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 // <random>
10 
11 // class bernoulli_distribution
12 
13 // template<class _URNG> result_type operator()(_URNG& g);
14 
15 #include <random>
16 #include <numeric>
17 #include <vector>
18 #include <cassert>
19 #include <cstddef>
20 
21 #include "test_macros.h"
22 
23 template <class T>
24 inline
25 T
sqr(T x)26 sqr(T x)
27 {
28     return x * x;
29 }
30 
main(int,char **)31 int main(int, char**)
32 {
33     {
34         typedef std::bernoulli_distribution D;
35         typedef std::minstd_rand G;
36         G g;
37         D d(.75);
38         const int N = 100000;
39         std::vector<D::result_type> u;
40         for (int i = 0; i < N; ++i)
41             u.push_back(d(g));
42         double mean = std::accumulate(u.begin(), u.end(),
43                                               double(0)) / u.size();
44         double var = 0;
45         double skew = 0;
46         double kurtosis = 0;
47         for (std::size_t i = 0; i < u.size(); ++i)
48         {
49             double dbl = (u[i] - mean);
50             double d2 = sqr(dbl);
51             var += d2;
52             skew += dbl * d2;
53             kurtosis += d2 * d2;
54         }
55         var /= u.size();
56         double dev = std::sqrt(var);
57         skew /= u.size() * dev * var;
58         kurtosis /= u.size() * var * var;
59         kurtosis -= 3;
60         double x_mean = d.p();
61         double x_var = d.p()*(1-d.p());
62         double x_skew = (1 - 2 * d.p())/std::sqrt(x_var);
63         double x_kurtosis = (6 * sqr(d.p()) - 6 * d.p() + 1)/x_var;
64         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
65         assert(std::abs((var - x_var) / x_var) < 0.01);
66         assert(std::abs((skew - x_skew) / x_skew) < 0.01);
67         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.02);
68     }
69     {
70         typedef std::bernoulli_distribution D;
71         typedef std::minstd_rand G;
72         G g;
73         D d(.25);
74         const int N = 100000;
75         std::vector<D::result_type> u;
76         for (int i = 0; i < N; ++i)
77             u.push_back(d(g));
78         double mean = std::accumulate(u.begin(), u.end(),
79                                               double(0)) / u.size();
80         double var = 0;
81         double skew = 0;
82         double kurtosis = 0;
83         for (std::size_t i = 0; i < u.size(); ++i)
84         {
85             double dbl = (u[i] - mean);
86             double d2 = sqr(dbl);
87             var += d2;
88             skew += dbl * d2;
89             kurtosis += d2 * d2;
90         }
91         var /= u.size();
92         double dev = std::sqrt(var);
93         skew /= u.size() * dev * var;
94         kurtosis /= u.size() * var * var;
95         kurtosis -= 3;
96         double x_mean = d.p();
97         double x_var = d.p()*(1-d.p());
98         double x_skew = (1 - 2 * d.p())/std::sqrt(x_var);
99         double x_kurtosis = (6 * sqr(d.p()) - 6 * d.p() + 1)/x_var;
100         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
101         assert(std::abs((var - x_var) / x_var) < 0.01);
102         assert(std::abs((skew - x_skew) / x_skew) < 0.01);
103         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.02);
104     }
105 
106   return 0;
107 }
108