Asymptotic behavior of infinite dimensional stochastic differential equations by anticipative variation of constants formula (Q5955598)

Given two infinite-dimensional Hilbert spaces \(H\), \(K\), a generator \(A\) of a continuous semigroup in \(H\), a linear map \(B: H\to{\mathcal L}_2(K, H)\) and a Lipschitz function \(D: H\to{\mathcal L}_2(K, H)\), the authors study the long time behavior of the infinite-dimensional evolution equation \(dY_t= AY_t dt+ BY_t dW_t+ D(Y_t) dW_t\), \(t\geq s\), \(Y_s= x\in H\). New estimates on the operator associating adapted processes \(\psi\) to the solution \(\zeta\) of the above equation with \(D(Y_t)\) replaced by \(\psi\) and \(\zeta_0= 0\), are developed. They are based on a stochastic variation of constants formula established by \textit{S. Bonaccorsi} [Stochastic Anal. Appl. 17, No. 4, 509-528 (1999; Zbl 0947.60067)] for infinite-dimensional equations and involving a Skorokhod anticipative integral. Such estimates, under the assumption of mean-square stability for the linear part of the equation, lead directly to sufficient conditions for the exponential stability of the problem. It is proved that, under suitable conditions, the equation has a unique invariant measure that is strongly mixing. The paper is completed by some interesting examples.