Selection of a Metropolis-Hastings simulation strategy via an entropy convergence criterion - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2004

Selection of a Metropolis-Hastings simulation strategy via an entropy convergence criterion

Résumé

Various convergence properties of the Metropolis-Hastings (MH) algorithm have been proved under conditions relative to the instrumental (proposal) distribution. To implement this algorithm, many different choices for the instrumental distribution can be made, associated in practice to unknown rates of convergence. In this paper we propose a simulation-based methodology to compare these rates of convergence, grounded on an entropy criterion computed from parallel (i.i.d.) simulated Markov chains coming from each candidate strategy. Theoretically, we give general conditions under which the successive densities of the MH algorithm satisfy adequate smoothness and tail properties, so that this entropy criterion can be estimated consistently using kernel density estimate and Monte Carlo integration. Examples are provided to illustrate this convergence criterion.
Fichier principal
Vignette du fichier
Chauveau_EntropyHM04.pdf (350.71 Ko) Télécharger le fichier

Dates et versions

hal-00019174 , version 1 (17-02-2006)
hal-00019174 , version 2 (10-05-2006)
hal-00019174 , version 3 (10-04-2007)

Identifiants

  • HAL Id : hal-00019174 , version 1

Citer

Didier Chauveau, Pierre Vandekerkhove. Selection of a Metropolis-Hastings simulation strategy via an entropy convergence criterion. 2004. ⟨hal-00019174v1⟩
228 Consultations
1243 Téléchargements

Partager

Gmail Facebook X LinkedIn More