Perturbative semiclassical entropy of dynamical black holes

Abstract: We consider perturbative quantum gravity as a quantum field theory of linearized metric perturbation on an asymptotically flat spacetime with a bifurcate Killing horizon. We include the perturbative gravitational constraints into the algebra of observables restricted to the right half of the future horizon of the spacetime. We use the boundary charge, associated to the horizon Killing field, as an auxiliary "observer" degree of freedom. The observables "dressed" with the additional charge are invariant under the Killing symmetry and generate a Type-$\text{II}_{\infty}$ von Neumann factor. We compute the von Neumann entropy of the reduced density matrix of a classical-quantum coherent state constructed from the metric perturbations and the "observer wavefunction". This von Neumann entropy satisfies an analogue of the first law of thermodynamics. We further show that this entropy is related to Hollands-Wald-Zhang entropy of the (second order) perturbed dynamical black hole through the flux of perturbations through the horizon and future null infinity.

• The paper constructs a Type-II∞ crossed-product algebra by adjoining the asymptotic HWZ boundary charge to the Type-III horizon algebra, enabling a renormalized von Neumann entropy for perturbative quantum-gravity states.
• The entropy of coherent perturbations satisfies S_vN = −S(ω_h|ω_0) + β⟨X⟩ + S(f) + log β, with relative entropy equal to the classical horizon flux under the slowly varying observer-wavefunction approximation.
• The resulting entropy equals the expected quadratic Hollands–Wald–Zhang entropy at any horizon cut minus the flux already absorbed by the horizon, while trace normalization, radiation, soft modes, and noncommuting observer states remain open issues.

Overview and setting

The paper computes a perturbative semiclassical entropy for dynamical black holes by combining algebraic quantum field theory on bifurcate Killing horizons with the crossed-product construction of Type-II von Neumann algebras. The authors work in linearized quantum gravity on an asymptotically flat background spacetime $(\mathcal{M}, g_{ab})$ admitting a Killing field $\xi^a$ with bifurcate horizon $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$ of constant surface gravity $\kappa$, inverse temperature $\beta = 2\pi/\kappa$. Two structural assumptions are made: the region $\mathcal{P} \cup \mathcal{R}$ is globally hyperbolic, and free-field data can be specified independently on the horizon $\mathcal{H}$ and at $i_R^+ \cup \mathscr{I}_R^+$; for most of the paper the latter data is set to zero (no radiation through null infinity), an assumption relaxed only in a later remark.

Gauge conditions are imposed so that the event horizon coincides with $\mathcal{H}$ to first order, with $\delta\vartheta = 0$, $\xi^a$0, and $\xi^a$1 on $\xi^a$2. The free data on the horizon is then the perturbed shear $\xi^a$3, where $\xi^a$4 is the affine parameter along the null generators. The second variation of the Raychaudhuri equation gives $\xi^a$5, which underlies the dynamical correction to black hole entropy.

Horizon phase space and quantization

The covariant phase space symplectic form of general relativity, evaluated on $\xi^a$6, reduces to a symplectic product of shear data:

$\xi^a$7

Smeared observables $\xi^a$8 carry the Poisson bracket inherited from this symplectic structure. The Hamiltonian generating the Killing flow is the flux of the modified canonical energy,

$\xi^a$9

which can equivalently be written as the integral of an exact 2-form built from the Noether charge $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$0 and the symplectic potential $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$1: $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$2.

Quantization proceeds à la algebraic QFT: the $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$3-algebra generated by smeared shear operators with CCR determined by $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$4, in a centred Gaussian Hadamard vacuum state $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$5 whose two-point function is distributional on the horizon. Restricted to the right future horizon algebra $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$6, $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$7 is KMS at temperature $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$8; the authors note that global bulk KMS states need not exist (e.g., on Kerr), so only horizon KMS property is required. The weak closure of Weyl operators supported on $\mathcal{H} = \mathcal{H}^+ \cup \mathcal{H}^-$9 yields a von Neumann factor $\kappa$0 which, by Araki's results, is Type-III — hence density matrices and von Neumann entropy are not defined on it directly, though relative entropy remains well-defined.

Crossed product and the observer charge

To obtain a tractable entropy, the authors invoke the gravitational constraint at linearized order. Using the Hollands–Wald–Zhang (HWZ) dynamical entropy prescription,

$\kappa$1

the flux decomposes as $\kappa$2, assuming sufficiently fast decay of perturbations at early and late times. The boundary charge $\kappa$3 plays the role of an auxiliary "observer" Hamiltonian, represented as a multiplication operator on $\kappa$4, extending the Hilbert space to $\kappa$5. The constraint operator $\kappa$6 acts as minus the total Hamiltonian; observables commuting with it are the dressed Weyl operators $\kappa$7 together with $\kappa$8. By the Bisognano–Wichmann theorem the modular Hamiltonian satisfies $\kappa$9, so this extended algebra is precisely the crossed product of $\beta = 2\pi/\kappa$0 with its modular automorphism group; by Takesaki's theorem it is a Type-II factor.

A key point concerns the factor type. Under the no-flux assumption through $\beta = 2\pi/\kappa$1, the charge equals the second variation of the ADM Hamiltonian associated with $\beta = 2\pi/\kappa$2 at spatial infinity, $\beta = 2\pi/\kappa$3, whose spectrum is generally unbounded above. Consequently $\beta = 2\pi/\kappa$4 and the extended algebra is Type-$\beta = 2\pi/\kappa$5, not Type-$\beta = 2\pi/\kappa$6 as in the de Sitter observer case of Chandrasekaran–LongO–Penington–Witten. This has two consequences stated plainly by the authors: the trace is defined only up to scale (producing a state-independent additive ambiguity in the entropy), and there exists no maximal-entropy state since finite projections have traces spanning $\beta = 2\pi/\kappa$7.

Entropy of classical-quantum coherent states

A classical metric perturbation $\beta = 2\pi/\kappa$8 on $\beta = 2\pi/\kappa$9 defines an algebraic coherent state $\mathcal{P} \cup \mathcal{R}$0 via conjugation by a Weyl-type unitary. Since this state is defined only on the right-horizon subalgebra, its GNS vector is non-unique; the authors select the unique representative $\mathcal{P} \cup \mathcal{R}$1 in the natural cone $\mathcal{P} \cup \mathcal{R}$2 of the pair $\mathcal{P} \cup \mathcal{R}$3, constructed using the modular conjugation $\mathcal{P} \cup \mathcal{R}$4. Tensoring with an "observer wavefunction" $\mathcal{P} \cup \mathcal{R}$5 yields the classical-quantum state $\mathcal{P} \cup \mathcal{R}$6, which admits a renormalized density matrix $\mathcal{P} \cup \mathcal{R}$7 in the Type-$\mathcal{P} \cup \mathcal{R}$8 factor.

The central technical result, derived in the appendix under the assumption that $\mathcal{P} \cup \mathcal{R}$9 is slowly varying (its Fourier transform sharply peaked at zero momentum, so that commutators between $\mathcal{H}$0 and the relative modular operator are negligible), is:

$\mathcal{H}$1

For coherent states the relative entropy equals the classical flux into the black hole, $\mathcal{H}$2, so dividing by $\mathcal{H}$3 the relation takes the form of a thermodynamic first law for quantized gravitational perturbations, up to the state-independent constants $\mathcal{H}$4 and $\mathcal{H}$5.

Relation to the HWZ dynamical entropy

The main physical claim connects this von Neumann entropy to the HWZ entropy of the dynamical black hole. For any cut $\mathcal{H}$6 of $\mathcal{H}$7, integrating the exact 2-form expression for the flux between $\mathcal{H}$8 and $\mathcal{H}$9 gives $i_R^+ \cup \mathscr{I}_R^+$0, and substituting into the master formula yields:

$i_R^+ \cup \mathscr{I}_R^+$1

Thus the entanglement-style entropy of the perturbed quantum state on the entire right future horizon equals the expectation value of the quadratic HWZ entropy at an arbitrary cut, corrected by the classical flux that has fallen through the horizon up to that cut plus additive constants. This substantiates the interpretation of $i_R^+ \cup \mathscr{I}_R^+$2 as a genuine entropy of the perturbed state at leading (second) order in perturbation theory — notably, while each individual $i_R^+ \cup \mathscr{I}_R^+$3 depends on second-order metric data and is not itself an observable on the linearized phase space, the combination appearing here is well-defined through the flux difference.

Two extensions are recorded. If radiative flux through $i_R^+ \cup \mathscr{I}_R^+$4 is included (the news algebra being a Type-$i_R^+ \cup \mathscr{I}_R^+$5 factor with BMS-invariant vacuum), the identity becomes $i_R^+ \cup \mathscr{I}_R^+$6. Matter fields can be appended: massless matter adds flux terms at both the horizon and null infinity, while massive matter modifies the boundary charge $i_R^+ \cup \mathscr{I}_R^+$7 at $i_R^+ \cup \mathscr{I}_R^+$8 without null-infinity contribution.

Limitations and open questions

Several assumptions qualify the results. The slowly varying approximation on the observer wavefunction $i_R^+ \cup \mathscr{I}_R^+$9 is essential to the entropy computation; without it, non-commutativity between $\mathcal{H}$0 and the relative modular operator prevents an additive logarithm, and the master formula may fail. The identification $\mathcal{H}$1 relies on vanishing symplectic flux through $\mathcal{H}$2, which the authors note should follow from linearized stability results for gravitational perturbations but which they have not verified in detail. Soft radiation and memory effects are neglected throughout; the authors expect their framework extends via recently developed Tomita–Takesaki techniques incorporating soft modes, but this extension is not carried out. Finally, the trace normalization ambiguity inherent to Type-$\mathcal{H}$3 factors leaves an undetermined additive constant in the entropy, and the absence of a maximal-entropy state means no canonical reference entropy exists in this setting.

Conclusion

The paper constructs a well-defined von Neumann entropy for perturbed states of linearized quantum gravity on the right future horizon of a bifurcate Killing horizon, by adjoining the asymptotic HWZ boundary charge as an observer degree of freedom and passing to the Type-$\mathcal{H}$4 crossed product algebra. The resulting entropy obeys a first-law-type relation and is equal, at leading order, to the expectation value of the Hollands–Wald–Zhang dynamical black hole entropy at any horizon cut, up to a classical flux term and state-independent constants. This provides a direct bridge between algebraic-QFT entropy constructions and Noether-charge-based prescriptions for dynamical black hole entropy within perturbative quantum gravity.