1 // Ceres Solver - A fast non-linear least squares minimizer
2 // Copyright 2014 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: richie.stebbing@gmail.com (Richard Stebbing)
30 
31 #include "ceres/dynamic_compressed_row_sparse_matrix.h"
32 
33 #include "ceres/casts.h"
34 #include "ceres/compressed_row_sparse_matrix.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/triplet_sparse_matrix.h"
40 #include "gtest/gtest.h"
41 
42 namespace ceres {
43 namespace internal {
44 
45 class DynamicCompressedRowSparseMatrixTest : public ::testing::Test {
46  protected:
SetUp()47   virtual void SetUp() {
48     num_rows = 7;
49     num_cols = 4;
50 
51     // The number of additional elements reserved when `Finalize` is called
52     // should have no effect on the number of rows, columns or nonzeros.
53     // Set this to some nonzero value to be sure.
54     num_additional_elements = 13;
55 
56     expected_num_nonzeros = num_rows * num_cols - min(num_rows, num_cols);
57 
58     InitialiseDenseReference();
59     InitialiseSparseMatrixReferences();
60 
61     dcrsm.reset(new DynamicCompressedRowSparseMatrix(num_rows,
62                                                      num_cols,
63                                                      0));
64   }
65 
Finalize()66   void Finalize() {
67     dcrsm->Finalize(num_additional_elements);
68   }
69 
InitialiseDenseReference()70   void InitialiseDenseReference() {
71     dense.resize(num_rows, num_cols);
72     dense.setZero();
73     int num_nonzeros = 0;
74     for (int i = 0; i < (num_rows * num_cols); ++i) {
75       const int r = i / num_cols, c = i % num_cols;
76       if (r != c) {
77         dense(r, c) = i + 1;
78         ++num_nonzeros;
79       }
80     }
81     ASSERT_EQ(num_nonzeros, expected_num_nonzeros);
82   }
83 
InitialiseSparseMatrixReferences()84   void InitialiseSparseMatrixReferences() {
85     std::vector<int> rows, cols;
86     std::vector<double> values;
87     for (int i = 0; i < (num_rows * num_cols); ++i) {
88       const int r = i / num_cols, c = i % num_cols;
89       if (r != c) {
90         rows.push_back(r);
91         cols.push_back(c);
92         values.push_back(i + 1);
93       }
94     }
95     ASSERT_EQ(values.size(), expected_num_nonzeros);
96 
97     tsm.reset(new TripletSparseMatrix(num_rows,
98                                       num_cols,
99                                       expected_num_nonzeros));
100     std::copy(rows.begin(), rows.end(), tsm->mutable_rows());
101     std::copy(cols.begin(), cols.end(), tsm->mutable_cols());
102     std::copy(values.begin(), values.end(), tsm->mutable_values());
103     tsm->set_num_nonzeros(values.size());
104 
105     Matrix dense_from_tsm;
106     tsm->ToDenseMatrix(&dense_from_tsm);
107     ASSERT_TRUE((dense.array() == dense_from_tsm.array()).all());
108 
109     crsm.reset(new CompressedRowSparseMatrix(*tsm));
110     Matrix dense_from_crsm;
111     crsm->ToDenseMatrix(&dense_from_crsm);
112     ASSERT_TRUE((dense.array() == dense_from_crsm.array()).all());
113   }
114 
InsertNonZeroEntriesFromDenseReference()115   void InsertNonZeroEntriesFromDenseReference() {
116     for (int r = 0; r < num_rows; ++r) {
117       for (int c = 0; c < num_cols; ++c) {
118         const double& v = dense(r, c);
119         if (v != 0.0) {
120           dcrsm->InsertEntry(r, c, v);
121         }
122       }
123     }
124   }
125 
ExpectEmpty()126   void ExpectEmpty() {
127     EXPECT_EQ(dcrsm->num_rows(), num_rows);
128     EXPECT_EQ(dcrsm->num_cols(), num_cols);
129     EXPECT_EQ(dcrsm->num_nonzeros(), 0);
130 
131     Matrix dense_from_dcrsm;
132     dcrsm->ToDenseMatrix(&dense_from_dcrsm);
133     EXPECT_EQ(dense_from_dcrsm.rows(), num_rows);
134     EXPECT_EQ(dense_from_dcrsm.cols(), num_cols);
135     EXPECT_TRUE((dense_from_dcrsm.array() == 0.0).all());
136   }
137 
ExpectEqualToDenseReference()138   void ExpectEqualToDenseReference() {
139     Matrix dense_from_dcrsm;
140     dcrsm->ToDenseMatrix(&dense_from_dcrsm);
141     EXPECT_TRUE((dense.array() == dense_from_dcrsm.array()).all());
142   }
143 
ExpectEqualToCompressedRowSparseMatrixReference()144   void ExpectEqualToCompressedRowSparseMatrixReference() {
145     typedef Eigen::Map<const Eigen::VectorXi> ConstIntVectorRef;
146 
147     ConstIntVectorRef crsm_rows(crsm->rows(), crsm->num_rows() + 1);
148     ConstIntVectorRef dcrsm_rows(dcrsm->rows(), dcrsm->num_rows() + 1);
149     EXPECT_TRUE((crsm_rows.array() == dcrsm_rows.array()).all());
150 
151     ConstIntVectorRef crsm_cols(crsm->cols(), crsm->num_nonzeros());
152     ConstIntVectorRef dcrsm_cols(dcrsm->cols(), dcrsm->num_nonzeros());
153     EXPECT_TRUE((crsm_cols.array() == dcrsm_cols.array()).all());
154 
155     ConstVectorRef crsm_values(crsm->values(), crsm->num_nonzeros());
156     ConstVectorRef dcrsm_values(dcrsm->values(), dcrsm->num_nonzeros());
157     EXPECT_TRUE((crsm_values.array() == dcrsm_values.array()).all());
158   }
159 
160   int num_rows;
161   int num_cols;
162 
163   int num_additional_elements;
164 
165   int expected_num_nonzeros;
166 
167   Matrix dense;
168   scoped_ptr<TripletSparseMatrix> tsm;
169   scoped_ptr<CompressedRowSparseMatrix> crsm;
170 
171   scoped_ptr<DynamicCompressedRowSparseMatrix> dcrsm;
172 };
173 
TEST_F(DynamicCompressedRowSparseMatrixTest,Initialization)174 TEST_F(DynamicCompressedRowSparseMatrixTest, Initialization) {
175   ExpectEmpty();
176 
177   Finalize();
178   ExpectEmpty();
179 }
180 
TEST_F(DynamicCompressedRowSparseMatrixTest,InsertEntryAndFinalize)181 TEST_F(DynamicCompressedRowSparseMatrixTest, InsertEntryAndFinalize) {
182   InsertNonZeroEntriesFromDenseReference();
183   ExpectEmpty();
184 
185   Finalize();
186   ExpectEqualToDenseReference();
187   ExpectEqualToCompressedRowSparseMatrixReference();
188 }
189 
TEST_F(DynamicCompressedRowSparseMatrixTest,ClearRows)190 TEST_F(DynamicCompressedRowSparseMatrixTest, ClearRows) {
191   InsertNonZeroEntriesFromDenseReference();
192   Finalize();
193   ExpectEqualToDenseReference();
194   ExpectEqualToCompressedRowSparseMatrixReference();
195 
196   dcrsm->ClearRows(0, 0);
197   Finalize();
198   ExpectEqualToDenseReference();
199   ExpectEqualToCompressedRowSparseMatrixReference();
200 
201   dcrsm->ClearRows(0, num_rows);
202   ExpectEqualToCompressedRowSparseMatrixReference();
203 
204   Finalize();
205   ExpectEmpty();
206 
207   InsertNonZeroEntriesFromDenseReference();
208   dcrsm->ClearRows(1, 2);
209   Finalize();
210   dense.block(1, 0, 2, num_cols).setZero();
211   ExpectEqualToDenseReference();
212 
213   InitialiseDenseReference();
214 }
215 
216 }  // namespace internal
217 }  // namespace ceres
218