Assessing the impact of the electron ion collider in China on Deeply Virtual Compton Scattering

TL;DR

This work assesses how DVCS measurements at the planned Electron-Ion collider in China (EicC) can sharpen the extraction of Compton Form Factors (CFFs) by employing a flexible neural-network parametrization and replica-based uncertainty quantification, fitted to global DVCS data with Gepard. By simulating realistic EicC detector performance and polarized asymmetries, the study shows substantial reductions in CFF uncertainties across all regions, with the largest gains in the sea-quark domain, enabling improved GPD tomography and insights into QCD-scale evolution. The results validate the NN-based approach for multi-observable global fits and highlight the significant scientific impact of EicC on nucleon structure studies, while pointing to the need for further theoretical refinements and higher-order analyses.

Abstract

We assess the impact of future measurements of deeply virtual Compton scattering (DVCS) off protons using the planned detector at the Electron-Ion Collider in China (EicC), proposed as an upgrade to the High Intensity heavy-ion Accelerator Facility (HIAF). We develop a neural-network architecture to flexibly parameterize the Compton Form Factors (CFFs), extrapolate reliably into unmeasured kinematic regions, and provide robust uncertainty estimates through the replica method. The framework is fitted to the available worldwide DVCS data using the Gepard software. We find a significant reduction in the uncertainties of all CFFs after incorporating pseudo-data from single and double polarization asymmetries at the EicC, with particularly strong improvements in the sea-quark region.

Figures (5)

• Figure 1: The deeply virtual Compton scattering (left) and the accompanying Bethe-Heitler process (middle and right). The $Q^2 = -q^2$ is the negative four-momentum squared of the virtual photon. Other kinematic variables are: $x$, the average longitudinal momentum fraction of the active quark; $\xi$, the half longitudinal momentum fraction transferred to the nucleon; and $t$, the squared four-momentum transferred to the nucleon. The $F_{1,2}$ are elastic form factors in the Dirac-Pauli representation.
• Figure 2: A single neural network parameterizes the imaginary part of one of the CFFs, while a similar net represents the real part. The input values ($\xi$, $t$, $Q^2$) after linearization and normalization are fed to three input neurons, and then propagated through the four hidden layers with ninety neurons each, so architecture is (3 $\to 90 \to 90 \to 90 \to 90 \to$ 1).
• Figure 3: The extraction of CFFs versus skewness $\xi$ at $Q^2 =$ 4.0 GeV$^2$ and -$t =$ 0.4 GeV$^2$. The blue and red error bands are uncertainties before and after including the pseudodata of asymmetries at EicC.
• Figure 4: The extraction of CFFs versus $-t$ at $Q^2 =$ 4.0 GeV$^2$ and $\xi =$ 0.01 GeV$^2$.
• Figure 5: The extraction of CFFs versus $Q^2$ at $\xi =$ 0.01 and -$t =$ 0.4 GeV$^2$.