**Y**esterday, Giulia Carallo arXived the paper on generalised Poisson difference autoregressive processes that is a component of her Ph.D. thesis at Ca’ Foscari Universita di Venezia and to which I contributed while visiting Venezia last Spring. The stochastic process under study is integer valued as a difference of two generalised Poisson variates, made dependent by an INGARCH process that expresses the mean as a regression over past values of the process and past means. Which can be easily simulated as a difference of (correlated) Poisson variates. These two variates can in their turn be (re)defined through a thinning operator that I find most compelling, namely as a sum of Poisson variates with a number of terms being a (quasi-) Binomial variate depending on the previous value. This representation proves useful in establishing stationarity conditions on the process. Beyond establishing various properties of the process, the paper also examines how to conduct Bayesian inference in this context, with specialised Gibbs samplers in action. And comparing models on real datasets via Geyer‘s (1994) logistic approximation to Bayes factors.

## Archive for PhD thesis

## generalised Poisson difference autoregressive processes

Posted in pictures, Statistics, Travel, University life with tags autoregressive model, Bayes factors, Ca' Foscari University, Charlie Geyer, PhD thesis, Poisson distribution, stationarity, thinning, Venezia on February 14, 2020 by xi'an## MCMC, with common misunderstandings

Posted in Books, pictures, R, Statistics, University life with tags ABC, Bayesian computing, computational statistics, Gibbs sampling, Handbook of Computational Statistics and Data Science, HMC, IMS Lawrence D. Brown PhD Student Award, MCMC, PhD thesis, Q&A format, Statistics and Computing, survey, variational Bayes methods on January 27, 2020 by xi'an**A**s I was asked to write a chapter on MCMC methods for an incoming *Handbook of Computational Statistics and Data Science*, published by Wiley, rather than cautiously declining!, I decided to recycle the answers I wrote on X validated to what I considered to be the most characteristic misunderstandings about MCMC and other computing methods, using as background the introduction produced by Wu Changye in his PhD thesis. Waiting for the opinion of the editors of the *Handbook* on this Q&A style. The outcome is certainly lighter than other recent surveys like the one we wrote with Peter Green, Krys Latuszinski, and Marcelo Pereyra, for Statistics and Computing, or the one with Victor Elvira, Nick Tawn, and Changye Wu.

## Froebenius coin problem

Posted in pictures, R, Statistics with tags bills, codegolf, combn(), Ferdinand Frobenius, james Joseph Sylvester, Journal of Number Theory, mathematical puzzle, Norway, PhD thesis, Sylvester's formula, The Riddler on November 29, 2019 by xi'an**A** challenge from The Riddler last weekend came out as the classical Frobenius coin problem, namely to find the largest amount that cannot be obtained using only n coins of specified coprime denominations (i.e., with gcd equal to one). There is always such a largest value. For the units a=19 and b=538, I ran a basic R code that returned 9665 as the largest impossible value, which happens to be 19×538-538-19, the Sylvester solution to the problem when n=2. A recent paper by Tripathi (2017) manages the case n=3, for “almost all triples”, which decomposes into a myriad of sub-cases. (As an aside, Tripathi (2017) thanks a PhD student, Prof. Thomas W. Cusick, for contributing to the proof, which constitutes a part of his dissertation, but does not explain why he did not join as co-author.) The specific case when a=19, b=101, and c=538 suggested by The Riddler happens to fall in one of the simplest categories since, as ⌊cb⁻¹⌋ and ⌊cb⁻¹⌋ (a) are equal and gcd(a,b)=1 (Lemma 2), the solution is then the same as for the pair (a,b), namely 1799. As this was quite a light puzzle, I went looking for a codegolf challenge that addressed this problem and lo and behold! found one. And proposed the condensed R function

function(a)max((1:(b<-prod(a)))[-apply(combn(outer(a,0:b,"*"),sum(!!a))),2,sum)])

that assumes no duplicate and ordering in the input a. (And learned about *combn* from Robin.) It is of course very inefficient—to the point of crashing R—to look at the upper bound

for the Frobenius number since

is already an upper bound, by Sylvester’s formula. But coding (2) would alas take much more space…

## ABC-SAEM

Posted in Books, Statistics, University life with tags ABC, ABC-Gibbs, ABC-MCMC, Alan Turing, École Polytechnique, EM, JSM 2015, MAP estimators, MCMC, MCMC-SAEM, Monolix, Paris-Saclay campus, PhD thesis, SAEM, Seattle, simulated annealing, stochastic approximation, University of Warwick, well-tempered algorithm on October 8, 2019 by xi'an**I**n connection with the recent PhD thesis defence of Juliette Chevallier, in which I took a somewhat virtual part for being physically in Warwick, I read a paper she wrote with Stéphanie Allassonnière on stochastic approximation versions of the EM algorithm. Computing the MAP estimator can be done via some adapted for simulated annealing versions of EM, possibly using MCMC as for instance in the Monolix software and its MCMC-SAEM algorithm. Where SA stands sometimes for stochastic approximation and sometimes for simulated annealing, originally developed by Gilles Celeux and Jean Diebolt, then reframed by Marc Lavielle and Eric Moulines [friends and coauthors]. With an MCMC step because the simulation of the latent variables involves an untractable normalising constant. (Contrary to this paper, Umberto Picchini and Adeline Samson proposed in 2015 a genuine ABC version of this approach, paper that I thought I missed—although I now remember discussing it with Adeline at JSM in Seattle—, ABC is used as a substitute for the conditional distribution of the latent variables given data and parameter. To be used as a substitute for the Q step of the (SA)EM algorithm. One more approximation step and one more simulation step and we would reach a form of ABC-Gibbs!) In this version, there are very few assumptions made on the approximation sequence, except that it converges with the iteration index to the true distribution (for a fixed observed sample) if convergence of ABC-SAEM is to happen. The paper takes as an illustrative sequence a collection of tempered versions of the true conditionals, but this is quite formal as I cannot fathom a feasible simulation from the tempered version and not from the untempered one. It is thus much more a version of tempered SAEM than truly connected with ABC (although a genuine ABC-EM version could be envisioned).

## noise contrastive estimation

Posted in Statistics with tags Bayesian GANs, CREST, doubly intractable problems, Electronic Journal of Statistics, ENSAE, Langevin MCMC algorithm, noise-contrastive estimation, Paris-Saclay campus, PhD thesis, thesis defence on July 15, 2019 by xi'an**A**s I was attending Lionel Riou-Durand’s PhD thesis defence in ENSAE-CREST last week, I had a look at his papers (!). The 2018 noise contrastive paper is written with Nicolas Chopin (both authors share the CREST affiliation with me). Which compares Charlie Geyer’s 1994 bypassing the intractable normalising constant problem by virtue of an artificial logit model with additional simulated data from another distribution *ψ*.

“Geyer (1994) established the asymptotic properties of the MC-MLE estimates under general conditions; in particular that the x’s are realisations of an ergodic process. This is remarkable, given that most of the theory on M-estimation (i.e.estimation obtained by maximising functions) is restricted to iid data.”

Michael Guttman and Aapo Hyvärinen also use additional simulated data in another likelihood of a logistic classifier, called noise contrastive estimation. Both methods replace the unknown ratio of normalising constants with an unbiased estimate based on the additional simulated data. The major and impressive result in this paper [now published in the Electronic Journal of Statistics] is that the noise contrastive estimation approach always enjoys a smaller variance than Geyer’s solution, at an equivalent computational cost when the actual data observations are iid. And the artificial data simulations ergodic. The difference between both estimators is however negligible against the Monte Carlo error (Theorem 2).

This may be a rather naïve question, but I wonder at the choice of the alternative distribution *ψ*. With a vague notion that it could be optimised in a GANs perspective. A side result of interest in the paper is to provide a minimal (re)parameterisation of the truncated multivariate Gaussian distribution, if only as an exercise for future exams. Truncated multivariate Gaussian for which the normalising constant is of course unknown.