12.18.05 · quantum / gauge-and-symmetry

The chiral (Adler-Bell-Jackiw) anomaly from the triangle diagram

shipped3 tiersLean: none

Anchor (Master): Weinberg, S., *The Quantum Theory of Fields, Vol. 2: Modern Applications* (Cambridge, 1996), Ch. 22; Bertlmann, R. A., *Anomalies in Quantum Field Theory* (Oxford, 1996), Chs. 2-4; Adler, S. L., *Phys. Rev.* 177 (1969), 2426

Intuition Beginner

Some quantities in physics are supposed to be perfectly conserved: whatever you start with, you end with. A current that obeys such a rule is like water in a sealed pipe, the same amount flowing out as flowed in. The handedness of certain particles, called their chirality, looks at first like one of these sealed quantities. Spin a massless particle one way and it stays left-handed; spin it the other way and it stays right-handed. Classically nothing converts one into the other.

The surprise is that the quantum world quietly drills a hole in this pipe. When you account for the cloud of particle pairs that quantum theory says is always flickering in and out of the vacuum, the handedness count stops being exactly conserved. A precise, unchangeable trickle leaks out, and the size of the leak is set by the electric and magnetic fields the particle moves through. You cannot patch the hole by being more careful. The leak is built into the theory.

Why care? Because this leak explains a real measurement: how fast a neutral pion decays into two photons. The plain theory predicts almost nothing; the leak predicts the right number.

Visual Beginner

Picture three particle lines meeting in a triangle, like a closed loop with three corners. Two corners attach to photons, the carriers of light. The third corner attaches to the handedness current you hoped was sealed. This little triangle is the smallest quantum process that lets the photons feel the handedness, and it is exactly where the leak comes from.

The triangle is symmetric: you could try to make it respect the handedness rule, or you could try to make it respect the rule that electric charge is conserved, but you cannot do both at once. Physicists keep the charge rule, because charge conservation is sacred, and let the handedness rule take the hit. The amount it loses is the leak, and it is the same no matter how you set up the calculation.

Worked example Beginner

Suppose a background field is switched on for a moment, and the handedness rule would have kept the left-minus-right particle count fixed at its starting value. Start with 4 left-handed and 4 right-handed particles, so left-minus-right is 0.

Step 1. Fix the leak rule. The quantum leak says the left-minus-right count changes by a fixed amount tied to the field, and suppose the field is set so the change is exactly 2.

Step 2. Apply the field. The count is forced to shift by 2, so left-minus-right goes from 0 to 2.

Step 3. Read off the new populations. One consistent outcome is 5 left-handed and 3 right-handed: 5 − 3 = 2, matching.

Step 4. Double the field. If you double the field strength in the right way, the forced change becomes 4 instead of 2, giving for example 6 left-handed and 2 right-handed.

What this tells us: the handedness imbalance is not chosen by hand; the field forces a definite whole-number mismatch. That whole number is the same object that, in deeper treatments, counts solutions of an equation and connects this leak to the geometry of the field.

Check your understanding Beginner

Formal definition Intermediate+

Let $ψ$ be a Dirac field 12.05.05 coupled to an abelian gauge field $A_{μ}$ with charge $e$ and field strength $F_{μν} = \partial_{μ} A_{ν} - \partial_{ν} A_{μ}$ . Metric signature is mostly-minus, $γ^{5} = i γ^{0} γ^{1} γ^{2} γ^{3}$ , and $ε^{0123} = + 1$ . The theory has two Noether currents: the vector current $j^{μ} = \overset{ˉ}{ψ} γ^{μ} ψ$ and the axial (chiral) current $j_{5}^{μ} = \overset{ˉ}{ψ} γ^{μ} γ^{5} ψ$ .

Classically, the equations of motion give $$ \partial_\mu j^\mu = 0, \qquad \partial_\mu j^\mu_5 = 2im,\bar\psi\gamma^5\psi, $$ so the vector current is conserved and the axial current is conserved in the massless limit $m \to 0$ . The classical axial symmetry is the chiral rotation $ψ \mapsto e^{i α γ^{5}} ψ$ , under which left- and right-handed components $ψ_{L, R} = \frac{1}{2} (1 \mp γ^{5}) ψ$ rotate by opposite phases.

The Adler-Bell-Jackiw anomaly is the statement that this last conservation law fails at one loop. With the regularisation chosen to preserve the vector Ward identity (gauge invariance), the quantum axial current obeys $$ \partial_\mu \langle j^\mu_5\rangle = 2im,\bar\psi\gamma^5\psi ;+; \frac{e^2}{16\pi^2},\varepsilon^{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma}. $$ The second term is the anomaly. It is local, finite, independent of the regulator details, and cannot be removed by any local counterterm without spoiling $\partial_{μ} j^{μ} = 0$ .

The object computing it is the triangle (AVV) amplitude: the one-loop fermion-loop three-point function with one axial vertex $γ^{μ} γ^{5}$ and two vector vertices $γ^{ν}, γ^{ρ}$ , $$ T^{\mu\nu\rho}(k_1,k_2) = \int!\frac{d^4\ell}{(2\pi)^4};\operatorname{tr}!\Big[\gamma^\mu\gamma^5,\frac{1}{\slashed{\ell}-m},\gamma^\nu,\frac{1}{\slashed{\ell}-\slashed{k}_1-m},\gamma^\rho,\frac{1}{\slashed{\ell}-\slashed{k}_1-\slashed{k}_2-m}\Big] + (k_1\nu \leftrightarrow k_2\rho). $$ This integral is linearly divergent. Its value is finite once a regulator is fixed, but a linearly divergent integral has a finite, routing-dependent ambiguity: shifting the loop momentum $ℓ \to ℓ + a$ changes $T^{μν ρ}$ by a surface term. That ambiguity is exactly enough to move the anomaly between the vector and the axial Ward identities, and no choice sets both to zero.

Counterexamples to common slips

The anomaly is not a divergence one renormalises away. The integral is divergent, but the anomalous divergence of the current is finite and physical. Subtracting a local counterterm can shift the anomaly from one Ward identity to another, never delete it.
The anomaly does not vanish as $m \to 0$ ; it is $m$ -independent. The naive mass term $2 im \overset{ˉ}{ψ} γ^{5} ψ$ does vanish, but the $F \tilde{F}$ piece survives, which is why the massless theory still has a broken axial symmetry.
A nonzero axial divergence is not a failure of the classical symmetry. Classically the symmetry is exact; the breaking is a property of the regularised measure, equivalently of the triangle's surface term, and exists only at the quantum level.

Key derivation Intermediate+

Theorem (the AVV anomaly; vector-conserving regularisation). In massless spinor electrodynamics the triangle amplitude cannot satisfy both Ward identities $k_{1 ν} T^{μν ρ} = k_{2 ρ} T^{μν ρ} = 0$ (vector conservation) and $(k_{1} + k_{2})_{μ} T^{μν ρ} = 0$ (axial conservation). Imposing the vector identities fixes $$ (k_1+k_2)\mu,T^{\mu\nu\rho}(k_1,k_2) = -\frac{1}{2\pi^2},\varepsilon^{\nu\rho\alpha\beta}k{1\alpha}k_{2\beta}, $$ which in position space is the anomalous divergence $\partial_{μ} j_{5}^{μ} = \frac{e ^{2}}{16 π ^{2}} ε^{μν ρ σ} F_{μν} F_{ρ σ}$ .

Proof. The two external vector legs carry momenta $k_{1}, k_{2}$ and the axial leg carries $q = k_{1} + k_{2}$ . Contract the bare amplitude with $q_{μ}$ and use the operator identity $\slashed q γ^{5} = (\slashed ℓ - m) γ^{5} + γ^{5} (\slashed ℓ - \slashed q - m) + 2 m γ^{5}$ , where $ℓ$ and $ℓ - q$ are the propagator momenta adjacent to the axial vertex. The first two terms cancel the neighbouring propagator denominators, collapsing the triangle to two formally distinct difference-of-bubbles integrals; the $2 m γ^{5}$ term reproduces the naive mass divergence. Were the integrals convergent, the two bubble pieces would cancel by a shift of the loop momentum and one would conclude $q_{μ} T^{μν ρ} = 2 m (\dots)$ , the naive Ward identity.

The integrals are not convergent. Each bubble is linearly divergent, so the shift $ℓ \to ℓ + a$ relating them is not free: it produces a finite surface term $$ \Delta^\mu(a) = \int!\frac{d^4\ell}{(2\pi)^4}\big[f^\mu(\ell+a) - f^\mu(\ell)\big] = \frac{i}{32\pi^2},a^\mu ,(\text{angular factor}), $$ evaluated by Gauss's theorem on the three-sphere at infinity, $\int d^{4} ℓ \partial_{α} f = \oint_{S^{3}} d S_{α} f$ , with $f \sim ℓ / ℓ^{4}$ giving a nonzero flux. Carrying the antisymmetric trace $tr (γ^{5} γ^{μ} γ^{ν} γ^{ρ} γ^{σ}) = - 4 i ε^{μν ρ σ}$ through, the surface term is proportional to $ε^{ν ρ α β} k_{1 α} k_{2 β}$ .

The undetermined piece is the loop-momentum routing $a = μ k_{1} + ν k_{2}$ left free by the linear divergence. The three Ward identities — one axial and two vector — impose three linear conditions on the single free constant. They are mutually inconsistent: the sum of the anomalous coefficients across the three identities is fixed at $- \frac{1}{2 π ^{2}} ε^{ν ρ α β} k_{1 α} k_{2 β}$ and cannot be made to vanish in all three. Choosing the routing that sets both vector identities to zero (Bose symmetry between the two photons plus gauge invariance fixes it uniquely) loads the entire fixed coefficient onto the axial identity: $$ q_\mu T^{\mu\nu\rho} = 2m(\cdots) - \frac{1}{2\pi^2},\varepsilon^{\nu\rho\alpha\beta}k_{1\alpha}k_{2\beta}. $$ Translating the momentum-space contraction back to position space, $i k_{1 α} \to \partial_{α} A$ on each photon leg, gives $\partial_{μ} j_{5}^{μ} = 2 im \overset{ˉ}{ψ} γ^{5} ψ + \frac{e ^{2}}{16 π ^{2}} ε^{μν ρ σ} F_{μν} F_{ρ σ}$ ^{[tong standardmodel]}. $□$

Bridge. This computation builds toward the path-integral and index-theorem readings of the same coefficient, and it appears again wherever chiral fermions are gauged. The foundational reason the symmetry breaks is that the regularised fermion measure is not chiral-invariant, and that is the central content connecting the triangle's surface term to Fujikawa's Jacobian and to the Atiyah-Singer index density of 03.07.32. Putting these together, the $F \tilde{F}$ on the right-hand side is the same density that, integrated, gives the instanton number weighting the $θ$ -vacuum of 12.18.04; the chiral rotation that shifts $θ$ there is exactly the anomalous transformation derived here. The single coefficient $e^{2} /16 π^{2}$ is therefore shared across the triangle, the measure Jacobian, and the index, and is fixed to one loop by the Adler-Bardeen theorem proved below.

Exercises Intermediate+

Exercise 4 (medium, symbolic).

Show that the anomaly coefficient of the vector-vector-axial triangle is proportional to $tr ({T^{a}, T^{b}} T_{5}^{c})$ for non-abelian charges, and explain why an abelian axial current with abelian vector charges gives the simple coefficient of the text.

Hint

The two vector vertices contribute $T^{a}, T^{b}$ symmetrised by Bose statistics; the axial vertex contributes the axial generator.

Answer

Bose symmetry between the two identical vector legs symmetrises their generators, giving $tr ({T^{a}, T^{b}} T_{5}^{c}) = \frac{1}{2} d_{axial}^{ab c}$ , the symmetric invariant. For a single abelian gauge field and a single Dirac fermion all generators are the unit charge, the trace reduces to the charge factor $e^{2}$ , and the group-theoretic structure collapses to the numerical coefficient $e^{2} /16 π^{2}$ . The non-abelian case is the structure that controls gauge-anomaly cancellation in 03.07.32.

Exercise 5 (medium, derivation).

Using the trace identity $tr (γ^{5} γ^{μ} γ^{ν} γ^{ρ} γ^{σ}) = - 4 i ε^{μν ρ σ}$ , explain why the triangle anomaly is necessarily proportional to $ε^{μν ρ σ}$ and hence parity-odd, while a vector-vector-vector triangle in the same theory has a vanishing anomaly.

Hint

The single $γ^{5}$ at the axial vertex is what produces the $ε$ -tensor; without it the trace is parity-even and the surface terms cancel by symmetry.

Answer

The axial vertex inserts one $γ^{5}$ . The only nonvanishing trace of $γ^{5}$ with four other gammas is the totally antisymmetric $ε$ -tensor, so any anomalous piece must carry $ε^{μν ρ σ}$ and is parity-odd. A VVV triangle has no $γ^{5}$ , its trace is parity-even, and the two routings related by charge-conjugation (Furry's theorem) cancel, leaving the vector Ward identity exact. The anomaly is intrinsically tied to the chirality-distinguishing $γ^{5}$ .

Exercise 6 (hard, derivation).

Carry out the Fujikawa computation schematically: under $ψ \mapsto e^{i α (x) γ^{5}} ψ$ the measure changes by $exp (- 2 i \int α J)$ with $J (x) = lim_{M \to \infty} tr [γ^{5} e^{- \slashed D^{2} / M^{2}}] (x, x)$ . Identify which term of the heat-kernel expansion of $\slashed D^{2}$ survives the $γ^{5}$ trace and gives the anomaly.

Hint

$\slashed D^{2} = D^{2} + \frac{e}{2} σ^{μν} F_{μν}$ ; the $γ^{5}$ trace needs four gamma matrices, so two powers of $σ F$ are required.

Answer

Write $\slashed D^{2} = D^{2} + \frac{e}{2} σ^{μν} F_{μν}$ (Lichnerowicz). The trace $tr γ^{5} (\dots)$ vanishes unless four gamma matrices appear, so only the term with two powers of $\frac{e}{2} σ F$ in the expansion of $e^{- \slashed D^{2} / M^{2}}$ survives. The Gaussian momentum integral against $e^{- ℓ^{2} / M^{2}}$ converts the $M$ -dependence into a finite limit, leaving $J = \frac{1}{32 π ^{2}} ε^{μν ρ σ} F_{μν} F_{ρ σ}$ from $tr (γ^{5} σ^{μν} σ^{ρ σ}) = - 4 i ε^{μν ρ σ}$ . The Jacobian density equals the triangle anomaly: $\partial_{μ} j_{5}^{μ} = 2 J = \frac{e ^{2}}{16 π ^{2}} F \tilde{F}$ .

Lean formalization Intermediate+

lean_status: none — Mathlib has no regularised fermion-loop amplitude, no Dirac operator on Minkowski space, and no heat-kernel regularisation of operator traces, so neither the triangle integral nor the Fujikawa Jacobian can be stated.

-- Pseudocode only: the anomaly is not formalisable in current Mathlib.
axiom Spacetime : Type*
axiom AxialCurrent  : Spacetime → ℝ          -- ∂_μ j^μ_5 as a density
axiom FieldStrength : Spacetime → Type*       -- F_{μν}
axiom FdualF : (Spacetime → FieldStrength) → (Spacetime → ℝ)  -- ε F F density

-- The Adler-Bell-Jackiw statement, schematically:
axiom abj_anomaly (m : ℝ) (A : Spacetime → FieldStrength) (x : Spacetime) :
    AxialCurrent x = (2 * m) * (0 : ℝ) /- pseudoscalar bilinear -/
      + (1 / (16 * Real.pi^2)) * FdualF A x

-- Adler-Bardeen: the coefficient is loop-order independent (no formal content here)
axiom adler_bardeen : True

The missing formalization work is the regularised one-loop trace of $γ^{5} e^{- \slashed D^{2} / M^{2}}$ and its heat-kernel (Seeley-DeWitt) coefficient, the surface-term ambiguity of the linearly divergent triangle, and the equality of the resulting density with the Atiyah-Singer index density itemised in 03.07.32.

Advanced results Master

The PCAC puzzle and $π^{0} \to γ γ$ . Before the anomaly, the partially-conserved-axial-current hypothesis modelled $\partial_{μ} j_{5}^{μ}$ as proportional to the pion field, $\partial_{μ} j_{5}^{μ} = f_{π} m_{π}^{2} π^{0}$ . Combined with the soft-pion theorems this predicted a $π^{0} \to γ γ$ amplitude vanishing as $m_{π} \to 0$ , in flat contradiction with the observed decay rate. Bell and Jackiw, and Adler, resolved the puzzle: the triangle adds the anomalous term, and the physical amplitude is fixed by the anomaly coefficient rather than by PCAC alone ^{[tong standardmodel]}. The decay width is $$ \Gamma(\pi^0\to\gamma\gamma) = \frac{\alpha^2 m_\pi^3}{64\pi^3 f_\pi^2},\Big(N_c\sum_q Q_q^2\Big)^2, $$ and quantitative agreement with experiment requires $N_{c} \sum_{q} Q_{q}^{2} = N_{c} (Q_{u}^{2} - Q_{d}^{2}) = 1$ , hence $N_{c} = 3$ . The neutral pion decay is the cleanest low-energy measurement of the colour number, and historically one of the first quantitative confirmations of the anomaly itself.

The Adler-Bardeen nonrenormalisation theorem. Adler and Bardeen proved that the anomaly coefficient is saturated at one loop: to all orders in perturbation theory, no diagram beyond the single triangle corrects the $F \tilde{F}$ term ^{[Adler-Bardeen 1969]}. The argument isolates the anomaly as a property of the superficially-divergent triangle subgraph; higher-loop dressings are convergent and respect the naive Ward identity, so they cannot generate new anomalous pieces. The theorem is what makes the anomaly a precision tool: the coefficient computed from one diagram is the exact coefficient, which is also why it can equal a topological integer.

Fujikawa's measure derivation. In the Euclidean path integral the fields $ψ, \overset{ˉ}{ψ}$ are independent Grassmann variables, and a chiral rotation $ψ \mapsto e^{i α γ^{5}} ψ$ is a change of integration variables. The Jacobian is not unity. Expanding in eigenmodes $\slashed D φ_{n} = λ_{n} φ_{n}$ , the rotation multiplies the measure by $\prod_{n} e^{- i α ⟨ φ_{n} ∣2 γ^{5} ∣ φ_{n} ⟩}$ , an ill-defined product regulated by the Gaussian cutoff $e^{- λ_{n}^{2} / M^{2}}$ . The regulated exponent is $- 2 i α lim_{M \to \infty} Tr γ^{5} e^{- \slashed D^{2} / M^{2}}$ , and the Lichnerowicz expansion $\slashed D^{2} = D^{2} + \frac{e}{2} σ \cdot F$ makes the surviving heat-kernel coefficient exactly $\frac{1}{32 π ^{2}} εF F$ ^{[tong gaugetheory]}. The anomalous Ward identity is the statement that the path-integral measure transforms by this density. Fujikawa's derivation is regulator-transparent: the same coefficient that the triangle's surface term produces is here the heat-kernel trace of $γ^{5}$ , which is manifestly the integrand of the Atiyah-Singer index.

The index reading and the unification. The trace $Tr γ^{5} e^{- \slashed D^{2} / M^{2}}$ is independent of $M$ because only zero modes contribute to $Tr γ^{5}$ (nonzero modes pair into chirality doublets), giving $Tr γ^{5} e^{- \slashed D^{2} / M^{2}} = n_{+} - n_{-} = ind (\slashed D)$ . Integrating the anomaly therefore yields $\int \partial_{μ} j_{5}^{μ} = 2 ind (\slashed D)$ , twice the net chirality of Dirac zero modes — the same statement derived geometrically in 03.07.32 from the descent equations and the second Chern form. The triangle, the measure Jacobian, and the index are three computations of one coefficient.

Covariant versus consistent anomaly. The $\frac{e ^{2}}{16 π ^{2}} F \tilde{F}$ form derived here is the covariant anomaly, the one that transforms covariantly under gauge transformations and is symmetric in all three legs. The consistent anomaly of 03.07.32, satisfying the Wess-Zumino condition, differs by a local Bardeen-Zumino counterterm; the two coincide for the abelian single-current case but differ in the non-abelian theory. The triangle in its symmetric (Bose-symmetric) regularisation computes the covariant form, and the choice of which Ward identity to violate is precisely the choice between these two bases.

Synthesis. The foundational fact is that no regularisation of the fermion loop preserves both the vector and the axial Ward identity, and this is the central content that ties the triangle's routing ambiguity to Fujikawa's measure Jacobian and to the index density of 03.07.32. The single coefficient $e^{2} /16 π^{2}$ does several jobs at once: it is the surface term of the linearly divergent AVV triangle; it is the heat-kernel trace $Tr γ^{5} e^{- \slashed D^{2} / M^{2}}$ of the rotated measure; and integrated it is twice the Atiyah-Singer index $n_{+} - n_{-}$ . This builds toward the topological reading in which the integrated anomaly is an integer, and it appears again in 12.18.04, where the same $F \tilde{F}$ is the instanton density weighting the $θ$ -vacuum and the anomalous chiral rotation is what shifts $θ$ . The Adler-Bardeen theorem protects the coefficient at one loop, consistent with its equality to a discrete index. The phenomenology confirms the coefficient: $π^{0} \to γ γ$ measures $N_{c} \sum_{q} Q_{q}^{2} = 1$ , the cleanest determination of three colours. Putting these together, the abelian ABJ result $\partial_{μ} j_{5}^{μ} = \frac{e ^{2}}{16 π ^{2}} F \tilde{F}$ generalises to the non-abelian descent tower of 03.07.32, whose top rung is the same characteristic class.

Full proof set Master

Proposition (classical axial divergence). For a Dirac field obeying $(i \slashed \partial - m) ψ = 0$ , $\partial_{μ} j_{5}^{μ} = 2 im \overset{ˉ}{ψ} γ^{5} ψ$ .

Proof. Differentiate $j_{5}^{μ} = \overset{ˉ}{ψ} γ^{μ} γ^{5} ψ$ : $$ \partial_\mu j^\mu_5 = (\partial_\mu\bar\psi)\gamma^\mu\gamma^5\psi + \bar\psi\gamma^\mu\gamma^5(\partial_\mu\psi). $$ The equations of motion give $γ^{μ} \partial_{μ} ψ = - im ψ$ and $\partial_{μ} \overset{ˉ}{ψ} γ^{μ} = im \overset{ˉ}{ψ}$ . Substituting and using $γ^{μ} γ^{5} = - γ^{5} γ^{μ}$ in the first term, $$ \partial_\mu j^\mu_5 = (im\bar\psi)\gamma^5\psi - \bar\psi\gamma^5(-im\psi)\cdot(-1)\cdots = 2im,\bar\psi\gamma^5\psi, $$ where the anticommutation collects the two contributions with the same sign. At $m = 0$ the right side vanishes, so the massless axial current is classically conserved. $□$

Proposition (the integrated anomaly is an even integer). On a compactified Euclidean background, $\int d^{4} x \frac{e ^{2}}{16 π ^{2}} ε^{μν ρ σ} F_{μν} F_{ρ σ} = 2 ind (\slashed D) \in 2 Z$ .

Proof. By Fujikawa, the anomaly density equals $2 J (x)$ with $J (x) = lim_{M \to \infty} ⟨ x ∣ tr γ^{5} e^{- \slashed D^{2} / M^{2}} ∣ x ⟩$ . Integrating over $x$ gives $\int J = Tr γ^{5} e^{- \slashed D^{2} / M^{2}}$ . Because $\slashed D$ anticommutes with $γ^{5}$ , every nonzero eigenvalue $λ$ of $\slashed D$ has its $γ^{5}$ -partner at $- λ$ , and these pair off so that nonzero modes contribute zero to $Tr γ^{5} e^{- \slashed D^{2} / M^{2}}$ . Only zero modes survive, giving $n_{+} - n_{-} = ind (\slashed D)$ , independent of $M$ . Thus $\int \frac{e ^{2}}{16 π ^{2}} εF F = 2 (n_{+} - n_{-})$ , an even integer. This matches the second-Chern-number computation of 03.07.32. $□$

Proposition (Ward-identity incompatibility). The bare triangle amplitude $T^{μν ρ}$ cannot simultaneously satisfy $k_{1 ν} T^{μν ρ} = 0$ , $k_{2 ρ} T^{μν ρ} = 0$ , and $(k_{1} + k_{2})_{μ} T^{μν ρ} = 0$ at $m = 0$ .

Proof. Write $T^{μν ρ} (a) = T_{0}^{μν ρ} + S^{μν ρ} (a)$ , where $S$ is the finite surface term generated by the loop-momentum routing $a = α k_{1} + β k_{2}$ permitted by the linear divergence. Contracting with each of the three external momenta gives three linear functionals of $(α, β)$ . Computing the surface integrals, the three contractions sum to a routing-independent constant: $k_{1 ν} T^{μν ρ} + k_{2 ρ} T^{μν ρ} ↾ + (k_{1} + k_{2})_{μ} T^{μν ρ} \supset - \frac{1}{2 π ^{2}} ε (\dots) k_{1} k_{2} \neq = 0$ , the anomaly. Since the total is a fixed nonzero number independent of $(α, β)$ , no choice makes all three contractions vanish. Imposing the two vector identities (forced by gauge invariance and Bose symmetry) determines $(α, β)$ uniquely and places the full constant on the axial identity. $□$

Proposition (one-loop exactness, Adler-Bardeen, statement). The coefficient of the $F \tilde{F}$ anomaly receives no corrections beyond one loop; higher orders renormalise the operators in the Ward identity but not the anomaly coefficient. Stated without full proof here — the complete diagrammatic argument is in Adler-Bardeen 1969 ^{[Adler-Bardeen 1969]}; the topological consistency is the integrality of $ind (\slashed D)$ proved above, which forbids a smooth coupling-dependent correction to an integer-valued integrated anomaly. $□$

Connections Master

Anomalies via descent equations and the Atiyah-Singer index theorem 03.07.32. This unit is the QFT-side, triangle-diagram derivation of the anomaly whose geometric content 03.07.32 develops through the descent equations. The $\frac{e ^{2}}{16 π ^{2}} F \tilde{F}$ computed here from the AVV triangle is the abelian, four-dimensional bottom rung of the descent tower $Str (F^{n + 1})$ of 03.07.32; the integrated anomaly equals $2 ind (\slashed D)$ , the same Dirac index that the descent's top polynomial computes. The covariant anomaly of this unit and the consistent anomaly of 03.07.32 differ by the Bardeen-Zumino counterterm.
Theta-vacua, the vacuum angle, and the strong-CP problem 12.18.04. The anomaly density $F \tilde{F}$ derived here is exactly the integrand of the Pontryagin index that labels the $θ$ -vacua of 12.18.04; the anomalous chiral rotation $ψ \mapsto e^{i α γ^{5}} ψ$ is the transformation that shifts $θ \to θ + α$ , and the fact that a massless quark would make $\overset{ˉ}{θ}$ unobservable is a direct corollary of the anomaly being $m$ -independent. The 't Hooft vertex and the $U (1)_{A}$ resolution of 12.18.04 are the instanton-background incarnation of the same axial-current non-conservation.
Free Dirac (spin-1/2) quantum field 12.05.05. The axial current $j_{5}^{μ} = \overset{ˉ}{ψ} γ^{μ} γ^{5} ψ$ , the chirality projectors $P_{L, R} = \frac{1}{2} (1 \mp γ^{5})$ , and the trace technology $tr (γ^{5} γ^{μ} γ^{ν} γ^{ρ} γ^{σ}) = - 4 i ε^{μν ρ σ}$ that the triangle computation rests on are all introduced in 12.05.05; the anomaly is the quantum statement that the classical conservation law of that current fails.

Historical & philosophical context Master

The anomaly was found in 1969 by Adler, working in spinor electrodynamics, and independently by Bell and Jackiw, who arrived at it through the $π^{0} \to γ γ$ puzzle of the partially-conserved-axial-current hypothesis. Adler computed the axial-vector vertex and showed that the triangle diagram produces a finite, regularisation-independent term proportional to $ε^{μν ρ σ} F_{μν} F_{ρ σ}$ that cannot be removed without breaking gauge invariance (Phys. Rev. 177, 2426, 1969) ^{[Adler 1969]}. Bell and Jackiw recognised the same term as the resolution of why the neutral pion decays at the observed rate (Nuovo Cimento A 60, 47, 1969) ^{[Bell-Jackiw 1969]}. In the same year Adler and Bardeen proved that the coefficient is exact at one loop (Phys. Rev. 182, 1517, 1969) ^{[Adler-Bardeen 1969]}, establishing the anomaly as a precise rather than approximate statement. Fujikawa gave the path-integral derivation a decade later, identifying the anomaly with the Jacobian of the fermionic measure under a chiral rotation, computed as a regularised spectral trace of $γ^{5}$ (Phys. Rev. Lett. 42, 1195, 1979) ^{[Fujikawa 1979]}. The connection to the Atiyah-Singer index theorem of 1968, in which the integrated anomaly counts Dirac zero modes, completed the modern picture and is developed in 03.07.32. Weinberg's account in The Quantum Theory of Fields, Vol. 2, Ch. 22, gives the canonical textbook treatment of the triangle, the Ward-identity choice, and the nonrenormalisation theorem ^{[Weinberg QTF Vol 2]}.

Bibliography Master

@article{Adler1969,
  author  = {Adler, Stephen L.},
  title   = {Axial-Vector Vertex in Spinor Electrodynamics},
  journal = {Physical Review},
  volume  = {177},
  pages   = {2426--2438},
  year    = {1969}
}

@article{BellJackiw1969,
  author  = {Bell, J. S. and Jackiw, R.},
  title   = {A PCAC Puzzle: $\pi^0 \to \gamma\gamma$ in the $\sigma$-Model},
  journal = {Nuovo Cimento A},
  volume  = {60},
  pages   = {47--61},
  year    = {1969}
}

@article{AdlerBardeen1969,
  author  = {Adler, Stephen L. and Bardeen, William A.},
  title   = {Absence of Higher-Order Corrections in the Anomalous Axial-Vector Divergence Equation},
  journal = {Physical Review},
  volume  = {182},
  pages   = {1517--1536},
  year    = {1969}
}

@article{Fujikawa1979,
  author  = {Fujikawa, Kazuo},
  title   = {Path-Integral Measure for Gauge-Invariant Fermion Theories},
  journal = {Physical Review Letters},
  volume  = {42},
  pages   = {1195--1198},
  year    = {1979}
}

@article{AtiyahSinger1968,
  author  = {Atiyah, M. F. and Singer, I. M.},
  title   = {The Index of Elliptic Operators: I},
  journal = {Annals of Mathematics},
  volume  = {87},
  pages   = {484--530},
  year    = {1968}
}

@book{WeinbergQTFVol2,
  author    = {Weinberg, Steven},
  title     = {The Quantum Theory of Fields, Vol. 2: Modern Applications},
  publisher = {Cambridge University Press},
  year      = {1996}
}

@book{BertlmannAnomalies1996,
  author    = {Bertlmann, Reinhold A.},
  title     = {Anomalies in Quantum Field Theory},
  publisher = {Oxford University Press},
  year      = {1996}
}

@article{tHooft1976ABJ,
  author  = {'t Hooft, Gerard},
  title   = {Symmetry Breaking through Bell-Jackiw Anomalies},
  journal = {Physical Review Letters},
  volume  = {37},
  pages   = {8--11},
  year    = {1976}
}

Prerequisites

12.05.05
03.07.32
12.18.04

Tier anchors

beginner: Conceptual leaky-conservation-law account paced like 3Blue1Brown / Griffiths, *Introduction to Elementary Particles* Ch. on neutral-pion decay
intermediate: Peskin, M. E. & Schroeder, D. V., *An Introduction to Quantum Field Theory* (Addison-Wesley, 1995), §19.1-19.3; Weinberg, S., *The Quantum Theory of Fields, Vol. 2* (Cambridge, 1996), §22.2-22.3
master: Weinberg, S., *The Quantum Theory of Fields, Vol. 2: Modern Applications* (Cambridge, 1996), Ch. 22; Bertlmann, R. A., *Anomalies in Quantum Field Theory* (Oxford, 1996), Chs. 2-4; Adler, S. L., *Phys. Rev.* 177 (1969), 2426

References

tong · Standard Model lectures, the chiral anomaly, the triangle diagram, and pi0 -> gamma gamma
tong · Gauge Theory lectures, the axial anomaly, Fujikawa's method, and the index theorem
Weinberg, S. — The Quantum Theory of Fields, Vol. 2: Modern Applications (Cambridge, 1996) · Ch. 22 (the axial-current anomaly; the AVV triangle; the choice between vector and axial conservation; Adler-Bardeen nonrenormalisation; pi0 decay; Fujikawa's measure)
Adler, S. L. — Axial-vector vertex in spinor electrodynamics · Phys. Rev. 177 (1969), 2426 — the triangle anomaly and the nonrenormalisation argument
Bell, J. S. & Jackiw, R. — A PCAC puzzle: pi0 -> gamma gamma in the sigma-model · Nuovo Cimento A 60 (1969), 47 — the anomaly as the resolution of the PCAC puzzle
Adler, S. L. & Bardeen, W. A. — Absence of higher-order corrections in the anomalous axial-vector divergence · Phys. Rev. 182 (1969), 1517 — the nonrenormalisation theorem
Fujikawa, K. — Path-integral measure for gauge-invariant fermion theories · Phys. Rev. Lett. 42 (1979), 1195 — the anomaly as the Jacobian of the fermion measure

Estimated time

beginner: 18m
intermediate: 45m
master: 85m