Some Ergodic Theorems for a Parabolic Anderson Model
Yong Liu,Feng Xia Yang
DOI: https://doi.org/10.1007/s10114-012-0672-4
2012-01-01
Acta Mathematica Sinica English Series
Abstract:In this paper, we study some ergodic theorems of a class of linear systems of interacting diffusions, which is a parabolic Anderson model. First, under the assumption that the transition kernel a = ( a ( i, j )) i,j ∈ S is doubly stochastic, we obtain the long-time convergence to an invariant probability measure ν h starting from a bounded a -harmonic function h based on self-duality property, and then we show the convergence to the invariant probability measure ν h holds for a broad class of initial distributions. Second, if ( a ( i, j )) i,j∈S is transient and symmetric, and the diffusion parameter c remains below a threshold, we are able to determine the set of extremal invariant probability measures with finite second moment. Finally, in the case that the transition kernel ( a ( i, j )) i,j ∈ S is doubly stochastic and satisfies Case I (see Case I in [Shiga, T.: An interacting system in population genetics. J. Math. Kyoto Univ. , 20 , 213–242 (1980)]), we show that this parabolic Anderson model locally dies out independent of the diffusion parameter c .