Speaker
Description
Understanding the excitation spectrum of the nucleon remains a central topic in hadron physics, providing essential insight into the non-perturbative regime of Quantum Chromodynamics (QCD). In this work, we present a data-driven analysis of meson electroproduction in the Δ(1232) and N(1440) resonance regions aimed at improving the extraction of nucleon resonance parameters and transition amplitudes.
The analysis is performed using a Bayesian inference framework combined with machine learning techniques to fit electroproduction cross sections and helicity-dependent observables. The Bayesian approach ensures a consistent treatment of statistical and systematic uncertainties, while machine learning algorithms enhance the stability of multi-parameter optimization in high-dimensional parameter. space
The method is tested on simulated electroproduction data and benchmarked against standard partial wave analysis (PWA) frameworks, including model-based fits used in baryon spectroscopy. The results show a reduction of approximately 20–30% in the uncertainty of the extracted helicity amplitudes and an improvement of 10–20% in the global fit quality (χ²/d.o.f.), while maintaining consistency with established resonance parameter values within one standard deviation.
These results suggest improved stability in the extraction of nucleon resonance parameters in regions with strong channel coupling and overlapping resonances, indicating that Bayesian–machine learning approaches may provide a valuable complementary constraint in future global amplitude analyses.
This study demonstrates the potential of modern statistical learning techniques as a complementary tool to traditional amplitude analysis methods in ongoing and future experiments at Jefferson Lab and MAMI.