Calendar
- 27September 27, 2020No events
- 28September 28, 2020
CMSA Math Science Literature Lecture Series
TITLE: From Deep Learning to Deep Understanding
ABSTRACT: In this talk I will discuss a couple of research directions for robust AI beyond deep neural networks. The first is the need to understand what we are learning, by shifting the focus from targeting effects to understanding causes. The second is the need for a hybrid neural/symbolic approach that leverages both commonsense knowledge and massive amount of data. Specifically, as an example, I will present some latest work at Microsoft Research on building a pre-trained grounded text generator for task-oriented dialog. It is a hybrid architecture that employs a large-scale Transformer-based deep learning model, and symbol manipulation modules such as business databases, knowledge graphs and commonsense rules. Unlike GPT or similar language models learnt from data, it is a multi-turn decision making system which takes user input, updates the belief state, retrieved from the database via symbolic reasoning, and decides how to complete the task with grounded response.
Written articles will accompany each lecture in this series and be available as part of the publication “History and Literature of Mathematical Science.”
For more information, please visit the event page.
Register here to attend.
CMSA Mathematical Physics Seminar: Wilson loops as matrix product states
In this talk, I will discuss a reformulation of the Wilson loop in large N gauge theories in terms of matrix product states. The construction is motivated by the analysis of supersymmetric Wilson loops in the maximally super Yang–Mills theory in four dimensions, but can be applied to any other large N gauge theories and matrix models, although less effective. For the maximally super Yang–Mills theory, one can further perform the computation exactly as a function of ‘t Hooft coupling by combining our formulation with the relation to integrable spin chains.
CMSA Math Science Literature Lecture Series
TITLE: A personal story of the 4D Poincare conjecture.
ABSTRACT: The proof of PC4 involved the convergence of several historical streams. To get started: high dimensional manifold topology (Smale), a new idea on how to study 4-manifolds (Casson), wild “Texas” topology (Bing). Once inside the proof: there are three submodules: Casson towers come to life (in the sense of reproduction), a very intricate explicit shrinking argument (provided by Edwards), and the “blind fold” shrinking argument (which in retrospect is in the linage of Brown’s proof of the Schoenflies theorem). Beyond those mentioned: Kirby, Cannon, Ancel, Quinn, and Starbird helped me understand my proof. I will discuss the main points and how they fit together.
Talk Chair: Peter Kronheimer
Written articles will accompany each lecture in this series and be available as part of the publication “History and Literature of Mathematical Science.”
For more information, please visit the event page.
Register here to attend.
- 29September 29, 2020
K-theory of Operator Algebras, Orbifolds, and Conformal Field Theory
Subfactors and K-theory are useful mechanisms for understanding modular tensor categories and conformal field theories CFT. As part of this, one issue is to try and construct or reconstruct a conformal field theory as the representation theory of a conformal net of algebras, or as a vertex operator algebra from a given abstractly presented modular tensor category. Freed, Hopkins and Teleman realized the chiral Verlinde rings of WZW models as twisted equivariant K-theory. I will describe work which has led to represent the full CFT and modular invariant partition function K-theoretically and descriptions of Verlinde rings as Hilbert modules over an operator algebra, and higher equivariant twists with bundles beyond compact operators. Orbifold models play an important role and orbifolds of Tambara-Yamagami systems are relevant to understanding the double of the Haagerup as a conformal field theory. This is joint work with Andreas Aaserud, Terry Gannon and Ulrich Pennig.
Zoom: https://harvard.zoom.us/j/779283357?pwd=MitXVm1pYUlJVzZqT3lwV2pCT1ZUQT09
CMSA Computer Science for Mathematicians: Testing Positive Semi-Definiteness via Random Submatrices
Given an n x n matrix A with bounded entries, we study the problem of testing whether A is positive semi-definite (PSD) via only a small number of queries to the entries of A. While in general one must read the entire matrix A to determine if it is PSD, we demonstrate that testing whether A is PSD or “far” from PSD (under some norm) is indeed possible with query complexity which only depends on the distance to the PSD cone. This setup is commonly referred to as the property testing framework. We consider two natural norms of n x n matrices: the spectral norm and the Frobenius (Euclidean) norm. We give a query-optimal algorithm for the former, and present new upper and lower bounds for the latter.
Both of these algorithms have a very simple structure: they randomly sample a collection of principal submatrices and check whether these submatrices are PSD. Thus, our problem can phrased purely as a question in random matrix theory: given a (entry-wise bounded) matrix A which is at distance D (in some norm) from the PSD cone, what is the probability that a random k x k principal submatrix is not PSD? For the spectral norm, this problem can be tightly answered by classical arguments (e.g. scalar valued concentration), however the case of the Euclidean norm appears to be more involved, and will require matrix concentration based arguments.
In this talk, we will discuss the analysis of eigenvalues of random submatrices which lead to these algorithms, and touch on several open questions related to spectral concentration of random submatrices.
Joint work with Ainesh Bakshi and Nadiia Chepurko.
Talk based on the paper https://arxiv.org/abs/2005.06441, to appear in FOCS 2020. Algebraic Braids and Transcendental Retractions
If a complex, integral, projective curve C has only planar singularities, then its Jacobian admits a natural compactification with interesting topology. Work of Oblomkov, Shende, and others suggests the existence of a variety, stratified by algebraic tori and defined solely in terms of the topology of C, that retracts transcendentally onto this compactified Jacobian. We expect this retraction to be a new type of nonabelian Hodge correspondence: in particular, at the level of cohomology, it should map a (halved) weight filtration onto a filtration defined via perverse sheaves. I will construct a candidate for the larger variety using the combinatorics of braids and flag varieties, related to but ultimately different from a construction of Shende-Treumann-Zaslow. I will present evidence that the entire story is the SL_n case of a recipe that works for any semisimple group G, and that in a precise sense, these retractions should respect the isomorphism between the unipotent locus of G and the nilpotent locus of Lie(G). The key ingredient is a map from elements of the loop Lie algebra to conjugacy classes in a generalized braid group. The latter are the “algebraic braids” of the title.
Zoom: https://harvard.zoom.us/j/91794282895?pwd=VFZxRWdDQ0VNT0hsVTllR0JCQytoZz09
SYZ mirror symmetry for del Pezzo surfaces and rational elliptic surfaces
I will discuss some aspects of SYZ mirror symmetry for pairs (X,D) where X is a del Pezzo surface or a rational elliptic surface and D is an anti-canonical divisor which is either smooth or a wheel of rational curves. In particular I will explain the existence of special Lagrangian fibrations and mirror symmetry for (suitably interpreted) Hodge numbers. If time permits, I will describe a proof of SYZ mirror symmetry for del Pezzo surfaces. This is joint work with A. Jacob and Y.-S. Lin.
Zoom: https://harvard.zoom.us/j/96709211410?pwd=SHJyUUc4NzU5Y1d0N2FKVzIwcmEzdz09
CMSA Math Science Literature Lecture Series
TITLE: Hodge structures and the topology of algebraic varieties
ABSTRACT: We review the major progress made since the 50’s in our understanding of the topology of complex algebraic varieties. Most of the results we will discuss rely on Hodge theory, which has some analytic aspects giving the Hodge and Lefschetz decompositions, and the Hodge-Riemann relations. We will see that a crucial ingredient, the existence of a polarization, is missing in the general Kaehler context.
We will also discuss some results and problems related to algebraic cycles and motives.
Talk chair: Joe Harris
Written articles will accompany each lecture in this series and be available as part of the publication “History and Literature of Mathematical Science.”
For more information, please visit the event page.
Register here to attend.
CMSA Quantum Matter in Mathematics and Physics: Gravitational Constrained Instantons and Random Matrix Theory
We discover a wide range of new nonperturbative effects in quantum gravity, namely moduli spaces of constrained instantons of the Einstein-Hilbert action. We find these instantons in all spacetime dimensions, for AdS and dS. Many can be written in closed form and are quadratically stable. In 3D AdS, where the full gravitational path integral is more tractable, we study constrained instantons corresponding to Euclidean wormholes. We show that they encode the energy level statistics of microstates of BTZ black holes, which precisely agrees with a quantitative prediction from random matrix theory.
CMSA Math Science Literature Lecture Series
TITLE: Immersions of manifolds and homotopy theory
ABSTRACT: The interface between the study of the topology of differentiable manifolds and algebraic topology has been one of the richest areas of work in topology since the 1950’s. In this talk I will focus on one aspect of that interface: the problem of studying embeddings and immersions of manifolds using homotopy theoretic techniques. I will discuss the history of this problem, going back to the pioneering work of Whitney, Thom, Pontrjagin, Wu, Smale, Hirsch, and others. I will discuss the historical applications of this homotopy theoretic perspective, going back to Smale’s eversion of the 2-sphere in 3-space. I will then focus on the problems of finding the smallest dimension Euclidean space into which every n-manifold embeds or immerses. The embedding question is still very much unsolved, and the immersion question was solved in the 1980’s. I will discuss the homotopy theoretic techniques involved in the solution of this problem, and contributions in the 60’s, 70’s and 80’s of Massey, Brown, Peterson, and myself. I will also discuss questions regarding the best embedding and immersion dimensions of specific manifolds, such has projective spaces. Finally, I will end by discussing more modern approaches to studying spaces of embeddings due to Goodwillie, Weiss, and others. This talk will be geared toward a general mathematical audience.
Talk chair: Michael Hopkins
Written articles will accompany each lecture in this series and be available as part of the publication “History and Literature of Mathematical Science.”
For more information, please visit the event page.
Register here to attend.
Pointwise Bound for $\ell$-torsion of Class Groups
$\ell$-torsion conjecture states that $\ell$-torsion of the class group $|\text{Cl}_K[\ell]|$ for every number field $K$ is bounded by $\text{Disc}(K)^{\epsilon}$. It follows from a classical result of Brauer-Siegel, or even earlier result of Minkowski that the class number $|\text{Cl}_K|$ of a number field $K$ are always bounded by $\text{Disc}(K)^{1/2+\epsilon}
Zoom: https://harvard.zoom.us/j/96767001802
Password: The order of the permutation group on 9 elements.
The Combinatorics of Rhombic Polygon Tilings
The geometry of rhombic tilings and tessellations like the Penrose tiling have captivated mathematicians and artists alike. Hidden in the geometry of certain rhombic tilings of certain polygons, though, is an unexpected combinatorial structure that not only lends itself to some combinatorial objects, but also is often rather useful for their enumeration. In this talk, we will highlight the connection between one type of rhombic tiling and the world of plane partitions, monotone discrete functions, and stacks of cubes; and another type of rhombic tiling to the world of permutations, Coxeter groups, and reduced words.
Come learn some neat mathematics connecting permutations, polygons, geometry, and groups.
Zoom link is posted here:
https://calendar.college.harvard.edu/event/math_table
CMSA Quantum Matter in Mathematics and Physics: Intrinsic sign problems in topological matter
The infamous sign problem leads to an exponential complexity in Monte Carlo simulations of generic many-body quantum systems. Nevertheless, many phases of matter are known to admit a sign-problem-free representative, allowing efficient simulations on classical computers. Motivated by long standing open problems in many-body physics, as well as fundamental questions in quantum complexity, the possibility of intrinsic sign problems, where a phase of matter admits no sign-problem-free representative, was recently raised but remains largely unexplored. I will describe results establishing the existence of intrinsic sign problems in a broad class of topologically ordered phases in 2+1 dimensions. Within this class, these results exclude the possibility of ‘stoquastic’ Hamiltonians for bosons, and of sign-problem-free determinantal Monte Carlo algorithms for fermions. The talk is based on arxiv: 2005.05566 and 2005.05343.
CMSA Math Science Literature Lecture Series
TITLE: Birational geometry
ABSTRACT: About main achievements in birational geometry during the last fifty years.
Written articles will accompany each lecture in this series and be available as part of the publication “History and Literature of Mathematical Science.”
For more information, please visit the event page.