Distribution of the random walk conditioned on survival among quenched Bernoulli obstacles (Q2227714)

The paper under review deals with the localization properties of a simple random walk on \(\mathbb{Z}^d\) (\(d\geq 2\)) in the presence of independent Bernoulli obstacles. These obstacles are (i) hard, in the sense that the walk is killed upon visiting any of the obstacle, and (ii) quenched, with respect to the probability measure that is conditioned on the origin being in an infinite open cluster. It was previously established by two of the authors [\textit{J. Ding} and \textit{C. Xu}, Commun. Math. Phys. 375, No. 2, 949--1001 (2020; Zbl 1440.60083)] that the walk conditioned to survive for a large time \(n\) gets eventually localized in a ball whose radius is deterministic, explicit and of logarithmic order in \(n\), and whose center is random and dependent on the obstacle set. The present paper contains two remarkably sharp results. First, the authors prove that with probability converging to one, as \(n\) goes to infinity, the localization ball is (almost) free from obstacles. Then, they provide a local limit theorem for the position of the random walk conditioned to survive, both at the endpoint of the trajectory and inside the bulk. The corresponding limiting laws are the \(\ell^1\) and \(\ell^2\)-normalized eigenfunctions associated to the principal Dirichlet eigenvalue inside the localization ball. The proof builds upon previously established lemmas (such as the existence of a localization pocket with large eigenvalue inside which a low obstacle density region looks almost like a ball) and new intermediate results, including a ball-clearing lemma. The latter is obtained by estimating the impact of removing obstacles on the principal eigenvalue of the killed random walk inside a ball.