Hirofumi Shiba

司馬博文
Ph.D. student in Statistical Computation
Google Scholar arXiv Github

Institute of Statistical Mathematics (ISM), Tokyo, Japan

Feel free to call me Hiro. I'm advised by Kengo Kamatani.

I work mainly in Bayesian computation and am especially interested in analyzing algorithmic dynamics as stochastic processes.

Interests

  1. Markov chain Monte Carlo, piecewise deterministic Markov process, diffusion models, …
  2. Sequential Monte Carlo, interacting particle systems, message passing, …
  3. Stochastic gradient descent, variational inference, Wasserstein gradient flow, …

My primary interest lies in the dynamical properties of these algorithms, especially their connections to natural phenomena.

I approach these questions through stochastic processes and functional analysis, fields in which I was trained at the University of Tokyo before my Ph.D.

My surname Shiba (司馬) has different kanji from the Shiba (柴) dog, although they are equally charming.

Software

I maintain the Julia package PDMPFlux.jl, and also contribute to the R package YUIMA.

Selected Publications

Diffusive Scaling Limits of Forward Event-Chain Monte Carlo: Provably Efficient Exploration with Partial Refreshment

Hirofumi Shiba, Kengo Kamatani. Submitted to Annals of Applied Probability. 2026

Many PDMP methods have been proposed. We develop a high-dimensional scaling framework to compare their asymptotic efficiency (effective sample size per event).
No matching items