Optimal control of point processes with noisy observations: the maximum principle (Q1599466)

The optimal control problem is investigated for point processes (same as solutions to SDEs with pure jumps), given observations under a Wiener noise. A maximum principle is established using backward stochastic differential equations (BSDEs) for diffusion processes with jumps. The system of equations is represented in a form corresponding to a fully observable case, but with a restricted set of admissible controls; then a BSDE method is used. Recently a similar approach was applied to other (diffusion) settings by one of the authors. Auxiliary adjoint fields satisfying backward stochastic integral PDEs are introduced: in a heuristic way, they are used to represent adjoint processes. An example with an explicit solution is given.