Past

Peel Ports operate a number of locks that allow ships to enter and leave the port. The lock gates comprise a single caisson structure which blocks the waterway when closed and retracts into the dockside as the gate opens. Build up of silt ahead of the opening lock gate can prevent it from fully opening or requiring excessive power to move. If the lock is not able to fully open, ships are unable to enter the port, leading to significant operational impacts for the whole port. Peel ports are interested in understanding, and mitigating, this silt build up. 

In joint work with Jacob Tsimerman we study when the primitive
of a given algebraic function can be constructed using primitives
from some given finite set of algebraic functions, their inverses,
algebraic functions, and composition. When the given finite set is just {1/x}
this is the classical problem of "elementary integrability".
We establish some results, including a decision procedure for this problem.

The design and analysis of finite element methods for high Reynolds flow remains a challenging task, not least because of the difficulties associated with turbulence. In this talk we will first revisit some theoretical results on interior penalty methods using equal order interpolation for smooth solutions of the Navier-Stokes’ equations at high Reynolds number and show some recent computational results for turbulent flows.

Then we will focus on so called pressure robust methods, i.e. methods where the smoothness of the pressure does not affect the upper bound of error estimates for the velocity of the Stokes’ system. We will discuss how convection can be stabilized for such methods in the high Reynolds regime and, for the lowest order case, show an interesting connection to turbulence modelling.

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 aimed at PhD students and post-docs but everyone is welcome

Numerous applications in geometric, visual, and scientific computing rely on the ability to nicely distribute points in space according to a repulsive potential.  In contrast, there has been relatively little work on equidistribution of higher-dimensional geometry like curves and surfaces—which in many contexts must not pass through themselves or each other.  This talk explores methods for optimization of curve and surface geometry while avoiding (self-)collision. The starting point is the tangent-point energy of Buck & Orloff, which penalizes pairs of points that are close in space but distant with respect to geodesic distance. We develop a discretization of this energy, and introduce a novel preconditioning scheme based on a fractional Sobolev inner product.  We further accelerate this scheme via hierarchical approximation, and describe how to incorporate into a constrained optimization framework. Finally, we explore how this machinery can be applied to problems in mathematical visualization, geometric modeling, and geometry processing.

When considering compatifications of heterotic string theory down to 3D, the heterotic $G_2$ system arises naturally. It is a system for both geometric fields and gauge fields over a manifold with a $G_2$-structure. In particular, it asks for the $G_2$-structure to be coclosed. We will begin this talk defining this system and giving a description of the geometry of cohomogeneity one manifolds. Then, we will look for coclosed $G_2$-structures in the cohomogeneity one setting. We will end up by proving the existence of a family of coclosed $G_2$-structures which are invariant under a cohomogeneity one action of $\text{SU}(2)^2$ on certain seven-dimensional simply connected manifolds.

One of the most important constructions in operator algebras is the tracial ultrapower for a tracial state on a C*-algebra. This tracial ultrapower is a finite von Neumann algebra, and it appears in seminal work of McDuff, Connes, and more recently by Matui-Sato and many others for studying the structure and classification of nuclear C*-algebras. I will talk about how to generalise this to unbounded traces (such as the standard trace on B(H)). Here the induced tracial ultrapower is not a finite von Neumann algebra, but its multiplier algebra is a semifinite von Neumann algebra.

The introduction to the talk will review basics of $G_{2}$ geometry in seven dimensions, and associative and co-associative submanifolds. In one part of the talk we will explain how fibrations with co-associative fibres, near the “adiabatic limit” when the fibres are very small,  give insights into various questions about moduli spaces of $G_{2}$ structures and singularity formation. This part is mostly speculative. In the other part of the talk we discuss some analysis questions which enter when setting up the foundations of this adiabatic theory. These can be seen as codimension 2 analogues of free boundary problems and related questions have arisen in a number of areas of differential geometry recently.

The Jacobi Ensembles of random matrices have joint distribution of eigenvalues proportional to the integration measure in the Selberg integral. They can also be realised as the singular values of principal submatrices of random unitaries. In this talk we will review some old and new results concerning the distribution of the largest and smallest eigenvalues.

In this talk we discuss solution of the well-known problem of Erdős and Sauer from 1975 which asks for the maximum number of edges an $n$-vertex graph can have without containing a $k$-regular subgraph, for some fixed integer $k\geq 3$. We prove that any $n$-vertex graph with average degree at least $C_k\log\log n$ contains a $k$-regular subgraph. This matches the lower bound of Pyber, Rödl and Szemerédi and substantially
improves an old result of Pyber, who showed that average degree at least $C_k\log n$ is enough.

Our method can also be used to settle asymptotically a problem raised by Erdős and Simonovits in 1970 on almost regular subgraphs of sparse graphs and to make progress on the well-known question of Thomassen from 1983 on finding subgraphs with large girth and large average degree.

Joint work with Oliver Janzer

Motivated by the physical relevance of many Hodge Laplace-type PDEs from the finite element exterior calculus, we analyse the Hodge Laplacian boundary value problem arising from the strain space in the linear elasticity complex, an exact sequence of function spaces naturally arising in several areas of continuum mechanics. We propose a discretisation based on the adaptation of discontinuous Galerkin FEM for the incompatibility operator $\mathrm{inc} := \mathrm{rot}\circ\mathrm{rot}$, using the symmetric-tensor-valued Regge finite element to discretise  the strain field; via the 'Regge calculus', this element has already been successfully applied to discretise another metric tensor, namely that arising in general relativity. Of central interest is the characterisation of the associated Sobolev space $H(\mathrm{inc};\mathbb{R}^{d\times d}_{\mathrm{sym}})$. Building on the pioneering work of van Goethem and coauthors, we also discuss promising connections between functional analysis of the $\mathrm{inc}$ operator and Kröner's theory of intrinsic elasticity in the presence of defects.

This is based on ongoing work with Dr Kaibo Hu.

By a classical theorem of Jordan, every faithful transitive action of a nontrivial finite group admits a derangement (an element with no fixed points). More recently, the existence of derangements with additional properties has attracted much attention, especially for primitive actions of almost simple groups. Surprisingly, there exist almost simple groups with elements that are derangements in every faithful primitive action; we say that these elements are totally deranged. I'll talk about ongoing work to classify the totally deranged elements of almost simple groups, and I'll mention how this solves a question of Garzoni about invariable generating sets for simple groups.

In this talk, I review some recent results obtained for black holes using
effective field theory methods applied to quantum gravity, in particular the
unique effective action. Black holes are complex thermodynamical objects
that not only have a temperature but also have a pressure. Furthermore, they
have quantum hair which provides a solution to the black hole information
paradox.

The aim of this talk is to present some novel inequalities for the k-plane transform acting on the modulus square of solutions of the linear time-dependent Schrodinger equation. Our motivation for studying these tomographic expressions comes for virial identities in the context of Schrodinger equations, where tomographic Strichartz estimates of the type we will discuss here appear naturally.

I'll introduce the classical arithmetic topology dictionary of Mumford-Manin-Mazur-Morishita-etc. I'll then present an interesting instance of parallel phenomena related to symplectic structures on moduli spaces of certain bundles. The arithmetic side turns out to be an application of Poitou-Tate duality. Depending on time, I'll delve into the delicate details which make the analogy useful for Diophantine geometers.

Noether’s theorem plays a fundamental role in modern physics by relating symmetries of a Lagrangian to conserved quantities of the Euler-Lagrange equations. In ideal fluid dynamics, the theorem relates the particle labeling symmetry to a Kelvin circulation law. Circulation is conserved for incompressible flows and, otherwise, is generated by advected variables through the momentum map due to a broken symmetry. We will introduce variational principles for fluid dynamics that constrain advection to be the sum of a smooth and geometric rough-in-time vector field. The corresponding rough Euler-Poincare equations satisfy a Kelvin circulation theorem and lead to a natural framework to develop parsimonious non-Markovian parameterizations of subgrid-scale dynamics.

I will describe joint work with Tyler Kelly and Ran Tessler. FJRW (Fan-Jarvis-Ruan-Witten) theory is an enumerative theory of quasi-homogeneous singularities, or alternatively, of Landau-Ginzburg models. It associates to a potential W:C^n -> C given by a quasi-homogeneous polynomial moduli spaces of (orbi-)curves of some genus and marked points along with some extra structure, and these moduli spaces carry virtual fundamental classes as constructed by Fan-Jarvis-Ruan. Here we specialize to the case W=x^r+y^s and construct an analogous enumerative theory for disks. We show that these open invariants provide perturbations of the potential W in such a way that mirror symmetry becomes manifest. Further, these invariants are dependent on certain choices of boundary conditions, but satisfy a beautiful wall-crossing formalism.

Predicting a protein’s structure from its primary sequence has been a grand challenge in biology for the past 50 years, holding the promise to bridge the gap between the pace of genomics discovery and resulting structural characterization. In this talk, we will describe work at DeepMind to develop AlphaFold, a new deep learning-based system for structure prediction that achieves high accuracy across a wide range of targets. We demonstrated our system in the 14th biennial Critical Assessment of Protein Structure Prediction (CASP14) across a wide range of difficult targets, where the assessors judged our predictions to be at an accuracy “competitive with experiment” for approximately 2/3rds of proteins. The talk will cover both the underlying machine learning ideas and the implications for biological research as well as some promising further work.

In this talk, I will investigate the origin of Euclidean wormholes in the gravitational part integral in the context of AdS/CFT. These geometries are confusing since they prevent products of partition functions to factorize, as they should in any quantum mechanical system. I will briefly review the different proposals for the origin of these wormholes, one of which is that one should consider ensemble of average of boundary systems instead of a fixed quantum system with a fixed Hamiltonian. I will explain that it seems unlikely that one can average over CFTs and present a new idea: averaging over approximate CFTs, which I will define. I will then study the variance of the crossing equation in an ensemble relevant for 3d gravity. Based on work in progress with de Boer, Jafferis, Nayak and Sonner.

I will describe recent work with Zhibeodov on the boundary interpretation of orbits around an AdS black hole. When the orbits are far away from the black hole, these orbits describe heavy-light double-twist operators on the boundary. I will discuss how the dimensions of these operators can be computed exactly in terms of quasinormal modes in the bulk, using techniques from a paper to appear soon with Grassi, Iossa, Lichtig, and Zhiboedov. Then I will explain how these results are related to the concept of quantum scars, which are eigenstates that do not obey ETH. 

Melanie Weber 

Title: Geometric Methods for Machine Learning and Optimization

Abstract: A key challenge in machine learning and optimization is the identification of geometric structure in high-dimensional data. Such structural understanding is of great value for the design of efficient algorithms and for developing fundamental guarantees for their performance. Motivated by the observation that many applications involve non-Euclidean data, such as graphs, strings, or matrices, we discuss how Riemannian geometry can be exploited in Machine Learning and Optimization. First, we consider the task of learning a classifier in hyperbolic space. Such spaces have received a surge of interest for representing large-scale, hierarchical data, since they achieve better representation accuracy with fewer dimensions. Secondly, we consider the problem of optimizing a function on a Riemannian manifold. Specifically, we will consider classes of optimization problems where exploiting Riemannian geometry can deliver algorithms that are computationally superior to standard (Euclidean) approaches.

Francesca Panero

Title: A general overview of the different projects explored during my DPhil in Statistics.

Abstract: In the first half of the talk, I will present my work on statistical models for complex networks. I will propose a model to describe sparse spatial random graph underpinned by the Bayesian nonparametric theory and asymptotic properties of a more general class of these models, regarding sparsity, degree distribution and clustering coefficients.

The second half will be devoted to the statistical quantification of the risk of disclosure, a quantity used to evaluate the level of privacy that can be achieved by publishing a microdata file without modifications. I propose two ways to estimate the risk of disclosure, using both frequentist and Bayes nonparametric statistics.