I am using CVX through Matlab to solve an optimization problem. I am implementing a scheduling framework that solves multiple instances for the optimization problem before making an allocation decision. Unfortunately these instances are solved slowly. I was wondering if there are any general hints about how to increase the solver speed.

My objective function is a Max of a set of convex functions and all my constraints (equality and inequality) are affine.

Is most of the time spent between cvx_begin and cvx_end (during the model building process) or from cvx_end until the solver completes? If it’s the latter then there really isn’t anything to be done about it but try other solvers.

Thanks for your reply. Could you advise on how I can change the solver and could you explain breifly what is the difference among the available solvers?

I would like to know which one you think is the lighter one.