Multicriteria impulsive control of jump Markov processes (Q706378)

Multiple criteria optimization problems for jump Markov processes are considered. Let \(\omega = (T_0, x_0,{\hat \theta}_0, a_0, T_1, x_1, \ldots )\), where \(x_n\) are the (piece-wise constant) states, \(T_n\) are the jump epochs, \({\hat \theta}_n \in R_+^\infty\) are the intervals (delays) between \(T_n\) and the next moment of intervention denoted by \(a_n.\) The intervention takes place only if there have been no jumps in the interval \((T_n, T_n + {\hat \theta}_n).\) The sequence \(\{T_n, x_n, n=0, 1, \ldots\}\) is called a marked point process. A control strategy \(\pi\) is a sequence of measurable stochastic kernels \(\pi_n (T_0, x_0, {\hat \theta}_0, a_0, T_1, x_1, {\hat \theta}_1, a_1, \dots, T_n, x_n).\) The author considers Markov and stationary strategies. Let \(R(x, y)\) and \(R(x, a, y)\) be given functions. Here \(R(x, y)\) describes a ``natural'' jump and \(R(x,a, y)\) is associated with a jump due to an intervention \(a \in A.\) Let \[ \begin{align*}{ R(P^\pi) = E^\pi \biggl[ \sum_{n=1}^\infty & \biggl\{ r(x_{n-1}) (T_n - T_{n-1}) + R(x_{n-1}, x_n) I \{T_n - T_{n-1} < {\hat \theta}_{n-1}\} \cr & + R(x_{n-1}, a_{n-1}, x_n) I \{T_n - T_{n-1} = {\hat \theta}_{n-1}\} \biggr\} \biggr].}\end{align*} \] The goal of this paper is to construct a strategy providing the minimum \(R(P^\pi) \to \inf_\pi\) (\(\pi\) is a fixed control strategy). This theory is illustrated by an example of optimal control of epidemic transmission.