Compadre  1.5.5
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Pages
GMLS_NeumannGradScalar.cpp
Go to the documentation of this file.
1 #include <iostream>
2 #include <string>
3 #include <vector>
4 #include <map>
5 #include <stdlib.h>
6 #include <cstdio>
7 #include <random>
8 
9 #include <Compadre_Config.h>
10 #include <Compadre_GMLS.hpp>
11 #include <Compadre_Evaluator.hpp>
13 
14 #include "GMLS_Tutorial.hpp"
15 #include "CommandLineProcessor.hpp"
16 
17 #ifdef COMPADRE_USE_MPI
18 #include <mpi.h>
19 #endif
20 
21 #include <Kokkos_Timer.hpp>
22 #include <Kokkos_Core.hpp>
23 
24 using namespace Compadre;
25 
26 //! [Parse Command Line Arguments]
27 
28 // called from command line
29 int main (int argc, char* args[]) {
30 
31 // initializes MPI (if available) with command line arguments given
32 #ifdef COMPADRE_USE_MPI
33 MPI_Init(&argc, &args);
34 #endif
35 
36 // initializes Kokkos with command line arguments given
37 Kokkos::initialize(argc, args);
38 
39 // becomes false if the computed solution not within the failure_threshold of the actual solution
40 bool all_passed = true;
41 
42 // code block to reduce scope for all Kokkos View allocations
43 // otherwise, Views may be deallocating when we call Kokkos finalize() later
44 {
45 
46  CommandLineProcessor clp(argc, args);
47  auto order = clp.order;
48  auto dimension = clp.dimension;
49  auto number_target_coords = clp.number_target_coords;
50  auto constraint_name = clp.constraint_name;
51  auto solver_name = clp.solver_name;
52  auto problem_name = clp.problem_name;
53  auto number_of_batches = clp.number_of_batches;
54 
55  // the functions we will be seeking to reconstruct are in the span of the basis
56  // of the reconstruction space we choose for GMLS, so the error should be very small
57  const double failure_tolerance = 1e-9;
58 
59  // minimum neighbors for unisolvency is the same as the size of the polynomial basis
60  const int min_neighbors = Compadre::GMLS::getNP(order, dimension);
61 
62  //! [Parse Command Line Arguments]
63  Kokkos::Timer timer;
64  Kokkos::Profiling::pushRegion("Setup Point Data");
65  //! [Setting Up The Point Cloud]
66 
67  // approximate spacing of source sites
68  double h_spacing = 0.05;
69  int n_neg1_to_1 = 2*(1/h_spacing) + 1; // always odd
70 
71  // number of source coordinate sites that will fill a box of [-1,1]x[-1,1]x[-1,1] with a spacing approximately h
72  const int number_source_coords = std::pow(n_neg1_to_1, dimension);
73 
74  // coordinates of source sites
75  Kokkos::View<double**, Kokkos::DefaultExecutionSpace> source_coords_device("source coordinates",
76  number_source_coords, 3);
77  Kokkos::View<double**>::HostMirror source_coords = Kokkos::create_mirror_view(source_coords_device);
78 
79  // coordinates of target sites
80  Kokkos::View<double**, Kokkos::DefaultExecutionSpace> target_coords_device ("target coordinates", number_target_coords, 3);
81  Kokkos::View<double**>::HostMirror target_coords = Kokkos::create_mirror_view(target_coords_device);
82 
83  // tangent bundle for each target sites
84  Kokkos::View<double***, Kokkos::DefaultExecutionSpace> tangent_bundles_device ("tangent bundles", number_target_coords, dimension, dimension);
85  Kokkos::View<double***>::HostMirror tangent_bundles = Kokkos::create_mirror_view(tangent_bundles_device);
86 
87  // fill source coordinates with a uniform grid
88  int source_index = 0;
89  double this_coord[3] = {0,0,0};
90  for (int i=-n_neg1_to_1/2; i<n_neg1_to_1/2+1; ++i) {
91  this_coord[0] = i*h_spacing;
92  for (int j=-n_neg1_to_1/2; j<n_neg1_to_1/2+1; ++j) {
93  this_coord[1] = j*h_spacing;
94  for (int k=-n_neg1_to_1/2; k<n_neg1_to_1/2+1; ++k) {
95  this_coord[2] = k*h_spacing;
96  if (dimension==3) {
97  source_coords(source_index,0) = this_coord[0];
98  source_coords(source_index,1) = this_coord[1];
99  source_coords(source_index,2) = this_coord[2];
100  source_index++;
101  }
102  }
103  if (dimension==2) {
104  source_coords(source_index,0) = this_coord[0];
105  source_coords(source_index,1) = this_coord[1];
106  source_coords(source_index,2) = 0;
107  source_index++;
108  }
109  }
110  if (dimension==1) {
111  source_coords(source_index,0) = this_coord[0];
112  source_coords(source_index,1) = 0;
113  source_coords(source_index,2) = 0;
114  source_index++;
115  }
116  }
117 
118  // fill target coords somewhere inside of [-0.5,0.5]x[-0.5,0.5]x[-0.5,0.5]
119  for(int i=0; i<number_target_coords; i++){
120 
121  // first, we get a uniformly random distributed direction
122  double rand_dir[3] = {0,0,0};
123 
124  for (int j=0; j<dimension; ++j) {
125  // rand_dir[j] is in [-0.5, 0.5]
126  rand_dir[j] = ((double)rand() / (double) RAND_MAX) - 0.5;
127  }
128 
129  // then we get a uniformly random radius
130  for (int j=0; j<dimension; ++j) {
131  target_coords(i,j) = rand_dir[j];
132  }
133  // target_coords(i, 2) = 1.0;
134 
135  // Set tangent bundles
136  if (dimension == 2) {
137  tangent_bundles(i, 0, 0) = 0.0;
138  tangent_bundles(i, 0, 1) = 0.0;
139  tangent_bundles(i, 1, 0) = 1.0/(sqrt(2.0));
140  tangent_bundles(i, 1, 1) = 1.0/(sqrt(2.0));
141  } else if (dimension == 3) {
142  tangent_bundles(i, 0, 0) = 0.0;
143  tangent_bundles(i, 0, 1) = 0.0;
144  tangent_bundles(i, 0, 2) = 0.0;
145  tangent_bundles(i, 1, 0) = 0.0;
146  tangent_bundles(i, 1, 1) = 0.0;
147  tangent_bundles(i, 1, 2) = 0.0;
148  tangent_bundles(i, 2, 0) = 1.0/(sqrt(3.0));
149  tangent_bundles(i, 2, 1) = 1.0/(sqrt(3.0));
150  tangent_bundles(i, 2, 2) = 1.0/(sqrt(3.0));
151  }
152  }
153 
154  //! [Setting Up The Point Cloud]
155 
156  Kokkos::Profiling::popRegion();
157  Kokkos::Profiling::pushRegion("Creating Data");
158 
159  //! [Creating The Data]
160 
161 
162  // source coordinates need copied to device before using to construct sampling data
163  Kokkos::deep_copy(source_coords_device, source_coords);
164 
165  // target coordinates copied next, because it is a convenient time to send them to device
166  Kokkos::deep_copy(target_coords_device, target_coords);
167 
168  // tangent bundles copied next, because it is a convenient time to send them to device
169  Kokkos::deep_copy(tangent_bundles_device, tangent_bundles);
170 
171  // need Kokkos View storing true solution
172  Kokkos::View<double*, Kokkos::DefaultExecutionSpace> sampling_data_device("samples of true solution",
173  source_coords_device.extent(0));
174 
175  Kokkos::parallel_for("Sampling Manufactured Solutions", Kokkos::RangePolicy<Kokkos::DefaultExecutionSpace>
176  (0,source_coords.extent(0)), KOKKOS_LAMBDA(const int i) {
177 
178  // coordinates of source site i
179  double xval = source_coords_device(i,0);
180  double yval = (dimension>1) ? source_coords_device(i,1) : 0;
181  double zval = (dimension>2) ? source_coords_device(i,2) : 0;
182 
183  // data for targets with scalar input
184  sampling_data_device(i) = trueSolution(xval, yval, zval, order, dimension);
185  });
186 
187  //! [Creating The Data]
188 
189  Kokkos::Profiling::popRegion();
190  Kokkos::Profiling::pushRegion("Neighbor Search");
191 
192  //! [Performing Neighbor Search]
193 
194 
195  // Point cloud construction for neighbor search
196  // CreatePointCloudSearch constructs an object of type PointCloudSearch, but deduces the templates for you
197  auto point_cloud_search(CreatePointCloudSearch(source_coords, dimension));
198 
199  // each row is a neighbor list for a target site, with the first column of each row containing
200  // the number of neighbors for that rows corresponding target site
201  double epsilon_multiplier = 1.8;
202  int estimated_upper_bound_number_neighbors =
203  point_cloud_search.getEstimatedNumberNeighborsUpperBound(min_neighbors, dimension, epsilon_multiplier);
204 
205  Kokkos::View<int**, Kokkos::DefaultExecutionSpace> neighbor_lists_device("neighbor lists",
206  number_target_coords, estimated_upper_bound_number_neighbors); // first column is # of neighbors
207  Kokkos::View<int**>::HostMirror neighbor_lists = Kokkos::create_mirror_view(neighbor_lists_device);
208 
209  // each target site has a window size
210  Kokkos::View<double*, Kokkos::DefaultExecutionSpace> epsilon_device("h supports", number_target_coords);
211  Kokkos::View<double*>::HostMirror epsilon = Kokkos::create_mirror_view(epsilon_device);
212 
213  // query the point cloud to generate the neighbor lists using a kdtree to produce the n nearest neighbor
214  // to each target site, adding (epsilon_multiplier-1)*100% to whatever the distance away the further neighbor used is from
215  // each target to the view for epsilon
216  point_cloud_search.generate2DNeighborListsFromKNNSearch(false /*not dry run*/, target_coords, neighbor_lists,
217  epsilon, min_neighbors, epsilon_multiplier);
218 
219  //! [Performing Neighbor Search]
220 
221  Kokkos::Profiling::popRegion();
222  Kokkos::fence(); // let call to build neighbor lists complete before copying back to device
223  timer.reset();
224 
225  //! [Setting Up The GMLS Object]
226 
227 
228  // Copy data back to device (they were filled on the host)
229  // We could have filled Kokkos Views with memory space on the host
230  // and used these instead, and then the copying of data to the device
231  // would be performed in the GMLS class
232  Kokkos::deep_copy(neighbor_lists_device, neighbor_lists);
233  Kokkos::deep_copy(epsilon_device, epsilon);
234 
235  // initialize an instance of the GMLS class
237  PointSample,
238  order, dimension,
239  solver_name.c_str(), problem_name.c_str(), constraint_name.c_str(),
240  0 /*manifold order*/);
241 
242  // pass in neighbor lists, source coordinates, target coordinates, and window sizes
243  //
244  // neighbor lists have the format:
245  // dimensions: (# number of target sites) X (# maximum number of neighbors for any given target + 1)
246  // the first column contains the number of neighbors for that rows corresponding target index
247  //
248  // source coordinates have the format:
249  // dimensions: (# number of source sites) X (dimension)
250  // entries in the neighbor lists (integers) correspond to rows of this 2D array
251  //
252  // target coordinates have the format:
253  // dimensions: (# number of target sites) X (dimension)
254  // # of target sites is same as # of rows of neighbor lists
255  //
256  my_GMLS.setProblemData(neighbor_lists_device, source_coords_device, target_coords_device, epsilon_device);
257  my_GMLS.setTangentBundle(tangent_bundles_device);
258 
259  // create a vector of target operations
260  TargetOperation lro;
262 
263  // and then pass them to the GMLS class
264  my_GMLS.addTargets(lro);
265 
266  // sets the weighting kernel function from WeightingFunctionType
267  my_GMLS.setWeightingType(WeightingFunctionType::Power);
268 
269  // power to use in that weighting kernel function
270  my_GMLS.setWeightingParameter(2);
271 
272  // generate the alphas that to be combined with data for each target operation requested in lro
273  my_GMLS.generateAlphas(number_of_batches);
274 
275  //! [Setting Up The GMLS Object]
276 
277  double instantiation_time = timer.seconds();
278  std::cout << "Took " << instantiation_time << "s to complete normal vectors generation." << std::endl;
279  Kokkos::fence(); // let generateNormalVectors finish up before using alphas
280  Kokkos::Profiling::pushRegion("Apply Alphas to Data");
281 
282  //! [Apply GMLS Alphas To Data]
283 
284  // it is important to note that if you expect to use the data as a 1D view, then you should use double*
285  // however, if you know that the target operation will result in a 2D view (vector or matrix output),
286  // then you should template with double** as this is something that can not be infered from the input data
287  // or the target operator at compile time. Additionally, a template argument is required indicating either
288  // Kokkos::HostSpace or Kokkos::DefaultExecutionSpace::memory_space()
289 
290  // The Evaluator class takes care of handling input data views as well as the output data views.
291  // It uses information from the GMLS class to determine how many components are in the input
292  // as well as output for any choice of target functionals and then performs the contactions
293  // on the data using the alpha coefficients generated by the GMLS class, all on the device.
294  Evaluator gmls_evaluator(&my_GMLS);
295 
296  auto output_value = gmls_evaluator.applyAlphasToDataAllComponentsAllTargetSites<double*, Kokkos::HostSpace>
297  (sampling_data_device, LaplacianOfScalarPointEvaluation);
298 
299  Kokkos::fence(); // let application of alphas to data finish before using results
300  Kokkos::Profiling::popRegion();
301  // times the Comparison in Kokkos
302  Kokkos::Profiling::pushRegion("Comparison");
303 
304  //! [Check That Solutions Are Correct]
305 
306  // loop through the target sites
307  for (int i=0; i<number_target_coords; i++) {
308  // target site i's coordinate
309  double xval = target_coords(i,0);
310  double yval = (dimension>1) ? target_coords(i,1) : 0;
311  double zval = (dimension>2) ? target_coords(i,2) : 0;
312 
313  // 0th entry is # of neighbors, which is the index beyond the last neighbor
314  int num_neigh_i = neighbor_lists(i, 0);
315  double b_i = my_GMLS.getSolutionSetHost()->getAlpha0TensorTo0Tensor(lro, i, num_neigh_i);
316 
317  // load value from output
318  double GMLS_value = output_value(i);
319 
320  // obtain the real Laplacian
321  double actual_Laplacian = trueLaplacian(xval, yval, zval, order, dimension);
322 
323  // calculate value of g to reconstruct the computed Laplacian
324  double actual_Gradient[3] = {0,0,0}; // initialized for 3, but only filled up to dimension
325  trueGradient(actual_Gradient, xval, yval, zval, order, dimension);
326  double g = (dimension == 3) ? (1.0/sqrt(3.0))*(actual_Gradient[0] + actual_Gradient[1] + actual_Gradient[2])
327  : (1.0/sqrt(2.0))*(actual_Gradient[0] + actual_Gradient[1]);
328  double adjusted_value = GMLS_value + b_i*g;
329 
330  // check actual function value
331  if(GMLS_value!=GMLS_value || std::abs(actual_Laplacian - adjusted_value) > failure_tolerance) {
332  all_passed = false;
333  std::cout << i << " Failed Actual by: " << std::abs(actual_Laplacian - adjusted_value) << std::endl;
334  }
335  }
336 
337  //! [Check That Solutions Are Correct]
338  // popRegion hidden from tutorial
339  // stop timing comparison loop
340  Kokkos::Profiling::popRegion();
341  //! [Finalize Program]
342 } // end of code block to reduce scope, causing Kokkos View de-allocations
343 // otherwise, Views may be deallocating when we call Kokkos finalize() later
344 
345 // finalize Kokkos and MPI (if available)
346 Kokkos::finalize();
347 #ifdef COMPADRE_USE_MPI
348 MPI_Finalize();
349 #endif
350 
351 // output to user that test passed or failed
352 if(all_passed) {
353  fprintf(stdout, "Passed test \n");
354  return 0;
355 } else {
356  fprintf(stdout, "Failed test \n");
357  return -1;
358 }
359 
360 } // main
361 
362 
363 //! [Finalize Program]
Lightweight Evaluator Helper This class is a lightweight wrapper for extracting and applying all rele...
Point evaluation of the laplacian of a scalar (could be on a manifold or not)
PointCloudSearch< view_type > CreatePointCloudSearch(view_type src_view, const local_index_type dimensions=-1, const local_index_type max_leaf=-1)
CreatePointCloudSearch allows for the construction of an object of type PointCloudSearch with templat...
int main(int argc, char **argv)
TargetOperation
Available target functionals.
static KOKKOS_INLINE_FUNCTION int getNP(const int m, const int dimension=3, const ReconstructionSpace r_space=ReconstructionSpace::ScalarTaylorPolynomial)
Returns size of the basis for a given polynomial order and dimension General to dimension 1...
KOKKOS_INLINE_FUNCTION void trueGradient(double *ans, double x, double y, double z, int order, int dimension)
Scalar polynomial basis centered at the target site and scaled by sum of basis powers e...
KOKKOS_INLINE_FUNCTION double trueLaplacian(double x, double y, double z, int order, int dimension)
Generalized Moving Least Squares (GMLS)
Kokkos::View< output_data_type, output_array_layout, output_memory_space > applyAlphasToDataAllComponentsAllTargetSites(view_type_input_data sampling_data, TargetOperation lro, const SamplingFunctional sro_in=PointSample, bool scalar_as_vector_if_needed=true, const int evaluation_site_local_index=0) const
Transformation of data under GMLS (allocates memory for output)
void setProblemData(view_type_1 neighbor_lists, view_type_2 source_coordinates, view_type_3 target_coordinates, view_type_4 epsilons)
Sets basic problem data (neighbor lists, source coordinates, and target coordinates) ...
constexpr SamplingFunctional PointSample
Available sampling functionals.
KOKKOS_INLINE_FUNCTION double trueSolution(double x, double y, double z, int order, int dimension)