Xi'an's Og

I was informed yesterday that my former colleague and collaborator Dominique Cellier had passed away. I had met him when starting my PhD—under the supervision of Jean-Pierre Raoult—at the Université de Rouen in 1985, where he was a freshly recruited Maître de Conférence, along with his colleague and collaborator Dominique Fourdrinier. The three of us started working together on the derivation of James-Stein (or shrinkage) estimators and I published one of my very first papers with them, about the robustness of the Stein phenomenon across elliptically symmetric distributions (thanks to a mixture decomposition of these distributions). In the Journal of Multivariate Analysis. We continued collaborating on shrinkage and MCMC over the years I was a professor at Rouen (1992-1994 and 1998-2000), with Dominique explaining me the specifics of coupling and minorisation in Markov chains, since his thesis was closely connected to Nummelin’s splitting technique. Thanks to these conversations, I wrote Convergence Control Methods for Markov Chain Monte Carlo Algorithms, published in Statistical Science in 1995. After joining Dauphine, I lost touch with him, but I keep the memory of a warm, enthusiast, dedicated, colleague, a charismatic teacher, with strong moral principles and an infinite readiness to engage into societal and political actions. For instance, he was a member of the LCR, the NPA, ATTAC Rouen, Amis de la Conf’, Alternatiba Rouen, Coalition Climat 21 Haute Normandie. and strongly involved into the Fondation Sciences Citoyennes. Au revoir, Dom!

This entry was posted on October 23, 2026 at 12:26 am and is filed under Books, Statistics, University life with tags James-Stein estimator, Journal of Multivariate Analysis, MCMC, obituary, renewal process, Rouen, Sciences Citoyennes, shrinkage estimation, splitting technique, Université de Rouen. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.