/[escript]/branches/domexper/dudley/src/Assemble_PDE_Single2_3D.c
ViewVC logotype

Annotation of /branches/domexper/dudley/src/Assemble_PDE_Single2_3D.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3136 - (hide annotations)
Thu Sep 2 03:48:29 2010 UTC (8 years, 8 months ago) by jfenwick
File MIME type: text/plain
File size: 19473 byte(s)
Removed numSubElements
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_{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 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 = 3 x 3 */
29     /* B = 3 */
30     /* C = 3 */
31     /* D = scalar */
32     /* X = 3 */
33     /* Y = scalar */
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_Single2_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 jfenwick 3136 dim_t e;
55 gross 2748 __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;
59     register double rtmp, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22, rtmp0, rtmp1, rtmp2;
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;
71     dim_t len_EM_F=p.row_numShapesTotal;
72 gross 798
73 jfenwick 3136 #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,rtmp, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22, rtmp0, rtmp1, rtmp2,add_EM_F, add_EM_S)
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 jfenwick 3136 Vol=&(p.row_jac->volume[INDEX3(0,0,e, p.numQuadSub,1)]);
95     DSDX=&(p.row_jac->DSDX[INDEX5(0,0,0,0,e,p.row_numShapesTotal,DIM,p.numQuadSub,1)]);
96 gross 2748 for (q=0;q<len_EM_S;++q) EM_S[q]=0;
97     for (q=0;q<len_EM_F;++q) EM_F[q]=0;
98     add_EM_F=FALSE;
99     add_EM_S=FALSE;
100    
101     /**************************************************************/
102     /* process A: */
103     /**************************************************************/
104     if (NULL!=A_p) {
105     add_EM_S=TRUE;
106     if (extendedA) {
107 jfenwick 3136 A_q=&(A_p[INDEX4(0,0,0,0, DIM,DIM,p.numQuadSub)]);
108 gross 2748 for (s=0;s<p.row_numShapes;s++) {
109     for (r=0;r<p.col_numShapes;r++) {
110     rtmp=0;
111     for (q=0;q<p.numQuadSub;q++) {
112     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)]
113     + 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)]
114     + DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(0,2,q,DIM,DIM)]*DSDX[INDEX3(r,2,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     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(1,2,q,DIM,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]
118     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(2,0,q,DIM,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
119     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(2,1,q,DIM,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
120     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(2,2,q,DIM,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
121     }
122     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
123     }
124     }
125     } else {
126     for (s=0;s<p.row_numShapes;s++) {
127     for (r=0;r<p.col_numShapes;r++) {
128     rtmp00=0;
129     rtmp01=0;
130     rtmp02=0;
131     rtmp10=0;
132     rtmp11=0;
133     rtmp12=0;
134     rtmp20=0;
135     rtmp21=0;
136     rtmp22=0;
137     for (q=0;q<p.numQuadSub;q++) {
138    
139     rtmp0=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
140     rtmp00+=rtmp0*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
141     rtmp01+=rtmp0*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
142     rtmp02+=rtmp0*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
143    
144     rtmp1=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
145     rtmp10+=rtmp1*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
146     rtmp11+=rtmp1*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
147     rtmp12+=rtmp1*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
148    
149     rtmp2=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
150     rtmp20+=rtmp2*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
151     rtmp21+=rtmp2*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
152     rtmp22+=rtmp2*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
153     }
154     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp00*A_p[INDEX2(0,0,DIM)]
155     +rtmp01*A_p[INDEX2(0,1,DIM)]
156     +rtmp02*A_p[INDEX2(0,2,DIM)]
157     +rtmp10*A_p[INDEX2(1,0,DIM)]
158     +rtmp11*A_p[INDEX2(1,1,DIM)]
159     +rtmp12*A_p[INDEX2(1,2,DIM)]
160     +rtmp20*A_p[INDEX2(2,0,DIM)]
161     +rtmp21*A_p[INDEX2(2,1,DIM)]
162     +rtmp22*A_p[INDEX2(2,2,DIM)];
163     }
164     }
165     }
166     }
167     /**************************************************************/
168     /* process B: */
169     /**************************************************************/
170     if (NULL!=B_p) {
171     add_EM_S=TRUE;
172     if (extendedB) {
173 jfenwick 3136 B_q=&(B_p[INDEX3(0,0,0, DIM,p.numQuadSub)]);
174 gross 2748 for (s=0;s<p.row_numShapes;s++) {
175     for (r=0;r<p.col_numShapes;r++) {
176     rtmp=0;
177     for (q=0;q<p.numQuadSub;q++) {
178     rtmp+=Vol[q]*S[INDEX2(r,q,p.row_numShapes)]*
179     ( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(0,q,DIM)]
180     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(1,q,DIM)]
181     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(2,q,DIM)]);
182     }
183     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
184     }
185     }
186     } else {
187     for (s=0;s<p.row_numShapes;s++) {
188     for (r=0;r<p.col_numShapes;r++) {
189     rtmp0=0;
190     rtmp1=0;
191     rtmp2=0;
192     for (q=0;q<p.numQuadSub;q++) {
193     rtmp=Vol[q]*S[INDEX2(r,q,p.row_numShapes)];
194     rtmp0+=rtmp*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
195     rtmp1+=rtmp*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
196     rtmp2+=rtmp*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
197     }
198     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*B_p[0]+rtmp1*B_p[1]+rtmp2*B_p[2];
199     }
200     }
201 gross 853 }
202 gross 798 }
203 gross 2748 /**************************************************************/
204     /* process C: */
205     /**************************************************************/
206     if (NULL!=C_p) {
207     add_EM_S=TRUE;
208     if (extendedC) {
209 jfenwick 3136 C_q=&(C_p[INDEX3(0,0,0, DIM, p.numQuadSub)]);
210 gross 2748 for (s=0;s<p.row_numShapes;s++) {
211     for (r=0;r<p.col_numShapes;r++) {
212     rtmp=0;
213     for (q=0;q<p.numQuadSub;q++) {
214     rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*
215     ( C_q[INDEX2(0,q,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
216     + C_q[INDEX2(1,q,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
217     + C_q[INDEX2(2,q,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
218     }
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     rtmp0=0;
226     rtmp1=0;
227     rtmp2=0;
228     for (q=0;q<p.numQuadSub;q++) {
229     rtmp=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
230     rtmp0+=rtmp*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
231     rtmp1+=rtmp*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
232     rtmp2+=rtmp*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
233     }
234     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp0*C_p[0]+rtmp1*C_p[1]+rtmp2*C_p[2];
235     }
236     }
237     }
238     }
239     /************************************************************* */
240     /* process D */
241     /**************************************************************/
242     if (NULL!=D_p) {
243     add_EM_S=TRUE;
244     if (extendedD) {
245 jfenwick 3136 D_q=&(D_p[INDEX2(0,0, p.numQuadSub)]);
246 gross 2748 for (s=0;s<p.row_numShapes;s++) {
247     for (r=0;r<p.col_numShapes;r++) {
248     rtmp=0;
249     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)];
250     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
251     }
252     }
253     } else {
254     for (s=0;s<p.row_numShapes;s++) {
255     for (r=0;r<p.col_numShapes;r++) {
256     rtmp=0;
257     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*S[INDEX2(r,q,p.row_numShapes)];
258     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp*D_p[0];
259     }
260     }
261     }
262     }
263     /**************************************************************/
264     /* process X: */
265     /**************************************************************/
266     if (NULL!=X_p) {
267     add_EM_F=TRUE;
268     if (extendedX) {
269 jfenwick 3136 X_q=&(X_p[INDEX3(0,0,0, DIM,p.numQuadSub)]);
270 gross 2748 for (s=0;s<p.row_numShapes;s++) {
271     rtmp=0;
272     for (q=0;q<p.numQuadSub;q++) {
273     rtmp+=Vol[q]*( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(0,q,DIM)]
274     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(1,q,DIM)]
275     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(2,q,DIM)]);
276     }
277     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
278 gross 853 }
279 gross 2748 } else {
280     for (s=0;s<p.row_numShapes;s++) {
281     rtmp0=0;
282     rtmp1=0;
283     rtmp2=0;
284     for (q=0;q<p.numQuadSub;q++) {
285     rtmp0+=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
286     rtmp1+=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
287     rtmp2+=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
288     }
289     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp0*X_p[0]+rtmp1*X_p[1]+rtmp2*X_p[2];
290 gross 853 }
291 gross 2748 }
292 gross 853 }
293 gross 2748 /**************************************************************/
294     /* process Y: */
295     /**************************************************************/
296     if (NULL!=Y_p) {
297     add_EM_F=TRUE;
298     if (extendedY) {
299 jfenwick 3136 Y_q=&(Y_p[INDEX2(0,0, p.numQuadSub)]);
300 gross 2748 for (s=0;s<p.row_numShapes;s++) {
301     rtmp=0;
302     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*Y_q[q];
303     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
304 gross 853 }
305 gross 2748 } else {
306     for (s=0;s<p.row_numShapes;s++) {
307 gross 853 rtmp=0;
308 gross 2748 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
309     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp*Y_p[0];
310 gross 853 }
311     }
312 gross 2748 }
313     /***********************************************************************************************/
314     /* add the element matrices onto the matrix and right hand side */
315     /***********************************************************************************************/
316    
317 jfenwick 3136 for (q=0;q<p.row_numShapesTotal;q++) row_index[q]=p.row_DOF[elements->Nodes[INDEX2(p.row_node[INDEX2(q,0,p.row_numShapesTotal)],e,p.NN)]];
318 gross 2748
319 jfenwick 3086 if (add_EM_F) Dudley_Util_AddScatter(p.row_numShapesTotal,row_index,p.numEqu,EM_F,F_p, p.row_DOF_UpperBound);
320     if (add_EM_S) Dudley_Assemble_addToSystemMatrix(Mat,p.row_numShapesTotal,row_index,p.numEqu,p.col_numShapesTotal,row_index,p.numComp,EM_S);
321 gross 2748
322 gross 853 } /* end color check */
323     } /* end element loop */
324     } /* end color loop */
325    
326     THREAD_MEMFREE(EM_S);
327     THREAD_MEMFREE(EM_F);
328     THREAD_MEMFREE(row_index);
329 gross 798
330 gross 853 } /* end of pointer check */
331 gross 798 } /* end parallel region */
332     }

  ViewVC Help
Powered by ViewVC 1.1.26