Random Sampling (cupy.random)

Differences between cupy.random and numpy.random:

• CuPy provides Legacy Random Generation API (see also: NumPy 1.16 Reference). The new random generator API (numpy.random.Generator class) introduced in NumPy 1.17 has not been implemented yet.

• Most functions under cupy.random support the dtype option, which do not exist in the corresponding NumPy APIs. This option enables generation of float32 values directly without any space overhead.

• CuPy does not guarantee that the same number generator is used across major versions. This means that numbers generated by cupy.random by new major version may not be the same as the previous one, even if the same seed and distribution are used.

Simple random data

cupy.random.rand	Returns an array of uniform random values over the interval [0, 1).
cupy.random.randn	Returns an array of standard normal random values.
cupy.random.randint	Returns a scalar or an array of integer values over [low, high).
cupy.random.random_integers	Return a scalar or an array of integer values over [low, high]
cupy.random.random_sample	Returns an array of random values over the interval [0, 1).
cupy.random.random	Returns an array of random values over the interval [0, 1).
cupy.random.ranf	Returns an array of random values over the interval [0, 1).
cupy.random.sample	Returns an array of random values over the interval [0, 1).
cupy.random.choice	Returns an array of random values from a given 1-D array.
cupy.random.bytes	Returns random bytes.

Permutations

cupy.random.shuffle	Shuffles an array.
cupy.random.permutation	Returns a permuted range or a permutation of an array.

Distributions

cupy.random.beta	Beta distribution.
cupy.random.binomial	Binomial distribution.
cupy.random.chisquare	Chi-square distribution.
cupy.random.dirichlet	Dirichlet distribution.
cupy.random.exponential	Exponential distribution.
cupy.random.f	F distribution.
cupy.random.gamma	Gamma distribution.
cupy.random.geometric	Geometric distribution.
cupy.random.gumbel	Returns an array of samples drawn from a Gumbel distribution.
cupy.random.hypergeometric	hypergeometric distribution.
cupy.random.laplace	Laplace distribution.
cupy.random.logistic	Logistic distribution.
cupy.random.lognormal	Returns an array of samples drawn from a log normal distribution.
cupy.random.logseries	Log series distribution.
cupy.random.multinomial	Returns an array from multinomial distribution.
cupy.random.multivariate_normal	Multivariate normal distribution.
cupy.random.negative_binomial	Negative binomial distribution.
cupy.random.noncentral_chisquare	Noncentral chisquare distribution.
cupy.random.noncentral_f	Noncentral F distribution.
cupy.random.normal	Returns an array of normally distributed samples.
cupy.random.pareto	Pareto II or Lomax distribution.
cupy.random.poisson	Poisson distribution.
cupy.random.power	Power distribution.
cupy.random.rayleigh	Rayleigh distribution.
cupy.random.standard_cauchy	Standard cauchy distribution.
cupy.random.standard_exponential	Standard exponential distribution.
cupy.random.standard_gamma	Standard gamma distribution.
cupy.random.standard_normal	Returns an array of samples drawn from the standard normal distribution.
cupy.random.standard_t	Standard Student’s t distribution.
cupy.random.triangular	Triangular distribution.
cupy.random.uniform	Returns an array of uniformly-distributed samples over an interval.
cupy.random.vonmises	von Mises distribution.
cupy.random.wald	Wald distribution.
cupy.random.weibull	weibull distribution.
cupy.random.zipf	Zipf distribution.

Random generator

cupy.random.RandomState	Portable container of a pseudo-random number generator.
cupy.random.seed	Resets the state of the random number generator with a seed.
cupy.random.get_random_state	Gets the state of the random number generator for the current device.
cupy.random.set_random_state	Sets the state of the random number generator for the current device.

Note

CuPy does not provide cupy.random.get_state nor cupy.random.set_state at this time. Use cupy.random.get_random_state() and cupy.random.set_random_state() instead. Note that these functions use cupy.random.RandomState instance to represent the internal state, which cannot be serialized.