Moment and almost sure Lyapunov exponents of mild solutions of stochastic evolution equations with variable delays via approximation approaches (Q700279)

The authors consider stochastic evolution equations of the form \[ \begin{aligned} dX_t &=(A X_t + A(t,X_t,X_{t-\tau(t)}))dt + B(t,X_t,X_{t-\tau(t)}) dW(t)\;\forall t \in [0,+\infty),\\ X_t &= \varphi(t),\;t \in [-h,0]\end{aligned} \tag{1} \] on the Hilbert space \(H\). The operator \(A\) is the infinitesimal generator of a certain \(C_0\)-semigroup \(S(t)\) over \(H\), and \(A(t,.,.)\) and \(B(t,.,.)\) are nonlinear mappings from \(H \times H\) to \(H\) and \(H \times H\) to \({\mathcal L}(K,H)\), respectively. \(W_t\) is a \(K\)-valued Wiener process and the initial function \(\varphi\) is such that \(\varphi(t)\) is \(\;F_0\)-measurable and \(\sup_{-h\leq r\leq 0} E |\varphi(r)|^2<\infty\). The lag \(\tau(t)\) is continuously differentiable with \(\tau'(t)\leq M,\;0\leq M < 1\). They also introduce approximating systems to (1) of the form \[ \begin{aligned} dX_t &=(A X_t + R(n)A(t,X_t,X_{t-\tau(t)}))dt + R(n)B(t,X_t,X_{t-\tau(t)}) dW(t)\;\forall t \in [0,+\infty),\\ X_0 &= R(n)x, X_t= R(n)\varphi(t),\;t \in [-h,0],\end{aligned}\tag{2} \] where \(n_0 \leq n \in \rho(A)\), \(\rho(A)\) the resolvent set of \(A\), and \(R(n) = nR(n,A)\). Results are cited that give existence and uniqueness of a mild solution of equation (1) and a strong solution of (2), as well as the convergence of the solutions of the systems (2) to the solution of (1). In addition an appropriate Itô-formula is provided, which can be applied to the strong solution of the approximating system. The purpose of the paper is to carry out a Lyapunov function programme to investigate exponential stability in mean square and almost surely of the mild solution of (1). Applications to stochastic evolution equations without delay and constant delay are provided. An extension of the main theorems to a fractional power type stability condition concludes the article.