Coupling for Ornstein–Uhlenbeck Processes with Jumps
Feng-Yu Wang
DOI: https://doi.org/10.3150/10-bej308
IF: 1.5
2011-01-01
Bernoulli
Abstract:Consider the linear stochastic differential equation (SDE) on R(n):dX(t) = AX(t)dt + BdL(t),where A is a real n x n matrix, B is a real n x d real matrix and L(t) is a Levy process with Levy measure v on R(d). Assume that v(dz) >= rho(0)(z)dz for some rho(0) >= 0. If A <= 0, Rank(B) = n and integral({vertical bar z - z0} <= epsilon})rho(0)(z)(-1) dz < infinity holds for some z(0) is an element of R(d) and some epsilon > 0, then the associated Markov transition probability P(t)(x, dy) satisfiesparallel to P(t)(x, .) - P(t)(y, .)parallel to(var) <= C (1 + vertical bar x - y vertical bar)/root t, x, y is an element of R(d), t > 0,for some constant C > 0, which is sharp for large t and implies that the process has successful couplings. The Harnack inequality, ultracontractivity and the strong Feller property are also investigated for the (conditional) transition semigroup.