Positive operators, Z-operators, Lyapunov rank, and linear games on closed convex cones

Author/Creator ORCID

Date

2017-01-01

Department

Mathematics and Statistics

Program

Mathematics, Applied

Citation of Original Publication

Rights

This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
Distribution Rights granted to UMBC by the author.

Abstract

Given a closed convex cone K with dual H in a finite-dimensional real Hilbert space, the linear operator L is positive on K if L(K) is a subset of K, and a Z-operator on K if <L(x),s> is nonpositive for all x in K and s in H with <x,s> = 0. If both L and -L are Z-operators on K, then L is Lyapunov-like on K. These concepts generalize (respectively) the nonnegative, Z, and diagonal matrices. They appear in various fields including dynamical systems, optimization, economics, and game theory. Our contribution is to extend results for proper cones to general closed convex cones. We extend a result of Tam describing the dual of the cone of positive operators. We compute its largest linear subspace, and then use the exponential map to connect the positive and Z-operators. In particular, we show that both families share the same dimension and polyhedrality. Motivated by optimization considerations, the Lyapunov rank of K is defined as the dimension of the space of all Lyapunov-like operators on K. We extend this concept from proper to closed convex cones, and provide an algorithm to efficiently compute the Lyapunov rank in that case. We further show that the Lyapunov rank of K is the dimension of the Lie algebra of the automorphism group of K in the general setting. We then improve an existing upper bound for the Lyapunov rank. We introduce linear games on proper cones, and extend some results that Gowda and Ravindran formulated for self-dual cones. The more-general setting allows us to place linear games in the framework of conic programming.