Rare-event analysis for extremal eigenvalues of white wishart matrices

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In this paper, we consider the extreme behavior of the extremal eigenvalues of white Wishart matrices, which plays an important role in multivariate analysis. In particular, we focus on the case when the dimension of the feature p is much larger than or comparable to the number of observations n, a common situation in modern data analysis. We provide asymptotic approximations and bounds for the tail probabilities of the extremal eigenvalues. Moreover, we construct efficient Monte Carlo simulation algorithms to compute the tail probabilities. Simulation results show that our method has the best performance among known approximation approaches, and furthermore provides an efficient and accurate way for evaluating the tail probabilities in practice.

Original languageEnglish (US)
Pages (from-to)1609-1637
Number of pages29
JournalAnnals of Statistics
Volume45
Issue number4
DOIs
StatePublished - Aug 1 2017

Fingerprint

Wishart Matrix
Tail Probability
Rare Events
Eigenvalue
Multivariate Analysis
Asymptotic Approximation
Data analysis
Extremes
Monte Carlo Simulation
Approximation
Eigenvalues
Rare events
Tail probability
Simulation

Keywords

  • Extremal eigenvalues
  • Importance sampling
  • Random matrix
  • β-Laguerre ensemble

Cite this

Rare-event analysis for extremal eigenvalues of white wishart matrices. / Jiang, Tiefeng; Leder, Kevin; Xu, Gongjun.

In: Annals of Statistics, Vol. 45, No. 4, 01.08.2017, p. 1609-1637.

Research output: Contribution to journalArticle

@article{2d43287f6f324a8595ed00928fa851e3,
title = "Rare-event analysis for extremal eigenvalues of white wishart matrices",
abstract = "In this paper, we consider the extreme behavior of the extremal eigenvalues of white Wishart matrices, which plays an important role in multivariate analysis. In particular, we focus on the case when the dimension of the feature p is much larger than or comparable to the number of observations n, a common situation in modern data analysis. We provide asymptotic approximations and bounds for the tail probabilities of the extremal eigenvalues. Moreover, we construct efficient Monte Carlo simulation algorithms to compute the tail probabilities. Simulation results show that our method has the best performance among known approximation approaches, and furthermore provides an efficient and accurate way for evaluating the tail probabilities in practice.",
keywords = "Extremal eigenvalues, Importance sampling, Random matrix, β-Laguerre ensemble",
author = "Tiefeng Jiang and Kevin Leder and Gongjun Xu",
year = "2017",
month = "8",
day = "1",
doi = "10.1214/16-AOS1502",
language = "English (US)",
volume = "45",
pages = "1609--1637",
journal = "Annals of Statistics",
issn = "0090-5364",
publisher = "Institute of Mathematical Statistics",
number = "4",

}

TY - JOUR

T1 - Rare-event analysis for extremal eigenvalues of white wishart matrices

AU - Jiang, Tiefeng

AU - Leder, Kevin

AU - Xu, Gongjun

PY - 2017/8/1

Y1 - 2017/8/1

N2 - In this paper, we consider the extreme behavior of the extremal eigenvalues of white Wishart matrices, which plays an important role in multivariate analysis. In particular, we focus on the case when the dimension of the feature p is much larger than or comparable to the number of observations n, a common situation in modern data analysis. We provide asymptotic approximations and bounds for the tail probabilities of the extremal eigenvalues. Moreover, we construct efficient Monte Carlo simulation algorithms to compute the tail probabilities. Simulation results show that our method has the best performance among known approximation approaches, and furthermore provides an efficient and accurate way for evaluating the tail probabilities in practice.

AB - In this paper, we consider the extreme behavior of the extremal eigenvalues of white Wishart matrices, which plays an important role in multivariate analysis. In particular, we focus on the case when the dimension of the feature p is much larger than or comparable to the number of observations n, a common situation in modern data analysis. We provide asymptotic approximations and bounds for the tail probabilities of the extremal eigenvalues. Moreover, we construct efficient Monte Carlo simulation algorithms to compute the tail probabilities. Simulation results show that our method has the best performance among known approximation approaches, and furthermore provides an efficient and accurate way for evaluating the tail probabilities in practice.

KW - Extremal eigenvalues

KW - Importance sampling

KW - Random matrix

KW - β-Laguerre ensemble

UR - http://www.scopus.com/inward/record.url?scp=85021410031&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021410031&partnerID=8YFLogxK

U2 - 10.1214/16-AOS1502

DO - 10.1214/16-AOS1502

M3 - Article

AN - SCOPUS:85021410031

VL - 45

SP - 1609

EP - 1637

JO - Annals of Statistics

JF - Annals of Statistics

SN - 0090-5364

IS - 4

ER -