About the first eigenvector of the simple random walk killed upon exiting a large bounded domain
Quentin Berger,Nicolas Bouchot
2024-08-28
Abstract:In this article, we study a discrete version of a Dirichlet problem in an open bounded set $D\subset \mathbb{R}^d$, in dimension $d\geq 2$. More precisely, we consider the simple random walk on $\mathbb{Z}^d$, $d\geq 2$, killed upon exiting the large (bounded) domain $D_N :=(ND)\cap \mathbb{Z}^d$. We denote by $P_N$ the corresponding transition matrix and we study the properties of its ($L^2$-normalized) principal eigenvector $\phi_N$ -- one of our motivation is that the random walk ``conditioned to stay in $D_N$'' is a random walk among conductances $c_N(x,y) = \phi_N(x)\phi_N(y)$. With probabilistic arguments and under mild assumptions on the domain, we show that $\phi_N$ varies regularly, with a uniform control inside $D_N$. We derive several corollaries, among which a uniform convergence of $\phi_N$ to the first eigenfunction of the corresponding continuous Dirichlet problem. Our results may not be new, but our proofs use (simple) probabilistic ideas that could be helpful in other contexts.
Probability,Analysis of PDEs,Numerical Analysis