smc_samplers

SMC samplers.

Overview

This module implements:

  • StaticModel: a base class for defining static models (and the corresponding target distributions).
  • FeynmanKac sub-classes that correspond to the following SMC samplers:
  • ThetaParticles classes: classes to represent a collection of N particles, together with extra fields, such as the posterior log-density. The particle systems generated by SMC samplers are objects of these classes.

For more information and examples, see the following notebook tutorial.

Module summary

StaticModel([data, prior]) Base class for static models.
ThetaParticles(**kwargs) Base class for particle systems for SMC samplers.
MetroParticles([theta, lpost, acc_rates]) Particles that may be moved through a Metropolis step.
FKSMCsampler(model[, mh_options]) Base FeynmanKac class for SMC samplers.
IBIS(model[, mh_options]) FeynmanKac class for IBIS algorithm.
Tempering(model[, mh_options, exponents]) FeynmanKac class for tempering SMC.
AdaptiveTempering(model[, mh_options, ESSrmin]) Feynman-Kac class for adaptive tempering SMC.
SMC2([ssm_cls, prior, data, smc_options, …]) Feynman-Kac subclass for the SMC^2 algorithm.