Curvature, Frequency, and Adiabatic Hierarchies

A local expansion of a massive field’s vacuum polarization requires slowly varying geometry and external probes: $mL_{\mathrm{curv}}\gg1$, $m\tau_{\mathrm{evol}}\gg1$, and external frequencies well below the mass threshold. These are not restrictions that every real particle have energy below its own mass. Occupied massive modes have local on-shell energy at least $m$; their state-dependent stress must be retained separately. Low curvature alone is insufficient: a weakly curved spacetime can carry a rapidly varying state, while a stationary state can have large observer-dependent proper frequencies near a horizon.

Required background. Einstein Equations supplies the curvature tensors and source dimensions used to compare successive terms.

Helpful background. Quantum Fields in Curved Spacetime supplies adiabatic and Hadamard state constructions, while Black-Hole Geometry and Mechanics supplies the invariant horizon and observer distinctions needed for local frequency estimates.

Curvature radii and derivative counting

At a point, define a conservative curvature length by the largest measured orthonormal component,

$$L_{\mathrm{curv}}^{-2} =\max\!\left( \lvert R_{\hat a\hat b\hat c\hat d}\rvert, \sqrt{\lvert\nabla_{\hat e}\nabla_{\hat f}R_{\hat a\hat b\hat c\hat d}\rvert}, \ldots \right),$$

with enough derivatives included to test the order being retained. This is not a unique scalar definition; it is a bound on all local components in the frame relevant to the measurement. A scalar invariant that vanishes in a special geometry need not bound tidal components, so one checks a complete enough set for the problem.

For a massive field, the local part of the one-loop effective action has the schematic heat-kernel expansion

$$\Gamma_{\mathrm{loc}} \sim\int d^4x\sqrt{-g}\left[ m^4a_0+m^2a_1+a_2\ln\!\frac{m^2}{\mu^2} +\frac{a_3}{m^2}+\cdots\right],$$

where $a_n$ contains curvature and derivatives with overall mass dimension $2n$. After the divergent and scale-dependent $a_0,a_1,a_2$ pieces renormalize gravitational couplings, successive finite terms are suppressed by powers of $(mL_{\mathrm{curv}})^{-2}$ when derivatives share that long scale. Rapid weak-amplitude variations require a separate derivative test: small curvature components alone do not bound ratios such as $\Box R/(m^2R)$. Shapiro explains this organization and the limits of massive-field decoupling in Shapiro 2008, §§ 2–3 and 6.

Frequency and state scales

The geometry does not determine the state spectrum. Given a timelike observer $u^\mu$, a mode with covector $k_\mu$ has measured frequency

$$\omega_{\mathrm{loc}}=-u^\mu k_\mu.$$

The relevant frequency hierarchy for resolved excitations is $\omega_{\mathrm{loc}}\ll\Lambda_{\mathrm{EFT}}$ throughout the sampling region. In a local inertial frame a massive on-shell excitation satisfies $\omega_{\mathrm{loc}}^2=\boldsymbol k_{\mathrm{loc}}^2+m^2$, so $\omega_{\mathrm{loc}}\ge m$. By contrast, a probe or background modulation with frequency $\Omega\ll m$ can admit a local expansion after virtual massive fields are integrated out. In a stationary spacetime a conserved Killing frequency can remain modest while the locally measured frequency becomes large for a highly boosted observer. That observation diagnoses the chosen observable and frame; it does not by itself prove that an invariant low-energy prediction fails.

For a mode reduced to $\ddot q+\omega(t)^2q=0$ in a stated time coordinate, an adiabatic hierarchy begins with

$$\epsilon_1=\left\lvert\frac{\dot\omega}{\omega^2}\right\rvert\ll1, \qquad \epsilon_2=\left\lvert\frac{\ddot\omega}{\omega^3}\right\rvert\ll1,$$

and continues to the order needed by the subtraction or evolution scheme. The leading approximation is $q\simeq(2\omega)^{-1/2}\exp[-i\int^t\omega(s)ds]$: the amplitude and frequency must change little in one oscillation. Small $\epsilon_1$ does not guarantee small $\epsilon_2$, and an asymptotic adiabatic series need not converge. The first omitted term, not the name of the approximation, sets the error estimate.

A hierarchy table

Question	Dimensionless control	What it suppresses	What it does not suppress
Are local higher-curvature operators small?	$L_{\mathrm{cut}}^2/L_{\mathrm{curv}}^2$	local EFT corrections	nonlocal state response
Can a massive field be expanded locally?	$(mL_{\mathrm{curv}})^{-2}$	higher heat-kernel coefficients	on-shell particle production
Is evolution adiabatic?	$\lvert\dot\omega/\omega^2\rvert$	transitions between instantaneous modes	large accumulated phase or secular drift
Is a measured mode below cutoff?	$\omega_{\mathrm{loc}}/\Lambda_{\mathrm{EFT}}$	trans-cutoff sensitivity in that frame	large long-wavelength occupation
Is backreaction weak?	$G\lvert\langle T\rangle\rvert L_{\mathrm{curv}}^2$	change of the mean geometry	connected stress fluctuations

Independent limits and failure cases

In flat spacetime, $L_{\mathrm{curv}}\to\infty$ removes curvature corrections but leaves state-dependent excitations; this checks that curvature and occupation were not conflated. In a stationary constant-curvature state, time derivatives vanish but the ratio $mL_{\mathrm{curv}}$ can still be small; this checks that adiabaticity and locality were not conflated. Hu and Verdaguer similarly separate characteristic length, response, and fluctuation criteria in Hu and Verdaguer 2008, §§ 3.3 and 6.5.3.

The expansion must be reorganized for massless fields, thresholds, resonances, secular times of order the inverse small parameter, or nonlocal particle-production effects. Fixed-background computation of those effects belongs to Quantum Fields in Curved Spacetime; here the hierarchy is used to decide when their stress tensor may consistently backreact. If no ordering keeps both geometry and state below the cutoff, the problem has crossed the semiclassical boundary.

Exercise

A field of mass $m$ is integrated out on a geometry with curvature radius $L$ and evolution time $\tau$. An external probe has frequency $\Omega$. List the independent small parameters and decide whether $mL=100$, $m\tau=2$, and $\Omega/m=0.01$ justify a local adiabatic approximation. Could $\Omega$ instead be the local energy of a real massive particle?

Solution

The basic ratios are $(mL)^{-2}=10^{-4}$ for curvature, $(m\tau)^{-1}=1/2$ for temporal derivatives, and $\Omega/m=10^{-2}$ for the external probe, supplemented by the gravitational cutoff ratios. The curvature amplitude and probe frequency are favorable, but $m\tau=2$ supplies no strong derivative or adiabatic separation. A finite local derivative expansion of the entire response is therefore not justified; rapid time dependence may require the nonlocal kernel. Nor can $\Omega=0.01m$ be the local on-shell energy of a real particle of mass $m$, since that energy is at least $m$.

References

• Hu, B. L., and Verdaguer, E. 2008. “Stochastic Gravity: Theory and Applications.” Living Reviews in Relativity 11, 3. DOI.
• Shapiro, I. L. 2008. “Effective Action of Vacuum: Semiclassical Approach.” Classical and Quantum Gravity 25, 103001. Open PDF.