Speaker
Description
Symbolic regression (SR) yields closed-form models of generalized parton distributions (GPDs), yet a good fit alone does not ensure consistency with known GPD physics. We utilize SR to model the $x$ and $t$ dependence of the flavor isovector combination $H_{u-d}(x,t,\xi)$ at $\xi=0$ under physics constraints. These models are fitted to GPD results from Lattice QCD at the physical pion mass. Candidate constraints are extracted from the GPD literature by a large language model. Each constraint is implemented as a numerical check. Each run repeats a three-step loop for many rounds: run SR from a set of parent expressions, rank the parents and their offspring together by these physics constraint checks and the fitting error, and keep the best ones as the parents of the next SR. Combining the results of many independent runs, we obtain models that fit the lattice data and pass all constraint checks. Beyond the data, our models show a smaller spread than the earlier SR generated models (Dotson et al. 2026) without considering these physics constraints. We further discuss whether these models contain new functional forms or reduce to established parametrizations.