Minimizing with fminbnd - how to use it for a matrix?

Hello,

I want to minimize the difference between theory value and measured value to calculate the parameter x.

Here's my function to minimize:

function difference = FUN_NAME(x,MEASURE,Qs,Qe,g)
omega = Qs/Qe-eps;
Rinf = (((sqrt(1-omega*g))-(sqrt(1-omega))))/(((sqrt(1-omega*g))+(sqrt(1-omega))));
gamma = 2*(sqrt(1-omega))*(sqrt(1-omega*g));
THEORY = (Rinf*(exp(gamma*x)-exp(-gamma*x)))/(exp(gamma*x)-Rinf^2*exp(-gamma*x));
difference = (THEORY-MEASURE)^2;

And everything works fine if I use fminbnd for any specific pixel of my MEASURE matrix (201x451), for example:

x=fminbnd('FUN_NAME',0,100,[],0.5189,2.1669,2.1669,0.83)

it gives me proper value of x. But I want to use it for my whole matrix and here's the problem. I'm trying to do it in this way:

for i=201
    for j=451
        x(i,j)=fminbnd('FUN_NAME',0,100,[],MEASURE(i,j),2.1669,2.1669,0.83)
    end
end

Unfortunately it gives me something like this:

wrong value 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0

....................

0 0 0 0 0 0 wrong value

There's always wrong value of x in the first and in the last pixel, the rest pixels are 0. Is there soemthing worng with this for loop?

Thanks for all the help.

Answer 1

[Matt J]

One problem I see is that your loop range is only the single pair i=201, j=451. Perhaps you really meant

 x=nan(201,451);

 for i=1:201
     for j=1:451

        fun=@(z) FUN_NAME(z, MEASURE(i,j),2.1669,2.1669,0.83);
        x(i,j)=fminbnd(fun,0,100);

     end
 end

Note also the more modern way of attaching fixed parameters to functions.

http://www.mathworks.com/help/optim/ug/passing-extra-parameters.html

[Afsafsa]

for i=1:201 for j=1:451

I guess I need to get some sleep. Thanks :D