Solving Spinful Fermi Systems with a Self-Attention Ansatz
Date: Thu, November 13, 2025
Time: 1:30pm - 2:30pm
Location: Holmes Hall 411
Speaker: Alexander Avdoshkin, Postdoc (Physics), Massachusetts Institute of Technology
(hosted by Prof. Igor Molybog, College of Engineering, ECE Department)
ECE Graduate Students: This will count towards your seminar credit.
Abstract
Recently, neural-network-based variational methods have shown success in groundstate search for problems in quantum chemistry and condensed-matter physics. This approach relies on variational Monte Carlo optimization with a real-space representation of the many-electron wavefunction built from a neural network. In our work (arXiv:2510.18621), we extend the approach to Fermi systems that contain spin or other discrete degrees of freedom in addition to continuous positions. Our solution combines Markov chain Monte Carlo sampling for energy estimation— adapted to cover the enlarged configuration space—with a transformer-based wavefunction to represent fermionic states. We validate the method on a range of two-dimensional material problems: a two-dimensional electron gas with Rashba spin–orbit coupling, a noncollinear spin texture, and a quantum antiferromagnet in a honeycomb moiré potential. This development enables the study of various correlated phases in quantum materials with layer or spin degrees of freedom.
Biography
Alex Avdoshkin is a computational and theoretical physicist working on correlated materials and quantum dynamics. He earned his Ph.D. in physics from the University of California, Berkeley, and is now a postdoctoral associate at MIT. His research spans deep-learning approaches to electronic structure, the foundations of quantum statistical mechanics, and geometric methods for wavefunction effects in semiconductors.