Detailed Description

template<class HESS_Ap, class HESS_f, class HESS_Apf>
class mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >

A generic symbolic and numeric Schur-complement handler for builing reduced systems of equations.

Definition at line 17 of file schur.h.

#include <mrpt/srba/impl/schur.h>

Inheritance diagram for mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >:

Classes
struct	TGradApSymbolicEntry

struct	THApSymbolicEntry
	Info for each block in HAp. More...

struct	TInfoPerHfBlock

Public Member Functions
	SchurComplement (HESS_Ap &_HAp, HESS_f &_Hf, HESS_Apf &_HApf, double _minus_grad_Ap, double _minus_grad_f)
	Constructor: builds the symbolic representations Note: HApf must be in row-compressed form; HAp & Hf in column-compressed form. More...

void	realize_HAp_changed ()
	Must be called after the numerical values of the Hessian HAp change, typically after an optimization update which led to re-evaluation of the Jacobians around a new linearization point. More...

size_t	getNumFeatures () const
	After calling numeric_build_reduced_system() one can get the stats on how many features are actually estimable. More...

size_t	getNumFeaturesFullRank () const

void	numeric_build_reduced_system (const double lambda)
	Replace the HAp matrix with its Schur reduced version. More...

void	numeric_solve_for_features (double in_deltas_Ap, double out_deltas_feats)

bool	was_ith_feature_invertible (const size_t i) const

Private Types
typedef Eigen::Map< Eigen::Matrix< double, HESS_Ap::matrix_t::RowsAtCompileTime, 1 > >	vector_Ap_t

typedef Eigen::Map< Eigen::Matrix< double, HESS_f::matrix_t::RowsAtCompileTime, 1 > >	vector_f_t

typedef mrpt::aligned_containers< TInfoPerHfBlock >::vector_t	TInfoPerHfBlock_vector_t

Private Attributes
HESS_Ap	HAp_original

HESS_Ap &	HAp

HESS_f &	Hf

HESS_Apf &	HApf

double *	minus_grad_Ap

double *	minus_grad_f

const size_t	nUnknowns_Ap

const size_t	nUnknowns_f

size_t	nHf_invertible_blocks
	for stats, the number of Hf diagonal blocks which are not rank-deficient. More...

TInfoPerHfBlock_vector_t	m_Hf_blocks_info

std::deque< THApSymbolicEntry >	m_sym_HAp_reduce
	All the required operations needed to update HAp into its reduced system. More...

std::vector< TGradApSymbolicEntry >	m_sym_GradAp_reduce
	All the required operations needed to update Gradient of Ap into its reduced system. More...

Member Typedef Documentation

template<class HESS_Ap, class HESS_f, class HESS_Apf>

typedef mrpt::aligned_containers<TInfoPerHfBlock>::vector_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::TInfoPerHfBlock_vector_t

private

Definition at line 344 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

typedef Eigen::Map<Eigen::Matrix<double,HESS_Ap::matrix_t::RowsAtCompileTime,1> > mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::vector_Ap_t

private

Definition at line 329 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

typedef Eigen::Map<Eigen::Matrix<double,HESS_f::matrix_t::RowsAtCompileTime,1> > mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::vector_f_t

private

Definition at line 330 of file schur.h.

Constructor & Destructor Documentation

template<class HESS_Ap, class HESS_f, class HESS_Apf>

mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::SchurComplement	(	HESS_Ap &	_HAp,
		HESS_f &	_Hf,
		HESS_Apf &	_HApf,
		double *	_minus_grad_Ap,
		double *	_minus_grad_f
	)

inline

Constructor: builds the symbolic representations Note: HApf must be in row-compressed form; HAp & Hf in column-compressed form.

Parameters

[in] _minus_grad_f Can be NULL if there're no observations of landmarks with unknown positions (may still be of LMs with known ones).

Definition at line 25 of file schur.h.

Member Function Documentation

template<class HESS_Ap, class HESS_f, class HESS_Apf>

size_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::getNumFeatures ( ) const

inline

After calling numeric_build_reduced_system() one can get the stats on how many features are actually estimable.

Definition at line 171 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

size_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::getNumFeaturesFullRank ( ) const

inline

Definition at line 172 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

void mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::numeric_build_reduced_system ( const double lambda )

inline

Replace the HAp matrix with its Schur reduced version.

The lambda value is used to sum it to the diagonal of Hf (features), but it's NOT added to the output HAp. This is intentional, so the caller can add different lamdba values as needed in different trials if an ill conditioned matrix is found.

Definition at line 180 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

void mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::numeric_solve_for_features	(	double *	in_deltas_Ap,
		double *	out_deltas_feats
	)

inline

Definition at line 271 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

void mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::realize_HAp_changed ( )

inline

Must be called after the numerical values of the Hessian HAp change, typically after an optimization update which led to re-evaluation of the Jacobians around a new linearization point.

This method latches the current value of HAp for reseting it again if we later need to recompute the reduced Schur system for different values of lambda.

Definition at line 165 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

bool mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::was_ith_feature_invertible ( const size_t i ) const

inline

Definition at line 314 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_solve_for_features().

Member Data Documentation

template<class HESS_Ap, class HESS_f, class HESS_Apf>

HESS_Ap& mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::HAp

private

Definition at line 319 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

HESS_Ap mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::HAp_original

private

Definition at line 318 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

HESS_Apf& mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::HApf

private

Definition at line 321 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_solve_for_features(), and mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::SchurComplement().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

HESS_f& mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::Hf

private

Definition at line 320 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::SchurComplement().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

TInfoPerHfBlock_vector_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::m_Hf_blocks_info

private

Definition at line 346 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

std::vector<TGradApSymbolicEntry> mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::m_sym_GradAp_reduce

private

All the required operations needed to update Gradient of Ap into its reduced system.

Definition at line 410 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_build_reduced_system(), and mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::SchurComplement().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

std::deque<THApSymbolicEntry> mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::m_sym_HAp_reduce

private

All the required operations needed to update HAp into its reduced system.

Definition at line 383 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_build_reduced_system(), and mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::SchurComplement().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

double* mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::minus_grad_Ap

private

Definition at line 322 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_build_reduced_system().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

double* mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::minus_grad_f

private

Definition at line 323 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_build_reduced_system(), and mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_solve_for_features().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

size_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::nHf_invertible_blocks

private

for stats, the number of Hf diagonal blocks which are not rank-deficient.

Definition at line 327 of file schur.h.

Referenced by mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::getNumFeaturesFullRank(), and mrpt::srba::SchurComplement< typename hessian_traits_t::TSparseBlocksHessian_Ap, typename hessian_traits_t::TSparseBlocksHessian_f, typename hessian_traits_t::TSparseBlocksHessian_Apf >::numeric_build_reduced_system().

template<class HESS_Ap, class HESS_f, class HESS_Apf>

const size_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::nUnknowns_Ap

private

Definition at line 325 of file schur.h.

template<class HESS_Ap, class HESS_f, class HESS_Apf>

const size_t mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >::nUnknowns_f

private

Definition at line 326 of file schur.h.

Detailed Description

template<class HESS_Ap, class HESS_f, class HESS_Apf> class mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >

Classes

Public Member Functions

Private Types

Private Attributes

Member Typedef Documentation

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation

template<class HESS_Ap, class HESS_f, class HESS_Apf>
class mrpt::srba::SchurComplement< HESS_Ap, HESS_f, HESS_Apf >