Geometric ergodicity and the spectral gap of non-reversible Markov chains

被引:0
|
作者
I. Kontoyiannis
S. P. Meyn
机构
[1] Athens University of Economics and Business,Department of Informatics
[2] University of Illinois at Urbana-Champaign,Department of Electrical and Computer Engineering and the Coordinated Sciences Laboratory
来源
关键词
Markov chain; Geometric ergodicity; Spectral theory; Stochastic Lyapunov function; Reversibility; Spectral gap; 60J05; 60J10; 37A30; 37A25;
D O I
暂无
中图分类号
学科分类号
摘要
We argue that the spectral theory of non-reversible Markov chains may often be more effectively cast within the framework of the naturally associated weighted-L∞ space \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^V}$$\end{document} , instead of the usual Hilbert space L2 = L2(π), where π is the invariant measure of the chain. This observation is, in part, based on the following results. A discrete-time Markov chain with values in a general state space is geometrically ergodic if and only if its transition kernel admits a spectral gap in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^V}$$\end{document} . If the chain is reversible, the same equivalence holds with L2 in place of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^V}$$\end{document} . In the absence of reversibility it fails: There are (necessarily non-reversible, geometrically ergodic) chains that admit a spectral gap in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^V}$$\end{document} but not in L2. Moreover, if a chain admits a spectral gap in L2, then for any \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${h\in L_2}$$\end{document} there exists a Lyapunov function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V_h\in L_1}$$\end{document} such that Vh dominates h and the chain admits a spectral gap in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^{V_h}}$$\end{document} . The relationship between the size of the spectral gap in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L_\infty^V}$$\end{document} or L2, and the rate at which the chain converges to equilibrium is also briefly discussed.
引用
收藏
页码:327 / 339
页数:12
相关论文
共 50 条
  • [1] Geometric ergodicity and the spectral gap of non-reversible Markov chains
    Kontoyiannis, I.
    Meyn, S. P.
    PROBABILITY THEORY AND RELATED FIELDS, 2012, 154 (1-2) : 327 - 339
  • [2] Spectral clustering for non-reversible Markov chains
    Fackeldey, K.
    Sikorski, A.
    Weber, M.
    COMPUTATIONAL & APPLIED MATHEMATICS, 2018, 37 (05): : 6376 - 6391
  • [3] Spectral clustering for non-reversible Markov chains
    K. Fackeldey
    A. Sikorski
    M. Weber
    Computational and Applied Mathematics, 2018, 37 : 6376 - 6391
  • [4] A Local Limit Theorem for a Family of Non-Reversible Markov Chains
    Elizabeth L. Wilmer
    Journal of Theoretical Probability, 2003, 16 : 751 - 770
  • [5] Hard-disk dipoles and non-reversible Markov chains
    Hoellmer, Philipp
    Maggs, A. C.
    Krauth, Werner
    JOURNAL OF CHEMICAL PHYSICS, 2022, 156 (08):
  • [6] Variational principles of hitting times for non-reversible Markov chains
    Huang, Lu-Jing
    Mao, Yong-Hua
    JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 2018, 468 (02) : 959 - 975
  • [7] A ROBUST SPECTRAL METHOD FOR FINDING LUMPINGS AND META STABLE STATES OF NON-REVERSIBLE MARKOV CHAINS
    Jacobi, Martin Nilsson
    ELECTRONIC TRANSACTIONS ON NUMERICAL ANALYSIS, 2010, 37 : 296 - 306
  • [8] A local limit theorem for a family of non-reversible Markov chains
    Wilmer, EL
    JOURNAL OF THEORETICAL PROBABILITY, 2003, 16 (03) : 751 - 770
  • [9] Analysis of non-reversible Markov chains via similarity orbits
    Choi, Michael C. H.
    Patie, Pierre
    COMBINATORICS PROBABILITY & COMPUTING, 2020, 29 (04): : 508 - 536
  • [10] Metropolis-Hastings reversiblizations of non-reversible Markov chains
    Choi, Michael C. H.
    STOCHASTIC PROCESSES AND THEIR APPLICATIONS, 2020, 130 (02) : 1041 - 1073