Linear Response, Fluctuations, and Stochastic Gravity

The semiclassical Einstein equation is a mean-field equation. This chapter asks the next question: how does the geometry respond to a perturbation of the metric and to the irreducible fluctuations of quantum matter? The answer joins a retarded stress-response kernel, a positive stress noise kernel, and the linearized gravitational operator in the Einstein–Langevin equation. The resulting framework tests stability and computes gauge-invariant metric covariances without claiming a complete quantum theory of gravity.

Helpful background. Quantum Fields in Curved Spacetime supplies states and renormalized stress operators, Einstein Equations supplies linearized geometry and constraints, Black-Hole Geometry and Mechanics supplies horizon boundary conditions, and Semiclassical Gravity supplies the mean-field regime being extended.

Enter the response problem

Begin with a state–geometry pair $(\rho,\bar g_{ab})$ that solves the renormalized semiclassical equation. A perturbation $h_{ab}$ changes the stress expectation by

$$\delta\langle T_{ab}(x)\rangle_{\mathrm{ren}} =\int dV_y\, \Pi^{\mathrm{ret}}_{ab}{}^{cd}(x,y)h_{cd}(y).$$

This derivative holds the initial state fixed. The kernel has local operator and subtraction contact terms, plus nonlocal state-dependent polarization; local couplings assigned to the gravitational operator must not also be included in the stress kernel. Retarded support is essential: an in–out effective action generally gives a time-ordered, complex kernel rather than a real causal initial-value equation. Ward identities tie the stress response to the linearized Bianchi identity and prevent a pure gauge perturbation from masquerading as a physical instability. Anderson, Molina-París, and Mottola develop this causal, gauge-invariant criterion in Anderson, Molina-París, and Mottola 2003, §§ II–V, PDF.

The mean response is only half of the problem. With

$$\hat t_{ab} =\hat T^{\mathrm{ren}}_{ab} -\langle\hat T^{\mathrm{ren}}_{ab}\rangle\hat I,$$

the symmetrized connected correlator

$$N_{ab\,c'd'}(x,y) =\frac12\langle\{\hat t_{ab}(x),\hat t_{c'd'}(y)\}\rangle$$

defines the noise kernel. It is a positive bitensor distribution after admissible smearing, not a generally finite pointwise variance.

The Einstein–Langevin synthesis

Let $L$ include the linearized renormalized geometric operator and the retarded stress response. Introduce a conserved Gaussian source with

$$\mathbb E[\xi_{ab}]=0, \qquad \mathbb E[\xi_{ab}(x)\xi_{c'd'}(y)] =N_{ab\,c'd'}(x,y).$$

Then

$$(Lh)_{ab}=8\pi G\,\xi_{ab}.$$

The retarded solution separates an intrinsic homogeneous perturbation $h^{(0)}$ from an induced term $8\pi G\,G^{\mathrm{ret}}*\xi$. Its covariance is therefore

$$C_h=C_{\mathrm{intr}} +(8\pi G)^2G^{\mathrm{ret}}*N*(G^{\mathrm{ret}})^{T}.$$

The first term tests the initial metric state and homogeneous stability. The second tests how matter fluctuations are filtered and possibly amplified. At leading nontrivial order in the standard large-$N$ scaling, this stochastic covariance reproduces the symmetrized quantum metric two-point function when the intrinsic initial data are matched. Hu and Verdaguer derive the framework and its limits in Hu and Verdaguer 2008, §§ 3.2–3.3 and 4.1–4.3.

In a stationary KMS state, the noise and dissipative spectra obey a fluctuation–dissipation relation. With commutator spectrum $\rho(\omega)$,

$$\widetilde N(\omega) =\frac12\coth\!\left(\frac{\beta\omega}{2}\right)\rho(\omega).$$

For $\omega\ne0$, this follows from thermal detailed balance. A zero-frequency spectral atom from a conserved observable is a separate contribution, as the fluctuation–dissipation proof explains. The thermal factor does not hold globally for a generic nonequilibrium state, including an evaporating Unruh-state black hole.

Route through the chapter

Page	Central question	Result to carry forward
Metric Perturbations and Stress–Tensor Response	How does $\langle T_{ab}\rangle$ vary with the metric?	A renormalized retarded polarization kernel with Ward identities and a declared cutoff regime.
The Stress–Energy Noise Kernel	How are connected stress fluctuations represented?	A conserved, positive distribution defined through physical smearing.
The Einstein–Langevin Equation	How does stress noise produce metric covariance?	Retarded propagation and the split into intrinsic and induced fluctuations.
Fluctuation–Dissipation Relations	When do noise and absorption determine one another?	An exact KMS spectral theorem with explicit hypotheses and counterexample.
Linear Stability of Semiclassical Solutions	Which growing modes are physical and below the cutoff?	Gauge-invariant pole tests, order-reduced branch selection, and a reproducible convergence benchmark.
Validity of Stochastic and Mean-Field Gravity	What do two-point tests establish, and what remains open?	A dated hierarchy of mean, response, induced-noise, large-$N$, and higher-cumulant claims.

Read in order. The response page fixes causality before the noise page defines the source; the Einstein–Langevin page then combines them. The thermal theorem is a special consistency relation, not a prerequisite for nonequilibrium stochastic gravity. The last two pages turn the formalism into stability and validity diagnostics.

How to make a defensible calculation

Specify the self-consistent background, quantum state, renormalization prescription, and finite gravitational couplings. State the frequency-curvature window and the cutoff. Construct the causal response from that same state, solve the constraints or choose gauge-invariant observables, and retain only boundary conditions appropriate to the retarded problem.

For fluctuations, define a smooth detector or averaging profile before evaluating the noise kernel. Propagate both intrinsic and induced covariances. Check positivity, Ward identities, retarded support, dimensions, and sensitivity to the smearing scale. If a derivative expansion introduces high-frequency branches, compare them with the cutoff and with a consistent order-reduction prescription rather than deleting roots after inspecting the answer. Parker and Simon explain this perturbative branch logic in Parker and Simon 1993, §§ II–VI and Appendix, PDF.

Three conclusions must remain distinct. A bounded homogeneous response is a stability statement. A small induced covariance is a fluctuation statement for a named observable. Leading large-$N$ agreement is a two-point quantum correspondence. None alone proves nonlinear stability, controls higher cumulants, or resolves black-hole information.

Review the chapter

Consider a stationary scalar channel with retarded transfer function

$$G^{\mathrm{ret}}(\omega) =\frac{1}{\Omega^2-\omega^2-i\gamma\omega}, \qquad \Omega\gt 0,\qquad\gamma\gt 0,$$

and a stationary noise spectrum $\widetilde N(\omega)$. Derive the induced equal-time variance, state the pole stability test for the convention $e^{-i\omega t}$, and explain why a finite variance still does not by itself validate a semiclassical background.

Solution

The stationary solution is $h(\omega)=8\pi G\,G^{\mathrm{ret}}(\omega)\xi(\omega)$. With

$$\mathbb E[\xi(\omega)\xi(\omega')] =2\pi\delta(\omega+\omega')\widetilde N(\omega),$$

the induced equal-time variance is

$$\operatorname{Var}_{\mathrm{ind}}h =(8\pi G)^2 \int_{-\infty}^{\infty}\frac{d\omega}{2\pi}\, \left\lvert G^{\mathrm{ret}}(\omega)\right\rvert^2 \widetilde N(\omega).$$

The poles solve $\Omega^2-\omega^2-i\gamma\omega=0$ and lie in the lower half-plane for $\gamma\gt 0$, so homogeneous modes decay with $e^{-i\omega t}$. This establishes retarded linear stability in the scalar channel. Validity additionally requires that the integral exist for the declared smearing and cutoff, that its square root be small relative to the resolved background scale, and that gauge, other polarizations, intrinsic fluctuations, higher cumulants, and secular nonlinear effects remain controlled. A finite number without those comparisons is not a complete validity test.

References

• Anderson, P. R., Molina-París, C., and Mottola, E. 2003. “Linear Response, Validity of Semi-Classical Gravity, and the Stability of Flat Space.” Physical Review D 67, 024026. DOI. Open PDF.
• Hu, B. L., and Verdaguer, E. 2008. “Stochastic Gravity: Theory and Applications.” Living Reviews in Relativity 11, 3. DOI.
• Parker, L., and Simon, J. Z. 1993. “Einstein Equation with Quantum Corrections Reduced to Second Order.” Physical Review D 47, 1339–1355. DOI. Open PDF.