Today

  • Incentives for data sharing in federated learning

    CMSA Room G10 CMSA, 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Han Shao, Harvard CMSA Title: Incentives for data sharing in federated learning Abstract: Federated learning has recently emerged as a powerful approach for enabling collaboration across large populations of learning agents. However, agents may have incentives to defect from the collaboration—that is, to withdraw or contribute less data than expected—due to the […]

  • Asset pricing with heterogeneous agents

    CMSA Room G10 CMSA, 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Sergiy Verstyuk, Harvard CMSA Title: Asset pricing with heterogeneous agents Abstract: This talk will introduce the basics of continuous-time finance, discuss important existing theories and models, as well as present some new asset pricing results in a setting with many heterogeneous investors. (Joint work with Puskar Mondal.)  

  • Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining

    CMSA Room G10 CMSA, 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Samy Jelassi, CMSA Title: Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining Abstract: Reinforcement Learning has become a crucial step in training state-of-the-art language models such as DeepSeek-R1 for solving mathematical problems. In this talk, I will first review the mechanisms of Reinforcement Learning fine-tuning. Then, I will present a systematic end-to-end […]

  • Top-Down Perspectives on Symmetry Theories

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Max Hubner Title: Top-Down Perspectives on Symmetry Theories Abstract: I will review the construction and utility of symmetry theories for string constructed quantum field theories. Symmetry theories are extra-dimensional auxiliary theories separating aspects of a quantum field theory's symmetries from many of its more messy features. For QFTs with extra-dimensional string constructions the […]

  • Sections of fibrations onto curves in characteristic p>0

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Iacopo Brivio Title: Sections of fibrations onto curves in characteristic p>0 Abstract: This talk is based on joint work in progress with Ben Church. Using symplectic geometry, Pieloch showed that every smooth fibration $f\colon X\to \mathbb{P}^1$ of complex projective varieties always admits a section. I will explain how this theorem can be recovered using […]

  • Local Donaldson-Scaduto conjecture

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Saman Habibi Esfahani Title: Local Donaldson-Scaduto conjecture Abstract: This talk is based on joint works with Gora Bera and Yang Li. Motivated by collapsing Calabi-Yau 3-folds and G2-manifolds with Lefschetz K3 fibrations in the adiabatic setting, Donaldson and Scaduto conjectured the existence and uniqueness of a special Lagrangian pair-of-pants in the Calabi-Yau […]

  • The Rozansky-Witten field theory in the functorial TQFT formalism

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Lorenzo Riva Title: The Rozansky-Witten field theory in the functorial TQFT formalism Abstract: This will be a broad talk about the topic of my PhD thesis. We will discuss a particular example of a 3D field theory from physics called Rozansky-Witten which is interesting from both a physical and a mathematical point […]

  • DMFT, Two Point Correlations of Resolvents, and Applications to Machine Learning Theory

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Blake Bordelon Title: DMFT, Two Point Correlations of Resolvents, and Applications to Machine Learning Theory Abstract: Machine learning algorithms evolve the parameters of a model in a high dimensional and disordered loss landscape. To characterize the effects of random initialization of model parameters, randomly sampled training data, and the effect of SGD noise, it […]

  • Analytic Spread of Binomial Edge Ideals

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Stephen Landsittel, CMSA Title: Analytic Spread of Binomial Edge Ideals Abstract: To an ideal J in a polynomial ring R over a field K we associate its analytic spread \ell(J), which is the dimension of the fiber cone F(J) of J. When J is graded and generated in a single degree d, then […]

  • Skein remain the same

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Sunghyuk Park, CMSA Title: Skein remain the same Abstract: The count of holomorphic curves in a Calabi-Yau 3-fold ending on a Lagrangian is famously not deformation invariant, but Ekholm and Shende have shown that it can be made invariant by counting in the skein. Given a 3-manifold M and a branched cover arising from […]

  • Classification of 2D Stabilizer States

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Bowen Yang Title: Classification of 2D Stabilizer States Abstract: I will explain how translation-invariant two-dimensional stabilizer states are completely classified by finite abelian groups with nondegenerate quadratic forms—that is, by abelian anyon theories. The proof uses the algebraic structure of stabilizer codes as modules over Laurent polynomial rings, revealing how their physical […]

  • Optimal learning protocols via statistical physics and control theory

    Common Room, CMSA 20 Garden Street, Cambridge, MA, United States

    Member Seminar Speaker: Francesco Mori, CMSA Title: Optimal learning protocols via statistical physics and control theory Abstract: Behind the impressive performance of modern machine learning lies a toolkit of training tricks, from tuning learning rates to curating training data. These heuristics are powerful but hard to interpret and possibly suboptimal, leaving open the challenge of finding general […]