A weekly online seminar on all aspects of Monte Carlo methods.
Join us weekly to explore recent developments and connect with a diverse community of passionate researchers in this exciting field!
Videos of past talks available on YouTube.
For announcements and Zoom invitations please subscribe to our mailing list.
Upcoming Seminar Presentations
Starting in fall 2026, all seminars are on Thursdays at [ 8:30 am PT ] = [ 11:30 am ET ]. Times in other regions vary with daylight saving time; please check the individual talk or calendar.
Subscribe to our mailing list and calendar for up-to-date schedule!
Thursday, September 24, 2026
Speaker: Pierre Monmarché (Université Gustave Eiffel) [Zoom]
Title: Local convergence rates for sampling mean-field free energy minimisers
Abstract: Many models (in statistical physics or biology) or high-dimensional algorithms can be interpreted as the problem of minimizing a so-called free energy over a space of probability measures. The corresponding Wasserstein gradient flow is approximated in practice by a system of mean-field interacting particles, following a Langevin dynamics. When the free energy has several minimizers, the particle system is metastable, so that its ergodic convergence to equilibrium is out of reach for practical simulation times. We will discuss how it is possible to get convergence results of practical interest in this situation.
Links: YouTube
Thursday, October 1, 2026
Speaker: Nikolas Nüsken (King's College London) [Zoom]
Title: Acceleration in underdamped Langevin dynamics via Fenchel games
Abstract: A common viewpoint in computational statistics is to cast sampling as the minimisation of a functional on the space of probability distributions, with the Bayesian posterior arising as a free-energy minimiser. Gradient-flow methods, including Wasserstein gradient flows, naturally lead to Langevin-type algorithms and connect sampling to optimisation. In this talk, I will discuss recent work that broadens this optimisation perspective beyond gradient flows through saddle-point formulations and duality-gap techniques, encompassing underdamped Langevin dynamics. Joint work with Alexandra Borkowski (KCL), https://arxiv.org/abs/2609.07812v1.
Links: YouTube
Thursday, October 8, 2026
Speaker: Andreas Eberle (University of Bonn) [Zoom]
Title: Relaxation times of non-reversible Markov processes
Abstract: We develop a systematic approach to quantify relaxation times for non-reversible Markov processes based on the singular value gap of the generator introduced by Chatterjee. The inverse of the singular value gap is equivalent to the relaxation time of the time-averaged transition semigroup. We show that, moreover, the inverse singular value gap of the two-point motion is equivalent to the relaxation time of the non-averaged transition semigroup, and that it provides a lower bound on the usual spectral gap of the generator.
We then introduce a method for deriving lower bounds on singular value gaps for Markov processes with degenerate noise that is based on the concept of a first- and second-order collapse of the generator. It follows ideas from hypocoercivity developed in a previous series of works but is simpler and more broadly applicable. In contrast to previous results, it includes settings with non-vanishing first-order collapse, and thus applies directly to Markov chains, but also to diffusion processes and piecewise-deterministic Markov processes.
Our approach yields sharp upper and lower bounds for several classes of examples. First applications include the proof of a conjecture by Diaconis and Miclo on a square-root speed-up for lifted random walks on abelian groups, as well as bounds on relaxation times of switching flows perturbed by noise and of non-reversible diffusion processes. Joint work with Francis Lörler.
Link: YouTube
Thursday, October 15, 2026
Speaker: Jianfeng Lu (Duke University) [Zoom]
Title: Bouncy Particle Samplers
Abstract: In this talk, we will discuss some recent works on the bouncy particle samplers in the context of log-concave sampling. We will establish O(\kappa^{1/2} d) up to logarithmic factors for gradient query complexity for a version of bouncy particle sampler from warm-start. We will also introduce a recent PDMP sampler, the proximal bouncy particle sampler, which combines the idea of proximal sampler and bouncy particle sampler to achieve O(\kappa^{1/2} d^{1/4}) gradient query complexity. Time permits, we will discuss some aspects of algorithmic warm-start generation. Based on joint work with Fan Chen (MIT), Sinho Chewi (Yale), Yinchen Luo (Duke), and Matthew S. Zhang (MIT).
Thursday, October 22, 2026
Speaker: Chris Oates (Newcastle University)
Title: A Computable Measure of Suboptimality for Entropy-Regularised Variational Objectives
Abstract: Several methods in statistics and machine learning target a probability distribution for which an entropy-regularised variational objective is minimised. This increased flexibility introduces a computational challenge, as one loses access to an explicit unnormalised density for the target. To mitigate this difficulty, we introduce a novel measure of suboptimality called 'gradient discrepancy', and in particular a kernel gradient discrepancy (KGD) that can be explicitly computed. In the Bayesian statistics context, KGD coincides with the kernel Stein discrepancy (KSD), and we obtain a novel characterisation of KSD as measuring the size of a variational gradient. Outside this familiar setting, KGD enables novel sampling algorithms to be developed and compared, even when unnormalised densities cannot be obtained. To illustrate this point several novel algorithms are proposed and studied, including a natural generalisation of Stein variational gradient descent, with applications to mean-field neural networks and predictively oriented posteriors presented. On the theoretical side, our principal contribution is to establish sufficient conditions for desirable properties of KGD, such as continuity and convergence control.
Thursday, October 29, 2026
Speaker: Yair Shenfeld (Brown University) [Zoom]
Time: October 29, 2026 [ 8:30 am PT ] = [ 11:30 am ET ] = [ 3:30 pm London ] = [ 4:30 pm Paris ] = [ 11:30 pm Beijing ]
Title: Discrete diffusion models and measure transport on ordinal structures
Abstract: Finding good ways to transport between two probability distributions is a task which naturally arises in generative modeling as well as probability and analysis. In diffusion models, the goal is to transport noise to the data distribution, and then reverse the process in order to get new samples. In probability and analysis, the goal is to transfer properties from canonical distributions to more complex ones. In this talk I will focus on both problems in the context of distributions on ordinal structures. I will show how new constructions lead to optimal bounds for ultra log-concave distributions, as well as competitive and scalable discrete diffusion models.
Thursday, November 5, 2026
Speaker: Jason Xu (University of California, Los Angeles) [Zoom]
Time: November 5, 2026 [ 8:30 am PT ] = [ 11:30 am ET ] = [ 4:30 pm London ] = [ 5:30 pm Paris ] = [ 12:30 am Beijing, November 6 ]
Title: Gradient-based sampling for stochastic epidemic models via dynamical survival analysis
Abstract: Stochastic compartmental models are prevalent tools for describing disease spread, but inference under these models is challenging for many types of surveillance data when the marginal likelihood function becomes intractable due to missing information. Statisticians tend to fit these models either using computationally intensive simulation-based method, or bespoke data-augmented MCMC due to the complex, high-dimensional latent space of possible unobserved configurations. To address this, we develop a closed-form likelihood for discretely observed incidence data arising in SIR models and their variants, using the dynamical survival analysis (DSA) paradigm. The method approximates the stochastic population-level hazard by a large population limit while retaining a count-valued stochastic model, and enables canonical posterior sampling procedures that make use of off-the-shelf software, and are highly efficient yet flexible to model generalizations. We show that parameter estimation is competitive with recent exact but computationally expensive likelihood-based methods in partially observed settings, and that the method is extensible to heterogeneous and network-based model variants.
Thursday, November 12, 2026
Speaker: Yuanqi Du (Microsoft Research New England)
Thursday, November 19, 2026
Speaker: Andre Wibisono (Yale University)
Format and Rules
The seminars are held on Zoom and last 60 minutes.
A moderator collects audience questions in Q&A section.
Moderators may ask you to unmute yourself to participate in the discussion. Please note that you may be recorded if you activate your audio or video during the seminar.
Organizers
Valentin De Bortoli (Google DeepMind)
Yang Chen (University of Michigan)
Nianqiao Ju (Dartmouth College)
Sifan Liu (Duke University)
Sam Power (University of Bristol)
Qian Qin (University of Minnesota)
Guanyang Wang (Rutgers University)
Feedback and Suggestions
If you have feedback or suggestions or want to propose a speaker, email us at internationalmontecarloseminar@gmail.com, or fill out this form.
Acknowledgements
Sincerely thank Ying Jin for the patient guidance
Gratefully inspired by the Online Causal Inference Seminar