Kengo Kamatani

Bayesian computation · Monte Carlo methods

Kengo Kamatani

Professor at the Institute of Statistical Mathematics, Japan

My research focuses on the theoretical and computational foundations of Bayesian inference, with particular interests in Markov chain Monte Carlo, particle methods, and stochastic processes.

Explore my research Email

Research

Computation for modern Bayesian statistics

Bayesian computation makes posterior inference possible when analytic calculation is out of reach. My work studies why Monte Carlo algorithms work, how they behave in high dimensions, and how their design can be improved.

I also work on statistical inference for stochastic processes and on computational tools connecting theory with practical analysis.

For broader perspectives, see Doing Bayesian Data Analysis, Bayesian Data Analysis, and the International Society for Bayesian Analysis. I also follow the research blogs of Andrew Gelman and Christian Robert.

See also our seminar series on Bayesian Computation.

01

Markov chain Monte Carlo

Ergodicity, scaling limits, heavy-tailed targets, and non-reversible transition kernels.

02

Particle methods

Sequential Monte Carlo and particle filters for complex, high-dimensional state-space models.

03

Stochastic processes

Inference and computation for continuous-time models, including stochastic differential equations.

Selected publications

Recent and representative work

Complete list on ORCID

    News

    Recent activity

      Kengo Kamatani giving a talk

      Research network

      Projects and collaboration

      My research connects the theory of Monte Carlo methods with collaborative work on stochastic-process models, statistical software, and practical inference.

      JST CREST

      I serve as the ISM group leader in the JST CREST project on statistics for stochastic systems and large-scale spatiotemporal dependence (Oct 2021–Mar 2027), led by Nakahiro Yoshida. I previously led the Osaka group in the CREST area Modeling Methods allied with Modern Mathematics (Oct 2014–Sep 2020).

      Visit the project site

      Software and collaboration

      I contribute to the development of YUIMA, a computational framework for simulation and inference for stochastic differential equations. The related yuimaBayes package connects that framework with Bayesian workflows. Other work includes collaboration with the Railway Technical Research Institute.

      YUIMA Project yuimaBayes on GitHub Railway Technical Research Institute

      Academic record

      Funding, roles, and awards

      Research funding · JSPS KAKENHI

        Editorial roles

          Awards

            Profile

            Education and academic career

            Education

              Professional societies

                Academic positions

                  Contact

                  Get in touch

                  kamatani@ism.ac.jp
                  D610B, Institute of Statistical Mathematics
                  10-3 Midoricho, Tachikawa
                  Tokyo 190-8562, Japan