Help, my code is always Unbounded

1

Help. I need help. I have been debugging for several days. But it is always Unbounded. The last constraint is ineffective, and the optimization variable is beyond the domain of definition. Thanks a lot.

cvx_clear
cvx_begin sdp
cvx_expert true
variable W(Nt,Nt) hermitian semidefinite
variable gama nonnegative
variable I nonnegative
maximize I
subject to
I-2log(sigma^2+sigmareal(trace((E_hvh+E_gvg)W))+gama)+real(s_tildei+trace(delta1’(W-Wi))+trace(delta2’*(gama-gamai)))<=0;

gama+square_pos(norm(E_gvhW,‘fro’))-real(pi+trace(deltap’(W-Wi)))<=0;

trace(W)<=Pt(t_p);

   cvx_end
   cvx_status
2

The advice at Debugging unbounded models - YALMIP also applies to CVX.

4

Thank you very much. Can having parameters that are too large or too small lead to this situation?

5

Yes, that is addressed in “3. Numerical nonsense” of the link in my previous post.

6

Thank you so much for your help!