1 | //===----------------------------------------------------------------------===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | // |
9 | // REQUIRES: long_tests |
10 | |
11 | // <random> |
12 | |
13 | // template<class RealType = double> |
14 | // class weibull_distribution |
15 | |
16 | // template<class _URNG> result_type operator()(_URNG& g); |
17 | |
18 | #include <random> |
19 | #include <cassert> |
20 | #include <vector> |
21 | #include <numeric> |
22 | #include <cstddef> |
23 | |
24 | #include "test_macros.h" |
25 | |
26 | template <class T> |
27 | inline |
28 | T |
29 | sqr(T x) |
30 | { |
31 | return x * x; |
32 | } |
33 | |
34 | int main(int, char**) |
35 | { |
36 | { |
37 | typedef std::weibull_distribution<> D; |
38 | typedef std::mt19937 G; |
39 | G g; |
40 | D d(0.5, 2); |
41 | const int N = 1000000; |
42 | std::vector<D::result_type> u; |
43 | for (int i = 0; i < N; ++i) |
44 | { |
45 | D::result_type v = d(g); |
46 | assert(d.min() <= v); |
47 | u.push_back(x: v); |
48 | } |
49 | double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); |
50 | double var = 0; |
51 | double skew = 0; |
52 | double kurtosis = 0; |
53 | for (std::size_t i = 0; i < u.size(); ++i) |
54 | { |
55 | double dbl = (u[i] - mean); |
56 | double d2 = sqr(dbl); |
57 | var += d2; |
58 | skew += dbl * d2; |
59 | kurtosis += d2 * d2; |
60 | } |
61 | var /= u.size(); |
62 | double dev = std::sqrt(x: var); |
63 | skew /= u.size() * dev * var; |
64 | kurtosis /= u.size() * var * var; |
65 | kurtosis -= 3; |
66 | double x_mean = d.b() * std::tgamma(1 + 1/d.a()); |
67 | double x_var = sqr(d.b()) * std::tgamma(1 + 2/d.a()) - sqr(x_mean); |
68 | double x_skew = (sqr(d.b())*d.b() * std::tgamma(1 + 3/d.a()) - |
69 | 3*x_mean*x_var - sqr(x_mean)*x_mean) / |
70 | (std::sqrt(x: x_var)*x_var); |
71 | double x_kurtosis = (sqr(sqr(d.b())) * std::tgamma(1 + 4/d.a()) - |
72 | 4*x_skew*x_var*sqrt(x: x_var)*x_mean - |
73 | 6*sqr(x_mean)*x_var - sqr(sqr(x_mean))) / sqr(x_var) - 3; |
74 | assert(std::abs((mean - x_mean) / x_mean) < 0.01); |
75 | assert(std::abs((var - x_var) / x_var) < 0.01); |
76 | assert(std::abs((skew - x_skew) / x_skew) < 0.01); |
77 | assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.03); |
78 | } |
79 | { |
80 | typedef std::weibull_distribution<> D; |
81 | typedef std::mt19937 G; |
82 | G g; |
83 | D d(1, .5); |
84 | const int N = 1000000; |
85 | std::vector<D::result_type> u; |
86 | for (int i = 0; i < N; ++i) |
87 | { |
88 | D::result_type v = d(g); |
89 | assert(d.min() <= v); |
90 | u.push_back(v); |
91 | } |
92 | double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); |
93 | double var = 0; |
94 | double skew = 0; |
95 | double kurtosis = 0; |
96 | for (std::size_t i = 0; i < u.size(); ++i) |
97 | { |
98 | double dbl = (u[i] - mean); |
99 | double d2 = sqr(dbl); |
100 | var += d2; |
101 | skew += dbl * d2; |
102 | kurtosis += d2 * d2; |
103 | } |
104 | var /= u.size(); |
105 | double dev = std::sqrt(x: var); |
106 | skew /= u.size() * dev * var; |
107 | kurtosis /= u.size() * var * var; |
108 | kurtosis -= 3; |
109 | double x_mean = d.b() * std::tgamma(1 + 1/d.a()); |
110 | double x_var = sqr(d.b()) * std::tgamma(1 + 2/d.a()) - sqr(x_mean); |
111 | double x_skew = (sqr(d.b())*d.b() * std::tgamma(1 + 3/d.a()) - |
112 | 3*x_mean*x_var - sqr(x_mean)*x_mean) / |
113 | (std::sqrt(x: x_var)*x_var); |
114 | double x_kurtosis = (sqr(sqr(d.b())) * std::tgamma(1 + 4/d.a()) - |
115 | 4*x_skew*x_var*sqrt(x: x_var)*x_mean - |
116 | 6*sqr(x_mean)*x_var - sqr(sqr(x_mean))) / sqr(x_var) - 3; |
117 | assert(std::abs((mean - x_mean) / x_mean) < 0.01); |
118 | assert(std::abs((var - x_var) / x_var) < 0.01); |
119 | assert(std::abs((skew - x_skew) / x_skew) < 0.01); |
120 | assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); |
121 | } |
122 | { |
123 | typedef std::weibull_distribution<> D; |
124 | typedef std::mt19937 G; |
125 | G g; |
126 | D d(2, 3); |
127 | const int N = 1000000; |
128 | std::vector<D::result_type> u; |
129 | for (int i = 0; i < N; ++i) |
130 | { |
131 | D::result_type v = d(g); |
132 | assert(d.min() <= v); |
133 | u.push_back(v); |
134 | } |
135 | double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); |
136 | double var = 0; |
137 | double skew = 0; |
138 | double kurtosis = 0; |
139 | for (std::size_t i = 0; i < u.size(); ++i) |
140 | { |
141 | double dbl = (u[i] - mean); |
142 | double d2 = sqr(dbl); |
143 | var += d2; |
144 | skew += dbl * d2; |
145 | kurtosis += d2 * d2; |
146 | } |
147 | var /= u.size(); |
148 | double dev = std::sqrt(x: var); |
149 | skew /= u.size() * dev * var; |
150 | kurtosis /= u.size() * var * var; |
151 | kurtosis -= 3; |
152 | double x_mean = d.b() * std::tgamma(1 + 1/d.a()); |
153 | double x_var = sqr(d.b()) * std::tgamma(1 + 2/d.a()) - sqr(x_mean); |
154 | double x_skew = (sqr(d.b())*d.b() * std::tgamma(1 + 3/d.a()) - |
155 | 3*x_mean*x_var - sqr(x_mean)*x_mean) / |
156 | (std::sqrt(x: x_var)*x_var); |
157 | double x_kurtosis = (sqr(sqr(d.b())) * std::tgamma(1 + 4/d.a()) - |
158 | 4*x_skew*x_var*sqrt(x: x_var)*x_mean - |
159 | 6*sqr(x_mean)*x_var - sqr(sqr(x_mean))) / sqr(x_var) - 3; |
160 | assert(std::abs((mean - x_mean) / x_mean) < 0.01); |
161 | assert(std::abs((var - x_var) / x_var) < 0.01); |
162 | assert(std::abs((skew - x_skew) / x_skew) < 0.01); |
163 | assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.03); |
164 | } |
165 | |
166 | return 0; |
167 | } |
168 | |