Assessing Symbolic Regression for use in Compton Form Factor Extraction

In this study, we explore the applicability of the nascent technique of symbolic regression (SR) to analyzing the structure of the proton in the context of deeply-virtual Compton scattering (DVCS). Using extant DVCS data, we apply SR to extract the Compton Form Factors (CFFs) that characterize the DVCS process. SR, with its potential for enhanced interpretability, allows for the formulation of analytic expressions that may offer insight into the underlying physics of proton structure. We use test functions of the CFFs and generate pseudodata over the phase space, then systematically compare the regression with the initial test function, assessing accuracy and interpretability. Our preliminary analysis includes error quantification for each test used to evaluated SR's potential as a tool for determining a parsimonious representation of the CFF functions from a purely data-driven methodoloy.