Phillips-Inspired Machine Learning for Band Gap and Exciton Binding Energy Prediction

Abstract

In this work, inspired by Phillips's ionicity theory in solid-state physics, we directly sort out the critical factors of the band gap's feature correlations in the machine learning architected with the Lasso algorithm. Even based on a small 2D materials data set, we can fundamentally approach an accurate and rational model about the band gap and exciton binding energy with robust transferability to other databases. Our machine learning outputs can reveal the exact physics pictures behind the predicted quantity as well as the "secondary understanding" of the correlation between the approximated physics models in exciton. This work stresses the significant value of physics endorsement on the machine learning (ML) algorithm and provides a symbolic regression solution for the "few-shot" training scheme for ML technology in materials science. Moreover, physics-inspired secondary understanding could be an essential supplement for ML in scientific research fields.

Recent progress on discovery and properties prediction of energy materials: Simple machine learning meets complex quantum chemistry

“…5c). Inspired by Phillips's ionicity theory of solid-state physics, Liang et al [108] demonstrated a least absolute shrinkage and selection operator to predict the band energy and band gap. This method shows high accuracy and can be further extended to other databases ( Fig.…”

QM-symex, update of the QM-sym database with excited state information for 173 kilo molecules

“…Another vital application of accelerated development is artificial intelligence. Checking the excited-state properties of each molecule experimentally is time and energy consuming, and thus the use of quantum mechanical computation (QM) or machine learning algorithm (ML) is necessary in enabling scholars to study the structure and properties of material molecules more efficiently [11][12][13][14][15][16][17] and to compile large databases. However, quantum mechanical computation and machine learning algorithms, especially neural networks, are able to come up with relatively good performance only if large databases are utilized in training and debugging models.…”

Representing individual electronic states for machine learning GW band structures of 2D materials

“…Rajan et al used different regressions methods to predict bandgaps of MXene crystals using a training set of 76 G 0 W 0 bandgaps and a representation encoding atomic and structural properties 17 . Liang et al used a representation based on atomic ionicity descriptors to predict GW bandgaps of a set of 2D semiconductors 18 . In all these previous studies, the ML model was trained to predict the size of the bandgap rather than the full k -resolved band structure.…”