[Ipopt] Without providing the Hessian matrix?

Andreas Waechter andreasw at watson.ibm.com
Fri Jan 28 16:34:39 EST 2011

Hi Lewis,

Using the Hessian approximation does not change Ipopt's termination 
criteria, so you should overall still get a similar degree of accuracy. 
However, it might take longer, and it might also be that the (looser) 
"acceptable" termination criteria are kicking in (see the documentation).

Also, for non-convex problems with more than one local solution or 
stationary point, the optimizer might take a different path and end up at 
a different point.



On Wed, 26 Jan 2011, Lewis I wrote:

> Hi
> I know that ipopt allows the users to perform optimization without the 
> Hessian matrix (by approximate it automatically). But, will it affect 
> the optimization ability of ipopt without providing the Hessian matrix? 
> such as feasible accuracy, optimization accuracy.
> Thank you.
> Regards,
> lewis
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt

More information about the Ipopt mailing list