MATLAB: Fminbnd Error using .^ Matrix dimensions must agree.

Question

I'm trying to use fminbnd to find a scalar k that minimizes the difference between three sets of values, but receive the above error.

My Function:

function[f] = min_diff(k,sample1,sample2,sample3)
  alpha1 = 5 ./ (sample1 .^ k);
  alpha2 = 7 ./ (sample2 .^ k);
  alpha3 = 10 ./ (sample3 .^ k);
  mean1 = mean(alpha1);
  mean2 = mean(alpha2);
  mean3 = mean(alpha3);
  f = abs(mean1-mean2) + abs(mean1-mean3) + abs(mean2-mean3);

Code to run function:

[k_estimate,fval,exitflag] = fminbnd(@(k) min_diff(k,sample1,sample2,sample3),0,[],options);

Full Error Message:

Error using  .^ 
  Matrix dimensions must agree.
   Error in min_diff (line 3)
    alpha1 = 5 ./ (sample1 .^ k);
   Error in @(k)min_diff(k,sample1,sample2,sample3)
   Error in fminbnd (line 214)
   x= xf; fx = funfcn(x,varargin{:});

I'm using dot-operators everywhere. Does this have something to do with how fminbnd runs?

Answer 1

You are trying to use fminbnd with [] as your x2. That is giving you an empty k.

If you want 0 to be the lower bound and you do not want an upper bound then use inf for the upper bound.