Conditionally Resampled Sliding-Window Count Kernels: Spectral-Gap Bounds and Poincaré Inequalities

2026-08-09

AI Summary

This work investigates the spectral gap and Poincaré inequality for the non-Markovian counting kernel induced by sliding-window count processes in stationary, finite-state, reversible Markov chains. By constructing a conditional resampling Markov kernel on the count vectors and combining path-space martingale oscillation inequalities with Dirichlet form comparison techniques, the authors establish, for the first time, a Poincaré inequality for this non-Markovian process that explicitly depends on the spectral gap of the underlying chain. The main contribution is a sharp lower bound on the spectral gap of the counting kernel, namely $\mathrm{Gap}(\mathcal{P}_n) = \Theta_p(1/n)$, which yields variance bounds for sliding-window count statistics and operator-norm concentration for matrix-valued empirical averages, thereby enabling effective control of global variance through local counting information.

Abstract

We study the conditionally resampled sliding-window count kernel associated with the empirical counts of length-$n$ windows from a stationary finite-state reversible Markov chain. Although the resulting count process is generally not Markov, its stationary one-step conditional law defines a genuine Markov kernel. For every fixed strictly positive reversible kernel \(P\) on a finite state space, we present a Poincaré inequality for the induced count kernel $\tP_n$ of length $n$. In other words, we derive the lower bound of the spectral gap $\Gap(\tP_n)$ of $\tP_n$ as \[ \Gap(\tP_n)\ge \frac{c(P)}{n}, \] where \(c(P)>0\) depends only on \(P\). The proof combines a martingale oscillation inequality for the stationary path law with a direct comparison of coordinate oscillations to the Dirichlet form of the count kernel. A linear statistic of the count vector gives the matching \(O(1/n)\) upper bound, so for every fixed strictly positive reversible \(P\) one has \(\Gap(\tP_n)=Θ_P(1/n)\). The resulting count-space Poincaré inequality yields a local-to-global variance bound for finite-window count statistics and, together with a general matrix-concentration principle, operator-norm concentration for matrix-valued empirical averages.

Problem

• spectral gap
• Poincaré inequality
• Markov chain
• sliding-window count
• reversible kernel

Innovation

• Poincaré inequality
• spectral gap
• sliding-window count kernel
• reversible Markov chain
• matrix concentration