arXiv subjects

Peter Challenor

Publications and source records attributed to Peter Challenor.

At least 19 recordsLinked to original sources

arXiv2026

Bayesian Emulation of Multi-fidelity Earth System Modelling Using Hierarchical Gaussian Processes

Multi-fidelity Earth system models provide simulations at different levels of complexity and computational cost, but exhaustive exploration of the parameter space at the highest fidelity is often prohibitively expensive. Multi-fidelity emulators can reduce this burden by combining abundant lower-fidelity simulations with limited high-fidelity evaluations. We compare four Gaussian-process-based multi-fidelity approaches: the Kennedy--O'Hagan autoregressive model (K&O), hierarchical kriging (HK), Bayesian hierarchical emulation for multi-level models (BayHEm), and multi-fidelity deep Gaussian processes (MF-DGP). We evaluate the methods using two contrasting applications: a three-fidelity tsunami simulator and a two-fidelity implementation of the Joint UK Land Environment Simulator (JULES). Performance is assessed using leave-one-out predictive accuracy, uncertainty representation, design requirements, and computational characteristics. In the tsunami application, BayHEm gives the lowest normalised root mean square error (NRMSE = 0.031) and highest SCORE (3.025), while MF-DGP performs worse than the single-fidelity baseline when only 10 high-fidelity simulations are available. In the JULES application, MF-DGP gives the lowest NRMSE (0.079) and highest SCORE (3.032), with all 30 held-out high-fidelity observations lying within their nominal 95\% predictive intervals. These contrasting results show that no single multi-fidelity emulator is uniformly superior. Instead, method choice should reflect the complexity of the inter-fidelity relationship, the amount of high-fidelity information available, and the importance placed on predictive accuracy and uncertainty quantification.

stat.AP↗

arXiv2025

Bayesian Hierarchical Emulators for Multi-Level Models: BayHEm

Decision making often uses complex computer codes run at the exa-scale (10e18 flops). Such computer codes or models are often run in a hierarchy of different levels of fidelity ranging from the basic to the very sophisticated. The top levels in this hierarchy are expensive to run, limiting the number of possible runs. To make use of runs over all levels, and crucially improve emulation at the top level, we use multi-level Gaussian process emulators (GPs). We will present a new method of building GP emulators from hierarchies of models. In order to share information across the different levels, l=1,...,L, we define the form of the prior of the l+1th level to be the posterior of the lth level, hence building a Bayesian hierarchical structure for the top Lth level. This enables us to not only learn about the GP hyperparameters as we move up the multi-level hierarchy, but also allows us to limit the total number of parameters in the full model, whilst maintaining accuracy.

stat.ME↗

arXiv2024

Uncertainty Quantification for Agent Based Models: A Tutorial

We explore the application of uncertainty quantification methods to agent-based models (ABMs) using a simple sheep and wolf predator-prey model. This work serves as a tutorial on how techniques like emulation can be powerful tools in this context. We also highlight the importance of advanced statistical methods in effectively utilising computationally expensive ABMs. Specifically, we implement stochastic Gaussian processes, Gaussian process classification, sequential design, and history matching to address uncertainties in model input parameters and outputs. Our results show that these methods significantly enhance the robustness, accuracy, and predictive power of ABMs.

stat.OT↗

arXiv2023

Propagating moments in probabilistic graphical models with polynomial regression forms for decision support systems

Probabilistic graphical models are widely used to model complex systems under uncertainty. Traditionally, Gaussian directed graphical models are applied for analysis of large networks with continuous variables as they can provide conditional and marginal distributions in closed form simplifying the inferential task. The Gaussianity and linearity assumptions are often adequate, yet can lead to poor performance when dealing with some practical applications. In this paper, we model each variable in graph G as a polynomial regression of its parents to capture complex relationships between individual variables and with a utility function of polynomial form. We develop a message-passing algorithm to propagate information throughout the network solely using moments which enables the expected utility scores to be calculated exactly. Our propagation method scales up well and enables to perform inference in terms of a finite number of expectations. We illustrate how the proposed methodology works with examples and in an application to decision problems in energy planning and for real-time clinical decision support.

stat.ME↗

arXiv2023

Cross-Validation Based Adaptive Sampling for Multi-Level Gaussian Process Models

Complex computer codes or models can often be run in a hierarchy of different levels of complexity ranging from the very basic to the sophisticated. The top levels in this hierarchy are typically expensive to run, which limits the number of possible runs. To make use of runs over all levels, and crucially improve predictions at the top level, we use multi-level Gaussian process emulators (GPs). The accuracy of the GP greatly depends on the design of the training points. In this paper, we present a multi-level adaptive sampling algorithm to sequentially increase the set of design points to optimally improve the fit of the GP. The normalised expected leave-one-out cross-validation error is calculated at all unobserved locations, and a new design point is chosen using expected improvement combined with a repulsion function. This criterion is calculated for each model level weighted by an associated cost for the code at that level. Hence, at each iteration, our algorithm optimises for both the new point location and the model level. The algorithm is extended to batch selection as well as single point selection, where batches can be designed for single levels or optimally across all levels.

stat.ME↗

arXiv2022

Modelling Numerical Systems with Two Distinct Labelled Output Classes

We present a new method of modelling numerical systems where there are two distinct output solution classes, for example tipping points or bifurcations. Gaussian process emulation is a useful tool in understanding these complex systems and provides estimates of uncertainty, but we aim to include systems where there are discontinuities between the two output solutions. Due to continuity assumptions, we consider current methods of classification to split our input space into two output regions. Classification and logistic regression methods currently rely on drawing from an independent Bernoulli distribution, which neglects any information known in the neighbouring area. We build on this by including correlation between our input points. Gaussian processes are still a vital element, but used in latent space to model the two regions. Using the input values and an associated output class label, the latent variable is estimated using MCMC sampling and a unique likelihood. A threshold (usually at zero) defines the boundary. We apply our method to a motivating example provided by the hormones associated with the reproductive system in mammals, where the two solutions are associated with high and low rates of reproduction.

stat.ME↗

arXiv2022

Modelling Correlated Bernoulli Data Part II: Inference

Binary data are highly common in many applications, however it is usually modelled with the assumption that the data are independently and identically distributed. This is typically not the case in many real-world examples and such the probability of a success can be dependent on the outcome successes of past events. The de Bruijn process (DBP) was introduced in Kimpton et al. [2022]. This is a correlated Bernoulli process which can be used to model binary data with known correlation. The correlation structures are included through the use of de Bruijn graphs, giving an extension to Markov chains. Given the DBP and an observed sequence of binary data, we present a method of inference using Bayes' factors. Results are applied to the Oxford and Cambridge annual boat race.

stat.ME↗

arXiv2022

Binary De Bruijn Processes

Binary time series data are very common in many applications, and are typically modelled independently via a Bernoulli process with a single probability of success. However, the probability of a success can be dependent on the outcome successes of past events. Presented here is a novel approach for modelling binary time series data called a binary de Bruijn process which takes into account temporal correlation. The structure is derived from de Bruijn Graphs - a directed graph, where given a set of symbols, V, and a 'word' length, m, the nodes of the graph consist of all possible sequences of V of length m. De Bruijn Graphs are equivalent to mth order Markov chains, where the 'word' length controls the number of states that each individual state is dependent on. This increases correlation over a wider area. To quantify how clustered a sequence generated from a de Bruijn process is, the run lengths of letters are observed along with run length properties. Inference is also presented along with two application examples: precipitation data and the Oxford and Cambridge boat race.

stat.ME↗

arXiv2022

Sequential adaptive design for emulating costly computer codes

Gaussian processes (GPs) are generally regarded as the gold standard surrogate model for emulating computationally expensive computer-based simulators. However, the problem of training GPs as accurately as possible with a minimum number of model evaluations remains challenging. We address this problem by suggesting a novel adaptive sampling criterion called VIGF (variance of improvement for global fit). The improvement function at any point is a measure of the deviation of the GP emulator from the nearest observed model output. At each iteration of the proposed algorithm, a new run is performed where VIGF is the largest. Then, the new sample is added to the design and the emulator is updated accordingly. A batch version of VIGF is also proposed which can save the user time when parallel computing is available. Additionally, VIGF is extended to the multi-fidelity case where the expensive high-fidelity model is predicted with the assistance of a lower fidelity simulator. This is performed via hierarchical kriging. The applicability of our method is assessed on a bunch of test functions and its performance is compared with several sequential sampling strategies. The results suggest that our method has a superior performance in predicting the benchmark functions in most cases. An implementation of VIGF is available in the dgpsi R package, which can be found on CRAN.

stat.ME↗

arXiv2022

Variance-based global sensitivity analysis of numerical models using R

Sensitivity analysis plays an important role in the development of computer models/simulators through identifying the contribution of each (uncertain) input factor to the model output variability. This report investigates different aspects of the variance-based global sensitivity analysis in the context of complex black-box computer codes. The analysis is mainly conducted using two R packages, namely sensobol (Puy et al., 2021) and sensitivity (Iooss et al., 2021). While the package sensitivity is equipped with a rich set of methods to conduct sensitivity analysis, especially in the case of models with dependent inputs, the package sensobol offers a bunch of user-friendly tools for the visualisation purposes. Several illustrative examples are supplied that allow the user to learn both packages easily and benefit from their features.

stat.CO↗

arXiv2022

A Bayesian Decision Support System in Energy Systems Planning

Gaussian Process (GP) emulators are widely used to approximate complex computer model behaviour across the input space. Motivated by the problem of coupling computer models, recently progress has been made in the theory of the analysis of networks of connected GP emulators. In this paper, we combine these recent methodological advances with classical state-space models to construct a Bayesian decision support system. This approach gives a coherent probability model that produces predictions with the measure of uncertainty in terms of two first moments and enables the propagation of uncertainty from individual decision components. This methodology is used to produce a decision support tool for a UK county council considering low carbon technologies to transform its infrastructure to reach a net-zero carbon target. In particular, we demonstrate how to couple information from an energy model, a heating demand model, and gas and electricity price time-series to quantitatively assess the impact on operational costs of various policy choices and changes in the energy market.

stat.AP↗

arXiv2021

Cross-validation based adaptive sampling for Gaussian process models

In many real-world applications, we are interested in approximating black-box, costly functions as accurately as possible with the smallest number of function evaluations. A complex computer code is an example of such a function. In this work, a Gaussian process (GP) emulator is used to approximate the output of complex computer code. We consider the problem of extending an initial experiment (set of model runs) sequentially to improve the emulator. A sequential sampling approach based on leave-one-out (LOO) cross-validation is proposed that can be easily extended to a batch mode. This is a desirable property since it saves the user time when parallel computing is available. After fitting a GP to training data points, the expected squared LOO (ES-LOO) error is calculated at each design point. ES-LOO is used as a measure to identify important data points. More precisely, when this quantity is large at a point it means that the quality of prediction depends a great deal on that point and adding more samples nearby could improve the accuracy of the GP. As a result, it is reasonable to select the next sample where ES-LOO is maximised. However, ES-LOO is only known at the experimental design and needs to be estimated at unobserved points. To do this, a second GP is fitted to the ES-LOO errors and where the maximum of the modified expected improvement (EI) criterion occurs is chosen as the next sample. EI is a popular acquisition function in Bayesian optimisation and is used to trade-off between local/global search. However, it has a tendency towards exploitation, meaning that its maximum is close to the (current) "best" sample. To avoid clustering, a modified version of EI, called pseudo expected improvement, is employed which is more explorative than EI yet allows us to discover unexplored regions. Our results show that the proposed sampling method is promising.

stat.CO↗

arXiv2021

Stochastic Downscaling to Chaotic Weather Regimes using Spatially Conditioned Gaussian Random Fields with Adaptive Covariance

Downscaling aims to link the behaviour of the atmosphere at fine scales to properties measurable at coarser scales, and has the potential to provide high resolution information at a lower computational and storage cost than numerical simulation alone. This is especially appealing for targeting convective scales, which are at the edge of what is possible to simulate operationally. Since convective scale weather has a high degree of independence from larger scales, a generative approach is essential. We here propose a statistical method for downscaling moist variables to convective scales using conditional Gaussian random fields, with an application to wet bulb potential temperature (WBPT) data over the UK. Our model uses an adaptive covariance estimation to capture the variable spatial properties at convective scales. We further propose a method for the validation, which has historically been a challenge for generative models.

physics.ao-ph↗

arXiv2021

Emulating complex dynamical simulators with random Fourier features

A Gaussian process (GP)-based methodology is proposed to emulate complex dynamical computer models (or simulators). The method relies on emulating the numerical flow map of the system over an initial (short) time step, where the flow map is a function that describes the evolution of the system from an initial condition to a subsequent value at the next time step. This yields a probabilistic distribution over the entire flow map function, with each draw offering an approximation to the flow map. The model output times series is then predicted (under the Markov assumption) by drawing a sample from the emulated flow map (i.e., its posterior distribution) and using it to iterate from the initial condition ahead in time. Repeating this procedure with multiple such draws creates a distribution over the time series. The mean and variance of this distribution at a specific time point serve as the model output prediction and the associated uncertainty, respectively. However, drawing a GP posterior sample that represents the underlying function across its entire domain is computationally infeasible, given the infinite-dimensional nature of this object. To overcome this limitation, one can generate such a sample in an approximate manner using random Fourier features (RFF). RFF is an efficient technique for approximating the kernel and generating GP samples, offering both computational efficiency and theoretical guarantees. The proposed method is applied to emulate several dynamic nonlinear simulators including the well-known Lorenz and van der Pol models. The results suggest that our approach has a promising predictive performance and the associated uncertainty can capture the dynamics of the system appropriately.

stat.ME↗

arXiv2021

Diagnostics for Stochastic Gaussian Process Emulators

Computer models, also known as simulators, can be computationally expensive to run, and for this reason statistical surrogates, known as emulators, are often used. Any statistical model, including an emulator, should be validated before being used, otherwise resulting decisions can be misguided. We discuss how current methods for validating Gaussian process emulators of deterministic models are insufficient for emulators of stochastic computer models and develop a framework for diagnosing problems in stochastic emulators. These diagnostics are based on independently validating the mean and variance predictions using out-of-sample, replicated, simulator runs. We then also use a building performance simulator as a case study example.

stat.ME↗

arXiv2020

Future Proofing a Building Design Using History Matching Inspired Level-Set Techniques

History Matching is a technique used to calibrate complex computer models, that is, finding the input settings which lead to the simulated output matching up with real world observations. Key to this technique is the construction of emulators, which provide fast probabilistic predictions of future simulations. In this work, we adapt the History Matching framework to tackle the problem of level set estimation, that is, finding input settings where the output is below (or above) some threshold. The developed methodology is heavily motivated by a specific case study: how can one design a building that will be sufficiently protected against overheating and sufficiently energy efficient, whilst considering the expected increases in temperature due to climate change? We successfully manage to address this - greatly reducing a large initial set of candidate building designs down to a small set of acceptable potential buildings.

stat.AP↗

arXiv2020

Emulating computer models with step-discontinuous outputs using Gaussian processes

In many real-world applications we are interested in approximating costly functions that are analytically unknown, e.g. complex computer codes. An emulator provides a fast approximation of such functions relying on a limited number of evaluations. Gaussian processes (GPs) are commonplace emulators due to their statistical properties such as the ability to estimate their own uncertainty. GPs are essentially developed to fit smooth, continuous functions. However, the assumptions of continuity and smoothness is unwarranted in many situations. For example, in computer models where bifurcations or tipping points occur, the outputs can be discontinuous. This work examines the capacity of GPs for emulating step-discontinuous functions. Several approaches are proposed for this purpose. Two special covariance functions/kernels are adapted with the ability to model discontinuities. They are the neural network and Gibbs kernels whose properties are demonstrated using several examples. Another approach, which is called warping, is to transform the input space into a new space where a GP with a standard kernel, such as the Matern family, is able to predict the function well. The transformation is perform by a parametric map whose parameters are estimated by maximum likelihood. The results show that the proposed approaches have superior performance to GPs with standard kernels in capturing sharp jumps in the true function.

stat.ME↗

arXiv2020

Classification of Computer Models with Labelled Outputs

Classification is a vital tool that is important for modelling many complex numerical models. A model or system may be such that, for certain areas of input space, the output either does not exist, or is not in a quantifiable form. Here, we present a new method for classification where the model outputs are given distinct classifying labels, which we model using a latent Gaussian process (GP). The latent variable is estimated using MCMC sampling, a unique likelihood and distinct prior specifications. Our classifier is then verified by calculating a misclassification rate across the input space. Comparisons are made with other existing classification methods including logistic regression, which models the probability of being classified into one of two regions. To make classification predictions we draw from an independent Bernoulli distribution, meaning that distance correlation is lost from the independent draws and so can result in many misclassifications. By modelling the labels using a latent GP, this problem does not occur in our method. We apply our novel method to a range of examples including a motivating example which models the hormones associated with the reproductive system in mammals, where the two labelled outputs are high and low rates of reproduction.

stat.ME↗