Loop, Curvature, and Nonperturbative Corrections

String corrections do not form a single series. A black-hole observable can receive string-loop or genus corrections, alpha-prime derivative corrections, compactification effects, and exponentially small nonperturbative terms, each controlled by a different dimensionless quantity. The correct approximation is therefore a hierarchy, not the slogan “large black hole” or “small coupling.” This page separates the expansions and shows how one can fail while another remains controlled.

Required background. Gauge Theory and String-Theory Preparation establishes worldsheet topology, the dilaton coupling, and the effective string action whose separate expansions are compared here.

Helpful background. Quantum Gravity supplies effective-field-theory power counting; Quantum Information is useful when the observable is an entropy or index; and Black-Hole Geometry and Mechanics provides invariant horizon and curvature scales.

Genus and spacetime loops

For a closed oriented worldsheet of genus $h$, the constant dilaton contributes

$$\mathcal A_h\propto g_s^{\,2h-2}.$$

After extracting the tree contribution, each additional closed-string handle nominally costs $g_s^2$. In a background with $N$ flux units or many light species, the actual loop parameter can be enhanced. For AdS$_5\times S^5$, the dictionary reorganizes the genus series as a $1/N^2$ expansion at fixed ’t Hooft coupling, as reviewed in Aharony et al. 2000, §1.2, pp. 10–15.

The relevant coupling is local: $e^{\Phi(x)}$ may grow toward a core even if $g_s=e^{\Phi_\infty}$ is small. Duality can provide a weakly coupled frame, but only after transforming the entire charge, radius, and observable package. A small asymptotic $g_s$ by itself does not establish loop control.

Alpha-prime and curvature corrections

Integrating out massive string modes produces a derivative expansion,

$$S_{\mathrm{eff}} =\frac{1}{16\pi G_D}\int\sqrt{-g}\, \left[R+\alpha'c_1\mathcal R^2+\alpha'^2c_2\mathcal R^3+\cdots\right].$$

Supersymmetry can forbid the schematic first terms; in type II theories the familiar leading purely gravitational correction is of $\alpha'^3R^4$ type. The parametric lesson is unchanged. If $L_{\mathcal R}$ is the shortest invariant curvature radius, each additional derivative pair contributes a power of

$$\epsilon_{\alpha'}=\frac{\alpha'}{L_{\mathcal R}^2}.$$

Weak string coupling does not imply small $\epsilon_{\alpha'}$, and weak curvature does not imply a small string-loop parameter. The distinction is especially important along charged-brane flows, where the dilaton and curvature scale vary oppositely. Peet 2000, §4.4, pp. 43–45 uses D$p$-brane limits to exhibit precisely this loss of control.

Field redefinitions can move local higher-derivative operators among one another. Consequently, an individual coefficient or corrected coordinate radius need not be invariant. Compare observables such as conserved charges, temperature at fixed ensemble, scattering data, or Wald entropy after translating the fields and boundary terms consistently.

Nonperturbative sectors

Nonperturbative corrections are not higher powers in either perturbative series. Typical weights include

$$e^{-A_\Sigma/(2\pi\alpha')}, \qquad e^{-2\pi/g_s}, \qquad e^{-S_{\mathrm{E,grav}}},$$

for a worldsheet instanton wrapping area $A_\Sigma$, a D-instanton-like sector, and a spacetime Euclidean saddle. A term can be exponentially small throughout a parametric limit yet dominate an exactly vanishing perturbative contribution. Conversely, when a cycle collapses, the worldsheet instanton action need not be large even though bulk curvature is mild.

The distinction between perturbative and nonperturbative is tied to the expansion variable. A D-brane effect nonperturbative in $g_s$ may become perturbative in a dual open-string or gauge-theory description. The physical statement names the chosen frame and the observable rather than assigning an absolute label.

A practical comparison

For an observable $\mathcal O$ in a charged background, organize the estimate as

$$\frac{\delta\mathcal O}{\mathcal O_0} \sim a\,\epsilon_{\alpha'}^{\,k} +b\,\epsilon_{\mathrm{loop}} +c\,e^{-S_{\mathrm{inst}}} +d\,\epsilon_{\mathrm{KK}}+\cdots .$$

Each coefficient may depend on protected charges and moduli. The table is meaningful only after declaring the local curvature, dilaton, flux scaling, compactification spectrum, ensemble, and whether supersymmetry protects the observable. A BPS index can be invariant while the local metric has large corrections; protection of one quantity does not repair an uncontrolled derivative expansion for another.

Two limiting checks are decisive. Holding $g_s\ll1$ while shrinking $L_{\mathcal R}$ toward $\ell_s$ suppresses loops but destroys the alpha-prime series. Holding $L_{\mathcal R}\gg\ell_s$ while raising $e^\Phi$ suppresses derivative corrections but destroys the genus series. Any claimed all-purpose “supergravity regime” must exclude both paths.

Exercises

An observable has estimated corrections $0.8(\alpha'/L^2)^3$, $5g_s^2N$, and $20e^{-2\pi/g_s}$. For $L/\ell_s=3$, $g_s=0.08$, and $N=10$, compute and compare the three corrections, then determine whether a one-percent calculation is controlled.

Solution

The derivative term is $0.8/3^6\approx1.10\times10^{-3}$. The loop estimate is $5(0.08)^2(10)=0.32$, so the charge enhancement defeats the small bare coupling. The nonperturbative term is $20e^{-2\pi/0.08}$, far below any displayed decimal precision. Thus the ordering is loop, derivative, then nonperturbative. Because the loop correction is roughly thirty-two percent, the proposed one-percent calculation is not controlled even though curvature and instanton effects are small. The comparison also shows why quoting $g_s=0.08$ without the effective multiplicity would give the wrong conclusion.

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. Open manuscript.
• Peet, Amanda W. “TASI Lectures on Black Holes in String Theory.” 2000. arXiv:hep-th/0008241. Open manuscript.