arXiv · 2611.18840
Gapless Nature of Spin Liquid in the $S=1/2$ Kagome Heisenberg Antiferromagnet with Neural-Network Quantum States
Abstract
Whether the spin-$1/2$ kagome Heisenberg antiferromagnet hosts a gapped $\mathbb{Z}_2$ or a gapless $U(1)$ Dirac spin liquid remains a central open question in frustrated magnetism. Here, we address this question by directly computing the triplet gap with a neural-network quantum state that does not presuppose any specific spin-liquid structure. Applying a time-reversal symmetric transformer-based ansatz to clusters of up to $768$ sites, we obtain state-of-the-art variational energies on two-dimensional lattices with periodic boundary conditions and estimate an energy per site of $-0.43873(12)$ in the thermodynamic limit. Finite-size scaling of the magnetic order parameter and triplet excitation gap shows that both extrapolate to zero in the thermodynamic limit. Together, these results provide independent and direct evidence of a gapless spin-liquid ground state, consistent with the $U(1)$ Dirac spin liquid. More broadly, they demonstrate that neural-network quantum states can resolve ground-state properties of frustrated quantum magnets that remain challenging for conventional numerical methods.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Andreas Raikos, Alessandro Sinibaldi, Riccardo Rende, Giuseppe Carleo, Federico Becca, Luciano Loris Viteritti. 2026-11-13. Gapless Nature of Spin Liquid in the $S=1/2$ Kagome Heisenberg Antiferromagnet with Neural-Network Quantum States. https://searcharxiv.com/abs/2611.18840
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.