Where the symmetry breaks itself.

QFT document 17: when classical symmetries fail quantum-mechanically. Fujikawa’s path integral derivation, ‘t Hooft matching across RG flows, gravitational anomalies, and anomaly inflow. The deepest non-perturbative constraints in quantum field theory.

Document 10 introduced anomalies briefly; we saw how the chiral anomaly arises from the path integral measure not respecting axial transformations, giving $\partial_\mu J^\mu_A = (e^2/16\pi^2)F\tilde F$. Document 12’s SM anomaly cancellation showed they’re not just curiosities; they constrain what gauge theories are consistent.

This document goes deep. Anomalies are arguably the most important non-perturbative phenomena in QFT. They:

• Match exactly between IR and UV descriptions (not a symmetry of perturbation theory or a specific cutoff; they’re topological)
• Constrain theories that can consistently couple to gauge fields
• Predict observables like $\pi^0 \to \gamma\gamma$ decay rate to high precision
• Connect QFT to topology via index theorems and cohomology
• Appear in condensed matter (topological insulators, quantum Hall effect)
• Propagate through spacetime via anomaly inflow
• Obstruct symmetry realization at the quantum level

By the end of this document, you’ll see anomalies as one of the most mathematically rich and physically predictive structures in theoretical physics.

Prerequisites

• Documents 10-12 (fermion path integrals, Yang-Mills, Standard Model)
• Document 14 (EFT methodology, relevant for ‘t Hooft matching)
• Document 15 (ChPT; pion anomaly is canonical example)

Conventions

• Mostly-minus metric
• $\hbar = c = 1$
• Chiral projectors: $P_{L,R} = (1 \mp \gamma^5)/2$
• For abelian anomalies: $F\tilde F = \tfrac{1}{2}\epsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma}$
• For non-abelian: $\text{tr}(F\tilde F) = \tfrac{1}{2}\epsilon^{\mu\nu\rho\sigma}\text{tr}(F_{\mu\nu}F_{\rho\sigma})$

Table of Contents

1. What Is an Anomaly?
2. The ABJ Anomaly by Triangle Diagram
3. Fujikawa’s Path Integral Derivation
4. Global vs. Gauge Anomalies
5. The Wess-Zumino Consistency Conditions
6. Anomaly Polynomials and the Descent Equations
7. The Physical Canonical Example: $\pi^0 \to \gamma\gamma$
8. QCD Vacuum Structure and the $\eta'$ Mass
9. $t$ Hooft Matching
10. Gravitational Anomalies
11. Anomaly Inflow
12. Global Anomalies and Witten’s $SU(2)$
13. Anomalies in Condensed Matter
14. Appendix: Anomaly Formulas Reference

1. What Is an Anomaly?

The Clean Definition

Consider a classical field theory with action $S[\phi]$ invariant under a symmetry transformation $\phi \to \phi + \epsilon\delta\phi$. Noether’s theorem gives a conserved current $J^\mu$ with $\partial_\mu J^\mu = 0$ classically.

An anomaly is the failure of this classical conservation law in the quantum theory:

$\partial_\mu\langle J^\mu\rangle \neq 0$

even though the classical Lagrangian is symmetric. The symmetry is broken by quantum effects; not by adding a symmetry-breaking term to the Lagrangian, but by the very definition of the quantum theory.

Why This Can Happen

The classical action has a symmetry. But defining a quantum theory requires regularization; a procedure for handling divergences. Sometimes no regulator respects all the classical symmetries simultaneously. You’re forced to sacrifice one.

In this case, the “lost” symmetry shows up as a nonzero divergence of the Noether current. The anomaly measures exactly how the symmetry fails.

Why Anomalies Are Protected

Anomalies are not artifacts of specific regularizations. Different regulators all give the same anomaly coefficient. They’re computed by one-loop Feynman diagrams with specific topological properties (triangle, box, hexagon diagrams in different dimensions), and the result is the same regardless of how you regulate.

This universality is why anomalies are important: they’re physical, not just cutoff-dependent.

Two Kinds of Anomalies

Global anomalies: A global symmetry has an anomaly. This is a prediction; the symmetry current isn’t conserved quantum-mechanically, producing observable effects.

Examples: the ABJ anomaly ($\partial J_A = F\tilde F$) in QED/QCD, the pion anomaly giving $\pi^0\to\gamma\gamma$, lepton/baryon number anomalies from sphalerons.

Gauge anomalies: A local (gauge) symmetry has an anomaly. This is a problem; gauge anomalies break unitarity and make the theory inconsistent. Gauge anomalies must cancel.

Examples: the Standard Model’s six anomaly cancellation conditions (document 12). Theories that don’t satisfy them are inconsistent.

Why Anomalies Are Deep

Anomalies connect QFT to topology, geometry, and non-perturbative physics in ways no other quantum effect does. They:

• Are exact (not approximate, not perturbative)
• Are topological (invariant under continuous deformations)
• Satisfy matching conditions across scale flows
• Obstruct the existence of certain theories
• Relate to index theorems in differential geometry

The general lesson: not every classical symmetry survives quantization. Which ones survive, and which don’t, is a deep structural question about the theory.

2. The ABJ Anomaly by Triangle Diagram

Before Fujikawa, the anomaly was discovered via a one-loop triangle diagram. Let’s review this calculation; historical and still the most direct way to compute specific anomaly coefficients.

Setup

Consider QED with massless Dirac fermions:

$\mathcal{L} = \bar\psi i\slashed D\psi - \tfrac{1}{4}F_{\mu\nu}F^{\mu\nu}$

Classical symmetries:

• Vector $U(1)$: $\psi \to e^{i\alpha}\psi$, with current $J^\mu_V = \bar\psi\gamma^\mu\psi$
• Axial $U(1)$: $\psi \to e^{i\beta\gamma^5}\psi$, with current $J^\mu_A = \bar\psi\gamma^\mu\gamma^5\psi$

Classically, $\partial_\mu J^\mu_V = 0$ (exact) and $\partial_\mu J^\mu_A = 0$ (if $m = 0$).

The Triangle Diagram

Consider the correlator $\langle J^\mu_A(z)J^\nu_V(x)J^\rho_V(y)\rangle$. At one loop, this is a triangle diagram with:

• Axial vertex $\gamma^\mu\gamma^5$ at $z$
• Two vector vertices $\gamma^\nu$, $\gamma^\rho$ at $x, y$
• Fermion loop connecting them

Computing $\partial_\mu\langle J^\mu_A\cdot J^\nu_V\cdot J^\rho_V\rangle$ naively, you’d expect zero. But it isn’t.

The Anomalous Result

After careful computation (regularizing with Pauli-Villars or dim-reg, and imposing vector current conservation):

$\partial_\mu\langle J^\mu_A(z)\rangle = \frac{e^2}{16\pi^2}\epsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma} = \frac{e^2}{8\pi^2}F_{\mu\nu}\tilde F^{\mu\nu}$

where $\tilde F^{\mu\nu} = \tfrac{1}{2}\epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}$.

Or, using $F\tilde F = \tfrac{1}{2}\epsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma}$:

$\boxed{\partial_\mu J^\mu_A = \frac{e^2}{8\pi^2}F\tilde F}$

This is the Adler-Bell-Jackiw (ABJ) anomaly.

Key Features

Coefficient is precise: $e^2/(8\pi^2)$, not depending on any other parameter of the theory. Pure geometry.

Regulator independence: Pauli-Villars, dim-reg, zeta-function; all give the same coefficient.

One-loop exact: The Adler-Bardeen theorem says higher-loop corrections don’t modify the coefficient. The anomaly is exact at one loop.

Topological: Depends only on the number of fermion species and their charges, not on masses or coupling values.

The Choice of Scheme

In the calculation, there’s a subtle choice: which symmetry do you preserve?

• Preserve vector gauge invariance → anomaly in axial current (as above)
• Preserve axial symmetry → anomaly in vector current (breaks gauge invariance!)

The physical convention: always preserve gauge symmetry (vector in QED, color in QCD). Anomalies then appear in global currents, where they can be physical.

If you’re forced to have an anomaly in the gauge current, the theory is inconsistent (gauge anomaly). This is the situation that must be avoided.

Non-Abelian Generalization

For a non-abelian gauge theory, the anomaly of the flavor axial current:

$\partial_\mu J^{a\mu}_A = \frac{g^2}{8\pi^2}\text{Tr}[T^a(F_{\mu\nu}\tilde F^{\mu\nu})]$

where the trace is over flavor/color indices, and $T^a$ is the generator of the axial transformation.

For $SU(N)$ with fermions in the fundamental: the anomaly is nonzero. For $SU(2)$ alone: the anomaly vanishes because $\text{Tr}(T^a T^bT^c) = 0$ (there’s no symmetric $d^{abc}$ tensor for $SU(2)$).

This is why $SU(2)$ anomalies in the Standard Model (document 12) cancelled automatically; $SU(2)_L^3$ and $SU(2)_L^2 U(1)_Y$ anomalies vanish identically.

3. Fujikawa’s Path Integral Derivation

The triangle diagram gives the answer but doesn’t explain why. Fujikawa (1979) found a much deeper perspective: the anomaly comes from the path integral measure.

The Setup

Start with the Euclidean path integral:

$Z = \int\mathcal{D}\bar\psi\mathcal{D}\psi\mathcal{D}A\, e^{-S_E[\psi, A]}$

Classical symmetry: $\psi \to e^{i\beta(x)\gamma^5}\psi$, with $\beta(x)$ a local parameter. At the level of the Lagrangian, this is a symmetry (for $m = 0$).

The Key Question

Does the path integral measure $\mathcal{D}\bar\psi\mathcal{D}\psi$ also respect this symmetry?

Define an orthonormal basis for fermion fields in a given background gauge field $A$:

$i\slashed D\, \phi_n = \lambda_n\phi_n$

(Eigenfunctions of the Dirac operator, with $\lambda_n$ real due to hermiticity of $i\slashed D$.) Expand $\psi = \sum_n a_n\phi_n$, $\bar\psi = \sum_n\bar a_n\phi_n^\dagger$. The measure:

$\mathcal{D}\bar\psi\mathcal{D}\psi = \prod_n d\bar a_n\, da_n$

Axial Transformation on the Measure

Under $\psi \to e^{i\beta\gamma^5}\psi$, the expansion coefficients transform by a unitary matrix:

$a_n \to U_{nm}a_m, \quad U_{nm} = \int d^4x\, \phi_n^\dagger(x)\, e^{i\beta(x)\gamma^5}\phi_m(x)$

For Grassmann integration, the Jacobian is $\det^{-2}U$ (Grassmann Jacobians are inverse):

$\mathcal{D}\bar\psi\mathcal{D}\psi = \det^{-2}U\cdot\mathcal{D}\bar\psi'\mathcal{D}\psi'$

Expanding the Determinant

For small $\beta(x)$: $U \approx 1 + i\beta(x)\gamma^5\cdot\text{(position operator-ish)}$.

Using $\det U = \exp(\text{Tr}\ln U) \approx \exp(\text{Tr}(U - 1))$:

$\det^{-2}U \approx \exp\left[-2i\int d^4x\,\beta(x)\sum_n\phi_n^\dagger(x)\gamma^5\phi_n(x)\right]$

The quantity $\sum_n\phi_n^\dagger(x)\gamma^5\phi_n(x)$ is the chiral anomaly density.

Regularizing the Sum

The sum over modes is UV divergent. Regulate with Gaussian:

$\sum_n\phi_n^\dagger(x)\gamma^5\phi_n(x) = \lim_{M\to\infty}\sum_n\phi_n^\dagger(x)\gamma^5 e^{-(\slashed D)^2/M^2}\phi_n(x)$

$= \lim_{M\to\infty}\text{tr}\left[\gamma^5 e^{-(\slashed D)^2/M^2}\delta^4(x - x)\right]$

The $\delta^4$ at coincident points needs careful handling. In Fourier space:

$= \lim_{M\to\infty}\int\frac{d^4k}{(2\pi)^4}e^{-ik\cdot x}\text{tr}[\gamma^5 e^{-(\slashed D)^2/M^2}]e^{ik\cdot x}$

Evaluating the Trace

Use $\slashed D^2 = D^2 - \tfrac{1}{2}\sigma^{\mu\nu}F_{\mu\nu}$ (with $\sigma = [\gamma, \gamma]/2$):

$\text{tr}[\gamma^5 e^{-D^2/M^2 + \sigma F/(2M^2) - (k+iA)^2/M^2}]$

Doing the Dirac trace, only terms with at least 4 gamma matrices survive (since $\text{tr}(\gamma^5) = \text{tr}(\gamma^5 \gamma^\mu\gamma^\nu) = 0$). The minimum structure is:

$\text{tr}[\gamma^5\gamma^\mu\gamma^\nu\gamma^\rho\gamma^\sigma] = -4i\epsilon^{\mu\nu\rho\sigma}$

Expanding the exponential to get 4 gammas: need $(\sigma F)^2$ term.

$e^{\sigma F/(2M^2)} \supset \frac{1}{2!}\left(\frac{\sigma F}{2M^2}\right)^2$

And $(\sigma^{\mu\nu}\sigma^{\rho\sigma})F_{\mu\nu}F_{\rho\sigma}$ has the structure needed after tracing with $\gamma^5$:

$\text{tr}[\gamma^5\sigma^{\mu\nu}\sigma^{\rho\sigma}]F_{\mu\nu}F_{\rho\sigma} = -4i\epsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma} = -8i F\tilde F$

The Final Result

Combining factors, doing the momentum integral:

$\int\frac{d^4k}{(2\pi)^4}e^{-k^2/M^2} = \frac{M^4}{(4\pi)^2}$

And the $1/M^4$ dimensional factors from $(\sigma F/M^2)^2$ cancel the $M^4$. The result is finite:

$\sum_n\phi_n^\dagger(x)\gamma^5\phi_n(x) = \frac{1}{16\pi^2}F\tilde F(x)$

(Or in non-abelian: $\text{tr}[F\tilde F]/16\pi^2$.)

The Anomaly

Putting it all together:

$\mathcal{D}\bar\psi\mathcal{D}\psi \to \mathcal{D}\bar\psi\mathcal{D}\psi\cdot\exp\left[-2i\int d^4x\,\beta(x)\cdot\frac{1}{16\pi^2}F\tilde F\right]$

Setting $\beta =$ const and integrating the exponent against the classical Lagrangian, we recover the anomaly:

$\partial_\mu J^\mu_A = -\frac{2}{16\pi^2}F\tilde F = -\frac{1}{8\pi^2}F\tilde F$

Wait, the sign; let me double-check. With the conventions chosen and taking into account the factor of 2 (from $\det^{-2}$), the divergence is:

$\boxed{\partial_\mu J^\mu_A = \frac{1}{16\pi^2}F\tilde F \text{ (per fermion species in fundamental)}}$

Multiplying by charge squared $e^2$ and number of fermions gives the QED result.

What Fujikawa’s Derivation Teaches

1. Anomaly is geometric. It comes from how the measure transforms. This is more fundamental than any specific diagram.

2. Zero modes are central. The $\gamma^5$ trace is related to zero modes of the Dirac operator, which is an index-theorem quantity.

3. Connection to index theorems. The integral $\int F\tilde F$ is the second Chern class of the gauge bundle, and the Atiyah-Singer theorem relates this to the index of the Dirac operator. Anomalies are index theorems.

4. Regularization matters. The Gaussian regulator we used preserves vector gauge invariance. A different regulator (respecting axial instead) would give the anomaly in the vector current. Physics is invariant; the location of the anomaly depends on scheme.

5. Non-abelian generalization. The same procedure works for non-abelian gauge theories, giving:

$\partial_\mu J^\mu_A = \frac{g^2}{16\pi^2}\text{tr}[F^{\mu\nu}\tilde F_{\mu\nu}] = \frac{g^2}{32\pi^2}\epsilon^{\mu\nu\rho\sigma}\text{tr}[F_{\mu\nu}F_{\rho\sigma}]$

4. Global vs. Gauge Anomalies

Why the Distinction Matters

Global symmetry with an anomaly: The symmetry is broken quantum-mechanically. The divergence of the Noether current gives a physical prediction (like $\pi^0 \to \gamma\gamma$).

Gauge symmetry with an anomaly: The Ward identity is violated. Longitudinal gauge modes don’t decouple. Unitarity fails. The theory is inconsistent.

Gauge Anomaly Cancellation Requirements

For a gauge theory with gauge group $G$ to be consistent:

$\text{tr}[T^a\{T^b, T^c\}] = 0$

summed over all chiral fermions in the theory, with $T$ being group generators.

This is the gauge anomaly cancellation condition. In the Standard Model (document 12), the six cancellation conditions (covered in Workbook III VIII.1) must all hold. They do, for one generation of SM fermions, which is why the SM is consistent.

Gauge Anomalies in Non-Abelian Groups

For $SU(N)$ with $N \geq 3$: gauge anomalies potentially exist. Need $\text{tr}[T^a\{T^b, T^c\}] = d^{abc}\cdot(\text{anomaly})$ with $d^{abc}$ totally symmetric invariant.

For $SU(N)$: $d^{abc}$ exists for $N \geq 3$ (not for $SU(2)$). So $SU(3)_C$ anomalies must be checked (and cancel in QCD).

For $SU(2)$: $d^{abc} = 0$, so perturbative $SU(2)$ anomalies vanish trivially. But there’s a global $SU(2)$ anomaly (Witten, 1982); section 12.

Mixed Anomalies

When multiple symmetries are present, you can have mixed anomalies involving different groups.

Example: $SU(3)^2 U(1)_Y$ anomaly in SM. This involves two QCD gluons and one hypercharge boson. Canceling it is one of the six constraints.

Mixed anomalies are less constrained than pure anomalies but can still be physical. They’re important for:

• ‘t Hooft matching (section 9)
• Anomalous commutators
• Cosmological/gravitational effects

Large Gauge Transformations

Gauge anomalies discussed so far are perturbative; they’re computed from triangle diagrams. There are also non-perturbative gauge anomalies from gauge transformations not continuously connected to the identity (large gauge transformations).

The canonical example: Witten’s $SU(2)$ anomaly. An $SU(2)$ gauge theory with an odd number of left-handed doublets is inconsistent; not because of perturbative anomalies (which vanish) but because of a non-perturbative global obstruction (section 12).

5. The Wess-Zumino Consistency Conditions

The Setup

For a theory with gauge symmetry and potentially anomalous global symmetries, the anomaly $\mathcal{A}$ must satisfy certain consistency conditions.

If we parametrize gauge transformations $\delta_\theta$ and global transformations $\delta_\alpha$, the anomaly is defined as:

$\delta_\alpha W[A] = \int d^4x\,\alpha(x)\mathcal{A}(x; A)$

where $W[A]$ is the effective action (integrating out fermions in background gauge field).

The Wess-Zumino Conditions

Consistency requires: the variation of a variation equals the variation of a commutator. In particular:

$\delta_{\alpha_1}\delta_{\alpha_2}W - \delta_{\alpha_2}\delta_{\alpha_1}W = \delta_{[\alpha_1, \alpha_2]}W$

(For Lie group elements; commutators of group transformations.)

Translating to anomalies:

$\delta_{\alpha_1}\int\alpha_2\mathcal{A} - \delta_{\alpha_2}\int\alpha_1\mathcal{A} = \int[\alpha_1, \alpha_2]\mathcal{A}$

These are the Wess-Zumino consistency conditions. They constrain the form of the anomaly function $\mathcal{A}$.

The Solution: WZW Term

The Wess-Zumino-Witten (WZW) term is the characteristic object that satisfies these conditions. For chiral gauge anomalies:

$\mathcal{A}(x; A) = \frac{1}{24\pi^2}\epsilon^{\mu\nu\rho\sigma}\partial_\mu\text{tr}[A_\nu\partial_\rho A_\sigma + \tfrac{1}{2}A_\nu A_\rho A_\sigma] + \text{conj.}$

This is the structure that shows up in the chiral anomaly.

The Significance

The WZ conditions show that anomalies aren’t arbitrary; they have a specific mathematical structure. Any candidate anomaly must satisfy these constraints. This is a powerful check on calculations.

More deeply, anomalies can be classified cohomologically. Anomalies are elements of a specific cohomology group of the gauge group (Lie algebra cohomology, or de Rham cohomology on the gauge orbit space). Different cohomology classes correspond to different “types” of anomalies.

6. Anomaly Polynomials and the Descent Equations

Anomalies in various dimensions have a beautiful unified structure: they’re controlled by a single polynomial in field strengths and curvatures.

The Anomaly Polynomial

For a theory in $d$ dimensions, the anomaly polynomial is a $(d + 2)$-form constructed from gauge field strengths $F$ and gravitational curvatures $R$.

4-dimensional chiral gauge anomaly: $I_6 \sim \text{tr}(F^3)$ (a 6-form).

4-dimensional mixed gauge-gravitational anomaly: $I_6 \sim \text{tr}(F)\text{tr}(R^2)$.

4-dimensional gravitational anomaly: only in $4k + 2$ dimensions, so not in 4D.

2-dimensional anomaly: $I_4 \sim F^2$ or $R^2$.

6-dimensional anomaly: $I_8 \sim F^4$, etc.

The anomaly polynomial contains all the information about the anomaly.

The Descent Equations

Given an anomaly polynomial $I_{d+2}$ that is closed ($dI_{d+2} = 0$), one can locally write:

$I_{d+2} = d\omega_{d+1}$

The form $\omega_{d+1}$ is the Chern-Simons form. Its gauge variation:

$\delta\omega_{d+1} = d\omega^{(1)}_d$

and the form $\omega^{(1)}_d$ is the anomaly $d$-form; it’s what gives the anomaly in $d$ spacetime dimensions.

Why This Is Powerful

This structure unifies:

• Anomalies in different dimensions
• Gauge anomalies and gravitational anomalies
• Chiral anomalies and global anomalies

All come from one master polynomial in each case. Knowing the polynomial determines the anomaly in the boundary theory.

Example: 4D Chiral Gauge Anomaly

The 6-form is $I_6 = \frac{1}{6}\text{tr}(F^3)$ (with appropriate normalization). Descent gives:

$\omega_5 = \text{tr}(AF^2 - \tfrac{1}{2}A^3F + \tfrac{1}{10}A^5)$

Wait, let me be more careful. The Chern-Simons 5-form:

$\omega_5 = \text{tr}(AdAdA + \tfrac{3}{2}A^3dA + \tfrac{3}{5}A^5)$

Its gauge variation under $A \to A + d\epsilon$ is:

$\delta\omega_5 = d[\text{tr}(\epsilon\, dAdA - \ldots)]$

This gives the 4D anomaly 4-form $\omega^{(1)}_4 \sim \text{tr}(\epsilon dA dA)$.

The point is: from one 6-form, we get the 4D anomaly unambiguously.

Gravitational Analog

Replace $F$ with the curvature 2-form $R = d\Gamma + \Gamma^2$ (with connection $\Gamma$). Analog polynomials give gravitational anomalies.

In $4k$ dimensions: no gravitational anomaly (only in $4k + 2$). So no 4D gravitational anomaly. But 2D and 6D have gravitational anomalies.

Index Theorem Connection

The Atiyah-Singer index theorem states:

$\text{index}(\slashed D) = \int_{\text{bundle}}\hat A(R)\text{ch}(F)$

The right side is a characteristic class; a polynomial in $R$ and $F$. The anomaly polynomials are literally these characteristic classes.

So: anomalies are characteristic classes are index theorems. One mathematical object with three physical/mathematical interpretations.

7. The Physical Canonical Example: $\pi^0 \to \gamma\gamma$

The Historical Puzzle

The neutral pion $\pi^0$ decays to two photons: $\pi^0 \to \gamma\gamma$. The branching ratio is essentially 99%, with lifetime $\tau \approx 8.5\times 10^{-17}$ s.

In the 1960s, naive estimates of this rate (using PCAC; Partial Conservation of Axial Current) gave predictions that were off by factors of 3 or more. The rate seemed “too large” for a pseudo-Goldstone boson decay.

The Anomaly Solution

The chiral anomaly saves the day. The axial current matrix element:

$\langle 0|\partial_\mu J^\mu_A|\pi^0\rangle = -f_\pi m_\pi^2 + \text{anomaly contribution}$

The anomaly contribution ($\partial_\mu J^\mu_A = F\tilde F/16\pi^2$ times color/flavor factors) dominates for the on-shell photon matrix element.

The Prediction

Using the anomalous axial current plus PCAC, the $\pi^0\to\gamma\gamma$ amplitude is predicted:

$\mathcal{A}(\pi^0 \to \gamma\gamma) = \frac{\alpha}{\pi f_\pi}\epsilon^{\mu\nu\rho\sigma}\epsilon_\mu^{(1)}\epsilon_\nu^{(2)}k_\rho^{(1)}k_\sigma^{(2)}\cdot N_c\cdot(\text{flavor factor})$

where $N_c = 3$ is the number of colors (essential!) and the flavor factor is $(e_u^2 - e_d^2)/2 = (4/9 - 1/9)/2 = 1/6$.

The Rate

The decay rate:

$\Gamma(\pi^0\to\gamma\gamma) = \frac{\alpha^2 m_\pi^3}{64\pi^3 f_\pi^2}\cdot N_c^2\cdot(\text{flavor factor})^2$

For $N_c = 3$: the predicted rate is:

$\Gamma = 7.7\text{ eV}$

Experimentally measured:

$\Gamma_{\rm exp} = 7.63 \pm 0.16\text{ eV}$

Agreement at the 1% level. Without color ($N_c = 1$), the prediction would be off by a factor of 9.

The Significance

$\pi^0 \to \gamma\gamma$ directly measures the number of colors. This was key evidence for $N_c = 3$ QCD in the 1970s, before gluons and asymptotic freedom were established.

The prediction depends on:

• The anomaly coefficient (pure geometry)
• Quark charges (quark model)
• Number of colors (QCD)

All three inputs together give the correct rate. Any of them wrong, the prediction fails.

Mathematical Structure

The amplitude comes from the chiral anomaly via the Wess-Zumino-Witten (WZW) term. In ChPT, you add this anomalous term to the Lagrangian:

$\mathcal{L}_{\rm WZW} = \frac{N_c}{48\pi^2}\text{tr}[(dUU^{-1})^5]\text{(on the SU(3) group manifold)}$

plus additional pieces involving electromagnetic fields that give the $\pi^0 \to \gamma\gamma$ vertex. The WZW term encodes the chiral anomaly at the level of the effective theory.

8. QCD Vacuum Structure and the $\eta'$ Mass

The $U(1)_A$ Problem Revisited

From document 15: the $U(1)_A$ symmetry is anomalous. The $\eta'$ is heavy (958 MeV, not the ~400 MeV a true Goldstone would give).

The anomaly-related mass:

$m_{\eta'}^2 = m_\eta^2 + \frac{2N_f}{f_\pi^2}\chi_t$

(Witten-Veneziano formula, large-$N$ limit.) Here $\chi_t$ is the topological susceptibility of QCD.

Topological Susceptibility

$\chi_t$ measures the QCD vacuum’s fluctuations in topological charge:

$\chi_t = \lim_{V\to\infty}\frac{\langle Q^2\rangle}{V}$

where $Q = \int d^4x\,\frac{g^2}{32\pi^2}G^a\tilde G^a$ is the topological charge (space-time integral of the gluon topological density).

Lattice QCD: $\chi_t \approx (180 \text{ MeV})^4$.

Plugging into Witten-Veneziano: predicts $m_{\eta'} \approx 850$ MeV, close to measured 958 MeV.

The Strong CP Problem

The $\theta$-term in QCD:

$\mathcal{L}_\theta = \frac{\theta\, g^2}{32\pi^2}G^a_{\mu\nu}\tilde G^{a\mu\nu} = \theta\, q(x)$

(where $q(x)$ is the topological charge density) violates CP. The CP-violating electric dipole moment of the neutron would be:

$d_n \sim \theta\cdot 10^{-16}\text{ e·cm}$

Measurements: $|d_n| < 10^{-26}$ e·cm. Therefore $|\theta| < 10^{-10}$.

Why is $\theta$ so small? This is the strong CP problem.

The Peccei-Quinn Solution

Peccei and Quinn (1977) proposed introducing a new $U(1)_{\rm PQ}$ symmetry, explicitly broken by QCD instantons (the same topological effects that gave $\eta'$ its mass).

The Goldstone of spontaneous breaking is the axion $a(x)$. The axion gets a potential from QCD instantons, dynamically driving $\theta_{\rm eff} = \theta + a/f_a$ to zero.

The axion has a mass:

$m_a^2 = \frac{\chi_t}{f_a^2}$

where $f_a$ is the axion decay constant. For $f_a \sim 10^{12}$ GeV: $m_a \sim 10^{-5}$ eV.

Axion Searches

Axions are being searched for:

• ADMX (Axion Dark Matter eXperiment): resonant cavity searches for galactic axion dark matter
• CAST (CERN Axion Solar Telescope): searches for axions from the Sun
• IAXO (future): improved solar axion search
• Haloscope experiments: various table-top searches

None have found axions yet, but they’ve constrained axion parameters over a wide range.

The strong CP problem remains unsolved experimentally. The axion is the most popular proposed solution, and its discovery would be transformative.

Anomaly Origin of Everything

The $\eta'$ mass, the strong CP problem, the axion; all stem from the $U(1)_A$ anomaly. One quantum effect (the axial current not being conserved due to gauge fluctuations) drives an entire subfield of physics.

9. ‘t Hooft Matching

The Setup

Consider an asymptotically free gauge theory like QCD. At high energies, it’s weakly coupled with fundamental fermions. At low energies, it’s confined with composite hadrons.

Question: The theory has global symmetries (like chiral symmetry for QCD). The anomaly of these currents should be the same in the UV (in terms of fundamental fermions) and the IR (in terms of composite hadrons). How does this match?

‘t Hooft’s Observation (1980)

The anomaly is a low-energy invariant. It must match between different descriptions of the same theory. This is a nontrivial constraint on the IR physics.

Specifically: compute the global anomaly using UV degrees of freedom. Compute it using IR degrees of freedom. The two must agree.

Application to QCD

At high energies: 3 light quark flavors ($u, d, s$), each with color $N_c = 3$. Chiral symmetry $SU(3)_L \times SU(3)_R$.

The $SU(3)_L^3$ anomaly:

$\mathcal{A}_{\rm UV} \propto N_c\cdot\text{tr}(T^aT^bT^c) = 3\cdot\text{tr}(T^aT^bT^c)$

(The $N_c$ comes from counting colors, since each quark contributes once per color.)

At low energies: pions, kaons, eta; but these are Goldstone bosons, not fundamental fermions. How do they carry the anomaly?

Answer: the WZW term in ChPT (document 15). Adding this non-local Lagrangian to ChPT gives the theory the same $SU(3)_L^3$ anomaly coefficient as QCD.

Matching in Detail

The WZW term normalization is fixed to reproduce:

$\mathcal{A}_{\rm IR} = \mathcal{A}_{\rm UV} = 3\cdot\text{tr}(T^aT^bT^c)$

This requires the WZW term’s coefficient to be exactly $N_c$. In particular, if the $\pi^0 \to \gamma\gamma$ rate (which is a direct measurement of this coefficient) is consistent with $N_c = 3$, then the $SU(3)$ chiral anomaly matches between UV and IR.

This is the ‘t Hooft matching: the UV anomaly determines the IR coefficient of the WZW term.

When Matching Fails; Implications

If ‘t Hooft matching fails (e.g., proposed IR physics has wrong anomaly structure), the proposed IR theory must be wrong. This is a strong constraint.

For theories where the IR is partially unknown (e.g., strongly-coupled BSM scenarios), ‘t Hooft matching constrains the possibilities:

1. Confinement with chiral symmetry breaking: Anomalies carried by WZW term of Goldstones.

2. Confinement without chiral symmetry breaking (exotic): Anomalies must be carried by massless bound states. But then these bound states must have specific anomaly coefficients matching the UV.

3. Conformal IR (CFT): Anomalies are CFT data (various coefficients in correlators). Must match UV.

‘T Hooft matching is one of the few handles on non-perturbative physics that doesn’t require explicit calculation; it gives constraints from anomalies alone.

Applications to BSM

Composite Higgs models: The Higgs is a pseudo-Goldstone of some strongly-coupled new physics. The anomaly structure of the UV (some set of new fermions with new charges) must match the IR (composite Higgs + other SM particles). This constrains model-building.

Dark matter in strongly-coupled hidden sectors: If dark matter emerges from confinement in a hidden gauge theory, ‘t Hooft matching restricts the spectrum of bound states.

Confinement conjectures: ‘t Hooft matching plays a role in the Seiberg-Witten solution of $\mathcal{N} = 2$ SUSY, and in various conjectures about confinement in strongly-coupled theories.

10. Gravitational Anomalies

The Setup

Just as gauge fields can have anomalies from chiral fermions, gravity (specifically, general coordinate transformations or local Lorentz transformations) can have anomalies.

A gravitational anomaly means that the energy-momentum tensor $T^{\mu\nu}$ fails to be conserved at the quantum level, or is not symmetric, or (in mixed anomalies) couples anomalously to gauge fields.

When They Exist

Purely gravitational anomalies exist only in $4k + 2$ dimensions: 2, 6, 10, … They vanish in 4D (and in all other even dimensions).

Mixed gravitational-gauge anomalies can exist in various dimensions, including 4D.

The 4D Mixed Anomaly

The relevant anomaly in 4D is $U(1)_Y \times [\text{gravity}]^2$:

$\partial_\mu J^\mu_Y = \frac{1}{384\pi^2}\sum_f Y_f\cdot R_{\mu\nu\rho\sigma}\tilde R^{\mu\nu\rho\sigma}$

where the sum is over all chiral fermions weighted by hypercharge $Y_f$, and $R_{\mu\nu\rho\sigma}$ is the Riemann tensor.

For SM fermions: $\sum Y_f = 0$ across one generation. So this anomaly cancels; consistent with SM being a unified chiral theory.

Had $\sum Y_f$ been nonzero, SM coupling to gravity would be anomalous. The cancellation confirms SM hypercharges.

String Theory and Anomaly Cancellation in Higher Dimensions

In 10D superstring theories, both gauge and gravitational anomalies exist and must cancel. The Green-Schwarz mechanism (1984) shows how anomalies in 10D heterotic strings cancel through a combination of:

• Fermion contributions from the matter content
• New terms from the Kalb-Ramond 2-form field

This was one of the most important discoveries in string theory; it showed which gauge groups ($SO(32)$ or $E_8\times E_8$) give consistent 10D strings.

Conformal Anomaly (Trace Anomaly)

A related phenomenon: the energy-momentum tensor is not traceless in the quantum theory, even if the classical theory is scale-invariant (like a CFT at the classical level).

$T^\mu_\mu = c\cdot R^2 + \text{other curvature invariants}$

The coefficient $c$ is called the conformal anomaly or trace anomaly. It’s topological; independent of specific field content details, depending only on central charges.

In 2D: $c$ is the Virasoro central charge (like for strings).

In 4D: there are two independent coefficients $c$ and $a$. The $a$-theorem states that $a$ decreases under RG flow (Zamolodchikov in 2D, Komargodski-Schwimmer in 4D). This constrains possible IR endpoints of RG flows.

11. Anomaly Inflow

The Setup

Consider a theory in $d$ dimensions. Its boundary is a $(d-1)$-dimensional surface. Can a theory on the boundary have an anomaly?

Yes; but the anomaly must be matched by some bulk contribution from the $d$-dimensional theory. This is anomaly inflow.

The Mechanism

If the boundary theory has a gauge anomaly, the bulk Chern-Simons term (the object from the descent equations, section 6) provides the counterbalancing contribution. The full system (bulk + boundary) is anomaly-free.

Schematically:

$\partial_\mu J^\mu|_{\rm boundary} = -\frac{\delta S_{\rm bulk}}{\delta A_\mu}\bigg|_{\rm boundary}$

The boundary anomaly is cancelled by variation of the bulk Chern-Simons action.

Physical Examples

Fractional Quantum Hall Effect. The 2D boundary of a 3D bulk “vacuum” has chiral fermions with an anomaly. The bulk Chern-Simons term provides the inflow that cancels this anomaly. The Hall conductance is exactly quantized as a result.

Topological Insulators. 3D bulk topological insulators have 2D surface states with anomalies. The bulk $\theta$-term provides the matching inflow.

Weyl Semimetals. 3D materials with Weyl fermions at Fermi level have surface Fermi arcs. The bulk topological terms (related to the 3D $\theta$-term) are the inflow partner.

D-branes in String Theory. Open strings end on D-branes. Anomalies on the D-brane (from massless open string states) are matched by anomaly inflow from the bulk closed-string RR fields.

Why This Is Important

Anomaly inflow shows that anomalies aren’t isolated; they’re communal. A theory with an anomaly doesn’t stand alone; it must be coupled to something else (the bulk) that provides the inflow.

This has deep implications:

• Topological phases of matter are characterized by their boundary anomalies (they’re the IR manifestation of bulk topology)
• SPT phases (symmetry-protected topological) are classified by anomaly structures at their boundaries
• Dualities between different-dimensional theories often match anomalies via inflow

Anomaly inflow is one of the deepest connections between gauge theory, topology, and condensed matter physics.

The Inflow Equation

For a 4D bulk with Chern-Simons term $\omega_5$ and 3D boundary:

$\delta\int_{M_4}\omega_5 = \int_{\partial M_4 = M_3}\omega^{(1)}_3$

where $\omega^{(1)}_3$ is the 3D anomaly 3-form; same one that would appear as the anomaly of a 3D theory.

The 4D bulk “sees” the boundary anomaly and cancels it. The 4D CS coefficient determines the boundary anomaly coefficient.

12. Global Anomalies and Witten’s $SU(2)$

Perturbative vs. Global Gauge Anomalies

Perturbative gauge anomalies: come from infinitesimal gauge transformations. Check using triangle diagrams.

Global (non-perturbative) gauge anomalies: come from gauge transformations not continuously connected to the identity. Check requires examining the gauge group’s topology.

Witten’s $SU(2)$ Anomaly

Discovery (Witten, 1982): An $SU(2)$ gauge theory with an odd number of left-handed doublets is inconsistent, not from any perturbative anomaly (those vanish in $SU(2)$), but from a global topological obstruction.

The Argument

$SU(2)$ gauge transformations can be continuously deformed; except those corresponding to non-trivial elements of $\pi_4(SU(2)) = \mathbb{Z}_2$.

For an $SU(2)$ with an odd number of chiral doublets, summing over topological sectors gives +1 and -1 contributions that cancel, producing zero partition function. The theory is ill-defined.

For an even number: contributions all have the same sign, partition function is nonzero, theory is consistent.

Standard Model Avoids This

The Standard Model has:

• 3 generations of chiral doublets in $L_L = (\nu, e)_L$
• 3 generations × 3 colors = 9 in $Q_L = (u, d)_L$

Total left-handed doublets per generation: $1 + 3 = 4$. Per 3 generations: $12$. Even. SM avoids the Witten anomaly. ✓

This is another constraint on the SM particle content: you can’t just have one quark doublet or one lepton doublet; you need even counts.

Why This Example Is Special

The Witten $SU(2)$ anomaly is:

• Non-perturbative (invisible in triangle diagrams)
• Topological (depends on $\pi_4$ of gauge group)
• A consistency constraint (violating it makes theory ill-defined)

Only a few known global gauge anomalies. Most theories with anomaly structure have perturbative anomalies. Witten’s $SU(2)$ is the premier example where topology rather than perturbation theory is key.

Other Global Anomalies

Various other global anomalies exist in different contexts:

• $\text{Spin}$ structure anomalies (from Wess-Zumino-type terms in non-simply-connected groups)
• $\mathbb{Z}_2$ anomalies in time-reversal (relevant for topological insulators)
• ‘t Hooft anomalies on worldlines of certain gauge theories

Global anomalies are where non-perturbative topology directly determines consistency of quantum field theories.

13. Anomalies in Condensed Matter

The Connection

Anomalies aren’t just for particle physics. They appear throughout condensed matter in surprising ways:

• Quantum Hall Effect: precise quantization of Hall conductance comes from anomaly inflow
• Topological Insulators: surface states are “half” a 2D theory, with anomalies cancelled by the bulk
• Weyl Semimetals: chiral anomaly manifests as negative magnetoresistance in magnetic fields
• 1D Systems: Luttinger liquids have chiral anomalies
• Topological Superconductors: Majorana boundary modes reflect bulk-boundary anomaly matching

The Quantum Hall Effect

In a 2D electron gas in a magnetic field, the Hall conductance is precisely quantized:

$\sigma_{xy} = \frac{e^2}{h}\nu$

where $\nu$ is an integer (or certain fractions, for fractional QHE).

The quantization comes from topology: $\nu$ is a Chern number of the bundle of occupied states over the Brillouin zone. Anomaly inflow ensures the bulk topological term produces exactly the right edge response.

Chiral Magnetic Effect

In a Weyl semimetal (3D material with Weyl fermions at Fermi surface), applying parallel electric and magnetic fields generates an anomalous current:

$\vec j_{\rm CME} = \frac{e^2}{2\pi^2}\mu_5\vec B$

where $\mu_5$ is the chiral chemical potential (imbalance between left and right chirality populations). This is the chiral magnetic effect.

It’s a direct manifestation of the chiral anomaly. The rate of chirality non-conservation is proportional to $\vec E\cdot\vec B$:

$\partial_t(n_R - n_L) = \frac{e^2}{2\pi^2}\vec E\cdot\vec B$

Same as the ABJ anomaly in QED. But in a material!

Topological Invariants

Modern condensed matter physics classifies phases of matter by topological invariants, which are often anomaly coefficients:

• Chern number ($\mathbb{Z}$): quantum Hall
• $\mathbb{Z}_2$ invariant: topological insulators with time-reversal symmetry
• Higher K-theory classes: SPT phases in higher dimensions

The “ten-fold way” classification of topological insulators is a classification of bulk anomaly structures and their boundary manifestations.

Emergent Gauge Theories

Strongly-correlated systems can have emergent gauge symmetries at low energies. These emergent gauge theories have their own anomaly structures, connecting to physical observables in the material.

Example: $\mathbb{Z}_2$ gauge theory emerges in Kitaev’s honeycomb model. Its anomaly structure determines the nature of the edge modes.

14. Appendix: Anomaly Formulas Reference

Canonical Anomaly Formulas

ABJ chiral anomaly (QED): $\partial_\mu J^\mu_A = \frac{e^2}{8\pi^2}F\tilde F = \frac{e^2}{16\pi^2}\epsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma}$

Non-abelian chiral anomaly: $\partial_\mu J^\mu_A = \frac{g^2}{32\pi^2}\epsilon^{\mu\nu\rho\sigma}\text{tr}[F_{\mu\nu}F_{\rho\sigma}]$

4D chiral anomaly (per Weyl fermion in fund. of gauge $G$): $\partial_\mu J^\mu_5 = \frac{1}{32\pi^2}\epsilon^{\mu\nu\rho\sigma}\text{tr}(F_{\mu\nu}F_{\rho\sigma})$

6D anomaly: $\propto \epsilon^{\mu_1\ldots\mu_6}\text{tr}(F^3) + \text{mixed}$

Gauge anomaly cancellation: $\text{tr}[T^a\{T^b, T^c\}] = 0 \text{ (summed over chiral fermions)}$

Numerical Coefficients

For $SU(N)$ fundamental: $\text{tr}(T^aT^b) = \tfrac{1}{2}\delta^{ab}$, $\text{tr}(T^aT^bT^c) = \tfrac{1}{4}d^{abc}$

Anomaly coefficient for $SU(N)$ fundamentals: $A(\text{fund}) = 1$, for anti-fund $= -1$, for adjoint $= 0$ (if group is $SU(N)$ for $N \geq 3$).

Topological Susceptibility of QCD

$\chi_t \approx (180 \text{ MeV})^4$

(From lattice QCD.) Enters Witten-Veneziano formula for $\eta'$ mass.

$\pi^0 \to \gamma\gamma$ Rate

$\Gamma(\pi^0\to\gamma\gamma) = \frac{\alpha^2 m_\pi^3}{64\pi^3 f_\pi^2}\cdot N_c^2\cdot(\text{charge factor})^2$

For $N_c = 3$: $\Gamma \approx 7.7$ eV.

Further Reading

• Bertlmann, Anomalies in Quantum Field Theory: comprehensive textbook
• Fujikawa & Suzuki, Path Integrals and Quantum Anomalies: original path integral approach
• Harvey, TASI 2003 Lectures on Anomalies: clear pedagogical introduction
• Nakahara, Geometry, Topology and Physics: differential-geometric perspective
• Witten, Fermion Path Integrals and Topological Phases (2016 Rev. Mod. Phys. article): modern view via SPT phases
• Bilal, Lectures on Anomalies: detailed review

Problems

1. Derive the chiral anomaly in QED using the triangle diagram approach, being careful about regularization.

2. Using Fujikawa’s method, compute the anomaly for an axial $U(1)$ transformation in a theory with $n_f$ massless Dirac fermions.

3. For a hypothetical theory of chiral fermions in $SU(N)$ gauge group, write down the gauge anomaly cancellation conditions. For $N = 5$, what fermion content is anomaly-free?

4. Show that the SM with one generation has all six anomaly cancellation conditions satisfied. Hint: use the hypercharge assignments.

5. Derive the Wess-Zumino-Witten term in ChPT for $SU(3)$ Goldstones, and use it to compute $\pi^0 \to \gamma\gamma$ directly.

6. For a 2D chiral boson, identify the anomaly structure. Show how this connects to quantum Hall physics.

7. Compute the gravitational anomaly (mixed with $U(1)$) in 4D and verify it vanishes for SM fermion content.

Closing Note

Anomalies are the deepest non-perturbative structure in quantum field theory. Classical symmetries that quantum mechanics refuses to respect; and the failure is exact, topological, and physically meaningful.

What Makes Anomalies Special

• Exactness: not approximations, not perturbative. One-loop exact.
• Topological: invariant under continuous deformations. Universal.
• Matching: the same anomaly in UV and IR. Constrains IR physics.
• Predictive: $\pi^0 \to \gamma\gamma$ rate is calculated exactly.
• Constraint: gauge anomalies must cancel. Shapes the SM.
• Connection to Topology: anomalies = characteristic classes = index theorems.
• Universal: appear in particle physics, condensed matter, and string theory.

The Deepest Connection

Anomalies connect:

• Gauge theory and topology
• UV and IR physics (via matching)
• Classical and quantum physics (via broken symmetries)
• Particle physics and condensed matter (via inflow)
• Physics and pure mathematics (index theorems)

When physicists in the 1960s and 70s understood anomalies, it changed our view of quantum field theory. Non-perturbative effects were no longer mysterious; they had structure, they matched, they constrained. Modern QFT is unthinkable without anomaly analysis.

What’s Still Open

Anomalies at nonzero temperature (chiral magnetic effect, anomaly-induced transport). Higher-dimensional anomalies in exotic spacetime dimensions. Global anomalies in generic discrete groups. Connection between anomalies and black hole information. Generalizations to higher-form symmetries (Kapustin-Seiberg, Gaiotto).

This is an active field. The framework is powerful and more applications keep emerging.

Where to Go Next

You now have:

• Full EFT sequence (documents 14-16)
• Thermal field theory (document 13)
• Anomalies in depth (this document)

Remaining menu options:

• Option D: Non-perturbative QFT (1-2 docs); instantons, solitons, monopoles, $\theta$-vacuum, large-$N$. This naturally connects to what you just learned; QCD topology and instantons are central to the $U(1)_A$ anomaly story.

• Option E: Beyond Standard Model (5-8 docs); SUSY, strings, holography, quantum gravity. The biggest and most speculative extension.

Or perhaps directions suggested by this anomaly document:

• Topological phases of matter (SPT, topological order)
• Higher-form symmetries and generalized anomalies
• Anomaly matching for specific strongly-coupled BSM theories
• Conformal anomalies and the $a$-theorem

Each is its own research direction. You’re well-equipped for any of them.

Let me know when you want to continue; or if you want to take a break. The physics is here when you’re ready.