1 //===----------------------------------------------------------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is dual licensed under the MIT and the University of Illinois Open 6 // Source Licenses. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 10 // <random> 11 12 // class bernoulli_distribution 13 14 // template<class _URNG> result_type operator()(_URNG& g); 15 16 #include <random> 17 #include <numeric> 18 #include <vector> 19 #include <cassert> 20 #include <cstddef> 21 22 template <class T> 23 inline 24 T sqr(T x)25sqr(T x) 26 { 27 return x * x; 28 } 29 main()30int main() 31 { 32 { 33 typedef std::bernoulli_distribution D; 34 typedef std::minstd_rand G; 35 G g; 36 D d(.75); 37 const int N = 100000; 38 std::vector<D::result_type> u; 39 for (int i = 0; i < N; ++i) 40 u.push_back(d(g)); 41 double mean = std::accumulate(u.begin(), u.end(), 42 double(0)) / u.size(); 43 double var = 0; 44 double skew = 0; 45 double kurtosis = 0; 46 for (std::size_t i = 0; i < u.size(); ++i) 47 { 48 double dbl = (u[i] - mean); 49 double d2 = sqr(dbl); 50 var += d2; 51 skew += dbl * d2; 52 kurtosis += d2 * d2; 53 } 54 var /= u.size(); 55 double dev = std::sqrt(var); 56 skew /= u.size() * dev * var; 57 kurtosis /= u.size() * var * var; 58 kurtosis -= 3; 59 double x_mean = d.p(); 60 double x_var = d.p()*(1-d.p()); 61 double x_skew = (1 - 2 * d.p())/std::sqrt(x_var); 62 double x_kurtosis = (6 * sqr(d.p()) - 6 * d.p() + 1)/x_var; 63 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 64 assert(std::abs((var - x_var) / x_var) < 0.01); 65 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 66 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.02); 67 } 68 { 69 typedef std::bernoulli_distribution D; 70 typedef std::minstd_rand G; 71 G g; 72 D d(.25); 73 const int N = 100000; 74 std::vector<D::result_type> u; 75 for (int i = 0; i < N; ++i) 76 u.push_back(d(g)); 77 double mean = std::accumulate(u.begin(), u.end(), 78 double(0)) / u.size(); 79 double var = 0; 80 double skew = 0; 81 double kurtosis = 0; 82 for (std::size_t i = 0; i < u.size(); ++i) 83 { 84 double dbl = (u[i] - mean); 85 double d2 = sqr(dbl); 86 var += d2; 87 skew += dbl * d2; 88 kurtosis += d2 * d2; 89 } 90 var /= u.size(); 91 double dev = std::sqrt(var); 92 skew /= u.size() * dev * var; 93 kurtosis /= u.size() * var * var; 94 kurtosis -= 3; 95 double x_mean = d.p(); 96 double x_var = d.p()*(1-d.p()); 97 double x_skew = (1 - 2 * d.p())/std::sqrt(x_var); 98 double x_kurtosis = (6 * sqr(d.p()) - 6 * d.p() + 1)/x_var; 99 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 100 assert(std::abs((var - x_var) / x_var) < 0.01); 101 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 102 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.02); 103 } 104 } 105