Originally posted by: veronicads

Hello!

I'm implementing a sort of Fourier-Motzkin elimination using the Python API and CPLEX 12.4. This means that i'm adding a huge number of constraints to my problem but most likely some of them are redundant. Is there a way to detect redundant constraints for the formulation? That is, is there a way I can check if a constraint cuts off a part of the feasible region?

Thanks for your help,

Veronica

This does not answer your question but you could define your constraints as lazy constraints so that they are only pulled into the model when they are violated. Maybe this is already good enough for you?

Originally posted by: veronicads

Is it possible to add lazy constraints also if my problem has only continuous variables? 

Yes, that should work. It will solve your problem as a MIP but that should not matter since the LP relaxation is then always integer feasible. Note that in the C-API you would have to explicitly declare your problem to be a MIP but since you are using Python the solve() method will do that for you.