Samuel Livingstone
**The rmcmc package for robust and flexible mcmc sampling in R is now available on CRAN. See here or below for more details.
I am an associate professor at University College London (UCL). In my research I use tools from probability and mathematical analysis to study algorithms in Statistics and Machine Learning. The field is often called CSML (computational statistics and machine learning). I also do some more applied/methodological work in probabilistic modelling, particularly for health applications (see my research page for more).
I currently hold an EPSRC New Investigator Award entitled 'Robust and scalable Markov chain Monte Carlo for heterogeneous models'. See here for more details.
In 2021 I was honoured to become the first UK recipient of the Blackwell--Rosenbluth award from the International Society for Bayesian Analysis. Press release here.
I joined the department of Statistical Science in January 2018, after being a postdoctoral researcher at the University of Bristol, under Christophe Andrieu, as part of the i-like project, and before that a PhD student in Statistical Science at UCL, supervised by Mark Girolami and Alex Beskos.
I believe in Stigler's law of eponymy, and that the concept of multiple discovery/simultaneous invention is more closely aligned with the reality of scientific research than the heroic theory of invention that is dominant in popular culture. That being said, I still have a romantic view of academic life and strive for originality in my work.
NEWS
Jan 2024. First release of the rmcmc package for robust sampling via Markov chain Monte Carlo is now on CRAN. See here for details. The package contains flexible implementations of the Barker proposal, alongside random walk Metropolis, MALA and HMC. There is lots of functionality to use different forms of adaptive MCMC to adjust algorithmic tuning parameters, allowing fast prototyping and comparison of different sampling approaches, as well as interface with the Stan modelling language through BridgeStan. This is a project in collaboration with Matthew Graham @UCL Advanced Research Computing, and funded by EPSRC. Stay tuned for more updates in future, and please get in touch if interested in contributing.
July 2024. From July - December 2024 I am participating in the Isaac Newton Institute research programme 'Stochastic systems for anomalous diffusion'. I am organising the specific workshop: 'Monte Carlo sampling: beyond the diffusive regime' on 25-29 November.
June 2024. 'The London meeting on computational statistics' was held 11-13 June 2024 at UCL, funded by EPSRC. More details can be found here. Thank you to all speakers and participants, it was a great success!
June 2024. New preprint 'Averaging polyhazard models using Piecewise deterministic Monte Carlo with applications to data with long-term survivors' together with my student Luke Hardcastle and co-supervisor Gianluca Baio.
May 2024. New preprint 'Skew-symmetric schemes for stochastic differential equations with non-Lipschitz drift: an unadjusted Barker algorithm' together with my postdoc Giorgos Vasdekis and collaborators Nikolas Nuesken and Ruiyang Zhang.
May 2024. In May I held a visiting position at the University of Torino In Turin, Italy, and taught a graduate level course on Monte Carlo. Thanks to Matteo Ruggiero for the invitation, it was a pleasure to spend time there and teach some very talented students!
Dec 2023. I co-organised the '3rd workshop on Monte Carlo in Warsaw' together with Blazej Miasojedow, Iwona Chlebicka & Zuzanna Szymanska on 13-15 Dec at MIMUW.
Dec 2023. New preprint 'Quantifying the effectiveness of linear preconditioning in Markov chain Monte Carlo' together with my student Max Hird.
Dec 2023. Our paper 'Structure learning with adaptive random neighbourhood informed MCMC' has been accepted for publication in NeurIPS2023.
Jun 2023. Our paper 'Sampling algorithms in statistical physics: a guide for statistics and machine learning' has been accepted for publication in Statistical Science.
Sep 2022. Our paper 'Optimal design of the Barker proposal and other locally-balanced Metropolis--Hastings algorithms' has been accepted for publication in Biometrika.
Aug 2022. New preprint 'Sampling algorithms in statistical physics: a guide for statistics and machine learning' (in collaboration with Michael Faulkner at University of Bristol).
Aug 2022. Our work on Bayesian variable selection has been accepted for publication in Statistics and Computing. Congratulations Xitong!
Aug 2022. Georgios Vasdekis has joined our group as a postdoctoral researcher.
Nov 2021. I am hiring a postdoctoral researcher to work on MCMC theory and methodology. The job advert is here: https://worldsbestpostdoc.com. Informal enquiries welcome, closing date 3rd Jan 2022.
Oct 2021. My PhD student Xitong Liang has arXived a new preprint on MCMC for Bayesian variable selection, written together with myself and Jim Griffin. Avaiable here: https://arxiv.org/abs/2110.11747
Oct 2021. I am honoured to receive one of the inaugural Blackwell--Rosenbluth awards from the International Society for Bayesian Analysis. Press release: https://j-isba.github.io/blackwell-rosenbluth.html
Oct 2021. I am delighted to receive an EPSRC New Investigator Award to develop robust and scalable Monte Carlo methods for heterogeneous models. As part of the grant I will be hiring a postdoctoral researcher, informal enquiries welcome. More info: https://gow.epsrc.ukri.org/NGBOViewGrant.aspx?GrantRef=EP/V055380/1