epyc is a library to manage large sets of computational experiments.

Computational science often needs us to perform a lot of repetitions of experiments, for example to reduce variance or to cancel out strange artefacts. Typically one might also want to perform a set of experiments across a parameter space, varying one or more parameters to see their effect. All these tasks generate a lot of repetitions that need to be managed while they’re on-going, and a lot of result data that needs to be kept archived.

These activities are so structured, so repetitive, and so common, that they cry out for automation — but there didn’t seem to be an automation library for doing this sort of thing in Python. So I wrote one.

epyc is being used to manage experiments with complex networks, but it can manage any kind of experiment written in Python, and can run experiments in parallel using an IPython compute cluster, or locally taking advantage of a multicore workstation.


Recent articles about epyc

All articles