/[escript]/branches/doubleplusgood/finley/src/Assemble_PDE_Single2_3D.cpp
ViewVC logotype

Annotation of /branches/doubleplusgood/finley/src/Assemble_PDE_Single2_3D.cpp

Parent Directory Parent Directory | Revision Log Revision Log


Revision 4327 - (hide annotations)
Wed Mar 20 05:09:11 2013 UTC (6 years, 7 months ago) by jfenwick
File size: 20036 byte(s)
some finley memory
1 gross 798
2 jfenwick 3981 /*****************************************************************************
3 ksteube 1811 *
4 jfenwick 4154 * Copyright (c) 2003-2013 by University of Queensland
5 jfenwick 3981 * http://www.uq.edu.au
6 ksteube 1811 *
7     * Primary Business: Queensland, Australia
8     * Licensed under the Open Software License version 3.0
9     * http://www.opensource.org/licenses/osl-3.0.php
10     *
11 jfenwick 3981 * Development until 2012 by Earth Systems Science Computational Center (ESSCC)
12     * Development since 2012 by School of Earth Sciences
13     *
14     *****************************************************************************/
15 ksteube 1312
16 ksteube 1811
17 jfenwick 3981 /************************************************************************************/
18 gross 798
19 caltinay 3639 /* assembles the system of numEq PDEs into the stiffness matrix S and right hand side F */
20 gross 798
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 jfenwick 3981 /************************************************************************************/
37 gross 798
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 jfenwick 3981 /************************************************************************************/
47 gross 798
48 jfenwick 3259 void Finley_Assemble_PDE_Single2_3D(Finley_Assemble_Parameters p,
49     Finley_ElementFile* elements,
50 gross 798 Paso_SystemMatrix* Mat, escriptDataC* F,
51     escriptDataC* A, escriptDataC* B, escriptDataC* C, escriptDataC* D, escriptDataC* X, escriptDataC* Y) {
52    
53     #define DIM 3
54     index_t color;
55 gross 2748 dim_t e, isub;
56     __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;
57 jfenwick 2271 double *EM_S, *EM_F, *Vol, *DSDX;
58 gross 853 index_t *row_index;
59     register dim_t q, s,r;
60     register double rtmp, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22, rtmp0, rtmp1, rtmp2;
61     bool_t add_EM_F, add_EM_S;
62    
63 gross 798 bool_t extendedA=isExpanded(A);
64     bool_t extendedB=isExpanded(B);
65     bool_t extendedC=isExpanded(C);
66     bool_t extendedD=isExpanded(D);
67     bool_t extendedX=isExpanded(X);
68     bool_t extendedY=isExpanded(Y);
69 jfenwick 2271 double *F_p=(requireWrite(F), getSampleDataRW(F,0)); /* use comma, to get around the mixed code and declarations thing */
70 gross 2748 double *S=p.row_jac->BasisFunctions->S;
71     dim_t len_EM_S=p.row_numShapesTotal*p.col_numShapesTotal;
72     dim_t len_EM_F=p.row_numShapesTotal;
73 gross 798
74 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,rtmp, rtmp00, rtmp01, rtmp02, rtmp10, rtmp11, rtmp12, rtmp20, rtmp21, rtmp22, rtmp0, rtmp1, rtmp2,add_EM_F, add_EM_S, isub)
75 gross 798 {
76 jfenwick 4327 EM_S=new double[len_EM_S];
77     EM_F=new double[len_EM_F];
78     row_index=new index_t[p.row_numShapesTotal];
79 gross 853
80     if (!Finley_checkPtr(EM_S) && !Finley_checkPtr(EM_F) && !Finley_checkPtr(row_index) ) {
81 gross 798
82 gross 853 for (color=elements->minColor;color<=elements->maxColor;color++) {
83     /* open loop over all elements: */
84     #pragma omp for private(e) schedule(static)
85     for(e=0;e<elements->numElements;e++){
86     if (elements->Color[e]==color) {
87 gross 2748
88 jfenwick 2770 A_p=getSampleDataRO(A,e);
89     B_p=getSampleDataRO(B,e);
90     C_p=getSampleDataRO(C,e);
91     D_p=getSampleDataRO(D,e);
92     X_p=getSampleDataRO(X,e);
93     Y_p=getSampleDataRO(Y,e);
94 gross 2748
95     for (isub=0; isub<p.numSub; isub++) {
96    
97     Vol=&(p.row_jac->volume[INDEX3(0,isub,e, p.numQuadSub,p.numSub)]);
98     DSDX=&(p.row_jac->DSDX[INDEX5(0,0,0,isub,e,p.row_numShapesTotal,DIM,p.numQuadSub,p.numSub)]);
99     for (q=0;q<len_EM_S;++q) EM_S[q]=0;
100     for (q=0;q<len_EM_F;++q) EM_F[q]=0;
101     add_EM_F=FALSE;
102     add_EM_S=FALSE;
103    
104 jfenwick 3981 /************************************************************************************/
105 gross 2748 /* process A: */
106 jfenwick 3981 /************************************************************************************/
107 gross 2748 if (NULL!=A_p) {
108     add_EM_S=TRUE;
109     if (extendedA) {
110     A_q=&(A_p[INDEX4(0,0,0,isub, DIM,DIM,p.numQuadSub)]);
111     for (s=0;s<p.row_numShapes;s++) {
112     for (r=0;r<p.col_numShapes;r++) {
113     rtmp=0;
114     for (q=0;q<p.numQuadSub;q++) {
115     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)]
116     + 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)]
117     + 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)]
118     + 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)]
119     + 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)]
120     + 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)]
121     + 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)]
122     + 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)]
123     + 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)]);
124     }
125     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
126     }
127     }
128     } else {
129     for (s=0;s<p.row_numShapes;s++) {
130     for (r=0;r<p.col_numShapes;r++) {
131     rtmp00=0;
132     rtmp01=0;
133     rtmp02=0;
134     rtmp10=0;
135     rtmp11=0;
136     rtmp12=0;
137     rtmp20=0;
138     rtmp21=0;
139     rtmp22=0;
140     for (q=0;q<p.numQuadSub;q++) {
141    
142     rtmp0=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
143     rtmp00+=rtmp0*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
144     rtmp01+=rtmp0*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
145     rtmp02+=rtmp0*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
146    
147     rtmp1=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
148     rtmp10+=rtmp1*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
149     rtmp11+=rtmp1*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
150     rtmp12+=rtmp1*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
151    
152     rtmp2=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
153     rtmp20+=rtmp2*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
154     rtmp21+=rtmp2*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
155     rtmp22+=rtmp2*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
156     }
157     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp00*A_p[INDEX2(0,0,DIM)]
158     +rtmp01*A_p[INDEX2(0,1,DIM)]
159     +rtmp02*A_p[INDEX2(0,2,DIM)]
160     +rtmp10*A_p[INDEX2(1,0,DIM)]
161     +rtmp11*A_p[INDEX2(1,1,DIM)]
162     +rtmp12*A_p[INDEX2(1,2,DIM)]
163     +rtmp20*A_p[INDEX2(2,0,DIM)]
164     +rtmp21*A_p[INDEX2(2,1,DIM)]
165     +rtmp22*A_p[INDEX2(2,2,DIM)];
166     }
167     }
168     }
169     }
170 jfenwick 3981 /************************************************************************************/
171 gross 2748 /* process B: */
172 jfenwick 3981 /************************************************************************************/
173 gross 2748 if (NULL!=B_p) {
174     add_EM_S=TRUE;
175     if (extendedB) {
176     B_q=&(B_p[INDEX3(0,0,isub, DIM,p.numQuadSub)]);
177     for (s=0;s<p.row_numShapes;s++) {
178     for (r=0;r<p.col_numShapes;r++) {
179     rtmp=0;
180     for (q=0;q<p.numQuadSub;q++) {
181     rtmp+=Vol[q]*S[INDEX2(r,q,p.row_numShapes)]*
182     ( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(0,q,DIM)]
183     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(1,q,DIM)]
184     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*B_q[INDEX2(2,q,DIM)]);
185     }
186     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
187     }
188     }
189     } else {
190     for (s=0;s<p.row_numShapes;s++) {
191     for (r=0;r<p.col_numShapes;r++) {
192     rtmp0=0;
193     rtmp1=0;
194     rtmp2=0;
195     for (q=0;q<p.numQuadSub;q++) {
196     rtmp=Vol[q]*S[INDEX2(r,q,p.row_numShapes)];
197     rtmp0+=rtmp*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
198     rtmp1+=rtmp*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
199     rtmp2+=rtmp*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
200     }
201     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];
202     }
203     }
204 gross 853 }
205 gross 798 }
206 jfenwick 3981 /************************************************************************************/
207 gross 2748 /* process C: */
208 jfenwick 3981 /************************************************************************************/
209 gross 2748 if (NULL!=C_p) {
210     add_EM_S=TRUE;
211     if (extendedC) {
212     C_q=&(C_p[INDEX3(0,0,isub, DIM, p.numQuadSub)]);
213     for (s=0;s<p.row_numShapes;s++) {
214     for (r=0;r<p.col_numShapes;r++) {
215     rtmp=0;
216     for (q=0;q<p.numQuadSub;q++) {
217     rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*
218     ( C_q[INDEX2(0,q,DIM)]*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)]
219     + C_q[INDEX2(1,q,DIM)]*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)]
220     + C_q[INDEX2(2,q,DIM)]*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)]);
221     }
222     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
223     }
224     }
225     } else {
226     for (s=0;s<p.row_numShapes;s++) {
227     for (r=0;r<p.col_numShapes;r++) {
228     rtmp0=0;
229     rtmp1=0;
230     rtmp2=0;
231     for (q=0;q<p.numQuadSub;q++) {
232     rtmp=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
233     rtmp0+=rtmp*DSDX[INDEX3(r,0,q,p.row_numShapesTotal,DIM)];
234     rtmp1+=rtmp*DSDX[INDEX3(r,1,q,p.row_numShapesTotal,DIM)];
235     rtmp2+=rtmp*DSDX[INDEX3(r,2,q,p.row_numShapesTotal,DIM)];
236     }
237     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];
238     }
239     }
240     }
241     }
242 jfenwick 3981 /*********************************************************************************** */
243 gross 2748 /* process D */
244 jfenwick 3981 /************************************************************************************/
245 gross 2748 if (NULL!=D_p) {
246     add_EM_S=TRUE;
247     if (extendedD) {
248     D_q=&(D_p[INDEX2(0,isub, p.numQuadSub)]);
249     for (s=0;s<p.row_numShapes;s++) {
250     for (r=0;r<p.col_numShapes;r++) {
251     rtmp=0;
252     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)];
253     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp;
254     }
255     }
256     } else {
257     for (s=0;s<p.row_numShapes;s++) {
258     for (r=0;r<p.col_numShapes;r++) {
259     rtmp=0;
260     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*S[INDEX2(r,q,p.row_numShapes)];
261     EM_S[INDEX4(0,0,s,r,p.numEqu,p.numComp,p.row_numShapesTotal)]+=rtmp*D_p[0];
262     }
263     }
264     }
265     }
266 jfenwick 3981 /************************************************************************************/
267 gross 2748 /* process X: */
268 jfenwick 3981 /************************************************************************************/
269 gross 2748 if (NULL!=X_p) {
270     add_EM_F=TRUE;
271     if (extendedX) {
272     X_q=&(X_p[INDEX3(0,0,isub, DIM,p.numQuadSub)]);
273     for (s=0;s<p.row_numShapes;s++) {
274     rtmp=0;
275     for (q=0;q<p.numQuadSub;q++) {
276     rtmp+=Vol[q]*( DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(0,q,DIM)]
277     + DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(1,q,DIM)]
278     + DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)]*X_q[INDEX2(2,q,DIM)]);
279     }
280     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
281 gross 853 }
282 gross 2748 } else {
283     for (s=0;s<p.row_numShapes;s++) {
284     rtmp0=0;
285     rtmp1=0;
286     rtmp2=0;
287     for (q=0;q<p.numQuadSub;q++) {
288     rtmp0+=Vol[q]*DSDX[INDEX3(s,0,q,p.row_numShapesTotal,DIM)];
289     rtmp1+=Vol[q]*DSDX[INDEX3(s,1,q,p.row_numShapesTotal,DIM)];
290     rtmp2+=Vol[q]*DSDX[INDEX3(s,2,q,p.row_numShapesTotal,DIM)];
291     }
292     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp0*X_p[0]+rtmp1*X_p[1]+rtmp2*X_p[2];
293 gross 853 }
294 gross 2748 }
295 gross 853 }
296 jfenwick 3981 /************************************************************************************/
297 gross 2748 /* process Y: */
298 jfenwick 3981 /************************************************************************************/
299 gross 2748 if (NULL!=Y_p) {
300     add_EM_F=TRUE;
301     if (extendedY) {
302     Y_q=&(Y_p[INDEX2(0,isub, p.numQuadSub)]);
303     for (s=0;s<p.row_numShapes;s++) {
304     rtmp=0;
305     for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)]*Y_q[q];
306     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp;
307 gross 853 }
308 gross 2748 } else {
309     for (s=0;s<p.row_numShapes;s++) {
310 gross 853 rtmp=0;
311 gross 2748 for (q=0;q<p.numQuadSub;q++) rtmp+=Vol[q]*S[INDEX2(s,q,p.row_numShapes)];
312     EM_F[INDEX2(0,s,p.numEqu)]+=rtmp*Y_p[0];
313 gross 853 }
314     }
315 gross 2748 }
316 jfenwick 3981 /*********************************************************************************************************************/
317 gross 2748 /* add the element matrices onto the matrix and right hand side */
318 jfenwick 3981 /*********************************************************************************************************************/
319 gross 2748
320     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)]];
321    
322     if (add_EM_F) Finley_Util_AddScatter(p.row_numShapesTotal,row_index,p.numEqu,EM_F,F_p, p.row_DOF_UpperBound);
323     if (add_EM_S) Finley_Assemble_addToSystemMatrix(Mat,p.row_numShapesTotal,row_index,p.numEqu,p.col_numShapesTotal,row_index,p.numComp,EM_S);
324    
325     } /* end of isub */
326 gross 853 } /* end color check */
327     } /* end element loop */
328     } /* end color loop */
329    
330 jfenwick 4327 delete[] EM_S;
331     delete[] EM_F;
332     delete[] row_index;
333 gross 798
334 gross 853 } /* end of pointer check */
335 gross 798 } /* end parallel region */
336     }

  ViewVC Help
Powered by ViewVC 1.1.26