1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
|
// @HEADER
// ************************************************************************
//
// Intrepid2 Package
// Copyright (2007) Sandia Corporation
//
// Under terms of Contract DE-AC04-94AL85000, there is a non-exclusive
// license for use of this work by or on behalf of the U.S. Government.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact Kyungjoo Kim (kyukim@sandia.gov),
// Mauro Perego (mperego@sandia.gov), or
// Nate Roberts (nvrober@sandia.gov)
//
// ************************************************************************
// @HEADER
/** \file TensorViewFunctorTests.cpp
\brief Tests to verify the TensorViewFunctor used in TensorBasis.
\author Created by N.V. Roberts.
*/
#include "Teuchos_UnitTestHarness.hpp"
#include "Intrepid2_TensorBasis.hpp"
#include "Intrepid2_Types.hpp"
#include "Intrepid2_TestUtils.hpp"
#include "Kokkos_Core.hpp"
namespace
{
using namespace Intrepid2;
template<typename ScalarViewType>
void runTensorViewFunctorTest(ScalarViewType tensor_expected, ScalarViewType view1, ScalarViewType view2, double weight, bool tensorPoints,
Teuchos::FancyOStream &out, bool &success)
{
double tol = 1e-15;
using namespace Intrepid2;
using ExecutionSpace = typename ScalarViewType::execution_space;
using Scalar = typename ScalarViewType::value_type;
const bool hasADType = false;
const int vectorSize = hasADType ? FAD_VECTOR_SIZE : VECTOR_SIZE;
auto policy = Kokkos::TeamPolicy<ExecutionSpace>(view1.extent_int(0),Kokkos::AUTO(),vectorSize);
using FunctorType = TensorViewFunctor<ExecutionSpace, Scalar, ScalarViewType>;
ScalarViewType tensor_actual;
// TODO: figure out a better way to initialize tensor_actual to be the same size/shape as tensor_expected
if (tensor_expected.rank() == 1)
{
tensor_actual = ScalarViewType("tensor actual", tensor_expected.extent_int(0));
}
else if (tensor_expected.rank() == 2)
{
tensor_actual = ScalarViewType("tensor actual", tensor_expected.extent_int(0),tensor_expected.extent_int(1));
}
else if (tensor_expected.rank() == 3)
{
tensor_actual = ScalarViewType("tensor actual", tensor_expected.extent_int(0),tensor_expected.extent_int(1), tensor_expected.extent_int(2));
}
else if (tensor_expected.rank() == 4)
{
tensor_actual = ScalarViewType("tensor actual", tensor_expected.extent_int(0),tensor_expected.extent_int(1), tensor_expected.extent_int(2), tensor_expected.extent_int(2));
}
else
{
TEUCHOS_TEST_FOR_EXCEPTION(true, std::invalid_argument, "Test does not yet support this output rank");
}
FunctorType functor(tensor_actual, view1, view2, tensorPoints, weight);
Kokkos::parallel_for( policy , functor, "TensorViewFunctor");
switch (tensor_expected.rank())
{
case 1: testFloatingEquality1(tensor_actual,tensor_expected,tol,tol,out,success); break;
case 2: testFloatingEquality2(tensor_actual,tensor_expected,tol,tol,out,success); break;
case 3: testFloatingEquality3(tensor_actual,tensor_expected,tol,tol,out,success); break;
case 4: testFloatingEquality4(tensor_actual,tensor_expected,tol,tol,out,success); break;
default:
TEUCHOS_TEST_FOR_EXCEPTION(true, std::invalid_argument, "Test does not yet support this output rank");
}
}
template<typename Scalar>
void runTensorViewFunctorTests(Teuchos::FancyOStream &out, bool &success)
{
using DeviceType = DefaultTestDeviceType;
using ScalarViewType = ViewType<Scalar,DeviceType>;
// TEST 1: simple contraction
// we'll use trivial fields so as to factor out problems in the tensor product logic
out << "TEST 1: simple contraction.\n";
int num_fields1 = 1;
int num_fields2 = 1;
int num_fields = num_fields1 * num_fields2;
int num_points = 1;
int space_dim = 3;
ScalarViewType tensor_expected("expected_tensor",num_fields,num_points);
ScalarViewType view1("view1",num_fields,num_points,space_dim);
ScalarViewType view2("view2",num_fields,num_points,space_dim);
auto tensor_expected_host = getHostCopy(tensor_expected);
auto view1_host = getHostCopy(view1);
auto view2_host = getHostCopy(view2);
view1_host(0,0,0) = 3.0;
view1_host(0,0,1) = 2.0;
view1_host(0,0,2) = 0.0;
view2_host(0,0,0) = 1.0;
view2_host(0,0,1) = 0.5;
view2_host(0,0,2) = 3.0;
tensor_expected_host(0,0,0) = 0;
for (int d=0; d<space_dim; d++)
{
tensor_expected_host(0,0,0) += view1_host(0,0,d) * view2_host(0,0,d);
}
Kokkos::deep_copy(view1, view1_host);
Kokkos::deep_copy(view2, view2_host);
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
double weight = 1.0;
bool tensor_points = false; // does not matter for the single-point case
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
// TEST 2: tensor product ordering
out << "TEST 2: tensor product ordering.\n";
num_fields1 = 2;
num_fields2 = 2;
num_fields = num_fields1 * num_fields2;
num_points = 1;
Kokkos::resize(tensor_expected,num_fields,num_points);
Kokkos::resize(view1, num_fields1,num_points);
Kokkos::resize(view2, num_fields2,num_points);
Kokkos::resize(tensor_expected_host,num_fields,num_points);
Kokkos::resize(view1_host, num_fields1,num_points);
Kokkos::resize(view2_host, num_fields2,num_points);
view1_host(0,0) = 3.0;
view1_host(1,0) = 2.0;
view2_host(0,0) = 1.0;
view2_host(1,0) = 0.5;
// view1 is the fastest-moving: tensor entry 1 corresponds to view1(1,0) and view2(0,0)
tensor_expected_host(0,0) = view1_host(0,0) * view2_host(0,0);
tensor_expected_host(1,0) = view1_host(1,0) * view2_host(0,0);
tensor_expected_host(2,0) = view1_host(0,0) * view2_host(1,0);
tensor_expected_host(3,0) = view1_host(1,0) * view2_host(1,0);
Kokkos::deep_copy(view1, view1_host);
Kokkos::deep_copy(view2, view2_host);
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
weight = 1.0;
tensor_points = false; // does not matter for the single-point case
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
// TEST 3: like TEST 2, but include non-trivial weight
out << "TEST 3: like TEST 2, but include non-trivial weight.\n";
weight = 2.0;
for (int i=0; i<num_fields; i++)
{
tensor_expected_host(i,0) *= weight;
}
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
// TEST 4: scalar times vector
out << "TEST 4: scalar times vector.\n";
num_fields1 = 2;
num_fields2 = 2;
num_fields = num_fields1 * num_fields2;
num_points = 1;
Kokkos::resize(tensor_expected,num_fields, num_points, space_dim);
Kokkos::resize(view1, num_fields1, num_points);
Kokkos::resize(view2, num_fields2, num_points, space_dim);
Kokkos::resize(tensor_expected_host, num_fields, num_points,space_dim);
Kokkos::resize(view1_host, num_fields1, num_points);
Kokkos::resize(view2_host, num_fields2, num_points,space_dim);
view1_host(0,0) = 3.0;
view1_host(1,0) = 2.0;
view2_host(0,0,0) = 1.0;
view2_host(1,0,0) = 0.5;
view2_host(0,0,1) = 2.0;
view2_host(1,0,1) = 1.0;
view2_host(0,0,2) = 4.0;
view2_host(1,0,2) = 2.0;
for (int d=0; d<space_dim; d++)
{
// view1 is the fastest-moving: tensor entry 1 corresponds to view1(1,…) and view2(0,…)
tensor_expected_host(0,0,d) = view1_host(0,0) * view2_host(0,0,d);
tensor_expected_host(1,0,d) = view1_host(1,0) * view2_host(0,0,d);
tensor_expected_host(2,0,d) = view1_host(0,0) * view2_host(1,0,d);
tensor_expected_host(3,0,d) = view1_host(1,0) * view2_host(1,0,d);
}
Kokkos::deep_copy(view1, view1_host);
Kokkos::deep_copy(view2, view2_host);
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
weight = 1.0;
tensor_points = false; // does not matter for the single-point case
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
// TEST 5: scalar times vector, nontrivial points, but still matching in point dimension
out << "TEST 5: scalar times vector, nontrivial points, but still matching in point dimension.\n";
num_fields1 = 2;
num_fields2 = 2;
num_fields = num_fields1 * num_fields2;
num_points = 2;
Kokkos::resize(tensor_expected, num_fields, num_points, space_dim);
Kokkos::resize(view1, num_fields1, num_points);
Kokkos::resize(view2, num_fields2, num_points, space_dim);
Kokkos::resize(tensor_expected_host, num_fields, num_points, space_dim);
Kokkos::resize(view1_host, num_fields1, num_points);
Kokkos::resize(view2_host, num_fields2, num_points, space_dim);
view1_host(0,0) = 3.0;
view1_host(1,0) = 2.0;
view2_host(0,0,0) = 1.0;
view2_host(1,0,0) = 0.5;
view2_host(0,0,1) = 2.0;
view2_host(1,0,1) = 1.0;
view2_host(0,0,2) = 4.0;
view2_host(1,0,2) = 2.0;
view1_host(0,1) = 1.0;
view1_host(1,1) = 1.0;
view2_host(0,1,0) = 1.0;
view2_host(1,1,0) = 2.0;
view2_host(0,1,1) = 3.0;
view2_host(1,1,1) = 4.0;
view2_host(0,1,2) = 5.0;
view2_host(1,1,2) = 6.0;
for (int point_ordinal=0; point_ordinal<num_points; point_ordinal++)
{
for (int d=0; d<space_dim; d++)
{
// view1 is the fastest-moving: tensor entry 1 corresponds to view1(1,…) and view2(0,…)
tensor_expected_host(0,point_ordinal,d) = view1_host(0,point_ordinal) * view2_host(0,point_ordinal,d);
tensor_expected_host(1,point_ordinal,d) = view1_host(1,point_ordinal) * view2_host(0,point_ordinal,d);
tensor_expected_host(2,point_ordinal,d) = view1_host(0,point_ordinal) * view2_host(1,point_ordinal,d);
tensor_expected_host(3,point_ordinal,d) = view1_host(1,point_ordinal) * view2_host(1,point_ordinal,d);
}
}
Kokkos::deep_copy(view1, view1_host);
Kokkos::deep_copy(view2, view2_host);
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
weight = 1.0;
tensor_points = false; // does not matter for the single-point case
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
// TEST 6: like TEST 2 above, but with different field counts
out << "TEST 6: like TEST 2 above, but with different field counts.\n";
num_fields1 = 2;
num_fields2 = 3;
num_fields = num_fields1 * num_fields2;
num_points = 1;
Kokkos::resize(tensor_expected,num_fields,num_points);
Kokkos::resize(view1, num_fields1,num_points);
Kokkos::resize(view2, num_fields2,num_points);
Kokkos::resize(tensor_expected_host, num_fields, num_points);
Kokkos::resize(view1_host, num_fields1, num_points);
Kokkos::resize(view2_host, num_fields2, num_points);
view1_host(0,0) = 3.0;
view1_host(1,0) = 2.0;
view2_host(0,0) = 1.0;
view2_host(1,0) = 0.5;
view2_host(2,0) = 1.5;
// view1 is the fastest-moving: tensor entry 1 corresponds to view1(1,0) and view2(0,0)
tensor_expected_host(0,0) = view1_host(0,0) * view2_host(0,0);
tensor_expected_host(1,0) = view1_host(1,0) * view2_host(0,0);
tensor_expected_host(2,0) = view1_host(0,0) * view2_host(1,0);
tensor_expected_host(3,0) = view1_host(1,0) * view2_host(1,0);
tensor_expected_host(4,0) = view1_host(0,0) * view2_host(2,0);
tensor_expected_host(5,0) = view1_host(1,0) * view2_host(2,0);
Kokkos::deep_copy(view1, view1_host);
Kokkos::deep_copy(view2, view2_host);
Kokkos::deep_copy(tensor_expected, tensor_expected_host);
weight = 1.0;
tensor_points = false; // does not matter for the single-point case
runTensorViewFunctorTest(tensor_expected, view1, view2, weight, tensor_points, out, success);
}
TEUCHOS_UNIT_TEST( TensorViewFunctor, MultipleTests )
{
using Scalar = double;
runTensorViewFunctorTests<Scalar>(out, success);
}
} // namespace
|
