Ben Calderhead
Also known as B. Calderhead
Ben Calderhead is a computational statistician whose work links Bayesian inference with machine-learning research and scientific modelling. He studied population MCMC at Glasgow and developed Riemannian, parallel and probabilistic methods for complex systems. At Imperial, he applied parallel sampling to scientific models; at Cervest, he served as Chief Scientific Advisor to a climate-intelligence company whose platform combines machine learning and statistical models. As of October 2026, he is an active director of Purpose Driven Foods and a co-founder of Good Phats, launched with Tom Redwood in 2024.
Timeline
-
2008
Glasgow study and Bayesian inference
At Glasgow's School of Computing Science, he completes an MSc(R) thesis in 2008 on population MCMC for comparing nonlinear ODE models. With Mark Girolami and Neil D. Lawrence, he presents a NeurIPS 2008 method that uses Gaussian processes to accelerate Bayesian inference of parameters in those equations.
-
2011
A PhD in computational statistics
He earns a PhD in 2011 at Glasgow's School of Computing Science under Mark Girolami. His thesis develops Riemannian-manifold Monte Carlo methods for Bayesian inference in high-dimensional models.
-
2012
Sparse manifold MCMC
In 2012, with Mátyás A. Sustik, he develops a sampling-based method to approximate the metric tensor when exact calculations are unavailable. They use sparse regularisation to reduce computation in Riemannian MCMC and test it on a nonlinear ODE model.
-
2014
Parallel Metropolis-Hastings at Imperial
In 2014, at Imperial College London's Department of Mathematics, he introduces a Metropolis-Hastings construction that calculates several proposed states in parallel. Imperial reports applications in models of fly vision and the soil beneath the V&A extension in London.
-
2016–2018
Probabilistic ODE solvers
With Onur Teymur and Kostas Zygalakis, he develops probabilistic Adams-Bashforth and Adams-Moulton methods for ODEs from a Gaussian-process framework in 2016. A 2018 NeurIPS paper with Teymur, Han Cheng Lie and Tim Sullivan introduces implicit probabilistic integrators for initial-value problems.
-
2021
Rao-Blackwellised parallel MCMC
With Tobias Schwedes, he publishes Rao-Blackwellised parallel MCMC at AISTATS in 2021. They extend adaptive MCMC to multiple proposals and prove asymptotic unbiasedness, with numerical studies showing improved sampling efficiency.
-
2021
Climate intelligence at Cervest
In September 2021, he serves as Cervest's Chief Scientific Advisor. The company's climate-intelligence platform combines climate data, machine learning and statistical models to assess risk across millions of assets.
-
As of October 2026
Purpose Driven Foods director
He is an active director of Purpose Driven Foods, a role that begins in May 2022. With Tom Redwood, he co-founds Good Phats in 2024, a brand for cooking fats and oils.
Career
- Since 2024Good PhatsCo-founder
- Since 2022Purpose Driven FoodsDirector
- From 2021CervestChief Scientific Advisor
Education
- 2011University of GlasgowPhD
- 2008University of GlasgowMSc(R)
Colleagues
Sources and credits
Photo: We have no published photo of Ben Calderhead yet. Send us a free-licensed photo.
- A study of Population MCMC for estimating Bayes Factors over nonlinear ODE models
- Accelerating Bayesian Inference over Nonlinear Differential Equations with Gaussian Processes
- Differential geometric MCMC methods and applications
- Sparse Approximate Manifolds for Differential Geometric MCMC
- Imperial mathematician sheds new light on 50 year old algorithm
- Probabilistic Linear Multistep Methods
- Implicit Probabilistic Integrators for ODEs
- Rao-Blackwellised parallel MCMC
- Microsoft Deputy CTO joins Cervest as Chairperson
- Good Phats – Sous Chef UK
- PURPOSE DRIVEN FOODS LTD people