MATLAB: Not enough input arguments error using fminunc

Question

while using fminunc with a function file, I keep running into the error "not enough input arguments". I want to minimize my objective function with respect to three parameters, which I represented with the vector 'm'

my minimization code:

x0 = [0.2 1 0.01];
options = optimoptions(@fminunc,'Display','iter','Algorithm','quasi-newton');
[x,fval,exitflag,output,grad,hessian] = fminunc(@(m) myfun,x0,options)

my function file:

function logl = myfun(m);
rs = m(2)*100*(0*puse(16,1)) + m(3)*(0-euse(16,1))*eprc(16,1)/100;
(I have a bunch of code in between which includes m(1), but the above line is where the error occured)
end

To fix the error I tried fixing the minimization code to:

[x,fval,exitflag,output,grad,hessian] = fminunc(@(x) myfun,x0,options)
[x,fval,exitflag,output,grad,hessian] = fminunc(myfun,x0,options)

I also tried defining m in the minimization code:

m = [0 0 0];
x0 = [0.2 1 0.01];
options = optimoptions(@fminunc,'Display','iter','Algorithm','quasi-newton');
[x,fval,exitflag,output,grad,hessian] = fminunc(@(m) myfun,x0,options)

in the function file, I tried eliminating the input argument,

function logl = myfun;

and changing the order by which elements of 'm' appear

   rs = m(1)*100*(0*puse(16,1)) + m(2)*(0-euse(16,1))*eprc(16,1)/100;

The function file itself works as it is. Before, I had defined 'm' inside the function file (like m = [1 1 1]), but I think this caused the optimization routine to stay at that point.

Answer 1

I think that instead of

 fminunc(@(m) myfun,x0,options)

you should call

 fminunc(@myfun,x0,options)

or

 fminunc(@(m) myfun(m),x0,options)

Alan Weiss

MATLAB mathematical toolbox documentation