arXiv subjects

Nicola D. Walker

Publications and source records attributed to Nicola D. Walker.

2 recordsLinked to original sources

arXiv2026

Exploring Pareto smoothing in sequential Monte Carlo

A popular technique for reducing the variance of importance sampling (IS) estimators is to modify the weights of some importance points. One approach is to truncate the largest weights, which reduces variance but can introduce substantial bias. Pareto smoothed importance sampling (PSIS), by contrast, reduces the variance of the weights by fitting a generalised Pareto distribution to the upper tail of the weight distribution and replacing the weights in this tail with the corresponding expected quantiles from the fitted distribution. PSIS can therefore also reduce variance, but typically with less bias, and has been used successfully in IS-based approximations for Bayesian cross-validation. This paper explores the use of PSIS steps within sequential Monte Carlo (SMC) samplers, with a particular focus on approximate Bayesian computation (ABC)-SMC algorithms, where we aim to use Pareto smoothing to reduce the use of Markov chain Monte Carlo (MCMC) moves, each of which requires simulation from a model that is often computationally expensive. Our empirical investigation suggests that there are only minimal benefits to using Pareto smoothing in SMC, since the variance reduction through using a sequence of targets dominates the impact of the weight adjustment.

stat.CO↗

arXiv2020

Fish should not be in isolation: Calculating maximum sustainable yield using an ensemble model

Many jurisdictions have a legal requirement to manage fish stocks to maximum sustainable yield (MSY). Generally, MSY is calculated on a single-species basis, however in reality, the yield of one species depends, not only on its own fishing level, but that of other species. We show that bold assumptions about the effect of interacting species on MSY are made when managing on a single-species basis, often leading to inconsistent and conflicting advice, demonstrating the requirement of a multispecies MSY (MMSY). Although there are several definitions of MMSY, there is no consensus. Furthermore, calculating a MMSY can be difficult as there are many models, of varying complexity, each with their own strengths and weaknesses, and the value if MMSY can be sensitive to the model used. Here, we use an ensemble model to combine different multispecies models, exploiting their individual strengths and quantifying their uncertainties and discrepancies, to calculate a more robust MMSY. We demonstrate this by calculating a MMSY for nine species in the North Sea. We found that it would be impossible to fish at single-species MSY and that MMSY led to higher yields and revenues than current levels.

stat.AP↗