Show
Ignore:
Timestamp:
06/10/10 21:40:09 (14 years ago)
Author:
mido
Message:

another update of doc - all bayesian models until bdm::MultiModel? finished
also MixEF::MixEF_options renamed just to MixEF::Options

Files:
1 modified

Legend:

Unmodified
Added
Removed

  • library/bdm/estim/arx.h

    r1064 r1077  
    2727Regression of the following kind: 
    2828\f[ 
    29 y_t = \theta_1 \psi_1 + \theta_2 + \psi_2 +\ldots + \theta_n \psi_n + r e_t 
     29y_t =     heta_1 \psi_1 +     heta_2 + \psi_2 +\ldots +     heta_n \psi_n + r e_t 
    3030\f] 
    31 where unknown parameters \c rv are \f$[\theta r]\f$, regression vector \f$\psi=\psi(y_{1:t},u_{1:t})\f$ is a known function of past outputs and exogeneous variables \f$u_t\f$. Distrubances \f$e_t\f$ are supposed to be normally distributed: 
     31where unknown parameters \c rv are \f$[    heta r]\f$, regression vector \f$\psi=\psi(y_{1:t},u_{1:t})\f$ is a known function of past outputs and exogeneous variables \f$u_t\f$. Distrubances \f$e_t\f$ are supposed to be normally distributed: 
    3232\f[ 
    3333e_t \sim \mathcal{N}(0,1). 
    … …  
    3939\include arx_simple.cpp 
    4040 
    41         \todo sort out constant terms - bayes should accept vec without additional 1s 
     41            odo sort out constant terms - bayes should accept vec without additional 1s 
    4242*/ 
    4343class ARX: public BMEF { 
    … …  
    132132    //!@} 
    133133 
    134     /*! UI for ARX estimator 
     134    /*! Create object from the following structure 
    135135 
    136136    \code 
    137137    class = 'ARX'; 
    138     yrv   = RV({names_of_dt} )                 // description of output variables 
    139     rgr   = RV({names_of_regressors}, [-1,-2]} // description of regressor variables 
    140     constant = 1;                              // 0/1 switch if the constant term is modelled or not 
    141  
    142     --- optional --- 
    143     prior = {class='egiw',...};                // Prior density, when given default is used instead 
    144     alternative = {class='egiw',...};          // Alternative density in stabilized estimation, when not given prior is used 
    145  
    146     frg = 1.0;                                 // forgetting, default frg=1.0 
    147  
    148     rv  = RV({names_of_parameters}}            // description of parametetr names 
    149                                                                                    // default: [""] 
     138    rgr = RV({'names',...},[sizes,...],[times,...]);   % description of regressor variables 
     139    --- optional fields ---     
     140    prior = configuration of bdm::egiw;                % any offspring of eqiw for prior density, bdm::egiw::from_setting 
     141    alternative = configuration of bdm::egiw;          % any offspring of eqiw for alternative density in stabilized estimation of prior density     
     142    constant = [];                                     % 0/1 switch if the constant term is modelled or not 
     143    --- inherited fields --- 
     144    bdm::BMEF::from_setting 
     145    \endcode 
     146    If the optional fields are not given, they will be filled as follows: 
     147    \code 
     148    prior = posterior;                                % when prior is not given the posterior is used (TODO it is unclear) 
     149    alternative = prior;                              % when alternative is not given the prior is used 
     150    constant = 1;                                     % constant term is modelled on default 
    150151    \endcode 
    151152    */ 
    … …  
    208209SHAREDPTR ( ARX ); 
    209210 
    210 /*! ARX model conditined by knowledge of the forgetting factor 
    211 \f[ f(\theta| d_1 \ldots d_t , \phi_t) \f] 
     211/*! \brief ARX model conditined by knowledge of the forgetting factor 
     212\f[ f(    heta| d_1 \ldots d_t , \phi_t) \f] 
    212213 
    213214The symbol \f$ \phi \f$ is assumed to be the last of the conditioning variables.