Ergodic problem for the Hamilton-Jacobi-Bellman equation. II (Q1386843)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Ergodic problem for the Hamilton-Jacobi-Bellman equation. II
scientific article

    Statements

    Ergodic problem for the Hamilton-Jacobi-Bellman equation. II (English)
    0 references
    0 references
    9 December 1998
    0 references
    Consider the O.D.E. \(\dot x_\alpha=b(x_\alpha(t), \alpha(t))\), \(x_\alpha(0)=x\in \overline\Omega\), \(x_\alpha(t)\in \overline\Omega\) for all \(t\geq 0\), with periodic boundary conditions if \(\overline\Omega\) is a \(n\)-dimensional torus (or with Neumann boundary condition if \(\Omega\) is a bounded open subset in \(R^n\)). In this setting the control variable \(\alpha\) is a measurable function from \([0,\infty)\) to some metric space \(A\). Let \(u_\lambda(x)\) be the value function \(u_\lambda(x)=\text{ inf}_\alpha \int_0^\infty e^{-\lambda s} f(x_\alpha(s),\alpha(s)) ds\), where \(f\) is Lipschitz on \(\overline\Omega\times A\). It is proved that if any point \(x\in \overline\Omega\) is exactly uniformly controllable, then there exists a constant \(d_f\) such that \(\text{ lim}_{\lambda \downarrow 0}\lambda u_\lambda(x)=d_f\) uniformly in \(x\in\overline\Omega\). Similar results are given when the exact controllability assumption is replaced by an approximate controllability condition. The finite horizon problem is also studied. The same kind of result is proved when \(x\) is controllable to any point \(y\in Z_1\), where \(Z_1\) is a nonempty closed invariant subset strictly contained in \(\overline\Omega\). This paper is the continuation of [\textit{M. Arisawa}, Ann. Inst. Henri Poincaré, Anal. Non Linéaire 14, No. 4, 415-438 (1997; Zbl 0892.49015)].
    0 references
    0 references
    0 references
    0 references
    0 references
    Hamilton-Jacobi equations
    0 references
    ergodic attractor
    0 references
    0 references