Skip to main content

SST::Random

SST includes a few random number generators (RNGs) that can be used by SST objects. These generators produce identical and deterministic output across all platforms and compilers that SST supports. The RNGs have a shared API.

Generators include:

  • Marsaglia
    • SST::RNG::MarsagliaRNG
    • An RNG using the Marsaglia method. This is computationally cheap and provides a reasonable distribution of random numbers. The Mersenne RNG is an alternative for a stronger distribution.
  • Mersenne
    • SST::RNG::MersenneRNG
    • A Mersenne-based RNG. This provides better "randomness" compared to Marsaglia but is more computationally expensive.
  • XOR-Shift
    • SST::RNG::XORShift
    • A lightweight and computationally inexpensive RNG based on xor-shift operations. Implements the algorithm described here.

API

Random number generators implement the following API:

nextUniform

virtual double nextUniform();

Return the next random number in the range[0,1).

generateNextUInt32

virtual uint32_t generateNextUint32();

Generate the next random number as a 32-bit unsigned integer.

generateNextUInt64

virtual uint64_t generateNextUint64();

Generate the next random number as a 64-bit unsigned integer.

generateNextInt32

virtual int32_t generateNextInt32();

Generate the next random number as a 32-bit signed integer.

generateNextInt64

virtual int64_t generateNextInt64();

Generate the next random number as a 64-bit signed integer.

#include <sst/core/rng.h>