Optimize problem without objective function
CVX does not require an objective function. In such case, the solver finds whatever feasible solution “it feels like”, and CVX returns that to the user (or the problem is declared infeasible, or the optimization fails due to numerical issues).
Whether your alternating variables scheme converges to anything, let alone a global, or even local minimum of the original problem, is another matter, and is out of scope of the forum,.
