Jan 23 | First-Principles Deep Learning for Quantum Many-body Physics

[Tatev Vardanyan]

Dear friends,

This Friday, January 23, at 4:30 PM we are hosting a guest seminar on "First-Principles Deep Learning for Quantum Many-body Physics" by Khachatur Nazaryan a PhD student at MIT. 

Abstract

First-principles approaches to quantum many-body physics face a fundamental challenge: the exponential complexity of strongly correlated wave functions. In this talk, I discuss how modern deep learning techniques can be combined with variational Monte Carlo to construct highly expressive, symmetry-respecting wave-function ansatze directly from the microscopic Hamiltonian. I will highlight recent advances in neural quantum states that accurately capture fermionic sign structure, correlations, and emergent phases without relying on effective models or perturbative assumptions. Particular emphasis will be placed on architectural design, physically motivated loss functions, and scalability to large two-dimensional systems. I will conclude by outlining open challenges and future directions at the intersection of machine learning and many-body quantum physics.

Date: Friday, January 23, 2025
Time: 4:30 PM - 6:00 PM
Location: YSU Krisp-AI Lab

Looking forward to seeing you!

Regards,
Tatev