Girolami, Mark; Calderhead, Ben [Sanz-Serna, J. M.; Rasmussen, Carl Edward; Doucet, Arnaud; Jacob, Pierre; Johansen, Adam M.; Honkela, Antti; Guillas, Serge; Barthelmé, Simon; Chopin, Nicolas; Filippone, Maurizio; Critchley, Frank; Friel, N.; Wyse, J.; Stathopoulos, Vassilios; Robert, Christian P.; Fearnhead, Paul; Beskos, A.; Stuart, A. M.; Skilling, John; Anaya-Izquierdo, Karim; Marriott, Paul; Archambeau, Cedric; Bouchard, Guillaume; Beffy, Magali; Bhadra, Anindya; Bornn, Luke; Cornebise, Julien; Campbell, David A.; Cao, Jiguo; Wang, Liangliang; Chin, Siu A.; Coolen, A. C. C.; Peters, Gareth; Cox, D. R.; Draper, David; Dryden, Ian; Eguchi, Shinto; Gelman, Andrew; Golightly, Andrew; Boys, Richard J.; Griffin, Jim; Gripton, Adam; Christie, Mike; Guerrera, Thiago; Rue, Håvard; Simpson, Daniel; Higham, Desmond J.; Holmes, Chris; Husmeier, Dirk; Ikeda, Shiro; Jasra, Ajay; Singh, Sumeetpal; Jupp, Peter; Kypraios, Theodore; Łatuszyński, Krzysztof; Roberts, Gareth O.; Thiéry, Alexandre; Wolny, Katarzyna; Mansinghka, V.; Marin, Jean-Michel; Meng, Xiao-Li; Mira, Antonietta; Haario, Heikki; Murray, Iain; Adams, Ryan Prescott; Parry, Matthew F.; Penny, W. D.; Pettitt, Anthony; Pillai, Natesh S.; Rao, C. R.; Roy, Daniel M.; Salimans, Tim; Schmidl, D.; Theis, F. J.; Sermaidis, Giorgos; Srivastava, Anuj; Stephens, David A.; Titsias, M. K.; Transtrum, Mark K.; Chen, Yanjiun; Machta, Benjamin B.; Sethna, James P.; Gutenkunst, Ryan; Vehtari, Aki; Vanhatalo, Jarno; Welling, Max; Winther, Ole; Opper, Manfred] Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors’ reply. (English) Zbl 1411.62071 J. R. Stat. Soc., Ser. B, Stat. Methodol. 73, No. 2, 123-214 (2011). Summary: The paper proposes Metropolis adjusted Langevin and Hamiltonian Monte Carlo sampling methods defined on the Riemann manifold to resolve the shortcomings of existing Monte Carlo algorithms when sampling from target densities that may be high dimensional and exhibit strong correlations. The methods provide fully automated adaptation mechanisms that circumvent the costly pilot runs that are required to tune proposal densities for Metropolis-Hastings or indeed Hamiltonian Monte Carlo and Metropolis adjusted Langevin algorithms. This allows for highly efficient sampling even in very high dimensions where different scalings may be required for the transient and stationary phases of the Markov chain. The methodology proposed exploits the Riemann geometry of the parameter space of statistical models and thus automatically adapts to the local structure when simulating paths across this manifold, providing highly efficient convergence and exploration of the target density. The performance of these Riemann manifold Monte Carlo methods is rigorously assessed by performing inference on logistic regression models, log-Gaussian Cox point processes, stochastic volatility models and Bayesian estimation of dynamic systems described by non-linear differential equations. Substantial improvements in the time-normalized effective sample size are reported when compared with alternative sampling approaches. MATLAB code that is available from http://www.ucl.ac.uk/statistics/research/rmhmc allows replication of all the results reported. Cited in 249 Documents MSC: 62F15 Bayesian inference 62H11 Directional data; spatial statistics 65C60 Computational problems in statistics (MSC2010) 65C05 Monte Carlo methods 62-02 Research exposition (monographs, survey articles) pertaining to statistics Keywords:Bayesian inference; geometry in statistics; Hamiltonian Monte Carlo methods; Langevin diffusion; Markov chain Monte Carlo methods; Riemann manifolds Software:Matlab PDFBibTeX XMLCite \textit{M. Girolami} and \textit{B. Calderhead}, J. R. Stat. Soc., Ser. B, Stat. Methodol. 73, No. 2, 123--214 (2011; Zbl 1411.62071) Full Text: DOI