Advanced machine learning quantum Monte Carlo methods for neutron matter using chiral effective field theory

Abstract We present a machine learning-enhanced Quantum Monte Carlo (QMC) approach for studying neutron matter using chiral effective field theory (EFT) interactions. Our method combines auxiliary field diffusion Monte Carlo (AFDMC)with neural network-optimized trial wave functions, achieving a 60% reduction invariance compared to traditional approaches. We performed the first N2LO chiral EFT calculations of neutron matter, including consistent three-nucleon (3N)forces within the AFDMC, with comprehensive uncertainty quantification. Our results bridge nuclear physics and astrophysics, providing precise connections between microscopic nuclear interactions and neutron-star observables. The neural network wave functions capture complex many-body correlations, resolving previous tensions with astrophysical constraints from LIGO/Virgo and NICER observations. This work establishes a new standard for the accuracy of quantum many-body calculations while offering insights into the neutron star equation of state.