The central limit theorem in Lipschitz domains (Q461407)

To catch the issue of the paper, one should consider a real-valued centered random walk \(S_n\), stopped at the first time when \(S_n\geq \varphi(n)\) for appropriate \(\varphi\) having Lipschitz property. A version of the central limit theorem for \(S_n\) is given for this stopped walk. More precisely, an optimal estimate is given for the difference between the prelimit transition probability and the transition density of the normalized Brownian motion stopped upon exit from the same domain. The optimal estimate is a generalization of the Edgeworth expansion. A higher dimensional analogue is also considered. The author operates in terms of homogenization theory and formulates the expansion results in terms of the Green function for an elliptic second-order differential operator \(\Delta-\frac{\partial}{\partial t}\). Tools such as the Harnack inequality and the parabolic boundary principle are applied to prove the results.