Actual source code: sbaij2.c
1: /*$Id: sbaij2.c,v 1.32 2001/08/07 03:03:01 balay Exp $*/
3: #include src/mat/impls/baij/seq/baij.h
4: #include src/vec/vecimpl.h
5: #include src/inline/spops.h
6: #include src/inline/ilu.h
7: #include petscbt.h
8: #include src/mat/impls/sbaij/seq/sbaij.h
10: int MatIncreaseOverlap_SeqSBAIJ(Mat A,int is_max,IS *is,int ov)
11: {
13: SETERRQ(1,"Function not yet written for SBAIJ format");
14: /* return(0); */
15: }
17: int MatGetSubMatrix_SeqSBAIJ_Private(Mat A,IS isrow,IS iscol,int cs,MatReuse scall,Mat *B)
18: {
19: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data,*c;
20: int *smap,i,k,kstart,kend,ierr,oldcols = a->mbs,*lens;
21: int row,mat_i,*mat_j,tcol,*mat_ilen;
22: int *irow,nrows,*ssmap,bs=a->bs,bs2=a->bs2;
23: int *aj = a->j,*ai = a->i;
24: MatScalar *mat_a;
25: Mat C;
26: PetscTruth flag;
29:
30: if (isrow != iscol) SETERRQ(1,"MatGetSubmatrices_SeqSBAIJ: For symm. format, iscol must equal isro");
31: ISSorted(iscol,(PetscTruth*)&i);
32: if (!i) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"IS is not sorted");
34: ISGetIndices(isrow,&irow);
35: ISGetSize(isrow,&nrows);
36:
37: ierr = PetscMalloc((1+oldcols)*sizeof(int),&smap);
38: ssmap = smap;
39: ierr = PetscMalloc((1+nrows)*sizeof(int),&lens);
40: ierr = PetscMemzero(smap,oldcols*sizeof(int));
41: for (i=0; i<nrows; i++) smap[irow[i]] = i+1; /* nrows = ncols */
42: /* determine lens of each row */
43: for (i=0; i<nrows; i++) {
44: kstart = ai[irow[i]];
45: kend = kstart + a->ilen[irow[i]];
46: lens[i] = 0;
47: for (k=kstart; k<kend; k++) {
48: if (ssmap[aj[k]]) {
49: lens[i]++;
50: }
51: }
52: }
53: /* Create and fill new matrix */
54: if (scall == MAT_REUSE_MATRIX) {
55: c = (Mat_SeqSBAIJ *)((*B)->data);
57: if (c->mbs!=nrows || c->bs!=bs) SETERRQ(PETSC_ERR_ARG_SIZ,"Submatrix wrong size");
58: PetscMemcmp(c->ilen,lens,c->mbs *sizeof(int),&flag);
59: if (flag == PETSC_FALSE) {
60: SETERRQ(PETSC_ERR_ARG_SIZ,"Cannot reuse matrix. wrong no of nonzeros");
61: }
62: PetscMemzero(c->ilen,c->mbs*sizeof(int));
63: C = *B;
64: } else {
65: MatCreateSeqSBAIJ(A->comm,bs,nrows*bs,nrows*bs,0,lens,&C);
66: }
67: c = (Mat_SeqSBAIJ *)(C->data);
68: for (i=0; i<nrows; i++) {
69: row = irow[i];
70: kstart = ai[row];
71: kend = kstart + a->ilen[row];
72: mat_i = c->i[i];
73: mat_j = c->j + mat_i;
74: mat_a = c->a + mat_i*bs2;
75: mat_ilen = c->ilen + i;
76: for (k=kstart; k<kend; k++) {
77: if ((tcol=ssmap[a->j[k]])) {
78: *mat_j++ = tcol - 1;
79: ierr = PetscMemcpy(mat_a,a->a+k*bs2,bs2*sizeof(MatScalar));
80: mat_a += bs2;
81: (*mat_ilen)++;
82: }
83: }
84: }
85:
86: /* Free work space */
87: PetscFree(smap);
88: PetscFree(lens);
89: MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);
90: MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);
91:
92: ISRestoreIndices(isrow,&irow);
93: *B = C;
94: return(0);
95: }
97: int MatGetSubMatrix_SeqSBAIJ(Mat A,IS isrow,IS iscol,int cs,MatReuse scall,Mat *B)
98: {
99: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
100: IS is1;
101: int *vary,*iary,*irow,nrows,i,ierr,bs=a->bs,count;
104: if (isrow != iscol) SETERRQ(1,"MatGetSubmatrices_SeqSBAIJ: For symm. format, iscol must equal isro");
105:
106: ISGetIndices(isrow,&irow);
107: ISGetSize(isrow,&nrows);
108:
109: /* Verify if the indices corespond to each element in a block
110: and form the IS with compressed IS */
111: PetscMalloc(2*(a->mbs+1)*sizeof(int),&vary);
112: iary = vary + a->mbs;
113: PetscMemzero(vary,(a->mbs)*sizeof(int));
114: for (i=0; i<nrows; i++) vary[irow[i]/bs]++;
115:
116: count = 0;
117: for (i=0; i<a->mbs; i++) {
118: if (vary[i]!=0 && vary[i]!=bs) SETERRQ(1,"Index set does not match blocks");
119: if (vary[i]==bs) iary[count++] = i;
120: }
121: ISCreateGeneral(PETSC_COMM_SELF,count,iary,&is1);
122:
123: ISRestoreIndices(isrow,&irow);
124: PetscFree(vary);
126: MatGetSubMatrix_SeqSBAIJ_Private(A,is1,is1,cs,scall,B);
127: ISDestroy(is1);
128: return(0);
129: }
131: int MatGetSubMatrices_SeqSBAIJ(Mat A,int n,IS *irow,IS *icol,MatReuse scall,Mat **B)
132: {
133: int ierr,i;
136: if (scall == MAT_INITIAL_MATRIX) {
137: PetscMalloc((n+1)*sizeof(Mat),B);
138: }
140: for (i=0; i<n; i++) {
141: MatGetSubMatrix_SeqSBAIJ(A,irow[i],icol[i],PETSC_DECIDE,scall,&(*B)[i]);
142: }
143: return(0);
144: }
146: /* -------------------------------------------------------*/
147: /* Should check that shapes of vectors and matrices match */
148: /* -------------------------------------------------------*/
149: #include petscblaslapack.h
151: int MatMult_SeqSBAIJ_1(Mat A,Vec xx,Vec zz)
152: {
153: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
154: PetscScalar *x,*z,*xb,x1,zero=0.0;
155: MatScalar *v;
156: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
159: VecSet(&zero,zz);
160: VecGetArray(xx,&x);
161: VecGetArray(zz,&z);
163: v = a->a;
164: xb = x;
165:
166: for (i=0; i<mbs; i++) {
167: n = ai[1] - ai[0]; /* length of i_th row of A */
168: x1 = xb[0];
169: ib = aj + *ai;
170: jmin = 0;
171: if (*ib == i) { /* (diag of A)*x */
172: z[i] += *v++ * x[*ib++];
173: jmin++;
174: }
175: for (j=jmin; j<n; j++) {
176: cval = *ib;
177: z[cval] += *v * x1; /* (strict lower triangular part of A)*x */
178: z[i] += *v++ * x[*ib++]; /* (strict upper triangular part of A)*x */
179: }
180: xb++; ai++;
181: }
183: VecRestoreArray(xx,&x);
184: VecRestoreArray(zz,&z);
185: PetscLogFlops(2*(a->s_nz*2 - A->m) - A->m); /* s_nz = (nz+m)/2 */
186: return(0);
187: }
189: int MatMult_SeqSBAIJ_2(Mat A,Vec xx,Vec zz)
190: {
191: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
192: PetscScalar *x,*z,*xb,x1,x2,zero=0.0;
193: MatScalar *v;
194: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
198: VecSet(&zero,zz);
199: VecGetArray(xx,&x);
200: VecGetArray(zz,&z);
201:
202: v = a->a;
203: xb = x;
205: for (i=0; i<mbs; i++) {
206: n = ai[1] - ai[0]; /* length of i_th block row of A */
207: x1 = xb[0]; x2 = xb[1];
208: ib = aj + *ai;
209: jmin = 0;
210: if (*ib == i){ /* (diag of A)*x */
211: z[2*i] += v[0]*x1 + v[2]*x2;
212: z[2*i+1] += v[2]*x1 + v[3]*x2;
213: v += 4; jmin++;
214: }
215: for (j=jmin; j<n; j++) {
216: /* (strict lower triangular part of A)*x */
217: cval = ib[j]*2;
218: z[cval] += v[0]*x1 + v[1]*x2;
219: z[cval+1] += v[2]*x1 + v[3]*x2;
220: /* (strict upper triangular part of A)*x */
221: z[2*i] += v[0]*x[cval] + v[2]*x[cval+1];
222: z[2*i+1] += v[1]*x[cval] + v[3]*x[cval+1];
223: v += 4;
224: }
225: xb +=2; ai++;
226: }
228: VecRestoreArray(xx,&x);
229: VecRestoreArray(zz,&z);
230: PetscLogFlops(8*(a->s_nz*2 - A->m) - A->m);
231: return(0);
232: }
234: int MatMult_SeqSBAIJ_3(Mat A,Vec xx,Vec zz)
235: {
236: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
237: PetscScalar *x,*z,*xb,x1,x2,x3,zero=0.0;
238: MatScalar *v;
239: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
243: VecSet(&zero,zz);
244: VecGetArray(xx,&x);
245: VecGetArray(zz,&z);
246:
247: v = a->a;
248: xb = x;
250: for (i=0; i<mbs; i++) {
251: n = ai[1] - ai[0]; /* length of i_th block row of A */
252: x1 = xb[0]; x2 = xb[1]; x3 = xb[2];
253: ib = aj + *ai;
254: jmin = 0;
255: if (*ib == i){ /* (diag of A)*x */
256: z[3*i] += v[0]*x1 + v[3]*x2 + v[6]*x3;
257: z[3*i+1] += v[3]*x1 + v[4]*x2 + v[7]*x3;
258: z[3*i+2] += v[6]*x1 + v[7]*x2 + v[8]*x3;
259: v += 9; jmin++;
260: }
261: for (j=jmin; j<n; j++) {
262: /* (strict lower triangular part of A)*x */
263: cval = ib[j]*3;
264: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3;
265: z[cval+1] += v[3]*x1 + v[4]*x2 + v[5]*x3;
266: z[cval+2] += v[6]*x1 + v[7]*x2 + v[8]*x3;
267: /* (strict upper triangular part of A)*x */
268: z[3*i] += v[0]*x[cval] + v[3]*x[cval+1]+ v[6]*x[cval+2];
269: z[3*i+1] += v[1]*x[cval] + v[4]*x[cval+1]+ v[7]*x[cval+2];
270: z[3*i+2] += v[2]*x[cval] + v[5]*x[cval+1]+ v[8]*x[cval+2];
271: v += 9;
272: }
273: xb +=3; ai++;
274: }
276: VecRestoreArray(xx,&x);
277: VecRestoreArray(zz,&z);
278: PetscLogFlops(18*(a->s_nz*2 - A->m) - A->m);
279: return(0);
280: }
282: int MatMult_SeqSBAIJ_4(Mat A,Vec xx,Vec zz)
283: {
284: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
285: PetscScalar *x,*z,*xb,x1,x2,x3,x4,zero=0.0;
286: MatScalar *v;
287: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
290: VecSet(&zero,zz);
291: VecGetArray(xx,&x);
292: VecGetArray(zz,&z);
293:
294: v = a->a;
295: xb = x;
297: for (i=0; i<mbs; i++) {
298: n = ai[1] - ai[0]; /* length of i_th block row of A */
299: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3];
300: ib = aj + *ai;
301: jmin = 0;
302: if (*ib == i){ /* (diag of A)*x */
303: z[4*i] += v[0]*x1 + v[4]*x2 + v[8]*x3 + v[12]*x4;
304: z[4*i+1] += v[4]*x1 + v[5]*x2 + v[9]*x3 + v[13]*x4;
305: z[4*i+2] += v[8]*x1 + v[9]*x2 + v[10]*x3 + v[14]*x4;
306: z[4*i+3] += v[12]*x1+ v[13]*x2+ v[14]*x3 + v[15]*x4;
307: v += 16; jmin++;
308: }
309: for (j=jmin; j<n; j++) {
310: /* (strict lower triangular part of A)*x */
311: cval = ib[j]*4;
312: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4;
313: z[cval+1] += v[4]*x1 + v[5]*x2 + v[6]*x3 + v[7]*x4;
314: z[cval+2] += v[8]*x1 + v[9]*x2 + v[10]*x3 + v[11]*x4;
315: z[cval+3] += v[12]*x1 + v[13]*x2 + v[14]*x3 + v[15]*x4;
316: /* (strict upper triangular part of A)*x */
317: z[4*i] += v[0]*x[cval] + v[4]*x[cval+1]+ v[8]*x[cval+2] + v[12]*x[cval+3];
318: z[4*i+1] += v[1]*x[cval] + v[5]*x[cval+1]+ v[9]*x[cval+2] + v[13]*x[cval+3];
319: z[4*i+2] += v[2]*x[cval] + v[6]*x[cval+1]+ v[10]*x[cval+2]+ v[14]*x[cval+3];
320: z[4*i+3] += v[3]*x[cval] + v[7]*x[cval+1]+ v[11]*x[cval+2]+ v[15]*x[cval+3];
321: v += 16;
322: }
323: xb +=4; ai++;
324: }
326: VecRestoreArray(xx,&x);
327: VecRestoreArray(zz,&z);
328: PetscLogFlops(32*(a->s_nz*2 - A->m) - A->m);
329: return(0);
330: }
332: int MatMult_SeqSBAIJ_5(Mat A,Vec xx,Vec zz)
333: {
334: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
335: PetscScalar *x,*z,*xb,x1,x2,x3,x4,x5,zero=0.0;
336: MatScalar *v;
337: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
340: VecSet(&zero,zz);
341: VecGetArray(xx,&x);
342: VecGetArray(zz,&z);
343:
344: v = a->a;
345: xb = x;
347: for (i=0; i<mbs; i++) {
348: n = ai[1] - ai[0]; /* length of i_th block row of A */
349: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4];
350: ib = aj + *ai;
351: jmin = 0;
352: if (*ib == i){ /* (diag of A)*x */
353: z[5*i] += v[0]*x1 + v[5]*x2 + v[10]*x3 + v[15]*x4+ v[20]*x5;
354: z[5*i+1] += v[5]*x1 + v[6]*x2 + v[11]*x3 + v[16]*x4+ v[21]*x5;
355: z[5*i+2] += v[10]*x1 +v[11]*x2 + v[12]*x3 + v[17]*x4+ v[22]*x5;
356: z[5*i+3] += v[15]*x1 +v[16]*x2 + v[17]*x3 + v[18]*x4+ v[23]*x5;
357: z[5*i+4] += v[20]*x1 +v[21]*x2 + v[22]*x3 + v[23]*x4+ v[24]*x5;
358: v += 25; jmin++;
359: }
360: for (j=jmin; j<n; j++) {
361: /* (strict lower triangular part of A)*x */
362: cval = ib[j]*5;
363: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4 + v[4]*x5;
364: z[cval+1] += v[5]*x1 + v[6]*x2 + v[7]*x3 + v[8]*x4 + v[9]*x5;
365: z[cval+2] += v[10]*x1 + v[11]*x2 + v[12]*x3 + v[13]*x4+ v[14]*x5;
366: z[cval+3] += v[15]*x1 + v[16]*x2 + v[17]*x3 + v[18]*x4+ v[19]*x5;
367: z[cval+4] += v[20]*x1 + v[21]*x2 + v[22]*x3 + v[23]*x4+ v[24]*x5;
368: /* (strict upper triangular part of A)*x */
369: z[5*i] +=v[0]*x[cval]+v[5]*x[cval+1]+v[10]*x[cval+2]+v[15]*x[cval+3]+v[20]*x[cval+4];
370: z[5*i+1] +=v[1]*x[cval]+v[6]*x[cval+1]+v[11]*x[cval+2]+v[16]*x[cval+3]+v[21]*x[cval+4];
371: z[5*i+2] +=v[2]*x[cval]+v[7]*x[cval+1]+v[12]*x[cval+2]+v[17]*x[cval+3]+v[22]*x[cval+4];
372: z[5*i+3] +=v[3]*x[cval]+v[8]*x[cval+1]+v[13]*x[cval+2]+v[18]*x[cval+3]+v[23]*x[cval+4];
373: z[5*i+4] +=v[4]*x[cval]+v[9]*x[cval+1]+v[14]*x[cval+2]+v[19]*x[cval+3]+v[24]*x[cval+4];
374: v += 25;
375: }
376: xb +=5; ai++;
377: }
379: VecRestoreArray(xx,&x);
380: VecRestoreArray(zz,&z);
381: PetscLogFlops(50*(a->s_nz*2 - A->m) - A->m);
382: return(0);
383: }
386: int MatMult_SeqSBAIJ_6(Mat A,Vec xx,Vec zz)
387: {
388: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
389: PetscScalar *x,*z,*xb,x1,x2,x3,x4,x5,x6,zero=0.0;
390: MatScalar *v;
391: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
394: VecSet(&zero,zz);
395: VecGetArray(xx,&x);
396: VecGetArray(zz,&z);
397:
398: v = a->a;
399: xb = x;
401: for (i=0; i<mbs; i++) {
402: n = ai[1] - ai[0]; /* length of i_th block row of A */
403: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4]; x6=xb[5];
404: ib = aj + *ai;
405: jmin = 0;
406: if (*ib == i){ /* (diag of A)*x */
407: z[6*i] += v[0]*x1 + v[6]*x2 + v[12]*x3 + v[18]*x4+ v[24]*x5 + v[30]*x6;
408: z[6*i+1] += v[6]*x1 + v[7]*x2 + v[13]*x3 + v[19]*x4+ v[25]*x5 + v[31]*x6;
409: z[6*i+2] += v[12]*x1 +v[13]*x2 + v[14]*x3 + v[20]*x4+ v[26]*x5 + v[32]*x6;
410: z[6*i+3] += v[18]*x1 +v[19]*x2 + v[20]*x3 + v[21]*x4+ v[27]*x5 + v[33]*x6;
411: z[6*i+4] += v[24]*x1 +v[25]*x2 + v[26]*x3 + v[27]*x4+ v[28]*x5 + v[34]*x6;
412: z[6*i+5] += v[30]*x1 +v[31]*x2 + v[32]*x3 + v[33]*x4+ v[34]*x5 + v[35]*x6;
413: v += 36; jmin++;
414: }
415: for (j=jmin; j<n; j++) {
416: /* (strict lower triangular part of A)*x */
417: cval = ib[j]*6;
418: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4+ v[4]*x5 + v[5]*x6;
419: z[cval+1] += v[6]*x1 + v[7]*x2 + v[8]*x3 + v[9]*x4+ v[10]*x5 + v[11]*x6;
420: z[cval+2] += v[12]*x1 + v[13]*x2 + v[14]*x3 + v[15]*x4+ v[16]*x5 + v[17]*x6;
421: z[cval+3] += v[18]*x1 + v[19]*x2 + v[20]*x3 + v[21]*x4+ v[22]*x5 + v[23]*x6;
422: z[cval+4] += v[24]*x1 + v[25]*x2 + v[26]*x3 + v[27]*x4+ v[28]*x5 + v[29]*x6;
423: z[cval+5] += v[30]*x1 + v[31]*x2 + v[32]*x3 + v[33]*x4+ v[34]*x5 + v[35]*x6;
424: /* (strict upper triangular part of A)*x */
425: z[6*i] +=v[0]*x[cval]+v[6]*x[cval+1]+v[12]*x[cval+2]+v[18]*x[cval+3]+v[24]*x[cval+4]+v[30]*x[cval+5];
426: z[6*i+1] +=v[1]*x[cval]+v[7]*x[cval+1]+v[13]*x[cval+2]+v[19]*x[cval+3]+v[25]*x[cval+4]+v[31]*x[cval+5];
427: z[6*i+2] +=v[2]*x[cval]+v[8]*x[cval+1]+v[14]*x[cval+2]+v[20]*x[cval+3]+v[26]*x[cval+4]+v[32]*x[cval+5];
428: z[6*i+3] +=v[3]*x[cval]+v[9]*x[cval+1]+v[15]*x[cval+2]+v[21]*x[cval+3]+v[27]*x[cval+4]+v[33]*x[cval+5];
429: z[6*i+4] +=v[4]*x[cval]+v[10]*x[cval+1]+v[16]*x[cval+2]+v[22]*x[cval+3]+v[28]*x[cval+4]+v[34]*x[cval+5];
430: z[6*i+5] +=v[5]*x[cval]+v[11]*x[cval+1]+v[17]*x[cval+2]+v[23]*x[cval+3]+v[29]*x[cval+4]+v[35]*x[cval+5];
431: v += 36;
432: }
433: xb +=6; ai++;
434: }
436: VecRestoreArray(xx,&x);
437: VecRestoreArray(zz,&z);
438: PetscLogFlops(72*(a->s_nz*2 - A->m) - A->m);
439: return(0);
440: }
441: int MatMult_SeqSBAIJ_7(Mat A,Vec xx,Vec zz)
442: {
443: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
444: PetscScalar *x,*z,*xb,x1,x2,x3,x4,x5,x6,x7,zero=0.0;
445: MatScalar *v;
446: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
449: VecSet(&zero,zz);
450: VecGetArray(xx,&x);
451: VecGetArray(zz,&z);
452:
453: v = a->a;
454: xb = x;
456: for (i=0; i<mbs; i++) {
457: n = ai[1] - ai[0]; /* length of i_th block row of A */
458: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4]; x6=xb[5]; x7=xb[6];
459: ib = aj + *ai;
460: jmin = 0;
461: if (*ib == i){ /* (diag of A)*x */
462: z[7*i] += v[0]*x1 + v[7]*x2 + v[14]*x3 + v[21]*x4+ v[28]*x5 + v[35]*x6+ v[42]*x7;
463: z[7*i+1] += v[7]*x1 + v[8]*x2 + v[15]*x3 + v[22]*x4+ v[29]*x5 + v[36]*x6+ v[43]*x7;
464: z[7*i+2] += v[14]*x1+ v[15]*x2 +v[16]*x3 + v[23]*x4+ v[30]*x5 + v[37]*x6+ v[44]*x7;
465: z[7*i+3] += v[21]*x1+ v[22]*x2 +v[23]*x3 + v[24]*x4+ v[31]*x5 + v[38]*x6+ v[45]*x7;
466: z[7*i+4] += v[28]*x1+ v[29]*x2 +v[30]*x3 + v[31]*x4+ v[32]*x5 + v[39]*x6+ v[46]*x7;
467: z[7*i+5] += v[35]*x1+ v[36]*x2 +v[37]*x3 + v[38]*x4+ v[39]*x5 + v[40]*x6+ v[47]*x7;
468: z[7*i+6] += v[42]*x1+ v[43]*x2 +v[44]*x3 + v[45]*x4+ v[46]*x5 + v[47]*x6+ v[48]*x7;
469: v += 49; jmin++;
470: }
471: for (j=jmin; j<n; j++) {
472: /* (strict lower triangular part of A)*x */
473: cval = ib[j]*7;
474: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4+ v[4]*x5 + v[5]*x6+ v[6]*x7;
475: z[cval+1] += v[7]*x1 + v[8]*x2 + v[9]*x3 + v[10]*x4+ v[11]*x5 + v[12]*x6+ v[13]*x7;
476: z[cval+2] += v[14]*x1 + v[15]*x2 + v[16]*x3 + v[17]*x4+ v[18]*x5 + v[19]*x6+ v[20]*x7;
477: z[cval+3] += v[21]*x1 + v[22]*x2 + v[23]*x3 + v[24]*x4+ v[25]*x5 + v[26]*x6+ v[27]*x7;
478: z[cval+4] += v[28]*x1 + v[29]*x2 + v[30]*x3 + v[31]*x4+ v[32]*x5 + v[33]*x6+ v[34]*x7;
479: z[cval+5] += v[35]*x1 + v[36]*x2 + v[37]*x3 + v[38]*x4+ v[39]*x5 + v[40]*x6+ v[41]*x7;
480: z[cval+6] += v[42]*x1 + v[43]*x2 + v[44]*x3 + v[45]*x4+ v[46]*x5 + v[47]*x6+ v[48]*x7;
481: /* (strict upper triangular part of A)*x */
482: z[7*i] +=v[0]*x[cval]+v[7]*x[cval+1]+v[14]*x[cval+2]+v[21]*x[cval+3]+v[28]*x[cval+4]+v[35]*x[cval+5]+v[42]*x[cval+6];
483: z[7*i+1]+=v[1]*x[cval]+v[8]*x[cval+1]+v[15]*x[cval+2]+v[22]*x[cval+3]+v[29]*x[cval+4]+v[36]*x[cval+5]+v[43]*x[cval+6];
484: z[7*i+2]+=v[2]*x[cval]+v[9]*x[cval+1]+v[16]*x[cval+2]+v[23]*x[cval+3]+v[30]*x[cval+4]+v[37]*x[cval+5]+v[44]*x[cval+6];
485: z[7*i+3]+=v[3]*x[cval]+v[10]*x[cval+1]+v[17]*x[cval+2]+v[24]*x[cval+3]+v[31]*x[cval+4]+v[38]*x[cval+5]+v[45]*x[cval+6];
486: z[7*i+4]+=v[4]*x[cval]+v[11]*x[cval+1]+v[18]*x[cval+2]+v[25]*x[cval+3]+v[32]*x[cval+4]+v[39]*x[cval+5]+v[46]*x[cval+6];
487: z[7*i+5]+=v[5]*x[cval]+v[12]*x[cval+1]+v[19]*x[cval+2]+v[26]*x[cval+3]+v[33]*x[cval+4]+v[40]*x[cval+5]+v[47]*x[cval+6];
488: z[7*i+6]+=v[6]*x[cval]+v[13]*x[cval+1]+v[20]*x[cval+2]+v[27]*x[cval+3]+v[34]*x[cval+4]+v[41]*x[cval+5]+v[48]*x[cval+6];
489: v += 49;
490: }
491: xb +=7; ai++;
492: }
493: VecRestoreArray(xx,&x);
494: VecRestoreArray(zz,&z);
495: PetscLogFlops(98*(a->s_nz*2 - A->m) - A->m);
496: return(0);
497: }
499: /*
500: This will not work with MatScalar == float because it calls the BLAS
501: */
502: int MatMult_SeqSBAIJ_N(Mat A,Vec xx,Vec zz)
503: {
504: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
505: PetscScalar *x,*x_ptr,*z,*z_ptr,*xb,*zb,*work,*workt,zero=0.0;
506: MatScalar *v;
507: int ierr,mbs=a->mbs,i,*idx,*aj,*ii,bs=a->bs,j,n,bs2=a->bs2;
508: int ncols,k;
511: VecSet(&zero,zz);
512: VecGetArray(xx,&x); x_ptr=x;
513: VecGetArray(zz,&z); z_ptr=z;
515: aj = a->j;
516: v = a->a;
517: ii = a->i;
519: if (!a->mult_work) {
520: PetscMalloc((A->m+1)*sizeof(PetscScalar),&a->mult_work);
521: }
522: work = a->mult_work;
523:
524: for (i=0; i<mbs; i++) {
525: n = ii[1] - ii[0]; ncols = n*bs;
526: workt = work; idx=aj+ii[0];
528: /* upper triangular part */
529: for (j=0; j<n; j++) {
530: xb = x_ptr + bs*(*idx++);
531: for (k=0; k<bs; k++) workt[k] = xb[k];
532: workt += bs;
533: }
534: /* z(i*bs:(i+1)*bs-1) += A(i,:)*x */
535: Kernel_w_gets_w_plus_Ar_times_v(bs,ncols,work,v,z);
536:
537: /* strict lower triangular part */
538: idx = aj+ii[0];
539: if (*idx == i){
540: ncols -= bs; v += bs2; idx++; n--;
541: }
542:
543: if (ncols > 0){
544: workt = work;
545: ierr = PetscMemzero(workt,ncols*sizeof(PetscScalar));
546: Kernel_w_gets_w_plus_trans_Ar_times_v(bs,ncols,x,v,workt);
547: for (j=0; j<n; j++) {
548: zb = z_ptr + bs*(*idx++);
549: for (k=0; k<bs; k++) zb[k] += workt[k] ;
550: workt += bs;
551: }
552: }
553: x += bs; v += n*bs2; z += bs; ii++;
554: }
555:
556: VecRestoreArray(xx,&x);
557: VecRestoreArray(zz,&z);
558: PetscLogFlops(2*(a->s_nz*2 - A->m)*bs2 - A->m);
559: return(0);
560: }
562: int MatMultAdd_SeqSBAIJ_1(Mat A,Vec xx,Vec yy,Vec zz)
563: {
564: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
565: PetscScalar *x,*y,*z,*xb,x1;
566: MatScalar *v;
567: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
570: VecGetArray(xx,&x);
571: if (yy != xx) {
572: VecGetArray(yy,&y);
573: } else {
574: y = x;
575: }
576: if (zz != yy) {
577: VecCopy(yy,zz);
578: VecGetArray(zz,&z);
579: } else {
580: z = y;
581: }
583: v = a->a;
584: xb = x;
586: for (i=0; i<mbs; i++) {
587: n = ai[1] - ai[0]; /* length of i_th row of A */
588: x1 = xb[0];
589: ib = aj + *ai;
590: jmin = 0;
591: if (*ib == i) { /* (diag of A)*x */
592: z[i] += *v++ * x[*ib++]; jmin++;
593: }
594: for (j=jmin; j<n; j++) {
595: cval = *ib;
596: z[cval] += *v * x1; /* (strict lower triangular part of A)*x */
597: z[i] += *v++ * x[*ib++]; /* (strict upper triangular part of A)*x */
598: }
599: xb++; ai++;
600: }
602: VecRestoreArray(xx,&x);
603: if (yy != xx) VecRestoreArray(yy,&y);
604: if (zz != yy) VecRestoreArray(zz,&z);
605:
606: PetscLogFlops(2*(a->s_nz*2 - A->m));
607: return(0);
608: }
610: int MatMultAdd_SeqSBAIJ_2(Mat A,Vec xx,Vec yy,Vec zz)
611: {
612: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
613: PetscScalar *x,*y,*z,*xb,x1,x2;
614: MatScalar *v;
615: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
618: VecGetArray(xx,&x);
619: if (yy != xx) {
620: VecGetArray(yy,&y);
621: } else {
622: y = x;
623: }
624: if (zz != yy) {
625: VecCopy(yy,zz);
626: VecGetArray(zz,&z);
627: } else {
628: z = y;
629: }
631: v = a->a;
632: xb = x;
634: for (i=0; i<mbs; i++) {
635: n = ai[1] - ai[0]; /* length of i_th block row of A */
636: x1 = xb[0]; x2 = xb[1];
637: ib = aj + *ai;
638: jmin = 0;
639: if (*ib == i){ /* (diag of A)*x */
640: z[2*i] += v[0]*x1 + v[2]*x2;
641: z[2*i+1] += v[2]*x1 + v[3]*x2;
642: v += 4; jmin++;
643: }
644: for (j=jmin; j<n; j++) {
645: /* (strict lower triangular part of A)*x */
646: cval = ib[j]*2;
647: z[cval] += v[0]*x1 + v[1]*x2;
648: z[cval+1] += v[2]*x1 + v[3]*x2;
649: /* (strict upper triangular part of A)*x */
650: z[2*i] += v[0]*x[cval] + v[2]*x[cval+1];
651: z[2*i+1] += v[1]*x[cval] + v[3]*x[cval+1];
652: v += 4;
653: }
654: xb +=2; ai++;
655: }
657: VecRestoreArray(xx,&x);
658: if (yy != xx) VecRestoreArray(yy,&y);
659: if (zz != yy) VecRestoreArray(zz,&z);
661: PetscLogFlops(4*(a->s_nz*2 - A->m));
662: return(0);
663: }
665: int MatMultAdd_SeqSBAIJ_3(Mat A,Vec xx,Vec yy,Vec zz)
666: {
667: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
668: PetscScalar *x,*y,*z,*xb,x1,x2,x3;
669: MatScalar *v;
670: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
673: VecGetArray(xx,&x);
674: if (yy != xx) {
675: VecGetArray(yy,&y);
676: } else {
677: y = x;
678: }
679: if (zz != yy) {
680: VecCopy(yy,zz);
681: VecGetArray(zz,&z);
682: } else {
683: z = y;
684: }
686: v = a->a;
687: xb = x;
689: for (i=0; i<mbs; i++) {
690: n = ai[1] - ai[0]; /* length of i_th block row of A */
691: x1 = xb[0]; x2 = xb[1]; x3 = xb[2];
692: ib = aj + *ai;
693: jmin = 0;
694: if (*ib == i){ /* (diag of A)*x */
695: z[3*i] += v[0]*x1 + v[3]*x2 + v[6]*x3;
696: z[3*i+1] += v[3]*x1 + v[4]*x2 + v[7]*x3;
697: z[3*i+2] += v[6]*x1 + v[7]*x2 + v[8]*x3;
698: v += 9; jmin++;
699: }
700: for (j=jmin; j<n; j++) {
701: /* (strict lower triangular part of A)*x */
702: cval = ib[j]*3;
703: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3;
704: z[cval+1] += v[3]*x1 + v[4]*x2 + v[5]*x3;
705: z[cval+2] += v[6]*x1 + v[7]*x2 + v[8]*x3;
706: /* (strict upper triangular part of A)*x */
707: z[3*i] += v[0]*x[cval] + v[3]*x[cval+1]+ v[6]*x[cval+2];
708: z[3*i+1] += v[1]*x[cval] + v[4]*x[cval+1]+ v[7]*x[cval+2];
709: z[3*i+2] += v[2]*x[cval] + v[5]*x[cval+1]+ v[8]*x[cval+2];
710: v += 9;
711: }
712: xb +=3; ai++;
713: }
715: VecRestoreArray(xx,&x);
716: if (yy != xx) VecRestoreArray(yy,&y);
717: if (zz != yy) VecRestoreArray(zz,&z);
719: PetscLogFlops(18*(a->s_nz*2 - A->m));
720: return(0);
721: }
723: int MatMultAdd_SeqSBAIJ_4(Mat A,Vec xx,Vec yy,Vec zz)
724: {
725: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
726: PetscScalar *x,*y,*z,*xb,x1,x2,x3,x4;
727: MatScalar *v;
728: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
731: VecGetArray(xx,&x);
732: if (yy != xx) {
733: VecGetArray(yy,&y);
734: } else {
735: y = x;
736: }
737: if (zz != yy) {
738: VecCopy(yy,zz);
739: VecGetArray(zz,&z);
740: } else {
741: z = y;
742: }
744: v = a->a;
745: xb = x;
747: for (i=0; i<mbs; i++) {
748: n = ai[1] - ai[0]; /* length of i_th block row of A */
749: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3];
750: ib = aj + *ai;
751: jmin = 0;
752: if (*ib == i){ /* (diag of A)*x */
753: z[4*i] += v[0]*x1 + v[4]*x2 + v[8]*x3 + v[12]*x4;
754: z[4*i+1] += v[4]*x1 + v[5]*x2 + v[9]*x3 + v[13]*x4;
755: z[4*i+2] += v[8]*x1 + v[9]*x2 + v[10]*x3 + v[14]*x4;
756: z[4*i+3] += v[12]*x1+ v[13]*x2+ v[14]*x3 + v[15]*x4;
757: v += 16; jmin++;
758: }
759: for (j=jmin; j<n; j++) {
760: /* (strict lower triangular part of A)*x */
761: cval = ib[j]*4;
762: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4;
763: z[cval+1] += v[4]*x1 + v[5]*x2 + v[6]*x3 + v[7]*x4;
764: z[cval+2] += v[8]*x1 + v[9]*x2 + v[10]*x3 + v[11]*x4;
765: z[cval+3] += v[12]*x1 + v[13]*x2 + v[14]*x3 + v[15]*x4;
766: /* (strict upper triangular part of A)*x */
767: z[4*i] += v[0]*x[cval] + v[4]*x[cval+1]+ v[8]*x[cval+2] + v[12]*x[cval+3];
768: z[4*i+1] += v[1]*x[cval] + v[5]*x[cval+1]+ v[9]*x[cval+2] + v[13]*x[cval+3];
769: z[4*i+2] += v[2]*x[cval] + v[6]*x[cval+1]+ v[10]*x[cval+2]+ v[14]*x[cval+3];
770: z[4*i+3] += v[3]*x[cval] + v[7]*x[cval+1]+ v[11]*x[cval+2]+ v[15]*x[cval+3];
771: v += 16;
772: }
773: xb +=4; ai++;
774: }
776: VecRestoreArray(xx,&x);
777: if (yy != xx) VecRestoreArray(yy,&y);
778: if (zz != yy) VecRestoreArray(zz,&z);
780: PetscLogFlops(32*(a->s_nz*2 - A->m));
781: return(0);
782: }
784: int MatMultAdd_SeqSBAIJ_5(Mat A,Vec xx,Vec yy,Vec zz)
785: {
786: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
787: PetscScalar *x,*y,*z,*xb,x1,x2,x3,x4,x5;
788: MatScalar *v;
789: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
792: VecGetArray(xx,&x);
793: if (yy != xx) {
794: VecGetArray(yy,&y);
795: } else {
796: y = x;
797: }
798: if (zz != yy) {
799: VecCopy(yy,zz);
800: VecGetArray(zz,&z);
801: } else {
802: z = y;
803: }
805: v = a->a;
806: xb = x;
808: for (i=0; i<mbs; i++) {
809: n = ai[1] - ai[0]; /* length of i_th block row of A */
810: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4];
811: ib = aj + *ai;
812: jmin = 0;
813: if (*ib == i){ /* (diag of A)*x */
814: z[5*i] += v[0]*x1 + v[5]*x2 + v[10]*x3 + v[15]*x4+ v[20]*x5;
815: z[5*i+1] += v[5]*x1 + v[6]*x2 + v[11]*x3 + v[16]*x4+ v[21]*x5;
816: z[5*i+2] += v[10]*x1 +v[11]*x2 + v[12]*x3 + v[17]*x4+ v[22]*x5;
817: z[5*i+3] += v[15]*x1 +v[16]*x2 + v[17]*x3 + v[18]*x4+ v[23]*x5;
818: z[5*i+4] += v[20]*x1 +v[21]*x2 + v[22]*x3 + v[23]*x4+ v[24]*x5;
819: v += 25; jmin++;
820: }
821: for (j=jmin; j<n; j++) {
822: /* (strict lower triangular part of A)*x */
823: cval = ib[j]*5;
824: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4 + v[4]*x5;
825: z[cval+1] += v[5]*x1 + v[6]*x2 + v[7]*x3 + v[8]*x4 + v[9]*x5;
826: z[cval+2] += v[10]*x1 + v[11]*x2 + v[12]*x3 + v[13]*x4+ v[14]*x5;
827: z[cval+3] += v[15]*x1 + v[16]*x2 + v[17]*x3 + v[18]*x4+ v[19]*x5;
828: z[cval+4] += v[20]*x1 + v[21]*x2 + v[22]*x3 + v[23]*x4+ v[24]*x5;
829: /* (strict upper triangular part of A)*x */
830: z[5*i] +=v[0]*x[cval]+v[5]*x[cval+1]+v[10]*x[cval+2]+v[15]*x[cval+3]+v[20]*x[cval+4];
831: z[5*i+1] +=v[1]*x[cval]+v[6]*x[cval+1]+v[11]*x[cval+2]+v[16]*x[cval+3]+v[21]*x[cval+4];
832: z[5*i+2] +=v[2]*x[cval]+v[7]*x[cval+1]+v[12]*x[cval+2]+v[17]*x[cval+3]+v[22]*x[cval+4];
833: z[5*i+3] +=v[3]*x[cval]+v[8]*x[cval+1]+v[13]*x[cval+2]+v[18]*x[cval+3]+v[23]*x[cval+4];
834: z[5*i+4] +=v[4]*x[cval]+v[9]*x[cval+1]+v[14]*x[cval+2]+v[19]*x[cval+3]+v[24]*x[cval+4];
835: v += 25;
836: }
837: xb +=5; ai++;
838: }
840: VecRestoreArray(xx,&x);
841: if (yy != xx) VecRestoreArray(yy,&y);
842: if (zz != yy) VecRestoreArray(zz,&z);
844: PetscLogFlops(50*(a->s_nz*2 - A->m));
845: return(0);
846: }
847: int MatMultAdd_SeqSBAIJ_6(Mat A,Vec xx,Vec yy,Vec zz)
848: {
849: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
850: PetscScalar *x,*y,*z,*xb,x1,x2,x3,x4,x5,x6;
851: MatScalar *v;
852: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
855: VecGetArray(xx,&x);
856: if (yy != xx) {
857: VecGetArray(yy,&y);
858: } else {
859: y = x;
860: }
861: if (zz != yy) {
862: VecCopy(yy,zz);
863: VecGetArray(zz,&z);
864: } else {
865: z = y;
866: }
868: v = a->a;
869: xb = x;
871: for (i=0; i<mbs; i++) {
872: n = ai[1] - ai[0]; /* length of i_th block row of A */
873: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4]; x6=xb[5];
874: ib = aj + *ai;
875: jmin = 0;
876: if (*ib == i){ /* (diag of A)*x */
877: z[6*i] += v[0]*x1 + v[6]*x2 + v[12]*x3 + v[18]*x4+ v[24]*x5 + v[30]*x6;
878: z[6*i+1] += v[6]*x1 + v[7]*x2 + v[13]*x3 + v[19]*x4+ v[25]*x5 + v[31]*x6;
879: z[6*i+2] += v[12]*x1 +v[13]*x2 + v[14]*x3 + v[20]*x4+ v[26]*x5 + v[32]*x6;
880: z[6*i+3] += v[18]*x1 +v[19]*x2 + v[20]*x3 + v[21]*x4+ v[27]*x5 + v[33]*x6;
881: z[6*i+4] += v[24]*x1 +v[25]*x2 + v[26]*x3 + v[27]*x4+ v[28]*x5 + v[34]*x6;
882: z[6*i+5] += v[30]*x1 +v[31]*x2 + v[32]*x3 + v[33]*x4+ v[34]*x5 + v[35]*x6;
883: v += 36; jmin++;
884: }
885: for (j=jmin; j<n; j++) {
886: /* (strict lower triangular part of A)*x */
887: cval = ib[j]*6;
888: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4+ v[4]*x5 + v[5]*x6;
889: z[cval+1] += v[6]*x1 + v[7]*x2 + v[8]*x3 + v[9]*x4+ v[10]*x5 + v[11]*x6;
890: z[cval+2] += v[12]*x1 + v[13]*x2 + v[14]*x3 + v[15]*x4+ v[16]*x5 + v[17]*x6;
891: z[cval+3] += v[18]*x1 + v[19]*x2 + v[20]*x3 + v[21]*x4+ v[22]*x5 + v[23]*x6;
892: z[cval+4] += v[24]*x1 + v[25]*x2 + v[26]*x3 + v[27]*x4+ v[28]*x5 + v[29]*x6;
893: z[cval+5] += v[30]*x1 + v[31]*x2 + v[32]*x3 + v[33]*x4+ v[34]*x5 + v[35]*x6;
894: /* (strict upper triangular part of A)*x */
895: z[6*i] +=v[0]*x[cval]+v[6]*x[cval+1]+v[12]*x[cval+2]+v[18]*x[cval+3]+v[24]*x[cval+4]+v[30]*x[cval+5];
896: z[6*i+1] +=v[1]*x[cval]+v[7]*x[cval+1]+v[13]*x[cval+2]+v[19]*x[cval+3]+v[25]*x[cval+4]+v[31]*x[cval+5];
897: z[6*i+2] +=v[2]*x[cval]+v[8]*x[cval+1]+v[14]*x[cval+2]+v[20]*x[cval+3]+v[26]*x[cval+4]+v[32]*x[cval+5];
898: z[6*i+3] +=v[3]*x[cval]+v[9]*x[cval+1]+v[15]*x[cval+2]+v[21]*x[cval+3]+v[27]*x[cval+4]+v[33]*x[cval+5];
899: z[6*i+4] +=v[4]*x[cval]+v[10]*x[cval+1]+v[16]*x[cval+2]+v[22]*x[cval+3]+v[28]*x[cval+4]+v[34]*x[cval+5];
900: z[6*i+5] +=v[5]*x[cval]+v[11]*x[cval+1]+v[17]*x[cval+2]+v[23]*x[cval+3]+v[29]*x[cval+4]+v[35]*x[cval+5];
901: v += 36;
902: }
903: xb +=6; ai++;
904: }
906: VecRestoreArray(xx,&x);
907: if (yy != xx) VecRestoreArray(yy,&y);
908: if (zz != yy) VecRestoreArray(zz,&z);
910: PetscLogFlops(72*(a->s_nz*2 - A->m));
911: return(0);
912: }
914: int MatMultAdd_SeqSBAIJ_7(Mat A,Vec xx,Vec yy,Vec zz)
915: {
916: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
917: PetscScalar *x,*y,*z,*xb,x1,x2,x3,x4,x5,x6,x7;
918: MatScalar *v;
919: int mbs=a->mbs,i,*aj=a->j,*ai=a->i,n,ierr,*ib,cval,j,jmin;
922: VecGetArray(xx,&x);
923: if (yy != xx) {
924: VecGetArray(yy,&y);
925: } else {
926: y = x;
927: }
928: if (zz != yy) {
929: VecCopy(yy,zz);
930: VecGetArray(zz,&z);
931: } else {
932: z = y;
933: }
935: v = a->a;
936: xb = x;
938: for (i=0; i<mbs; i++) {
939: n = ai[1] - ai[0]; /* length of i_th block row of A */
940: x1 = xb[0]; x2 = xb[1]; x3 = xb[2]; x4 = xb[3]; x5=xb[4]; x6=xb[5]; x7=xb[6];
941: ib = aj + *ai;
942: jmin = 0;
943: if (*ib == i){ /* (diag of A)*x */
944: z[7*i] += v[0]*x1 + v[7]*x2 + v[14]*x3 + v[21]*x4+ v[28]*x5 + v[35]*x6+ v[42]*x7;
945: z[7*i+1] += v[7]*x1 + v[8]*x2 + v[15]*x3 + v[22]*x4+ v[29]*x5 + v[36]*x6+ v[43]*x7;
946: z[7*i+2] += v[14]*x1+ v[15]*x2 +v[16]*x3 + v[23]*x4+ v[30]*x5 + v[37]*x6+ v[44]*x7;
947: z[7*i+3] += v[21]*x1+ v[22]*x2 +v[23]*x3 + v[24]*x4+ v[31]*x5 + v[38]*x6+ v[45]*x7;
948: z[7*i+4] += v[28]*x1+ v[29]*x2 +v[30]*x3 + v[31]*x4+ v[32]*x5 + v[39]*x6+ v[46]*x7;
949: z[7*i+5] += v[35]*x1+ v[36]*x2 +v[37]*x3 + v[38]*x4+ v[39]*x5 + v[40]*x6+ v[47]*x7;
950: z[7*i+6] += v[42]*x1+ v[43]*x2 +v[44]*x3 + v[45]*x4+ v[46]*x5 + v[47]*x6+ v[48]*x7;
951: v += 49; jmin++;
952: }
953: for (j=jmin; j<n; j++) {
954: /* (strict lower triangular part of A)*x */
955: cval = ib[j]*7;
956: z[cval] += v[0]*x1 + v[1]*x2 + v[2]*x3 + v[3]*x4+ v[4]*x5 + v[5]*x6+ v[6]*x7;
957: z[cval+1] += v[7]*x1 + v[8]*x2 + v[9]*x3 + v[10]*x4+ v[11]*x5 + v[12]*x6+ v[13]*x7;
958: z[cval+2] += v[14]*x1 + v[15]*x2 + v[16]*x3 + v[17]*x4+ v[18]*x5 + v[19]*x6+ v[20]*x7;
959: z[cval+3] += v[21]*x1 + v[22]*x2 + v[23]*x3 + v[24]*x4+ v[25]*x5 + v[26]*x6+ v[27]*x7;
960: z[cval+4] += v[28]*x1 + v[29]*x2 + v[30]*x3 + v[31]*x4+ v[32]*x5 + v[33]*x6+ v[34]*x7;
961: z[cval+5] += v[35]*x1 + v[36]*x2 + v[37]*x3 + v[38]*x4+ v[39]*x5 + v[40]*x6+ v[41]*x7;
962: z[cval+6] += v[42]*x1 + v[43]*x2 + v[44]*x3 + v[45]*x4+ v[46]*x5 + v[47]*x6+ v[48]*x7;
963: /* (strict upper triangular part of A)*x */
964: z[7*i] +=v[0]*x[cval]+v[7]*x[cval+1]+v[14]*x[cval+2]+v[21]*x[cval+3]+v[28]*x[cval+4]+v[35]*x[cval+5]+v[42]*x[cval+6];
965: z[7*i+1]+=v[1]*x[cval]+v[8]*x[cval+1]+v[15]*x[cval+2]+v[22]*x[cval+3]+v[29]*x[cval+4]+v[36]*x[cval+5]+v[43]*x[cval+6];
966: z[7*i+2]+=v[2]*x[cval]+v[9]*x[cval+1]+v[16]*x[cval+2]+v[23]*x[cval+3]+v[30]*x[cval+4]+v[37]*x[cval+5]+v[44]*x[cval+6];
967: z[7*i+3]+=v[3]*x[cval]+v[10]*x[cval+1]+v[17]*x[cval+2]+v[24]*x[cval+3]+v[31]*x[cval+4]+v[38]*x[cval+5]+v[45]*x[cval+6];
968: z[7*i+4]+=v[4]*x[cval]+v[11]*x[cval+1]+v[18]*x[cval+2]+v[25]*x[cval+3]+v[32]*x[cval+4]+v[39]*x[cval+5]+v[46]*x[cval+6];
969: z[7*i+5]+=v[5]*x[cval]+v[12]*x[cval+1]+v[19]*x[cval+2]+v[26]*x[cval+3]+v[33]*x[cval+4]+v[40]*x[cval+5]+v[47]*x[cval+6];
970: z[7*i+6]+=v[6]*x[cval]+v[13]*x[cval+1]+v[20]*x[cval+2]+v[27]*x[cval+3]+v[34]*x[cval+4]+v[41]*x[cval+5]+v[48]*x[cval+6];
971: v += 49;
972: }
973: xb +=7; ai++;
974: }
976: VecRestoreArray(xx,&x);
977: if (yy != xx) VecRestoreArray(yy,&y);
978: if (zz != yy) VecRestoreArray(zz,&z);
980: PetscLogFlops(98*(a->s_nz*2 - A->m));
981: return(0);
982: }
984: int MatMultAdd_SeqSBAIJ_N(Mat A,Vec xx,Vec yy,Vec zz)
985: {
986: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
987: PetscScalar *x,*x_ptr,*y,*z,*z_ptr=0,*xb,*zb,*work,*workt;
988: MatScalar *v;
989: int ierr,mbs=a->mbs,i,*idx,*aj,*ii,bs=a->bs,j,n,bs2=a->bs2;
990: int ncols,k;
993: VecGetArray(xx,&x); x_ptr=x;
994: if (yy != xx) {
995: VecGetArray(yy,&y);
996: } else {
997: y = x;
998: }
999: if (zz != yy) {
1000: VecCopy(yy,zz);
1001: VecGetArray(zz,&z); z_ptr=z;
1002: } else {
1003: z = y;
1004: }
1006: aj = a->j;
1007: v = a->a;
1008: ii = a->i;
1010: if (!a->mult_work) {
1011: PetscMalloc((A->m+1)*sizeof(PetscScalar),&a->mult_work);
1012: }
1013: work = a->mult_work;
1014:
1015:
1016: for (i=0; i<mbs; i++) {
1017: n = ii[1] - ii[0]; ncols = n*bs;
1018: workt = work; idx=aj+ii[0];
1020: /* upper triangular part */
1021: for (j=0; j<n; j++) {
1022: xb = x_ptr + bs*(*idx++);
1023: for (k=0; k<bs; k++) workt[k] = xb[k];
1024: workt += bs;
1025: }
1026: /* z(i*bs:(i+1)*bs-1) += A(i,:)*x */
1027: Kernel_w_gets_w_plus_Ar_times_v(bs,ncols,work,v,z);
1029: /* strict lower triangular part */
1030: idx = aj+ii[0];
1031: if (*idx == i){
1032: ncols -= bs; v += bs2; idx++; n--;
1033: }
1034: if (ncols > 0){
1035: workt = work;
1036: ierr = PetscMemzero(workt,ncols*sizeof(PetscScalar));
1037: Kernel_w_gets_w_plus_trans_Ar_times_v(bs,ncols,x,v,workt);
1038: for (j=0; j<n; j++) {
1039: zb = z_ptr + bs*(*idx++);
1040: /* idx++; */
1041: for (k=0; k<bs; k++) zb[k] += workt[k] ;
1042: workt += bs;
1043: }
1044: }
1046: x += bs; v += n*bs2; z += bs; ii++;
1047: }
1049: VecRestoreArray(xx,&x);
1050: if (yy != xx) VecRestoreArray(yy,&y);
1051: if (zz != yy) VecRestoreArray(zz,&z);
1053: PetscLogFlops(2*(a->s_nz*2 - A->m));
1054: return(0);
1055: }
1057: int MatMultTranspose_SeqSBAIJ(Mat A,Vec xx,Vec zz)
1058: {
1060: SETERRQ(1,"Matrix is symmetric. Call MatMult().");
1061: /* return(0); */
1062: }
1064: int MatMultTransposeAdd_SeqSBAIJ(Mat A,Vec xx,Vec yy,Vec zz)
1066: {
1068: SETERRQ(1,"Matrix is symmetric. Call MatMultAdd().");
1069: /* return(0); */
1070: }
1072: int MatScale_SeqSBAIJ(PetscScalar *alpha,Mat inA)
1073: {
1074: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)inA->data;
1075: int one = 1,totalnz = a->bs2*a->s_nz;
1078: BLscal_(&totalnz,alpha,a->a,&one);
1079: PetscLogFlops(totalnz);
1080: return(0);
1081: }
1083: int MatNorm_SeqSBAIJ(Mat A,NormType type,PetscReal *norm)
1084: {
1085: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1086: MatScalar *v = a->a;
1087: PetscReal sum_diag = 0.0, sum_off = 0.0, *sum;
1088: int i,j,k,bs = a->bs,bs2=a->bs2,k1,mbs=a->mbs,*aj=a->j;
1089: int *jl,*il,jmin,jmax,ierr,nexti,ik,*col;
1090:
1092: if (type == NORM_FROBENIUS) {
1093: for (k=0; k<mbs; k++){
1094: jmin = a->i[k]; jmax = a->i[k+1];
1095: col = aj + jmin;
1096: if (*col == k){ /* diagonal block */
1097: for (i=0; i<bs2; i++){
1098: #if defined(PETSC_USE_COMPLEX)
1099: sum_diag += PetscRealPart(PetscConj(*v)*(*v)); v++;
1100: #else
1101: sum_diag += (*v)*(*v); v++;
1102: #endif
1103: }
1104: jmin++;
1105: }
1106: for (j=jmin; j<jmax; j++){ /* off-diagonal blocks */
1107: for (i=0; i<bs2; i++){
1108: #if defined(PETSC_USE_COMPLEX)
1109: sum_off += PetscRealPart(PetscConj(*v)*(*v)); v++;
1110: #else
1111: sum_off += (*v)*(*v); v++;
1112: #endif
1113: }
1114: }
1115: }
1116: *norm = sqrt(sum_diag + 2*sum_off);
1118: } else if (type == NORM_INFINITY) { /* maximum row sum */
1119: PetscMalloc(mbs*sizeof(int),&il);
1120: PetscMalloc(mbs*sizeof(int),&jl);
1121: PetscMalloc(bs*sizeof(PetscReal),&sum);
1122: for (i=0; i<mbs; i++) {
1123: jl[i] = mbs; il[0] = 0;
1124: }
1126: *norm = 0.0;
1127: for (k=0; k<mbs; k++) { /* k_th block row */
1128: for (j=0; j<bs; j++) sum[j]=0.0;
1130: /*-- col sum --*/
1131: i = jl[k]; /* first |A(i,k)| to be added */
1132: /* jl[k]=i: first nozero element in row i for submatrix A(1:k,k:n) (active window)
1133: at step k */
1134: while (i<mbs){
1135: nexti = jl[i]; /* next block row to be added */
1136: ik = il[i]; /* block index of A(i,k) in the array a */
1137: for (j=0; j<bs; j++){
1138: v = a->a + ik*bs2 + j*bs;
1139: for (k1=0; k1<bs; k1++) {
1140: sum[j] += PetscAbsScalar(*v); v++;
1141: }
1142: }
1143: /* update il, jl */
1144: jmin = ik + 1; /* block index of array a: points to the next nonzero of A in row i */
1145: jmax = a->i[i+1];
1146: if (jmin < jmax){
1147: il[i] = jmin;
1148: j = a->j[jmin];
1149: jl[i] = jl[j]; jl[j]=i;
1150: }
1151: i = nexti;
1152: }
1153:
1154: /*-- row sum --*/
1155: jmin = a->i[k]; jmax = a->i[k+1];
1156: for (i=jmin; i<jmax; i++) {
1157: for (j=0; j<bs; j++){
1158: v = a->a + i*bs2 + j;
1159: for (k1=0; k1<bs; k1++){
1160: sum[j] += PetscAbsScalar(*v);
1161: v += bs;
1162: }
1163: }
1164: }
1165: /* add k_th block row to il, jl */
1166: col = aj+jmin;
1167: if (*col == k) jmin++;
1168: if (jmin < jmax){
1169: il[k] = jmin;
1170: j = a->j[jmin];
1171: jl[k] = jl[j]; jl[j] = k;
1172: }
1173: for (j=0; j<bs; j++){
1174: if (sum[j] > *norm) *norm = sum[j];
1175: }
1176: }
1177: PetscFree(il);
1178: PetscFree(jl);
1179: PetscFree(sum);
1180: } else {
1181: SETERRQ(PETSC_ERR_SUP,"No support for this norm yet");
1182: }
1183: return(0);
1184: }
1186: int MatEqual_SeqSBAIJ(Mat A,Mat B,PetscTruth* flg)
1187: {
1188: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data,*b = (Mat_SeqSBAIJ *)B->data;
1189: int ierr;
1190: PetscTruth flag;
1193: PetscTypeCompare((PetscObject)A,MATSEQBAIJ,&flag);
1194: if (!flag) SETERRQ(PETSC_ERR_ARG_INCOMP,"Matrices must be same type");
1196: /* If the matrix/block dimensions are not equal, or no of nonzeros or shift */
1197: if ((A->m != B->m) || (A->n != B->n) || (a->bs != b->bs)|| (a->s_nz != b->s_nz)) {
1198: *flg = PETSC_FALSE;
1199: return(0);
1200: }
1201:
1202: /* if the a->i are the same */
1203: PetscMemcmp(a->i,b->i,(a->mbs+1)*sizeof(int),flg);
1204: if (*flg == PETSC_FALSE) {
1205: return(0);
1206: }
1207:
1208: /* if a->j are the same */
1209: PetscMemcmp(a->j,b->j,(a->s_nz)*sizeof(int),flg);
1210: if (*flg == PETSC_FALSE) {
1211: return(0);
1212: }
1213: /* if a->a are the same */
1214: PetscMemcmp(a->a,b->a,(a->s_nz)*(a->bs)*(a->bs)*sizeof(PetscScalar),flg);
1215: return(0);
1216:
1217: }
1219: int MatGetDiagonal_SeqSBAIJ(Mat A,Vec v)
1220: {
1221: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1222: int ierr,i,j,k,n,row,bs,*ai,*aj,ambs,bs2;
1223: PetscScalar *x,zero = 0.0;
1224: MatScalar *aa,*aa_j;
1227: if (A->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1228: bs = a->bs;
1229: aa = a->a;
1230: ai = a->i;
1231: aj = a->j;
1232: ambs = a->mbs;
1233: bs2 = a->bs2;
1235: VecSet(&zero,v);
1236: VecGetArray(v,&x);
1237: VecGetLocalSize(v,&n);
1238: if (n != A->m) SETERRQ(PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
1239: for (i=0; i<ambs; i++) {
1240: j=ai[i];
1241: if (aj[j] == i) { /* if this is a diagonal element */
1242: row = i*bs;
1243: aa_j = aa + j*bs2;
1244: for (k=0; k<bs2; k+=(bs+1),row++) x[row] = aa_j[k];
1245: }
1246: }
1247: VecRestoreArray(v,&x);
1248: return(0);
1249: }
1251: int MatDiagonalScale_SeqSBAIJ(Mat A,Vec ll,Vec rr)
1252: {
1253: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1254: PetscScalar *l,*r,x,*li,*ri;
1255: MatScalar *aa,*v;
1256: int ierr,i,j,k,lm,rn,M,m,*ai,*aj,mbs,tmp,bs,bs2;
1259: ai = a->i;
1260: aj = a->j;
1261: aa = a->a;
1262: m = A->m;
1263: bs = a->bs;
1264: mbs = a->mbs;
1265: bs2 = a->bs2;
1267: if (ll != rr) {
1268: SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"For symmetric format, left and right scaling vectors must be samen");
1269: }
1270: if (ll) {
1271: VecGetArray(ll,&l);
1272: VecGetLocalSize(ll,&lm);
1273: if (lm != m) SETERRQ(PETSC_ERR_ARG_SIZ,"Left scaling vector wrong length");
1274: for (i=0; i<mbs; i++) { /* for each block row */
1275: M = ai[i+1] - ai[i];
1276: li = l + i*bs;
1277: v = aa + bs2*ai[i];
1278: for (j=0; j<M; j++) { /* for each block */
1279: for (k=0; k<bs2; k++) {
1280: (*v++) *= li[k%bs];
1281: }
1282: #ifdef CONT
1283: /* will be used to replace the above loop */
1284: ri = l + bs*aj[ai[i]+j];
1285: for (k=0; k<bs; k++) { /* column value */
1286: x = ri[k];
1287: for (tmp=0; tmp<bs; tmp++) (*v++) *= li[tmp]*x;
1288: }
1289: #endif
1291: }
1292: }
1293: VecRestoreArray(ll,&l);
1294: PetscLogFlops(2*a->s_nz);
1295: }
1296: /* will be deleted */
1297: if (rr) {
1298: VecGetArray(rr,&r);
1299: VecGetLocalSize(rr,&rn);
1300: if (rn != m) SETERRQ(PETSC_ERR_ARG_SIZ,"Right scaling vector wrong length");
1301: for (i=0; i<mbs; i++) { /* for each block row */
1302: M = ai[i+1] - ai[i];
1303: v = aa + bs2*ai[i];
1304: for (j=0; j<M; j++) { /* for each block */
1305: ri = r + bs*aj[ai[i]+j];
1306: for (k=0; k<bs; k++) {
1307: x = ri[k];
1308: for (tmp=0; tmp<bs; tmp++) (*v++) *= x;
1309: }
1310: }
1311: }
1312: VecRestoreArray(rr,&r);
1313: PetscLogFlops(a->s_nz);
1314: }
1315: return(0);
1316: }
1318: int MatGetInfo_SeqSBAIJ(Mat A,MatInfoType flag,MatInfo *info)
1319: {
1320: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1323: info->rows_global = (double)A->m;
1324: info->columns_global = (double)A->m;
1325: info->rows_local = (double)A->m;
1326: info->columns_local = (double)A->m;
1327: info->block_size = a->bs2;
1328: info->nz_allocated = a->s_maxnz; /*num. of nonzeros in upper triangular part */
1329: info->nz_used = a->bs2*a->s_nz; /*num. of nonzeros in upper triangular part */
1330: info->nz_unneeded = (double)(info->nz_allocated - info->nz_used);
1331: info->assemblies = A->num_ass;
1332: info->mallocs = a->reallocs;
1333: info->memory = A->mem;
1334: if (A->factor) {
1335: info->fill_ratio_given = A->info.fill_ratio_given;
1336: info->fill_ratio_needed = A->info.fill_ratio_needed;
1337: info->factor_mallocs = A->info.factor_mallocs;
1338: } else {
1339: info->fill_ratio_given = 0;
1340: info->fill_ratio_needed = 0;
1341: info->factor_mallocs = 0;
1342: }
1343: return(0);
1344: }
1347: int MatZeroEntries_SeqSBAIJ(Mat A)
1348: {
1349: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1350: int ierr;
1353: PetscMemzero(a->a,a->bs2*a->i[a->mbs]*sizeof(MatScalar));
1354: return(0);
1355: }
1357: int MatGetRowMax_SeqSBAIJ(Mat A,Vec v)
1358: {
1359: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data;
1360: int ierr,i,j,n,row,col,bs,*ai,*aj,mbs;
1361: PetscReal atmp;
1362: MatScalar *aa;
1363: PetscScalar zero = 0.0,*x;
1364: int ncols,brow,bcol,krow,kcol;
1367: if (A->factor) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Not for factored matrix");
1368: bs = a->bs;
1369: aa = a->a;
1370: ai = a->i;
1371: aj = a->j;
1372: mbs = a->mbs;
1374: VecSet(&zero,v);
1375: VecGetArray(v,&x);
1376: VecGetLocalSize(v,&n);
1377: if (n != A->m) SETERRQ(PETSC_ERR_ARG_SIZ,"Nonconforming matrix and vector");
1378: for (i=0; i<mbs; i++) {
1379: ncols = ai[1] - ai[0]; ai++;
1380: brow = bs*i;
1381: for (j=0; j<ncols; j++){
1382: bcol = bs*(*aj);
1383: for (kcol=0; kcol<bs; kcol++){
1384: col = bcol + kcol; /* col index */
1385: for (krow=0; krow<bs; krow++){
1386: atmp = PetscAbsScalar(*aa); aa++;
1387: row = brow + krow; /* row index */
1388: /* printf("val[%d,%d]: %gn",row,col,atmp); */
1389: if (PetscRealPart(x[row]) < atmp) x[row] = atmp;
1390: if (*aj > i && PetscRealPart(x[col]) < atmp) x[col] = atmp;
1391: }
1392: }
1393: aj++;
1394: }
1395: }
1396: VecRestoreArray(v,&x);
1397: return(0);
1398: }