A Markov property for set-indexed processes (Q1866056)

To define a Markov property for processes indexed by an uncountable partially ordered set is not a straightforward task for someone who has as declared goals to prove that: (i) all processes with independent increments possess this property; (ii) there exists a systematic procedure which allows us to construct a general process which enjoys this property; (iii) we can define a generator which completely characterizes the finite-dimensional distributions of the process. The paper represents a successful manner of approaching this problem. The authors consider the version of Markov processes that they named as ``set-Markov property''. It is less general than the Markov property introduced previously by Ivanoff and Merzbach but attains the three goals (i)--(iii) and implies sharp Markov property. It captures the essence of the Markov property in terms of the increments, allowing us to have a perfect analogy with the classical case, and the process is set-Markov if and only if it becomes Markov in the classical sense when it is transported by a ``flow''. The existence of a set-Markov process is considered and it is proved that one can construct such a process if we specify the laws characterizing some transitions.