Assume there is a parameter vector g = [x,y,xy,1].
The CWLS constraints are as follows:
g(3) = g(1)*g(2)
g(4) = 1
which are non-convex.
So what is the SDP constraints that will convexify it? Is it:
[G(3,3) G(1,1); G(2,2) 1] >= 0 (positive semidefinite, rank constraint is dropped)
G(4) == 1
Or can i put additional constraints that will shrink parameter space while not breaking convexity?