Physics-Informed Neural Networks construct lattice Dirac operators satisfying the Ginsparg-Wilson relation, reproducing overlap fermions to high accuracy and discovering a Fujikawa-type generalized relation via algebraic search.
New fermion discretizations and their applications
3 Pith papers cite this work. Polarity classification is still indexing.
abstract
We review the recent progress in new lattice fermion formulations. We focus on the following three types which have possibility of improving lattice simulations. (1) Flavored-mass fermions are a generalization of Wilson fermions with species-splitting mass terms. In particular, staggered-Wilson fermions initiated by Adams have possibilities of reducing numerical costs in overlap fermions and the influence of taste-breaking in staggered fermions. (2) Central-branch Wilson fermions, in which additive mass renormalization is forbidden by extra axial symmetry, could enable us to perform Wilson-fermion lattice QCD without fine-tuning. (3) Minimally doubled fermions, which reduce the number of species by species-dependent chemical potential terms, realizes a ultra-local chiral fermion at the price of hypercubic symmetry. These setups reveal unknown aspects of lattice fermions, and we obtain a deeper understanding of lattice field theory.
citation-role summary
citation-polarity summary
fields
hep-lat 3verdicts
UNVERDICTED 3roles
background 1polarities
background 1representative citing papers
A kink in a one-link mass term for 3+1D staggered fermions creates a 2+1D domain wall with two-flavor massless Dirac fermions protected by SU(2) and parity, realizing the parity anomaly from the UV lattice Hamiltonian.
Minimal-doubling lattice fermion Hamiltonians yield single-Weyl phases when supplemented by a species-splitting mass term, but one-parameter symmetry-preserving deformations introduce additional Weyl nodes above a critical value.
citing papers explorer
-
Lattice fermion formulation via Physics-Informed Neural Networks: Ginsparg-Wilson relation and Overlap fermions
Physics-Informed Neural Networks construct lattice Dirac operators satisfying the Ginsparg-Wilson relation, reproducing overlap fermions to high accuracy and discovering a Fujikawa-type generalized relation via algebraic search.
-
Taste-splitting mass and edge modes in $3+1$ D staggered fermions
A kink in a one-link mass term for 3+1D staggered fermions creates a 2+1D domain wall with two-flavor massless Dirac fermions protected by SU(2) and parity, realizing the parity anomaly from the UV lattice Hamiltonian.
-
Minimal-doubling and single-Weyl Hamiltonians
Minimal-doubling lattice fermion Hamiltonians yield single-Weyl phases when supplemented by a species-splitting mass term, but one-parameter symmetry-preserving deformations introduce additional Weyl nodes above a critical value.