Problem
Research questions and friction points this paper is trying to address.
Flow-based sampling for multimodal distributions
Targets with extended/continuous modes
Application in lattice field theory
Flow-based sampling for multimodal and extended-mode distributions in lattice field theory
🤖 AI Summary
To address the challenge of efficient sampling in lattice field theories exhibiting multiple vacua or continuous degeneracy (e.g., spontaneous symmetry-breaking phases), this work proposes a vacuum-aware normalizing flow framework. We design a symmetry-guided loss function to explicitly encode physical constraints and introduce a vacuum identification mechanism to guide flow training, mitigating modeling failure under multimodal and extended distributions. Furthermore, we develop a hybrid flow–Hamiltonian Monte Carlo (HMC) sampling paradigm that balances sample quality and physical fidelity. Experiments on two-dimensional real and complex scalar field models demonstrate high-precision, low-autocorrelation sampling, significantly accelerating convergence compared to conventional MCMC methods hampered by topological freezing. Our approach provides a scalable, generative sampling strategy for strongly correlated quantum field systems.
📝 Abstract
Recent results have demonstrated that samplers constructed with flow-based generative models are a promising new approach for configuration generation in lattice field theory. In this paper, we present a set of training- and architecture-based methods to construct flow models for targets with multiple separated modes (i.e.~vacua) as well as targets with extended/continuous modes. We demonstrate the application of these methods to modeling two-dimensional real and complex scalar field theories in their symmetry-broken phases. In this context we investigate different flow-based sampling algorithms, including a composite sampling algorithm where flow-based proposals are occasionally augmented by applying updates using traditional algorithms like HMC.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Flow-based generative models
Composite sampling algorithm
Training- and architecture-based methods
🔎 Similar Papers
No similar papers found.
