On large deviation theorems for Markov processes on a compact domain

京都工芸繊維大学

2009-03-13T05:37:01Z

2009-03-13T05:37:01Z

2000-02-25

Large deviation theorems of the Donsker-Varadhan type are studied. Those theorems for a family of stochastic processes converging to a Markov process have already been obtained by the present author. In this paper, these theorems are modified so that they cover the case where each process is killed on exiting a compact domain. The general theory of large deviations of such a type is mentioned at two levels; the state-space level and the path-space level, and applied to the study of the principal eigenvalue of the generator of a Markov process. It is shown that the principal eigenvalue converges as the probability law of the corresponding Markov process converges. As a typical example, the converging family of Markov processes in the homogenization problem is investigated.

2073712 bytes

application/pdf

京都工芸繊維大学　工芸学部研究報告　第４８巻　理工・欧文(1999) pp.11-22

http://hdl.handle.net/10212/1875

en

publisher

0110

Journal Article

Article