Sharp entrywise perturbation bounds for markov chains

Erik Thiede, Brian Van Koten, Jonathan Weare

Research output: Contribution to journalArticle

Abstract

For many Markov chains of practical interest, the invariant distribution is extremely sensitive to perturbations of some entries of the transition matrix, but insensitive to others; we give an example of such a chain, motivated by a problem in computational statistical physics. We have derived perturbation bounds on the relative error of the invariant distribution that reveal these variations in sensitivity. Our bounds are sharp, we do not impose any structural assumptions on the transition matrix or on the perturbation, and computing the bounds has the same complexity as computing the invariant distribution or computing other bounds in the literature. Moreover, our bounds have a simple interpretation in terms of hitting times, which can be used to draw intuitive but rigorous conclusions about the sensitivity of a chain to various types of perturbations.

Original languageEnglish (US)
Pages (from-to)917-941
Number of pages25
JournalSIAM Journal on Matrix Analysis and Applications
Volume36
Issue number3
DOIs
StatePublished - Jan 1 2015

Fingerprint

Perturbation Bound
Sharp Bound
Invariant Distribution
Markov chain
Transition Matrix
Perturbation
Computing
Hitting Time
Statistical Physics
Relative Error
Intuitive

Keywords

  • Condition numbers
  • Markov chains
  • Perturbation bounds
  • Sensitivity analysis
  • Stochastic matrices

ASJC Scopus subject areas

  • Analysis

Cite this

Sharp entrywise perturbation bounds for markov chains. / Thiede, Erik; Van Koten, Brian; Weare, Jonathan.

In: SIAM Journal on Matrix Analysis and Applications, Vol. 36, No. 3, 01.01.2015, p. 917-941.

Research output: Contribution to journalArticle

Thiede, Erik ; Van Koten, Brian ; Weare, Jonathan. / Sharp entrywise perturbation bounds for markov chains. In: SIAM Journal on Matrix Analysis and Applications. 2015 ; Vol. 36, No. 3. pp. 917-941.
@article{d6b530fb33404f0db9b714ac8e3c3dfe,
title = "Sharp entrywise perturbation bounds for markov chains",
abstract = "For many Markov chains of practical interest, the invariant distribution is extremely sensitive to perturbations of some entries of the transition matrix, but insensitive to others; we give an example of such a chain, motivated by a problem in computational statistical physics. We have derived perturbation bounds on the relative error of the invariant distribution that reveal these variations in sensitivity. Our bounds are sharp, we do not impose any structural assumptions on the transition matrix or on the perturbation, and computing the bounds has the same complexity as computing the invariant distribution or computing other bounds in the literature. Moreover, our bounds have a simple interpretation in terms of hitting times, which can be used to draw intuitive but rigorous conclusions about the sensitivity of a chain to various types of perturbations.",
keywords = "Condition numbers, Markov chains, Perturbation bounds, Sensitivity analysis, Stochastic matrices",
author = "Erik Thiede and {Van Koten}, Brian and Jonathan Weare",
year = "2015",
month = "1",
day = "1",
doi = "10.1137/140987900",
language = "English (US)",
volume = "36",
pages = "917--941",
journal = "SIAM Journal on Matrix Analysis and Applications",
issn = "0895-4798",
publisher = "Society for Industrial and Applied Mathematics Publications",
number = "3",

}

TY - JOUR

T1 - Sharp entrywise perturbation bounds for markov chains

AU - Thiede, Erik

AU - Van Koten, Brian

AU - Weare, Jonathan

PY - 2015/1/1

Y1 - 2015/1/1

N2 - For many Markov chains of practical interest, the invariant distribution is extremely sensitive to perturbations of some entries of the transition matrix, but insensitive to others; we give an example of such a chain, motivated by a problem in computational statistical physics. We have derived perturbation bounds on the relative error of the invariant distribution that reveal these variations in sensitivity. Our bounds are sharp, we do not impose any structural assumptions on the transition matrix or on the perturbation, and computing the bounds has the same complexity as computing the invariant distribution or computing other bounds in the literature. Moreover, our bounds have a simple interpretation in terms of hitting times, which can be used to draw intuitive but rigorous conclusions about the sensitivity of a chain to various types of perturbations.

AB - For many Markov chains of practical interest, the invariant distribution is extremely sensitive to perturbations of some entries of the transition matrix, but insensitive to others; we give an example of such a chain, motivated by a problem in computational statistical physics. We have derived perturbation bounds on the relative error of the invariant distribution that reveal these variations in sensitivity. Our bounds are sharp, we do not impose any structural assumptions on the transition matrix or on the perturbation, and computing the bounds has the same complexity as computing the invariant distribution or computing other bounds in the literature. Moreover, our bounds have a simple interpretation in terms of hitting times, which can be used to draw intuitive but rigorous conclusions about the sensitivity of a chain to various types of perturbations.

KW - Condition numbers

KW - Markov chains

KW - Perturbation bounds

KW - Sensitivity analysis

KW - Stochastic matrices

UR - http://www.scopus.com/inward/record.url?scp=84944528006&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84944528006&partnerID=8YFLogxK

U2 - 10.1137/140987900

DO - 10.1137/140987900

M3 - Article

VL - 36

SP - 917

EP - 941

JO - SIAM Journal on Matrix Analysis and Applications

JF - SIAM Journal on Matrix Analysis and Applications

SN - 0895-4798

IS - 3

ER -