sparse_ref.cpp
Go to the documentation of this file.
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 20015 Gael Guennebaud <gael.guennebaud@inria.fr>
5 //
6 // This Source Code Form is subject to the terms of the Mozilla
7 // Public License v. 2.0. If a copy of the MPL was not distributed
8 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9 
10 // This unit test cannot be easily written to work with EIGEN_DEFAULT_TO_ROW_MAJOR
11 #ifdef EIGEN_DEFAULT_TO_ROW_MAJOR
12 #undef EIGEN_DEFAULT_TO_ROW_MAJOR
13 #endif
14 
15 static long int nb_temporaries;
16 
17 inline void on_temporary_creation() {
18  // here's a great place to set a breakpoint when debugging failures in this test!
20 }
21 
22 #define EIGEN_SPARSE_CREATE_TEMPORARY_PLUGIN { on_temporary_creation(); }
23 
24 #include "main.h"
25 #include <Eigen/SparseCore>
26 
27 #define VERIFY_EVALUATION_COUNT(XPR,N) {\
28  nb_temporaries = 0; \
29  CALL_SUBTEST( XPR ); \
30  if(nb_temporaries!=N) std::cerr << "nb_temporaries == " << nb_temporaries << "\n"; \
31  VERIFY( (#XPR) && nb_temporaries==N ); \
32  }
33 
34 template<typename PlainObjectType> void check_const_correctness(const PlainObjectType&)
35 {
36  // verify that ref-to-const don't have LvalueBit
37  typedef typename internal::add_const<PlainObjectType>::type ConstPlainObjectType;
38  VERIFY( !(internal::traits<Ref<ConstPlainObjectType> >::Flags & LvalueBit) );
39  VERIFY( !(internal::traits<Ref<ConstPlainObjectType, Aligned> >::Flags & LvalueBit) );
42 }
43 
44 template<typename B>
45 EIGEN_DONT_INLINE void call_ref_1(Ref<SparseMatrix<float> > a, const B &b) { VERIFY_IS_EQUAL(a.toDense(),b.toDense()); }
46 
47 template<typename B>
48 EIGEN_DONT_INLINE void call_ref_2(const Ref<const SparseMatrix<float> >& a, const B &b) { VERIFY_IS_EQUAL(a.toDense(),b.toDense()); }
49 
50 template<typename B>
52  VERIFY(a.isCompressed());
53  VERIFY_IS_EQUAL(a.toDense(),b.toDense());
54 }
55 
56 template<typename B>
57 EIGEN_DONT_INLINE void call_ref_4(Ref<SparseVector<float> > a, const B &b) { VERIFY_IS_EQUAL(a.toDense(),b.toDense()); }
58 
59 template<typename B>
60 EIGEN_DONT_INLINE void call_ref_5(const Ref<const SparseVector<float> >& a, const B &b) { VERIFY_IS_EQUAL(a.toDense(),b.toDense()); }
61 
62 void call_ref()
63 {
64  SparseMatrix<float> A = MatrixXf::Random(10,10).sparseView(0.5,1);
65  SparseMatrix<float,RowMajor> B = MatrixXf::Random(10,10).sparseView(0.5,1);
66  SparseMatrix<float> C = MatrixXf::Random(10,10).sparseView(0.5,1);
67  C.reserve(VectorXi::Constant(C.outerSize(), 2));
68  const SparseMatrix<float>& Ac(A);
69  Block<SparseMatrix<float> > Ab(A,0,1, 3,3);
70  const Block<SparseMatrix<float> > Abc(A,0,1,3,3);
71  SparseVector<float> vc = VectorXf::Random(10).sparseView(0.5,1);
72  SparseVector<float,RowMajor> vr = VectorXf::Random(10).sparseView(0.5,1);
73  SparseMatrix<float> AA = A*A;
74 
75 
77 // VERIFY_EVALUATION_COUNT( call_ref_1(Ac, Ac), 0); // does not compile on purpose
80  VERIFY_EVALUATION_COUNT( call_ref_2(A.transpose(), A.transpose()), 1);
81  VERIFY_EVALUATION_COUNT( call_ref_3(A.transpose(), A.transpose()), 1);
88  VERIFY_EVALUATION_COUNT( call_ref_2(B.transpose(), B.transpose()), 0);
89  VERIFY_EVALUATION_COUNT( call_ref_3(B.transpose(), B.transpose()), 0);
92 
93  VERIFY(!C.isCompressed());
95 
97  VERIFY_IS_APPROX(Ar+Ar, A+A);
100 
102  VERIFY_EVALUATION_COUNT( call_ref_1(Br.transpose(), Br.transpose()), 0);
103  VERIFY_EVALUATION_COUNT( call_ref_2(Br, Br), 1);
104  VERIFY_EVALUATION_COUNT( call_ref_2(Br.transpose(), Br.transpose()), 0);
105 
107 // VERIFY_EVALUATION_COUNT( call_ref_1(Arc, Arc), 0); // does not compile on purpose
108  VERIFY_EVALUATION_COUNT( call_ref_2(Arc, Arc), 0);
109 
110  VERIFY_EVALUATION_COUNT( call_ref_2(A.middleCols(1,3), A.middleCols(1,3)), 0);
111 
112  VERIFY_EVALUATION_COUNT( call_ref_2(A.col(2), A.col(2)), 0);
113  VERIFY_EVALUATION_COUNT( call_ref_2(vc, vc), 0);
114  VERIFY_EVALUATION_COUNT( call_ref_2(vr.transpose(), vr.transpose()), 0);
115  VERIFY_EVALUATION_COUNT( call_ref_2(vr, vr.transpose()), 0);
116 
117  VERIFY_EVALUATION_COUNT( call_ref_2(A.block(1,1,3,3), A.block(1,1,3,3)), 1); // should be 0 (allocate starts/nnz only)
118 
119  VERIFY_EVALUATION_COUNT( call_ref_4(vc, vc), 0);
120  VERIFY_EVALUATION_COUNT( call_ref_4(vr, vr.transpose()), 0);
121  VERIFY_EVALUATION_COUNT( call_ref_5(vc, vc), 0);
122  VERIFY_EVALUATION_COUNT( call_ref_5(vr, vr.transpose()), 0);
123  VERIFY_EVALUATION_COUNT( call_ref_4(A.col(2), A.col(2)), 0);
124  VERIFY_EVALUATION_COUNT( call_ref_5(A.col(2), A.col(2)), 0);
125  // VERIFY_EVALUATION_COUNT( call_ref_4(A.row(2), A.row(2).transpose()), 1); // does not compile on purpose
126  VERIFY_EVALUATION_COUNT( call_ref_5(A.row(2), A.row(2).transpose()), 1);
127 }
128 
130 {
131  for(int i = 0; i < g_repeat; i++) {
135 
138  }
139 }
Eigen::SparseMatrix
A versatible sparse matrix representation.
Definition: SparseMatrix.h:96
B
Matrix< SCALARB, Dynamic, Dynamic, opt_B > B
Definition: bench_gemm.cpp:49
Eigen::Block
Expression of a fixed-size or dynamic-size block.
Definition: Block.h:103
gtsam.examples.DogLegOptimizerExample.type
type
Definition: DogLegOptimizerExample.py:111
nb_temporaries
static long int nb_temporaries
Definition: sparse_ref.cpp:15
VERIFY_IS_EQUAL
#define VERIFY_IS_EQUAL(a, b)
Definition: main.h:386
b
Scalar * b
Definition: benchVecAdd.cpp:17
call_ref_1
EIGEN_DONT_INLINE void call_ref_1(Ref< SparseMatrix< float > > a, const B &b)
Definition: sparse_ref.cpp:45
call_ref_4
EIGEN_DONT_INLINE void call_ref_4(Ref< SparseVector< float > > a, const B &b)
Definition: sparse_ref.cpp:57
VERIFY_EVALUATION_COUNT
#define VERIFY_EVALUATION_COUNT(XPR, N)
Definition: sparse_ref.cpp:27
call_ref_5
EIGEN_DONT_INLINE void call_ref_5(const Ref< const SparseVector< float > > &a, const B &b)
Definition: sparse_ref.cpp:60
A
Matrix< SCALARA, Dynamic, Dynamic, opt_A > A
Definition: bench_gemm.cpp:48
CALL_SUBTEST_3
#define CALL_SUBTEST_3(FUNC)
Definition: split_test_helper.h:16
CALL_SUBTEST_1
#define CALL_SUBTEST_1(FUNC)
Definition: split_test_helper.h:4
Eigen::LvalueBit
const unsigned int LvalueBit
Definition: Constants.h:144
call_ref
void call_ref()
Definition: sparse_ref.cpp:62
Eigen::g_repeat
static int g_repeat
Definition: main.h:169
Eigen::StandardCompressedFormat
@ StandardCompressedFormat
Definition: SparseRef.h:16
call_ref_2
EIGEN_DONT_INLINE void call_ref_2(const Ref< const SparseMatrix< float > > &a, const B &b)
Definition: sparse_ref.cpp:48
CALL_SUBTEST_2
#define CALL_SUBTEST_2(FUNC)
Definition: split_test_helper.h:10
VERIFY_IS_APPROX
#define VERIFY_IS_APPROX(a, b)
Definition: integer_types.cpp:15
a
ArrayXXi a
Definition: Array_initializer_list_23_cxx11.cpp:1
Eigen::Ref
A matrix or vector expression mapping an existing expression.
Definition: Ref.h:281
C
Matrix< Scalar, Dynamic, Dynamic > C
Definition: bench_gemm.cpp:50
check_const_correctness
void check_const_correctness(const PlainObjectType &)
Definition: sparse_ref.cpp:34
main.h
call_ref_3
EIGEN_DONT_INLINE void call_ref_3(const Ref< const SparseMatrix< float >, StandardCompressedFormat > &a, const B &b)
Definition: sparse_ref.cpp:51
Eigen::SparseVector
a sparse vector class
Definition: SparseUtil.h:54
EIGEN_DECLARE_TEST
EIGEN_DECLARE_TEST(sparse_ref)
Definition: sparse_ref.cpp:129
exampleQR::Ab
Matrix Ab
Definition: testNoiseModel.cpp:207
i
int i
Definition: BiCGSTAB_step_by_step.cpp:9
on_temporary_creation
void on_temporary_creation()
Definition: sparse_ref.cpp:17
VERIFY
#define VERIFY(a)
Definition: main.h:380
EIGEN_DONT_INLINE
#define EIGEN_DONT_INLINE
Definition: Macros.h:940


gtsam
Author(s):
autogenerated on Sun Dec 22 2024 04:13:30