Gravitational Replica Saddles

The gravitational replica integral sums over bulk fields and metrics compatible with the replicated boundary data. Consequently, a connected bulk geometry can be an admissible saddle even though the nongravitating construction began with $n$ nominal copies. Replica symmetry then turns the quotient fixed locus into a cosmic brane. The calculation is controlled only after the action, topology class, symmetry, contour, state preparation, and semiclassical regime are stated.

Required background. Quantum Information supplies normalized moments, Rényi entropy, and the cyclic permutation observable.

Helpful background. Semiclassical Gravity supplies Euclidean saddle expansions and generalized entropy; Quantum Fields in Curved Spacetime supplies the regulated matter determinant; Black-Hole Information supplies the QES and Page-curve questions.

What is allowed to connect

Let $\mathcal B_n(R)$ denote the replicated asymptotic boundary conditions obtained by cyclically gluing the retained nongravitating region $R$. A semiclassical gravitational moment is organized as

$$Z_n[R] \simeq \sum_{\alpha\in\mathcal S_n} \mathcal N_\alpha(n) e^{-I_{\mathrm{ren}}[g_{n,\alpha}]} Z_{\mathrm{matter}}[g_{n,\alpha};R],$$

where $\mathcal S_n$ is the declared saddle set. Here $I_{\mathrm{ren}}$ is the gravitational action; $Z_{\mathrm{matter}}$ includes the matter fluctuations; and $\mathcal N_\alpha$ contains the remaining gravitational/ghost determinants, zero-mode measures, and multiplicities. If matter has instead been integrated into an effective action, its determinant must not be multiplied in a second time. The asymptotic boundaries remain those prescribed by $\mathcal B_n(R)$. What can connect is the integrated gravitating interior.

On a fixed background, there is no integral over topology or metric, so this new option is absent. Likewise, a nongravitating bath is not joined merely because a gravitational interior is connected. The gluing of bath fields remains fixed by the density-matrix trace.

A formal solution is only one ingredient. It contributes if it obeys all boundary and regularity conditions and lies on, or is reached by deforming, the chosen integration contour. Dominance then depends on the real part of the complete effective action, not just the classical Einstein term.

Replica quotient and the fixed locus

Assume a dominant saddle $g_n$ is invariant under the cyclic group $\mathbb Z_n$. The quotient $\widehat g_n=g_n/\mathbb Z_n$ has the original asymptotic boundary but a codimension-two fixed locus $C_n$ with opening angle $2\pi/n$. It can be described by a cosmic brane of tension

$$T_n=\frac{n-1}{4nG}.$$

The brane backreacts for $n\ne1$. Dong’s refined Rényi relation is

$$\widetilde S_n := n^2\partial_n\left(\frac{n-1}{n}S_n\right) = \frac{\operatorname{Area}(C_n)}{4G} +\text{quantum corrections},$$

with the higher-derivative replacement understood when Einstein gravity is not the action. The brane prescription and backreaction are derived in Dong 2016, Eqs. (4)–(10). In the $n\to1$ limit, $T_n\to0$; requiring a regular family of saddles gives extremality of the limiting fixed locus, as in Lewkowycz and Maldacena 2013, §§2–4.

The tension has a direct geometric check: the deficit angle is $8\pi G T_n=2\pi(1-1/n)$, leaving opening angle $2\pi/n$. The area computes the refined Rényi quantity displayed above, not $S_n$ itself. Recovering $S_n$ requires integrating this relation in $n$ with the proper normalization.

This quotient argument assumes replica symmetry. A symmetry-breaking saddle cannot be represented by one quotient cone and one brane, although it can still contribute to the integer moment.

Regularity is equally consequential. A conical description is permitted in the quotient because it represents a smooth fixed point upstairs; an unrelated curvature singularity is not automatically an admissible saddle. Boundary terms at asymptotic ends, interfaces, and the fixed locus must be included consistently before differentiating the action with respect to $n$.

Saddle competition and interpretation

For a replica-disconnected saddle, the gravitating interior is the disjoint union of the one-copy geometry. A connected saddle has a different topology and generally a different action. Defining

$$\Delta I_n =I_{\mathrm{conn}}(n)-I_{\mathrm{disc}}(n),$$

the connected-to-disconnected magnitude is approximately $\lvert\mathcal N_{\mathrm{conn}}/\mathcal N_{\mathrm{disc}}\rvert e^{-\operatorname{Re}\Delta I_n}$. Negative $\operatorname{Re}\Delta I_n$ favors the connected saddle, provided both saddles lie on the contour. Near equality, determinants and other saddles cannot be discarded.

The permissive interpretation treats connected saddles as genuine contributions to the fine-grained entropy of the gravitational system. A conservative interpretation regards the saddle calculation as reliable in the specified low-dimensional model while withholding a claim about a unique microscopic boundary theory. An ensemble interpretation says that connected correlators are natural because the path integral computes an average over theories or couplings. These are competing positions about the object computed; the semiclassical action comparison does not choose among them.

Results and open questions as of 2026-08-19

This discussion reflects results available through 2026-08-19.

Evidence or result	Model or assumptions	Limitation or adverse issue
Smooth replica saddle and quotient cone, Lewkowycz and Maldacena 2013, §§2–4	Replica-symmetric semiclassical holographic saddle with a regular $n\to1$ family	Symmetry breaking and uniqueness of continuation are not established generically
Cosmic-brane refined Rényi entropy, Dong 2016, Eqs. (4)–(10)	Classical Einstein gravity with brane backreaction, plus stated quantum corrections	A formal quotient solution still needs an allowed contour and complete saddle comparison
Connected evaporation saddles, Almheiri et al. 2020, §§2–5	JT gravity with specified baths and matter preparation	Controlled low-dimensional evidence, not a universal four-dimensional derivation
Entropy review synthesis, Almheiri et al. 2021, §§10–11	Semiclassical replica and island framework	Microscopic single-theory meaning and individual density-matrix elements remain unresolved

The sharp adverse check is to repeat the setup with the metric held fixed. The cyclic boundary condition still computes $\operatorname{Tr}\rho_R^n$, but no connected metric saddle appears; connectedness is therefore a consequence of gravitational integration, not of the trace identity alone. A second adverse check is contour exclusion: even a smooth connected solution contributes zero if its thimble has zero intersection with the original contour.

The conclusions could change after a nonperturbative contour prescription includes or excludes the relevant connected saddle, after a replica-symmetry-breaking solution wins near $n=1$, or after a microscopic fixed theory reproduces the moments with a demonstrably different saddle organization.

Exercises

At $n=2$, suppose $I_{\mathrm{disc}}=200$, $I_{\mathrm{conn}}=198$, and the determinant ratio is $\mathcal N_{\mathrm{conn}}/\mathcal N_{\mathrm{disc}}=e^{-3}$. Which saddle is larger in magnitude? What further fact is required before calling it dominant?

Solution

The magnitude ratio is $e^{-3}e^{-(198-200)}=e^{-1}$, so the disconnected saddle is larger despite the connected saddle’s smaller classical action. One must also know that both saddles lie on the chosen integration contour, including their phases and thimble coefficients. If the connected saddle has zero contour coefficient, the action comparison is irrelevant; if phases cause cancellations, magnitude alone does not determine the full sum.

References

• Almheiri et al. 2020, Ahmed Almheiri, Thomas Hartman, Juan Maldacena, Edgar Shaghoulian, and Amirhossein Tajdini, “Replica Wormholes and the Entropy of Hawking Radiation,” Journal of High Energy Physics 05 (2020) 013.
• Almheiri et al. 2021, Ahmed Almheiri, Thomas Hartman, Juan Maldacena, Edgar Shaghoulian, and Amirhossein Tajdini, “The Entropy of Hawking Radiation,” Reviews of Modern Physics 93 (2021) 035002.
• Dong 2016, Xi Dong, “The Gravity Dual of Rényi Entropy,” Nature Communications 7 (2016) 12472.
• Lewkowycz and Maldacena 2013, Aitor Lewkowycz and Juan Maldacena, “Generalized Gravitational Entropy,” Journal of High Energy Physics 08 (2013) 090.