The variable range is Hmin<=H(n)<=Hmax, however the cvx output is H(n)=0

@Mark_L_Stone again to bother you, :joy:. My Cvx solver is Mosek.

You haven’t told us what the value of Hmin is. Nevertheless, I’m not sure any variable values are meaningful when the problem is unbounded, as in this case. Even if variable values are supposed to be feasible within tolerance for an unbounded problem, the “optimal” (which is meaningless if unbounded), values of expressions may or may not be populated. Only declared variables are guaranteed to be populated with their optimal values.when the problem is solved (If it is solved).

Try following the advice at Debugging unbounded models - YALMIP . if these bounds are your attempt at “fixing” SCA, perhaps the advice in that link doesn’t really apply.