/[escript]/trunk/ripley/src/Rectangle.cpp
ViewVC logotype

Annotation of /trunk/ripley/src/Rectangle.cpp

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3759 - (hide annotations)
Fri Jan 6 06:54:51 2012 UTC (7 years, 10 months ago) by caltinay
Original Path: branches/ripleygmg_from_3668/ripley/src/Rectangle.cpp
File size: 105801 byte(s)
Implemented ownSample for face elements.

1 caltinay 3691
2     /*******************************************************
3     *
4     * Copyright (c) 2003-2011 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     #include <ripley/Rectangle.h>
15     extern "C" {
16 caltinay 3748 #include "paso/SystemMatrix.h"
17 caltinay 3691 }
18    
19     #if USE_SILO
20     #include <silo.h>
21     #ifdef ESYS_MPI
22     #include <pmpio.h>
23     #endif
24     #endif
25    
26     #include <iomanip>
27    
28     using namespace std;
29    
30     namespace ripley {
31    
32     Rectangle::Rectangle(int n0, int n1, double l0, double l1, int d0, int d1) :
33     RipleyDomain(2),
34     m_gNE0(n0),
35     m_gNE1(n1),
36     m_l0(l0),
37     m_l1(l1),
38     m_NX(d0),
39     m_NY(d1)
40     {
41     // ensure number of subdivisions is valid and nodes can be distributed
42     // among number of ranks
43     if (m_NX*m_NY != m_mpiInfo->size)
44     throw RipleyException("Invalid number of spatial subdivisions");
45    
46 caltinay 3752 if ((n0+1)%m_NX > 0 || (n1+1)%m_NY > 0)
47     throw RipleyException("Number of elements+1 must be separable into number of ranks in each dimension");
48 caltinay 3691
49 caltinay 3752 if ((m_NX > 1 && (n0+1)/m_NX<2) || (m_NY > 1 && (n1+1)/m_NY<2))
50     throw RipleyException("Too few elements for the number of ranks");
51    
52     // local number of elements (including overlap)
53     m_NE0 = (m_NX>1 ? (n0+1)/m_NX : n0);
54     if (m_mpiInfo->rank%m_NX>0 && m_mpiInfo->rank%m_NX<m_NX-1)
55     m_NE0++;
56     m_NE1 = (m_NY>1 ? (n1+1)/m_NY : n1);
57     if (m_mpiInfo->rank/m_NX>0 && m_mpiInfo->rank/m_NX<m_NY-1)
58     m_NE1++;
59    
60     // local number of nodes
61 caltinay 3691 m_N0 = m_NE0+1;
62     m_N1 = m_NE1+1;
63 caltinay 3752
64 caltinay 3691 // bottom-left node is at (offset0,offset1) in global mesh
65 caltinay 3752 m_offset0 = (n0+1)/m_NX*(m_mpiInfo->rank%m_NX);
66 caltinay 3753 if (m_offset0 > 0)
67 caltinay 3752 m_offset0--;
68     m_offset1 = (n1+1)/m_NY*(m_mpiInfo->rank/m_NX);
69 caltinay 3753 if (m_offset1 > 0)
70 caltinay 3752 m_offset1--;
71    
72 caltinay 3691 populateSampleIds();
73 caltinay 3756 createPattern();
74 caltinay 3691 }
75    
76     Rectangle::~Rectangle()
77     {
78     }
79    
80     string Rectangle::getDescription() const
81     {
82     return "ripley::Rectangle";
83     }
84    
85     bool Rectangle::operator==(const AbstractDomain& other) const
86     {
87 caltinay 3744 const Rectangle* o=dynamic_cast<const Rectangle*>(&other);
88     if (o) {
89     return (RipleyDomain::operator==(other) &&
90     m_gNE0==o->m_gNE0 && m_gNE1==o->m_gNE1
91     && m_l0==o->m_l0 && m_l1==o->m_l1
92     && m_NX==o->m_NX && m_NY==o->m_NY);
93     }
94 caltinay 3691
95     return false;
96     }
97    
98     void Rectangle::dump(const string& fileName) const
99     {
100     #if USE_SILO
101     string fn(fileName);
102     if (fileName.length() < 6 || fileName.compare(fileName.length()-5, 5, ".silo") != 0) {
103     fn+=".silo";
104     }
105    
106     const int NUM_SILO_FILES = 1;
107     const char* blockDirFmt = "/block%04d";
108     int driver=DB_HDF5;
109     string siloPath;
110     DBfile* dbfile = NULL;
111    
112     #ifdef ESYS_MPI
113     PMPIO_baton_t* baton = NULL;
114     #endif
115    
116     if (m_mpiInfo->size > 1) {
117     #ifdef ESYS_MPI
118     baton = PMPIO_Init(NUM_SILO_FILES, PMPIO_WRITE, m_mpiInfo->comm,
119     0x1337, PMPIO_DefaultCreate, PMPIO_DefaultOpen,
120     PMPIO_DefaultClose, (void*)&driver);
121     // try the fallback driver in case of error
122     if (!baton && driver != DB_PDB) {
123     driver = DB_PDB;
124     baton = PMPIO_Init(NUM_SILO_FILES, PMPIO_WRITE, m_mpiInfo->comm,
125     0x1338, PMPIO_DefaultCreate, PMPIO_DefaultOpen,
126     PMPIO_DefaultClose, (void*)&driver);
127     }
128     if (baton) {
129     char str[64];
130     snprintf(str, 64, blockDirFmt, PMPIO_RankInGroup(baton, m_mpiInfo->rank));
131     siloPath = str;
132     dbfile = (DBfile*) PMPIO_WaitForBaton(baton, fn.c_str(), siloPath.c_str());
133     }
134     #endif
135     } else {
136     dbfile = DBCreate(fn.c_str(), DB_CLOBBER, DB_LOCAL,
137     getDescription().c_str(), driver);
138     // try the fallback driver in case of error
139     if (!dbfile && driver != DB_PDB) {
140     driver = DB_PDB;
141     dbfile = DBCreate(fn.c_str(), DB_CLOBBER, DB_LOCAL,
142     getDescription().c_str(), driver);
143     }
144     }
145    
146     if (!dbfile)
147     throw RipleyException("dump: Could not create Silo file");
148    
149     /*
150     if (driver==DB_HDF5) {
151     // gzip level 1 already provides good compression with minimal
152     // performance penalty. Some tests showed that gzip levels >3 performed
153     // rather badly on escript data both in terms of time and space
154     DBSetCompression("ERRMODE=FALLBACK METHOD=GZIP LEVEL=1");
155     }
156     */
157    
158     boost::scoped_ptr<double> x(new double[m_N0]);
159     boost::scoped_ptr<double> y(new double[m_N1]);
160     double* coords[2] = { x.get(), y.get() };
161 caltinay 3697 pair<double,double> xdx = getFirstCoordAndSpacing(0);
162     pair<double,double> ydy = getFirstCoordAndSpacing(1);
163 caltinay 3691 #pragma omp parallel
164     {
165 caltinay 3722 #pragma omp for nowait
166 caltinay 3691 for (dim_t i0 = 0; i0 < m_N0; i0++) {
167 caltinay 3697 coords[0][i0]=xdx.first+i0*xdx.second;
168 caltinay 3691 }
169 caltinay 3722 #pragma omp for nowait
170 caltinay 3691 for (dim_t i1 = 0; i1 < m_N1; i1++) {
171 caltinay 3697 coords[1][i1]=ydy.first+i1*ydy.second;
172 caltinay 3691 }
173     }
174 caltinay 3697 IndexVector dims = getNumNodesPerDim();
175    
176     // write mesh
177     DBPutQuadmesh(dbfile, "mesh", NULL, coords, &dims[0], 2, DB_DOUBLE,
178 caltinay 3691 DB_COLLINEAR, NULL);
179    
180 caltinay 3697 // write node ids
181     DBPutQuadvar1(dbfile, "nodeId", "mesh", (void*)&m_nodeId[0], &dims[0], 2,
182     NULL, 0, DB_INT, DB_NODECENT, NULL);
183    
184     // write element ids
185     dims = getNumElementsPerDim();
186     DBPutQuadvar1(dbfile, "elementId", "mesh", (void*)&m_elementId[0],
187     &dims[0], 2, NULL, 0, DB_INT, DB_ZONECENT, NULL);
188    
189     // rank 0 writes multimesh and multivar
190 caltinay 3691 if (m_mpiInfo->rank == 0) {
191     vector<string> tempstrings;
192 caltinay 3697 vector<char*> names;
193 caltinay 3691 for (dim_t i=0; i<m_mpiInfo->size; i++) {
194     stringstream path;
195     path << "/block" << setw(4) << setfill('0') << right << i << "/mesh";
196     tempstrings.push_back(path.str());
197 caltinay 3697 names.push_back((char*)tempstrings.back().c_str());
198 caltinay 3691 }
199 caltinay 3697 vector<int> types(m_mpiInfo->size, DB_QUAD_RECT);
200 caltinay 3691 DBSetDir(dbfile, "/");
201 caltinay 3697 DBPutMultimesh(dbfile, "multimesh", m_mpiInfo->size, &names[0],
202     &types[0], NULL);
203     tempstrings.clear();
204     names.clear();
205     for (dim_t i=0; i<m_mpiInfo->size; i++) {
206     stringstream path;
207     path << "/block" << setw(4) << setfill('0') << right << i << "/nodeId";
208     tempstrings.push_back(path.str());
209     names.push_back((char*)tempstrings.back().c_str());
210     }
211     types.assign(m_mpiInfo->size, DB_QUADVAR);
212     DBPutMultivar(dbfile, "nodeId", m_mpiInfo->size, &names[0],
213     &types[0], NULL);
214     tempstrings.clear();
215     names.clear();
216     for (dim_t i=0; i<m_mpiInfo->size; i++) {
217     stringstream path;
218     path << "/block" << setw(4) << setfill('0') << right << i << "/elementId";
219     tempstrings.push_back(path.str());
220     names.push_back((char*)tempstrings.back().c_str());
221     }
222     DBPutMultivar(dbfile, "elementId", m_mpiInfo->size, &names[0],
223     &types[0], NULL);
224 caltinay 3691 }
225    
226     if (m_mpiInfo->size > 1) {
227     #ifdef ESYS_MPI
228     PMPIO_HandOffBaton(baton, dbfile);
229     PMPIO_Finish(baton);
230     #endif
231     } else {
232     DBClose(dbfile);
233     }
234    
235     #else // USE_SILO
236     throw RipleyException("dump(): no Silo support");
237     #endif
238     }
239    
240 caltinay 3697 const int* Rectangle::borrowSampleReferenceIDs(int fsType) const
241 caltinay 3691 {
242 caltinay 3697 switch (fsType) {
243 caltinay 3691 case Nodes:
244 caltinay 3748 case ReducedNodes: //FIXME: reduced
245 caltinay 3691 return &m_nodeId[0];
246 caltinay 3750 case DegreesOfFreedom:
247     case ReducedDegreesOfFreedom: //FIXME: reduced
248     return &m_dofId[0];
249 caltinay 3691 case Elements:
250 caltinay 3733 case ReducedElements:
251 caltinay 3691 return &m_elementId[0];
252     case FaceElements:
253 caltinay 3733 case ReducedFaceElements:
254 caltinay 3691 return &m_faceId[0];
255     default:
256     break;
257     }
258    
259     stringstream msg;
260     msg << "borrowSampleReferenceIDs() not implemented for function space type "
261 caltinay 3701 << functionSpaceTypeAsString(fsType);
262 caltinay 3691 throw RipleyException(msg.str());
263     }
264    
265 caltinay 3757 bool Rectangle::ownSample(int fsType, index_t id) const
266 caltinay 3691 {
267 caltinay 3759 if (getMPISize()==1)
268     return true;
269    
270 caltinay 3757 switch (fsType) {
271     case Nodes:
272     case ReducedNodes: //FIXME: reduced
273     return (m_dofMap[id] < getNumDOF());
274     case DegreesOfFreedom:
275     case ReducedDegreesOfFreedom:
276     return true;
277     case Elements:
278     case ReducedElements:
279     // check ownership of element's bottom left node
280     return (m_dofMap[id%m_NE0+m_N0*(id/m_NE0)] < getNumDOF());
281     case FaceElements:
282     case ReducedFaceElements:
283 caltinay 3759 {
284     // check ownership of face element's first node
285     const IndexVector faces = getNumFacesPerBoundary();
286     dim_t n=0;
287     for (size_t i=0; i<faces.size(); i++) {
288     n+=faces[i];
289     if (id<n) {
290     index_t k;
291     if (i==1)
292     k=m_N0-1;
293     else if (i==3)
294     k=m_N0*(m_N1-1);
295     else
296     k=0;
297     // determine whether to move right or up
298     const index_t delta=(i/2==0 ? m_N0 : 1);
299     return (m_dofMap[k+(id-n+faces[i])*delta] < getNumDOF());
300     }
301     }
302     return false;
303     }
304 caltinay 3757 default:
305     break;
306 caltinay 3702 }
307 caltinay 3757
308     stringstream msg;
309     msg << "ownSample() not implemented for "
310     << functionSpaceTypeAsString(fsType);
311     throw RipleyException(msg.str());
312 caltinay 3691 }
313    
314 caltinay 3702 void Rectangle::setToGradient(escript::Data& out, const escript::Data& cIn) const
315 caltinay 3691 {
316 caltinay 3702 escript::Data& in = *const_cast<escript::Data*>(&cIn);
317 caltinay 3701 const dim_t numComp = in.getDataPointSize();
318 caltinay 3702 const double h0 = m_l0/m_gNE0;
319     const double h1 = m_l1/m_gNE1;
320 caltinay 3724 const double cx0 = -1./h0;
321     const double cx1 = -.78867513459481288225/h0;
322     const double cx2 = -.5/h0;
323     const double cx3 = -.21132486540518711775/h0;
324     const double cx4 = .21132486540518711775/h0;
325     const double cx5 = .5/h0;
326     const double cx6 = .78867513459481288225/h0;
327     const double cx7 = 1./h0;
328     const double cy0 = -1./h1;
329     const double cy1 = -.78867513459481288225/h1;
330     const double cy2 = -.5/h1;
331     const double cy3 = -.21132486540518711775/h1;
332     const double cy4 = .21132486540518711775/h1;
333     const double cy5 = .5/h1;
334     const double cy6 = .78867513459481288225/h1;
335     const double cy7 = 1./h1;
336    
337 caltinay 3702 if (out.getFunctionSpace().getTypeCode() == Elements) {
338 caltinay 3724 /*** GENERATOR SNIP_GRAD_ELEMENTS TOP */
339 caltinay 3697 #pragma omp parallel for
340 caltinay 3707 for (index_t k1=0; k1 < m_NE1; ++k1) {
341     for (index_t k0=0; k0 < m_NE0; ++k0) {
342 caltinay 3702 const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,k1, m_N0));
343     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,k1+1, m_N0));
344     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,k1+1, m_N0));
345     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,k1, m_N0));
346     double* o = out.getSampleDataRW(INDEX2(k0,k1,m_NE0));
347     for (index_t i=0; i < numComp; ++i) {
348 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_00[i]*cx1 + f_01[i]*cx3 + f_10[i]*cx6 + f_11[i]*cx4;
349     o[INDEX3(i,1,0,numComp,2)] = f_00[i]*cy1 + f_01[i]*cy6 + f_10[i]*cy3 + f_11[i]*cy4;
350     o[INDEX3(i,0,1,numComp,2)] = f_00[i]*cx1 + f_01[i]*cx3 + f_10[i]*cx6 + f_11[i]*cx4;
351     o[INDEX3(i,1,1,numComp,2)] = f_00[i]*cy3 + f_01[i]*cy4 + f_10[i]*cy1 + f_11[i]*cy6;
352     o[INDEX3(i,0,2,numComp,2)] = f_00[i]*cx3 + f_01[i]*cx1 + f_10[i]*cx4 + f_11[i]*cx6;
353     o[INDEX3(i,1,2,numComp,2)] = f_00[i]*cy1 + f_01[i]*cy6 + f_10[i]*cy3 + f_11[i]*cy4;
354     o[INDEX3(i,0,3,numComp,2)] = f_00[i]*cx3 + f_01[i]*cx1 + f_10[i]*cx4 + f_11[i]*cx6;
355     o[INDEX3(i,1,3,numComp,2)] = f_00[i]*cy3 + f_01[i]*cy4 + f_10[i]*cy1 + f_11[i]*cy6;
356 caltinay 3702 } /* end of component loop i */
357     } /* end of k0 loop */
358     } /* end of k1 loop */
359     /* GENERATOR SNIP_GRAD_ELEMENTS BOTTOM */
360 caltinay 3711 } else if (out.getFunctionSpace().getTypeCode() == ReducedElements) {
361 caltinay 3724 /*** GENERATOR SNIP_GRAD_REDUCED_ELEMENTS TOP */
362 caltinay 3711 #pragma omp parallel for
363     for (index_t k1=0; k1 < m_NE1; ++k1) {
364     for (index_t k0=0; k0 < m_NE0; ++k0) {
365     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,k1, m_N0));
366     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,k1+1, m_N0));
367     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,k1+1, m_N0));
368     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,k1, m_N0));
369     double* o = out.getSampleDataRW(INDEX2(k0,k1,m_NE0));
370     for (index_t i=0; i < numComp; ++i) {
371 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = cx5*(f_10[i] + f_11[i]) + cx2*(f_00[i] + f_01[i]);
372     o[INDEX3(i,1,0,numComp,2)] = cy2*(f_00[i] + f_10[i]) + cy5*(f_01[i] + f_11[i]);
373 caltinay 3711 } /* end of component loop i */
374     } /* end of k0 loop */
375     } /* end of k1 loop */
376     /* GENERATOR SNIP_GRAD_REDUCED_ELEMENTS BOTTOM */
377 caltinay 3707 } else if (out.getFunctionSpace().getTypeCode() == FaceElements) {
378 caltinay 3722 #pragma omp parallel
379     {
380 caltinay 3724 /*** GENERATOR SNIP_GRAD_FACES TOP */
381 caltinay 3722 if (m_faceOffset[0] > -1) {
382     #pragma omp for nowait
383     for (index_t k1=0; k1 < m_NE1; ++k1) {
384     const register double* f_10 = in.getSampleDataRO(INDEX2(1,k1, m_N0));
385     const register double* f_11 = in.getSampleDataRO(INDEX2(1,k1+1, m_N0));
386     const register double* f_01 = in.getSampleDataRO(INDEX2(0,k1+1, m_N0));
387     const register double* f_00 = in.getSampleDataRO(INDEX2(0,k1, m_N0));
388     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
389     for (index_t i=0; i < numComp; ++i) {
390 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_00[i]*cx1 + f_01[i]*cx3 + f_10[i]*cx6 + f_11[i]*cx4;
391     o[INDEX3(i,1,0,numComp,2)] = f_00[i]*cy0 + f_01[i]*cy7;
392     o[INDEX3(i,0,1,numComp,2)] = f_00[i]*cx3 + f_01[i]*cx1 + f_10[i]*cx4 + f_11[i]*cx6;
393     o[INDEX3(i,1,1,numComp,2)] = f_00[i]*cy0 + f_01[i]*cy7;
394 caltinay 3722 } /* end of component loop i */
395     } /* end of k1 loop */
396     } /* end of face 0 */
397     if (m_faceOffset[1] > -1) {
398     #pragma omp for nowait
399     for (index_t k1=0; k1 < m_NE1; ++k1) {
400     const register double* f_10 = in.getSampleDataRO(INDEX2(m_N0-1,k1, m_N0));
401     const register double* f_11 = in.getSampleDataRO(INDEX2(m_N0-1,k1+1, m_N0));
402     const register double* f_01 = in.getSampleDataRO(INDEX2(m_N0-2,k1+1, m_N0));
403     const register double* f_00 = in.getSampleDataRO(INDEX2(m_N0-2,k1, m_N0));
404     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
405     for (index_t i=0; i < numComp; ++i) {
406 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_00[i]*cx1 + f_01[i]*cx3 + f_10[i]*cx6 + f_11[i]*cx4;
407     o[INDEX3(i,1,0,numComp,2)] = f_10[i]*cy0 + f_11[i]*cy7;
408     o[INDEX3(i,0,1,numComp,2)] = f_00[i]*cx3 + f_01[i]*cx1 + f_10[i]*cx4 + f_11[i]*cx6;
409     o[INDEX3(i,1,1,numComp,2)] = f_10[i]*cy0 + f_11[i]*cy7;
410 caltinay 3722 } /* end of component loop i */
411     } /* end of k1 loop */
412     } /* end of face 1 */
413     if (m_faceOffset[2] > -1) {
414     #pragma omp for nowait
415     for (index_t k0=0; k0 < m_NE0; ++k0) {
416     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,0, m_N0));
417     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,0, m_N0));
418     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,1, m_N0));
419     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,1, m_N0));
420     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
421     for (index_t i=0; i < numComp; ++i) {
422 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_00[i]*cx0 + f_10[i]*cx7;
423     o[INDEX3(i,1,0,numComp,2)] = f_00[i]*cy1 + f_01[i]*cy6 + f_10[i]*cy3 + f_11[i]*cy4;
424     o[INDEX3(i,0,1,numComp,2)] = f_00[i]*cx0 + f_10[i]*cx7;
425     o[INDEX3(i,1,1,numComp,2)] = f_00[i]*cy3 + f_01[i]*cy4 + f_10[i]*cy1 + f_11[i]*cy6;
426 caltinay 3722 } /* end of component loop i */
427     } /* end of k0 loop */
428     } /* end of face 2 */
429     if (m_faceOffset[3] > -1) {
430     #pragma omp for nowait
431     for (index_t k0=0; k0 < m_NE0; ++k0) {
432     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,m_N1-1, m_N0));
433     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,m_N1-1, m_N0));
434     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,m_N1-2, m_N0));
435     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,m_N1-2, m_N0));
436     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
437     for (index_t i=0; i < numComp; ++i) {
438 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_01[i]*cx0 + f_11[i]*cx7;
439     o[INDEX3(i,1,0,numComp,2)] = f_00[i]*cy1 + f_01[i]*cy6 + f_10[i]*cy3 + f_11[i]*cy4;
440     o[INDEX3(i,0,1,numComp,2)] = f_01[i]*cx0 + f_11[i]*cx7;
441     o[INDEX3(i,1,1,numComp,2)] = f_00[i]*cy3 + f_01[i]*cy4 + f_10[i]*cy1 + f_11[i]*cy6;
442 caltinay 3722 } /* end of component loop i */
443     } /* end of k0 loop */
444     } /* end of face 3 */
445     /* GENERATOR SNIP_GRAD_FACES BOTTOM */
446     } // end of parallel section
447 caltinay 3711 } else if (out.getFunctionSpace().getTypeCode() == ReducedFaceElements) {
448 caltinay 3722 #pragma omp parallel
449     {
450 caltinay 3724 /*** GENERATOR SNIP_GRAD_REDUCED_FACES TOP */
451 caltinay 3722 if (m_faceOffset[0] > -1) {
452     #pragma omp for nowait
453     for (index_t k1=0; k1 < m_NE1; ++k1) {
454     const register double* f_10 = in.getSampleDataRO(INDEX2(1,k1, m_N0));
455     const register double* f_11 = in.getSampleDataRO(INDEX2(1,k1+1, m_N0));
456     const register double* f_01 = in.getSampleDataRO(INDEX2(0,k1+1, m_N0));
457     const register double* f_00 = in.getSampleDataRO(INDEX2(0,k1, m_N0));
458     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
459     for (index_t i=0; i < numComp; ++i) {
460 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = cx5*(f_10[i] + f_11[i]) + cx2*(f_00[i] + f_01[i]);
461     o[INDEX3(i,1,0,numComp,2)] = f_00[i]*cy0 + f_01[i]*cy7;
462 caltinay 3722 } /* end of component loop i */
463     } /* end of k1 loop */
464     } /* end of face 0 */
465     if (m_faceOffset[1] > -1) {
466     #pragma omp for nowait
467     for (index_t k1=0; k1 < m_NE1; ++k1) {
468     const register double* f_10 = in.getSampleDataRO(INDEX2(m_N0-1,k1, m_N0));
469     const register double* f_11 = in.getSampleDataRO(INDEX2(m_N0-1,k1+1, m_N0));
470     const register double* f_01 = in.getSampleDataRO(INDEX2(m_N0-2,k1+1, m_N0));
471     const register double* f_00 = in.getSampleDataRO(INDEX2(m_N0-2,k1, m_N0));
472     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
473     for (index_t i=0; i < numComp; ++i) {
474 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = cx5*(f_10[i] + f_11[i]) + cx2*(f_00[i] + f_01[i]);
475     o[INDEX3(i,1,0,numComp,2)] = f_10[i]*cy0 + f_11[i]*cy7;
476 caltinay 3722 } /* end of component loop i */
477     } /* end of k1 loop */
478     } /* end of face 1 */
479     if (m_faceOffset[2] > -1) {
480     #pragma omp for nowait
481     for (index_t k0=0; k0 < m_NE0; ++k0) {
482     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,0, m_N0));
483     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,0, m_N0));
484     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,1, m_N0));
485     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,1, m_N0));
486     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
487     for (index_t i=0; i < numComp; ++i) {
488 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_00[i]*cx0 + f_10[i]*cx7;
489     o[INDEX3(i,1,0,numComp,2)] = cy2*(f_00[i] + f_10[i]) + cy5*(f_01[i] + f_11[i]);
490 caltinay 3722 } /* end of component loop i */
491     } /* end of k0 loop */
492     } /* end of face 2 */
493     if (m_faceOffset[3] > -1) {
494     #pragma omp for nowait
495     for (index_t k0=0; k0 < m_NE0; ++k0) {
496     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,m_N1-1, m_N0));
497     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,m_N1-1, m_N0));
498     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,m_N1-2, m_N0));
499     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,m_N1-2, m_N0));
500     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
501     for (index_t i=0; i < numComp; ++i) {
502 caltinay 3724 o[INDEX3(i,0,0,numComp,2)] = f_01[i]*cx0 + f_11[i]*cx7;
503     o[INDEX3(i,1,0,numComp,2)] = cy5*(f_01[i] + f_11[i]) + cy2*(f_00[i] + f_10[i]);
504 caltinay 3722 } /* end of component loop i */
505     } /* end of k0 loop */
506     } /* end of face 3 */
507     /* GENERATOR SNIP_GRAD_REDUCED_FACES BOTTOM */
508     } // end of parallel section
509 caltinay 3702 } else {
510 caltinay 3707 stringstream msg;
511     msg << "setToGradient() not implemented for "
512     << functionSpaceTypeAsString(out.getFunctionSpace().getTypeCode());
513     throw RipleyException(msg.str());
514 caltinay 3702 }
515 caltinay 3701 }
516 caltinay 3697
517 caltinay 3713 void Rectangle::setToIntegrals(vector<double>& integrals, const escript::Data& arg) const
518     {
519     escript::Data& in = *const_cast<escript::Data*>(&arg);
520     const dim_t numComp = in.getDataPointSize();
521     const double h0 = m_l0/m_gNE0;
522     const double h1 = m_l1/m_gNE1;
523     if (arg.getFunctionSpace().getTypeCode() == Elements) {
524     const double w_0 = h0*h1/4.;
525     #pragma omp parallel
526     {
527     vector<double> int_local(numComp, 0);
528 caltinay 3722 #pragma omp for nowait
529 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
530     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
531     const double* f = in.getSampleDataRO(INDEX2(k0, k1, m_NE0));
532     for (index_t i=0; i < numComp; ++i) {
533     const register double f_0 = f[INDEX2(i,0,numComp)];
534     const register double f_1 = f[INDEX2(i,1,numComp)];
535     const register double f_2 = f[INDEX2(i,2,numComp)];
536     const register double f_3 = f[INDEX2(i,3,numComp)];
537     int_local[i]+=(f_0+f_1+f_2+f_3)*w_0;
538     } /* end of component loop i */
539     } /* end of k0 loop */
540     } /* end of k1 loop */
541    
542     #pragma omp critical
543     for (index_t i=0; i<numComp; i++)
544     integrals[i]+=int_local[i];
545 caltinay 3722 } // end of parallel section
546 caltinay 3713 } else if (arg.getFunctionSpace().getTypeCode() == ReducedElements) {
547     const double w_0 = h0*h1;
548     #pragma omp parallel
549     {
550     vector<double> int_local(numComp, 0);
551 caltinay 3722 #pragma omp for nowait
552 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
553     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
554     const double* f = in.getSampleDataRO(INDEX2(k0, k1, m_NE0));
555     for (index_t i=0; i < numComp; ++i) {
556     int_local[i]+=f[i]*w_0;
557     } /* end of component loop i */
558     } /* end of k0 loop */
559     } /* end of k1 loop */
560    
561     #pragma omp critical
562     for (index_t i=0; i<numComp; i++)
563     integrals[i]+=int_local[i];
564 caltinay 3722 } // end of parallel section
565 caltinay 3713 } else if (arg.getFunctionSpace().getTypeCode() == FaceElements) {
566     const double w_0 = h0/2.;
567     const double w_1 = h1/2.;
568     #pragma omp parallel
569     {
570     vector<double> int_local(numComp, 0);
571     if (m_faceOffset[0] > -1) {
572 caltinay 3722 #pragma omp for nowait
573 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
574     const double* f = in.getSampleDataRO(m_faceOffset[0]+k1);
575     for (index_t i=0; i < numComp; ++i) {
576     const register double f_0 = f[INDEX2(i,0,numComp)];
577     const register double f_1 = f[INDEX2(i,1,numComp)];
578     int_local[i]+=(f_0+f_1)*w_1;
579     } /* end of component loop i */
580     } /* end of k1 loop */
581     }
582    
583     if (m_faceOffset[1] > -1) {
584 caltinay 3722 #pragma omp for nowait
585 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
586     const double* f = in.getSampleDataRO(m_faceOffset[1]+k1);
587     for (index_t i=0; i < numComp; ++i) {
588     const register double f_0 = f[INDEX2(i,0,numComp)];
589     const register double f_1 = f[INDEX2(i,1,numComp)];
590     int_local[i]+=(f_0+f_1)*w_1;
591     } /* end of component loop i */
592     } /* end of k1 loop */
593     }
594    
595     if (m_faceOffset[2] > -1) {
596 caltinay 3722 #pragma omp for nowait
597 caltinay 3713 for (index_t k0 = 0; k0 < m_NE0; ++k0) {
598     const double* f = in.getSampleDataRO(m_faceOffset[2]+k0);
599     for (index_t i=0; i < numComp; ++i) {
600     const register double f_0 = f[INDEX2(i,0,numComp)];
601     const register double f_1 = f[INDEX2(i,1,numComp)];
602     int_local[i]+=(f_0+f_1)*w_0;
603     } /* end of component loop i */
604     } /* end of k0 loop */
605     }
606    
607     if (m_faceOffset[3] > -1) {
608 caltinay 3722 #pragma omp for nowait
609 caltinay 3713 for (index_t k0 = 0; k0 < m_NE0; ++k0) {
610     const double* f = in.getSampleDataRO(m_faceOffset[3]+k0);
611     for (index_t i=0; i < numComp; ++i) {
612     const register double f_0 = f[INDEX2(i,0,numComp)];
613     const register double f_1 = f[INDEX2(i,1,numComp)];
614     int_local[i]+=(f_0+f_1)*w_0;
615     } /* end of component loop i */
616     } /* end of k0 loop */
617     }
618    
619     #pragma omp critical
620     for (index_t i=0; i<numComp; i++)
621     integrals[i]+=int_local[i];
622 caltinay 3722 } // end of parallel section
623 caltinay 3713 } else if (arg.getFunctionSpace().getTypeCode() == ReducedFaceElements) {
624     #pragma omp parallel
625     {
626     vector<double> int_local(numComp, 0);
627     if (m_faceOffset[0] > -1) {
628 caltinay 3722 #pragma omp for nowait
629 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
630     const double* f = in.getSampleDataRO(m_faceOffset[0]+k1);
631     for (index_t i=0; i < numComp; ++i) {
632     int_local[i]+=f[i]*h1;
633     } /* end of component loop i */
634     } /* end of k1 loop */
635     }
636    
637     if (m_faceOffset[1] > -1) {
638 caltinay 3722 #pragma omp for nowait
639 caltinay 3713 for (index_t k1 = 0; k1 < m_NE1; ++k1) {
640     const double* f = in.getSampleDataRO(m_faceOffset[1]+k1);
641     for (index_t i=0; i < numComp; ++i) {
642     int_local[i]+=f[i]*h1;
643     } /* end of component loop i */
644     } /* end of k1 loop */
645     }
646    
647     if (m_faceOffset[2] > -1) {
648 caltinay 3722 #pragma omp for nowait
649 caltinay 3713 for (index_t k0 = 0; k0 < m_NE0; ++k0) {
650     const double* f = in.getSampleDataRO(m_faceOffset[2]+k0);
651     for (index_t i=0; i < numComp; ++i) {
652     int_local[i]+=f[i]*h0;
653     } /* end of component loop i */
654     } /* end of k0 loop */
655     }
656    
657     if (m_faceOffset[3] > -1) {
658 caltinay 3722 #pragma omp for nowait
659 caltinay 3713 for (index_t k0 = 0; k0 < m_NE0; ++k0) {
660     const double* f = in.getSampleDataRO(m_faceOffset[3]+k0);
661     for (index_t i=0; i < numComp; ++i) {
662     int_local[i]+=f[i]*h0;
663     } /* end of component loop i */
664     } /* end of k0 loop */
665     }
666    
667     #pragma omp critical
668     for (index_t i=0; i<numComp; i++)
669     integrals[i]+=int_local[i];
670 caltinay 3722 } // end of parallel section
671 caltinay 3713 } else {
672     stringstream msg;
673     msg << "setToIntegrals() not implemented for "
674     << functionSpaceTypeAsString(arg.getFunctionSpace().getTypeCode());
675     throw RipleyException(msg.str());
676     }
677     }
678    
679 caltinay 3722 void Rectangle::setToNormal(escript::Data& out) const
680     {
681     if (out.getFunctionSpace().getTypeCode() == FaceElements) {
682     #pragma omp parallel
683     {
684     if (m_faceOffset[0] > -1) {
685     #pragma omp for nowait
686     for (index_t k1 = 0; k1 < m_NE1; ++k1) {
687     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
688     // set vector at two quadrature points
689     *o++ = -1.;
690     *o++ = 0.;
691     *o++ = -1.;
692     *o = 0.;
693     }
694     }
695    
696     if (m_faceOffset[1] > -1) {
697     #pragma omp for nowait
698     for (index_t k1 = 0; k1 < m_NE1; ++k1) {
699     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
700     // set vector at two quadrature points
701     *o++ = 1.;
702     *o++ = 0.;
703     *o++ = 1.;
704     *o = 0.;
705     }
706     }
707    
708     if (m_faceOffset[2] > -1) {
709     #pragma omp for nowait
710     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
711     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
712     // set vector at two quadrature points
713     *o++ = 0.;
714     *o++ = -1.;
715     *o++ = 0.;
716     *o = -1.;
717     }
718     }
719    
720     if (m_faceOffset[3] > -1) {
721     #pragma omp for nowait
722     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
723     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
724     // set vector at two quadrature points
725     *o++ = 0.;
726     *o++ = 1.;
727     *o++ = 0.;
728     *o = 1.;
729     }
730     }
731     } // end of parallel section
732     } else if (out.getFunctionSpace().getTypeCode() == ReducedFaceElements) {
733     #pragma omp parallel
734     {
735     if (m_faceOffset[0] > -1) {
736     #pragma omp for nowait
737     for (index_t k1 = 0; k1 < m_NE1; ++k1) {
738     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
739     *o++ = -1.;
740     *o = 0.;
741     }
742     }
743    
744     if (m_faceOffset[1] > -1) {
745     #pragma omp for nowait
746     for (index_t k1 = 0; k1 < m_NE1; ++k1) {
747     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
748     *o++ = 1.;
749     *o = 0.;
750     }
751     }
752    
753     if (m_faceOffset[2] > -1) {
754     #pragma omp for nowait
755     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
756     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
757     *o++ = 0.;
758     *o = -1.;
759     }
760     }
761    
762     if (m_faceOffset[3] > -1) {
763     #pragma omp for nowait
764     for (index_t k0 = 0; k0 < m_NE0; ++k0) {
765     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
766     *o++ = 0.;
767     *o = 1.;
768     }
769     }
770     } // end of parallel section
771    
772     } else {
773     stringstream msg;
774     msg << "setToNormal() not implemented for "
775     << functionSpaceTypeAsString(out.getFunctionSpace().getTypeCode());
776     throw RipleyException(msg.str());
777     }
778     }
779    
780 caltinay 3691 Paso_SystemMatrixPattern* Rectangle::getPattern(bool reducedRowOrder,
781     bool reducedColOrder) const
782     {
783     if (reducedRowOrder || reducedColOrder)
784     throw RipleyException("getPattern() not implemented for reduced order");
785    
786 caltinay 3756 return m_pattern;
787 caltinay 3691 }
788    
789     void Rectangle::Print_Mesh_Info(const bool full) const
790     {
791     RipleyDomain::Print_Mesh_Info(full);
792     if (full) {
793     cout << " Id Coordinates" << endl;
794     cout.precision(15);
795     cout.setf(ios::scientific, ios::floatfield);
796 caltinay 3697 pair<double,double> xdx = getFirstCoordAndSpacing(0);
797     pair<double,double> ydy = getFirstCoordAndSpacing(1);
798 caltinay 3691 for (index_t i=0; i < getNumNodes(); i++) {
799     cout << " " << setw(5) << m_nodeId[i]
800 caltinay 3697 << " " << xdx.first+(i%m_N0)*xdx.second
801     << " " << ydy.first+(i/m_N0)*ydy.second << endl;
802 caltinay 3691 }
803     }
804     }
805    
806 caltinay 3697 IndexVector Rectangle::getNumNodesPerDim() const
807     {
808     IndexVector ret;
809     ret.push_back(m_N0);
810     ret.push_back(m_N1);
811     return ret;
812     }
813    
814     IndexVector Rectangle::getNumElementsPerDim() const
815     {
816     IndexVector ret;
817     ret.push_back(m_NE0);
818     ret.push_back(m_NE1);
819     return ret;
820     }
821    
822     IndexVector Rectangle::getNumFacesPerBoundary() const
823     {
824     IndexVector ret(4, 0);
825     //left
826     if (m_offset0==0)
827     ret[0]=m_NE1;
828     //right
829     if (m_mpiInfo->rank%m_NX==m_NX-1)
830     ret[1]=m_NE1;
831     //bottom
832     if (m_offset1==0)
833     ret[2]=m_NE0;
834     //top
835     if (m_mpiInfo->rank/m_NX==m_NY-1)
836     ret[3]=m_NE0;
837     return ret;
838     }
839    
840     pair<double,double> Rectangle::getFirstCoordAndSpacing(dim_t dim) const
841     {
842     if (dim==0) {
843     return pair<double,double>((m_l0*m_offset0)/m_gNE0, m_l0/m_gNE0);
844     } else if (dim==1) {
845     return pair<double,double>((m_l1*m_offset1)/m_gNE1, m_l1/m_gNE1);
846     }
847     throw RipleyException("getFirstCoordAndSpacing(): invalid argument");
848     }
849    
850 caltinay 3691 //protected
851 caltinay 3750 dim_t Rectangle::getNumDOF() const
852     {
853 caltinay 3752 return (m_gNE0+1)/m_NX*(m_gNE1+1)/m_NY;
854 caltinay 3750 }
855    
856     //protected
857 caltinay 3691 dim_t Rectangle::getNumFaceElements() const
858     {
859 caltinay 3699 const IndexVector faces = getNumFacesPerBoundary();
860 caltinay 3691 dim_t n=0;
861 caltinay 3699 for (size_t i=0; i<faces.size(); i++)
862     n+=faces[i];
863 caltinay 3691 return n;
864     }
865    
866     //protected
867     void Rectangle::assembleCoordinates(escript::Data& arg) const
868     {
869     escriptDataC x = arg.getDataC();
870     int numDim = m_numDim;
871     if (!isDataPointShapeEqual(&x, 1, &numDim))
872     throw RipleyException("setToX: Invalid Data object shape");
873     if (!numSamplesEqual(&x, 1, getNumNodes()))
874     throw RipleyException("setToX: Illegal number of samples in Data object");
875    
876 caltinay 3697 pair<double,double> xdx = getFirstCoordAndSpacing(0);
877     pair<double,double> ydy = getFirstCoordAndSpacing(1);
878 caltinay 3691 arg.requireWrite();
879     #pragma omp parallel for
880     for (dim_t i1 = 0; i1 < m_N1; i1++) {
881     for (dim_t i0 = 0; i0 < m_N0; i0++) {
882     double* point = arg.getSampleDataRW(i0+m_N0*i1);
883 caltinay 3697 point[0] = xdx.first+i0*xdx.second;
884     point[1] = ydy.first+i1*ydy.second;
885 caltinay 3691 }
886     }
887     }
888    
889 caltinay 3756 //protected
890     dim_t Rectangle::insertNeighbourNodes(IndexVector& index, index_t node) const
891     {
892     const dim_t nDOF0 = (m_gNE0+1)/m_NX;
893     const dim_t nDOF1 = (m_gNE1+1)/m_NY;
894     const int x=node%nDOF0;
895     const int y=node/nDOF0;
896     dim_t num=0;
897     // loop through potential neighbours and add to index if positions are
898     // within bounds
899     for (int i1=-1; i1<2; i1++) {
900     for (int i0=-1; i0<2; i0++) {
901     // skip node itself
902     if (i0==0 && i1==0)
903     continue;
904     // location of neighbour node
905     const int nx=x+i0;
906     const int ny=y+i1;
907     if (nx>=0 && ny>=0 && nx<nDOF0 && ny<nDOF1) {
908     index.push_back(ny*nDOF0+nx);
909     num++;
910     }
911     }
912     }
913    
914     return num;
915     }
916    
917     //protected
918     void Rectangle::nodesToDOF(escript::Data& out, escript::Data& in) const
919     {
920     const dim_t numComp = in.getDataPointSize();
921     out.requireWrite();
922    
923     const index_t left = (m_offset0==0 ? 0 : 1);
924     const index_t bottom = (m_offset1==0 ? 0 : 1);
925     const dim_t nDOF0 = (m_gNE0+1)/m_NX;
926     const dim_t nDOF1 = (m_gNE1+1)/m_NY;
927     #pragma omp parallel for
928     for (index_t i=0; i<nDOF1; i++) {
929     for (index_t j=0; j<nDOF0; j++) {
930     const index_t n=j+left+(i+bottom)*m_N0;
931     const double* src=in.getSampleDataRO(n);
932     copy(src, src+numComp, out.getSampleDataRW(j+i*nDOF0));
933     }
934     }
935     }
936    
937     //protected
938     void Rectangle::dofToNodes(escript::Data& out, escript::Data& in) const
939     {
940     const dim_t numComp = in.getDataPointSize();
941     Paso_Coupler* coupler = Paso_Coupler_alloc(m_connector, numComp);
942     in.requireWrite();
943     Paso_Coupler_startCollect(coupler, in.getSampleDataRW(0));
944    
945     const dim_t numDOF = getNumDOF();
946     out.requireWrite();
947     const double* buffer = Paso_Coupler_finishCollect(coupler);
948    
949     #pragma omp parallel for
950     for (index_t i=0; i<getNumNodes(); i++) {
951     const double* src=(m_dofMap[i]<numDOF ?
952     in.getSampleDataRO(m_dofMap[i])
953     : &buffer[(m_dofMap[i]-numDOF)*numComp]);
954     copy(src, src+numComp, out.getSampleDataRW(i));
955     }
956     }
957    
958 caltinay 3691 //private
959     void Rectangle::populateSampleIds()
960     {
961 caltinay 3752 // identifiers are ordered from left to right, bottom to top globablly.
962 caltinay 3697
963     // build node distribution vector first.
964     // rank i owns m_nodeDistribution[i+1]-nodeDistribution[i] nodes
965     m_nodeDistribution.assign(m_mpiInfo->size+1, 0);
966 caltinay 3752 const dim_t numDOF=getNumDOF();
967     for (dim_t k=1; k<m_mpiInfo->size; k++) {
968     m_nodeDistribution[k]=k*numDOF;
969 caltinay 3697 }
970     m_nodeDistribution[m_mpiInfo->size]=getNumDataPointsGlobal();
971 caltinay 3691 m_nodeId.resize(getNumNodes());
972 caltinay 3753 m_dofId.resize(numDOF);
973     m_elementId.resize(getNumElements());
974     m_faceId.resize(getNumFaceElements());
975 caltinay 3697
976 caltinay 3753 #pragma omp parallel
977     {
978     // nodes
979     #pragma omp for nowait
980     for (dim_t i1=0; i1<m_N1; i1++) {
981     for (dim_t i0=0; i0<m_N0; i0++) {
982     m_nodeId[i0+i1*m_N0] = (m_offset1+i1)*(m_gNE0+1)+m_offset0+i0;
983     }
984 caltinay 3697 }
985    
986 caltinay 3753 // degrees of freedom
987     #pragma omp for nowait
988     for (dim_t k=0; k<numDOF; k++)
989     m_dofId[k] = m_nodeDistribution[m_mpiInfo->rank]+k;
990 caltinay 3752
991 caltinay 3753 // elements
992     #pragma omp for nowait
993 caltinay 3755 for (dim_t i1=0; i1<m_NE1; i1++) {
994     for (dim_t i0=0; i0<m_NE0; i0++) {
995     m_elementId[i0+i1*m_NE0]=(m_offset1+i1)*m_gNE0+m_offset0+i0;
996     }
997     }
998 caltinay 3753
999     // face elements
1000     #pragma omp for
1001     for (dim_t k=0; k<getNumFaceElements(); k++)
1002     m_faceId[k]=k;
1003     } // end parallel section
1004    
1005 caltinay 3735 m_nodeTags.assign(getNumNodes(), 0);
1006     updateTagsInUse(Nodes);
1007 caltinay 3697
1008 caltinay 3735 m_elementTags.assign(getNumElements(), 0);
1009     updateTagsInUse(Elements);
1010 caltinay 3697
1011 caltinay 3722 // generate face offset vector and set face tags
1012 caltinay 3704 const IndexVector facesPerEdge = getNumFacesPerBoundary();
1013 caltinay 3722 const index_t LEFT=1, RIGHT=2, BOTTOM=10, TOP=20;
1014     const index_t faceTag[] = { LEFT, RIGHT, BOTTOM, TOP };
1015 caltinay 3704 m_faceOffset.assign(facesPerEdge.size(), -1);
1016 caltinay 3722 m_faceTags.clear();
1017 caltinay 3704 index_t offset=0;
1018     for (size_t i=0; i<facesPerEdge.size(); i++) {
1019     if (facesPerEdge[i]>0) {
1020     m_faceOffset[i]=offset;
1021     offset+=facesPerEdge[i];
1022 caltinay 3722 m_faceTags.insert(m_faceTags.end(), facesPerEdge[i], faceTag[i]);
1023 caltinay 3704 }
1024     }
1025 caltinay 3722 setTagMap("left", LEFT);
1026     setTagMap("right", RIGHT);
1027     setTagMap("bottom", BOTTOM);
1028     setTagMap("top", TOP);
1029     updateTagsInUse(FaceElements);
1030 caltinay 3691 }
1031    
1032 caltinay 3699 //private
1033 caltinay 3756 void Rectangle::createPattern()
1034 caltinay 3699 {
1035 caltinay 3756 const dim_t nDOF0 = (m_gNE0+1)/m_NX;
1036     const dim_t nDOF1 = (m_gNE1+1)/m_NY;
1037     const index_t left = (m_offset0==0 ? 0 : 1);
1038     const index_t bottom = (m_offset1==0 ? 0 : 1);
1039    
1040     // populate node->DOF mapping with own degrees of freedom.
1041     // The rest is assigned in the loop further down
1042     m_dofMap.assign(getNumNodes(), 0);
1043     #pragma omp parallel for
1044     for (index_t i=bottom; i<m_N1; i++) {
1045     for (index_t j=left; j<m_N0; j++) {
1046     m_dofMap[i*m_N0+j]=(i-bottom)*nDOF0+j-left;
1047     }
1048     }
1049    
1050     // build list of shared components and neighbours by looping through
1051     // all potential neighbouring ranks and checking if positions are
1052 caltinay 3754 // within bounds
1053 caltinay 3756 const dim_t numDOF=nDOF0*nDOF1;
1054     vector<IndexVector> colIndices(numDOF); // for the couple blocks
1055     RankVector neighbour;
1056     IndexVector offsetInShared(1,0);
1057     IndexVector sendShared, recvShared;
1058     int numShared=0;
1059     const int x=m_mpiInfo->rank%m_NX;
1060     const int y=m_mpiInfo->rank/m_NX;
1061 caltinay 3754 for (int i1=-1; i1<2; i1++) {
1062     for (int i0=-1; i0<2; i0++) {
1063 caltinay 3756 // skip this rank
1064 caltinay 3754 if (i0==0 && i1==0)
1065     continue;
1066 caltinay 3756 // location of neighbour rank
1067 caltinay 3754 const int nx=x+i0;
1068     const int ny=y+i1;
1069 caltinay 3756 if (nx>=0 && ny>=0 && nx<m_NX && ny<m_NY) {
1070     neighbour.push_back(ny*m_NX+nx);
1071     if (i0==0) {
1072     // sharing top or bottom edge
1073     const int firstDOF=(i1==-1 ? 0 : numDOF-nDOF0);
1074     const int firstNode=(i1==-1 ? left : m_N0*(m_N1-1)+left);
1075     offsetInShared.push_back(offsetInShared.back()+nDOF0);
1076     for (dim_t i=0; i<nDOF0; i++, numShared++) {
1077     sendShared.push_back(firstDOF+i);
1078     recvShared.push_back(numDOF+numShared);
1079     if (i>0)
1080     colIndices[firstDOF+i-1].push_back(numShared);
1081     colIndices[firstDOF+i].push_back(numShared);
1082     if (i<nDOF0-1)
1083     colIndices[firstDOF+i+1].push_back(numShared);
1084     m_dofMap[firstNode+i]=numDOF+numShared;
1085     }
1086     } else if (i1==0) {
1087     // sharing left or right edge
1088     const int firstDOF=(i0==-1 ? 0 : nDOF0-1);
1089     const int firstNode=(i0==-1 ? bottom*m_N0 : (bottom+1)*m_N0-1);
1090     offsetInShared.push_back(offsetInShared.back()+nDOF1);
1091     for (dim_t i=0; i<nDOF1; i++, numShared++) {
1092     sendShared.push_back(firstDOF+i*nDOF0);
1093     recvShared.push_back(numDOF+numShared);
1094     if (i>0)
1095     colIndices[firstDOF+(i-1)*nDOF0].push_back(numShared);
1096     colIndices[firstDOF+i*nDOF0].push_back(numShared);
1097     if (i<nDOF1-1)
1098     colIndices[firstDOF+(i+1)*nDOF0].push_back(numShared);
1099     m_dofMap[firstNode+i*m_N0]=numDOF+numShared;
1100     }
1101     } else {
1102     // sharing a node
1103     const int dof=(i0+1)/2*(nDOF0-1)+(i1+1)/2*(numDOF-nDOF0);
1104     const int node=(i0+1)/2*(m_N0-1)+(i1+1)/2*m_N0*(m_N1-1);
1105     offsetInShared.push_back(offsetInShared.back()+1);
1106     sendShared.push_back(dof);
1107     recvShared.push_back(numDOF+numShared);
1108     colIndices[dof].push_back(numShared);
1109     m_dofMap[node]=numDOF+numShared;
1110     ++numShared;
1111     }
1112 caltinay 3754 }
1113 caltinay 3699 }
1114     }
1115 caltinay 3754
1116 caltinay 3756 // create connector
1117     Paso_SharedComponents *snd_shcomp = Paso_SharedComponents_alloc(
1118     numDOF, neighbour.size(), &neighbour[0], &sendShared[0],
1119     &offsetInShared[0], 1, 0, m_mpiInfo);
1120     Paso_SharedComponents *rcv_shcomp = Paso_SharedComponents_alloc(
1121     numDOF, neighbour.size(), &neighbour[0], &recvShared[0],
1122     &offsetInShared[0], 1, 0, m_mpiInfo);
1123     m_connector = Paso_Connector_alloc(snd_shcomp, rcv_shcomp);
1124     Paso_SharedComponents_free(snd_shcomp);
1125     Paso_SharedComponents_free(rcv_shcomp);
1126 caltinay 3754
1127 caltinay 3756 // create main and couple blocks
1128     Paso_Pattern *mainPattern = createMainPattern();
1129     Paso_Pattern *colPattern, *rowPattern;
1130     createCouplePatterns(colIndices, numShared, &colPattern, &rowPattern);
1131 caltinay 3754
1132 caltinay 3756 // allocate paso distribution
1133     Paso_Distribution* distribution = Paso_Distribution_alloc(m_mpiInfo,
1134     const_cast<index_t*>(&m_nodeDistribution[0]), 1, 0);
1135 caltinay 3755
1136 caltinay 3756 // finally create the system matrix
1137     m_pattern = Paso_SystemMatrixPattern_alloc(MATRIX_FORMAT_DEFAULT,
1138     distribution, distribution, mainPattern, colPattern, rowPattern,
1139     m_connector, m_connector);
1140 caltinay 3755
1141 caltinay 3756 Paso_Distribution_free(distribution);
1142 caltinay 3755
1143 caltinay 3756 // useful debug output
1144     /*
1145     cout << "--- rcv_shcomp ---" << endl;
1146     cout << "numDOF=" << numDOF << ", numNeighbors=" << neighbour.size() << endl;
1147     for (size_t i=0; i<neighbour.size(); i++) {
1148     cout << "neighbor[" << i << "]=" << neighbour[i]
1149     << " offsetInShared[" << i+1 << "]=" << offsetInShared[i+1] << endl;
1150 caltinay 3699 }
1151 caltinay 3756 for (size_t i=0; i<recvShared.size(); i++) {
1152     cout << "shared[" << i << "]=" << recvShared[i] << endl;
1153     }
1154     cout << "--- snd_shcomp ---" << endl;
1155     for (size_t i=0; i<sendShared.size(); i++) {
1156     cout << "shared[" << i << "]=" << sendShared[i] << endl;
1157     }
1158     cout << "--- dofMap ---" << endl;
1159     for (size_t i=0; i<m_dofMap.size(); i++) {
1160     cout << "m_dofMap[" << i << "]=" << m_dofMap[i] << endl;
1161     }
1162     cout << "--- colIndices ---" << endl;
1163     for (size_t i=0; i<colIndices.size(); i++) {
1164     cout << "colIndices[" << i << "].size()=" << colIndices[i].size() << endl;
1165     }
1166     */
1167 caltinay 3754
1168 caltinay 3756 /*
1169     cout << "--- main_pattern ---" << endl;
1170     cout << "M=" << mainPattern->numOutput << ", N=" << mainPattern->numInput << endl;
1171     for (size_t i=0; i<mainPattern->numOutput+1; i++) {
1172     cout << "ptr[" << i << "]=" << mainPattern->ptr[i] << endl;
1173     }
1174     for (size_t i=0; i<mainPattern->ptr[mainPattern->numOutput]; i++) {
1175     cout << "index[" << i << "]=" << mainPattern->index[i] << endl;
1176     }
1177     */
1178 caltinay 3754
1179 caltinay 3756 /*
1180     cout << "--- colCouple_pattern ---" << endl;
1181     cout << "M=" << colPattern->numOutput << ", N=" << colPattern->numInput << endl;
1182     for (size_t i=0; i<colPattern->numOutput+1; i++) {
1183     cout << "ptr[" << i << "]=" << colPattern->ptr[i] << endl;
1184     }
1185     for (size_t i=0; i<colPattern->ptr[colPattern->numOutput]; i++) {
1186     cout << "index[" << i << "]=" << colPattern->index[i] << endl;
1187     }
1188     */
1189 caltinay 3754
1190 caltinay 3756 /*
1191     cout << "--- rowCouple_pattern ---" << endl;
1192     cout << "M=" << rowPattern->numOutput << ", N=" << rowPattern->numInput << endl;
1193     for (size_t i=0; i<rowPattern->numOutput+1; i++) {
1194     cout << "ptr[" << i << "]=" << rowPattern->ptr[i] << endl;
1195 caltinay 3699 }
1196 caltinay 3756 for (size_t i=0; i<rowPattern->ptr[rowPattern->numOutput]; i++) {
1197     cout << "index[" << i << "]=" << rowPattern->index[i] << endl;
1198     }
1199     */
1200    
1201     Paso_Pattern_free(mainPattern);
1202     Paso_Pattern_free(colPattern);
1203     Paso_Pattern_free(rowPattern);
1204 caltinay 3699 }
1205    
1206 caltinay 3702 //protected
1207 caltinay 3711 void Rectangle::interpolateNodesOnElements(escript::Data& out,
1208     escript::Data& in, bool reduced) const
1209 caltinay 3702 {
1210     const dim_t numComp = in.getDataPointSize();
1211 caltinay 3711 if (reduced) {
1212 caltinay 3724 /*** GENERATOR SNIP_INTERPOLATE_REDUCED_ELEMENTS TOP */
1213     const double c0 = .25;
1214 caltinay 3702 #pragma omp parallel for
1215 caltinay 3711 for (index_t k1=0; k1 < m_NE1; ++k1) {
1216     for (index_t k0=0; k0 < m_NE0; ++k0) {
1217     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,k1+1, m_N0));
1218     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,k1, m_N0));
1219     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,k1, m_N0));
1220     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,k1+1, m_N0));
1221     double* o = out.getSampleDataRW(INDEX2(k0,k1,m_NE0));
1222     for (index_t i=0; i < numComp; ++i) {
1223 caltinay 3724 o[INDEX2(i,numComp,0)] = c0*(f_00[i] + f_01[i] + f_10[i] + f_11[i]);
1224 caltinay 3711 } /* end of component loop i */
1225     } /* end of k0 loop */
1226     } /* end of k1 loop */
1227     /* GENERATOR SNIP_INTERPOLATE_REDUCED_ELEMENTS BOTTOM */
1228     } else {
1229 caltinay 3724 /*** GENERATOR SNIP_INTERPOLATE_ELEMENTS TOP */
1230     const double c0 = .16666666666666666667;
1231     const double c1 = .044658198738520451079;
1232     const double c2 = .62200846792814621559;
1233 caltinay 3711 #pragma omp parallel for
1234     for (index_t k1=0; k1 < m_NE1; ++k1) {
1235     for (index_t k0=0; k0 < m_NE0; ++k0) {
1236     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,k1, m_N0));
1237     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,k1+1, m_N0));
1238     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,k1+1, m_N0));
1239     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,k1, m_N0));
1240     double* o = out.getSampleDataRW(INDEX2(k0,k1,m_NE0));
1241     for (index_t i=0; i < numComp; ++i) {
1242 caltinay 3724 o[INDEX2(i,numComp,0)] = f_00[i]*c2 + f_11[i]*c1 + c0*(f_01[i] + f_10[i]);
1243     o[INDEX2(i,numComp,1)] = f_01[i]*c1 + f_10[i]*c2 + c0*(f_00[i] + f_11[i]);
1244     o[INDEX2(i,numComp,2)] = f_01[i]*c2 + f_10[i]*c1 + c0*(f_00[i] + f_11[i]);
1245     o[INDEX2(i,numComp,3)] = f_00[i]*c1 + f_11[i]*c2 + c0*(f_01[i] + f_10[i]);
1246 caltinay 3711 } /* end of component loop i */
1247     } /* end of k0 loop */
1248     } /* end of k1 loop */
1249     /* GENERATOR SNIP_INTERPOLATE_ELEMENTS BOTTOM */
1250     }
1251 caltinay 3702 }
1252    
1253     //protected
1254 caltinay 3711 void Rectangle::interpolateNodesOnFaces(escript::Data& out, escript::Data& in,
1255     bool reduced) const
1256 caltinay 3702 {
1257 caltinay 3704 const dim_t numComp = in.getDataPointSize();
1258 caltinay 3711 if (reduced) {
1259 caltinay 3724 const double c0 = .5;
1260     #pragma omp parallel
1261     {
1262     /*** GENERATOR SNIP_INTERPOLATE_REDUCED_FACES TOP */
1263     if (m_faceOffset[0] > -1) {
1264     #pragma omp for nowait
1265     for (index_t k1=0; k1 < m_NE1; ++k1) {
1266     const register double* f_00 = in.getSampleDataRO(INDEX2(0,k1, m_N0));
1267     const register double* f_01 = in.getSampleDataRO(INDEX2(0,k1+1, m_N0));
1268     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
1269     for (index_t i=0; i < numComp; ++i) {
1270     o[INDEX2(i,numComp,0)] = c0*(f_00[i] + f_01[i]);
1271     } /* end of component loop i */
1272     } /* end of k1 loop */
1273     } /* end of face 0 */
1274     if (m_faceOffset[1] > -1) {
1275     #pragma omp for nowait
1276     for (index_t k1=0; k1 < m_NE1; ++k1) {
1277     const register double* f_10 = in.getSampleDataRO(INDEX2(m_N0-1,k1, m_N0));
1278     const register double* f_11 = in.getSampleDataRO(INDEX2(m_N0-1,k1+1, m_N0));
1279     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
1280     for (index_t i=0; i < numComp; ++i) {
1281     o[INDEX2(i,numComp,0)] = c0*(f_10[i] + f_11[i]);
1282     } /* end of component loop i */
1283     } /* end of k1 loop */
1284     } /* end of face 1 */
1285     if (m_faceOffset[2] > -1) {
1286     #pragma omp for nowait
1287     for (index_t k0=0; k0 < m_NE0; ++k0) {
1288     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,0, m_N0));
1289     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,0, m_N0));
1290     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
1291     for (index_t i=0; i < numComp; ++i) {
1292     o[INDEX2(i,numComp,0)] = c0*(f_00[i] + f_10[i]);
1293     } /* end of component loop i */
1294     } /* end of k0 loop */
1295     } /* end of face 2 */
1296     if (m_faceOffset[3] > -1) {
1297     #pragma omp for nowait
1298     for (index_t k0=0; k0 < m_NE0; ++k0) {
1299     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,m_N1-1, m_N0));
1300     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,m_N1-1, m_N0));
1301     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
1302     for (index_t i=0; i < numComp; ++i) {
1303     o[INDEX2(i,numComp,0)] = c0*(f_01[i] + f_11[i]);
1304     } /* end of component loop i */
1305     } /* end of k0 loop */
1306     } /* end of face 3 */
1307     /* GENERATOR SNIP_INTERPOLATE_REDUCED_FACES BOTTOM */
1308     } // end of parallel section
1309 caltinay 3711 } else {
1310 caltinay 3724 const double c0 = 0.21132486540518711775;
1311     const double c1 = 0.78867513459481288225;
1312     #pragma omp parallel
1313     {
1314     /*** GENERATOR SNIP_INTERPOLATE_FACES TOP */
1315     if (m_faceOffset[0] > -1) {
1316     #pragma omp for nowait
1317     for (index_t k1=0; k1 < m_NE1; ++k1) {
1318     const register double* f_01 = in.getSampleDataRO(INDEX2(0,k1+1, m_N0));
1319     const register double* f_00 = in.getSampleDataRO(INDEX2(0,k1, m_N0));
1320     double* o = out.getSampleDataRW(m_faceOffset[0]+k1);
1321     for (index_t i=0; i < numComp; ++i) {
1322     o[INDEX2(i,numComp,0)] = f_00[i]*c1 + f_01[i]*c0;
1323     o[INDEX2(i,numComp,1)] = f_00[i]*c0 + f_01[i]*c1;
1324     } /* end of component loop i */
1325     } /* end of k1 loop */
1326     } /* end of face 0 */
1327     if (m_faceOffset[1] > -1) {
1328     #pragma omp for nowait
1329     for (index_t k1=0; k1 < m_NE1; ++k1) {
1330     const register double* f_10 = in.getSampleDataRO(INDEX2(m_N0-1,k1, m_N0));
1331     const register double* f_11 = in.getSampleDataRO(INDEX2(m_N0-1,k1+1, m_N0));
1332     double* o = out.getSampleDataRW(m_faceOffset[1]+k1);
1333     for (index_t i=0; i < numComp; ++i) {
1334     o[INDEX2(i,numComp,0)] = f_10[i]*c1 + f_11[i]*c0;
1335     o[INDEX2(i,numComp,1)] = f_10[i]*c0 + f_11[i]*c1;
1336     } /* end of component loop i */
1337     } /* end of k1 loop */
1338     } /* end of face 1 */
1339     if (m_faceOffset[2] > -1) {
1340     #pragma omp for nowait
1341     for (index_t k0=0; k0 < m_NE0; ++k0) {
1342     const register double* f_10 = in.getSampleDataRO(INDEX2(k0+1,0, m_N0));
1343     const register double* f_00 = in.getSampleDataRO(INDEX2(k0,0, m_N0));
1344     double* o = out.getSampleDataRW(m_faceOffset[2]+k0);
1345     for (index_t i=0; i < numComp; ++i) {
1346     o[INDEX2(i,numComp,0)] = f_00[i]*c1 + f_10[i]*c0;
1347     o[INDEX2(i,numComp,1)] = f_00[i]*c0 + f_10[i]*c1;
1348     } /* end of component loop i */
1349     } /* end of k0 loop */
1350     } /* end of face 2 */
1351     if (m_faceOffset[3] > -1) {
1352     #pragma omp for nowait
1353     for (index_t k0=0; k0 < m_NE0; ++k0) {
1354     const register double* f_11 = in.getSampleDataRO(INDEX2(k0+1,m_N1-1, m_N0));
1355     const register double* f_01 = in.getSampleDataRO(INDEX2(k0,m_N1-1, m_N0));
1356     double* o = out.getSampleDataRW(m_faceOffset[3]+k0);
1357     for (index_t i=0; i < numComp; ++i) {
1358     o[INDEX2(i,numComp,0)] = f_01[i]*c1 + f_11[i]*c0;
1359     o[INDEX2(i,numComp,1)] = f_01[i]*c0 + f_11[i]*c1;
1360     } /* end of component loop i */
1361     } /* end of k0 loop */
1362     } /* end of face 3 */
1363     /* GENERATOR SNIP_INTERPOLATE_FACES BOTTOM */
1364     } // end of parallel section
1365 caltinay 3711 }
1366 caltinay 3702 }
1367    
1368 caltinay 3748 //protected
1369     void Rectangle::assemblePDESingle(Paso_SystemMatrix* mat,
1370     escript::Data& rhs, const escript::Data& A, const escript::Data& B,
1371     const escript::Data& C, const escript::Data& D,
1372     const escript::Data& X, const escript::Data& Y,
1373     const escript::Data& d, const escript::Data& y) const
1374     {
1375     const double h0 = m_l0/m_gNE0;
1376     const double h1 = m_l1/m_gNE1;
1377 caltinay 3754 /*** GENERATOR SNIP_PDE_SINGLE_PRE TOP */
1378 caltinay 3748 const double w0 = -0.1555021169820365539*h1/h0;
1379     const double w1 = 0.041666666666666666667;
1380     const double w10 = -0.041666666666666666667*h0/h1;
1381     const double w11 = 0.1555021169820365539*h1/h0;
1382     const double w12 = 0.1555021169820365539*h0/h1;
1383     const double w13 = 0.01116454968463011277*h0/h1;
1384     const double w14 = 0.01116454968463011277*h1/h0;
1385     const double w15 = 0.041666666666666666667*h1/h0;
1386     const double w16 = -0.01116454968463011277*h0/h1;
1387     const double w17 = -0.1555021169820365539*h0/h1;
1388     const double w18 = -0.33333333333333333333*h1/h0;
1389     const double w19 = 0.25000000000000000000;
1390     const double w2 = -0.15550211698203655390;
1391     const double w20 = -0.25000000000000000000;
1392     const double w21 = 0.16666666666666666667*h0/h1;
1393     const double w22 = -0.16666666666666666667*h1/h0;
1394     const double w23 = -0.16666666666666666667*h0/h1;
1395     const double w24 = 0.33333333333333333333*h1/h0;
1396     const double w25 = 0.33333333333333333333*h0/h1;
1397     const double w26 = 0.16666666666666666667*h1/h0;
1398     const double w27 = -0.33333333333333333333*h0/h1;
1399     const double w28 = -0.032861463941450536761*h1;
1400     const double w29 = -0.032861463941450536761*h0;
1401     const double w3 = 0.041666666666666666667*h0/h1;
1402     const double w30 = -0.12264065304058601714*h1;
1403     const double w31 = -0.0023593469594139828636*h1;
1404     const double w32 = -0.008805202725216129906*h0;
1405     const double w33 = -0.008805202725216129906*h1;
1406     const double w34 = 0.032861463941450536761*h1;
1407     const double w35 = 0.008805202725216129906*h1;
1408     const double w36 = 0.008805202725216129906*h0;
1409     const double w37 = 0.0023593469594139828636*h1;
1410     const double w38 = 0.12264065304058601714*h1;
1411     const double w39 = 0.032861463941450536761*h0;
1412     const double w4 = 0.15550211698203655390;
1413     const double w40 = -0.12264065304058601714*h0;
1414     const double w41 = -0.0023593469594139828636*h0;
1415     const double w42 = 0.0023593469594139828636*h0;
1416     const double w43 = 0.12264065304058601714*h0;
1417     const double w44 = -0.16666666666666666667*h1;
1418     const double w45 = -0.083333333333333333333*h0;
1419     const double w46 = 0.083333333333333333333*h1;
1420     const double w47 = 0.16666666666666666667*h1;
1421     const double w48 = 0.083333333333333333333*h0;
1422     const double w49 = -0.16666666666666666667*h0;
1423     const double w5 = -0.041666666666666666667;
1424     const double w50 = 0.16666666666666666667*h0;
1425     const double w51 = -0.083333333333333333333*h1;
1426     const double w52 = 0.025917019497006092316*h0*h1;
1427     const double w53 = 0.0018607582807716854616*h0*h1;
1428     const double w54 = 0.0069444444444444444444*h0*h1;
1429     const double w55 = 0.09672363354357992482*h0*h1;
1430     const double w56 = 0.00049858867864229740201*h0*h1;
1431     const double w57 = 0.055555555555555555556*h0*h1;
1432     const double w58 = 0.027777777777777777778*h0*h1;
1433     const double w59 = 0.11111111111111111111*h0*h1;
1434     const double w6 = -0.01116454968463011277*h1/h0;
1435     const double w60 = -0.19716878364870322056*h1;
1436     const double w61 = -0.19716878364870322056*h0;
1437     const double w62 = -0.052831216351296779436*h0;
1438     const double w63 = -0.052831216351296779436*h1;
1439     const double w64 = 0.19716878364870322056*h1;
1440     const double w65 = 0.052831216351296779436*h1;
1441     const double w66 = 0.19716878364870322056*h0;
1442     const double w67 = 0.052831216351296779436*h0;
1443     const double w68 = -0.5*h1;
1444     const double w69 = -0.5*h0;
1445     const double w7 = 0.011164549684630112770;
1446     const double w70 = 0.5*h1;
1447     const double w71 = 0.5*h0;
1448     const double w72 = 0.1555021169820365539*h0*h1;
1449     const double w73 = 0.041666666666666666667*h0*h1;
1450     const double w74 = 0.01116454968463011277*h0*h1;
1451     const double w75 = 0.25*h0*h1;
1452     const double w8 = -0.011164549684630112770;
1453     const double w9 = -0.041666666666666666667*h1/h0;
1454     /* GENERATOR SNIP_PDE_SINGLE_PRE BOTTOM */
1455    
1456     rhs.requireWrite();
1457     #pragma omp parallel
1458     {
1459 caltinay 3754 for (index_t k1_0=0; k1_0<2; k1_0++) { // colouring
1460 caltinay 3748 #pragma omp for
1461     for (index_t k1=k1_0; k1<m_NE1; k1+=2) {
1462     for (index_t k0=0; k0<m_NE0; ++k0) {
1463     bool add_EM_S=false;
1464     bool add_EM_F=false;
1465     vector<double> EM_S(4*4, 0);
1466     vector<double> EM_F(4, 0);
1467     const index_t e = k0 + m_NE0*k1;
1468 caltinay 3754 /*** GENERATOR SNIP_PDE_SINGLE TOP */
1469 caltinay 3748 ///////////////
1470     // process A //
1471     ///////////////
1472     if (!A.isEmpty()) {
1473     add_EM_S=true;
1474     const double* A_p=const_cast<escript::Data*>(&A)->getSampleDataRO(e);
1475     if (A.actsExpanded()) {
1476     const register double A_00_0 = A_p[INDEX3(0,0,0,2,2)];
1477     const register double A_01_0 = A_p[INDEX3(0,1,0,2,2)];
1478     const register double A_10_0 = A_p[INDEX3(1,0,0,2,2)];
1479     const register double A_11_0 = A_p[INDEX3(1,1,0,2,2)];
1480     const register double A_00_1 = A_p[INDEX3(0,0,1,2,2)];
1481     const register double A_01_1 = A_p[INDEX3(0,1,1,2,2)];
1482     const register double A_10_1 = A_p[INDEX3(1,0,1,2,2)];
1483     const register double A_11_1 = A_p[INDEX3(1,1,1,2,2)];
1484     const register double A_00_2 = A_p[INDEX3(0,0,2,2,2)];
1485     const register double A_01_2 = A_p[INDEX3(0,1,2,2,2)];
1486     const register double A_10_2 = A_p[INDEX3(1,0,2,2,2)];
1487     const register double A_11_2 = A_p[INDEX3(1,1,2,2,2)];
1488     const register double A_00_3 = A_p[INDEX3(0,0,3,2,2)];
1489     const register double A_01_3 = A_p[INDEX3(0,1,3,2,2)];
1490     const register double A_10_3 = A_p[INDEX3(1,0,3,2,2)];
1491     const register double A_11_3 = A_p[INDEX3(1,1,3,2,2)];
1492     const register double tmp4_0 = A_10_1 + A_10_2;
1493     const register double tmp12_0 = A_11_0 + A_11_2;
1494     const register double tmp2_0 = A_11_0 + A_11_1 + A_11_2 + A_11_3;
1495     const register double tmp10_0 = A_01_3 + A_10_3;
1496     const register double tmp14_0 = A_01_0 + A_01_3 + A_10_0 + A_10_3;
1497     const register double tmp0_0 = A_01_0 + A_01_3;
1498     const register double tmp13_0 = A_01_2 + A_10_1;
1499     const register double tmp3_0 = A_00_2 + A_00_3;
1500     const register double tmp11_0 = A_11_1 + A_11_3;
1501     const register double tmp18_0 = A_01_1 + A_10_1;
1502     const register double tmp1_0 = A_00_0 + A_00_1;
1503     const register double tmp15_0 = A_01_1 + A_10_2;
1504     const register double tmp5_0 = A_00_0 + A_00_1 + A_00_2 + A_00_3;
1505     const register double tmp16_0 = A_10_0 + A_10_3;
1506     const register double tmp6_0 = A_01_3 + A_10_0;
1507     const register double tmp17_0 = A_01_1 + A_01_2;
1508     const register double tmp9_0 = A_01_0 + A_10_0;
1509     const register double tmp7_0 = A_01_0 + A_10_3;
1510     const register double tmp8_0 = A_01_1 + A_01_2 + A_10_1 + A_10_2;
1511     const register double tmp19_0 = A_01_2 + A_10_2;
1512     const register double tmp14_1 = A_10_0*w8;
1513     const register double tmp23_1 = tmp3_0*w14;
1514     const register double tmp35_1 = A_01_0*w8;
1515     const register double tmp54_1 = tmp13_0*w8;
1516     const register double tmp20_1 = tmp9_0*w4;
1517     const register double tmp25_1 = tmp12_0*w12;
1518     const register double tmp2_1 = A_01_1*w4;
1519     const register double tmp44_1 = tmp7_0*w7;
1520     const register double tmp26_1 = tmp10_0*w4;
1521     const register double tmp52_1 = tmp18_0*w8;
1522     const register double tmp48_1 = A_10_1*w7;
1523     const register double tmp46_1 = A_01_3*w8;
1524     const register double tmp50_1 = A_01_0*w2;
1525     const register double tmp8_1 = tmp4_0*w5;
1526     const register double tmp56_1 = tmp19_0*w8;
1527     const register double tmp9_1 = tmp2_0*w10;
1528     const register double tmp19_1 = A_10_3*w2;
1529     const register double tmp47_1 = A_10_2*w4;
1530     const register double tmp16_1 = tmp3_0*w0;
1531     const register double tmp18_1 = tmp1_0*w6;
1532     const register double tmp31_1 = tmp11_0*w12;
1533     const register double tmp55_1 = tmp15_0*w2;
1534     const register double tmp39_1 = A_10_2*w7;
1535     const register double tmp11_1 = tmp6_0*w7;
1536     const register double tmp40_1 = tmp11_0*w17;
1537     const register double tmp34_1 = tmp15_0*w8;
1538     const register double tmp33_1 = tmp14_0*w5;
1539     const register double tmp24_1 = tmp11_0*w13;
1540     const register double tmp3_1 = tmp1_0*w0;
1541     const register double tmp5_1 = tmp2_0*w3;
1542     const register double tmp43_1 = tmp17_0*w5;
1543     const register double tmp15_1 = A_01_2*w4;
1544     const register double tmp53_1 = tmp19_0*w2;
1545     const register double tmp27_1 = tmp3_0*w11;
1546     const register double tmp32_1 = tmp13_0*w2;
1547     const register double tmp10_1 = tmp5_0*w9;
1548     const register double tmp37_1 = A_10_1*w4;
1549     const register double tmp38_1 = tmp5_0*w15;
1550     const register double tmp17_1 = A_01_1*w7;
1551     const register double tmp12_1 = tmp7_0*w4;
1552     const register double tmp22_1 = tmp10_0*w7;
1553     const register double tmp57_1 = tmp18_0*w2;
1554     const register double tmp28_1 = tmp9_0*w7;
1555     const register double tmp29_1 = tmp1_0*w14;
1556     const register double tmp51_1 = tmp11_0*w16;
1557     const register double tmp42_1 = tmp12_0*w16;
1558     const register double tmp49_1 = tmp12_0*w17;
1559     const register double tmp21_1 = tmp1_0*w11;
1560     const register double tmp1_1 = tmp0_0*w1;
1561     const register double tmp45_1 = tmp6_0*w4;
1562     const register double tmp7_1 = A_10_0*w2;
1563     const register double tmp6_1 = tmp3_0*w6;
1564     const register double tmp13_1 = tmp8_0*w1;
1565     const register double tmp36_1 = tmp16_0*w1;
1566     const register double tmp41_1 = A_01_3*w2;
1567     const register double tmp30_1 = tmp12_0*w13;
1568     const register double tmp4_1 = A_01_2*w7;
1569     const register double tmp0_1 = A_10_3*w8;
1570     EM_S[INDEX2(0,1,4)]+=tmp0_1 + tmp1_1 + tmp2_1 + tmp3_1 + tmp4_1 + tmp5_1 + tmp6_1 + tmp7_1 + tmp8_1;
1571     EM_S[INDEX2(1,2,4)]+=tmp10_1 + tmp11_1 + tmp12_1 + tmp13_1 + tmp9_1;
1572     EM_S[INDEX2(3,2,4)]+=tmp14_1 + tmp15_1 + tmp16_1 + tmp17_1 + tmp18_1 + tmp19_1 + tmp1_1 + tmp5_1 + tmp8_1;
1573     EM_S[INDEX2(0,0,4)]+=tmp13_1 + tmp20_1 + tmp21_1 + tmp22_1 + tmp23_1 + tmp24_1 + tmp25_1;
1574     EM_S[INDEX2(3,3,4)]+=tmp13_1 + tmp26_1 + tmp27_1 + tmp28_1 + tmp29_1 + tmp30_1 + tmp31_1;
1575     EM_S[INDEX2(3,0,4)]+=tmp10_1 + tmp32_1 + tmp33_1 + tmp34_1 + tmp9_1;
1576     EM_S[INDEX2(3,1,4)]+=tmp35_1 + tmp36_1 + tmp37_1 + tmp38_1 + tmp39_1 + tmp40_1 + tmp41_1 + tmp42_1 + tmp43_1;
1577     EM_S[INDEX2(2,1,4)]+=tmp10_1 + tmp13_1 + tmp44_1 + tmp45_1 + tmp9_1;
1578     EM_S[INDEX2(0,2,4)]+=tmp36_1 + tmp38_1 + tmp43_1 + tmp46_1 + tmp47_1 + tmp48_1 + tmp49_1 + tmp50_1 + tmp51_1;
1579     EM_S[INDEX2(2,0,4)]+=tmp0_1 + tmp15_1 + tmp17_1 + tmp1_1 + tmp38_1 + tmp49_1 + tmp51_1 + tmp7_1 + tmp8_1;
1580     EM_S[INDEX2(1,3,4)]+=tmp14_1 + tmp19_1 + tmp1_1 + tmp2_1 + tmp38_1 + tmp40_1 + tmp42_1 + tmp4_1 + tmp8_1;
1581     EM_S[INDEX2(2,3,4)]+=tmp16_1 + tmp18_1 + tmp35_1 + tmp36_1 + tmp41_1 + tmp43_1 + tmp47_1 + tmp48_1 + tmp5_1;
1582     EM_S[INDEX2(2,2,4)]+=tmp24_1 + tmp25_1 + tmp27_1 + tmp29_1 + tmp33_1 + tmp52_1 + tmp53_1;
1583     EM_S[INDEX2(1,0,4)]+=tmp36_1 + tmp37_1 + tmp39_1 + tmp3_1 + tmp43_1 + tmp46_1 + tmp50_1 + tmp5_1 + tmp6_1;
1584     EM_S[INDEX2(0,3,4)]+=tmp10_1 + tmp33_1 + tmp54_1 + tmp55_1 + tmp9_1;
1585     EM_S[INDEX2(1,1,4)]+=tmp21_1 + tmp23_1 + tmp30_1 + tmp31_1 + tmp33_1 + tmp56_1 + tmp57_1;
1586     } else { /* constant data */
1587     const register double A_00 = A_p[INDEX2(0,0,2)];
1588     const register double A_01 = A_p[INDEX2(0,1,2)];
1589     const register double A_10 = A_p[INDEX2(1,0,2)];
1590     const register double A_11 = A_p[INDEX2(1,1,2)];
1591     const register double tmp0_0 = A_01 + A_10;
1592     const register double tmp0_1 = A_00*w18;
1593     const register double tmp10_1 = A_01*w20;
1594     const register double tmp12_1 = A_00*w26;
1595     const register double tmp4_1 = A_00*w22;
1596     const register double tmp8_1 = A_00*w24;
1597     const register double tmp13_1 = A_10*w19;
1598     const register double tmp9_1 = tmp0_0*w20;
1599     const register double tmp3_1 = A_11*w21;
1600     const register double tmp11_1 = A_11*w27;
1601     const register double tmp1_1 = A_01*w19;
1602     const register double tmp6_1 = A_11*w23;
1603     const register double tmp7_1 = A_11*w25;
1604     const register double tmp2_1 = A_10*w20;
1605     const register double tmp5_1 = tmp0_0*w19;
1606     EM_S[INDEX2(0,1,4)]+=tmp0_1 + tmp1_1 + tmp2_1 + tmp3_1;
1607     EM_S[INDEX2(1,2,4)]+=tmp4_1 + tmp5_1 + tmp6_1;
1608     EM_S[INDEX2(3,2,4)]+=tmp0_1 + tmp1_1 + tmp2_1 + tmp3_1;
1609     EM_S[INDEX2(0,0,4)]+=tmp5_1 + tmp7_1 + tmp8_1;
1610     EM_S[INDEX2(3,3,4)]+=tmp5_1 + tmp7_1 + tmp8_1;
1611     EM_S[INDEX2(3,0,4)]+=tmp4_1 + tmp6_1 + tmp9_1;
1612     EM_S[INDEX2(3,1,4)]+=tmp10_1 + tmp11_1 + tmp12_1 + tmp13_1;
1613     EM_S[INDEX2(2,1,4)]+=tmp4_1 + tmp5_1 + tmp6_1;
1614     EM_S[INDEX2(0,2,4)]+=tmp10_1 + tmp11_1 + tmp12_1 + tmp13_1;
1615     EM_S[INDEX2(2,0,4)]+=tmp11_1 + tmp12_1 + tmp1_1 + tmp2_1;
1616     EM_S[INDEX2(1,3,4)]+=tmp11_1 + tmp12_1 + tmp1_1 + tmp2_1;
1617     EM_S[INDEX2(2,3,4)]+=tmp0_1 + tmp10_1 + tmp13_1 + tmp3_1;
1618     EM_S[INDEX2(2,2,4)]+=tmp7_1 + tmp8_1 + tmp9_1;
1619     EM_S[INDEX2(1,0,4)]+=tmp0_1 + tmp10_1 + tmp13_1 + tmp3_1;
1620     EM_S[INDEX2(0,3,4)]+=tmp4_1 + tmp6_1 + tmp9_1;
1621     EM_S[INDEX2(1,1,4)]+=tmp7_1 + tmp8_1 + tmp9_1;
1622     }
1623     }
1624     ///////////////
1625     // process B //
1626     ///////////////
1627     if (!B.isEmpty()) {
1628     add_EM_S=true;
1629     const double* B_p=const_cast<escript::Data*>(&B)->getSampleDataRO(e);
1630     if (B.actsExpanded()) {
1631     const register double B_0_0 = B_p[INDEX2(0,0,2)];
1632     const register double B_1_0 = B_p[INDEX2(1,0,2)];
1633     const register double B_0_1 = B_p[INDEX2(0,1,2)];
1634     const register double B_1_1 = B_p[INDEX2(1,1,2)];
1635     const register double B_0_2 = B_p[INDEX2(0,2,2)];
1636     const register double B_1_2 = B_p[INDEX2(1,2,2)];
1637     const register double B_0_3 = B_p[INDEX2(0,3,2)];
1638     const register double B_1_3 = B_p[INDEX2(1,3,2)];
1639     const register double tmp3_0 = B_0_0 + B_0_2;
1640     const register double tmp1_0 = B_1_2 + B_1_3;
1641     const register double tmp2_0 = B_0_1 + B_0_3;
1642     const register double tmp0_0 = B_1_0 + B_1_1;
1643     const register double tmp63_1 = B_1_1*w42;
1644     const register double tmp79_1 = B_1_1*w40;
1645     const register double tmp37_1 = tmp3_0*w35;
1646     const register double tmp8_1 = tmp0_0*w32;
1647     const register double tmp71_1 = B_0_1*w34;
1648     const register double tmp19_1 = B_0_3*w31;
1649     const register double tmp15_1 = B_0_3*w34;
1650     const register double tmp9_1 = tmp3_0*w34;
1651     const register double tmp35_1 = B_1_0*w36;
1652     const register double tmp66_1 = B_0_3*w28;
1653     const register double tmp28_1 = B_1_0*w42;
1654     const register double tmp22_1 = B_1_0*w40;
1655     const register double tmp16_1 = B_1_2*w29;
1656     const register double tmp6_1 = tmp2_0*w35;
1657     const register double tmp55_1 = B_1_3*w40;
1658     const register double tmp50_1 = B_1_3*w42;
1659     const register double tmp7_1 = tmp1_0*w29;
1660     const register double tmp1_1 = tmp1_0*w32;
1661     const register double tmp57_1 = B_0_3*w30;
1662     const register double tmp18_1 = B_1_1*w32;
1663     const register double tmp53_1 = B_1_0*w41;
1664     const register double tmp61_1 = B_1_3*w36;
1665     const register double tmp27_1 = B_0_3*w38;
1666     const register double tmp64_1 = B_0_2*w30;
1667     const register double tmp76_1 = B_0_1*w38;
1668     const register double tmp39_1 = tmp2_0*w34;
1669     const register double tmp62_1 = B_0_1*w31;
1670     const register double tmp56_1 = B_0_0*w31;
1671     const register double tmp49_1 = B_1_1*w36;
1672     const register double tmp2_1 = B_0_2*w31;
1673     const register double tmp23_1 = B_0_2*w33;
1674     const register double tmp38_1 = B_1_1*w43;
1675     const register double tmp74_1 = B_1_2*w41;
1676     const register double tmp43_1 = B_1_1*w41;
1677     const register double tmp58_1 = B_0_2*w28;
1678     const register double tmp67_1 = B_0_0*w33;
1679     const register double tmp33_1 = tmp0_0*w39;
1680     const register double tmp4_1 = B_0_0*w28;
1681     const register double tmp20_1 = B_0_0*w30;
1682     const register double tmp13_1 = B_0_2*w38;
1683     const register double tmp65_1 = B_1_2*w43;
1684     const register double tmp0_1 = tmp0_0*w29;
1685     const register double tmp41_1 = tmp3_0*w33;
1686     const register double tmp73_1 = B_0_2*w37;
1687     const register double tmp69_1 = B_0_0*w38;
1688     const register double tmp48_1 = B_1_2*w39;
1689     const register double tmp59_1 = B_0_1*w33;
1690     const register double tmp17_1 = B_1_3*w41;
1691     const register double tmp5_1 = B_0_3*w33;
1692     const register double tmp3_1 = B_0_1*w30;
1693     const register double tmp21_1 = B_0_1*w28;
1694     const register double tmp42_1 = B_1_0*w29;
1695     const register double tmp54_1 = B_1_2*w32;
1696     const register double tmp60_1 = B_1_0*w39;
1697     const register double tmp32_1 = tmp1_0*w36;
1698     const register double tmp10_1 = B_0_1*w37;
1699     const register double tmp14_1 = B_0_0*w35;
1700     const register double tmp29_1 = B_0_1*w35;
1701     const register double tmp26_1 = B_1_2*w36;
1702     const register double tmp30_1 = B_1_3*w43;
1703     const register double tmp70_1 = B_0_2*w35;
1704     const register double tmp34_1 = B_1_3*w39;
1705     const register double tmp51_1 = B_1_0*w43;
1706     const register double tmp31_1 = B_0_2*w34;
1707     const register double tmp45_1 = tmp3_0*w28;
1708     const register double tmp11_1 = tmp1_0*w39;
1709     const register double tmp52_1 = B_1_1*w29;
1710     const register double tmp44_1 = B_1_3*w32;
1711     const register double tmp25_1 = B_1_1*w39;
1712     const register double tmp47_1 = tmp2_0*w33;
1713     const register double tmp72_1 = B_1_3*w29;
1714     const register double tmp40_1 = tmp2_0*w28;
1715     const register double tmp46_1 = B_1_2*w40;
1716     const register double tmp36_1 = B_1_2*w42;
1717     const register double tmp24_1 = B_0_0*w37;
1718     const register double tmp77_1 = B_0_3*w35;
1719     const register double tmp68_1 = B_0_3*w37;
1720     const register double tmp78_1 = B_0_0*w34;
1721     const register double tmp12_1 = tmp0_0*w36;
1722     const register double tmp75_1 = B_1_0*w32;
1723     EM_S[INDEX2(0,1,4)]+=tmp0_1 + tmp1_1 + tmp2_1 + tmp3_1 + tmp4_1 + tmp5_1;
1724     EM_S[INDEX2(1,2,4)]+=tmp6_1 + tmp7_1 + tmp8_1 + tmp9_1;
1725     EM_S[INDEX2(3,2,4)]+=tmp10_1 + tmp11_1 + tmp12_1 + tmp13_1 + tmp14_1 + tmp15_1;
1726     EM_S[INDEX2(0,0,4)]+=tmp16_1 + tmp17_1 + tmp18_1 + tmp19_1 + tmp20_1 + tmp21_1 + tmp22_1 + tmp23_1;
1727     EM_S[INDEX2(3,3,4)]+=tmp24_1 + tmp25_1 + tmp26_1 + tmp27_1 + tmp28_1 + tmp29_1 + tmp30_1 + tmp31_1;
1728     EM_S[INDEX2(3,0,4)]+=tmp32_1 + tmp33_1 + tmp6_1 + tmp9_1;
1729     EM_S[INDEX2(3,1,4)]+=tmp34_1 + tmp35_1 + tmp36_1 + tmp37_1 + tmp38_1 + tmp39_1;
1730     EM_S[INDEX2(2,1,4)]+=tmp32_1 + tmp33_1 + tmp40_1 + tmp41_1;
1731     EM_S[INDEX2(0,2,4)]+=tmp42_1 + tmp43_1 + tmp44_1 + tmp45_1 + tmp46_1 + tmp47_1;
1732     EM_S[INDEX2(2,0,4)]+=tmp45_1 + tmp47_1 + tmp48_1 + tmp49_1 + tmp50_1 + tmp51_1;
1733     EM_S[INDEX2(1,3,4)]+=tmp37_1 + tmp39_1 + tmp52_1 + tmp53_1 + tmp54_1 + tmp55_1;
1734     EM_S[INDEX2(2,3,4)]+=tmp11_1 + tmp12_1 + tmp56_1 + tmp57_1 + tmp58_1 + tmp59_1;
1735     EM_S[INDEX2(2,2,4)]+=tmp60_1 + tmp61_1 + tmp62_1 + tmp63_1 + tmp64_1 + tmp65_1 + tmp66_1 + tmp67_1;
1736     EM_S[INDEX2(1,0,4)]+=tmp0_1 + tmp1_1 + tmp68_1 + tmp69_1 + tmp70_1 + tmp71_1;
1737     EM_S[INDEX2(0,3,4)]+=tmp40_1 + tmp41_1 + tmp7_1 + tmp8_1;
1738     EM_S[INDEX2(1,1,4)]+=tmp72_1 + tmp73_1 + tmp74_1 + tmp75_1 + tmp76_1 + tmp77_1 + tmp78_1 + tmp79_1;
1739     } else { /* constant data */
1740     const register double B_0 = B_p[0];
1741     const register double B_1 = B_p[1];
1742     const register double tmp6_1 = B_1*w50;
1743     const register double tmp1_1 = B_1*w45;
1744     const register double tmp5_1 = B_1*w49;
1745     const register double tmp4_1 = B_1*w48;
1746     const register double tmp0_1 = B_0*w44;
1747     const register double tmp2_1 = B_0*w46;
1748     const register double tmp7_1 = B_0*w51;
1749     const register double tmp3_1 = B_0*w47;
1750     EM_S[INDEX2(0,1,4)]+=tmp0_1 + tmp1_1;
1751     EM_S[INDEX2(1,2,4)]+=tmp1_1 + tmp2_1;
1752     EM_S[INDEX2(3,2,4)]+=tmp3_1 + tmp4_1;
1753     EM_S[INDEX2(0,0,4)]+=tmp0_1 + tmp5_1;
1754     EM_S[INDEX2(3,3,4)]+=tmp3_1 + tmp6_1;
1755     EM_S[INDEX2(3,0,4)]+=tmp2_1 + tmp4_1;
1756     EM_S[INDEX2(3,1,4)]+=tmp2_1 + tmp6_1;
1757     EM_S[INDEX2(2,1,4)]+=tmp4_1 + tmp7_1;
1758     EM_S[INDEX2(0,2,4)]+=tmp5_1 + tmp7_1;
1759     EM_S[INDEX2(2,0,4)]+=tmp6_1 + tmp7_1;
1760     EM_S[INDEX2(1,3,4)]+=tmp2_1 + tmp5_1;
1761     EM_S[INDEX2(2,3,4)]+=tmp0_1 + tmp4_1;
1762     EM_S[INDEX2(2,2,4)]+=tmp0_1 + tmp6_1;
1763     EM_S[INDEX2(1,0,4)]+=tmp1_1 + tmp3_1;
1764     EM_S[INDEX2(0,3,4)]+=tmp1_1 + tmp7_1;
1765     EM_S[INDEX2(1,1,4)]+=tmp3_1 + tmp5_1;
1766     }
1767     }
1768     ///////////////
1769     // process C //
1770     ///////////////
1771     if (!C.isEmpty()) {
1772     add_EM_S=true;
1773     const double* C_p=const_cast<escript::Data*>(&C)->getSampleDataRO(e);
1774     if (C.actsExpanded()) {
1775     const register double C_0_0 = C_p[INDEX2(0,0,2)];
1776     const register double C_1_0 = C_p[INDEX2(1,0,2)];
1777     const register double C_0_1 = C_p[INDEX2(0,1,2)];
1778     const register double C_1_1 = C_p[INDEX2(1,1,2)];
1779     const register double C_0_2 = C_p[INDEX2(0,2,2)];
1780     const register double C_1_2 = C_p[INDEX2(1,2,2)];
1781     const register double C_0_3 = C_p[INDEX2(0,3,2)];
1782     const register double C_1_3 = C_p[INDEX2(1,3,2)];
1783     const register double tmp2_0 = C_0_1 + C_0_3;
1784     const register double tmp1_0 = C_1_2 + C_1_3;
1785     const register double tmp3_0 = C_0_0 + C_0_2;
1786     const register double tmp0_0 = C_1_0 + C_1_1;
1787     const register double tmp64_1 = C_0_2*w30;
1788     const register double tmp14_1 = C_0_2*w28;
1789     const register double tmp19_1 = C_0_3*w31;
1790     const register double tmp22_1 = C_1_0*w40;
1791     const register double tmp37_1 = tmp3_0*w35;
1792     const register double tmp29_1 = C_0_1*w35;
1793     const register double tmp73_1 = C_0_2*w37;
1794     const register double tmp74_1 = C_1_2*w41;
1795     const register double tmp52_1 = C_1_3*w39;
1796     const register double tmp25_1 = C_1_1*w39;
1797     const register double tmp62_1 = C_0_1*w31;
1798     const register double tmp79_1 = C_1_1*w40;
1799     const register double tmp43_1 = C_1_1*w36;
1800     const register double tmp27_1 = C_0_3*w38;
1801     const register double tmp28_1 = C_1_0*w42;
1802     const register double tmp63_1 = C_1_1*w42;
1803     const register double tmp59_1 = C_0_3*w34;
1804     const register double tmp72_1 = C_1_3*w29;
1805     const register double tmp40_1 = tmp2_0*w35;
1806     const register double tmp13_1 = C_0_3*w30;
1807     const register double tmp51_1 = C_1_2*w40;
1808     const register double tmp54_1 = C_1_2*w42;
1809     const register double tmp12_1 = C_0_0*w31;
1810     const register double tmp2_1 = tmp1_0*w32;
1811     const register double tmp68_1 = C_0_2*w31;
1812     const register double tmp75_1 = C_1_0*w32;
1813     const register double tmp49_1 = C_1_1*w41;
1814     const register double tmp4_1 = C_0_2*w35;
1815     const register double tmp66_1 = C_0_3*w28;
1816     const register double tmp56_1 = C_0_1*w37;
1817     const register double tmp5_1 = C_0_1*w34;
1818     const register double tmp38_1 = tmp2_0*w34;
1819     const register double tmp76_1 = C_0_1*w38;
1820     const register double tmp21_1 = C_0_1*w28;
1821     const register double tmp69_1 = C_0_1*w30;
1822     const register double tmp53_1 = C_1_0*w36;
1823     const register double tmp42_1 = C_1_2*w39;
1824     const register double tmp32_1 = tmp1_0*w29;
1825     const register double tmp45_1 = C_1_0*w43;
1826     const register double tmp33_1 = tmp0_0*w32;
1827     const register double tmp35_1 = C_1_0*w41;
1828     const register double tmp26_1 = C_1_2*w36;
1829     const register double tmp67_1 = C_0_0*w33;
1830     const register double tmp31_1 = C_0_2*w34;
1831     const register double tmp20_1 = C_0_0*w30;
1832     const register double tmp70_1 = C_0_0*w28;
1833     const register double tmp8_1 = tmp0_0*w39;
1834     const register double tmp30_1 = C_1_3*w43;
1835     const register double tmp0_1 = tmp0_0*w29;
1836     const register double tmp17_1 = C_1_3*w41;
1837     const register double tmp58_1 = C_0_0*w35;
1838     const register double tmp9_1 = tmp3_0*w33;
1839     const register double tmp61_1 = C_1_3*w36;
1840     const