# Characterization of photoexcited states in the half-filled one-dimensional extended Hubbard model assisted by machine learning

## TL;DR

This paper demonstrates how machine learning can effectively characterize photoexcited states in the half-filled one-dimensional extended Hubbard model, revealing enhanced bond-spin-density wave order through neural network analysis.

## Contribution

It introduces a supervised machine learning approach using entanglement spectra to identify and analyze photoexcited states in the 1DEHM, highlighting the enhancement of BSDW order.

## Key findings

- ML predicts enhancement of BSDW order in photoexcited states
- Neural network trained on entanglement spectra successfully characterizes quantum states
- Correlation functions confirm ML predictions of BSDW enhancement

## Abstract

Photoinduced nonequilibrium states can provide new insight into dynamical properties of strongly correlated electron systems. One of the typical and extensively studied systems is the half-filled one-dimensional extended Hubbard model (1DEHM). Here, we propose that the supervised machine learning (ML) can provide useful information for characterizing photoexcited states in 1DEHM. Using entanglement spectra as a training dataset, we construct neural network. Judging from the trained network, we find that bond-spin-density wave (BSDW) order can be enhanced in photoexcited states if the frequency of a driving pulse nearly resonates with gap. We separately calculate the time evolution of local and non-local order parameters and confirm that the correlation functions of BSDW are actually enhanced by photoexcitation as predicted by ML. The successful prediction of BSDW demonstrates the advantage of ML to assist characterizing photoexcited quantum states.