A Fast Metropolis-Hastings Method for Generating Random Correlation Matrices

Publikation: Bidrag til bog/antologi/rapportBidrag til bog/antologiForskningfagfællebedømt

  • Irene Córdoba
  • Gherardo Varando
  • Concha Bielza
  • Pedro Larrañaga
We propose a novel Metropolis-Hastings algorithm to sample uniformly from the space of correlation matrices. Existing methods in the literature are based on elaborated representations of a correlation matrix, or on complex parametrizations of it. By contrast, our method is intuitive and simple, based the classical Cholesky factorization of a positive definite matrix and Markov chain Monte Carlo theory. We perform a detailed convergence analysis of the resulting Markov chain, and show how it benefits from fast convergence, both theoretically and empirically. Furthermore, in numerical experiments our algorithm is shown to be significantly faster than the current alternative approaches, thanks to its simple yet principled approach.
OriginalsprogEngelsk
TitelDistributions and operators Gerd Grubb : 19th International Conference Madrid, Spain, November 21–23, 2018
RedaktørerHujun Yin, David Camacho, Paulo Novais, Antonio J. Tallón-Ballesteros
Antal sider8
Vol/bind1
ForlagSpringer
Publikationsdato2018
Sider117-124
ISBN (Trykt)9783030034924
DOI
StatusUdgivet - 2018
Begivenhed19th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2018 - Madrid, Spanien
Varighed: 21 nov. 201823 nov. 2018

Konference

Konference19th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2018
LandSpanien
ByMadrid
Periode21/11/201823/11/2018
NavnLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Vol/bind11314 LNCS

ID: 215036886