Lattice Weyl Fermion on a Single Spherical Domain-Wall * Abstract We investigate a single spherical domain-wall embedded in a three-dimensional Euclidean lattice. We employ the Shamir-type domain-wall fermion formulation, where the negative mass region is confined inside the S 2 S^{2} domain-wall, while the positive mass outside is taken to the infinite limit, allowing the exterior region to be neglected effectively. In the absence of a gauge potential, Weyl fermions emerge as edge-localized modes along the S 2 S^{2} domain-wall. When a nontrivial U ⁡ ( 1 ) U(1) gauge potential is present, additional zero modes with opposite chirality appears, localized at the center near the monopole. This centrally localized mode originates from the effective mass contribution of the Wilson term, which induces a domain-wall near the monopole. Preprint number: UT-Komaba/25-2, OU-HET-1264 [a]Shoto Aoki [b]Naoto Kan 1 Introduction In the conventional flat domain-wall fermion framework [ 1 , 2 , 3 ] , two domain walls are required on a five-dimensional lattice to localize massless left- and right-handed fermions. This setup has been explored as a potential foundation for constructing chiral gauge theories on the lattice. However, decoupling unwanted modes from gauge fields presents significant challenges [ 4 ] . For instance, the cobordism invariance of instanton backgrounds ensures that both domain-walls host an equal number of zero modes, making it difficult to isolate purely chiral dynamics in the low-energy theory. Recently, symmetric mass generation mechanism is studied to gap out these unwanted modes without breaking chiral symmetry [ 5 ] . However, this method relies on highly nonperturbative dynamics, which remain challenging to fully understand and implement. When the domain-wall is allowed to have nontrivial curvature, it does not have to appear in left and right pairs. In particular, when the domain-wall is a closed manifold like a sphere, it can be embedded into a one-dimension higher flat space without any other domain-walls. Nash embedding theorems [ 6 ] guarantee that every curved manifold can be isometrically embedded into a finite-dimensional Euclidean spaces. Moreover, the metric and spin connection are uniquely determined up to the local Lorentz transformation by the embedding. Two of the authors have embodied this concept in lattice gauge theory [ 7 , 8 ] . They have confirmed that the spherical domain-wall fermion systems on a periodic flat lattice hold edge-localized modes, which feel gravity through the induced spin connection. The obtained eigenvalue spectrum is consistent with the continuum theory prediction. Moreover, automatic recovery of the rotational invariance is also observed. Recently, a similar system was discussed in Refs. [ 9 , 10 , 11 , 12 ] . Since the single spherical domain-wall fermion can have an isolated left- or right-handed edge fermion, they argued that chiral gauge theory can be formulated on the lattice. Although this proposal is appealing in the free fermion case, it is nontrivial to reproduce the Weyl fermion anomaly and its cancellation. In particular, it was pointed out in Ref. [ 13 ] that an axial current, which should be anomalous, can become conserved, which contradicts with the continuum theory. In this work, we consider a two-dimensional spherical domain-wall embedded in a three-dimensional square lattice. The result was already published in Ref. [ 14 ] . As is expected, the free fermion Dirac operator spectrum is consistent with that of a single Weyl fermion. The gravitational effect through the induced spin connection is confirmed in the gap of the eigenvalues. However, when a topologically nontrivial U(1) gauge background is given, we find an additional zero mode localized at the center of the sphere, where the gauge field becomes singular. We identify this additional zero mode as an edge-localized state on a small but finite domain-wall near the center, which is dynamically created by the Wilson term 1 1 1 In Ref. [ 15 ] , we discussed that the same mechanism can explain the Witten effect, in which the monopole background captures an electron zero mode and becomes a dyon. . 2 Without gauge fields In this paper, we consider Dirac operators \displaystyle D= \displaystyle\sum_{i}\sigma^{i}\quantity(\pdv{}{x^i}-iA_i)+m(r) (1) on a three-dimensional Euclidean space \mathbb{R}^{3} . Here, \sigma_{i} is a Pauli matrix, A_{i} is a U(1) gauge field in the i -th direction, which will be considered in a later section, and m(r) is a mass term. To put an S^{2} domain-wall with the radius r_{0} , the form of m(r) is \displaystyle m(r)=\begin{cases}-m&(r\leq r_{0})\\ +M&(r>r_{0})\end{cases}, (2) where m and M are positive constants. We denote the negative mass region as \mathbb{D}^{3} . Taking the limit of M\to+\infty . It is equivalent to ignoring the outside of \mathbb{D}^{3} and imposing a boundary condition \displaystyle\sigma_{r}\psi(r_{0})=+\psi(r_{0}), (3) where the chirality operator \sigma_{r} is given by \displaystyle\sigma_{r}=\sum_{i}\sigma^{i}x^{i}/r. (4) Then, we obtain a Shamir domain-wall fermion system. If we employ the Hermitian conjugate operator D^{\dagger} , we obtain a boundary condition with \sigma_{r}=-1 . We first address the free fermion case with \textbf{A}=0 . Since the system is rotational symmetric, the Dirac operator D commutes with the total angular momentum operators \displaystyle J_{a}=L_{a}+\frac{1}{2}\sigma_{a}, (5) where L_{a} is the orbital contribution \displaystyle L_{a}=-i\epsilon_{abc}x^{b}\partialderivative{x^c}, (6) and \frac{1}{2}\sigma_{a} is the spin part. Let j be the total angular momentum quantum number. Then, we have \displaystyle j=\frac{1}{2},~\frac{3}{2},~\frac{5}{2},\cdots. (7) Respecting the rotational symmetry, we rewrite D as \displaystyle D=\sigma_{r}\quantity(\pdv{}{r} +\frac{1}{r} -\frac{iD^{S^2}}{r} )-m, (8) where D^{S^{2}} is an effective Dirac operator given by \displaystyle iD^{S^{2}}=\sigma^{a}\quantity(L_a )+1=\sigma^{a}J_{a}-\frac{1}{2}. (9) D^{S^{2}} and \sigma_{r} commute with the total angular momentum operator J_{a} , but anti-commute with each other. Since the square of D^{S^{2}} is give by \displaystyle(iD^{S^{2}})^{2}=\mathbf{J}^{2}+\frac{1}{4}, (10) the eigenvalues of D^{S^{2}} are not zero. We find the spinors \chi_{j,j_{3},\pm} which satisfy \displaystyle\mathbf{J}^{2}\chi_{j,j_{3},\pm} \displaystyle=j(j+1)\chi_{j,j_{3},\pm}, (11) \displaystyle J_{3}\chi_{j,j_{3},\pm} \displaystyle=j_{3}\chi_{j,j_{3},\pm}\quad(j_{3}=-j,-j+1,\cdots,j-1,j), (12) \displaystyle iD^{S^{2}}\chi_{j,j_{3},\pm} \displaystyle=\pm\nu\chi_{j,j_{3},\pm}\quad(\nu=j+\frac{1}{2}), (13) \displaystyle\sigma_{r}\chi_{j,j_{3},\pm} \displaystyle=\chi_{j,j_{3},\mp}. (14) The eigenstate of D^{\dagger}D\psi=E^{2}\psi and DD^{\dagger}\psi=E^{2}\psi are expanded by \chi_{j,j_{3},\pm} . Assuming that E is positive, we obtain the edge localized modes as \displaystyle\psi_{j,j_{3}}^{D^{\dagger}D} \displaystyle=\frac{1}{\sqrt{r}}\quantity( \sqrt{m-E} I_{\nu-1/2}(\sqrt{m^2-E^2}r) \chi_{j,j_3,+} + \sqrt{m+E} I_{\nu+1/2}(\sqrt{m^2-E^2}r)\chi_{j,j_3,-} ) (15) \displaystyle\psi_{j,j_{3}}^{DD^{\dagger}} \displaystyle=\frac{1}{\sqrt{r}}\quantity( \sqrt{m-E} I_{\nu-1/2}(\sqrt{m^2-E^2}r) \chi_{j,j_3,+} - \sqrt{m+E} I_{\nu+1/2}(\sqrt{m^2-E^2}r)\chi_{j,j_3,-} ) (16) for E