Exact Random Distributions

Author

Charles F. F. Karney (charl.nosp@m.es.k.nosp@m.arney.nosp@m.@sri.nosp@m..com)

Version

3.0

Date

2020-12-27

This library provides an implementation of the algorithms described in the paper

• C. F. F. Karney,
  Sampling exactly from the normal distribution,
  ACM Trans. Math. Software 42(1), 3:1–14 (Jan. 2016).
  doi: 10.1145/2710016
  preprint: arXiv:1303.6257v2

It provides classes for sampling exactly from the continuous unit normal and exponential distributions and from the discrete normal distribution (given a source of uniformly distribution random digits). The library requires a C++11 compiler and consists only of header files.

Version 3.0 incorporates also improvements suggested by

• Y. Du, B. Fan, and B. Wei,
  An improved exact sampling algorithm for the standard normal distribution,
  preprint: arXiv:2008.03855v1 (Aug. 2020).

for sampling the unit normal and discrete normal distributions. See Description of the algorithm for details.

It is licensed under the MIT/X11 License; see LICENSE.txt for the terms.

Download

The main project page is at

• http://sourceforge.net/projects/exrandom

You can download the library as a tar.gz or a zip file from

• exrandom-3.0.tar.gz
• exrandom-3.0.zip

The version of the code included with the original, March 2013, version of the paper is available at

• exrandom-1.0.tar.gz

Contents

• Simple examples
• Getting started
• Compiling all the examples
• Code organization
• Examples using u-rands
• Multi-precision floating-point types
• MPFR functions
• Other examples
• Description of the algorithm
• A brief history of the algorithms

Forward to Simple examples.