it appears that SINR and sinr are affine (linear) in the CVX variable ptilda, so use of rel_entr in the objective and constraint should be allowed. But if so, why do you have an error message in Disciplined convex programming error: Cannot perform the operation: {positive constant} ./ {convex}…
You can always use for loops. Your mathematical oprtmization problem formulation is not presented clearly enough for me to figure out whether there is a better way. If I were you, I;d worry about getting a correct model correctly implemented. If you have done that, you can then worry about efficiency improvements. if CVX calls the solver quickly enough, then don;t worry about getting rid of for loops.
You really need to make a clear and self-contained mathematical optimization problem statement. Show the complete code, and if you can, the input data which resulted in whatever error messages you are getting. And please don;t keep starting new threads on the same problem. When you show little bits here and there, and they keep changing, it doesn’t serve anyone’s interest. Maintaining a disciplined approach to your modeling coding, and data management is as or more valuable than mathematical optimization theory knowledge if your goal is to solve and implement solutions to actual problems.