The ARES3 package comes with a basic flavor within the binary program "ares3". "ares3" is an implementation of the algorithm outlined in the paper "Matrix free Large scale Bayesian inference" (Jasche & Lavaux 2014)
ARES3 mathematical framework is detailed in "ARES model".
The ARES3 serves as a basis for number of extensions and modules. The minimal extension is the foreground sampler mechanism (here published), that allows to fit some model of foreground contamination in large scale structure data. The second main module is the HADES sampler, which incorporates the HMC base definition and implementation alongside some likelihood models. The third module is the BORG sampler. It is a much more advanced likelihood analysis which incorporates non-linear dynamics of the Large scale structures. The fourth major module is the VIRBIUS (Lavaux 2016) model implemented as a BORG likelihood. It has been a major effort principally by Florian Führer. The final auxiliary module offers both a smaller variant of the HMC algorithm and a Julia bind to quickly test new likelihood and forward models.