A large-deviation result for the range of random walk and for the Wiener sausage (Q5944091)

Let \(\{S_n\}\) be a random walk on \(Z^{d}\), \(d\geq 2\), and let \(R_n\) be the number of different points among \(0,S_1,\ldots,S_{n-1}\). Then \[ \psi(x)=\lim_{n\to\infty} - {1\over n} \log P\{R_n\geq nx\} \] exists for all \(x\), and \(\psi(x)\) has the following properties \[ \psi(x)=0 \;\text{ for } x\leq\pi,\quad 0<\psi(x)<\infty \text{ for }\pi<x\leq 1,\quad \psi(x)=\infty \text{ for }x>1, \] \(x\mapsto\psi(x)\) is convex on \([0,1],\) \(x\mapsto\psi(x)\) is strictly increasing on \([\pi,1]\). Here \(\pi=P\{S_n\neq 0\) for all \(n\geq 1\}\) and \(R_n/n\to\pi\) a.s. Similar results hold also for the Wiener sausage (with drift). Let \(B(t)\) be a \(d\)-dimensional Brownian motion with constant drift, and for a fixed bounded set \(A\subset R^{d}\) let \(\lambda_t\) be the \(d\)-dimensional measure of the ``sausage'' \(\bigcup_{0\leq s<t}(B(s)+A)\). Then \(\varphi(x)=\lim_{t\to\infty}-{1\over t}\) if \(P(\lambda_t\geq tx)\) exists for \(x\geq 0\) and has similar properties as \(\psi\).