A limit theorem for symmetric statistics of Brownian particles (Q1807261)

Asymptotic distributions for a family of time-varying symmetric statistics formed from an infinite particle system are derived, and a representation for the limit is obtained in terms of multiple stochastic integrals. This family arises from a system of Brownian particles diffusing in \(R\) where the initial configuration is given via a Poisson point process on \(R\). Namely, let \(\{X_j,j\geq 1\}\) be the atoms of a Poisson point process on \(R\) with parameter \(\lambda \). Let \(X_j(t) = X_j+B_j(t)\) for \(j\geq 1\) and \(0\leq t\leq T\), where \(\{B_j(t)\}\) is a sequence of independent martingales starting at origin. The statistics to be studied is defined as follows \[ U_\lambda ^p(t,\Phi) = {1\over \lambda ^{p/2}} \sum \xi_{i_1}\cdots \xi_{i_p} \Phi (X_{i_1}(t),\ldots ,X_{i_p}(t)), \] where \(\Phi \) is a symmetric function and \(\{\xi_i\}\) is an independent Rademacher sequence. Then the limit for \(\lambda \to \infty \) is proved to satisfy a stochastic partial differential equation which is given explicitly as the main result of the paper.