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 // template<class RealType = double> 13 // class extreme_value_distribution 14 15 // template<class _URNG> result_type operator()(_URNG& g); 16 17 #include <random> 18 #include <cassert> 19 #include <vector> 20 #include <numeric> 21 22 template <class T> 23 inline 24 T 25 sqr(T x) 26 { 27 return x * x; 28 } 29 30 int main() 31 { 32 { 33 typedef std::extreme_value_distribution<> D; 34 typedef D::param_type P; 35 typedef std::mt19937 G; 36 G g; 37 D d(0.5, 2); 38 const int N = 1000000; 39 std::vector<D::result_type> u; 40 for (int i = 0; i < N; ++i) 41 { 42 D::result_type v = d(g); 43 u.push_back(v); 44 } 45 double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); 46 double var = 0; 47 double skew = 0; 48 double kurtosis = 0; 49 for (int i = 0; i < u.size(); ++i) 50 { 51 double d = (u[i] - mean); 52 double d2 = sqr(d); 53 var += d2; 54 skew += d * d2; 55 kurtosis += d2 * d2; 56 } 57 var /= u.size(); 58 double dev = std::sqrt(var); 59 skew /= u.size() * dev * var; 60 kurtosis /= u.size() * var * var; 61 kurtosis -= 3; 62 double x_mean = d.a() + d.b() * 0.577215665; 63 double x_var = sqr(d.b()) * 1.644934067; 64 double x_skew = 1.139547; 65 double x_kurtosis = 12./5; 66 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 67 assert(std::abs((var - x_var) / x_var) < 0.01); 68 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 69 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 70 } 71 { 72 typedef std::extreme_value_distribution<> D; 73 typedef D::param_type P; 74 typedef std::mt19937 G; 75 G g; 76 D d(1, 2); 77 const int N = 1000000; 78 std::vector<D::result_type> u; 79 for (int i = 0; i < N; ++i) 80 { 81 D::result_type v = d(g); 82 u.push_back(v); 83 } 84 double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); 85 double var = 0; 86 double skew = 0; 87 double kurtosis = 0; 88 for (int i = 0; i < u.size(); ++i) 89 { 90 double d = (u[i] - mean); 91 double d2 = sqr(d); 92 var += d2; 93 skew += d * d2; 94 kurtosis += d2 * d2; 95 } 96 var /= u.size(); 97 double dev = std::sqrt(var); 98 skew /= u.size() * dev * var; 99 kurtosis /= u.size() * var * var; 100 kurtosis -= 3; 101 double x_mean = d.a() + d.b() * 0.577215665; 102 double x_var = sqr(d.b()) * 1.644934067; 103 double x_skew = 1.139547; 104 double x_kurtosis = 12./5; 105 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 106 assert(std::abs((var - x_var) / x_var) < 0.01); 107 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 108 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 109 } 110 { 111 typedef std::extreme_value_distribution<> D; 112 typedef D::param_type P; 113 typedef std::mt19937 G; 114 G g; 115 D d(1.5, 3); 116 const int N = 1000000; 117 std::vector<D::result_type> u; 118 for (int i = 0; i < N; ++i) 119 { 120 D::result_type v = d(g); 121 u.push_back(v); 122 } 123 double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); 124 double var = 0; 125 double skew = 0; 126 double kurtosis = 0; 127 for (int i = 0; i < u.size(); ++i) 128 { 129 double d = (u[i] - mean); 130 double d2 = sqr(d); 131 var += d2; 132 skew += d * d2; 133 kurtosis += d2 * d2; 134 } 135 var /= u.size(); 136 double dev = std::sqrt(var); 137 skew /= u.size() * dev * var; 138 kurtosis /= u.size() * var * var; 139 kurtosis -= 3; 140 double x_mean = d.a() + d.b() * 0.577215665; 141 double x_var = sqr(d.b()) * 1.644934067; 142 double x_skew = 1.139547; 143 double x_kurtosis = 12./5; 144 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 145 assert(std::abs((var - x_var) / x_var) < 0.01); 146 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 147 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 148 } 149 { 150 typedef std::extreme_value_distribution<> D; 151 typedef D::param_type P; 152 typedef std::mt19937 G; 153 G g; 154 D d(3, 4); 155 const int N = 1000000; 156 std::vector<D::result_type> u; 157 for (int i = 0; i < N; ++i) 158 { 159 D::result_type v = d(g); 160 u.push_back(v); 161 } 162 double mean = std::accumulate(u.begin(), u.end(), 0.0) / u.size(); 163 double var = 0; 164 double skew = 0; 165 double kurtosis = 0; 166 for (int i = 0; i < u.size(); ++i) 167 { 168 double d = (u[i] - mean); 169 double d2 = sqr(d); 170 var += d2; 171 skew += d * d2; 172 kurtosis += d2 * d2; 173 } 174 var /= u.size(); 175 double dev = std::sqrt(var); 176 skew /= u.size() * dev * var; 177 kurtosis /= u.size() * var * var; 178 kurtosis -= 3; 179 double x_mean = d.a() + d.b() * 0.577215665; 180 double x_var = sqr(d.b()) * 1.644934067; 181 double x_skew = 1.139547; 182 double x_kurtosis = 12./5; 183 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 184 assert(std::abs((var - x_var) / x_var) < 0.01); 185 assert(std::abs((skew - x_skew) / x_skew) < 0.01); 186 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 187 } 188 } 189