This repo is a placeholder. The previous code for creating a Gaussian / normal distribution has been lost but I believe broadly followed the SO discussion here, provisioning a custom implementation of the below methods. With a new Gaussian object it would be nice to control the fall-off and have an option to set clipping on the output range.
TODO: Test, benchmark and (potentially) add implementations for the following methods. Prefer npm libs where available.
- Box-Muller transform
- Central limit theorem
- Marsaglia polar method [random-normal package]
- Ziggurat
- Ratio-of-uniforms
Useful references: