clc,clear;
close all;
N=10;
eta=rand(1);
Pmax=rand(1);
B=rand(N)+1i*rand(N);
B=B*B';
C=rand(N)+1i*rand(N);
C=C*C';
u_t=0.1;
cvx_begin
variable W(N,N) semidefinite hermitian
maximize (real(trace(W*B)))
subject to
real(trace(W))<=Pmax
suqare_pos(norm(sqrtm(W+u_t*diag(diag(W)))*C,'fro'))<=eta
cvx_end
sqrtm
is not supported by CVX. If this is a convex optimization problem, you will need some other formulation.
In the future, before asking how to use a function in CVX, pleas look to see if it is listed in the CVX Users’ Guide.
Extension of MATLAB functions for CVX
New functions for CVX