control variates [seminar]

Today, Petros Dellaportas (whom I have know since the early days of MCMC, when we met in CIRM) gave a seminar at the Warwick algorithm seminar on control variates for MCMC, reminding me of his 2012 JRSS paper. Based on the Poisson equation and using a second control variate to stabilise the Monte Carlo approximation do the first control variate. The difference with usual control variates is finding a first approximate G(x)-q(y|x)G(Y) to F-πF. And the first Poisson equation is using α(x,y)q(y|x) rather than π. Then the second expands log α(x,y)q(y|x) to achieve a manageable term.

Abstract: We provide a general methodology to construct control variates for any discrete time random walk Metropolis and Metropolis-adjusted Langevin algorithm Markov chains that can achieve, in a post-processing manner and with a negligible additional computational cost, impressive variance reduction when compared to the standard MCMC ergodic averages. Our proposed estimators are based on an approximate solution of the Poisson equation for a multivariate Gaussian target densities of any dimension.

I wonder if there were a neural network version that would first build G from scratch and later optimise it towards solving the Poisson equation. As in this recent arXival I haven’t read (yet).

intimate touch [wildlife photograph of the year]

Bayesians at the helm!

Just read the announcement that my friend (and former colleague at Warwick U) Mark Girolami became the Chief Scientist at The Alan Turing Institute, joining forces with Adrian Smith, currently Director and Chief Executive of the Turing Institute, into a Bayesian leadership!

Our Lady of the Van

As I came across this Lady in the van film on a lazy evening, I gave it a go since it featured Maggie Smith, London, and apparently an eccentric homeless old-lady. Afterwards, since I was somewhat reserved about the story and the plot, if not by Smith’s and Jennings’ acting, and surprised at the highly positive reviews it received. I looked at the background, only to discover that this was a slightly modified version of a real story where the English writer Alan Bennett let Margaret Fairchild live in a van on his property, in Northern London, in the 70’s and 80’s, until she died in 1989. If ignoring the heavy pathos permeating the film throughout to concentrate on the (light) social criticism of the bo-bo Gloucester Crescent residents and on the very British satire behind essentially every character, first and foremost the writer himself, there are some enjoyable aspects.  But I remain perturbed by the somewhat exploitative way Bennett turned this story into a 1989 essay, then a 1990 book, then a 1999 play (already involving Maggie Smith), and a 2009 BBC radio play, before adapting the play for the film. And somewhat shocked that over 15 years Margaret Fairchild was let to live in such conditions. In a rusted van, in the middle of London…

Shades of magic [book review]

After seeing the books in a Denver bookstore (in the summer of 2019), I eventually came to try one, then the others. Even though the setting is somewhat simplistic, or may intended for young adults, with ideas also found in earlier novels, it makes for a pleasant read. The underlying concept is having several Londons set in different universes and connected by magic for the happy few able to travel between them. One of them is “our” Victorian London. Labelled as Grey London. Then there are White, Red, and Black Londons… With some pivotal pubs existing in all (?) of them. This reminded me very much of Neverwhere, one of the few Gaimans I deeply appreciated. Or of Pullman’s Oxfords. The first volume sets the scene, with two main characters, (Grey) Lila and (Red) Kell, whose paths will come to cross, some villains in the least privileged London, and some sudden existential threat on Red London. The latest being the least convincing part of the plot as lacking subtlety. The second volume mostly takes place in Red London and the first part sounds a wee bit like the female part of Red seas under red skys. That is, a smart thief at sea. And a smarter captain. With on top of it a magic competition where all main characters cross path. Again a poor part of the plot as the competition feels like Harry Potter’s Goblet of Fire, while a new danger is building up to bring the fodder for the third volume. Not completely uninteresting (I read most of it over Xmas day, by a log fire), but somewhat two-dimensional (with a surprising lack of moral reticence to kill people, most surprising for a YA series). The third volume, A conjuring of light,  is a bit more predictable, including the deaths of some major characters (one or two more would have helped). And the ending could have been less all-inclusive and rosy!,  but this was an enjoyable conclusion nonetheless.