[Ipopt] Defining the inverse of the hessian

Andreas Waechter awaechter.iems at gmail.com
Mon Sep 12 11:27:36 EDT 2011

Hi Ajo,

The step is computed from the solution of a linear system that is 
considerably more complicated than just the Hessian of the Lagrangian 
function (it involves the primal-dual Hessian of the barrier terms as 
well as the constraint Jacobians) - see Eq.(11) in the Ipopt 
implementation paper in Math Prog.  Having the inverse of the Hessian 
would not really help.  In addition, the inverse would usually be much 
more dense than the original (hopefully sparse) Hessian.

So, the answer to your questions is: No.

Hope this helps,


On 09/11/2011 01:04 AM, Ajo Fod wrote:
> Hello,
> I guess the computation of the solution of the equation involving the
> hessian is computationally the most expensive step in the optimization
> process.
> Is there a way of passing a function to calculate the inverse of the
> hessian instead of the hessian to the optimizer?
> Cheers,
> -Ajo.
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt

More information about the Ipopt mailing list