A note on self-dual cones in Hilbert spaces (Q2928686)

A subset of an inner product space is called a pointed self-dual cone if it is closed with respect to addition and under scalar multiplication by nonnegative scalars, contains no line segment passing through the origin and equals its dual cone. The dual of a cone is the set of all linear functionals that assume nonnegative values over the cone. Let \(H_1\) and \(H_2\) be Hilbert spaces, let \(A\) be a bounded linear map from \(H_1\) into \(H_2\) and \(T\) be a bounded linear map in the other direction such that both \(A\) and \(T\) are nonnegative (relative to the respective pairs of cones) and also satisfy \(ATA=A\) and \(TAT=T\). Let \(R(A)\) be closed in \(H_2\). The author shows that there is a complete inner product \(\langle .,. \rangle\) on \(R(A)\) such that the induced norm is equivalent to the induced norm on \(H_2\) and that the image of the cone of \(H_1\) under \(A\) is a self-dual cone relative to the inner product \(\langle .,. \rangle\).