Aspects of QCD perturbative evolution

TL;DR

This thesis develops high-precision techniques in perturbative QCD to extract physical parameters from deep inelastic scattering data, combining truncated moments and neural networks for unbiased analysis.

Contribution

It introduces a novel approach combining truncated moments and neural networks for parameter extraction in perturbative QCD, improving accuracy and bias control.

Findings

Effective extraction of the strong coupling constant.

Neural network parametrization preserves experimental error information.

Truncated moments facilitate solving the Altarelli-Parisi equation.

Abstract

This thesis is devoted to the study of some aspects of perturbative QCD, and in particular to the development of high-precision techniques for the extraction of physical parameters such as structure functions, parton distributions, and the strong coupling from the analysis of deep inelastic scattering data. First, we will discuss scaling violations of singlet and nonsinglet truncated moments, and the use of truncated momets to solve the Altarelli-Parisi equation. Then we will suggest an approach based on neural networks to the parametrization and interpolation of experimental data, which retains information on experimental errors and correlations. The method of truncated moments can be combined with the neural network fit to extract various quantities of phenomenological interest in a bias-free way. As an example of such application, we will discuss the determination of the strong coupling constant.