Some inequalities for reversible Markov chains and branching random walks via spectral optimization

TL;DR

This paper establishes new inequalities relating mixing times, hitting times, and intersection times of reversible Markov chains and branching random walks, revealing conditions for mean-field behavior and spectral optimization.

Contribution

It introduces novel inequalities connecting mixing and hitting times via spectral gap analysis, and resolves a conjecture on mean-field behavior in transitive reversible Markov chains.

Findings

Maximal expected hitting time is larger than mixing time by a universal constant.

Under transitivity, intersection time of two BRWs relates similarly to mixing time.

The inequality $t_{mix}^{( infty)} \, \le \, \frac{1}{gap} \log(et_{hit} \cdot gap)$ characterizes mean-field behavior.

Abstract

We present results relating mixing times to the intersection time of branching random walk (BRW) in which the logarithm of the expected number of particles grows at rate of the spectral-gap $\mathrm{gap}$ . This is a finite state space analog of a critical branching process. Namely, we show that the maximal expected hitting time of a state by such a BRW is up to a universal constant larger than the $L_{\infty}$ mixing-time, whereas under transitivity the same is true for the intersection time of two independent such BRWs. Using the same methodology, we show that for a sequence of reversible Markov chains, the $L_{\infty}$ mixing-times $t_{\mathrm{mix}}^{(\infty)} $ are of smaller order than the maximal hitting times $t_{\mathrm{hit}}$ iff the product of the spectral-gap and $t_{\mathrm{hit}}$ diverges, by establishing the inequality $t_{\mathrm{mix}}^{(\infty)} \le \frac{1}{\mathrm{gap}}\log(et_{\mathrm{hit}} \cdot \mathrm{gap}) $. This resolves a conjecture of Aldous and Fill (Reversible Markov chains and random walks on graphs, Open Problem 14.12) asserting that under transitivity the condition that $ t_{\mathrm{hit}} \gg \frac{1}{\mathrm{gap}} $ implies mean-field behavior for the coalescing time of coalescing random walks.