KalmanLinearAlgebra.h File Reference

Kalman filter linear algebra typedefs. More...

#include <cmath>
#include "boost/serialization/array_wrapper.hpp"
#include "boost/numeric/ublas/vector.hpp"
#include "boost/numeric/ublas/matrix.hpp"
#include "boost/numeric/ublas/symmetric.hpp"
#include "boost/numeric/ublas/lu.hpp"

Go to the source code of this file.

Classes
struct	trkf::KVector< N >
	Define a shortened alias for ublas namespace. More...
struct	trkf::KSymMatrix< N >
	Symmetric matrix, dimension NxN. More...
struct	trkf::KMatrix< N, M >
	General matrix, dimension NxM. More...
struct	trkf::KHMatrix< N >
	Kalman H-matrix, dimension Nx5. More...
struct	trkf::KGMatrix< N >
	Kalman gain matrix, dimension 5xN. More...

Namespaces
	trkf

Typedefs
typedef KVector< 5 >::type	trkf::TrackVector
	Track state vector, dimension 5. More...
typedef KSymMatrix< 5 >::type	trkf::TrackError
	Track error matrix, dimension 5x5. More...
typedef KMatrix< 5, 5 >::type	trkf::TrackMatrix
	General 5x5 matrix. More...

Functions
template<class T , class TRI , class L , class A >
bool	trkf::syminvert (ublas::symmetric_matrix< T, TRI, L, A > &m)
template<class T , class L , class A >
bool	trkf::invert (ublas::matrix< T, L, A > &m)
template<class M >
M::value_type	trkf::trace (const M &m)

Detailed Description

Kalman filter linear algebra typedefs.

Author

H. Greenlee

There are various linear algebra typedefs defined in this header:

1. KVector<N>::type - Vector, dimension N.
2. KSymMatrix<N>::type - Symmetric matrix, dimension NxN.
3. KMatrix<N,M>::type - A matrix with dimension NxM.
4. KHMatrix<N>::type - Matrix with dimension Nx5 (H-matrix).
5. KGMatrix<N>::type - Matrix with dimension 5xN (gain matrix).
6. TrackVector - Track state vector, fixed dimension 5
7. TrackError - Track error matrix, fixed dimension 5x5.
8. TrackMatrix - General matrix, fixed dimension 5x5.

The above typedefs refer to some ublas (boost/numeric/ublas) linear algebra class. The templated typedefs are defined as a member of a template class because c++ doesn't have template typedefs.

All linear algebra objects use the following storage model.

1. Matrices are stored in row major order (normal c/c++ convention).
2. Symmetric matrices are stored in lower triangular format.
3. Amount of preallocated stack memory is specified at compilation time (the actual size of objects must still be specified at run time).

Surprisingly, the ublas linear algebra package does not have any built-in symmetric matrix inverse function. We provide one here as free function syminvert.

Definition in file KalmanLinearAlgebra.h.