Past

Session to be run by Five Rings

Calculus and geometry are ubiquitous in the theoretical modelling of scientific phenomena, but have historically been very challenging to apply directly to real data as statistics. Diffusion geometry is a new theory that reformulates classical calculus and geometry in terms of a diffusion process, allowing these theories to generalise beyond manifolds and be computed from data. In this talk, I will describe a new, simple computational framework for diffusion geometry that substantially broadens its practical scope and improves its precision, robustness to noise, and computational complexity. We present a range of new computational methods, including all the standard objects from vector calculus and Riemannian geometry, and apply them to solve spatial PDEs and vector field flows, find geodesic (intrinsic) distances, curvature, and several new topological tools like de Rham cohomology, circular coordinates, and Morse theory. These methods are data-driven, scalable, and can exploit highly optimised numerical tools for linear algebra.

Dr Jenny Power (Heriot-Watt University) will share some of the personal and unexpected lessons she learnt while completing her PhD. We’ll then discuss how a “perfect PhD student” doesn’t exist!

Bring interesting problems (relating to your research or otherwise) for a unique brainstorming session

Certain models of collective dynamics exhibit deceptively simple patterns that are surprisingly difficult to explain. These patterns often arise from phase transitions within the underlying dynamics. However, these phase transitions can be explained only when one derives continuum equations from the corresponding individual-based models. In this talk, I will explore this subtle yet rich phenomenon and discuss advances and open problems.

I will present new results, joint with Krzysztof Klosin (CUNY), on the modularity of residually reducible Galois representations with 3 residual pieces. This will be applied to prove the p-adic modularity of Picard curves.

Speaker Estefania Loayza Romero will talk about:  A Riemannian Approach for PDE-Constrained Shape Optimization Using Outer Metrics

In PDE-constrained shape optimisation, shapes are traditionally viewed as elements of a Riemannian manifold, specifically as embeddings of the unit circle into the plane, modulo reparameterizations. The standard approach employs the Steklov-Poincaré metric to compute gradients for Riemannian optimisation methods. A significant limitation of current methods is the absence of explicit expressions for the geodesic equations associated with this metric. Consequently, algorithms have relied on retractions (often equivalent to the perturbation of identity method in shape optimisation) rather than true geodesic paths. Previous research suggests that incorporating geodesic equations, or better approximations thereof, can substantially enhance algorithmic performance. This talk presents numerical evidence demonstrating that using outer metrics, defined on the space of diffeomorphisms with known geodesic expressions, improves Riemannian gradient-based optimisation by significantly reducing the number of required iterations and preserving mesh quality throughout the optimisation process.

This talk is hosted at RAL. 

For this JC, I will review the recently much debated puzzles that arise in holographic systems with baby universes. After describing the original set-up of Antonini-Sasieta-Swingle, I will explain the paradox raised by Antonini-Rath, which suggests the existence of a single CFT state that can have two distinct holographic descriptions in the bulk: one with a closed baby universe and one without. I will discuss various proposed resolutions of this puzzle, which may (or may not) require us to rethink the holographic dictionary in AdS/CFT.

We study two-dimensional characteristic free-boundary problems in ideal compressible magnetohydrodynamics. For current-vortex sheets, surface-wave effects yield derivative loss and only weak (neutral) stability; under a sufficient stability condition on the background state we obtain anisotropic weighted Sobolev energy estimates and prove local-in-time existence and nonlinear stability via a Nash-Moser scheme, confirming stabilization by strong magnetic fields against Kelvin-Helmholtz instability. For the plasma-vacuum interface, coupling hyperbolic MHD with elliptic pre-Maxwell dynamics, we establish local existence and uniqueness provided at least one magnetic field is nonzero along the initial interface.

Dr Jari Fowkes will talk about; 'A Very Short Introduction to Ptychographic Image Reconstruction'

I will present a very short introduction to the mathematics behind the scientific imaging technique known as ptychography, starting with a brief overview of the physics model and the various simplifications required, before moving on to the main ptychography inverse problem and the three principal classes of optimization algorithms currently being used in practice. 

Fracture is usually treated as an outcome to be avoided; here we see it as something we may write into a lattice's microstructure. Maxwell lattices sit at the edge of mechanical stability, where robust topological properties provide a way on how stress localises and delocalises across the structure with directional preference. Building on this, we propose a direct relationship between lattice topology and damage propagation. We identify a set of topology- and geometry-dependent parameters that gives a simple, predictive framework for nonideal Maxwell lattices and their damage processes. We will discuss how topological polarisation and domain walls steer and arrest damage in a repeatable way. Experiments confirm the theoretical predicted localisation and the resulting tuneable progression of damage and show how this control mechanism can be used to enhance dissipation and raise the apparent fracture energy.

The restricted Burnside problem asks whether, for each natural numbers r and n, there are only finitely many finite r-generated groups of exponent n. The solution of this problem was given by Kostrikin in the 1960s for prime exponent, then by Efim Zelmanov in 1991, for which he was awarded the Fields medal in 1994. In fact, both Kostrikin and Zelmanov results concern Lie algebras, and are a perfect illustration of Lie methods in group theory: how to reduce questions on groups to questions on Lie algebras. Starting from a finitely generated group, one may construct an "associated Lie algebra" which, for the case of exponent p, is n-Engel, i.e. satisfies the n-Engel identity: [x,y,y,...,y] = 0 (n times). For that case, the restricted Burnside problem reduces to proving that every finitely generated n-Engel Lie algebra is nilpotent.

In 1988, Zelmanov proved the ultimate generalization of Engel's classical result: every n-Engel Lie algebra over a field of characteristic 0 is nilpotent. This theorem has the following consequence: for every n there exists N such that every n-Engel Lie algebra of characteristic p>N is nilpotent. It also has consequences for Engel groups.

The proof is rather involved and consists mainly of some intense Lie algebra computations, sprinkled with several beautiful tricks. In particular, the surprising use of the representation theory of the symmetric group has inspired several other authors since then.

In this talk, I will present a little bit of all this. For instance, we will study the case of 3-Engel Lie algebras and I will explain how some part of Zelmanov's proof was re-used by Vaughan-Lee and Traustason to reduce the algorithmic complexity of computing in 4-Engel Lie algebras.

Dimension 4 is the first dimension in which exotic smooth manifold pairs appear — manifolds which are topologically the same but for which there is no smooth deformation of one into the other. On the other hand, smooth and PL manifolds (manifolds which can be described discretely) do coincide in dimension 4. Despite this, there has been comparatively little work done towards gaining an understanding of smooth 4-manifolds from the discrete and algorithmic perspective. The aim of this talk will be to give a gentle introduction to some of the tools, techniques, and ideas, which inform a computational approach to 4-manifold topology.

First introduced by Fomin and Zelevinsky, cluster algebras are commutative rings that have many combinatorial properties. They have had many applications to both mathematics and physics. In this talk, I will first introduce cluster algebras and explore some of their properties. I will then move on to their applications, starting with dilogarithm identities and then moving to integrable systems and the thermodynamic Bethe ansatz (TBA). Time permitting, I will connect some of these ideas to the ODE/IM correspondence. 

Recently, a close parallel emerged between conformal field theory in general dimension and the theory of automorphic forms. I will review this connection and explain how it can be leveraged to make rigorous progress on central open problems of number theory, using methods borrowed from the conformal bootstrap. In particular, I will use the crossing equation to prove new subconvex bounds on L-functions. Based on work with Adve, Bonifacio, Kravchuk, Pal, Radcliffe, and Rogelberg: https://arxiv.org/abs/2508.20576.

In 1962, Horn raised the following problem: Let A and B be n-by-n Hermitian matrices with respective eigenvalues a_1,...,a_n and b_1,...,b_n. What can we say about the possible eigenvalues c_1,...,c_n of A + B?

The deterministic perspective is that the set of possible values for c_1,...,c_n are described by a collection of inequalities known as the Horn inequalities.

Free probability offers the following alternative perspective on the problem: if (A_n) and (B_n) are independent sequences of n-by-n random matrices with empirical spectra converging to probability measures mu and nu respectively, then the random empirical spectrum of A_n + B_n converges to the free convolution of mu and nu.

But how are these two perspectives related?

In this talk Samuel Johnston will discuss approaches to free probability that bridge between the two perspectives. More broadly, Samuel will discuss how the fundamental operations of free probability (such as free convolution and free compression) arise out of statistical physics mechanics of corresponding finite representation theory objects (hives, Gelfand-Tsetlin patterns, characteristic polynomials, Horn inequalities, permutations etc).

This talk is based on joint work with Octavio Arizmendi (CIMAT, Mexico), Colin McSWiggen (Academia Sinica, Taiwan) and Joscha Prochno (Passau, Germany).

In this talk I will describe how bornological spaces give a foundation for derived geometries. This works over any Banach ring allowing to define analytic and differential geometry over the integers. I will discuss applications of this approach such as the representability of certain moduli spaces and Galois actions on the cohomology of differetiable manifolds admitting a \Q-form.

It is a classical result that a random permutation of $n$ elements has, on average, about $\log n$ cycles. We generalise this fact to all directed $d$-regular graphs on $n$ vertices by showing that, on average, a random cycle-factor of such a graph has $\mathcal{O}((n\log d)/d)$ cycles. This is tight up to the constant factor and improves the best previous bound of the form $\mathcal{O}({n/\sqrt{\log d}})$ due to Vishnoi. It also yields randomised polynomial-time algorithms for finding such a cycle-factor and for finding a tour of length $(1+\mathcal{O}((\log d)/d)) \cdot n$ if the graph is connected. The latter result makes progress on a restriction of the Traveling Salesman Problem to regular graphs, a problem studied by Vishnoi and by Feige, Ravi, and Singh. Our proof uses the language of entropy to exploit the fact that the upper and lower bounds on the number of perfect matchings in regular bipartite graphs are extremely close.

This talk is based on joint work with Micha Christoph, Nemanja Draganić, António Girão, Eoin Hurley, and Alp Müyesser.

Partition diversity refers to the concept that for some networks there may be multiple, similarly plausible ways to group the nodes, rather than one single best partition. In this talk, I will present two projects that address this idea from different but complementary angles. The first introduces the benchmark stochastic cross-block model (SCBM), a generative model designed to create synthetic networks with two distinct 'ground-truth' partitions. This allows us to study the extent to which existing methods for partition detection are able to reveal the coexistence of multiple underlying structures. The second project builds on this benchmark and paves the way for a Bayesian inference framework to directly detect coexisting partitions in empirical networks. By formulating this model as a microcanonical variant of the SCBM, we can evaluate how well it fits a given network compared to existing models. We find that our method more reliably detects partition diversity in synthetic networks with planted coexisting partitions, compared to methods designed to detect a single optimal partition. Together, the two projects contribute to a broader understanding of partition diversity by offering tools to explore the ambiguity of network structure.

In the talk, I'll first describe a more general context of Sárközy-type problems and interesting directions in which they can be pursued. Then, I'll focus on the specific case of bounding the size of sets A s. t. A - A + 1 contains no prime. After describing the progress on the problem for integers, I'll pass on to considering an analogous question for function fields and (after a general introduction to function fields) I'll speak about my recent result in this area.

Speaker Aaron Mishkin will talk about; 'Convex Analysis of Non-Convex Neural Networks' 

One of the key themes in modern optimization is the boundary between convex and non-convex problems. While convex problems can often be solved efficiently, many non-convex programs are NP-Hard and formally difficult. 

In this talk, we show how to break the barrier between convex and non-convex optimization by reformulating, or "lifting", neural networks into high-dimensional spaces where they become convex. These convex reformulations serve two purposes: as algorithmic tools to enable fast, global optimization for two-layer ReLU networks; and as a convex proxy to study variational properties of the original non-convex problem. In particular, we show that shallow ReLU networks are equivalent to models with simple "gated ReLU" activations, derive the set of all critical points for two-layer ReLU networks, and give the first polynomial-time algorithm for optimal neuron pruning. We conclude with extensions to ReLU networks of arbitrary depth using a novel layer-elimination argument.

Details to be finalised

• as holomorphic functions on certain symplectic spaces
• as matrix coefficients of the Heisenberg (and metaplectic) representation,
• as sections of line bundles on abelian varieties.

Based on mathematical ecological models, this report reviews the impact of spatially heterogeneous environments on the persistence and distribution of biological populations. The report aims to elucidate the interplay between population density and key factors, including diffusion coefficients, resource availability, and habitat structure. The study first investigates the ecological consequences of dispersal strategies within environments characterized by uneven resource distribution, demonstrating the monotonic dependence of peak population densities on diffusion rates. Furthermore, analysis of a consumer-resource system indicates that under resource-limited conditions, the ecosystem converges to a globally asymptotically stable state of coexistence. Building on these findings, the report additionally explores the constraints imposed by domain geometry on the spatial patterning of populations.

Approximate subrings are subsets $A$ of a ring $R$ satisfying \[ A + A + AA \subset F + A \] for some finite $F \subset R$. They encode the failure of sum-product phenomena, much like approximate subgroups encode failure of growth in groups.

I will discuss how approximate subrings mirror approximate subgroups and how model-theoretic tools, such as a stabilizer lemma for approximate subrings due to Krupiński, lead to structural results implying a general, non-effective sum-product phenomenon in arbitrary rings: either sets grow rapidly under sum and product, or nilpotent ideals govern their structure. I will also outline related results for infinite approximate subrings and conjectures unifying known (effective) sum-product phenomena.

Based on joint work with Krzysztof Krupiński.

Professor Kaibo Hu will be talking about: 'Finite element form-valued forms'

Some of the most successful vector-valued finite elements in computational electromagnetics and fluid mechanics, such as the Nédélec and Raviart-Thomas elements, are recognized as special cases of Whitney’s discrete differential forms. Recent efforts aim to go beyond differential forms and establish canonical discretizations for more general tensors. An important class is that of form-valued forms, or double forms, which includes the metric tensor (symmetric (1,1)-forms) and the curvature tensor (symmetric (2,2)-forms). Like the differential structure of forms is encoded in the de Rham complex, that of double forms is encoded in the Bernstein–Gelfand–Gelfand (BGG) sequences and their cohomologies. Important examples include the Calabi complex in geometry and the Kröner complex in continuum mechanics.
These constructions aim to address the problem of discretizing tensor fields with general symmetries on a triangulation, with a particular focus on establishing discrete differential-geometric structures and compatible tensor decompositions in 2D, 3D, and higher dimensions.
 

We all sleep. But what determines when and for how long? In this talk I’ll describe some of the fundamental mechanisms that regulate sleep. I’ll introduce the nonsmooth coupled oscillator systems that form the basis of current mathematical models of sleep-wake regulation and discuss their dynamical behaviour. I will describe how we are using models to unravel environmental, societal and physiological factors that determine sleep timing and outline how constructing digital-twins could enable us to create personalised light interventions for sleep timing disorders.

The family of right-angled Artin groups (RAAGs) interpolates 
between free groups and free abelian groups. These groups are defined by 
a simplicial graph: the vertices correspond to generators, and two 
generators commute if and only if they are connected by an edge in the 
defining graph. A key feature of RAAGs is that many of their algebraic 
properties can be detected purely in terms of the combinatorics of the 
defining graph.
The family of outer automorphism groups of RAAGs similarly interpolates 
between Out(F_n) and GL(n, Z). While the l2-Betti numbers of GL(n, Z) 
are well understood, those of Out(F_n) remain largely mysterious. The 
aim of this talk is to introduce automorphism groups of RAAGs and to 
present a combinatorial criterion, expressed in terms of the defining 
graph, that characterizes when the first l2-Betti number of Out(RAAG) 
vanishes.
If time permits, we will also discuss higher l2-Betti numbers and 
algebraic fibring properties of these group

Koszul duality is a powerful mathematical construction.  In this talk, I will take a physical perspective to demonstrate one instance of this duality: an algebraic approach to coupling quantum field theories to a quantum mechanical system on a line.  I will explain how a Lagrangian coupling results in an algebraic object, called a Maurer-Cartan element, and show that there is a sense in which the Koszul dual to the algebra of local operators gives a “universal coupling”.  I will then describe what Koszul duality really “is”, and why many other mathematical constructions deserve the same name.

Junior Strings is a seminar series where DPhil students present topics of common interest that do not necessarily overlap with their own research area. This is primarily aimedl at PhD students and post-docs but everyone is welcome.

We give an introduction to a rigorous renormalization group analysis of the sine-Gordon model with a focus on deriving the lowest-order beta function.

This talk by Tim Austin, at the University of Warwick, will be an introduction to "almost periodic entropy".  This quantity is defined for positive definite functions on a countable group, or more generally for positive functionals on a separable C*-algebra.  It is an analog of Lewis Bowen's "sofic entropy" from ergodic theory.  This analogy extends to many of its properties, but some important differences also emerge.  Tim will not assume any prior knowledge about sofic entropy.

After setting up the basic definition, Tim will focus on the special case of finitely generated free groups, about which the most is known.  For free groups, results include a large deviations principle in a fairly strong topology for uniformly random representations.  This, in turn, offers a new proof of the Collins—Male theorem on strong convergence of independent tuples of random unitary matrices, and a large deviations principle for operator norms to accompany that theorem.

S-duality is an intriguing symmetry of (twisted) N=4 supersymmetric Yang-Mills theory on a four-manifold. When the four-manifold underlies a complex projective surface, it leads to the Vafa-Witten invariants defined by Tanaka-Thomas in 2017. I will discuss some developments related to Azumaya algebras, universality, Seiberg-Witten invariants, wall-crossing for Nakajima quiver varieties, the structure of S-duality, and modular curves (including relations to the Rogers-Ramanujan continued fraction and Klein quartic).

The theory of JSJ decomposition plays a key role in the classification of hyperbolic groups, in analogy with the case of 3-manifolds. While this theory can be extended to larger families of groups, the JSJ decomposition displays significant flexibility in general, making a complete understanding of its behaviour more challenging. In this talk, Dario Ascari explores this flexibility, with an emphasis on the case of generalized Baumslag-Solitar groups.

A temporal graph $G$ is a sequence of graphs $G_1, G_2, \ldots, G_t$ on the same vertex set. In this talk, we are interested in the analogue of the Travelling Salesman Problem for temporal graphs. It is referred to in the literature as the Temporal Exploration Problem, and asks for the minimum length of an exploration of the graph, that is, a sequence of vertices such that at each time step $t$, one either stays at the same vertex or moves along a single edge of $G_t$.

One natural and still open case is when each graph $G_t$ is connected and has bounded maximum degree. We present a short proof that any such graph admits an exploration in $O(n^{3/2}\sqrt{\log n})$ time steps. In fact, we deduce this result from a more general statement by introducing the notion of average temporal maximum degree. This more general statement improves the previous best bounds, under a unified approach, for several studied exploration problems.

This is based on joint work with Carla Groenland, Lukas Michel and Clément Rambaud.

Let g be the Lie algebra of a simple algebraic group over an algebraically closed field of characteristic p. When p=0 the celebrated Jacobson-Morozov Theorem promises that every non-zero nilpotent element of g is contained in a simple 3-dimensional subalgebra of g (an sl2). This has been extended to odd primes but what about p=2? There is still a unique 3-dimensional simple Lie algebra, known colloquially as fake sl2, but there are other very sensible candidates like sl2 and pgl2. In this talk, Adam Thomas from the University of Warwick will discuss recent joint work with David Stewart (Manchester) determining which nilpotent elements of g live in subalgebras isomorphic to one of these three Lie algebras. There will be an abundance of concrete examples, calculations with small matrices and even some combinatorics.