[Ipopt] Does the Hessian need to be continuous?

Andreas Waechter andreasw at watson.ibm.com
Wed Nov 24 11:08:59 EST 2010


Hi Chia Weng,

Discontinuous second derivatives might not be too harmful, specifically if 
you don't expect the solution to be close to such discontinuities, the 
algorithm is much less sensitive to non-smoothness in the second 
derivatives than in the first derivatives.  I would just try what happens 
with your original smaller formulation.  If you have convergence issues, 
you might also just try to see what happens if you use the quasi-Newton 
Hessian approximation instead of your Hessians.

Andreas


On Wed, 24 Nov 2010, ChiaWeng Boon wrote:

> Hi,
>
> I have a step function (discontinuity ) in my 2nd derivatives.  I can
> remove the discontinuity but will increase the number of constraints
> substantially.  Is it necessary to remove the discontinuity?
>
>
> Chia Weng
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>
>



More information about the Ipopt mailing list