I will review recent progress on extending the Minimal Model Program to algebraically integrable foliations, focusing on applications such as the canonical bundle formula and recent results toward the boundedness of Fano foliations.
Differentiable programming is the underpinning technology for the AI revolution. It allows neural networks to be programmed in very high level user code while still achieving very high performance for both the evaluation of the network and, crucially, its derivatives. The Firedrake project applies exactly the same concepts to the simulation of physical phenomena modelled with partial differential equations (PDEs). By exploiting the high level mathematical abstraction offered by the finite element method, users are able to write mathematical operators for the problem they wish to solve in Python. The high performance parallel implementations of these operators are then automatically generated, and composed with the PETSc solver framework to solve the resulting PDE. However, because the symbolic differential operators are available as code, it is possible to reason symbolically about them before the numerical evaluation. In particular, the operators can be differentiated with respect to their inputs, and the resulting derivative operators composed in forward or reverse order. This creates a differentiable programming paradigm congruent with (and compatible with) machine learning frameworks such as Pytorch and JAX.
In this presentation, David Ham will present Firedrake in the context of differentiable programming, and show how this enables productivity, capability and performance to be combined in a unique way. I will also touch on the mechanism that enables Firedrake to be coupled with Pytorch and JAX.
Please note this talk will take place at Rutherford Appleton Laboratory, Harwell Campus, Didcot.
A way to study rational points on a variety is by looking at their image in the p-adic points. Some natural questions that arise are the following: is there any obstruction to weak approximation on the variety? Which primes might be involved in it? I will explain how primes of good reduction can play a role in the Brauer-Manin obstruction to weak approximation, with particular emphasis on the case of K3 surfaces.
Many optimal control, estimation and design problems can be formulated as so-called dynamic optimization problems, which are optimization problems with differential equations and other constraints. State-of-the-art methods based on collocation, which enforce the differential equations at only a finite set of points, can struggle to solve certain dynamic optimization problems, such as those with high-index differential algebraic equations, consistent overdetermined constraints or problems with singular arcs. We show how numerical methods based on integrating the differential equation residuals can be used to solve dynamic optimization problems where collocation methods fail. Furthermore, we show that integrated residual methods can be computationally more efficient than direct collocation.
This seminar takes place at RAL (Rutherford Appleton Lab).
Persistent homology is infeasible to compute when a dataset is very large. Inspired by the bootstrapping method, Chazal et al. (2014) proposed a multiple subsampling approach to approximate the persistence landscape of a massive dataset. In this talk, I will present an extension of the multiple subsampling method to a broader class of vectorizations of persistence diagrams and to persistence diagrams directly. First, I will review the statistical foundation of the multiple subsampling approach as applied to persistence landscapes in Chazal et al. (2014). Next, I will talk about how this analysis extends to a class of vectorized persistence diagrams called Hölder continuous vectorizations. Finally, I will address the challenges in applying this method to raw persistence diagrams for two measures of centrality: the mean persistence measure and the Fréchet mean of persistence diagrams. I will demonstrate these methods through simulation results and applications in estimating data shapes.
Let $K$ be an unramified extension of $\mathbb{Q}_p$ for a prime $p > 3$. The reduced part of the Emerton-Gee stack for $\mathrm{GL}_{2}$ can be viewed as parameterizing two-dimensional mod $p$ Galois representations of the absolute Galois group of $K$. In this talk, we will consider the extremely non-generic irreducible components of this reduced part and see precisely which ones are smooth or normal, and which have Gorenstein normalizations. We will see that the normalizations of the irreducible components admit smooth-local covers by resolution-rational schemes. We will also determine the singular loci on the components, and use these results to update expectations about the conjectural categorical $p$-adic Langlands correspondence. This is based on recent joint work with Ben Savoie.
In this talk, I will describe recent work in the application of machine learning to explore questions in algebraic geometry, specifically in the context of the study of Q-Fano varieties. These are Q-factorial terminal Fano varieties, and they are the key players in the Minimal Model Program. In this work, we ask and answer if machine learning can determine if a toric Fano variety has terminal singularities. We build a high-accuracy neural network that detects this, which has two consequences. Firstly, it inspires the formulation and proof of a new global, combinatorial criterion to determine if a toric variety of Picard rank two has terminal singularities. Secondly, the machine learning model is used directly to give the first sketch of the landscape of Q-Fano varieties in dimension eight. This is joint work with Tom Coates and Al Kasprzyk.