The order of convergence in the averaging principle for slow-fast systems of stochastic evolution equations in Hilbert spaces (Q6166352)

The equations \begin{align*} dU^\varepsilon&=[A_1U^\varepsilon+F(U^\varepsilon,V^\varepsilon)]\,dt+dW^1,&U^\varepsilon(0)&=u\in H \\ dV^\varepsilon&=\varepsilon^{-1}[A_2V^\varepsilon+G(V^\varepsilon)]\,dt+\varepsilon^{-1/2}dW^2,&V^\varepsilon(0)&=v\in K \end{align*} are considered in Hilbert spaces \(K\) and \(H\) for \(\varepsilon>0\). Here \(A_1\) is a selfadjoint negative operator on \(H\) with eigenvalues \(\{-\alpha_k\}\) satisfying \[ \sum_k\left[\alpha_k^{-\zeta}+\alpha_k^{n(\zeta+2\beta-1)-\zeta}\right]<\infty \] for some \(\zeta>0\), \(n\ge 2\) and \(\frac{1}{2n}<\beta<\frac{1}{3}\), \(A_2\) is an infinitesimal generator of a \(C_0\)-semigroup \(e^{A_2t}\) on \(K\) satisfying \[ \|e^{A_2t}\|\le e^{-\lambda t},\qquad t\ge 0, \] for some \(\lambda>0\), \(F:H\times K\to H\) and \(G:K\to K\) are Lipschitz continuous mappings and the Lipschitz constant \(L_G\) for the mapping \(G\) satisfies \(L_G<\lambda\), \[ \|e^{A_1t}Q_1^{1/2}\|_{HS(H)}+\|e^{A_2t}Q_2^{1/2}\|_{HS(K)}\le C(t\land 1)^{-\gamma},\qquad t\ge 0, \] where \(HS\) stands for the Hilbert-Schmidt norm, \(W^1\), \(W^2\) are independent Wiener processes on \(H\) and \(K\) with covariance operators \(Q_1\) and \(Q_2\), respectively and \(Q_2\) is invertible with a bounded inverse operator on \(K\). It is proved that the equation \[ dV=[A_2V+G(V)]\,dt+dW^2 \] has a unique invariant probability measure \(\mu\), the equation \[ dU=[A_1U+\overline F(U)]\,dt+dW^1,\qquad U(0)=u\in H, \] where \[ \overline F(z)=\int_KF(z,\cdot)\,d\mu,\qquad z\in H \] is well posed on \(H\) and \[ \mathbb E\left[\sup_{t\in[0,T]}|U^\varepsilon_t-U_t|^2\right]\le C\varepsilon,\qquad\varepsilon>0, \] where the constant \(C\) depends on \(T\) and on the norms of \(u\) and \(v\).