Problem with Powell's Minimization


Yeraze
11-15-2005, 01:52 PM
I've implemented the 10.5 Powell's methods, but it seems my Minimization is actually a Maximization.

I'm working with very small numbers (0.0 to -0.0002), all negative, and when I minimize I still want to find the "lowest" number (eg, -1 < -0.0002), not the number closest to zero, which it what it seems to be doing. Is this a known "flaw" in the algorithm as presented?

Is this an easy thing to overcome?

Yeraze
11-15-2005, 02:13 PM
nevermind... My mistake.