• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 // REQUIRES: long_tests
11 
12 // <random>
13 
14 // template<class _IntType = int>
15 // class uniform_int_distribution
16 
17 // template<class _URNG> result_type operator()(_URNG& g);
18 
19 #include <random>
20 #include <cassert>
21 #include <vector>
22 #include <numeric>
23 #include <cstddef>
24 
25 template <class T>
26 inline
27 T
sqr(T x)28 sqr(T x)
29 {
30     return x * x;
31 }
32 
main()33 int main()
34 {
35     {
36         typedef std::uniform_int_distribution<> D;
37         typedef std::minstd_rand0 G;
38         G g;
39         D d;
40         const int N = 100000;
41         std::vector<D::result_type> u;
42         for (int i = 0; i < N; ++i)
43         {
44             D::result_type v = d(g);
45             assert(d.a() <= v && v <= d.b());
46             u.push_back(v);
47         }
48         double mean = std::accumulate(u.begin(), u.end(),
49                                               double(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(var);
63         skew /= u.size() * dev * var;
64         kurtosis /= u.size() * var * var;
65         kurtosis -= 3;
66         double x_mean = ((double)d.a() + d.b()) / 2;
67         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
68         double x_skew = 0;
69         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
70                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
71         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
72         assert(std::abs((var - x_var) / x_var) < 0.01);
73         assert(std::abs(skew - x_skew) < 0.01);
74         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
75     }
76     {
77         typedef std::uniform_int_distribution<> D;
78         typedef std::minstd_rand G;
79         G g;
80         D d;
81         const int N = 100000;
82         std::vector<D::result_type> u;
83         for (int i = 0; i < N; ++i)
84         {
85             D::result_type v = d(g);
86             assert(d.a() <= v && v <= d.b());
87             u.push_back(v);
88         }
89         double mean = std::accumulate(u.begin(), u.end(),
90                                               double(0)) / u.size();
91         double var = 0;
92         double skew = 0;
93         double kurtosis = 0;
94         for (std::size_t i = 0; i < u.size(); ++i)
95         {
96             double dbl = (u[i] - mean);
97             double d2 = sqr(dbl);
98             var += d2;
99             skew += dbl * d2;
100             kurtosis += d2 * d2;
101         }
102         var /= u.size();
103         double dev = std::sqrt(var);
104         skew /= u.size() * dev * var;
105         kurtosis /= u.size() * var * var;
106         kurtosis -= 3;
107         double x_mean = ((double)d.a() + d.b()) / 2;
108         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
109         double x_skew = 0;
110         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
111                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
112         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
113         assert(std::abs((var - x_var) / x_var) < 0.01);
114         assert(std::abs(skew - x_skew) < 0.01);
115         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
116     }
117     {
118         typedef std::uniform_int_distribution<> D;
119         typedef std::mt19937 G;
120         G g;
121         D d;
122         const int N = 100000;
123         std::vector<D::result_type> u;
124         for (int i = 0; i < N; ++i)
125         {
126             D::result_type v = d(g);
127             assert(d.a() <= v && v <= d.b());
128             u.push_back(v);
129         }
130         double mean = std::accumulate(u.begin(), u.end(),
131                                               double(0)) / u.size();
132         double var = 0;
133         double skew = 0;
134         double kurtosis = 0;
135         for (std::size_t i = 0; i < u.size(); ++i)
136         {
137             double dbl = (u[i] - mean);
138             double d2 = sqr(dbl);
139             var += d2;
140             skew += dbl * d2;
141             kurtosis += d2 * d2;
142         }
143         var /= u.size();
144         double dev = std::sqrt(var);
145         skew /= u.size() * dev * var;
146         kurtosis /= u.size() * var * var;
147         kurtosis -= 3;
148         double x_mean = ((double)d.a() + d.b()) / 2;
149         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
150         double x_skew = 0;
151         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
152                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
153         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
154         assert(std::abs((var - x_var) / x_var) < 0.01);
155         assert(std::abs(skew - x_skew) < 0.01);
156         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
157     }
158     {
159         typedef std::uniform_int_distribution<> D;
160         typedef std::mt19937_64 G;
161         G g;
162         D d;
163         const int N = 100000;
164         std::vector<D::result_type> u;
165         for (int i = 0; i < N; ++i)
166         {
167             D::result_type v = d(g);
168             assert(d.a() <= v && v <= d.b());
169             u.push_back(v);
170         }
171         double mean = std::accumulate(u.begin(), u.end(),
172                                               double(0)) / u.size();
173         double var = 0;
174         double skew = 0;
175         double kurtosis = 0;
176         for (std::size_t i = 0; i < u.size(); ++i)
177         {
178             double dbl = (u[i] - mean);
179             double d2 = sqr(dbl);
180             var += d2;
181             skew += dbl * d2;
182             kurtosis += d2 * d2;
183         }
184         var /= u.size();
185         double dev = std::sqrt(var);
186         skew /= u.size() * dev * var;
187         kurtosis /= u.size() * var * var;
188         kurtosis -= 3;
189         double x_mean = ((double)d.a() + d.b()) / 2;
190         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
191         double x_skew = 0;
192         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
193                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
194         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
195         assert(std::abs((var - x_var) / x_var) < 0.01);
196         assert(std::abs(skew - x_skew) < 0.01);
197         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
198     }
199     {
200         typedef std::uniform_int_distribution<> D;
201         typedef std::ranlux24_base G;
202         G g;
203         D d;
204         const int N = 100000;
205         std::vector<D::result_type> u;
206         for (int i = 0; i < N; ++i)
207         {
208             D::result_type v = d(g);
209             assert(d.a() <= v && v <= d.b());
210             u.push_back(v);
211         }
212         double mean = std::accumulate(u.begin(), u.end(),
213                                               double(0)) / u.size();
214         double var = 0;
215         double skew = 0;
216         double kurtosis = 0;
217         for (std::size_t i = 0; i < u.size(); ++i)
218         {
219             double dbl = (u[i] - mean);
220             double d2 = sqr(dbl);
221             var += d2;
222             skew += dbl * d2;
223             kurtosis += d2 * d2;
224         }
225         var /= u.size();
226         double dev = std::sqrt(var);
227         skew /= u.size() * dev * var;
228         kurtosis /= u.size() * var * var;
229         kurtosis -= 3;
230         double x_mean = ((double)d.a() + d.b()) / 2;
231         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
232         double x_skew = 0;
233         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
234                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
235         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
236         assert(std::abs((var - x_var) / x_var) < 0.01);
237         assert(std::abs(skew - x_skew) < 0.01);
238         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
239     }
240     {
241         typedef std::uniform_int_distribution<> D;
242         typedef std::ranlux48_base G;
243         G g;
244         D d;
245         const int N = 100000;
246         std::vector<D::result_type> u;
247         for (int i = 0; i < N; ++i)
248         {
249             D::result_type v = d(g);
250             assert(d.a() <= v && v <= d.b());
251             u.push_back(v);
252         }
253         double mean = std::accumulate(u.begin(), u.end(),
254                                               double(0)) / u.size();
255         double var = 0;
256         double skew = 0;
257         double kurtosis = 0;
258         for (std::size_t i = 0; i < u.size(); ++i)
259         {
260             double dbl = (u[i] - mean);
261             double d2 = sqr(dbl);
262             var += d2;
263             skew += dbl * d2;
264             kurtosis += d2 * d2;
265         }
266         var /= u.size();
267         double dev = std::sqrt(var);
268         skew /= u.size() * dev * var;
269         kurtosis /= u.size() * var * var;
270         kurtosis -= 3;
271         double x_mean = ((double)d.a() + d.b()) / 2;
272         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
273         double x_skew = 0;
274         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
275                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
276         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
277         assert(std::abs((var - x_var) / x_var) < 0.01);
278         assert(std::abs(skew - x_skew) < 0.01);
279         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
280     }
281     {
282         typedef std::uniform_int_distribution<> D;
283         typedef std::ranlux24 G;
284         G g;
285         D d;
286         const int N = 100000;
287         std::vector<D::result_type> u;
288         for (int i = 0; i < N; ++i)
289         {
290             D::result_type v = d(g);
291             assert(d.a() <= v && v <= d.b());
292             u.push_back(v);
293         }
294         double mean = std::accumulate(u.begin(), u.end(),
295                                               double(0)) / u.size();
296         double var = 0;
297         double skew = 0;
298         double kurtosis = 0;
299         for (std::size_t i = 0; i < u.size(); ++i)
300         {
301             double dbl = (u[i] - mean);
302             double d2 = sqr(dbl);
303             var += d2;
304             skew += dbl * d2;
305             kurtosis += d2 * d2;
306         }
307         var /= u.size();
308         double dev = std::sqrt(var);
309         skew /= u.size() * dev * var;
310         kurtosis /= u.size() * var * var;
311         kurtosis -= 3;
312         double x_mean = ((double)d.a() + d.b()) / 2;
313         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
314         double x_skew = 0;
315         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
316                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
317         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
318         assert(std::abs((var - x_var) / x_var) < 0.01);
319         assert(std::abs(skew - x_skew) < 0.01);
320         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
321     }
322     {
323         typedef std::uniform_int_distribution<> D;
324         typedef std::ranlux48 G;
325         G g;
326         D d;
327         const int N = 100000;
328         std::vector<D::result_type> u;
329         for (int i = 0; i < N; ++i)
330         {
331             D::result_type v = d(g);
332             assert(d.a() <= v && v <= d.b());
333             u.push_back(v);
334         }
335         double mean = std::accumulate(u.begin(), u.end(),
336                                               double(0)) / u.size();
337         double var = 0;
338         double skew = 0;
339         double kurtosis = 0;
340         for (std::size_t i = 0; i < u.size(); ++i)
341         {
342             double dbl = (u[i] - mean);
343             double d2 = sqr(dbl);
344             var += d2;
345             skew += dbl * d2;
346             kurtosis += d2 * d2;
347         }
348         var /= u.size();
349         double dev = std::sqrt(var);
350         skew /= u.size() * dev * var;
351         kurtosis /= u.size() * var * var;
352         kurtosis -= 3;
353         double x_mean = ((double)d.a() + d.b()) / 2;
354         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
355         double x_skew = 0;
356         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
357                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
358         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
359         assert(std::abs((var - x_var) / x_var) < 0.01);
360         assert(std::abs(skew - x_skew) < 0.01);
361         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
362     }
363     {
364         typedef std::uniform_int_distribution<> D;
365         typedef std::knuth_b G;
366         G g;
367         D d;
368         const int N = 100000;
369         std::vector<D::result_type> u;
370         for (int i = 0; i < N; ++i)
371         {
372             D::result_type v = d(g);
373             assert(d.a() <= v && v <= d.b());
374             u.push_back(v);
375         }
376         double mean = std::accumulate(u.begin(), u.end(),
377                                               double(0)) / u.size();
378         double var = 0;
379         double skew = 0;
380         double kurtosis = 0;
381         for (std::size_t i = 0; i < u.size(); ++i)
382         {
383             double dbl = (u[i] - mean);
384             double d2 = sqr(dbl);
385             var += d2;
386             skew += dbl * d2;
387             kurtosis += d2 * d2;
388         }
389         var /= u.size();
390         double dev = std::sqrt(var);
391         skew /= u.size() * dev * var;
392         kurtosis /= u.size() * var * var;
393         kurtosis -= 3;
394         double x_mean = ((double)d.a() + d.b()) / 2;
395         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
396         double x_skew = 0;
397         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
398                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
399         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
400         assert(std::abs((var - x_var) / x_var) < 0.01);
401         assert(std::abs(skew - x_skew) < 0.01);
402         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
403     }
404     {
405         typedef std::uniform_int_distribution<> D;
406         typedef std::minstd_rand0 G;
407         G g;
408         D d(-6, 106);
409         for (int i = 0; i < 10000; ++i)
410         {
411             int u = d(g);
412             assert(-6 <= u && u <= 106);
413         }
414     }
415     {
416         typedef std::uniform_int_distribution<> D;
417         typedef std::minstd_rand G;
418         G g;
419         D d(5, 100);
420         const int N = 100000;
421         std::vector<D::result_type> u;
422         for (int i = 0; i < N; ++i)
423         {
424             D::result_type v = d(g);
425             assert(d.a() <= v && v <= d.b());
426             u.push_back(v);
427         }
428         double mean = std::accumulate(u.begin(), u.end(),
429                                               double(0)) / u.size();
430         double var = 0;
431         double skew = 0;
432         double kurtosis = 0;
433         for (std::size_t i = 0; i < u.size(); ++i)
434         {
435             double dbl = (u[i] - mean);
436             double d2 = sqr(dbl);
437             var += d2;
438             skew += dbl * d2;
439             kurtosis += d2 * d2;
440         }
441         var /= u.size();
442         double dev = std::sqrt(var);
443         skew /= u.size() * dev * var;
444         kurtosis /= u.size() * var * var;
445         kurtosis -= 3;
446         double x_mean = ((double)d.a() + d.b()) / 2;
447         double x_var = (sqr((double)d.b() - d.a() + 1) - 1) / 12;
448         double x_skew = 0;
449         double x_kurtosis = -6. * (sqr((double)d.b() - d.a() + 1) + 1) /
450                             (5. * (sqr((double)d.b() - d.a() + 1) - 1));
451         assert(std::abs((mean - x_mean) / x_mean) < 0.01);
452         assert(std::abs((var - x_var) / x_var) < 0.01);
453         assert(std::abs(skew - x_skew) < 0.01);
454         assert(std::abs((kurtosis - x_kurtosis) / x_kurtosis) < 0.01);
455     }
456 }
457