MS Talk Series
‘How is it to do an MS thesis?’, ‘What does writing a paper feel like?’, ‘What are the topics and exciting areas of research out there?’ These questions are always there in the minds of BS-MS students. This will be a series of talks intended to give the general audience an appreciation for the work carried out during the MS thesis. 5th-year Mathematics BS-MS students who are going to present their MS thesis work this year are going to deliver these talks.
LHC 101
3rd-16th April 2023 , 07:00pm
Talks of 2023
Loop Groups and Simplicial Homotopy Theory - Karthik Vasisht
Abstract: In this talk, we introduce simplicial homotopy theory and its relation to "classical" algebraic topology. The talk will begin with the basic notions of homotopy and homotopy groups from algebraic topology. We then introduce the notion of simplicial sets and the simplicial versions of homotopy and homotopy groups. Finally, we'll look at a situation where working with simplicial sets is more advantageous as compared to working with topological spaces. The talk will involve many examples to make it easier to understand the concepts that are being developed. I will also talk about my experiences in conducting research for the past year.
Date & Time: 3rd April 2023, 07:00pm
The 'Lonely Runner Conjecture' - Hrishikesh V
Abstract: Imagine that you are running on a circular track. For company, you have a few friends with you. No two of you run at the same speed and all of you have a constant, non-zero speed. If you all run in the same direction, will there ever be a moment when you are a set minimum distance apart from the friend nearest to you? Does the same thing happen with each of your friends? In this talk, we shall look at different formulations of the conjecture, and the various approaches that have been taken to tackle the problem. I’ll then explain, in brief, how the conjecture can be viewed geometrically using the concept of polyhedra. Finally, we see some of the important known results, especially those proven using the polyhedral theory approach.
Date & Time : 4th April 2023, 07:00pm
How large can homogeneous sets in Graphs be? - Mihir Neve
Abstract: Consider the party problem - Given 6 people at a party, you can always find 3 people who are all either mutually friends or mutually strangers. Such symmetric relations are best captured in math by the notion of 'Graphs' - a set of nodes, where two nodes are joined by an edge if they are related in some way. No, these aren't the same graphs which we usually see in our calculus courses, but are interesting nevertheless. In a graph, if a subset of nodes are all pairwise connected or pairwise disconnected, like the 3 people above, then it is said to be a homogeneous set. We ask the question: "Given a graph, how large homogeneous sets can it have?" In this talk, we shall look at some basics of graphs, and take a dive into the world of Ramsey Theory and Structural Graph Theory while trying to answer this question. We shall look at the Erdös-Hajnal conjecture, and as we shall see, help comes from a surprising place.
Date & Time: 5th April 2023, 07:00pm
Characterization of Uniform Pro-p Groups - A P Aravintakshan
Abstract: Profinite Groups are a fascinating family of topological groups which can be also defined using the notion of an inverse limit. We start by discuss this equivalence, and using this theory we try to understand various different kinds of profinite groups such as Pro-p Groups, Powerful Pro-p Groups, and Uniform Pro-p Groups. Uniform Pro-p Groups have a well-defined structure to them being torsion free and homoeomorphic to 𝐙pd, where d is the number of topological generators of the group, and has far reaching applications in the field of Number Theory. We then elaborate on the characteristics of various different families of Uniform Pro-p Groups.
Date & Time: 10th April 2023, 07:00pm
A tour of graph transformation problems - Anirudh Rachuri
Abstract: Consider two clusters of computers, that have constant communication with each other. Due to an increase in malware in one of the clusters, you want to cut off communication between the two. However, these computers are performing vital tasks, and you can only switch off a limited number of them. How do you decide which computers to switch off? The field of graph theory is filled with a plethora of such algorithmic problems. Many of these are deemed “too hard to solve”, however, there exist numerous ways to get around this issue. One of the more popular ways is to only allow inputs satisfying a certain property. By restricting our inputs, we can exploit their properties, and perhaps come up with efficient solutions for the problem at hand. In this talk, I will focus on one of the most prevalent and well-studied graph class, namely the class of chordal graphs. We will explore some of its beautiful properties, and understand how one can use them to build efficient algorithms.
Date & Time: 11th April 2023, 07:00pm
Simplicial Complexes on Seifert Surfaces of Knots - A Tale of Two Structures - Ipsa Bezbarua
Abstract: Over the last century, knot theory has developed into an active area of research in topology. This is partly due to its connections with several other areas of science and technology. In the domain of mathematics, low-dimensional topologists look to results and techniques from knot theory to gain a better understanding of 3-manifolds. One of the fundamental structures studied in knot theory is a compact connected surface whose boundary is the knot under consideration, called a Seifert surface. In this talk, we will learn various properties related to two well-known simplicial complexes constructed using the Seifert surfaces of knots - the incompressible complex and the Kakimizu complex (named after Osamu Kakimizu). We will look at some classic properties of the Kakimizu complex, the more popular of the two, like contractibility and local infiniteness. I will then talk about my experiences of trying to extend some of these properties to the seemingly less cool incompressible complex, who (SPOILER ALERT) may be a lot more interesting than we originally thought. I will also recount my experiences of reasoning with (aka, researching) these complexes.
Date & Time: 12th April 2023, 07:00pm
Exact Exponential Algorithms - Ajaykrishnan ES
Abstract: Imagine yourself to be a tiny fish frantically looking for directions to” P.Sherman, 42 Wallaby Way, Sydney”. Since time is of the essence, you and your friend collects various pieces of information which might help find the way. Unfortunately, many of them end up contradicting each other and you are left with no choice but to ignore a few. In the talk we shall see how hopeless our situation can be, but considering how crucial our task is, we will do the best we can to solve the problem. Along the way, we shall learn what P and NP are and why a lot of people believe them to not be equal. We will observe that this conjecture (if true) leaves us with a lot of problems that cannot be solved efficiently and explore a way to cope with this intractability. In the end, we could also briefly look at another problem called Knot-Free Vertex Deletion which is central to my Masters thesis and go over some ideas which were helpful in designing a faster algorithm for the same.
Date & Time: 13th April 2023, 07:00pm
Computational Complexity of the Pilot Assignment problem in Cell-Free Massive MIMO - Shruthi Prusty
Abstract: Wireless networks is an essential technology for enabling flexible communication and connectivity between individuals (or machines) across regions. In addition, it is transforming every sector of the economy (transportation, healthcare, education, etc.), and powerful new technologies (artificial intelligence, internet of things, etc.) are being built upon it. Cellular networks is the technology that 1G to 5G relies on. However, as the number of devices that depend on wireless communication networks continues to grow, each needing a high connection rate and better coverage with minimal interference, this technology will no longer be suitable. For future wireless communications (e.g., 6G), the key technology that has the potential to enhance connectivity and provide better coverage for billions of users is referred to as Cell-Free Massive Multiple-Input Multiple-Output (CF-mMIMO). CF-mMIMO is accompanied by many challenges, one of which is how to efficiently manage limited resources, giving rise to the resource allocation problem in wireless networks. This talk will describe a major problem that hinders resource allocation in wireless networks, namely the Pilot Assignment problem (PA). To design heuristics that can find reasonably good solutions to PA, the problem has been mapped to several well-known graph theory problems which fall in the class of NP-hard problems. However, there is no literature that formally states the PA problem as a mathematical problem, let alone explicitly prove complexity results for it. The aim of this talk is to look at the pilot assignment problem from a theoretical computer science perspective and present a few results on the complexity and approximability of this problem. We discuss our findings in the context of the algorithms already proposed in the literature to tackle this problem.
Date & Time: 14th April 2023, 07:00pm
Formalising mathematics and algorithms using computers - Anand Tadipatri
Abstract: Proof assistants are softwares capable of verifying mathematical arguments down to their foundational details. Those built on the mathematical foundation of type theory can also be used as programming languages, due to a deep correspondence between programs and proofs. The first half of this talk will introduce proof assistants and their various capabilities, with a particular focus on the Lean4 proof assistant. In the second half, I will speak about a recent formalisation work done jointly with my MS advisor Dr. Siddhartha Gadgil, which illustrates how a proof assistant may be used to formalise results that involve both proofs and computations. The talk will conclude with a demonstration of a tool for automatically formalising mathematical statements to Lean code.
Date & Time: 15th April 2023, 07:00pm
The geometry of curves and the shapes of minimal surfaces joining them - Sreedev M
Abstract: Minimal surfaces are zero mean curvature surfaces that appear in nature as idealized soap films. Given two curves one can ask for a minimal surface connecting them. In this talk, we would look at the shapes of such curves and how it controls the image of the minimal surfaces interpolating them. Besides containing lots of creative arguments involving beautiful pictures, this talk would also explore various other different branches such as Fourier analysis, complex analysis, split-complex analysis, functional analysis, and so on. This is a joint work with Prof. Rukmini Dey (ICTS Bangalore).
Date & Time: 16th April 2023, 07:00pm