à la journée François Perron
Wonderful day celebrating my long-term friend François Perron’s career, at the Université de Montréal! In full autumnal glory! 
With several long-time-no-see friends attending and presenting their work and connecting with François’ achievements, incl. Éric Marchand on Stein prediction for spherically symmetric distributions, where the predicted vector y appears in the same quadratic form as the observed one x, with the same location parameter, a setting that turns prediction into estimation of that parameter, a 2019 paper with Dominique Fourdrinier and Bill Strawderman,A
Mylène Bédard on a generalised optimal scaling of annealed MALA, with higher speeds in transient cases than known so far,t.
Yves Atchadé on cyclical MCMC, a rejection free version of Radford Neal’s parallel tempering, that sometimes fails to converge
and Alex Bouchard-Côté on escaping the curse of dimensionality by interpolations of the target, which is a variant of simulated tempering (quite the theme of the day!), and happened to be his invited lecture for the 2024 CRM-SSC prize, awarded right after. Unfortunately I had to catch my plane and face the notorious jams to the airport. (And, fortunately, I had attended Saifuddin Syed’s related talk at Mostly MC the week before!)
October 27, 2024 at 2:29 pm
A correction… I did not invent Parallel Tempering. It was I believe first devised by Charles Geyer, in the following paper:
Geyer, C.J. (1991) “Markov chain Monte Carlo maximum likelihood”,
in E.~M.~Keramidas (editor), Computing Science and Statistics:
Proceedings of the 23rd Symposium on the Interface, pp.~156-163,
Interface Foundation.
I think it was later independently re-invented by some physicists, from whom the “Parallel Tempering” name may originate.
The paper of mine from 1996 that you link to is about a different tempering scheme, which relates to my later work on Annealed Importance Sampling.