Probability of Feasibility where the function is actually not possible to evaluate

[Hans Leonard van Brueggemann]

Dear all,

I have a question regarding the constraint optimization.

In the example provided here, the constraint optimization defines a non-feasible region while, at the same time, the objective function defines values for that region.

What if values for that regions are not defined at all? Let's say for example that's a region where I can't return a value to the optimizer because the objective function runs out of memory.

I tried to return an arbitrary value instead, but that will change the result of the evaluation, of course.

Is there another way to solve the problem?