I try to solve a complex problem as fast as possible using various optimization techniques. With the Genetic Algorithm, some problems arise: I have to combine equality, inequality and integer constraints (only part of the X vector), which is not possible with the ga function. The equality constraints deal with the non-integer variables. This far, I've used some techniques to work around this, but they didn't prove very succesful yet:
1) Let GA look for the integers only, In the evaluation of the fitness of individuals, call a linprog which optimizes the other variables and makes sure the equality constraints are ok. The linprog makes fitness evaluation really slow.
2) Replace the equality constraint by two inequality constr (Ax<=b and Ax>=b), but this renders the problem with a lot of constraints and no feasible solution is found in a 'normal' timeframe.
Are other techniques possible or does anyone have tips which might help to solve this complex problem with much constraints?