[Ipopt-tickets] [Ipopt] #45: Suggestion
Ipopt
coin-trac at coin-or.org
Sun Sep 30 17:27:22 EDT 2007
#45: Suggestion
------------------------------+---------------------------------------------
Reporter: kapurash | Owner: ipopt-team
Type: enhancement | Status: new
Priority: normal | Component: Ipopt
Version: 3.3 (C++ Version) | Severity: normal
Keywords: |
------------------------------+---------------------------------------------
Just a suggestion and I don't know enough about the inner workings of
Ipopt to know how feasible this is.
Basically there are problems where the hessian is large and dense but it
has structure so it's very easy and fast to compute the product of the
hessian and an arbitrary vector, both for the hessian of the objective
function and the constraints. Clearly in this situation it's possible to
not use second derivatives and ipopt does very well vs commercial solvers
I also use when both use only first derivatives. However, two of the
commercial solvers also have the option of using hessian vector products
instead of the hessians themselves and perform much better than ipopt
either with or without the hessian when used like this.
Just wondering if this is something the developers had considered and how
hard it would be. I'd be happy to help.
--
Ticket URL: <https://projects.coin-or.org/Ipopt/ticket/45>
Ipopt <http://projects.coin-or.org/Ipopt>
Interior-point optimizer for nonlinear programs.
More information about the Ipopt-tickets
mailing list