/[escript]/trunk/paso/src/Solver_SchurComplement.c
ViewVC logotype

Contents of /trunk/paso/src/Solver_SchurComplement.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1388 - (show annotations)
Fri Jan 11 07:45:58 2008 UTC (11 years, 9 months ago) by trankine
File MIME type: text/plain
File size: 10657 byte(s)
And get the *(&(*&(* name right
1
2 /* $Id$ */
3
4 /*******************************************************
5 *
6 * Copyright 2003-2007 by ACceSS MNRF
7 * Copyright 2007 by University of Queensland
8 *
9 * http://esscc.uq.edu.au
10 * Primary Business: Queensland, Australia
11 * Licensed under the Open Software License version 3.0
12 * http://www.opensource.org/licenses/osl-3.0.php
13 *
14 *******************************************************/
15
16 /**************************************************************/
17
18 /* Paso: updates A_CC <- ACC-ACF AFF^{-1} AFC */
19
20 /* no check of consistency of matrices !!!! */
21
22 /**************************************************************/
23
24 /* Copyrights by ACcESS Australia 2003,2004,2005 */
25 /* Author: gross@access.edu.au */
26
27 /**************************************************************/
28
29 #include "Paso.h"
30 #include "SparseMatrix.h"
31 #include "Solver.h"
32
33 /**************************************************************/
34
35
36
37 void Paso_Solver_updateIncompleteSchurComplement(Paso_SparseMatrix* A_CC,Paso_SparseMatrix *A_CF,double* invA_FF,index_t* A_FF_pivot,Paso_SparseMatrix *A_FC) {
38
39 index_t iPtr_CC,*index_CC,col_CF,col_FC, *where_p,iPtr_CC_2,i,iPtr_CF,iPtr_FC;
40 dim_t index_CC_len;
41 bool_t set_A;
42 dim_t n_loc_rows=A_CC->numRows;
43 dim_t n_block=A_CC->row_block_size;
44 register double A_CF_11,A_CF_21,A_CF_31,A_CF_12,A_CF_22,A_CF_32,A_CF_13,A_CF_23,A_CF_33,
45 invA_FF_11,invA_FF_21,invA_FF_31,invA_FF_12,invA_FF_22,invA_FF_32,invA_FF_13,invA_FF_23,invA_FF_33,
46 A11,A21,A31,A12,A22,A32,A13,A23,A33,A_FC_11,A_FC_21,A_FC_31,A_FC_12,A_FC_22,A_FC_32,A_FC_13,A_FC_23,A_FC_33;
47 if (n_block==1) {
48 #pragma omp parallel for private(i,iPtr_CC,index_CC,index_CC_len,col_CF,set_A,iPtr_CF,iPtr_FC,col_FC,where_p,A11) schedule(static)
49 for (i = 0; i < n_loc_rows;++i) {
50 iPtr_CC=A_CC->pattern->ptr[i];
51 index_CC=&(A_CC->pattern->index[iPtr_CC]);
52 index_CC_len=(size_t)(A_CC->pattern->ptr[i+1]-A_CC->pattern->ptr[i]);
53 /* now we run through the columns of A_CF in row i */
54 for (iPtr_CF = A_CF->pattern->ptr[i]; iPtr_CF < A_CF->pattern->ptr[i + 1]; ++iPtr_CF) {
55 col_CF=A_CF->pattern->index[iPtr_CF];
56 set_A=TRUE;
57 for (iPtr_FC = A_FC->pattern->ptr[col_CF]; iPtr_FC < A_FC->pattern->ptr[col_CF + 1]; ++iPtr_FC) {
58 col_FC=A_FC->pattern->index[iPtr_FC];
59 /* is (i,col_FC) in the shape of A_CC ? */
60 where_p=(index_t*)bsearch(&col_FC,index_CC,index_CC_len,sizeof(index_t),Paso_comparIndex);
61 if (where_p!=NULL) {
62 if (set_A) {
63 A11=A_CF->val[iPtr_CF]*invA_FF[col_CF];
64 set_A=FALSE;
65 }
66 A_CC->val[iPtr_CC+(index_t)(where_p-index_CC)]-=A11*A_FC->val[iPtr_FC];
67 }
68 } /* end of iPtr_FC loop */
69 } /* end of iPtr_CF loop */
70 } /* end of irow loop */
71 } else if (n_block==2) {
72 #pragma omp parallel for private(i,iPtr_CC,index_CC,index_CC_len,iPtr_CF,col_CF,iPtr_FC,col_FC,where_p,iPtr_CC_2,set_A,A_CF_11,A_CF_21,A_CF_12,A_CF_22,invA_FF_11,invA_FF_21,invA_FF_12,invA_FF_22,A11,A21,A12,A22,A_FC_11,A_FC_21,A_FC_12,A_FC_22) schedule(static)
73 for (i = 0; i < n_loc_rows;++i) {
74 iPtr_CC=A_CC->pattern->ptr[i];
75 index_CC=&(A_CC->pattern->index[iPtr_CC]);
76 index_CC_len=(size_t)(A_CC->pattern->ptr[i+1]-A_CC->pattern->ptr[i]);
77 /* now we run through the columns of A_CF in row i */
78 for (iPtr_CF = A_CF->pattern->ptr[i]; iPtr_CF < A_CF->pattern->ptr[i + 1]; ++iPtr_CF) {
79 col_CF=A_CF->pattern->index[iPtr_CF];
80 set_A=TRUE;
81 for (iPtr_FC = A_FC->pattern->ptr[col_CF]; iPtr_FC < A_FC->pattern->ptr[col_CF + 1]; ++iPtr_FC) {
82 col_FC=A_FC->pattern->index[iPtr_FC];
83 /* is (i,col_FC) in the shape of A_CC ? */
84 where_p=(index_t*)bsearch(&col_FC,index_CC,index_CC_len,sizeof(index_t),Paso_comparIndex);
85 if (where_p!=NULL) {
86 iPtr_CC_2=iPtr_CC+(index_t)(where_p-index_CC);
87 /* this calculutes A_CF*invA_FF(i,col_CF) */
88 if (set_A) {
89 A_CF_11=A_CF->val[4*iPtr_CF ];
90 A_CF_21=A_CF->val[4*iPtr_CF+1];
91 A_CF_12=A_CF->val[4*iPtr_CF+2];
92 A_CF_22=A_CF->val[4*iPtr_CF+3];
93
94 invA_FF_11=invA_FF[4*col_CF ];
95 invA_FF_21=invA_FF[4*col_CF+1];
96 invA_FF_12=invA_FF[4*col_CF+2];
97 invA_FF_22=invA_FF[4*col_CF+3];
98
99 A11=A_CF_11*invA_FF_11+A_CF_12*invA_FF_21;
100 A21=A_CF_21*invA_FF_11+A_CF_22*invA_FF_21;
101 A12=A_CF_11*invA_FF_12+A_CF_12*invA_FF_22;
102 A22=A_CF_21*invA_FF_12+A_CF_22*invA_FF_22;
103
104 set_A=FALSE;
105 }
106
107 A_FC_11=A_FC->val[4*iPtr_FC ];
108 A_FC_21=A_FC->val[4*iPtr_FC+1];
109 A_FC_12=A_FC->val[4*iPtr_FC+2];
110 A_FC_22=A_FC->val[4*iPtr_FC+3];
111
112 A_CC->val[4*iPtr_CC_2 ]-=A11*A_FC_11+A12*A_FC_21;
113 A_CC->val[4*iPtr_CC_2+1]-=A21*A_FC_11+A22*A_FC_21;
114 A_CC->val[4*iPtr_CC_2+2]-=A11*A_FC_12+A12*A_FC_22;
115 A_CC->val[4*iPtr_CC_2+3]-=A21*A_FC_12+A22*A_FC_22;
116
117 }
118 } /* end of iPtr_FC loop */
119 } /* end of iPtr_CF loop */
120 } /* end of irow loop */
121 } else if (n_block==3) {
122 #pragma omp parallel for private(i,iPtr_CC,index_CC,index_CC_len,iPtr_CF,col_CF,iPtr_FC,col_FC,where_p,iPtr_CC_2,set_A,A_CF_11,A_CF_21,A_CF_31,A_CF_12,A_CF_22,A_CF_32,A_CF_13,A_CF_23,A_CF_33,invA_FF_11,invA_FF_21,invA_FF_31,invA_FF_12,invA_FF_22,invA_FF_32,invA_FF_13,invA_FF_23,invA_FF_33,A11,A21,A31,A12,A22,A32,A13,A23,A33,A_FC_11,A_FC_21,A_FC_31,A_FC_12,A_FC_22,A_FC_32,A_FC_13,A_FC_23,A_FC_33) schedule(static)
123 for (i = 0; i < n_loc_rows;++i) {
124 iPtr_CC=A_CC->pattern->ptr[i];
125 index_CC=&(A_CC->pattern->index[iPtr_CC]);
126 index_CC_len=(size_t)(A_CC->pattern->ptr[i+1]-A_CC->pattern->ptr[i]);
127 /* now we run through the columns of A_CF in row i */
128 for (iPtr_CF = A_CF->pattern->ptr[i]; iPtr_CF < A_CF->pattern->ptr[i + 1]; ++iPtr_CF) {
129 col_CF=A_CF->pattern->index[iPtr_CF];
130 set_A=TRUE;
131 for (iPtr_FC = A_FC->pattern->ptr[col_CF]; iPtr_FC < A_FC->pattern->ptr[col_CF + 1]; ++iPtr_FC) {
132 col_FC=A_FC->pattern->index[iPtr_FC];
133 /* is (i,col_FC) in the shape of A_CC ? */
134 where_p=(index_t*)bsearch(&col_FC,index_CC,index_CC_len,sizeof(index_t),Paso_comparIndex);
135 if (where_p!=NULL) {
136 iPtr_CC_2=iPtr_CC+(index_t)(where_p-index_CC);
137 /* this calculutes A_CF*invA_FF(i,col_CF) */
138 if (set_A) {
139 A_CF_11=A_CF->val[9*iPtr_CF ];
140 A_CF_21=A_CF->val[9*iPtr_CF+1];
141 A_CF_31=A_CF->val[9*iPtr_CF+2];
142 A_CF_12=A_CF->val[9*iPtr_CF+3];
143 A_CF_22=A_CF->val[9*iPtr_CF+4];
144 A_CF_32=A_CF->val[9*iPtr_CF+5];
145 A_CF_13=A_CF->val[9*iPtr_CF+6];
146 A_CF_23=A_CF->val[9*iPtr_CF+7];
147 A_CF_33=A_CF->val[9*iPtr_CF+8];
148
149 invA_FF_11=invA_FF[9*col_CF ];
150 invA_FF_21=invA_FF[9*col_CF+1];
151 invA_FF_31=invA_FF[9*col_CF+2];
152 invA_FF_12=invA_FF[9*col_CF+3];
153 invA_FF_22=invA_FF[9*col_CF+4];
154 invA_FF_32=invA_FF[9*col_CF+5];
155 invA_FF_13=invA_FF[9*col_CF+6];
156 invA_FF_23=invA_FF[9*col_CF+7];
157 invA_FF_33=invA_FF[9*col_CF+8];
158
159 A11=A_CF_11*invA_FF_11+A_CF_12*invA_FF_21+A_CF_13*invA_FF_31;
160 A21=A_CF_21*invA_FF_11+A_CF_22*invA_FF_21+A_CF_23*invA_FF_31;
161 A31=A_CF_31*invA_FF_11+A_CF_32*invA_FF_21+A_CF_33*invA_FF_31;
162 A12=A_CF_11*invA_FF_12+A_CF_12*invA_FF_22+A_CF_13*invA_FF_32;
163 A22=A_CF_21*invA_FF_12+A_CF_22*invA_FF_22+A_CF_23*invA_FF_32;
164 A32=A_CF_31*invA_FF_12+A_CF_32*invA_FF_22+A_CF_33*invA_FF_32;
165 A13=A_CF_11*invA_FF_13+A_CF_12*invA_FF_23+A_CF_13*invA_FF_33;
166 A23=A_CF_21*invA_FF_13+A_CF_22*invA_FF_23+A_CF_23*invA_FF_33;
167 A33=A_CF_31*invA_FF_13+A_CF_32*invA_FF_23+A_CF_33*invA_FF_33;
168
169 set_A=FALSE;
170 }
171
172 A_FC_11=A_FC->val[9*iPtr_FC ];
173 A_FC_21=A_FC->val[9*iPtr_FC+1];
174 A_FC_31=A_FC->val[9*iPtr_FC+2];
175 A_FC_12=A_FC->val[9*iPtr_FC+3];
176 A_FC_22=A_FC->val[9*iPtr_FC+4];
177 A_FC_32=A_FC->val[9*iPtr_FC+5];
178 A_FC_13=A_FC->val[9*iPtr_FC+6];
179 A_FC_23=A_FC->val[9*iPtr_FC+7];
180 A_FC_33=A_FC->val[9*iPtr_FC+8];
181
182 A_CC->val[9*iPtr_CC_2 ]-=A11*A_FC_11+A12*A_FC_21+A13*A_FC_31;
183 A_CC->val[9*iPtr_CC_2+1]-=A21*A_FC_11+A22*A_FC_21+A23*A_FC_31;
184 A_CC->val[9*iPtr_CC_2+2]-=A31*A_FC_11+A32*A_FC_21+A33*A_FC_31;
185 A_CC->val[9*iPtr_CC_2+3]-=A11*A_FC_12+A12*A_FC_22+A13*A_FC_32;
186 A_CC->val[9*iPtr_CC_2+4]-=A21*A_FC_12+A22*A_FC_22+A23*A_FC_32;
187 A_CC->val[9*iPtr_CC_2+5]-=A31*A_FC_12+A32*A_FC_22+A33*A_FC_32;
188 A_CC->val[9*iPtr_CC_2+6]-=A11*A_FC_13+A12*A_FC_23+A13*A_FC_33;
189 A_CC->val[9*iPtr_CC_2+7]-=A21*A_FC_13+A22*A_FC_23+A23*A_FC_33;
190 A_CC->val[9*iPtr_CC_2+8]-=A31*A_FC_13+A32*A_FC_23+A33*A_FC_33;
191
192 }
193 } /* end of iPtr_FC loop */
194 } /* end of iPtr_CF loop */
195 } /* end of irow loop */
196 }
197 }
198 /*
199 * $Log$
200 * Revision 1.2 2005/09/15 03:44:40 jgs
201 * Merge of development branch dev-02 back to main trunk on 2005-09-15
202 *
203 * Revision 1.1.2.1 2005/09/05 06:29:50 gross
204 * These files have been extracted from finley to define a stand alone libray for iterative
205 * linear solvers on the ALTIX. main entry through Paso_solve. this version compiles but
206 * has not been tested yet.
207 *
208 *
209 */

Properties

Name Value
svn:eol-style native
svn:keywords Author Date Id Revision

  ViewVC Help
Powered by ViewVC 1.1.26