pocketpy/docs/modules/random.md

---

icon: package

label: random

random

Pseudo-random numbers from a Mersenne Twister generator. Module-level functions use a shared generator in the current VM; Random creates independent state. This generator is for simulations and games, not cryptographic secrets.

API	Behavior
seed(value)	Seed with an integer, or None for the clock. Integer seeds use the low 32 bits.
random()	Float in [0.0, 1.0).
randint(a, b)	Integer including both endpoints; require a <= b.
uniform(a, b)	Float between the endpoints; floating-point rounding can affect the boundary.
choice(sequence)	One item from a nonempty list, tuple, or string.
shuffle(items)	Shuffle a list in place; return None.
choices(population, weights=None, k=1)	List of k selections with replacement; population and weights are lists or tuples.
getstate(), setstate(state)	Save/restore the generator's state as bytes.
Random(value=None)	New generator, optionally initialized with an integer seed or saved state bytes.

For weighted choices, supply finite nonnegative weights with matching length and total greater than 1e-6; use a nonnegative k. Unseeded generators initialize from the clock on first use.

Repeatable independent draws

from random import Random

rng = Random(7)
state = rng.getstate()
rolls = [rng.randint(1, 6) for _ in range(5)]
rng.setstate(state)
assert rolls == [rng.randint(1, 6) for _ in range(5)]

loot = rng.choices(['coin', 'gem'], weights=[9, 1], k=3)
assert len(loot) == 3
assert all([item in ['coin', 'gem'] for item in loot])

State bytes use pocketpy's internal representation, not CPython's state tuple. Do not assume sequences or saved states are interchangeable with CPython or across future runtime versions.

A generator can also be pickled:

import pickle
from random import Random

original = Random(7)
restored = pickle.loads(pickle.dumps(original))
assert original.random() == restored.random()