[Ipopt] infeasible problem

Andreas Waechter andreasw at watson.ibm.com
Tue Nov 18 23:49:46 EST 2008


I ran the feas problem with Ipopt, and Ipopt does not do a single 
iteration because the starting point you provide is such that the 
functions or their derivatives cannot be computed at that point. 
Therefore, Ipopt stops with an error message.

I also ran the infeas problem, and Ipopt converges to a point at which the 
constraints are violated by 4.469e-08.  For this run, I set 
the "bound_relax_factor" option to 0, so that Ipopt does not modify the 
bounds (by default, Ipopt relaxes the bounds a little bit, see the 
documentation of that option).  This final violation of the constraints is 
very small, and within the typical tolerances of NLP solvers and therefore 
declared feasible.  I'm not sure what you mean by Ipopt returning an 
incorrect result.


On Sat, 8 Nov 2008, Ali Baharev wrote:

> With the feas.mod Minos concludes that the problem is feasible, the
> solution vector is also given.
> I add variable bounds which are (or at least should be) consistent
> with the solution vector but Minos states the problem is now
> infeasible. Ipopt returns an incorrect result.
> Is it my fault? How can i identify what went wrong?
> The problem is known to be numerically challenging, the function is
> "flat" that may confuse the solver when checking the stopping
> criteria.
> Any help is greatly appreciated.
> Ali

More information about the Ipopt mailing list