2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2008, The GROMACS development team.
6 * Copyright (c) 2010,2014, by the GROMACS development team, led by
7 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
8 * and including many others, as listed in the AUTHORS file in the
9 * top-level source directory and at http://www.gromacs.org.
11 * GROMACS is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public License
13 * as published by the Free Software Foundation; either version 2.1
14 * of the License, or (at your option) any later version.
16 * GROMACS is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with GROMACS; if not, see
23 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 * If you want to redistribute modifications to GROMACS, please
27 * consider that scientific software is very special. Version
28 * control is crucial - bugs must be traceable. We will be happy to
29 * consider code for inclusion in the official distribution, but
30 * derived work must not be called official GROMACS. Details are found
31 * in the README & COPYING files - if they are missing, get the
32 * official version at http://www.gromacs.org.
34 * To help us fund GROMACS development, we humbly ask that you cite
35 * the research papers on the package. Check out http://www.gromacs.org.
38 #ifndef _GMX_RANDOM_H_
39 #define _GMX_RANDOM_H_
42 #include "types/simple.h"
48 /*! Fixed random number seeds for different types of RNG */
49 #define RND_SEED_UPDATE 1 /*!< For coordinate update (sd, bd, ..) */
50 #define RND_SEED_REPLEX 2 /*!< For replica exchange */
51 #define RND_SEED_VRESCALE 3 /*!< For V-rescale thermostat */
52 #define RND_SEED_ANDERSEN 4 /*!< For Andersen thermostat */
53 #define RND_SEED_TPI 5 /*!< For test particle insertion */
54 #define RND_SEED_EXPANDED 6 /*!< For expanded emseble methods */
56 /*! \brief Abstract datatype for a random number generator
58 * This is a handle to the full state of a random number generator.
59 * You can not access anything inside the gmx_rng structure outside this
62 typedef struct gmx_rng *
66 /*! \brief Returns the size of the RNG integer data structure
68 * Returns the size of the RNG integer data structure.
75 /*! \brief Create a new RNG, seeded from a single integer.
77 * If you dont want to pick a seed, just call it as
78 * rng=gmx_rng_init(gmx_rng_make_seed()) to seed it from
79 * the system time or a random device.
81 * \param seed Random seed, unsigned 32-bit integer.
83 * \return Reference to a random number generator, or NULL if there was an
89 gmx_rng_init(unsigned int seed);
92 /*! \brief Generate a 'random' RNG seed.
94 * This routine tries to get a seed from /dev/random if present,
95 * and if not it uses time-of-day and process id to generate one.
97 * \return 32-bit unsigned integer random seed.
99 * Tip: If you use this in your code, it is a good idea to write the
100 * returned random seed to a logfile, so you can recreate the exact sequence
101 * of random number if you need to reproduce your run later for one reason
107 gmx_rng_make_seed(void);
110 /*! \brief Initialize a RNG with 624 integers (>32 bits of entropy).
112 * The Mersenne twister RNG used in Gromacs has an extremely long period,
113 * but when you only initialize it with a 32-bit integer there are only
114 * 2^32 different possible sequences of number - much less than the generator
117 * If you really need the full entropy, this routine makes it possible to
118 * initialize the RNG with up to 624 32-bit integers, which will give you
119 * up to 2^19968 bits of entropy.
121 * \param seed Array of unsigned integers to form a seed
122 * \param seed_length Number of integers in the array, up to 624 are used.
124 * \return Reference to a random number generator, or NULL if there was an
130 gmx_rng_init_array(unsigned int seed[],
134 /*! \brief Release resources of a RNG
136 * This routine destroys a random number generator and releases all
137 * resources allocated by it.
139 * \param rng Handle to random number generator previously returned by
140 * gmx_rng_init() or gmx_rng_init_array().
142 * \threadsafe Function itself is threadsafe, but you should only destroy a
143 * certain RNG once (i.e. from one thread).
146 gmx_rng_destroy(gmx_rng_t rng);
149 /*! \brief Get the state of a RNG
151 * This routine stores the random state in mt and mti, mt should have
152 * a size of at least 624, mt of 1.
154 * \param rng Handle to random number generator previously returned by
155 * gmx_rng_init() or gmx_rng_init_array().
158 gmx_rng_get_state(gmx_rng_t rng, unsigned int *mt, int *mti);
161 /*! \brief Set the state of a RNG
163 * This routine sets the random state from mt and mti, mt should have
164 * a size of at least 624.
166 * \param rng Handle to random number generator previously returned by
167 * gmx_rng_init() or gmx_rng_init_array().
170 gmx_rng_set_state(gmx_rng_t rng, unsigned int *mt, int mti);
173 /*! \brief Random 32-bit integer from a uniform distribution
175 * This routine returns a random integer from the random number generator
176 * provided, and updates the state of that RNG.
178 * \param rng Handle to random number generator previously returned by
179 * gmx_rng_init() or gmx_rng_init_array().
181 * \return 32-bit unsigned integer from a uniform distribution.
183 * \threadsafe Function yes, input data no. You should not call this function
184 * from two different threads using the same RNG handle at the
185 * same time. For performance reasons we cannot lock the handle
186 * with a mutex every time we need a random number - that would
187 * slow the routine down a factor 2-5. There are two simple
188 * solutions: either use a mutex and lock it before calling
189 * the function, or use a separate RNG handle for each thread.
192 gmx_rng_uniform_uint32(gmx_rng_t rng);
195 /*! \brief Random gmx_real_t 0<=x<1 from a uniform distribution
197 * This routine returns a random floating-point number from the
198 * random number generator provided, and updates the state of that RNG.
200 * \param rng Handle to random number generator previously returned by
201 * gmx_rng_init() or gmx_rng_init_array().
203 * \return floating-point number 0<=x<1 from a uniform distribution.
205 * \threadsafe Function yes, input data no. You should not call this function
206 * from two different threads using the same RNG handle at the
207 * same time. For performance reasons we cannot lock the handle
208 * with a mutex every time we need a random number - that would
209 * slow the routine down a factor 2-5. There are two simple
210 * solutions: either use a mutex and lock it before calling
211 * the function, or use a separate RNG handle for each thread.
214 gmx_rng_uniform_real(gmx_rng_t rng);
217 /*! \brief Random gmx_real_t from a gaussian distribution
219 * This routine returns a random floating-point number from the
220 * random number generator provided, and updates the state of that RNG.
222 * The Box-Muller algorithm is used to provide gaussian random numbers. This
223 * is not the fastest known algorithm for gaussian numbers, but in contrast
224 * to the alternatives it is very well studied and you can trust the returned
225 * random numbers to have good properties and no correlations.
227 * \param rng Handle to random number generator previously returned by
228 * gmx_rng_init() or gmx_rng_init_array().
230 * \return Gaussian random floating-point number with average 0.0 and
231 * standard deviation 1.0. You can get any average/mean you want
232 * by first multiplying with the desired average and then adding
233 * the average you want.
235 * \threadsafe Function yes, input data no. You should not call this function
236 * from two different threads using the same RNG handle at the
237 * same time. For performance reasons we cannot lock the handle
238 * with a mutex every time we need a random number - that would
239 * slow the routine down a factor 2-5. There are two simple
240 * solutions: either use a mutex and lock it before calling
241 * the function, or use a separate RNG handle for each thread.
243 * It works perfectly to mix calls to get uniform and gaussian random numbers
244 * from the same generator, but since it will affect the sequence of returned
245 * numbers it is probably better to use separate random number generator
249 gmx_rng_gaussian_real(gmx_rng_t rng);
253 /* Return a new gaussian random number with expectation value
254 * 0.0 and standard deviation 1.0. This routine uses a table
255 * lookup for maximum speed.
257 * WARNING: The lookup table is 16k by default, which means
258 * the granularity of the random numbers is coarser
259 * than what you get from gmx_rng_gauss_real().
260 * In most cases this is no problem whatsoever,
261 * and it is particularly true for BD/SD integration.
262 * Note that you will NEVER get any really extreme
263 * numbers: the maximum absolute value returned is
269 gmx_rng_gaussian_table(gmx_rng_t rng);
272 /* The stateless cycle based random number generators below,
273 * which all use threefry2x64, take the following arguments:
275 * ctr1: In mdrun the step counter, in tools the frame(-step)
276 * counter, so we can ensure reproducible results, even
277 * we starting at different steps/frames. Might need to be
278 * multiplied by a constant if we need more random numbers.
279 * ctr2: A local counter, in mdrun often a global atom index.
280 * If any algorithm needs a variable number of random numbers,
281 * the second counter is usually a function of the local
283 * key1: A user provided random seed.
284 * key2: A fixed seed which is particular for the algorithm,
285 * as defined at the top of this file, to ensure different
286 * random sequences when the same user seed is used for
287 * different algorithms.
290 /* Return two uniform random numbers with 2^53 equally
291 * probable values between 0 and 1 - 2^-53.
292 * It uses a stateless counter based random number generator
296 gmx_rng_cycle_2uniform(gmx_int64_t ctr1, gmx_int64_t ctr2,
297 gmx_int64_t key1, gmx_int64_t key2,
300 /* Return three Gaussian random numbers with expectation value
301 * 0.0 and standard deviation 1.0. This routine uses a table
302 * lookup for maximum speed. It uses a stateless counter
303 * based random number generator (threefry2x64). See
304 * gmx_rng_gaussian_table for a warning about accuracy of the table.
309 gmx_rng_cycle_3gaussian_table(gmx_int64_t ctr1, gmx_int64_t ctr2,
310 gmx_int64_t key1, gmx_int64_t key2,
313 /* As gmx_rng_3gaussian_table, but returns 6 Gaussian numbers. */
315 gmx_rng_cycle_6gaussian_table(gmx_int64_t ctr1, gmx_int64_t ctr2,
316 gmx_int64_t key1, gmx_int64_t key2,
323 #endif /* _GMX_RANDOM_H_ */