**W**ith Charly Andral (PhD, Paris Dauphine), Randal Douc, and Hugo Marival (PhD, Telecom SudParis), we just arXived a paper on importance Markov chains that merges importance sampling and MCMC. An idea already mentioned in Hastings (1970) and even earlier in Fodsick (1963), and later exploited in Liu et al. (2003) for instance. And somewhat dual of the vanilla Rao-Backwellisation paper Randal and I wrote a (long!) while ago. Given a target π with a dominating measure π⁰≥Mπ, using a Markov kernel to simulate from this dominating measure and subsampling by the importance weight ρ does produce a new Markov chain with the desired target measure as invariant distribution. However, the domination assumption is rather unrealistic and a generic approach can be implemented without it, by defining an extended Markov chain, with the addition of the number N of replicas as the supplementary term… And a transition kernel R(n|x) on N with expectation ρ, which is a minimal(ist) assumption for the validation of the algorithm.. While this initially defines a semi-Markov chain, an extended Markov representation is also feasible, by decreasing N one by one until reaching zero, and this is most helpful in deriving convergence properties for the resulting chain, including a CLT. While the choice of the kernel R is free, the optimal choice is associated with residual sampling, where only the fractional part of ρ is estimated by a Bernoulli simulation.

## Archive for geometric ergodicity

## important Markov chains

Posted in Books, Statistics, University life with tags arXiv, CLT, geometric ergodicity, importance sampling, Law of Large Numbers, Markov kernel, MCMC, PhD students, residual sampling, semi-Markov chain, Université Paris Dauphine, vanilla Rao-Blackwellisation on July 21, 2022 by xi'an## Siem Reap conference

Posted in Kids, pictures, Travel, University life with tags Arkhangelsk, Bayes factors, Bayesian model choice, Bayesian model comparison, Cambodia, conference, CREST, Data Science and Finance conference, geometric ergodicity, group picture, Hyvärinen score, India, krama, Langevin MCMC algorithm, NGO, Pakistan, Sala Baï school, Siem Reap, Wasserstein distance on March 8, 2019 by xi'an**A**s I returned from the conference in Siem Reap. on a flight avoiding India and Pakistan and their [brittle and bristling!] boundary on the way back, instead flying far far north, near Arkhangelsk (but with nothing to show for it, as the flight back was fully in the dark), I reflected how enjoyable this conference had been, within a highly friendly atmosphere, meeting again with many old friends (some met prior to the creation of CREST) and new ones, a pleasure not hindered by the fabulous location near Angkor of course. (The above picture is the “last hour” group picture, missing a major part of the participants, already gone!)

Among the many talks, Stéphane Shao gave a great presentation on a paper [to appear in JASA] jointly written with Pierre Jacob, Jie Ding, and Vahid Tarokh on the Hyvärinen score and its use for Bayesian model choice, with a highly intuitive representation of this divergence function (which I first met in Padua when Phil Dawid gave a talk on this approach to Bayesian model comparison). Which is based on the use of a divergence function based on the squared error difference between the gradients of the true log-score and of the model log-score functions. Providing an alternative to the Bayes factor that can be shown to be consistent, even for some non-iid data, with some gains in the experiments represented by the above graph.

Arnak Dalalyan (CREST) presented a paper written with Lionel Riou-Durand on the convergence of non-Metropolised Langevin Monte Carlo methods, with a new discretization which leads to a substantial improvement of the upper bound on the sampling error rate measured in Wasserstein distance. Moving from p/ε to √p/√ε in the requested number of steps when p is the dimension and ε the target precision, for smooth and strongly log-concave targets.

This post gives me the opportunity to advertise for the NGO Sala Baï hostelry school, which the whole conference visited for lunch and which trains youths from underprivileged backgrounds towards jobs in hostelery, supported by donations, companies (like Krama Krama), or visiting the Sala Baï restaurant and/or hotel while in Siem Reap.

## non-reversible Langevin samplers

Posted in Books, pictures, Statistics, Travel, University life with tags England, Foster-Lyapunov assumptions, geometric ergodicity, Langevin diffusion, Lie-Trotter splitting, MCMC, Metropolis-Hastings algorithm, non-reversible diffusion, Oxford, reversibility, train on February 6, 2017 by xi'an**I**n the train to Oxford yesterday night, I read through the recently arXived Duncan et al.’s *Nonreversible Langevin Samplers: Splitting Schemes, Analysis and Implementation*. Standing up the whole trip in the great tradition of British trains.

The paper is fairly theoretical and full of Foster-Lyapunov assumptions but aims at defending an approach based on a non-reversible diffusion. One idea is that the diffusion based on the drift {∇ log π(x) + γ(x)} is associated with the target π provided

∇ . {π(x)γ(x)} = 0

which holds for the Langevin diffusion when γ(x)=0, but produces a non-reversible process in the alternative. The Langevin choice γ(x)=0 happens to be the worst possible when considering the asymptotic variance. In practice however the diffusion need be discretised, which induces an approximation that may be catastrophic for convergence if not corrected, and a relapse into reversibility if corrected by Metropolis. The proposal in the paper is to use a Lie-Trotter splitting I had never heard of before to split between reversible [∇ log π(x)] and non-reversible [γ(x)] parts of the process. The deterministic part is chosen as γ(x)=∇ log π(x) [but then what is the point since this is Langevin?] or as the gradient of a power of π(x). Although I was mostly lost by that stage, the paper then considers the error induced by a numerical integrator related with this deterministic part, towards deriving asymptotic mean and variance for the splitting scheme. On the unit hypercube. Although the paper includes a numerical example for the warped normal target, I find it hard to visualise the implementation of this scheme. Having obviously not heeded Nicolas’ and James’ advice, the authors also analyse the Pima Indian dataset by a logistic regression!)

## Sampling latent states for high-dimensional non-linear state space models with the embedded HMM method

Posted in Books, pictures, Statistics, University life with tags ABC, arXiv, ensemble Monte Carlo, forward-backward formula, geometric ergodicity, hidden Markov models, HMM, Radford Neal, University of Toronto on March 17, 2016 by xi'an**P**reviously, I posted a comment on a paper by Alex Shestopaloff and Radford Neal, after my visit to Toronto two years ago, using a particular version of ensemble Monte Carlo. A new paper by the same authors was recently arXived, as an refinement of the embedded HMM paper of Neal (2003), in that the authors propose a new and more efficient way to generate from the (artificial) embedded hidden Markov sampler that is central to their technique of propagating a set of pool states. The method exploits both forward and backward representations of HMMs in an alternating manner. And propagates the pool states from one observation time to the next. The paper also exploits latent Gaussian structures to make autoregressive proposals, as well as flip proposals from x to -x [which seem to only make sense when 0 is a central value for the target, i.e. when the observables y only depend on |x|]. All those modifications bring the proposal quite close to (backward) particle Gibbs, the difference being in using Metropolis rather than importance steps. And in an improvement brought by the embedded HMM approach, even though it is always delicate to generalise those comparisons when some amount of calibration is required by both algorithms under comparison. (Especially delicate when it is rather remote from my area of expertise!) Anyway, I am still intrigued [in a positive way] by the embedded HMM idea as it remains mysterious that a finite length HMM simulation can improve the convergence performances that much. And wonder at a potential connection with an earlier paper of Anthony Lee and Krys Latuszynski using a random number of auxiliary variables. Presumably a wrong impression from a superficial memory…

## stability of noisy Metropolis-Hastings

Posted in Books, Statistics, Travel, University life with tags geometric ergodicity, Monte Carlo within Metropolis, noisy Metropolis-Hastings algorithm, pseudo-marginal MCMC on April 3, 2015 by xi'an**F**elipe Medina-Aguayo, Anthony Lee and Gareths Roberts, all from Warwick, arXived last Thursday a paper on the stability properties of noisy Metropolis-Hastings algorithms. The validation of unbiased estimators of the target à la Andrieu and Roberts (2009, AoS)—often discussed here—is in fact obvious when following the auxiliary variable representation of Andrieu and Vihola (2015, AoAP). Assuming the unbiased estimator of the target is generated conditional on the proposed value in the original Markov chain. The noisy version of the above means refreshing the unbiased estimator at each iteration. It also goes under the name of Monte Carlo within Metropolis. The difficulty with this noisy version is that it is not exact, i.e., does not enjoy the true target as its marginal stationary distribution. The paper by Medina-Aguayo, Lee and Roberts focusses on its validation or invalidation (with examples of transient noisy versions). Under geometric ergodicity of the marginal chain, plus some stability in the weights, the noisy version is also geometrically ergodic. A drift condition on the proposal kernel is also sufficient. Under (much?) harder conditions, the limiting distribution of the noisy chain is asymptotically in the number of unbiased estimators the true target. The result is thus quite interesting in that it provides sufficient convergence conditions, albeit not always easy to check in realistic settings.