root/library/bdm/estim/ekf_template.h @ 1025

/*!
  \file
  \brief Bayesian Filtering for linear Gaussian models (Kalman Filter) and extensions
  \author Vaclav Smidl.

  -----------------------------------
  BDM++ - C++ library for Bayesian Decision Making under Uncertainty

  Using IT++ for numerical operations
  -----------------------------------
*/

#ifndef EKF_TEMPLATE_H
#define EKF_TEMPLATE_H

#include "kalman.h"

namespace bdm {

//!Extended Kalman filter with unknown \c Q and \c R
class EKFful_unQR : public EKFfull  {
public:
        void condition ( const vec &QR0 ) {
                Q = diag ( QR0 ( 0, dimension() - 1 ) );
                R = diag ( QR0 ( dimension(), dimension() + dimy - 1 ) );
        };
};

//!Extended Kalman filter in Choleski form with unknown diagonal \c Q
class EKFCh_dQ : public EKFCh {
public:
        root* _copy_() const {
                return new EKFCh_dQ(*this);
        }

        //! new bayes, expects cond = [ut, Qt]
        void bayes( const vec &yt , const vec &cond) {
                vec ut=cond.left(dimc-dimension());
                vec dQt=cond.right(dimension());
                Q.setD ( dQt, 0 );
                //from EKF
                preA.set_submatrix ( dimy + dimension(), dimy, Q._Ch() );
                
                EKFCh::bayes(yt,ut);
        };
        void validate() {
                EKFCh::validate();
                dimc += dimension();
        }
};

//!Extended Kalman filter in Choleski form with unknown \c Q
class EKFCh_chQ : public EKFCh {
public:
        void condition ( const vec &chQ0 ) {
                Q.setCh ( chQ0 );
                //from EKF
                preA.set_submatrix ( dimy + dimension(), dimension(), Q._Ch() );
        };
};

////!Extended Kalman filter with unknown parameters in \c IM
// class EKFCh_cond : public EKFCh  {
// public:
//      void condition ( const vec &val ) {
//              pfxu->condition ( val );
//      };
// };

}
#endif //EKF_TEMP_H