I’m working on a robust optimisation problem wherein I first solve the nominal problem. I then check if the solution obtained from the nominal problem violates the robust constraints for a particular uncertain scenario (lets say the worst case realisation of the uncertain parameters) by solving a separation problem. If so, I add the violated inequality to the nominal problem and resolve. My question is, can I embed the separation problem within the original optimisation problem (just like in simplelazy.jl) and solve it or write a function to solve the separation problem outside the scope of the variables of the problem?

TIA