root/library/bdm/estim/arx.cpp @ 639

#include "arx.h"
namespace bdm {

void ARX::bayes ( const vec &dt, const double w ) {
        double lnc;
                
        if ( frg < 1.0 ) {
                est.pow ( frg );
                
                //stabilize
                ldmat V0(eye(V.rows()));
                V0*=(1-frg)*1e-3;
                V += V0; //stabilization
                nu +=(1-frg)*(0.1 + V.rows() + 1* dimx + 2);
                
                // recompute loglikelihood of "prior"
                if ( evalll ) {
                        last_lognc = est.lognc();
                }
        }
        if (have_constant) {
                _dt.set_subvector(0,dt);
                V.opupdt ( _dt, w );
        } else {
                V.opupdt ( dt, w );
        }
        nu += w;

        // log(sqrt(2*pi)) = 0.91893853320467
        if ( evalll ) {
                lnc = est.lognc();
                ll = lnc - last_lognc - 0.91893853320467;
                last_lognc = lnc;
        }
}

double ARX::logpred ( const vec &dt ) const {
        egiw pred ( est );
        ldmat &V = pred._V();
        double &nu = pred._nu();

        double lll;

        if ( frg < 1.0 ) {
                pred.pow ( frg );
                lll = pred.lognc();
        } else//should be save: last_lognc is changed only by bayes;
                if ( evalll ) {
                        lll = last_lognc;
                } else {
                        lll = pred.lognc();
                }

        V.opupdt ( dt, 1.0 );
        nu += 1.0;
        // log(sqrt(2*pi)) = 0.91893853320467
        return pred.lognc() - lll - 0.91893853320467;
}

ARX* ARX::_copy_ ( ) const {
        ARX* Tmp = new ARX ( *this );
        return Tmp;
}

void ARX::set_statistics ( const BMEF* B0 ) {
        const ARX* A0 = dynamic_cast<const ARX*> ( B0 );

        bdm_assert_debug ( V.rows() == A0->V.rows(), "ARX::set_statistics Statistics  differ" );
        set_statistics ( A0->dimx, A0->V, A0->nu );
}

enorm<ldmat>* ARX::epredictor ( const vec &rgr ) const {
        int dim = dimx;//est.dimension();
        mat mu ( dim, V.rows() - dim );
        mat R ( dim, dim );

        enorm<ldmat>* tmp;
        tmp = new enorm<ldmat> ( );
        //TODO: too hackish
        if ( drv._dsize() > 0 ) {
        }

        est.mean_mat ( mu, R ); //mu =
        //correction for student-t  -- TODO check if correct!!
        //R*=nu/(nu-2);
        mat p_mu = mu.T() * rgr;        //the result is one column
        tmp->set_parameters ( p_mu.get_col ( 0 ), ldmat ( R ) );
        return tmp;
}

mlnorm<ldmat>* ARX::predictor ( ) const {
        int dim = est.dimension();
        
        mat mu ( dim, V.rows() - dim );
        mat R ( dim, dim );
        mlnorm<ldmat>* tmp;
        tmp = new mlnorm<ldmat> ( );

        est.mean_mat ( mu, R ); //mu =
        mu = mu.T();
        //correction for student-t  -- TODO check if correct!!

        if ( have_constant) { // constant term
                //Assume the constant term is the last one:
                tmp->set_parameters ( mu.get_cols ( 0, mu.cols() - 2 ), mu.get_col ( mu.cols() - 1 ), ldmat ( R ) );
        } else {
                tmp->set_parameters ( mu, zeros ( dim ), ldmat ( R ) );
        }
        return tmp;
}

mlstudent* ARX::predictor_student ( ) const {
        int dim = est.dimension();

        mat mu ( dim, V.rows() - dim );
        mat R ( dim, dim );
        mlstudent* tmp;
        tmp = new mlstudent ( );

        est.mean_mat ( mu, R ); //
        mu = mu.T();

        int xdim = dimx;
        int end = V._L().rows() - 1;
        ldmat Lam ( V._L() ( xdim, end, xdim, end ), V._D() ( xdim, end ) );  //exp val of R


        if ( have_constant) { // no constant term
                //Assume the constant term is the last one:
                if ( mu.cols() > 1 ) {
                        tmp->set_parameters ( mu.get_cols ( 0, mu.cols() - 2 ), mu.get_col ( mu.cols() - 1 ), ldmat ( R ), Lam );
                } else {
                        tmp->set_parameters ( mat ( dim, 0 ), mu.get_col ( mu.cols() - 1 ), ldmat ( R ), Lam );
                }
        } else {
                // no constant term
                tmp->set_parameters ( mu, zeros ( xdim ), ldmat ( R ), Lam );
        }
        return tmp;
}



/*! \brief Return the best structure
@param Eg a copy of GiW density that is being examined
@param Eg0 a copy of prior GiW density before estimation
@param Egll likelihood of the current Eg
@param indeces current indeces
\return best likelihood in the structure below the given one
*/
double egiw_bestbelow ( egiw Eg, egiw Eg0, double Egll, ivec &indeces ) { //parameter Eg is a copy!
        ldmat Vo = Eg._V(); //copy
        ldmat Vo0 = Eg._V(); //copy
        ldmat& Vp = Eg._V(); // pointer into Eg
        ldmat& Vp0 = Eg._V(); // pointer into Eg
        int end = Vp.rows() - 1;
        int i;
        mat Li;
        mat Li0;
        double maxll = Egll;
        double tmpll = Egll;
        double belll = Egll;

        ivec tmpindeces;
        ivec maxindeces = indeces;


        cout << "bb:(" << indeces << ") ll=" << Egll << endl;

        //try to remove only one rv
        for ( i = 0; i < end; i++ ) {
                //copy original
                Li = Vo._L();
                Li0 = Vo0._L();
                //remove stuff
                Li.del_col ( i + 1 );
                Li0.del_col ( i + 1 );
                Vp.ldform ( Li, Vo._D() );
                Vp0.ldform ( Li0, Vo0._D() );
                tmpll = Eg.lognc() - Eg0.lognc(); // likelihood is difference of norm. coefs.

                cout << "i=(" << i << ") ll=" << tmpll << endl;

                //
                if ( tmpll > Egll ) { //increase of the likelihood
                        tmpindeces = indeces;
                        tmpindeces.del ( i );
                        //search for a better match in this substructure
                        belll = egiw_bestbelow ( Eg, Eg0, tmpll, tmpindeces );
                        if ( belll > maxll ) { //better match found
                                maxll = belll;
                                maxindeces = tmpindeces;
                        }
                }
        }
        indeces = maxindeces;
        return maxll;
}

ivec ARX::structure_est ( egiw est0 ) {
        ivec ind = linspace ( 1, est.dimension() - 1 );
        egiw_bestbelow ( est, est0, est.lognc() - est0.lognc(), ind );
        return ind;
}



ivec ARX::structure_est_LT ( egiw est0 ) {
        //some stuff with beliefs etc.
//      ivec ind = bdm::straux1(V,nu, est0._V(), est0._nu());
        return ivec();//ind;
}

void ARX::from_setting ( const Setting &set ) {
        shared_ptr<RV> yrv = UI::build<RV> ( set, "rv", UI::compulsory );
        shared_ptr<RV> rrv = UI::build<RV> ( set, "rgr", UI::compulsory );
        int ylen = yrv->_dsize();
        // rgrlen - including constant!!!
        int rgrlen = rrv->_dsize();
        
        set_rv ( *yrv, *rrv );
        
        string opt;
        if ( UI::get(opt, set,  "options", UI::optional) ) {
                BM::set_options(opt);
        }
        int constant;
        if (!UI::get(constant, set, "constant", UI::optional)){
                have_constant=true;
        } else {
                have_constant=constant>0;
        }
        if (have_constant) {rgrlen++;_dt=ones(rgrlen+ylen);}

        //init
        mat V0;
        vec dV0;
        if (!UI::get(V0, set, "V0",UI::optional)){
                if ( !UI::get ( dV0, set, "dV0" ) )
                        dV0 = concat ( 1e-3 * ones ( ylen ), 1e-5 * ones ( rgrlen ) );
                V0 = diag ( dV0 );
        }
        double nu0;
        if ( !UI::get ( nu0, set, "nu0" ) )
                nu0 = rgrlen + ylen + 2;

        double frg;
        if ( !UI::get ( frg, set, "frg" ) )
                frg = 1.0;

        set_parameters ( frg );
        set_statistics ( ylen, V0, nu0 );
        
        //name results (for logging)
        shared_ptr<RV> rv_par=UI::build<RV>(set, "rv_param",UI::optional );
        if (!rv_par){
                est.set_rv ( RV ( "{theta r }", vec_2 ( ylen*rgrlen, ylen*ylen ) ) );
        } else {
                est.set_rv ( *rv_par );
        }
        validate();
}

}