Generate seeded random numbers

Edit on Github

Math.random gives you no way to set a seed, so test runs and simulations that need reproducible randomness cannot use it. The @std/random module provides a seeded pseudo-random number generator (PCG32) plus helpers for ranges, sampling, and shuffling that accept any generator. Do not use seeded generators for security; for tokens and keys, see the secure random values example.

import {
  randomIntegerBetween,
  randomSeeded,
  sample,
  shuffle,
} from "jsr:@std/random";

randomSeeded creates a generator from a bigint seed. It is a drop-in replacement for Math.random: a function returning floats in [0, 1).

const prng = randomSeeded(42n);

console.log(prng()); // 0.7916327826678753
console.log(prng()); // 0.7799280444160104

The same seed always produces the same sequence, which is the point: a failing test that used seed 42 can be replayed exactly.

const replay = randomSeeded(42n);
console.log(replay()); // 0.7916327826678753

The helpers take the generator through the prng option. Without it they fall back to Math.random.

console.log(randomIntegerBetween(1, 6, { prng })); // 1

sample picks a random element, shuffle returns a new shuffled array.

const deck = ["ace", "king", "queen", "jack"];
console.log(sample(deck, { prng })); // jack
console.log(shuffle(deck, { prng })); // [ "ace", "queen", "jack", "king" ]

For seeded bytes, getRandomValuesSeeded builds the seeded counterpart of crypto.getRandomValues: it takes a seed and returns a function that fills typed arrays.

import { getRandomValuesSeeded } from "jsr:@std/random";
const getRandomValues = getRandomValuesSeeded(42n);
console.log(getRandomValues(new Uint8Array(4)));

Run this example locally using the Deno CLI:

deno run https://docs.deno.com/examples/scripts/seeded_random.ts

Generate seeded random numbers

Additional resources

Did you find what you needed?