Eigenfunctions and optimal orbits (Q801132)

The eigenvector-eigenvalue problem in the algebraic structure \(({\mathbb{R}},\max,+)\) is usually discussed and applied in the finite- dimensional case. Here, a continuous analogue is considered and related to a minimal-cost orbit problem. Given a function \(A(x,y):{\mathbb{R}}^ n\times {\mathbb{R}}^ n\to {\mathbb{R}}\), the problem consists in finding an eigenfunction \(f:{\mathbb{R}}^ n\to {\mathbb{R}}\) and an eigenvalue \(\lambda\in {\mathbb{R}}\) such that \[ \max_{y\in {\mathbb{R}}^ n}(A(x,y)+f(y))=\lambda +f(x) \] for all \(x\in {\mathbb{R}}^ n\). Under certain analytic assumptions on A and on the optimal solution \(\theta\) of the above maximization for each \(x\in {\mathbb{R}}^ n\), a formal solution is derived. If \(\theta\) has a fixed point on a certain compact set C then \(\lambda =\max \{A(z,z)| \quad z\in C\}.\) Sufficient conditions for the existence of such an eigenvalue and its corresponding eigenfunction are given. Further, an explicit solution is derived for strictly concave quadratic functions A(x,y).