The existence of black holes perfectly embodies and necessitates the synchronicity between quantum mechanics and general relativity. In particular, the non-vanishing black-hole entropy indicates that there may exist a more fundamental set of microstates accounting for this entropy. However, the explicit form of these microstates may be hard to realize if one assumes only the degrees of freedom found in general relativity.

A framework which offers additional degrees of freedom is string theory. In particular, the fuzzball paradigm [1, 2, 3] states that a classical black hole should be replaced by objects in string theory, called fuzzballs, which are horizonless and, when averaged over, reproduce the same observable effects as a black hole of the same mass, angular momentum, and charges.¹¹1However, for extremal black holes with vanishing horizon radius the corresponding averaged configurations are larger than the stretched horizon [4, 5]. In general, fuzzballs are highly quantum and stringy configurations. However, analogous to how a system of atoms, through processes of stimulated emission, can be configured to produce a coherent monochromatic beam of photons, a laser, one can also consider very coherent fuzzball states which admit a smooth, geometric description. This subset of fuzzballs, often called microstate geometries [6, 7], can be described using just supergravity, the low-energy limit of string theory, and is the focus of this paper.

We work in the D1-D5 system: Starting with Type IIB string theory on²²2$\mathcal{M}_{4}$ can be either $T^{4}$ or $K3$. In this paper we use exclusively $T^{4}$. $\mathbb{R}^{4,1}\times S_{y}^{1}\times\mathcal{M}_{4}$ one wraps $N_{1}$ D1-branes along the $y$-circle and $N_{5}$ D5-branes along $S_{y}^{1}\times\mathcal{M}_{4}$. Such two-charge configurations preserve 8 out of the total 32 supercharges and we refer to them as 1/4-BPS. By adding momentum charge, $P$, along $S_{y}^{1}$, one breaks 4 additional supercharges leading to 1/8-BPS configurations. The degeneracy of bound states with such charges, as calculated at weak string coupling, matches the (exponent of) the Bekenstein-Hawking entropy of the corresponding D1-D5-P black hole [8, 9]. While this provides a match between the ensembles of black hole microstates and ensembles of brane configurations as the string coupling (or the gravitational coupling, $G_{N}$) is varied, we are interested in the fate of individual microstates.

The first $1/4$-BPS microstates were constructed in [10, 11, 12] by wrapping fundamental strings along $S_{y}^{1}$ and allowing for transverse oscillations with which the string carries momentum. These configurations can then be, through a series of S and T-dualities, related to the D1-D5 system [13, 14, 15, 16]. In the latter frame, the near-horizon region of (geometric) microstates contains an AdS₃ factor, which, through the AdS/CFT correspondence [17], allows us to relate the geometries to Ramond-Ramond (RR) ground states in the dual two-dimensional D1-D5 CFT [10, 11, 13, 14, 15].

The maximally spinning round supertube, perhaps the simplest of such solutions, is the starting point of our construction. By adding a small perturbation on top of this solution one can generate the additional momentum charge. Utilising the linear and upper-triangular structure of the relevant BPS equations [18, 19, 20, 21], one is then able to find the fully backreacted, $1/8$-BPS microstate geometries, called superstrata [22, 23, 24, 25, 26, 27, 28, 29]. When viewed as solutions of six-dimensional supergravity, in most superstrata the momentum is carried by a field in the tensor multiplet. However, when counted, such solutions represent a parametrically small contribution to the entropy of the D1-D5-P black hole [30, 31].

Recently, it has been shown that the momentum charge can also be carried by vector fields [7] with the corresponding set of microstate geometries being dubbed “vector superstrata” [32]. In this paper, we construct a new family of such vector superstrata. Our work aims to complete the set of $1/8$-BPS microstate geometries of the D1-D5-P black hole which can be constructed in non-orbifolded AdS₃ [23, 33]. We identify their dual CFT states and perform some rudimentary checks that support our identification. Furthermore, we develop a systematic method to solve the BPS equations and present an algorithmic way to solve a Laplace equation in spheroidal coordinates with arbitrarily complicated sources. We do so by making an informed ansatz based on smoothness of the geometries in the interior and their prescribed asymptotic behaviour. We use an appropriate basis expansion, which reduces the partial differential equation to a system of algebraic equations for the basis coefficients. We hope that the techniques presented will prove helpful for future constructions of microstate geometries and related computations.

In Section 2 we describe CFT states which are dual to the vector superstrata that we construct in this paper. In Section 3 we briefly review the relevant supersymmetric ansatz and the associated BPS equations that are solved by the new microstate geometries. In Section 4 we start with a vector perturbation and show how to find a fully backreacted solution. In Section 5 we employ a type of ‘bootstrap’ technique to find the explicit solutions. In the final section we discuss our results and conclude with some future directions. The appendices contain details omitted in the main body of the paper. In Appendix A we summarise the conventions used in the CFT picture. Appendix B contains the detailed calculations that explicitly show how the constructed geometries solve the aforementioned BPS equations. Finally, in Appendix C, we present in full detail a particular solution as an application of our newly developed method to solve the full system of BPS equations.

The bulk of this paper deals with solving BPS equations of six-dimensional supergravity in order to obtain regular solutions that asymptote to AdS${}_{3}\times S^{3}$. Through the AdS/CFT correspondence [17], such solutions have an equivalent description in terms of heavy states in the dual D1-D5 CFT. This is a two-dimensional supersymmetric CFT whose central charge is $c=\bar{c}=6\,N$, where $N=N_{1}\,N_{5}$ is the product of the integer number of D1-branes and D5-branes used to construct the system. In this section, we present states that are dual to the geometries that we construct in subsequent sections. We limit ourselves only to the information that is immediate to our construction, purposefully omitting a lot of details (or relegating them to Appendix A, where we also summarise our conventions).³³3A more thorough analysis can be found for example in [14, 15, 16, 34, 30, 29], while states related to microstate geometries with vector fields were discussed in detail in [32].

To describe the geometries in the CFT language, we use the free orbifold point. In the bulk this free CFT corresponds to strings on AdS₃ in the tensionless limit [35, 36, 37], while the dual of the supergravity theory is a strongly coupled CFT, over which we have little control. However, since all of our states preserve at least 4 supercharges, we rely on non-renormalisation theorems [38] to protect some properties of supersymmetric states – in particular, the charges under the symmetry group – from changing as we move in the moduli space of the theory.

The target space of the theory with $\mathcal{M}_{4}=T^{4}$ at the symmetric orbifold point is $(T^{4})^{N}/S_{N}$ – $N$ copies of a free, $c=\bar{c}=6$ CFT with target space $T^{4}$ identified under the permutation group $S_{N}$. Each individual copy can be thought of as living on an effective closed string, which we refer to as a strand, described by $t$ and $y$ coordinates that parameterise the boundary of AdS₃. The theory, including its supersymmetry group, $SU(1,1|2)_{L}\times SU(1,1|2)_{R}$, naturally splits into left-moving and right-moving sectors. The state on each strand, labelled by $(r)$, is then written as $|\Psi_{(r)},{\widetilde{\Psi}}_{(r)}\rangle_{k_{r}}$, where $\Psi$ and ${\widetilde{\Psi}}$ are the states in the left-moving and right-moving sectors, respectively. The integer $k_{r}$ denotes the length of the effective string: Namely, one is able to make an effectively longer strand by changing the boundary conditions on the periodic coordinate $y$. The strand length can be arbitrary as long as the total length of all strands combined is equal to $N$. The full state is then the permutation invariant product of states in all copies $(r)$, subject to this constraint (for a schematic representation see Figure 1).

In what follows, the invariance under permutations does not play a critical role, so we suppress the copy index $(r)$. We can thus write the full state as a product over all possible individual states on all possible strand lengths, which we can schematically write as

where $N_{k}^{\Psi,{\widetilde{\Psi}}}$ denotes the number of states $|\Psi,{\widetilde{\Psi}}\rangle_{k}$. The state (2.1) is constrained by

which ensures that the total strand length is $N$.

We are interested in supersymmetric heavy states in this theory – states which preserve a finite number of supercharges and whose conformal dimensions scale with the total central charge, $\Delta\sim c\sim N$. In the main text we work primarily in the Ramond sector of the theory, since states in this sector arise naturally when one considers the near horizon region of asymptotically flat supersymmetric black holes [39]. In particular, the 1/4-BPS Ramond ground states have a large degeneracy and their counting reproduces the entropy of the two-charge D1-D5 black hole [8, 40, 10]. These ground states are the starting point of our construction.

On each strand, one can have 16 different states, 8 of them are bosonic while the remaining 8 are fermionic. Let us begin by focusing on the set of four bosonic states which, following [30, 29], can be labelled by their eigenvalues under the $R$-symmetry group $SU(2)_{L}\times SU(2)_{R}$⁴⁴4The $R$-symmetry $SU(2)_{L}\times SU(2)_{R}\simeq SO(4)$ is associated with the isometries of the $S^{3}$ in the bulk. Intuitively one can think of labelling states with the $j$ and ${\overline{j}}$ eigenvalues as labelling them according to their angular momentum in two orthogonal planes of rotation in $S^{3}$.

with the eigenvalues under the $SL(2,\mathbb{R})_{L}\times SL(2,\mathbb{R})_{R}$ symmetry being $h={\bar{h}}=k/4$.⁵⁵5The eigenvalues under the conformal symmetries are connected to the isometries of the AdS₃ space and are related to the mass and linear momentum in this spacetime. In particular $\Delta=h+{\bar{h}}$ and $n_{P}=h-{\bar{h}}$. Coherent superpositions of states containing only these strands can be realised in the bulk as Lunin-Mathur geometries [10, 12, 13, 14, 15, 22, 25]. Perhaps the simplest geometry is the maximally spinning supertube, which is dual to a state consisting only of ${\left|{++}\right\rangle}_{1}$ strands

with the conformal dimension, $\Delta=h+{\bar{h}}$, indeed scaling with the central charge.

In this paper we construct new 1/8-BPS solutions via the standard six-dimensional superstratum method [22, 23, 24, 25, 26, 27, 28, 29, 32]. One begins by adding a small perturbation to the supertube geometry – in the CFT this corresponds to adding to the supertube state (2.4) a small amount of strands dual to the perturbation in the bulk. Then one acts on this new state with the generators $L_{n}$, $J_{n}^{i}$, $G_{n}^{\alpha A}$ of the left-moving supersymmetry group, $SU(1,1|2)_{L}$. This increases the momentum charge of the state, $n_{P}=h-{\bar{h}}$, while preserving some supersymmetry due the right-moving sector remaining unchanged. Finally, to obtain the fully backreacted solution in the bulk, one needs to solve the full equations of supergravity. On the CFT side, this corresponds to increasing the number of strands associated with the perturbation and taking a coherent superposition of such states – this ensures that the state has a valid description within supergravity.

The deformation that is central for this paper is described by the state⁶⁶6It is natural to arrive at this state by starting in the Neveu-Schwarz sector and then spectral flowing to the Ramond sector. We present this path in Appendix A.

whose quantum numbers are given by

and $k=2,3,\ldots$, $m=1,2,\ldots k-1$, and $n=0,1,2\ldots$. This state is based on a fermionic 1/4-BPS state $|+\dot{A}\rangle_{k}$ on a strand of length $k$. The index $\dot{A}=1,2$ denotes a doublet under the $SU(2)_{C}$, which is part of the symmetry related to the isometry of the $T^{4}$, $SU(2)_{B}\times SU(2)_{C}\simeq SO(4)_{I}$. The residual $SU(2)_{B}$ appears through the index $A=1,2$ of the supersymmetry generator $G_{-1}^{+A}$. The action of this generator is two-fold: It transforms the fermonic perturbation into a bosonic one and the two indices $A$ and $\dot{A}$ combine into a vector index of $SO(4)_{I}$. As a consequence, the gravity dual of (2.5) is described by a vector field when viewed in six-dimensional supergravity compactified on $T^{4}$.

The full state that we consider is then given by a coherent sum (see also Figure 2)

where

and $\sum{}^{\!{}^{\prime}}$ denotes the coherent sum.

As discussed in the introduction, generic distributions of $N_{a}$ and $N_{b}$ describe a highly quantum state which cannot be reliably described within supergravity. Taking a coherent superposition of such quantum states is necessary for the dual of (2.7) to have a valid description in terms of a geometric solution [14, 41, 15, 42, 25]. However, since the supergravity description is valid in the limit where $N_{a},N_{b}\gg 1$ the coherent sum is sharply peaked around its average values. To a good approximation we can then use these average values to describe the states (2.7) and its dual geometry. To lighten the notation, we still use $N_{a}$ and $N_{b}$ to denote these averages. In terms of these, the total conserved charges of the state (2.7) are

Since, $h>{\bar{h}}$, such states carry a non-trivial momentum charge

which should also be seen from the dual gravity description. Indeed, this will be a sanity check of our fully backreacted solutions.

Coherent superpositions of heavy states in the D1-D5 CFT have a dual representation in terms of supergravity solutions that are asymptotically AdS${}_{3}\times S^{3}\times T^{4}$. While in general this requires us to work with ten-dimensional supergravity, one usually assumes that the geometries do not depend on the coordinates of the four-torus, in which case the system reduces to six-dimensional supergravity.

The minimal field content that needs to be included in the lower-dimensional theory depends on the states one wants to describe. For the state (2.7), it is sufficient to look at minimal supergravity coupled to a tensor and a vector multiplet. The bosonic fields contained in such a theory are the metric $g_{MN}$, a two-form gauge field $B_{MN}$, a one-form vector field $A_{M}$, and a dilaton $\phi$.

Supersymmetric solutions of such theories are completely characterised [43, 44, 20]. The six-dimensional metric in the Einstein frame can be decomposed as

where $u$ and $v$ are null-coordinates that are related to the usual time, $t$, and the coordinate parametrising the asymptotic circle of AdS₃, $y\sim y+2\pi\,R_{y}$, through

The $ds_{4}^{2}$ denotes the line-element of a four-dimensional Euclidean space, which we refer to as the base space. The remaining fields in the theory can be written as

where $Z_{1}$, $Z_{2}$, $Z_{A}$, and ${\cal F}$ are scalar functions, while $\beta$, $\omega$, ${\tilde{A}}$ and $a_{2}$ are one-forms and $\gamma_{1}$ is a two-form on the base space. Supersymmetry also demands that (the components of) these ansatz quantities are independent of one null coordinate, which we choose to be $u$, so that the ansatz quantities can be functions of $v$ and the base space coordinates only.

The power of this decomposition is that one can recast the BPS equations as differential equations for the ansatz quantities on the four-dimensional base space. These equations can then be organised in several layers with an upper-triangular structure: The solutions to the previous layer act as (at most) quadratic sources for the next layer [18, 19, 20, 21]. Anticipating the form of the solutions, we now present a simplified version of the BPS equations that are relevant in the construction that follows.

The initial BPS equations that one needs to solve are in the so-called “zeroth layer”, which determine the base space and the one-form $\beta$. We take the base space to be given by flat $\mathbb{R}^{4}$ and the one-form to be $v$-independent, $\dot{\beta}=0$. This automatically satisfies all the equations in the zeroth layer, apart from the self-duality condition

where $d_{4}$ is the exterior derivative restricted to the base space and $*_{4}$ is the Hodge dual.⁷⁷7Our conventions for the Hodge dual of a $p$-form in $D$-dimensions are $\displaystyle*_{D}X_{p}~{}\equiv~{}\frac{1}{p!(D-p)!}\,\epsilon_{m_{1}\ldots m_{D-p},n_{D-p+1}\ldots n_{D}}\,X^{n_{D-p+1}\ldots n_{D}}\,e^{m_{1}}\wedge\ldots\wedge e^{m_{D-p}}\,,$ where $e^{m}$ denotes an orthonormal frame. This choice for $\beta$ and the base space metric, which we justify below, effectively reduces the remaining BPS equations to Laplace equations in four-dimensional flat space.

For the remaining layers, one can use the gauge symmetry of the theory to make convenient choices for the ansatz quantities [21]. However, we choose to work with gauge-invariant components of the field strengths

where it is convenient to introduce an additional two form

Using the form of the gauge field potentials and the definitions of the field strengths, one can show that the latter can be written as

$\widehat{G}_{1,2}$ are two-forms on the four-dimensional base space given by⁸⁸8In all expressions we have already assumed $\dot{\beta}=0$.

where we used a differential operator

The gauge-invariant forms $\Theta^{1,2}$, $\omega_{F}$, and ${\tilde{F}}$, that we introduced⁹⁹9$\Theta^{2}$ in this paper is denoted by $\widetilde{\Theta}^{2}$ in [21, 32]. The same difference in notation applies for ${\cal F}$ and $a_{2}$. are given by

and

We now turn to the construction of the bulk dual of the state (2.7). We start by finding the perturbation around the maximally spinning supertube that is dual to the strands (2.5). Within six-dimensional supergravity this perturbation is encoded in a one-form gauge field, $A$. By utilising the upper-triangular structure of the BPS equations, we find the fully backreacted solutions, corresponding to smooth horizonless microstate geometries. However, to obtain explicit solutions, one has to solve two partial differential equations whose closed-form solutions are not known. To remedy this, in the following section we provide a bootstrap-like method which in principle allows us to find all explicit solutions.