2020

Adaptive Path Sampling in Metastable Posterior Distributions

Yao, Yuling, Cademartori, Collin, Vehtari, Aki et al.

Understand

The normalizing constant plays an important role in Bayesian computation, and there is a large literature on methods for computing or approximating normalizing constants that cannot be evaluated in closed form.

When the normalizing constant varies by orders of magnitude, methods based on importance sampling can require many rounds of tuning.
We present an improved approach using adaptive path sampling, iteratively reducing gaps between the base and target.
Using this adaptive strategy, we develop two metastable sampling schemes.

Built on

Implicitly adaptive importance sampling
Original
Paananen, T., Piironen, J., Bürkner, P.-C., and Vehtari, A. (2020) · 1906
Earlier work this paper cites.
Statistical mechanics of fluid mixtures
Kirkwood, J. G. (1935) · 1935
Earlier work this paper cites.
A stochastic approximation method
Robbins, H. and Monro, S. (1951) · 1951
Earlier work this paper cites.
Bayesian inference in econometric models using Monte Carlo integration
Geweke, J. (1989) · 1989
Earlier work this paper cites.
A Monte Carlo method for high dimensional integration
Ogata, Y. (1989) · 1989
Earlier work this paper cites.
Methods for minimizing errors in linear thermodynamic integration
Schlitter, J. (1991) · 1991
Earlier work this paper cites.
Simulated tempering: A new Monte Carlo scheme
Marinari, E. and Parisi, G. (1992) · 1992
Earlier work this paper cites.
Probabilistic inference using Markov chain Monte Carlo methods
Neal, R. M. (1993) · 1993
Earlier work this paper cites.
Annealing Markov chain Monte Carlo with applications to ancestral inference
Geyer, C. J. and Thompson, E. A. (1995) · 1995
Earlier work this paper cites.
Simulating ratios of normalizing constants via a simple identity: A theoretical exploration
Meng, X.-L. and Wong, W. H. (1996) · 1996
Earlier work this paper cites.
Nonequilibrium equality for free energy differences
Jarzynski, C. (1997) · 1997
Earlier work this paper cites.
Simulating normalizing constants: From importance sampling to bridge sampling to path sampling
Gelman, A. and Meng, X.-L. (1998) · 1998
Earlier work this paper cites.
Automated docking using a lamarckian genetic algorithm and an empirical binding free energy function
Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey, R., Hart, W. E., Belew, R. K., and Olson, A. J. (1998) · 1998
Earlier work this paper cites.
Safe and effective importance sampling
Owen, A. and Zhou, Y. (2000) · 2000
Earlier work this paper cites.
Annealed importance sampling
Neal, R. M. (2001) · 2001
Earlier work this paper cites.

Similar

On the swapping algorithm
Madras, N. and Zheng, Z. (2003) · 2003
Cited alongside, same era.
Torpid mixing of simulated tempering on the Potts model
Bhatnagar, N. and Randall, D. (2004) · 2004
Cited alongside, same era.
Ensemble variance in free energy calculations by thermodynamic integration: theory, optimal “alchemical” path, and practical solutions
Blondel, A. (2004) · 2004
Cited alongside, same era.
Stacking for non-mixing Bayesian computations: The curse and blessing of multimodal posteriors
Original
Yao, Y., Vehtari, A., and Gelman, A. (2020) · 2006
Cited alongside, same era.
The folk theorem of statistical computing
Gelman, A. (2008) · 2008
Cited alongside, same era.
Statistically optimal analysis of samples from multiple equilibrium states
Shirts, M. R. and Chodera, J. D. (2008) · 2008
Cited alongside, same era.
Conditions for rapid mixing of parallel and simulated tempering on multimodal distributions
Woodard, D. B., Schmidler, S. C., and Huber, M. (2009) · 2009
Cited alongside, same era.
Free Energy Computation: A Mathematical Perspective
Lelievre, T., Rousset, M., and Stoltz, G. (2010) · 2010
Cited alongside, same era.
Bayesian Data Analysis
Gelman, A., Carlin, J. B., Stern, H. S., Dunson, D. B., Vehtari, A., and Rubin, D. B. (2013) · 2013
Cited alongside, same era.
Monte Carlo Statistical Methods
Robert, C. and Casella, G. (2013) · 2013
Cited alongside, same era.
The no-U-turn sampler: Adaptively setting path lengths in hamiltonian monte carlo
Hoffman, M. D. and Gelman, A. (2014) · 2014
Cited alongside, same era.
Kernel adaptive Metropolis-Hastings
Sejdinovic, D., Strathmann, H., Garcia, M. L., Andrieu, C., and Gretton, A. (2014) · 2014
Cited alongside, same era.

Then

A conceptual introduction to Hamiltonian Monte Carlo
Original
Betancourt, M. (2017) · 2017
Later among the works it cites.
Adaptive importance sampling: The past, the present, and the future
Bugallo, M. F., Elvira, V., Martino, L., Luengo, D., Miguez, J., and Djuric, P. M. (2017) · 2017
Later among the works it cites.
Stan: A probabilistic programming language
Carpenter, B., Gelman, A., Hoffman, M., Lee, D., Goodrich, B., Betancourt, M., Brubaker, M. A., Guo, J., Li, P., and Ridell, A. (2017) · 2017
Later among the works it cites.
Continuously tempered Hamiltonian Monte Carlo
Graham, M. M. and Storkey, A. J. (2017) · 2017
Later among the works it cites.
The sample size required in importance sampling
Chatterjee, S. and Diaconis, P. (2018) · 2018
Later among the works it cites.
Does Hamiltonian Monte Carlo mix faster than a random walk on multimodal densities?
Original
Mangoubi, O., Pillai, N. S., and Smith, A. (2018) · 2018
Later among the works it cites.
Unbiased estimation of log normalizing constants with applications to Bayesian cross-validation
Original
Rischard, M., Jacob, P. E., and Pillai, N. (2018) · 2018
Later among the works it cites.
Yes, but did it work?: Evaluating variational inference
Yao, Y., Vehtari, A., Simpson, D., and Gelman, A. (2018) · 2018
Later among the works it cites.
Pseudo-extended Markov chain Monte Carlo
Nemeth, C., Lindsten, F., Filippone, M., and Hensman, J. (2019) · 2019
Later among the works it cites.
Pareto smoothed importance sampling
Original
Vehtari, A., Simpson, D., Gelman, A., Yao, Y., and Gabry, J. (2019) · 2019
Later among the works it cites.
Automatic reparameterisation of probabilistic programs
Gorinova, M. I., Moore, D., and Hoffman, M. D. (2020) · 2020
Closest in time.
R: A language and environment for statistical computing
R Core Team (2020) · 2020
Closest in time.
Stan modeling language user’s guide
Stan Development Team (2020) · 2020
Closest in time.
Rank-normalization, folding, and localization: An improved R ^ \widehat{R} for assessing convergence of MCMC
Vehtari, A., Gelman, A., Simpson, D., Carpenter, B., and Bürkner, P.-C. (2020) · 2020
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…