stochastic flows with interaction and measure-valued processes随机流相互作用和measure-valued流程.pdf

IJMMS 2003:63, 3963–3977 PII. S0161171203301073 © Hindawi Publishing Corp. STOCHASTIC FLOWS WITH INTERACTION AND MEASURE-VALUED PROCESSES ANDREY A. DOROGOVTSEV Received 5 January 2003 We consider the new class of the Markov measure-valued stochastic processes with constant mass. We give the construction of such processes with the family of the probabilities which describe the motion of single particles. We also consider examples related to stochastic flows with the interactions and the local times for such processes. 2000 Mathematics Subject Classification: 60H15, 60H10. 1. Introduction. This paper is devoted to one type of the Markov measure- valued processes with constant mass. We consider the measure-valued pro- cesses {µt ; t ≥ 0} which have the following representation: −1 µt = µ ◦ft . (1.1) Here the random mapping ft on the state space depends on the initial state µ0 . In the representation (1.1), ft can describe the motion on the phase space which carries the mass µt . In such a model, the behavior of the trajectory which starts from a certain point depends on the whole mass distribution on the space. These processes are different from the well-known Fleming-Viot processes with constant mass. Consider, for example, the equation for the evolution of the discrete system of the particles, which was proposed in [3]. Example 1.1. Let a1,...,aN be positive numbers and let {w(t); t ≥ 0} be a Wiener process in Rd . Consider the following system of stochastic differential equations: N dx (t) = a h x (t),x