Pseudorandom number generator suite

Definitely still a Work In Progress!

This TypeScript module provides:

• an interface for pseudorandom number generators (PRNG),
• an implementation of a linear congruential generator (LCG),
• a wrapper around PRNGs to obtain common probability distributions.

PRNG

type rng = () => number

Functions marked as rng must return a value in the range [0, 1) and produce pseudorandom numbers (number sequences that look like true randomness).

type RngCreator = (seed?: number) => rng

Functions marked as RngCreator are factories for rng functions. They are used to initialize and seed the rng.

Implementations

GetDefaultGenerator() is an RngCreator that returns the environment's Math.random() generator. Note that Math.random() cannot be seeded and its implementation can vary.

GetLCG(seed?) is an RngCreator that returns a seeded linear congruential generator. It provides a random seed if none is supplied.

RngProvider

The class RngProvider is a wrapper around an rng that provides the following distributions:

• Bernoulli(p?) (coin toss) returns true with probability p (default 0.5) and false with probability 1-p.
• UniformInt() (random integer) returns any integer in a range with equal probability. It has the following variants:
  • UniformInt(): range [0, MAX_SAFE_INTEGER).
  • UniformInt(n): range [0, n) (n excluded).
  • UniformInt(min, max): range [min, max] (max included).
• Uniform() (random float) returns any number in a range with equal probability. It has the following variants:
  • Uniform(): range [0, 1).
  • Uniform(x): range [0, x).
  • Uniform(min, max): range [min, max).
• Dice(dice, faces) (rolling dice) returns the result of rolling dice dice of faces faces each.
• Bates(min, max, n?) (bounded bell) uses a bell-shaped distribution that takes values in [min, max]. It is calculated by averaging n uniform distributions (default 4).
• Pick(options) returns a random element from an array, with equal probability. It preserves the original array.
• Pop(options) removes and returns a random element from an array, with equal probability.
• PickWeightedIndex(weights) returns a random index in the range [0, weights.length) with probability of i proportional to weights[i].
• PickWeighted(options, weights) returns a random element from an array, with probability proportional to its weight. It preserves the original array.
• Shuffle(values) randomly shuffles an array in-place.

Examples

const rng = GetLCG(92740287); // Same seed => same results.
const provider = new RngProvider(rng);
const also = new RngProvide(); // Defaults are provided.

const bernoulli: boolean = provider.Bernoulli(0.3); // 30% chance of true.
const n = provider.UniformInt(3, 5); // 3, 4 or 5.
const x = provider.Uniform(); // 0 <= x < 1.
const roll = provider.Dice(2, 6); // 2 and 12 very unlikely, 7 very likely.
const bell = provider.Bates(8, 10); // Probably close to 9.

const array = [0, 1, 2, 3];
const entry = provider.Pick(array); // 0, 1, 2, or 3.
const another = provider.Pop(array); // Now array is smaller.
provider.PickWeightedIndex([1, 1, 4.5]); // 2 is more likely.
provider.PickWeighted(['a', 'b', 'c'], [1, 1, 4.5]); // 'c' is more likely.
const remix = [4, 5, 6];
provider.Shuffle(remix); // Any re-ordering, e.g. [5, 4, 6].