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
|
#include "../../algorithms/medianwindow.h"
#include "../../util/stopwatch.h"
#include <iostream>
#include <limits>
#include <random>
#include <vector>
#include <type_traits>
#include <boost/test/unit_test.hpp>
// The result of the test have been validated with Octave
// https://www.mathworks.com/help/matlab/ref/movmedian.html
static SampleRow CreateSamples(const std::vector<num_t>& values) {
SampleRow result{values.size()};
for (size_t i = 0; i < values.size(); ++i) {
result.SetValue(i, values[i]);
}
return result;
}
BOOST_AUTO_TEST_SUITE(meanwindow, *boost::unit_test::label("algorithms") *
boost::unit_test::tolerance(num_t(1e-6)))
// movmedian([1, 2, 3, 4, 5, 6, 7, 8, 9, 10], 3)
BOOST_AUTO_TEST_CASE(median_num_elements_even_window_size_odd) {
SampleRow samples{
CreateSamples(std::vector<num_t>{1, 2, 3, 4, 5, 6, 7, 8, 9, 10})};
algorithms::MedianWindow<num_t>::SubtractMedian(samples, 3);
BOOST_TEST(samples.Value(0) == 1 - 1.5);
BOOST_TEST(samples.Value(1) == 2 - 2);
BOOST_TEST(samples.Value(2) == 3 - 3);
BOOST_TEST(samples.Value(3) == 4 - 4);
BOOST_TEST(samples.Value(4) == 5 - 5);
BOOST_TEST(samples.Value(5) == 6 - 6);
BOOST_TEST(samples.Value(6) == 7 - 7);
BOOST_TEST(samples.Value(7) == 8 - 8);
BOOST_TEST(samples.Value(8) == 9 - 9);
BOOST_TEST(samples.Value(9) == 10 - 9.5);
}
// movmedian([1, 2, 3, 4, 5, 6, 7, 8, 9], 3)
BOOST_AUTO_TEST_CASE(median_num_elements_odd_window_size_odd) {
SampleRow samples{
CreateSamples(std::vector<num_t>{1, 2, 3, 4, 5, 6, 7, 8, 9})};
algorithms::MedianWindow<num_t>::SubtractMedian(samples, 3);
BOOST_TEST(samples.Value(0) == 1 - 1.5);
BOOST_TEST(samples.Value(1) == 2 - 2);
BOOST_TEST(samples.Value(2) == 3 - 3);
BOOST_TEST(samples.Value(3) == 4 - 4);
BOOST_TEST(samples.Value(4) == 5 - 5);
BOOST_TEST(samples.Value(5) == 6 - 6);
BOOST_TEST(samples.Value(6) == 7 - 7);
BOOST_TEST(samples.Value(7) == 8 - 8);
BOOST_TEST(samples.Value(8) == 9 - 8.5);
}
// movmedian([1, 2, 3, 4, 5, 6, 7, 8, 9, 10], 4)
BOOST_AUTO_TEST_CASE(median_num_elements_even_window_size_even) {
SampleRow samples{
CreateSamples(std::vector<num_t>{1, 2, 3, 4, 5, 6, 7, 8, 9, 10})};
algorithms::MedianWindow<num_t>::SubtractMedian(samples, 4);
BOOST_TEST(samples.Value(0) == 1 - 1.5);
BOOST_TEST(samples.Value(1) == 2 - 2);
BOOST_TEST(samples.Value(2) == 3 - 2.5);
BOOST_TEST(samples.Value(3) == 4 - 3.5);
BOOST_TEST(samples.Value(4) == 5 - 4.5);
BOOST_TEST(samples.Value(5) == 6 - 5.5);
BOOST_TEST(samples.Value(6) == 7 - 6.5);
BOOST_TEST(samples.Value(7) == 8 - 7.5);
BOOST_TEST(samples.Value(8) == 9 - 8.5);
BOOST_TEST(samples.Value(9) == 10 - 9);
}
// movmedian([1, 2, 3, 4, 5, 6, 7, 8, 9], 4)
BOOST_AUTO_TEST_CASE(median_num_elements_odd_window_size_even) {
SampleRow samples{
CreateSamples(std::vector<num_t>{1, 2, 3, 4, 5, 6, 7, 8, 9})};
algorithms::MedianWindow<num_t>::SubtractMedian(samples, 4);
BOOST_TEST(samples.Value(0) == 1 - 1.5);
BOOST_TEST(samples.Value(1) == 2 - 2);
BOOST_TEST(samples.Value(2) == 3 - 2.5);
BOOST_TEST(samples.Value(3) == 4 - 3.5);
BOOST_TEST(samples.Value(4) == 5 - 4.5);
BOOST_TEST(samples.Value(5) == 6 - 5.5);
BOOST_TEST(samples.Value(6) == 7 - 6.5);
BOOST_TEST(samples.Value(7) == 8 - 7.5);
BOOST_TEST(samples.Value(8) == 9 - 8);
}
BOOST_AUTO_TEST_CASE(median_non_finite_elements) {
SampleRow samples{CreateSamples(
std::vector<num_t>{1, 2, 3, -std::numeric_limits<num_t>::infinity(), 4, 5,
6, std::numeric_limits<num_t>::infinity(), 7, 8, 9,
std::numeric_limits<num_t>::quiet_NaN(), 10, 11, 12})};
algorithms::MedianWindow<num_t>::SubtractMedian(samples, 3);
BOOST_TEST(samples.Value(0) == 1 - 1.5);
BOOST_TEST(samples.Value(1) == 2 - 2);
BOOST_TEST(samples.Value(2) == 3 - 2.5);
BOOST_CHECK(samples.Value(3) == -std::numeric_limits<num_t>::infinity());
BOOST_TEST(samples.Value(4) == 4 - 4.5);
BOOST_TEST(samples.Value(5) == 5 - 5);
BOOST_TEST(samples.Value(6) == 6 - 5.5);
BOOST_CHECK(samples.Value(7) == std::numeric_limits<num_t>::infinity());
BOOST_TEST(samples.Value(8) == 7 - 7.5);
BOOST_TEST(samples.Value(9) == 8 - 8);
BOOST_TEST(samples.Value(10) == 9 - 8.5);
BOOST_CHECK(std::isnan(samples.Value(11)));
BOOST_TEST(samples.Value(12) == 10 - 10.5);
BOOST_TEST(samples.Value(13) == 11 - 11);
BOOST_TEST(samples.Value(14) == 12 - 11.5);
}
static std::vector<num_t> MakeData(size_t size) {
std::random_device seed;
std::mt19937 generator(seed());
std::uniform_int_distribution<std::conditional<
sizeof(num_t) == sizeof(uint32_t), uint32_t, uint64_t>::type>
distribution;
std::vector<num_t> result;
std::generate_n(std::back_inserter(result), size, [&] {
#if defined(__has_builtin)
#if __has_builtin(__builtin_bit_cast)
#define HAS_BUILTIN_BIT_CAST
#endif
#endif
#ifdef HAS_BUILTIN_BIT_CAST
// This is the same implementation as C++20's std::bit_cast.
return __builtin_bit_cast(num_t, distribution(generator));
#else
const auto value = distribution(generator);
num_t result;
memcpy(&result, &value, sizeof(result));
return result;
#endif
});
return result;
}
BOOST_AUTO_TEST_CASE(speed, *boost::unit_test::disabled()) {
// The size of samples is choosen so the tests are done in a reasonable time.
const size_t kSize = 4 * 1024 * 1024;
const SampleRow samples{CreateSamples(MakeData(kSize))};
std::cout << "SubtractMedian for " << kSize << " elements\n";
for (int i = 1; i < 16; ++i) {
// An odd sized window has the median value in the middle of the window.
const int size = (2 << i) - 1;
SampleRow s = samples;
Stopwatch stopwatch(true);
algorithms::MedianWindow<num_t>::SubtractMedian(s, size);
stopwatch.Pause();
std::cout << "\tWindow " << size << '\t' << stopwatch.ToString() << '\n';
}
}
BOOST_AUTO_TEST_SUITE_END()
|
