[Coin-ipopt] Ipopt for nonlinear least-squares

[damien at khubla.com]

Herve,

when I was at another company we used IPOPT for nonlinear regression of
distillation column models, using the QN Hessian.  Some problems had
nonlinear constraints.  It works quite well, but it isn't as efficient as
a good LM on problems with no constraints.  Make sure you get your scaling
right on the x's for the x^2 calculation, once x < 1 progress drops off
dramatically as x^2 -> 0, and that one variable doesn't swamp all the
others, like x1 >> x2.

Damien

> Hi,
>
> I already use Ipopt (with success so far) for solving
> nonlinear optimization problems, and I am thinking to use it
> for nonlinear data fitting.
>
> So, in the case there are no constraints, I was thinking of
> giving to Ipopt the Gauss-Newton approximation to the
> hessian. I would be interested to know if it is valid to use
> Ipopt for this kind of minimization. Will Ipopt be as
> efficient as more specialized methods like
> Levenberg-Marquardt ? Are there special considerations about
> scaling of the problem with this approach ?
>
> Thanks,
> Hervé
>
>
>
> --------------------- ALICE SECURITE ENFANTS ---------------------
> Protégez vos enfants des dangers d'Internet en installant Sécurité
> Enfants, le contrôle parental d'Alice.
> http://www.aliceadsl.fr/securitepc/default_copa.asp
>
>
>
> _______________________________________________
> Coin-ipopt mailing list
> Coin-ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-ipopt
>