Unbiased Monte Carlo Cluster Updates with Autoregressive Neural Networks

Paper link: arXiv:2105.05650 | Phys. Rev. Research 3, L042024 (2021)

The code requires Python >= 3.8 and JAX >= 0.2.12.

train.py trains a network, and sample_ncus.py generates samples using neural cluster update with symmetries (NCUS). args.py contains all the configurations.

reproduce.sh contains the commands to reproduce the results in Figs. 2~4. In practice, you may run these commands in parallel on multiple GPUs, and set your output directory.