Optimization least square approach

Hi everybody

I'm trying to find the optimum value of a vector of parameters to approximate an experimental set of data starting from a simplified vector x_0

x_0= [0.13084 ; 0 ; 0; 0; 0; 0 ]
 par=fminsearch(@(par)errHjk(par,frq_redHjk,csi_i,w0i,Hjk_red_exp),x_0,options);

where has been defined the fun :

function err = errHjk(par, frq_redHjk,csi_i,w0i, Hjk_red_exp)
 
Aj=par(1,1);
Bj=par(2,1);
Cj=par(3,1);
Dj=par(4,1);
Ej=par(5,1);
Fj=par(6,1);
 
 
omega_red= frq_redHjk*2*pi;
Hjki=(Aj+1i*Bj)./(-omega_red.^2+2i*w0i*csi_i+w0i^2) + Cj+1i*Dj +(Ej+1i*Fj)./omega_red.^2 ;
 
%Output the error 
 
error= (Hjki-Hjk_red_exp);
err = sum (real(error.^2)) + sum(imag(error.^2)) ;
 
end

instead of find the minimum value the error found by the program continues to increase

Do you know what is wrong with this code ?

Answer 1

[Matt J]

Shouldn't it be

err = sum (real(error).^2) + sum(imag(error).^2) ;

If so, then this is a linear least squares problem, and you should probably just use mldivide() instead of fminsearch.