Rao-Blackwellized/Marginal Particle Filter with inner Kalman Filter objectss. More...

#include <rbpf.h>

Inheritance diagram for pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >:

Collaboration diagram for pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >:

Public Types
using	sssv = Eigen::Matrix< float_t, dimss, 1 >

using	nsssv = Eigen::Matrix< float_t, dimnss, 1 >

using	osv = Eigen::Matrix< float_t, dimy, 1 >

using	Mat = Eigen::Matrix< float_t, Eigen::Dynamic, Eigen::Dynamic >

using	nsssMat = Eigen::Matrix< float_t, dimnss, dimnss >

using	cfModType = kalman< dimnss, dimy, 0, float_t, debug >

using	arrayMod = std::array< cfModType, nparts >

using	arrayVec = std::array< sssv, nparts >

using	arrayfloat_t = std::array< float_t, nparts >

Public Types inherited from pf::bases::rbpf_base< float_t, dimss, dimnss, dimy >
using	obs_sized_vec = Eigen::Matrix< float_t, dimobs, 1 >

using	sampled_state_sized_vec = Eigen::Matrix< float_t, dim_s_state, 1 >

using	not_sampled_state_sized_vec = Eigen::Matrix< float_t, dim_ns_state, 1 >

using	dynamic_matrix = Eigen::Matrix< float_t, Eigen::Dynamic, Eigen::Dynamic >

using	func = std::function< const dynamic_matrix(const not_sampled_state_sized_vec &, const sampled_state_sized_vec &)>

using	func_vec = std::vector< func >

Public Member Functions
	rbpf_kalman (const unsigned int &resamp_sched=1)
	The constructor. More...

void	filter (const osv &data, const std::vector< std::function< const Mat(const nsssv &x1t, const sssv &x2t)> > &fs=std::vector< std::function< const Mat(const nsssv &x1t, const sssv &x2t)> >())
	Filter! More...

float_t	getLogCondLike () const
	Get the latest log conditional likelihood. More...

std::vector< Mat >	getExpectations () const
	Get the latest filtered expectation E[h(x_1t, x_2t) \| y_{1:t}]. More...

virtual float_t	logMuEv (const sssv &x21)=0
	Evaluates the first time state density. More...

virtual sssv	q1Samp (const osv &y1)=0
	Sample from the first time's proposal distribution. More...

virtual nsssv	initKalmanMean (const sssv &x21)=0
	Provides the initial mean vector for each Kalman filter object. More...

virtual nsssMat	initKalmanVar (const sssv &x21)=0
	Provides the initial covariance matrix for each Kalman filter object. More...

virtual sssv	qSamp (const sssv &x2tm1, const osv &yt)=0
	Samples the time t second component. More...

virtual float_t	logQ1Ev (const sssv &x21, const osv &y1)=0
	Evaluates the proposal density of the second state component at time 1. More...

virtual float_t	logFEv (const sssv &x2t, const sssv &x2tm1)=0
	Evaluates the state transition density for the second state component. More...

virtual float_t	logQEv (const sssv &x2t, const sssv &x2tm1, const osv &yt)=0
	Evaluates the proposal density at time t > 1. More...

virtual void	updateKalman (cfModType &kMod, const osv &yt, const sssv &x2t)=0
	How to update your inner Kalman filter object at each time. More...

Public Member Functions inherited from pf::bases::rbpf_base< float_t, dimss, dimnss, dimy >
virtual void	filter (const obs_sized_vec &data, const func_vec &fs=func_vec())=0
	the filtering function that must be defined More...

virtual	~rbpf_base ()
	virtual destructor

Private Attributes
unsigned int	m_rs

arrayMod	m_p_innerMods

arrayVec	m_p_samps

arrayfloat_t	m_logUnNormWeights

unsigned int	m_now

float_t	m_lastLogCondLike

resamp_t	m_resampler

std::vector< Mat >	m_expectations

Additional Inherited Members
Static Public Attributes inherited from pf::bases::rbpf_base< float_t, dimss, dimnss, dimy >
static constexpr unsigned int	dim_sampled_state

static constexpr unsigned int	dim_not_sampled_state

static constexpr unsigned int	dim_obs

Detailed Description

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t, typename float_t, bool debug = false>
class pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >

Rao-Blackwellized/Marginal Particle Filter with inner Kalman Filter objectss.

Author: t

Member Typedef Documentation

◆ arrayfloat_t

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::arrayfloat_t = std::array<float_t,nparts>

array of weights

◆ arrayMod

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::arrayMod = std::array<cfModType,nparts>

array of model objects

◆ arrayVec

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::arrayVec = std::array<sssv,nparts>

array of samples

◆ cfModType

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::cfModType = kalman<dimnss,dimy,0,float_t,debug>

closed-form model type

◆ Mat

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::Mat = Eigen::Matrix<float_t,Eigen::Dynamic,Eigen::Dynamic>

dynamic size matrices

◆ nsssMat

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::nsssMat = Eigen::Matrix<float_t,dimnss,dimnss>

"not sampled state size matrix"

◆ nsssv

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::nsssv = Eigen::Matrix<float_t,dimnss,1>

"not sampled state size vector"

◆ osv

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::osv = Eigen::Matrix<float_t,dimy,1>

"observation size vector"

◆ sssv

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

using pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::sssv = Eigen::Matrix<float_t,dimss,1>

"sampled state size vector"

Constructor & Destructor Documentation

◆ rbpf_kalman()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug>

pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::rbpf_kalman ( const unsigned int & resamp_sched = 1 )

The constructor.

Parameters

resamp_sched how often you want to resample (e.g once every resamp_sched time points)

Member Function Documentation

◆ filter()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug>

void pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::filter	(	const osv &	data,
		const std::vector< std::function< const Mat(const nsssv &x1t, const sssv &x2t)> > &	fs = `std::vector<std::function<const Mat(const nsssv &x1t, const sssv &x2t)> >()`
	)

Filter!

The workhorse function

Parameters

data	the most recent observable portion of the time series.
fs	a vector of functions computing E[h(x_1t, x_2t^i)\| x_2t^i,y_1:t]. to be averaged to yield E[h(x_1t, x_2t)\|,y_1:t]

◆ getExpectations()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug>

auto pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::getExpectations

Get the latest filtered expectation E[h(x_1t, x_2t) | y_{1:t}].

Get the expectations you're keeping track of.

Returns: a vector of Mats

◆ getLogCondLike()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug>

float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::getLogCondLike

Get the latest log conditional likelihood.

Returns: the latest log conditional likelihood.

◆ initKalmanMean()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual nsssv pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::initKalmanMean ( const sssv & x21 )

pure virtual

Provides the initial mean vector for each Kalman filter object.

provides the initial mean vector for each Kalman filter object.

Parameters

x21	the second state componenent at time 1.

Returns: a nsssv representing the unconditional mean.

◆ initKalmanVar()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual nsssMat pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::initKalmanVar ( const sssv & x21 )

pure virtual

Provides the initial covariance matrix for each Kalman filter object.

provides the initial covariance matrix for each Kalman filter object.

Parameters

x21	the second state component at time 1.

Returns: a covariance matrix.

◆ logFEv()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::logFEv	(	const sssv &	x2t,
		const sssv &	x2tm1
	)

pure virtual

Evaluates the state transition density for the second state component.

Parameters

x2t	the current second state component.
x2tm1	the previous second state component.

Returns: a float_t evaluation.

◆ logMuEv()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::logMuEv ( const sssv & x21 )

pure virtual

Evaluates the first time state density.

evaluates log mu(x21).

Parameters

x21	component two at time 1

Returns: a float_t evaluation

◆ logQ1Ev()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::logQ1Ev	(	const sssv &	x21,
		const osv &	y1
	)

pure virtual

Evaluates the proposal density of the second state component at time 1.

Parameters

x21	the second state component at time 1 you sampled.
y1	time 1 observation.

Returns: a float_t evaluation of the density.

◆ logQEv()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::logQEv	(	const sssv &	x2t,
		const sssv &	x2tm1,
		const osv &	yt
	)

pure virtual

Evaluates the proposal density at time t > 1.

Parameters

x2t	the current second state component.
x2tm1	the previous second state component.
yt	the current time series observation.

Returns: a float_t evaluation.

◆ q1Samp()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual sssv pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::q1Samp ( const osv & y1 )

pure virtual

Sample from the first time's proposal distribution.

samples the second component of the state at time 1.

Parameters

y1	most recent datum.

Returns: a Vec sample for x21.

◆ qSamp()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual sssv pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::qSamp	(	const sssv &	x2tm1,
		const osv &	yt
	)

pure virtual

Samples the time t second component.

Parameters

x2tm1	the previous time's second state component.
yt	the current observation.

Returns: a sssv sample of the second state component at the current time.

◆ updateKalman()

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

virtual void pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::updateKalman	(	cfModType &	kMod,
		const osv &	yt,
		const sssv &	x2t
	)

pure virtual

How to update your inner Kalman filter object at each time.

Parameters

kMod	a Kalman filter object describing the conditional closed-form model.
yt	the current time series observation.
x2t	the current second state component.

Member Data Documentation

◆ m_expectations

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

std::vector<Mat> pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_expectations

private

expectations

◆ m_lastLogCondLike

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

float_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_lastLogCondLike

private

log p(y_t|y_{1:t-1}) or log p(y1)

◆ m_logUnNormWeights

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

arrayfloat_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_logUnNormWeights

private

the array of the (log of) unnormalized weights

◆ m_now

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

unsigned int pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_now

private

the current time period

◆ m_p_innerMods

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

arrayMod pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_p_innerMods

private

the array of inner Kalman filter objects

◆ m_p_samps

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

arrayVec pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_p_samps

private

the array of particle samples

◆ m_resampler

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

resamp_t pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_resampler

private

resampler object

◆ m_rs

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t , typename float_t , bool debug = false>

unsigned int pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >::m_rs

private

the resamplign schedule

The documentation for this class was generated from the following file:

include/pf/rbpf.h

Public Types

Public Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t, typename float_t, bool debug = false> class pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >

Member Typedef Documentation

◆ arrayfloat_t

◆ arrayMod

◆ arrayVec

◆ cfModType

◆ Mat

◆ nsssMat

◆ nsssv

◆ osv

◆ sssv

Constructor & Destructor Documentation

◆ rbpf_kalman()

Member Function Documentation

◆ filter()

◆ getExpectations()

◆ getLogCondLike()

◆ initKalmanMean()

◆ initKalmanVar()

◆ logFEv()

◆ logMuEv()

◆ logQ1Ev()

◆ logQEv()

◆ q1Samp()

◆ qSamp()

◆ updateKalman()

Member Data Documentation

◆ m_expectations

◆ m_lastLogCondLike

◆ m_logUnNormWeights

◆ m_now

◆ m_p_innerMods

◆ m_p_samps

◆ m_resampler

◆ m_rs

template<size_t nparts, size_t dimnss, size_t dimss, size_t dimy, typename resamp_t, typename float_t, bool debug = false>
class pf::filters::rbpf_kalman< nparts, dimnss, dimss, dimy, resamp_t, float_t, debug >