A utilização de Redes Neurais Artificiais no estudo da Constante de Acoplamento da Cromodinâmica Quântica

Abstract

Computational algorithms and their advancements have brought increased comfort to society. One standout algorithm is machine learning, utilized not only in everyday tasks such as search mechanisms and text translation but also in the analysis of soy crops. Although its use in the field of particle physics and high-energy physics is currently limited, it holds promise. Quantum Chromodynamics is the theory of the Standard Model that explores the interaction of color-carrying particles: quarks and gluons. Despite being well-established in the perturbative regime, there is still much to study in the non-perturbative regime. From this perspective, the present work explores the application of machine learning algorithms, specifically neural networks, to investigate the coupling constant in the non-perturbative regime. Different network architectures, including Feed Forward and Long Short-Term Memory (LSTM), were studied using activation functions such as ReLU, Sigmoid, Hyperbolic Tangent, and ELU.