genfit Won't Converge

Try it again. Do a search in the Collab for:

genfit easy

TTFN

OK, I spent the day in the Collab reading. Seems like the main problem being solved is finding automated ways to generate the vector of derivitives. Lot of discussion about one of the easiest parts of this process.

What I get is "Can't converge to a solution" and I'm looking for a way to make MathCad give me more information. I've set the convergence tolerance to 1.0 (the max). I've picked my guess values at +/- infinity and most values in between.

One of many things that I don't understand is that the function turns red instantly after I change something. When genfit works on a dataset of any size it always takes a moment to decide if it wants to answer your question. With this it is instant. If I try linfit, pwrfit, expfit or lnfit it doesn't come close to matching the data, but it takes 3-5 seconds to decide on the constants. With genfit it is immediate.

Any ideas on how to get it to go down a different road to try to converge?

David

wHY NOT TRY EXCELS SOLVER?

I'm trying to solve for five constants. Excel just runs out of gas. I finally pulled my head out and took the time to understand the "Easy Genfit" thread on the Collab and it works wonderfully.

David

TK Solver Premium has an optimizer in it that is the latest version of solver> Newer and more robust than what is in excel

Cool, the version of TK I have is 10 years old and anything but robust. If I find anything that MathCad v11 won't handle I'll look into it.

David