Extended framework for the hybrid Monte Carlo in lattice gauge theory
arXiv:2412.19904 · doi:10.1103/k6dm-lj56
Abstract
We develop an extended framework for the hybrid Monte Carlo (HMC) algorithm in lattice gauge theory by embedding the group into the space of general complex matrices,. Auxiliary directions will be completely factorized in the path integral, and the embedding does not alter the expectation values of the original theory. We perform the molecular dynamics updates by using the matrix elements of as the dynamical variables without group theoretic constraints. The framework enables us to introduce non-separable Hamiltonians for the HMC in lattice gauge theory exactly, whose immediate application includes the Riemannian manifold HMC.
arXiv comment11 pages, 3 figures. v2: Reversibility arguments improved; to appear in Phys. Rev. D
References in corpus (10)
- Trivializing maps, the Wilson flow and the HMC algorithm
- Equivariant flow-based sampling for lattice gauge theory
- Testing and tuning symplectic integrators for Hybrid Monte Carlo algorithm in lattice QCD
- Irreversible Monte Carlo Algorithms for Efficient Sampling
- On the construction of high-order force gradient algorithms for integration of motion in classical and quantum systems
- Choice of Integrator in the Hybrid Monte Carlo Algorithm
- Shadow Hamiltonians, Poisson Brackets, and Gauge Theories
- Testing algorithms for critical slowing down
- Gauge-Fixed Fourier Acceleration
- Self-learning Monte Carlo with equivariant Transformer