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 _IntType = int> 13 // class uniform_int_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::uniform_int_distribution<> D; 34 typedef std::minstd_rand0 G; 35 G g; 36 D d; 37 const int N = 100000; 38 std::vector<D::result_type> u; 39 for (int i = 0; i < N; ++i) 40 { 41 D::result_type v = d(g); 42 assert(d.a() <= v && v <= d.b()); 43 u.push_back(v); 44 } 45 double mean = std::accumulate(u.begin(), u.end(), 46 double(0)) / u.size(); 47 double var = 0; 48 double skew = 0; 49 double kurtosis = 0; 50 for (int i = 0; i < u.size(); ++i) 51 { 52 double d = (u[i] - mean); 53 double d2 = sqr(d); 54 var += d2; 55 skew += d * d2; 56 kurtosis += d2 * d2; 57 } 58 var /= u.size(); 59 double dev = std::sqrt(var); 60 skew /= u.size() * dev * var; 61 kurtosis /= u.size() * var * var; 62 kurtosis -= 3; 63 double x_mean = ((double)d.a() + d.b()) / 2; 64 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 65 double x_skew = 0; 66 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 67 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 68 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 69 assert(std::abs((var - x_var) / x_var) < 0.01); 70 assert(std::abs(skew - x_skew) < 0.01); 71 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 72 } 73 { 74 typedef std::uniform_int_distribution<> D; 75 typedef std::minstd_rand G; 76 G g; 77 D d; 78 const int N = 100000; 79 std::vector<D::result_type> u; 80 for (int i = 0; i < N; ++i) 81 { 82 D::result_type v = d(g); 83 assert(d.a() <= v && v <= d.b()); 84 u.push_back(v); 85 } 86 double mean = std::accumulate(u.begin(), u.end(), 87 double(0)) / u.size(); 88 double var = 0; 89 double skew = 0; 90 double kurtosis = 0; 91 for (int i = 0; i < u.size(); ++i) 92 { 93 double d = (u[i] - mean); 94 double d2 = sqr(d); 95 var += d2; 96 skew += d * d2; 97 kurtosis += d2 * d2; 98 } 99 var /= u.size(); 100 double dev = std::sqrt(var); 101 skew /= u.size() * dev * var; 102 kurtosis /= u.size() * var * var; 103 kurtosis -= 3; 104 double x_mean = ((double)d.a() + d.b()) / 2; 105 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 106 double x_skew = 0; 107 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 108 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 109 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 110 assert(std::abs((var - x_var) / x_var) < 0.01); 111 assert(std::abs(skew - x_skew) < 0.01); 112 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 113 } 114 { 115 typedef std::uniform_int_distribution<> D; 116 typedef std::mt19937 G; 117 G g; 118 D d; 119 const int N = 100000; 120 std::vector<D::result_type> u; 121 for (int i = 0; i < N; ++i) 122 { 123 D::result_type v = d(g); 124 assert(d.a() <= v && v <= d.b()); 125 u.push_back(v); 126 } 127 double mean = std::accumulate(u.begin(), u.end(), 128 double(0)) / u.size(); 129 double var = 0; 130 double skew = 0; 131 double kurtosis = 0; 132 for (int i = 0; i < u.size(); ++i) 133 { 134 double d = (u[i] - mean); 135 double d2 = sqr(d); 136 var += d2; 137 skew += d * d2; 138 kurtosis += d2 * d2; 139 } 140 var /= u.size(); 141 double dev = std::sqrt(var); 142 skew /= u.size() * dev * var; 143 kurtosis /= u.size() * var * var; 144 kurtosis -= 3; 145 double x_mean = ((double)d.a() + d.b()) / 2; 146 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 147 double x_skew = 0; 148 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 149 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 150 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 151 assert(std::abs((var - x_var) / x_var) < 0.01); 152 assert(std::abs(skew - x_skew) < 0.01); 153 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 154 } 155 { 156 typedef std::uniform_int_distribution<> D; 157 typedef std::mt19937_64 G; 158 G g; 159 D d; 160 const int N = 100000; 161 std::vector<D::result_type> u; 162 for (int i = 0; i < N; ++i) 163 { 164 D::result_type v = d(g); 165 assert(d.a() <= v && v <= d.b()); 166 u.push_back(v); 167 } 168 double mean = std::accumulate(u.begin(), u.end(), 169 double(0)) / u.size(); 170 double var = 0; 171 double skew = 0; 172 double kurtosis = 0; 173 for (int i = 0; i < u.size(); ++i) 174 { 175 double d = (u[i] - mean); 176 double d2 = sqr(d); 177 var += d2; 178 skew += d * d2; 179 kurtosis += d2 * d2; 180 } 181 var /= u.size(); 182 double dev = std::sqrt(var); 183 skew /= u.size() * dev * var; 184 kurtosis /= u.size() * var * var; 185 kurtosis -= 3; 186 double x_mean = ((double)d.a() + d.b()) / 2; 187 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 188 double x_skew = 0; 189 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 190 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 191 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 192 assert(std::abs((var - x_var) / x_var) < 0.01); 193 assert(std::abs(skew - x_skew) < 0.01); 194 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 195 } 196 { 197 typedef std::uniform_int_distribution<> D; 198 typedef std::ranlux24_base G; 199 G g; 200 D d; 201 const int N = 100000; 202 std::vector<D::result_type> u; 203 for (int i = 0; i < N; ++i) 204 { 205 D::result_type v = d(g); 206 assert(d.a() <= v && v <= d.b()); 207 u.push_back(v); 208 } 209 double mean = std::accumulate(u.begin(), u.end(), 210 double(0)) / u.size(); 211 double var = 0; 212 double skew = 0; 213 double kurtosis = 0; 214 for (int i = 0; i < u.size(); ++i) 215 { 216 double d = (u[i] - mean); 217 double d2 = sqr(d); 218 var += d2; 219 skew += d * d2; 220 kurtosis += d2 * d2; 221 } 222 var /= u.size(); 223 double dev = std::sqrt(var); 224 skew /= u.size() * dev * var; 225 kurtosis /= u.size() * var * var; 226 kurtosis -= 3; 227 double x_mean = ((double)d.a() + d.b()) / 2; 228 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 229 double x_skew = 0; 230 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 231 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 232 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 233 assert(std::abs((var - x_var) / x_var) < 0.01); 234 assert(std::abs(skew - x_skew) < 0.01); 235 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 236 } 237 { 238 typedef std::uniform_int_distribution<> D; 239 typedef std::ranlux48_base G; 240 G g; 241 D d; 242 const int N = 100000; 243 std::vector<D::result_type> u; 244 for (int i = 0; i < N; ++i) 245 { 246 D::result_type v = d(g); 247 assert(d.a() <= v && v <= d.b()); 248 u.push_back(v); 249 } 250 double mean = std::accumulate(u.begin(), u.end(), 251 double(0)) / u.size(); 252 double var = 0; 253 double skew = 0; 254 double kurtosis = 0; 255 for (int i = 0; i < u.size(); ++i) 256 { 257 double d = (u[i] - mean); 258 double d2 = sqr(d); 259 var += d2; 260 skew += d * d2; 261 kurtosis += d2 * d2; 262 } 263 var /= u.size(); 264 double dev = std::sqrt(var); 265 skew /= u.size() * dev * var; 266 kurtosis /= u.size() * var * var; 267 kurtosis -= 3; 268 double x_mean = ((double)d.a() + d.b()) / 2; 269 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 270 double x_skew = 0; 271 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 272 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 273 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 274 assert(std::abs((var - x_var) / x_var) < 0.01); 275 assert(std::abs(skew - x_skew) < 0.01); 276 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 277 } 278 { 279 typedef std::uniform_int_distribution<> D; 280 typedef std::ranlux24 G; 281 G g; 282 D d; 283 const int N = 100000; 284 std::vector<D::result_type> u; 285 for (int i = 0; i < N; ++i) 286 { 287 D::result_type v = d(g); 288 assert(d.a() <= v && v <= d.b()); 289 u.push_back(v); 290 } 291 double mean = std::accumulate(u.begin(), u.end(), 292 double(0)) / u.size(); 293 double var = 0; 294 double skew = 0; 295 double kurtosis = 0; 296 for (int i = 0; i < u.size(); ++i) 297 { 298 double d = (u[i] - mean); 299 double d2 = sqr(d); 300 var += d2; 301 skew += d * d2; 302 kurtosis += d2 * d2; 303 } 304 var /= u.size(); 305 double dev = std::sqrt(var); 306 skew /= u.size() * dev * var; 307 kurtosis /= u.size() * var * var; 308 kurtosis -= 3; 309 double x_mean = ((double)d.a() + d.b()) / 2; 310 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 311 double x_skew = 0; 312 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 313 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 314 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 315 assert(std::abs((var - x_var) / x_var) < 0.01); 316 assert(std::abs(skew - x_skew) < 0.01); 317 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 318 } 319 { 320 typedef std::uniform_int_distribution<> D; 321 typedef std::ranlux48 G; 322 G g; 323 D d; 324 const int N = 100000; 325 std::vector<D::result_type> u; 326 for (int i = 0; i < N; ++i) 327 { 328 D::result_type v = d(g); 329 assert(d.a() <= v && v <= d.b()); 330 u.push_back(v); 331 } 332 double mean = std::accumulate(u.begin(), u.end(), 333 double(0)) / u.size(); 334 double var = 0; 335 double skew = 0; 336 double kurtosis = 0; 337 for (int i = 0; i < u.size(); ++i) 338 { 339 double d = (u[i] - mean); 340 double d2 = sqr(d); 341 var += d2; 342 skew += d * d2; 343 kurtosis += d2 * d2; 344 } 345 var /= u.size(); 346 double dev = std::sqrt(var); 347 skew /= u.size() * dev * var; 348 kurtosis /= u.size() * var * var; 349 kurtosis -= 3; 350 double x_mean = ((double)d.a() + d.b()) / 2; 351 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 352 double x_skew = 0; 353 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 354 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 355 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 356 assert(std::abs((var - x_var) / x_var) < 0.01); 357 assert(std::abs(skew - x_skew) < 0.01); 358 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 359 } 360 { 361 typedef std::uniform_int_distribution<> D; 362 typedef std::knuth_b G; 363 G g; 364 D d; 365 const int N = 100000; 366 std::vector<D::result_type> u; 367 for (int i = 0; i < N; ++i) 368 { 369 D::result_type v = d(g); 370 assert(d.a() <= v && v <= d.b()); 371 u.push_back(v); 372 } 373 double mean = std::accumulate(u.begin(), u.end(), 374 double(0)) / u.size(); 375 double var = 0; 376 double skew = 0; 377 double kurtosis = 0; 378 for (int i = 0; i < u.size(); ++i) 379 { 380 double d = (u[i] - mean); 381 double d2 = sqr(d); 382 var += d2; 383 skew += d * d2; 384 kurtosis += d2 * d2; 385 } 386 var /= u.size(); 387 double dev = std::sqrt(var); 388 skew /= u.size() * dev * var; 389 kurtosis /= u.size() * var * var; 390 kurtosis -= 3; 391 double x_mean = ((double)d.a() + d.b()) / 2; 392 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 393 double x_skew = 0; 394 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 395 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 396 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 397 assert(std::abs((var - x_var) / x_var) < 0.01); 398 assert(std::abs(skew - x_skew) < 0.01); 399 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 400 } 401 { 402 typedef std::uniform_int_distribution<> D; 403 typedef std::minstd_rand0 G; 404 G g; 405 D d(-6, 106); 406 for (int i = 0; i < 10000; ++i) 407 { 408 int u = d(g); 409 assert(-6 <= u && u <= 106); 410 } 411 } 412 { 413 typedef std::uniform_int_distribution<> D; 414 typedef std::minstd_rand G; 415 G g; 416 D d(5, 100); 417 const int N = 100000; 418 std::vector<D::result_type> u; 419 for (int i = 0; i < N; ++i) 420 { 421 D::result_type v = d(g); 422 assert(d.a() <= v && v <= d.b()); 423 u.push_back(v); 424 } 425 double mean = std::accumulate(u.begin(), u.end(), 426 double(0)) / u.size(); 427 double var = 0; 428 double skew = 0; 429 double kurtosis = 0; 430 for (int i = 0; i < u.size(); ++i) 431 { 432 double d = (u[i] - mean); 433 double d2 = sqr(d); 434 var += d2; 435 skew += d * d2; 436 kurtosis += d2 * d2; 437 } 438 var /= u.size(); 439 double dev = std::sqrt(var); 440 skew /= u.size() * dev * var; 441 kurtosis /= u.size() * var * var; 442 kurtosis -= 3; 443 double x_mean = ((double)d.a() + d.b()) / 2; 444 double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12; 445 double x_skew = 0; 446 double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) / 447 (5. * (sqr((double)d.b() - d.a() + 1) - 1)); 448 assert(std::abs((mean - x_mean) / x_mean) < 0.01); 449 assert(std::abs((var - x_var) / x_var) < 0.01); 450 assert(std::abs(skew - x_skew) < 0.01); 451 assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01); 452 } 453 } 454
