Fluctuations in the occupation time of a site in the asymmetric simple exclusion process. (Q1879871): Difference between revisions

Consider the simple exclusion process \( (\eta_t)_{t\geq 0} \) evolving on the configuration space \( \Omega=\{0,1\}^{\mathbb{Z}^d} \). Here \( \eta(x) \) is \(1\) or \(0\) depending on whether the site \( x\) is occupied or not. Let the finite-range transition probability \( p:\mathbb{Z}^d\to[0,1] \) be given with mean \( m \) and symmetrization \( s(\cdot)=\frac{1}{2}(p(\cdot)+p(-\cdot)) \). The process is called symmetric if \( s=p \) and asymmetric if \( s\not=p \). The process is conservative in the sense that no particles are created or destroyed. For each density \( \rho\in[0,1], \) the Bernoulli product measure over \( \Omega \) with density \( \rho \) is invariant for the process, in the symmetric case it is also reversible. The article studies the asymmetric simple exclusion process in dimension \(d=1,2 \) under the Bernoulli product measure with density \( \rho \). The main object of interest is the variance of the occupation time functional \( \int_0^t(\eta_s(0)-\rho)ds, \) which measures the occupation at a fixed site different from the given density \( \rho \). The author proves the following two results: 1) For \( d=2 \) the limiting variance given by \(\lim_{t\to\infty}t^{-1}\mathbb{E}[\int_0^t(\eta_s(0)-\rho)ds]\) remains finite if \( \rho\not=\frac{1}{2} \) and \( m\not=0\). 2) For \( d=1 \) and density \( \frac{1}{2} \) and nonzero mean the limiting variance is infinite. In the introduction there are links to recent developments in this field. Before proving his two main results the author describes the usage of the duality of the simple exclusion process and the approximation by free particles.