Sam Power
Hello! My name is Sam, and I am a researcher in Statistics.
I am currently Lecturer in Statistical Science at the University of Bristol.
Prior to this role, I was a Senior Research Associate (also at the University of Bristol) working with Prof. Christophe Andrieu on the Bayes4Health grant, and also collaborating closely with Prof. Anthony Lee.
Even further in the past, I was a PhD student at the University of Cambridge, working with Dr. Sergio Bacallado. You can find an online copy of my dissertation here.
News
I am currently advertising a PhD project through the Probabilistic AI Hub on the topic of "Comparison Theory for MCMC Algorithms, with application to Subsampling MCMC". See the project description and details on how to apply; note that the funding is only available for UK students. The deadline for applications is 30 June 2024, though note that applications will be regularly reviewed and may be filled well in advance of this date.
Keywords
Computational Statistics
Monte Carlo Methods
Numerical Analysis
Bayesian Modelling
Research Interests
My research interests center around the design and analysis of stochastic algorithms, with applications mainly to statistics. I am particularly interested in Monte Carlo methods, such as Markov Chain Monte Carlo and Sequential Monte Carlo, and how the implementation of these methods can be made automatic, robust, and efficient.
Manuscripts
R. Caprio, J. Kuntz, S. Power, A.M. Johansen - Error bounds for particle gradient descent, and extensions of the log-Sobolev and Talagrand inequalities - arXiv
S. Power, D. Rudolf, B. Sprungk, A.Q. Wang - Weak Poincaré inequality comparisons for ideal and hybrid slice sampling - arXiv
C. Andrieu, A. Lee, S. Power, A.Q. Wang - Weak Poincaré Inequalities for Markov chains: theory and applications - arXiv
J.N. Lim, J. Kuntz, S. Power, A.M. Johansen - Momentum Particle Maximum Likelihood - arXiv
S. Duffield, S. Power, L. Rimella - A State-Space Perspective on Modelling and Inference for Online Skill Rating - arXiv
C. Andrieu, A. Lee, S. Power, A.Q. Wang - Explicit convergence bounds for Metropolis Markov chains: isoperimetry, spectral gaps and profiles - to appear in The Annals of Applied Probability, arXiv
L. Riou-Durand, P. Sountsov, J. Vogrinc, C.C. Margossian, S. Power - Adaptive Tuning for Metropolis Adjusted Langevin Trajectories - Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, PMLR 206:8102-8116, 2023, [arXiv]
C. Andrieu, A. Lee, S. Power, A.Q. Wang - Poincaré inequalities for Markov chains: a meeting with Cheeger, Lyapunov and Metropolis - Technical Report, arXiv
F. Pagani, A. Chevallier, S. Power, S. Cotter, T. House - NuZZ: Numerical Zig-Zag Sampling for General Models - Statistics and Computing, [arXiv]
C. Andrieu, A. Lee, S. Power, A.Q. Wang - Comparison of Markov chains via weak Poincaré inequalities with application to pseudo−marginal MCMC - The Annals of Statistics, 50(6), 2022, [arXiv]
A. Chevallier, S. Power, A.Q. Wang, P. Fearnhead - PDMP Monte Carlo methods for piecewise-smooth densities - Advances in Applied Probability, [arXiv]
S. Power, J. Vorstrup Goldman - Accelerated Sampling on Discrete Spaces with Non−Reversible Markov Processes - arXiv, GitHub, YouTube
See also my Google Scholar profile.
Collaborators (in alphabetical order)
Christophe Andrieu, Rocco Caprio, Augustin Chevallier, Simon Cotter, Sam Duffield, Paul Fearnhead, Jacob Vorstrup Goldman, Thomas House, Adam Johansen, Juan Kuntz, Anthony Lee, Jen Ning Lim, Charles Margossian, Filippo Pagani, Lorenzo Rimella, Lionel Riou-Durand, Daniel Rudolf, Pavel Sountsov, Bjoern Sprungk, Jure Vogrinc, Andi Q. Wang
Slides
I have given research talks about several of the works listed above, and I am generally very happy to share the slides which I use in these talks. Where possible, I tailor the structure and content of my presentations to each specific audience. As a result, each work is typically associated with a few distinct slide decks. If you are interested in seeing my slides corresponding to any of these works, then I would be very happy for you to email me about this, so that I may provide you with the most suitable variant of the slides for your purposes.
Notes
Education and Positions
(2024-): Lecturer in Statistical Science University of Bristol
(2020-2023): Postdoctoral Research Associate University of Bristol
(2016-2020): PhD University of Cambridge
(2010-2014): MMath University of Oxford
I am happy to be contacted about my work, or other research-related topics, and generally prefer to begin a dialogue over email. Historically, I have also been known to tweet fairly regularly about research. I am happy to be contacted there in a less formal capacity.
Some of my postings are about publicly-available reference materials, some of which I have catalogued here.