1 // Ceres Solver - A fast non-linear least squares minimizer
2 // Copyright 2010, 2011, 2012 Google Inc. All rights reserved.
3 // http://code.google.com/p/ceres-solver/
4 //
5 // Redistribution and use in source and binary forms, with or without
6 // modification, are permitted provided that the following conditions are met:
7 //
8 // * Redistributions of source code must retain the above copyright notice,
9 // this list of conditions and the following disclaimer.
10 // * Redistributions in binary form must reproduce the above copyright notice,
11 // this list of conditions and the following disclaimer in the documentation
12 // and/or other materials provided with the distribution.
13 // * Neither the name of Google Inc. nor the names of its contributors may be
14 // used to endorse or promote products derived from this software without
15 // specific prior written permission.
16 //
17 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
18 // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 // ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
21 // LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
22 // CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
23 // SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
24 // INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
25 // CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
26 // ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
27 // POSSIBILITY OF SUCH DAMAGE.
28 //
29 // Author: sameeragarwal@google.com (Sameer Agarwal)
30
31 #include "ceres/partitioned_matrix_view.h"
32
33 #include <vector>
34 #include "ceres/block_structure.h"
35 #include "ceres/casts.h"
36 #include "ceres/internal/eigen.h"
37 #include "ceres/internal/scoped_ptr.h"
38 #include "ceres/linear_least_squares_problems.h"
39 #include "ceres/random.h"
40 #include "ceres/sparse_matrix.h"
41 #include "glog/logging.h"
42 #include "gtest/gtest.h"
43
44 namespace ceres {
45 namespace internal {
46
47 const double kEpsilon = 1e-14;
48
49 class PartitionedMatrixViewTest : public ::testing::Test {
50 protected :
SetUp()51 virtual void SetUp() {
52 scoped_ptr<LinearLeastSquaresProblem> problem(
53 CreateLinearLeastSquaresProblemFromId(2));
54 CHECK_NOTNULL(problem.get());
55 A_.reset(problem->A.release());
56
57 num_cols_ = A_->num_cols();
58 num_rows_ = A_->num_rows();
59 num_eliminate_blocks_ = problem->num_eliminate_blocks;
60 }
61
62 int num_rows_;
63 int num_cols_;
64 int num_eliminate_blocks_;
65
66 scoped_ptr<SparseMatrix> A_;
67 };
68
TEST_F(PartitionedMatrixViewTest,DimensionsTest)69 TEST_F(PartitionedMatrixViewTest, DimensionsTest) {
70 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
71 num_eliminate_blocks_);
72 EXPECT_EQ(m.num_col_blocks_e(), num_eliminate_blocks_);
73 EXPECT_EQ(m.num_col_blocks_f(), num_cols_ - num_eliminate_blocks_);
74 EXPECT_EQ(m.num_cols_e(), num_eliminate_blocks_);
75 EXPECT_EQ(m.num_cols_f(), num_cols_ - num_eliminate_blocks_);
76 EXPECT_EQ(m.num_cols(), A_->num_cols());
77 EXPECT_EQ(m.num_rows(), A_->num_rows());
78 }
79
TEST_F(PartitionedMatrixViewTest,RightMultiplyE)80 TEST_F(PartitionedMatrixViewTest, RightMultiplyE) {
81 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
82 num_eliminate_blocks_);
83
84 srand(5);
85
86 Vector x1(m.num_cols_e());
87 Vector x2(m.num_cols());
88 x2.setZero();
89
90 for (int i = 0; i < m.num_cols_e(); ++i) {
91 x1(i) = x2(i) = RandDouble();
92 }
93
94 Vector y1 = Vector::Zero(m.num_rows());
95 m.RightMultiplyE(x1.data(), y1.data());
96
97 Vector y2 = Vector::Zero(m.num_rows());
98 A_->RightMultiply(x2.data(), y2.data());
99
100 for (int i = 0; i < m.num_rows(); ++i) {
101 EXPECT_NEAR(y1(i), y2(i), kEpsilon);
102 }
103 }
104
TEST_F(PartitionedMatrixViewTest,RightMultiplyF)105 TEST_F(PartitionedMatrixViewTest, RightMultiplyF) {
106 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
107 num_eliminate_blocks_);
108
109 srand(5);
110
111 Vector x1(m.num_cols_f());
112 Vector x2 = Vector::Zero(m.num_cols());
113
114 for (int i = 0; i < m.num_cols_f(); ++i) {
115 x1(i) = RandDouble();
116 x2(i + m.num_cols_e()) = x1(i);
117 }
118
119 Vector y1 = Vector::Zero(m.num_rows());
120 m.RightMultiplyF(x1.data(), y1.data());
121
122 Vector y2 = Vector::Zero(m.num_rows());
123 A_->RightMultiply(x2.data(), y2.data());
124
125 for (int i = 0; i < m.num_rows(); ++i) {
126 EXPECT_NEAR(y1(i), y2(i), kEpsilon);
127 }
128 }
129
TEST_F(PartitionedMatrixViewTest,LeftMultiply)130 TEST_F(PartitionedMatrixViewTest, LeftMultiply) {
131 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
132 num_eliminate_blocks_);
133
134 srand(5);
135
136 Vector x = Vector::Zero(m.num_rows());
137 for (int i = 0; i < m.num_rows(); ++i) {
138 x(i) = RandDouble();
139 }
140
141 Vector y = Vector::Zero(m.num_cols());
142 Vector y1 = Vector::Zero(m.num_cols_e());
143 Vector y2 = Vector::Zero(m.num_cols_f());
144
145 A_->LeftMultiply(x.data(), y.data());
146 m.LeftMultiplyE(x.data(), y1.data());
147 m.LeftMultiplyF(x.data(), y2.data());
148
149 for (int i = 0; i < m.num_cols(); ++i) {
150 EXPECT_NEAR(y(i),
151 (i < m.num_cols_e()) ? y1(i) : y2(i - m.num_cols_e()),
152 kEpsilon);
153 }
154 }
155
TEST_F(PartitionedMatrixViewTest,BlockDiagonalEtE)156 TEST_F(PartitionedMatrixViewTest, BlockDiagonalEtE) {
157 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
158 num_eliminate_blocks_);
159
160 scoped_ptr<BlockSparseMatrix>
161 block_diagonal_ee(m.CreateBlockDiagonalEtE());
162 const CompressedRowBlockStructure* bs = block_diagonal_ee->block_structure();
163
164 EXPECT_EQ(block_diagonal_ee->num_rows(), 2);
165 EXPECT_EQ(block_diagonal_ee->num_cols(), 2);
166 EXPECT_EQ(bs->cols.size(), 2);
167 EXPECT_EQ(bs->rows.size(), 2);
168
169 EXPECT_NEAR(block_diagonal_ee->values()[0], 10.0, kEpsilon);
170 EXPECT_NEAR(block_diagonal_ee->values()[1], 155.0, kEpsilon);
171 }
172
TEST_F(PartitionedMatrixViewTest,BlockDiagonalFtF)173 TEST_F(PartitionedMatrixViewTest, BlockDiagonalFtF) {
174 PartitionedMatrixView m(*down_cast<BlockSparseMatrix*>(A_.get()),
175 num_eliminate_blocks_);
176
177 scoped_ptr<BlockSparseMatrix>
178 block_diagonal_ff(m.CreateBlockDiagonalFtF());
179 const CompressedRowBlockStructure* bs = block_diagonal_ff->block_structure();
180
181 EXPECT_EQ(block_diagonal_ff->num_rows(), 3);
182 EXPECT_EQ(block_diagonal_ff->num_cols(), 3);
183 EXPECT_EQ(bs->cols.size(), 3);
184 EXPECT_EQ(bs->rows.size(), 3);
185 EXPECT_NEAR(block_diagonal_ff->values()[0], 70.0, kEpsilon);
186 EXPECT_NEAR(block_diagonal_ff->values()[1], 17.0, kEpsilon);
187 EXPECT_NEAR(block_diagonal_ff->values()[2], 37.0, kEpsilon);
188 }
189
190 } // namespace internal
191 } // namespace ceres
192