Please note that the list below only shows forthcoming events, which may not include regular events that have not yet been entered for the forthcoming term. Please see the past events page for a list of all seminar series that the department has on offer.

 

Wed, 28 Sep 2022 09:00 -
Wed, 30 Jun 2027 17:00
Mathematical Institute

Cascading Principles - a major mathematically inspired art exhibition by Conrad Shawcross

Further Information

Oxford Mathematics is delighted to be hosting one of the largest exhibitions by the artist Conrad Shawcross in the UK. The exhibition, Cascading Principles: Expansions within Geometry, Philosophy, and Interference, brings together over 40 of Conrad's mathematically inspired works from the past seventeen years. Rather than in a gallery, they are placed in the working environment of the practitioners of the subject that inspired them, namely mathematics.

Conrad Shawcross models scientific thought and reasoning within his practice. Drawn to mathematics, physics, and philosophy from the early stages of his artistic career, Shawcross combines these disciplines in his work. He places a strong emphasis on the nature of matter, and on the relativity of gravity, entropy, and the nature of time itself. Like a scientist working in a laboratory, he conceives each work as an experiment. Modularity is key to his process and many works are built from a single essential unit or building block. If an atom or electron is a basic unit for physicists, his unit is the tetrahedron.

Unlike other shapes, a tetrahedron cannot tessellate with itself. It cannot cover or form a surface through its repetition - one tetrahedron is unable to fit together with others of its kind. Whilst other shapes can sit alongside one another without creating gaps or overlapping, tetrahedrons cannot resolve in this way. Shawcross’ Schisms are a perfect demonstration of this failure to tessellate. They bring twenty tetrahedrons together to form a sphere, which results in a deep crack and ruptures that permeate its surface. This failure of its geometry means that it cannot succeed as a scientific model, but it is this very failure that allows it to succeed as an art work, the cracks full of broad and potent implications.

The show includes all Conrad's manifold geometric and philosophical investigations into this curious, four-surfaced, triangular prism to date. These include the Paradigms, the Lattice Cubes, the Fractures, the Schisms, and The Dappled Light of the Sun. The latter was first shown in the courtyard of the Royal Academy and subsequently travelled all across the world, from east to west, China to America.

The show also contains the four Beacons. Activated like a stained-glass window by the light of the sun, they are composed of two coloured, perforated disks moving in counter rotation to one another, patterning the light through the non-repeating pattern of holes, and conveying a message using semaphoric language. These works are studies for the Ramsgate Beacons commission in Kent, as part of Pioneering Places East Kent.

The exhibition Cascading Principles: Expansions within Geometry, Philosophy, and Interference is curated by Fatoş Üstek, and is organised in collaboration with Oxford Mathematics. 

The exhibition is open 9am-5pm, Monday to Friday. Some of the works are in the private part of the building and we shall be arranging regular tours of that area. If you wish to join a tour please email @email.

The exhibition runs until 30 June 2026. You can see and find out more here.

Watch the four public talks centred around the exhibition (featuring Conrad himself).

The exhibition is generously supported by our longstanding partner XTX Markets.

Images clockwise from top left of Schism, Fracture, Paradigm and Axiom

Schism Fracture

Axiom Paradigm

Mon, 08 Jun 2026 09:00 -
Thu, 31 Dec 2026 17:00
Mathematical Institute

Paul Ouwerkerk - The Oxford Variations

Further Information

We are delighted to introduce our latest exhibition in the Andrew Wiles Building. Visual artist Paul Ouwerkerk has created 30 new paintings where he plays with the perspective plane in paintings that are generated from self-composed number sequences. The handcrafted canvases are the result of a process in which the artist, after defining a rigid grid as starting point, leaves space for intuition and industrious manual application to elaborate towards the final result.

Visually these paintings can often be interpreted as unfolded polyhedra, dissolving into mathematical landscape perspectives. The rule-based compositions are sometimes derailed purposefully during the painting process, as if to ‘break-the-code’. Painting techniques and materials play a pivotal role in the creation of these works and the materialisation of these abstract illusions.

Paul Ouwerkerk lives and works in Amsterdam. He has a background in art, photography and design. His previous work experience is intermingled with the world of architecture, urbanism and landscape design. Since 2017 he has been painting his abstract ‘Dynamic Geometry’ series.

9 a.m. - 5 p.m. Monday to Friday.

Image of one of the works
 

Mon, 23 Nov 2026
14:15
L4

Dirac operators twisted by ramified Euclidean line bundles

Gorapada Bera
(University College London)
Abstract
Z/2 harmonic spinors arise naturally in gauge theory and calibrated geometry, notably as limiting objects in compactness problems. They are harmonic sections of Dirac operators twisted by Euclidean line bundles ramified along codimension-two subsets. Understanding the analysis of these operators is an important step towards developing a deformation theory of Z/2 harmonic spinors. In this talk, I will discuss the closed extensions of such Dirac operators and their characterization in terms of the Gelfand–Robbin quotient. When the branching locus is a smooth codimension-two submanifold, we construct a geometric realization of this quotient, allowing us to describe Fredholm extensions through residue conditions along the branching locus. I will also discuss elliptic regularity in adapted Sobolev spaces. This is joint work with Thomas Walpuski (arXiv:2503.01392).

 
Mon, 23 Nov 2026
15:00
L5

Stably exotic phenomena in 4-manifold topology

Mark Powell
(University of Glasgow)
Abstract

Exotic phenomena in 4-manifold topology include exotic 4-manifolds, diffeomorphisms,  and surfaces.  A famous result of Wall implies that homeomorphic simply-connected, smooth 4-manifolds are stably diffeomorphic, meaning that they become diffeomorphic after connected sum with enough copies of S^2 x S^2.  In other words, there are no stably exotic simply-connected 4-manfolds. I will discuss the extent to which stable exotica are known to exist in other contexts, with a focus on the case of surfaces. 
 

Mon, 23 Nov 2026

15:30 - 16:30
L3

Around confined planar Coulomb gases

Fredrik Viklund
(KTH Royal Institute of Technology)
Abstract

I will discuss probabilistic and analytic aspects of two-dimensional Coulomb gases confined to Jordan domains, curves, and arcs. In particular, I will describe the asymptotic behavior of particle distributions as the number of particles tends to infinity. I will also explain how the Loewner energy and related quantities arise in the asymptotic expansion of the free energy. A key analytical tool throughout is the Grunsky operator. 

 

The talk is based on joint works with Kurt Johansson.

Mon, 23 Nov 2026

16:30 - 17:30
L4

Stable gravitational collapse and (naked) singularity formation

Matthew Schrecker
(University of Bath)
Abstract

Stars are commonly modelled as a spherically symmetric gas under the influence of either Newtonian or relativistic gravitational forces, giving rise to either the Euler-Poisson or Einstein-Euler equations. In certain regimes, it is known that smooth initial data may give rise to blow-up solutions, corresponding to the collapse of a star under its own gravity. In the relativistic setting, in certain regimes, the blow-up creates a so-called naked singularity. In this talk, I will present the nonlinear stability of a particular family of these blow-up profiles, the (relativistic) Larson-Penston solution. This is based on joint works with Yan Guo, Mahir Hadzic, and Juhi Jang.

Tue, 24 Nov 2026
13:00
L2

Emergent Hamiltonians from boundary symmetries

Jose Garre Rubio
(Vienna)
Abstract

We introduce a systematic framework for constructing emergent local commuting Hamiltonians in (d+1) dimensions from symmetries defined on a d-dimensional boundary. Starting from a generic d-dimensional symmetry, we gauge it by introducing layers of gauge fields carrying dual symmetries, which can in turn be gauged. The fixed points of this iterative gauging procedure yield (d+1)-dimensional stabilizer Hamiltonians built from local Gauss-law constraints. These Hamiltonians encode the emergent bulk order while realizing the original symmetry on the boundary.

I will show how, starting in one dimension, this construction reproduces familiar two-dimensional topological codes, such as the toric code, and naturally realizes all gapped boundary conditions as different one-dimensional symmetric phases at the boundary. The framework extends to higher-form and subsystem symmetries in arbitrary dimensions, yielding, for example, three-dimensional fracton models.

If time permits, I will also discuss how these symmetries can be classified within the framework of tensor networks.

Based on arXiv: 2403.07575, 2410.09044, and 2203.12563.
 

Tue, 24 Nov 2026
14:00
L6

The Geometry of Intertwining Operators

Johannes Carl Droschl
(Aarhus University)
Abstract

Given a p-adic group G, one would like to find a systematic way of constructing all its irreducible representations. As a first approximation, one can parabolically induce irreducible representations of Levi subgroups of G, however, the resulting representations are often reducible. A more refined approach is to consider (analytic) intertwining operators between such induced representations, whose images turn out to be irreducible in many cases. These intertwining operators have proven to be highly useful in a variety of problems, yet their general behavior remains far from fully understood. In the case of general linear groups, conjectures relate their behavior to the geometry of the cotangent bundle of the space of Langlands parameters.

This talk offers a gentle introduction to these topics and reports on recent progress towards these conjectures.

Tue, 24 Nov 2026
14:00
L4

TBA

Michał Pilipczuk
(University of Warsaw)
Abstract

TBA

Tue, 24 Nov 2026
15:00
L6

The ideal Poisson–Voronoi tessellation and some applications

Amanda Wilkens
(University of North Carolina)
Abstract

We motivate and define the ideal Poisson–Voronoi tessellation (IPVT), a new random object with intriguing geometric properties when considered on a semisimple symmetric space (the hyperbolic plane, for example). In joint work with Mikolaj Frączyk and Sam Mellick, we use the IPVT to prove a result on the relationship between the volume of a manifold and the number of generators of its fundamental group (for higher rank semisimple Lie groups, the minimum number of generators in a lattice is sublinear in the covolume). In this talk we give some ideas of the proof. Time permitting, we will discuss related percolation results, joint with Matteo D'Achille, Jan Grebik, Ali Khezeli, and Konstantin Recke.

Tue, 24 Nov 2026

15:00 - 16:00
C3

Which links matter most? Sparsifying network dynamics with effective resistance

Prof. Cristopher Moore
(Santa Fe Institute)
Abstract
“Sparsification” is the act of reducing a network to a subset of its edges while approximately preserving its properties: either to reduce the computational cost of solving problems about it, or to identify which edges are the most important in some sense. Computer scientists have developed beautiful techniques for sparsifying a graph using physics-related ideas like the effective resistance. However, while these methods preserve the spectral properties of the Laplacian, it is not obvious to what extent they preserve the behavior of nonlinear dynamical systems. Using a mobility network from the United States as an example, I’ll show that they do very well for the SIR epidemic model, including the probability each node becomes infected and its distribution of arrival times, even when the sparse network includes less than 10% of the original edges. Choosing edges using purely topological methods, or by thresholding edge weights, does not perform nearly as well. I will end by discussing the possibility of using sparsification to “denoise” networks from bioinformatics, and present some preliminary results on the Kuramoto model of coupled oscillators.
 
This is joint work with Alexander Mercier (Harvard School of Public Health), Emmie Fitz-Gibbons (Brown), and Sam Scarpino (Northeastern).
Tue, 24 Nov 2026
15:30
L4

TBA

Federico Ardila
(Queen Mary University of London)
Tue, 24 Nov 2026
16:00
L5

A Model Theorem for Complex Symmetric Operators

Ryan O'Loughlin
(University of Reading)
Abstract

The spectral theorem is a fundamental result in operator theory, asserting that every normal operator can, after a unitary change of coordinates, be represented as a multiplication operator. Complex symmetric operators are the infinite-dimensional analogue of symmetric (i.e. self-transpose) matrices and form a class that contains all normal operators. 

In this talk, I will present recent results aimed at extending aspects of the spectral theorem beyond normal operators to the broader setting of complex symmetric operators.

Tue, 24 Nov 2026
16:00
L6

Scaling limits of critical FK-decorated maps at q=4

Ellen Powell
(University of Durham)
Abstract

The critical Fortuin–Kasteleyn random planar map with parameter q>0 is a model of random (discretised) surfaces decorated by loops, related to the q-state Potts model. For q<4, Sheffield established a scaling limit result for these discretised surfaces, where the limit is described by a so-called Liouville quantum gravity surface decorated by a conformal loop ensemble. At q=4 a phase transition occurs, and the correct rescaling needed to obtain a limit has so far remained unclear. I will talk about joint work with William Da Silva, XinJiang Hu, and Mo Dick Wong, where we identify the right rescaling at this critical value and prove a number of convergence results.

Wed, 25 Nov 2026

11:00 - 13:00
L4

Geometric SPDE and gauge fixing

Ilya Chevyrev
(SISSA)
Abstract

In this talk, I will give a gentle introduction to geometric SPDEs. I will in particular describe works in the last years on the stochastic gradient decent of the Dirichlet energy (motion of a random string) and of the Yang-Mills energy. I will also present a recent joint work with Tom Klose and Abdulwahab Mohamed on gauge-fixing the non-Abelian Yang-Mills measure, which involves analysis of an elliptic geometric SPDE. This work develops an approach based on implicit function theorem and continuity methods to solve geometric SPDEs, which differs from previous works.

Wed, 25 Nov 2026

13:00 - 14:00
C4

From Generalised Symmetries to Asymptotic Symmetries in Electromagnetism and Gravity

Finn Gagliano
(Durham University)
Abstract

Since the early days of the generalised symmetries literature, it has been noted that the charges generating the electric 1-form symmetry of electromagnetism are related to asymptotic charges, yet this connection has remained poorly understood. Interest in this relationship has been revived by recent work deriving soft photon theorems from the Ward identities of the 1-form symmetry. In this talk, I will review this relationship and extend it by showing that the memory effect can also be described by these symmetries. This brings the third corner of the infrared triangle into the picture, turning it into an infrared tetrahedron, with generalised symmetries as its new vertex. I will also show how this story plays out in (linearised) gravity, where biform symmetries take on the role of the new vertex. Based on upcoming work with Max Velásquez Cotini Hutt. 

Wed, 25 Nov 2026

14:00 - 15:00
Quillen Room N3.12

DPhil Applications Q&A

(Mathematrix)
Abstract

Your chance to ask Mathematrix DPhil students about the process of applying to PhD programs, including written stages and interviews!

Thu, 26 Nov 2026
11:00
C6

A Lean-verified counterexample to Vaught's Conjecture for ${\mathcal L}_{\omega_1,\omega}$

Robin Knight
(Oxford University)
Abstract

Joint work with Nate Ackerman, Harvard and Cameron Freer, MIT.

The negation of Vaught's Conjecture for the infinitary logic ${\mathcal L}_{\omega_1,\omega}$ is the statement that, if the continuum hypothesis is false, there is a sentence $T$ of this logic which has exactly $\aleph_1$ countable models. This is equivalent to the existence of a continuous action of the permutation group $S_\infty$ of the integers on a Polish space (such as the irrationals) with exactly $\aleph_1$ orbits. Both statements can be phrased in such a way as to avoid mention of the continuum hypothesis.

In October, the authors announced such a counterexample that had been expressed and verified in the theorem-proving language Lean. This uses ideas deriving from the proposed counterexample to Vaught's Conjecture in first-order logic circulated by the speaker in 2002.

Thu, 26 Nov 2026

12:00 - 13:00
L3

From Physics to Factory: Using Mathematical Modelling to Accelerate Battery Manufacturing

Dr Helen Walker
(UK Battery Industrialisation Centre)

The join button will be shown 30 minutes before the seminar starts.

Abstract

The rapid growth of the battery industry has created an urgent need for faster process development, reduced experimental burden, and improved manufacturing efficiency. Mathematical modelling provides a powerful framework for understanding and optimising processes across the battery manufacturing chain.

 

This talk will explore how modelling can support battery manufacturing across a range of levels of complexity, from high-fidelity multiphysics simulations to reduced-order models, numerical solutions of governing transport equations, and simpler analytical tools used directly by engineers. Using examples from electrode manufacture, it will demonstrate how different approaches can be applied according to the question being addressed, balancing accuracy, computational effort and practical utility. The talk will show how these complementary techniques can be used to gain process insight, guide experimentation, explore large design spaces and support decisions from early-stage development through to industrial scale-up.

 

Drawing on examples from industrial practice, the talk will illustrate how modelling can shorten development timelines, reduce reliance on costly experimental programmes, and enable more effective scale-up of battery manufacturing processes. It concludes with a discussion of emerging opportunities for applied mathematics in next-generation battery production and digital manufacturing.

Thu, 26 Nov 2026

12:00 - 12:30
Lecture Room 4, Mathematical Institute

Multicomponent flow in domain with moving boundary

Lenka Košárková
(Charles University)
Abstract

The flow of mixtures in deforming domains arises in a range of physical and biological processes. A prominent example is the airflow in lungs, where distinguishing between individual species becomes essential when tracking the distribution of oxygen through the airway branches.
We consider a multicomponent flow in which the bulk momentum transport is described by the Navier-Stokes equations and diffusion is described by the Stefan-Maxwell model, which captures cross-diffusive effects between species. We also account for the motion of the domain boundary and reformulate the equations in the Arbitrary Eulerian-Lagrangian framework. We then discuss the discretization of the coupled governing equations and present numerical simulations of the system in some simplified settings.

Thu, 26 Nov 2026

14:00 - 15:00
Lecture Room 3

Which Sparse Solution Does Gradient Descent Find? Implicit Bias and Convergence Rates for Diagonal Linear Networks Under General Initialization

Assistant Professor Dominik Stöger
(Catholic University of Eichstätt-Ingolstadt, Germany)
Abstract

Assistant Professor Dominik Stöger will give a talk about: 'Which Sparse Solution Does Gradient Descent Find? Implicit Bias and Convergence Rates for Diagonal Linear Networks Under General Initialization'

 

Modern machine learning models are often overparametrized, with more parameters than training samples. The implicit bias of gradient descent determines which of the many solutions that fit the data is selected. Diagonal linear networks, in which a linear predictor is parametrized as an entrywise product of several weight vectors, are a tractable model for studying this bias in underdetermined linear regression. A well-known result states that gradient descent converges to the solution with minimum l1-norm as the scale of initialization tends to zero. This connects training dynamics to sparse recovery. However, the result assumes that all parameters are initialized to the same value. In this talk, we remove this assumption and address two questions: (a) What is the implicit bias when parameters are not initialized identically? (b) How fast is the implicit bias approached as the scale of initialization decreases?

For (a), we show that gradient flow converges to the minimizer of a weighted l1-type penalty whose weights depend on both the initialization and the sign of each coordinate of the solution. Our proof relies on a new characterization of the gradient dynamics in terms of a Bregman divergence. For (b), we derive tight convergence rates. In particular, we show that for two-layer networks the limit of gradient flow converges to this minimizer at a sublinear rate in the scale of initialization, whereas for more layers the rate of convergence is linear. Surprisingly, the constants governing these rates are connected to the null-space constants of compressed sensing. Our rates improve on previous results in the identical initialization setting and are, to our knowledge, the first in the general initialization setting.

Joint work with Hannes Matt (KU Eichstätt-Ingolstadt).

 

 

 

Further Information

To visit Dominik Stögers’ webpage, please click here.

Thu, 26 Nov 2026

16:00 - 17:00
L5

Trading on a Regime-Change Alarm: Optimal Execution and Portfolio Choice

Horace Yiu
((Mathematical Institute University of Oxford))
Abstract
Markets change regime. A trend fades, order flow turns, and a signal that made money last week starts to lose it. Traders often monitor these breaks with a sequential alarm, which fires once the evidence against the current regime is strong enough. But alarms can be late or simply wrong, so how much should a trader trust one? Our main result is a general theorem that characterises the optimal strategy for a broad class of trading problems in which decisions rely on such an alarm. We apply it to two central problems in finance, namely executing a large order with price impact and allocating a portfolio between risky assets and cash. In both cases, the theorem gives explicit strategies. In execution, the trader does not wait for the alarm, and positions shift ahead of it based on how likely it is to fire before the horizon ends. We also show that the alarm should be calibrated to the trading objective rather than to a purely statistical false-alarm rate, because false alarms and late alarms have different costs in each setting. Finally, simulations compare these strategies with simple rules a trading desk might use, showing where the extra complexity pays off and where it does not.


 

Thu, 26 Nov 2026
17:00
L3

The Zilber-Pink conjecture for generic subvarieties of C^n

Guy Fowler
(Manchester University)
Abstract
The Zilber-Pink conjecture is an important conjecture in Diophantine geometry, motivated in part by Zilber’s work on the model theory of complex exponentiation. In this talk, I will outline a proof of the conjecture for all subvarieties of C^n that are sufficiently “generic”. Our proof combines techniques from the model theory of differential fields with some arithmetic inputs. This is joint work with Vahagn Aslanyan and Sebastian Eterović.
Fri, 27 Nov 2026

11:00 - 12:00
L4

Mathematical modelling of naturally occurring epigenetic barcodes as a tool to resolve clonal dynamics in cancer

Prof Calum Gabbutt
(Department of Immunology and Inflammation Imperial College London)
Abstract

Evolution underlies the transformation of a normal cell to a cancer, yet learning the parameters defining this dynamic process from single-timepoint bulk samples is an open challenge. To understand how cancer cells evolve in vivo, we must rely on naturally occurring, heritable lineage tracing markers that encode the evolutionary history of a population of cells. Here, I shall introduce our work on identifying selectively neutral “epigenetic barcodes” and employing them as a molecular clock. By coupling this process with mathematical modelling and Bayesian inference, we characterised the evolutionary history of almost 2000 lymphoid cancers (Gabbutt et al., 2025). Across a broad range of cancer types, we demonstrated that tumour growth rates and malignancy ages differed by orders of magnitude. In 2 independent cohorts of patients with chronic lymphocytic leukaemia (CLL), a typically indolent and slow growing cancer, the inferred growth rates were highly prognostic. I shall further discuss recent work applying this approach to resolve the clonal relationship between acute myeloid leukaemia (AML) blasts and differentiated neutrophils in patients without bone marrow failure. 

Fri, 27 Nov 2026

11:00 - 12:00
L1

What to do over the vacation?

Abstract

In this session, we will discuss how best to use the Christmas vacation to consolidate your learning, to prepare for collections, and of course to have a rest!

This session is likely to be most relevant for first-year undergraduates, but all are welcome.

Fri, 27 Nov 2026
12:00
Quillen Room N3.12

$\operatorname{SL}_2$ plethysms and binomial identities

Michał Szwej
(University of Bristol)
Abstract
Apply a symmetric power to the natural representation of $\operatorname{SL}_n(\mathbb{C})$ once, and you obtain an irreducible representation. Apply it twice, and you get a `plethystic representation', whose decomposition is a major open problem in algebraic combinatorics. 'If you can't solve a problem, generalise it': we focus on $\operatorname{SL}_2$, but allow the ground field $F$ to be arbitrary.
 
When $F=\mathbb{C}$, q-analogues of many familiar binomial identities correspond to isomorphisms of $\operatorname{SL}_2$-plethysms via character theory, which alone does not suffice over arbitrary fields. In this talk I will present several identities which do lift to $\operatorname{SL}_2(F)$-isomorphisms. I will conclude with a surprising connection between such plethystic representations and the quantum group of $\mathfrak{sl}_2$. No specialist background knowledge will be assumed. This is joint work with Álvaro Gutiérrez, Álvaro L. Martínez, and Mark Wildon. 
Fri, 27 Nov 2026
12:00
C1

Heavenly equations in de Sitter space

Tim Moy
(Cambridge/Sheffield)
Abstract
In joint work with Maciej Dunajski, it was shown that the Einstein equations for the general analytic ASD metric may be formulated as a single non-linear second-order PDE, found earlier by Lipstein and Nagy. Firstly, I will present surprisingly simple, rational solutions that correspond to the (analytic continuations) of some well-known metrics. Secondly, I will relate the equation to an older formulation of the ASD Einstein condition due to Przanowski. Thirdly, I will explain how to obtain solutions of the linearised equation, hence ASD Einstein metric deformations, from contact Hamiltonians on twistor space, and vice-versa. 
 
Fri, 27 Nov 2026
13:00
L4

TBC

María Sánchez-Muñiz
(City College New York)
Fri, 27 Nov 2026
13:30
C6

TBC

Qiu Shi Wang
((Mathematical Institute University of Oxford))
Fri, 27 Nov 2026

14:00 - 15:00
L1

Introduction to LaTeX

Eamonn Gaffney
Abstract
No prior experience of LaTeX is assumed.
 
This one-hour session introduces LaTeX, a highly-recommended document preparation system widely used for technical writing, particularly for documents containing mathematical equations, figures and references.
 
We will very briefly mention local LaTeX installations (e.g. MiKTeX and TeXmaker), but the focus will be the cloud-based option, Overleaf. Please note that Oxford students have free access to an Overleaf subscription, and Overleaf is recommended.
 
We will extensively follow Parts 1 and 2 of Overleaf's online introduction to LaTeX, creating and structuring documents, formatting text, writing mathematical expressions, inserting figures and tables, and managing references. We also briefly discuss the history functionality of Overleaf, backing up your work and the Mathematical Institute thesis template (for those in Parts B or C writing a project, essay or dissertation).  
 
You are encouraged to bring a laptop with an Overleaf account set up in advance, by signing up at
 
 
Please use your SSO email in the sign-up to access the institutional subscription, but the basic, free Overleaf account will work for the session.   
 
Resources:
Mon, 30 Nov 2026

14:00 - 15:00
Lecture Room 3

Physics-informed deep generative models: Applications to computational sensing

Professor Marcelo Pereyra
(Heriot-Watt University, Edinburgh)
Abstract

Professor Pereyra will talk about; 'Physics-informed deep generative models: Applications to computational sensing'

This talk introduces a novel mathematical and computational framework for constructing high-dimensional Bayesian inversion methods that leverage state-of-the-art generative denoising diffusion models as highly informative priors. A central innovation is the construction of physics-informed generative models using Langevin diffusion processes and Markov chain Monte Carlo (MCMC) sampling techniques to develop stochastic neural network architectures capable of near-exact sampling. The obtained networks are modular and composed of interpretable layers that are directly related to statistical image priors and data likelihoods derived from forward observation models. The layers encoding the data likelihood function are designed for flexibility, enabling scene and instrument model parameters to be specified at inference time and seamlessly integrated with pre-trained foundational generative priors. To achieve high computational efficiency, we employ adversarial model distillation, which yields excellent sampling performance with as few as four Markov chain Monte Carlo steps, even in problems exceeding one million dimensions. Our approach is validated through non-asymptotic convergence analysis and extensive numerical experiments in computational image and video restoration. We conclude by discussing unsupervised training strategies that allow the models to be fine-tuned directly from measurement data, thereby bypassing the need for clean reference data.

The talk is based on recent work in physics-informed generative AI for Bayesian imaging: https://arxiv.org/abs/2503.12615 (ICCV 2025), which uses a distilled latent Stable Diffusion XL model trained on five billion clean images as a zero-shot prior, and  https://arxiv.org/pdf/2507.02686, which integrates pixel-based diffusion models with deep unfolding and diffusion distillation (TMLR 2025). The extension to video restoration is presented in https://arxiv.org/abs/2510.01339 (ICLR 2025). Our approach to unsupervised training of diffusion models is introduced in https://arxiv.org/abs/2510.11964.

 

 

Further Information

Bio:
Marcelo Pereyra is a Professor in Statistics and UKRI EPSRC Open Research Fellow at the School of Mathematical and Computer Sciences of Heriot-Watt University & Maxwell Institute for Mathematical Sciences. He leads pioneering research advancing the statistical foundations of quantitative and scientific imaging, shaping how image data are used as rigorous quantitative evidence, and forging deep connections between statistical, variational, and machine learning approaches to imaging. His leadership and contributions have been recognized through multiple prestigious awards, most recently a five-year fulltime EPSRC Open Fellowship to drive the next generation of breakthroughs in statistical imaging sciences based on physics-informed generative artificial intelligence. Prof. Pereyra will join Imperial College London in 2027 as Chair in Statistical Machine Learning in the Department of Mathematics.

Prof. Pereyra received the SIAM SIGEST Award in Imaging Sciences for his contributions to Bayesian imaging in 2022. He has held Invited Professor positions at Institut Henri Poincaré (Paris, 2019), Université Paris Cité (2022), Ecole Normale Superiéure Lyon (2023), Université Paris Cité (2024) and Centralle Lille (2025). He is also the recipient of a UKRI EPSRC Open Research Fellowship (2025), a Marie Curie Intra-European Fellowship for Career Development (2013), a Brunel Postdoctoral Research Fellowship in Statistics (2012), a Postdoctoral Research Fellowship from French Ministry of Defence (2012), and a Leopold Escande PhD Thesis award from the University of Toulouse (2012).

Mon, 30 Nov 2026
14:15
L4

On metric connections with parallel and closed skew torsion

Paul Schwahn
((Mathematical Institute University of Oxford))
Abstract

This talk is divided in three parts. First, I review structural properties and examples of geometries with parallel skew torsion. Second, I take a closer look at those geometries where the torsion is in addition closed; this equips every tangent space with the structure of a metric Lie algebra. These manifolds always locally split as a product into well understood factors, allowing a complete local classification. Third, I investigate various G-structures defining a geometry of this type, with a focus on almost Hermitian structures, extending recent results of Barbaro-Pediconi-Tardini. This is joint work with Andrei Moroianu.

Mon, 30 Nov 2026

15:30 - 16:30
L3

Agentic Asset Pricing

Kay Giesecke
(University of Potsdam)
Abstract

As AI agents become capable of autonomously conducting empirical research, a basic question arises: when can we trust the discoveries they produce? Autonomous systems can formulate hypotheses, implement and evaluate them, learn from intermediate results, and repeatedly redirect their search. The resulting hypotheses, their number, and their dependence structure are themselves outcomes of a random and highly adaptive research process, while multiplicity and information leakage can make spurious findings appear convincing. We develop a reusable statistical methodology for rigorous autonomous asset-pricing research that separates essentially unrestricted exploration from formal post-registration confirmation. A time-stamped boundary ledger records and freezes the realized research output and its confirmation designs; fresh post-boundary data are then reserved for statistical certification.

The key insight is that adaptive exploration need not compromise statistical validity once the realized search output is frozen and subsequent evidence is valid conditional on that output. We formalize this principle through conditional anytime e-closed testing for random, dependent hypothesis families. Our central result controls conditional expected loss at level $\alpha$, even when the reporting time is data dependent and the registered family and boundary-specified design choices are random. Different loss functions recover FDR, FWER, FDX, and other error criteria within a common framework; for indicator losses, the guarantees hold simultaneously across confirmation checkpoints.

We instantiate the exploration layer with SEADS, an autonomous system that proposes, implements, tests, and refines predictors of daily U.S. equity returns. Using two fresh years of post-registration data, we subject its discoveries to FDR- and FWER-controlled confirmation using the same evidence paths. Without being supplied known anomaly formulas or labels, SEADS reconstructs much of the established short-horizon structure of the equity cross section from primitive building blocks, including reversal, low-risk and volatility effects, turnover, and profitability. The results demonstrate how open-ended autonomous exploration can be coupled with rigorous fresh-data confirmation to produce statistically credible asset-pricing discoveries.

This is joint work with Xiaowei Ding (Nanjing), Jason Pan (Stanford), and Markus Pelger (Stanford).

 
 
Mon, 30 Nov 2026
15:30
L5

TBC

Victor Roca i Lucio
(Université Paris Cité)
Mon, 30 Nov 2026

16:30 - 17:30
L4

Aleksandrov estimate in kinetic equations

Amelie Loher
((Mathematical Institute University of Oxford))
Abstract

We discuss Aleksandrov estimates for kinetic equations in non-divergence form with a rough diffusion coefficient. We present two different mechanisms, each of which has its advantages.

Tue, 01 Dec 2026
14:00
L6

TBC

Mario Marcos Losada
(Mathematical Institute University of Oxford )
Abstract

to follow

Tue, 01 Dec 2026
14:00
L4

TBA

Annika Heckel
(Uppsala University)
Abstract

TBA

Tue, 01 Dec 2026
15:00
L6

The conjugacy problem in twisted right-angled Artin groups

Gemma Crowe
(Manchester)
Abstract
There exists a vast literature around the idea of constructing groups (and monoids) via some correspondence with finite simple graphs. One well-known example is right-angled Artin groups (RAAGs), where edges in our graph correspond to commuting generators in our group.
 
In this talk, we will discuss an adaptation of this group, known as twisted RAAGs (T-RAAGs), where we introduce the option of ‘Klein-bottle’ relations, to correspond to directed edges of our graph. We will then survey some results about these groups (including many open questions!) and show T-RAAGs have a decidable conjugacy problem. This is based on joint work with Islam Foniqi.
Tue, 01 Dec 2026

15:00 - 16:00
C3

Mind the sign: what we miss when we ignore antagonistic interactions

Fernando Diaz-Diaz
(University of Madrid Carlos III)
Abstract

Many complex systems have both cooperative and antagonistic interactions. For example, countries form alliances and wage wars, financial assets move together or in opposite directions, genes activate or inhibit each other, and users of online platforms endorse or reject each other. We can represent these systems as signed networks, in which each edge has a positive or a negative sign. However, most tools of network science assume non-negative weights. The common practices of discarding negative edges, taking absolute values, or treating signs as ordinary weights can thus lead to wrong or incomplete conclusions. In this talk, we first review where signed networks appear and which problems arise when we apply standard methods to them. Then we discuss three types of analysis that account for signs explicitly: measuring structural balance at different scales, detecting mesoscale structure, and constructing network embeddings. Throughout, we illustrate how incorporating the signs into the analysis reveals structural properties that conventional approaches can overlook.

Tue, 01 Dec 2026
16:00
L5

TBC

Jesse Reimann
(Delft University)
Abstract

to follow

Tue, 01 Dec 2026
16:00
L6

On the complex divisor function, random permutations, and branching random walks

Louis-Pierre Arguin
(The University of Oxford and the City University of New York)
Abstract

I will detail the surprising connections between seemingly unrelated problems pertaining to the characterististic polynomial of random permutation matrices, the complex divisor function on a short interval, and branching random walks in random environment. Such questions arose in the work of Cook & Zeitouni, Hall & Tenenbaum, Mallein & Milos, and more recently Cook & Gu. This is joint with P. Bourgade and D. Koukoulopoulos.

Wed, 02 Dec 2026

11:00 - 13:00
L4

Invitation to Random Tensor Models: from random geometry to enumeration of tensor invariants

Reiko Toriumi
(Okinawa Institute of Science and Technology Graduate University)
Abstract

I will introduce random tensor models by first reviewing their motivation coming from random geometric approach to quantum gravity. I will also give some survey on the current topics around random tensors. Then, I will present recent results on enumeration of graphs representing tensor invariants based on the paper ''Counting U(N)⊗r⊗O(N)⊗q invariants and tensor model observables''.