Controlled Markov Chains with AVaR Criteria for Unbounded Costs

  • Kerem Ugurlu
    • Categories Applications - OR and Management Sciences Tags , , Short URL: https://optimization-online.org/?p=13641

    In this paper, we consider the control problem with the Average-Value-at-Risk (AVaR) criteria of the possibly unbounded $L^{1}$-costs in infinite horizon on a Markov Decision Process (MDP). With a suitable state aggregation and by choosing a priori a global variable $s$ heuristically, we show that there exist optimal policies for the infinite horizon problem. To our knowledge, this is the first work of deriving dynamic programming equations with $L^1$-unbounded costs via AVaR-operator.

    Article

    Download

    View Controlled Markov Chains with AVaR Criteria for Unbounded Costs