Abstract:
Using a reduction to discrete time we prove two properties of jump Markov control
models with Borel state and action spaces: 1) the value of the model is analytic; 2) almost everywhere $\varepsilon$-optimal strategies (maybe non-Markovian) do exist.