George

George is a fast and flexible Python library for Gaussian Process (GP) Regression. A full introduction to the theory of Gaussian Processes is beyond the scope of this documentation but the best resource is available for free online: Rasmussen & Williams (2006).

Unlike some other GP implementations, george is focused on efficiently evaluating the marginalized likelihood of a dataset under a GP prior, even as this dataset gets Big™. As you’ll see in these pages of documentation, the module exposes quite a few other features but it is designed to be used alongside your favorite non-linear optimization or posterior inference library for the best results.

George is being actively developed in a public repository on GitHub so if you have any trouble, open an issue there.

https://img.shields.io/badge/GitHub-dfm%2Fgeorge-blue.svg?style=flat http://img.shields.io/badge/license-MIT-blue.svg?style=flat http://img.shields.io/travis/dfm/george/master.svg?style=flat https://ci.appveyor.com/api/projects/status/xy4ts3v3sk5lo5ll?svg=true&style=flat https://coveralls.io/repos/github/dfm/george/badge.svg?branch=master&style=flat

License & Attribution

Copyright 2012-2017 Daniel Foreman-Mackey and contributors.

George is being developed by Dan Foreman-Mackey in a public GitHub repository. The source code is made available under the terms of the MIT license.

If you make use of this code, please cite the paper which is currently on the ArXiv:

@article{hodlr,
    author = {{Ambikasaran}, S. and {Foreman-Mackey}, D. and
              {Greengard}, L. and {Hogg}, D.~W. and {O'Neil}, M.},
     title = "{Fast Direct Methods for Gaussian Processes}",
      year = 2014,
     month = mar,
       url = http://arxiv.org/abs/1403.6015
}

Changelog

0.3.0 (2017-07-12)

  • New modeling protocol for parameter fitting
  • Rewritten HODLR algorithm - more stable and better performance
  • New interface for “easily” implementing new kernel functions

pre-0.3

  • This project has had a long and tumultuous history