# Postgraduate Seminar

Hi! Welcome to the Warwick Mathematics Postgraduate Seminar page. Every week during term time we meet to discuss both our research and the topics that interest us. The aim is to deliver this content in an accessible format for a general audience.

The talks will be held in person, however can also be viewed live online here. For those who join in person, **lunch** will be provided after each talk in the postgraduate common room.

**Time:** 12:00-13:00 Wednesday

**Location:** B3.02 Zeeman

**Online link: **join teams

**Organisers:** Cam HeatherLink opens in a new window and Marc TruterLink opens in a new window

**Term 2 2023/2024**

**Week 1****: 10th January**

Andrew Nugent** Approaches to modelling opinion formation**

Do you prefer an ODE or a PDE? A computer simulation or a proof? An analytic approximation or a data-driven discovery? Whatever your preference it's (probably) got an application to opinion dynamics. This talk will give a whistle-stop tour through various techniques that have been used to model opinion formation, with a focus on how different models are connected and complementary. Throughout we'll be trying to answer the big question: when do people agree? And if not, why not?

**Week 2****: 17th January**

Joris Koefler (Max-Planck Institute)** Exploring positive geometries and their link to Physics**

Join us for an accessible introduction to positive geometries—an emerging field blending algebraic/polytopal geometry, combinatorics, and particle Physics. We will begin by exploring key concepts of positive geometry and in the latter part we will showcase its surprising connection to Quantum Field theories.

**Week 3****: 24th January**

Glen Salter **Continued Fractions, Chaos and The Modular Surface**

** **

**Week 4****: 31st January**

William O'Regan** Introduction to Kolmogorov complexity and applications to problems in fractal geometry**

**Week 5****: 7th February**

Philip Holdridge ** I see threes of green: Colouring results in number theory**

**Week 6****: 14****th February**

Nicola Rosetti** Random walks and electrical circuits**

Did you know that electrons in a circuit and drunk people on the street have something in common? Well this is not properly true, however the so called "drunkard walk" and electric circuits, share somehow some common features; indeed in the course of the seminar, we will see how both can be modelled by a random walk on a graph and we will discover how such connection will help us proving a beautiful theorem: Polya's recurrence and transience theorem. Approaching the problem from the point of view of resistor networks will in fact allow us to prove the result in a more elegant way, moreover it will lead us to prove the same result in the more general setting where the conductances of the network are random.

**Week 7****: 21st**** February**

Tarek Acila** Qualitative Mathematical Modelling of Inverse Blebbing**

Biological cells live in a hostile environment in which they are constantly exposed to external forces. To preserve their functionality, they employ their mechanical and chemical abilities to change their shapes accordingly, thanks to their elastic membranes. Shape change is a biophysical mechanism used by cells to counteract any applied forces. In this talk, we will focus on a specific type of shape change known as Inverse Blebbing. We will start with a biological introduction and then slowly make our way to the mathematical set-up of this problem. We derive a qualitative continuum mathematical PDE model that describes inverse blebbing using techniques from differential geometry and continuum mechanics. To conclude our presentation, we discuss why and how this model can be improved in the future.

**Week 8****: 28****th February**

Kristian Romano** Real Time Telemetric Monitoring of the Circadian Rhythm via Hidden Markov Models for Advanced Pancreatic Cancer Patients undergoing Chemotherapy**

Disruptions of the Circadian Timing System in cancer patients are associated with poorer treatment outcomes, and short progression-free and overall-survival. The MultiDom clinical trial (NCT04263948) telemonitors the physical activity and body temperature of patients with advanced or metastatic pancreatic cancer undergoing chemotherapy aiming to reduce toxicity-related emergency hospitalizations (Bouchahda et al. 2023BMJ Open). We develop methodology based on Hidden Markov Models (HMM) that allows for near-real time estimation of circadian parameters dynamics to inform proactive decision making by the medical team. For that aim we developed a HMM with a time changing transition matrix that we estimated via state of the art gradient based methods. As we assumed Zero Inflated Gamma emissions, we will employ a Metropolis-Hastings for the shape parameter with an educated proposal distribution. The work presented provides a first step into the reproducible and fully automatic Bayesian computation in real time of circadian parameters of interest for the purpose of telemonitoring. The approach also facilitates a systematically updated daily quantification of uncertainty. We will demonstrate the use of our methodology in the context of tele monitoring patients undergoing chemotherapeutic treatment, thus highlighting the potential for circadian digital and precision medicine.

**Week 9****: 6th March**

**Week 10****: 13th March**

**Term 1 2023/2024**

**Week 1 : 4th October**

Katerina Santicola** Reverse-Engineering **

The simplest way to prove something exists is not always by constructing it. Such proofs are called ineffective proofs. Falting's Theorem is an example of an ineffective result: it tells us that every curve (of genus at least 2) "belongs" to a finite set of rational points. In this talk, I will convince you that the converse is true: that every finite set of rational points "belongs" to a curve. Our proof is, by contrast, effective. I will also convince you that this has nothing to do with reverse-engineering.

**Week 2: 11th October**

Luke Murray Kearney** ****Data-driven age structured contact networks**

Network structure provides a powerful mechanism through which to analyse social interactions at an individual level. Many respiratory and sexually transmitted diseases travel through these social connections, allowing epidemiologists to use contact networks to analyse disease transmission in a population. Much work in network epidemiology attempts to reproduce an exact contact network from outbreak data to aid in control or surveillance activities such as contact tracing or aims to create a representative network for use in outbreak simulation. However, there is no standardised approach to creating a suitable network from the type of individual data that are routinely collected. In this project, we propose two novel approaches to create age-specific network representations from real world data that are adaptable to times of social restriction in response to an outbreak. We test these approaches using two periods of varied lockdown strictness during the Covid-19 pandemic in the UK, using the degree distributions of the egocentric data provided by the CoMix study and the between demographic contact rates. We further analyse the scale-ability and characteristics of our approaches through their computation times and the structure of the resulting structure of the network.

Nathan Van der Riet **From pancakes to cups: the physically driven structure of the Golgi Apparatus**** **

The Golgi Apparatus is a key organelle found in eukaryotic cells responsible for modifying proteins and lipids and distributing them to their target destinations. Akin to an amped up post office, the Golgi is in a constant state of flux, receiving and distributing packages, and yet it has a remarkably stable structure. This structure is of interest to biologists and now, hopefully, to mathematicians. This is due to its two distinct compartments which can be characterised by their distinct Gaussian curvatures. We seek to explain this distinction by means of a minimal physical model, assuming that the Golgi’s lipid membrane consists of two molecules with opposing affinities for Gaussian curvature. By following through a chain of reasoning on vesiculation events and by modelling the Golgi as a dynamical system we come up with a plausible explanation for the existence of the negative-Gaussian curvature structure. Encouraged, we seek then to produce from a handful of assumptions and deductions a physically motivated dynamical system whose equilibrium is consistent with qualitative observations of the Golgi’s structure.

**Week 3: 18th October**

Layne Hall** Between chaotic flows and combinatorics**

A flow is a continuous dynamical system, like a set of ODEs. Especially in dimension three, there are deep relationships between the dynamics of flows and the topological properties of the underlying space. We will introduce, through examples, a large class of flows which exemplify this connection. We will see that these flows have robust, chaotic dynamics. Then, we will discuss how the flows can be elegantly characterised through combinatorial topology.

**Week 4: 25th October**

Abigail Hollingsworth** Hyperbolic Knot Theory**

** **

A key question in knot theory is how to differentiate two knots. The knot invariant we will study is the knot complement. Knots can be classified into three groups, one of which is hyperbolic knots, which have hyperbolic knot complements. To understand when a knot complement is hyperbolic, we will see an overview of hyperbolic geometry in two and three dimensions and define the three key equations that decide if a knot complement is hyperbolic. These equations define how the tetrahedral decomposition of the knot complement glues together. This talk is inspired by Jessica Purcell’s book on Hyperbolic Knot Theory.

**Week 5****: 1st November**

Hefin Lambley **What does it mean to be most likely?**

The concept of a most likely value, or mode, appears frequently in statistics and applied mathematics. Sometimes the definition is obvious. Often it is not, e.g. in large-deviation theory or the Bayesian approach to inverse problems.

This talk will introduce a general theory for modes of probability distributions on metric spaces. In most cases, this theory works very well, and admits an intuitive order-theoretic formulation. However, we'll also see some pathological distributions where the definition of a mode is very unclear, and where there might not even exist a good approximation to a mode.**Week 6****: 8th November**

Patience Ablett ** How many points to verify a Hilbert polynomial?**

**Week 7****: 15th November**

Yorick Fuhrmann** From formal groups to elliptic cohomology**

Give me a ring and a formal group law over it, then I'll give you a cohomology theory! Although... I have to admit, your ring should be Landweber flat as an -module. In this talk we will find out what all that means and how complex cobordism is able to classify cohomology theories which have Euler classes for complex vector bundles. This is a celebrated result of Daniel Quillen which opened up the world of chromatic homotopy theory. We will use it to rephrase topological -theory in terms of complex cobordism and see how it can be utilized to get new cohomology theories from elliptic curves.

**Week 8: 22nd November**

David Hubbard** Torsion Points on Curves in the Torus**

Lang posed the following question on solutions of Diophantine equations which consist of pairs of roots of unity:

“If f(x,y)=0 is a Diophantine equation, then when is the set {(a,b) | f(a,b)=0 a,b roots of unity} finite?” Geometrically, this set is the intersection of the curve X={f=0} with torsion points of the algebraic 2-torus (C*)^2. Generically, this set is finite unless X is ‘special’.This problem belongs to the ‘Unlikely Intersections’ of arithmetic geometry which contains the Mordell, André-Oort and Manin-Mumford Conjecture. A great deal of recent research has aimed to find uniform versions of these conjectures, and I will present the uniform version of Lang’s problem by Beukers and Smyth using methods from convex geometry.

**Week 9****: 29th November**

Laura Bradby** If I made a list of every doughnut, would it be compact?**

Suppose I wanted to make a list of all compact genus 1 surfaces. How many are there? Well, given such a torus (or indeed, any genus g surface), if you consider its topological structure and work up to homeomorphism, then there’s only one. How many if you consider smooth structures, up to diffeomorphism? Surprisingly, still just one. What if you care about complex structure, and work up to biholomorphism? Uncountably many! The last case is the only interesting one from a list-making perspective, and has the major problem that uncountably many things are famously impossible to list. So instead, we extend our idea of “list”, and do what any sensible person would do with uncountably many things: shove them all together and give that a topology. This gives us the concept of a *moduli space*, and in this talk we will discuss the moduli space of the torus, and then the moduli spaces of surfaces of higher genus, using the genus 1 case as a guiding example. Our aim will be to see how to build these spaces in a sensible way, and to begin to understand what they look like, including how “close” they are to being compact.

**Week 10****: 6th December**

Muhammad Manji **Local-Global Principles**

When studying geometric objects, sometimes we care about questions which exist only globally, for example; where is the maximum of a function, if such a thing exists? At other times we care about questions which are only local; what is the derivative at a point of interest? If a property holds everywhere locally, will it necessarily give us a global condition? By patching together local information we can sometimes tackle global problems, and at other times we can't. We will discuss some of these questions and deal with the basics of arithmetic geometry.