Cvx DCP error:affine.*convex

This is the expression of the objective function，the error is at L2，tV(n) is variable.

``````for n = 1:N
f(n) = 0;
P_n(n) = gamma*P(n);
numerator = P_n(n)*(uV(n)-u(n));
denominator = log(2)*(u(n)^2+P_n(n)*u(n));
L1 = numerator/denominator;
L2 = tV(n)*pow_p((tV(n)+P_n(n)),-1);
f(n) = f(n)-L1-(rel_entr(1,L2));
end
``````

Actually,the third term of f(n) is log(1+Pn/tV(n)).If expressed directly, there will be a log(convex) error.
So I chose the above expression method, but there is a mistake of affine.*convex.
I really appreciate it if you could help me with this question!thank you!

Have you proven this is a convex optimization problem? I will assume it is not unless/until you show otherwise.

This is a proved convex optimization problem after transformation. The problem we have now is that the expression does not know how to conform to CVX rules.Or maybe there are some built-in functions that I don’t know very well for this expression.

Please show us the transformation and the proof. if I had a dollar for every false claim of convexity on this forum, I’d have a triple digit number of dollars.

Please make very clear what are optimization variables, and what is input data, together with any pertinent facts (such as their sign) about the input data bearing on the convexity of the problem.

A few conic models of \log(1+1/x) are given in sections 5.2.4-5.2.10 of https://docs.mosek.com/modeling-cookbook/expo.html#relative-entropy

In particular

\log(1+1/x) = relentr(1+x,x) + relentr(x,1+x)