class pymc.RecursiveDAProposal(step_method_below, model_below, tune, subsampling_rate)[source]#

Recursive Delayed Acceptance proposal to be used with MLDA step sampler. Recursively calls an MLDA sampler if level > 0 and calls MetropolisMLDA or DEMetropolisZMLDA sampler if level = 0. The sampler generates self.subsampling_rate samples and returns the sample with index self.subchain_selection to be used as a proposal. Results in a hierarchy of chains each of which is used to propose samples to the chain above.