models.hpp

Go to the documentation of this file.

#ifndef _BAYES_FILTER__MODELS
#define _BAYES_FILTER__MODELS

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

/*
 * Predict and Observe models
 *  These models extend, adapt and simplify the fundamental Bayesian filter models
 *  Simple : Simplify model construction and use
 *  General: Generalise a model so it include properties of more then one model
 *  Adapted: Adapt one model type into another
 */
#include <boost/function.hpp>

/* Filter namespace */
namespace Bayesian_filter
{

typedef boost::function1<const FM::Vec&, const FM::Vec&> State_function;
// A generalised function of state. Compatible with predict and observe models


class Simple_additive_predict_model : public Additive_predict_model
// Additive predict model initialised from function and model matricies
{
    State_function ff;
public:
    Simple_additive_predict_model (State_function f_init, const FM::Matrix& G_init, const FM::Vec& q_init);
    // Precondition: G, q are conformantly dimensioned (not checked)

    // No default assignment operator

    virtual const FM::Vec& f(const FM::Vec& x) const
    {   return ff(x);
    }
};

class Simple_linrz_predict_model : public Linrz_predict_model
// Linrz predict model initialised from function and model matrices
{
    State_function ff;
public:
    Simple_linrz_predict_model (State_function f_init, const FM::Matrix& Fx_init, const FM::Matrix& G_init, const FM::Vec& q_init);
    // Precondition: Fx, G, q are conformantly dimensioned (not checked)

    // No default assignment operator

    virtual const FM::Vec& f(const FM::Vec& x) const
    {   return ff(x);
    }
};

class Simple_linear_predict_model : public Linear_predict_model
// Linear predict model initialised from model matricies
{
public:
    Simple_linear_predict_model (const FM::Matrix& Fx_init, const FM::Matrix& G_init, const FM::Vec& q_init);
    // Precondition: Fx, q and G are conformantly dimensioned (not checked)
};


class Simple_linrz_correlated_observe_model : public Linrz_correlated_observe_model
// Linrz observe model initialised from function and model matrices
{
    State_function ff;
public:
    Simple_linrz_correlated_observe_model (State_function f_init, const FM::Matrix& Hx_init, const FM::SymMatrix& Z_init);
    // Precondition: Hx, Z are conformantly dimensioned (not checked)
    // No default assignment operator

    virtual const FM::Vec& h(const FM::Vec& x) const
    {   return ff(x);
    }
};

class Simple_linrz_uncorrelated_observe_model : public Linrz_uncorrelated_observe_model
// Linrz observe model initialised from function and model matrices
{
    State_function ff;
public:
    Simple_linrz_uncorrelated_observe_model (State_function f_init, const FM::Matrix& Hx_init, const FM::Vec& Zv_init);
    // Precondition: Hx, Zv are conformantly dimensioned (not checked)
    // No default assignment operator

    virtual const FM::Vec& h(const FM::Vec& x) const
    {   return ff(x);
    }
};

class Simple_linear_correlated_observe_model : public Linear_correlated_observe_model
// Linear observe model initialised from model matrices
{
public:
    Simple_linear_correlated_observe_model (const FM::Matrix& Hx_init, const FM::SymMatrix& Z_init);
    // Precondition: Hx, Z are conformantly dimensioned (not checked)
};

class Simple_linear_uncorrelated_observe_model : public Linear_uncorrelated_observe_model
// Linear observe model initialised from model matrices
{
public:
    Simple_linear_uncorrelated_observe_model (const FM::Matrix& Hx_init, const FM::Vec& Zv_init);
    // Precondition: Hx, Zv are conformantly dimensioned (not checked)
};



/*
 * Model Adaptors: Constructed with a reference to another model
 */


class Adapted_Correlated_additive_observe_model : public Correlated_additive_observe_model
/*
 * Adapt Uncorrelated_additive_observe_model to an equivalent
 * Correlated_additive_observe_model_adaptor
 */
{
public:
    Adapted_Correlated_additive_observe_model (Uncorrelated_additive_observe_model& adapt);
    const FM::Vec& h(const FM::Vec& x) const
    {
        return unc.h(x);
    }
    inline void normalise (FM::Vec& z_denorm, const FM::Vec& z_from) const
    {
        unc.normalise (z_denorm, z_from);
    };
private:
    Uncorrelated_additive_observe_model& unc;
};

class Adapted_Linrz_correlated_observe_model : public Linrz_correlated_observe_model
/*
 * Adapt Linrz_uncorrelated_observe_model to an equivalent
 * Linrz_correlated_observe_model
 */
{
public:
    Adapted_Linrz_correlated_observe_model (Linrz_uncorrelated_observe_model& adapt);
    const FM::Vec& h(const FM::Vec& x) const
    {
        return unc.h(x);
    }
    inline void normalise (FM::Vec& z_denorm, const FM::Vec& z_from) const
    {
        unc.normalise (z_denorm, z_from);
    };
protected:
    Linrz_uncorrelated_observe_model& unc;
};


/*
 * Generalised Models: generalise a model so it include properties of more then one model.
 */

// General Linearised Uncorrelated Additive and Likelihood observe model
class General_LzUnAd_observe_model : public Linrz_uncorrelated_observe_model, public Likelihood_observe_model
{
public:
    General_LzUnAd_observe_model (std::size_t x_size, std::size_t z_size) :
        Linrz_uncorrelated_observe_model(x_size, z_size),
        Likelihood_observe_model(z_size),
        li(z_size)
    {}
    virtual Float L(const FM::Vec& x) const
    // Definition of likelihood for additive noise model given zz
    {   return li.L(*this, z, h(x));
    }
    virtual void Lz (const FM::Vec& zz)
    // Fix the observation zz about which to evaluate the Likelihood function
    // Zv is also fixed
    {   Likelihood_observe_model::z = zz;
        li.Lz(*this);
    }
private:
    friend class General_LiUnAd_observe_model;
    struct Likelihood_uncorrelated
    {
        Likelihood_uncorrelated(std::size_t z_size) :
            zInnov(z_size), Zv_inv(z_size)
        {   zset = false;
        }
        mutable FM::Vec zInnov; // Normailised innovation, temporary for L(x)
        FM::Vec Zv_inv;         // Inverse Noise Covariance given zz
        Float logdetZ;          // log(det(Z))
        bool zset;
        Float L(const Uncorrelated_additive_observe_model& model, const FM::Vec& z, const FM::Vec& zp) const;
        // Definition of likelihood for additive noise model given zz
        void Lz(const Uncorrelated_additive_observe_model& model);
    };
    Likelihood_uncorrelated li;
};

// General Linear Uncorrelated Additive and Likelihood observe model
class General_LiUnAd_observe_model : public Linear_uncorrelated_observe_model, public Likelihood_observe_model
{
public:
    General_LiUnAd_observe_model (std::size_t x_size, std::size_t z_size) :
        Linear_uncorrelated_observe_model(x_size, z_size),
        Likelihood_observe_model(z_size),
        li(z_size)
    {}
    virtual Float L(const FM::Vec& x) const
    // Definition of likelihood for additive noise model given zz
    {   return li.L(*this, z, h(x));
    }
    virtual void Lz (const FM::Vec& zz)
    // Fix the observation zz about which to evaluate the Likelihood function
    // Zv is also fixed
    {   Likelihood_observe_model::z = zz;
        li.Lz(*this);
    }

private:
    General_LzUnAd_observe_model::Likelihood_uncorrelated li;
};

// General Linearised Correlated Additive and Likelihood observe model
class General_LzCoAd_observe_model : public Linrz_correlated_observe_model, public Likelihood_observe_model
{
public:
    General_LzCoAd_observe_model (std::size_t x_size, std::size_t z_size) :
        Linrz_correlated_observe_model(x_size, z_size),
        Likelihood_observe_model(z_size),
        li(z_size)
    {}
    virtual Float L(const FM::Vec& x) const
    // Definition of likelihood for additive noise model given zz
    {   return li.L(*this, z, h(x));
    }
    virtual void Lz (const FM::Vec& zz)
    // Fix the observation zz about which to evaluate the Likelihood function
    // Zv is also fixed
    {   Likelihood_observe_model::z = zz;
        li.Lz(*this);
    }

private:
    friend class General_LiCoAd_observe_model;
    struct Likelihood_correlated
    {
        Likelihood_correlated(std::size_t z_size) :
            zInnov(z_size), Z_inv(z_size,z_size)
        {   zset = false;
        }
        mutable FM::Vec zInnov; // Normalised innovation, temporary for L(x)
        FM::SymMatrix Z_inv;    // Inverse Noise Covariance
        Float logdetZ;          // log(det(Z)
        bool zset;  
        static Float scaled_vector_square(const FM::Vec& v, const FM::SymMatrix& V);
        Float L(const Correlated_additive_observe_model& model, const FM::Vec& z, const FM::Vec& zp) const;
        // Definition of likelihood for additive noise model given zz
        void Lz(const Correlated_additive_observe_model& model);
    };
    Likelihood_correlated li;
};

// General Linear Correlated Additive and Likelihood observe model
class General_LiCoAd_observe_model : public Linear_correlated_observe_model, public Likelihood_observe_model
{
public:
    General_LiCoAd_observe_model (std::size_t x_size, std::size_t z_size) :
        Linear_correlated_observe_model(x_size, z_size),
        Likelihood_observe_model(z_size),
        li(z_size)
    {}
    virtual Float L(const FM::Vec& x) const
    // Definition of likelihood for additive noise model given zz
    {   return li.L(*this, z, h(x));
    }
    virtual void Lz (const FM::Vec& zz)
    // Fix the observation zz about which to evaluate the Likelihood function
    // Zv is also fixed
    {   Likelihood_observe_model::z = zz;
        li.Lz(*this);
    }

private:
    General_LzCoAd_observe_model::Likelihood_correlated li;
};


}// namespace

#endif