tulogo

Homepage of Joris Bierkens

TUDelft/EEMCS/DIAM
Mekelweg 4
2628 CD Delft
Room: 6.270
Phone: +31 15 27 84593
E-mail: Joris.Bierkens@tudelft.nl

Assistant professor
dr. ir. G.N.J.C. Bierkens

My interest lies in the computational challenges arising in Bayesian statistics and statistical physics. This means that I am mostly developing and analyzing new Markov Chain Monte Carlo (MCMC) algorithms. My philosophy in MCMC is that it is always worth to aim for algorithms which sample from the exact distribution of interest (and introduce no computational bias, something which can often be avoided).

Recently we have discovered (alongside other groups) that Piecewise Deterministic Markov Processes can be used for Monte Carlo purposes. This discovery opens up an entirely new family of elegant algorithms which can be extremely efficient in challenging settings. Naturally, this discovery also leads to many further open questions. See our paper on the Zig-Zag Process.

Projects for students

Feel free to contact me by e-mail if you would like to carry out a research project (BSc/MSc) in the direction of machine learning, computational physics, computational statistics or MCMC. Possible MSc projects are listed at this website.

Information for companies

R&D professionals with research questions concerning Bayesian statistics and in particular Markov Chain Monte Carlo methods are welcome to contact me.

Links

Statistics at TU Delft