kwin-wang / chaospy

Source code and documentation for the Chaospy package for uncertainty quantification.

Chaospy

Chaospy is a numerical tool for performing uncertainty quantification using polynomial chaos expansions and advanced Monte Carlo methods.

A article in Elsevier Journal of Computational Science has been published introducing the paper: (http://dx.doi.org/10.1016/j.jocs.2015.08.008)

Requirements

python numpy scipy networkx

Optional packages

For regression analysis:

scikit-learn

Prerequisite in Debian/Ubuntu

To install the prerequisite on a Debian/Ubuntu machine:

sudo apt-get install python-scipy python-networkx cython gcc

For scikit-learn:

sudo apt-get install build-essential python-dev \ python-setuptools libatlas-dev libatlas3gf-base

Installation

To install in the site-packages directory and make it importable from anywhere.

Automated download and installation can be done by running the following as super user:

pip install -e git+https://github.com/hplgit/chaospy.git#egg=chaospy

Alternative, download the Github folder and run the following command as super user in the root folder:

python setup.py install

For scikit-learn:

pip install --user --install-option="--prefix=" -U scikit-learn

License

The core code base is licensed under BSD terms. Files with deviating license have their own license written on top of the file.