Eigenvalues of convex processes and convergence properties of differential inclusions (Q1344274): Difference between revisions

The first part of the paper concerns characterizations of the extremal eigenvalues of a convex process \(G(.): K\to 2^{\mathbb{R}^n}\) (whose graph is a convex cone in \(\mathbb{R}^n\times \mathbb{R}^n\)) and the existence of corresponding eigenvectors. Extending previous results of \textit{R. T. Rockafellar} [``Monotone processes of convex and concave type'', Mem. Am. Math. Soc. 77 (1967; Zbl 0189.196)], the first main result states that if \(K^0\) denotes the polar of the convex cone \(K\subset \mathbb{R}^n\) then the number defined by: \[ \lambda_0:= \min_{\eta\in K^0} \max_{x\in K} \{\langle \eta, v\rangle/ \langle \eta, x\rangle; v\in G(x)\} \] is greater than the maximal eigenvalue of \(G(.)\) and coincides with it if the minimum is attained at \(\text{int } K^0\); if, in addition, a certain tangency condition is satisfied then there exists an eigenvector \(x_0\in K\backslash \{0\}\) of \(G(.)\) corresponding to \(\lambda_0\) (i.e., \(\lambda_0 x_0\in G(x_0)\)). The second part of the paper concerns asymptotic properties of solutions of the differential inclusion \(x'(t)\in G(x(t))\), \(x(0)= x_0\), \(x(t)\in K \forall t\geq 0\), where \(x_0\in K\) is the unique eigenvector of \(G(.)\) corresponding to \(\lambda_0\).