What do you want to do with h(jd) ? Maybe you can multiply through the denominator when you use itm, rather than explicitly forming h(jd). Of course, for that to work, your optimization problem needs to be convex. You haven’t shown us what that problem is, so the forum readers don’t know. Your first task is to prove your optimization problem, whatever it is, is convex. if it is not convex, you can’t use CVX, but might be able to try YALMIP.