`altar.bayesian.MPIAnnealing`

Module Contents

Classes

class altar.bayesian.MPIAnnealing.MPIAnnealing(annealer, worker, communicator=None, **kwds)

Bases: altar.bayesian.AnnealingMethod.AnnealingMethod

A distributed implementation of the annealing method that uses MPI

manager = 0

worker

initialize(self, application): Initialize me and my parts given an {application} context

start(self, annealer): Start the annealing process

top(self, annealer): Notification that we are at the beginning of a β update

cool(self, annealer): Push my state forward along the cooling schedule

walk(self, annealer): Explore configuration space by walking the Markov chains

resample(self, annealer, statistics): Analyze the acceptance statistics and take the problem state to the end of the annealing step

archive(self, annealer, scaling, stats): Notify archiver to record annealer information

bottom(self, annealer): Notification that we are at the end of a β update

finish(self, annealer): Shut down the annealing process

collect(self): Assemble my global state

partition(self): Distribute my global state