Machine Learning Prediction of Optical Absorption in GaAs 2D Nanostructure under Hydrostatic Pressure

Abstract

This work focuses on the modeling and prediction of the optical absorption coefficient in GaAs 2D nanostructure subjected to hydrostatic pressure. The database is generated from numerical calculations describing the optical absorption of exciton confined in GaAs 2D nanostructure. The obtained data are exploited to develop predictive models based on machine learning such as Decision Tree and Gradient Boosting techniques, in order to establish an accurate relationship between the incident photon energies and the optical absorption coefficient. The quality of the proposed approaches is examined using standard statistical metrics, considering the mean absolute, squared errors, and root mean squared error, as well as the coefficient of determination. The obtained results demonstrate an excellent agreement between the predicted and numerical values, with very low errors and strong generalization capability. these approaches therefore represent an efficient alternative to conventional numerical methods for the rapid prediction of the optical properties of discoidal quantum dots and offers promising perspectives for the optimization of optoelectronic devices.