Invariant convex sets and functions in Lie algebras (Q1924522): Difference between revisions

Let \(W\) be an invariant (under the adjoint action) convex cone in a Lie algebra \({\mathfrak g}\). The cone \(W\) is said to be elliptic if its interior consists of elliptic elements (elements \(X\) for which \(\text{ad }X\) is semisimple and has purely imaginary spectrum). The principal result of the paper is a characterization of the local convexity of a subset \(D\) of \(W\) and the invariant locally convex functions on \(D\) in terms of their restrictions to \(D\cap {\mathfrak t}\), where \({\mathfrak t}\) is a compactly embedded Cartan algebra. It is also shown that the restriction mapping from smooth invariant functions on the interior of \(W\) to Weyl group invariant functions on its intersection with \({\mathfrak t}\) is a bijection. In the earlier part of the paper the author sharpens known results about invariant cones to give a characterization of those Lie algebras which contain elliptic invariant cones.