• First Proof, Second Batch: Results

    Virtual

    https://youtu.be/pAlKAFC5u64 First Proof, Second Batch: Results Date: June 10, 2026 Time: 1:00–2:00 pm ET Location: via Webinar The First Proof Editors present the results of their Second Batch benchmark testing on AI systems. First Proof Editors Mohammed Abouzaid (Stanford) Nikhil Srivastava (UC Berkeley) Rachel Ward (UT Austin) Lauren Williams (Harvard)

  • Introduction to First Proof: A conversation

    Virtual

    https://youtu.be/fNrR4lTiScQ Introduction to First Proof: A conversation Date: June 3, 2026 Time: 1:00–2:00 pm Location: via Webinar Harvard CMSA Director Dan Freed will lead a dialogue with First Proof Editors Mohammed Abouzaid (Stanford), Nikhil Srivastava (UC Berkeley), Rachel Ward (UT Austin), and Lauren Williams (Harvard) to explore the origins and goals of First Proof, sample […]

  • Separation of timescales controls feature learning and overfitting in large neural networks

    Virtual
    Virtual Event

    New Technologies in Mathematics Seminar Speaker: Pierfrancesco Urbani, Universite Paris-Saclay, CNRS, CEA, Institut de physique theorique Title: Separation of timescales controls feature learning and overfitting in large neural networks Abstract: To understand the inductive bias and generalization capabilities of large, overparameterized machine learning models, it is essential to analyze the dynamics of their training algorithms. […]

  • New directions in synthetic data

    Virtual
    Virtual Event

    New Technologies in Mathematics Seminar Speaker: Tatsunori Hashimoto, Stanford Title: New directions in synthetic data Abstract: Synthetic data has been an effective, if boring set of techniques: prompt some language model to restructure your corpus to match some downstream task, with occasionally some distillation. In this talk, we will take a more expansive view of […]

  • Compression Is All You Need: Modeling Mathematics

    Virtual

    Freedman Seminar Speaker: Mike Freedman, Harvard CMSA Title: Compression Is All You Need: Modeling Mathematics Abstract: The talk will exposit a recent eponymous arXiv posting with coauthors Vitaly Aksenov, Eve Bodnia, and Mike Mulligan. The approach is to think like a physicist and model a seemingly complex bit of reality: mathematics, by a simple toy […]

  • Exotic R^4’s are unclassifiable

    Virtual

    Freedman Seminar Speaker: Robert Gompf, UT Austin Title: Exotic R^4's are unclassifiable Abstract: We will use descriptive set theory to show that there is a precise sense in which exotic R^4's are unclassifiable. For other open manifolds, we can reach a much higher level of unclassifiability. This is work in progress with Aristotelis Panagiotopoulos.

  • Dynamic reasoning

    Virtual

    New Technologies in Mathematics Seminar Speaker: Emmanuel Abbé, EPFL, Institute of Mathematics and School of Computer and Communication Sciences & Apple Title: Dynamic reasoning Abstract: In the current AI landscape, reasoning is frequently equated with the generation of intermediate "thinking traces". However, these traces are merely a mechanism, not the ultimate objective. Relying solely on the presence of […]

  • The categorical ‘t Hooft expansion

    Virtual

    Quantum Field Theory and Physical Mathematics Seminar Speaker: Davide Gaiotto (Perimeter Institute) Title: The categorical 't Hooft expansion Abstract: The 't Hooft expansion is the key structure underlying dualities between gauge theories of large matrices and string theories. I will review categorical aspects of the 't Hooft expansion, matching the formal deformation space of certain […]

  • ReLU and Softplus neural nets as zero-sum, turn-based, stopping games

    Virtual

    New Technologies in Mathematics Seminar Speaker: Yiannis Vlassopoulos, Athena Research Center Title: ReLU and Softplus neural nets as zero-sum, turn-based, stopping games Abstract: Neural networks are for the most part treated as black boxes. In an effort to begin elucidating the mathematical structure they encode, we will explain how ReLU neural nets can be interpreted as […]

  • Machine learning tools for mathematical discovery

    Virtual

    New Technologies in Mathematics Seminar Speaker: Adam Zsolt Wagner, Google DeepMind Title: Machine learning tools for mathematical discovery Abstract: I will discuss various ML tools we can use today to try to find interesting constructions to various mathematical problems. I will briefly mention simple reinforcement learning setups and PatternBoost, but the talk will mainly focus […]

  • Tropical-Topological(Tropological) Sigma Models

    Virtual

    Differential Geometry and Physics Seminar Speaker: Andrés Franco Valiente, UC Berkeley Title: Tropical-Topological (Tropological) Sigma Models Abstract: Tropical geometry provides a powerful bridge between complex and combinatorial worlds, allowing certain curve-counting invariants to be computed in a piecewise-linear “tropical” limit. Building on Mikhalkin’s insight that Gromov–Witten invariants can be recovered from tropical curves, this talk revisits Mikhalkin's […]

  • Freedman Seminar: Michael Freedman, CMSA & Bowen Yang, CMSA

    Virtual

    Freedman Seminar Speaker: Michael Freedman, Harvard CMSA Title: Sullivan’s work on Lipschitz structures Part II (but self-contained)   Speaker: Bowen Yang, CMSA Title: Deligne and Sullivan's work on complex bundles with discrete structure group