Optimisation KKT formulating into CVX

Nonconvex
Dec 3, 2019
P
#1

Capture

I need to find PR where the rest of it (Ak,Bk,Ck,Dk,Ek,Fk,Gk,Hk, q1,K) are numbers. How to formulate these equations into CVX? Thanks

M
#2

The first constraint is not convex. For instance, choosing all the input constants to be 1, then the 2nd derivative of log(1+sqrt(p)/(p+1)) is negative for p = 2 and positive for p = 3.

Unless with the combination of all input values over the whole summation, it all nets together somehow to be a convex constraint. But certainly not convex in general.

And KKT really doesn’t come into play when trying to formulate this in CVX.