What version of MATLAB are you using? Can you set the options Display to 'iter' (via optimset) and paste the output?
m!ke wrote: > I have an optimization problem with about 19 dimensions. For some > reason, using the medium-scale algorithm and fminunc, it does not > calculate a gradient at even the first iteration and declares that > there was no change in gradient as it bails. When I look at my > parameter file, it is clear that no finite-difference gradient calcs > have been made since the parameter file remains unchanged. However, > when I run it with a two-dimension version it works fine. > > Any suggestions on why the gradient might not be calcualted are mich > appreciated! > > peace, > m!ke