Voodoo Ranger IPA

another electoral map

the year(s) with no conferences

This week, Nature has an article on “A year without conferences? How the coronavirus pandemic could change research”, where the journalist predicts a potential halt to scientific conferences. Taking as example the cancelled American Physical Society (APS) March Meeting, to quote

“many of them rapidly set up platforms to hold virtual sessions for the meeting, inviting their speakers to present by webcam or to upload their presentations to online repositories. Researchers who hadn’t been in a position to fly to Denver found themselves able to participate from afar in what became the Virtual APS March Meeting.”

On this same day I should have been traveling from Brussels to Grenoble for the ABC meeting there. Instead, I had a four day virtual panel meeting from home and there is no virtual version of the ABC in Gre[e]noble workshop. As no one seemed particularly eager to animate a few local talks with no guarantee of spectators. As things deteriorated to home confinement,  it was actually better not to spend more efforts on the project. Since this confinement is bound to last much longer, it would however become more obvious that the community and the academic societies need plan virtual conference and invent different channels to gather members and disseminate innovation.

a problem that did not need ABC in the end

While in Denver, at JSM, I came across [across validated!] this primarily challenging problem of finding the posterior of the 10³ long probability vector of a Multinomial M(10⁶,p) when only observing the range of a realisation of M(10⁶,p). This sounded challenging because the distribution of the pair (min,max) is not available in closed form. (Although this allowed me to find a paper on the topic by the late Shanti Gupta, who was chair at Purdue University when I visited 32 years ago…) This seemed to call for ABC (especially since I was about to give an introductory lecture on the topic!, law of the hammer…), but the simulation of datasets compatible with the extreme values of both minimum and maximum, m=80 and M=12000, proved difficult when using a uniform Dirichlet prior on the probability vector, since these extremes called for both small and large values of the probabilities. However, I later realised that the problem could be brought down to a Multinomial with only three categories and the observation (m,M,n-m-M), leading to an obvious Dirichlet posterior and a predictive for the remaining 10³-2 realisations.

R wins COPSS Award!

Hadley Wickham from RStudio has won the 2019 COPSS Award, which expresses a rather radical switch from the traditional recipient of this award in that this recognises his many contributions to the R language and in particular to RStudio. The full quote for the nomination is his  “influential work in statistical computing, visualisation, graphics, and data analysis” including “making statistical thinking and computing accessible to a large audience”. With the last part possibly a recognition of the appeal of Open Source… (I was not in Denver for the awards ceremony, having left after the ABC session on Monday morning. Unfortunately, this session only attracted a few souls, due to the competition of twentysome other sessions, including, excusez du peu!, David Dunson’s Medallion Lecture and Michael Lavine’s IOL on the likelihood principle. And Marco Ferreira’s short-course on Bayesian time series. This is the way the joint meeting goes, but it is disappointing to reach so few people.)