A novel energy reconstruction method for wide-field-of-view imaging atmospheric Cherenkov technique experiments

Abstract

The High Altitude Detection of Astronomical Radiation (HADAR) experiment employs an innovative Cherenkov observation technique, boasting an expansive Field-of-View (FOV), and is specifically designed to capture the prompt emissions from Gamma Ray Bursts (GRBs).We propose a novel method for energy reconstruction of Very-high-energy (VHE) γ-rays in the HADAR experiment, based on the Boosted Decision Trees (BDTs) model in machine learning algorithms, referred to as BDTs-Erec. We discuss this energy reconstruction method in detail. A training dataset is generated through Monte Carlo simulation, and the TMVA tool in the ROOT framework is utilized to implement the BDTs model. This model minimizes prediction errors by incrementally adding decision trees and finally constructs 3000 BDTs, thus optimizing the accuracy of energy reconstruction. Performance comparisons are made against the traditional energy reconstruction method based on Look-Up-Tables (denoted as LUTs-Erec), indicating that BDTs-Erec significantly outperforms LUTs-Erec in prediction performance with increasing energy, while it exhibits poorer performance in the low-energy range.