Fusion at CIRM

Today is the first day of the FUSION workshop Rémi Bardenet and myself organised. Due to schedule clashes, I will alas not be there, since [no alas!] at the BNP conference in Chili. The program and collection of participants is quite exciting and I hope more fusion will result from this meeting. Enjoy! (And beware of boars, cold water, and cliffs!!!)

likelihood-free nested sampling

Last week, I came by chance across a paper by Jan Mikelson and Mustafa Khammash on a likelihood-free version of nested sampling (a popular keyword on the ‘Og!). Published in 2020 in PLoS Comput Biol. The setup is a parameterised and hidden state-space model, which allows for an approximation of the (observed) likelihood function L(θ|y) by means of a particle filter. An immediate issue with this proposal is that a novel  filter need be produced for a new value of the parameter θ, which makes it enormously expensive. It then gets more bizarre as the [Monte Carlo] distribution of the particle filter approximation ô(θ|y) is agglomerated with the original prior π(θ) as a joint “prior” [despite depending on the observed y] and a nested sampling is conducted with level sets of the form

ô(θ|y)>ε.

Actually, if the Monte Carlo error was null, that is, if the number of particles was infinite,

ô(θ|y)=L(θ|y)

implies that this is indeed the original nested sampler. Simulation from the restricted region is done by constructing an extra density estimator of the constrained distribution (in θ)…

“We have shown how using a Monte Carlo estimate over the livepoints not only results in an unbiased estimator of the Bayesian evidence Z, but also allows us to derive a formulation for a lower bound on the achievable variance in each iteration (…)”

As shown by the above the authors insist on the unbiasedness of the particle approximation, but since nested sampling is not producing an unbiased estimator of the evidence Z, the point is somewhat moot. (I am also rather surprised by the reported lack of computing time benefit in running ABC-SMC.)

no filter [no jatp]

IMS workshop [day 4]

While I did not repeat the mistake of yesterday morning, just as well because the sun was unbearably strong!, I managed this time to board a bus headed in the wrong direction and as a result went through several remote NUS campi! Missing the first talk of the day as a result. By Youssef Marzouk, with a connection between sequential Monte Carlo and optimal transport. Transport for sampling, that is. The following talk by Tiangang Cui was however related, with Marzouk a co-author, as it aimed at finding linear transforms towards creating Normal approximations to the target to be used as proposals in Metropolis algorithms. Which may sound like something already tried a zillion times in the MCMC literature, except that the setting was rather specific to some inverse problems, imposing a generalised Normal structure on the transform, then optimised by transport arguments. It is unclear to me [from just attending the talk] how complex this derivation is and how dimension steps in, but the produced illustrations were quite robust to an increase in dimension.

The remaining talks for the day were mostly particular, from Anthony Lee introducing a new and almost costless way of producing variance estimates in particle filters, exploiting only the ancestry of particles, to Mike Pitt discussing the correlated pseudo-marginal algorithm developed with George Deligiannidis and Arnaud Doucet. Which somewhat paradoxically managed to fight the degeneracy [i.e., the need for a number of terms increasing like the time index T] found in independent pseudo-marginal resolutions, moving down to almost log(T)… With an interesting connection to the quasi SMC approach of Mathieu and Nicolas. And Sebastian Reich also stressed the links with optimal transport in a talk about data assimilation that was way beyond my reach. The day concluded with fireworks, through a magistral lecture by Professeur Del Moral on a continuous time version of PMCMC using the Feynman-Kac terminology. Pierre did a superb job during his lecture towards leading the whole room to the conclusion.

unbiased consistent nested sampling via sequential Monte Carlo [a reply]

Rob Salomone sent me the following reply on my comments of yesterday about their recently arXived paper.

Our main goal in the paper was to show that Nested Sampling (when interpreted a certain way) is really just a member of a larger class of SMC algorithms, and exploring the consequences of that. We should point out that the section regarding calibration applies generally to SMC samplers, and hope that people give those techniques a try regardless of their chosen SMC approach.

Regarding your question about “whether or not it makes more sense to get completely SMC and forego any nested sampling flavour!”, this is an interesting point. After all, if Nested Sampling is just a special form of SMC, why not just use more standard SMC approaches? It seems that the Nested Sampling’s main advantage is its ability to cope with problems that have “phase transition’’ like behaviour, and thus is robust to a wider range of difficult problems than annealing approaches. Nevertheless, we hope this way of looking at NS (and showing that there may be variations of SMC with certain advantages) leads to improved NS and SMC methods down the line.  

Regarding your post, I should clarify a point regarding unbiasedness. The largest likelihood bound is actually set to infinity. Thus, for the fixed version of NS—SMC, one has an unbiased estimator of the “final” band. Choosing a final band prematurely will of course result in very high variance. However, the estimator is unbiased. For example, consider NS—SMC with only one strata. Then, the method reduces to simply using the prior as an importance sampling distribution for the posterior (unbiased, but often high variance).

Comments related to two specific parts of your post are below (your comments in italicised bold):

“Which never occurred as the number one difficulty there, as the simplest implementation runs a Markov chain from the last removed entry, independently from the remaining entries. Even stationarity is not an issue since I believe that the first occurrence within the level set is distributed from the constrained prior.”

This is an interesting point that we had not considered! In practice, and in many papers that apply Nested Sampling with MCMC, the common approach is to start the MCMC at one of the randomly selected “live points”, so the discussion related to independence was in regard to these common implementations.

Regarding starting the chain from outside of the level set. This is likely not done in practice as it introduces an additional difficulty of needing to propose a sample inside the required region (Metropolis–Hastings will have non—zero probability of returning a sample that is still outside the constrained region for any fixed number of iterations). Forcing the continuation of MCMC until a valid point is proposed I believe will be a subtle violation of detailed balance. Of course, the bias of such a modification may be small in practice, but it is an additional awkwardness introduced by the requirement of sample independence!

“And then, in a twist that is not clearly explained in the paper, the focus moves to an improved nested sampler that moves one likelihood value at a time, with a particle step replacing a single  particle. (Things get complicated when several particles may take the very same likelihood value, but randomisation helps.) At this stage the algorithm is quite similar to the original nested sampler. Except for the unbiased estimation of the constants, the  final constant, and the replacement of exponential weights exp(-t/N) by powers of (N-1/N)”

Thanks for pointing out that this isn’t clear, we will try to do better in the next revision! The goal of this part of the paper wasn’t necessarily to propose a new version of nested sampling. Our focus here was to demonstrate that NS–SMC is not simply the Nested Sampling idea with an SMC twist, but that the original NS algorithm with MCMC (and restarting the MCMC sampling at one of the “live points’” as people do in practice) actually is a special case of SMC (with the weights replaced with a suboptimal choice).

The most curious thing is that, as you note, the estimates of remaining prior mass in the SMC context come out as powers of (N-1)/N and not exp(-t/N). In the paper by Walter (2017), he shows that the former choice is actually superior in terms of bias and variance. It was a nice touch that the superior choice of weights came out naturally in the SMC interpretation! 

That said, as the fixed version of NS-SMC is the one with the unbiasedness and consistency properties, this was the version we used in the main statistical examples.