pytutorial/numpy_random/README.md

Numpy -- Random numbers the non-legacy way

Goal

If you don't see something like np.random.default_rng() in your code then you are probably using the old [Legacy Random Generation] (https://numpy.org/doc/stable/reference/random/legacy.html#legacy-random-generation).

Don't use the legacy methods for new source code!!!

Do it like this:

import numpy as np

rng = np.random.default_rng()
random_values = rng.random(size=(2, 10))

Questions to David Rotermund

Random Generator

Typical usage

import numpy as np

rng = np.random.default_rng()
random_values = rng.random(size=(2, 10))

print(random_values)

Output:

[[0.81103943 0.1110591  0.42978062 0.47818377 0.91138636 0.47051031
  0.08662082 0.1643707  0.48717037 0.17870536]
 [0.94499902 0.74089677 0.12221184 0.61603001 0.91198789 0.33900609
  0.75832792 0.74465679 0.19940125 0.56674595]]

With seed:

import numpy as np

rng = np.random.default_rng(seed=23)
random_values = rng.random(size=(2, 10))

print(random_values)

Output:

[[0.69393308 0.64145822 0.12864422 0.11370805 0.65334552 0.85345711
  0.20177913 0.21801864 0.71658464 0.47069967]
 [0.41522193 0.3491478  0.06385375 0.45466617 0.30145328 0.38907675
  0.54029782 0.68358969 0.62475238 0.74270445]]

Changing the random number generator

Default

import numpy as np

rng = np.random.default_rng()
print(rng)  # -> Generator(PCG64)

If you don't like it there are other options:

MT19937
PCG64
PCG64DXSM
Philox
SFC64

Distributions (you will use)

The most important ones are in bold. If you see a function argument out, then you can reuse an existing np array (i.e. in-place operation) as target.

integers(low[, high, size, dtype, endpoint])	Return random integers from low (inclusive) to high (exclusive), or if endpoint=True, low (inclusive) to high (inclusive).
random([size, dtype, out])	Return random floats in the half-open interval [0.0, 1.0).
choice(a[, size, replace, p, axis, shuffle])	Generates a random sample from a given array
bytes(length)	Return random bytes.
binomial(n, p[, size])	Draw samples from a binomial distribution.
multinomial(n, pvals[, size])	Draw samples from a multinomial distribution.
multivariate_normal(mean, cov[, size, ...])	Draw random samples from a multivariate normal distribution.
normal([loc, scale, size])	Draw random samples from a normal (Gaussian) distribution.
poisson([lam, size])	Draw samples from a Poisson distribution.
standard_normal([size, dtype, out])	Draw samples from a standard Normal distribution (mean=0, stdev=1).
uniform([low, high, size])	Draw samples from a uniform distribution.

random

import numpy as np

rng = np.random.default_rng()
random_values = rng.random(size=(2, 10))
print(random_values)

Output:

[[0.75309105 0.15751286 0.49454759 0.18204807 0.88459006 0.78685769
  0.68525047 0.4000365  0.45317167 0.62412358]
 [0.01082224 0.13257961 0.75638974 0.84886965 0.19755022 0.18697649
  0.47064409 0.66128207 0.30285691 0.53465021]]

integers

import numpy as np

rng = np.random.default_rng()
random_values = rng.integers(
    low=1, high=3, size=(2, 10), dtype=np.uint64, endpoint=True
)
print(random_values)

Output:

[[2 3 3 2 1 3 1 1 2 2]
 [3 3 2 3 3 2 3 3 1 3]]