/[escript]/trunk/finley/src/Assemble_PDE_Single2_2D.c
ViewVC logotype

Contents of /trunk/finley/src/Assemble_PDE_Single2_2D.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3259 - (show annotations)
Mon Oct 11 01:48:14 2010 UTC (8 years, 11 months ago) by jfenwick
File MIME type: text/plain
File size: 16144 byte(s)
Merging dudley and scons updates from branches

1
2 /*******************************************************
3 *
4 * Copyright (c) 2003-2010 by University of Queensland
5 * Earth Systems Science Computational Center (ESSCC)
6 * http://www.uq.edu.au/esscc
7 *
8 * Primary Business: Queensland, Australia
9 * Licensed under the Open Software License version 3.0
10 * http://www.opensource.org/licenses/osl-3.0.php
11 *
12 *******************************************************/
13
14
15 /**************************************************************/
16
17 /* assembles the system of numEq PDEs into the stiffness matrix S right hand side F */
18 /* the shape functions for test and solution must be identical */
19
20
21 /* -(A_{i,j} u_,j)_i-(B_{i} u)_i+C_{j} u_,j-D u_m and -(X_,i)_i + Y */
22
23 /* in a 2D domain. The shape functions for test and solution must be identical */
24 /* and row_NS == row_NN */
25
26 /* Shape of the coefficients: */
27
28 /* A = 2 x 2 */
29 /* B = 2 */
30 /* C = 2 */
31 /* D = scalar */
32 /* X = 2 */
33 /* Y = scalar */
34
35
36 /**************************************************************/
37
38
39 #include "Assemble.h"
40 #include "Util.h"
41 #ifdef _OPENMP
42 #include <omp.h>
43 #endif
44
45 /**************************************************************/
46
47 void Finley_Assemble_PDE_Single2_2D(Finley_Assemble_Parameters p,
48 Finley_ElementFile* elements,
49 Paso_SystemMatrix* Mat, escriptDataC* F,
50 escriptDataC* A, escriptDataC* B, escriptDataC* C, escriptDataC* D, escriptDataC* X, escriptDataC* Y) {
51
52 #define DIM 2
53 index_t color;
54 dim_t e,isub;
55 __const double *A_p, *B_p, *C_p, *D_p, *X_p, *Y_p, *A_q, *B_q, *C_q, *D_q, *X_q, *Y_q;
56 double *EM_S, *EM_F, *Vol, *DSDX;
57 index_t *row_index;
58 register dim_t q, s,r;
59 register double rtmp00, rtmp01, rtmp10, rtmp11, rtmp, rtmp0, rtmp1;
60 bool_t add_EM_F, add_EM_S;
61
62 bool_t extendedA=isExpanded(A);
63 bool_t extendedB=isExpanded(B);
64 bool_t extendedC=isExpanded(C);
65 bool_t extendedD=isExpanded(D);
66 bool_t extendedX=isExpanded(X);
67 bool_t extendedY=isExpanded(Y);
68 double *F_p=(requireWrite(F), getSampleDataRW(F,0)); /* use comma, to get around the mixed code and declarations thing */
69 double *S=p.row_jac->BasisFunctions->S;
70 dim_t len_EM_S=p.row_numShapesTotal*p.col_numShapesTotal;
71 dim_t len_EM_F=p.row_numShapesTotal;
72
73 #pragma omp parallel private(color,EM_S, EM_F, Vol, DSDX, A_p, B_p, C_p, D_p, X_p, Y_p, A_q, B_q, C_q, D_q, X_q, Y_q,row_index,q, s,r,rtmp00, rtmp01, rtmp10, rtmp11, rtmp, rtmp0, rtmp1,add_EM_F, add_EM_S, isub)
74 {
75 EM_S=THREAD_MEMALLOC(len_EM_S,double);
76 EM_F=THREAD_MEMALLOC(len_EM_F,double);
77 row_index=THREAD_MEMALLOC(p.row_numShapesTotal,index_t);
78
79 if (!Finley_checkPtr(EM_S) && !Finley_checkPtr(EM_F) && !Finley_checkPtr(row_index) ) {
80
81 for (color=elements->minColor;color<=elements->maxColor;color++) {
82 /* open loop over all elements: */
83 #pragma omp for private(e) schedule(static)
84 for(e=0;e<elements->numElements;e++){
85 if (elements->Color[e]==color) {
86
87 A_p=getSampleDataRO(A,e);
88 B_p=getSampleDataRO(B,e);
89 C_p=getSampleDataRO(C,e);
90 D_p=getSampleDataRO(D,e);
91 X_p=getSampleDataRO(X,e);
92 Y_p=getSampleDataRO(Y,e);
93
94 for (isub=0; isub<p.numSub; isub++) {
95
96 Vol=&(p.row_jac->volume[INDEX3(0,isub,e, p.numQuadSub,p.numSub)]);
97 DSDX=&(p.row_jac->DSDX[INDEX5(0,0,0,isub,e, p.row_numShapesTotal,DIM,p.numQuadSub,p.numSub)]);
98 for (q=0;q<len_EM_S;++q) EM_S[q]=0;
99 for (q=0;q<len_EM_F;++q) EM_F[q]=0;
100 add_EM_F=FALSE;
101 add_EM_S=FALSE;
102 /**************************************************************/
103 /* process A: */
104 /**************************************************************/
105 if (NULL!=A_p) {
106 add_EM_S=TRUE;
107 if (extendedA) {
108 A_q=&(A_p[INDEX4(0,0,0,isub, DIM,DIM,p.numQuadSub)]);
109 for (s=0;s<p.row_numShapes;s++) {
110 for (r=0;r<p.col_numShapes;r++) {
111 rtmp=0;
112 for (q=0;q<p.numQuadSub;q++) {
113 rtmp+=Vol[q]*(DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(0,0,q,DIM,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
114 + DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(0,1,q,DIM,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
115 + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(1,0,q,DIM,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
116 + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(1,1,q,DIM,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)] );
117 }
118 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
119 }
120 }
121 } else {
122 for (s=0;s<p.row_numShapes;s++) {
123 for (r=0;r<p.col_numShapes;r++) {
124 rtmp00=0;
125 rtmp01=0;
126 rtmp10=0;
127 rtmp11=0;
128 for (q=0;q<p.numQuadSub;q++) {
129 rtmp0=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
130 rtmp1=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
131 rtmp00+=rtmp0*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
132 rtmp01+=rtmp0*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
133 rtmp10+=rtmp1*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
134 rtmp11+=rtmp1*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
135 }
136 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+= rtmp00*A_p[INDEX2(0,0,DIM)]
137 + rtmp01*A_p[INDEX2(0,1,DIM)]
138 + rtmp10*A_p[INDEX2(1,0,DIM)]
139 + rtmp11*A_p[INDEX2(1,1,DIM)];
140 }
141 }
142 }
143 }
144 /**************************************************************/
145 /* process B: */
146 /**************************************************************/
147 if (NULL!=B_p) {
148 add_EM_S=TRUE;
149 if (extendedB) {
150 B_q=&(B_p[INDEX3(0,0,isub, DIM,p.numQuadSub)]);
151 for (s=0;s<p.row_numShapes;s++) {
152 for (r=0;r<p.col_numShapes;r++) {
153 rtmp=0.;
154 for (q=0;q<p.numQuadSub;q++) {
155 rtmp+=Vol[q]*S[INDEX2(r,q,p.row_numShapes)]*(DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(0,q,DIM)]
156 + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(1,q,DIM)]);
157 }
158 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
159 }
160 }
161 } else {
162 for (s=0;s<p.row_numShapes;s++) {
163 for (r=0;r<p.col_numShapes;r++) {
164 rtmp0=0;
165 rtmp1=0;
166 for (q=0;q<p.numQuadSub;q++) {
167 rtmp=Vol[q]*S[INDEX2(r,q,p.row_numShapes)];
168 rtmp0+=rtmp*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
169 rtmp1+=rtmp*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
170 }
171 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*B_p[0]+rtmp1*B_p[1];
172 }
173 }
174 }
175 }
176 /**************************************************************/
177 /* process C: */
178 /**************************************************************/
179 if (NULL!=C_p) {
180 add_EM_S=TRUE;
181 if (extendedC) {
182 C_q=&(C_p[INDEX3(0,0,isub, DIM, p.numQuadSub)]);
183 for (s=0;s<p.row_numShapes;s++) {
184 for (r=0;r<p.col_numShapes;r++) {
185 rtmp=0;
186 for (q=0;q<p.numQuadSub;q++) {
187 rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*(C_q[INDEX2(0,q,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
188 + C_q[INDEX2(1,q,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]);
189 }
190 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
191 }
192 }
193 } else {
194 for (s=0;s<p.row_numShapes;s++) {
195 for (r=0;r<p.col_numShapes;r++) {
196 rtmp0=0;
197 rtmp1=0;
198 for (q=0;q<p.numQuadSub;q++) {
199 rtmp=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
200 rtmp0+=rtmp*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
201 rtmp1+=rtmp*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
202 }
203 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*C_p[0]+rtmp1*C_p[1];
204 }
205 }
206 }
207 }
208 /************************************************************* */
209 /* process D */
210 /**************************************************************/
211 if (NULL!=D_p) {
212 add_EM_S=TRUE;
213 if (extendedD) {
214 D_q=&(D_p[INDEX2(0,isub, p.numQuadSub)]);
215 for (s=0;s<p.row_numShapes;s++) {
216 for (r=0;r<p.col_numShapes;r++) {
217 rtmp=0;
218 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*D_q[q]*S[INDEX2(r,q,p.row_numShapes)];
219 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
220 }
221 }
222 } else {
223 for (s=0;s<p.row_numShapes;s++) {
224 for (r=0;r<p.col_numShapes;r++) {
225 rtmp=0;
226 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*S[INDEX2(r,q,p.row_numShapes)];
227 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp*D_p[0];
228 }
229 }
230 }
231 }
232 /**************************************************************/
233 /* process X: */
234 /**************************************************************/
235 if (NULL!=X_p) {
236 add_EM_F=TRUE;
237 if (extendedX) {
238 X_q=&(X_p[INDEX3(0,0,isub, DIM,p.numQuadSub)]);
239 for (s=0;s<p.row_numShapes;s++) {
240 rtmp=0.;
241 for (q=0;q<p.numQuadSub;q++) {
242 rtmp+=Vol[q]*( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(0,q,DIM)]
243 + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(1,q,DIM)]);
244 }
245 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
246 }
247 } else {
248 for (s=0;s<p.row_numShapes;s++) {
249 rtmp0=0.;
250 rtmp1=0.;
251 for (q=0;q<p.numQuadSub;q++) {
252 rtmp0+=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
253 rtmp1+=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
254 }
255 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp0*X_p[0]+rtmp1*X_p[1];
256 }
257 }
258 }
259 /**************************************************************/
260 /* process Y: */
261 /**************************************************************/
262 if (NULL!=Y_p) {
263 add_EM_F=TRUE;
264 if (extendedY) {
265 Y_q=&(Y_p[INDEX2(0,isub, p.numQuadSub)]);
266 for (s=0;s<p.row_numShapes;s++) {
267 rtmp=0;
268 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*Y_q[q];
269 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
270 }
271 } else {
272 for (s=0;s<p.row_numShapes;s++) {
273 rtmp=0;
274 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
275 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp*Y_p[0];
276 }
277 }
278 }
279 /***********************************************************************************************/
280 /* add the element matrices onto the matrix and right hand side */
281 /***********************************************************************************************/
282
283 for (q=0;q<p.row_numShapesTotal;q++) row_index[q]=p.row_DOF[elements->Nodes[INDEX2(p.row_node[INDEX2(q,isub,p.row_numShapesTotal)],e,p.NN)]];
284 if (add_EM_F) Finley_Util_AddScatter(p.row_numShapesTotal,row_index,p.numEqu,EM_F,F_p, p.row_DOF_UpperBound);
285 if (add_EM_S) Finley_Assemble_addToSystemMatrix(Mat,p.row_numShapesTotal,row_index,p.numEqu,p.col_numShapesTotal,row_index,p.numComp,EM_S);
286
287 } /* end of isub */
288 } /* end color check */
289 } /* end element loop */
290 } /* end color loop */
291
292 THREAD_MEMFREE(EM_S);
293 THREAD_MEMFREE(EM_F);
294 THREAD_MEMFREE(row_index);
295
296 } /* end of pointer check */
297 } /* end parallel region */
298 }

  ViewVC Help
Powered by ViewVC 1.1.26