/[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 3184 - (hide annotations)
Wed Sep 15 00:23:42 2010 UTC (8 years, 10 months ago) by jfenwick
File MIME type: text/plain
File size: 19476 byte(s)
Removed volume from ElementFile_Jacobeans

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 3184 double *EM_S, *EM_F, *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 3184 // Vol=&(p.row_jac->volume[INDEX3(0,0,e, p.numQuadTotal,1)]);
95     double vol=p.row_jac->absD[e]*p.row_jac->quadweight;
96 jfenwick 3137 DSDX=&(p.row_jac->DSDX[INDEX5(0,0,0,0,e,p.row_numShapesTotal,DIM,p.numQuadTotal,1)]);
97 gross 2748 for (q=0;q<len_EM_S;++q) EM_S[q]=0;
98     for (q=0;q<len_EM_F;++q) EM_F[q]=0;
99     add_EM_F=FALSE;
100     add_EM_S=FALSE;
101    
102     /**************************************************************/
103     /* process A: */
104     /**************************************************************/
105     if (NULL!=A_p) {
106     add_EM_S=TRUE;
107     if (extendedA) {
108 jfenwick 3137 A_q=&(A_p[INDEX4(0,0,0,0, DIM,DIM,p.numQuadTotal)]);
109 gross 2748 for (s=0;s<p.row_numShapes;s++) {
110     for (r=0;r<p.col_numShapes;r++) {
111     rtmp=0;
112 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
113 jfenwick 3184 rtmp+=vol*( 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 gross 2748 + 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,0,q,p.row_numShapesTotal,DIM)]*A_q[INDEX3(0,2,q,DIM,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]
116     + 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)]
117     + 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)]
118     + 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)]
119     + 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)]
120     + 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)]
121     + 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)]);
122     }
123     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
124     }
125     }
126     } else {
127     for (s=0;s<p.row_numShapes;s++) {
128     for (r=0;r<p.col_numShapes;r++) {
129     rtmp00=0;
130     rtmp01=0;
131     rtmp02=0;
132     rtmp10=0;
133     rtmp11=0;
134     rtmp12=0;
135     rtmp20=0;
136     rtmp21=0;
137     rtmp22=0;
138 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
139 gross 2748
140 jfenwick 3184 rtmp0=vol*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
141 gross 2748 rtmp00+=rtmp0*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
142     rtmp01+=rtmp0*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
143     rtmp02+=rtmp0*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
144    
145 jfenwick 3184 rtmp1=vol*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
146 gross 2748 rtmp10+=rtmp1*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
147     rtmp11+=rtmp1*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
148     rtmp12+=rtmp1*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
149    
150 jfenwick 3184 rtmp2=vol*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
151 gross 2748 rtmp20+=rtmp2*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
152     rtmp21+=rtmp2*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
153     rtmp22+=rtmp2*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
154     }
155     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp00*A_p[INDEX2(0,0,DIM)]
156     +rtmp01*A_p[INDEX2(0,1,DIM)]
157     +rtmp02*A_p[INDEX2(0,2,DIM)]
158     +rtmp10*A_p[INDEX2(1,0,DIM)]
159     +rtmp11*A_p[INDEX2(1,1,DIM)]
160     +rtmp12*A_p[INDEX2(1,2,DIM)]
161     +rtmp20*A_p[INDEX2(2,0,DIM)]
162     +rtmp21*A_p[INDEX2(2,1,DIM)]
163     +rtmp22*A_p[INDEX2(2,2,DIM)];
164     }
165     }
166     }
167     }
168     /**************************************************************/
169     /* process B: */
170     /**************************************************************/
171     if (NULL!=B_p) {
172     add_EM_S=TRUE;
173     if (extendedB) {
174 jfenwick 3137 B_q=&(B_p[INDEX3(0,0,0, DIM,p.numQuadTotal)]);
175 gross 2748 for (s=0;s<p.row_numShapes;s++) {
176     for (r=0;r<p.col_numShapes;r++) {
177     rtmp=0;
178 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
179 jfenwick 3184 rtmp+=vol*S[INDEX2(r,q,p.row_numShapes)]*
180 gross 2748 ( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(0,q,DIM)]
181     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(1,q,DIM)]
182     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(2,q,DIM)]);
183     }
184     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
185     }
186     }
187     } else {
188     for (s=0;s<p.row_numShapes;s++) {
189     for (r=0;r<p.col_numShapes;r++) {
190     rtmp0=0;
191     rtmp1=0;
192     rtmp2=0;
193 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
194 jfenwick 3184 rtmp=vol*S[INDEX2(r,q,p.row_numShapes)];
195 gross 2748 rtmp0+=rtmp*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
196     rtmp1+=rtmp*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
197     rtmp2+=rtmp*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
198     }
199     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];
200     }
201     }
202 gross 853 }
203 gross 798 }
204 gross 2748 /**************************************************************/
205     /* process C: */
206     /**************************************************************/
207     if (NULL!=C_p) {
208     add_EM_S=TRUE;
209     if (extendedC) {
210 jfenwick 3137 C_q=&(C_p[INDEX3(0,0,0, DIM, p.numQuadTotal)]);
211 gross 2748 for (s=0;s<p.row_numShapes;s++) {
212     for (r=0;r<p.col_numShapes;r++) {
213     rtmp=0;
214 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
215 jfenwick 3184 rtmp+=vol*S[INDEX2(s,q,p.row_numShapes)]*
216 gross 2748 ( C_q[INDEX2(0,q,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
217     + C_q[INDEX2(1,q,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
218     + C_q[INDEX2(2,q,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
219     }
220     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
221     }
222     }
223     } else {
224     for (s=0;s<p.row_numShapes;s++) {
225     for (r=0;r<p.col_numShapes;r++) {
226     rtmp0=0;
227     rtmp1=0;
228     rtmp2=0;
229 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
230 jfenwick 3184 rtmp=vol*S[INDEX2(s,q,p.row_numShapes)];
231 gross 2748 rtmp0+=rtmp*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
232     rtmp1+=rtmp*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
233     rtmp2+=rtmp*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
234     }
235     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];
236     }
237     }
238     }
239     }
240     /************************************************************* */
241     /* process D */
242     /**************************************************************/
243     if (NULL!=D_p) {
244     add_EM_S=TRUE;
245     if (extendedD) {
246 jfenwick 3137 D_q=&(D_p[INDEX2(0,0, p.numQuadTotal)]);
247 gross 2748 for (s=0;s<p.row_numShapes;s++) {
248     for (r=0;r<p.col_numShapes;r++) {
249     rtmp=0;
250 jfenwick 3184 for (q=0;q<p.numQuadTotal;q++) rtmp+=vol*S[INDEX2(s,q,p.row_numShapes)]*D_q[q]*S[INDEX2(r,q,p.row_numShapes)];
251 gross 2748 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
252     }
253     }
254     } else {
255     for (s=0;s<p.row_numShapes;s++) {
256     for (r=0;r<p.col_numShapes;r++) {
257     rtmp=0;
258 jfenwick 3184 for (q=0;q<p.numQuadTotal;q++) rtmp+=vol*S[INDEX2(s,q,p.row_numShapes)]*S[INDEX2(r,q,p.row_numShapes)];
259 gross 2748 EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp*D_p[0];
260     }
261     }
262     }
263     }
264     /**************************************************************/
265     /* process X: */
266     /**************************************************************/
267     if (NULL!=X_p) {
268     add_EM_F=TRUE;
269     if (extendedX) {
270 jfenwick 3137 X_q=&(X_p[INDEX3(0,0,0, DIM,p.numQuadTotal)]);
271 gross 2748 for (s=0;s<p.row_numShapes;s++) {
272     rtmp=0;
273 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
274 jfenwick 3184 rtmp+=vol*( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(0,q,DIM)]
275 gross 2748 + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(1,q,DIM)]
276     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(2,q,DIM)]);
277     }
278     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
279 gross 853 }
280 gross 2748 } else {
281     for (s=0;s<p.row_numShapes;s++) {
282     rtmp0=0;
283     rtmp1=0;
284     rtmp2=0;
285 jfenwick 3137 for (q=0;q<p.numQuadTotal;q++) {
286 jfenwick 3184 rtmp0+=vol*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
287     rtmp1+=vol*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
288     rtmp2+=vol*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
289 gross 2748 }
290     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp0*X_p[0]+rtmp1*X_p[1]+rtmp2*X_p[2];
291 gross 853 }
292 gross 2748 }
293 gross 853 }
294 gross 2748 /**************************************************************/
295     /* process Y: */
296     /**************************************************************/
297     if (NULL!=Y_p) {
298     add_EM_F=TRUE;
299     if (extendedY) {
300 jfenwick 3137 Y_q=&(Y_p[INDEX2(0,0, p.numQuadTotal)]);
301 gross 2748 for (s=0;s<p.row_numShapes;s++) {
302     rtmp=0;
303 jfenwick 3184 for (q=0;q<p.numQuadTotal;q++) rtmp+=vol*S[INDEX2(s,q,p.row_numShapes)]*Y_q[q];
304 gross 2748 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
305 gross 853 }
306 gross 2748 } else {
307     for (s=0;s<p.row_numShapes;s++) {
308 gross 853 rtmp=0;
309 jfenwick 3184 for (q=0;q<p.numQuadTotal;q++) rtmp+=vol*S[INDEX2(s,q,p.row_numShapes)];
310 gross 2748 EM_F[INDEX2(0,s,p.numEqu)]+=rtmp*Y_p[0];
311 gross 853 }
312     }
313 gross 2748 }
314     /***********************************************************************************************/
315     /* add the element matrices onto the matrix and right hand side */
316     /***********************************************************************************************/
317    
318 jfenwick 3144 for (q=0;q<p.row_numShapesTotal;q++) row_index[q]=p.row_DOF[elements->Nodes[INDEX2(q,e,p.NN)]];
319 gross 2748
320 jfenwick 3086 if (add_EM_F) Dudley_Util_AddScatter(p.row_numShapesTotal,row_index,p.numEqu,EM_F,F_p, p.row_DOF_UpperBound);
321     if (add_EM_S) Dudley_Assemble_addToSystemMatrix(Mat,p.row_numShapesTotal,row_index,p.numEqu,p.col_numShapesTotal,row_index,p.numComp,EM_S);
322 gross 2748
323 gross 853 } /* end color check */
324     } /* end element loop */
325     } /* end color loop */
326    
327     THREAD_MEMFREE(EM_S);
328     THREAD_MEMFREE(EM_F);
329     THREAD_MEMFREE(row_index);
330 gross 798
331 gross 853 } /* end of pointer check */
332 gross 798 } /* end parallel region */
333     }

  ViewVC Help
Powered by ViewVC 1.1.26