D-Branes and Open–Closed Descriptions

A D-brane is simultaneously a hypersurface on which open strings end and a charged source for closed-string supergravity fields. These are not competing models: they are two perturbative organizations of the same object. The open description is natural when backreaction is weak; the closed description is natural when many branes create a large, weakly curved throat. Matching tension, R–R charge, and exchange amplitudes provides the bridge.

Required background. Gauge Theory and String-Theory Preparation supplies open strings, Chan–Paton labels, and gauge fields; Differential Forms and Integration supplies pullbacks and the Wess–Zumino form coupling.

Helpful background. Quantum Gravity frames the dual perturbative descriptions; Quantum Information supports later state counting; and Exact Spacetimes provides the backreacted geometries.

The worldvolume description

For one D$p$-brane in string frame,

$$S_{\mathrm{DBI}} =-\mu_p\int d^{p+1}\xi\,e^{-\Phi} \sqrt{-\det\!\left(P[g+B]_{ab}+2\pi\alpha'F_{ab}\right)},$$

and

$$S_{\mathrm{WZ}}=\mu_p\int_{\mathcal W}P\!\left[\sum_qC_q\right] \wedge e^{B+2\pi\alpha'F}.$$

Here $\mu_p=[(2\pi)^p\alpha'^{(p+1)/2}]^{-1}$, while the physical asymptotic tension is $\mu_p/g_s$. The Wess–Zumino coefficient equals the R–R charge in this normalization, the key observation established by Polchinski 1995, Eqs. 8–10 and pp. 4725–4727.

For $N$ coincident branes, Chan–Paton factors give a $U(N)$ gauge theory. Expanding the determinant about flat space yields

$$S_{\mathrm{DBI}}\supset -\frac{1}{2g_{\mathrm{YM}}^2} \int d^{p+1}x\,\operatorname{Tr}F_{ab}F^{ab}, \qquad g_{\mathrm{YM}}^2=2(2\pi)^{p-2}g_s\alpha'^{(p-3)/2},$$

with the fundamental trace normalized by $\operatorname{Tr}(T^AT^B)=\delta^{AB}/2$. This gives the component action $-\int F^A_{ab}F^{A\,ab}/(4g_{\rm YM}^2)$, where $A$ is an adjoint gauge label and $a,b$ are worldvolume indices. In particular, $g_{\rm YM}^2=4\pi g_s$ for D3-branes, matching $L^4/\alpha'^2=\lambda=g_{\rm YM}^2N$ used throughout the D3 thread. The also-common matrix convention $-\operatorname{Tr}F^2/(4\widetilde g^2)$ has $\widetilde g^2=g_{\rm YM}^2/2$; the physical action is identical when both factors are translated. Higher powers of $F$, transverse scalar gradients, and commutators are controlled by $\alpha'$ and the separation scale.

The closed-string description

For $0\leq p\lt7$, the same stack sources the metric, dilaton, and a R–R $(p+1)$-form. Its extremal solution contains

$$H_p(r)=1+\frac{c_p g_sN\alpha'^{(7-p)/2}}{r^{7-p}},$$

where $c_p$ depends on sphere-volume conventions. The coefficient follows from the same tension and charge measured in the worldvolume action. At large $r$, linearized fields reproduce exchange of massless closed strings. At small $r$ and sufficiently large effective charge, the fields must be treated nonlinearly as a backreacted geometry.

The two descriptions and their parameter windows are compared in Aharony et al. 2000, §1.3.1–1.3.2, pp. 16–21. The open description can be strongly interacting even when individual string loops are small because planar diagrams scale with $g_sN$. The closed geometry can be classical when curvature and local dilaton are small, typically in a large-$N$ regime.

Open–closed channel duality

The one-loop vacuum diagram of an open string stretched between two branes is a cylinder. A modular transformation exchanges its long and short directions, reinterpreting the same worldsheet as tree-level closed-string propagation between boundary states:

$$\mathcal A_{\mathrm{open\ loop}}(t) =\mathcal A_{\mathrm{closed\ exchange}}(1/t).$$

At large separation only massless closed modes survive, and their force is determined by the brane tension and R–R charge. For parallel BPS branes, NS–NS attraction cancels R–R repulsion. This cancellation independently checks both the relative normalization and the preserved supersymmetry.

Open–closed duality does not say that a truncated weakly coupled gauge theory equals an arbitrary classical brane metric. One must take a decoupling limit and identify the retained energy, coupling, and boundary data before claiming an autonomous dual pair.

Probe and backreaction regimes

A probe approximation requires the added brane’s stress and charge to change the background only perturbatively. For a stack, the rough backreaction strength is $g_sN$ times a distance-dependent power. The comparison axes are:

• $g_s\ll1$ for fundamental closed-string loops;
• $g_sN$ or the effective ’t Hooft coupling for collective open-string interactions;
• $\alpha'|\mathcal R|\ll1$ for the supergravity derivative expansion;
• separation in string units for the mass of stretched open strings;
• energy below the open-string oscillator and unwanted Kaluza–Klein scales.

Thus $g_s\ll1$ and $g_sN\gg1$ can coexist. That is precisely the useful large-$N$ window: individual string interactions are weak, collective brane backreaction is strong, and a smooth throat may emerge.

The overlap is tested rather than assumed. Compute a protected tension or absorption amplitude in the open channel, translate every $2\pi$, trace, and state normalization, and compare it with flux or wave propagation in the closed solution. Agreement of one protected quantity does not imply that all finite-energy correlators agree outside the decoupling regime. Conversely, a singular classical core need not invalidate the open description; it can indicate that the open degrees of freedom are the better variables at that scale.

Exercises

Expand the DBI action to quadratic order in a slowly varying gauge field and derive the scaling of $g_{\mathrm{YM}}^2$. Then compare the result with the D3 case.

Solution

For $X^a{}_b=2\pi\alpha' g^{ac}F_{cb}$, $\sqrt{\det(1+X)}=1-\tfrac14\operatorname{tr}X^2+\cdots$, and $\operatorname{tr}X^2=-(2\pi\alpha')^2F_{ab}F^{ab}$. The DBI term is therefore $-\mu_p(2\pi\alpha')^2\operatorname{Tr}F^2/(4g_s)$. Comparing with $-\operatorname{Tr}F^2/(2g_{\rm YM}^2)$ gives $1/g_{\rm YM}^2=\mu_p(2\pi\alpha')^2/(2g_s)$, hence $g_{\rm YM}^2=2(2\pi)^{p-2}g_s\alpha'^{(p-3)/2}$. For $p=3$, $g_{\rm YM}^2=4\pi g_s$ is dimensionless and $\lambda=4\pi g_sN$. The independent dimensional check is $[g_{\rm YM}^2]=L^{p-3}$. The trace over spacetime indices in the determinant expansion and the gauge trace $\operatorname{Tr}$ play different roles.

Physically, the gauge field is a low-energy open-string excitation, so its kinetic coefficient inherits the disk tension. Changing the normalization of gauge generators changes the coupling label, while the action and measurable interaction strengths stay the same. This is why the numerical coupling cannot be compared across references without the trace convention.

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.
• Polchinski, Joseph. “Dirichlet-Branes and Ramond–Ramond Charges.” Physical Review Letters 75 (1995): 4724–4727. DOI.