Academia Sinica Logo

Global QCD analyses infer the quark and gluon structure of hadrons from measurements that probe partonic correlation functions through convolutions, QCD evolution, and incomplete kinematic coverage. I will discuss recent work on flexible, high-dimensional reconstruction methods designed to reduce parametrization bias and identify directions that remain weakly constrained by data. Examples will include finite-element-based methods for generalized parton distributions and pixel-based Bayesian imaging of transverse-momentum-dependent distributions using generative AI. I will also describe ongoing extensions to collinear parton distributions and fragmentation functions within the JAM global-analysis framework, as well as prospects for combining experimental and lattice-QCD information. These developments provide new tools for extracting the multidimensional structure of hadrons and maximizing the physics potential of the future Electron–Ion Collider.