Introduction

Supernest is a package for accelerating Bayesian inference for algorithms that are sensitive to the boundary between prior and likelihood.

WARNING

If this sounds like gibberish to you, perhaps you should start with McKay, and readon on about Bayesian inference.

This means Nested Sampling (todo: cite Skilling), and to a greater extent newer algorithms some of which are based on nested sampling and some of which were designed as a consequence of supernest.

How to use this

The simple answer is that you should view this as an interactive guided tour of what you can and should do in particular situations with Bayseian Inference.

The supernest package is integrated with particular tools, such as e.g. the Cobaya sampler, which shall be covered in a separate section both here and in the cobaya tutorial.

While there is planned integration with other tools, if it's not stated here, then the tool is not ready for production.

Licensing and Attribution

The main code for the supernest package is licensed under LGPLv3.

INFO

This means that you can use it at no extra cost in your commercial project.

WARNING

This also means that you would need to make any code improvements available upstream (that is, here).

WARNING

However, you must respect the attribution rights, that is you must cite the supernest paper, (once that is finalised, it will be linked here), if you used it.

Generally speaking if what you're doing is open source, just linking to this documentation, or mentioning me as the author should be sufficient. If you're writing an academic publication, citing me would greatly help justifying spedning more time on it.