math0702205/ESI.tex
1: \documentclass[10pt]{emsart}
2: \usepackage{amsmath,amssymb,amstext,array}
3: \usepackage[latin1]{inputenc}
4: %\usepackage[T1]{fontenc}
5: \usepackage{makeidx}
6: \makeindex
7: 
8: 
9: \contact[\texttt{j.m.figueroa@ed.ac.uk}]{José Figueroa-O'Farrill\\ School of
10:   Mathematics\\ and Maxwell Institute for Mathematical Sciences\\
11:   University of Edinburgh\\ United Kingdom} 
12: 
13: 
14: %%%%%%%%%%%
15: 
16: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
17: % Commands and environments
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19: \DeclareMathOperator{\Spin}{Spin}
20: \DeclareMathOperator{\CW}{CW}
21: \DeclareMathOperator{\NW}{NW}
22: \DeclareMathOperator{\KG}{KG}
23: \DeclareMathOperator{\rank}{rank}
24: \DeclareMathOperator{\dvol}{dvol}
25: \DeclareMathOperator{\im}{Im}
26: \DeclareMathOperator{\SU}{SU}
27: \DeclareMathOperator{\U}{U}
28: \DeclareMathOperator{\SO}{SO}
29: \DeclareMathOperator{\Ort}{O}
30: \DeclareMathOperator{\SL}{SL}
31: \DeclareMathOperator{\Sp}{Sp}
32: \DeclareMathOperator{\Mat}{Mat}
33: \DeclareMathOperator{\End}{End}
34: \DeclareMathOperator{\Hom}{Hom}
35: \DeclareMathOperator{\Ric}{Ric}
36: \DeclareMathOperator{\Riem}{Riem}
37: \DeclareMathOperator{\dS}{dS}
38: \DeclareMathOperator{\AdS}{AdS}
39: \DeclareMathOperator{\sech}{sech}
40: \DeclareMathOperator{\sS}{\mbox{\Large\textsf{S}}}
41: \DeclareMathOperator{\fS}{\mathfrak{S}}
42: \newcommand{\cyclicsum}{\sS\displaylimits}
43: \newcommand{\fd}{\mathfrak{d}}
44: \newcommand{\fg}{\mathfrak{g}}
45: \newcommand{\fm}{\mathfrak{m}}
46: \newcommand{\fn}{\mathfrak{n}}
47: \newcommand{\fh}{\mathfrak{h}}
48: \newcommand{\fk}{\mathfrak{k}}
49: \newcommand{\fp}{\mathfrak{p}}
50: \newcommand{\fso}{\mathfrak{so}}
51: \newcommand{\fsu}{\mathfrak{su}}
52: \newcommand{\fu}{\mathfrak{u}}
53: \renewcommand{\d}{\partial}
54: \newcommand{\eS}{\mathcal{S}}
55: \newcommand{\half}{\tfrac12}
56: \newcommand{\Cl}{\mathrm{C}\ell}
57: \newcommand{\CC}{\mathbb{C}}
58: \newcommand{\ZZ}{\mathbb{Z}}
59: \newcommand{\EE}{\mathbb{E}}
60: \newcommand{\RR}{\mathbb{R}}
61: \newcommand{\HH}{\mathbb{H}}
62: \newcommand{\bD}{\boldsymbol{D}}
63: \newcommand{\ebS}{\boldsymbol{\eS}}
64: \newcommand{\bS}{\boldsymbol{S}}
65: \newcommand{\blambda}{\boldsymbol{\lambda}}
66: \newcommand{\bzero}{\boldsymbol{0}}
67: \newcommand{\1}{\boldsymbol{1}}
68: \newcommand{\bH}{\boldsymbol{H}}
69: \newcommand{\be}{\boldsymbol{e}}
70: \newcommand{\bv}{\boldsymbol{v}}
71: \newcommand{\bw}{\boldsymbol{w}}
72: \newcommand{\bx}{\boldsymbol{x}}
73: \newcommand{\by}{\boldsymbol{y}}
74: \newcommand{\bz}{\boldsymbol{z}}
75: \newcommand{\eH}{\mathcal{H}}
76: \newcommand{\eL}{\mathcal{L}}
77: \newcommand{\eM}{\mathcal{M}}
78: \newcommand{\eV}{\mathcal{V}}
79: %
80: %\newcommand{\idx}[1]{\textbf{#1}\index{#1}}
81: \newcommand{\idx}[2]{\index{#2}{\bfseries #1}}
82: %
83: \numberwithin{equation}{section}
84: \newtheorem{theorem}{Theorem}[section]
85: \newtheorem{corollary}[theorem]{Corollary}
86: \newtheorem{lemma}[theorem]{Lemma}
87: \newtheorem{proposition}[theorem]{Proposition}
88: \newtheorem{conjecture}[theorem]{Conjecture}
89: 
90: \theoremstyle{definition}
91: \newtheorem{definition}[theorem]{Definition}
92: \newtheorem{remark}[theorem]{Remark}
93: 
94: 
95: \title[Lorentzian symmetric spaces in supergravity]{Lorentzian
96:   symmetric spaces in supergravity}
97: 
98: \author[José Figueroa-O'Farrill]{José Figueroa-O'Farrill}
99: 
100: \begin{document}
101: 
102: \begin{abstract}
103:   I will discuss the emergence of lorentzian symmetric spaces as
104:   supersymmetric supergravity backgrounds. I will focus on
105:   supergravity theories in dimension 11, 10, and 6, and will
106:   concentrate on the determination of the so-called maximally
107:   supersymmetric backgrounds, for which a classification exists up to
108:   local isometry.  A special class of lorentzian symmetric spaces also
109:   plays a rôle in the determination of parallelisable supergravity
110:   backgrounds in type II supergravity, which I will also summarise.\\
111:   (To appear in the proceedings of the programme \emph{Geometry of
112:     pseudo-riemannian manifolds with application to physics} hosted by
113:   the Erwin Schrödinger International Institute for Mathematical
114:   Physics.)
115: \end{abstract}
116: 
117: \begin{classification}
118:   Primary 53-B30; Secondary 83-50.
119: \end{classification}
120: 
121: \begin{keywords}
122:   Lorentzian geometry; supergravity.
123: \end{keywords}
124: 
125: \maketitle
126: \tableofcontents
127: 
128: \section{Introduction}
129: 
130: The purpose of this short review is to highlight the rôle of
131: lorentzian geometry in the supergravity limit of string theories.
132: That lorentzian geometry plays a rôle in such theories should not come
133: as a surprise, given that the supergravity theories in question
134: \emph{are} lorentzian theories; that is, some (if not all) of the
135: gravitational degrees of freedom are encoded in the form of a (local)
136: lorentzian metric.  What may be a little surprising is the fact that,
137: with notable exceptions, until relatively recently the lorentzian
138: nature of the solutions had not been fully exploited.  Indeed, most
139: early papers studying solutions to the supergravity field equations
140: concentrated on decomposable geometries $L \times R$, where $L$ is a
141: lorentzian space-form (e.g., Minkowski or (anti) de Sitter spacetime)
142: and $R$ a riemannian manifold, which would invariably become the focus
143: of the ensuing analysis.  This is not to say that such solutions are
144: geometrically or physically uninteresting.  In fact, they have
145: motivated the study of a large class of riemannian geometries which
146: otherwise might have remained largely in obscurity: Calabi--Yau
147: manifolds, manifolds with $G_2$ and $\Spin(7)$ holonomy, as well as
148: manifolds whose metric cones have such holonomies, e.g.,
149: Sasaki--Einstein and nearly Kähler manifolds, among others.  However
150: they do miss interesting solutions, as we will try to illustrate in
151: this review.
152: 
153: This review is organised as follows.  In Section~\ref{sec:geos} we
154: introduce the geometries of interest, namely lorentzian symmetric
155: spaces and in particular thoese which admit an absolute parallelism,
156: which translates into the question of which of these spaces are Lie
157: groups admitting bi-invariant lorentzian metrics.  In Section
158: \ref{sec:sugra} we discuss the geometrical aspects of supergravity
159: theories.  Much more could and should eventually be written about
160: this, but for the purposes of this review we will limit ourselves to
161: treat supergravity theories as collections of geometric PDEs whose
162: form is highly constrained, despite at first seeming ad hoc.  For
163: reasons explained in the body of the review, we will consider only the
164: following supergravity theories: eleven-dimensional supergravity,
165: ten-dimensional type IIB and the chiral six-dimensional $(1,0)$ and
166: $(2,0)$ supergravities.  In Section \ref{sec:maximal} we discuss the
167: classification of maximally supersymmetric solutions of the above
168: theories and also of ten-dimensional type IIA supergravity, which can
169: be obtained from eleven-dimensional supergravity via
170: Ka{\l}u\.za--Klein reduction, a technique we review in
171: Section~\ref{sec:kk}.  Finally in Section~\ref{sec:parallel} we
172: discuss parallelisable backgrounds in the common sector of type II
173: supergravity.
174: 
175: \subsection*{Acknowledgments}
176: 
177: Most of the work described in this review was obtained in
178: collaboration with a number of colleagues whom it is a pleasure to
179: thank and remember: Matthias Blau, Ali Chamseddine, Chris Hull, Kawano
180: Teruhiko, Patrick Meessen, George Papadopoulos, Simon Philip, Wafic
181: Sabra, Joan Simón, Sonia Stanciu and Yamaguchi Satoshi.  A lot of this
182: work was either done or started while I was a guest and/or
183: co-organiser of scientific programmes hosted by the Erwin Schrödinger
184: Institute for Mathematical Physics in Vienna and it is again my
185: pleasure to thank the ESI staff for all their support.  It is only
186: fitting that this review should appear in a volume born out of one
187: such programme and I would like to take this opportunity to reiterate
188: my thanks to Dmitri Alekseevsky and Helga Baum for the chance to
189: participate in it.
190: 
191: \section{The geometries of interest}
192: \label{sec:geos}
193: 
194: In this section we will quickly review the geometries of interest:
195: lorentzian symmetric spaces and, in particular, those which are
196: parallelisable.
197: 
198: \subsection{Lorentzian symmetric spaces}
199: \label{sec:symmetric}
200: 
201: We start by reviewing the classification of lorentzian symmetric
202: spaces.
203: 
204: The classification of symmetric spaces in indefinite signature is
205: hindered by the fact that there is no splitting theorem saying that if
206: the holonomy representation is reducible, the space is locally
207: isometric to a product.  In fact, local splitting implies both
208: reducibility \emph{and} a nondegeneracy condition on the factors
209: \cite{Wu}.  This means that one has to take into account reducible yet
210: indecomposable holonomy representations.  The general semi-riemannian
211: case is still open, but indecomposable lorentzian symmetric spaces
212: were classified by Cahen and Wallach \cite{CahenWallach} almost four
213: decades ago.  Indeed, they stated the following theorem
214: 
215: \begin{theorem}[Cahen--Wallach \cite{CahenWallach}]
216:   \label{thm:CahenWallach}
217:   Let $(M,g)$ be a simply-connected lorentzian symmetric space.  Then
218:   $M$ is isometric to the product of a simply-connected riemannian
219:   symmetric space and one of the following:
220:   \begin{itemize}
221:   \item $\RR$ with metric $-dt^2$;
222:   \item the universal cover of $n$-dimensional
223:     \idx{de~Sitter}{lorentzian symmetric space!de~Sitter} or
224:     \idx{anti~de~Sitter}{lorentzian symmetric space!anti~de~Sitter}
225:     spaces, where $n\geq 2$; or
226:   \item a \idx{Cahen--Wallach space}{lorentzian symmetric
227:       space!Cahen--Wallach} $\CW_n(A)$ with $n\geq 3$ and metric given
228:     by \eqref{eq:CWmetric} below.
229:   \end{itemize}
230: \end{theorem}
231: If we drop the hypothesis of simply-connectedness then this theorem
232: holds up to local isometry, which is the version of the theorem of
233: greater relevance in supergravity.
234: 
235: The $n$-dimensional Cahen--Wallach spaces $\CW_n(A)$ are constructed
236: as follows.  Let $V$ be a real vector space of dimension $n-2$ endowed
237: with a euclidean structure $\left<-,-\right>$.  Let $V^*$ denote its
238: dual. Let $Z$ be a real one-dimensional vector space and $Z^*$ its
239: dual.  We will identify $Z$ and $Z^*$ with $\RR$ via canonical dual
240: bases $\{e_+\}$ and $\{e_-\}$, respectively.  Let $A \in S^2V^*$ be a
241: symmetric bilinear form on $V$.  Using the euclidean structure on $V$
242: we can associate with $A$ an endomorphism of $V$ also denoted $A$:
243: \begin{equation*}
244:  \left< A(v), w\right> = A(v,w)\qquad\text{for all $v,w\in V$.}
245: \end{equation*}
246: We will also let $\flat:V \to V^*$ and $\sharp: V^* \to V$ denote the
247: musical isomorphisms associated to the euclidean structure on $V$.
248: 
249: Let $\fg_A$ be the Lie algebra with underlying vector space $V \oplus
250: V^* \oplus Z \oplus Z^*$ and with Lie brackets
251: \begin{equation}
252:   \label{eq:gA}
253:   \begin{aligned}[m]
254:     [e_-, v] &= v^\flat\\
255:     [e_-, \alpha] &= A(\alpha^\sharp)\\
256:     [\alpha, v] &= A(v,\alpha^\sharp) e_+~,
257:   \end{aligned}
258: \end{equation}
259: for all $v\in V$ and $\alpha \in V^*$.  All other brackets not
260: following from these are zero.  The Jacobi identity is satisfied by
261: virtue of $A$ being symmetric.  Notice that since its second derived
262: ideal is central, $\fg_A$ is (three-step) solvable.
263: 
264: Notice that $\fk_A=V^*$ is an abelian Lie subalgebra, and its
265: complementary subspace $\fp_A= V \oplus Z \oplus Z^*$ is acted on by
266: $\fk_A$.  Indeed, it follows easily from \eqref{eq:gA} that
267: \begin{equation*}
268:   [\fk_A,\fp_A] \subset \fp_A \qquad\text{and}\qquad [\fp_A,\fp_A]
269:   \subset \fk_A~,
270: \end{equation*}
271: whence $\fg_A = \fk_A \oplus \fp_A$ is a symmetric split.  Lastly, let
272: $B\in\left(S^2\fp_A^*\right)^{\fk_A}$ denote the invariant symmetric
273: bilinear form on $\fp_A$ defined by
274: \begin{equation*}
275:   B(v,w) = \left<v,w\right> \qquad\text{and}\qquad B(e_+,e_-)=1~,
276: \end{equation*}
277: for all $v,w\in V$.  This defines on $\fp_A$ a $\fk_A$-invariant
278: lorentzian scalar product of signature $(1,n-1)$.
279: 
280: We now have the required ingredients to construct a (lorentzian)
281: symmetric space.  Let $G_A$ denote the connected, simply-connected Lie
282: group with Lie algebra $\fg_A$ and let $K_A$ denote the Lie subgroup
283: corresponding to the subalgebra $\fk_A$.  The lorentzian scalar product
284: $B$ on $\fp_A$ induces a lorentzian metric $g$ on the space of cosets
285: \begin{equation*}
286:   M_A = G_A/K_A~,
287: \end{equation*}
288: turning it into a symmetric space.
289: 
290: Introducing coordinates $x^\pm, x^i$ naturally associated to
291: $e_\pm,e_i$, where $e_i$ is an orthonormal frame for $V$, we can write
292: the Cahen--Wallach metric explicitly as
293: \begin{equation}
294:   \label{eq:CWmetric}
295:   g = 2 dx^+ dx^- + \left(\sum_{i,j=1}^{n-2} A_{ij} x^i x^j\right)
296:   (dx^-)^2 + \sum_{i=1}^{n-2} \left(dx^i\right)^2~.
297: \end{equation}
298: 
299: \begin{proposition}[Cahen--Wallach \cite{CahenWallach}]
300:   The metric on $M_A$ defined above is indecomposable if and only if
301:   $A$ is nondegenerate.  Moreover, $M_A$ and $M_{A'}$ are isometric if
302:   and only if $A$ and $A'$ are related in the following way:
303:   \begin{equation*}
304:     A'(v,w) = c A(Ov, Ow) \qquad\text{for all $v,w\in V$,}
305:   \end{equation*}
306:   for some orthogonal transformation $O:V\to V$ and a positive scale
307:   $c>0$.
308: \end{proposition}
309: 
310: From this result one sees that the moduli space $\eM_n$ of
311: indecomposable such metrics in $n$ dimensions is given by
312: \begin{equation*}
313:   \eM_n = \left( S^{n-3} - \Delta \right) / \fS_{n-2}~,
314: \end{equation*}
315: where
316: \begin{equation*}
317:   \Delta = \left\{ (\lambda_1,\dots,\lambda_{n-2}) \in S^{n-3} \subset
318:     \RR^{n-2} \mid \lambda_1 \cdots \lambda_{n-2} = 0\right\}
319: \end{equation*}
320: is the singular locus consisting of eigenvalues of degenerate
321: $A$'s, and $\fS_{n-2}$ is the symmetric group in $n-2$ symbols, acting
322: by permutations on $S^{n-3} \subset \RR^{n-2}$.
323: 
324: \paragraph{Local isometric embeddings.}
325: \label{sec:embeddings}
326: \index{lorentzian symmetric space!isometric embedding}
327: 
328: Indecomposable lorentzian symmetric spaces in $d\geq 2$ are locally
329: isometric to algebraic varieties in pseudo-euclidean spaces.  This is
330: well-known for both de~Sitter and anti~de~Sitter spaces.  Indeed, let
331: $\kappa>0$.  Then the quadric in $\EE^{n,1}$ consisting of points
332: $(x_0,x_1,\dots,x_n)\in \RR^{n+1}$ such that
333: \begin{equation*}
334:   - x_0^2 + x_1^2 + x_2^2 + \cdots + x_n^2 = 1/\kappa^2
335: \end{equation*}
336: has constant sectional curvature $\kappa$ and hence is locally
337: isometric to a de~Sitter space, whereas the quadric in $\EE^{n-1,2}$
338: consisting of points $(x_0,x_1,\dots,x_n) \in \RR^{n+1}$ such that
339: \begin{equation*}
340:   -x_0^2 + x_1^2 + x_2^2 + \cdots + x_{n-1}^2 - x_n^2 =
341:   -1/\kappa^2
342: \end{equation*}
343: has constant section curvature $-\kappa$ and hence is locally
344: isometric to an anti~de~Sitter space.
345: 
346: Similarly, the $n$-dimensional Cahen--Wallach spaces are locally
347: isometric to the intersection of two quadrics in $\EE^{n,2}$.  Indeed,
348: let $(u^1,v^1,u^2,v^2,x^i)$, for $i=1,\dots,n-2$,
349: be flat coordinates in $\EE^{n,2}$ relative to which the metric takes
350: the form
351: \begin{equation}
352:   2 du^1 dv^1 + 2 du^2 dv^2 + \sum_{i=1}^9 dx^i dx^i~.
353: \end{equation}
354: Then the Cahen--Wallach space with matrix $A$ is locally isometric to
355: the induced metric on the intersection of the two quadrics
356: \begin{equation}
357:   (u^1)^2 + (u^2)^2 = 1 \qquad\text{and}\qquad 2 u^1 v^1 + 2 u^2 v^2 = 
358:   \sum_{i,j=1}^9 A_{ij} x^i x^j~.
359: \end{equation}
360: This was proven in \cite{Limits}.
361: 
362: \subsection{Lorentzian parallelisable manifolds}
363: \label{sec:parallelisable}
364: 
365: A subclass of the lorentzian symmetric spaces are the parallelisable
366: manifolds.
367: 
368: Recall that a differentiable manifold $M$ is said to admit an
369: \textbf{absolute parallelism} if it admits a smooth
370: trivialisation of the frame bundle.  Such a trivialisation consists of
371: a smooth global frame and hence also trivialises the tangent bundle;
372: whence manifolds admitting absolute parallelisms are parallelisable in
373: the topological sense.  The reduction theorem for connections on
374: principal bundles (see, for example,
375: \cite[Section~II.7]{KobayashiNomizu}) allows us to think of absolute
376: parallelisms in terms of holonomy groups of connections.  Indeed, an
377: absolute parallelism is equivalent to a smooth connection on the frame
378: bundle with trivial holonomy.  This implies, in particular, that the
379: connection is flat and if the manifold is simply-connected then
380: flatness is also sufficient.
381: 
382: So far these notions are purely (differential) topological and make no
383: mention of metrics or any other structure on the manifold.  The
384: question arises whether there is a metric on $M$ which is consistent
385: with a given absolute parallelism, so that parallel transport is an
386: isometry; or turning the question around, whether a given
387: pseudo-riemannian manifold $(M,g)$ admits an absolute parallelism
388: consistent with it.  In terms of connections, a consistent absolute
389: parallelism is equivalent to a metric connection with torsion with
390: trivial holonomy; or, locally, to a flat metric connection with
391: torsion.
392: 
393: Cartan and Schouten \cite{CartanSchouten1,CartanSchouten2} essentially
394: solved the riemannian case by generalising Clifford's parallelism on
395: the 3-sphere in two different ways.  The three-sphere can be
396: understood both as the unit-norm quaternions and also as the Lie group
397: $\SU(2)=\Sp(1)$.  The latter characterisation generalises to other
398: (semi)simple Lie groups, whereas the former gives rise to the
399: parallelism of the 7-sphere thought of as the unit-norm octonions.  It
400: follows from the results of Cartan and Schouten that a
401: simply-connected irreducible riemannian manifold admitting a
402: consistent absolute parallelism (equivalently a flat metric
403: connection) is isometric to one of the following: the real line, a
404: simple Lie group with the bi-invariant metric induced from a multiple
405: of the Killing form, or the round 7-sphere.
406: 
407: Their proofs might have had gaps which were addressed by Wolf
408: \cite{Wolf1,Wolf2}, who also generalised these results to arbitrary
409: signature, subject to an algebraic curvature condition saying that the
410: pseudo-riemannian manifold $(M,g)$ is of ``reductive type,'' a
411: condition which is automatically satisfied in the riemannian case.
412: (See Wolf's paper for the precise condition.)  In the case of
413: lorentzian signature, Cahen and Parker \cite{CahenParker} showed that
414: one can relax the ``reductive type'' condition; completing the
415: classification of absolute parallelisms consistent with a lorentzian
416: metric.
417: 
418: Wolf also showed that if one also assumes that the torsion is
419: parallel, then, in any signature, $(M,g)$ is locally isometric to a
420: Lie group with a bi-invariant metric.  In fact, as we will show below
421: in Section \ref{sec:flat}, one obtains the same result starting with
422: the weaker hypothesis that the torsion three-form is closed, which
423: will be the case needed in supergravity.
424: 
425: The results of Cahen and Parker \cite{CahenParker} actually show that
426: in lorentzian signature one gets for free that the torsion is
427: parallel.  Therefore it follows that an indecomposable lorentzian
428: manifold $(M,g)$ admits a consistent absolute parallelism if and only
429: if it is locally isometric to a lorentzian Lie group with bi-invariant
430: metric.\index{lorentzian symmetric space!parallelisable}
431: 
432: \subsection{Flat metric connections with closed torsion}
433: \label{sec:flat}
434: 
435: We will now show that a pseudo-riemannian manifold $(M,g)$ with a
436: flat metric connection with closed torsion three-form is locally
437: isometric to a Lie group admitting a bi-invariant metric.
438: 
439: Let $(M,g)$ be a pseudo-riemannian manifold and let $D$ be a metric
440: connection with torsion $T$. In other words, $D g = 0$ and for all
441: vector fields $X,Y$ on $M$, $T:\Lambda^2TM \to TM$ is defined by
442: \begin{equation*}
443:  T(X,Y) =  D_X Y - D_Y X - [X,Y]~.
444: \end{equation*}
445: In terms of the torsion-free Levi-Cività connection $\nabla$, we have
446: \begin{equation*}
447:   D_X Y = \nabla_X Y + \half T(X,Y)~.
448: \end{equation*}
449: Since both $Dg=0$ and $\nabla g = 0$, $T$ is skew-symmetric:
450: \begin{equation}
451:   \label{eq:skew}
452:   g(T(X,Y),Z) = - g(T(X,Z),Y)~,
453: \end{equation}
454: for all vector fields $X,Y,Z$ and gives rise to a \textbf{torsion
455:   three-form} $H\in\Omega^3(M)$, defined by
456: \begin{equation*}
457:   H(X,Y,Z) = g(T(X,Y),Z)~.
458: \end{equation*}
459: We will assume that $H$ is closed and in this section we will
460: characterise those manifolds for which $D$ is flat.
461: 
462: Let $R^D$ denote the curvature tensor of $D$, defined by
463: \begin{equation*}
464:   R^D(X,Y)Z = D_{[X,Y]}Z - D_X D_Y Z + D_Y
465:   D_X Z~.
466: \end{equation*}
467: Our strategy will be to consider the equation $R^D = 0$,
468: decompose it into types and solve the corresponding equations.  We
469: will find that $T$ is parallel with respect to both $\nabla$ and $D$,
470: and this will imply that $(M,g)$ is locally a Lie group with a
471: bi-invariant metric and $D$ the parallelising connection of Cartan and
472: Schouten \cite{CartanSchouten1}.
473: 
474: The curvature $R^D$ is given by
475: \begin{multline*}
476:   R^D(X,Y)Z = R(X,Y)Z - \half (\nabla_X T)(Y,Z) + \half (\nabla_Y
477:   T)(X,Z)\\
478:   - \tfrac14 T(X,T(Y,Z)) + \tfrac14 T(Y,T(X,Z))~,
479: \end{multline*}
480: where $R=R^\nabla$ is the curvature of the Levi-Cività connection.  The
481: tensor
482: \begin{equation*}
483:   R^D(X,Y,Z,W) := g(R^D(X,Y)Z,W)
484: \end{equation*}
485: takes the following form
486: \begin{multline*}
487:   R^D(X,Y,Z,W) = R(X,Y,Z,W)\\
488:   - \half g((\nabla_X T)(Y,Z),W) + \half g((\nabla_Y T)(X,Z),W)\\
489:   - \tfrac14 g(T(X,T(Y,Z)),W) + \tfrac14 g(T(Y,T(X,Z)),W)~,
490: \end{multline*}
491: where we have defined the Riemann tensor as usual:
492: \begin{equation*}
493:   R(X,Y,Z,W) := g(R(X,Y)Z,W)~.
494: \end{equation*}
495: Using equation \eqref{eq:skew} we can rewrite $R^D$ as
496: \begin{multline*}
497:   R^D(X,Y,Z,W) = R(X,Y,Z,W)\\
498:   - \half g((\nabla_X T)(Y,Z),W) + \half g((\nabla_Y T)(X,Z),W)\\
499:   + \tfrac14 g(T(X,W),T(Y,Z)) - \tfrac14 g(T(Y,W),T(X,Z))~,
500: \end{multline*}
501: which is manifestly skew-symmetric in $X,Y$ and in $Z,W$.  Observe
502: that unlike $R$, the torsion terms in $R^D$ do \emph{not} satisfy the
503: first Bianchi identity.  Therefore breaking $R^D$ into algebraic types
504: will give rise to more equations and will eventually allow us to
505: characterise the data $(M,g,T)$ for which $R^D = 0$.
506: 
507: Indeed, let $R^D = 0$ and consider the identity
508: \begin{equation*}
509:   \cyclicsum_{XYZ} R^D(X,Y,Z,W) = 0~,
510: \end{equation*}
511: where $\cyclicsum$ denotes signed permutations.  Since $R$ does obey the
512: Bianchi identity
513: \begin{equation*}
514:   \cyclicsum_{XYZ} R(X,Y,Z,W) = 0~,
515: \end{equation*}
516: we obtain the following identity
517: \begin{equation}
518:   \label{eq:bianchi}
519:   \cyclicsum_{XYZ} g((\nabla_X T)(Y,Z),W) =  - \tfrac12 \cyclicsum_{XYZ}
520:   g(T(W,X),T(Y,Z))~.
521: \end{equation}
522: Now we use the fact that the torsion three-form $H$ is closed, which
523: can be written as
524: \begin{multline*}
525:   g((\nabla_X T)(Y,Z),W) - g((\nabla_Y T)(X,Z),W)\\
526:   + g((\nabla_Z T)(X,Y),W) - g((\nabla_W T)(X,Y),Z)=0~,
527: \end{multline*}
528: or equivalently,
529: \begin{equation*}
530:   g((\nabla_W T)(X,Y),Z) = \half \cyclicsum_{XYZ} g((\nabla_X T)(Y,Z),W)~.
531: \end{equation*}
532: This turns equation \eqref{eq:bianchi} into
533: \begin{equation}
534:   \label{eq:bianchitoo}
535:   g((\nabla_W T)(X,Y),Z) = - \tfrac14 \cyclicsum_{XYZ} g(T(W,X),T(Y,Z))~.
536: \end{equation}
537: From this equation it follows that
538: \begin{equation*}
539:   g((\nabla_W T)(X,Y),Z) = - g((\nabla_X T)(W,Y),Z)~,
540: \end{equation*}
541: so that $g((\nabla_W T)(X,Y),Z)$ is totally skew-symmetric.  This
542: means that $\nabla H = dH = 0$, whence $H$ and hence $T$ are parallel.
543: Therefore equation \eqref{eq:bianchi} simplifies to
544: \begin{equation}
545:   \label{eq:Jacobi}
546:   \cyclicsum_{XYZ} g(T(W,X),T(Y,Z)) = 0~.
547: \end{equation}
548: 
549: Equation \eqref{eq:Jacobi} is the Jacobi identity for $T$.  Indeed,
550: notice that
551: \begin{multline*}
552:   g(T(W,X),T(Y,Z)) = H(W,X,T(Y,Z))\\
553:   = H(X,T(Y,Z),W) = g(T(X,T(Y,Z)),W)~,
554: \end{multline*}
555: whence equation \eqref{eq:Jacobi} is satisfied if and only if
556: \begin{equation}
557:   \label{eq:Jacobi}
558:     \cyclicsum_{XYZ} T(X,T(Y,Z)) = 0~.
559: \end{equation}
560: This means that the tangent space $T_pM$ of $M$ at every point $p$
561: becomes a Lie algebra where the Lie bracket is given by the
562: restriction of $T$ to $T_pM$.  More is true and the restriction to
563: $T_pM$ of the metric $g$ gives rise to an (ad-)invariant scalar
564: product:
565: \begin{equation*}
566:   g(T(X,Y),Z) = g(X,T(Y,Z))~.
567: \end{equation*}
568: 
569: By a theorem of Wolf \cite{Wolf1,Wolf2} (based on the earlier work of
570: Cartan and Schouten \cite{CartanSchouten1,CartanSchouten2}) if $(M,g)$
571: is complete then it is a discrete quotient of a Lie group with a
572: bi-invariant metric.  In general, we can say that $(M,g)$ is locally
573: isometric to a Lie group with a bi-invariant metric.
574: 
575: Indeed, since $D$ is flat, there exists locally a parallel frame
576: $\{\xi_i\}$ for $TM$.  Since $\xi_i$ is parallel, from the definition
577: of the torsion,
578: \begin{equation*}
579:   T(\xi_i,\xi_j) = - [\xi_i,\xi_j]~.
580: \end{equation*}
581: Moreover, since $T$ is parallel relative to $D$, we see that
582: $[\xi_i,\xi_j]$ is also parallel with respect to $D$, whence it can be
583: written as a linear combination of the $\xi_i$ with constant
584: coefficients.  In other words, they span a real Lie algebra $\fg$.
585: The homomorphism $\fg \to C^\infty(M,TM)$ whose image is the
586: subalgebra spanned by the $\{\xi_i\}$ integrates, once we choose a
587: point in $M$, to a local diffeomorphism $G\to M$.  This is also an
588: isometry if we use on $G$ the metric induced from the one on the Lie
589: algebra, whence we conclude that $(M,g)$ is locally isometric to a Lie
590: group with a bi-invariant metric.
591: 
592: \subsection{Bi-invariant lorentzian metrics on Lie groups}
593: \label{sec:lie}
594: 
595: In this section we will briefly review the structure of Lie groups
596: admitting a bi-invariant lorentzian metric.  It is well-known that
597: bi-invariant metrics on a Lie group are in bijective correspondence
598: with (ad-)invariant scalar products on the Lie algebra.  Therefore it
599: is enough to study those Lie algebras possessing an invariant
600: lorentzian scalar product.  We shall call them \idx{lorentzian Lie
601:   algebras}{Lie algebra!lorentzian} in this review.
602: 
603: It is well-known that reductive Lie algebras admit invariant scalar
604: products: Cartan's criterion allows us to use the Killing forms on the
605: simple factors and any scalar product on an abelian Lie algebra is
606: trivially invariant.  Another well-known example of Lie algebras
607: admitting an invariant scalar product are the classical doubles.  Let
608: $\fh$ be \emph{any} Lie algebra and let $\fh^*$ denote the dual space
609: on which $\fh$ acts via the coadjoint representation.  The definition
610: of the coadjoint representation is such that the dual pairing $\fh
611: \otimes \fh^* \to \RR$ is an invariant scalar product on the
612: semidirect product $\fh \ltimes \fh^*$ with $\fh^*$ an abelian ideal.
613: The Lie algebra $\fh \ltimes \fh^*$ is called the classical double of
614: $\fh$ and the invariant metric has split signature $(r,r)$ where $\dim
615: \fh = r$.
616: 
617: It turns out that all Lie algebras admitting an invariant scalar
618: product can be obtained by a mixture of these constructions.  Let
619: $\fg$ be a Lie algebra with an invariant scalar product
620: $\left<-,-\right>_\fg$.  Now let $\fh$ act on $\fg$ as skew-symmetric
621: derivations; that is, preserving both the Lie bracket and the scalar
622: product.  First of all, since $\fh$ acts on $\fg$ preserving
623: the  scalar product, we have a linear map
624: \begin{equation*}
625:   \fh \to \fso(\fg) \cong \Lambda^2 \fg~,
626: \end{equation*}
627: with dual map
628: \begin{equation*}
629:   c: \Lambda^2 \fg \to \fh^*~,
630: \end{equation*}
631: where we have used the invariant scalar product to identity $\fg$ and
632: $\fg^*$ equivariantly.  Since $\fh$ preserves the Lie bracket in
633: $\fg$, this map is a cocycle, whence it defines a class $[c]\in
634: H^2(\fg;\fh^*)$ in the second Lie algebra cohomology of $\fg$ with
635: coefficients in the trivial module $\fh^*$.  Let $\fg \times_c \fh^*$
636: denote the corresponding central extension.  The Lie bracket of $\fg
637: \times_c \fh^*$ is such that $\fh^*$ is central and if $X,Y\in\fg$,
638: then
639: \begin{equation*}
640:   [X,Y] = [X,Y]_\fg + c(X,Y)~,
641: \end{equation*}
642: where $[-,-]_\fg$ is the original Lie bracket on $\fg$.  Now $\fh$
643: acts naturally on $\fg \times_c \fh^*$ preserving the Lie bracket; the
644: action on $\fh^*$ being given by the coadjoint representation.  This
645: then allows us to define the \idx{double extension}{Lie algebra!double
646:   extension} of $\fg$ by $\fh$,
647: \begin{equation*}
648:   \fd(\fg,\fh) = \fh \ltimes (\fg \times_c \fh^*)
649: \end{equation*}
650: as a semidirect product.  Details of this construction can be found in
651: \cite{MedinaRevoy,FSsug}.  The remarkable fact is that $\fd(\fg,\fh)$
652: admits an invariant scalar product:
653: \begin{equation}
654:   \left<(X,h,\alpha), (Y,k,\beta)\right> = \left<X,Y\right>_\fg +
655:   \alpha(k) + \beta(h) + B(h,k)~,
656: \end{equation}
657: for all $X,Y\in\fg$, $h,k\in\fh$, $\alpha,\beta\in\fh^*$ and where $B$
658: is \emph{any} invariant symmetric bilinear form on $\fh$.
659: 
660: We say that a Lie algebra with an invariant scalar product is
661: \idx{indecomposable}{Lie algebra!indecomposable} if it cannot be
662: written as the direct product of two orthogonal ideals.  A theorem of
663: Medina and Revoy \cite{MedinaRevoy} (see also \cite{FSalgebra} for a
664: refinement) says that an indecomposable (finite-dimensional) Lie
665: algebra with an invariant scalar product is either one-dimensional,
666: simple, or a double extension $\fd(\fg,\fh)$ where $\fh$ is either
667: simple or one-dimensional and $\fg$ is a (possibly trivial) Lie
668: algebra with an invariant scalar product.  Any (finite-dimensional)
669: Lie algebra with an invariant scalar product is then a direct sum of
670: indecomposables.
671: 
672: If the scalar product on $\fg$ has signature $(s,t)$, then the scalar
673: product on the double extension $\fd(\fg,\fh)$ has signature
674: $(s+r,t+r)$, where $r = \dim\fh$.  This means that if we are
675: interested in lorentzian signature, we can double extend at most once
676: and by a one-dimensional $\fh$.
677: 
678: Therefore indecomposable lorentzian Lie algebras are either reductive
679: or double extensions $\fd(\fg,\fh)$ where $\fg$ has a
680: positive-definite invariant scalar product and $\fh$ is
681: one-dimensional.  In the reductive case, indecomposability means that
682: it has to be simple, whereas in the latter case, since the scalar
683: product on $\fg$ is positive-definite, $\fg$ must be reductive.  A
684: result of \cite{FSsug} (see also \cite{FSalgebra}) then says that any
685: semisimple factor in $\fg$ splits off resulting in a decomposable Lie
686: algebra.  Thus if the double extension is to be indecomposable, $\fg$
687: must be abelian.  In summary, an indecomposable lorentzian Lie algebra
688: is either simple or a double extension of an abelian Lie algebra by a
689: one-dimensional Lie algebra and hence solvable (see, e.g.,
690: \cite{MedinaRevoy}).
691: 
692: In summary, an indecomposable lorentzian Lie algebra is either
693: isomorphic to $\fso(1,2)$ with (a multiple of) the Killing form, or
694: else is solvable and can be described as a double extension
695: $\fd_{2n+2}:= \fd(\EE^{2n},\RR)$ of the abelian Lie algebra $\EE^{2n}$
696: with the (trivially invariant) euclidean ``dot'' product by a
697: one-dimensional Lie algebra acting on $\EE^{2n}$ via a non-degenerate
698: skew-symmetric linear map $J : \EE^{2n} \to \EE^{2n}$.  Let $\omega
699: \in \Lambda^2(\EE^{2n})^*$ denote the associated $2$-form:
700: $\omega(\bv,\bw) = \left<\bv, J\bw\right>$.
701: 
702: More concretely, the double extension $\fd_{2n+2}$ has underlying
703: vector space $V = \EE^{2(d-1)} \oplus \RR \oplus \RR$, and if
704: $(\bv,v^-,v^+),(\bw, w^-, w^+) \in V$, then their Lie bracket is
705: given by
706: \begin{equation*}
707:   [(\bv,v^-,v^+), (\bw, w^-, w^+)] = (v^- J(\bw) - w^- J(\bv), 0, \bv
708:   \cdot J(\bw))
709: \end{equation*}
710: and their inner product follows by polarisation from
711: \begin{equation*}
712:   \left|(\bv,v^-,v^+)\right|^2 = \bv \cdot \bv + 2 v^+ v^- + b
713:   (v^-)^2~,
714: \end{equation*}
715: where $b\in \RR$ is arbitrary.  One can however always set $b=0$ via a
716: Lie algebra automorphism and we will do so here; although there are
717: situations when one may wish to retain this freedom.
718: 
719: The unique simply-connected Lie group with Lie algebra
720: $\fd_{2n+2}$ is a solvable ($2n+2$)-dimensional Lie group admitting a
721: bi-invariant metric
722: \begin{equation}
723:   \label{eq:cwmetric}
724:   ds^2 = 2 dx^+ dx^- - \left<J\bx,J\bx\right> (dx^-)^2 +
725:   \left<d\bx,d\bx\right>~,
726: \end{equation}
727: relative to natural coordinates $(\bx, x^-, x^+)$.  The parallelising
728: torsion has $3$-form
729: \begin{equation}
730:   \label{eq:partor}
731:   H = dx^- \wedge \omega~.
732: \end{equation}
733: 
734: The non-degenerate skew-symmetric endomorphism $J$ can be brought to a
735: Jordan normal form consisting of nonzero $2\times 2$ blocks via an
736: orthogonal transformation.  The skew-eigenvalues
737: $\lambda_1,\dots,\lambda_n$, which are different from zero, can be
738: arranged so that they obey: $0 < \lambda_1 \leq \lambda_2 \leq \cdots
739: \leq \lambda_n$.  Finally a positive rescaling of $J$ can be absorbed
740: into reciprocal rescalings of $x^\pm$, so that we can set $\lambda_n$,
741: say, equal to $1$ without loss of generality.  Therefore we see that
742: the moduli space of metrics \eqref{eq:cwmetric} is given by an
743: ($n-1$)-tuple $\blambda = (\lambda_1,\ldots,\lambda_{n-1})$ where $0 <
744: \lambda_1 \leq \cdots \leq \lambda_{n-1} \leq 1$.  It is clear that
745: they are particular cases of the Cahen--Wallach spaces discussed in
746: Section~\ref{sec:symmetric}.
747: 
748: \section{Supergravity}
749: \label{sec:sugra}
750: 
751: Supergravity is one of the later jewels of 20th century theoretical
752: physics.  It started out as an attempt to `gauge' the supersymmetry of
753: certain quantum field theories, but it was quickly realised that it
754: provides a nontrivial extension of Einstein gravity.  Supergravity
755: theories are fairly rigid---their structure dictated largely by the
756: representation theory of the spin groups.  A good modern review of the
757: structure of supergravity theories is \cite{ToineReview}.  It is fair
758: to say, however, that supergravity theories are still somewhat
759: mysterious to most mathematicians and much remains to be done to make
760: this beautiful chapter of modern mathematical physics accessible to a
761: larger mathematical audience.  That, however, is a task for a
762: different occasion.  For our present purposes, each supergravity
763: theory will be a collection of geometric PDEs and our interest will be
764: in finding special types of solutions.
765: 
766: We shall be interested uniquely in lorentzian supergravity theories in
767: dimension $d\geq 4$.  There are supergravity theories in lower
768: dimensions and in other metric signatures, but we will not discuss
769: them here.  Neither will we discuss other types of supergravity
770: theories: heterotic, gauged, conformal, massive,...  The two-volume
771: set \cite{SalamSezgin} reprints many of the foundational supergravity
772: papers.
773: 
774: For reasons which are well-known, namely the otherwise non-existence
775: of nontrivial interacting theories, the dimension of the spacetime
776: will be bounded above by $11$.  Apart from the dimension of the
777: spacetime, the other important invariant is the ``number of
778: supercharges'', denoted $n$, which is an integer multiple of the
779: dimension of the smallest irreducible \emph{real} spinor
780: representation in that spacetime dimension.  For dimension $\geq 4$
781: the number of supercharges ranges from $4$ to $32$.
782: 
783: In Table \ref{tab:supergravities}, which is borrowed from
784: \cite{ToineReview}, we tabulate the different supergravity theories in
785: $d\geq 4$.  The seemingly baroque notation is not too important: M
786: refers to the unique eleven-dimensional supergravity theory
787: \cite{Nahm,CJS} which is a low-energy limit of M-theory (hence the
788: name), types I \cite{NilssonN=1,ChamseddineN=1}, IIA
789: \cite{CampbellWestIIA,HuqNamazieIIA,GianiPerniciIIA} and IIB
790: \cite{SchwarzIIB,SchwarzWestIIB,HoweWestIIB} supergravities are the
791: low-energy limits of the similarly named string theories, whereas the
792: notation $N=n$ or $(p,q)$ is historical and denotes the multiplicity
793: of the spinor (or half-spinor) representations in the corresponding
794: supersymmetry algebra.  The original supergravity theory
795: \cite{FvNF,DeserZumino} is the four-dimensional $N=1$ theory.  The top
796: entry in each column has been highlighted to indicate that upon
797: dimensional reduction it gives rise to all the theories below it in
798: the same column.  As we will explain below, this means that for many
799: purposes, especially the classification of solutions, it is generally
800: enough to understand the `top' theories and, indeed, we will
801: concentrate on those.
802: 
803: \begin{table}[h!]
804:   \centering
805:   \begin{tabular}{|*{8}{>{$}c<{$}|} }
806:     \hline
807:     d\downarrow~n\to & \multicolumn{4}{c|}{32} & \multicolumn{2}{c|}{24}  &
808:     20 \\
809:     \hline
810:     11  & \textbf{M} & \multicolumn{3}{c|}{} &\multicolumn{2}{c|}{} & \\
811:     10  & \text{IIA} & \textbf{IIB} & \multicolumn{2}{c|}{}&\multicolumn{2}{c|}{} &
812:     \\
813:     9  &  \multicolumn{2}{c|}{N=2} &\multicolumn{2}{c|}{ }&
814:     \multicolumn{2}{c|}{ }  & \\
815:     8  &  \multicolumn{2}{c|}{N=2}&\multicolumn{2}{c|}{ }&
816:     \multicolumn{2}{c|}{ }  & \\
817:     7  &  \multicolumn{2}{c|}{N=4} &\multicolumn{2}{c|}{
818:     }&\multicolumn{2}{c|}{ }  & \\
819:     6  &
820:     \multicolumn{2}{c|}{(2,2)}&\boldsymbol{(3,1)}&\boldsymbol{(4,0)}
821:     &\boldsymbol{(2,1)} & \boldsymbol{(3,0)}& \\
822:     5  &  \multicolumn{4}{c|}{N=8}  &\multicolumn{2}{c|}{N=6}  & \\
823:     4  &  \multicolumn{4}{c|}{N=8}  & \multicolumn{2}{c|}{N=6}&
824:     \boldsymbol{N=5} \\
825:     \hline
826:   \end{tabular}\\[20pt]
827:   \begin{tabular}{|*{6}{>{$}c<{$}|} }
828:     \hline
829:     d\downarrow~n\to & \multicolumn{2}{c|}{16}  & 12 & 8 & 4  \\
830:     \hline
831:     10  & \textbf{I} &  &  &  &  \\
832:     9  & N=1  &  &  &  &  \\
833:     8  & N=1 &  &  &  &  \\
834:     7  & N=2 & &  &  &  \\
835:     6  & (1,1) &  \boldsymbol{(2,0)} &  & \boldsymbol{(1,0)} &  \\
836:     5  & \multicolumn{2}{c|}{N=4}&  & N=2 &  \\
837:     4  & \multicolumn{2}{c|}{N=4}  & \boldsymbol{N=3} & N=2 & \boldsymbol{N=1}\\
838:     \hline
839:   \end{tabular}
840:   \label{tab:supergravities}
841:   \caption{Lorentzian supergravity theories in $d\geq 4$}
842: \end{table}
843: 
844: Indeed, supergravity theories in different dimensions may be related by a
845: procedure known as Ka{\l}u\.za--Klein reduction.  This can be read off
846: from Table \ref{tab:supergravities}: any supergravity theory in the
847: table can be obtained by Ka{\l}u\.za--Klein reduction from any theory
848: sitting above it in the same column.  In practice this means that a
849: solution to any of the supergravity theories in the table can be
850: lifted to a solution of any theory above it in the same column, should
851: there be any.  Conversely, any solution of a supergravity theory which
852: is invariant under a one-dimensional Lie group gives rise to a local
853: solution (and indeed global if the action is free and proper) of the
854: supergravity theory immediately below it in the same column.  We shall
855: be particularly interested in the reduction from $d=11$ supergravity
856: to $d=10$ type IIA supergravity, and in the reduction and subsequent
857: truncation from the $d=6$ $(1,0)$ supergravity to minimal $d=5$ $N=2$
858: supergravity.
859: 
860: We shall be interested in solutions of the field equations coming from
861: these supergravity theories.  Such a solution is described in
862: geometric terms by the following data:
863: \begin{itemize}
864: \item a $d$-dimensional lorentzian spin manifold $(M,g)$ with a
865:   (possibly twisted) real rank $n$ spinor bundle $\eS \to M$, and
866: \item certain additional geometric data, which will be different in
867:   each supergravity theory, consisting of differential forms or, more
868:   generally, sections of certain fibre bundles over $M$,
869: \end{itemize}
870: all subject to field equations which generalise the coupled
871: Einstein--Maxwell equations familiar from four-dimensional Physics.
872: 
873: The above geometric data defines a connection $D$ on the spinor bundle
874: $\eS$ as well as a (possibly empty) set of endomorphisms of $\eS$.
875: Together they define a class of sections of $\eS$, parallel with
876: respect to $D$ and in the kernel of the endomorphisms, which are
877: called Killing spinors.
878: 
879: We will be particularly interested in the cases where the connection
880: $D$ is flat, so that it admits the maximum number of parallel
881: sections.  In this case, the field equations are automatically
882: satisfied.  In general, the field equations are intimately related to
883: integrability conditions for the existence of parallel sections of
884: $D$.
885: 
886: 
887: \subsection{Eleven-dimensional supergravity}
888: \label{sec:d=11}
889: 
890: Eleven-dimensional supergravity was predicted by Nahm \cite{Nahm} and
891: constructed soon thereafter by Cremmer, Julia and Scherk \cite{CJS}.
892: We will only be concerned with the bosonic equations of motion.  The
893: geometrical data consists of $(M, g, F)$ where $(M,g)$ is an
894: eleven-dimensional lorentzian manifold with a spin structure and $F
895: \in \Omega^4(M)$ is a closed $4$-form.  The equations of motion
896: generalise the Einstein--Maxwell equations in four dimensions.  The
897: Einstein equation relates the Ricci curvature to the energy momentum
898: tensor of $F$.  More precisely, the equation is
899: \begin{equation}
900:   \label{eq:einstein}
901:     \Ric(g) = T(g,F)
902: \end{equation}
903: where the symmetric tensor
904: \begin{equation*}
905:   T(X,Y) = \half \left< \imath_X F, \imath_Y F\right> - \tfrac16
906:   g(X,Y) |F|^2~,
907: \end{equation*}
908: is related to the energy-momentum tensor of the (generalised) Maxwell
909: field $F$.  In the above formula, $\left<-,-\right>$ denotes the
910: scalar product on forms, which depends on $g$, and $|F|^2 =
911: \left<F,F\right>$ is the associated (indefinite) norm.  The
912: generalised Maxwell equations are now nonlinear:
913: \begin{equation}
914:   \label{eq:maxwell}
915:    d \star F = - \half F \wedge F~.
916: \end{equation}
917: 
918: \begin{definition}
919:   A triple $(M,g,F)$ satisfying the equations \eqref{eq:einstein} and
920:   \eqref{eq:maxwell} is called a (bosonic)
921:   \idx{background}{supergravity!$d{=}11$!background} of
922:   eleven-dimensional supergravity.
923: \end{definition}
924: 
925: Let $\eS$ denote the bundle of spinors on $M$.  It is a real vector
926: bundle of rank $32$ with a spin-invariant symplectic form
927: $\left(-,-\right)$.  A differential form on $M$ gives rise to an
928: endomorphism of the spinor bundle via the composition
929: \begin{equation*}
930:   c: \Lambda T^*M  \xrightarrow{\cong} \Cl(T^*M) \to \End\eS~,
931: \end{equation*}
932: where the first map is the bundle isomorphism induced by the vector
933: space isomorphism between the exterior and Clifford algebras, and the
934: second map is induced from the action of the Clifford algebra
935: $\Cl(1,10)$ on the spinor representation $S$ of $\Spin(1,10)$.  In
936: signature $(1,10)$ one has the algebra isomorphism
937: \begin{equation*}
938:   \Cl(1,10) \cong \Mat(32,\RR) \oplus \Mat(32,\RR)~,
939: \end{equation*}
940: hence the map $\Cl(1,10) \to \End S$ has kernel.  In other words, the
941: map $c$ defined above involves a choice.  This comes down to choosing
942: whether the (normalised) volume element in $\Cl(1,10)$ acts as $\pm$
943: the identity.  In our conventions, the volume element acts as minus
944: the identity.
945: 
946: \begin{definition}
947:   We say that a background $(M,g,F)$ is
948:   \idx{supersymmetric}{supergravity!$d{=}11$!supersymmetric
949:     background} if there exists a nonzero spinor $\varepsilon \in
950:   \Gamma(\eS)$ which is parallel with respect to the supersymmetric
951:   connection
952:   \begin{equation*}
953:     D : \Gamma(\eS) \to \Gamma( T^*M \otimes \eS)
954:   \end{equation*}
955:   defined, for all vector fields $X$, by
956:   \begin{equation}
957:     \label{eq:connd=11}
958:     D_X \varepsilon  = \nabla_X \varepsilon + \Omega_X(F)
959:     \varepsilon~,
960:   \end{equation}
961:   where $\nabla$ is the spin connection and $\Omega(F):TM \to \End\eS$
962:   is defined by
963:   \begin{equation*}
964:     \Omega_X(F) = \tfrac1{12} c(X^\flat \wedge F) - \tfrac16
965:     c(\imath_X F)~,
966:   \end{equation*}
967:   with $X^\flat$ the one-form dual to $X$.
968: \end{definition}
969: 
970: A nonzero spinor $\varepsilon$ which is parallel with respect to $D$
971: is called a \idx{Killing spinor}{supergravity!$d{=}11$!Killing
972:   spinor}.  This is a generalisation of the usual geometrical notion
973: of Killing spinor (see, for example, \cite{BFGK}).  The name is apt
974: because Killing spinors are ``square roots'' of Killing vectors.
975: Indeed, one has the following
976: 
977: \begin{proposition}
978:   Let $\varepsilon_i$, $i=1,2$ be Killing spinors: $D \varepsilon_i =
979:   0$.  Then the vector field $V$ defined, for all vector fields $X$,
980:   by
981:   \begin{equation*}
982:     g(V,X) = \left( \varepsilon_1, X \cdot \varepsilon_2\right)
983:   \end{equation*}
984:   is a Killing vector and moreover \cite{GauPak} preserves $F$.
985: \end{proposition}
986: 
987: The fundamental object in eleven-dimensional supergravity is the
988: connection $D$, whose curvature encodes the field equations.  Indeed,
989: the field equations are equivalent \cite{GauPak,Preons} to
990: \begin{equation*}
991:   e^i \cdot R^D_{X,e_i} = 0\qquad\text{for every vector field $X$,}
992: \end{equation*}
993: where $(e_i)$ is an orthonormal frame and $(e^i)$ is dual coframe and
994: $\cdot$ is Clifford multiplication.
995: 
996: Alas, $D$ is not induced from a connection on the tangent bundle and
997: in fact, it does not even preserve the symplectic structure.
998: Nevertheless one has the following
999: 
1000: \begin{proposition}\cite{HullHolonomy}
1001:   The holonomy of $D$ is contained in $\SL(32,\RR)$.
1002: \end{proposition}
1003: 
1004: An important open problem is to determine the possible holonomy groups
1005: of $D$ subject to the field equations.  In a way, the field equations
1006: play the rôle of the torsion-free condition in the holonomy problem
1007: for affine connections.  Except for the above result there are no
1008: other results of a general nature and although the infinitesimal
1009: holonomy of a number of solutions are known \cite{DuffLiuHol,BDLWHol},
1010: a general pattern has yet to emerge.
1011: 
1012: A coarser invariant than the holonomy of $D$ is the dimension of its
1013: kernel; that is, the dimension of the space of Killing spinors. It is
1014: customary to write this as a fraction
1015: \begin{equation*}
1016:   \nu = \frac{\dim \{\text{Killing spinors}\}}{\rank \eS}
1017: \end{equation*}
1018: which in this case is of the form $k/32$ for some integer
1019: $k=0,1,\dots,32$.
1020: 
1021: In Section~\ref{sec:maxd=11} we will review the classification of
1022: those bacgrounds with $\nu = 1$; that is, those backgrounds where $D$
1023: is flat.  We will see that they are all given by lorentzian symmetric
1024: spaces.  In fact, it was shown in \cite{FMPHom} that if $\nu > \tfrac34$,
1025: then $M$ is locally homogeneous and moreover it was conjectured that
1026: there exist backgrounds with $\nu = \tfrac34$ which are not locally
1027: homogeneous; although at present none have been constructed.  At the
1028: other end of the spectrum, the general form of $(g,F)$ which admit
1029: (at least) one Killing spinor is known \cite{GauPak,GauGutPak}.
1030: 
1031: \subsection{Ten-dimensional IIB supergravity}
1032: \label{sec:IIB}
1033: 
1034: Ten-dimensional IIB supergravity
1035: \cite{SchwarzIIB,SchwarzWestIIB,HoweWestIIB} is somewhat more
1036: complicated than eleven-dimensional supergravity due to the
1037: proliferation of dynamical fields and the fact that it cannot be
1038: obtained by Ka{\l}u\.za--Klein reduction from any higher-dimensional
1039: supergravity theory.
1040: 
1041: A type IIB supergravity background is described by the geometric data
1042: we describe presently.  First of all, we have a ten-dimensional
1043: lorentzian spin manifold $(M,g)$ together with a self-dual $5$-form
1044: $F$.  Now let $\eH$ be the complex upper half-plane, thought of as the
1045: riemannian symmetric space $\SU(1,1)/\U(1)$ and let $\tau: M\to \eH$
1046: be a smooth map.  We may think of $\SU(1,1)$ as the total space of a
1047: principle circle bundle over $\eH$ and we let $\eL\to\eH$ denote the
1048: associated complex line bundle.  Let $L_\tau = \tau^*\eL$ denote the
1049: pull-back bundle over $M$.  Choosing a section $\sigma : \eH \to
1050: \SU(1,1)$, we may pull back to $M$ the left-invariant Maurer--Cartan
1051: form on $\SU(1,1)$: its component along $\fu(1)$ defines a connection
1052: $A$ on $L_\tau$, whereas the component perpendicular to $\fu(1)$,
1053: relative to the invariant lorentzian scalar product on $\fsu(1,1)$,
1054: defines a one-form $B$ on $M$ with values in $L_\tau^2$.  Both $A$ and
1055: $B$ can be written explicitly in terms of $\tau$.  Indeed, if we let
1056: $z = (\tau -i)/(\tau + i)$ be the Cayley transform of $\tau$, so that
1057: $|z|<1$, then there is a choice of section $\sigma$ for which
1058: \begin{equation*}
1059:   A = \frac{\im(z d\overline z)}{1-|z|^2} \qquad\text{and}\qquad
1060:   B = \frac{dz}{1-|z|^2}~.
1061: \end{equation*}
1062: Finally, let $G$ be a $3$-form on $M$ with values in $L_\tau$.  On the
1063: bundles $\Lambda^p T^*M \otimes L_\tau^q$ we have connections
1064: $\nabla^{p,q}$ obtained from the Levi-Cività connection on $TM$ (and
1065: hence the tensor bundles) and the connection $A$ on $L_\tau$ (and
1066: hence its powers).  We will let
1067: \begin{equation*}
1068:   d^{\nabla^{p,q}} : \Omega^p(M;L_\tau^q) \to \Omega^{p+1}(M;L_\tau^q)
1069:   \quad\text{and}\quad
1070:   \delta^{\nabla^{p,q}} : \Omega^p(M;L_\tau^q) \to
1071:   \Omega^{p-1}(M;L_\tau^q)
1072: \end{equation*}
1073: denote the associated differential and co-differential on
1074: $L_\tau^q$-valued differential forms.  We will let $\left<-,-\right>$
1075: denote the natural pairing
1076: \begin{equation*}
1077:   \Omega^p(M;L_\tau^q) \otimes \Omega^p(M;L_\tau^{q'}) \to
1078:   \Omega^0(M;L_\tau^{q+q'})
1079: \end{equation*}
1080: induced from the metric $g$.  With these notational remarks behind us,
1081: we can finally define a IIB supergravity background.
1082: 
1083: \begin{definition}
1084:   The data $(M,g,\tau,F,G)$ described above defines a IIB supergravity
1085:   \idx{background}{supergravity!$d{=}10$ IIB!background} provided that
1086:   the following equations are satisfied:
1087:   \begin{align*}
1088:     \delta^{\nabla^{1,2}} B &= \tfrac14 |G|^2 \\
1089:     \delta^{\nabla^{3,1}} G(X,Y) &= \left<B,\imath_X\imath_Y \overline
1090:       G \right> - \tfrac{2i}3 \left<\imath_X\imath_YF,G\right> \\
1091:     d^{\nabla^{3,1}} G &= - B \wedge \overline G \\
1092:     d^{\nabla^{5,0}} F &= \tfrac{i}{8} G \wedge \overline G \\
1093:     \Ric(X,Y) &= B(X)\overline B(Y) + B(Y)\overline B(X) + 4
1094:     \left<\imath_XF,\imath_YF\right>\\
1095:     & \quad {} + \tfrac14 \left(\left<\imath_XG,\imath_Y\overline G\right> +
1096:       \left<\imath_YG,\imath_X\overline G\right>\right) -
1097:     \tfrac18 \left<G,\overline G\right> g(X,Y)~.
1098:   \end{align*}
1099: \end{definition}
1100: 
1101: Let $\eS_\pm$ denote the half-spinor bundles over $M$.  They are real,
1102: symplectic and have rank $16$.  Let $\eS := \eS_- \otimes
1103: L_\tau^{1/2}$, where $L_\tau^{1/2}$ is the square-root bundle of
1104: $L_\tau$.  Let $\overline\eS := \eS_- \otimes L_\tau^{-1/2}$.  Notice
1105: that if $\varepsilon \in \Omega^0(M;\eS)$, then $\overline\varepsilon
1106: \in \Omega^0(\overline\eS)$.  Furthermore the Clifford action of
1107: differential forms on spinors extends to an action
1108: \begin{equation*}
1109:   c : \Omega^p(M;L_\tau^q) \to \Hom\left(\Omega^0\left(M;\eS_\pm \otimes
1110:   L_\tau^r\right), \Omega^0\left(M;\eS_{(-1)^p} \otimes
1111:   L_\tau^{r+q}\right)\right)~.
1112: \end{equation*}
1113: Similarly we have a connection $\nabla^s$ acting on
1114: $\Omega^0(M;\eS_\pm \otimes L_\tau^s)$ which is defined using the spin
1115: connection and the connection $A$ on $L_\tau$.  We are now in a
1116: position to define a type IIB supergravity Killing spinor.
1117: 
1118: \begin{definition}
1119:   A IIB supergravity \idx{Killing spinor}{supergravity!$d{=}10$ IIB!Killing
1120:     spinor} is a nonzero section $\varepsilon$ of $\eS$ satisfying the
1121:   following two conditions
1122:   \begin{align}
1123:     c(B)\overline \varepsilon &= \tfrac14 c(G)\varepsilon \label{eq:dilatinoIIB}\\
1124:     \nabla^{1/2}_X \varepsilon &= - \tfrac{i}4 c(F) c(X^\flat)
1125:     \varepsilon - \tfrac1{32} \left( c(\imath_X G) - 2 c(X^\flat
1126:       \wedge G)\right) \overline \varepsilon~.\label{eq:gravitinoIIB}
1127:   \end{align}
1128: \end{definition}
1129: 
1130: Just like in eleven-dimensional supergravity, IIB Killing spinors are
1131: square roots of Killing vectors.  Indeed, the image of the natural map
1132: \begin{equation*}
1133:   \Omega^0(M;\eS) \otimes \Omega^0(M;\overline\eS) \to \text{vector
1134:     fields}
1135: \end{equation*}
1136: consists of Killing vectors which in addition preserve the geometric
1137: data of a background.  Again it is possible to show that if the space
1138: of Killing spinors of a supersymmetric background of type IIB
1139: supergravity has (real) dimension $>24$, then the background is
1140: locally homogeneous \cite{EHJGMHom}.
1141: 
1142: Type IIB supergravity backgrounds are acted upon by $\SU(1,1)$, which
1143: is the \idx{duality group}{supergravity!$d{=}10$ IIB!duality group} of type IIB
1144: supergravity.  The metric $g$ and the five-form $F$ are
1145: $\SU(1,1)$-invariant, whereas $\SU(1,1)$ acts on $z$ (hence on $\tau$)
1146: via fractional linear transformations:
1147: \begin{equation*}
1148:   \begin{pmatrix}
1149:     a & b \\ \overline b & \overline a
1150:   \end{pmatrix} \cdot z = \frac{a z + b}{\overline b z + \overline
1151:     a}~.
1152: \end{equation*}
1153: Moreover, the bundle $\eL \to \eH$ is a homogeneous bundle of
1154: $\SU(1,1)$ hence there is an action of $\SU(1,1)$ on sections of $\eL$
1155: and its powers.  Putting these two actions together we see that
1156: $\gamma\in\SU(1,1)$ sends sections of $L_\tau^p$ (and also differential
1157: forms and spinors with values in such a bundle) to sections of
1158: $L_{\gamma\tau}^p$.
1159: 
1160: The classification of the maximally supersymmetric background will be
1161: presented in Section~\ref{sec:maxIIB}, which is based on the papers
1162: \cite{FOPMax} and \cite{FOPPluecker}.  In the opposite extreme, there
1163: has been steady progress recently on the determination of the general
1164: form of the backgrounds admitting some supersymmetry
1165: \cite{GGP1,GGP2,GGPR}.
1166: 
1167: 
1168: \subsection{Six-dimensional $(2,0)$ and $(1,0)$ supergravities}
1169: \label{sec:sixdim}
1170: 
1171: We start by describing the field content and Killing spinor equations
1172: of $(1,0)$ \cite{NishinoSezgin10} and $(2,0)$
1173: \cite{Townsend20,Tanii20} chiral supergravities in six dimensions.  We
1174: start as usual by describing the relevant spinorial representations.
1175: The spin group $\Spin(1,5) \cong \SL(2,\HH)$, whence the irreducible
1176: spinorial representations are quaternionic of complex dimension $4$.
1177: There are two inequivalent representations $S_\pm$ which are
1178: distinguished by their chirality. Let $S_1$ denote the fundamental
1179: representation of $\Sp(1)$: it is a quaternionic representation of
1180: complex dimension $2$, and similarly let $S_2$ denote the fundamental
1181: representation of $\Sp(2)$, which is a quaternionic representation of
1182: complex dimension $4$.  The tensor products $S_+ \otimes S_1$ and $S_+
1183: \otimes S_2$ are complex representations of $\Spin(1,5) \times \Sp(1)$
1184: and $\Spin(1,5) \times \Sp(2)$, respectively, with a real structure.
1185: We will let
1186: \begin{equation*}
1187:   S = [S_+ \otimes S_1] \qquad\text{and}\qquad \bS = [S_+ \otimes
1188:   S_2]
1189: \end{equation*}
1190: denote the underlying real representations.  Clearly $S$ is a real
1191: representation of dimension $8$ and $\bS$ is a real representation of
1192: dimension $16$.  If $(M,g)$ is a six-dimensional lorentzian spin
1193: manifold, then we will let $\eS$ and $\ebS$ denote the bundles of
1194: spinors associated with the representations $S$ and $\bS$,
1195: respectively.  The groups $\Sp(1)$ and $\Sp(2)$ are the
1196: \idx{R-symmetry}{supergravity!$d{=}6$!R-symmetry groups} groups of
1197: these supergravity theories.
1198: 
1199: \begin{definition}
1200:   A $(1,0)$ supergravity \idx{background}{supergravity!$d{=}6$!$(1,0)$
1201:     background} consist of a six-dimensional lorentzian spin manifold
1202:   $(M,g)$ together with a closed antiself-dual $3$-form $H$ subject to
1203:   the Einstein equation
1204:   \begin{equation*}
1205:     \Ric(X,Y) = -\tfrac14 \left<\imath_XH,\imath_YH\right>~.
1206:   \end{equation*}
1207:   Such a background is said to be
1208:   \idx{supersymmetric}{supergravity!$d{=}6$!$(1,0)$ supersymmetric
1209:     background} if there are nonzero sections $\varepsilon$ of $\eS$
1210:   obeying
1211:   \begin{equation}
1212:     \label{eq:conn10}
1213:     D_X \varepsilon := \nabla_X \varepsilon - \half c(\imath_X H)
1214:     \varepsilon = 0~,
1215:   \end{equation}
1216:   for all vector fields $X$, where $c: \Omega(M) \to \Cl(T^*M) \to
1217:   \End(S)$ is the action of forms on sections of $S$, and $\nabla$ is
1218:   induced from the Levi-Cività connection.
1219: \end{definition}
1220: 
1221: We remark that the connection $D$ in equation \eqref{eq:conn10} is
1222: induced from a spin connection with torsion three-form $H$.
1223: 
1224: Similarly for $(2,0)$ supergravity, we have the following
1225: 
1226: \begin{definition}
1227:   A (2,0) supergravity \idx{background}{supergravity!$d{=}6$!$(2,0)$
1228:     background} consists of a
1229:   six-dimensional lorentzian spin manifold $(M,g)$, a $V$-valued
1230:   closed antiself-dual $3$-form $\bH$, where $V$ is the
1231:   five-dimensional real representation of the R-symmetry group
1232:   $\Sp(2)\cong \Spin(5)$ together with a $\Sp(2)$-invariant scalar
1233:   product, subject to the Einstein equation
1234:   \begin{equation*}
1235:     \Ric(X,Y) = -\tfrac14 \left<\imath_X\bH,\imath_Y\bH\right>~,
1236:   \end{equation*}
1237:   where $\left<-,-\right>$ now also includes the $\Sp(2)$-invariant
1238:   inner product on $V$.  Such a background is said to be
1239:   \idx{supersymmetric}{supergravity!$d{=}6$!$(2,0)$ supersymmetric
1240:     background} if there are nonzero sections $\varepsilon$ of $\ebS$
1241:   obeying
1242:   \begin{equation}
1243:     \label{eq:conn20}
1244:     \bD_X \varepsilon = \nabla_X \varepsilon - \half c(\imath_X \bH)
1245:     \varepsilon = 0~,
1246:   \end{equation}
1247:   for all vector fields $X$ and where $c: \Omega(M;V) \to \Cl(T^*M)
1248:   \otimes \Cl(V) \to \End(\ebS)$ is the action of $V$-valued forms on
1249:   sections of $\ebS$.
1250: \end{definition}
1251: 
1252: Notice that in $(2,0)$ supergravity, the anti-selfduality of $\bH$
1253: imply that $\bH \wedge \bH = 0$ in $\Omega^6(M; \Lambda^2 V)$.
1254: 
1255: Maximal supersymmetry implies that the connections $D$ acting on $\eS$
1256: and $\bD$ on $\ebS$ are flat.  In the case of $(1,0)$ supergravity,
1257: $D$ is a spin connection with torsion and maximally supersymmetric
1258: solutions correspond to six-dimensional lorentzian manifolds admitting
1259: a flat metric connection with anti-selfdual closed torsion three-form.
1260: We saw in Section \ref{sec:flat} that $(M,g)$ is locally isometric to
1261: a Lie group with a bi-invariant lorentzian metric.  In the case of
1262: $(2,0)$ supergravity, $\bD$ does not have such an obvious geometrical
1263: interpretation, but it is proven in \cite{CFOSchiral} that, up to the
1264: natural action of the R-symmetry group, the $(2,0)$ maximally
1265: supersymmetric backgrounds are in one-to-one correspondence with those
1266: of $(1,0)$ supergravity.
1267: 
1268: The general form of a supersymmetric background in $(1,0)$
1269: supergravity has been obtained in \cite{GMR}, who in particular also
1270: determine the maximally supersymmetric backgrounds by a different
1271: method, closely related to the one in \cite{FOPMax,FOPPluecker}.
1272: 
1273: \section{Maximally supersymmetric backgrounds}
1274: \label{sec:maximal}
1275: 
1276: In this section we review the known results about maximally
1277: supersymmetric backgrounds in a number of the more interesting
1278: supergravity theories.  Several of the theories under the
1279: consideration will be tackled directly: $d=11$ supergravity,
1280: $d=10$ IIB supergravity and the $d=6$ supergravities, whereas the
1281: maximally supersymmetric backgrounds of $d=10$ IIA and $d=5$ $N=2$
1282: supergravities will be obtained from those $d=11$ and $d=6$
1283: supergravities by the technique of Ka{\l}u\.za--Klein reduction.  As
1284: this technique is very useful, we will review it briefly now.
1285: 
1286: \subsection{Ka{\l}u\.za--Klein reduction}
1287: \label{sec:kk}
1288: 
1289: In this section we will briefly review the geometric underpinning of
1290: Ka{\l}u\.za--Klein reduction.  We start with a supergravity background
1291: $(M,g,F,\dots)$ which is invariant under a one-dimensional Lie group
1292: $\Gamma$, acting freely and properly on $M$ by isometries which in
1293: addition preserve any other geometric data $F,\dots$.  We shall let
1294: $\xi$ denote a Killing vector field for the $\Gamma$-action.  Since
1295: the action is free, $\xi$ is nowhere vanishing.  We will also assume
1296: that $\xi$ is spacelike; although this is not strictly necessary and
1297: indeed time-like reductions can be quite useful, especially in the
1298: context of topological field theories.
1299: 
1300: The original spacetime $M$ is to be thought of as the total space of a
1301: principal $\Gamma$-bundle $\pi:M \to N = M/\Gamma$, where $\pi$ the
1302: map taking a point in $M$ to the $\Gamma$-orbit on which it lies.  At
1303: every point $p$ in $M$, the tangent space $T_pM$ of $M$ at $p$
1304: decomposes into two orthogonal subspaces: $T_p M = \eV_p \oplus
1305: \eH_p$, where the \textbf{vertical subspace} $\eV_p = \ker \pi_*$
1306: consists of those vectors tangent to the $\Gamma$-orbit through $p$,
1307: and the \textbf{horizontal subspace} $\eH_p=\eV_p^\perp$ is its
1308: orthogonal complement relative to the metric $g$.  The resulting
1309: decomposition is indeed a direct sum by virtue of the
1310: nowhere-vanishing of the norm of $\xi$, whose value at $p$ spans
1311: $\eV_p$ for all $p$.  The derivative map $\pi_*$ sets up an
1312: isomorphism between $T_p M$ and $T_q N$, where $\pi(p) = q$.  As is
1313: well-known, there is a unique metric on $N$ for which this isomorphism
1314: is also an isometry and for which the map $\pi$ is a riemannian
1315: submersion.  We will call this metric $h$.
1316: 
1317: The horizontal sub-bundle $\eH$ gives rise to a connection one-form
1318: $\alpha$ on $M$ such that $\eH = \ker \alpha$ and such that
1319: $\alpha(\xi) = 1$.  We remark that $\alpha$ is invariant, so that
1320: $\eL_\xi\alpha = 0$.  This means that the curvature $2$-form $d\alpha$
1321: is both invariant and \textbf{horizontal}---that is, $\imath_\xi d\alpha
1322: = 0$.  Such forms are called \textbf{basic} and it is a basic fact that
1323: they define forms on $N$.  Hence $d\alpha$ defines a $2$-form on $N$.
1324: 
1325: Finally the norm $|\xi|$ of the Killing vector is itself
1326: $\Gamma$-invariant and hence defines a function on $N$.  Since $\xi$
1327: is spacelike, this function is positive and hence it is convenient to
1328: write it as the exponential of a real valued function $\phi:N \to \RR$
1329: which is (up to a constant multiple) called the \textbf{dilaton}.
1330: 
1331: In summary, and omitting the pull-backs on $h$ and $\phi$, we can
1332: write the metric $g$ as
1333: \begin{equation*}
1334:   g = h + e^{2\phi} \alpha^2~.
1335: \end{equation*}
1336: 
1337: The other geometric data also reduces.  For example, if $F$ is an
1338: invariant differential form on $M$, it gives rise to two differential
1339: forms on $N$ simply by decomposing
1340: \begin{equation*}
1341:   F = G - \alpha \wedge H~,
1342: \end{equation*}
1343: where $\alpha$ is the connection one-form defined above.  The forms
1344: $G$ and $H$ are basic and hence define differential forms on $N$.
1345: Indeed, it is clear from the above expression that $H = -\imath_\xi
1346: F$, so that it is manifestly horizontal.  Invariance of $F$ means that
1347: $H$ is closed, whence it is also invariant.  Finally, we observe that
1348: $G$ is also basic.  It is manifestly horizontal, and invariance
1349: follows by a simple calculation using that $G$, $H$ and $d\alpha$ are
1350: horizontal.
1351: 
1352: \subsection{Eleven-dimensional supergravity}
1353: \label{sec:maxd=11}
1354: 
1355: Maximal supersymmetry implies the flatness of the supersymmetric
1356: connection (\ref{eq:connd=11}).  Calculating the curvature of this
1357: connection and separating into types one arrives at the following
1358: conditions:
1359: \begin{itemize}
1360: \item $\nabla F = 0$;
1361: \item the Riemann curvature tensor is given by
1362:   \begin{equation}
1363:     \label{eq:riemd=11}
1364:     \Riem(g) = \tfrac1{12} T^{[4]} + \tfrac1{36} (g \odot T^{[2]}) -
1365:     \tfrac1{72} |F|^2 (g \odot g)~,
1366:   \end{equation}
1367:   where $\odot$ is the Kulkarni--Nomizu product (see, e.g.,
1368:   \cite[§1.G, 1.110]{Besse}), and the tensors $T^{[2k]} \in C^\infty(M, S^2
1369:   \Lambda^k T^*M)$, for $k=1,2$ are defined by
1370:   \begin{equation*}
1371:     \begin{aligned}[t]
1372:       T^{[2]}(X,Y) &:= \left< \iota_X F, \iota_Y F\right>\\
1373:       T^{[4]}(X,Y,W,Z) &:= \left<\iota_X \iota_Y F, \iota_W \iota_Z F
1374:       \right>
1375:     \end{aligned}
1376:   \end{equation*}
1377:   for all vector fields $X,Y,W,Z$; and
1378: \item $F$ obeys the Plücker identity:
1379:   \begin{equation*}
1380:     \imath_X \imath_Y \imath_Z F \wedge F = 0~,
1381:   \end{equation*}
1382:   for all vector fields $X,Y,Z$.
1383: \end{itemize}
1384: The first two conditions imply that the Riemann tensor is parallel,
1385: whence $(M,g)$ is locally symmetric, whence locally isometric to one
1386: of the spaces in Theorem \ref{thm:CahenWallach}.  Every such space is
1387: acted on transitively by a Lie group $G$ (the group of
1388: \textbf{transvections}), whence if we fix a point in $M$ (the
1389: \textbf{origin}) with isotropy $H$, $M$ is isomorphic to the space of
1390: cosets $G/H$.  Let $\fg$ denote the Lie algebra of $G$ and $\fh$ the
1391: Lie subalgebra corresponding to $H$.  Then $\fg$ admits a vector space
1392: decomposition $\fg=\fh\oplus \fm$, where $\fm$ is isomorphic to the
1393: tangent space of $M$ at the origin.  The Lie brackets are such that
1394: \begin{equation*}
1395:   [\fh,\fh] \subset \fh \qquad   [\fh,\fm] \subset \fm \qquad
1396:   [\fm,\fm] \subset \fh~.
1397: \end{equation*}
1398: The metric $g$ on $M$ is determined by an $\fh$-invariant inner
1399: product $B$ on $\fm$.
1400: 
1401: Since $F$ is parallel, it is $G$-invariant.  This means that it is
1402: uniquely specified by its value at the origin, which defines an
1403: $\fh$-invariant four-form on $\fm$.  For $F=0$, the right-hand side of
1404: equation~\eqref{eq:riemd=11} vanishes, and hence $g$ is flat.  We will
1405: therefore assume that $F\neq 0$.  The Plücker identity says that it is
1406: then decomposable, whence it determines a four-dimensional vector
1407: subspace $\fn \subset \fm$ as follows: if at the origin $F = \theta_1
1408: \wedge \theta_2 \wedge \theta_3 \wedge \theta_4$, then $\fm$ is the
1409: span of (the dual vectors to) the $\theta_i$.  Furthermore, because
1410: $F$ is invariant, we have that $H$ leaves the space $\fn$ invariant,
1411: whence $[\fh, \fn] \subset \fn$, which means that the holonomy group
1412: of $M$ (which is isomorphic to $H$) acts reducibly.  In lorentzian
1413: signature this does not imply that the space is locally isometric to a
1414: product, since the metric may be degenerate when restricted to $\fn$.
1415: Therefore we must distinguish between two cases, depending on whether
1416: or not the restriction $B|_{\fn}$ of $B$ to $\fn$ is or is not
1417: degenerate.
1418: 
1419: If $B|_{\fn}$ is non-degenerate, then it follows from the de~Rham--Wu
1420: decomposition theorem \cite{Wu} that the space is locally isometric to
1421: a product $N\times P$, with $N$ and $P$ locally symmetric spaces of
1422: dimensions four and seven, respectively.  Explicitly, we can see this
1423: as follows: there exists a $B$-orthogonal decomposition
1424: $\fm=\fn\oplus\fp$, with $\fp := \fm^\perp$, where $[\fh,\fp] \subset
1425: \fp$ because of the invariance of the inner product.  Let $\fg_N = \fh
1426: \oplus \fn$ and $\fg_P = \fh \oplus \fp$.  They are clearly both Lie
1427: subalgebras of $\fg$.  Let $G_N$ and $G_P$ denote the respective
1428: (connected, simply-connected) Lie groups.  Then $N$ will be locally
1429: isometric to $G_N/H$ and $P$ will be locally isometric to $G_P/H$, and
1430: $M$ will be locally isometric to the product.  The metrics on $N$ and
1431: $P$ are induced by the restrictions of $\fn$ and $\fp$ respectively of
1432: the inner product $B$ on $\fn \oplus \fp$, denoted
1433: \begin{equation}
1434:   \begin{aligned}
1435:     B_\fn&= B|_\fn
1436:     \\
1437:     B_\fp&=B|_\fp~.
1438:   \end{aligned}
1439: \end{equation}
1440: We shall denote the metrics on $N$ and $P$ induced from the above
1441: inner products by $h$ and $m$, respectively.
1442: 
1443: On the other hand if the restriction $B|_\fn$ is degenerate, so that
1444: $\fn$ is a null four-dimensional subspace of $\fm$, the four-form $F$
1445: is also null.  From Theorem~\ref{thm:CahenWallach} one sees (see,
1446: e.g., \cite{FOPflux}) that the only lorentzian symmetric spaces
1447: admitting parallel null forms are those which are locally isometric to
1448: a product $M=\CW_d(A)\times Q_{11-d}$, where $\CW_d(A)$ is a
1449: $d$-dimensional Cahen-Wallach space and $Q_{11-d}$ is an
1450: $(11{-}d)$-dimensional riemannian symmetric space.
1451: 
1452: In summary, there are two separate cases to consider:
1453: \begin{enumerate}
1454: \item $(M,g)= (N_4 \times P_7, h \oplus m)$ (locally), where
1455:   $(N,h)$ and $(P,m)$ are symmetric spaces and where $F$ is
1456:   proportional to (the pull-back of) the volume form on $(N,h)$; or
1457: \item $M = \CW_d(A) \times Q_{11-d}$ (locally) and $d\geq 3$, where
1458:   $Q_{11-d}$ is a riemannian symmetric space.
1459: \end{enumerate}
1460: 
1461: In \cite{FOPMax} these cases are analysed further, resulting in the
1462: following theorem.
1463: 
1464: \begin{theorem}[FO-Papadopoulos \cite{FOPMax}]
1465:   \label{th:d=11}
1466:   Let $(M,g,F)$ be a maximally supersymmetric solution of
1467:   eleven-dimensional supergravity.  Then it is locally isometric to
1468:   one of the following:
1469:   \begin{itemize}
1470:   \item $\AdS_7(-7R) \times S^4(8R)$ and $F = \sqrt{6 R} \dvol(S^4)$,
1471:     where $R>0$ is the constant scalar curvature of $M$;
1472:   \item $\AdS_4(8R) \times S^7(-7R)$ and $F = \sqrt{-6 R}
1473:     \dvol(\AdS_4)$, where $R<0$ is again the constant scalar curvature
1474:     of $M$; or
1475:   \item $\CW_{11}(A)$ with $A=-\frac{\mu^2}{36}\,
1476:     \text{diag}(4,4,4,1,1,1,1,1,1)$ and\\
1477:     $F = \mu\, dx^- \wedge dx^1\wedge dx^2 \wedge dx^3$.  One must
1478:     distinguish between two cases:
1479:     \begin{itemize}
1480:     \item $\mu=0$: which recovers the flat space solution $\EE^{1,10}$
1481:           with $F=0$; and
1482:     \item $\mu\neq 0$: all these are isometric and describe a
1483:       symmetric plane wave.
1484:     \end{itemize}
1485:   \end{itemize}
1486: \end{theorem}
1487: 
1488: The first two solutions are the well-known Freund--Rubin backgrounds
1489: \cite{FreundRubin} and \cite{AdS7S4}, whereas the plane wave was
1490: originally discovered by Kowalski-Glikman \cite{KG} and rediscovered
1491: subsequently in \cite{FOPflux}.  All of these solutions are locally
1492: isometric to the intersection of two quadrics in $\EE^{11,2}$.
1493: Moreover, as shown in \cite{ShortLimits,Limits} they are related by
1494: ``plane-wave limits'' \cite{PenrosePlaneWave,GuevenPlaneWave}.
1495: 
1496: \subsection{Ten-dimensional IIA supergravity}
1497: \label{sec:maxIIA}
1498: 
1499: Type IIA supergravity \cite{GianiPerniciIIA, CampbellWestIIA,
1500:   HuqNamazieIIA} is obtained by dimensional reduction from
1501: eleven-dimensional supergravity.  From the discussion in
1502: Section~\ref{sec:kk} and the fact that a $d=11$ background is
1503: characterised by a metric $g$ and a $4$-form $F$, it follows that a
1504: IIA supergravity background is characterised by a quintuplet
1505: $(h,\phi,\Omega,G,H)$ where $h$ is a lorentzian metric on a
1506: ten-dimensional spacetime $N$, $\phi$ is a real function on $N$,
1507: $\Omega$ a closed $2$-form which is the curvature of a principal
1508: $\Gamma$-bundle over $N$, $G$ a $4$-form on $N$ and $H$ a $3$-form on
1509: $N$.  The PDEs satisfied by these fields are obtained by reducing
1510: those in $d=11$ supergravity by the action of the group $\Gamma$.
1511: 
1512: It is a fundamental property of the Ka{\l}u\.za--Klein reduction, that
1513: any IIA supergravity background can be lifted (or ``oxidised'') to a
1514: background of eleven-dimensional supergravity possessing a
1515: one-parameter group symmetries.  If the IIA supergravity solution
1516: preserves some supersymmetry, its lift to eleven dimensions will
1517: preserve at least the same amount of supersymmetry. This means that a
1518: maximally supersymmetric solution of IIA supergravity will uplift to
1519: one of the maximally supersymmetric solutions of eleven-dimensional
1520: supergravity determined in the previous section.  Therefore the
1521: determination of the maximally supersymmetric IIA backgrounds reduces
1522: to classifying those dimensional reductions of the maximally
1523: supersymmetric eleven-dimensional backgrounds which preserve all
1524: supersymmetry.
1525: 
1526: As explained already in \cite{FOPflux}, the only such reductions are
1527: the reductions of the flat eleven-dimensional background by a
1528: translation subgroup of the Poincaré group.  In summary, one has
1529: 
1530: \begin{corollary}[FO-Papadopoulos \cite{FOPMax}]
1531:   \label{th:IIA}
1532:   Any maximally supersymmetric solution of type IIA supergravity is
1533:   locally isometric to $\EE^{1,9}$ with zero fluxes and constant
1534:   dilaton.
1535: \end{corollary}
1536: 
1537: \subsection{Ten-dimensional IIB supergravity}
1538: \label{sec:maxIIB}
1539: 
1540: A maximally supersymmetric background of IIB supergravity admits a
1541: (real) $32$-dimensional space of Killing spinors.  Since this is the
1542: (real) rank of the spinor bundle $\eS$ defined in
1543: Section~\ref{sec:IIB}, it means that at any given point, there is a
1544: basis for the spinor bundle consisting of Killing spinors.  These
1545: spinors satisfy equation \eqref{eq:dilatinoIIB}, whence $c(B)$ and
1546: $c(G)$ must vanish separately, which in turn imply the vanishing of
1547: $G$ and $B$.  In particular, this has a consequence that $z$ and hence
1548: $\tau$ are constant, whence the connection $A$ on $L_\tau$ also
1549: vanishes.  Maximally supersymmetric backgrounds have the form
1550: $(M,g,F)$ and are parametrised by the upper half-plane via the
1551: constant parameter $\tau$.  Maximally supersymmetry now implies the
1552: flatness of the connection $D$ defined by equation
1553: \eqref{eq:gravitinoIIB} and which takes the simplified form
1554: \begin{equation*}
1555:   D_X \varepsilon = \nabla_X \varepsilon + \tfrac{i}4 c(F) c(X^\flat)
1556:   \varepsilon~,
1557: \end{equation*}
1558: where $\nabla$ is the spin connection.  Notice that the equations of
1559: motion now say that $F$ is closed.
1560: 
1561: Computing the curvature of this connection, and separating into types,
1562: we arrive at the following conditions:
1563: \begin{itemize}
1564: \item $\nabla F = 0$;
1565: \item the Riemann curvature tensor is given by
1566:   \begin{equation}
1567:     \label{eq:RiemIIB}
1568:     R(X,Y,Z,W) = \left<\imath_X\imath_ZF,\imath_Y\imath_WF\right> - 
1569:     \left<\imath_X\imath_WF,\imath_Y\imath_ZF\right>~.
1570:   \end{equation}
1571:   Since $F$ is parallel, this means that so is the Riemann tensor,
1572:   whence $(M,g)$ is locally symmetric; and
1573: \item $F$ obeys an identity reminiscent of both the Plücker and Jacobi
1574:   identities:
1575:   \begin{equation}
1576:     \label{eq:plujac}
1577:     \lambda(\imath_X\imath_Y\imath_Z F) F = 0 \qquad\text{for all
1578:       vector fields $X,Y,Z$,}
1579:   \end{equation}
1580:   where $\lambda : \Omega^2(M) \to \End(\Lambda^5 T^*M)$ is the
1581:   composition of the (metric-induced) isomorphism $\Omega^2(TM) \cong
1582:   \fso(TM)$ between $2$-forms and skew-symmetric endomorphisms of the
1583:   tangent bundle and the action of such endomorphisms on the
1584:   $5$-forms.
1585: \end{itemize}
1586: 
1587: It was proved in \cite{FOPPluecker} that equation~\eqref{eq:plujac}
1588: implies that $F = G + \star G$, where $G = \theta_1 \wedge \theta_2
1589: \wedge \theta_3 \wedge \theta_4 \wedge \theta_5$ is a parallel
1590: decomposable form.
1591: 
1592: The ensuing analysis follows closely the case of eleven-dimensional
1593: supergravity and will not be repeated here.  We must distinguish
1594: between two cases, depending on whether or not the five-form $G$ is
1595: null.  First suppose that $G$ (and hence $F$) is not null.  Then the
1596: five-form $G$ induces a local decomposition of $(M,g)$ into a product
1597: $N_5\times P_5$ of two five-dimensional symmetric spaces $(N,h)$ and
1598: $(P,m)$, where $G \propto \dvol(N)$ and hence $\star G \propto
1599: \dvol(P)$.  Since $(M,g)$ is lorentzian, one of the spaces $(N,h)$ and
1600: $(P,m)$ is lorentzian and the other riemannian.  By interchanging $G$
1601: with $\star G$ if necessary, we can assume that $G$ has positive norm
1602: and hence that $N$ is riemannian.
1603: 
1604: In summary, there are two separate cases to consider:
1605: \begin{enumerate}
1606: \item $(M,g)= (N_5 \times P_5, h \oplus m)$ (locally), where
1607:   $(N,h)$ and $(P,m)$ are symmetric spaces and where $F=G + \star G$
1608:   and $G$ is proportional to (the pull-back of) the volume form on
1609:   $(N,h)$; or
1610: \item $M = \CW_d(A) \times Q_{10-d}$ (locally) and $d\geq 3$, where
1611:   $Q_{10-d}$ is a riemannian symmetric space.
1612: \end{enumerate}
1613: 
1614: In \cite{FOPMax} these cases are analysed further, resulting in the
1615: following theorem.
1616: 
1617: \begin{theorem}[FO-Papadopoulos \cite{FOPMax}]
1618:   \label{th:IIB}
1619:   Let $(M,g,F_5^+,...)$ be a maximally supersymmetric solution of
1620:   ten-dimensional type IIB supergravity.  Then it has constant
1621:   axi-dilaton (normalised so that $z=0$ in the formulas below), all fluxes 
1622:   vanish except for the one corresponding to the self-dual five-form,
1623:   and is locally isometric to one of the following:
1624:   \begin{itemize}
1625:   \item $\AdS_5(-R) \times S^5(R)$ and $F = 2 \sqrt{\frac{R}5}
1626:     \left(\dvol(\AdS_5) + \dvol(S^5)\right)$, where $\pm R$ are the
1627:     scalar curvatures of $\AdS_5$ and $S^5$, respectively; or
1628:   \item $\CW_{10}(A)$ with $A=-\mu^2 \1$ and $F = \half \mu\, dx^-
1629:     \wedge (dx^1\wedge dx^2 \wedge dx^3 \wedge dx^4 + dx^5\wedge dx^6
1630:     \wedge dx^7 \wedge dx^8)$.  One must distinguish between two
1631:     cases:
1632:     \begin{itemize}
1633:     \item $\mu=0$: which yields the flat space solution $\EE^{1,9}$
1634:       with zero fluxes; and
1635:     \item $\mu\neq 0$: all these are isometric and describe a
1636:       symmetric plane wave.
1637:     \end{itemize}
1638:   \end{itemize}
1639: \end{theorem}
1640: 
1641: The first solution is the well-known Freund--Rubin background
1642: mentioned originally in \cite{SchwarzIIB}.  The plane wave solution
1643: was discovered in \cite{NewIIB}.  As in eleven-dimensional
1644: supergravity, the solutions above are locally isometric to the
1645: intersection of two quadrics in $\EE^{10,2}$ and as shown in
1646: \cite{ShortLimits,Limits} they are related by plane-wave limits.
1647: 
1648: \subsection{Six-dimensional $(2,0)$ and $(1,0)$ supergravities}
1649: \label{sec:maxsixd}
1650: 
1651: In this case, maximal supersymmetry implies the flatness of the
1652: supersymmetric connection $D$ in \eqref{eq:conn10} which, as explained
1653: in Section~\ref{sec:sixdim}, is induced from a metric connection with
1654: closedd torsion $3$-form $H$.  In Section~\ref{sec:flat} we showed
1655: that $(M,g)$ is locally isometric to a six-dimensional Lie group with
1656: a bi-invariant lorentzian metric.  The only extra condition is that
1657: $H$, the canonical bi-invariant $3$-form associated to such a Lie
1658: group, should be self-dual.
1659: 
1660: We therefore look for Lie algebras with invariant lorentzian scalar
1661: products relative to which the canonical invariant $3$-form is
1662: anti-self dual.  As explained in Section~\ref{sec:lie}, such Lie
1663: algebra is a direct sum of indecomposables.  Furthermore, if the Lie
1664: algebra is indecomposable then it must be the double extension of an
1665: abelian Lie algebra by a one-dimensional Lie algebra and hence
1666: solvable (see, e.g., \cite{MedinaRevoy}).
1667: 
1668: These considerations make possible the following enumeration of
1669: six-dimensional lorentzian Lie algebras:
1670: \begin{enumerate}
1671: \item $\EE^{1,5}$
1672: \item $\EE^{1,2} \oplus \fso(3)$
1673: \item $\EE^3 \oplus \fso(1,2)$
1674: \item $\fso(1,2) \oplus \fso(3)$
1675: \item $\fd(\EE^4,\RR)$
1676: \end{enumerate}
1677: where the last case actually corresponds to a family of Lie algebras,
1678: depending on the action of $\RR$ on $\EE^4$, which is given by a
1679: homomorphism $\RR \to \fso(4)$.
1680: 
1681: Imposing the condition of anti-selfduality trivially discards cases
1682: (2) and (3) above.  Case (1) is the abelian Lie algebra with Minkowski
1683: metric.  The remaining two cases were investigated in
1684: \cite{CFOSchiral} (see also \cite{GMR}) in detail and we review this
1685: below.
1686: 
1687: \subsubsection{A six-dimensional Cahen--Wallach space}
1688: 
1689: Let $e_i$, $i=1,2,3,4$, be an orthonormal basis for $\EE^4$, and let
1690: $e_- \in \RR$ and $e_+\in\RR^*$, so that together they span
1691: $\fd(\EE^4,\RR)$.  The action of $\RR$ on $\EE^4$ defines a map $\rho:
1692: \RR \to \Lambda^2 \EE^4$, which can be brought to the form $\rho(e_-) =
1693: \alpha e_1 \wedge e_2 + \beta e_3 \wedge e_4$ via an orthogonal change
1694: of basis in $\EE^4$ which moreover preserves the orientation.  The Lie
1695: brackets of $\fd(\EE^4,\RR)$ are given by
1696: \begin{equation*}
1697:   \begin{aligned}[m]
1698:     [e_-,e_1] &= \alpha e_2\\
1699:     [e_-,e_2] &= -\alpha e_1\\
1700:     [e_1,e_2] &= \alpha e_+
1701:   \end{aligned}\qquad\qquad
1702:   \begin{aligned}[m]
1703:     [e_-,e_3] &= \beta e_4\\
1704:     [e_-,e_4] &= -\beta e_3\\
1705:     [e_3,e_4] &= \beta e_+
1706:   \end{aligned}~,
1707: \end{equation*}
1708: and the scalar product is given (up to scale) by
1709: \begin{equation*}
1710:   \left<e_-,e_-\right> = b \qquad \left<e_+,e_-\right> = 1 \qquad
1711:   \left<e_i,e_j\right> = \delta_{ij}~.
1712: \end{equation*}
1713: The first thing we notice is that we can set $b=0$ without loss of
1714: generality by the automorphism fixing all $e_i,e_+$ and mapping $e_-
1715: \mapsto e_- - \half b e_+$.  We will assume that this has been done
1716: and that $\left<e_-,e_-\right>=0$.  A straightforward calculation
1717: shows that the three-form $H$ is anti-selfdual if and only if $\beta =
1718: \alpha$.  Let us put $\beta=\alpha$ from now on.  We must distinguish
1719: between two cases: if $\alpha = 0$, then the resulting algebra is
1720: abelian and is precisely $\EE^{1,5}$.  On the other hand if
1721: $\alpha\neq 0$, then rescaling $e_\pm \mapsto \alpha^{\pm 1} e_\pm$ we
1722: can effectively set $\alpha = 1$ without changing the scalar product.
1723: Finally we notice that a constant rescaling of the scalar product can
1724: be undone by an automorphism of the algebra.  As a result we have two
1725: cases: $\EE^{1,5}$ (obtained from $\alpha =0$) and the algebra
1726: \begin{equation}
1727:   \label{eq:nw6}
1728:   \begin{aligned}[m]
1729:     [e_-,e_1] &= e_2\\
1730:     [e_-,e_2] &= - e_1\\
1731:     [e_1,e_2] &= e_+
1732:   \end{aligned}\qquad\qquad
1733:   \begin{aligned}[m]
1734:     [e_-,e_3] &= e_4\\
1735:     [e_-,e_4] &= - e_3\\
1736:     [e_3,e_4] &= e_+
1737:   \end{aligned}~,
1738: \end{equation}
1739: with scalar product given by
1740: \begin{equation}
1741:   \label{eq:nw6m}
1742:   \left<e_+,e_-\right> = 1 \qquad\text{and}\qquad
1743:   \left<e_i,e_j\right> = \delta_{ij}~.
1744: \end{equation}
1745: There is a unique simply-connected Lie group with the above Lie
1746: algebra which inherits a bi-invariant lorentzian metric.  This Lie
1747: group is a six-dimensional analogue of the Nappi--Witten group
1748: \cite{NW}, which is based on the double extension $\fd(\EE^2,\RR)$
1749: \cite{FSsug}.  This was denoted $\NW_6$ in \cite{FSPL}, where one
1750: can find a derivation of the metric on this six-dimensional group.
1751: The supergravity solution was discovered by Meessen \cite{Meessen} who
1752: called it KG6 by analogy with the maximally supersymmetric plane wave
1753: of eleven-dimensional supergravity discovered by Kowalski-Glikman
1754: \cite{KG} and rediscovered in \cite{FOPflux}.
1755: 
1756: The metric is easy to write down once we choose a parametrisation for
1757: the group.  The calculation is routine (see, for example, \cite{FSPL})
1758: and the result is
1759: \begin{equation}
1760:   \label{eq:CWmetric2}
1761:   g = 2 dx^+ dx^- - \tfrac14 \sum_i (x^i)^2 (dx^-)^2 + \sum_i
1762:   (dx^i)^2~.
1763: \end{equation}
1764: In these coordinates the three-form $H$ is given by
1765: \begin{equation*}
1766:   H = \tfrac23 dx^- \wedge (dx^1 \wedge dx^2 + dx^3 + dx^4)~.
1767: \end{equation*}
1768: 
1769: \subsubsection{The Freund--Rubin backgrounds}
1770: 
1771: Finally we discuss case (4), with Lie algebra $\fso(1,2) \oplus
1772: \fso(3)$.  Let $e_0,e_1,e_2$ be a pseudo-orthonormal basis for
1773: $\fso(1,2)$.  The Lie brackets are given by
1774: \begin{equation*}
1775:   [e_0,e_1] = -e_2\qquad   [e_0,e_2] = e_1\qquad   [e_1,e_2] =
1776:   e_0~.
1777: \end{equation*}
1778: Similarly let $e_3,e_4,e_5$ denote an orthonormal basis for
1779: $\fso(3)$, with Lie brackets
1780: \begin{equation*}
1781:   [e_5,e_3] = -e_4\qquad   [e_5,e_4] = e_3\qquad   [e_3,e_4] =
1782:   -e_5~.
1783: \end{equation*}
1784: The most general invariant lorentzian scalar product on
1785: $\fso(1,2)\oplus\fso(3)$ is labelled by two positive numbers $\alpha$
1786: and $\beta$ and is given by
1787: \begin{equation*}
1788:   \bordermatrix{& e_0 & e_1 & e_2 & e_3 & e_4 & e_5 \cr
1789:   e_0 &  -\alpha & 0  &  0  & 0 & 0 & 0 \cr
1790:   e_1 & 0 & \alpha & 0  & 0 & 0 & 0 \cr
1791:   e_2 & 0 & 0 & \alpha & 0 & 0 & 0 \cr
1792:   e_3 & 0 & 0 & 0 & \beta & 0 & 0 \cr
1793:   e_4 & 0 & 0 & 0 & 0 & \beta &  0 \cr
1794:   e_5 & 0 & 0 & 0 & 0 & 0 & \beta \cr}~.  
1795: \end{equation*}
1796: Anti-selfduality of the canonical three-form implies that $\beta =
1797: \alpha$.  There is a unique simply-connected Lie group with Lie
1798: algebra $\fso(1,2) \oplus \fso(3)$, namely $\widetilde{\SL(2,\RR)}
1799: \times \SU(2)$, where $\widetilde{\SL(2,\RR)}$ denotes the universal
1800: covering group of $\SL(2,\RR)$.  This group inherits a one-parameter
1801: family of bi-invariant metrics.  This solution is none other than the
1802: standard Freund--Rubin solution $\AdS_3 \times S^3$, with equal radii
1803: of curvature, where strictly speaking we should take the universal
1804: covering space of $\AdS_3$.
1805: 
1806: In summary, the following are the possible maximally supersymmetric
1807: backgrounds of $(1,0)$ supergravity, and of $(2,0)$ supergravity up to
1808: the action of the R-symmetry group.  First of all we have a
1809: one-parameter family of Freund-Rubin backgrounds locally isometric to
1810: $\AdS_3 \times S^3$, with equal radii of curvature.  The anti-selfdual
1811: three-form $H$ is then proportional to the difference of the volume
1812: forms of the two spaces.  Then we have a six-dimensional analogue
1813: $\NW_6$ of the Nappi--Witten group, locally isometric to a
1814: Cahen--Wallach symmetric space.  Finally there is flat Minkowski
1815: spaceteime $\EE^{1,5}$.  These backgrounds are related by Penrose
1816: limits which can be interpreted in this case as group contractions.
1817: The details appear in \cite{FSPL}.
1818: 
1819: 
1820: \subsection{Five-dimensional $N=2$ supergravity}
1821: \label{sec:maxN=2}
1822: 
1823: In this section we will review the dimensional reductions of the
1824: six-dimensional backgrounds just found.  Dimensional reduction usually
1825: breaks some supersymmetry: in the ten- and eleven-dimensional
1826: supergravity theories, only the flat background remains maximally
1827: supersymmetric after dimensional reduction and then only by a
1828: translation.  However for the six-dimensional backgrounds the
1829: situation is different.  Indeed, in \cite{LMO8} it was shown that the
1830: thereto known maximally supersymmetric backgrounds with eight
1831: supercharges in six, five and four dimensions are related by
1832: dimensional reduction and oxidation.  As we will see presently, this
1833: perhaps surprising phenomenon stems from the fact that the
1834: six-dimensional backgrounds are parallelised Lie groups.  Our results
1835: will also give an \emph{a priori} explanation to the empirical fact
1836: that these backgrounds are homogeneous \cite{ALO}.
1837: 
1838: We now explain the technical result which underlies this result.  Let
1839: $D$ be a metric connection with torsion $T$.  We observe that if a
1840: vector field $\xi$ is $D$-parallel then it is Killing.  Now let $\psi$
1841: be a Killing spinor; that is, $D\psi = 0$. Then the Lie derivative of
1842: $\psi$ along $\xi$ is well-defined (see, for example,
1843: \cite{JMFKilling}) and, furthermore, it vanishes identically.
1844: Moreover, if $\eL_\xi \psi = 0$ for \emph{all} Killing spinors then
1845: $D\xi = 0$.
1846: 
1847: For a parallelised Lie group $G$, the $D$-parallel vectors are either
1848: the left- or right-invariant vector fields, depending on the choice of
1849: parallelising connection.  For definiteness, we will choose the
1850: connection whose parallel sections are the left-invariant vector
1851: fields.  Left-invariant vector fields generate right translations and
1852: are in one-to-one correspondence with elements of the Lie algebra
1853: $\fg$.  Therefore every left-invariant vector field $\xi$ determines a
1854: one-parameter subgroup $K$, say, of $G$ and the orbits of such a
1855: vector field in $G$ are the right $K$-cosets.  The dimensional
1856: reduction along this vector field is smooth and diffeomorphic to the
1857: space of cosets $G/K$.  We will be interested in subgroups $K$ such
1858: that $G/K$ is a five-dimensional lorentzian spacetime, which requires
1859: that the right $K$-cosets are spacelike.  In other words, we require
1860: that the Killing vector $\xi$ be spacelike.  Bi-invariance of the
1861: metric guarantees that this is the case provided that the Lie algebra
1862: element $\xi(e)\in \fg$ is spacelike relative to the invariant scalar
1863: product.  Further notice that a constant rescaling of $\xi$ does not
1864: change its causal property nor the subgroup $K$ it generates: it is
1865: simply reparameterised.  Therefore, in order to classify all possible
1866: reductions we need to classify all spacelike elements of $\fg$ up to
1867: scale.  Moreover elements of $\fg$ which are related by isometric
1868: automorphisms (e.g., which are in the same adjoint orbit of $G$) give
1869: rise to isometric quotients.  Thus, to summarise, we want to classify
1870: spacelike elements of $\fg$ up to scale and up to automorphisms.
1871: 
1872: As discussed in Section~\ref{sec:kk}, the reduction of the
1873: six-dimensional metric to five dimensions gives rise to a metric $h$,
1874: a dilaton $\phi$ and a curvature 2-form $F$.  The dilaton $\phi$ is a
1875: logarithmic measure of the fibre metric $\|\xi_X\|$ which in our case
1876: is constant, and $F = d\alpha$ (omitting pullbacks).  We can give an
1877: explicit formula for $\Omega$ using the Maurer--Cartan structure
1878: equations.  Indeed,
1879: \begin{equation}
1880:   \label{eq:F}
1881:   F = d\alpha = \left<X,d\theta\right> = -\half
1882:   \left<X,[\theta,\theta]\right>~.
1883: \end{equation}
1884: In terms of this data, the metric on the $G$ is given by the usual
1885: Ka{\l}u\.za--Klein ansatz
1886: \begin{equation*}
1887:   ds^2 = h + \alpha^2~,
1888: \end{equation*}
1889: where we have set the dilaton to zero in agreement with the choice of
1890: normalisation for $\xi_X$.  More explicitly the metric on the
1891: five-dimensional quotient is given by
1892: \begin{equation*}
1893:   h = \left<\theta,\theta\right> - \left<X,\theta\right>^2~.
1894: \end{equation*}
1895: 
1896: To reduce the anti-selfdual three-form $H$ we first decompose it as
1897: \begin{equation*}
1898:   H = G_3 + \alpha \wedge G_2~,
1899: \end{equation*}
1900: where $G_2 = \iota_{\xi_X} H$ and $G_3$ are basic.  Because $dH=0$ it
1901: follows that $dG_2=0$ and that $dG_3 + F \wedge G_2=0$ where $F =
1902: d\alpha$ was defined above.  Finally because $H$ is anti-selfdual, it
1903: follows that $G_3$ and $G_2$ are related by Hodge duality in five
1904: dimensions: $G_3 = \star_h G_2$.  In other words, we have that
1905: \begin{equation*}
1906:   H = \star_h G_2 + \alpha \wedge G_2~,
1907: \end{equation*}
1908: where $dG_2 = 0$ and $d\star_h G_2 = -F \wedge G_2$.
1909: 
1910: In fact, in this case we have $F=G_2$.  Indeed, using that $H = -\tfrac16
1911: \left<\theta,[\theta,\theta]\right>$, we compute
1912: \begin{equation*}
1913:    G_2 = \iota_{\xi_X} H = -\half \left<X,[\theta,\theta]\right>~,
1914: \end{equation*}
1915: which agrees with the expression for $F$ derived in \eqref{eq:F}.
1916: 
1917: In summary, for the reductions under consideration, we obtain a
1918: maximally supersymmetric background of the minimal $N{=}2$
1919: supergravity with bosonic fields $(h,F)$ given by the reduction of
1920: $(g,H)$ where $F=d\alpha$, $h = g - \alpha^2$ and $H = \star_h F +
1921: \alpha \wedge F$.
1922: 
1923: The different reductions were classified in \cite{CFOSchiral}, to
1924: where we send the reader for details, hence obtaining all the
1925: maximally supersymmetric backgrounds of the minimal $N{=}2$
1926: supergravity and thus completing the classification of supersymmetric
1927: backgrounds in \cite{GGHPR}.  Among the maximally supersymmetric
1928: backgrounds one finds the near-horizon geometries \cite{GMT1} of the
1929: rotating black holes of \cite{BMPV,KRW}, the symmetric plane-wave of
1930: \cite{Meessen} and the Gödel-like background discovered in
1931: \cite{GGHPR}.
1932: 
1933: 
1934: \section{Parallelisable type II backgrounds}
1935: \label{sec:parallel}
1936: 
1937: In this section we will present a classification of parallelisable
1938: type II backgrounds,  by which we mean backgrounds of both type IIA
1939: and type IIB supergravity.  Since these theories contain different
1940: dynamical degrees of freedom, common backgrounds are necessarily very
1941: special.
1942: 
1943: \begin{definition}
1944:   A type II supergravity
1945:   \idx{background}{supergravity!$d{=}10$ type~II!background} consists of a
1946:   ten-dimensional lorentzian spin manifold $(M,g)$ together with a
1947:   closed $3$-form $H$ and a smooth function $\phi: M \to \RR$ subject
1948:   to the equations of motion obtained by varying the (formal) action
1949:   functional
1950:   \begin{equation}
1951:     \label{eq:typeIIaction}
1952:     \int_M e^{-2\phi} \left( R + 4 |d\phi|^2 - \half |H|^2 \right)
1953:     \dvol_g ~,
1954:   \end{equation}
1955:   where $R$ and $\dvol_g$ are the scalar curvature and the volume form
1956:   associated to $g$.  
1957: \end{definition}
1958: 
1959: We are interested in
1960: \idx{parallelisable}{supergravity!$d{=}10$ type~II!parallelisable background}
1961: backgrounds, for which the metric connection $D$ with torsion $3$-form
1962: $H$ is flat.  In that case, the equations of motion simplify to the
1963: following three conditions:
1964: \begin{equation}
1965:   \label{eq:eomspara}
1966:   \begin{aligned}[c]
1967:     \nabla d \phi &= 0\\
1968:     d\phi \wedge \star H &= 0\\
1969:     |d\phi|^2 - \tfrac14 |H|^2 &=0~.
1970:   \end{aligned}
1971: \end{equation}
1972: 
1973: To discuss supersymmetry, we need to distinguish whether we are in
1974: type IIA or type IIB supergravity, since the spinor bundles are
1975: different.  Let $S_\pm$ denote the real $16$-dimensional half-spin
1976: representations of $\Spin(1,9)$ and let $S_A = S_+ \oplus S_-$ and
1977: $S_B = S_+ \oplus S_+$.  Let $\eS_A$ and $\eS_B$ denote the spinor
1978: bundles on $M$ associated to $S_A$ and $S_B$, respectively.  We will
1979: let $\eS$ denote either $\eS_A$ or $\eS_B$, depending on which type II
1980: theory we are considering.
1981: 
1982: \begin{definition}
1983:   A type II background is
1984:   \idx{supersymmetric}{supergravity!$d{=}10$ type~II!supersymmetric background}
1985:   if there are nonzero sections $\varepsilon$ of $\eS$ satisfying the
1986:   two conditions:
1987:   \begin{equation*}
1988:     D \varepsilon = 0 \qquad\text{and}\qquad c\left(d\phi + \half
1989:       H\right) \varepsilon = 0~,
1990:   \end{equation*}
1991:   where $c: \Omega(M) \to \Cl(TM) \to \End(\eS)$ is the Clifford
1992:   action of forms on spinors.
1993: \end{definition}
1994: 
1995: The supersymmetric parallelisable type II backgrounds were classified
1996: in \cite{JMFPara,KYPara} and revisited in the context of heterotic
1997: supergravity in \cite{FKYHeterotic}, whose treatment we follow.
1998: 
1999: \subsection{Ten-dimensional parallelisable geometries}
2000: \label{sec:parallelisms}
2001: 
2002: As explained in Section \ref{sec:parallelisable}, it is possible to
2003: list all the simply-connected parallelisable lorentzian manifolds in
2004: any dimension.  The ingredients out of which we can make them are
2005: given in Table~\ref{tab:ingredients}, whose last column follows from
2006: equation \eqref{eq:eomspara}.
2007: 
2008: Indeed, in the case of a Lie group, that is, when $dH=0$, equation
2009: \eqref{eq:eomspara} says that $d\phi$ must be central, when thought of as
2010: an element in the Lie algebra. Since $\AdS_3$, $S^3$ and $\SU(3)$ are
2011: simple, their Lie algebras have no centre, whence $d\phi=0$.  In the
2012: case of an abelian group there are no conditions, and in the case of
2013: $\CW_{2n}(A)$, the Lie algebra has a one-dimensional centre
2014: corresponding to $\d_+$, whose dual one-form is $dx^-$.  This means
2015: that $d\phi$ must be proportional to $dx^-$, whence $\phi$ can only
2016: depend on $x^-$.  Finally for $S^7$, the equation of motion $\star H
2017: \wedge d\phi = 0$ implies that $d\phi=0$.  To see this, notice that
2018: the parallelised $S^7$ possesses a nearly parallel $G_2$ structure and
2019: the differential forms decompose into irreducible types under $G_2$.
2020: For example, the one-forms corresponding to the irreducible
2021: seven-dimensional irreducible representation $\fm$ of $G_2$ coming
2022: from the embedding $G_2 \subset \SO(7)$, whereas the two-forms
2023: decompose into $\fg_2 \oplus \fm$, where $\fg_2$ is the adjoint
2024: representation which is irreducible since $G_2$ is simple.  Now, $H$
2025: and $\star H$ both are $G_2$-invariant and hence the map
2026: $\Omega^1(S^7) \to \Omega^2(S^7)$ defined by $\theta \mapsto \star
2027: (\star H \wedge\theta)$ is $G_2$-equivariant.  Since it is not
2028: identically zero, it must be an isomorphism onto its image.  Hence if
2029: $\star H \wedge d\phi = 0$, then also in this case $d\phi = 0$.
2030: 
2031: \begin{table}[h!]
2032:   \centering
2033:   \setlength{\extrarowheight}{3pt}
2034:   \renewcommand{\arraystretch}{1.3}
2035:   \begin{small}
2036:     \begin{tabular}{|>{$}l<{$}|>{$}l<{$}|>{$}l<{$}|}\hline
2037:       \multicolumn{1}{|c|}{Space} & \multicolumn{1}{c|}{Torsion} &
2038:       \multicolumn{1}{c|}{Dilaton} \\
2039:       \hline\hline
2040:       \AdS_3 & dH=0 \quad |H|^2 < 0 & \text{constant}\\
2041:       \EE^{1,0} & H=0 & \text{unconstrained}\\
2042:       \EE^{0,1} & H=0 & \text{unconstrained}\\
2043:       S^3 & dH=0 \quad |H|^2 > 0 &  \text{constant}\\
2044:       S^7 & dH\neq 0 \quad |H|^2 > 0 & \text{constant}\\
2045:       \SU(3) & dH= 0 \quad |H|^2 > 0 & \text{constant}\\
2046:       \CW_{2n}(A) & dH=0 \quad |H|^2 = 0 & \phi(x^-)\\ \hline
2047:     \end{tabular}
2048:   \end{small}
2049:   \vspace{8pt}
2050:   \caption{Elementary parallelisable (lorentzian or riemannian) geometries}
2051:   \label{tab:ingredients}
2052: \end{table}
2053: 
2054: It is now a simple matter to put these ingredients together to make up
2055: all possible ten-dimensional combinations with lorentzian signature.
2056: Doing so, we arrive at Table~\ref{tab:geometries} (see also
2057: \cite{JMFPara}, where the entry corresponding to $\EE^{1,0} \times S^3
2058: \times S^3 \times S^3$ had been omitted inadvertently and where the
2059: entries with $S^7$ had also been omitted due to the fact that in type
2060: II string theory $dH=0$).
2061: 
2062: \begin{table}[h!]
2063:   \centering
2064:   \setlength{\extrarowheight}{3pt}
2065:   \renewcommand{\arraystretch}{1.3}
2066:   \begin{small}
2067:     \begin{tabular}{|>{$}l<{$}|>{$}l<{$}|}\hline
2068:       \multicolumn{1}{|c|}{Spacetime} & \multicolumn{1}{c|}{Spacetime}\\
2069:       \hline\hline
2070:       \AdS_3 \times S^7 & \AdS_3 \times S^3 \times S^3 \times \EE\\
2071:       \AdS_3 \times S^3 \times \EE^4 & \AdS_3 \times \EE^7\\
2072:       \EE^{1,0} \times S^3 \times S^3 \times S^3 & \EE^{1,1} \times \SU(3)\\
2073:       \EE^{1,2} \times S^7 & \EE^{1,3} \times S^3 \times S^3\\
2074:       \EE^{1,6} \times S^3 & \EE^{1,9}\\
2075:       \CW_{10}(A) & \CW_8(A) \times \EE^2\\
2076:       \CW_6(A) \times S^3 \times \EE & \CW_6(A) \times \EE^4\\
2077:       \CW_4(A) \times S^3 \times S^3 & \CW_4(A) \times
2078:       S^3 \times \EE^3\\
2079:       \CW_4(A) \times \EE^6 & \\\hline
2080:     \end{tabular}
2081:   \end{small}
2082:   \vspace{8pt}
2083:   \caption{Ten-dimensional simply-connected parallelisable spacetimes}
2084:   \label{tab:geometries}
2085: \end{table}
2086: 
2087: \subsection{Type II backgrounds}
2088: \label{sec:linear}
2089: 
2090: First of all we notice that $S^7$ cannot appear because $dH=0$.
2091: Therefore the allowed backgrounds follow \emph{mutatis mutandis} from
2092: the analysis of \cite{JMFPara,KYPara}.  We start by listing the
2093: possible backgrounds and then counting the amount of supersymmetry
2094: that each preserves.  The results are summarised in
2095: Table~\ref{tab:linear} and Table~\ref{tab:summary}, which also
2096: contains the analysis of the supersymmetry preserved by the background.
2097: 
2098: \subsubsection*{$\AdS_3 \times S^3 \times S^3 \times \EE$}
2099: 
2100: Here $d\phi$ can only have nonzero components along the flat
2101: direction, which is spacelike, whence $|d\phi|^2 \geq 0$.  Equation
2102: \eqref{eq:eomspara} says that $|H|^2 \geq 0$, so that if we call $R_0$,
2103: $R_1$ and $R_2$ the radii of curvature of $\AdS_3$ and of the two
2104: $3$-spheres, respectively, then
2105: \begin{equation*}
2106:   \frac{1}{R_1^2}  +  \frac{1}{R_2^2} \geq \frac{1}{R_0^2}~.
2107: \end{equation*}
2108: This bound is saturated if and only if the dilaton is constant.
2109: 
2110: \subsubsection*{$\AdS_3 \times S^3 \times \EE^4$}
2111: 
2112: This is the limit $R_2 \to \infty$ of the above case.
2113: 
2114: \subsubsection*{$\AdS_3 \times \EE^7$}
2115: 
2116: This would be the limit $R_1 \to \infty$ of the above case, but then
2117: the inequality $R_0^{-2} \leq 0$ cannot be satisfied.  Hence this
2118: geometry is not a background (with or without supersymmetry).
2119: 
2120: \subsubsection*{$\EE^{1,9}$}
2121: 
2122: In this case $H=0$, so $|d\phi|^2 =0$.  So we can take a linear
2123: dilaton along a null direction: $\phi = a + b x^-$, for some constants
2124: $a,b$ say.
2125: 
2126: \subsubsection*{$\EE^{1,0} \times S^3 \times S^3 \times S^3$}
2127: 
2128: The dilaton can only depend on the flat coordinate, which is timelike,
2129: so $|d\phi|^2 \leq 0$.  However $|H|^2 > 0$, whence this geometry
2130: is never a background (with or without supersymmetry).
2131: 
2132: \subsubsection*{$\EE^{1,1} \times \SU(3)$}
2133: 
2134: Here $|H|^2 > 0$, and $d\phi$ can have components along
2135: $\EE^{1,1}$.  Letting $(x^0,x^1)$ be flat coordinates for $\EE^{1,1}$,
2136: we can take $\phi = a + \half |H| x^1$, for some constant $a$,
2137: without loss of generality.
2138: 
2139: \subsubsection*{$\EE^{1,3} \times S^3 \times S^3$}
2140: 
2141: Here $|H|^2 > 0$ and $d\phi$ can have components along
2142: $\EE^{1,3}$ \cite{Khuri}.  With $(x^0,x^1,x^2,x^3)$ being flat
2143: coordinates for $\EE^{1,3}$, we take $\phi = a + \half |H| x^1$, for
2144: some constant $a$.
2145: 
2146: \subsubsection*{$\EE^{1,6} \times S^3$}
2147: 
2148: This is the limit $R_2 \to \infty$ of the above case, where $R_2$ is
2149: the radius of curvature of one of the spheres
2150: \cite{DuffLu5Brane,NS5Brane}.
2151: 
2152: \subsubsection*{$\CW_{2n}(A) \times \EE^{10-2n}$, $n=2,3,4,5$}
2153: 
2154: In these cases $|H|^2 = 0$ and hence $|d\phi|^2 = 0$, so that it
2155: cannot have components along the flat directions (if any).  This means
2156: $\phi = a + b x^-$, for constants $a,b$.
2157: 
2158: \subsubsection*{$\CW_{4}(A) \times S^3 \times S^3$}
2159: 
2160: Here $|d\phi|^2 = 0$, whereas $|H|^2 >0$, hence there are no
2161: backgrounds with this geometry.
2162: 
2163: \subsubsection*{$\CW_{2n}(A) \times S^3 \times \EE^{7-2n}$,
2164:   $n=2,3$}
2165: 
2166: Here $|H|^2 > 0$, whence $|d\phi|^2 > 0$.  This means that we can
2167: take $\phi = a + b x^- + \half |H| y$, where $y$ is any flat
2168: coordinate in $\EE^{7-2n}$ and $a,b$ are constants.
2169: 
2170: \begin{table}[h!]
2171:   \centering
2172:   \setlength{\extrarowheight}{3pt}
2173:   \renewcommand{\arraystretch}{1.3}
2174:   \begin{small}
2175:     \begin{tabular}{|>{$}l<{$}|>{$}l<{$}|}\hline
2176:       \multicolumn{1}{|c|}{Geometry} &
2177:       \multicolumn{1}{c|}{Dilaton} \\
2178:       \hline\hline
2179:       \AdS_3 \times S^3 \times S^3 \times \EE & \phi = a + \half |H| y\\
2180:       \AdS_3 \times S^3 \times \EE^4 & \phi = a + \half |H| y\\
2181:       \EE^{1,1} \times \SU(3) &  \phi = a + \half |H| y\\
2182:       \EE^{1,3} \times S^3 \times S^3 & \phi = a + \half |H| y\\
2183:       \EE^{1,6} \times S^3 & \phi = a + \half |H| y\\
2184:       \EE^{1,9} & \phi = a + b x^-\\
2185:       \CW_{10}(A) & \phi = a + b x^-\\
2186:       \CW_8(A) \times \EE^2 & \phi = a + b x^-\\
2187:       \CW_6(A) \times S^3 \times \EE & \phi = a + b x^- + \half
2188:       |H| y\\
2189:       \CW_6(A) \times \EE^4 & \phi = a + b x^-\\
2190:       \CW_4(A) \times S^3 \times \EE^3 & \phi = a + b x^- +
2191:       \half |H| y\\
2192:       \CW_4(A) \times \EE^6 & \phi = a + b x^-\\
2193:       \hline
2194:     \end{tabular}
2195:   \end{small}
2196:   \vspace{8pt}
2197:   \caption{Parallelisable backgrounds with a linear dilaton.  The
2198:     notation is such that $y$ is a spacelike flat coordinate.}
2199:   \label{tab:linear}
2200: \end{table}
2201: 
2202: \begin{table}[h!]
2203:   \centering
2204:   \setlength{\extrarowheight}{3pt}
2205:   \renewcommand{\arraystretch}{1.3}
2206:   \begin{small}
2207:     \begin{tabular}{|>{$}l<{$}|>{$}c<{$}|>{$}c<{$}|}\hline
2208:       \multicolumn{1}{|c|}{Parallelisable} &
2209:       \multicolumn{2}{c|}{Supersymmetries with dilaton being}\\
2210:       \multicolumn{1}{|c|}{geometry} &
2211:       \multicolumn{1}{c|}{constant} &
2212:       \multicolumn{1}{c|}{nonconstant} \\
2213:       \hline\hline
2214:       \AdS_3 \times S^3 \times S^3 \times \EE &  16 & 16 \\
2215:       \AdS_3 \times S^3 \times \EE^4 & 16 & 16  \\
2216:       \EE^{1,1} \times \SU(3) & \times & 16 \\
2217:       \EE^{1,3} \times S^3 \times S^3 & \times & 16\\
2218:       \EE^{1,6} \times S^3 & \times & 16 \\
2219:       \EE^{1,9} & 32 & 16 \\
2220:       \CW_{10}(A) & 16, 18(A), 20, 22(A), 24(B), 28(B) & 16 \\
2221:       \CW_8(A) \times \EE^2& 16, 20 & 16\\
2222:       \CW_6(A) \times S^3 \times \EE & \times & 16\\
2223:       \CW_6(A) \times \EE^4 & 16, 24 & 16 \\
2224:       \CW_4(A) \times S^3 \times \EE^3 & \times & 16 \\
2225:       \CW_4(A) \times \EE^6 & 16 & 16\\
2226:       \hline
2227:     \end{tabular}
2228:   \end{small}
2229:   \vspace{8pt}
2230:   \caption{Supersymmetric parallelisable backgrounds, with (A) or (B)
2231:     indicating IIA or IIB.}
2232:   \label{tab:summary}
2233: \end{table}
2234: 
2235: 
2236: \bibliographystyle{utphys}
2237: \bibliography{AdS3,AdS,ESYM,Sugra,Geometry}
2238: 
2239: %\printindex
2240: 
2241: \begin{theindex}
2242: 
2243:   \item Lie algebra
2244:     \subitem double extension, 10
2245:     \subitem indecomposable, 10
2246:     \subitem lorentzian, 9
2247:   \item lorentzian symmetric space
2248:     \subitem anti\nobreakspace  {}de\nobreakspace  {}Sitter, 3
2249:     \subitem Cahen--Wallach, 3
2250:     \subitem de\nobreakspace  {}Sitter, 3
2251:     \subitem isometric embedding, 5
2252:     \subitem parallelisable, 6
2253: 
2254:   \indexspace
2255: 
2256:   \item supergravity
2257:     \subitem $d{=}10$ IIB
2258:       \subsubitem background, 16
2259:       \subsubitem duality group, 17
2260:       \subsubitem Killing spinor, 17
2261:     \subitem $d{=}10$ type\nobreakspace  {}II
2262:       \subsubitem background, 29
2263:       \subsubitem parallelisable background, 29
2264:       \subsubitem supersymmetric background, 29
2265:     \subitem $d{=}11$
2266:       \subsubitem background, 14
2267:       \subsubitem Killing spinor, 14
2268:       \subsubitem supersymmetric background, 14
2269:     \subitem $d{=}6$
2270:       \subsubitem $(1,0)$ background, 18
2271:       \subsubitem $(1,0)$ supersymmetric background,\nobreakspace 18
2272:       \subsubitem $(2,0)$ background, 18
2273:       \subsubitem $(2,0)$ supersymmetric background,\nobreakspace 18
2274:       \subsubitem R-symmetry groups, 18
2275: 
2276: \end{theindex}
2277: 
2278: \end{document}
2279: 
2280: