[Ipopt] step (infeasibility)

Juan Carlos Lopez Alfonso jc.atlantis at gmail.com
Thu Jun 14 02:59:49 EDT 2012


Hi all,

I have a big optimization problem, where I have included a volumetric
constraint. I have made a test of the gradient of this constraint
(varying epsilon) and I am sure that the implementation is correct.

I have reduced the size of my optimization problem (number of
tetraedras) and I have implemented it in matlab. In both cases, matlab
and IPOPT, when I use the interior-point algorithm the problem in some
cases finished with this message: "EXIT: Converged to a point of local
infeasibility. Problem may be infeasible", while in matlab when I use
other algorithm the same problem is solved correctly. I think that the
problem is the way to compute the step between interations, but i cant
find in IPOPT the way to play with this topic.

In my problem is complicated to start with a feasible point, but in
all cases I am sure that my problem has a unique optimal solution, but
for some initial points in the same problem the optimal solution is
obtained and in others not.

Please, Could you give me some help, suggestions, comments and ideas?
Thank you in advance
Juan Carlos


More information about the Ipopt mailing list