Given the condition f<0, the following G(Rw) also proves to be a concave function, but it still gives an error when used. Is it because we need to write this G (Rw), which can recognize it as a concave function? What should I do?

Do you have a “constrictive” proof that the expression is concave, using operations allowed by CVX? Does the paper say that the expression can be formulated for CVX, CVXPY, or other DCP systems? Not every concave function can be entered in CVX.