Approximate Bayesian Computation (ABC) with differential evolution (de) moves and model evidence (Z) estimates.

ABCdeZ.jl offers Bayesian parameter estimation and model comparison/selection for inference problems with an intractable likelihood. Models only need to be simulated (instead of calculating the likelihood). Please visit the documentation (dev/stable) to get started.

The documentation will go through a minimal example (code also found in examples folder above) computing the posterior samples (Figure above) and evidences for two different models. Model evidences can be used to derive posterior model probabilities (Figure below) or Bayes Factors.

ABCdeZ.jl was developed @TSB by Maurice Langhinrichs and Nils Becker. This work is based on many people's previous achievements, particular some part of the code base was adapted from KissABC.jl; please visit the documentation for a complete list of references.