Quantifying theoretical uncertainties in the microscopic nuclear equation of state using GPJaxEmcee

Predicting the nuclear equation of state (EOS) with rigorously quantified uncertainties is a significant effort in ab initio many-body theory. In this talk, we present GPJaxEmcee, a novel machine-learning library based on Gaussian Processes (GP) and Google JAX, for uncertainty quantification (UQ) of nuclear matter properties. This tool allows us to optimize GPs with arbitrary kernel functions to microscopic EOS calculations and quantify and propagate theoretical uncertainties to derived quantities, such as the nuclear symmetry energy. We present results for low-density EOS parameters and the EOS of (beta-equilibrated) neutron star matter based on recent microscopic calculations with two- and three-body forces derived from chiral effective field theory (EFT). We also discuss our work in progress on applying GPJaxEmcee to model the high-density EOS in a data-driven way and studying the mass-radius relation of neutron stars with quantified uncertainties.