lyngby_lz_error

(export/lyngby/lyngby_lz_error.m)

Function Synopsis

[E, EY] = lyngby_lz_error(T, Y)	

Help text

 lyngby_lz_error      - Lange-Zeger residual signal (Quadratic error)

       function function [E, EY] = lyngby_lz_error(T, Y)	

       Input:  T    Target output (time domain)
               Y    Poisson filter prediction (time domain)

       Output:	EY   Error matrix for output units (T-Y)
               E    Qudratic error, normalized for the number of
                    examples

       The function computes this:   1/(Np) * (sum( (T-Y).^2 )).
       It expect that T and Y are in the time domain.

       If T has more rows than Y it will only use the last samples of
       T in the subtraction with Y, ie., it will assume that the
       first samples are not forwarded with lyngby_pois_forward.

       T and Y can be vectors or matrices.

       See also LYNGBY, LYNGBY_LZGS_MAIN, LYNGBY_LZIT_MAIN,
                LYNGBY_LZ_MASSCENT. 

 $Id: lyngby_lz_error.m,v 1.2 2003/04/28 15:08:34 fnielsen Exp $