Asymptotic behavior of the local time of a recurrent random walk (Q793441)

Let \(\{X_ k\), \(k\geq 1\}\) be a sequence of i.i.d. nondegenerate random variables taking values in the one dimensional integer lattice \({\mathbb{Z}}\). Put \(S_ 0=0\), \(S_ n=\sum^{n}_{k=1}X_ k\), \(n\geq 1\), and define the local time at x before time n to be the number of visits to x before time n: \(L_ n(x)=\sum^{n-1}_{j=0}1_{(x)}(S_ j)\), \(x\in {\mathbb{Z}}\), \(n\geq 1\). This paper establishes the rate of growth of the large values of \(L_ n(x)\) for fixed x and also for \(\sup_{x} L_ n(x).\) Let X be a random variable with the same distribution as \(X_ 1\) and F its distribution function. For \(x>0\) define \(G(x)=P[| X|>x]\), \(K(x)=x^{-2}\int_{| y| \leq x}y^ 2dF(y)\). Under the assumption \(\lim \sup_{x\to \infty}(G(x)/K(x))<1\) and \(EX=0\), the following theorems are proved: Theorem 1. There exists \(\theta_ 1\in(0,\infty)\) such that for all \(x\in {\mathbb{Z}} \lim \sup_{n\to \infty}(c_ n/n)L_ n(x)=\theta_ 1\) a.s. Theorem 2. Given \(\epsilon>0\), there exists \(\delta>0\) such that \(\lim \sup_{n\to \infty}\sup_{| x-y| \leq \delta c_ n}(c_ n/n)| L_ n(x)-L_ n(y)|<\epsilon\) a.s. Theorem 3. There exists \(\theta_ 2\in(0,\infty)\) such that \(\lim \sup_{n\to \infty}\sup_{x}(c_ n/n)L_ n(x)=\theta_ 2\) a.s. The proofs base on an estimate \(\sum^{\infty}_{n=0}| P[S_ n=z]-P[S_ n=z+x]|\leq C/(| x| Q(x))\), where C is a positive constant, and \(Q(x)=G(x)+K(x)\).