Skip to main content
Seminar Series Logo

AY 2026-2027 Semester 1 Seminar Schedule

Timetable

Date

Day

Time

Week no.

Presenter

Affil.

Dept. Contact

Title

2026-10-08

Thursday

11am

5

Wasgner Barreto de Souza

University College Dublin

James Sweeney

Generalised Linear Models Driven by Latent Processes: Asymptotic Theory and Applications

2026-10-15

Thursday

11am

6

Kate Finucane

University of Limerick

David OSullivan

Missing data imputation using a truncated Gaussian infinite factor model with application to metabolomics data

2026-10-22

Thursday

11am

7

Open

TBC

TBC

TBC

2026-10-29

Thursday

11am

8

Marc Calvo Schwarzwalder

Universitat Politècnica de Catalunya

Kevin Morney

Mathematical Modelling of Adsorption Columns State of the Art and Future Perspectives

2026-11-05

Thursday

11am

9

Open

TBC

TBC

TBC

2026-11-19

Thursday

11am

11

Santiago Lamata Otín

University of Zaragoza

David Osullivan

Viral Evolution and Epidemic Spreading: The Interplay of Timescales

2026-11-26

Thursday

11am

12

Ivan Christov

Purdue University

Ailish Hannigan

Soft channels, asymptotics and a little Bayes

2026-12-03

Thursday

11am

13

Stephen Jun Villejo

Imperial College London

James Sweeney

TBC

Abstracts

As seminar abstracts become available from the speakers this page will be updated.


Seminar week 5 by Wasgner Barreto de Souza

Date: 2026-10-08 at 11am
Speaker: Wasgner Barreto de Souza (University College Dublin)
Host: James Sweeney

Title: Generalised Linear Models Driven by Latent Processes: Asymptotic Theory and Applications

Abstract: This paper introduces a class of generalised linear models (GLMs) driven by latent processes for modelling count, real-valued, binary, and positive continuous time series. Extending earlier latent-process regression frameworks based on Poisson or one-parameter exponential family assumptions, we allow the conditional distribution of the response to belong to a bi-parameter exponential family, with the latent process entering the conditional mean multiplicatively. This formulation substantially broadens the scope of latent-process GLMs, for instance, it naturally accommodates gamma responses for positive continuous data, enables estimation of an unknown dispersion parameter via the method of moments, and avoids restrictive conditions on link functions that arise under existing formulations. We establish the asymptotic normality of the GLM estimators obtained from the GLM likelihood that ignores the latent process, and we derive the correct information matrix for valid inference. In addition, we provide a principled approach to prediction and forecasting in GLMs driven by latent processes, a topic not previously addressed in the literature. We present two real data applications on measles infections in North Rhine-Westphalia (Germany) and paleoclimatic glacial varves, which highlight the practical advantages and enhanced flexibility of the proposed modelling framework.


Seminar week 6 by Kate Finucane

Date: 2026-10-15 at 11am
Speaker: Kate Finucane (University of Limerick)
Host: David OSullivan

Title: Missing data imputation using a truncated Gaussian infinite factor model with application to metabolomics data

Abstract: Metabolomics is the study of small molecules in biological samples, where non-negative, high-dimensional, multicollinear data are common, typically with many more variables than observations. Missing data are prevalent, and imputation is critical as many analysis methods necessitate complete data. In such data, missing at random (MAR) missingness occurs due to acquisition or processing error, while missing not at random (MNAR) missingness occurs when true values lie below a threshold for detection.

A novel truncated Gaussian infinite factor analysis (TGIFA) model is proposed to perform statistically principled and physically realistic imputation in metabolomics data. Truncated Gaussian distributional assumptions ensure cognisance of the data’s physical constraints, and an infinite latent factor framework captures high-dimensional dependencies without pre-specifying the number of latent factors. Both MAR and MNAR missingness are addressed, and a Bayesian inferential approach provides uncertainty quantification for both imputed values and missingness types. A computationally efficient exchange algorithm enables scalable posterior inference via Markov Chain Monte Carlo. Performance of TGIFA is assessed via a simulation study and its utility is demonstrated in a motivating urinary metabolomics dataset, where it yields useful imputations, with associated uncertainty quantification.

Speaker’s Paper

Speaker’s Webpage


Seminar week 8 by Marc Calvo Schwarzwalder

Date: 2026-10-29 at 11am
Speaker: Marc Calvo Schwarzwalder (Universitat Politècnica de Catalunya)
Host: Kevin Morney

Title: Mathematical Modelling of Adsorption Columns State of the Art and Future Perspectives

Abstract: Column adsorption is an essential separation and purification process widely used in water treatment, greenhouse gas capture, air purification, and chromatography. While classical mathematical models for single-component systems under idealized conditions are well established, real-world applications present complex physical and chemical challenges that push standard models beyond their limitations. In this talk we provide a comprehensive overview of the current state of column adsorption modelling, focusing on recent theoretical advances that bridge the gap between simplified mathematical descriptions and realistic industrial environments. We will start from the general description for trace amounts, consisting of an advection-diffusion equation for mass transfer within the column and a kinetic equation for describing the adsorption mechanisms, and will discuss how this formulation is tweaked in different situations, such as the presence of large amounts of contaminant, consideration of multiple contaminant species to be removed simultaneously, increase of the size of the adsorbing particles, etc.

Speaker’s Google Scholar

Speaker’s Webpage


Seminar week 11 by Santiago Lamata Otín

Date: 2026-11-19 at 11am
Speaker: Santiago Lamata Otín (University of Zaragoza)
Host: David Osullivan

Title: Viral Evolution and Epidemic Spreading: The Interplay of Timescales

Abstract: Epidemic outbreaks impose a substantial burden on the health and wealth of societies. Epidemic models have long been used to understand outbreak trajectories and assess the impact of control policies over relatively short timescales. Viral evolution, however, adds another layer of complexity, particularly for rapidly evolving pathogens, whose evolutionary dynamics unfold on timescales comparable to those of epidemic spreading. In this seminar, we explore how different dimensions of viral evolution reshape epidemic dynamics. Evolution across antigenic space allows viruses to escape population immunity through the interplay between mutation and immunity timescales, thereby determining whether epidemic activity can persist as an endemic state. The structure of the underlying genotype network further shapes this evolutionary exploration and can, by itself, account for seasonal or steady patterns of endemicity. Evolution in infectivity, instead, alters the conditions for epidemic onset through the interplay between evolutionary and epidemic timescales. Importantly, this coupling also challenges conventional expectations about epidemic control. Under certain conditions, interventions can alter evolutionary dynamics in ways that lead to worse epidemic outcomes than taking no action.

References:

[1] Lamata-Otín, Santiago, et al. “Genotype networks drive oscillating endemicity and epidemic trajectories in viral evolution.” Communications Physics 8.1: 502 (2025).

[2] Lamata-Otín, Santiago, et al. “Interplay between evolutionary and epidemic time scales challenges the outcome of control policies.” Physical Review Letters, In press (arXiv preprint arXiv:2603.18801) (2026).

Speaker’s Google Scholar

Speaker’s Paper 1 Speaker’s Paper 2

Speaker’s Webpage