A polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise

Masaki Yamamoto, Shuji Kijima, Yasuko Matsui

Research output: Contribution to journalArticle

Abstract

We present a polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise. The Q-Ising, one of the generalized models of the Ising, arose in the context of Bayesian image restoration in statistical mechanics. We study the distribution of Q-Ising on a two-dimensional square lattice over n vertices, that is, we deal with a discrete state space {1,⋯,Q} n for a positive integer Q. Employing the Q-Ising (having a parameter β) as a prior distribution, and assuming a Gaussian noise (having another parameter α), a posterior is obtained from the Bayes' formula. Furthermore, we generalize it: the distribution of noise is not necessarily a Gaussian, but any vertex-independent noise. We first present a Gibbs sampler from our posterior, and also present a perfect sampler by defining a coupling via a monotone update function. Then, we show O(nlog∈n) mixing time of the Gibbs sampler for the generalized model under a condition that β is sufficiently small (whatever the distribution of noise is). In case of a Gaussian, we obtain another more natural condition for rapid mixing that α is sufficiently larger than β. Thereby, we show that the expected running time of our sampler is O(nlog∈n).

Original languageEnglish
Pages (from-to)392-408
Number of pages17
JournalJournal of Combinatorial Optimization
Volume22
Issue number3
DOIs
Publication statusPublished - Jan 1 2011

Fingerprint

Ising
Polynomial time
Polynomials
Statistical mechanics
Vertex of a graph
Image reconstruction
Gibbs Sampler
Q-integers
Bayes' Formula
Mixing Time
Image Restoration
Gaussian Noise
Prior distribution
Square Lattice
Statistical Mechanics
Monotone
State Space
Update
Generalise
Model

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Discrete Mathematics and Combinatorics
  • Control and Optimization
  • Computational Theory and Mathematics
  • Applied Mathematics

Cite this

A polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise. / Yamamoto, Masaki; Kijima, Shuji; Matsui, Yasuko.

In: Journal of Combinatorial Optimization, Vol. 22, No. 3, 01.01.2011, p. 392-408.

Research output: Contribution to journalArticle

@article{3b716c3da4994f9184db391772827d5c,
title = "A polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise",
abstract = "We present a polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise. The Q-Ising, one of the generalized models of the Ising, arose in the context of Bayesian image restoration in statistical mechanics. We study the distribution of Q-Ising on a two-dimensional square lattice over n vertices, that is, we deal with a discrete state space {1,⋯,Q} n for a positive integer Q. Employing the Q-Ising (having a parameter β) as a prior distribution, and assuming a Gaussian noise (having another parameter α), a posterior is obtained from the Bayes' formula. Furthermore, we generalize it: the distribution of noise is not necessarily a Gaussian, but any vertex-independent noise. We first present a Gibbs sampler from our posterior, and also present a perfect sampler by defining a coupling via a monotone update function. Then, we show O(nlog∈n) mixing time of the Gibbs sampler for the generalized model under a condition that β is sufficiently small (whatever the distribution of noise is). In case of a Gaussian, we obtain another more natural condition for rapid mixing that α is sufficiently larger than β. Thereby, we show that the expected running time of our sampler is O(nlog∈n).",
author = "Masaki Yamamoto and Shuji Kijima and Yasuko Matsui",
year = "2011",
month = "1",
day = "1",
doi = "10.1007/s10878-010-9309-7",
language = "English",
volume = "22",
pages = "392--408",
journal = "Journal of Combinatorial Optimization",
issn = "1382-6905",
publisher = "Springer Netherlands",
number = "3",

}

TY - JOUR

T1 - A polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise

AU - Yamamoto, Masaki

AU - Kijima, Shuji

AU - Matsui, Yasuko

PY - 2011/1/1

Y1 - 2011/1/1

N2 - We present a polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise. The Q-Ising, one of the generalized models of the Ising, arose in the context of Bayesian image restoration in statistical mechanics. We study the distribution of Q-Ising on a two-dimensional square lattice over n vertices, that is, we deal with a discrete state space {1,⋯,Q} n for a positive integer Q. Employing the Q-Ising (having a parameter β) as a prior distribution, and assuming a Gaussian noise (having another parameter α), a posterior is obtained from the Bayes' formula. Furthermore, we generalize it: the distribution of noise is not necessarily a Gaussian, but any vertex-independent noise. We first present a Gibbs sampler from our posterior, and also present a perfect sampler by defining a coupling via a monotone update function. Then, we show O(nlog∈n) mixing time of the Gibbs sampler for the generalized model under a condition that β is sufficiently small (whatever the distribution of noise is). In case of a Gaussian, we obtain another more natural condition for rapid mixing that α is sufficiently larger than β. Thereby, we show that the expected running time of our sampler is O(nlog∈n).

AB - We present a polynomial-time perfect sampler for the Q-Ising with a vertex-independent noise. The Q-Ising, one of the generalized models of the Ising, arose in the context of Bayesian image restoration in statistical mechanics. We study the distribution of Q-Ising on a two-dimensional square lattice over n vertices, that is, we deal with a discrete state space {1,⋯,Q} n for a positive integer Q. Employing the Q-Ising (having a parameter β) as a prior distribution, and assuming a Gaussian noise (having another parameter α), a posterior is obtained from the Bayes' formula. Furthermore, we generalize it: the distribution of noise is not necessarily a Gaussian, but any vertex-independent noise. We first present a Gibbs sampler from our posterior, and also present a perfect sampler by defining a coupling via a monotone update function. Then, we show O(nlog∈n) mixing time of the Gibbs sampler for the generalized model under a condition that β is sufficiently small (whatever the distribution of noise is). In case of a Gaussian, we obtain another more natural condition for rapid mixing that α is sufficiently larger than β. Thereby, we show that the expected running time of our sampler is O(nlog∈n).

UR - http://www.scopus.com/inward/record.url?scp=85028127494&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028127494&partnerID=8YFLogxK

U2 - 10.1007/s10878-010-9309-7

DO - 10.1007/s10878-010-9309-7

M3 - Article

VL - 22

SP - 392

EP - 408

JO - Journal of Combinatorial Optimization

JF - Journal of Combinatorial Optimization

SN - 1382-6905

IS - 3

ER -