Validity Beyond Einstein Vacuum Gravity

The four laws do not survive a change of gravitational theory by replacing “area” with a new symbol and leaving every proof untouched. Stationary first laws extend remarkably far through diffeomorphism symmetry and Noether charge. Dynamical second laws are much less universal: their increasing functional, energy condition, perturbative order, and horizon class depend on the theory. This page compares classical results only. “Entropy” below names the geometric Noether-charge functional that enters the mechanics law; no microscopic or semiclassical interpretation is assumed.

Required background. The Einstein Equation provides the reference theory and its null-focusing relation.

Helpful background. Exact Solutions of Einstein’s Equations supplies stationary solution families; Causality and Global Structure supplies event horizons and predictability; Flat-Background Hilbert Stress-Energy from Metric Variation supplies minimally coupled matter flux; and Black-Hole Geometry and Mechanics supplies Killing horizons and surface gravity.

The Noether-charge replacement for area

Consider a classical, diffeomorphism-covariant Lagrangian $D$-form depending on a metric, curvature, matter fields, and possibly covariant derivatives of these quantities. Define the Euler derivative of its Lagrangian scalar with respect to the Riemann tensor by $E^{abcd}$. On a stationary bifurcate Killing horizon, the Wald functional is

$$S_{\mathrm W} =-2\pi\int_B E^{abcd}\epsilon_{ab}\epsilon_{cd}\,dA,$$

where $\epsilon_{ab}\epsilon^{ab}=-2$. For the Einstein–Hilbert Lagrangian, $S_{\mathrm W}=A/(4G)$, so

$$\frac{\kappa}{2\pi}\,\delta S_{\mathrm W} =\frac{\kappa}{8\pi G}\,\delta A.$$

The general first law follows from equality of the covariant phase-space boundary terms at the horizon and infinity. Iyer and Wald construct the symplectic and Noether-charge forms in Iyer and Wald 1994, §§III–IV, Open PDF, printed pp. 848–854. For an asymptotically flat stationary background with a bifurcate Killing horizon, Iyer and Wald 1994, Theorem 6.1 and Eqs. (92)–(102), arXiv PDF pp. 17–18 establish the first law for an arbitrary smooth, nonsingular perturbation satisfying the linearized equations and asymptotic conditions; the perturbation need not itself be stationary. The boundary Hamiltonians must be well defined and integrable, while boundary conditions, couplings, and gauge-charge conventions remain fixed. Diffeomorphism covariance is the source of this breadth; it does not by itself prove that $S_{\mathrm W}$ increases during nonlinear evolution.

Stationary Noether-charge ambiguities vanish on a bifurcation surface and do not change the stationary entropy. Away from stationarity, extrinsic-curvature terms that vanish on a Killing horizon can distinguish candidate dynamical functionals. Jacobson, Kang, and Myers analyze the stationary invariance and field-redefinition behavior in Jacobson, Kang, and Myers 1994, §§2–3. Thus “use Wald entropy” is a complete stationary instruction but not always a unique nonlinear one.

What survives in different theories

Theory and data	Mechanics result	Increasing or varied quantity	Exact limitation
Einstein gravity with minimally coupled matter	Hawking area theorem	$A$	Nonlinear, but needs null convergence plus global predictability or generator completeness and regularity.
Stationary Einstein–Maxwell	Zeroth and first laws	$A/(4G)$ with $\Phi_H\delta Q$	Requires a fixed charge and gauge normalization; the first law is linear in perturbations.
General diffeomorphism-covariant gravity	Equilibrium first law	$S_{\mathrm W}$	Stationary bifurcate background, admissible linearized perturbation, and integrable boundary charges.
General higher-curvature metric–scalar theory with minimally coupled matter	Linearized second law	$S_{\mathrm W}$ plus fixed extrinsic-curvature ambiguity terms	Arbitrary dimension, linear perturbations of a stationary bifurcate Killing horizon, null-energy matter, and future settling.
Covariant higher-curvature gravity coupled nonminimally to scalar and vector fields	Linearized second law	A Jacobson–Kang–Myers-invariant dynamical entropy; explicit vector-dependent terms in the stated $f(\mathrm{Riemann})$ class	Linear perturbations of a bifurcate Killing horizon. The explicit formula assumes at most one derivative on each vector and is not obtained by copying the pure-curvature expression.
Selected theories such as suitable $f(R)$ models	Theory-specific nonlinear second laws	Model-dependent corrected entropy	Extra sign, regularity, and field-redefinition assumptions; not a theorem for arbitrary curvature actions.
Lovelock merger examples	Standard stationary entropy can drop at a topology-changing merger	No universal monotone supplied by the stationary formula alone	Counterexample concerns a particular theory, compactification, entropy prescription, and nonsmooth merger event.

The zeroth law also splits into two routes. Constancy of $\kappa$ on a bifurcate Killing horizon follows geometrically under the corresponding regularity and symmetry assumptions. A field-equation proof on a general stationary horizon must be rebuilt for the chosen higher-curvature equations and matter couplings. One cannot automatically reuse the Einstein dominant- energy argument after moving curvature terms to an “effective stress tensor.”

Adding minimally coupled Maxwell fields is comparatively controlled: the stationary first law acquires potential–charge work terms, while the Einstein area theorem uses the null energy of the total classical matter. Nonminimal couplings are different. They redistribute terms between the gravitational equations and matter, change $E^{abcd}$, and can make the relevant monotone a weighted area rather than area itself. Positivity of a convention-dependent effective stress tensor is not a substitute for proving the required focusing inequality.

The linearized second-law frontier

Wall proves that for a very general higher-curvature metric–scalar action with minimally coupled matter obeying the null energy condition, one can choose the dynamical ambiguity terms so that a local horizon functional increases to first order about a stationary bifurcate Killing horizon. The proof fixes a regular horizon gauge, uses the higher-curvature null equation to write the linearized source as two affine derivatives, and imposes future stationarity. The horizon sections are compact, or the perturbation has sufficient compact support for transverse boundary terms to vanish. The matter sector is minimally coupled in this 2015 theorem. The hypotheses and ambiguity problem appear in Wall 2015, Open PDF, pp. 2–3, equations (3)–(7); the monotonicity construction and entropy formula are in Wall 2015, Open PDF, pp. 4–5, equations (8)–(14).

Wall and Yan extend the linearized theorem to covariant higher-curvature gravity coupled to scalar and vector fields with arbitrary nonminimal metric couplings; the vectors need not be gauge fields. Their covariant-phase-space entropy is invariant under the Jacobson–Kang–Myers ambiguities. For $f(\mathrm{Riemann})$ theories with at most one derivative on each vector, the explicit entropy contains three additional terms involving the symmetric vector derivative, all of which vanish for gauge fields. Thus a Maxwell-like field can simplify the formula, but a generic vector cannot be handled by reusing the pure-curvature entropy Wall and Yan 2024, §§1 and 3–5.

This is a theorem, but it is a linearized theorem. It does not control terms quadratic in the perturbation, violent mergers, or arbitrary effective theories treated outside their perturbative regime. By contrast, Jacobson, Kang, and Myers establish quasi-stationary increase quite broadly and a nonlinear second law for restricted Ricci-polynomial theories in Jacobson, Kang, and Myers 1995, pp. 3518–3528. These are complementary domains, not competing universal answers.

There are also genuine warning examples. Sarkar and Wall construct Kaluza–Klein Lovelock merger scenarios in which the standard stationary entropy has an instantaneous decrease Sarkar and Wall 2011, §§2–3, pp. 124048-2–124048-5. Such examples do not show that every Lovelock process lacks every possible generalized functional. They do show that a stationary first-law expression and a linearized second law do not entail a general nonlinear second law.

The third law is still less portable. Even within Einstein–Maxwell–Vlasov, smooth critical collapse can form an extremal black hole Kehle and Unger 2024, §1.6, Theorem 2, Open PDF. A higher-curvature third law must therefore state its matter model, horizon notion, regularity, energy inequalities, and admissible process; no Noether-charge identity fills in those dynamical hypotheses.

Formulation-choice diagnostic in quadratic curvature gravity

Take

$$I=\frac{1}{16\pi G}\int d^Dx\sqrt{-g}\,f(R), \qquad f(R)=R+\alpha R^2.$$

On a stationary horizon cross-section with constant $R=R_H$, determine the horizon functional and its first-law variation when $\delta R_H=0$. Explain why varying area alone is generally insufficient.

Solution

The Noether-charge functional is

$$S_{\mathrm W} =\frac{1}{4G}\int_B f'(R)\,dA =\frac{1}{4G}\int_B(1+2\alpha R)\,dA.$$

Suppose $R=R_H$ is constant on $B$ and an admissible variation has $\delta R_H=0$. Then

$$\frac{\kappa}{2\pi}\delta S_{\mathrm W} =\frac{\kappa}{8\pi G}(1+2\alpha R_H)\delta A.$$

The Einstein area term is recovered when $\alpha=0$ or $R_H=0$. Otherwise, using $\kappa\delta A/(8\pi G)$ would miss the factor $f'(R_H)$. If $\delta R_H\ne0$, there is an additional area-weighted curvature variation; the correct first law follows from varying the full functional, not from guessing an effective Newton constant after the calculation.

Present boundary of the claim

This comparison covers results available through 2026-08-12. Across the common axes—dimension, theory, asymptotics, horizon regularity, data class, boundary conditions, observable, and perturbative order—the stable conclusion is: stationary Noether-charge first laws are broad; linearized second laws are broader than known nonlinear ones; and nonlinear monotonicity remains theory-specific, with counterexamples to naive extensions. Whether a general nonlinear monotone exists remains an open question, not a settled universal verdict. This assessment should be revised if a general nonlinear theorem, or a counterexample under explicit higher-curvature equations and regularity conditions, changes that separation.

Continue to Stationary Classification, Rigidity, and No-Hair Results to see how symmetry and regularity turn mechanics data into classification theorems.

References

• Iyer, Vivek, and Robert M. Wald. “Some Properties of Noether Charge and a Proposal for Dynamical Black Hole Entropy.” Physical Review D 50 (1994): 846–864. DOI.
• Jacobson, Ted, Gungwon Kang, and Robert C. Myers. “Increase of Black Hole Entropy in Higher Curvature Gravity.” Physical Review D 52 (1995): 3518–3528. DOI.
• Jacobson, Ted, Gungwon Kang, and Robert C. Myers. “On Black Hole Entropy.” Physical Review D 49 (1994): 6587–6598. DOI.
• Kehle, Christoph, and Ryan Unger. “Extremal Black Hole Formation as a Critical Phenomenon.” arXiv:2402.10190 [gr-qc] (2024). arXiv.
• Sarkar, Sudipta, and Aron C. Wall. “Second Law Violations in Lovelock Gravity for Black Hole Mergers.” Physical Review D 83 (2011): 124048. DOI.
• Wall, Aron C. “A Second Law for Higher Curvature Gravity.” International Journal of Modern Physics D 24 (2015): 1544014. DOI.
• Wall, Aron C., and Zihan Yan. “Linearised Second Law for Higher Curvature Gravity and Non-Minimally Coupled Vector Fields.” Physical Review D 110 (2024): 084005. DOI. Open PDF.