/[escript]/branches/domexper/dudley/src/Assemble_PDE_System2_3D.c
ViewVC logotype

Annotation of /branches/domexper/dudley/src/Assemble_PDE_System2_3D.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3086 - (hide annotations)
Thu Aug 5 05:07:58 2010 UTC (8 years, 10 months ago) by jfenwick
File MIME type: text/plain
File size: 23477 byte(s)
Another pass at removing finley

1 gross 798
2 ksteube 1312 /*******************************************************
3 ksteube 1811 *
4 jfenwick 2881 * Copyright (c) 2003-2010 by University of Queensland
5 ksteube 1811 * 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 ksteube 1312
14 ksteube 1811
15 gross 798 /**************************************************************/
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_{k,i,m,j} u_m,j)_i-(B_{k,i,m} u_m)_i+C_{k,m,j} u_m,j-D_{k,m} u_m and -(X_{k,i})_i + Y_k */
22    
23     /* u has p.numComp components in a 3D 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 = p.numEqu x 3 x p.numComp x 3 */
29     /* B = 3 x p.numEqu x p.numComp */
30     /* C = p.numEqu x 3 x p.numComp */
31     /* D = p.numEqu x p.numComp */
32     /* X = p.numEqu x 3 */
33     /* Y = p.numEqu */
34    
35    
36     /**************************************************************/
37    
38    
39     #include "Assemble.h"
40     #include "Util.h"
41 gross 853 #ifdef _OPENMP
42     #include <omp.h>
43     #endif
44 gross 798
45 gross 853
46 gross 798 /**************************************************************/
47    
48 jfenwick 3086 void Dudley_Assemble_PDE_System2_3D(Assemble_Parameters p, Dudley_ElementFile* elements,
49 gross 798 Paso_SystemMatrix* Mat, escriptDataC* F,
50     escriptDataC* A, escriptDataC* B, escriptDataC* C, escriptDataC* D, escriptDataC* X, escriptDataC* Y) {
51    
52     #define DIM 3
53     index_t color;
54 gross 2748 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 jfenwick 2271 double *EM_S, *EM_F, *Vol, *DSDX;
57 gross 853 index_t *row_index;
58     register dim_t q, s,r,k,m;
59     register double rtmp, rtmp0, rtmp1, rtmp2, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22;
60     bool_t add_EM_F, add_EM_S;
61    
62 gross 798 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 jfenwick 2271 double *F_p=(requireWrite(F), getSampleDataRW(F,0)); /* use comma, to get around the mixed code and declarations thing */
69 gross 2748 double *S=p.row_jac->BasisFunctions->S;
70     dim_t len_EM_S=p.row_numShapesTotal*p.col_numShapesTotal*p.numEqu*p.numComp;
71     dim_t len_EM_F=p.row_numShapesTotal*p.numEqu;
72 gross 798
73 gross 2748 #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,k,m,rtmp, rtmp0, rtmp1, rtmp2, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22,add_EM_F, add_EM_S, isub)
74 gross 798 {
75 gross 853 EM_S=THREAD_MEMALLOC(len_EM_S,double);
76     EM_F=THREAD_MEMALLOC(len_EM_F,double);
77 gross 2748 row_index=THREAD_MEMALLOC(p.row_numShapesTotal,index_t);
78 gross 853
79 jfenwick 3086 if (!Dudley_checkPtr(EM_S) && !Dudley_checkPtr(EM_F) && !Dudley_checkPtr(row_index) ) {
80 gross 798
81 gross 853 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 gross 2748
87 jfenwick 2770 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 gross 2748
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     /**************************************************************/
104     /* process A: */
105     /**************************************************************/
106     if (NULL!=A_p) {
107     add_EM_S=TRUE;
108     if (extendedA) {
109     A_q=&(A_p[INDEX6(0,0,0,0,0,isub, p.numEqu,DIM,p.numComp,DIM,p.numQuadSub)]);
110     for (s=0;s<p.row_numShapes;s++) {
111     for (r=0;r<p.col_numShapes;r++) {
112     for (k=0;k<p.numEqu;k++) {
113 gross 798 for (m=0;m<p.numComp;m++) {
114 gross 2748 rtmp=0;
115     for (q=0;q<p.numQuadSub;q++) {
116     rtmp+=Vol[q]*(
117     DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,0,m,0,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
118     +DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,0,m,1,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
119     +DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,0,m,2,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]
120     +DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,1,m,0,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
121     +DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,1,m,1,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
122     +DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,1,m,2,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]
123     +DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,2,m,0,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
124     +DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,2,m,1,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
125     +DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX5(k,2,m,2,q, p.numEqu,DIM,p.numComp,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
126    
127     }
128     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
129 gross 798 }
130 gross 2748 }
131     }
132     }
133     } else {
134     for (s=0;s<p.row_numShapes;s++) {
135     for (r=0;r<p.col_numShapes;r++) {
136     rtmp00=0;
137     rtmp01=0;
138     rtmp02=0;
139     rtmp10=0;
140     rtmp11=0;
141     rtmp12=0;
142     rtmp20=0;
143     rtmp21=0;
144     rtmp22=0;
145     for (q=0;q<p.numQuadSub;q++) {
146     rtmp0=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
147     rtmp00+=rtmp0*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
148     rtmp01+=rtmp0*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
149     rtmp02+=rtmp0*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
150    
151     rtmp1=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
152     rtmp10+=rtmp1*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
153     rtmp11+=rtmp1*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
154     rtmp12+=rtmp1*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
155    
156     rtmp2=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
157     rtmp20+=rtmp2*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
158     rtmp21+=rtmp2*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
159     rtmp22+=rtmp2*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
160 gross 853 }
161     for (k=0;k<p.numEqu;k++) {
162     for (m=0;m<p.numComp;m++) {
163 gross 2748 EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+= rtmp00*A_p[INDEX4(k,0,m,0, p.numEqu,DIM,p.numComp)]
164     +rtmp01*A_p[INDEX4(k,0,m,1, p.numEqu,DIM,p.numComp)]
165     +rtmp02*A_p[INDEX4(k,0,m,2, p.numEqu,DIM,p.numComp)]
166     +rtmp10*A_p[INDEX4(k,1,m,0, p.numEqu,DIM,p.numComp)]
167     +rtmp11*A_p[INDEX4(k,1,m,1, p.numEqu,DIM,p.numComp)]
168     +rtmp12*A_p[INDEX4(k,1,m,2, p.numEqu,DIM,p.numComp)]
169     +rtmp20*A_p[INDEX4(k,2,m,0, p.numEqu,DIM,p.numComp)]
170     +rtmp21*A_p[INDEX4(k,2,m,1, p.numEqu,DIM,p.numComp)]
171     +rtmp22*A_p[INDEX4(k,2,m,2, p.numEqu,DIM,p.numComp)];
172 gross 853 }
173     }
174 gross 2748 }
175     }
176     }
177     }
178     /**************************************************************/
179     /* process B: */
180     /**************************************************************/
181     if (NULL!=B_p) {
182     add_EM_S=TRUE;
183     if (extendedB) {
184     B_q=&(B_p[INDEX5(0,0,0,0,isub, p.numEqu,DIM,p.numComp,p.numQuadSub)]);
185     for (s=0;s<p.row_numShapes;s++) {
186     for (r=0;r<p.col_numShapes;r++) {
187     for (k=0;k<p.numEqu;k++) {
188     for (m=0;m<p.numComp;m++) {
189     rtmp=0;
190     for (q=0;q<p.numQuadSub;q++) {
191     rtmp+=Vol[q]*S[INDEX2(r,q,p.row_numShapes)]*(
192     DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*B_q[INDEX4(k,0,m,q,p.numEqu,DIM,p.numComp)]
193     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*B_q[INDEX4(k,1,m,q,p.numEqu,DIM,p.numComp)]
194     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*B_q[INDEX4(k,2,m,q,p.numEqu,DIM,p.numComp)] );
195     }
196     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
197     }
198     }
199     }
200     }
201     } else {
202     for (s=0;s<p.row_numShapes;s++) {
203     for (r=0;r<p.col_numShapes;r++) {
204     rtmp0=0;
205     rtmp1=0;
206     rtmp2=0;
207     for (q=0;q<p.numQuadSub;q++) {
208     rtmp=Vol[q]*S[INDEX2(r,q,p.row_numShapes)];
209     rtmp0+=rtmp*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
210     rtmp1+=rtmp*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
211     rtmp2+=rtmp*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
212     }
213     for (k=0;k<p.numEqu;k++) {
214     for (m=0;m<p.numComp;m++) {
215     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*B_p[INDEX3(k,0,m,p.numEqu,DIM)]
216     +rtmp1*B_p[INDEX3(k,1,m,p.numEqu,DIM)]
217     +rtmp2*B_p[INDEX3(k,2,m,p.numEqu,DIM)];
218     }
219     }
220     }
221     }
222 gross 853 }
223 gross 798 }
224 gross 2748 /**************************************************************/
225     /* process C: */
226     /**************************************************************/
227     if (NULL!=C_p) {
228     add_EM_S=TRUE;
229     if (extendedC) {
230     C_q=&(C_p[INDEX5(0,0,0,0,isub ,p.numEqu,p.numComp,DIM, p.numQuadSub)]);
231     for (s=0;s<p.row_numShapes;s++) {
232     for (r=0;r<p.col_numShapes;r++) {
233     for (k=0;k<p.numEqu;k++) {
234     for (m=0;m<p.numComp;m++) {
235     rtmp=0;
236     for (q=0;q<p.numQuadSub;q++) {
237     rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*(
238     C_q[INDEX4(k,m,0,q,p.numEqu,p.numComp,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
239     +C_q[INDEX4(k,m,1,q,p.numEqu,p.numComp,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
240     +C_q[INDEX4(k,m,2,q,p.numEqu,p.numComp,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
241     }
242     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
243 gross 853 }
244 gross 2748 }
245     }
246     }
247     } else {
248     for (s=0;s<p.row_numShapes;s++) {
249     for (r=0;r<p.col_numShapes;r++) {
250     rtmp0=0;
251     rtmp1=0;
252     rtmp2=0;
253     for (q=0;q<p.numQuadSub;q++) {
254     rtmp=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
255     rtmp0+=rtmp*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
256     rtmp1+=rtmp*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
257     rtmp2+=rtmp*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
258     }
259     for (k=0;k<p.numEqu;k++) {
260     for (m=0;m<p.numComp;m++) {
261     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*C_p[INDEX3(k,m,0,p.numEqu,p.numComp)]
262     +rtmp1*C_p[INDEX3(k,m,1,p.numEqu,p.numComp)]
263     +rtmp2*C_p[INDEX3(k,m,2,p.numEqu,p.numComp)];
264     }
265     }
266     }
267     }
268     }
269     }
270     /************************************************************* */
271     /* process D */
272     /**************************************************************/
273     if (NULL!=D_p) {
274     add_EM_S=TRUE;
275     if (extendedD) {
276     D_q=&(D_p[INDEX4(0,0,0,isub ,p.numEqu,p.numComp, p.numQuadSub)]);
277     for (s=0;s<p.row_numShapes;s++) {
278     for (r=0;r<p.col_numShapes;r++) {
279     for (k=0;k<p.numEqu;k++) {
280     for (m=0;m<p.numComp;m++) {
281     rtmp=0;
282     for (q=0;q<p.numQuadSub;q++) {
283     rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*D_q[INDEX3(k,m,q,p.numEqu,p.numComp)]*S[INDEX2(r,q,p.row_numShapes)];
284     }
285     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
286     }
287     }
288     }
289     }
290     } else {
291     for (s=0;s<p.row_numShapes;s++) {
292     for (r=0;r<p.col_numShapes;r++) {
293     rtmp=0;
294     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*S[INDEX2(r,q,p.row_numShapes)];
295 gross 853 for (k=0;k<p.numEqu;k++) {
296 gross 2748 for (m=0;m<p.numComp;m++) {
297     EM_S[INDEX4(k,m,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp*D_p[INDEX2(k,m,p.numEqu)];
298     }
299 gross 853 }
300 gross 2748 }
301     }
302     }
303     }
304     /**************************************************************/
305     /* process X: */
306     /**************************************************************/
307     if (NULL!=X_p) {
308     add_EM_F=TRUE;
309     if (extendedX) {
310     X_q=&(X_p[INDEX4(0,0,0,isub, p.numEqu,DIM,p.numQuadSub)]);
311     for (s=0;s<p.row_numShapes;s++) {
312     for (k=0;k<p.numEqu;k++) {
313 gross 853 rtmp=0;
314 gross 2748 for (q=0;q<p.numQuadSub;q++) {
315     rtmp+=Vol[q]* ( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*X_q[INDEX3(k,0,q,p.numEqu,DIM)]
316     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*X_q[INDEX3(k,1,q,p.numEqu,DIM)]
317     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*X_q[INDEX3(k,2,q,p.numEqu,DIM)]);
318 gross 853 }
319 gross 2748 EM_F[INDEX2(k,s,p.numEqu)]+=rtmp;
320     }
321     }
322     } else {
323     for (s=0;s<p.row_numShapes;s++) {
324     rtmp0=0;
325     rtmp1=0;
326     rtmp2=0;
327     for (q=0;q<p.numQuadSub;q++) {
328     rtmp0+=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
329     rtmp1+=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
330     rtmp2+=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
331 gross 853 }
332 gross 2748 for (k=0;k<p.numEqu;k++) {
333     EM_F[INDEX2(k,s,p.numEqu)]+=rtmp0*X_p[INDEX2(k,0,p.numEqu)]
334     +rtmp1*X_p[INDEX2(k,1,p.numEqu)]
335     +rtmp2*X_p[INDEX2(k,2,p.numEqu)];
336     }
337 gross 853 }
338 gross 2748 }
339 gross 853 }
340 gross 2748 /**************************************************************/
341     /* process Y: */
342     /**************************************************************/
343     if (NULL!=Y_p) {
344     add_EM_F=TRUE;
345     if (extendedY) {
346     Y_q=&(Y_p[INDEX3(0,0,isub, p.numEqu,p.numQuadSub)]);
347     for (s=0;s<p.row_numShapes;s++) {
348     for (k=0;k<p.numEqu;k++) {
349     rtmp=0.;
350     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*Y_q[INDEX2(k,q,p.numEqu)];
351     EM_F[INDEX2(k,s,p.numEqu)]+=rtmp;
352 gross 798 }
353 gross 853 }
354 gross 2748 } else {
355     for (s=0;s<p.row_numShapes;s++) {
356     rtmp=0;
357     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
358     for (k=0;k<p.numEqu;k++) EM_F[INDEX2(k,s,p.numEqu)]+=rtmp*Y_p[k];
359 gross 798 }
360 gross 2748 }
361     }
362    
363     /***********************************************************************************************/
364     /* add the element matrices onto the matrix and right hand side */
365     /***********************************************************************************************/
366     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)]];
367    
368 jfenwick 3086 if (add_EM_F) Dudley_Util_AddScatter(p.row_numShapesTotal,row_index,p.numEqu,EM_F,F_p, p.row_DOF_UpperBound);
369     if (add_EM_S) Dudley_Assemble_addToSystemMatrix(Mat,p.row_numShapesTotal,row_index,p.numEqu,p.col_numShapesTotal,row_index,p.numComp,EM_S);
370 gross 2748 } /* end of subn */
371 gross 853 } /* end color check */
372     } /* end element loop */
373     } /* end color loop */
374    
375     THREAD_MEMFREE(EM_S);
376     THREAD_MEMFREE(EM_F);
377     THREAD_MEMFREE(row_index);
378 gross 798
379 gross 853 } /* end of pointer check */
380 gross 798 } /* end parallel region */
381     }

  ViewVC Help
Powered by ViewVC 1.1.26