```
cvx_solver mosek
cvx_save_prefs
cvx_begin
variable u
variable X(M+1,M+1) hermitian
maximize real(trace(G_u*X)+u*(h_u+1))
subject to
real(trace(G_e*X)+u*(h_e+1))-1>=0;
real(trace(G_e*X)+u*(h_e+1))-1<=0;
u>=0;
trace(E*X)-u>=0;
trace(E*X)-u<=0;
X == hermitian_semidefinite(M+1);
cvx_end
```

You haven’t provided input data, so the problem is not reproducible. You haven’t shown the solver or CVX output, so we don’t know what happened or what the issue is.

You have 2 equality constrains disguised as pairs of inequality constraints.

Perhaps the problem was reported infeasible? if so, all but section 1 of https://yalmip.github.io/debugginginfeasible also applies to CVX.

So? What are the results? What is your issue?

BTW, iit is better to express equality constrains with `==`

, rather than a pair of `<=, >=`

.

thanks,your answer is helpful