Skip to main content

Access modules for the MultiNest and Cuba integration libraries

Project description

This library provides programmatic access to MultiNest and PyCuba.

What is MultiNest?

MultiNest is a program and a sampling technique. As a Bayesian inference technique, it allows parameter estimation and model selection. (find out more in the MultiNest paper,, or in a classic MCMC sampler, ). Recently, MultiNest added Importance Nested Sampling (INS, see which is now also supported.

The efficient Monte Carlo algorithm for sampling the parameter space is based on nested sampling and the idea of disjoint multi-dimensional ellipse sampling.

For the scientific community, where Python is becoming the new lingua franca (luckily), I provide an interface to MultiNest.

The automatic build makes sure both Python 2.7 and Python 3 are working correctly with MultiNest and Cuba. It also tests that MultiNest works with MPI enabled.

What does PyMultiNest do?


  • provides an easy-to-use interface to MultiNest and Cuba integration algorithms

  • allows connecting with your existing scientific Python code (numpy, scipy)

  • allows Prior & LogLikelihood functions written in Python.

  • Easy plotting, visualization and summary of MultiNest results.

  • Running MultiNest with MPI

The plotting can also be run on existing MultiNest output, and when not using PyMultiNest for running MultiNest.

Citing PyMultiNest


Plotting results, corner and trace plots

If you got pymultinest running with your likelihood (based on the pymultinest_demo*.py examples), you can create plots of the marginal probability distributions.

If you set outputfiles_basename=”myprefix-” in the run, you need to create a file myprefix-params.json which gives the names of each parameters, for example:


Then you can run:

$ python pymultinest-folder/ myprefix-

which will create marginal plots for you.

Recently I also added support for (needs to be installed):

$ python pymultinest-folder/ myprefix-

Also possible is trace and corner plots:

$ python pymultinest-folder/ myprefix-

This last one has some potential drawbacks however: The code I borrowed here from dynesty is not meant for multi-modal nested sampling which reorders the points, so the trace plot may not be 100% correct for multi-modal problems.

Questions and Problems

For any questions or problems with the software, please open an issue. This helps other people google the same question.

Using MultiNest with Python?

Look at the documentation available at

What is PyCuba?

Cuba (, is a multidimensional numerical integration library for low dimensions. PyCuba allows integration of Python functions, providing an advanced alternative to the basic functions provided in scipy.integrate.

In the Bayesian sense, it is possible to use Cuba for model selection.

Q: Python callback functions are too slow!

If you really identified that your callback functions are too slow, even when using the usual tricks (numpy, etc.), you can implement and compile them as C functions.

You still have the neat python interface (default parameters, etc.), but achieve full execution speed, as only native code is executed while MultiNest runs.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pymultinest-2.11.tar.gz (883.3 kB view hashes)

Uploaded source

Built Distributions

pymultinest-2.11-py3.8.egg (62.4 kB view hashes)

Uploaded 3 8

pymultinest-2.11-py2.py3-none-any.whl (50.7 kB view hashes)

Uploaded 3 8

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page