A Machine Learning Approach to Trapped Many-Fermion Systems

TL;DR

This paper introduces a neural network-based variational approach to efficiently compute the ground states of trapped many-fermion systems, demonstrating rapid convergence and adaptability to different interaction strengths.

Contribution

It presents a novel application of neural network variational ansatzes to many-fermion systems in harmonic traps, improving computational efficiency and flexibility.

Findings

Quick convergence in weakly coupled regimes

Efficient handling of higher couplings by adjusting interaction strength during training

Neural network approach outperforms traditional methods in speed

Abstract

We apply a variational Ansatz based on neural networks to the problem of spin-$1/2$ fermions in a harmonic trap interacting through a short distance potential. We showed that standard machine learning techniques lead to a quick convergence to the ground state, especially in weakly coupled cases. Higher couplings can be handled efficiently by increasing the strength of interactions during "training".