Spacetime Winding Number: Topological Invariants

• Spacetime winding number is a topological invariant that records the integer wrapping of fields, wavefunctions, or operators around closed cycles in spacetime or parameter space.
• It appears in various frameworks, including orbifold fixed points with magnetic flux, frequency–momentum Green's functions in Weyl systems, lattice unitary flows, and Wilson loop traversals in gauge theories.
• This invariant underpins robust phenomena such as chiral anomalies, quantized transport, and topological response coefficients, remaining stable under smooth deformations until a singular event occurs.

Spacetime winding number denotes a topological integer that records how a field, wavefunction, Green’s function, or holonomy wraps around a nontrivial cycle in spacetime, compactified spacetime, or a spacetime-augmented parameter space. In current usage the term is not unique: on magnetized orbifolds it refers to local phase monodromy of fermion zero modes at conical singularities; for Weyl Hamiltonians it is a Green’s-function invariant on frequency–momentum space; in lattice and Floquet settings it appears as a real-space flow of unitary operators; and in Yang–Mills theory it can mean the number of times a Wilson line traverses a closed contour (Imai et al., 2022, Elbistan, 2016, Hamano et al., 2024, Matsudo et al., 2017).

1. Conceptual scope and mathematical forms

Across these settings, the common structure is an integer extracted from a closed-cycle dependence. The cycle may be a small loop around an orbifold fixed point, a compactified $(\omega,\mathbf{p})$-surface surrounding a Weyl node, a spatial cut crossed by a locality-preserving unitary, or a spacetime contour followed repeatedly by a Wilson line. The underlying maps are therefore different, and the associated winding numbers are not interchangeable.

Several standard realizations illustrate this diversity. On $T^2/\mathbb{Z}_N$ orbifolds with magnetic flux, the local integer is defined by

$$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$

so it measures the phase winding of an orbifold eigenfunction around a conical singularity (Imai et al., 2022). For Weyl Hamiltonians in $3+1$ dimensions, the Green’s-function winding is

$$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$

which is defined on compactified frequency–momentum space and evaluates the chirality of a Weyl node (Elbistan, 2016). In $SU(N)$ Yang–Mills theory, the multiple-winding Wilson loop

$W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$

uses the integer $m$ itself as the spacetime winding number, namely the number of times the same contour is traversed (Matsudo et al., 2017). In real-space lattice formulations, Kitaev’s one-dimensional flow

$$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$

is an integer for locality-preserving unitaries and reproduces the conventional winding number in translationally invariant systems (Hamano et al., 2024).

This variety suggests a family resemblance rather than a single universal definition. What remains invariant across the literature is the role of winding as an integer-valued obstruction that cannot change under smooth deformations unless a singularity, gap closing, or topological event intervenes.

2. Orbifold fixed points, magnetic flux, and chiral index relations

A particularly explicit notion of spacetime winding number arises for fermions on magnetized two-dimensional orbifolds $T^2/\mathbb{Z}_N$, $T^2/\mathbb{Z}_N$0, where the local data at fixed points controls the chiral index (Imai et al., 2022). The torus is parametrized by $T^2/\mathbb{Z}_N$1, with identifications $T^2/\mathbb{Z}_N$2, and supports a uniform Abelian magnetic flux

$T^2/\mathbb{Z}_N$3

The orbifold adds the identification $T^2/\mathbb{Z}_N$4, $T^2/\mathbb{Z}_N$5, together with Scherk–Schwarz twist phases in the torus boundary conditions.

For $T^2/\mathbb{Z}_N$6, the torus Dirac operator has $T^2/\mathbb{Z}_N$7 right-chiral zero modes and no left-chiral zero modes; orbifold eigenstates are obtained by summing torus modes over $T^2/\mathbb{Z}_N$8 images with eigenvalue $T^2/\mathbb{Z}_N$9 (Imai et al., 2022). The spacetime winding number at a fixed point $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$0 is then the integer $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$1 extracted from the phase monodromy of the $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$2-eigenfunction around a small anticlockwise loop $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$3. In this context, the winding is not a string worldsheet winding along torus cycles. It is a local phase-winding of the fermion wavefunction around a conical singularity induced by the orbifold rotation, the gauge background, and the Scherk–Schwarz twists.

Summing the local windings gives

$$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$4

with a universal sum rule

$$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$5

The chiral index on the orbifold then takes the form

$$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$6

For $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$7 this relation is derived directly by the trace formula,

$$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$8

while for $$\chi_p \equiv \frac{1}{2\pi i}\oint_{C_p} d\boldsymbol{\ell}\cdot \nabla \log \psi,$$9 the same structure is obtained under a generalized trace-identity assumption for arbitrary $3+1$0 (Imai et al., 2022).

The significance of this result is that the smooth bulk contribution $3+1$1 does not by itself determine chirality. The fixed points contribute localized singular terms encoded exactly by the winding numbers. This realizes, in a concrete two-dimensional setting, the broader index-theoretic principle that conical singularities modify the naive bulk index by localized topological data.

3. Frequency–momentum winding for Weyl Hamiltonians

In Weyl semimetals and related chiral systems, spacetime winding number refers to a topological invariant of the Green’s function on compactified frequency–momentum space $3+1$2 (Elbistan, 2014, Elbistan, 2016). For the $3+1$3-dimensional Weyl Hamiltonian $3+1$4, the Green’s-function invariant is the integral $3+1$5 written above. Its evaluation proceeds by inserting the positive- and negative-energy projectors $3+1$6 and performing the $3+1$7-integration. The result reduces to a total divergence in momentum space,

$3+1$8

The vector field $3+1$9 is the field of a Dirac monopole in momentum space centered at the Weyl node. Consequently, the winding has unit magnitude and equals the monopole charge carried by the node; reversing chirality flips the sign (Elbistan, 2016). The same structure persists in higher even spacetime dimensions, where the generalized invariant $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$0 reduces to the divergence of

$$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$1

again giving unit topological charge (Elbistan, 2016).

The same papers show that this Green’s-function winding is equal, up to convention-sensitive sign, to the Chern number obtained from the Berry curvature of the positive-energy band bundle (Elbistan, 2014, Elbistan, 2016). In $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$2 dimensions this is the first Chern number on a surrounding $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$3; in $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$4 dimensions it is the second Chern number on $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$5. The relation is convention-sensitive because the overall sign depends on chirality choice, projector choice, and orientation conventions.

The physical relevance is direct. A chirally coupled effective action contains the term

$$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$6

so the anomalous Hall effect, chiral magnetic effect, gauge anomaly, and topological magnetoelectric response inherit coefficients proportional to the same winding number (Elbistan, 2016). In this sense, the spacetime winding is not merely classificatory; it fixes response coefficients.

4. Real-space flow, disorder, and Floquet reinterpretation

A different but closely related formulation appears in the real-space topology of unitary matrices. For a locality-preserving unitary $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$7 on an infinite one-dimensional lattice, the flow

$$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$8

is an integer, where $$C_3=\frac{1}{24\pi^2}\int d\omega\, d^3p\, \epsilon^{\mu\nu\rho\sigma}\, \mathrm{Tr}\!\left[(G\partial_\mu G^{-1})(G\partial_\nu G^{-1})(G\partial_\rho G^{-1})(G\partial_\sigma G^{-1})\right],$$9 projects onto a half-space (Hamano et al., 2024). Locality ensures that $SU(N)$0 is supported near the cut, so the trace is well-defined. In translationally invariant systems this equals the conventional momentum-space winding

$SU(N)$1

The same paper derives a three-dimensional real-space invariant,

$SU(N)$2

where the $SU(N)$3 partition space into four regions. This is proved to equal the conventional three-dimensional winding number

$SU(N)$4

for translationally invariant systems (Hamano et al., 2024).

The Floquet interpretation is especially important. For a one-period unitary $SU(N)$5, the one-dimensional flow computes the Thouless pump integer in real space. For a $SU(N)$6-dimensional Floquet loop unitary, treating time as a third coordinate converts the Rudner spacetime winding into a three-dimensional winding, and the real-space flow furnishes a disorder-robust evaluation strategy (Hamano et al., 2024). This removes the need for crystal momentum and preserves quantization as long as the unitary remains locality-preserving.

The conceptual shift is substantial: here the winding is not attached to a field configuration in continuous spacetime but to a local unitary evolution operator acting on a lattice. Nevertheless, its function is analogous—an integer-valued obstruction stable under homotopies that preserve locality and unitarity.

5. Geometric phases, curve geometry, and loop–surface representations

The literature also contains geometric and operator-theoretic reformulations that encode winding information without using the standard degree-of-map language directly. In the self-dual massive Kalb–Ramond–Klein–Gordon model in $SU(N)$7 dimensions, the Noether generator of duality rotations,

$SU(N)$8

is a topological, metric-independent quantity (Iñiguez et al., 28 May 2025). In the loop/surface representation, its second term becomes a two-dimensional invariant equal to the signed count of points piercing a surface,

$SU(N)$9

The paper interprets this invariant as encoding the same information as the standard winding number of planar vortices enclosed by a curve (Iñiguez et al., 28 May 2025). The winding data is therefore realized as an intersection number between a $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$0-surface and a $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$1-surface.

A related geometric reformulation expresses topological invariants through intrinsic curve geometry. For a unit vector field represented by evolving space curves, the $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$2-dimensional winding number is

$W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$3

so in $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$4D it is determined entirely by the torsions of two evolving curves (Balakrishnan et al., 2023). In $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$5D, by contrast, the corresponding invariant requires the total twist density $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$6, and the global anholonomy takes the helicity-like form

$W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$7

(Balakrishnan et al., 2023).

The same framework admits a spacetime extension by replacing $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$8 with $W_m(C)=\frac{1}{N}\mathrm{Tr}[U(C)]^m$9, giving the natural analogue

$m$0

The paper directly establishes the spatial formulas and presents the spacetime replacement as an extension of the same geometric-phase mechanism (Balakrishnan et al., 2023). This is one of the clearest examples in which “spacetime winding number” is literally obtained by promoting a spatial topological density to a mixed space–time sheet.

6. Wilson-loop winding, supersymmetric indices, and recurrent distinctions

In non-Abelian gauge theory, spacetime winding number can mean the number of times a Wilson line traces a closed contour in spacetime. If $m$1, then

$m$2

is the $m$3-winding Wilson loop, with $m$4 the spacetime winding number (Matsudo et al., 2017). This notion is combinatorial rather than homotopic: it counts repeated traversal of a contour. Its physical effect is controlled by the decomposition of $m$5 into irreducible $m$6 characters and hence by the string tensions of those representations.

The resulting area laws are markedly group-dependent. For $m$7, double winding exhibits the difference-of-areas law because $m$8 contains a singlet. For $m$9, the double-winding area law is neither difference-of-areas nor sum-of-areas, and for $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$0 the difference-of-areas law is excluded while large $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$1 approaches a sum-of-areas form under Casimir scaling (Matsudo et al., 2017). Here the “winding number” controls which representation content is probed by the loop rather than a localized phase singularity or a map degree.

Supersymmetric index theory offers yet another use of winding. In theories with two real supercharges in a Majorana doublet, the twisted partition function of matter-only models reduces to an exact Gaussian integral in the variables $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$2, and the Witten index becomes the degree of the asymptotic map

$$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$3

that is, a winding number of $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$4 on the sphere at infinity (Ghim et al., 2019). In $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$5 $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$6 Chern–Simons–matter systems, the $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$7-dependent part of the index is controlled by the neutral sector’s winding of $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$8, while holonomy saddles organize the remaining contributions (Ghim et al., 2019). The winding here lives in path-integral zero-mode space induced by compactified spacetime, not in ordinary physical space.

A persistent misconception is therefore that “spacetime winding number” denotes one invariant with one canonical formula. The literature does not support that reading. The term names several topological integers associated with wrapping, monodromy, or repeated traversal, but the domain can be a local loop around an orbifold singularity, compactified $$\mathcal{F}_1(U)=\mathrm{Tr}\big(U^\dagger[\Pi,U]\big)$$9-space, a lattice cut, a spacetime contour, or a zero-mode configuration space. What unifies these constructions is not identical definition but identical function: they encode robust, integer-valued information that survives smooth deformations until a singular event changes the topology.