Actual source code: aij.c
1: #define PETSCMAT_DLL
3: /*
4: Defines the basic matrix operations for the AIJ (compressed row)
5: matrix storage format.
6: */
8: #include src/mat/impls/aij/seq/aij.h
9: #include src/inline/spops.h
10: #include src/inline/dot.h
11: #include petscbt.h
15: PetscErrorCode MatDiagonalSet_SeqAIJ(Mat Y,Vec D,InsertMode is)
16: {
18: Mat_SeqAIJ *aij = (Mat_SeqAIJ*) Y->data;
19: PetscInt i,*diag, m = Y->rmap.n;
20: PetscScalar *v,*aa = aij->a;
21: PetscTruth missing;
24: if (Y->assembled) {
25: MatMissingDiagonal_SeqAIJ(Y,&missing,PETSC_NULL);
26: if (!missing) {
27: diag = aij->diag;
28: VecGetArray(D,&v);
29: if (is == INSERT_VALUES) {
30: for (i=0; i<m; i++) {
31: aa[diag[i]] = v[i];
32: }
33: } else {
34: for (i=0; i<m; i++) {
35: aa[diag[i]] += v[i];
36: }
37: }
38: VecRestoreArray(D,&v);
39: return(0);
40: }
41: }
42: MatDiagonalSet_Default(Y,D,is);
43: return(0);
44: }
48: PetscErrorCode MatGetRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscTruth symmetric,PetscInt *m,PetscInt *ia[],PetscInt *ja[],PetscTruth *done)
49: {
50: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
52: PetscInt i,ishift;
53:
55: *m = A->rmap.n;
56: if (!ia) return(0);
57: ishift = 0;
58: if (symmetric && !A->structurally_symmetric) {
59: MatToSymmetricIJ_SeqAIJ(A->rmap.n,a->i,a->j,ishift,oshift,ia,ja);
60: } else if (oshift == 1) {
61: PetscInt nz = a->i[A->rmap.n];
62: /* malloc space and add 1 to i and j indices */
63: PetscMalloc((A->rmap.n+1)*sizeof(PetscInt),ia);
64: PetscMalloc((nz+1)*sizeof(PetscInt),ja);
65: for (i=0; i<nz; i++) (*ja)[i] = a->j[i] + 1;
66: for (i=0; i<A->rmap.n+1; i++) (*ia)[i] = a->i[i] + 1;
67: } else {
68: *ia = a->i; *ja = a->j;
69: }
70: return(0);
71: }
75: PetscErrorCode MatRestoreRowIJ_SeqAIJ(Mat A,PetscInt oshift,PetscTruth symmetric,PetscInt *n,PetscInt *ia[],PetscInt *ja[],PetscTruth *done)
76: {
78:
80: if (!ia) return(0);
81: if ((symmetric && !A->structurally_symmetric) || oshift == 1) {
82: PetscFree(*ia);
83: PetscFree(*ja);
84: }
85: return(0);
86: }
90: PetscErrorCode MatGetColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscTruth symmetric,PetscInt *nn,PetscInt *ia[],PetscInt *ja[],PetscTruth *done)
91: {
92: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
94: PetscInt i,*collengths,*cia,*cja,n = A->cmap.n,m = A->rmap.n;
95: PetscInt nz = a->i[m],row,*jj,mr,col;
96:
98: *nn = n;
99: if (!ia) return(0);
100: if (symmetric) {
101: MatToSymmetricIJ_SeqAIJ(A->rmap.n,a->i,a->j,0,oshift,ia,ja);
102: } else {
103: PetscMalloc((n+1)*sizeof(PetscInt),&collengths);
104: PetscMemzero(collengths,n*sizeof(PetscInt));
105: PetscMalloc((n+1)*sizeof(PetscInt),&cia);
106: PetscMalloc((nz+1)*sizeof(PetscInt),&cja);
107: jj = a->j;
108: for (i=0; i<nz; i++) {
109: collengths[jj[i]]++;
110: }
111: cia[0] = oshift;
112: for (i=0; i<n; i++) {
113: cia[i+1] = cia[i] + collengths[i];
114: }
115: PetscMemzero(collengths,n*sizeof(PetscInt));
116: jj = a->j;
117: for (row=0; row<m; row++) {
118: mr = a->i[row+1] - a->i[row];
119: for (i=0; i<mr; i++) {
120: col = *jj++;
121: cja[cia[col] + collengths[col]++ - oshift] = row + oshift;
122: }
123: }
124: PetscFree(collengths);
125: *ia = cia; *ja = cja;
126: }
127: return(0);
128: }
132: PetscErrorCode MatRestoreColumnIJ_SeqAIJ(Mat A,PetscInt oshift,PetscTruth symmetric,PetscInt *n,PetscInt *ia[],PetscInt *ja[],PetscTruth *done)
133: {
137: if (!ia) return(0);
139: PetscFree(*ia);
140: PetscFree(*ja);
141:
142: return(0);
143: }
147: PetscErrorCode MatSetValuesRow_SeqAIJ(Mat A,PetscInt row,const PetscScalar v[])
148: {
149: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
150: PetscInt *ai = a->i;
154: PetscMemcpy(a->a+ai[row],v,(ai[row+1]-ai[row])*sizeof(PetscScalar));
155: return(0);
156: }
158: #define CHUNKSIZE 15
162: PetscErrorCode MatSetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode is)
163: {
164: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
165: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N;
166: PetscInt *imax = a->imax,*ai = a->i,*ailen = a->ilen;
168: PetscInt *aj = a->j,nonew = a->nonew,lastcol = -1;
169: PetscScalar *ap,value,*aa = a->a;
170: PetscTruth ignorezeroentries = ((a->ignorezeroentries && is == ADD_VALUES) ? PETSC_TRUE:PETSC_FALSE);
171: PetscTruth roworiented = a->roworiented;
174: for (k=0; k<m; k++) { /* loop over added rows */
175: row = im[k];
176: if (row < 0) continue;
177: #if defined(PETSC_USE_DEBUG)
178: if (row >= A->rmap.n) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap.n-1);
179: #endif
180: rp = aj + ai[row]; ap = aa + ai[row];
181: rmax = imax[row]; nrow = ailen[row];
182: low = 0;
183: high = nrow;
184: for (l=0; l<n; l++) { /* loop over added columns */
185: if (in[l] < 0) continue;
186: #if defined(PETSC_USE_DEBUG)
187: if (in[l] >= A->cmap.n) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap.n-1);
188: #endif
189: col = in[l];
190: if (roworiented) {
191: value = v[l + k*n];
192: } else {
193: value = v[k + l*m];
194: }
195: if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue;
197: if (col <= lastcol) low = 0; else high = nrow;
198: lastcol = col;
199: while (high-low > 5) {
200: t = (low+high)/2;
201: if (rp[t] > col) high = t;
202: else low = t;
203: }
204: for (i=low; i<high; i++) {
205: if (rp[i] > col) break;
206: if (rp[i] == col) {
207: if (is == ADD_VALUES) ap[i] += value;
208: else ap[i] = value;
209: goto noinsert;
210: }
211: }
212: if (value == 0.0 && ignorezeroentries) goto noinsert;
213: if (nonew == 1) goto noinsert;
214: if (nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero at (%D,%D) in the matrix",row,col);
215: MatSeqXAIJReallocateAIJ(A,A->rmap.n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew);
216: N = nrow++ - 1; a->nz++; high++;
217: /* shift up all the later entries in this row */
218: for (ii=N; ii>=i; ii--) {
219: rp[ii+1] = rp[ii];
220: ap[ii+1] = ap[ii];
221: }
222: rp[i] = col;
223: ap[i] = value;
224: noinsert:;
225: low = i + 1;
226: }
227: ailen[row] = nrow;
228: }
229: A->same_nonzero = PETSC_FALSE;
230: return(0);
231: }
236: PetscErrorCode MatGetValues_SeqAIJ(Mat A,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],PetscScalar v[])
237: {
238: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
239: PetscInt *rp,k,low,high,t,row,nrow,i,col,l,*aj = a->j;
240: PetscInt *ai = a->i,*ailen = a->ilen;
241: PetscScalar *ap,*aa = a->a,zero = 0.0;
244: for (k=0; k<m; k++) { /* loop over rows */
245: row = im[k];
246: if (row < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",row);
247: if (row >= A->rmap.n) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",row,A->rmap.n-1);
248: rp = aj + ai[row]; ap = aa + ai[row];
249: nrow = ailen[row];
250: for (l=0; l<n; l++) { /* loop over columns */
251: if (in[l] < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative column: %D",in[l]);
252: if (in[l] >= A->cmap.n) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[l],A->cmap.n-1);
253: col = in[l] ;
254: high = nrow; low = 0; /* assume unsorted */
255: while (high-low > 5) {
256: t = (low+high)/2;
257: if (rp[t] > col) high = t;
258: else low = t;
259: }
260: for (i=low; i<high; i++) {
261: if (rp[i] > col) break;
262: if (rp[i] == col) {
263: *v++ = ap[i];
264: goto finished;
265: }
266: }
267: *v++ = zero;
268: finished:;
269: }
270: }
271: return(0);
272: }
277: PetscErrorCode MatView_SeqAIJ_Binary(Mat A,PetscViewer viewer)
278: {
279: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
281: PetscInt i,*col_lens;
282: int fd;
285: PetscViewerBinaryGetDescriptor(viewer,&fd);
286: PetscMalloc((4+A->rmap.n)*sizeof(PetscInt),&col_lens);
287: col_lens[0] = MAT_FILE_COOKIE;
288: col_lens[1] = A->rmap.n;
289: col_lens[2] = A->cmap.n;
290: col_lens[3] = a->nz;
292: /* store lengths of each row and write (including header) to file */
293: for (i=0; i<A->rmap.n; i++) {
294: col_lens[4+i] = a->i[i+1] - a->i[i];
295: }
296: PetscBinaryWrite(fd,col_lens,4+A->rmap.n,PETSC_INT,PETSC_TRUE);
297: PetscFree(col_lens);
299: /* store column indices (zero start index) */
300: PetscBinaryWrite(fd,a->j,a->nz,PETSC_INT,PETSC_FALSE);
302: /* store nonzero values */
303: PetscBinaryWrite(fd,a->a,a->nz,PETSC_SCALAR,PETSC_FALSE);
304: return(0);
305: }
307: EXTERN PetscErrorCode MatSeqAIJFactorInfo_Matlab(Mat,PetscViewer);
311: PetscErrorCode MatView_SeqAIJ_ASCII(Mat A,PetscViewer viewer)
312: {
313: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
314: PetscErrorCode ierr;
315: PetscInt i,j,m = A->rmap.n,shift=0;
316: const char *name;
317: PetscViewerFormat format;
320: PetscObjectGetName((PetscObject)A,&name);
321: PetscViewerGetFormat(viewer,&format);
322: if (format == PETSC_VIEWER_ASCII_MATLAB) {
323: PetscInt nofinalvalue = 0;
324: if ((a->i[m] == a->i[m-1]) || (a->j[a->nz-1] != A->cmap.n-!shift)) {
325: nofinalvalue = 1;
326: }
327: PetscViewerASCIIUseTabs(viewer,PETSC_NO);
328: PetscViewerASCIIPrintf(viewer,"%% Size = %D %D \n",m,A->cmap.n);
329: PetscViewerASCIIPrintf(viewer,"%% Nonzeros = %D \n",a->nz);
330: PetscViewerASCIIPrintf(viewer,"zzz = zeros(%D,3);\n",a->nz+nofinalvalue);
331: PetscViewerASCIIPrintf(viewer,"zzz = [\n");
333: for (i=0; i<m; i++) {
334: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
335: #if defined(PETSC_USE_COMPLEX)
336: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e + %18.16ei \n",i+1,a->j[j]+!shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));
337: #else
338: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",i+1,a->j[j]+!shift,a->a[j]);
339: #endif
340: }
341: }
342: if (nofinalvalue) {
343: PetscViewerASCIIPrintf(viewer,"%D %D %18.16e\n",m,A->cmap.n,0.0);
344: }
345: PetscViewerASCIIPrintf(viewer,"];\n %s = spconvert(zzz);\n",name);
346: PetscViewerASCIIUseTabs(viewer,PETSC_YES);
347: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO || format == PETSC_VIEWER_ASCII_INFO) {
348: return(0);
349: } else if (format == PETSC_VIEWER_ASCII_COMMON) {
350: PetscViewerASCIIUseTabs(viewer,PETSC_NO);
351: for (i=0; i<m; i++) {
352: PetscViewerASCIIPrintf(viewer,"row %D:",i);
353: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
354: #if defined(PETSC_USE_COMPLEX)
355: if (PetscImaginaryPart(a->a[j]) > 0.0 && PetscRealPart(a->a[j]) != 0.0) {
356: PetscViewerASCIIPrintf(viewer," (%D, %G + %G i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));
357: } else if (PetscImaginaryPart(a->a[j]) < 0.0 && PetscRealPart(a->a[j]) != 0.0) {
358: PetscViewerASCIIPrintf(viewer," (%D, %G - %G i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));
359: } else if (PetscRealPart(a->a[j]) != 0.0) {
360: PetscViewerASCIIPrintf(viewer," (%D, %G) ",a->j[j]+shift,PetscRealPart(a->a[j]));
361: }
362: #else
363: if (a->a[j] != 0.0) {PetscViewerASCIIPrintf(viewer," (%D, %G) ",a->j[j]+shift,a->a[j]);}
364: #endif
365: }
366: PetscViewerASCIIPrintf(viewer,"\n");
367: }
368: PetscViewerASCIIUseTabs(viewer,PETSC_YES);
369: } else if (format == PETSC_VIEWER_ASCII_SYMMODU) {
370: PetscInt nzd=0,fshift=1,*sptr;
371: PetscViewerASCIIUseTabs(viewer,PETSC_NO);
372: PetscMalloc((m+1)*sizeof(PetscInt),&sptr);
373: for (i=0; i<m; i++) {
374: sptr[i] = nzd+1;
375: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
376: if (a->j[j] >= i) {
377: #if defined(PETSC_USE_COMPLEX)
378: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) nzd++;
379: #else
380: if (a->a[j] != 0.0) nzd++;
381: #endif
382: }
383: }
384: }
385: sptr[m] = nzd+1;
386: PetscViewerASCIIPrintf(viewer," %D %D\n\n",m,nzd);
387: for (i=0; i<m+1; i+=6) {
388: if (i+4<m) {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]);}
389: else if (i+3<m) {PetscViewerASCIIPrintf(viewer," %D %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3],sptr[i+4]);}
390: else if (i+2<m) {PetscViewerASCIIPrintf(viewer," %D %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2],sptr[i+3]);}
391: else if (i+1<m) {PetscViewerASCIIPrintf(viewer," %D %D %D\n",sptr[i],sptr[i+1],sptr[i+2]);}
392: else if (i<m) {PetscViewerASCIIPrintf(viewer," %D %D\n",sptr[i],sptr[i+1]);}
393: else {PetscViewerASCIIPrintf(viewer," %D\n",sptr[i]);}
394: }
395: PetscViewerASCIIPrintf(viewer,"\n");
396: PetscFree(sptr);
397: for (i=0; i<m; i++) {
398: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
399: if (a->j[j] >= i) {PetscViewerASCIIPrintf(viewer," %D ",a->j[j]+fshift);}
400: }
401: PetscViewerASCIIPrintf(viewer,"\n");
402: }
403: PetscViewerASCIIPrintf(viewer,"\n");
404: for (i=0; i<m; i++) {
405: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
406: if (a->j[j] >= i) {
407: #if defined(PETSC_USE_COMPLEX)
408: if (PetscImaginaryPart(a->a[j]) != 0.0 || PetscRealPart(a->a[j]) != 0.0) {
409: PetscViewerASCIIPrintf(viewer," %18.16e %18.16e ",PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));
410: }
411: #else
412: if (a->a[j] != 0.0) {PetscViewerASCIIPrintf(viewer," %18.16e ",a->a[j]);}
413: #endif
414: }
415: }
416: PetscViewerASCIIPrintf(viewer,"\n");
417: }
418: PetscViewerASCIIUseTabs(viewer,PETSC_YES);
419: } else if (format == PETSC_VIEWER_ASCII_DENSE) {
420: PetscInt cnt = 0,jcnt;
421: PetscScalar value;
423: PetscViewerASCIIUseTabs(viewer,PETSC_NO);
424: for (i=0; i<m; i++) {
425: jcnt = 0;
426: for (j=0; j<A->cmap.n; j++) {
427: if (jcnt < a->i[i+1]-a->i[i] && j == a->j[cnt]) {
428: value = a->a[cnt++];
429: jcnt++;
430: } else {
431: value = 0.0;
432: }
433: #if defined(PETSC_USE_COMPLEX)
434: PetscViewerASCIIPrintf(viewer," %7.5e+%7.5e i ",PetscRealPart(value),PetscImaginaryPart(value));
435: #else
436: PetscViewerASCIIPrintf(viewer," %7.5e ",value);
437: #endif
438: }
439: PetscViewerASCIIPrintf(viewer,"\n");
440: }
441: PetscViewerASCIIUseTabs(viewer,PETSC_YES);
442: } else {
443: PetscViewerASCIIUseTabs(viewer,PETSC_NO);
444: for (i=0; i<m; i++) {
445: PetscViewerASCIIPrintf(viewer,"row %D:",i);
446: for (j=a->i[i]+shift; j<a->i[i+1]+shift; j++) {
447: #if defined(PETSC_USE_COMPLEX)
448: if (PetscImaginaryPart(a->a[j]) > 0.0) {
449: PetscViewerASCIIPrintf(viewer," (%D, %G + %G i)",a->j[j]+shift,PetscRealPart(a->a[j]),PetscImaginaryPart(a->a[j]));
450: } else if (PetscImaginaryPart(a->a[j]) < 0.0) {
451: PetscViewerASCIIPrintf(viewer," (%D, %G - %G i)",a->j[j]+shift,PetscRealPart(a->a[j]),-PetscImaginaryPart(a->a[j]));
452: } else {
453: PetscViewerASCIIPrintf(viewer," (%D, %G) ",a->j[j]+shift,PetscRealPart(a->a[j]));
454: }
455: #else
456: PetscViewerASCIIPrintf(viewer," (%D, %G) ",a->j[j]+shift,a->a[j]);
457: #endif
458: }
459: PetscViewerASCIIPrintf(viewer,"\n");
460: }
461: PetscViewerASCIIUseTabs(viewer,PETSC_YES);
462: }
463: PetscViewerFlush(viewer);
464: return(0);
465: }
469: PetscErrorCode MatView_SeqAIJ_Draw_Zoom(PetscDraw draw,void *Aa)
470: {
471: Mat A = (Mat) Aa;
472: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
473: PetscErrorCode ierr;
474: PetscInt i,j,m = A->rmap.n,color;
475: PetscReal xl,yl,xr,yr,x_l,x_r,y_l,y_r,maxv = 0.0;
476: PetscViewer viewer;
477: PetscViewerFormat format;
480: PetscObjectQuery((PetscObject)A,"Zoomviewer",(PetscObject*)&viewer);
481: PetscViewerGetFormat(viewer,&format);
483: PetscDrawGetCoordinates(draw,&xl,&yl,&xr,&yr);
484: /* loop over matrix elements drawing boxes */
486: if (format != PETSC_VIEWER_DRAW_CONTOUR) {
487: /* Blue for negative, Cyan for zero and Red for positive */
488: color = PETSC_DRAW_BLUE;
489: for (i=0; i<m; i++) {
490: y_l = m - i - 1.0; y_r = y_l + 1.0;
491: for (j=a->i[i]; j<a->i[i+1]; j++) {
492: x_l = a->j[j] ; x_r = x_l + 1.0;
493: #if defined(PETSC_USE_COMPLEX)
494: if (PetscRealPart(a->a[j]) >= 0.) continue;
495: #else
496: if (a->a[j] >= 0.) continue;
497: #endif
498: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
499: }
500: }
501: color = PETSC_DRAW_CYAN;
502: for (i=0; i<m; i++) {
503: y_l = m - i - 1.0; y_r = y_l + 1.0;
504: for (j=a->i[i]; j<a->i[i+1]; j++) {
505: x_l = a->j[j]; x_r = x_l + 1.0;
506: if (a->a[j] != 0.) continue;
507: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
508: }
509: }
510: color = PETSC_DRAW_RED;
511: for (i=0; i<m; i++) {
512: y_l = m - i - 1.0; y_r = y_l + 1.0;
513: for (j=a->i[i]; j<a->i[i+1]; j++) {
514: x_l = a->j[j]; x_r = x_l + 1.0;
515: #if defined(PETSC_USE_COMPLEX)
516: if (PetscRealPart(a->a[j]) <= 0.) continue;
517: #else
518: if (a->a[j] <= 0.) continue;
519: #endif
520: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
521: }
522: }
523: } else {
524: /* use contour shading to indicate magnitude of values */
525: /* first determine max of all nonzero values */
526: PetscInt nz = a->nz,count;
527: PetscDraw popup;
528: PetscReal scale;
530: for (i=0; i<nz; i++) {
531: if (PetscAbsScalar(a->a[i]) > maxv) maxv = PetscAbsScalar(a->a[i]);
532: }
533: scale = (245.0 - PETSC_DRAW_BASIC_COLORS)/maxv;
534: PetscDrawGetPopup(draw,&popup);
535: if (popup) {PetscDrawScalePopup(popup,0.0,maxv);}
536: count = 0;
537: for (i=0; i<m; i++) {
538: y_l = m - i - 1.0; y_r = y_l + 1.0;
539: for (j=a->i[i]; j<a->i[i+1]; j++) {
540: x_l = a->j[j]; x_r = x_l + 1.0;
541: color = PETSC_DRAW_BASIC_COLORS + (PetscInt)(scale*PetscAbsScalar(a->a[count]));
542: PetscDrawRectangle(draw,x_l,y_l,x_r,y_r,color,color,color,color);
543: count++;
544: }
545: }
546: }
547: return(0);
548: }
552: PetscErrorCode MatView_SeqAIJ_Draw(Mat A,PetscViewer viewer)
553: {
555: PetscDraw draw;
556: PetscReal xr,yr,xl,yl,h,w;
557: PetscTruth isnull;
560: PetscViewerDrawGetDraw(viewer,0,&draw);
561: PetscDrawIsNull(draw,&isnull);
562: if (isnull) return(0);
564: PetscObjectCompose((PetscObject)A,"Zoomviewer",(PetscObject)viewer);
565: xr = A->cmap.n; yr = A->rmap.n; h = yr/10.0; w = xr/10.0;
566: xr += w; yr += h; xl = -w; yl = -h;
567: PetscDrawSetCoordinates(draw,xl,yl,xr,yr);
568: PetscDrawZoom(draw,MatView_SeqAIJ_Draw_Zoom,A);
569: PetscObjectCompose((PetscObject)A,"Zoomviewer",PETSC_NULL);
570: return(0);
571: }
575: PetscErrorCode MatView_SeqAIJ(Mat A,PetscViewer viewer)
576: {
578: PetscTruth issocket,iascii,isbinary,isdraw;
581: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_SOCKET,&issocket);
582: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);
583: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_BINARY,&isbinary);
584: PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);
585: if (iascii) {
586: MatView_SeqAIJ_ASCII(A,viewer);
587: #if defined(PETSC_USE_SOCKET_VIEWER)
588: } else if (issocket) {
589: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
590: PetscViewerSocketPutSparse_Private(viewer,A->rmap.n,A->cmap.n,a->nz,a->a,a->i,a->j);
591: #endif
592: } else if (isbinary) {
593: MatView_SeqAIJ_Binary(A,viewer);
594: } else if (isdraw) {
595: MatView_SeqAIJ_Draw(A,viewer);
596: } else {
597: SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported by SeqAIJ matrices",((PetscObject)viewer)->type_name);
598: }
599: MatView_Inode(A,viewer);
600: return(0);
601: }
605: PetscErrorCode MatAssemblyEnd_SeqAIJ(Mat A,MatAssemblyType mode)
606: {
607: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
609: PetscInt fshift = 0,i,j,*ai = a->i,*aj = a->j,*imax = a->imax;
610: PetscInt m = A->rmap.n,*ip,N,*ailen = a->ilen,rmax = 0;
611: PetscScalar *aa = a->a,*ap;
612: PetscReal ratio=0.6;
615: if (mode == MAT_FLUSH_ASSEMBLY) return(0);
617: if (m) rmax = ailen[0]; /* determine row with most nonzeros */
618: for (i=1; i<m; i++) {
619: /* move each row back by the amount of empty slots (fshift) before it*/
620: fshift += imax[i-1] - ailen[i-1];
621: rmax = PetscMax(rmax,ailen[i]);
622: if (fshift) {
623: ip = aj + ai[i] ;
624: ap = aa + ai[i] ;
625: N = ailen[i];
626: for (j=0; j<N; j++) {
627: ip[j-fshift] = ip[j];
628: ap[j-fshift] = ap[j];
629: }
630: }
631: ai[i] = ai[i-1] + ailen[i-1];
632: }
633: if (m) {
634: fshift += imax[m-1] - ailen[m-1];
635: ai[m] = ai[m-1] + ailen[m-1];
636: }
637: /* reset ilen and imax for each row */
638: for (i=0; i<m; i++) {
639: ailen[i] = imax[i] = ai[i+1] - ai[i];
640: }
641: a->nz = ai[m];
643: MatMarkDiagonal_SeqAIJ(A);
644: PetscInfo4(A,"Matrix size: %D X %D; storage space: %D unneeded,%D used\n",m,A->cmap.n,fshift,a->nz);
645: PetscInfo1(A,"Number of mallocs during MatSetValues() is %D\n",a->reallocs);
646: PetscInfo1(A,"Maximum nonzeros in any row is %D\n",rmax);
647: a->reallocs = 0;
648: A->info.nz_unneeded = (double)fshift;
649: a->rmax = rmax;
651: /* check for zero rows. If found a large number of zero rows, use CompressedRow functions */
652: Mat_CheckCompressedRow(A,&a->compressedrow,a->i,m,ratio);
653: A->same_nonzero = PETSC_TRUE;
655: MatAssemblyEnd_Inode(A,mode);
656: return(0);
657: }
661: PetscErrorCode MatRealPart_SeqAIJ(Mat A)
662: {
663: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
664: PetscInt i,nz = a->nz;
665: PetscScalar *aa = a->a;
668: for (i=0; i<nz; i++) aa[i] = PetscRealPart(aa[i]);
669: return(0);
670: }
674: PetscErrorCode MatImaginaryPart_SeqAIJ(Mat A)
675: {
676: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
677: PetscInt i,nz = a->nz;
678: PetscScalar *aa = a->a;
681: for (i=0; i<nz; i++) aa[i] = PetscImaginaryPart(aa[i]);
682: return(0);
683: }
687: PetscErrorCode MatZeroEntries_SeqAIJ(Mat A)
688: {
689: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
693: PetscMemzero(a->a,(a->i[A->rmap.n])*sizeof(PetscScalar));
694: return(0);
695: }
699: PetscErrorCode MatDestroy_SeqAIJ(Mat A)
700: {
701: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
705: #if defined(PETSC_USE_LOG)
706: PetscLogObjectState((PetscObject)A,"Rows=%D, Cols=%D, NZ=%D",A->rmap.n,A->cmap.n,a->nz);
707: #endif
708: MatSeqXAIJFreeAIJ(A,&a->a,&a->j,&a->i);
709: if (a->row) {
710: ISDestroy(a->row);
711: }
712: if (a->col) {
713: ISDestroy(a->col);
714: }
715: PetscFree(a->diag);
716: PetscFree2(a->imax,a->ilen);
717: PetscFree(a->idiag);
718: PetscFree(a->solve_work);
719: if (a->icol) {ISDestroy(a->icol);}
720: PetscFree(a->saved_values);
721: if (a->coloring) {ISColoringDestroy(a->coloring);}
722: PetscFree(a->xtoy);
723: if (a->compressedrow.use){PetscFree(a->compressedrow.i);}
725: MatDestroy_Inode(A);
727: PetscFree(a);
729: PetscObjectChangeTypeName((PetscObject)A,0);
730: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatSeqAIJSetColumnIndices_C","",PETSC_NULL);
731: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatStoreValues_C","",PETSC_NULL);
732: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatRetrieveValues_C","",PETSC_NULL);
733: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatConvert_seqaij_seqsbaij_C","",PETSC_NULL);
734: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatConvert_seqaij_seqbaij_C","",PETSC_NULL);
735: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatConvert_seqaij_seqcsrperm_C","",PETSC_NULL);
736: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatIsTranspose_C","",PETSC_NULL);
737: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatSeqAIJSetPreallocation_C","",PETSC_NULL);
738: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatSeqAIJSetPreallocationCSR_C","",PETSC_NULL);
739: PetscObjectComposeFunctionDynamic((PetscObject)A,"MatReorderForNonzeroDiagonal_C","",PETSC_NULL);
740: return(0);
741: }
745: PetscErrorCode MatCompress_SeqAIJ(Mat A)
746: {
748: return(0);
749: }
753: PetscErrorCode MatSetOption_SeqAIJ(Mat A,MatOption op)
754: {
755: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
759: switch (op) {
760: case MAT_ROW_ORIENTED:
761: a->roworiented = PETSC_TRUE;
762: break;
763: case MAT_KEEP_ZEROED_ROWS:
764: a->keepzeroedrows = PETSC_TRUE;
765: break;
766: case MAT_COLUMN_ORIENTED:
767: a->roworiented = PETSC_FALSE;
768: break;
769: case MAT_COLUMNS_SORTED:
770: a->sorted = PETSC_TRUE;
771: break;
772: case MAT_COLUMNS_UNSORTED:
773: a->sorted = PETSC_FALSE;
774: break;
775: case MAT_NO_NEW_NONZERO_LOCATIONS:
776: a->nonew = 1;
777: break;
778: case MAT_NEW_NONZERO_LOCATION_ERR:
779: a->nonew = -1;
780: break;
781: case MAT_NEW_NONZERO_ALLOCATION_ERR:
782: a->nonew = -2;
783: break;
784: case MAT_YES_NEW_NONZERO_LOCATIONS:
785: a->nonew = 0;
786: break;
787: case MAT_IGNORE_ZERO_ENTRIES:
788: a->ignorezeroentries = PETSC_TRUE;
789: break;
790: case MAT_USE_COMPRESSEDROW:
791: a->compressedrow.use = PETSC_TRUE;
792: break;
793: case MAT_DO_NOT_USE_COMPRESSEDROW:
794: a->compressedrow.use = PETSC_FALSE;
795: break;
796: case MAT_ROWS_SORTED:
797: case MAT_ROWS_UNSORTED:
798: case MAT_YES_NEW_DIAGONALS:
799: case MAT_IGNORE_OFF_PROC_ENTRIES:
800: case MAT_USE_HASH_TABLE:
801: PetscInfo(A,"Option ignored\n");
802: break;
803: case MAT_NO_NEW_DIAGONALS:
804: SETERRQ(PETSC_ERR_SUP,"MAT_NO_NEW_DIAGONALS");
805: default:
806: break;
807: }
808: MatSetOption_Inode(A,op);
809: return(0);
810: }
814: PetscErrorCode MatGetDiagonal_SeqAIJ(Mat A,Vec v)
815: {
816: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
818: PetscInt i,j,n;
819: PetscScalar *x,zero = 0.0;
822: VecSet(v,zero);
823: VecGetArray(v,&x);
824: VecGetLocalSize(v,&n);
825: if (n != A->rmap.n) SETERRQ(PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
826: for (i=0; i<A->rmap.n; i++) {
827: for (j=a->i[i]; j<a->i[i+1]; j++) {
828: if (a->j[j] == i) {
829: x[i] = a->a[j];
830: break;
831: }
832: }
833: }
834: VecRestoreArray(v,&x);
835: return(0);
836: }
840: PetscErrorCode MatMultTransposeAdd_SeqAIJ(Mat A,Vec xx,Vec zz,Vec yy)
841: {
842: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
843: PetscScalar *x,*y;
844: PetscErrorCode ierr;
845: PetscInt m = A->rmap.n;
846: #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
847: PetscScalar *v,alpha;
848: PetscInt n,i,*idx,*ii,*ridx=PETSC_NULL;
849: Mat_CompressedRow cprow = a->compressedrow;
850: PetscTruth usecprow = cprow.use;
851: #endif
854: if (zz != yy) {VecCopy(zz,yy);}
855: VecGetArray(xx,&x);
856: VecGetArray(yy,&y);
858: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTTRANSPOSEAIJ)
859: fortranmulttransposeaddaij_(&m,x,a->i,a->j,a->a,y);
860: #else
861: if (usecprow){
862: m = cprow.nrows;
863: ii = cprow.i;
864: ridx = cprow.rindex;
865: } else {
866: ii = a->i;
867: }
868: for (i=0; i<m; i++) {
869: idx = a->j + ii[i] ;
870: v = a->a + ii[i] ;
871: n = ii[i+1] - ii[i];
872: if (usecprow){
873: alpha = x[ridx[i]];
874: } else {
875: alpha = x[i];
876: }
877: while (n-->0) {y[*idx++] += alpha * *v++;}
878: }
879: #endif
880: PetscLogFlops(2*a->nz);
881: VecRestoreArray(xx,&x);
882: VecRestoreArray(yy,&y);
883: return(0);
884: }
888: PetscErrorCode MatMultTranspose_SeqAIJ(Mat A,Vec xx,Vec yy)
889: {
890: PetscScalar zero = 0.0;
894: VecSet(yy,zero);
895: MatMultTransposeAdd_SeqAIJ(A,xx,yy,yy);
896: return(0);
897: }
902: PetscErrorCode MatMult_SeqAIJ(Mat A,Vec xx,Vec yy)
903: {
904: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
905: PetscScalar *x,*y,*aa;
907: PetscInt m=A->rmap.n,*aj,*ii;
908: PetscInt n,i,j,*ridx=PETSC_NULL;
909: PetscScalar sum;
910: PetscTruth usecprow=a->compressedrow.use;
911: #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ)
912: PetscInt jrow;
913: #endif
915: #if defined(PETSC_HAVE_PRAGMA_DISJOINT)
916: #pragma disjoint(*x,*y,*aa)
917: #endif
920: VecGetArray(xx,&x);
921: VecGetArray(yy,&y);
922: aj = a->j;
923: aa = a->a;
924: ii = a->i;
925: if (usecprow){ /* use compressed row format */
926: m = a->compressedrow.nrows;
927: ii = a->compressedrow.i;
928: ridx = a->compressedrow.rindex;
929: for (i=0; i<m; i++){
930: n = ii[i+1] - ii[i];
931: aj = a->j + ii[i];
932: aa = a->a + ii[i];
933: sum = 0.0;
934: for (j=0; j<n; j++) sum += (*aa++)*x[*aj++];
935: y[*ridx++] = sum;
936: }
937: } else { /* do not use compressed row format */
938: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTAIJ)
939: fortranmultaij_(&m,x,ii,aj,aa,y);
940: #else
941: for (i=0; i<m; i++) {
942: jrow = ii[i];
943: n = ii[i+1] - jrow;
944: sum = 0.0;
945: for (j=0; j<n; j++) {
946: sum += aa[jrow]*x[aj[jrow]]; jrow++;
947: }
948: y[i] = sum;
949: }
950: #endif
951: }
952: PetscLogFlops(2*a->nz - m);
953: VecRestoreArray(xx,&x);
954: VecRestoreArray(yy,&y);
955: return(0);
956: }
960: PetscErrorCode MatMultAdd_SeqAIJ(Mat A,Vec xx,Vec yy,Vec zz)
961: {
962: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
963: PetscScalar *x,*y,*z,*aa;
965: PetscInt m = A->rmap.n,*aj,*ii;
966: #if !defined(PETSC_USE_FORTRAN_KERNEL_MULTADDAIJ)
967: PetscInt n,i,jrow,j,*ridx=PETSC_NULL;
968: PetscScalar sum;
969: PetscTruth usecprow=a->compressedrow.use;
970: #endif
973: VecGetArray(xx,&x);
974: VecGetArray(yy,&y);
975: if (zz != yy) {
976: VecGetArray(zz,&z);
977: } else {
978: z = y;
979: }
981: aj = a->j;
982: aa = a->a;
983: ii = a->i;
984: #if defined(PETSC_USE_FORTRAN_KERNEL_MULTADDAIJ)
985: fortranmultaddaij_(&m,x,ii,aj,aa,y,z);
986: #else
987: if (usecprow){ /* use compressed row format */
988: if (zz != yy){
989: PetscMemcpy(z,y,m*sizeof(PetscScalar));
990: }
991: m = a->compressedrow.nrows;
992: ii = a->compressedrow.i;
993: ridx = a->compressedrow.rindex;
994: for (i=0; i<m; i++){
995: n = ii[i+1] - ii[i];
996: aj = a->j + ii[i];
997: aa = a->a + ii[i];
998: sum = y[*ridx];
999: for (j=0; j<n; j++) sum += (*aa++)*x[*aj++];
1000: z[*ridx++] = sum;
1001: }
1002: } else { /* do not use compressed row format */
1003: for (i=0; i<m; i++) {
1004: jrow = ii[i];
1005: n = ii[i+1] - jrow;
1006: sum = y[i];
1007: for (j=0; j<n; j++) {
1008: sum += aa[jrow]*x[aj[jrow]]; jrow++;
1009: }
1010: z[i] = sum;
1011: }
1012: }
1013: #endif
1014: PetscLogFlops(2*a->nz);
1015: VecRestoreArray(xx,&x);
1016: VecRestoreArray(yy,&y);
1017: if (zz != yy) {
1018: VecRestoreArray(zz,&z);
1019: }
1020: return(0);
1021: }
1023: /*
1024: Adds diagonal pointers to sparse matrix structure.
1025: */
1028: PetscErrorCode MatMarkDiagonal_SeqAIJ(Mat A)
1029: {
1030: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1032: PetscInt i,j,m = A->rmap.n;
1035: if (!a->diag) {
1036: PetscMalloc(m*sizeof(PetscInt),&a->diag);
1037: }
1038: for (i=0; i<A->rmap.n; i++) {
1039: a->diag[i] = a->i[i+1];
1040: for (j=a->i[i]; j<a->i[i+1]; j++) {
1041: if (a->j[j] == i) {
1042: a->diag[i] = j;
1043: break;
1044: }
1045: }
1046: }
1047: return(0);
1048: }
1050: /*
1051: Checks for missing diagonals
1052: */
1055: PetscErrorCode MatMissingDiagonal_SeqAIJ(Mat A,PetscTruth *missing,PetscInt *d)
1056: {
1057: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1058: PetscInt *diag,*jj = a->j,i;
1061: *missing = PETSC_FALSE;
1062: if (A->rmap.n > 0 && !jj) {
1063: *missing = PETSC_TRUE;
1064: if (d) *d = 0;
1065: PetscInfo(A,"Matrix has no entries therefor is missing diagonal");
1066: } else {
1067: diag = a->diag;
1068: for (i=0; i<A->rmap.n; i++) {
1069: if (jj[diag[i]] != i) {
1070: *missing = PETSC_TRUE;
1071: if (d) *d = i;
1072: PetscInfo1(A,"Matrix is missing diagonal number %D",i);
1073: }
1074: }
1075: }
1076: return(0);
1077: }
1081: PetscErrorCode MatRelax_SeqAIJ(Mat A,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx)
1082: {
1083: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1084: PetscScalar *x,d,*xs,sum,*t,scale,*idiag=0,*mdiag;
1085: const PetscScalar *v = a->a, *b, *bs,*xb, *ts;
1086: PetscErrorCode ierr;
1087: PetscInt n = A->cmap.n,m = A->rmap.n,i;
1088: const PetscInt *idx,*diag;
1091: its = its*lits;
1092: if (its <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits);
1094: diag = a->diag;
1095: if (!a->idiag) {
1096: PetscMalloc(3*m*sizeof(PetscScalar),&a->idiag);
1097: a->ssor = a->idiag + m;
1098: mdiag = a->ssor + m;
1100: v = a->a;
1102: /* this is wrong when fshift omega changes each iteration */
1103: if (omega == 1.0 && !fshift) {
1104: for (i=0; i<m; i++) {
1105: mdiag[i] = v[diag[i]];
1106: a->idiag[i] = 1.0/v[diag[i]];
1107: }
1108: PetscLogFlops(m);
1109: } else {
1110: for (i=0; i<m; i++) {
1111: mdiag[i] = v[diag[i]];
1112: a->idiag[i] = omega/(fshift + v[diag[i]]);
1113: }
1114: PetscLogFlops(2*m);
1115: }
1116: }
1117: t = a->ssor;
1118: idiag = a->idiag;
1119: mdiag = a->idiag + 2*m;
1121: VecGetArray(xx,&x);
1122: if (xx != bb) {
1123: VecGetArray(bb,(PetscScalar**)&b);
1124: } else {
1125: b = x;
1126: }
1128: /* We count flops by assuming the upper triangular and lower triangular parts have the same number of nonzeros */
1129: xs = x;
1130: if (flag == SOR_APPLY_UPPER) {
1131: /* apply (U + D/omega) to the vector */
1132: bs = b;
1133: for (i=0; i<m; i++) {
1134: d = fshift + a->a[diag[i]];
1135: n = a->i[i+1] - diag[i] - 1;
1136: idx = a->j + diag[i] + 1;
1137: v = a->a + diag[i] + 1;
1138: sum = b[i]*d/omega;
1139: SPARSEDENSEDOT(sum,bs,v,idx,n);
1140: x[i] = sum;
1141: }
1142: VecRestoreArray(xx,&x);
1143: if (bb != xx) {VecRestoreArray(bb,(PetscScalar**)&b);}
1144: PetscLogFlops(a->nz);
1145: return(0);
1146: }
1149: /* Let A = L + U + D; where L is lower trianglar,
1150: U is upper triangular, E is diagonal; This routine applies
1152: (L + E)^{-1} A (U + E)^{-1}
1154: to a vector efficiently using Eisenstat's trick. This is for
1155: the case of SSOR preconditioner, so E is D/omega where omega
1156: is the relaxation factor.
1157: */
1159: if (flag == SOR_APPLY_LOWER) {
1160: SETERRQ(PETSC_ERR_SUP,"SOR_APPLY_LOWER is not implemented");
1161: } else if (flag & SOR_EISENSTAT) {
1162: /* Let A = L + U + D; where L is lower trianglar,
1163: U is upper triangular, E is diagonal; This routine applies
1165: (L + E)^{-1} A (U + E)^{-1}
1167: to a vector efficiently using Eisenstat's trick. This is for
1168: the case of SSOR preconditioner, so E is D/omega where omega
1169: is the relaxation factor.
1170: */
1171: scale = (2.0/omega) - 1.0;
1173: /* x = (E + U)^{-1} b */
1174: for (i=m-1; i>=0; i--) {
1175: n = a->i[i+1] - diag[i] - 1;
1176: idx = a->j + diag[i] + 1;
1177: v = a->a + diag[i] + 1;
1178: sum = b[i];
1179: SPARSEDENSEMDOT(sum,xs,v,idx,n);
1180: x[i] = sum*idiag[i];
1181: }
1183: /* t = b - (2*E - D)x */
1184: v = a->a;
1185: for (i=0; i<m; i++) { t[i] = b[i] - scale*(v[*diag++])*x[i]; }
1187: /* t = (E + L)^{-1}t */
1188: ts = t;
1189: diag = a->diag;
1190: for (i=0; i<m; i++) {
1191: n = diag[i] - a->i[i];
1192: idx = a->j + a->i[i];
1193: v = a->a + a->i[i];
1194: sum = t[i];
1195: SPARSEDENSEMDOT(sum,ts,v,idx,n);
1196: t[i] = sum*idiag[i];
1197: /* x = x + t */
1198: x[i] += t[i];
1199: }
1201: PetscLogFlops(6*m-1 + 2*a->nz);
1202: VecRestoreArray(xx,&x);
1203: if (bb != xx) {VecRestoreArray(bb,(PetscScalar**)&b);}
1204: return(0);
1205: }
1206: if (flag & SOR_ZERO_INITIAL_GUESS) {
1207: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP){
1208: #if defined(PETSC_USE_FORTRAN_KERNEL_RELAXAIJ)
1209: fortranrelaxaijforwardzero_(&m,&omega,x,a->i,a->j,(PetscInt*)diag,idiag,a->a,(void*)b);
1210: #else
1211: for (i=0; i<m; i++) {
1212: n = diag[i] - a->i[i];
1213: idx = a->j + a->i[i];
1214: v = a->a + a->i[i];
1215: sum = b[i];
1216: SPARSEDENSEMDOT(sum,xs,v,idx,n);
1217: x[i] = sum*idiag[i];
1218: }
1219: #endif
1220: xb = x;
1221: PetscLogFlops(a->nz);
1222: } else xb = b;
1223: if ((flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP) &&
1224: (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP)) {
1225: for (i=0; i<m; i++) {
1226: x[i] *= mdiag[i];
1227: }
1228: PetscLogFlops(m);
1229: }
1230: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP){
1231: #if defined(PETSC_USE_FORTRAN_KERNEL_RELAXAIJ)
1232: fortranrelaxaijbackwardzero_(&m,&omega,x,a->i,a->j,(PetscInt*)diag,idiag,a->a,(void*)xb);
1233: #else
1234: for (i=m-1; i>=0; i--) {
1235: n = a->i[i+1] - diag[i] - 1;
1236: idx = a->j + diag[i] + 1;
1237: v = a->a + diag[i] + 1;
1238: sum = xb[i];
1239: SPARSEDENSEMDOT(sum,xs,v,idx,n);
1240: x[i] = sum*idiag[i];
1241: }
1242: #endif
1243: PetscLogFlops(a->nz);
1244: }
1245: its--;
1246: }
1247: while (its--) {
1248: if (flag & SOR_FORWARD_SWEEP || flag & SOR_LOCAL_FORWARD_SWEEP){
1249: #if defined(PETSC_USE_FORTRAN_KERNEL_RELAXAIJ)
1250: fortranrelaxaijforward_(&m,&omega,x,a->i,a->j,(PetscInt*)diag,a->a,(void*)b);
1251: #else
1252: for (i=0; i<m; i++) {
1253: d = fshift + a->a[diag[i]];
1254: n = a->i[i+1] - a->i[i];
1255: idx = a->j + a->i[i];
1256: v = a->a + a->i[i];
1257: sum = b[i];
1258: SPARSEDENSEMDOT(sum,xs,v,idx,n);
1259: x[i] = (1. - omega)*x[i] + (sum + mdiag[i]*x[i])*idiag[i];
1260: }
1261: #endif
1262: PetscLogFlops(a->nz);
1263: }
1264: if (flag & SOR_BACKWARD_SWEEP || flag & SOR_LOCAL_BACKWARD_SWEEP){
1265: #if defined(PETSC_USE_FORTRAN_KERNEL_RELAXAIJ)
1266: fortranrelaxaijbackward_(&m,&omega,x,a->i,a->j,(PetscInt*)diag,a->a,(void*)b);
1267: #else
1268: for (i=m-1; i>=0; i--) {
1269: d = fshift + a->a[diag[i]];
1270: n = a->i[i+1] - a->i[i];
1271: idx = a->j + a->i[i];
1272: v = a->a + a->i[i];
1273: sum = b[i];
1274: SPARSEDENSEMDOT(sum,xs,v,idx,n);
1275: x[i] = (1. - omega)*x[i] + (sum + mdiag[i]*x[i])*idiag[i];
1276: }
1277: #endif
1278: PetscLogFlops(a->nz);
1279: }
1280: }
1281: VecRestoreArray(xx,&x);
1282: if (bb != xx) {VecRestoreArray(bb,(PetscScalar**)&b);}
1283: return(0);
1284: }
1288: PetscErrorCode MatGetInfo_SeqAIJ(Mat A,MatInfoType flag,MatInfo *info)
1289: {
1290: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1293: info->rows_global = (double)A->rmap.n;
1294: info->columns_global = (double)A->cmap.n;
1295: info->rows_local = (double)A->rmap.n;
1296: info->columns_local = (double)A->cmap.n;
1297: info->block_size = 1.0;
1298: info->nz_allocated = (double)a->maxnz;
1299: info->nz_used = (double)a->nz;
1300: info->nz_unneeded = (double)(a->maxnz - a->nz);
1301: info->assemblies = (double)A->num_ass;
1302: info->mallocs = (double)a->reallocs;
1303: info->memory = A->mem;
1304: if (A->factor) {
1305: info->fill_ratio_given = A->info.fill_ratio_given;
1306: info->fill_ratio_needed = A->info.fill_ratio_needed;
1307: info->factor_mallocs = A->info.factor_mallocs;
1308: } else {
1309: info->fill_ratio_given = 0;
1310: info->fill_ratio_needed = 0;
1311: info->factor_mallocs = 0;
1312: }
1313: return(0);
1314: }
1318: PetscErrorCode MatZeroRows_SeqAIJ(Mat A,PetscInt N,const PetscInt rows[],PetscScalar diag)
1319: {
1320: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1321: PetscInt i,m = A->rmap.n - 1,d;
1323: PetscTruth missing;
1326: if (a->keepzeroedrows) {
1327: for (i=0; i<N; i++) {
1328: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
1329: PetscMemzero(&a->a[a->i[rows[i]]],a->ilen[rows[i]]*sizeof(PetscScalar));
1330: }
1331: if (diag != 0.0) {
1332: MatMissingDiagonal_SeqAIJ(A,&missing,&d);
1333: if (missing) SETERRQ1(PETSC_ERR_ARG_WRONGSTATE,"Matrix is missing diagonal entry in row %D",d);
1334: for (i=0; i<N; i++) {
1335: a->a[a->diag[rows[i]]] = diag;
1336: }
1337: }
1338: A->same_nonzero = PETSC_TRUE;
1339: } else {
1340: if (diag != 0.0) {
1341: for (i=0; i<N; i++) {
1342: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
1343: if (a->ilen[rows[i]] > 0) {
1344: a->ilen[rows[i]] = 1;
1345: a->a[a->i[rows[i]]] = diag;
1346: a->j[a->i[rows[i]]] = rows[i];
1347: } else { /* in case row was completely empty */
1348: MatSetValues_SeqAIJ(A,1,&rows[i],1,&rows[i],&diag,INSERT_VALUES);
1349: }
1350: }
1351: } else {
1352: for (i=0; i<N; i++) {
1353: if (rows[i] < 0 || rows[i] > m) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"row %D out of range", rows[i]);
1354: a->ilen[rows[i]] = 0;
1355: }
1356: }
1357: A->same_nonzero = PETSC_FALSE;
1358: }
1359: MatAssemblyEnd_SeqAIJ(A,MAT_FINAL_ASSEMBLY);
1360: return(0);
1361: }
1365: PetscErrorCode MatGetRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1366: {
1367: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1368: PetscInt *itmp;
1371: if (row < 0 || row >= A->rmap.n) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Row %D out of range",row);
1373: *nz = a->i[row+1] - a->i[row];
1374: if (v) *v = a->a + a->i[row];
1375: if (idx) {
1376: itmp = a->j + a->i[row];
1377: if (*nz) {
1378: *idx = itmp;
1379: }
1380: else *idx = 0;
1381: }
1382: return(0);
1383: }
1385: /* remove this function? */
1388: PetscErrorCode MatRestoreRow_SeqAIJ(Mat A,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v)
1389: {
1391: return(0);
1392: }
1396: PetscErrorCode MatNorm_SeqAIJ(Mat A,NormType type,PetscReal *nrm)
1397: {
1398: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1399: PetscScalar *v = a->a;
1400: PetscReal sum = 0.0;
1402: PetscInt i,j;
1405: if (type == NORM_FROBENIUS) {
1406: for (i=0; i<a->nz; i++) {
1407: #if defined(PETSC_USE_COMPLEX)
1408: sum += PetscRealPart(PetscConj(*v)*(*v)); v++;
1409: #else
1410: sum += (*v)*(*v); v++;
1411: #endif
1412: }
1413: *nrm = sqrt(sum);
1414: } else if (type == NORM_1) {
1415: PetscReal *tmp;
1416: PetscInt *jj = a->j;
1417: PetscMalloc((A->cmap.n+1)*sizeof(PetscReal),&tmp);
1418: PetscMemzero(tmp,A->cmap.n*sizeof(PetscReal));
1419: *nrm = 0.0;
1420: for (j=0; j<a->nz; j++) {
1421: tmp[*jj++] += PetscAbsScalar(*v); v++;
1422: }
1423: for (j=0; j<A->cmap.n; j++) {
1424: if (tmp[j] > *nrm) *nrm = tmp[j];
1425: }
1426: PetscFree(tmp);
1427: } else if (type == NORM_INFINITY) {
1428: *nrm = 0.0;
1429: for (j=0; j<A->rmap.n; j++) {
1430: v = a->a + a->i[j];
1431: sum = 0.0;
1432: for (i=0; i<a->i[j+1]-a->i[j]; i++) {
1433: sum += PetscAbsScalar(*v); v++;
1434: }
1435: if (sum > *nrm) *nrm = sum;
1436: }
1437: } else {
1438: SETERRQ(PETSC_ERR_SUP,"No support for two norm");
1439: }
1440: return(0);
1441: }
1445: PetscErrorCode MatTranspose_SeqAIJ(Mat A,Mat *B)
1446: {
1447: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1448: Mat C;
1450: PetscInt i,*aj = a->j,*ai = a->i,m = A->rmap.n,len,*col;
1451: PetscScalar *array = a->a;
1454: if (!B && m != A->cmap.n) SETERRQ(PETSC_ERR_ARG_SIZ,"Square matrix only for in-place");
1455: PetscMalloc((1+A->cmap.n)*sizeof(PetscInt),&col);
1456: PetscMemzero(col,(1+A->cmap.n)*sizeof(PetscInt));
1457:
1458: for (i=0; i<ai[m]; i++) col[aj[i]] += 1;
1459: MatCreate(A->comm,&C);
1460: MatSetSizes(C,A->cmap.n,m,A->cmap.n,m);
1461: MatSetType(C,A->type_name);
1462: MatSeqAIJSetPreallocation_SeqAIJ(C,0,col);
1463: PetscFree(col);
1464: for (i=0; i<m; i++) {
1465: len = ai[i+1]-ai[i];
1466: MatSetValues_SeqAIJ(C,len,aj,1,&i,array,INSERT_VALUES);
1467: array += len;
1468: aj += len;
1469: }
1471: MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);
1472: MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);
1474: if (B) {
1475: *B = C;
1476: } else {
1477: MatHeaderCopy(A,C);
1478: }
1479: return(0);
1480: }
1485: PetscErrorCode MatIsTranspose_SeqAIJ(Mat A,Mat B,PetscReal tol,PetscTruth *f)
1486: {
1487: Mat_SeqAIJ *aij = (Mat_SeqAIJ *) A->data,*bij = (Mat_SeqAIJ*) A->data;
1488: PetscInt *adx,*bdx,*aii,*bii,*aptr,*bptr; PetscScalar *va,*vb;
1490: PetscInt ma,na,mb,nb, i;
1493: bij = (Mat_SeqAIJ *) B->data;
1494:
1495: MatGetSize(A,&ma,&na);
1496: MatGetSize(B,&mb,&nb);
1497: if (ma!=nb || na!=mb){
1498: *f = PETSC_FALSE;
1499: return(0);
1500: }
1501: aii = aij->i; bii = bij->i;
1502: adx = aij->j; bdx = bij->j;
1503: va = aij->a; vb = bij->a;
1504: PetscMalloc(ma*sizeof(PetscInt),&aptr);
1505: PetscMalloc(mb*sizeof(PetscInt),&bptr);
1506: for (i=0; i<ma; i++) aptr[i] = aii[i];
1507: for (i=0; i<mb; i++) bptr[i] = bii[i];
1509: *f = PETSC_TRUE;
1510: for (i=0; i<ma; i++) {
1511: while (aptr[i]<aii[i+1]) {
1512: PetscInt idc,idr;
1513: PetscScalar vc,vr;
1514: /* column/row index/value */
1515: idc = adx[aptr[i]];
1516: idr = bdx[bptr[idc]];
1517: vc = va[aptr[i]];
1518: vr = vb[bptr[idc]];
1519: if (i!=idr || PetscAbsScalar(vc-vr) > tol) {
1520: *f = PETSC_FALSE;
1521: goto done;
1522: } else {
1523: aptr[i]++;
1524: if (B || i!=idc) bptr[idc]++;
1525: }
1526: }
1527: }
1528: done:
1529: PetscFree(aptr);
1530: if (B) {
1531: PetscFree(bptr);
1532: }
1533: return(0);
1534: }
1539: PetscErrorCode MatIsSymmetric_SeqAIJ(Mat A,PetscReal tol,PetscTruth *f)
1540: {
1543: MatIsTranspose_SeqAIJ(A,A,tol,f);
1544: return(0);
1545: }
1549: PetscErrorCode MatDiagonalScale_SeqAIJ(Mat A,Vec ll,Vec rr)
1550: {
1551: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1552: PetscScalar *l,*r,x,*v;
1554: PetscInt i,j,m = A->rmap.n,n = A->cmap.n,M,nz = a->nz,*jj;
1557: if (ll) {
1558: /* The local size is used so that VecMPI can be passed to this routine
1559: by MatDiagonalScale_MPIAIJ */
1560: VecGetLocalSize(ll,&m);
1561: if (m != A->rmap.n) SETERRQ(PETSC_ERR_ARG_SIZ,"Left scaling vector wrong length");
1562: VecGetArray(ll,&l);
1563: v = a->a;
1564: for (i=0; i<m; i++) {
1565: x = l[i];
1566: M = a->i[i+1] - a->i[i];
1567: for (j=0; j<M; j++) { (*v++) *= x;}
1568: }
1569: VecRestoreArray(ll,&l);
1570: PetscLogFlops(nz);
1571: }
1572: if (rr) {
1573: VecGetLocalSize(rr,&n);
1574: if (n != A->cmap.n) SETERRQ(PETSC_ERR_ARG_SIZ,"Right scaling vector wrong length");
1575: VecGetArray(rr,&r);
1576: v = a->a; jj = a->j;
1577: for (i=0; i<nz; i++) {
1578: (*v++) *= r[*jj++];
1579: }
1580: VecRestoreArray(rr,&r);
1581: PetscLogFlops(nz);
1582: }
1583: return(0);
1584: }
1588: PetscErrorCode MatGetSubMatrix_SeqAIJ(Mat A,IS isrow,IS iscol,PetscInt csize,MatReuse scall,Mat *B)
1589: {
1590: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data,*c;
1592: PetscInt *smap,i,k,kstart,kend,oldcols = A->cmap.n,*lens;
1593: PetscInt row,mat_i,*mat_j,tcol,first,step,*mat_ilen,sum,lensi;
1594: PetscInt *irow,*icol,nrows,ncols;
1595: PetscInt *starts,*j_new,*i_new,*aj = a->j,*ai = a->i,ii,*ailen = a->ilen;
1596: PetscScalar *a_new,*mat_a;
1597: Mat C;
1598: PetscTruth stride;
1601: ISSorted(isrow,(PetscTruth*)&i);
1602: if (!i) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"ISrow is not sorted");
1603: ISSorted(iscol,(PetscTruth*)&i);
1604: if (!i) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"IScol is not sorted");
1606: ISGetIndices(isrow,&irow);
1607: ISGetLocalSize(isrow,&nrows);
1608: ISGetLocalSize(iscol,&ncols);
1610: ISStrideGetInfo(iscol,&first,&step);
1611: ISStride(iscol,&stride);
1612: if (stride && step == 1) {
1613: /* special case of contiguous rows */
1614: PetscMalloc((2*nrows+1)*sizeof(PetscInt),&lens);
1615: starts = lens + nrows;
1616: /* loop over new rows determining lens and starting points */
1617: for (i=0; i<nrows; i++) {
1618: kstart = ai[irow[i]];
1619: kend = kstart + ailen[irow[i]];
1620: for (k=kstart; k<kend; k++) {
1621: if (aj[k] >= first) {
1622: starts[i] = k;
1623: break;
1624: }
1625: }
1626: sum = 0;
1627: while (k < kend) {
1628: if (aj[k++] >= first+ncols) break;
1629: sum++;
1630: }
1631: lens[i] = sum;
1632: }
1633: /* create submatrix */
1634: if (scall == MAT_REUSE_MATRIX) {
1635: PetscInt n_cols,n_rows;
1636: MatGetSize(*B,&n_rows,&n_cols);
1637: if (n_rows != nrows || n_cols != ncols) SETERRQ(PETSC_ERR_ARG_SIZ,"Reused submatrix wrong size");
1638: MatZeroEntries(*B);
1639: C = *B;
1640: } else {
1641: MatCreate(A->comm,&C);
1642: MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);
1643: MatSetType(C,A->type_name);
1644: MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);
1645: }
1646: c = (Mat_SeqAIJ*)C->data;
1648: /* loop over rows inserting into submatrix */
1649: a_new = c->a;
1650: j_new = c->j;
1651: i_new = c->i;
1653: for (i=0; i<nrows; i++) {
1654: ii = starts[i];
1655: lensi = lens[i];
1656: for (k=0; k<lensi; k++) {
1657: *j_new++ = aj[ii+k] - first;
1658: }
1659: PetscMemcpy(a_new,a->a + starts[i],lensi*sizeof(PetscScalar));
1660: a_new += lensi;
1661: i_new[i+1] = i_new[i] + lensi;
1662: c->ilen[i] = lensi;
1663: }
1664: PetscFree(lens);
1665: } else {
1666: ISGetIndices(iscol,&icol);
1667: PetscMalloc((1+oldcols)*sizeof(PetscInt),&smap);
1668:
1669: PetscMalloc((1+nrows)*sizeof(PetscInt),&lens);
1670: PetscMemzero(smap,oldcols*sizeof(PetscInt));
1671: for (i=0; i<ncols; i++) smap[icol[i]] = i+1;
1672: /* determine lens of each row */
1673: for (i=0; i<nrows; i++) {
1674: kstart = ai[irow[i]];
1675: kend = kstart + a->ilen[irow[i]];
1676: lens[i] = 0;
1677: for (k=kstart; k<kend; k++) {
1678: if (smap[aj[k]]) {
1679: lens[i]++;
1680: }
1681: }
1682: }
1683: /* Create and fill new matrix */
1684: if (scall == MAT_REUSE_MATRIX) {
1685: PetscTruth equal;
1687: c = (Mat_SeqAIJ *)((*B)->data);
1688: if ((*B)->rmap.n != nrows || (*B)->cmap.n != ncols) SETERRQ(PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong size");
1689: PetscMemcmp(c->ilen,lens,(*B)->rmap.n*sizeof(PetscInt),&equal);
1690: if (!equal) {
1691: SETERRQ(PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong no of nonzeros");
1692: }
1693: PetscMemzero(c->ilen,(*B)->rmap.n*sizeof(PetscInt));
1694: C = *B;
1695: } else {
1696: MatCreate(A->comm,&C);
1697: MatSetSizes(C,nrows,ncols,PETSC_DETERMINE,PETSC_DETERMINE);
1698: MatSetType(C,A->type_name);
1699: MatSeqAIJSetPreallocation_SeqAIJ(C,0,lens);
1700: }
1701: c = (Mat_SeqAIJ *)(C->data);
1702: for (i=0; i<nrows; i++) {
1703: row = irow[i];
1704: kstart = ai[row];
1705: kend = kstart + a->ilen[row];
1706: mat_i = c->i[i];
1707: mat_j = c->j + mat_i;
1708: mat_a = c->a + mat_i;
1709: mat_ilen = c->ilen + i;
1710: for (k=kstart; k<kend; k++) {
1711: if ((tcol=smap[a->j[k]])) {
1712: *mat_j++ = tcol - 1;
1713: *mat_a++ = a->a[k];
1714: (*mat_ilen)++;
1716: }
1717: }
1718: }
1719: /* Free work space */
1720: ISRestoreIndices(iscol,&icol);
1721: PetscFree(smap);
1722: PetscFree(lens);
1723: }
1724: MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);
1725: MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);
1727: ISRestoreIndices(isrow,&irow);
1728: *B = C;
1729: return(0);
1730: }
1732: /*
1733: */
1736: PetscErrorCode MatILUFactor_SeqAIJ(Mat inA,IS row,IS col,MatFactorInfo *info)
1737: {
1738: Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data;
1740: Mat outA;
1741: PetscTruth row_identity,col_identity;
1744: if (info->levels != 0) SETERRQ(PETSC_ERR_SUP,"Only levels=0 supported for in-place ilu");
1745: ISIdentity(row,&row_identity);
1746: ISIdentity(col,&col_identity);
1747: if (!row_identity || !col_identity) {
1748: SETERRQ(PETSC_ERR_ARG_WRONG,"Row and column permutations must be identity for in-place ILU");
1749: }
1751: outA = inA;
1752: inA->factor = FACTOR_LU;
1753: a->row = row;
1754: a->col = col;
1755: PetscObjectReference((PetscObject)row);
1756: PetscObjectReference((PetscObject)col);
1758: /* Create the inverse permutation so that it can be used in MatLUFactorNumeric() */
1759: if (a->icol) {ISDestroy(a->icol);} /* need to remove old one */
1760: ISInvertPermutation(col,PETSC_DECIDE,&a->icol);
1761: PetscLogObjectParent(inA,a->icol);
1763: if (!a->solve_work) { /* this matrix may have been factored before */
1764: PetscMalloc((inA->rmap.n+1)*sizeof(PetscScalar),&a->solve_work);
1765: }
1767: MatMarkDiagonal_SeqAIJ(inA);
1768: MatLUFactorNumeric_SeqAIJ(inA,info,&outA);
1769: return(0);
1770: }
1772: #include petscblaslapack.h
1775: PetscErrorCode MatScale_SeqAIJ(Mat inA,PetscScalar alpha)
1776: {
1777: Mat_SeqAIJ *a = (Mat_SeqAIJ*)inA->data;
1778: PetscBLASInt bnz = (PetscBLASInt)a->nz,one = 1;
1779: PetscScalar oalpha = alpha;
1784: BLASscal_(&bnz,&oalpha,a->a,&one);
1785: PetscLogFlops(a->nz);
1786: return(0);
1787: }
1791: PetscErrorCode MatGetSubMatrices_SeqAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[])
1792: {
1794: PetscInt i;
1797: if (scall == MAT_INITIAL_MATRIX) {
1798: PetscMalloc((n+1)*sizeof(Mat),B);
1799: }
1801: for (i=0; i<n; i++) {
1802: MatGetSubMatrix_SeqAIJ(A,irow[i],icol[i],PETSC_DECIDE,scall,&(*B)[i]);
1803: }
1804: return(0);
1805: }
1809: PetscErrorCode MatIncreaseOverlap_SeqAIJ(Mat A,PetscInt is_max,IS is[],PetscInt ov)
1810: {
1811: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1813: PetscInt row,i,j,k,l,m,n,*idx,*nidx,isz,val;
1814: PetscInt start,end,*ai,*aj;
1815: PetscBT table;
1818: m = A->rmap.n;
1819: ai = a->i;
1820: aj = a->j;
1822: if (ov < 0) SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"illegal negative overlap value used");
1824: PetscMalloc((m+1)*sizeof(PetscInt),&nidx);
1825: PetscBTCreate(m,table);
1827: for (i=0; i<is_max; i++) {
1828: /* Initialize the two local arrays */
1829: isz = 0;
1830: PetscBTMemzero(m,table);
1831:
1832: /* Extract the indices, assume there can be duplicate entries */
1833: ISGetIndices(is[i],&idx);
1834: ISGetLocalSize(is[i],&n);
1835:
1836: /* Enter these into the temp arrays. I.e., mark table[row], enter row into new index */
1837: for (j=0; j<n ; ++j){
1838: if(!PetscBTLookupSet(table,idx[j])) { nidx[isz++] = idx[j];}
1839: }
1840: ISRestoreIndices(is[i],&idx);
1841: ISDestroy(is[i]);
1842:
1843: k = 0;
1844: for (j=0; j<ov; j++){ /* for each overlap */
1845: n = isz;
1846: for (; k<n ; k++){ /* do only those rows in nidx[k], which are not done yet */
1847: row = nidx[k];
1848: start = ai[row];
1849: end = ai[row+1];
1850: for (l = start; l<end ; l++){
1851: val = aj[l] ;
1852: if (!PetscBTLookupSet(table,val)) {nidx[isz++] = val;}
1853: }
1854: }
1855: }
1856: ISCreateGeneral(PETSC_COMM_SELF,isz,nidx,(is+i));
1857: }
1858: PetscBTDestroy(table);
1859: PetscFree(nidx);
1860: return(0);
1861: }
1863: /* -------------------------------------------------------------- */
1866: PetscErrorCode MatPermute_SeqAIJ(Mat A,IS rowp,IS colp,Mat *B)
1867: {
1868: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1870: PetscInt i,nz,m = A->rmap.n,n = A->cmap.n,*col;
1871: PetscInt *row,*cnew,j,*lens;
1872: IS icolp,irowp;
1873: PetscInt *cwork;
1874: PetscScalar *vwork;
1877: ISInvertPermutation(rowp,PETSC_DECIDE,&irowp);
1878: ISGetIndices(irowp,&row);
1879: ISInvertPermutation(colp,PETSC_DECIDE,&icolp);
1880: ISGetIndices(icolp,&col);
1881:
1882: /* determine lengths of permuted rows */
1883: PetscMalloc((m+1)*sizeof(PetscInt),&lens);
1884: for (i=0; i<m; i++) {
1885: lens[row[i]] = a->i[i+1] - a->i[i];
1886: }
1887: MatCreate(A->comm,B);
1888: MatSetSizes(*B,m,n,m,n);
1889: MatSetType(*B,A->type_name);
1890: MatSeqAIJSetPreallocation_SeqAIJ(*B,0,lens);
1891: PetscFree(lens);
1893: PetscMalloc(n*sizeof(PetscInt),&cnew);
1894: for (i=0; i<m; i++) {
1895: MatGetRow_SeqAIJ(A,i,&nz,&cwork,&vwork);
1896: for (j=0; j<nz; j++) { cnew[j] = col[cwork[j]];}
1897: MatSetValues_SeqAIJ(*B,1,&row[i],nz,cnew,vwork,INSERT_VALUES);
1898: MatRestoreRow_SeqAIJ(A,i,&nz,&cwork,&vwork);
1899: }
1900: PetscFree(cnew);
1901: (*B)->assembled = PETSC_FALSE;
1902: MatAssemblyBegin(*B,MAT_FINAL_ASSEMBLY);
1903: MatAssemblyEnd(*B,MAT_FINAL_ASSEMBLY);
1904: ISRestoreIndices(irowp,&row);
1905: ISRestoreIndices(icolp,&col);
1906: ISDestroy(irowp);
1907: ISDestroy(icolp);
1908: return(0);
1909: }
1913: PetscErrorCode MatCopy_SeqAIJ(Mat A,Mat B,MatStructure str)
1914: {
1918: /* If the two matrices have the same copy implementation, use fast copy. */
1919: if (str == SAME_NONZERO_PATTERN && (A->ops->copy == B->ops->copy)) {
1920: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1921: Mat_SeqAIJ *b = (Mat_SeqAIJ*)B->data;
1923: if (a->i[A->rmap.n] != b->i[B->rmap.n]) {
1924: SETERRQ(PETSC_ERR_ARG_INCOMP,"Number of nonzeros in two matrices are different");
1925: }
1926: PetscMemcpy(b->a,a->a,(a->i[A->rmap.n])*sizeof(PetscScalar));
1927: } else {
1928: MatCopy_Basic(A,B,str);
1929: }
1930: return(0);
1931: }
1935: PetscErrorCode MatSetUpPreallocation_SeqAIJ(Mat A)
1936: {
1940: MatSeqAIJSetPreallocation_SeqAIJ(A,PETSC_DEFAULT,0);
1941: return(0);
1942: }
1946: PetscErrorCode MatGetArray_SeqAIJ(Mat A,PetscScalar *array[])
1947: {
1948: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
1950: *array = a->a;
1951: return(0);
1952: }
1956: PetscErrorCode MatRestoreArray_SeqAIJ(Mat A,PetscScalar *array[])
1957: {
1959: return(0);
1960: }
1964: PetscErrorCode MatFDColoringApply_SeqAIJ(Mat J,MatFDColoring coloring,Vec x1,MatStructure *flag,void *sctx)
1965: {
1966: PetscErrorCode (*f)(void*,Vec,Vec,void*) = (PetscErrorCode (*)(void*,Vec,Vec,void *))coloring->f;
1968: PetscInt k,N,start,end,l,row,col,srow,**vscaleforrow,m1,m2;
1969: PetscScalar dx,*y,*xx,*w3_array;
1970: PetscScalar *vscale_array;
1971: PetscReal epsilon = coloring->error_rel,umin = coloring->umin;
1972: Vec w1,w2,w3;
1973: void *fctx = coloring->fctx;
1974: PetscTruth flg;
1977: if (!coloring->w1) {
1978: VecDuplicate(x1,&coloring->w1);
1979: PetscLogObjectParent(coloring,coloring->w1);
1980: VecDuplicate(x1,&coloring->w2);
1981: PetscLogObjectParent(coloring,coloring->w2);
1982: VecDuplicate(x1,&coloring->w3);
1983: PetscLogObjectParent(coloring,coloring->w3);
1984: }
1985: w1 = coloring->w1; w2 = coloring->w2; w3 = coloring->w3;
1987: MatSetUnfactored(J);
1988: PetscOptionsHasName(coloring->prefix,"-mat_fd_coloring_dont_rezero",&flg);
1989: if (flg) {
1990: PetscInfo(coloring,"Not calling MatZeroEntries()\n");
1991: } else {
1992: PetscTruth assembled;
1993: MatAssembled(J,&assembled);
1994: if (assembled) {
1995: MatZeroEntries(J);
1996: }
1997: }
1999: VecGetOwnershipRange(x1,&start,&end);
2000: VecGetSize(x1,&N);
2002: /*
2003: This is a horrible, horrible, hack. See DMMGComputeJacobian_Multigrid() it inproperly sets
2004: coloring->F for the coarser grids from the finest
2005: */
2006: if (coloring->F) {
2007: VecGetLocalSize(coloring->F,&m1);
2008: VecGetLocalSize(w1,&m2);
2009: if (m1 != m2) {
2010: coloring->F = 0;
2011: }
2012: }
2014: if (coloring->F) {
2015: w1 = coloring->F;
2016: coloring->F = 0;
2017: } else {
2019: (*f)(sctx,x1,w1,fctx);
2021: }
2023: /*
2024: Compute all the scale factors and share with other processors
2025: */
2026: VecGetArray(x1,&xx);xx = xx - start;
2027: VecGetArray(coloring->vscale,&vscale_array);vscale_array = vscale_array - start;
2028: for (k=0; k<coloring->ncolors; k++) {
2029: /*
2030: Loop over each column associated with color adding the
2031: perturbation to the vector w3.
2032: */
2033: for (l=0; l<coloring->ncolumns[k]; l++) {
2034: col = coloring->columns[k][l]; /* column of the matrix we are probing for */
2035: dx = xx[col];
2036: if (dx == 0.0) dx = 1.0;
2037: #if !defined(PETSC_USE_COMPLEX)
2038: if (dx < umin && dx >= 0.0) dx = umin;
2039: else if (dx < 0.0 && dx > -umin) dx = -umin;
2040: #else
2041: if (PetscAbsScalar(dx) < umin && PetscRealPart(dx) >= 0.0) dx = umin;
2042: else if (PetscRealPart(dx) < 0.0 && PetscAbsScalar(dx) < umin) dx = -umin;
2043: #endif
2044: dx *= epsilon;
2045: vscale_array[col] = 1.0/dx;
2046: }
2047: }
2048: vscale_array = vscale_array + start;VecRestoreArray(coloring->vscale,&vscale_array);
2049: VecGhostUpdateBegin(coloring->vscale,INSERT_VALUES,SCATTER_FORWARD);
2050: VecGhostUpdateEnd(coloring->vscale,INSERT_VALUES,SCATTER_FORWARD);
2052: /* VecView(coloring->vscale,PETSC_VIEWER_STDOUT_WORLD);
2053: VecView(x1,PETSC_VIEWER_STDOUT_WORLD);*/
2055: if (coloring->vscaleforrow) vscaleforrow = coloring->vscaleforrow;
2056: else vscaleforrow = coloring->columnsforrow;
2058: VecGetArray(coloring->vscale,&vscale_array);
2059: /*
2060: Loop over each color
2061: */
2062: for (k=0; k<coloring->ncolors; k++) {
2063: coloring->currentcolor = k;
2064: VecCopy(x1,w3);
2065: VecGetArray(w3,&w3_array);w3_array = w3_array - start;
2066: /*
2067: Loop over each column associated with color adding the
2068: perturbation to the vector w3.
2069: */
2070: for (l=0; l<coloring->ncolumns[k]; l++) {
2071: col = coloring->columns[k][l]; /* column of the matrix we are probing for */
2072: dx = xx[col];
2073: if (dx == 0.0) dx = 1.0;
2074: #if !defined(PETSC_USE_COMPLEX)
2075: if (dx < umin && dx >= 0.0) dx = umin;
2076: else if (dx < 0.0 && dx > -umin) dx = -umin;
2077: #else
2078: if (PetscAbsScalar(dx) < umin && PetscRealPart(dx) >= 0.0) dx = umin;
2079: else if (PetscRealPart(dx) < 0.0 && PetscAbsScalar(dx) < umin) dx = -umin;
2080: #endif
2081: dx *= epsilon;
2082: if (!PetscAbsScalar(dx)) SETERRQ(PETSC_ERR_PLIB,"Computed 0 differencing parameter");
2083: w3_array[col] += dx;
2084: }
2085: w3_array = w3_array + start; VecRestoreArray(w3,&w3_array);
2087: /*
2088: Evaluate function at x1 + dx (here dx is a vector of perturbations)
2089: */
2092: (*f)(sctx,w3,w2,fctx);
2094: VecAXPY(w2,-1.0,w1);
2096: /*
2097: Loop over rows of vector, putting results into Jacobian matrix
2098: */
2099: VecGetArray(w2,&y);
2100: for (l=0; l<coloring->nrows[k]; l++) {
2101: row = coloring->rows[k][l];
2102: col = coloring->columnsforrow[k][l];
2103: y[row] *= vscale_array[vscaleforrow[k][l]];
2104: srow = row + start;
2105: MatSetValues_SeqAIJ(J,1,&srow,1,&col,y+row,INSERT_VALUES);
2106: }
2107: VecRestoreArray(w2,&y);
2108: }
2109: coloring->currentcolor = k;
2110: VecRestoreArray(coloring->vscale,&vscale_array);
2111: xx = xx + start; VecRestoreArray(x1,&xx);
2112: MatAssemblyBegin(J,MAT_FINAL_ASSEMBLY);
2113: MatAssemblyEnd(J,MAT_FINAL_ASSEMBLY);
2114: return(0);
2115: }
2117: #include petscblaslapack.h
2120: PetscErrorCode MatAXPY_SeqAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str)
2121: {
2123: PetscInt i;
2124: Mat_SeqAIJ *x = (Mat_SeqAIJ *)X->data,*y = (Mat_SeqAIJ *)Y->data;
2125: PetscBLASInt one=1,bnz = (PetscBLASInt)x->nz;
2128: if (str == SAME_NONZERO_PATTERN) {
2129: PetscScalar alpha = a;
2130: BLASaxpy_(&bnz,&alpha,x->a,&one,y->a,&one);
2131: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
2132: if (y->xtoy && y->XtoY != X) {
2133: PetscFree(y->xtoy);
2134: MatDestroy(y->XtoY);
2135: }
2136: if (!y->xtoy) { /* get xtoy */
2137: MatAXPYGetxtoy_Private(X->rmap.n,x->i,x->j,PETSC_NULL, y->i,y->j,PETSC_NULL, &y->xtoy);
2138: y->XtoY = X;
2139: }
2140: for (i=0; i<x->nz; i++) y->a[y->xtoy[i]] += a*(x->a[i]);
2141: PetscInfo3(0,"ratio of nnz(X)/nnz(Y): %d/%d = %G\n",x->nz,y->nz,(PetscReal)(x->nz)/y->nz);
2142: } else {
2143: MatAXPY_Basic(Y,a,X,str);
2144: }
2145: return(0);
2146: }
2150: PetscErrorCode MatSetBlockSize_SeqAIJ(Mat A,PetscInt bs)
2151: {
2153: return(0);
2154: }
2158: PetscErrorCode MatConjugate_SeqAIJ(Mat mat)
2159: {
2160: #if defined(PETSC_USE_COMPLEX)
2161: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
2162: PetscInt i,nz;
2163: PetscScalar *a;
2166: nz = aij->nz;
2167: a = aij->a;
2168: for (i=0; i<nz; i++) {
2169: a[i] = PetscConj(a[i]);
2170: }
2171: #else
2173: #endif
2174: return(0);
2175: }
2179: PetscErrorCode MatGetRowMax_SeqAIJ(Mat A,Vec v)
2180: {
2181: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
2183: PetscInt i,j,m = A->rmap.n,*ai,*aj,ncols,n;
2184: PetscReal atmp;
2185: PetscScalar *x,zero = 0.0;
2186: MatScalar *aa;
2189: if (A->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2190: aa = a->a;
2191: ai = a->i;
2192: aj = a->j;
2194: VecSet(v,zero);
2195: VecGetArray(v,&x);
2196: VecGetLocalSize(v,&n);
2197: if (n != A->rmap.n) SETERRQ(PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
2198: for (i=0; i<m; i++) {
2199: ncols = ai[1] - ai[0]; ai++;
2200: for (j=0; j<ncols; j++){
2201: atmp = PetscAbsScalar(*aa); aa++;
2202: if (PetscAbsScalar(x[i]) < atmp) x[i] = atmp;
2203: aj++;
2204: }
2205: }
2206: VecRestoreArray(v,&x);
2207: return(0);
2208: }
2210: /* -------------------------------------------------------------------*/
2211: static struct _MatOps MatOps_Values = {MatSetValues_SeqAIJ,
2212: MatGetRow_SeqAIJ,
2213: MatRestoreRow_SeqAIJ,
2214: MatMult_SeqAIJ,
2215: /* 4*/ MatMultAdd_SeqAIJ,
2216: MatMultTranspose_SeqAIJ,
2217: MatMultTransposeAdd_SeqAIJ,
2218: MatSolve_SeqAIJ,
2219: MatSolveAdd_SeqAIJ,
2220: MatSolveTranspose_SeqAIJ,
2221: /*10*/ MatSolveTransposeAdd_SeqAIJ,
2222: MatLUFactor_SeqAIJ,
2223: 0,
2224: MatRelax_SeqAIJ,
2225: MatTranspose_SeqAIJ,
2226: /*15*/ MatGetInfo_SeqAIJ,
2227: MatEqual_SeqAIJ,
2228: MatGetDiagonal_SeqAIJ,
2229: MatDiagonalScale_SeqAIJ,
2230: MatNorm_SeqAIJ,
2231: /*20*/ 0,
2232: MatAssemblyEnd_SeqAIJ,
2233: MatCompress_SeqAIJ,
2234: MatSetOption_SeqAIJ,
2235: MatZeroEntries_SeqAIJ,
2236: /*25*/ MatZeroRows_SeqAIJ,
2237: MatLUFactorSymbolic_SeqAIJ,
2238: MatLUFactorNumeric_SeqAIJ,
2239: MatCholeskyFactorSymbolic_SeqAIJ,
2240: MatCholeskyFactorNumeric_SeqAIJ,
2241: /*30*/ MatSetUpPreallocation_SeqAIJ,
2242: MatILUFactorSymbolic_SeqAIJ,
2243: MatICCFactorSymbolic_SeqAIJ,
2244: MatGetArray_SeqAIJ,
2245: MatRestoreArray_SeqAIJ,
2246: /*35*/ MatDuplicate_SeqAIJ,
2247: 0,
2248: 0,
2249: MatILUFactor_SeqAIJ,
2250: 0,
2251: /*40*/ MatAXPY_SeqAIJ,
2252: MatGetSubMatrices_SeqAIJ,
2253: MatIncreaseOverlap_SeqAIJ,
2254: MatGetValues_SeqAIJ,
2255: MatCopy_SeqAIJ,
2256: /*45*/ 0,
2257: MatScale_SeqAIJ,
2258: 0,
2259: MatDiagonalSet_SeqAIJ,
2260: MatILUDTFactor_SeqAIJ,
2261: /*50*/ MatSetBlockSize_SeqAIJ,
2262: MatGetRowIJ_SeqAIJ,
2263: MatRestoreRowIJ_SeqAIJ,
2264: MatGetColumnIJ_SeqAIJ,
2265: MatRestoreColumnIJ_SeqAIJ,
2266: /*55*/ MatFDColoringCreate_SeqAIJ,
2267: 0,
2268: 0,
2269: MatPermute_SeqAIJ,
2270: 0,
2271: /*60*/ 0,
2272: MatDestroy_SeqAIJ,
2273: MatView_SeqAIJ,
2274: 0,
2275: 0,
2276: /*65*/ 0,
2277: 0,
2278: 0,
2279: 0,
2280: 0,
2281: /*70*/ MatGetRowMax_SeqAIJ,
2282: 0,
2283: MatSetColoring_SeqAIJ,
2284: #if defined(PETSC_HAVE_ADIC)
2285: MatSetValuesAdic_SeqAIJ,
2286: #else
2287: 0,
2288: #endif
2289: MatSetValuesAdifor_SeqAIJ,
2290: /*75*/ 0,
2291: 0,
2292: 0,
2293: 0,
2294: 0,
2295: /*80*/ 0,
2296: 0,
2297: 0,
2298: 0,
2299: MatLoad_SeqAIJ,
2300: /*85*/ MatIsSymmetric_SeqAIJ,
2301: 0,
2302: 0,
2303: 0,
2304: 0,
2305: /*90*/ MatMatMult_SeqAIJ_SeqAIJ,
2306: MatMatMultSymbolic_SeqAIJ_SeqAIJ,
2307: MatMatMultNumeric_SeqAIJ_SeqAIJ,
2308: MatPtAP_Basic,
2309: MatPtAPSymbolic_SeqAIJ,
2310: /*95*/ MatPtAPNumeric_SeqAIJ,
2311: MatMatMultTranspose_SeqAIJ_SeqAIJ,
2312: MatMatMultTransposeSymbolic_SeqAIJ_SeqAIJ,
2313: MatMatMultTransposeNumeric_SeqAIJ_SeqAIJ,
2314: MatPtAPSymbolic_SeqAIJ_SeqAIJ,
2315: /*100*/MatPtAPNumeric_SeqAIJ_SeqAIJ,
2316: 0,
2317: 0,
2318: MatConjugate_SeqAIJ,
2319: 0,
2320: /*105*/MatSetValuesRow_SeqAIJ,
2321: MatRealPart_SeqAIJ,
2322: MatImaginaryPart_SeqAIJ,
2323: 0,
2324: 0,
2325: /*110*/MatMatSolve_SeqAIJ
2326: };
2331: PetscErrorCode MatSeqAIJSetColumnIndices_SeqAIJ(Mat mat,PetscInt *indices)
2332: {
2333: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
2334: PetscInt i,nz,n;
2338: nz = aij->maxnz;
2339: n = mat->cmap.n;
2340: for (i=0; i<nz; i++) {
2341: aij->j[i] = indices[i];
2342: }
2343: aij->nz = nz;
2344: for (i=0; i<n; i++) {
2345: aij->ilen[i] = aij->imax[i];
2346: }
2348: return(0);
2349: }
2354: /*@
2355: MatSeqAIJSetColumnIndices - Set the column indices for all the rows
2356: in the matrix.
2358: Input Parameters:
2359: + mat - the SeqAIJ matrix
2360: - indices - the column indices
2362: Level: advanced
2364: Notes:
2365: This can be called if you have precomputed the nonzero structure of the
2366: matrix and want to provide it to the matrix object to improve the performance
2367: of the MatSetValues() operation.
2369: You MUST have set the correct numbers of nonzeros per row in the call to
2370: MatCreateSeqAIJ(), and the columns indices MUST be sorted.
2372: MUST be called before any calls to MatSetValues();
2374: The indices should start with zero, not one.
2376: @*/
2377: PetscErrorCode MatSeqAIJSetColumnIndices(Mat mat,PetscInt *indices)
2378: {
2379: PetscErrorCode ierr,(*f)(Mat,PetscInt *);
2384: PetscObjectQueryFunction((PetscObject)mat,"MatSeqAIJSetColumnIndices_C",(void (**)(void))&f);
2385: if (f) {
2386: (*f)(mat,indices);
2387: } else {
2388: SETERRQ(PETSC_ERR_SUP,"Wrong type of matrix to set column indices");
2389: }
2390: return(0);
2391: }
2393: /* ----------------------------------------------------------------------------------------*/
2398: PetscErrorCode MatStoreValues_SeqAIJ(Mat mat)
2399: {
2400: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
2402: size_t nz = aij->i[mat->rmap.n];
2405: if (aij->nonew != 1) {
2406: SETERRQ(PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NO_NEW_NONZERO_LOCATIONS);first");
2407: }
2409: /* allocate space for values if not already there */
2410: if (!aij->saved_values) {
2411: PetscMalloc((nz+1)*sizeof(PetscScalar),&aij->saved_values);
2412: }
2414: /* copy values over */
2415: PetscMemcpy(aij->saved_values,aij->a,nz*sizeof(PetscScalar));
2416: return(0);
2417: }
2422: /*@
2423: MatStoreValues - Stashes a copy of the matrix values; this allows, for
2424: example, reuse of the linear part of a Jacobian, while recomputing the
2425: nonlinear portion.
2427: Collect on Mat
2429: Input Parameters:
2430: . mat - the matrix (currently only AIJ matrices support this option)
2432: Level: advanced
2434: Common Usage, with SNESSolve():
2435: $ Create Jacobian matrix
2436: $ Set linear terms into matrix
2437: $ Apply boundary conditions to matrix, at this time matrix must have
2438: $ final nonzero structure (i.e. setting the nonlinear terms and applying
2439: $ boundary conditions again will not change the nonzero structure
2440: $ MatSetOption(mat,MAT_NO_NEW_NONZERO_LOCATIONS);
2441: $ MatStoreValues(mat);
2442: $ Call SNESSetJacobian() with matrix
2443: $ In your Jacobian routine
2444: $ MatRetrieveValues(mat);
2445: $ Set nonlinear terms in matrix
2446:
2447: Common Usage without SNESSolve(), i.e. when you handle nonlinear solve yourself:
2448: $ // build linear portion of Jacobian
2449: $ MatSetOption(mat,MAT_NO_NEW_NONZERO_LOCATIONS);
2450: $ MatStoreValues(mat);
2451: $ loop over nonlinear iterations
2452: $ MatRetrieveValues(mat);
2453: $ // call MatSetValues(mat,...) to set nonliner portion of Jacobian
2454: $ // call MatAssemblyBegin/End() on matrix
2455: $ Solve linear system with Jacobian
2456: $ endloop
2458: Notes:
2459: Matrix must already be assemblied before calling this routine
2460: Must set the matrix option MatSetOption(mat,MAT_NO_NEW_NONZERO_LOCATIONS); before
2461: calling this routine.
2463: When this is called multiple times it overwrites the previous set of stored values
2464: and does not allocated additional space.
2466: .seealso: MatRetrieveValues()
2468: @*/
2469: PetscErrorCode MatStoreValues(Mat mat)
2470: {
2471: PetscErrorCode ierr,(*f)(Mat);
2475: if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2476: if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2478: PetscObjectQueryFunction((PetscObject)mat,"MatStoreValues_C",(void (**)(void))&f);
2479: if (f) {
2480: (*f)(mat);
2481: } else {
2482: SETERRQ(PETSC_ERR_SUP,"Wrong type of matrix to store values");
2483: }
2484: return(0);
2485: }
2490: PetscErrorCode MatRetrieveValues_SeqAIJ(Mat mat)
2491: {
2492: Mat_SeqAIJ *aij = (Mat_SeqAIJ *)mat->data;
2494: PetscInt nz = aij->i[mat->rmap.n];
2497: if (aij->nonew != 1) {
2498: SETERRQ(PETSC_ERR_ORDER,"Must call MatSetOption(A,MAT_NO_NEW_NONZERO_LOCATIONS);first");
2499: }
2500: if (!aij->saved_values) {
2501: SETERRQ(PETSC_ERR_ORDER,"Must call MatStoreValues(A);first");
2502: }
2503: /* copy values over */
2504: PetscMemcpy(aij->a,aij->saved_values,nz*sizeof(PetscScalar));
2505: return(0);
2506: }
2511: /*@
2512: MatRetrieveValues - Retrieves the copy of the matrix values; this allows, for
2513: example, reuse of the linear part of a Jacobian, while recomputing the
2514: nonlinear portion.
2516: Collect on Mat
2518: Input Parameters:
2519: . mat - the matrix (currently on AIJ matrices support this option)
2521: Level: advanced
2523: .seealso: MatStoreValues()
2525: @*/
2526: PetscErrorCode MatRetrieveValues(Mat mat)
2527: {
2528: PetscErrorCode ierr,(*f)(Mat);
2532: if (!mat->assembled) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for unassembled matrix");
2533: if (mat->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
2535: PetscObjectQueryFunction((PetscObject)mat,"MatRetrieveValues_C",(void (**)(void))&f);
2536: if (f) {
2537: (*f)(mat);
2538: } else {
2539: SETERRQ(PETSC_ERR_SUP,"Wrong type of matrix to retrieve values");
2540: }
2541: return(0);
2542: }
2545: /* --------------------------------------------------------------------------------*/
2548: /*@C
2549: MatCreateSeqAIJ - Creates a sparse matrix in AIJ (compressed row) format
2550: (the default parallel PETSc format). For good matrix assembly performance
2551: the user should preallocate the matrix storage by setting the parameter nz
2552: (or the array nnz). By setting these parameters accurately, performance
2553: during matrix assembly can be increased by more than a factor of 50.
2555: Collective on MPI_Comm
2557: Input Parameters:
2558: + comm - MPI communicator, set to PETSC_COMM_SELF
2559: . m - number of rows
2560: . n - number of columns
2561: . nz - number of nonzeros per row (same for all rows)
2562: - nnz - array containing the number of nonzeros in the various rows
2563: (possibly different for each row) or PETSC_NULL
2565: Output Parameter:
2566: . A - the matrix
2568: Notes:
2569: If nnz is given then nz is ignored
2571: The AIJ format (also called the Yale sparse matrix format or
2572: compressed row storage), is fully compatible with standard Fortran 77
2573: storage. That is, the stored row and column indices can begin at
2574: either one (as in Fortran) or zero. See the users' manual for details.
2576: Specify the preallocated storage with either nz or nnz (not both).
2577: Set nz=PETSC_DEFAULT and nnz=PETSC_NULL for PETSc to control dynamic memory
2578: allocation. For large problems you MUST preallocate memory or you
2579: will get TERRIBLE performance, see the users' manual chapter on matrices.
2581: By default, this format uses inodes (identical nodes) when possible, to
2582: improve numerical efficiency of matrix-vector products and solves. We
2583: search for consecutive rows with the same nonzero structure, thereby
2584: reusing matrix information to achieve increased efficiency.
2586: Options Database Keys:
2587: + -mat_no_inode - Do not use inodes
2588: . -mat_inode_limit <limit> - Sets inode limit (max limit=5)
2589: - -mat_aij_oneindex - Internally use indexing starting at 1
2590: rather than 0. Note that when calling MatSetValues(),
2591: the user still MUST index entries starting at 0!
2593: Level: intermediate
2595: .seealso: MatCreate(), MatCreateMPIAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays()
2597: @*/
2598: PetscErrorCode MatCreateSeqAIJ(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt nz,const PetscInt nnz[],Mat *A)
2599: {
2603: MatCreate(comm,A);
2604: MatSetSizes(*A,m,n,m,n);
2605: MatSetType(*A,MATSEQAIJ);
2606: MatSeqAIJSetPreallocation_SeqAIJ(*A,nz,(PetscInt*)nnz);
2607: return(0);
2608: }
2612: /*@C
2613: MatSeqAIJSetPreallocation - For good matrix assembly performance
2614: the user should preallocate the matrix storage by setting the parameter nz
2615: (or the array nnz). By setting these parameters accurately, performance
2616: during matrix assembly can be increased by more than a factor of 50.
2618: Collective on MPI_Comm
2620: Input Parameters:
2621: + B - The matrix
2622: . nz - number of nonzeros per row (same for all rows)
2623: - nnz - array containing the number of nonzeros in the various rows
2624: (possibly different for each row) or PETSC_NULL
2626: Notes:
2627: If nnz is given then nz is ignored
2629: The AIJ format (also called the Yale sparse matrix format or
2630: compressed row storage), is fully compatible with standard Fortran 77
2631: storage. That is, the stored row and column indices can begin at
2632: either one (as in Fortran) or zero. See the users' manual for details.
2634: Specify the preallocated storage with either nz or nnz (not both).
2635: Set nz=PETSC_DEFAULT and nnz=PETSC_NULL for PETSc to control dynamic memory
2636: allocation. For large problems you MUST preallocate memory or you
2637: will get TERRIBLE performance, see the users' manual chapter on matrices.
2639: Developers: Use nz of MAT_SKIP_ALLOCATION to not allocate any space for the matrix
2640: entries or columns indices
2642: By default, this format uses inodes (identical nodes) when possible, to
2643: improve numerical efficiency of matrix-vector products and solves. We
2644: search for consecutive rows with the same nonzero structure, thereby
2645: reusing matrix information to achieve increased efficiency.
2647: Options Database Keys:
2648: + -mat_no_inode - Do not use inodes
2649: . -mat_inode_limit <limit> - Sets inode limit (max limit=5)
2650: - -mat_aij_oneindex - Internally use indexing starting at 1
2651: rather than 0. Note that when calling MatSetValues(),
2652: the user still MUST index entries starting at 0!
2654: Level: intermediate
2656: .seealso: MatCreate(), MatCreateMPIAIJ(), MatSetValues(), MatSeqAIJSetColumnIndices(), MatCreateSeqAIJWithArrays()
2658: @*/
2659: PetscErrorCode MatSeqAIJSetPreallocation(Mat B,PetscInt nz,const PetscInt nnz[])
2660: {
2661: PetscErrorCode ierr,(*f)(Mat,PetscInt,const PetscInt[]);
2664: PetscObjectQueryFunction((PetscObject)B,"MatSeqAIJSetPreallocation_C",(void (**)(void))&f);
2665: if (f) {
2666: (*f)(B,nz,nnz);
2667: }
2668: return(0);
2669: }
2674: PetscErrorCode MatSeqAIJSetPreallocation_SeqAIJ(Mat B,PetscInt nz,PetscInt *nnz)
2675: {
2676: Mat_SeqAIJ *b;
2677: PetscTruth skipallocation = PETSC_FALSE;
2679: PetscInt i;
2682:
2683: if (nz == MAT_SKIP_ALLOCATION) {
2684: skipallocation = PETSC_TRUE;
2685: nz = 0;
2686: }
2688: B->rmap.bs = B->cmap.bs = 1;
2689: PetscMapInitialize(B->comm,&B->rmap);
2690: PetscMapInitialize(B->comm,&B->cmap);
2692: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 5;
2693: if (nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"nz cannot be less than 0: value %d",nz);
2694: if (nnz) {
2695: for (i=0; i<B->rmap.n; i++) {
2696: if (nnz[i] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be less than 0: local row %d value %d",i,nnz[i]);
2697: if (nnz[i] > B->cmap.n) SETERRQ3(PETSC_ERR_ARG_OUTOFRANGE,"nnz cannot be greater than row length: local row %d value %d rowlength %d",i,nnz[i],B->cmap.n);
2698: }
2699: }
2701: B->preallocated = PETSC_TRUE;
2702: b = (Mat_SeqAIJ*)B->data;
2704: if (!skipallocation) {
2705: PetscMalloc2(B->rmap.n,PetscInt,&b->imax,B->rmap.n,PetscInt,&b->ilen);
2706: if (!nnz) {
2707: if (nz == PETSC_DEFAULT || nz == PETSC_DECIDE) nz = 10;
2708: else if (nz <= 0) nz = 1;
2709: for (i=0; i<B->rmap.n; i++) b->imax[i] = nz;
2710: nz = nz*B->rmap.n;
2711: } else {
2712: nz = 0;
2713: for (i=0; i<B->rmap.n; i++) {b->imax[i] = nnz[i]; nz += nnz[i];}
2714: }
2716: /* b->ilen will count nonzeros in each row so far. */
2717: for (i=0; i<B->rmap.n; i++) { b->ilen[i] = 0;}
2719: /* allocate the matrix space */
2720: PetscMalloc3(nz,PetscScalar,&b->a,nz,PetscInt,&b->j,B->rmap.n+1,PetscInt,&b->i);
2721: b->i[0] = 0;
2722: for (i=1; i<B->rmap.n+1; i++) {
2723: b->i[i] = b->i[i-1] + b->imax[i-1];
2724: }
2725: b->singlemalloc = PETSC_TRUE;
2726: b->free_a = PETSC_TRUE;
2727: b->free_ij = PETSC_TRUE;
2728: } else {
2729: b->free_a = PETSC_FALSE;
2730: b->free_ij = PETSC_FALSE;
2731: }
2733: b->nz = 0;
2734: b->maxnz = nz;
2735: B->info.nz_unneeded = (double)b->maxnz;
2736: return(0);
2737: }
2740: #undef __FUNCT__
2742: /*@C
2743: MatSeqAIJSetPreallocationCSR - Allocates memory for a sparse sequential matrix in AIJ format.
2745: Input Parameters:
2746: + B - the matrix
2747: . i - the indices into j for the start of each row (starts with zero)
2748: . j - the column indices for each row (starts with zero) these must be sorted for each row
2749: - v - optional values in the matrix
2751: Contributed by: Lisandro Dalchin
2753: Level: developer
2755: .keywords: matrix, aij, compressed row, sparse, sequential
2757: .seealso: MatCreate(), MatCreateSeqAIJ(), MatSetValues(), MatSeqAIJSetPreallocation(), MatCreateSeqAIJ(), SeqAIJ
2758: @*/
2759: PetscErrorCode MatSeqAIJSetPreallocationCSR(Mat B,const PetscInt i[],const PetscInt j[],const PetscScalar v[])
2760: {
2761: PetscErrorCode (*f)(Mat,const PetscInt[],const PetscInt[],const PetscScalar[]);
2766: PetscObjectQueryFunction((PetscObject)B,"MatSeqAIJSetPreallocationCSR_C",(void (**)(void))&f);
2767: if (f) {
2768: (*f)(B,i,j,v);
2769: }
2770: return(0);
2771: }
2774: #undef __FUNCT__
2776: PetscErrorCode MatSeqAIJSetPreallocationCSR_SeqAIJ(Mat B,const PetscInt Ii[],const PetscInt J[],const PetscScalar v[])
2777: {
2778: PetscInt i;
2779: PetscInt m,n;
2780: PetscInt nz;
2781: PetscInt *nnz, nz_max = 0;
2782: PetscScalar *values;
2786: MatGetSize(B, &m, &n);
2788: if (Ii[0]) {
2789: SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE, "Ii[0] must be 0 it is %D", Ii[0]);
2790: }
2791: PetscMalloc((m+1) * sizeof(PetscInt), &nnz);
2792: for(i = 0; i < m; i++) {
2793: nz = Ii[i+1]- Ii[i];
2794: nz_max = PetscMax(nz_max, nz);
2795: if (nz < 0) {
2796: SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE, "Local row %D has a negative number of columns %D", i, nnz);
2797: }
2798: nnz[i] = nz;
2799: }
2800: MatSeqAIJSetPreallocation(B, 0, nnz);
2801: PetscFree(nnz);
2803: if (v) {
2804: values = (PetscScalar*) v;
2805: } else {
2806: PetscMalloc((nz_max+1)*sizeof(PetscScalar), &values);
2807: PetscMemzero(values, nz_max*sizeof(PetscScalar));
2808: }
2810: MatSetOption(B,MAT_COLUMNS_SORTED);
2812: for(i = 0; i < m; i++) {
2813: nz = Ii[i+1] - Ii[i];
2814: MatSetValues_SeqAIJ(B, 1, &i, nz, J+Ii[i], values + (v ? Ii[i] : 0), INSERT_VALUES);
2815: }
2817: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
2818: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
2819: MatSetOption(B,MAT_COLUMNS_UNSORTED);
2821: if (!v) {
2822: PetscFree(values);
2823: }
2824: return(0);
2825: }
2828: /*MC
2829: MATSEQAIJ - MATSEQAIJ = "seqaij" - A matrix type to be used for sequential sparse matrices,
2830: based on compressed sparse row format.
2832: Options Database Keys:
2833: . -mat_type seqaij - sets the matrix type to "seqaij" during a call to MatSetFromOptions()
2835: Level: beginner
2837: .seealso: MatCreateSeqAIJ(), MatSetFromOptions(), MatSetType(), MatCreate(), MatType
2838: M*/
2847: PetscErrorCode MatCreate_SeqAIJ(Mat B)
2848: {
2849: Mat_SeqAIJ *b;
2851: PetscMPIInt size;
2854: MPI_Comm_size(B->comm,&size);
2855: if (size > 1) SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"Comm must be of size 1");
2857: PetscNew(Mat_SeqAIJ,&b);
2858: B->data = (void*)b;
2859: PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));
2860: B->factor = 0;
2861: B->mapping = 0;
2862: b->row = 0;
2863: b->col = 0;
2864: b->icol = 0;
2865: b->reallocs = 0;
2866: b->sorted = PETSC_FALSE;
2867: b->ignorezeroentries = PETSC_FALSE;
2868: b->roworiented = PETSC_TRUE;
2869: b->nonew = 0;
2870: b->diag = 0;
2871: b->solve_work = 0;
2872: B->spptr = 0;
2873: b->saved_values = 0;
2874: b->idiag = 0;
2875: b->ssor = 0;
2876: b->keepzeroedrows = PETSC_FALSE;
2877: b->xtoy = 0;
2878: b->XtoY = 0;
2879: b->compressedrow.use = PETSC_FALSE;
2880: b->compressedrow.nrows = B->rmap.n;
2881: b->compressedrow.i = PETSC_NULL;
2882: b->compressedrow.rindex = PETSC_NULL;
2883: b->compressedrow.checked = PETSC_FALSE;
2884: B->same_nonzero = PETSC_FALSE;
2886: PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);
2887: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatSeqAIJSetColumnIndices_C",
2888: "MatSeqAIJSetColumnIndices_SeqAIJ",
2889: MatSeqAIJSetColumnIndices_SeqAIJ);
2890: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatStoreValues_C",
2891: "MatStoreValues_SeqAIJ",
2892: MatStoreValues_SeqAIJ);
2893: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatRetrieveValues_C",
2894: "MatRetrieveValues_SeqAIJ",
2895: MatRetrieveValues_SeqAIJ);
2896: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatConvert_seqaij_seqsbaij_C",
2897: "MatConvert_SeqAIJ_SeqSBAIJ",
2898: MatConvert_SeqAIJ_SeqSBAIJ);
2899: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatConvert_seqaij_seqbaij_C",
2900: "MatConvert_SeqAIJ_SeqBAIJ",
2901: MatConvert_SeqAIJ_SeqBAIJ);
2902: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatConvert_seqaij_seqcsrperm_C",
2903: "MatConvert_SeqAIJ_SeqCSRPERM",
2904: MatConvert_SeqAIJ_SeqCSRPERM);
2905: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatConvert_seqaij_seqcrl_C",
2906: "MatConvert_SeqAIJ_SeqCRL",
2907: MatConvert_SeqAIJ_SeqCRL);
2908: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatIsTranspose_C",
2909: "MatIsTranspose_SeqAIJ",
2910: MatIsTranspose_SeqAIJ);
2911: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatSeqAIJSetPreallocation_C",
2912: "MatSeqAIJSetPreallocation_SeqAIJ",
2913: MatSeqAIJSetPreallocation_SeqAIJ);
2914: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatSeqAIJSetPreallocationCSR_C",
2915: "MatSeqAIJSetPreallocationCSR_SeqAIJ",
2916: MatSeqAIJSetPreallocationCSR_SeqAIJ);
2917: PetscObjectComposeFunctionDynamic((PetscObject)B,"MatReorderForNonzeroDiagonal_C",
2918: "MatReorderForNonzeroDiagonal_SeqAIJ",
2919: MatReorderForNonzeroDiagonal_SeqAIJ);
2920: MatCreate_Inode(B);
2921: PetscObjectChangeTypeName((PetscObject)B,MATSEQAIJ);
2922: return(0);
2923: }
2928: PetscErrorCode MatDuplicate_SeqAIJ(Mat A,MatDuplicateOption cpvalues,Mat *B)
2929: {
2930: Mat C;
2931: Mat_SeqAIJ *c,*a = (Mat_SeqAIJ*)A->data;
2933: PetscInt i,m = A->rmap.n;
2936: *B = 0;
2937: MatCreate(A->comm,&C);
2938: MatSetSizes(C,A->rmap.n,A->cmap.n,A->rmap.n,A->cmap.n);
2939: MatSetType(C,A->type_name);
2940: PetscMemcpy(C->ops,A->ops,sizeof(struct _MatOps));
2941:
2942: PetscMapCopy(A->comm,&A->rmap,&C->rmap);
2943: PetscMapCopy(A->comm,&A->cmap,&C->cmap);
2945: c = (Mat_SeqAIJ*)C->data;
2947: C->factor = A->factor;
2949: c->row = 0;
2950: c->col = 0;
2951: c->icol = 0;
2952: c->reallocs = 0;
2954: C->assembled = PETSC_TRUE;
2955:
2956: PetscMalloc2(m,PetscInt,&c->imax,m,PetscInt,&c->ilen);
2957: for (i=0; i<m; i++) {
2958: c->imax[i] = a->imax[i];
2959: c->ilen[i] = a->ilen[i];
2960: }
2962: /* allocate the matrix space */
2963: PetscMalloc3(a->i[m],PetscScalar,&c->a,a->i[m],PetscInt,&c->j,m+1,PetscInt,&c->i);
2964: c->singlemalloc = PETSC_TRUE;
2965: PetscMemcpy(c->i,a->i,(m+1)*sizeof(PetscInt));
2966: if (m > 0) {
2967: PetscMemcpy(c->j,a->j,(a->i[m])*sizeof(PetscInt));
2968: if (cpvalues == MAT_COPY_VALUES) {
2969: PetscMemcpy(c->a,a->a,(a->i[m])*sizeof(PetscScalar));
2970: } else {
2971: PetscMemzero(c->a,(a->i[m])*sizeof(PetscScalar));
2972: }
2973: }
2975: c->sorted = a->sorted;
2976: c->ignorezeroentries = a->ignorezeroentries;
2977: c->roworiented = a->roworiented;
2978: c->nonew = a->nonew;
2979: if (a->diag) {
2980: PetscMalloc((m+1)*sizeof(PetscInt),&c->diag);
2981: PetscLogObjectMemory(C,(m+1)*sizeof(PetscInt));
2982: for (i=0; i<m; i++) {
2983: c->diag[i] = a->diag[i];
2984: }
2985: } else c->diag = 0;
2986: c->solve_work = 0;
2987: c->saved_values = 0;
2988: c->idiag = 0;
2989: c->ssor = 0;
2990: c->keepzeroedrows = a->keepzeroedrows;
2991: c->free_a = PETSC_TRUE;
2992: c->free_ij = PETSC_TRUE;
2993: c->xtoy = 0;
2994: c->XtoY = 0;
2996: c->nz = a->nz;
2997: c->maxnz = a->maxnz;
2998: C->preallocated = PETSC_TRUE;
3000: c->compressedrow.use = a->compressedrow.use;
3001: c->compressedrow.nrows = a->compressedrow.nrows;
3002: c->compressedrow.checked = a->compressedrow.checked;
3003: if ( a->compressedrow.checked && a->compressedrow.use){
3004: i = a->compressedrow.nrows;
3005: PetscMalloc((2*i+1)*sizeof(PetscInt),&c->compressedrow.i);
3006: c->compressedrow.rindex = c->compressedrow.i + i + 1;
3007: PetscMemcpy(c->compressedrow.i,a->compressedrow.i,(i+1)*sizeof(PetscInt));
3008: PetscMemcpy(c->compressedrow.rindex,a->compressedrow.rindex,i*sizeof(PetscInt));
3009: } else {
3010: c->compressedrow.use = PETSC_FALSE;
3011: c->compressedrow.i = PETSC_NULL;
3012: c->compressedrow.rindex = PETSC_NULL;
3013: }
3014: C->same_nonzero = A->same_nonzero;
3015: MatDuplicate_Inode(A,cpvalues,&C);
3017: *B = C;
3018: PetscFListDuplicate(A->qlist,&C->qlist);
3019: return(0);
3020: }
3024: PetscErrorCode MatLoad_SeqAIJ(PetscViewer viewer, MatType type,Mat *A)
3025: {
3026: Mat_SeqAIJ *a;
3027: Mat B;
3029: PetscInt i,sum,nz,header[4],*rowlengths = 0,M,N;
3030: int fd;
3031: PetscMPIInt size;
3032: MPI_Comm comm;
3033:
3035: PetscObjectGetComm((PetscObject)viewer,&comm);
3036: MPI_Comm_size(comm,&size);
3037: if (size > 1) SETERRQ(PETSC_ERR_ARG_SIZ,"view must have one processor");
3038: PetscViewerBinaryGetDescriptor(viewer,&fd);
3039: PetscBinaryRead(fd,header,4,PETSC_INT);
3040: if (header[0] != MAT_FILE_COOKIE) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"not matrix object in file");
3041: M = header[1]; N = header[2]; nz = header[3];
3043: if (nz < 0) {
3044: SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format on disk,cannot load as SeqAIJ");
3045: }
3047: /* read in row lengths */
3048: PetscMalloc(M*sizeof(PetscInt),&rowlengths);
3049: PetscBinaryRead(fd,rowlengths,M,PETSC_INT);
3051: /* check if sum of rowlengths is same as nz */
3052: for (i=0,sum=0; i< M; i++) sum +=rowlengths[i];
3053: if (sum != nz) SETERRQ2(PETSC_ERR_FILE_READ,"Inconsistant matrix data in file. no-nonzeros = %d, sum-row-lengths = %d\n",nz,sum);
3055: /* create our matrix */
3056: MatCreate(comm,&B);
3057: MatSetSizes(B,PETSC_DECIDE,PETSC_DECIDE,M,N);
3058: MatSetType(B,type);
3059: MatSeqAIJSetPreallocation_SeqAIJ(B,0,rowlengths);
3060: a = (Mat_SeqAIJ*)B->data;
3062: /* read in column indices and adjust for Fortran indexing*/
3063: PetscBinaryRead(fd,a->j,nz,PETSC_INT);
3065: /* read in nonzero values */
3066: PetscBinaryRead(fd,a->a,nz,PETSC_SCALAR);
3068: /* set matrix "i" values */
3069: a->i[0] = 0;
3070: for (i=1; i<= M; i++) {
3071: a->i[i] = a->i[i-1] + rowlengths[i-1];
3072: a->ilen[i-1] = rowlengths[i-1];
3073: }
3074: PetscFree(rowlengths);
3076: MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);
3077: MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);
3078: *A = B;
3079: return(0);
3080: }
3084: PetscErrorCode MatEqual_SeqAIJ(Mat A,Mat B,PetscTruth* flg)
3085: {
3086: Mat_SeqAIJ *a = (Mat_SeqAIJ *)A->data,*b = (Mat_SeqAIJ *)B->data;
3090: /* If the matrix dimensions are not equal,or no of nonzeros */
3091: if ((A->rmap.n != B->rmap.n) || (A->cmap.n != B->cmap.n) ||(a->nz != b->nz)) {
3092: *flg = PETSC_FALSE;
3093: return(0);
3094: }
3095:
3096: /* if the a->i are the same */
3097: PetscMemcmp(a->i,b->i,(A->rmap.n+1)*sizeof(PetscInt),flg);
3098: if (!*flg) return(0);
3099:
3100: /* if a->j are the same */
3101: PetscMemcmp(a->j,b->j,(a->nz)*sizeof(PetscInt),flg);
3102: if (!*flg) return(0);
3103:
3104: /* if a->a are the same */
3105: PetscMemcmp(a->a,b->a,(a->nz)*sizeof(PetscScalar),flg);
3107: return(0);
3108:
3109: }
3113: /*@
3114: MatCreateSeqAIJWithArrays - Creates an sequential AIJ matrix using matrix elements (in CSR format)
3115: provided by the user.
3117: Collective on MPI_Comm
3119: Input Parameters:
3120: + comm - must be an MPI communicator of size 1
3121: . m - number of rows
3122: . n - number of columns
3123: . i - row indices
3124: . j - column indices
3125: - a - matrix values
3127: Output Parameter:
3128: . mat - the matrix
3130: Level: intermediate
3132: Notes:
3133: The i, j, and a arrays are not copied by this routine, the user must free these arrays
3134: once the matrix is destroyed
3136: You cannot set new nonzero locations into this matrix, that will generate an error.
3138: The i and j indices are 0 based
3140: .seealso: MatCreate(), MatCreateMPIAIJ(), MatCreateSeqAIJ(), MatCreateMPIAIJWithArrays(), MatMPIAIJSetPreallocationCSR()
3142: @*/
3143: PetscErrorCode MatCreateSeqAIJWithArrays(MPI_Comm comm,PetscInt m,PetscInt n,PetscInt* i,PetscInt*j,PetscScalar *a,Mat *mat)
3144: {
3146: PetscInt ii;
3147: Mat_SeqAIJ *aij;
3150: if (i[0]) {
3151: SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"i (row indices) must start with 0");
3152: }
3153: MatCreate(comm,mat);
3154: MatSetSizes(*mat,m,n,m,n);
3155: MatSetType(*mat,MATSEQAIJ);
3156: MatSeqAIJSetPreallocation_SeqAIJ(*mat,MAT_SKIP_ALLOCATION,0);
3157: aij = (Mat_SeqAIJ*)(*mat)->data;
3158: PetscMalloc2(m,PetscInt,&aij->imax,m,PetscInt,&aij->ilen);
3160: aij->i = i;
3161: aij->j = j;
3162: aij->a = a;
3163: aij->singlemalloc = PETSC_FALSE;
3164: aij->nonew = -1; /*this indicates that inserting a new value in the matrix that generates a new nonzero is an error*/
3165: aij->free_a = PETSC_FALSE;
3166: aij->free_ij = PETSC_FALSE;
3168: for (ii=0; ii<m; ii++) {
3169: aij->ilen[ii] = aij->imax[ii] = i[ii+1] - i[ii];
3170: #if defined(PETSC_USE_DEBUG)
3171: if (i[ii+1] - i[ii] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Negative row length in i (row indices) row = %d length = %d",ii,i[ii+1] - i[ii]);
3172: #endif
3173: }
3174: #if defined(PETSC_USE_DEBUG)
3175: for (ii=0; ii<aij->i[m]; ii++) {
3176: if (j[ii] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Negative column index at location = %d index = %d",ii,j[ii]);
3177: if (j[ii] > n - 1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column index to large at location = %d index = %d",ii,j[ii]);
3178: }
3179: #endif
3181: MatAssemblyBegin(*mat,MAT_FINAL_ASSEMBLY);
3182: MatAssemblyEnd(*mat,MAT_FINAL_ASSEMBLY);
3183: return(0);
3184: }
3188: PetscErrorCode MatSetColoring_SeqAIJ(Mat A,ISColoring coloring)
3189: {
3191: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3194: if (coloring->ctype == IS_COLORING_LOCAL) {
3195: ISColoringReference(coloring);
3196: a->coloring = coloring;
3197: } else if (coloring->ctype == IS_COLORING_GHOSTED) {
3198: PetscInt i,*larray;
3199: ISColoring ocoloring;
3200: ISColoringValue *colors;
3202: /* set coloring for diagonal portion */
3203: PetscMalloc((A->cmap.n+1)*sizeof(PetscInt),&larray);
3204: for (i=0; i<A->cmap.n; i++) {
3205: larray[i] = i;
3206: }
3207: ISGlobalToLocalMappingApply(A->mapping,IS_GTOLM_MASK,A->cmap.n,larray,PETSC_NULL,larray);
3208: PetscMalloc((A->cmap.n+1)*sizeof(ISColoringValue),&colors);
3209: for (i=0; i<A->cmap.n; i++) {
3210: colors[i] = coloring->colors[larray[i]];
3211: }
3212: PetscFree(larray);
3213: ISColoringCreate(PETSC_COMM_SELF,coloring->n,A->cmap.n,colors,&ocoloring);
3214: a->coloring = ocoloring;
3215: }
3216: return(0);
3217: }
3219: #if defined(PETSC_HAVE_ADIC)
3221: #include "adic/ad_utils.h"
3226: PetscErrorCode MatSetValuesAdic_SeqAIJ(Mat A,void *advalues)
3227: {
3228: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3229: PetscInt m = A->rmap.n,*ii = a->i,*jj = a->j,nz,i,j,nlen;
3230: PetscScalar *v = a->a,*values = ((PetscScalar*)advalues)+1;
3231: ISColoringValue *color;
3234: if (!a->coloring) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Coloring not set for matrix");
3235: nlen = PetscADGetDerivTypeSize()/sizeof(PetscScalar);
3236: color = a->coloring->colors;
3237: /* loop over rows */
3238: for (i=0; i<m; i++) {
3239: nz = ii[i+1] - ii[i];
3240: /* loop over columns putting computed value into matrix */
3241: for (j=0; j<nz; j++) {
3242: *v++ = values[color[*jj++]];
3243: }
3244: values += nlen; /* jump to next row of derivatives */
3245: }
3246: return(0);
3247: }
3248: #endif
3252: PetscErrorCode MatSetValuesAdifor_SeqAIJ(Mat A,PetscInt nl,void *advalues)
3253: {
3254: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3255: PetscInt m = A->rmap.n,*ii = a->i,*jj = a->j,nz,i,j;
3256: PetscScalar *v = a->a,*values = (PetscScalar *)advalues;
3257: ISColoringValue *color;
3260: if (!a->coloring) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Coloring not set for matrix");
3261: color = a->coloring->colors;
3262: /* loop over rows */
3263: for (i=0; i<m; i++) {
3264: nz = ii[i+1] - ii[i];
3265: /* loop over columns putting computed value into matrix */
3266: for (j=0; j<nz; j++) {
3267: *v++ = values[color[*jj++]];
3268: }
3269: values += nl; /* jump to next row of derivatives */
3270: }
3271: return(0);
3272: }
3274: /*
3275: Special version for direct calls from Fortran
3276: */
3277: #if defined(PETSC_HAVE_FORTRAN_CAPS)
3278: #define matsetvaluesseqaij_ MATSETVALUESSEQAIJ
3279: #elif !defined(PETSC_HAVE_FORTRAN_UNDERSCORE)
3280: #define matsetvaluesseqaij_ matsetvaluesseqaij
3281: #endif
3283: /* Change these macros so can be used in void function */
3284: #undef CHKERRQ
3285: #define CHKERRQ(ierr) CHKERRABORT(A->comm,ierr)
3286: #undef SETERRQ2
3287: #define SETERRQ2(ierr,b,c,d) CHKERRABORT(A->comm,ierr)
3292: void PETSC_STDCALL matsetvaluesseqaij_(Mat *AA,PetscInt *mm,const PetscInt im[],PetscInt *nn,const PetscInt in[],const PetscScalar v[],InsertMode *isis, PetscErrorCode *_ierr)
3293: {
3294: Mat A = *AA;
3295: PetscInt m = *mm, n = *nn;
3296: InsertMode is = *isis;
3297: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
3298: PetscInt *rp,k,low,high,t,ii,row,nrow,i,col,l,rmax,N;
3299: PetscInt *imax,*ai,*ailen;
3301: PetscInt *aj,nonew = a->nonew,lastcol = -1;
3302: PetscScalar *ap,value,*aa;
3303: PetscTruth ignorezeroentries = ((a->ignorezeroentries && is == ADD_VALUES) ? PETSC_TRUE:PETSC_FALSE);
3304: PetscTruth roworiented = a->roworiented;
3307: MatPreallocated(A);
3308: imax = a->imax;
3309: ai = a->i;
3310: ailen = a->ilen;
3311: aj = a->j;
3312: aa = a->a;
3314: for (k=0; k<m; k++) { /* loop over added rows */
3315: row = im[k];
3316: if (row < 0) continue;
3317: #if defined(PETSC_USE_DEBUG)
3318: if (row >= A->rmap.n) SETERRABORT(A->comm,PETSC_ERR_ARG_OUTOFRANGE,"Row too large");
3319: #endif
3320: rp = aj + ai[row]; ap = aa + ai[row];
3321: rmax = imax[row]; nrow = ailen[row];
3322: low = 0;
3323: high = nrow;
3324: for (l=0; l<n; l++) { /* loop over added columns */
3325: if (in[l] < 0) continue;
3326: #if defined(PETSC_USE_DEBUG)
3327: if (in[l] >= A->cmap.n) SETERRABORT(A->comm,PETSC_ERR_ARG_OUTOFRANGE,"Column too large");
3328: #endif
3329: col = in[l];
3330: if (roworiented) {
3331: value = v[l + k*n];
3332: } else {
3333: value = v[k + l*m];
3334: }
3335: if (value == 0.0 && ignorezeroentries && (is == ADD_VALUES)) continue;
3337: if (col <= lastcol) low = 0; else high = nrow;
3338: lastcol = col;
3339: while (high-low > 5) {
3340: t = (low+high)/2;
3341: if (rp[t] > col) high = t;
3342: else low = t;
3343: }
3344: for (i=low; i<high; i++) {
3345: if (rp[i] > col) break;
3346: if (rp[i] == col) {
3347: if (is == ADD_VALUES) ap[i] += value;
3348: else ap[i] = value;
3349: goto noinsert;
3350: }
3351: }
3352: if (value == 0.0 && ignorezeroentries) goto noinsert;
3353: if (nonew == 1) goto noinsert;
3354: if (nonew == -1) SETERRABORT(A->comm,PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero in the matrix");
3355: MatSeqXAIJReallocateAIJ(A,A->rmap.n,1,nrow,row,col,rmax,aa,ai,aj,rp,ap,imax,nonew);
3356: N = nrow++ - 1; a->nz++; high++;
3357: /* shift up all the later entries in this row */
3358: for (ii=N; ii>=i; ii--) {
3359: rp[ii+1] = rp[ii];
3360: ap[ii+1] = ap[ii];
3361: }
3362: rp[i] = col;
3363: ap[i] = value;
3364: noinsert:;
3365: low = i + 1;
3366: }
3367: ailen[row] = nrow;
3368: }
3369: A->same_nonzero = PETSC_FALSE;
3370: PetscFunctionReturnVoid();
3371: }