Anomalies and Quantum Effective Actions

Quantum effective actions compress the response of fluctuating fields into a functional of the background geometry. Their ultraviolet expansion identifies local counterterms, their Weyl and chiral variations expose anomalies, and their nonlocal form factors retain propagation and threshold information that no finite local expansion can reproduce. This chapter develops that chain at one loop on a prescribed curved spacetime. The central discipline is to distinguish universal data from choices of subtraction, state, Green function, and boundary condition.

Helpful background. Flat-Spacetime Quantum Field Theory provides Gaussian functional integrals, loop counting, and renormalization; Partial Differential Equations and Green Functions supplies elliptic kernels, spectral conditions, and inverse operators; Causality and Global Structure establishes the Lorentzian causal structures used to select propagators; Quantum Fields in Curved Spacetime provides states, observables, and local covariance; Gravitational Actions and Boundary Terms defines the geometric couplings and boundary completions that receive loop corrections.

The chain from an operator to an anomaly

For a quadratic fluctuation operator $D$, the Euclidean one-loop functional is schematically

$$\Gamma_E^{(1)}=\frac{\sigma}{2}\operatorname{Tr}\ln(D/\mu^2) =-\frac{\sigma}{2}\int_0^\infty\frac{ds}{s}\,\operatorname{Tr}e^{-sD},$$

where $\sigma=+1$ for a real bosonic determinant and the fermionic sign and multiplicity must be derived from the actual Grassmann operator. The small-$s$ heat-kernel expansion sorts ultraviolet divergences by local geometric dimension. Renormalization absorbs those divergences into gravitational couplings. A Weyl or chiral variation then asks whether the renormalized measure and action preserve a classical symmetry. If they do not, the remaining finite variation is an anomaly.

This reasoning has hypotheses. The heat-kernel proof begins with an elliptic operator of Laplace type, appropriate domain data, and a controlled asymptotic limit; gauge systems first require gauge fixing and ghosts. Boundaries introduce additional coefficients rather than merely modifying bulk coefficients. A Lorentzian hyperbolic operator is therefore not inserted naively into an elliptic theorem. One computes a Euclidean invariant under stated continuation assumptions and translates the final covariant expression back to the site convention $(-,+,+,+)$. Vassilevich 2003, §§2.1–2.3 and §4, PDF gives the operator reduction, asymptotic expansion, and bulk coefficients on which this chain rests.

A six-stage route through the chapter

The pages are designed to be read in the following order.

1. Heat Kernels and Seeley–DeWitt Coefficients constructs the diagonal kernel, derives the first local coefficients, and separates pointwise densities from integrated bulk and boundary terms.
2. One-Loop Effective Actions in Curved Spacetime turns the proper-time trace into determinants, divergences, and finite form factors while distinguishing Euclidean, in–out, and causal in–in functionals.
3. Trace Anomalies classifies the four-dimensional Weyl anomaly into Euler, Weyl-squared, and total-derivative terms, showing why the coefficient of $\Box R$ is scheme-dependent.
4. Chiral and Gravitational Anomalies connects the regulated fermion measure to index densities and descent, then distinguishes consistent from covariant currents through the Bardeen–Zumino shift.
5. Anomaly-Induced Nonlocal Actions integrates an anomalous variation with the Paneitz operator and makes explicit the conformally invariant, state, boundary, and Green-function data left undetermined.
6. Decoupling, Mass Scales, and Scheme Dependence studies massive form factors across ultraviolet and infrared regimes and explains which decoupling claims are secure and which remain prescription-sensitive.

The order matters. Heat-kernel coefficients alone do not specify a physical real-time response; an anomaly alone does not determine a complete effective action; a mass-independent beta function alone does not describe a threshold. Each later page retains the hypotheses and ambiguities exposed by the earlier one.

What is universal, local, or state-dependent

Three levels of information recur throughout the chapter:

Quantity	What fixes it	What can still vary
Short-time coefficient of a specified elliptic boundary problem	Operator, bundle connection, endomorphism, dimension, and boundary condition	Total divergences can move between equivalent density representatives; changing the domain changes boundary coefficients
Nontrivial anomaly class	Field content, chirality, representation, and cohomology class	Trivial anomalies shift under finite local counterterms; consistent and covariant representatives differ by a local polynomial
Full nonlocal effective action or stress response	Spectrum plus a physical prescription	Weyl-invariant functionals, state, contour, Green function, and homogeneous solutions remain additional data

For example, for a massless conformal field in four dimensions the anomalous trace may be organized as

$$\langle T^\mu{}_{\mu}\rangle =\frac{1}{(4\pi)^2}\left(cC_{\mu\nu\rho\sigma}C^{\mu\nu\rho\sigma}-aE_4+b\Box R\right).$$

The type-A and type-B coefficients represent nontrivial anomaly data for fixed field content, whereas adding a finite $R^2$ counterterm changes $b$. This is not a contradiction between calculations; it is the expected freedom in a trivial anomaly. A mass term also breaks Weyl symmetry explicitly, so the complete massive trace is not exhausted by this geometric expression. The cohomological classification and the separation of the Euler and Weyl-invariant structures are developed in Deser and Schwimmer 1993, §§2–3, PDF.

The analogous distinction for chiral symmetry is between the consistent current, obtained by varying one effective action and therefore satisfying the Wess–Zumino condition, and the covariant current, obtained after a Bardeen–Zumino improvement. They cannot be interchanged inside a conservation law without also changing the definition of the current. The index and descent analysis requires smooth fields and appropriate global and boundary hypotheses; zero modes or boundary flux can invalidate a purely local inference. Alvarez-Gaumé and Witten 1984, §§3–5 develops the characteristic-class calculation and descent structure.

From local coefficients to nonlocal physics

Ultraviolet divergences are local because they arise from arbitrarily short proper times. Finite physics need not be. Logarithms such as $C\ln(-\Box/\mu^2)C$, inverse conformal operators, and massive threshold form factors remember propagation over finite distances. Their Lorentzian meaning depends on the analytic prescription: an in–out kernel computes a transition amplitude and can be complex, while a Schwinger–Keldysh construction gives causal expectation-value equations. A Euclidean determinant by itself does not choose between them.

Integrating the trace anomaly illustrates the same limit of local information. A Riegert-type functional is one particular solution of the Weyl-variation equation. One may add any Weyl-invariant functional, and choosing an inverse Paneitz operator requires boundary conditions and a Green function. Auxiliary-field localization transfers rather than removes this freedom: homogeneous auxiliary solutions encode it. Mazur and Mottola 2001, §§2–4, PDF explains the anomaly action, its auxiliary representation, and the role of additional solutions.

Mass introduces a second comparison scale. When $p^2/m^2\ll1$ and all curvature and boundary scales are likewise small, massive form factors admit a derivative expansion and curvature-squared response decouples by powers of $p^2/m^2$. In the opposite regime the logarithmic ultraviolet running reappears. This claim is strongest for form factors that can be tied to nonlocal curvature-squared structures. It is weaker for the cosmological and Einstein–Hilbert terms, whose momentum dependence is difficult to isolate invariantly on a generic curved background. The explicit mass-dependent calculation and its limitations appear in Gorbar and Shapiro 2003, §§3–5, PDF.

Limits of the chapter

The background geometry is fixed: varying the effective action can produce a renormalized stress tensor, but solving the semiclassical Einstein equation and controlling its stability require additional state and backreaction analysis. The chapter does not claim that every Lorentzian spacetime admits a useful Euclidean section, that an asymptotic heat-kernel series converges, or that a local large-mass expansion captures particle creation. Nor does equality of integrated coefficients imply equality of local densities in the presence of total derivatives or boundaries.

A decisive check is therefore question-dependent. Ultraviolet poles should match the appropriate local heat-kernel coefficient. Anomaly representatives should satisfy consistency conditions and display their counterterm freedom. Real-time response must obey the intended causal support. Massive form factors should reproduce both ultraviolet logarithms and infrared power suppression in sectors where those limits are defined. A failure of any one check cannot be repaired by changing terminology or subtraction conventions.

References

• Alvarez-Gaumé, Luis, and Edward Witten. “Gravitational Anomalies.” Nuclear Physics B 234 (1984): 269–330. DOI.
• Deser, Stanley, and Adam Schwimmer. “Geometric Classification of Conformal Anomalies in Arbitrary Dimensions.” Physics Letters B 309 (1993): 279–284. DOI. Open PDF, arXiv:hep-th/9302047.
• Gorbar, Eduard V., and Ilya L. Shapiro. “Renormalization Group and Decoupling in Curved Space.” Journal of High Energy Physics 2003, no. 02 (2003): 021. DOI. Open PDF, arXiv:hep-ph/0210388.
• Mazur, Pawel O., and Emil Mottola. “Weyl Cohomology and the Effective Action for Conformal Anomalies.” Physical Review D 64 (2001): 104022. DOI. Open PDF, arXiv:hep-th/0106151.
• Vassilevich, Dmitri V. “Heat Kernel Expansion: User’s Manual.” Physics Reports 388 (2003): 279–360. DOI. Open PDF, arXiv:hep-th/0306138.

Integrative diagnostic

A conformally coupled scalar of mass $m$ is integrated out on a smooth four-dimensional background without boundary. A Euclidean heat-kernel calculation finds an $a_4$ logarithmic divergence, and a colleague proposes three conclusions: the Euclidean determinant already gives the causal Lorentzian stress response; the anomaly uniquely fixes the finite effective action; and the same minimal-subtraction beta function describes physics at $p\ll m$. Which conclusions follow, and what additional data or calculation is needed?

Solution

None of the three conclusions follows as stated. The $a_4$ coefficient fixes the local logarithmic divergence, including mass-dependent counterterms. Its massless conformal limit gives the geometric anomaly after the Euclidean-to-Lorentzian sign translation; at finite $m$, the trace also includes explicit mass breaking. A finite $R^2$ counterterm can still shift the scheme-dependent $\Box R$ term. A causal stress response requires an in–in contour and its retarded kernel, not merely the Euclidean or in–out determinant. Integrating the anomaly determines only a particular functional: a Weyl-invariant functional and the state, boundary, Green-function, or homogeneous-solution data remain free. Finally, minimal subtraction records ultraviolet logarithms without resolving the physical threshold. One must compute a mass-dependent form factor and expand it at $p^2/m^2\ll1$; curvature-squared response should then show power suppression, while claims about the cosmological or Newton coupling need a separately defined observable and subtraction prescription.