[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ScanTesla program - basically working!!



Original poster: "Sean Taylor" <sstaylor@xxxxxxxx>


Marco has suggested some "smart" ways to search the parameter space so
the program would zero in on the best parameters and avoid searching
obviously hopeless ones.  But I have not gotten that far yet.

I was going to ask about this, but someone beat me to it :-) Basically, I was going to suggest doing a gradient ascent algorithm to look for what direction the output voltage increases. Perhaps an even "grid" of test points first, then calculate a gradient field, then choose the best point and go in the best direction. Each subsequent step can have the size changed in order to not overshoot the peaks. Of course, in this situation, we know generally what the space looks like, and there won't be any hiding tiny maxima or minima: mainly large peaks and valleys.

Sean Taylor
Urbana, IL