Past

A longstanding folklore conjecture in combinatorial number theory is the following: given an additive set $S$ not containing the identity, $S$ can be ordered as $s_1, \ldots, s_k$ so that the partial sums $s_1+\cdots+s_j$ are distinct for each $j\in[k]$. We discuss a recent resolution of this conjecture in the finite field model (where the ambient group is $\mathbb{F}_2^n$, or more generally, any bounded exponent abelian group). This is joint work with B. Bedert, M. Bucic, N. Kravitz, and R. Montgomery.

Introduced by Bestvina and Brady in 1997, Bestvina—Brady groups form an important class of examples in geometric group theory and topology, known for exhibiting unusual finiteness properties. In this talk, I will focus on the Dehn functions of finitely presented Bestvina—Brady groups. Very roughly speaking, the Dehn function of a group measures how difficult it is to fill loops by discs in spaces associated to the group, and captures geometric information that is invariant under coarse equivalence. After reviewing known results, I will present a classification of the Dehn functions of Bestvina—Brady groups. This talk is based on joint work with Yu-Chan Chang and Matteo Migliorini.

Fundamentally motivated by the two opposing phenomena of fragmentation and coalescence, we introduce a new stochastic object which is both a process and a geometry. The Brownian marble is built from coalescing Brownian motions on the real line, with further coalescing Brownian motions introduced through time in the gaps between yet to coalesce Brownian paths. The instantaneous rate at which we introduce more Brownian paths is given by λ/g^2  where g is the gap between two adjacent existing Brownian paths. We show that the process "comes down from infinity" when 0<λ<6  and the resulting space-time graph of the process is a strict subset of the Brownian Web on R×[0,∞) . When λ≥6 , the resulting process "does not come down from infinity" and the resulting range of the process agrees with the Brownian Web.

Symplectic capacities are symplectic invariants that measure the “size” of symplectic manifolds and are designed to capture phenomena of symplectic rigidity.

In this talk, I will focus on symplectic capacities of fiberwise convex domains in cotangent bundles. This setting provides a natural link to the systolic geometry of the base manifold. I will survey current results and discuss the variety of techniques used to compute symplectic capacities, ranging from billiard dynamics to pseudoholomorphic curves and symplectic homology. I will illustrate these techniques using disk tangent bundles of ellipsoids as an example.

In this talk, we are interested in neural network approximations for Hamilton–Jacobi–Bellman equations.These are non linear PDEs for which the solution should be considered in the viscosity sense. The solutions also corresponds to value functions of deterministic or stochastic optimal control problems. For these equations, it is well known that solving the PDE almost everywhere may lead to wrong solutions. 

We present a new method for approximating these PDEs using neural networks. We will closely follow a previous work by C. Esteve-Yagüe, R. Tsai and A. Massucco (2025), while extending the versatility of the approach. 

We will first show the existence and unicity of a general monotone abstract scheme (that can be chosen in a consistent way to the PDE), and that includes implicit schemes. Then, rather than directly approximating the PDE -- as is done in methods such as PINNs (Physics-Informed Neural Networks) or DGM (Deep Galerkin Method) -- we incorporate the monotone numerical scheme into the definition of the loss function. 

Finally, we can show that the critical point of the loss function is unique and corresponds to solving the desired scheme. When coupled with neural networks, this strategy allows for a (more) rigorous convergence analysis and accommodates a broad class of schemes. Preliminary numerical results are presented to support our theoretical findings.

This is joint work with C. Esteve-Yagüe and R. Tsai.

We will describe aspects of the geometry of non-minuscule rigid analytic period domains and their covering spaces, and pose some questions about p-adic period mappings and period images by analogy with the complex analytic theory.

Deep neural networks are often seen as different from other model classes by defying conventional notions of generalization. Popular examples of anomalous generalization behaviour include benign overfitting, double descent, and the success of overparametrization. We argue that these phenomena are not distinct to neural networks, or particularly mysterious. Moreover, this generalization behaviour can be intuitively understood, and rigorously characterized using long-standing generalization frameworks such as PAC-Bayes and countable hypothesis bounds. We present soft inductive biases as a key unifying principle in explaining these phenomena: rather than restricting the hypothesis space to avoid overfitting, embrace a flexible hypothesis space, with a soft preference for simpler solutions that are  consistent with the data. This principle can be encoded in many model classes, and thus deep learning is not as mysterious or different from other model classes as it might seem. However, we also highlight how deep learning is relatively distinct in other ways, such as its ability for representation learning, phenomena such as mode connectivity, and its relative universality.

Bio: Andrew Gordon Wilson is a Professor at the Courant Institute of Mathematical Sciences and Center for Data Science at New York University. He is interested in developing a prescriptive foundation for building intelligent systems. His work includes loss landscapes, optimization, Bayesian model selection, equivariances, generalization theory, and scientific applications. 
His website is https://cims.nyu.edu/~andrewgw.

The dynamical Phi^4_3 model is a stochastic partial differential equation that arises in quantum field theory and statistical physics. Owing to the singular nature of the driving noise and the presence of a nonlinear term, the equation is inherently ill-posed. Nevertheless, it can be given a rigorous meaning, for example, through the framework of regularity structures. On compact domains, standard arguments show that any solution converges to the equilibrium state described by the unique invariant measure. Extending this result to infinite volume is highly nontrivial: even for the lattice version of the model, uniqueness holds only in the high-temperature regime, whereas at low temperatures multiple phases coexist.

We prove that, when the mass is sufficiently large or the coupling constant sufficiently small (that is, in the high-temperature regime), all solutions of the dynamical Phi^4_3 model in infinite volume converge exponentially fast to the unique stationary solution, uniformly over all initial conditions. In particular, this result implies that the invariant measure of the dynamics is unique, exhibits exponential decay of correlations, and is invariant under translations, rotations, and reflections.

Joint work with Martin Hairer, Jaeyun Yi, and Wenhao Zhao.

Some molecules exhibit a peculiar behavior during collisions, called resonant: they exchange separately kinetic and internal energies. If the molecules of a gas undergo only resonant collisions, the equilibrium distribution exhibits two distinct temperatures, a kinetic and an internal one. To account for more realistic scenarios, we consider ‘’quasi’’-resonant collisions, where a very tiny exchange between kinetic and internal energies is allowed. We propose a mathematical framework for the notion of quasi-resonance, which leads to a Boltzmann model where the distribution is known at all times, a two-temperature Maxwellian, and converges towards a one-temperature Maxwellian. With this feature at hand, we derive so-called Landau-Teller equations, allowing us to replace the complicated Boltzmann equation by a simple ODE system of two equations.

I will discuss the paper "Construction of Gross-Neveu model using Polchinski flow equation" by Pawel Duch (https://arxiv.org/abs/2403.18562).

I will present short, new proofs of Dowker duality using various poset fiber lemmas. I will introduce modifications of joins and products of simplicial complexes called relational join and relational product complexes. Using the relational product complex, I will then discuss generalizations of Dowker duality to settings of relations among three (or more) sets.

Blood vessels are among the most vital structures in the human body, forming intricate networks that connect and support various organ systems. Remarkably, during early embryonic development—before any blood vessels are visible—their precursor cells are arranged in stereotypical patterns throughout the embryo. We hypothesize that these patterns guide the directional growth and fusion of precursor cells into hollow tubes formed from initially solid clusters. Further analysis of cells within these clusters reveals unique organization that may influence their differentiation into endothelial and blood cells. In this work, I revisit the problem of pattern formation through the lens of active matter physics, using both developing embryonic systems and in vitro cell culture models where similar patterns are observed during tissue budding. These different systems exhibit similar patterning behavior, driven by changes in cellular activity, adhesion and motility.

In 1977 Mazur proved that the rational torsion subgroup of the Jacobian of the modular curve $X_0(N)$, $N > 5$ prime, is generated by the linear equivalence class of the difference of the two cusps. More generally, it is conjectured that for a general $N$, the rational torsion subgroup of the Jacobian of $X_0(N)$ is generated by cusps.  In this talk, we'll discuss a generalisation of this to other modular curves, namely certain covers of $X_0(N)$, indexed by subgroups of $(\mathbf{Z}/N\mathbf{Z})^\times$.

The ’t Hooft expansion is a powerful organizational framework for understanding QFTs as perturbations away from the large N limit and has deep connections to string theory and holography. In this talk, I will discuss categorical aspects of the ’t Hooft expansion, i.e. what one learns about topological defects from the ’t Hooft expansion and, correspondingly, topological strings and twisted holography. This talk is based off the paper arXiv:2411.00760 from last year as well as the more recent review paper arXiv:2511.19776.

In this talk, we'll consider the numerical approximation of singularly perturbed reaction-diffusion partial differential equations, by finite element methods (FEMs).

Solutions to such problems feature boundary layers, the width of which depends on the magnitude of the perturbation parameter. For many hears, some numerical analysts have been preoccupied with constructing methods that can resolve any layers present, and for which one can establish an error estimate that is  independent of the perturbation parameter. Such methods are called "parameter robust", or (in some norms) "uniformly convergent".

In this talk we'll begin with the simplest possible parameter robust FEM: a standard Galerkin finite element method (FEM) applied on a suitably constructed  mesh using a priori information. However, from a practical point of view, not very scalable. To resolve this issue we consider the application of sparse grid techniques. These methods have many variants, two of which we'll consider: the hierarchical basis approach (e.g., Zenger, 1991) and the
two-scale method (e.g., many papers by Aihui Zhou and co-authors). The former can be more efficient, while the latter is considered simpler in both theory and practice.

Our goal is to try to unify these two approaches (at least in two dimensions), and then extend to three-dimensional problems, and, moreover, to other FEMs.
 

In this talk, I will discuss an approach to free boundary minimal surfaces which comes out of recent work by Struwe on a non-local energy, called the half-energy. I will introduce the gradient flow of this functional and its theory in the already studied case of disc type domains, covering existence, uniqueness, regularity and singularity analysis and highlighting the striking parallels with the theory of the classical harmonic map flow. Then I will go on to present new work, joint with Melanie Rupflin and Michael Struwe, which extends this theory to all compact surfaces with boundary. This relies upon combining the above ideas with those of the Teichmüller harmonic map flow introduced by Rupflin and Topping.

In this talk, we employ a level-set method to define complex computational domains and propose a ghost nodal finite element strategy tailored for two distinct applications. In the first part, we introduce a model for a Poisson-Nernst-Planck system that accounts for the correlated motion of positive and negative ions through Coulomb interactions. For very short Debye lengths, one can adopt the so called Quasi-Neutral limit which drastically simplifies the system, reducing it to a diffusion equation for a single carriers with effective diffusion coefficient. This approach, while simplifying the mathematical model, can limit the scope of numerical simulations, as it may not capture the full range of behaviors near the Quasi-Neutral limit. Our goal is therefore to design an Asymptotic Preserving (AP)  to handle both regimes: the full system when the Debye length is small but non-negligible, and the Quasi-Neutral regime as the Debye length approaches zero. In the second part, we study the formation of biological transportation networks governed by a nonlinear elliptic equation for the pressure coupled with a reaction-diffusion parabolic equation for the conductivity tensor. We compute numerical solutions using the proposed ghost nodal finite element method, which shows that the network becomes highly intricate and its branches extend over large portions of the domain.

Wave energy has the theoretical potential to meet global electricity demand, but it remains less mature and less cost-competitive than wind or solar power. A key barrier is the absence of engineering convergence on an optimal wave energy converter (WEC) design. In this work, I demonstrate how geometry optimisation can deliver step-change improvements in WEC performance. I present methodology and results from optimisations of two types of WECs: an axisymmetric point-absorber WEC and a top-hinged WEC. I show how the two types need different optimisation frameworks due to the differing physics of how they make waves. For axisymmetric WECs, optimisation achieves a 69% reduction in surface area (a cost proxy) while preserving power capture and motion constraints. For top-hinged WECs, optimisation reduces the reaction moment (another cost proxy) by 35% with only a 12% decrease in power. These result show that geometry optimisation can substantially improve performance and reduce costs of WECs.

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

Data assimilation plays a crucial role in modern weather prediction, providing a systematic way to incorporate observational data into complex dynamical models. The paper addresses continuous data assimilation for a model arising as a singular limit of the three-dimensional compressible Navier-Stokes-Fourier system with rotation driven by temperature gradient. The limit system preserves the essential physical mechanisms of the original model,  while exhibiting a reduced, effectively two-and-a-half-dimensional structure. This simplified framework allows for a rigorous analytical study of the data assimilation process while maintaining a direct physical connection to the full compressible model.  We establish well posedness of global-in-time solutions and a compact trajectory attractor, followed by the stability and convergence results for the nudging scheme applied to the limiting system. Finally, we demonstrate how these results can be combined with a relative entropy argument to extend the assimilation framework to the full three-dimensional compressible setting, thereby establishing a rigorous connection between the reduced and physically complete models.

In this talk, we will explore the emerging role of generative AI in mathematical research. Building on insights from the “Malliavin–Stein experiment”, carried out in collaboration with Charles-Philippe Diez and Luis Da Maia, we will discuss our experience and reflect on how AI might influence the way mathematics is conceived, proven, and created.

The simplex of traces of a unital C*-algebra has long been regarded as a central invariant in the theory. Likewise, from the group-theoretic perspective, the simplex of traces of a discrete group (namely, the simplex of traces of its maximal C*-algebra) is a fundamental object in harmonic analysis, and the study of this simplex led to many applications in recent years.

Itamar Vigdorovich , UCSD, will discuss several results describing the simplex of traces in concrete and significant cases. These include Property (T) groups and especially higher rank lattices, for which the simplex of traces is as tame as possible. In contrast, for free products, the simplex is typically as wild as possible, yet still admits a canonical and universal structure—the Poulsen simplex. In ongoing work, an analogous result is obtained for the space of traces on the fundamental group of a closed surface of genus g≥2.

Itamar presents these results, outlines the main ideas behind the proofs, and gives an overview of the central concepts. The talk is based on joint works with Gao, Ioana, Levit, Orovitz, Slutsky, and Spaas.

The weak noise theory (WNT) provides a framework for accessing large deviations in models of the Kardar-Parisi-Zhang (KPZ) universality class, probing the regime where randomness is small, fluctuations are rare, and atypical events dominate. Historically, two methods have been available: asymptotic analysis of Fredholm determinant formulas—applicable only for special initial data—and variational or saddle-point formulations leading to nonlinear evolution equations, which were mostly accessible perturbatively.

This talk explains how these approaches can be unified: the weak-noise saddle equations of KPZ-class models form classically integrable systems, admitting Lax pairs, conserved quantities, and an inverse scattering framework. In this setting, the large-deviation rate functions arise directly from the conserved charges of the associated integrable dynamics.

The discussion will focus on three examples:

1. The scalar Strict-Weak polymer ;
2. A matrix Strict-Weak polymer driven by Wishart noise ;
3. If time permits, the continuous-time q-TASEP.

In the 1990s, physicists introduced an ideal way to count curves inside a Calabi-Yau 3-fold, called the Gopakumar-Vafa (GV) theory. Building on several previous attempts, Maulik-Toda recently gave a mathematical rigorous definition of the GV invariants. We expect that the GV invariants and the Gromov-Witten (GW) invariants are related by an explicit formula, but this stands as a challenging open problem. In this talk, I will explain recent mathematical developments on the GV theory, especially for local curves, including the cohomological chi-independence theorem and the GV/GW correspondence in a special case.

An important invariant in the complex representation theory of reductive p-adic groups is the wavefront set, because it contains information about the character of such a representation. In this talk, Mick Gielen will introduce a new invariant called the canonical dimension, which can be said to measure the size of a representation and which has a close relation to the wavefront set.  He will then state some results he has obtained about the canonical dimensions of compactly induced representations and show how they teach us something new about the wavefront set. This illustrates a completely new approach to studying the wavefront set, because the methods used to obtain these results are very different from the ones usually used.

How many vectors are needed to simultaneously generate $m$ complete flags in $\mathbb{R}^d$, in the worst-case scenario?  A classical linear algebra fact, essentially equivalent to the Bruhat cell decomposition for $\text{GL}_d$, says that the answer is $d$ when $m=2$.  We obtain a precise answer for all values of $m$ and $d$.  Joint work with Federico Glaudo and Chayim Lowen.

TBA

Surfactants are chemicals that preferentially reside at interfaces. Once surfactant molecules have adsorbed to an interface, they reduce the surface tension between the two neighbouring fluids and may induce fluid flow. Surfactants have many household applications, such as in cleaning products and cosmetics, as well as industrial applications, like mineral processing and agriculture. Thus, understanding the dynamics of surfactant solutions is particularly important with regards to improving the efficacy of their applications as well as highlighting how they work. In this seminar, we will explore the spreading of a droplet over a substrate, in which there is constant injection of liquid and soluble surfactant through a slot in the substrate. Firstly, we will see how the inclusion of surfactant alters the spreading of the droplet. We will then investigate the early- and late-time behaviour of our model and compare this with numerical simulations. We shall conclude by briefly examining the effect of changing the geometry of the inflow slot.

Estimating the correlation $\sum_{n \le x} d_k(n)d(n+h)$ is a central problem in analytic number theory. In this talk, I will present a method to obtain an asymptotic formula for a smoothed version of this sum. A key feature of the result is a power-saving error term whose exponent does not depend on $k$, improving earlier bounds where the quality of the saving deteriorates with $k$. The argument relies on balancing three distinct bounds for the remainder term according to the sizes of the factors of $n$.

With the classification theory of simple and nuclear C*-algebras of real rank zero advanced to a level which may very well be final, it is natural to wonder what happens when one allows ideals, but not too many of them. Contrasting the simple case, the K-theoretical classification theory for real rank zero C*-algebras with finitely many ideals is only satisfactorily developed in subcases, and in many settings it is even unclear and/or disputed which flavor of K-theory to use.

Restricting throughout to the setting of real rank zero, Søren Eilers will compare what is known of the classification of graph C*-algebras and of approximately subhomogeneous C*-algebras, with an emphasis on what kind of conclusion can be extracted from restrictions on the complexity of the ideal lattice. The results presented are either more than a decade old or joint with An, Liu and Gong.

I will discuss a uniform waist inequality in codimension 2 for the family of finite covers of a Riemannian manifold whose fundamental group has Kazhdan‘s property T. I will describe a general strategy to prove waist inequalities based on a higher property T for Banach spaces. The general strategy can be implemented in codimension 2 but is conjectural in higher codimension. We speculate about the situation for lattices in semisimple Lie groups. Based on joint work with Uri Bader

As discovered by Perelman, the study of ancient Ricci flows which are $\kappa$-noncollapsed is a crucial prerequisite to understanding the singularity behaviour of more general Ricci flows. In dimension three, these so-called "$\kappa$-solutions" have been fully classified through the groundbreaking work of Brendle, Daskalopoulos, and Šešum. Their classification result can be extended to higher dimensions, but only for those Ricci flows that have uniformly positive isotropic curvature (PIC), as well as weakly-positive isotropic curvature of the second type (PIC2); it appears the classification result fails with only minor modifications to the curvature assumption. Indeed, with the alternative assumption of non-negative curvature operator, a rich variety of new examples emerge, as recently constructed by Buttsworth, Lai, and Haslhofer; Haslhofer himself has conjectured that this list of non-negatively curved $\kappa$-solutions is now exhaustive in dimension four. In this talk, we will discuss some recent progress towards resolving Haslhofer's conjecture, including a compactness result for non-negatively curved $\kappa$-solutions in dimension four, and a symmetry improvement result for bubble-sheet regions. This is joint work with Anusha Krishnan and Timothy Buttsworth. 

 Hyper-Kähler manifolds are rigid geometric structures. They have three different symplectic and complex structures, in direct analogy with the quaternions. Being Ricci-flat, they solve the vacuum Einstein equations, and so there has been considerable interest among physicists to explicitly construct such spaces. We will discuss in detail the examples arising from the Gibbons-Hawking ansatz. These give concrete descriptions of the metric, giving many examples to work with. They also lead to the generalised classification as hyper-Kähler quotients by P.B. Kronheimer, with one such space for each finite subgroup of SU(2). Finally, we will look at the McKay correspondence, relating the finite subgroups of SU(2) with the simple Lie algebras of type A,D,E.

Merge trees are a topological descriptor of a filtered space that enriches the degree zero barcode with its merge structure. The space of merge trees comes equipped with an interleaving distance d_I , which prompts a naive question: is the interleaving distance between two merge trees equal to the bottleneck distance between their corresponding barcodes? As the map from merge trees to barcodes is not injective, the answer as posed is no, but as proposed by Gasparovic et al., we explore intrinsic metrics d_I and d_B realized by infinitesimal path length in merge tree space, which do indeed coincide. This result suggests that in some special cases the bottleneck distance (which can be computed quickly) can be substituted for the interleaving distance (in general, NP-hard).

Fusion systems are a generalisation of finite groups designed in a way to capture local structure at a prime motivated by the existence of "exotic" fusion systems; local structures that do not appear in any finite group. In this talk I will give a brief introduction to fusion systems with emphasis on how they relate to groups. I will then discuss recent work done on fusion invariant character theory, concluding with a short excursion into biset functor theory to state a character value formula for "induction" between fusion systems and a Frobenius reciprocity analogue.

Join us on Friday Week 7 for some chill board games! Meet in N4.01 at 12pm for a Taylors sandwich lunch and positive end-of-term vibes.

The universal infinitesimal symmetry of a holomorphic field theory is the Lie algebra of holomorphic vector fields. We introduce the higher-dimensional Virasoro algebra and prove a local, universal, form of the Riemann–Roch theorem using Feynman diagrams. We use the concept of a (Jouanoulou) higher-dimensional chiral algebra as developed recently with Gui and Wang. We will remark on applications to superconformal field theory. This project is joint work with Zhengping Gui.