1 | // $Id: random.h 443 2005-12-15 15:28:37Z jari $ |
---|
2 | |
---|
3 | #ifndef _theplu_random_ |
---|
4 | #define _theplu_random_ |
---|
5 | |
---|
6 | #include <c++_tools/statistics/Histogram.h> |
---|
7 | |
---|
8 | #include <gsl/gsl_rng.h> |
---|
9 | #include <gsl/gsl_randist.h> |
---|
10 | |
---|
11 | #include <string> |
---|
12 | |
---|
13 | namespace theplu { |
---|
14 | namespace random { |
---|
15 | |
---|
16 | /// |
---|
17 | /// @brief Random Number Generator |
---|
18 | /// |
---|
19 | /// The RNG class is wrapper to the GSL random number generator |
---|
20 | /// (rng). This class provides a single global instance of the rng, |
---|
21 | /// and makes sure there is only one point of access to the |
---|
22 | /// generator. |
---|
23 | /// |
---|
24 | /// There is information about how to change seeding and generators |
---|
25 | /// at run time without recompilation using environment |
---|
26 | /// variables. RNG of course support seeding at compile time if you |
---|
27 | /// don't want to bother about environment variables and GSL. |
---|
28 | /// |
---|
29 | /// There are many different rng's available in GSL. Currently only |
---|
30 | /// the default generator is implemented and no other one is |
---|
31 | /// choosable through the class interface. This means that you have |
---|
32 | /// to fall back to the use of environment variables as described in |
---|
33 | /// the GSL documentation, or be bold and request support for other |
---|
34 | /// rng's through the class interface. |
---|
35 | /// |
---|
36 | /// Not all GSL functionality is implemented, we'll add |
---|
37 | /// functionality when needed and may do it when requested. Better |
---|
38 | /// yet, supply us with code and we will probably add it to the code |
---|
39 | /// (BUT remember to implement reasonable tests for your code and |
---|
40 | /// follow the coding style.) |
---|
41 | /// |
---|
42 | /// This implementation may be thread safe (according to GSL |
---|
43 | /// documentation), but should be checked to be so before trusting |
---|
44 | /// thread safety. |
---|
45 | /// |
---|
46 | /// @see Design Patterns (the singleton and adapter pattern). GSL |
---|
47 | /// documentation. |
---|
48 | /// |
---|
49 | /// @todo Get those exceptions in! Should classes be considered |
---|
50 | /// const if underlying structures are changed such as GSL stuff? |
---|
51 | /// |
---|
52 | class RNG |
---|
53 | { |
---|
54 | public: |
---|
55 | |
---|
56 | virtual ~RNG(void); |
---|
57 | |
---|
58 | /// |
---|
59 | /// Get an instance of the random number generator. If the random |
---|
60 | /// number generator is not already created, the call will create |
---|
61 | /// a new generator and use the default seed. I needed, the seed |
---|
62 | /// must be changed with the seed or seed_from_devurandom member |
---|
63 | /// functions. |
---|
64 | /// |
---|
65 | /// @brief Get an instance of the random number generator. |
---|
66 | /// |
---|
67 | /// @return A pointer to the random number generator. |
---|
68 | /// |
---|
69 | /// @see seed and seed_from_devurandom |
---|
70 | /// |
---|
71 | static RNG* instance(void) |
---|
72 | { if (!instance_) instance_=new RNG; return instance_; } |
---|
73 | |
---|
74 | /// |
---|
75 | /// @brief Returns the largest number that the random number |
---|
76 | /// generator can return. |
---|
77 | /// |
---|
78 | inline u_long max(void) const { return gsl_rng_max(rng_); } |
---|
79 | |
---|
80 | /// |
---|
81 | /// @brief Returns the smallest number that the random number |
---|
82 | /// generator can return. |
---|
83 | /// |
---|
84 | inline u_long min(void) const { return gsl_rng_min(rng_); } |
---|
85 | |
---|
86 | /// |
---|
87 | /// @brief Returns the name of the random number generator |
---|
88 | /// |
---|
89 | inline std::string name(void) const { return gsl_rng_name(rng_); } |
---|
90 | |
---|
91 | inline const gsl_rng* rng(void) const { return rng_; } |
---|
92 | |
---|
93 | /// |
---|
94 | /// Set the seed \a s for the rng. If \a s is zero, a default |
---|
95 | /// value from the rng's original implementation is used (cf. GSL |
---|
96 | /// documentation). |
---|
97 | /// |
---|
98 | /// @brief Set the seed \a s for the rng. |
---|
99 | /// |
---|
100 | /// @see seed_from_devurandom |
---|
101 | /// |
---|
102 | inline void seed(u_long s) const { gsl_rng_set(rng_,s); } |
---|
103 | |
---|
104 | /// |
---|
105 | /// @brief Seed the rng using the /dev/urandom device. |
---|
106 | /// |
---|
107 | /// @return The seed acquired from /dev/urandom. |
---|
108 | /// |
---|
109 | u_long seed_from_devurandom(void); |
---|
110 | |
---|
111 | private: |
---|
112 | RNG(void); |
---|
113 | |
---|
114 | static RNG* instance_; |
---|
115 | gsl_rng* rng_; |
---|
116 | }; |
---|
117 | |
---|
118 | // --------------------- Discrete distribtuions --------------------- |
---|
119 | |
---|
120 | /// |
---|
121 | /// @brief Discrete random number distributions. |
---|
122 | /// |
---|
123 | /// Abstract base class for discrete random number |
---|
124 | /// distributions. Given K discrete events with different |
---|
125 | /// probabilities \f$ P[k] \f$, produce a random value k consistent |
---|
126 | /// with its probability. |
---|
127 | /// |
---|
128 | class Discrete |
---|
129 | { |
---|
130 | public: |
---|
131 | /// |
---|
132 | /// @brief Constructor |
---|
133 | /// |
---|
134 | inline Discrete(void) { rng_=RNG::instance(); } |
---|
135 | |
---|
136 | inline virtual ~Discrete(void) { } |
---|
137 | |
---|
138 | /// |
---|
139 | /// Set the seed to \a s in the underlying rng. If \a s is zero, a |
---|
140 | /// default value from the rng's original implementation is used |
---|
141 | /// (cf. GSL documentation). |
---|
142 | /// |
---|
143 | /// @brief Set the seed to \a s. |
---|
144 | /// |
---|
145 | /// @see seed_from_devurandom, RNG::seed_from_devurandom, RNG::seed |
---|
146 | /// |
---|
147 | inline void seed(u_long s) const { rng_->seed(s); } |
---|
148 | |
---|
149 | /// |
---|
150 | /// @brief Set the seed using the /dev/urandom device. |
---|
151 | /// |
---|
152 | /// @return The seed acquired from /dev/urandom. |
---|
153 | /// |
---|
154 | /// @see seed, RNG::seed_from_devurandom, RNG::seed |
---|
155 | /// |
---|
156 | u_long seed_from_devurandom(void) { return rng_->seed_from_devurandom(); } |
---|
157 | |
---|
158 | /// |
---|
159 | /// @return A random number. |
---|
160 | /// |
---|
161 | virtual u_long operator()(void) const = 0; |
---|
162 | |
---|
163 | protected: |
---|
164 | RNG* rng_; |
---|
165 | }; |
---|
166 | |
---|
167 | /// |
---|
168 | /// @brief General |
---|
169 | /// |
---|
170 | class DiscreteGeneral : public Discrete |
---|
171 | { |
---|
172 | public: |
---|
173 | /// |
---|
174 | /// @brief Constructor |
---|
175 | /// |
---|
176 | /// @param hist is a Histogram defining the probability distribution |
---|
177 | /// |
---|
178 | DiscreteGeneral(const statistics::Histogram& hist); |
---|
179 | |
---|
180 | /// |
---|
181 | /// @brief Destructor |
---|
182 | /// |
---|
183 | ~DiscreteGeneral(void); |
---|
184 | |
---|
185 | /// |
---|
186 | /// The generated number is an integer and proportinal to the |
---|
187 | /// frequency in the corresponding histogram bin. In other words, |
---|
188 | /// the probability that 0 is returned is proportinal to the size |
---|
189 | /// of the first bin. |
---|
190 | /// |
---|
191 | /// @return A random number. |
---|
192 | /// |
---|
193 | inline u_long |
---|
194 | operator()(void) const { return gsl_ran_discrete(rng_->rng(), gen_); } |
---|
195 | |
---|
196 | private: |
---|
197 | gsl_ran_discrete_t* gen_; |
---|
198 | }; |
---|
199 | |
---|
200 | /// |
---|
201 | /// @brief Discrete uniform distribution |
---|
202 | /// |
---|
203 | /// Discrete uniform distribution also known as the "equally likely |
---|
204 | /// outcomes" distribution. Each outcome, in this case an integer |
---|
205 | /// from [0,n-1] , have equal probability to occur. |
---|
206 | /// |
---|
207 | /// Distribution function \f$ p(k) = \frac{1}{n+1} \f$ for \f$ 0 \le |
---|
208 | /// k < n \f$ \n |
---|
209 | /// Expectation value: \f$ \frac{n-1}{2} \f$ \n |
---|
210 | /// Variance: \f$ \frac{1}{12}(n-1)(n+1) \f$ |
---|
211 | /// |
---|
212 | class DiscreteUniform : public Discrete |
---|
213 | { |
---|
214 | public: |
---|
215 | /// |
---|
216 | /// @brief Default constructor. |
---|
217 | /// |
---|
218 | DiscreteUniform(void) : range_(rng_->max()) {} |
---|
219 | |
---|
220 | /// |
---|
221 | /// @brief Constructor. |
---|
222 | /// |
---|
223 | /// The generator will generate integers from \f$ [0,n-1] \f$. If |
---|
224 | /// \a n is larger than the maximum number the random number |
---|
225 | /// generator can return, then (currently) \a n is adjusted |
---|
226 | /// appropriately. |
---|
227 | /// |
---|
228 | /// @todo If a too large \a n is given an exception should be |
---|
229 | /// thrown, i.e. the behaviour of this class will change. The case |
---|
230 | /// when argument is 0 is not treated gracefully (underlying GSL |
---|
231 | /// functionality will not return). |
---|
232 | /// |
---|
233 | DiscreteUniform(const u_long n) : range_(n) |
---|
234 | { if ( range_>rng_->max() ) range_=rng_->max(); } |
---|
235 | |
---|
236 | /// |
---|
237 | /// This function returns a random integer from 0 to n-1 |
---|
238 | /// inclusive. All integers in the range [0,n-1] are equally |
---|
239 | /// likely. n is set in constructor. |
---|
240 | /// |
---|
241 | inline u_long |
---|
242 | operator()(void) const { return gsl_rng_uniform_int(rng_->rng(), range_); } |
---|
243 | |
---|
244 | /// |
---|
245 | /// This function returns a random integer from 0 to n-1 |
---|
246 | /// inclusive. All integers in the range [0,n-1] are equally |
---|
247 | /// likely. |
---|
248 | /// |
---|
249 | inline u_long operator()(const u_long n) const |
---|
250 | { return gsl_rng_uniform_int(rng_->rng(), n); } |
---|
251 | |
---|
252 | private: |
---|
253 | u_long range_; |
---|
254 | }; |
---|
255 | |
---|
256 | /// |
---|
257 | /// @brief Poisson Distribution |
---|
258 | /// |
---|
259 | /// Having a Poisson process (i.e. no memory), number of occurences |
---|
260 | /// within a given time window is Poisson distributed. This |
---|
261 | /// distribution is the limit of a Binomial distribution when number |
---|
262 | /// of attempts is large, and the probability for one attempt to be |
---|
263 | /// succesful is small (in such a way that the expected number of |
---|
264 | /// succesful attempts is \f$ m \f$. |
---|
265 | /// |
---|
266 | /// Probability function \f$ p(k) = e^{-m}\frac{m^k}{k!} \f$ for \f$ 0 \le |
---|
267 | /// k \f$ \n |
---|
268 | /// Expectation value: \f$ m \f$ \n |
---|
269 | /// Variance: \f$ m \f$ |
---|
270 | /// |
---|
271 | class Poisson : public Discrete |
---|
272 | { |
---|
273 | public: |
---|
274 | /// |
---|
275 | /// @brief Constructor |
---|
276 | /// |
---|
277 | /// @param m is expectation value |
---|
278 | inline Poisson(const double m=1) : m_(m) {} |
---|
279 | |
---|
280 | /// |
---|
281 | /// @return A Poisson distributed number. |
---|
282 | /// |
---|
283 | inline u_long |
---|
284 | operator()(void) const { return gsl_ran_poisson(rng_->rng(), m_); } |
---|
285 | |
---|
286 | /// |
---|
287 | /// @return A Poisson distributed number with expectation value \a |
---|
288 | /// m |
---|
289 | /// |
---|
290 | /// @note this operator ignores parameters set in Constructor |
---|
291 | /// |
---|
292 | inline u_long |
---|
293 | operator()(const double m) const { return gsl_ran_poisson(rng_->rng(), m); } |
---|
294 | |
---|
295 | private: |
---|
296 | double m_; |
---|
297 | }; |
---|
298 | |
---|
299 | // --------------------- Continuous distribtuions --------------------- |
---|
300 | |
---|
301 | /// |
---|
302 | /// @brief Continuous random number distributions. |
---|
303 | /// |
---|
304 | /// Abstract base class for continuous random number distributions. |
---|
305 | /// |
---|
306 | class Continuous |
---|
307 | { |
---|
308 | public: |
---|
309 | |
---|
310 | /// |
---|
311 | /// @brief Constructor |
---|
312 | /// |
---|
313 | inline Continuous(void) { rng_=RNG::instance(); } |
---|
314 | |
---|
315 | inline virtual ~Continuous(void) { } |
---|
316 | |
---|
317 | /// |
---|
318 | /// Set the seed to \a s in the underlying rng. If \a s is zero, a |
---|
319 | /// default value from the rng's original implementation is used |
---|
320 | /// (cf. GSL documentation). |
---|
321 | /// |
---|
322 | /// @brief Set the seed to \a s. |
---|
323 | /// |
---|
324 | /// @see seed_from_devurandom, RNG::seed_from_devurandom, RNG::seed |
---|
325 | /// |
---|
326 | inline void seed(u_long s) const { rng_->seed(s); } |
---|
327 | |
---|
328 | /// |
---|
329 | /// @brief Set the seed using the /dev/urandom device. |
---|
330 | /// |
---|
331 | /// @return The seed acquired from /dev/urandom. |
---|
332 | /// |
---|
333 | /// @see seed, RNG::seed_from_devurandom, RNG::seed |
---|
334 | /// |
---|
335 | u_long seed_from_devurandom(void) { return rng_->seed_from_devurandom(); } |
---|
336 | |
---|
337 | /// |
---|
338 | /// @return A random number |
---|
339 | /// |
---|
340 | virtual double operator()(void) const = 0; |
---|
341 | |
---|
342 | protected: |
---|
343 | RNG* rng_; |
---|
344 | }; |
---|
345 | |
---|
346 | /// |
---|
347 | /// @brief Uniform distribution |
---|
348 | /// |
---|
349 | /// Class for generating a random number from a uniform distribution |
---|
350 | /// in the range [0,1), i.e. zero is included but not 1. |
---|
351 | /// |
---|
352 | /// Distribution function \f$ f(x) = 1 \f$ for \f$ 0 \le x < 1 \f$ \n |
---|
353 | /// Expectation value: 0.5 \n |
---|
354 | /// Variance: \f$ \frac{1}{12} \f$ |
---|
355 | /// |
---|
356 | class ContinuousUniform : public Continuous |
---|
357 | { |
---|
358 | public: |
---|
359 | inline double operator()(void) const { return gsl_rng_uniform(rng_->rng());} |
---|
360 | }; |
---|
361 | |
---|
362 | /// |
---|
363 | /// Class to generate numbers from a histogram in a continuous manner. |
---|
364 | /// |
---|
365 | class ContinuousGeneral : public Continuous |
---|
366 | { |
---|
367 | public: |
---|
368 | /// |
---|
369 | /// @brief Constructor |
---|
370 | /// |
---|
371 | /// @param hist is a Histogram defining the probability distribution |
---|
372 | /// |
---|
373 | inline ContinuousGeneral(const statistics::Histogram& hist) |
---|
374 | : discrete_(DiscreteGeneral(hist)), hist_(hist) {} |
---|
375 | |
---|
376 | /// |
---|
377 | /// @brief Destructor |
---|
378 | /// |
---|
379 | ~ContinuousGeneral(void); |
---|
380 | |
---|
381 | /// |
---|
382 | /// The number is generated in a two step process. First the bin |
---|
383 | /// in the histogram is randomly selected (see |
---|
384 | /// DiscreteGeneral). Then a number is generated uniformly from |
---|
385 | /// the interval defined by the bin. |
---|
386 | /// |
---|
387 | /// @return A random number. |
---|
388 | /// |
---|
389 | inline double operator()(void) const |
---|
390 | { return hist_.observation_value(discrete_())+(u_()-0.5)*hist_.spacing(); } |
---|
391 | |
---|
392 | private: |
---|
393 | const DiscreteGeneral discrete_; |
---|
394 | const statistics::Histogram& hist_; |
---|
395 | ContinuousUniform u_; |
---|
396 | }; |
---|
397 | |
---|
398 | /// |
---|
399 | /// @brief Generator of random numbers from an exponential |
---|
400 | /// distribution. |
---|
401 | /// |
---|
402 | /// The distribution function is \f$ f(x) = \frac{1}{m}\exp(-x/a) |
---|
403 | /// \f$ for \f$ x \f$ with the expectation value \f$ m \f$ and |
---|
404 | /// variance \f$ m^2 \f$ |
---|
405 | /// |
---|
406 | class Exponential : public Continuous |
---|
407 | { |
---|
408 | public: |
---|
409 | /// |
---|
410 | /// @brief Constructor |
---|
411 | /// |
---|
412 | /// @param \a m is the expectation value of the distribution. |
---|
413 | /// |
---|
414 | inline Exponential(const double m=1) : m_(m) {} |
---|
415 | |
---|
416 | /// |
---|
417 | /// @return A random number from exponential distribution. |
---|
418 | /// |
---|
419 | inline double |
---|
420 | operator()(void) const { return gsl_ran_exponential(rng_->rng(), m_); } |
---|
421 | |
---|
422 | /// |
---|
423 | /// @return A random number from exponential distribution, with |
---|
424 | /// expectation value \a m |
---|
425 | /// |
---|
426 | /// @note This operator ignores parameters given in constructor. |
---|
427 | /// |
---|
428 | inline double operator()(const double m) const |
---|
429 | { return gsl_ran_exponential(rng_->rng(), m); } |
---|
430 | |
---|
431 | private: |
---|
432 | double m_; |
---|
433 | }; |
---|
434 | |
---|
435 | /// |
---|
436 | /// @brief Gaussian distribution |
---|
437 | /// |
---|
438 | /// Class for generating a random number from a Gaussian |
---|
439 | /// distribution between zero and unity. Utilizes the Box-Muller |
---|
440 | /// algorithm, which needs two calls to random generator. |
---|
441 | /// |
---|
442 | /// Distribution function \f$ f(x) = |
---|
443 | /// \frac{1}{\sqrt{2\pi\sigma^2}}\exp(-\frac{(x-\mu)^2}{2\sigma^2}) |
---|
444 | /// \f$ \n |
---|
445 | /// Expectation value: \f$ \mu \f$ \n |
---|
446 | /// Variance: \f$ \sigma^2 \f$ |
---|
447 | /// |
---|
448 | class Gaussian : public Continuous |
---|
449 | { |
---|
450 | public: |
---|
451 | /// |
---|
452 | /// @brief Constructor |
---|
453 | /// @param s is the standard deviation \f$ \sigma \f$ of distribution |
---|
454 | /// m is the expectation value \f$ \mu \f$ of the distribution |
---|
455 | /// |
---|
456 | inline Gaussian(const double s=1, const double m=0) : m_(m), s_(s) {} |
---|
457 | |
---|
458 | /// |
---|
459 | /// @return A random Gaussian number |
---|
460 | /// |
---|
461 | inline double |
---|
462 | operator()(void) const { return gsl_ran_gaussian(rng_->rng(), s_)+m_; } |
---|
463 | |
---|
464 | /// |
---|
465 | /// @return A random Gaussian number with standard deviation \a s |
---|
466 | /// and expectation value 0. |
---|
467 | /// @note this operator ignores parameters given in Constructor |
---|
468 | /// |
---|
469 | inline double |
---|
470 | operator()(const double s) const { return gsl_ran_gaussian(rng_->rng(), s); } |
---|
471 | |
---|
472 | /// |
---|
473 | /// @return A random Gaussian number with standard deviation \a s |
---|
474 | /// and expectation value \a m. |
---|
475 | /// @note this operator ignores parameters given in Constructor |
---|
476 | /// |
---|
477 | inline double operator()(const double s, const double m) const |
---|
478 | { return gsl_ran_gaussian(rng_->rng(), s)+m; } |
---|
479 | |
---|
480 | private: |
---|
481 | double m_; |
---|
482 | double s_; |
---|
483 | }; |
---|
484 | |
---|
485 | |
---|
486 | }} // of namespace random and namespace theplu |
---|
487 | |
---|
488 | #endif |
---|