CPF_RB: Conditional Particle Filter with RB

In pierrejacob/CoupledCPF: Coupling of Conditional Particle Filters

Description

Runs a conditional particle filter, with or without ancestor sampling, and with Rao-Blackwellization

Usage

CPF_RB(nparticles, model, theta, observations, ref_trajectory = NULL,
  with_as = FALSE, h = function(trajectory) {     return(trajectory) })

Arguments

nparticles	number of particles
model	a list representing a model, for instance as given by get_ar.
theta	a parameter to give to the model functions
observations	a matrix of observations of size datalength x dimension(observation)
ref_trajectory	a reference trajectory, of size dimension(process) x datalength; if missing, runs a standard particle filter.
with_as	whether ancestor sampling should be used (TRUE/FALSE)
h	test function, which we want to integrate with respect to the smoothing distribution

Value

A list with 'new_trajectory' and 'estimate'

pierrejacob/CoupledCPF documentation built on May 25, 2019, 6:05 a.m.