ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/OpenMD/branches/development/src/math/ConvexHull.cpp
(Generate patch)

Comparing:
trunk/src/math/ConvexHull.cpp (file contents), Revision 1141 by chuckv, Wed May 30 18:47:04 2007 UTC vs.
branches/development/src/math/ConvexHull.cpp (file contents), Revision 1465 by chuckv, Fri Jul 9 23:08:25 2010 UTC

# Line 1 | Line 1
1 < /* Copyright (c) 2006 The University of Notre Dame. All Rights Reserved.
1 > /* Copyright (c) 2010 The University of Notre Dame. All Rights Reserved.
2   *
3   * The University of Notre Dame grants you ("Licensee") a
4   * non-exclusive, royalty free, license to use, modify and
5   * redistribute this software in source and binary code form, provided
6   * that the following conditions are met:
7   *
8 < * 1. Acknowledgement of the program authors must be made in any
9 < *    publication of scientific results based in part on use of the
10 < *    program.  An acceptable form of acknowledgement is citation of
11 < *    the article in which the program was described (Matthew
12 < *    A. Meineke, Charles F. Vardeman II, Teng Lin, Christopher
13 < *    J. Fennell and J. Daniel Gezelter, "OOPSE: An Object-Oriented
14 < *    Parallel Simulation Engine for Molecular Dynamics,"
15 < *    J. Comput. Chem. 26, pp. 252-271 (2005))
16 < *
17 < * 2. Redistributions of source code must retain the above copyright
8 > * 1. Redistributions of source code must retain the above copyright
9   *    notice, this list of conditions and the following disclaimer.
10   *
11 < * 3. Redistributions in binary form must reproduce the above copyright
11 > * 2. Redistributions in binary form must reproduce the above copyright
12   *    notice, this list of conditions and the following disclaimer in the
13   *    documentation and/or other materials provided with the
14   *    distribution.
# Line 37 | Line 28
28   * University of Notre Dame has been advised of the possibility of
29   * such damages.
30   *
31 + * SUPPORT OPEN SCIENCE!  If you use OpenMD or its source code in your
32 + * research, please cite the appropriate papers when you publish your
33 + * work.  Good starting points are:
34 + *                                                                      
35 + * [1]  Meineke, et al., J. Comp. Chem. 26, 252-271 (2005).            
36 + * [2]  Fennell & Gezelter, J. Chem. Phys. 124, 234104 (2006).          
37 + * [3]  Sun, Lin & Gezelter, J. Chem. Phys. 128, 24107 (2008).          
38 + * [4]  Vardeman & Gezelter, in progress (2009).                        
39   *
40 + *
41   *  ConvexHull.cpp
42   *
43   *  Purpose: To calculate convexhull, hull volume libqhull.
44   *
45   *  Created by Charles F. Vardeman II on 11 Dec 2006.
46   *  @author  Charles F. Vardeman II
47 < *  @version $Id: ConvexHull.cpp,v 1.3 2007-05-30 18:47:03 chuckv Exp $
47 > *  @version $Id$
48   *
49   */
50  
51 + /* Standard includes independent of library */
52 +
53   #include <iostream>
54   #include <fstream>
55 + #include <list>
56 + #include <algorithm>
57 + #include <iterator>
58   #include "math/ConvexHull.hpp"
59 + #include "utils/simError.h"
60  
61 < using namespace oopse;
61 > #ifdef IS_MPI
62 > #include <mpi.h>
63 > #endif
64  
65 < ConvexHull::ConvexHull() : dim_(3), options_("qhull Qt FA") {
65 > using namespace OpenMD;
66 >
67 > #ifdef HAVE_QHULL
68 > extern "C"
69 > {
70 > #include <qhull/qhull.h>
71 > #include <qhull/mem.h>
72 > #include <qhull/qset.h>
73 > #include <qhull/geom.h>
74 > #include <qhull/merge.h>
75 > #include <qhull/poly.h>
76 > #include <qhull/io.h>
77 > #include <qhull/stat.h>
78   }
79  
80 < bool ConvexHull::genHull(std::vector<Vector3d> pos) {
81 <  FILE *outfile = stdout;
82 <  FILE *errfile = stderr;
80 > ConvexHull::ConvexHull() : Hull(), dim_(3), options_("qhull Qt Pp") {
81 > }
82 >
83 > void ConvexHull::computeHull(std::vector<StuntDouble*> bodydoubles) {
84 >
85 >  int numpoints = bodydoubles.size();
86 >
87 >  Triangles_.clear();
88 >  
89 >  vertexT *vertex, **vertexp;
90    facetT *facet;
91 <  int exitcode;
92 <  boolT ismalloc = False;
93 <  int curlong,totlong;
91 >  setT *vertices;
92 >  int curlong, totlong;
93 >  pointT *intPoint;
94    
95 <  int numpoints = pos.size();
96 <  
97 <  coordT points[numpoints][dim_];
98 <  
99 <  for (int i=0; i<numpoints; i++) {
100 <    points[i][0] = pos[i][0];
101 <    points[i][1] = pos[i][1];
102 <    points[i][2] = pos[i][2];          
95 >  std::vector<double> ptArray(numpoints*dim_);
96 >
97 >  // Copy the positon vector into a points vector for qhull.
98 >  std::vector<StuntDouble*>::iterator SD;
99 >  int i = 0;
100 >  for (SD =bodydoubles.begin(); SD != bodydoubles.end(); ++SD){
101 >    Vector3d pos = (*SD)->getPos();      
102 >    ptArray[dim_ * i] = pos.x();
103 >    ptArray[dim_ * i + 1] = pos.y();
104 >    ptArray[dim_ * i + 2] = pos.z();
105 >    i++;
106    }
107    
108 +  /* Clean up memory from previous convex hull calculations */
109 +  boolT ismalloc = False;
110    
111 +  /* compute the hull for our local points (or all the points for single
112 +     processor versions) */
113 +  if (qh_new_qhull(dim_, numpoints, &ptArray[0], ismalloc,
114 +                   const_cast<char *>(options_.c_str()), NULL, stderr)) {
115 +    
116 +    sprintf(painCave.errMsg, "ConvexHull: Qhull failed to compute convex hull");
117 +    painCave.isFatal = 1;
118 +    simError();
119 +    
120 +  } //qh_new_qhull
121 +
122 +
123 + #ifdef IS_MPI
124 +  //If we are doing the mpi version, set up some vectors for data communication
125    
126 <  qh_initflags (const_cast<char *>(options_.c_str()));
127 <  qh_init_B (points[0], numpoints, dim_, ismalloc);
128 <  qh_qhull();
129 <  qh_check_output();
126 >  int nproc = MPI::COMM_WORLD.Get_size();
127 >  int myrank = MPI::COMM_WORLD.Get_rank();
128 >  int localHullSites = 0;
129 >
130 >  std::vector<int> hullSitesOnProc(nproc, 0);
131 >  std::vector<int> coordsOnProc(nproc, 0);
132 >  std::vector<int> displacements(nproc, 0);
133 >  std::vector<int> vectorDisplacements(nproc, 0);
134 >
135 >  std::vector<double> coords;
136 >  std::vector<double> vels;
137 >  std::vector<int> indexMap;
138 >  std::vector<double> masses;
139 >
140 >  FORALLvertices{
141 >    localHullSites++;
142 >    
143 >    int idx = qh_pointid(vertex->point);
144 >
145 >    indexMap.push_back(idx);
146 >
147 >    coords.push_back(ptArray[dim_  * idx]);
148 >    coords.push_back(ptArray[dim_  * idx + 1]);
149 >    coords.push_back(ptArray[dim_  * idx + 2]);
150 >
151 >    StuntDouble* sd = bodydoubles[idx];
152 >
153 >    Vector3d vel = sd->getVel();
154 >    vels.push_back(vel.x());
155 >    vels.push_back(vel.y());
156 >    vels.push_back(vel.z());
157 >
158 >    masses.push_back(sd->getMass());
159 >  }
160 >
161 >  MPI::COMM_WORLD.Allgather(&localHullSites, 1, MPI::INT, &hullSitesOnProc[0],
162 >                            1, MPI::INT);
163 >
164 >  int globalHullSites = 0;
165 >  for (int iproc = 0; iproc < nproc; iproc++){
166 >    globalHullSites += hullSitesOnProc[iproc];
167 >    coordsOnProc[iproc] = dim_ * hullSitesOnProc[iproc];
168 >  }
169 >
170 >  displacements[0] = 0;
171 >  vectorDisplacements[0] = 0;
172    
173 +  for (int iproc = 1; iproc < nproc; iproc++){
174 +    displacements[iproc] = displacements[iproc-1] + hullSitesOnProc[iproc-1];
175 +    vectorDisplacements[iproc] = vectorDisplacements[iproc-1] + coordsOnProc[iproc-1];
176 +  }
177 +
178 +  std::vector<double> globalCoords(dim_ * globalHullSites);
179 +  std::vector<double> globalVels(dim_ * globalHullSites);
180 +  std::vector<double> globalMasses(globalHullSites);
181 +
182 +  int count = coordsOnProc[myrank];
183    
184 <  
185 <  qh_getarea(qh facet_list);
186 <  volume_ = qh totvol;                                                  
187 <  area_ = qh totarea;                                          
188 <  
189 <  
190 <  
184 >  MPI::COMM_WORLD.Allgatherv(&coords[0], count, MPI::DOUBLE, &globalCoords[0],
185 >                             &coordsOnProc[0], &vectorDisplacements[0],
186 >                             MPI::DOUBLE);
187 >
188 >  MPI::COMM_WORLD.Allgatherv(&vels[0], count, MPI::DOUBLE, &globalVels[0],
189 >                             &coordsOnProc[0], &vectorDisplacements[0],
190 >                             MPI::DOUBLE);
191 >
192 >  MPI::COMM_WORLD.Allgatherv(&masses[0], localHullSites, MPI::DOUBLE,
193 >                             &globalMasses[0], &hullSitesOnProc[0],
194 >                             &displacements[0], MPI::DOUBLE);
195 >
196 >  // Free previous hull
197    qh_freeqhull(!qh_ALL);
198 <  qh_memfreeshort (&curlong, &totlong);
199 <  if (curlong || totlong)
200 <    fprintf (errfile, "qhull internal warning (main): did not free %d bytes of long memory (%d pieces)\n",
201 <             totlong, curlong);
198 >  qh_memfreeshort(&curlong, &totlong);
199 >  if (curlong || totlong) {
200 >    sprintf(painCave.errMsg, "ConvexHull: qhull internal warning:\n"
201 >            "\tdid not free %d bytes of long memory (%d pieces)",
202 >            totlong, curlong);
203 >    painCave.isFatal = 1;
204 >    simError();
205 >  }
206    
207 <  
208 <  return true;
209 < }
207 >  if (qh_new_qhull(dim_, globalHullSites, &globalCoords[0], ismalloc,
208 >                   const_cast<char *>(options_.c_str()), NULL, stderr)){
209 >    
210 >    sprintf(painCave.errMsg, "ConvexHull: Qhull failed to compute global convex hull");
211 >    painCave.isFatal = 1;
212 >    simError();
213 >    
214 >  } //qh_new_qhull
215  
216 + #endif
217 +  intPoint = qh interior_point;
218 +  RealType calcvol = 0.0;
219 +  FORALLfacets {  
220 +    Triangle face;
221 +    //Qhull sets the unit normal in facet->normal
222 +    Vector3d V3dNormal(facet->normal[0], facet->normal[1], facet->normal[2]);
223 +    face.setUnitNormal(V3dNormal);
224 +    
225 +    RealType faceArea = qh_facetarea(facet);
226 +    face.setArea(faceArea);
227 +    
228 +    vertices = qh_facet3vertex(facet);
229 +      
230 +    coordT *center = qh_getcenter(vertices);
231 +    Vector3d V3dCentroid(center[0], center[1], center[2]);
232 +    face.setCentroid(V3dCentroid);
233  
234 < RealType ConvexHull::getVolume() {
235 <  return volume_;
234 >    Vector3d faceVel = V3Zero;
235 >    Vector3d p[3];
236 >    RealType faceMass = 0.0;
237 >
238 >    int ver = 0;
239 >
240 >    FOREACHvertex_(vertices){
241 >      int id = qh_pointid(vertex->point);
242 >      p[ver][0] = vertex->point[0];
243 >      p[ver][1] = vertex->point[1];
244 >      p[ver][2] = vertex->point[2];
245 >      Vector3d vel;
246 >      RealType mass;
247 >
248 > #ifdef IS_MPI
249 >      vel = Vector3d(globalVels[dim_ * id],
250 >                     globalVels[dim_ * id + 1],
251 >                     globalVels[dim_ * id + 2]);
252 >      mass = globalMasses[id];
253 >
254 >      // localID will be between 0 and hullSitesOnProc[myrank] if we
255 >      // own this guy.
256 >
257 >      int localID = id - displacements[myrank];
258 >
259 >
260 >      if (localID >= 0 && localID < hullSitesOnProc[myrank]){
261 >        face.addVertexSD(bodydoubles[indexMap[localID]]);
262 >      }else{
263 >        face.addVertexSD(NULL);
264 >      }
265 > #else
266 >      vel = bodydoubles[id]->getVel();
267 >      mass = bodydoubles[id]->getMass();
268 >      face.addVertexSD(bodydoubles[id]);      
269 > #endif  
270 >      faceVel = faceVel + vel;
271 >      faceMass = faceMass + mass;
272 >      ver++;      
273 >    } //Foreachvertex
274 >
275 >    face.addVertices(p[0], p[1], p[2]);
276 >    face.setFacetMass(faceMass);
277 >    face.setFacetVelocity(faceVel/3.0);
278 >    /*
279 >    RealType comparea = face.computeArea();
280 >    realT calcarea = qh_facetarea (facet);
281 >    Vector3d V3dCompNorm = -face.computeUnitNormal();
282 >    RealType thisOffset = ((0.0-p[0][0])*V3dCompNorm[0] + (0.0-p[0][1])*V3dCompNorm[1] + (0.0-p[0][2])*V3dCompNorm[2]);
283 >    RealType dist = facet->offset + intPoint[0]*V3dNormal[0] + intPoint[1]*V3dNormal[1] + intPoint[2]*V3dNormal[2];
284 >    std::cout << "facet offset and computed offset: " << facet->offset << "  " << thisOffset <<  std::endl;
285 >    calcvol +=  -dist*comparea/qh hull_dim;
286 >    */
287 >    Triangles_.push_back(face);
288 >    qh_settempfree(&vertices);      
289 >
290 >  } //FORALLfacets
291 >  
292 >  qh_getarea(qh facet_list);
293 >  volume_ = qh totvol;
294 >  area_ = qh totarea;
295 >  //  std::cout << "My volume is: " << calcvol << " qhull volume is:" << volume_ << std::endl;
296 >  qh_freeqhull(!qh_ALL);
297 >  qh_memfreeshort(&curlong, &totlong);
298 >  if (curlong || totlong) {
299 >    sprintf(painCave.errMsg, "ConvexHull: qhull internal warning:\n"
300 >            "\tdid not free %d bytes of long memory (%d pieces)",
301 >            totlong, curlong);
302 >    painCave.isFatal = 1;
303 >    simError();
304 >  }
305   }
306  
307 < void ConvexHull::geomviewHull(const std::string& geomFileName) {
307 > void ConvexHull::printHull(const std::string& geomFileName) {
308 >
309 > #ifdef IS_MPI
310 >  if (worldRank == 0)  {
311 > #endif
312 >  FILE *newGeomFile;
313    
110  std::ofstream newGeomFile;
111  
314    //create new .md file based on old .md file
315 <  newGeomFile.open(geomFileName.c_str());
315 >  newGeomFile = fopen(geomFileName.c_str(), "w");
316 >  qh_findgood_all(qh facet_list);
317 >  for (int i = 0; i < qh_PRINTEND; i++)
318 >    qh_printfacets(newGeomFile, qh PRINTout[i], qh facet_list, NULL, !qh_ALL);
319    
320 <  
321 <  newGeomFile.close();
320 >  fclose(newGeomFile);
321 > #ifdef IS_MPI
322 >  }
323 > #endif
324   }
325 + #endif //QHULL

Comparing:
trunk/src/math/ConvexHull.cpp (property svn:keywords), Revision 1141 by chuckv, Wed May 30 18:47:04 2007 UTC vs.
branches/development/src/math/ConvexHull.cpp (property svn:keywords), Revision 1465 by chuckv, Fri Jul 9 23:08:25 2010 UTC

# Line 0 | Line 1
1 + Author Id Revision Date

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines