Inference

BayesBay utilizes reversible-jump Markov chain Monte Carlo (MCMC) for sampling the posterior probability. This is achieved through the BayesianInversion and MarkovChain classes (which are subclasses of bayesbay.BaseBayesianInversion and bayesbay.BaseMarkovChain). A BayesianInversion instance serves as a bridge between the parameterization of the inference problem and its operational facets, such as the parallel execution of a specified number of Markov chains to gather posterior samples. Upon defining a BayesianInversion instance, one or multiple MarkovChain instances are automatically generated. Via a bayesbay.State, these encapsulate all information related to a given state of the inference problem, which can be accessed via a subclass of bayesbay.samplers.Sampler (see Samplers).

graph TD; BaseBayesianInversion-->BayesianInversion; BaseMarkovChain-->MarkovChain;

bayesbay.BayesianInversion	A high-level class for performing Bayesian inversion using Markov Chain Monte Carlo (McMC) methods.
bayesbay.MarkovChain	High-level interface for a Markov Chain.