatCreate(PetscObjectComm((PetscObject)A),B);CHKERRQ(ierr); 4481 ierr = MatSetSizes(*B,A->rmap->n,A->cmap->n,A->rmap->n,A->cmap->n);CHKERRQ(ierr); 4482 ierr = MatSetBlockSizes(*B,A->rmap->bs,A->cmap->bs);CHKERRQ(ierr); 4483 ierr = MatSetType(*B,((PetscObject)A)->type_name);CHKERRQ(ierr); 4484 ierr = MatDuplicateNoCreate_SeqAIJ(*B,A,cpvalues,PETSC_TRUE);CHKERRQ(ierr); 4485 PetscFunctionReturn(0); 4486 } 4487 4488 #undef __FUNCT__ 4489 #define __FUNCT__ "MatLoad_SeqAIJ" 4490 PetscErrorCode MatLoad_SeqAIJ(Mat newMat, PetscViewer viewer) 4491 { 4492 Mat_SeqAIJ *a; 4493 PetscErrorCode ierr; 4494 PetscInt i,sum,nz,header[4],*rowlengths = 0,M,N,rows,cols; 4495 int fd; 4496 PetscMPIInt size; 4497 MPI_Comm comm; 4498 PetscInt bs = 1; 4499 4500 PetscFunctionBegin; 4501 ierr = PetscObjectGetComm((PetscObject)viewer,&comm);CHKERRQ(ierr); 4502 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 4503 if (size > 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_SIZ,"view must have one processor"); 4504 4505 ierr = PetscOptionsBegin(comm,NULL,"Options for loading SEQAIJ matrix","Mat");CHKERRQ(ierr); 4506 ierr = PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,NULL);CHKERRQ(ierr); 4507 ierr = PetscOptionsEnd();CHKERRQ(ierr); 4508 if (bs > 1) {ierr = MatSetBlockSize(newMat,bs);CHKERRQ(ierr);} 4509 4510 ierr = PetscViewerBinaryGetDescriptor(viewer,&fd);CHKERRQ(ierr); 4511 ierr = PetscBinaryRead(fd,header,4,PETSC_INT);CHKERRQ(ierr); 4512 if (header[0] != MAT_FILE_CLASSID) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"not matrix object in file"); 4513 M = header[1]; N = header[2]; nz = header[3]; 4514 4515 if (nz < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format on disk,cannot load as SeqAIJ"); 4516 4517 /* read in row lengths */ 4518 ierr = PetscMalloc(M*sizeof(PetscInt),&rowlengths);CHKERRQ(ierr); 4519 ierr = PetscBinaryRead(fd,rowlengths,M,PETSC_INT);CHKERRQ(ierr); 4520 4521 /* check if sum of rowlengths is same as nz */ 4522 for (i=0,sum=0; i< M; i++) sum +=rowlengths[i]; 4523 if (sum != nz) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_FILE_READ,"Inconsistant matrix data in file. no-nonzeros = %d, sum-row-lengths = %d\n",nz,sum); 4524 4525 /* set global size if not set already*/ 4526 if (newMat->rmap->n < 0 && newMat->rmap->N < 0 && newMat->cmap->n < 0 && newMat->cmap->N < 0) { 4527 ierr = MatSetSizes(newMat,PETSC_DECIDE,PETSC_DECIDE,M,N);CHKERRQ(ierr); 4528 } else { 4529 /* if sizes and type are already set, check if the vector global sizes are correct */ 4530 ierr = MatGetSize(newMat,&rows,&cols);CHKERRQ(ierr); 4531 if (rows < 0 && cols < 0) { /* user might provide local size instead of global size */ 4532 ierr = MatGetLocalSize(newMat,&rows,&cols);CHKERRQ(ierr); 4533 } 4534 if (M != rows || N != cols) SETERRQ4(PETSC_COMM_SELF,PETSC_ERR_FILE_UNEXPECTED, "Matrix in file of different length (%d, %d) than the input matrix (%d, %d)",M,N,rows,cols); 4535 } 4536 ierr = MatSeqAIJSetPreallocation_SeqAIJ(newMat,0,rowlengths);CHKERRQ(ierr); 4537 a = (Mat_SeqAIJ*)newMat->data; 4538 4539 ierr = PetscBinaryRead(fd,a->j,nz,PETSC_INT);CHKERRQ(ierr); 4540 4541 /* read in nonzero values */ 4542 ierr = PetscBinaryRead(fd,a->a,nz,PETSC_SCALAR);CHKERRQ(ierr); 4543 4544 /* set matrix "i" values */ 4545 a->i[0] = 0; 4546 for (i=1; i<= M; i++) { 4547 a->i[i] = a->i[i-1] + rowlengths[i-1]; 4548 a->ilen[i-1] = rowlengths[i-1]; 4549 } 4550 ierr = PetscFree(rowlengths);CHKERRQ(ierr); 4551 4552 ierr = MatAssemblyBegin(newMat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4553 ierr = MatAssemblyEnd(newMat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4554 PetscFunctionReturn(0); 4555 } 4556 4557 #undef __FUNCT__ 4558 #define __FUNCT__ "MatEqual_SeqAIJ" 4559 PetscErrorCode MatEqual_SeqAIJ(Mat A,Mat B,PetscBool * flg) 4560 { 4561 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data,*b = (Mat_SeqAIJ*)B->data; 4562 PetscErrorCode ierr; 4563 #if defined(PETSC_USE_COMPLEX) 4564 PetscInt k; 4565 #endif 4566 4567 PetscFunctionBegin; 4568 /* If the matrix dimensions are not equal,or no of nonzeros */ 4569 if ((A->rmap->n != B->rmap->n) || (A->cmap->n != B->cmap->n) ||(a->nz != b->nz)) { 4570 *flg = PETSC_FALSE; 4571 PetscFunctionReturn(0); 4572 } 4573 4574 /* if the a->i are the same */ 4575 ierr = PetscMemcmp(a->i,b->i,(A->rmap->n+1)*sizeof(PetscInt),flg);CHKERRQ(ierr); 4576 if (!*flg) PetscFunctionReturn(0); 4577 4578 /* if a->j are the same */ 4579 ierr = PetscMemcmp(a->j,b->j,(a->nz)*sizeof(PetscInt),flg);CHKERRQ(ierr); 4580 if (!*flg) PetscFunctionReturn(0); 4581 4582 /* if a->a are the same */ 4583 #if defined(PETSC_USE_COMPLEX) 4584 for (k=0; k<a->nz; k++) { 4585 if (PetscRealPart(a->a[k]) != PetscRealPart(b->a[k]) || PetscImaginaryPart(a->a[k]) != PetscImaginaryPart(b->a[k])) { 4586 *flg = PETSC_FALSE; 4587 PetscFunctionReturn(0); 4588 } 4589 } 4590 #else 4591 ierr = PetscMemcmp(a->a,b->a,(a->nz)*sizeof(PetscScalar),flg);CHKERRQ(ierr); 4592 #endif 4593 PetscFunctionReturn(0); 4594 } 4595 4596 #undef __FUNCT__ 4597 #define __FUNCT__ "MatCreateSeqAIJWithArrays" 4598 /*@ 4599 MatCreateSeqAIJWithArrays - Creates an sequential AIJ matrix using matrix elements (in CSR format) 4600 provided by the user. 4601 4602 Collective on MPI_Comm 4603 4604 Input Parameters: 4605 + comm - must be an MPI communicator of size 1 4606 . m - number of rows 4607 . n - number of columns 4608 . i - row indices 4609 . j - column indices 4610 - a - matrix values 4611 4612 Output Parameter: 4613 . mat - the matrix 4614 4615 Level: intermediate 4616 4617 Notes: 4618 The i, j, and a arrays are not copied by this routine, the user must free these arrays 4619 once the matrix is destroyed and not before 4620 4621 You cannot set new nonzero locations into this matrix, that will generate an error. 4622 4623 The i and j indices are 0 based 4624 4625 The format which is used for the sparse matrix input, is equivalent to a 4626 row-major ordering.. i.e for the following matrix, the input data expected is 4627 as shown: 4628 4629 1 0 0 4630 2 0 3 4631 4 5 6 4632 4633 i = {0,1,3,6} [size = nrow+1 = 3+1] 4634 j = {0,0,2,0,1,2} [size = nz = 6]; values must be sorted for each row 4635 v = {1,2,3,4,5,6} [size = nz = 6] 4636 4637 4638 .seealso: MatCreate(), MatCreateAIJ(), MatCreateSeqAIJ(), MatCreateMPIAIJWithArrays(), MatMPIAIJSetPreallocationCSR() 4639 4640 @*/ 4641 PetscErrorCode MatCreateSeqAIJWithArrays(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt *i,PetscInt *j,PetscScalar *a,Mat *mat) 4642 { 4643 PetscErrorCode ierr; 4644 PetscInt ii; 4645 Mat_SeqAIJ *aij; 4646 #if defined(PETSC_USE_DEBUG) 4647 PetscInt jj; 4648 #endif 4649 4650 PetscFunctionBegin; 4651 if (i[0]) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0"); 4652 ierr = MatCreate(comm,mat);CHKERRQ(ierr); 4653 ierr = MatSetSizes(*mat,m,n,m,n);CHKERRQ(ierr); 4654 /* ierr = MatSetBlockSizes(*mat,,);CHKERRQ(ierr); */ 4655 ierr = MatSetType(*mat,MATSEQAIJ);CHKERRQ(ierr); 4656 ierr = MatSeqAIJSetPreallocation_SeqAIJ(*mat,MAT_SKIP_ALLOCATION,0);CHKERRQ(ierr); 4657 aij = (Mat_SeqAIJ*)(*mat)->data; 4658 ierr = PetscMalloc2(m,PetscInt,&aij->imax,m,PetscInt,&aij->ilen);CHKERRQ(ierr); 4659 4660 aij->i = i; 4661 aij->j = j; 4662 aij->a = a; 4663 aij->singlemalloc = PETSC_FALSE; 4664 aij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/ 4665 aij->free_a = PETSC_FALSE; 4666 aij->free_ij = PETSC_FALSE; 4667 4668 for (ii=0; ii<m; ii++) { 4669 aij->ilen[ii] = aij->imax[ii] = i[ii+1] - i[ii]; 4670 #if defined(PETSC_USE_DEBUG) 4671 if (i[ii+1] - i[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative row length in i (row indices) row = %d length = %d",ii,i[ii+1] - i[ii]); 4672 for (jj=i[ii]+1; jj<i[ii+1]; jj++) { 4673 if (j[jj] < j[jj-1]) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column entry number %D (actual colum %D) in row %D is not sorted",jj-i[ii],j[jj],ii); 4674 if (j[jj] == j[jj]-1) SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column entry number %D (actual colum %D) in row %D is identical to previous entry",jj-i[ii],j[jj],ii); 4675 } 4676 #endif 4677 } 4678 #if defined(PETSC_USE_DEBUG) 4679 for (ii=0; ii<aij->i[m]; ii++) { 4680 if (j[ii] < 0) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Negative column index at location = %d index = %d",ii,j[ii]); 4681 if (j[ii] > n - 1) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Column index to large at location = %d index = %d",ii,j[ii]); 4682 } 4683 #endif 4684 4685 ierr = MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4686 ierr = MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4687 PetscFunctionReturn(0); 4688 } 4689 #undef __FUNCT__ 4690 #define __FUNCT__ "MatCreateSeqAIJFromTriple" 4691 /*@C 4692 MatCreateSeqAIJFromTriple - Creates an sequential AIJ matrix using matrix elements (in COO format) 4693 provided by the user. 4694 4695 Collective on MPI_Comm 4696 4697 Input Parameters: 4698 + comm - must be an MPI communicator of size 1 4699 . m - number of rows 4700 . n - number of columns 4701 . i - row indices 4702 . j - column indices 4703 . a - matrix values 4704 . nz - number of nonzeros 4705 - idx - 0 or 1 based 4706 4707 Output Parameter: 4708 . mat - the matrix 4709 4710 Level: intermediate 4711 4712 Notes: 4713 The i and j indices are 0 based 4714 4715 The format which is used for the sparse matrix input, is equivalent to a 4716 row-major ordering.. i.e for the following matrix, the input data expected is 4717 as shown: 4718 4719 1 0 0 4720 2 0 3 4721 4 5 6 4722 4723 i = {0,1,1,2,2,2} 4724 j = {0,0,2,0,1,2} 4725 v = {1,2,3,4,5,6} 4726 4727 4728 .seealso: MatCreate(), MatCreateAIJ(), MatCreateSeqAIJ(), MatCreateSeqAIJWithArrays(), MatMPIAIJSetPreallocationCSR() 4729 4730 @*/ 4731 PetscErrorCode MatCreateSeqAIJFromTriple(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt *i,PetscInt *j,PetscScalar *a,Mat *mat,PetscInt nz,PetscBool idx) 4732 { 4733 PetscErrorCode ierr; 4734 PetscInt ii, *nnz, one = 1,row,col; 4735 4736 4737 PetscFunctionBegin; 4738 ierr = PetscMalloc(m*sizeof(PetscInt),&nnz);CHKERRQ(ierr); 4739 ierr = PetscMemzero(nnz,m*sizeof(PetscInt));CHKERRQ(ierr); 4740 for (ii = 0; ii < nz; ii++) { 4741 nnz[i[ii]] += 1; 4742 } 4743 ierr = MatCreate(comm,mat);CHKERRQ(ierr); 4744 ierr = MatSetSizes(*mat,m,n,m,n);CHKERRQ(ierr); 4745 /* ierr = MatSetBlockSizes(*mat,,);CHKERRQ(ierr); */ 4746 ierr = MatSetType(*mat,MATSEQAIJ);CHKERRQ(ierr); 4747 ierr = MatSeqAIJSetPreallocation_SeqAIJ(*mat,0,nnz);CHKERRQ(ierr); 4748 for (ii = 0; ii < nz; ii++) { 4749 if (idx) { 4750 row = i[ii] - 1; 4751 col = j[ii] - 1; 4752 } else { 4753 row = i[ii]; 4754 col = j[ii]; 4755 } 4756 ierr = MatSetValues(*mat,one,&row,one,&col,&a[ii],ADD_VALUES);CHKERRQ(ierr); 4757 } 4758 ierr = MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4759 ierr = MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 4760 ierr = PetscFree(nnz);CHKERRQ(ierr); 4761 PetscFunctionReturn(0); 4762 } 4763 4764 #undef __FUNCT__ 4765 #define __FUNCT__ "MatSetColoring_SeqAIJ" 4766 PetscErrorCode MatSetColoring_SeqAIJ(Mat A,ISColoring coloring) 4767 { 4768 PetscErrorCode ierr; 4769 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 4770 4771 PetscFunctionBegin; 4772 if (coloring->ctype == IS_COLORING_GLOBAL) { 4773 ierr = ISColoringReference(coloring);CHKERRQ(ierr); 4774 a->coloring = coloring; 4775 } else if (coloring->ctype == IS_COLORING_GHOSTED) { 4776 PetscInt i,*larray; 4777 ISColoring ocoloring; 4778 ISColoringValue *colors; 4779 4780 /* set coloring for diagonal portion */ 4781 ierr = PetscMalloc(A->cmap->n*sizeof(PetscInt),&larray);CHKERRQ(ierr); 4782 for (i=0; i<A->cmap->n; i++) larray[i] = i; 4783 ierr = ISGlobalToLocalMappingApply(A->cmap->mapping,IS_GTOLM_MASK,A->cmap->n,larray,NULL,larray);CHKERRQ(ierr); 4784 ierr = PetscMalloc(A->cmap->n*sizeof(ISColoringValue),&colors);CHKERRQ(ierr); 4785 for (i=0; i<A->cmap->n; i++) colors[i] = coloring->colors[larray[i]]; 4786 ierr = PetscFree(larray);CHKERRQ(ierr); 4787 ierr = ISColoringCreate(PETSC_COMM_SELF,coloring->n,A->cmap->n,colors,&ocoloring);CHKERRQ(ierr); 4788 a->coloring = ocoloring; 4789 } 4790 PetscFunctionReturn(0); 4791 } 4792 4793 #undef __FUNCT__ 4794 #define __FUNCT__ "MatSetValuesAdifor_SeqAIJ" 4795 PetscErrorCode MatSetValuesAdifor_SeqAIJ(Mat A,PetscInt nl,void *advalues) 4796 { 4797 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 4798 PetscInt m = A->rmap->n,*ii = a->i,*jj = a->j,nz,i,j; 4799 MatScalar *v = a->a; 4800 PetscScalar *values = (PetscScalar*)advalues; 4801 ISColoringValue *color; 4802 4803 PetscFunctionBegin; 4804 if (!a->coloring) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_WRONGSTATE,"Coloring not set for matrix"); 4805 color = a->coloring->colors; 4806 /* loop over rows */ 4807 for (i=0; i<m; i++) { 4808 nz = ii[i+1] - ii[i]; 4809 /* loop over columns putting computed value into matrix */ 4810 for (j=0; j<nz; j++) *v++ = values[color[*jj++]]; 4811 values += nl; /* jump to next row of derivatives */ 4812 } 4813 PetscFunctionReturn(0); 4814 } 4815 4816 #undef __FUNCT__ 4817 #define __FUNCT__ "MatSeqAIJInvalidateDiagonal" 4818 PetscErrorCode MatSeqAIJInvalidateDiagonal(Mat A) 4819 { 4820 Mat_SeqAIJ *a=(Mat_SeqAIJ*)A->data; 4821 PetscErrorCode ierr; 4822 4823 PetscFunctionBegin; 4824 a->idiagvalid = PETSC_FALSE; 4825 a->ibdiagvalid = PETSC_FALSE; 4826 4827 ierr = MatSeqAIJInvalidateDiagonal_Inode(A);CHKERRQ(ierr); 4828 PetscFunctionReturn(0); 4829 } 4830 4831 /* 4832 Special version for direct calls from Fortran 4833 */ 4834 #include <petsc-private/fortranimpl.h> 4835 #if defined(PETSC_HAVE_FORTRAN_CAPS) 4836 #define matsetvaluesseqaij_ MATSETVALUESSEQAIJ 4837 #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE) 4838 #define matsetvaluesseqaij_ matsetvaluesseqaij 4839 #endif 4840 4841 /* Change these macros so can be used in void function */ 4842 #undef CHKERRQ 4843 #define CHKERRQ(ierr) CHKERRABORT(PetscObjectComm((PetscObject)A),ierr) 4844 #undef SETERRQ2 4845 #define SETERRQ2(comm,ierr,b,c,d) CHKERRABORT(comm,ierr) 4846 #undef SETERRQ3 4847 #define SETERRQ3(comm,ierr,b,c,d,e) CHKERRABORT(comm,ierr) 4848 4849 #undef __FUNCT__ 4850 #define __FUNCT__ "matsetvaluesseqaij_" 4851 PETSC_EXTERN void PETSC_STDCALL matsetvaluesseqaij_(Mat *AA,PetscInt *mm,const PetscInt im[],PetscInt *nn,const PetscInt in[],const PetscScalar v[],InsertMode *isis, PetscErrorCode *_ierr) 4852 { 4853 Mat A = *AA; 4854 PetscInt m = *mm, n = *nn; 4855 InsertMode is = *isis; 4856 Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data; 4857 PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N; 4858 PetscInt *imax,*ai,*ailen; 4859 PetscErrorCode ierr; 4860 PetscInt *aj,nonew = a->nonew,lastcol = -1; 4861 MatScalar *ap,value,*aa; 4862 PetscBool ignorezeroentries = a->ignorezeroentries; 4863 PetscBool roworiented = a->roworiented; 4864 4865 PetscFunctionBegin; 4866 MatCheckPreallocated(A,1); 4867 imax = a->imax; 4868 ai = a->i; 4869 ailen = a->ilen; 4870 aj = a->j; 4871 aa = a->a; 4872 4873 for (k=0; k<m; k++) { /* loop over added rows */ 4874 row = im[k]; 4875 if (row < 0) continue; 4876 #if defined(PETSC_USE_DEBUG) 4877 if (row >= A->rmap->n) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Row too large"); 4878 #endif 4879 rp = aj + ai[row]; ap = aa + ai[row]; 4880 rmax = imax[row]; nrow = ailen[row]; 4881 low = 0; 4882 high = nrow; 4883 for (l=0; l<n; l++) { /* loop over added columns */ 4884 if (in[l] < 0) continue; 4885 #if defined(PETSC_USE_DEBUG) 4886 if (in[l] >= A->cmap->n) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Column too large"); 4887 #endif 4888 col = in[l]; 4889 if (roworiented) value = v[l + k*n]; 4890 else value = v[k + l*m]; 4891 4892 if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue; 4893 4894 if (col <= lastcol) low = 0; 4895 else high = nrow; 4896 lastcol = col; 4897 while (high-low > 5) { 4898 t = (low+high)/2; 4899 if (rp[t] > col) high = t; 4900 else low = t; 4901 } 4902 for (i=low; i<high; i++) { 4903 if (rp[i] > col) break; 4904 if (rp[i] == col) { 4905 if (is == ADD_VALUES) ap[i] += value; 4906 else ap[i] = value; 4907 goto noinsert; 4908 } 4909 } 4910 if (value == 0.0 && ignorezeroentries) goto noinsert; 4911 if (nonew == 1) goto noinsert; 4912 if (nonew == -1) SETERRABORT(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero in the matrix"); 4913 MatSeqXAIJReallocateAIJ(A,A->rmap->n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew,MatScalar); 4914 N = nrow++ - 1; a->nz++; high++; 4915 /* shift up all the later entries in this row */ 4916 for (ii=N; ii>=i; ii--) { 4917 rp[ii+1] = rp[ii]; 4918 ap[ii+1] = ap[ii]; 4919 } 4920 rp[i] = col; 4921 ap[i] = value; 4922 noinsert:; 4923 low = i + 1; 4924 } 4925 ailen[row] = nrow; 4926 } 4927 A->same_nonzero = PETSC_FALSE; 4928 PetscFunctionReturnVoid(); 4929 } 4930 4931 4932