| Titre : | MCMC via the Metropolis-Hastings algorithm and the Gibbs sampler. |
| Auteurs : | Samah GORMI, Auteur ; Mouloud Cherfaoui, Directeur de thèse |
| Type de document : | Monographie imprimée |
| Editeur : | Biskra [Algérie] : Faculté des Sciences Exactes et des Sciences de la Nature et de la Vie, Université Mohamed Khider, 2026 |
| Format : | 1vol.(37.p) / 30 CM |
| Langues: | Anglais |
| Mots-clés: | Bayesian estimation, Markov Chain Monte Carlo (MCMC), Metropolis–Hastings algorithm, Gibbs sampling, Simulation, MSE. |
| Résumé : |
This work aims to analyze several variants of the Metropolis-Hastings algorithm, particularly when this
algorithm is equipped with the Gibbs sampler. Simulation studies conducted on examples of parametric estimation via Bayesian inference have shown that the M-H algorithm with the Gibbs sampler offers us better results in terms of the speed of convergence to the target distribution. Furthermore, the quality of the estimates provided and the speed of convergence remain reasonable even when the parameter to be estimated is high-dimensional, unlike the Metropolis-Hastings algorithm, which can suffer from slowness in such cases. |
| Sommaire : |
Dedication i
Thanks ii Table of contents ii List of Tables v List of Figures vi Introduction 1 1 Introduction and overview of Bayesian statistics 3 1.1 Bayesian Estimation and Markov chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.1 General concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.2 Monte Carlo Method [7, 8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1.3 Markov Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2 Fundamental concepts of Bayesian statistics . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.1 Bayesian inference principle [8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.2 Likelihood Function [9, 8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.3 The Prior Distribution [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.4 The Bayes Theorem [16, 10] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2.5 The posterior law [8, 9, 5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2.6 The Bayesian Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.3 Examples of Bayesian estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 MCMC methods and their applications in Bayesian Statistics 12 2.1 Metropolis-Hastings (M-H) Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2 Variants of M-H Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 iii 2.2.1 Metropolis Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.2.2 Independent Metropolis-Hastings Algorithm . . . . . . . . . . . . . . . . . . . . . . 15 2.2.3 Dependent Metropolis-Hastings algorithm . . . . . . . . . . . . . . . . . . . . . . . 16 2.2.4 One-Variable-at-a-time Metropolis-Hastings Algorithm [3, 2] . . . . . . . . . . . . 16 2.3 Gibbs sampler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4 Difficulties and Challenges of MCMC [12] . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3 Comparative Study of the M-H Algorithm without and with Gibbs Sampling 21 3.1 Validation of the algorithms on a model with one parameter . . . . . . . . . . . . . . . . . 21 3.1.1 Application of Metropolis-Hastings Algorithm . . . . . . . . . . . . . . . . . . . . . 21 3.1.2 Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.1.3 Application of Gibbs Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2 Model with two Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2.1 Application of M-H Algorithm without Gibbs . . . . . . . . . . . . . . . . . . . . . 24 3.2.2 Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.2.3 Application of Gibbs Sampler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.4 Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.5 Comparative Analysis of the two variants of the M-H algorithm . . . . . . . . . . . 31 3.3 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Conclusion 34 Bibliography 35 Abbreviations and notations 37 iv |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| MM/1398 | Mémoire master | bibliothèque sciences exactes | Consultable |




