Title: Machine learning for physics: gauge-equivariant architectures
Abstract: As machine learning algorithms continue to enable and accelerate physics calculations in novel ways, the development of tailored physics-informed machine learning approaches is becoming more sophisticated, impactful, and important. I will give some broad context for this developing area, with a focus on the challenge of exact sampling from known probability distributions as relevant to lattice quantum field theory calculations in particle and nuclear physics. I will discuss in particular flow-based generative models, and describe how guarantees of exactness and the incorporation of complex symmetries (e.g., gauge symmetry) into model architectures can be achieved. I will show the results of proof-of-principle studies that demonstrate that sampling from generative models can be orders of magnitude more efficient than traditional Hamiltonian/hybrid Monte Carlo approaches in this context.
For help, please contact Webmaster.