Wald Entropy and Higher-Derivative Attractors

Higher-derivative interactions change both the extremal near-horizon fields and the entropy assigned to them. The area law is then replaced by a Noether-charge entropy, while Sen’s entropy function packages the corrected near-horizon equations into a finite-dimensional extremization. We derive both statements, check the Einstein limit, and mark the assumptions that fail for noncovariant terms or an uncontrolled derivative expansion.

Required background. Gauge Theory and String-Theory Preparation supplies gauge charges and effective interactions. Gravitational Actions and Boundary Terms supplies variational boundary terms and the covariant phase-space setting.

Helpful background. Quantum Gravity supplies the effective-field-theory interpretation, Quantum Information helps distinguish thermodynamic and fine-grained entropies, and Black-Hole Geometry and Mechanics supplies Killing horizons and the first law.

Noether-charge entropy

Let the diffeomorphism-invariant Lagrangian $D$-form depend locally on the metric, curvature, matter fields, and finitely many covariant derivatives. Its variation has the form

$$\delta L=E_\Phi\,\delta\Phi+d\Theta(\Phi,\delta\Phi).$$

For a vector field $\xi$, the Noether current is $J_\xi=\Theta(\Phi,\mathcal L_\xi\Phi)-\xi\mathbin{\lrcorner}L$. On shell, $J_\xi=dQ_\xi$. For a stationary black hole with a bifurcate Killing horizon, the horizon term in the first law identifies the entropy with the appropriately normalized charge on the bifurcation surface. Wald proves the Noether-charge construction in Wald 1993, pp. 5–10, Eqs. (2)–(20). Iyer and Wald give the general covariant variation and charge decomposition in Iyer and Wald 1994, §3, p. 6, Eq. (20), and §4, pp. 10–11, Eqs. (49)–(53), then prove the stationary bifurcate-horizon first law for arbitrary asymptotically flat linearized perturbations in Iyer and Wald 1994, §6, pp. 16–18, Theorem 6.1 and Eqs. (91)–(103).

When the Lagrangian scalar $\mathcal L$ depends algebraically on $R_{\mu\nu\rho\sigma}$ but not its derivatives, the result is

$$S_{\rm Wald} =-2\pi\int_{\mathcal H} \frac{\partial\mathcal L}{\partial R_{\mu\nu\rho\sigma}} \epsilon_{\mu\nu}\epsilon_{\rho\sigma}\sqrt h\,d^{D-2}x,$$

where $\epsilon_{\mu\nu}\epsilon^{\mu\nu}=-2$ for the binormal to a spatial horizon section. If $\mathcal L$ contains symmetrized covariant derivatives of the curvature through order $m$, replace the partial derivative by

$$\begin{aligned} E_R^{\mu\nu\rho\sigma} ={}&\frac{\partial\mathcal L}{\partial R_{\mu\nu\rho\sigma}} -\nabla_{\alpha_1} \frac{\partial\mathcal L} {\partial(\nabla_{\alpha_1}R_{\mu\nu\rho\sigma})} +\cdots\\ &+(-1)^m\nabla_{(\alpha_1}\cdots\nabla_{\alpha_m)} \frac{\partial\mathcal L} {\partial(\nabla_{(\alpha_1}\cdots\nabla_{\alpha_m)}R_{\mu\nu\rho\sigma})}. \end{aligned}$$

The entropy integrand then uses $E_R^{\mu\nu\rho\sigma}$ rather than $\partial\mathcal L/\partial R_{\mu\nu\rho\sigma}$. This alternating Euler derivative is Iyer and Wald 1994, §3, p. 7, Eq. (31). Their mostly-plus metric, curvature sign, and binormal norm match the global site package; their examples set $G=1$, so the factors of $G_D$ shown here must be restored. Reversing the binormal orientation leaves the quadratic stationary entropy unchanged.

Einstein-limit check

For

$$\mathcal L=\frac{1}{16\pi G_D}R, \qquad \frac{\partial\mathcal L}{\partial R_{\mu\nu\rho\sigma}} =\frac{1}{32\pi G_D} \left(g^{\mu\rho}g^{\nu\sigma}-g^{\mu\sigma}g^{\nu\rho}\right),$$

the binormal contraction equals $-4$. Substitution gives

$$S_{\rm Wald}=\frac{1}{4G_D}\int_{\mathcal H}\sqrt h\,d^{D-2}x =\frac{A_H}{4G_D}.$$

This is an indispensable normalization check. Metric and curvature conventions must be translated together with the action and binormal norm. Reversing the binormal orientation alone cannot change the entropy, because the binormal occurs twice.

Entropy function for an extremal throat

Assume an extremal, spherically symmetric near-horizon solution with $\mathrm{SO}(2,1)\times\mathrm{SO}(D-1)$ symmetry:

$$ds^2=v_1\left(-r^2dt^2+\frac{dr^2}{r^2}\right)+v_2d\Omega_{D-2}^2, \qquad \phi_s=u_s,\quad F^{(i)}_{rt}=e_i.$$

Magnetic charges $p_a$ fix the fluxes through the sphere. Evaluate the complete effective Lagrangian on this ansatz and integrate over the horizon:

$$f(u,v,e,p)=\int_{S^{D-2}}\sqrt{-g}\,\mathcal L\,d^{D-2}x.$$

The electric equations define $q_i=\partial f/\partial e_i$. The entropy function is the Legendre transform

$$\mathcal E(u,v,e;q,p)=2\pi\left(q_ie_i-f(u,v,e,p)\right).$$

The corrected throat is found from

$$\frac{\partial\mathcal E}{\partial u_s} =\frac{\partial\mathcal E}{\partial v_a} =\frac{\partial\mathcal E}{\partial e_i}=0, \qquad S_{\rm Wald}=\mathcal E_*.$$

Sen establishes these equations for general $\mathrm{AdS}_2\times S^{D-2}$ throats in Sen 2005, §2, pp. 4–9, Eqs. (2.1)–(2.29). Unlike the BPS first-order flow, this near-horizon derivation does not require supersymmetry. It does require the assumed isometries and an interpolation to a physically acceptable asymptotic solution.

Corrected attractors and charge definitions

Higher-derivative terms alter $f$, so they shift $u_s$, $v_a$, $e_i$, and the entropy simultaneously. Solving the corrected field values but inserting them into the area law is inconsistent. There is a useful first-order shortcut for the extremized entropy function at fixed charges: its leading first derivatives vanish, so the correction is obtained by evaluating the first-order entropy-function term at the uncorrected extremum, as the exercise below proves. This does not justify evaluating the entire Wald integrand on an uncorrected metric. In a direct Wald calculation, the leading area can change at first order and must be included along with the explicit higher-derivative integrand.

Chern–Simons terms require special care because the action may not be manifestly gauge invariant. One may dualize to gauge-invariant fields, reduce to a covariant lower-dimensional action, or use the appropriate generalized charge formalism. Maxwell, Page, and brane-source charges need not coincide. The $q_i$ in the Legendre transform must be the charge conjugate to the declared electric field in the chosen action.

Flat directions are possible: $\mathcal E$ may be stationary along a scalar direction while its extremal value remains fixed. Then entropy is attracted although that scalar is not. Multiple extrema can also exist, and the entropy function alone does not select the basin reached from a specified asymptotic boundary condition.

Regime and independent checks

The calculation is controlled when the retained derivative expansion is hierarchical. If a schematic correction $\alpha' R^2$ is comparable to $R$, omitted $\alpha'^2R^3$ terms may be equally important unless supersymmetry or an exact result supplies additional control. String loops and $\alpha'$ corrections are separate expansions.

Check the result by: reproducing $A/4G$ when higher-derivative couplings vanish; differentiating the full reduced function before imposing the extremum; matching the first law with the same asymptotic charges; verifying that field redefinitions change neither the on-shell entropy nor observables; and testing positivity and regularity of the corrected throat. The construction gives a stationary black-hole entropy. It is not by itself a proof of a nonlinear second law or a fine-grained von Neumann entropy; Iyer and Wald explicitly leave the general dynamical second-law question open.

Exercises

Suppose $\mathcal E(x;\lambda)=\mathcal E_0(x)+\lambda\mathcal E_1(x)$ and $x_0$ extremizes $\mathcal E_0$. Show that the entropy through first order is $\mathcal E_0(x_0)+\lambda\mathcal E_1(x_0)$ even though the corrected extremum is $x_*=x_0+O(\lambda)$.

Solution

Write $x_*=x_0+\lambda x_1+O(\lambda^2)$ and expand. The result is $\mathcal E(x_*;\lambda)=\mathcal E_0(x_0)+\lambda x_1\mathcal E_0'(x_0)+\lambda\mathcal E_1(x_0)+O(\lambda^2)$. Because $x_0$ is a leading-order extremum, $\mathcal E_0'(x_0)=0$, so the shift $x_1$ does not contribute at first order. It is still needed for corrected fields and for the entropy at second order. This reasoning fails if the leading extremum is singular or the expansion is nonuniform.

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. Open manuscript.
• Sen, Ashoke. “Black Hole Entropy Function and the Attractor Mechanism in Higher Derivative Gravity.” Journal of High Energy Physics 09 (2005): 038. DOI. Open manuscript.
• Wald, Robert M. “Black Hole Entropy Is the Noether Charge.” Physical Review D 48 (1993): R3427–R3431. DOI. Open manuscript.