infFlt.hpp

Go to the documentation of this file.

#ifndef _BAYES_FILTER_INFORMATION
#define _BAYES_FILTER_INFORMATION

/*
 * Bayes++ the Bayesian Filtering Library
 * Copyright (c) 2002 Michael Stevens
 * See accompanying Bayes++.htm for terms and conditions of use.
 *
 * $Id$
 */

/*
 * Information Filter Scheme.
 *  A possibly non-linear Information filter
 *
 * References
 * [1] "Stochastic Models, Estimation, and Control} Peter S Maybeck, Academic Press, ISBN 0-12-480701-1
 *      Section 5.7
 * [2] "Kalman Filtering, Theory and Practice", Mohinder S. Grewal, Angus P. Andrews ISBN 0-13-211335-X
 * To work with with Linear and Linrz models
 *  a) a separate state and information (via covariance) predict is used.
 *  b) a EIF modified innovation update algorithm is used.
 *
 * Two alternative algorithms are used for predict functions:
 *  For linrz models an extended predict form is used so information state 'y' is predicted via the
 *  non-linear function. This requires that X, and Y are invertible so 'x' can be computed.
 * For linear invertible models predict can be done directly without computing x
 * Discontinuous observe models require that predict is normailised with
 * respect to the observation.
 * NUMERICS
 *  The state x is represented by prod(X,y). This may be ill conditioned if the product is
 *  ill conditioned. At present only the conditioning of X if checked, if y has a large range the product
 *  may be ill conditioned. This is not checked
 *
 * The filter is operated by performing a
 *  predict, observe
 * cycle defined by the base class
 */
#include "bayesFlt.hpp"

/* Filter namespace */
namespace Bayesian_filter
{

class Information_scheme : public Extended_kalman_filter, virtual public Information_state_filter
{
public:
    Information_scheme (std::size_t x_size, std::size_t z_initialsize = 0);
    Information_scheme& operator= (const Information_scheme&);
    // Optimise copy assignment to only copy filter state

    struct Linear_predict_byproducts
    {
         Linear_predict_byproducts (std::size_t x_size, std::size_t q_size);
         FM::SymMatrix A;
         FM::Matrix tempG;
         FM::SymMatrix B;
         FM::Vec y;
    };


    void init ();
    void update ();
    void init_yY ();
    void update_yY ();
    // Covariance and information form state interface

    Float predict (Linear_invertable_predict_model& f, Linear_predict_byproducts& b);
    // Linear Prediction in information form as in Ref[2]
    Float predict (Linear_invertable_predict_model& f)
    {
        Linear_predict_byproducts b(f.Fx.size1(),f.q.size());
        return predict (f, b);  
    }

    Float predict (Linrz_predict_model& f);
    // Extended Prediction via state

    Float observe_innovation (Linrz_uncorrelated_observe_model& h, const FM::Vec& s);
    Float observe_innovation (Linrz_correlated_observe_model& h, const FM::Vec& s);

protected:
    bool update_required;   // Postcondition of update is not met

protected:                  // Permanently allocated temps
    FM::RowMatrix tempX;
    FM::Vec i;
    FM::SymMatrix I;
                    // allow fast operation if z_size remains constant
    FM::SymMatrix ZI;
    std::size_t last_z_size;
    void observe_size (std::size_t z_size);
};


}//namespace
#endif