Persistent PRNG

Persistent Pseudorandom Number Generator. Internally it uses CMWC4096 alogrithm by George Marsaglia. This algorithm gives good statistical results and quite performant.

This library implements persistent PRNG which means it doesn't mutate internal state, it just returns new state. So it's perfectlly ok to use for example in Redux reducers.

Prng.make( seed: number, uint32 = false ): Prng.Data

Create new PRNG instance. Initialization procedure is based on LCG and is borrowed from libtcod sources. For more compact representatin you can set uint32 = true for Uint32Array. By default uses standartJS arrays.

Prng.rand( prng: Prng.Data ): number

Returns current pseudorandom number from stored table. Value itself is unsigned 32bit integer.

Prng.random( prng: Prng.Data, min: number = 0.0, max: number = 1.0 ): number

Returns current pseudorandom number which uniformly distributed in [min,max). By default min = 0 and max = 1, which mimics Math.random(). Note that the number has only 32-bit precision. If min >= max returns min.

Prng.random64( prng: Prng.Data, min: number = 0.0, max: number = 1.0 ): number

Same as random but uses 2 numbers to form 64-bit precision float. Uses current and previous pseudorandom value, so don't forget to do next() twice before use this function.

Prng.next( prng: Prng.Data): Prng.Data

Generates next pseudorandom number for use by value or random. Returns new generator.

Performance

Naive benchmarking gave me on Node 7.9.0 on MacOSX x2.5-x3 perfomance drop compared to native Math.random.