Appl. Math. 0ptim. 9:1-24 (1982)
Applied Mathematics
and Optimization
Zero-sum Markov Games with Stopping and Impulsive Strategies
Lukasz Stettner
Institute of Mathematics, Polish Academyof Sciences,Sniadeckich8, 00-950 Warsaw, Poland
Communicated by A. V. Balakrishnan
Abstract. Three kinds of zero-sum Markov games with stopping and
impulsive strategies are considered. For these games we find the saddle point
strategies and prove that, the value of the game depends continuously on the
initial state.
In the paper the following three kinds of zero-sum, two-person Markov
games are considered: games with stopping, games with impulsive controls
and games with impulsive control and stopping. These games do not exhaust
the variety of the zero-sum games with stopping and impulsive strategies but
are typical in this theory.
The form of the associated cost functional depends on the kind of game.
For instance, for the game of the first type the functional is of the form:
Jx(%8)---- Ex{ fo~Aa e '~Sf(xs)ds+ x~<ae-~'~-Xt'l(X~)+ Xa<.~e-'~axP2(xa) }
If T and 8 are stopping times chosen by the first and second player
respectively, then the first player pays to the second one in average the total
amount equal to Jx(~-, 6).
A similar, but a more complicated functional is in the case of games with
impulsive controls, and games with impulsive control and stopping.
Under assumptions introduced by M. Robin in [8], we prove the
existence of the saddle point strategies for these games. We show also that the
values of the games depend continuously on the initial state. This way we
generalize Bismut results contained in [2] and solve a problem posed by M.
Robin in [8].
An analogous game to the one with impulsive controls was considered
independently and Under stronger assumptions by J. P. Lepeltier in [6].
0095-4616/82/0009-0001 $04.80
©1982 Springer-Verlag New York Inc.