Can a gauge-invariant measure for lattice chiral fermions be constructed nonperturbatively?

The missing piece is the measure phase

Lattice QCD works because the fermion determinant can be defined once the gauge field is fixed. Chiral gauge theories add a harder problem: the subspace of fermion fields being integrated over itself depends on the gauge field. Overlap fermions make this explicit. The modified chirality operator built from the overlap operator defines projectors that depend on the gauge links, so the chirality condition selects a different fermion subspace at every gauge configuration.

What the condition does not fix is the phase of the integration measure on that subspace. Choosing that phase consistently as the gauge field varies is the obstruction. If the phase is chosen badly, the lattice theory loses gauge invariance or fails to depend smoothly on the gauge field. The paper's target is therefore not the fermion action, which overlap fermions already provide, but the measure that multiplies it.

Admissibility keeps the overlap operator well defined

The overlap operator is built from the sign of a Hermitian Wilson operator. That sign is only a smooth function of the gauge links if the spectrum of the Hermitian Wilson operator stays away from zero. The paper restricts to admissible gauge fields, meaning plaquettes remain sufficiently close to the identity, which keeps the relevant eigenvalues separated from zero and makes the overlap operator a smooth function of the links.

This restriction is a working condition, not a cosmetic one. Every locality and smoothness statement in the construction is stated on admissible fields. Individual links and transports around periodic directions need not be close to the identity; the bound is placed on plaquette deviations. The paper also treats all four possible global structures of the Standard Model gauge group, since perturbation theory cannot distinguish them.

Comparing projectors across two lattice spacings

The central device is a comparison between chiral projectors at successive lattice spacings. The paper constructs a smooth connection from the original lattice links, samples it at a finer auxiliary spacing, and compares the coarse and fine overlap operators in a common color frame. The mismatch between the two defines a current, and that current is what enters the measure term.

The comparison is designed so that the result is exponentially local: contributions from distant regions decay exponentially with distance, with a rate that survives the physical normalization. The paper separates contributions by their range, shows that color factors cancel where they must, and bounds the sum over sites. The current's first three derivatives with respect to the original links vanish at the vacuum, and a full-inversion argument removes even powers of the lattice spacing, leaving the exact current suppressed more strongly than individual polynomial terms.

What Lüscher reconstruction actually requires

Lüscher's reconstruction builds a smooth, gauge-invariant fermion measure from a current satisfying a set of conditions. The paper organizes its proof around those conditions rather than around a single formula. Smoothness and locality are needed so the measure responds sensibly to small changes of the gauge field. Gauge invariance is needed so the measure transforms with the fermion fields, with the anomaly appearing in the covariant divergence equation rather than as a violation.

Two integrability conditions are also required. The differential condition ensures that the current is the derivative of a measure term, so the phase is path-independent for infinitesimal deformations. The full integrability condition ensures that the phase around closed loops is consistent, so the measure can be defined globally rather than only locally. The paper establishes these properties on admissible fields, including the loop phases and the behavior under rotations, reflections, and charge conjugation.

Global phases and disconnected gauge sectors

Local consistency is not enough for a measure defined on the whole space of admissible fields. The paper handles global integrability with an auxiliary connection and conjugate fermion pairs, transporting the determinant phase on a cylinder and comparing it with an auxiliary boundary phase in the refinement limit. This is what turns a locally defined current into a reconstructed measure.

The reconstruction still leaves a phase freedom for disconnected components of the admissible field space. The paper notes that fixing this remaining freedom through physical considerations is a direction for further study. It also notes that reflection positivity compatible with admissibility has not been established here. These are boundaries of the present result, not details of notation: they mark what a complete lattice definition would still need.

Which gauge theories the construction covers

The construction is not limited to the Standard Model. The paper identifies conditions on the fermion representation and then applies the same arguments to a class of nonabelian chiral gauge theories. For the SU(N) theories it covers, the adjoint representation has no invariant vector, which eliminates a constant current, and its invariant linear maps are scalar. Combined with the axial transformation law, the only candidate term through order four in the dimensionless current is the epsilon-tensor contraction of the covariant derivative of the field strength, which vanishes by the Bianchi identity.

That leaves an O(a^2) physical current and measure response for complete polynomial contributions, and an O(a^6) response for the exact current. The paper also states where the construction does not extend: it covers theories in the specified class and restricts to other groups where the required representation conditions fail. The Standard Model itself is treated through all four possible global gauge-group structures.

A definition of principle, not a simulation recipe

The paper is explicit that this is a formulation of principle rather than practice. Establishing optimal numerical admissibility bounds, distilling lessons for realistic lattice simulations, and working toward a practical implementation are listed as future work. The continuum interacting theory is also still open: a candidate lattice theory is a start, not a completed definition.

The finite-cutoff analysis shows how the approximation behaves. Truncating the polynomial representation of the current changes the measure only by a phase of unit modulus, so the approximate measure stays normalized, but it need not depend smoothly on the gauge field or satisfy the gauge-invariance condition. The section error is bounded by a constant times the lattice volume and the path length times a factor that decays with the polynomial cutoffs. At fixed lattice and endpoint, that error tends to zero as the cutoffs increase; in a joint continuum limit it tends to zero only if the whole right-hand side does, including the growth of the volume and the path length.

The priority is the measure, not the action

The most useful way to read this paper is as a relocation of the hard problem. Overlap fermions already supply exact lattice chiral symmetry at finite spacing; what was missing was a gauge-invariant measure on the gauge-dependent fermion subspace. The paper's contribution is to show that, on admissible fields and within its class of theories, the comparison current satisfies the locality, gauge, and integrability conditions needed for Lüscher reconstruction.

The practical threshold follows from the same structure. The construction's guarantees are tied to admissible plaquette bounds and to the polynomial truncation of the current, so the next questions are how tight the admissibility bound can be made and how the truncation error scales in a realistic simulation. Until those are answered, the result should be read as a candidate nonperturbative definition of the Standard Model and similar chiral gauge theories, not as a demonstrated computational framework.

Sources used in this answer