Finite-N, Stringy, and Nonperturbative Corrections

A classical AdS black hole is the leading term of several distinct expansions. Boundary $1/N$ corrections encode bulk quantum loops, inverse ’t Hooft coupling encodes higher-derivative string effects, and contributions exponential in N or the entropy can encode D-instantons, additional saddles, discreteness, or genuinely nonperturbative completion data. These orders must not be collapsed into one vague “quantum correction.” This assessment is current through 2026-08-20.

Required background. Gauge Theory and String-Theory Preparation supplies large-N topology, the ’t Hooft limit, and string perturbation theory used below.

Helpful background. Quantum Gravity frames semiclassical errors, Quantum Information clarifies finite-entropy recurrences, Black-Hole Geometry and Mechanics supplies the leading saddle, and Quantum Fields in Curved Spacetime distinguishes perturbative correlators from exact finite-volume evolution.

Three expansion parameters, not one

For AdS$_5\times S^5$ in a conventional normalization,

$$g_s\sim\frac{\lambda}{N}, \qquad \frac{\alpha'}{L^2}=\lambda^{-1/2}, \qquad \frac{G_5}{L^3}\sim\frac{1}{N^2}.$$

The planar, strong-coupling classical limit sends $N\to\infty$ and $\lambda\to\infty$ while retaining a hierarchy compatible with small $g_s$. A connected boundary observable has a topological expansion schematically

$$\mathcal A = \sum_{g=0}^{\infty}N^{2-2g}\, \mathcal A_g(\lambda), \qquad \text{at fixed }\lambda.$$

The genus label $g$ organizes closed-string loops. Each coefficient $\mathcal A_g(\lambda)$ has its own strong-coupling expansion, which may contain positive powers or logarithms of $\lambda$ as well as inverse powers. For example, relative to the classical action, a tree-level $R^4$ term scales as $\lambda^{-3/2}$, while its genus-one counterpart scales parametrically as $g_s^2\lambda^{-3/2}\sim\lambda^{1/2}/N^2$. This follows by translating both the genus weight and the string-length weight together. Restricting every $\mathcal A_g$ to nonpositive powers of $\lambda$ would lose that sector. Protected coefficients can vanish; the overall power of N also depends on the normalization of the chosen correlator.

For the near-extremal D3-brane free energy, the leading IIB $\alpha'^3R^4$ term produces a relative correction of order $\lambda^{-3/2}$. This is not a bulk loop. Conversely, a one-loop supergravity correction can survive at infinite $\lambda$ while being suppressed by $1/N^2$ relative to the classical result.

Results and limitations through 2026-08-20

Evidence or calculation	Controlled result	Limitation or adverse case
Aharony et al. 2000, §§1.2–1.3 and 3.1, pp. 10–21, 55–68	The ’t Hooft topological expansion maps planar diagrams to classical strings, handles to bulk loops, and large $\lambda$ to weak curvature in the D3 example.	The map is parametric and example-dependent; it does not specify an exact finite-N gravitational path integral or resum either asymptotic series.
Gubser, Klebanov, and Tseytlin 1998, §§2–3, pp. 214–222	The explicit $\alpha'^3$ correction to near-extremal D3 thermodynamics gives the leading $\lambda^{-3/2}$ change and checks the strong-coupling limit.	It is a leading derivative correction in a controlled background, not proof of convergence or a universal correction for every AdS black hole.
Maldacena 2003, §§2–4, pp. 4–15	Exact finite-entropy boundary evolution requires discreteness and late-time behavior absent from a single perturbative eternal-black-hole saddle.	Additional Euclidean saddles capture selected exponentially small effects but do not by themselves reproduce the exact quasiperiodic correlator or a unique interior completion.

These controlled calculations establish the perturbative hierarchy in declared regimes and demonstrate a real late-time mismatch. Proposed resurgent completions, complex saddles, ensemble descriptions, or exact finite-N interiors remain conjectural or model-dependent unless they reproduce a specified microscopic observable.

Exponential sectors and late times

Perturbation theory cannot see all relevant scales. A distinct saddle whose action differs by $\Delta I$ contributes $e^{-\Delta I}$; when $I$ scales as $N^2$, this is smaller than every power of $1/N$. D-instanton sectors instead carry weights set by $e^{-1/g_s}$, which become different functions of N and $\lambda$ in the ’t Hooft limit. Finite-entropy spectral effects can be of order $e^{-S}$ and become visible only at parametrically late times.

For a discrete spectrum,

$$G(t)=\frac{1}{Z}\sum_{m,n}e^{-\beta E_m} e^{-i(E_n-E_m)t}|\mathcal O_{mn}|^2$$

cannot decay smoothly to zero forever. The semiclassical quasinormal approximation is accurate over an intermediate time domain, not as an exact statement for arbitrary $t$. A late-time plateau, recurrence, or spectral form factor also depends on smoothing and ensemble choices; none is automatically a local interior observable.

Competing positions and failure of hierarchy

One position treats the classical saddle plus its genus and derivative expansions as a systematically improvable effective description, leaving exact completion to the boundary theory. Another seeks a bulk definition through a sum over additional real or complex saddles, resurgence, matrix integrals, or other nonperturbative structures. The competing positions agree on perturbative checks but differ on whether the bulk data alone uniquely specify the exponentially small sectors.

The hierarchy fails when nominal parameters cease to be small. If $\lambda$ is not large, massive string modes cannot be truncated. If $g_s\sim\lambda/N$ is not small, genus order is not suppressed even when curvature is. Near a phase transition, two saddle actions can be nearly equal, so a saddle labeled “exponentially subleading” away from the crossing becomes essential. Protected indices may be exact across these changes while unprotected correlators are not.

What remains unresolved

It remains open which bulk formulation gives a unique nonperturbative definition for a generic holographic CFT, how to identify all relevant complex saddles and contours, and which finite-N interior observables exist with state-independent accuracy. Evidence does not establish that one resummation prescription works for every observable or that an ensemble average equals a fixed boundary theory. The distinction between controlled asymptotic data and completion choices is therefore part of the scientific result.

What could change these conclusions

The conclusions could change if an exact finite-N boundary calculation is matched by a bulk construction beyond protected quantities, if an independent resurgence or saddle analysis fixes a previously ambiguous exponential sector, or if a counterexample shows that the stated genus or inverse-coupling hierarchy fails inside its claimed parameter regime. To interpret such a result, specify the observable, N and coupling limit, saddle contour, error scale, and whether it concerns a fixed theory or an ensemble.

Exercises

Determine how to classify a correction proportional to $N^0\lambda^{-3/2}$ relative to a leading free energy proportional to $N^2$.

Solution

Relative to $N^2$, the term is $N^{-2}\lambda^{-3/2}$: it lies at genus-one order in the stated topological normalization and is suppressed at strong coupling. These powers alone do not identify its bulk operator. In particular, the actual genus-one $R^4$ weight translates as $g_s^2\lambda^{-3/2}\sim\lambda^{1/2}/N^2$, so it is not this hypothetical term. Both limits must be taken inside a controlled hierarchy, such as $1\ll\lambda\ll N$; sending $\lambda\to\infty$ at fixed N eventually invalidates weak string coupling.

References

• Aharony, Ofer, Steven S. Gubser, Juan Maldacena, Hirosi Ooguri, and Yaron Oz. “Large N Field Theories, String Theory and Gravity.” Physics Reports 323 (2000): 183–386. DOI.
• Gubser, Steven S., Igor R. Klebanov, and Arkady A. Tseytlin. “Coupling Constant Dependence in the Thermodynamics of N=4 Supersymmetric Yang–Mills Theory.” Nuclear Physics B 534 (1998): 202–222. DOI.
• Maldacena, Juan. “Eternal Black Holes in Anti-de Sitter.” Journal of High Energy Physics 04 (2003): 021. DOI.