What a Rao-Blackwellized particle filter's update did, per particle and in total. More...

#include <EmrysFilter/Filters/RbParticleFilter.h>

Public Types
using	ArrayType = Eigen::Map<Eigen::Array<Scalar, Eigen::Dynamic, 1> const>
	a per-particle array
using	allocator_type = Allocator
	where the per-particle arrays come from

Public Member Functions
	RbParticleFilterUpdate (Allocator const &allocator={})
	An empty result whose arrays come from allocator.
ArrayType	logLikelihoods () const
ArrayType	nis () const
Eigen::Index	innovationSize () const
allocator_type	get_allocator () const
Scalar	logLikelihood () const

Public Attributes
bool	accepted = false
	whether any particle could explain the observation

Detailed Description

template<typename Scalar = double, typename Allocator = DefaultAllocator>
class Emrys::RbParticleFilterUpdate< Scalar, Allocator >

What a Rao-Blackwellized particle filter's update did, per particle and in total.

A particle whose update the policy rejected has log-likelihood \( -\infty \) and NIS NaN, and weight zero afterwards. When accepted is false, no particle could explain the observation, and the state was left as it was.

The result owns its per-particle arrays, in storage from Allocator. An update into a result (RbParticleFilter::update(measurement, result)) refills that storage, so a result kept across steps allocates nothing once it has held as many particles (see Allocation).