Matter Effective Actions in Curved Spacetime

A matter effective action summarizes the influence of quantum fields on a prescribed geometry. Determinants and heat kernels organize local curvature terms, while nonlocal response and causal evolution require additional information.

Enter this chapter

Begin with One-Loop Matter Effective Actions in Curved Space to define the matter operator, its domain, and the loop expansion.

Helpful background. Heat Kernels, Zeta Functions, and Spectral Determinants and The 1PI Effective Action and Mean-Field Equations develop the definitions used by the chapter’s opening pages.

Topics

1. One-Loop Matter Effective Actions in Curved Space — Defines what is integrated out and what is held classical.
2. Proper-Time, Zeta, and Determinant Prescriptions — Fixes domains, zero modes, scales, and spectral cuts.
3. Heat Kernels and the Schwinger–DeWitt Expansion — Extracts short-proper-time ultraviolet data.
4. Seeley–DeWitt Coefficients and Curvature Invariants — Derives the local coefficients and convention checks.
5. Heat Kernels with Boundaries and Conical Singularities — Adds boundary, corner, and defect contributions.
6. Worldline Methods in Curved Space — Rewrites the same trace as a particle path integral with a regulated measure.
7. Renormalization of Gravitational Couplings by Matter Loops — Matches ultraviolet poles onto local gravitational terms.
8. Mass Thresholds, Decoupling, and Curvature Expansions — States the hierarchy behind inverse-mass expansions.
9. Matter-Induced Nonlocal Form Factors — Retains momentum dependence and branch information.
10. Anomaly-Induced and Nonlocal Actions — Integrates the trace anomaly while exposing homogeneous ambiguities.
11. Imaginary Effective Actions and Vacuum Instability — Relates determinant phases to vacuum persistence only after an independent production check.
12. Effective-Action Variation, Stress Tensors, and Consistency Checks — Compares metric variations with conservation, anomaly, and point splitting.

How the ideas fit together

The short-proper-time heat-kernel expansion identifies local ultraviolet divergences and curvature counterterms. It does not determine the full spectrum, zero-mode contribution, or long-distance nonlocal response of the operator Vassilevich 2003, §§ 2.1–2.2.

The time contour also matters. A Euclidean determinant or in/out vacuum amplitude is not automatically a causal equation for an evolving expectation value; that requires an in/in construction or an equivalent causal prescription. Finally, integrating out matter on a classical metric includes matter loops, not metric and gravitational-ghost loops.

The comparison below separates the assumptions and tests needed for these different conclusions.

Domain and failure conditions

Method or output	Operator, contour, and global data	Ultraviolet treatment	Licensed regime and result	Threshold, imaginary, and causal meaning	Decisive downgrade
Matter one-loop determinant	Hessian of the integrated matter field; metric classical; state or Euclidean boundary data declared	Local counterterms through the required adiabatic/heat-kernel order	One-loop matter contribution to a specified 1PI or vacuum functional	No graviton/ghost loop; causal response only after in–in construction	Changing the integrated field, gauge fixing, or contour changes the object
Proper time and zeta	Positive elliptic $\mathcal L_E$ on a stated self-adjoint domain; zero modes removed explicitly	Analytic continuation plus scale $\mu$	Equivalent determinant after matching the same subtractions	Negative modes need a cut and can create a phase	An unreported kernel, cut, or multiplicative anomaly invalidates equality
Local heat-kernel asymptotics	Smooth Laplace-type operator, normally without singular strata	$s\downarrow0$ coefficients	Local ultraviolet divergences and large-mass asymptotics	Does not determine large-$s$ infrared physics or a state	Using a finite local series at large $s$ invents nonlocal information
Seeley–DeWitt recursion	Connection, endomorphism, Riemann convention, and invariant basis fixed	Coefficients $a_n$ or $a_{2n}$ with notation declared	Reproducible curvature polynomials and bundle traces	Total derivatives depend on boundaries; basis coefficients are not all observables	A sign translation applied to $R$ but not $\xi$ or $\mathcal E$ corrupts the result
Boundaries and cones	Elliptic boundary condition or self-adjoint extension; defect regularization fixed	Bulk plus half-integer, surface, corner, or tip terms	Localized counterterms for that domain	Replica interpretation requires a separate continuation and entropy prescription	Treating a cone as smooth misses defect terms and possible logarithms
Worldline representation	Same $\mathcal L_E$; periodic loop, center-of-mass zero mode, measure ghosts, regulator fixed	Finite scheme counterterm restores the intended Hamiltonian	Alternative evaluation of heat kernels and determinants	Not a replacement of QFT by fundamental particle mechanics	Dropping the measure or scheme counterterm produces false noncovariance
Gravitational-coupling matching	Matter $a_0,a_1,a_2$ in a chosen local basis	Renormalize $\Lambda$, $G$, and curvature-squared couplings	Matter contribution to running and matching	Pure-gravity beta functions remain absent	Field redefinitions move redundant coefficients; comparing them basis-free is meaningless
Heavy-mass expansion	Eigenvalues/momenta and curvatures small relative to $m^2$	Match local terms at a stated threshold	Series in $\nabla^2/m^2$ and $\mathcal R/m^2$ with truncation estimate	Mass-independent beta functions alone do not display physical decoupling	$\mathcal R/m^2\sim1$ or $p^2/m^2\sim1$ removes the expansion
Nonlocal form factors	Full spectral or momentum dependence and branch prescription retained	Local polynomial ambiguity subtracted separately	Kernels such as $\ln(-\Box/\mu^2)$ and threshold functions	Euclidean, Feynman, and retarded continuations answer different questions	Inserting a Feynman kernel into a causal equation violates the observable contract
Anomaly-induced action	Anomaly coefficients, conformal class, Green function, and boundary data fixed	Scheme-dependent $\Box R$ or local $R^2$ term declared	A particular solution of the Weyl-variation equation	Weyl-invariant functionals and homogeneous solutions remain free	The anomaly alone does not determine the full stress tensor or state
Imaginary in–out action	In/out vacua and $i0$ or spectral cut fixed	Real local counterterms do not remove physical absorptive parts	Vacuum-persistence exponent after mode normalization	$2\operatorname{Im}\Gamma_{\rm in-out}=-\ln P_0$ when the vacuum amplitude is defined	A lone negative mode or contour phase is not automatically particle production
Metric variation and response	Renormalized functional and allowed variations fixed	Finite curvature counterterms varied consistently	Conserved in–out/Euclidean stress and symmetric second variation	Retarded response and noise require in–in and connected correlators	Mixing schemes between action and stress breaks the Ward-identity comparison

Where to go next

Continue to Mean Semiclassical Backreaction to identify the renormalized source and couplings in the semiclassical Einstein equation.

References

• Vassilevich, Dmitri V. “Heat Kernel Expansion: User’s Manual.” Physics Reports 388 (2003): 279–360. DOI. Open preprint.