Stochastic perturbation of sweeping process and a convergence result for an associated numerical scheme (Q550027)

The authors prove several results. First, they prove existence and uniqueness of \(L^{4}\) solutions for the sweeping process for stochastically perturbed differential inclusions in \({\mathbb R}^{n}\) \[ dX_{t}+N(C(t),X_{t})\ni f(t,X_{t})dt+\sigma (t,X_{t})dB_{t}, \quad X_{t}\in C(t),\quad X_{0}=x_{0}\text{ for all }t\in I\equiv [ 0,T], \] where \(N\) represents the proximal normal cone appropriately defined in the stochastic setting, \(C:I\rightarrow {\mathcal P}({\mathbb R}^{n})\), \(f,\sigma :I\times {\mathbb R}^{n}\rightarrow {\mathbb R}^{n}\), \(B_{t}\) is real-valued Brownian motion, and \(x_{0}\in {\mathbb R}^{n}\) is deterministic. Among other conditions, \(C\) is assumed to be uniform prox-regular valued and regular, and \(f,\sigma \) are Lipschitz with respect to the second variable and are bounded. The authors note that historically, the uniformly prox-regularity assumption was used to weaken convexity. The proof uses some familiar tools such as the contraction fixed point theorem, Itô's lemma, Doob's inequality and a Gronwall-type lemma. A stability result is also proven (dependence on \(\sigma \)). The authors also prove a convergence result for the associated Euler scheme. This extends a result of \textit{Y. Saisho}, [``Stochastic differential equations for multi-dimensional domain with reflecting boundary'', Probab. Theory Relat. Fields 74, 455--477 (1987; Zbl 0591.60049)] for constant \(C\). An application of these results to a model of crowd motion is also given.