Random walks and the "Euclidean" association scheme in finite vector spaces

Abstract

In this paper, we provide an application to the random distance-$t$ walk in finite planes and derive asymptotic formulas (as $q\to \infty$) for the probability of return to start point after $\ell$ steps based on the "vertical" equidistribution of Kloosterman sums established by N. Katz. The application of these deep results from number theory allow a determination of the second order terms in the answers that simpler spectral gap/mixing rate methods do not. This work relies on a "Euclidean" association scheme studied in prior work of W.M.Kwok, E. Bannai, O. Shimabukuro and H. Tanaka. We also provide a self-contained discussion of the P-matrix and intersection numbers of this scheme for convenience in our application as well as a more explicit form for the intersection numbers in the planar case.