Speaker
Shunsuke Yasunaga
(Institute of Science Tokyo)
Description
Domain-wall fermions provide a good lattice realization of chiral fermions by introducing an additional fifth dimension. At finite fifth-dimensional extent, residual chiral symmetry breaking remains and is characterized by the residual mass. We propose a machine-learning-based parameter-optimization approach to reduce the residual mass while keeping the fifth dimension short. This method aims to emulate the effect of a longer fifth dimension through optimized domain-wall fermion parameters, thereby improving chiral symmetry without significantly increasing the computational cost.
Author
Shunsuke Yasunaga
(Institute of Science Tokyo)
Co-authors
Kenta Yoshimura
(Institute of Science Tokyo)
Akio Tomiya
(TWCU)
Yuki Nagai
(The University of Tokyo)