ChainConsumer

ChainConsumer is a python package designed to do one thing - consume the chains output from Monte Carlo processes like MCMC. ChainConsumer can utilise these chains to produce plots of the posterior surface inferred from the chain distributions, to plot the chains as walks (to check for mixing and convergence), and to output parameter summaries in the form of LaTeX tables. On top of all of this, if you have multiple models (chains), you can load them all in and perform some model comparison using AIC, BIC or DIC metrics.

To get things started, here is a basic example:

import numpy as np
from chainconsumer import ChainConsumer

mean = [0.0, 4.0]
data = np.random.multivariate_normal(mean, [[1.0, 0.7], [0.7, 1.5]], size=100000)

c = ChainConsumer()
c.add_chain(data, parameters=["$x_1$", "$x_2$"])
c.plotter.plot(filename="example.png", figsize="column", truth=mean)

The output figure is displayed below.

Or you can add more models and look at the summaries between them. Or a ton more, check the examples!

Check out the API and far more Examples below:

Contents

• Usage
  • The Process
  • Statistics
• Examples
  • Basic Usages
  • Customisations
  • More Complicated Examples
• Chain Consumer API
  • Basic Methods
  • Plotter Class
  • Analysis Class
  • Diagnostic Class
  • Model Comparison Class
  • Full Documentation

Installation

ChainConsumer requires the following dependencies, along with a LaTeX installation and dvipng (a maptlotlib dependency):

numpy
scipy
matplotlib
statsmodels

ChainConsumer can be installed as follows:

pip install chainconsumer

Common Issues

Users on some Linux platforms have reported issues rendering plots using ChainConsumer. The common error states that dvipng: not found, and as per StackOverflow post, it can be solved by explicitly install the matplotlib dependency dvipng via sudo apt-get install dvipng.

Citing

You can cite ChainConsumer using the following BibTeX:

@ARTICLE{Hinton2016,
   author = {{Hinton}, S.~R.},
    title = "{ChainConsumer}",
  journal = {The Journal of Open Source Software},
     year = 2016,
    month = aug,
   volume = 1,
      eid = {00045},
    pages = {00045},
      doi = {10.21105/joss.00045},
   adsurl = {http://adsabs.harvard.edu/abs/2016JOSS....1...45H},
}

Contributing

Users that wish to contribute to this project may do so in a number of ways. Firstly, for any feature requests, bugs or general ideas, please raise an issue via Github.

If you wish to contribute code to the project, please simple fork the project on Github and then raise a pull request. Pull requests will be reviewed to determine whether the changes are major or minor in nature, and to ensure all changes are tested.