PhD position on flexible structural causal modelling in extreme value statistics

Job description

In this project, we develop a new mathematical framework to efficiently model multivariate systems during extreme events like floods, heatwaves, or financial crashes. This PhD position offers a combination of theoretical research and the implementation of new statistical methodology and thus requires good knowledge of probability theory and mathematical statistics.

Many real-world applications exhibit cause-effect relations, where each variable in a system can be considered as an imperfect reflection of the influences of related variables. Structural causal models provide a flexible mathematical framework for efficient prediction, model selection and management in such settings. However, when interest is rather in the behavior of the system during extreme events, for example during floods in a river network or crashes of the stock market, current approaches for structural causal modeling in extremes are limited to scenarios where all variables are simultaneously large.

Such assumptions are often unrealistic, and can therefore result in inadequate models. This project proposes a generalized framework for extremal structural causal models on arbitrary directed acyclic graphs. Our new models will be able to incorporate non‑standard extreme directions, which permits the modeling of settings where only parts of a system are extreme. Special attention will be given to parametric families like the Hüsler–Reiss distribution, which lead to an extremal analogue of Gaussian structural causal models.

We further propose to develop scalable structure learning methods for these new models, including latent variable identification, and to demonstrate their effectiveness on real data. For this, we will study the model class of linear non‑Hüsler–Reiss acyclic models and their generalization that includes latent variables. Our proposal will fill a critical gap in statistical methodology, offering robust tools for prediction and management of extreme events in high‑dimensional, directed systems.

The project is funded by an M1 Open Competition grant from the Dutch research council (NWO). The successful candidate will be appointed in the Statistics group at the University of Twente, mainly supervised by Dr. Frank Röttger. The position comes with funding for active participation in scientific workshops and conferences, and includes research stays with international collaborators.

Starting date can be agreed upon after the interview, but will preferably be between October 2026 and early Spring 2027.

Your tasks:

• Perform daily PhD-level research.
• Publish results in journals and conference proceedings, and present these at (inter) national workshops and conferences.
• Contribute to teaching activities related to your work (at most 10% of your time).
• Actively participate in seminars, scientific discussions, and contribute to the good atmosphere in the group.
• Be a part of an excellent young research group.

We are an inclusive group and diversity is at the heart of our research principles. We care about a good working atmosphere and a good work‑life balance. Applications from all groups currently under‑represented in academic posts are especially encouraged. We particularly encourage women to apply.

• You have, or will shortly acquire a master’s degree in (applied) mathematics, statistics, or a related degree with a strong theoretical mathematical component (including probability theory and mathematical statistics). Enthusiasm for the topic of the project is more important than your choice of specialisation in your master’s.
• You enjoy scientific discussions and gatherings, and you want to develop your academic profile in mathematical statistics.
• You are proficient in English.
• Good scientific programming skills, e.g. in R, Python, C++, Julia, etc.

• As a PhD candidate at UT, you will be appointed to a full‑time position for four years, with a qualifier in the first year, within a very stimulating and exciting scientific environment;
• The Statistics group offers a dynamic ecosystem with enthusiastic colleagues;
• Your salary and associated conditions are in accordance with the collective labour agreement for…