Chern-Simons Theory

Definition

The Chern–Simons function. Take a connection $A$ on a bundle over a closed oriented 3-manifold $W$, and a 4-manifold $X$ with $\partial X=W$ over which the bundle and connection extend. Then

$$\mathrm{CS}(A)={\int }_X\frac{\mathrm{Tr}F\wedge F}{8{\pi }^2}\mathrm{mod}\mathbb{Z}$$

does not depend on the extension. Two extensions glue into a closed 4-manifold, and the difference is that manifold’s instanton number, an integer (Witten 2026, §1). It is a secondary characteristic class: locally it is given by a Chern–Simons form whose exterior derivative is the second Chern form.

• Abelian case. For a trivialized line bundle, $\mathrm{CS}(A)=\frac{1}{8{\pi }^2}{\int }_{W}A\wedge dA$.
• Non-abelian case. The local formula acquires the familiar cubic term. Sign conventions for the trace differ between hermitian and anti-hermitian $A$.
• Variation. In the abelian case $\delta \mathrm{CS}=\frac{1}{4{\pi }^2}\int \delta A\wedge F$, so the critical points are the flat connections.

Chern–Simons theory is the gauge theory with action $2\pi k\mathrm{CS}(A)$; for $U(1)$ this is $\frac{k}{4\pi }\int AdA$.

Level quantization. $\mathrm{CS}$ is defined only mod $\mathbb{Z}$ and $e^{iI}$ must be single-valued, so $k\in \mathbb{Z}$. For $U(1)$ there is a refinement.

• ${\int }_{X}F\wedge F/8{\pi }^2=\frac{1}{2}{\int }_X{c}_1^2$ is an integer only when the intersection form is even, i.e. on spin 4-manifolds.
• So $\mathrm{CS}(A)\mathrm{mod}\mathbb{Z}$ needs a spin structure on $W$ and depends on it. On $T^3=T^2\times S^1$, with $A$ pulled back from a degree-one connection on $T^2$, it is $0$ or $\frac{1}{2}$ according to the spin structure (Witten 2026, §3).
• Without a spin structure only $2\mathrm{CS}(A)$ is defined. A bosonic $U(1{)}_k$ therefore needs even $k$, and odd $k$ is a spin TQFT.
• The mixed invariant $\mathrm{CS}(A,a)$, defined via ${\int }_{X}F\wedge f/4{\pi }^2$, needs no spin structure and carries an integer coefficient (Witten 2026, §§3–4).

Structure of the theory.

• A topological field theory. No metric enters and classical solutions are flat. States on a closed surface ${\Sigma }_g$ form a finite-dimensional space, of dimension $|k{|}^g$ for $U(1{)}_k$. Anyons are Wilson lines: the charge-$n$ line has spin $h_n=n^2/2k\mathrm{mod}1$ and mutual braiding phase $e^{2\pi inm/k}$. These lines generate a ${\mathbb{Z}}_k$ 1-form symmetry whose ‘t Hooft anomaly is encoded in the spins (Hsin–Lam–Seiberg). Non-abelian Chern–Simons theory gives the Jones polynomial and 3-manifold invariants (Witten 1989).

• Maxwell–Chern–Simons (topologically massive). $\mathcal{L}=-\frac{1}{4e^2}{F}_{\mu \nu }{F}^{\mu \nu }+\frac{k}{4\pi }{ϵ}^{\mu \nu \rho }{A}_{\mu }{\partial }_{\nu }{A}_{\rho }$ has a single massive mode.

  • Its mass is $m=k{e}^2/2\pi$ and its spin is $\pm 1$, set by the sign of $k$ (Deser–Jackiw–Templeton).
  • At distances $\gg 1/m$ the Chern–Simons term dominates and the theory flows to the $U(1{)}_k$ TQFT.
  • On a spatial torus the holonomy zero modes form a Landau problem. Its cyclotron frequency is exactly $m$ and its $k$-fold degenerate lowest level is the TQFT sector (Dunne–Jackiw–Trugenberger; Dunne, hep-th/9902115).

• Duality with the self-dual model (Deser–Jackiw 1984). Maxwell–Chern–Simons is locally equivalent to ${\mathcal{L}}_{SD}=\frac{1}{2}{f}_{\mu }{f}^{\mu }-\frac{1}{2m}{ϵ}^{\mu \nu \rho }{f}_{\mu }{\partial }_{\nu }{f}_{\rho }$. The self-dual model is not a gauge theory and has no holonomy sector, so globally the equivalence should hold only up to the $U(1{)}_k$ TQFT factor. That statement is EXPECTED; no clean reference has been checked yet.

• The $K$-matrix (Wen–Zee 1992). The multi-component theory $\mathcal{L}=-\frac{1}{4\pi }{K}_{IJ}{a}^{I}d{a}^J+\frac{1}{2\pi }{t}_{I}Ad{a}^I$ has:

  • Hall conductance ${\sigma }_{xy}=t^{\mathsf{T}}{K}^{-1}t$ in units of $e^2/h$;
  • ground-state degeneracy $|\det K{|}^g$ on ${\Sigma }_g$;
  • anyon statistics read from $K^{-1}$.

  Witten’s $(k,s,r)$ model maps onto it as $r=-K$ and $s=t$, on top of a background level $k$. The generalized-symmetries course writes $+\frac{1}{4\pi }Kada$, the mirror image, where $r=K$ and ${\sigma }_{xy}=k-t^{\mathsf{T}}{K}^{-1}t$.

• Boundaries and inflow. On a 3-manifold with boundary, $e^{2\pi ik\mathrm{CS}}$ is a vector in a line determined by the boundary data (Ramadas–Singer–Weitsman). Chiral edge degrees of freedom must compensate, which is anomaly inflow. The Schwinger term (level) of the edge current algebra equals the Hall conductance.

• Four-dimensional origin. ${\mathrm{CS}}_X(A)={\int }_{X}F\wedge F/8{\pi }^2$ says that a region with $\theta =2\pi k$ bounded by $W$ induces $2\pi k\mathrm{CS}(A)$ on $W$. A θ-wall of height $\Delta \theta$ carries level $\Delta \theta /2\pi$, which is the bridge to axionic-electrodynamics.

Role in Research

The earliest line of the portfolio lives on this page (confinement-duality):

• Duality. Marcelo’s MSc thesis and PLB 605 (2005) extend the Maxwell–Chern–Simons/self-dual duality to noncommutative space. PLB 625 (2005) looks for the dual of Maxwell–Chern–Simons coupled to charged matter.
• Monopoles and Julia–Toulouse. PLB 674 (2009) studies monopoles in the presence of the Chern–Simons term via Julia–Toulouse. The Chern–Simons term forces monopole-instantons to carry electric charge, which spoils Polyakov confinement (Pisarski 1986; Affleck–Harvey–Palla–Semenoff 1989). PRD 88 (2013) puts the Chern–Simons term in a dual Josephson junction.
• Lorentz violation. The 2006–2009 papers on induced Chern–Simons-like (Carroll–Field–Jackiw) terms in Lorentz-violating QED concern a term that, for constant $b_{\mu }$, is axion electrodynamics with linear $\theta =2b\cdot x$. That is the anomalous Hall response of a Weyl semimetal (Zyuzin–Burkov 2012, ★ in Zotero). The historical Lorentz-violation work and the condensed-matter line therefore share one object.

It also enters current work in three places:

• Project 18 (non-invertible-symmetries-mcs). Any analysis of the global and non-invertible symmetries of Maxwell–Chern–Simons needs the data above: integer level, spin structure for odd $k$, the ${\mathbb{Z}}_k$ 1-form symmetry and its anomaly, the $|k{|}^g$ degeneracy, and the global caveat on the self-dual duality.
• condensed-matter-connections. This is the effective action of the quantum-hall-effect, the boundary term of θ-electrodynamics, and the source of chiral edge modes.
• Quantum information. The topological entanglement entropy of $U(1{)}_k$ is $\gamma =\ln \sqrt{|k|}$. Abelian link states are stabilizer states (Salton–Swingle–Walter 2017), so the TQFT is magic-free, whereas the local Maxwell–Chern–Simons theory is not (2026-benedetti-magic-in-qft).
• Courses. Generalized-symmetries course: Semester I Week 12 (the θ-term, its spin-structure periodicity and the Hall response of a θ-wall) and Semester II Week 10 ($K$-matrix, Laughlin $\nu =1/3$, in the mirror convention). Semester II Week 12 builds the modified-Villain lattice on which Jacobson and Sulejmanpasic construct $U(1{)}_k$ from a lattice θ-term, the lattice version of the formula above; the course itself does not construct $U(1{)}_k$.

Relations

• quantum-hall-effect: the Chern–Simons level is the Hall conductance.
• topological-order: abelian Chern–Simons theories are the continuum topological orders of fractional quantum Hall fluids.
• axionic-electrodynamics: the level across a θ-wall is $\Delta \theta /2\pi$.
• t-hooft-anomaly: the boundary of a Chern–Simons region carries an anomalous chiral theory (inflow).
• higher-form-symmetries: the ${\mathbb{Z}}_k$ 1-form symmetry generated by Wilson lines.
• wilson-loop: anyons are Wilson lines.
• julia-toulouse-mechanism: the group’s 2+1d condensation papers take the Chern–Simons term as an input; a condensation derivation of a Chern–Simons level would have to produce an integer.
• villain-action: modified-Villain lattice $U(1{)}_k$.
• non-invertible-symmetries-mcs: the open question on Maxwell–Chern–Simons symmetries.

Papers

• S.-S. Chern, J. Simons, Characteristic forms and geometric invariants, Ann. Math. 99 (1974) 48.
• A. Schwarz, CMP 67 (1979) 1. The abelian topological theory.
• S. Deser, R. Jackiw, S. Templeton, Ann. Phys. 140 (1982) 372. Topologically massive gauge theories.
• S. Deser, R. Jackiw, PLB 139 (1984) 371. The self-dual model.
• E. Witten, CMP 121 (1989) 351. Chern–Simons theory and the Jones polynomial.
• G. Dunne, R. Jackiw, C. Trugenberger, Ann. Phys. 194 (1989) 197. Maxwell–Chern–Simons on the torus. See also G. Dunne, Aspects of Chern–Simons theory, hep-th/9902115.
• I. Affleck, J. Harvey, L. Palla, G. Semenoff, NPB 328 (1989) 575. The Chern–Simons term versus the monopole.
• X.-G. Wen, A. Zee, PRB 46 (1992) 2290. The $K$-matrix.
• P.-S. Hsin, H. T. Lam, N. Seiberg, SciPost Phys. 6 (2019) 039 [arXiv:1812.04716]. 1-form symmetries in 3d and their anomalies.
• T. Jacobson, T. Sulejmanpasic, PRD 107 (2023) 125017 [arXiv:2303.06160]. Modified-Villain lattice $U(1{)}_k$.
• 2026-witten-chern-simons-quantum-hall. Level quantization, spin structure, inflow, and the $(k,s,r)$ model of the fractional effect.
• Group: MSc thesis (2005); PLB 605 (2005); PLB 625 (2005); PLB 674 (2009); PRD 88 (2013). See confinement-duality.

Notes

• Compactness is what quantizes. Level quantization is a statement about $U(1)$ as a compact group: line bundles, large gauge transformations, monopole configurations. For a non-compact gauge group, or in perturbation theory on ${\mathbb{R}}^3$, nothing forces $k\in \mathbb{Z}$. Much of the 1980s–2000s Maxwell–Chern–Simons literature, including the duality papers, works in that local, perturbative regime.
• Normalization ledger. Witten’s $\mathrm{CS}\in \mathbb{R}/\mathbb{Z}$, and the physics action is $2\pi k\mathrm{CS}=\frac{k}{4\pi }\int AdA$. After $A\to eA$ the Maxwell–Chern–Simons Lagrangian reads $-\frac{1}{4}{F}^2+\frac{k{e}^2}{4\pi }ϵA\partial A$. That is the Deser–Jackiw–Templeton form $-\frac{1}{4}{F}^2+\frac{m}{2}ϵA\partial A$ with $m=k{e}^2/2\pi$.
• On ${\mathbb{R}}^{2,1}$ the TQFT sector is invisible. It appears only with nontrivial spatial topology, with boundaries, or with anyon (Wilson-line) insertions.
• Witten’s argument implies that a purely bosonic lattice model can realize $U(1{)}_k$ only for even $k$. Odd levels need a lattice spin structure or fundamental fermions.