Chaospy is a numerical tool for performing uncertainty quantification using polynomial chaos expansions and advanced Monte Carlo methods implemented in Python 2 and 3.

A article in Elsevier Journal of Computational Science has been published introducing the software: here. If you are using this software in work that will be published, please cite this paper.

Installation should be straight forward:

pip install chaospy

And you should be ready to go.

Alternatively, to get the most current experimental version, the code can be installed from Github as follows:

git clone git@github.com:jonathf/chaospy.git
cd chaospy
pip install -r requirements.txt
python setup.py install

The last command might need sudo prefix, depending on your python setup.

Optionally, to support more regression methods, install the Scikit-learn package:

pip install scikit-learn

chaospy is created to be simple and modular. A simple script to implement point collocation method will look as follows:

import chaospy
import numpy

# your code wrapper goes here
def foo(coord, prm):
    """Function to do uncertainty quantification on."""
    return prm[0] * numpy.e ** (-prm[1] * numpy.linspace(0, 10, 100))

# bi-variate probability distribution
distribution = choaspy.J(cp.Uniform(1, 2), chaospy.Uniform(0.1, 0.2))

# polynomial chaos expansion
polynomial_expansion = chaospy.orth_ttr(8, distribution)

# samples:
samples = distribution.sample(1000)

# evaluations:
evals = [foo(sample) for sample in samples.T]

# polynomial approximation
foo_approx = chaospy.fit_regression(
    polynomial_expansion, samples, evals)

# statistical metrics
expected = chaospy.E(foo_approx, distribution)
deviation = chaospy.Std(foo_approx, distribution)

For a more extensive description of what going on, see the tutorial.

For a collection of recipes, see the cookbook.

For any problems and questions you might have related to chaospy, please feel free to file an issue.