blob: f0c4add61a19c0f2868565207c5740375fb9d286 [file] [log] [blame]
Georg Brandl116aa622007-08-15 14:28:22 +00001:mod:`random` --- Generate pseudo-random numbers
2================================================
3
4.. module:: random
5 :synopsis: Generate pseudo-random numbers with various common distributions.
6
Raymond Hettinger10480942011-01-10 03:26:08 +00007**Source code:** :source:`Lib/random.py`
Georg Brandl116aa622007-08-15 14:28:22 +00008
Raymond Hettinger4f707fd2011-01-10 19:54:11 +00009--------------
10
Georg Brandl116aa622007-08-15 14:28:22 +000011This module implements pseudo-random number generators for various
12distributions.
13
Raymond Hettingerb21dac12010-09-07 05:32:49 +000014For integers, there is uniform selection from a range. For sequences, there is
15uniform selection of a random element, a function to generate a random
16permutation of a list in-place, and a function for random sampling without
17replacement.
Georg Brandl116aa622007-08-15 14:28:22 +000018
19On the real line, there are functions to compute uniform, normal (Gaussian),
20lognormal, negative exponential, gamma, and beta distributions. For generating
21distributions of angles, the von Mises distribution is available.
22
23Almost all module functions depend on the basic function :func:`random`, which
24generates a random float uniformly in the semi-open range [0.0, 1.0). Python
25uses the Mersenne Twister as the core generator. It produces 53-bit precision
26floats and has a period of 2\*\*19937-1. The underlying implementation in C is
27both fast and threadsafe. The Mersenne Twister is one of the most extensively
28tested random number generators in existence. However, being completely
29deterministic, it is not suitable for all purposes, and is completely unsuitable
30for cryptographic purposes.
31
32The functions supplied by this module are actually bound methods of a hidden
33instance of the :class:`random.Random` class. You can instantiate your own
Raymond Hettinger28de64f2008-01-13 23:40:30 +000034instances of :class:`Random` to get generators that don't share state.
Georg Brandl116aa622007-08-15 14:28:22 +000035
36Class :class:`Random` can also be subclassed if you want to use a different
37basic generator of your own devising: in that case, override the :meth:`random`,
Raymond Hettingerafd30452009-02-24 10:57:02 +000038:meth:`seed`, :meth:`getstate`, and :meth:`setstate` methods.
Benjamin Petersond18de0e2008-07-31 20:21:46 +000039Optionally, a new generator can supply a :meth:`getrandbits` method --- this
Georg Brandl116aa622007-08-15 14:28:22 +000040allows :meth:`randrange` to produce selections over an arbitrarily large range.
41
Benjamin Peterson21896a32010-03-21 22:03:03 +000042The :mod:`random` module also provides the :class:`SystemRandom` class which
43uses the system function :func:`os.urandom` to generate random numbers
44from sources provided by the operating system.
Georg Brandl116aa622007-08-15 14:28:22 +000045
Georg Brandl116aa622007-08-15 14:28:22 +000046
Raymond Hettinger10480942011-01-10 03:26:08 +000047Bookkeeping functions:
Georg Brandl116aa622007-08-15 14:28:22 +000048
Raymond Hettingerf763a722010-09-07 00:38:15 +000049.. function:: seed([x], version=2)
Georg Brandl116aa622007-08-15 14:28:22 +000050
Raymond Hettingerf763a722010-09-07 00:38:15 +000051 Initialize the random number generator.
Georg Brandl116aa622007-08-15 14:28:22 +000052
Raymond Hettingerf763a722010-09-07 00:38:15 +000053 If *x* is omitted or ``None``, the current system time is used. If
54 randomness sources are provided by the operating system, they are used
55 instead of the system time (see the :func:`os.urandom` function for details
56 on availability).
Georg Brandl116aa622007-08-15 14:28:22 +000057
Raymond Hettingerf763a722010-09-07 00:38:15 +000058 If *x* is an int, it is used directly.
59
60 With version 2 (the default), a :class:`str`, :class:`bytes`, or :class:`bytearray`
Raymond Hettinger3149f9c2010-09-07 05:35:10 +000061 object gets converted to an :class:`int` and all of its bits are used. With version 1,
Raymond Hettingerf763a722010-09-07 00:38:15 +000062 the :func:`hash` of *x* is used instead.
63
64 .. versionchanged:: 3.2
65 Moved to the version 2 scheme which uses all of the bits in a string seed.
Georg Brandl116aa622007-08-15 14:28:22 +000066
67.. function:: getstate()
68
69 Return an object capturing the current internal state of the generator. This
70 object can be passed to :func:`setstate` to restore the state.
71
Georg Brandl116aa622007-08-15 14:28:22 +000072
73.. function:: setstate(state)
74
75 *state* should have been obtained from a previous call to :func:`getstate`, and
76 :func:`setstate` restores the internal state of the generator to what it was at
77 the time :func:`setstate` was called.
78
Georg Brandl116aa622007-08-15 14:28:22 +000079
Georg Brandl116aa622007-08-15 14:28:22 +000080.. function:: getrandbits(k)
81
Ezio Melotti0639d5a2009-12-19 23:26:38 +000082 Returns a Python integer with *k* random bits. This method is supplied with
Georg Brandl5c106642007-11-29 17:41:05 +000083 the MersenneTwister generator and some other generators may also provide it
Georg Brandl116aa622007-08-15 14:28:22 +000084 as an optional part of the API. When available, :meth:`getrandbits` enables
85 :meth:`randrange` to handle arbitrarily large ranges.
86
Georg Brandl116aa622007-08-15 14:28:22 +000087
88Functions for integers:
89
Georg Brandl116aa622007-08-15 14:28:22 +000090.. function:: randrange([start,] stop[, step])
91
92 Return a randomly selected element from ``range(start, stop, step)``. This is
93 equivalent to ``choice(range(start, stop, step))``, but doesn't actually build a
94 range object.
95
Raymond Hettinger05156612010-09-07 04:44:52 +000096 The positional argument pattern matches that of :func:`range`. Keyword arguments
97 should not be used because the function may use them in unexpected ways.
98
99 .. versionchanged:: 3.2
100 :meth:`randrange` is more sophisticated about producing equally distributed
101 values. Formerly it used a style like ``int(random()*n)`` which could produce
102 slightly uneven distributions.
Georg Brandl116aa622007-08-15 14:28:22 +0000103
104.. function:: randint(a, b)
105
Raymond Hettingerafd30452009-02-24 10:57:02 +0000106 Return a random integer *N* such that ``a <= N <= b``. Alias for
107 ``randrange(a, b+1)``.
Georg Brandl116aa622007-08-15 14:28:22 +0000108
Georg Brandl116aa622007-08-15 14:28:22 +0000109
Georg Brandl55ac8f02007-09-01 13:51:09 +0000110Functions for sequences:
Georg Brandl116aa622007-08-15 14:28:22 +0000111
112.. function:: choice(seq)
113
114 Return a random element from the non-empty sequence *seq*. If *seq* is empty,
115 raises :exc:`IndexError`.
116
117
118.. function:: shuffle(x[, random])
119
120 Shuffle the sequence *x* in place. The optional argument *random* is a
121 0-argument function returning a random float in [0.0, 1.0); by default, this is
122 the function :func:`random`.
123
124 Note that for even rather small ``len(x)``, the total number of permutations of
125 *x* is larger than the period of most random number generators; this implies
126 that most permutations of a long sequence can never be generated.
127
128
129.. function:: sample(population, k)
130
Raymond Hettinger1acde192008-01-14 01:00:53 +0000131 Return a *k* length list of unique elements chosen from the population sequence
132 or set. Used for random sampling without replacement.
Georg Brandl116aa622007-08-15 14:28:22 +0000133
Georg Brandl116aa622007-08-15 14:28:22 +0000134 Returns a new list containing elements from the population while leaving the
135 original population unchanged. The resulting list is in selection order so that
136 all sub-slices will also be valid random samples. This allows raffle winners
137 (the sample) to be partitioned into grand prize and second place winners (the
138 subslices).
139
Guido van Rossum2cc30da2007-11-02 23:46:40 +0000140 Members of the population need not be :term:`hashable` or unique. If the population
Georg Brandl116aa622007-08-15 14:28:22 +0000141 contains repeats, then each occurrence is a possible selection in the sample.
142
143 To choose a sample from a range of integers, use an :func:`range` object as an
144 argument. This is especially fast and space efficient for sampling from a large
145 population: ``sample(range(10000000), 60)``.
146
147The following functions generate specific real-valued distributions. Function
148parameters are named after the corresponding variables in the distribution's
149equation, as used in common mathematical practice; most of these equations can
150be found in any statistics text.
151
152
153.. function:: random()
154
155 Return the next random floating point number in the range [0.0, 1.0).
156
157
158.. function:: uniform(a, b)
159
Benjamin Petersonb58dda72009-01-18 22:27:04 +0000160 Return a random floating point number *N* such that ``a <= N <= b`` for
161 ``a <= b`` and ``b <= N <= a`` for ``b < a``.
Georg Brandl116aa622007-08-15 14:28:22 +0000162
Raymond Hettingerbe40db02009-06-11 23:12:14 +0000163 The end-point value ``b`` may or may not be included in the range
164 depending on floating-point rounding in the equation ``a + (b-a) * random()``.
Benjamin Peterson35e8c462008-04-24 02:34:53 +0000165
Christian Heimesfe337bf2008-03-23 21:54:12 +0000166.. function:: triangular(low, high, mode)
167
Benjamin Petersonb58dda72009-01-18 22:27:04 +0000168 Return a random floating point number *N* such that ``low <= N <= high`` and
Christian Heimescc47b052008-03-25 14:56:36 +0000169 with the specified *mode* between those bounds. The *low* and *high* bounds
170 default to zero and one. The *mode* argument defaults to the midpoint
171 between the bounds, giving a symmetric distribution.
Christian Heimesfe337bf2008-03-23 21:54:12 +0000172
Georg Brandl116aa622007-08-15 14:28:22 +0000173
174.. function:: betavariate(alpha, beta)
175
Benjamin Petersonb58dda72009-01-18 22:27:04 +0000176 Beta distribution. Conditions on the parameters are ``alpha > 0`` and
177 ``beta > 0``. Returned values range between 0 and 1.
Georg Brandl116aa622007-08-15 14:28:22 +0000178
179
180.. function:: expovariate(lambd)
181
Mark Dickinson2f947362009-01-07 17:54:07 +0000182 Exponential distribution. *lambd* is 1.0 divided by the desired
183 mean. It should be nonzero. (The parameter would be called
184 "lambda", but that is a reserved word in Python.) Returned values
185 range from 0 to positive infinity if *lambd* is positive, and from
186 negative infinity to 0 if *lambd* is negative.
Georg Brandl116aa622007-08-15 14:28:22 +0000187
188
189.. function:: gammavariate(alpha, beta)
190
Benjamin Petersonb58dda72009-01-18 22:27:04 +0000191 Gamma distribution. (*Not* the gamma function!) Conditions on the
192 parameters are ``alpha > 0`` and ``beta > 0``.
Georg Brandl116aa622007-08-15 14:28:22 +0000193
194
195.. function:: gauss(mu, sigma)
196
Benjamin Petersonb58dda72009-01-18 22:27:04 +0000197 Gaussian distribution. *mu* is the mean, and *sigma* is the standard
198 deviation. This is slightly faster than the :func:`normalvariate` function
199 defined below.
Georg Brandl116aa622007-08-15 14:28:22 +0000200
201
202.. function:: lognormvariate(mu, sigma)
203
204 Log normal distribution. If you take the natural logarithm of this
205 distribution, you'll get a normal distribution with mean *mu* and standard
206 deviation *sigma*. *mu* can have any value, and *sigma* must be greater than
207 zero.
208
209
210.. function:: normalvariate(mu, sigma)
211
212 Normal distribution. *mu* is the mean, and *sigma* is the standard deviation.
213
214
215.. function:: vonmisesvariate(mu, kappa)
216
217 *mu* is the mean angle, expressed in radians between 0 and 2\*\ *pi*, and *kappa*
218 is the concentration parameter, which must be greater than or equal to zero. If
219 *kappa* is equal to zero, this distribution reduces to a uniform random angle
220 over the range 0 to 2\*\ *pi*.
221
222
223.. function:: paretovariate(alpha)
224
225 Pareto distribution. *alpha* is the shape parameter.
226
227
228.. function:: weibullvariate(alpha, beta)
229
230 Weibull distribution. *alpha* is the scale parameter and *beta* is the shape
231 parameter.
232
233
Raymond Hettinger3cdf8712010-12-02 05:35:35 +0000234Alternative Generator:
Georg Brandl116aa622007-08-15 14:28:22 +0000235
Georg Brandl116aa622007-08-15 14:28:22 +0000236.. class:: SystemRandom([seed])
237
238 Class that uses the :func:`os.urandom` function for generating random numbers
239 from sources provided by the operating system. Not available on all systems.
Raymond Hettinger3cdf8712010-12-02 05:35:35 +0000240 Does not rely on software state, and sequences are not reproducible. Accordingly,
Raymond Hettingerafd30452009-02-24 10:57:02 +0000241 the :meth:`seed` method has no effect and is ignored.
Georg Brandl116aa622007-08-15 14:28:22 +0000242 The :meth:`getstate` and :meth:`setstate` methods raise
243 :exc:`NotImplementedError` if called.
244
Georg Brandl116aa622007-08-15 14:28:22 +0000245
Georg Brandl116aa622007-08-15 14:28:22 +0000246.. seealso::
247
248 M. Matsumoto and T. Nishimura, "Mersenne Twister: A 623-dimensionally
249 equidistributed uniform pseudorandom number generator", ACM Transactions on
250 Modeling and Computer Simulation Vol. 8, No. 1, January pp.3-30 1998.
251
Georg Brandl116aa622007-08-15 14:28:22 +0000252
Raymond Hettinger1fd32a62009-04-01 20:52:13 +0000253 `Complementary-Multiply-with-Carry recipe
Raymond Hettinger9743fd02009-04-03 05:47:33 +0000254 <http://code.activestate.com/recipes/576707/>`_ for a compatible alternative
255 random number generator with a long period and comparatively simple update
Raymond Hettinger1fd32a62009-04-01 20:52:13 +0000256 operations.
Raymond Hettinger435cb0f2010-09-06 23:36:31 +0000257
Raymond Hettinger3cdf8712010-12-02 05:35:35 +0000258
Raymond Hettinger435cb0f2010-09-06 23:36:31 +0000259Notes on Reproducibility
Antoine Pitroue72b5862010-12-12 20:13:31 +0000260------------------------
Raymond Hettinger435cb0f2010-09-06 23:36:31 +0000261
262Sometimes it is useful to be able to reproduce the sequences given by a pseudo
263random number generator. By re-using a seed value, the same sequence should be
264reproducible from run to run as long as multiple threads are not running.
265
266Most of the random module's algorithms and seeding functions are subject to
267change across Python versions, but two aspects are guaranteed not to change:
268
269* If a new seeding method is added, then a backward compatible seeder will be
270 offered.
271
272* The generator's :meth:`random` method will continue to produce the same
273 sequence when the compatible seeder is given the same seed.
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000274
Raymond Hettinger6e353942010-12-04 23:42:12 +0000275.. _random-examples:
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000276
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000277Examples and Recipes
Antoine Pitroue72b5862010-12-12 20:13:31 +0000278--------------------
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000279
Raymond Hettinger3cdf8712010-12-02 05:35:35 +0000280Basic usage::
281
282 >>> random.random() # Random float x, 0.0 <= x < 1.0
283 0.37444887175646646
284
285 >>> random.uniform(1, 10) # Random float x, 1.0 <= x < 10.0
286 1.1800146073117523
287
288 >>> random.randrange(10) # Integer from 0 to 9
289 7
290
291 >>> random.randrange(0, 101, 2) # Even integer from 0 to 100
292 26
293
294 >>> random.choice('abcdefghij') # Single random element
295 'c'
296
297 >>> items = [1, 2, 3, 4, 5, 6, 7]
298 >>> random.shuffle(items)
299 >>> items
300 [7, 3, 2, 5, 6, 4, 1]
301
302 >>> random.sample([1, 2, 3, 4, 5], 3) # Three samples without replacement
303 [4, 1, 5]
304
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000305A common task is to make a :func:`random.choice` with weighted probababilites.
306
307If the weights are small integer ratios, a simple technique is to build a sample
308population with repeats::
309
310 >>> weighted_choices = [('Red', 3), ('Blue', 2), ('Yellow', 1), ('Green', 4)]
311 >>> population = [val for val, cnt in weighted_choices for i in range(cnt)]
312 >>> random.choice(population)
313 'Green'
314
Raymond Hettinger3cdf8712010-12-02 05:35:35 +0000315A more general approach is to arrange the weights in a cumulative distribution
316with :func:`itertools.accumulate`, and then locate the random value with
317:func:`bisect.bisect`::
Raymond Hettinger2fdc7b12010-12-02 02:41:33 +0000318
319 >>> choices, weights = zip(*weighted_choices)
320 >>> cumdist = list(itertools.accumulate(weights))
321 >>> x = random.random() * cumdist[-1]
322 >>> choices[bisect.bisect(cumdist, x)]
323 'Blue'