Gauge-equivariant graph neural networks embed non-Abelian local symmetries directly into message passing for lattice gauge theories, enabling learning of nonlocal observables from local operations.
Title resolution pending
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
representative citing papers
A gauge-invariant GNN using Wilson loops as inputs accurately predicts observables and simulates dynamics in Z2 and U(1) lattice gauge models.
A machine learning Hamiltonian model trained on small 95-atom supercells enables linear-scaling structural relaxations and accurate formation energy predictions for oxygen vacancies in larger supercells of amorphous SiO2, with deviations below 50 meV from DFT due to error cancellation.
MD simulations with a machine-learned potential demonstrate that structural fracture energy in silica glass increases with crack velocity below the branching threshold due to both elevated intrinsic surface energy density and nanoscale roughening.
AiiDA-TrainsPot introduces an automated workflow for training neural-network interatomic potentials via calibrated active learning on carbon allotropes and alloy phase transitions.