Exponential tail bounds for loop-erased random walk in two dimensions (Q606636)

Let \(S[0,\sigma_{n}]\) be a simple random walk on \(\mathbb Z^2\) started at the origin and stopped at \(\sigma_{n}\), the first time it exits \(B_{n}\), the ball of radius \(n\) with center at the origin, let \(L(S[0,\sigma_{n}])\) denote the loop-erasure of \(S[0,\sigma_{n}]\), and let \(M_{n}\) be the number of steps of \(L(S[0,\sigma_{n}])\). Further, let \(p_{n}\) denote the probability that \(L(S[0,\sigma_{n}])\) and an independent random walk started at the origin do not intersect up to leaving \(B_{n}\). First, the authors show that there exists \(0<C<\infty\) such that \(E[M_{n}^{k}]\leq C^{k}k!(n^2p_{n})^{k}\), \(k\in\mathbb N\), and \(E[M_{n}]\geq Cn^2p_{n}\). This implies that there exists \(0<c<\infty\) such that \(P(M_{n}>\lambda\), they prove that, for \(0<\alpha<4/5\), there exist \(0<C(\alpha),c(\alpha)<\infty\) such that \(P(M_{n}<\lambda^{-1}E[M_{n}])\leq C(\alpha)\exp(-c(\alpha)\lambda^{\alpha})\), \(\lambda>0\).