A note on constructing e-optimal policies for controlled markov jump models with unbounded characteristics

Abstract

This note continues a recent paper (cf. [3]) on controlled Markov jump models with unbounded jump and cost rates. It is shown that the time-discretization used in [3] also leads to a computationally attractive construction of E-optimal policies by discrete-time dynamic programming. An order of the accuracy of this construction is provided

Keywords:

• Controlled Markov jump model
• unbounded jump rates
• Bellman equation
• E-optimal policy
• time-discretization