Toward a Scalable Upper Bound for a CVaR-LQ Problem

We study a linear-quadratic, optimal control problem on discrete, finite time horizon with distributional ambiguity, in which the cost is assessed via Conditional Value-at-Risk (CVaR). take steps toward deriving scalable dynamic programming approach to upper-bound value function for this problem. This program yields novel, tunable risk-averse policy, we compare existing state-of-the-art methods.