optimization with fminsearch error

Hi,

I am trying to optimize my function with fminsearch in order to find a minimum. My function is of a form:

    function v_max_diff = objFcn(ah25)
%OBJFCN
%    V_MAX_DIFF = OBJFCN(AH25)
%    This function was generated by the Symbolic Math Toolbox version 7.0.
%    28-Aug-2017 17:36:10
     v_max_diff = 5.0.^(-ah25).*6.25e6-1.25e6;

when I run fminsearch like:

    [x(:,i),fval(:,i)] = fminsearch(fun,1,options)

I get x=26.55 and fval=-1250000.

I want to minimize my fun v_max_diff by searching for an optimized ah25 so that the whole result approaches zero and not to fval of -125000 as in the present case.

I am new to optimization so any hint is more than appreciated.

thanks

Answer 1

[Alan Weiss]

Your function as written is monotone decreasing in the variable ah25, and asymptotically approaches -1.25e6 as ah25 gets large. I am not at all sure that this is what you intended your function to do, but that is what it does.

If, instead, you want to find a zero of your function, use fzero instead of fminsearch, because fminsearch looks for a minimum of your function, which, as I just told you, is approximately -1.25e6 for all large enough values of ah25.

Of course, your function is simple enough that you don't need to use fzero to find a root. Inspection shows that ah25 = 1 is the unique solution.

Alan Weiss

MATLAB mathematical toolbox documentation

[sensation]

Thanks Alan for this brief explanation. Actually I have many other points so fzero does the job:). Cheers