In convex optimization theory, if X is the optimization variable, the trace(X) should be an affine operation. However, in cvx, trace(X) is a convex operation. When I want to use the constraint trace(X) > gamma in cvx, how can I make it?
Trace is affine, so this shouldn’t be a problem. If it is, there is probably something you are not telling us, for example your X is actually something more complicated/nonlinear. If in doubt please show a reproducible example.
1 Like
hear is the problem.In cvx of matlab,The left side of constraint(3.17) is judged as convex.However,it should be affined in math,where H_um is constant matrix,the left side of constraint(3.17) only have one variable in Tr().
The left hand side is complex. An inequality between complex numbers makes no sense.
thankyou，I make a mistake