mixpp: bdm::StateFromARX Class Reference

bdm::StateFromARX Class Reference

#include <kalman.h>

Inheritance diagram for bdm::StateFromARX:

bdm::StateSpace< sq_T > List of all members.

Public Member Functions

void connect_mlnorm (const mlnorm< chmat > &ml, RV &xrv, RV &urv)
void update_from (const mlnorm< chmat > &ml)
 fast function to update parameters from ml - not checked for compatibility!!
bool _have_constant () const
 access function

Protected Member Functions

void diagonal_part (mat &A, int r, int c, int n)
 function adds n diagonal elements from given starting point r,c

Protected Attributes

datalink_part th2A
 remember connection from theta ->A
datalink_part th2B
 remember connection from theta ->B
bool have_constant
 similar to ARX.have_constant

Detailed Description

State-Space representation of multivariate autoregressive model. The original model:

\[ y_t = heta [\ldots y_{t-k}, \ldots u_{t-l}, \ldots z_{t-m}]' + \Sigma^{-1/2} e_t \]

where $ k,l,m $ are maximum delayes of corresponding variables in the regressor.

The transformed state is:

\[ x_t = [y_{t} \ldots y_{t-k-1}, u_{t} \ldots u_{t-l-1}, z_{t} \ldots z_{t-m-1}]\]

The state accumulates all delayed values starting from time $ t $ .


Member Function Documentation

void bdm::StateFromARX::connect_mlnorm ( const mlnorm< chmat > &  ml,
RV xrv,
RV urv 
)

set up this object to match given mlnorm Note that state-space and common mpdf use different meaning of $ _t $ in $ u_t $. While mlnorm typically assumes that $ u_t \rightarrow y_t $ in state space it is $ u_{t-1} \rightarrow y_t $ For consequences in notation of internal variable xt see arx2statespace_notes.lyx.

temporary RV for connection to ml.rvc, since notation of xrv and ml.rvc does not match


The documentation for this class was generated from the following files:
Generated on 2 Dec 2013 for mixpp by  doxygen 1.4.7