Averaging principle and normal deviations for multiscale stochastic systems (Q2021633)

This paper establishes asymptotic behavior for an inhomogeneous multiscale stochastic dynamical system with non-smooth coefficients, averaging principle and normal deviations for multiscale stochastic systems. The first few pages introduce all the necessary notation and terminology to understand the paper: considering a new Brownian motion; averaged equation involving new averaged drift coefficient; statement of the following inhomogeneous multiscale stochastic differential equations (SDE) in \(\mathbb{R}^{d_1+d_2}\): \begin{align*} \mathrm{d}X^{\epsilon}_{t}& =\alpha_\epsilon^{-2}b(X^{\epsilon}_{t},Y^{\epsilon}_{t})\mathrm{d}t+\beta_\epsilon^{-1}c(X^{\epsilon}_{t},Y^{\epsilon}_{t})\mathrm{d}t+\alpha_\epsilon^{-1}\sigma(X^{\epsilon}_{t},Y^{\epsilon}_{t})\mathrm{d}W^{1}_{t},\\ \mathrm{d}Y^{\epsilon}_{t}& =F(t,X^{\epsilon}_{t},Y^{\epsilon}_{t})\mathrm{d}t+\gamma_\epsilon^{-1}H(t,X^{\epsilon}_{t},Y^{\epsilon}_{t})\mathrm{d}t+G(t,Y^{\epsilon}_{t})\mathrm{d}W^{2}_{t},\\ X^{\epsilon}_{0} & =x \in \mathbb{R}^{d_{1}},\quad Y^{\epsilon}_{0}=y \in \mathbb{R}^{d_{2}}. \end{align*} The main results are stated in Section 2 by Theorems 2.1, 2.3, 2.5 and 2.7. The strong convergence in the averaging principle for SDE is first studied with non-smooth coefficients. The main focus is on the impact of noise on the averaging principle for the system. Two ingredients are crucial in the proof: Zvonkin's transformation and the Poisson equation in the whole space. First of all, due to the low regularity of the coefficients, Zvonkin's argument is used to transform the equations for \(Y^{\epsilon}_{t}\) and its average into new ones with better coefficients. Section 3 is devoted to the preparation of the main tools that will be used to prove the results. The proof of the aformentioned Theorem 2.1 (Strong convergence) in Section 4, Theorem 2.3 (Central limit theorem) in Section 5 and Theorem 2.5 (Central limit theorem: Regime 1) and 2.7 (Central limit theorem: Regime 2) in Section 6 are provided, respectively. Then, considering several independent topics which relate to the theory such as functional central limit theorem with homogenization and without homogenization are studied. Furthermore, the application of this theory to a Poisson equation will play an important role in the proof both for the strong convergence in the averaging principle and the central limit theorems. Using the technique of Poisson equation, some fluctuation estimates are derived. Then the strong convergence in the averaging principle of SDE in Section 4.2 by Zvonkin's transformation is proved.