Improved Constraints on Pion Fragmentation Functions from Simulated Electron-Ion Collider Data

TL;DR

This work quantifies how simulated Electron-Ion Collider (EIC) SIDIS data can sharpen the extraction of pion fragmentation functions (FFs) in a next-to-leading order (NLO) QCD framework using a neural-network parametrization for $D_i^{\pi}(z,Q^2)$. By performing global fits that combine real SIA and SIDIS data with EIC pseudo-data, the authors show substantial reductions in FF uncertainties, especially for the gluon FF and in the high-$z$ region, highlighting the EIC's potential to improve flavor separation and hadronization modeling. The methodology relies on time-like DGLAP evolution in a Zero-Mass VFNS, Monte Carlo uncertainty propagation, and a flexible NN architecture to minimize model bias. The results indicate that EIC measurements will meaningfully advance high-precision FF extractions and inform future theoretical and experimental developments in collider phenomenology.

Abstract

We present a quantitative assessment of the anticipated impact of future Electron-Ion Collider (EIC) measurements on the extraction of parton-to-pion fragmentation functions (FFs). Our analysis combines simulated semi-inclusive deep-inelastic scattering (SIDIS) pseudo-data at EIC energies of 45 GeV and 140 GeV with existing single-inclusive electron-positron annihilation (SIA) and SIDIS experimental data. The pion fragmentation functions are determined at next-to-leading-order (NLO) accuracy using a perturbative QCD framework and a neural network parametrization. Uncertainties are rigorously estimated through Monte Carlo sampling, accounting for both experimental errors and variations in input parton distribution functions. Our results demonstrate that incorporating EIC pseudo-data reduces their uncertainties, especially at medium to large momentum fractions ($z$). This improvement is particularly pronounced for the gluon and selected quark FFs, highlighting the substantial role that EIC measurements will play in achieving high-precision extractions of FFs and informing future experimental and theoretical developments in collider physics.

Figures (7)

• Figure 1: Relative total uncertainties $\delta\sigma/\sigma$ versus $z$ for the $\pi^+$ SIDIS data sets: HERMES, COMPASS, and EIC ($\sqrt{s}$ = 45 and 140 GeV).
• Figure 2: Same as Fig. \ref{['fig:relerr-piplus']}, but for the $\pi^-$ SIDIS data sets.
• Figure 3: Kinematic coverage in the $(z, Q)$ plane for the SIDIS, SIA, and EIC simulated SIDIS pseudo-data included in this analysis.
• Figure 4: Variation of $\chi^2$ per data point as a function of the applied $Q_{\text{min}}$ cut in the SIDIS datasets. The curves show $\chi^2/N_{\text{data}}$ for the total dataset (blue), HERMES (green), COMPASS (purple), and EIC pseudo-data (yellow) at different $Q_{\text{min}}$ thresholds.
• Figure 5: Comparison of the $\chi^2 / N_{\text{data}}$ values for three fits: Global Fit, Global Fit + EIC and MAPFF. The yellow bars represent the SIA datasets, the blue bars denote the SIDIS datasets and the orange bars show the EIC datasets. The topmost green bar summarizes the total $\chi^2 / N_{\text{data}}$ across all datasets combined.