English

Random reversible Markov matrices with tunable extremal eigenvalues

Probability 2015-09-09 v2

Abstract

Random sampling of large Markov matrices with a tunable spectral gap, a nonuniform stationary distribution, and a nondegenerate limiting empirical spectral distribution (ESD) is useful. Fix c>0c>0 and p>0p>0. Let AnA_n be the adjacency matrix of a random graph following G(n,p/n)\mathrm{G}(n, p/n), known as the Erd\H{o}s-R\'enyi distribution. Add c/nc/n to each entry of AnA_n and then normalize its rows. It is shown that the resulting Markov matrix has the desired properties. Its ESD weakly converges in probability to a symmetric nondegenerate distribution, and its extremal eigenvalues, other than 1, fall in [1/1+c/k,b][b,1/1+c/k][-1/\sqrt{1+c/k},-b]\cup [b,1/\sqrt{1+c/k}] for any 0<b<1/1+c0< b < 1/\sqrt{1+c}, where k=p+1k = \lfloor p \rfloor + 1. Thus, for p(0,1)p\in (0,1), the spectral gap tends to 11/1+c1-1/\sqrt{1+c}.

Keywords

Cite

@article{arxiv.1505.02086,
  title  = {Random reversible Markov matrices with tunable extremal eigenvalues},
  author = {Zhiyi Chi},
  journal= {arXiv preprint arXiv:1505.02086},
  year   = {2015}
}
R2 v1 2026-06-22T09:30:33.198Z