The Euclidean Spectrum of Kaplan's Lattice Chiral Fermions

Abstract

We consider the (2n+1)-dimensional euclidean Dirac operator with a mass term that looks like a domain wall, recently proposed by Kaplan to describe chiral fermions in $2n$ dimensions. In the continuum case we show that the euclidean spectrum contains {\it no} bound states with non-zero momentum. On the lattice, a bound state spectrum without energy gap exists only if $m$ is fine tuned to some special values, and the dispersion relation does not describe a relativistic fermion. In spite of these peculiarities, the fermionic propagator {\it has} the expected (1/p-slash) pole on the domain wall. But there may be a problem with the phase of the fermionic determinant at the non-perturbative level.