Rui-Yang Zhang

Liwen Xue, Axel Finke, Adam M. Johansen: Online Rolling Controlled Sequential Monte Carlo https://arxiv.org/abs/2508.00696 https://arxiv.org/pdf/2508.00696 https://arxiv.org/html/2508.00696

Really enjoyed listening to this interview with Mike Giles. Only knew him from his multilevel Monte Carlo work, and it was quite a nice surprise to learn about his contributions to CFD and experiences with industrial collaborations!

we're out here simulating, visualising, thriving

Congrats !!!

Gaussian Processes and Reproducing Kernels: Connections and Equivalences This monograph studies the relations between two approaches using positive definite kernels: probabilistic methods using Gaussian processes, and non-probabilistic methods using reproducing kernel Hilb...

We've written a monograph on Gaussian processes and reproducing kernel methods (with @philipphennig.bsky.social, @sejdino.bsky.social and Bharath Sriperumbudur).

arxiv.org/abs/2506.17366

Line chart titled ‘Weekly Runs of RStudio IDE’ showing usage data from 2023 to 2025. The y-axis ranges from 2,000,000 to 6,000,000 weekly runs. The chart displays a cyclical pattern with regular peaks around 5,000,000-6,000,000 runs and dramatic drops to approximately 2,000,000 runs that occur periodically during holiday periods.

Is #rstats dead? I don’t think so.

Is it just me or does Google Scholar forbid searches via Avanti’s WiFi?

The talks from the Post-Bayes workshop are now available online here - youtube.com/playlist?lis... - do take a look!

In the interim, I wanted to advertise our YouTube channel - youtube.com/@montecarlos... - which contains recordings for the bulk of our talks so far (sites.google.com/view/monte-c..., sites.google.com/view/monte-c...). I encourage you to catch up and enjoy them over the intervening months!

Starting from last October, we (@OnlineMCSeminar on Twitter, sites.google.com/view/monte-c...) have been running an online seminar on all aspects of Monte Carlo methods, with about ~30 talks so far. We are currently paused for the summer, expecting to return in September 2025.

Do you happen to have anything related to low-rank approximation / matrix sketching? Thanks !!

Luke Hardcastle, Samuel Livingstone, Gianluca Baio
Diffusion piecewise exponential models for survival extrapolation using Piecewise Deterministic Monte Carlo
https://arxiv.org/abs/2505.05932

ANN: A non-reversible rejection-free HMC sampler Hej! Have you ever wondered if momentum flips/refreshments are really needed in HMC or if we somehow can avoid to lose our sense of direction after each proposal step? Or even wondered if we could ge...

Demo for the sampler from our recent paper

discourse.julialang.org/t/ann-a-non-...

Keen to read this:

arxiv.org/abs/2504.13322
'Foundations of locally-balanced Markov processes'
- Samuel Livingstone, Giorgos Vasdekis, Giacomo Zanella

Inside arXiv—the Most Transformative Platform in All of Science Modern science wouldn’t exist without the online research repository known as arXiv. Three decades in, its creator still can’t let it go.

Modern science wouldn’t exist without the online research repository known as arXiv. Three decades in, its creator still can’t let it go.

With Gibbs you can arbitrary choose the order you go through the coords, but in your MH that is not possible. The MH acceptance simplifies to p(x,y’)p(x’)p(y) / p(y,x’)p(x)p(y’). In the case where your x and y are independent (so your MH is also arbitrary in coord order), your MH is Gibbs.

Is there a ref for this?

We also had some thoughts on maybe using the discrete KSD works to extend stein thinning to discrete distributions

Would that empirical average comparison / gauging then be equivalent to computing test statistics for hypothesis (goodness of fit) tests then?

Was hoping to establish some theoretical results on a good choice of window length but that did not go anywhere (yet

I was working with Lanya on some extensions to Stein thinning by doing a window based scan to detect convergence (as opposed to the greedy approximation of the full minimisation objective). We implemented it and it works, but is quite fiddly wrt window length parameter.

In fact, I am quite curious about how one would do convergence diagnostics for non continuous targets. Don’t recall reading about much work towards this direction

I think there are also more recent extensions to Stein thinning made by Chris Oates, Lester Mackey et al

I would recommend these two reviews: arxiv.org/abs/1909.11827 and arxiv.org/abs/2103.16048.

Subsequently, arxiv.org/abs/1802.09188 extended this idea and provided additional error bounds of ULA.

For ULA bias, there is this paper arxiv.org/abs/1802.08089 that describe it as a time discretisation of Wasserstein gradient flow, which helps us to explain the existence of bias.

Also was made aware of the Probabilistic Richardson Extrapolation work by Chris Oates et al by @adriencorenflos.bsky.social - is that related to multilevel?

Do you also know about variance reduction in the context of SDE numerical schemes? Other than the multilevel stuff

There seem to be many standard textbooks that have a chapter or two on variance reduction techniques. But am seeking a more modern coverage (if there exists any).