## On a directionally adjusted metropolis-hastings algorithm (2008)

Citations: | 1 - 0 self |

### BibTeX

@TECHREPORT{Bédard08ona,

author = {Mylène Bédard and D. A. S. Fraser},

title = {On a directionally adjusted metropolis-hastings algorithm},

institution = {},

year = {2008}

}

### OpenURL

### Abstract

We propose a new Metropolis-Hastings algorithm for sampling from smooth, unimodal distributions; a restriction to the method is that the target be optimizable. The method can be viewed as a mixture of two types of MCMC algorithm; specifically, we seek to combine the versatility of the random walk Metropolis and the efficiency of the independence sampler as found with various types of target distribution. This is achieved through a directional argument that allows us to adjust the thickness of the tails of the proposal density from one iteration to another. We discuss the relationship between the acceptance rate of the algorithm and its efficiency. We finally apply the method to a regression example concerning the cost of construction of nuclear power plants, and compare its performance to the random walk Metropolis algorithm with Gaussian proposal.

### Citations

2564 |
Equation of state calculations by fast computing machines
- Metropolis, Rosenbluth, et al.
- 1953
(Show Context)
Citation Context ...nterested in obtaining a sample from this density of interest, but that unfortunately there is no simple way to achieve this directly. We might then use the very general MetropolisHastings algorithm (=-=Metropolis et al. 1953-=-; Hastings 1970), which is implemented through the following procedure. Given that the current sample value is xj, we propose a new value for the sample by generating a value yj+1 from a preferred pro... |

1385 |
Monte Carlo Sampling Methods Using Markov Chains and Their
- Hastings
- 1970
(Show Context)
Citation Context ... sample from this density of interest, but that unfortunately there is no simple way to achieve this directly. We might then use the very general MetropolisHastings algorithm (Metropolis et al. 1953; =-=Hastings 1970-=-), which is implemented through the following procedure. Given that the current sample value is xj, we propose a new value for the sample by generating a value yj+1 from a preferred proposal distribut... |

370 | Bayesian Inference in Econometric Models Using Monte Carlo Integration.Econometrica - Geweke - 1989 |

170 |
Weak convergence and optimal scaling of random walk metropolis algorithms, The Annals of Applied Probability
- Roberts, Gelman, et al.
- 1997
(Show Context)
Citation Context ...e the variance of the proposal distribution in the case of a normal proposal, or its range in the case of a uniform proposal. There already exist guidelines in the literature to facilitate this step (=-=Roberts et al. 1997-=-; Roberts and Rosenthal 2001; Bédard 2006 and 2007). As a drawback to the wide applicability of RWM algorithms, we notice however that their convergence may be lengthy. This should not come as a surpr... |

167 | Rates of convergence of the Hastings and Metropolis algorithms, Ann - Mergersen, Tweedie - 1996 |

100 | Optimal Scaling for Various Metropolis-Hastings Algorithms - Roberts, Rosenthal - 2001 |

32 | Adaptive direction sampling - Gilks, Roberts, et al. - 1994 |

24 | Weak Convergence of Metropolis algorithms for non-i.i.d. target distributions - Bédard |

23 | Applied Statistics: Principles and Examples - Cox - 1981 |

21 | Optimal Acceptance Rates for Metropolis Algorithms: Moving Beyond 0.234
- Bédard
- 2008
(Show Context)
Citation Context ...e case of a normal proposal, or its range in the case of a uniform proposal. There already exist guidelines in the literature to facilitate this step (Roberts et al. 1997; Roberts and Rosenthal 2001; =-=Bédard 2006-=- and 2007). As a drawback to the wide applicability of RWM algorithms, we notice however that their convergence may be lengthy. This should not come as a surprise when taking into account the versatil... |

16 | Practical Small-Sample Parametric Inference - Brazzale - 2000 |

11 | Higher accuracy for Bayesian and frequentist inference: large sample theory for small sample likelihood
- Bédard, Fraser, et al.
- 2007
(Show Context)
Citation Context ...e, it may be of interest to determine an exact p-value; this may be the case when examining new or existing methods for computing p-values, or simply when performing a study about their accuracy (see =-=Bédard et al. 2007-=-). Common to all Metropolis-Hastings algorithms is the need for selecting a proposal density q, which is used to propose values to be potentially included in the sample. The characteristics of the cho... |

3 | Adaptive radial-based direction sampling: some flexible and robust Monte Carlo integration methods - Bauwens, Bos, et al. - 2004 |

1 | Applied Asymtotics: Case Studies in Small-Sample Statistics - Brazzale, Davison, et al. - 2007 |

1 | Applied Statistics: Principles and Examples - R, Snell - 1981 |