Strong convergence rates in averaging principle for slow-fast McKean-Vlasov SPDEs (Q2669916)

The paper studies a class of McKean-Vlasov stochastic partial differential equations (SPDEs) with slow and fast time-scales, which are given by \begin{align*} &dX^{\varepsilon}_t = (A_1 (X_t^{\varepsilon},\mathcal{L}_{X^{\varepsilon}_t})+f(X_t^{\varepsilon},\mathcal{L}_{X^{\varepsilon}_t},Y^{\varepsilon}_t))dt + B_1 (X_t^{\varepsilon},\mathcal{L}_{X^{\varepsilon}_t}) d W^1_t, \\ &dY^{\varepsilon}_t = \frac{1}{\varepsilon}A_2 (X_t^{\varepsilon},\mathcal{L}_{X^{\varepsilon}_t},Y^{\varepsilon}_t)dt + \frac{1}{\sqrt{\varepsilon}}B_2(X_t^{\varepsilon},\mathcal{L}_{X^{\varepsilon}_t},Y^{\varepsilon}_t) d W^2_t, \\ &X_0^{\varepsilon}=x,\,Y^{\varepsilon}_0=y, \end{align*} where \((W^i_t)_{t\in[0,T]}\), \(i=1,2\), are independent cylindrical Wiener processes and \(\mathcal{L}_{X^{\varepsilon}_t}\) denotes the law of \(X^{\varepsilon}_t\). With this time scale, the variable \(X^{\varepsilon}_t\) is referred to as the slow component and \(Y^{\varepsilon}_t\) as the fast component. Under suitable assumptions on the coefficients, notably supposing only some monotonicity and coercivity conditions, the existence und unique of a variational solution \((X^{\varepsilon}_t, Y^{\varepsilon}_t)_{t\in [0,T]}\) to the aforementioned McKean-Vlasov SPDE is established, and it is shown that the slow component strongly converges to the solution of an associated averaged equation. Moreover, the rate of convergence is provided. To obtain these results, the authors use the variational approach and the classical Khasminskii time discretization. As examples, the main results are applied to a stochastic porous media type equation, a stochastic p-Laplace type equation and some McKean-Vlasov stochastic differential equations.