A Dynamic Programming Algorithm for Finding an Optimal Sequence of Informative Measurements

. 2023 Jan 30;25(2):251. doi: 10.3390/e25020251

Algorithm 4 Generate base policies (one possibility based on an optimistic one-step lookahead)

1:
Input: $x_{k}^{'}, x_{k}$
2:
$w_{k} \sim p_{W_{k}}, m_{k} \sim p_{M_{k}}, m_{k + 1} \sim p_{M_{k + 1}}$
3:
$max_{u_{k} \in U_{k} (x_{k}^{'}, x_{k})} \{h_{k} (x_{k}, u_{k}, m_{k}) + max_{u_{k}^{'} \in U_{k}^{'} (x_{k}^{'})} max_{u_{k + 1} \in U_{k + 1} (v_{k}, f_{k})} \{h_{k + 1} (f_{k} (x_{k}, u_{k}, m_{k}), u_{k + 1}, m_{k + 1})\}\}$
4:
${\hat{μ}}_{k} (x_{k}^{'}, x_{k}) \leftarrow u_{k}^{*}, {\hat{μ}}_{k}^{'} (x_{k}^{'}) \leftarrow {u_{k}^{'}}^{*}$