The Multiple-Try Method and Local Optimization in Metropolis Sampling

2000

420 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistical and Nonlinear Physics, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/ff17c129a8d32bb7dc7206230da612e94bd24b9f

Citations in the corpus

• On the flexibility of the design of multiple try Metropolis schemes (2012, positive): “nd 0.6056 0.0012 = 10, ˙= 50 MTM-ind 0.5994 0.0010 = 100, ˙= 50 MTM-rw 0.5819 0.0050 ˙= 50 6.5 Dierent acceptance probabilities In this section, we consider again the bimodal target density in Eq. (17), i.e., po(x) /p(x) = exp (x2 4)2=4 , and we generate candidates from a random walk Gaussian densit”
• Markov chain Monte Carlo simulation using the DREAM software package: Theory, concepts, and MATLAB implementation (2016, positive): “The use of multiple chains has several desirable advantages, particularly when dealing with complex posterior distributions involving long tails, correlated parameters, multi-modality, and numerous local optima (Gilks et al., 1994; Liu et al., 2000; ter Braak, 2006; ter Braak and Vrugt, 2008; Vrugt ”
• Optimal scaling of Metropolis algorithms: Heading toward general target distributions (2008, neutral): “…algorithms; Bédard, Fort, and Moulines (2008) features a comparison of the asymptotic behaviour of the Multiple Try Metropolis algorithm and the Delayed Rejection Metropolis algorithm (for more information on these sampling methods , see Liu, Liang, and Wong (2000) and Mira (2001), respectively).”
• Bayesian CART – Prior Specification and Posterior Simulation – (2006, positive): “If convergence becomes a problem for large data sets, it is also possible to speed up the computations by adopting a Metropolis–Hastings algorithm where many potential new states are proposed in each iteration, see the discussion in Liu et al. (2000), Qin and Liu (2001) and Tjelmeland (2004).”