package prbnmcn-stats

  1. Overview
  2. Docs

Generic sampling loop.



val mcmc : verbosity:[ `Silent | `Progress | `Trace ] -> initial:X.t -> burn_in:int -> (X.t -> ('s, X.t) Stats_intf.gen) -> ('s, ('s, X.t) Stats_intf.gen) Stats_intf.gen

The sampler function produces a generative probability. More precisely: each sample from mcmc ~verbosity ~initial ~burn_in is a distinct Markov chain, which samples can in-turn be obtained by calling iteratively the closure returned by mcmc. Concretely, the outer `Stats.gen` corresponds to sampling the burn-in while the inner corresponds to sampling from the actual chain.