[Ipopt] Lower Bound for Objective Function?

[Uwe Nowak]

Hello!

I am using Ipopt as part of an optimization step within a local search 
heuristic.

This means I have a locally optimal solution, then perturbate the 
solution (which usually leads to an better but infeasible solution) and 
restart the optimization.
The new solution is accepted, if it is better then the old one. In most 
runs that is not the case.

I would like to stop, if it is likely, that the new solution is worse 
then the old one. Then I do not really care about about the exact optimum.

I know I can run the algorithm with high tolerance and rerun it with 
tighter tolerance it the solution seems to be near optimal.

However I would like to know if there is a way to get a lower bound out 
of Ipopt?

The model is 2 times continuous differentiable with convex objective and 
reverse convex constraints. Analytic gradient and hessian are provided.

Thank you,
Uwe