Passing variable array in fmincon

Hi all,

I am trying to do a non linear optimization for a function which takes parameters and variables that come from my dataset.

function [q] = negloglike(x, e, c, G, spelldur1, j2u, j2j)

% e, c, G, spelldur1, j2u, j2j are variables from my data - all 2263 X 1 double arrays

%%lambda0 = x(3)
%lambda1 = x(1)
%delta = x(2)

a = x(1)/x(2)
b1 = 1 + a.*G
b2 = G + a.*G
F_emp = b2./b1
tempp = x(2) + x(1) - x(1).*F_emp

q1 = -e.*(log(x(3))-log(x(3) + x(2)) + (1-c) .*log(tempp) ...
- spelldur1.*tempp + j2u.*log(x(2)) - j2u.*log(tempp) + j2j.*log(x(1)) + ...
 j2j.*log(1-F_emp) - j2j.*log(tempp)) - ...
(1-e).* (log(x(2)) - log(x(2) + x(3)) + (1-c).*log(x(3)) - spelldur1.* x(3))

q = sum(q1)
end

So, what I need is a negloglike as a function of just x, where then I can give initial values x = [0.1 0.1 0.1] and use fmincon to find the optimum values of parameter x.

Sorry if this is rudimentary. I am only a day old to Matlab! Thanks a lot in advance!

Yuvi

Answer 1

[Alan Weiss]

You are doing well for just a day with MATLAB! All you need to do is set your objective function as the following function handle:

 fun = @(x) negloglike(x, e, c, G, spelldur1, j2u, j2j)

The variables e, c, G, spelldur1, j2u, and j2j must already be in your workspace before you call this assignment. Then you can use fmincon to optimize fun. For more information, see Passing Extra Parameters.

Alan Weiss

MATLAB mathematical toolbox documentation

[yuvi]

Thanks Alan! This is exactly what I ended up doing. I created a m file for negloglike and then just called negloglike in my main script file.