Reproduce “Simulating Correlated Electrons with Symmetry-Enforced Normalizing Flows” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dominic Schuh, Janik Kreit, Evan Berkowitz, Lena Funcke, Thomas Luu, Kim A. Nicoli, Marcel Rodekamp
This research applies symmetry-enforced normalizing flows to accurately simulate the fermionic Hubbard model, demonstrating that this machine learning approach resolves long-standing ergodicity issues and outperforms traditional Hybrid Monte Carlo methods in sampling accuracy and computational efficiency, including for a four-site lattice.
Key insights
- Normalizing Flows (NFs) can effectively learn and sample from complex, multimodal probability distributions, such as the Boltzmann distribution of the fermionic Hubbard model.
- Explicitly enforcing physical symmetries within generative models, particularly through the Symmetry-Enforcing Stochastic Modulation (SESaMo) framework, is crucial for accurately capturing the target distribution and preventing issues like mode collapse.
- Generating independent and identically distributed (i.i.d.) samples via NFs offers a substantial advantage over Markov chain Monte Carlo methods, leading to significantly reduced autocorrelation times and increased effective sample size.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.