## unbiased consistent nested sampling via sequential Monte Carlo

Posted in pictures, Statistics, Travel with tags , , , , , , , , on June 12, 2018 by xi'an

“Moreover, estimates of the marginal likelihood are unbiased.” (p.2)

Rob Salomone, Leah South, Chris Drovandi and Dirk Kroese (from QUT and UQ, Brisbane) recently arXived a paper that frames the nested sampling in such a way that marginal likelihoods can be unbiasedly (and consistently) estimated.

“Why isn’t nested sampling more popular with statisticians?” (p.7)

A most interesting question, especially given its popularity in cosmology and other branches of physics. A first drawback pointed out in the c is the requirement of independence between the elements of the sample produced at each iteration. 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.

A second difficulty is the use of quadrature which turns integrand into step functions at random slices. Indeed, mixing Monte Carlo with numerical integration makes life much harder, as shown by the early avatars of nested sampling that only accounted for the numerical errors. (And which caused Nicolas and I to write our critical paper in Biometrika.) There are few studies of that kind in the literature, the only one I can think of being [my former PhD student] Anne Philippe‘s thesis twenty years ago.

The third issue stands with the difficulty in parallelising the method. Except by jumping k points at once, rather than going one level at a time. While I agree this makes life more complicated, I am also unsure about the severity of that issue as k nested sampling algorithms can be run in parallel and aggregated in the end, from simple averaging to something more elaborate.

The final blemish is that the nested sampling estimator has a stopping mechanism that induces a truncation error, again maybe a lesser problem given the overall difficulty in assessing the total error.

The paper takes advantage of the ability of SMC to produce unbiased estimates of a sequence of normalising constants (or of the normalising constants of a sequence of targets). For nested sampling, the sequence is made of the prior distribution restricted to an embedded sequence of level sets. With another sequence restricted to bands (likelihood between two likelihood boundaries). If all restricted posteriors of the second kind and their normalising constant are known, the full posterior is known. Apparently up to the main normalising constant, i.e. the marginal likelihood., , except that it is also the sum of all normalising constants. Handling this sequence by SMC addresses the four concerns of the four authors, apart from the truncation issue, since the largest likelihood bound need be set for running the algorithm.

When the sequence of likelihood bounds is chosen based on the observed likelihoods so far, the method becomes adaptive. Requiring again the choice of a stopping rule that may induce bias if stopping occurs too early. 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).

The remainder of this long paper (61 pages!) is dedicated to practical implementation, calibration and running a series of comparisons. A nice final touch is the thanks to the ‘Og for its series of posts on nested sampling, which “helped influence this work, and played a large part in inspiring it.”

In conclusion, this paper is certainly a worthy exploration of the nested sampler, providing further arguments towards a consistent version, with first and foremost an (almost?) unbiased resolution. The comparison with a wide range of alternatives remains open, in particular time-wise, if evidence is the sole target of the simulation. For instance, the choice of this sequence of targets in an SMC may be improved by another sequence, since changing one particle at a time does not sound efficient. The complexity of the implementation and in particular of the simulation from the prior under more and more stringent constraints need to be addressed.

Posted in Books, Kids, Statistics, University life with tags , , , , , , , , , , on December 5, 2017 by xi'an

I am a PhD student in biostatistics, and an avid reader of your work. I recently came across this blog post, where you review a text on statistical paradoxes, and I was struck by this section:

“For instance, the author considers the MLE being biased to be a paradox (p.117), while omitting the much more substantial “paradox” of the non-existence of unbiased estimators of most parameters—which simply means unbiasedness is irrelevant. Or the other even more puzzling “paradox” that the secondary MLE derived from the likelihood associated with the distribution of a primary MLE may differ from the primary. (My favourite!)”

I found this section provocative, but I am unclear on the nature of these “paradoxes”. I reviewed my stat inference notes and came across the classic example that there is no unbiased estimator for 1/p w.r.t. a binomial distribution, but I believe you are getting at a much more general result. If it’s not too much trouble, I would sincerely appreciate it if you could point me in the direction of a reference or provide a bit more detail for these two “paradoxes”.

The text is Chang’s Paradoxes in Scientific Inference, which I indeed reviewed negatively. To answer about the bias “paradox”, it is indeed a neglected fact that, while the average of any transform of a sample obviously is an unbiased estimator of its mean (!), the converse does not hold, namely, an arbitrary transform of the model parameter θ is not necessarily enjoying an unbiased estimator. In Lehmann and Casella, Chapter 2, Section 4, this issue is (just slightly) discussed. But essentially, transforms that lead to unbiased estimators are mostly the polynomial transforms of the mean parameters… (This also somewhat connects to a recent X validated question as to why MLEs are not always unbiased. Although the simplest explanation is that the transform of the MLE is the MLE of the transform!) In exponential families, I would deem the range of transforms with unbiased estimators closely related to the collection of functions that allow for inverse Laplace transforms, although I cannot quote a specific result on this hunch.

The other “paradox” is that, if h(X) is the MLE of the model parameter θ for the observable X, the distribution of h(X) has a density different from the density of X and, hence, its maximisation in the parameter θ may differ. An example (my favourite!) is the MLE of ||a||² based on x N(a,I) which is ||x||², a poor estimate, and which (strongly) differs from the MLE of ||a||² based on ||x||², which is close to (1-p/||x||²)²||x||² and (nearly) admissible [as discussed in the Bayesian Choice].

## unbiased HMC

Posted in Books, pictures, Statistics with tags , , , , , , , on September 25, 2017 by xi'an

Jeremy Heng and Pierre Jacob arXived last week a paper on unbiased Hamiltonian Monte Carlo by coupling, following the earlier paper of Pierre and co-authors on debiasing by coupling a few weeks ago. The coupling within the HMC amounts to running two HMC chains with common random numbers, plus subtleties!

“As with any other MCMC method, HMC estimators are justified in the limit of the number of iterations. Algorithms which rely on such asymptotics face the risk of becoming obsolete if computational power keeps increasing through the number of available processors and not through clock speed.”

The main difficulty here is to have both chains meet (exactly) with large probability, since coupled HMC can only bring these chain close to one another. The trick stands in using both coupled HMC and coupled Hastings-Metropolis kernels, since the coupled MH kernel allows for exact meetings when the chains are already close, after which they remain happily and forever together! The algorithm is implemented by choosing between the kernels at random at each iteration. (Unbiasedness follows by the Glynn-Rhee trick, which is eminently well-suited for coupling!) As pointed out from the start of the paper, the appeal of this unbiased version is that the algorithm can be (embarrassingly) parallelised since all processors in use return estimators that are iid copies of one another, hence easily merged into a better estimator.

## X-Outline of a Theory of Statistical Estimation

Posted in Books, Statistics, University life with tags , , , , , , , , , , on March 23, 2017 by xi'an

While visiting Warwick last week, Jean-Michel Marin pointed out and forwarded me this remarkable paper of Jerzy Neyman, published in 1937, and presented to the Royal Society by Harold Jeffreys.

“Leaving apart on one side the practical difficulty of achieving randomness and the meaning of this word when applied to actual experiments…”

“It may be useful to point out that although we are frequently witnessing controversies in which authors try to defend one or another system of the theory of probability as the only legitimate, I am of the opinion that several such theories may be and actually are legitimate, in spite of their occasionally contradicting one another. Each of these theories is based on some system of postulates, and so long as the postulates forming one particular system do not contradict each other and are sufficient to construct a theory, this is as legitimate as any other. “

This paper is fairly long in part because Neyman starts by setting Kolmogorov’s axioms of probability. This is of historical interest but also needed for Neyman to oppose his notion of probability to Jeffreys’ (which is the same from a formal perspective, I believe!). He actually spends a fair chunk on explaining why constants cannot have anything but trivial probability measures. Getting ready to state that an a priori distribution has no meaning (p.343) and that in the rare cases it does it is mostly unknown. While reading the paper, I thought that the distinction was more in terms of frequentist or conditional properties of the estimators, Neyman’s arguments paving the way to his definition of a confidence interval. Assuming repeatability of the experiment under the same conditions and therefore same parameter value (p.344).

“The advantage of the unbiassed [sic] estimates and the justification of their use lies in the fact that in cases frequently met the probability of their differing very much from the estimated parameters is small.”

“…the maximum likelihood estimates appear to be what could be called the best “almost unbiassed [sic]” estimates.”

It is also quite interesting to read that the principle for insisting on unbiasedness is one of producing small errors, because this is not that often the case, as shown by the complete class theorems of Wald (ten years later). And that maximum likelihood is somewhat relegated to a secondary rank, almost unbiased being understood as consistent. A most amusing part of the paper is when Neyman inverts the credible set into a confidence set, that is, turning what is random in a constant and vice-versa. With a justification that the credible interval has zero or one coverage, while the confidence interval has a long-run validity of returning the correct rate of success. What is equally amusing is that the boundaries of a credible interval turn into functions of the sample, hence could be evaluated on a frequentist basis, as done later by Dennis Lindley and others like Welch and Peers, but that Neyman fails to see this and turn the bounds into hard values. For a given sample.

“This, however, is not always the case, and in general there are two or more systems of confidence intervals possible corresponding to the same confidence coefficient α, such that for certain sample points, E’, the intervals in one system are shorter than those in the other, while for some other sample points, E”, the reverse is true.”

The resulting construction of a confidence interval is then awfully convoluted when compared with the derivation of an HPD region, going through regions of acceptance that are the dual of a confidence interval (in the sampling space), while apparently [from my hasty read] missing a rule to order them. And rejecting the notion of a confidence interval being possibly empty, which, while being of practical interest, clashes with its frequentist backup.

## exact, unbiased, what else?!

Posted in Books, Statistics, University life with tags , , , , , , , , on April 13, 2016 by xi'an

Last week, Matias Quiroz, Mattias Villani, and Robert Kohn arXived a paper on exact subsampling MCMC, a paper that contributes to the current literature on approximating MCMC samplers for large datasets, in connection with an earlier paper of Quiroz et al. discussed here last week.

The “exact” in the title is to be understood in the Russian roulette sense. By using Rhee and Glynn debiaising device, the authors achieve an unbiased estimator of the likelihood as in Bardenet et al. (2015). The central tool for the derivation of an unbiased and positive estimator is to find a control variate for each component of the log likelihood that is good enough for the difference between the component and the control to be lower bounded. By the constant a in the screen capture above. When the individual terms d in the product are iid unbiased estimates of the log likelihood difference. And q is the sum of the control variates. Or maybe more accurately of the cheap substitutes to the exact log likelihood components. Thus still of complexity O(n), which makes the application to tall data more difficult to contemplate.

The \$64 question is obviously how to produce cheap and efficient control variates that kill the curse of the tall data. (It still irks to resort to this term of control variate, really!) Section 3.2 in the paper suggests clustering the data and building an approximation for each cluster, which seems to imply manipulating the whole dataset at this early stage. At a cost of O(Knd). Furthermore, because finding a correct lower bound a is close to impossible in practice, the authors use a “soft lower bound”, meaning that it is only an approximation and thus that (3.4) above can get negative from time to time, which cancels the validation of the method as a pseudo-marginal approach. The resolution of this difficulty is to resort to the same proxy as in the Russian roulette paper, replacing the unbiased estimator with its absolute value, an answer I already discussed for the Russian roulette paper. An additional step is proposed by Quiroz et al., namely correlating the random numbers between numerator and denominator in their final importance sampling estimator, via a Gaussian copula as in Deligiannidis et al.

This paper made me wonder (idly wonder, mind!) anew how to get rid of the vexing unbiasedness requirement. From a statistical and especially from a Bayesian perspective, unbiasedness is a second order property that cannot be achieved for most transforms of the parameter θ. And that does not keep under reparameterisation. It is thus vexing and perplexing that unbiased is so central to the validation of our Monte Carlo technique and that any divergence from this canon leaves us wandering blindly with no guarantee of ever reaching the target of the simulation experiment…