An anomaly occurs when a symmetry of the classical Lagrangian is broken by quantum effects (loop corrections). The chiral anomaly breaks the classical conservation of the axial current and explains neutral pion decay. Gauge anomalies would destroy the consistency of a gauge theory; their cancellation constrains the particle content of the Standard Model.
An anomaly in quantum field theory occurs when a symmetry of the classical Lagrangian fails to survive quantization. The classical theory has a conserved current (by Noether's theorem), but quantum corrections (specifically, loop diagrams) generate a nonzero divergence of that current. The most important example is the chiral anomaly (or ABJ anomaly), discovered independently by Adler and by Bell and Jackiw in 1969.
Consider massless QED. The classical Lagrangian is invariant under both vector transformations (psi -> e^{i alpha} psi, conserving the vector current) and axial transformations (psi -> e^{i alpha gamma_5} psi, conserving the axial current). But the triangle diagram -- a fermion loop with one axial-current vertex and two vector-current vertices -- is ambiguous: you cannot regularize it in a way that preserves both symmetries simultaneously. The standard choice preserves the vector symmetry (essential for electric charge conservation) at the expense of the axial symmetry, giving the anomaly equation partial_mu j^mu_5 = (e^2)/(16 pi^2) F_{mu nu} F-tilde^{mu nu}. This is an exact result, receiving no corrections beyond one loop.
Anomalies are classified into two types with very different implications. Global anomalies (anomalies in global symmetries) are physically real and have observable consequences. The chiral anomaly explains the decay pi^0 -> gamma gamma: without it, this decay would be forbidden, and the predicted rate (proportional to the number of quark colors squared) agrees with experiment for N_c = 3. Gauge anomalies (anomalies in local gauge symmetries) would be fatal: they would destroy unitarity and renormalizability, making the theory mathematically inconsistent. Gauge anomaly cancellation is therefore a constraint on the allowed particle content.
In the Standard Model, gauge anomaly cancellation places tight constraints on the charges and representations of the particles. The anomaly coefficients for SU(3)^2 U(1)_Y, SU(2)^2 U(1)_Y, U(1)_Y^3, and the mixed gravitational-U(1)_Y anomaly must all vanish. Remarkably, they do -- but only when the quarks and leptons are included with their observed quantum numbers, and within each complete generation. This cancellation is one of the most compelling pieces of evidence that the Standard Model has a deeper structure, likely a grand unified theory in which quarks and leptons are unified into larger representations where anomaly cancellation is automatic.
No topics depend on this one yet.