English

Spectral norm bounds for block Markov chain random matrices

Probability 2021-11-12 v1

Abstract

This paper quantifies the asymptotic order of the largest singular value of a centered random matrix built from the path of a Block Markov Chain (BMC). In a BMC there are nn labeled states, each state is associated to one of KK clusters, and the probability of a jump depends only on the clusters of the origin and destination. Given a path X0,X1,,XTnX_0, X_1, \ldots, X_{T_n} started from equilibrium, we construct a random matrix N^\hat{N} that records the number of transitions between each pair of states. We prove that if ω(n)=Tn=o(n2)\omega(n) = T_n = o(n^2), then N^E[N^]=ΩP(Tn/n)\| \hat{N} - \mathbb{E}[\hat{N}] \| = \Omega_{\mathbb{P}}(\sqrt{T_n/n}). We also prove that if Tn=Ω(nlnn)T_n = \Omega(n \ln{n}), then N^E[N^]=OP(Tn/n)\| \hat{N} - \mathbb{E}[\hat{N}] \| = O_{\mathbb{P}}(\sqrt{T_n/n}) as nn \to \infty; and if Tn=ω(n)T_n = \omega(n), a sparser regime, then N^ΓE[N^]=OP(Tn/n)\| \hat{N}_\Gamma - \mathbb{E}[\hat{N}] \| = O_{\mathbb{P}}(\sqrt{T_n/n}). Here, N^Γ\hat{N}_{\Gamma} is a regularization that zeroes out entries corresponding to jumps to and from most-often visited states. Together this establishes that the order is ΘP(Tn/n)\Theta_{\mathbb{P}}(\sqrt{T_n/n}) for BMCs.

Keywords

Cite

@article{arxiv.2111.06201,
  title  = {Spectral norm bounds for block Markov chain random matrices},
  author = {Jaron Sanders and Albert Senen-Cerda},
  journal= {arXiv preprint arXiv:2111.06201},
  year   = {2021}
}

Comments

30 pages, 1 figure

R2 v1 2026-06-24T07:35:01.513Z