/tx-prng

An interface for pseudo random number generators compatible with C (C89 and up) and C++ (C++98 and up), header-only.

Primary LanguageC++MIT LicenseMIT

TX-PRNG

Build Status

PRNG-C provides an interface for pseudo random number generators compatible with C (C89 and up) and C++ (C++98 and up). This library is header-only.

Its primary intention is serve statistical simulations. Therefore, this library includes specific distributions which a random sample can be drawn from. For now this library will only include univariate distributions.

Table of Contents

Namespacing

This library is namespaced for both C and C++.

In C all functions and complex data structures are prefixed by PRNG_ if it is meant to be a consumer API or by prng_ if it is meant to be used internally.

In C++ the namespace prng is set. All functions and complex data structures are written in Pascal case if they are meant for consumer APIs and camelCase if they are used internally.

In the following only the C namespace is used.

The Pseudo Random Number Generator (PRNG) Interface

Every PRNG implemented will offer four basic functions:

  • Advancing to the next state and returning a 64 bit unsigned integer: PRNG_Next.
  • Jumping ahead which is equivalent to calling PRNG_Next a certain number of times: PRNG_Jump.
  • Setting the seed: PRNG_Seed.

List of PRNGs

All PRNGs used here will work on an output of unsigned 64 bit integers. The state of the PRNGs will be at least 64 bit large.

  • xorshift1024* (from) is given as struct PRNG_Xorshift1024Star. A state of 1024 bit for massively parallel applications.

  • xoshiro256** (from) is given as struct PRNG_Xorshiro256StarStar. A state of 256 bit for moderately parallel applications.

  • splitmix64 (from) is only used for seeding. It is given as struct prng_sm64.

List of Univariate Distributions

  • Uniform U[0,1] distribution: PRNG_Uniform.

  • Uniform U(0,1] distribution: PRNG_UniformPositive.

  • Normal distribution: PRNG_Normal.

  • Exponential distribution: PRNG_Exponential.

  • Gamma distribution: PRNG_Gamma.

  • Accept-reject method for any univariate distribution: PRNG_AcceptRejectSingle for a single value, PRNG_AcceptReject for an array of random values.

  • Bernoulli distribution: PRNG_Bernoulli.

  • Beta distribution: PRNG_Beta.

Examples

C as well as C++ examples can be found in the examples directory.

Changelog

  • Version 0.1: First version.