Attention-Based Foundation Model for Quantum States

TL;DR

This paper introduces an attention-based neural network architecture capable of learning and predicting quantum ground states across various parameters, enabling the construction of phase diagrams with minimal input data.

Contribution

The authors develop a novel foundation model architecture that accurately predicts quantum states from basis configurations and physical parameters, demonstrating its potential as a universal quantum matter model.

Findings

Accurately predicts ground state wavefunctions across different Hamiltonian parameters.

Successfully constructs phase diagrams for 2D square-lattice $t-V$ model.

Demonstrates the model's universality for various quantum systems.

Abstract

We present an attention-based foundation model architecture for learning and predicting quantum states across Hamiltonian parameters, system sizes, and physical systems. Using only basis configurations and physical parameters as inputs, our trained neural network is able to produce highly accurate ground state wavefunctions. For example, we build the phase diagram for the 2D square-lattice $t-V$ model with $N$ particles, from only 18 parameters $(V/t,N)$. Thus, our architecture provides a basis for building a universal foundation model for quantum matter.