Hi, concerning the discussions about finding the upper/lower limit of some data some days ago (http://article.gmane.org/gmane.comp.python.scientific.user/8926), I've enhanced my script and added constraints to the 2nd derivative of the function. This gives much smoother results, at least for that special type of data Bill has to deal with. That additional constraint seems to help the minimizer find a solution and the number of function evaluations goes down significantly. I'm really impressed how well fmin_cobyla works. Regards, Christian