Prev: Minimum gradient not reached ! please help
Next: simulation of self excited induction generator?
From: gpanterov Panterov on 10 Apr 2010 16:52
"Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message

> > I guess this specification poses a possible over parametrization. So I have played with imposing a constraint on sum(alpha)=30.
>
> Yes, that would be wise. Without this constraint. it's clear that if mu and alpha are solutions than c*alpha and mu/c are also solutions.

Yes ideally I would like to impose this constraint plus a sign restriction. I tried with fmincon and couldn't even get a simple set of solutions. With the unconstrained optimization I occasionally get what look like reasonable numbers but given how sensitive they are to everything I am afraid to use them.

Perhaps I could avoid fmincon by using some sort of a penalty in the likelihood with fminunc?
First  |  Prev  | 
Pages: 1 2
Prev: Minimum gradient not reached ! please help
Next: simulation of self excited induction generator?