Lecture Notes in Mathematics  2201

Lars Schäfer

Nearly
Pseudo-Kähler
Manifolds and
Related Special
Holonomies
Lecture Notes in Mathematics 2201
Editors-in-Chief:
Jean-Michel Morel, Cachan
Bernard Teissier, Paris

Advisory Board:
Michel Brion, Grenoble
Camillo De Lellis, Zurich
Alessio Figalli, Zurich
Davar Khoshnevisan, Salt Lake City
Ioannis Kontoyiannis, Athens
Gábor Lugosi, Barcelona
Mark Podolskij, Aarhus
Sylvia Serfaty, New York
Anna Wienhard, Heidelberg
More information about this series at http://www.springer.com/series/304
Lars SchRafer

Nearly Pseudo-KRahler
Manifolds and Related
Special Holonomies

123
Lars SchRafer
Institut Differentialgeometrie
Leibniz UniversitRat Hannover
Hannover, Germany

ISSN 0075-8434 ISSN 1617-9692 (electronic)

Lecture Notes in Mathematics
ISBN 978-3-319-65806-3 ISBN 978-3-319-65807-0 (eBook)
DOI 10.1007/978-3-319-65807-0

Library of Congress Control Number: 2017951220

Mathematics Subject Classification (2010): 53C10, 53C15, 53C29, 53C30, 53C25, 53C28

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.1 Stable Forms, Half-Flat Structures and Holonomy . . . . . . . . . . . . . . . . . . 1
1.2 Nearly Kähler Geometry . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 3
1.3 Special Kähler Manifolds . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 10
1.4 Summary .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 11
1.5 Zusammenfassung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 13
2 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
2.1 Stable Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
2.1.1 Real Forms of SL.3; C/ .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
2.1.2 Relation Between Real Forms of SL.3; C/ and GC 2 ........ 25
2.1.3 Relation Between Real Forms of GC 2 and Spin.7; C/.. . . . . . 27
2.2 Almost Pseudo-Hermitian and Almost Para-Hermitian
Geometry .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 28
2.3 Linear Algebra of Three-Forms in Dimension 8 and 10 .. . . . . . . . . . . . 32
2.4 Structure Reduction of Almost "-Hermitian Six-Manifolds .. . . . . . . . 34
2.5 Pseudo-Riemannian Submersions . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 36
2.6 Para-Sasaki Manifolds .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 37
2.6.1 The T-Dual Space . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 38
3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds . . . . . . . . . . . . . 41
3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds . . . . . . . . . . 41
3.1.1 General Properties . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 41
3.1.2 Characterisations by Exterior Differential Systems
in Dimension 6 .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 45
3.1.3 Curvature Identities for Nearly "-Kähler Manifolds . . . . . . . 50
3.2 Structure Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 52
3.2.1 Kähler Factors and the Structure in Dimension 8 . . . . . . . . . . 52
3.2.2 Einstein Condition Versus Reducible Holonomy .. . . . . . . . . . 53
3.3 Twistor Spaces over Quaternionic and Para-Quaternionic
Kähler Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 58

v
vi Contents

3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds . . . . . . . . . . . . . . 62

3.4.1 General Properties . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 62
3.4.2 Co-dimension Two . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
3.4.3 Six-Dimensional Nearly Pseudo-Kähler Manifolds .. . . . . . . 64
3.4.4 General Dimension . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 66
3.4.5 The Twistor Structure .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 71
3.5 A Class of Flat Pseudo-Riemannian Lie Groups .. . . . . . . . . . . . . . . . . . . . 72
3.6 Classification Results for Flat Nearly "-Kähler Manifolds . . . . . . . . . . 76
3.6.1 Classification Results for Flat Nearly Pseudo-Kähler
Manifolds .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
3.6.2 Classification of Flat Nearly Para-Kähler Manifolds .. . . . . . 79
3.7 Conical Ricci-Flat Nearly Para-Kähler Manifolds . . . . . . . . . . . . . . . . . . . 84
3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler
Six-Manifolds .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 89
3.8.1 Linear Algebra of Five-Dimensional Reductions
of SU.1; 2/-Structures . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 89
3.8.2 Evolution of Hypo Structures . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
3.8.3 Evolution of Nearly Hypo Structures . . .. . . . . . . . . . . . . . . . . . . . 95
3.9 Results in the Homogeneous Case . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 100
3.9.1 Consequences for Automorphism Groups . . . . . . . . . . . . . . . . . . 100
3.9.2 Left-Invariant Nearly "-Kähler Structures
on SL.2; R/  SL.2; R/ . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 102
3.9.3 Real Reducible Holonomy.. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 106
3.9.4 3-Symmetric Spaces . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 106
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds .. . . . 112
3.10.1 Definitions and Geometric Identities .. . .. . . . . . . . . . . . . . . . . . . . 113
3.10.2 Lagrangian Submanifolds in Nearly Kähler
Six-Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 116
3.10.3 The Splitting Theorem .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 118
3.10.4 Lagrangian Submanifolds in Twistor Spaces . . . . . . . . . . . . . . . 123
3.10.5 Deformations of Lagrangian Submanifolds
in Nearly Kähler Manifolds . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
4 Hitchin’s Flow Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 131
./
4.1 Half-Flat Structures and Parallel G2 -Structures . . . . . . . . . . . . . . . . . . . . 131
4.1.1 Remark on Completeness: Geodesically Complete
Conformal G2 -Metrics .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 136
4.1.2 Nearly Half-Flat Structures and Nearly Parallel
./
G2 -Structures . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 138
./
4.1.3 Cocalibrated G2 -Structures and Parallel Spin.7/-
and Spin0 .3; 4/-Structures .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 140
4.2 Evolution of Nearly "-Kähler Manifolds . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 142
4.2.1 Cones over Nearly "-Kähler Manifolds .. . . . . . . . . . . . . . . . . . . . 142
4.2.2 Sine Cones over Nearly "-Kähler Manifolds . . . . . . . . . . . . . . . 144
./
4.2.3 Cones over Nearly Parallel G2 -Structures.. . . . . . . . . . . . . . . . 145
Contents vii

4.3 The Evolution Equations on Nilmanifolds  n H3 H3 . . . . . . . . . . . . . . 146

4.3.1 Evolution of Invariant Half-Flat Structures
on Nilmanifolds .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 147
4.3.2 Left-Invariant Half-Flat Structures on H3  H3 . . . . . . . . . . . . 150
4.3.3 Solving the Evolution Equations on H3  H3 . . . . . . . . . . . . . . . 157
4.4 Special Geometry of Real Forms of the Symplectic
SL.6; C/-Module ^3 C6 . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 161
4.4.1 The Symplectic SL.6; C/-Module V D ^3 C6
and Its Lagrangian Cone C.X/
of Highest Weight Vectors .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 163
4.4.2 Real Forms .G; V0 / of the Complex Module
.SL.6; C/; V/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 163
4.4.3 Classification of Open G-Orbits
on the Grassmannian X and Corresponding
Special Kähler Manifolds.. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 164
4.4.4 The Homogeneous Projective Special Para-Kähler
Manifold SL.6; R/= S.GL.3; R/  GL.3; R// .. . . . . . . . . . . . . 170

References .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 175

Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 181
Chapter 1
Introduction

1.1 Stable Forms, Half-Flat Structures and Holonomy

Following Hitchin [76], a k-form ' on a differentiable manifold M is called stable

if the orbit of '. p/ under GL.Tp M/ is open in ƒk Tp M for all p 2 M. In this text
we are mainly concerned with six-dimensional manifolds M endowed with a stable
two-form ! and a stable three-form . A stable three-form defines an endomorphism
field J on M such that J2 D "id, see (2.6). We will assume the following algebraic
compatibility equations between ! and :

2 3
! ^  D 0; J  ^  D ! :
3

The pair .!; / defines an SU. p; q/-structure if " D 1 and an SL.3; R/-structure if
" D C1. In the former case, the pseudo-Riemannian metric !. J ; / has signature
.2p; 2q/. In the latter case it has signature .3; 3/. The structure is called half-flat if
the pair .!; / satisfies the following exterior differential system:

d! 2 D 0; d D 0:

In [76], Hitchin introduced the following evolution equations for a time-dependent

pair of stable forms .!.t/; .t// evolving from a half-flat SU.3/-structure
.!.0/; .0//:

@ @
 D d!; !O D d ;
O
@t @t
2
where !O D !2 and O D J . For compact manifolds M, he showed that these
equations are the flow equations of a certain Hamiltonian system and that any
solution defined on some interval 0 2 I  R defines a Riemannian metric on

© Springer International Publishing AG 2017 1

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0_1
2 1 Introduction

M  I with holonomy group in G2 . We give a new proof of this theorem, which

does not use the Hamiltonian system and does not assume that M is compact.
Moreover, our proof yields a similar result for all three types of half-flat G-
structures: G D SU.3/; SU.1; 2/ and SL.3; R/. For the noncompact groups G we
obtain a pseudo-Riemannian metric of signature .3; 4/ and holonomy group in G2
on M  I (see Theorem 4.1.3 of Chap. 4). As an application, we prove that any six-
manifold endowed with a real analytic half-flat G-structure can be extended to a
Ricci-flat seven-manifold with holonomy group in G2 or G2 , depending on whether
G is compact or noncompact, see Corollary 4.1.6 of Chap. 4. More generally, a G-
structure .!; / is called nearly half-flat if

d D !O

and a G2 - or G2 -structure defined by a three-form ' is called nearly parallel if

d' D ' ':

We
 prove in Theorem  4.1.12 of Chap. 4 that any solution I 3 t 7!
b
!.t/ D 2d.t/; .t/ of the evolution equation

@
 D d!  "O
@t

evolving from a nearly half-flat G-structure .!.0/; .0// on M defines a nearly

parallel G2 - or G2 -structure on M  I, depending on whether G is compact or
b For compact manifolds
noncompact, (see (2.3) of Chap. 4) for the definition of d.
M and G D SU.3/ this theorem was proven by Stock [114].
The above constructions are illustrated in Sect. 4.2 of Chap. 4, where we start
with a nearly pseudo-Kähler or a nearly para-Kähler six-manifold as initial structure.
These structures are both half-flat and nearly half-flat and the resulting parallel or
nearly parallel G2 - and G2 -structures induce cone or (hyperbolic) sine cone metrics.
In Sect. 4.3 of Chap. 4, we discuss the evolution of invariant half-flat structures
on nilmanifolds. Lemma 4.3.1 of Chap. 4 shows how to simplify effectively the
ansatz for a solution for a number of nilpotent Lie algebras including the direct
sum g D h3 ˚ h3 of two Heisenberg algebras. Focusing on this case, we
determine the orbits of the Aut.h3 ˚ h3 /-action on non-degenerate two-forms !
on g which satisfy d! 2 D 0. Based on this, we describe all left-invariant half-
flat structures .!; / on H3  H3 . A surprising phenomenon occurs in indefinite
signature. Under the assumption !.z; z/ D 0, which corresponds to the vanishing
of the projection of ! on a one-dimensional space, the geometry of the metric
induced by a half-flat structure .!; / is completely determined (Proposition 4.3.7
of Chap. 4) and the evolution turns out to be affine linear (Proposition 4.3.10 of
Chap. 4). However, this evolution produces only metrics that are decomposable and
have one-dimensional holonomy group. On the other hand, we give an explicit
1.2 Nearly Kähler Geometry 3

formula in Proposition 4.3.12 of Chap. 4 for the parallel three-form ' resulting
from the evolution for any half-flat structure .!; / with !.z; z/ ¤ 0. In fact,
the formula is completely algebraic such that the integration of the differential
equation is circumvented. In particular, we give a number of explicit examples of
half-flat structures of the second kind on h3 ˚ h3 which evolve to new metrics with
holonomy group equal to G2 and G2 . Moreover, we construct an eight-parameter
family of half-flat deformations of the half-flat examples which lift to an eight-
parameter family of deformations of the corresponding parallel stable three-forms in
./
dimension 7. Needless to say, those examples of G2 -metrics on M.a; b/ for which
.a; b/ 6D R are geodesically incomplete. However, for M compact with an SU.3/-
structure, a conformal transformation produces complete Riemannian metrics on
M  R that are conformally parallel G2 .
A G2 - or G2 -structure defined by a three-form ' is called cocalibrated if

d ' ' D 0:

Hitchin proposed the following equation for the evolution of a cocalibrated G2 -

structure '.0/:

@
.' '/ D d':
@t

He proved that any solution I 3 t 7! '.t/ on a compact manifold M defines a

Riemannian metric on M  I with holonomy group in Spin.7/. We generalise also
this theorem to noncompact manifolds and show that any solution of the evolution
equation starting from a cocalibrated G2 -structure defines a pseudo-Riemannian
metric of signature .4; 4/ and holonomy group in Spin0 .3; 4/, see Theorem 4.1.13
of Chap. 4.

1.2 Nearly Kähler Geometry

Nearly Kähler manifolds have been introduced and studied by Gray [64–67] in a
series of papers in the seventies in the context of weak holonomy. These manifolds
are almost Hermitian manifolds which are of type W1 in the well-known Gray-
Hervella classification [68] of almost Hermitian structures. The most prominent
example is the six-sphere endowed with the round metric and the almost complex
structure induced by the cross product coming from the octonions [62, 63]. New
interest in the subject came in the beginning of the nineties from the relation
between real Killing spinors on six-manifolds and nearly Kähler manifolds given
by Grunewald [70]. Shortly after this paper the classification of manifolds admitting
real Killing spinors was given in [12] by using a cone construction. More precisely,
in this reference it is shown that a real Killing spinor on M corresponds to
a parallel spinor on the metric cone over M: By a theorem of Gallot [60] it
4 1 Introduction

follows, that, if the holonomy of the cone M b (over a compact simply connected
manifold M) is reducible, then M b is flat and M is isometric to the standard sphere.
Excluding the sphere the reference [12] obtains a list of possible holonomies of
the cone over manifolds admitting real Killing spinors and analyses the induced
geometry on M: Let us observe, that in the semi-Riemannian setting the situation is
different, since Section 3 of [8] gives a number of examples, where the statement of
Gallot’s theorem is not true. Nonetheless cone constructions are useful in the semi-
Riemannian setting and are still an area of active research in Riemannian geometry.
For example, inspired by the classical correspondence between Ricci-flat Kähler
manifolds and Sasaki-Einstein manifolds (see Section 5 of [12] or Chapter 11.1
of [20]) the recent article [35] studies this type of correspondence in the setting
of ‘geometries with torsion’. More recently, the reference [1] gives a systematic
approach to cone constructions over manifolds carrying a G structure endowed with
a ‘characteristic connection’, i.e. a metric connection with skew symmetric torsion
preserving the G structure and obtains ‘torsionful’ generalisations of the above
mentioned results. In the same way nearly Kähler geometry is an area of active
research during the last decades [16, 25, 98, 99]. The classification of complete
simply connected strict nearly Kähler manifolds was reduced to dimension 6 by
Nagy [98, 99] using a result of Cleyton and Swann [34]. Butruille [25] has shown
under the same assumptions that all homogenous strict nearly Kähler manifolds are
3-symmetric. The first examples of complete inhomogeneous strict nearly Kähler
manifolds had to wait until the beginning of this year [55]! For interest and
applications of all these geometric structures in physics, namely supergravity and
string theories, we may refer to [49, 57, 115].
To our best knowledge only Gray considers pseudo-Riemannian metrics in his
paper on 3-symmetric spaces [66]. The notion of a nearly para-Kähler manifold
was recently introduced by Ivanov and Zamkovoy [78] and originates in the
study of Killing spinors in semi-Riemannian geometry by Kath [81, 82]. These
manifolds naturally appear as one class in the classification of almost para-
Hermitian manifolds [59, 86], which generalises the Gray-Hervella list. In contrast
to Riemannian geometry, where a Ricci-flat nearly Kähler manifold is automatically
a Kähler manifold (compare Remark 3.1.9), a Ricci-flat nearly para-Kähler manifold
needs not necessarily to be para-Kähler (In fact, it is not, as a posteriori shown by
our examples). Therefore Ivanov and Zamkovoy [78] ask for examples of this type
and the same authors observe, that at this time there were no known examples of
Ricci-flat (non-Kähler) nearly para-Kähler manifolds in the literature.
In addition, nearly para-Kähler metrics (or more generally almost para-Hermitian
metrics) are of split signature, i.e. signature .n; n/ and for this reason of special
interest in the study of semi-Riemannian holonomy [17]. More recently nearly para-
Kähler geometry is of interest in the setting of projective holonomy [61].
The interest in flat nearly pseudo-Kähler manifolds is also motivated by the
study of tt -structures (topological-antitopological fusion structures) on the tangent-
bundle. In fact, flat nearly pseudo-Kähler manifolds provide a class of tt -structures
on the tangent-bundle [107]. A second important class of solutions is given by
1.2 Nearly Kähler Geometry 5

special Kähler manifolds [40]. In other words, one can interpret tt -structures on
the tangent-bundle as a common generalisation of these two geometries. Therefore
we shortly review this subject.
Topological-antitopological fusion or tt -geometry is a topic of mathematical
and physical interest. In about 1990 physicists studied topological-field-theories and
their moduli spaces, in particular N D 2 supersymmetric field-theories and found a
special geometric structure called topological-antitopological fusion (see the works
of Cecotti and Vafa [30] and Dubrovin [48]). These geometries are realised on the
tangent bundle of some manifold and part of their data is a Riemannian metric.
One can replace the tangent bundle by an abstract vector bundle. This step allows
for instance to consider tt -bundles as a generalisation of variations of Hodge
structures, as it was done in Hertling’s paper [74].
Let us now shortly recall some notions of tt -geometry and refer to [42] for some
survey on this differential geometric viewpoint of the field. A tt*-bundle .E; D; S/
over an almost (para-)complex manifold .M; J/ is a real vector bundle E ! M
endowed with a connection D and a section S 2 .T  M ˝ End E/ which satisfy the
tt*-equation

R D 0 for all  2 R ; (1.1)

where R is the curvature tensor of the connection D defined by

DX WD DX C cos./SX C sin./SJX for all X 2 TM : (1.2)

A metric tt*-bundle .E; D; S; g/ is a tt*-bundle .E; D; S/ endowed with a possibly

indefinite D-parallel fibre metric g such that for all p 2 M

g.SX Y; Z/ D g. Y; SX Z/ for all X; Y; Z 2 Tp M : (1.3)

A symplectic tt*-bundle .E; D; S; !/ is a tt*-bundle .E; D; S/ endowed with the

structure of a symplectic vector bundle .E; !/; such that ! is D-parallel and S is
!-symmetric, i.e. for all p 2 M

!.SX Y; Z/ D !. Y; SX Z/ for all X; Y; Z 2 Tp M : (1.4)

In [106, 107] we have shown the following general construction principle of such
geometries.
Theorem 1.2.1 Given an almost (para-)complex manifold .M; J/ with a flat con-
nection r and a decomposition of r D D C S in a connection D and a section S
in T  M ˝ End .TM/; such that J is D-parallel, i.e. DJ D 0. If .TM; D; S/ defines a
tt -bundle, such that D is r ˛ with ˛ D ˙2; then D and S are uniquely determined
by S D  12 J 1 .rJ/ and D D r  S:
6 1 Introduction

Moreover, .TM; D; S/ as above defines a tt -bundle, such that D is r ˛ with

˛ D ˙2; if and only if J satisfies

.rJX J/ D ˙J.rX J/ (1.5)

and D and S are given by S D  12 J 1 .rJ/ and D D r  S:

As an application this theorem gives solutions of symplectic tt -bundles on the
tangent bundle, which are more general than nearly Kähler manifolds in the sense,
that we admit connections r having torsion, but more special in the sense, that our
connection r has to be flat:
Theorem 1.2.2 Given an almost (para-)Hermitian manifold .M; J; g/ with a flat
metric connection r; such that .r; J/ satisfies the nearly (para-)Kähler condition.
Define S; a section in T  M ˝ End .TM/ by

1
S WD  J 1 .rJ/; (1.6)
2

then .TM; D D r  S; S; ! D g.; J 1 // defines a symplectic tt -bundle. Moreover,

it holds DJ D 0 and T D D T r  2S:
At the moment, when the present author has found this new class of solutions
to the tt -equations, there were no known examples of Levi-Civita flat nearly
para-Kähler manifolds. This is the starting point of our interest in the subject. As
one of the results presented in the present text we discuss the classification of these
manifolds in Sect. 3.6 of Chap. 3. As mentioned above, at about the same time
Ivanov and Zamkovoy [78] asked for examples of Ricci-flat nearly para-Kähler
manifolds in six dimensions with DJ ¤ 0. As a consequence of our results we
show that there exists compact six-dimensional such manifolds with DJ ¤ 0 (see
Sect. 3.6 of Chap. 3).
Another class of almost Hermitian manifolds solving the condition (1.5) is
special Kähler manifolds, which shows, that there is a strong similarity to special
Kähler geometry. This duality can also be seen in the present work. We construct
flat nearly pseudo-Kähler manifolds of split signature from a certain constant three-
form, while in [14] flat special Kähler manifolds were constructed from a symmetric
three-tensor. The classification results are given in Sect. 3.6 of Chap. 3. The first one
is Theorem 3.6.4, which encodes a flat nearly pseudo-Kähler structure in a constant
three-form  subject to two constraints. An explicit formula for J in terms of  is
given. Next we analyse the constraints on . It turns out that the first is equivalent
to require that  has isotropic support (cf. Proposition 3.6.5) and the second is
equivalent to a type condition on : We explicitly solve the two constraints on the
real three-form  (in 4m variables) in terms of a freely specifiable complex three-
form  2 ƒ3 .Cm / . In particular, any such form  ¤ 0 defines a complete simply
connected strict flat nearly pseudo-Kähler manifold, see Corollary 3.6.11.
Further we show that any strict flat nearly pseudo-Kähler manifold is locally the
product of a flat pseudo-Kähler factor of maximal dimension and a strict flat nearly
1.2 Nearly Kähler Geometry 7

pseudo-Kähler manifold of dimension 4m  12 and split signature (Theorem 3.6.9).

This implies, in particular, the non-existence of strict flat nearly Kähler manifolds
with positive definite metric. This subsection finishes with the classification of
complete simply connected flat nearly Kähler manifolds up to isomorphism in terms
of GLm .C/-orbits on ƒ3 .Cm / , see Corollary 3.6.12.
This motivates our excursion to flat structures on Lie groups: Let V be a pseudo-
Euclidian vector space and  2 ƒ3 V. Contraction with  defines a linear map
ƒ2 V  ! V. The image is the support of . In Sect. 3.5 we show that any 3-
vector  2 ƒ3 V with isotropic support defines a simply connected 2-step nilpotent
Lie group L./ with a flat bi-invariant pseudo-Riemannian metric h of the same
signature as V. We prove that this exhausts all simply connected Lie groups with a
flat bi-invariant metric, see Theorem 3.5.2. Let dim V be even and fix J 2 so.V/;
such that J 2 D Id or J 2 D Id. We denote the corresponding left-invariant
endomorphism field on the group L./ again by J. Assume that  2 ƒ3 V has
isotropic support and has, in addition, type .3; 0/ C .0; 3/: Then .L./; h; J/ is a
flat nearly Kähler manifold if J 2 D Id and a flat nearly para-Kähler manifold if
J 2 D Id. This follows from our classification results (Theorems 3.6.4 and 3.6.17).
Moreover it is shown that any complete simply connected flat nearly (para-)Kähler
manifold is of this form. To sum up, we have shown that any simply connected
complete flat nearly (para-)Kähler manifold is a Lie group L./ with a left-invariant
nearly (para-)Kähler structure and bi-invariant metric. Conversely, it follows from
unpublished work of Cortés and Nagy, that a Lie group with a left-invariant nearly
(para-)Kähler structure and bi-invariant metric is necessarily flat and is therefore
covered by one of our groups L./. The proof of this statement uses the unicity
of the connection with totally skew-symmetric torsion preserving the nearly (para-)
Kähler structure and the Jacobi identity. Further, suppose that   L./ is a lattice.
Then the almost (para-)complex structure J on the group L./ induces an almost
(para-)complex structure J on the compact manifold M D M.; / D  n L./.
Therefore .M; h; J/ is a compact nearly (para-)Kähler manifold. However, the
(para-)complex structure is not L./-invariant, unless  D 0. Moreover, for  ¤ 0,
.M; h; J/ is an inhomogeneous nearly (para-)Kähler manifold, that is, it does not
admit any transitive group of automorphisms of the nearly (para-)Kähler structure.
Since J is not right-invariant, this follows from the fact that Isom0 .M/ is obtained
from the action of L./ by right-multiplication on M, see Corollary 3.5.6. The first
such non-trivial flat compact nearly para-Kähler nilmanifold M./ D  n L./ is
six-dimensional and is obtained from a non-zero element  2 ƒ3 V C Š R, where
V C  V D R3;3 is the C1-eigenspace of J.
The next natural question is the existence of non-flat Ricci-flat examples of
strict (i.e. non-Kähler) nearly para-Kähler six-manifolds. Let us shortly describe
the construction we give in Sect. 3.7: Starting with a Ricci-flat nearly para-Kähler
manifold .M; ; g/; we make an ansatz P depending on  and a vector field :
This P is a second almost para-complex structure if the Nijenhuis tensor of  has
isotropic support (see Definition 2.3.1) and is para-Kähler if  is a conical vector
field in the sense of Definition 3.7.1. Namely, we prove in Theorem 3.7.3 that the
data .M; P; g; / is a conical Ricci-flat para-Kähler manifold and give the inverse of
8 1 Introduction

this construction in Theorem 3.7.5. Space-like regular conical Ricci-flat para-Kähler

manifolds induce the structure of a para-Sasaki Einstein manifold on the level set
fg.; / D c2 ¤ 0g; cf. Proposition 3.7.8. As an application of our correspondence
we obtain the following result:
Theorem 1.2.3 Let .N 5 ; g; T/ be a para-Sasaki Einstein manifold of dimension
5 and denote by .M 6 D b N;b g; P; / the associated conical Ricci-flat para-Kähler
manifold on the cone M D b N over N: Then the cone M can be endowed with the
structure of a conical Ricci-flat strict nearly para-Kähler six-manifold .M; ;b g; /:
Moreover, M is flat if and only if N has constant curvature.
Further we like to mention that starting this construction with five-dimensional
para-Sasaki Einstein manifolds of non-constant curvature yields Ricci-flat nearly
para-Kähler six-manifolds, which are not flat and hence were not known in
the literature before. Different classes of five-dimensional para-Sasaki Einstein
manifolds are given in Sect. 2.6 of Chap. 2. The examples obtained from the present
method cannot be compact. Compact examples are only known by the above
described results for the flat case. To our knowledge the existence of compact (non-
flat) Ricci-flat examples remains still open.
A well-known characterisation [103] of a nearly Kähler structure on a six-
manifold is the one as an SU.3/-structure .!; C ;  / satisfying the exterior
system
C
d! D 3 ; (1.7)

d D ! ^! (1.8)

for a real constant which depends on sign and normalisation conventions.

As a consequence, nearly Kähler manifolds are for example a special class of
half-flat structures.
There exists a left-invariant nearly Kähler structure on S3  S3 which arises from
a classical construction of 3-symmetric spaces by Ledger and Obata [90] and this
exterior system is the main tool in the proof, that this nearly Kähler structure is
the only one on S3  S3 up to homothety ([25], compare also the extended English
version [26]). In fact, the proof of this uniqueness result has been the most difficult
step in the classification of homogeneous nearly Kähler structures in dimension 6
[25, 26].
This Riemannian remark is the starting point of the following development. The
construction of a 3-symmetric space from G D SL.2; R/ instead of SU.2/ defines a
left-invariant nearly pseudo-Kähler structure on SL.2; R/  SL.2; R/. We shortly
recall this construction explicitly. The group G  G  G admits a symmetry of
order three given by .g1 ; g2 ; g3 / 7! .g2 ; g3 ; g1 / which stabilises the diagonal
:
The tangent space of M 6 D G  G  G=
is identified with

p D f.X; Y; Z/ 2 g ˚ g ˚ g j X C Y C Z D 0g:
1.2 Nearly Kähler Geometry 9

Denote by Kg the Killing form of g and define an invariant scalar product on g˚g˚g
by g D Kg ˚ Kg ˚ Kg : This yields a naturally reductive metric on M 6 : Using
Proposition 5.6 of [66] this metric is nearly pseudo-Kähler. For completeness sake
we recall that the complex structure is given by

2 1
J.X; Y; Z/ D p .Z; X; Y/ C p .X; Y; Z/:
3 3

Considering Butruille’s results, it is natural to ask how many left-invariant nearly

pseudo-Kähler structures there are on SL.2; R/  SL.2; R/. Comparing with the
results mentioned in the last paragraph, the answer seems a priori hard to guess. The
result of Sect. 3.9.2 of Chap. 3 is the proof that there is a unique left-invariant nearly
pseudo-Kähler structure on all Lie groups with Lie algebra sl.2; R/ ˚ sl.2; R/. A
byproduct of the proof is the result that there are no nearly para-Kähler structures
on these Lie groups. We add the remark that there exist co-compact lattices for
these Lie groups. Indeed, the article [102] contains a complete list of the compact
quotients of Lie groups with Lie algebras sl.2; R/, which also give rise to compact
quotients on a direct product of such groups.
Another important application of the exterior system is the evolution of hypo-
structures to Calabi-Yau metrics by the Conti-Salamon evolution equations [36]
and that of nearly hypo-structures to nearly Kähler structures [53]. We extend these
constructions for all possible signatures in Sect. 3.8 of Chap. 3.
Let us now shortly sum up some of our structure results for pseudo-Riemannian
signature: A general condition to split off the Kähler factor from a nearly pseudo-
Kähler manifold is obtained in Theorem 3.2.1. A strict nearly pseudo-Kähler
six-manifold M 6 is shown to be Einstein in Theorem 3.2.8. An application of Theo-
rem 3.2.1 implies after using some linear algebra of three-forms the splitting result
(Theorem 3.2.4) for nice nearly pseudo-Kähler manifolds (cf. Definition 3.2.3) in
dimension 8. The argument also holds true for a Riemannian metric and gives an
alternative proof for the known result in positive signature.
If a nice nearly pseudo-Kähler ten-manifold is in addition decomposable (cf.
Definitions 3.2.3 and 3.2.6), we find two cases:
(1) In the first we can split off the Kähler factor and
(2) in the second the holonomy of rN is reducible with a complex one dimensional
factor.
The situation in (2) is one motivation to study twistor spaces: An interesting
class of nearly Kähler manifolds M 4nC2 can be constructed from twistor spaces over
positive quaternionic Kähler manifolds. This class is characterised [16, 98] by the
reducibility of the holonomy of the canonical connection rN to U.n/U.1/. We show
in Sect. 3.3 of Chap. 3 that the twistor spaces over negative quaternionic Kähler
manifolds and para-quaternionic Kähler manifolds carry a nearly pseudo-Kähler
structure and characterise the class of such examples by a holonomic condition
in Sect. 3.4 of Chap. 3. In the pseudo-Riemannian setting twistor spaces are a
10 1 Introduction

good source of examples, since quaternionic geometry is richer in negative scalar

curvature than in positive (cf. Remark 3.3.7) and since we have the additional class
of twistor spaces over para-quaternionic manifolds. In the last section we prove
that a nearly pseudo-Kähler manifold M of twistorial type (cf. Definition 3.4.11)
is obtained from the above mentioned construction on a twistor space. This is
done as follows: We prove that M comes from a pseudo-Riemannian submersion
 W M ! N: Then we use the nearly Kähler data on M to endow N with the
structure of a (para-)quaternionic manifold. The proof is finished by identifying the
twistor space of N with M: The former proofs [16, 98] in the Riemannian case all use
the inverse twistor construction of Penrose or LeBrun, which does not seem to be
developed for the situations occurring in this text. As the reader might observes, the
approach presented here holds also true for Riemannian metrics. As an application
of our results we get the following information:
Theorem 1.2.4 Let .M 10 ; J; g/ be a nice decomposable nearly pseudo-Kähler
manifold, then the universal cover of M is either the product of a pseudo-Kähler
surface and a (strict) nearly pseudo-Kähler manifold M 6 or a twistor space of an
eight-dimensional (para-)quaternionic Kähler manifold endowed with its canonical
nearly pseudo-Kähler structure.
Since the fundamental two-form of a nearly Kähler manifold is not closed,
Lagrangian or maximal totally real submanifolds are more rigid. As we shall
see in the Sect. 3.10.2 these submanifolds are minimal, or more precisely have
vanishing mean curvature in dimension 6. Moreover, we prove in Theorem 3.10.10
of Sect. 3.10.3, that such submanifolds split according to the decomposition of
the ambient space into its Kähler and strict nearly Kähler factor and apply this
information to obtain more precise structure results in dimension 8 and 10. In the
third subsection we show that for the class of nearly Kähler spaces coming from the
twistor construction Lagrangian submanifolds are also minimal. The last subsection
studies deformations of Lagrangian submanifolds.

1.3 Special Kähler Manifolds

As mentioned above special Kähler manifolds provide a class of tt -bundles

which is closely related to nearly Kähler geometry. Homogeneous projective
special pseudo-Kähler manifolds of semisimple groups with compact stabiliser were
classified in [4]. It follows that there is a unique homogeneous projective special
pseudo-Kähler manifold with compact stabiliser which admits a transitive action of
a real form of SL.3; C/ by automorphisms of the special Kähler structure, namely

SU.3; 3/
:
S.U.3/  U.3//
1.4 Summary 11

Its special Kähler metric is (negative) definite. The above manifold occurred in [4]
as an open orbit of SU.3; 3/ on the projectivised highest weight vector orbit of
SL.6; C/ on ƒ3 .C6 / . The space of stable three-forms  2 ƒ3 .R6 / , such that
J2 D 1, has also the structure of a special pseudo-Kähler manifold [76]. The
underlying projective special pseudo-Kähler manifold is the manifold

SL.6; R/
U.1/  SL.3; C/

which has noncompact stabiliser and indefinite special Kähler metric. Both man-
ifolds can be obtained from the space of stable three-forms  2 ƒ3 .C6 / by
imposing two different reality conditions. In the last section of this text we determine
all homogeneous spaces which can be obtained in this way and describe their
special geometric structures. In particular, we calculate the signature of the special
Kähler metrics. For the projective special pseudo-Kähler manifold SL.6; R/= .U.1/
SL.3; C//, for instance, we obtain the signature .6; 12/. Apart from the two above
examples, we find two additional special pseudo-Kähler manifolds and also a special
para-Kähler manifold. The latter is associated to the space of stable three-forms
 2 ƒ3 .R6 / , such that J2 D C1.

1.4 Summary

Let us shortly present the structure of this text. Chapter 2 collects some basics and
starts with an introduction to the formalism of stable forms in which we thoroughly
discuss its application to the different real forms of SL.3; C/; GC 2 and Spin.7; C/:
The second section focuses on the basic notions of (para-)Hermitian geometry and
U " . p; q/-structures and the third prepares some linear algebra of three forms, which
is needed in the third chapter to construct the above cited examples of Ricci-flat
nearly (para-)Kähler manifolds and to obtain the splitting results. The next section
of this chapter recalls some geometry of pseudo-Riemannian submersions. In the
last section of the first chapter we recall some notions on para-Sasaki geometries
and discuss the construction method given by T-duality.
In the third chapter we consider nearly para-Kähler manifolds. Starting with
the definition of a nearly "-Kähler manifold (Definition 3.1.1) we generalise its
characterisation in terms of d! and the Nijenhuis tensor (Proposition 3.1.2). Next
we introduce the characteristic connection rN of an almost "-Hermitian manifold
.M; J; g; !/ (Theorem 3.1.5). We point out the fact, that this connection has parallel
torsion (Proposition 3.1.7) and face for the first time the observation that one cannot
apply the stable form technique in the Ricci-flat setting (cf. Proposition 3.1.10),
since the support of the Nijenhuis tensor is isotropic. The first section ends
deducing some helpful consequences for the setting of isotropic Nijenhuis tensor.
The second subsection is devoted to the proof of the exterior differential system
characterising those SU" . p; q/-structures .!; C ;  / which correspond to nearly
12 1 Introduction

Kähler manifolds (Theorem 3.1.19). On the way to this result information on

more general exterior systems is derived. The last subsection collects a number of
curvature identities for nearly "-Kähler manifolds. In the second section we treat
structure results. More precisely, we split off the Kähler factor (Theorem 3.2.1), give
the relation between the Einstein condition and reducible holonomy (Theorem 3.2.5)
for the characteristic connection and apply these to the observations in dimension
8 and 10 (Theorems 3.2.4 and 3.2.10). One of the classes motivates to consider
the twistorial construction of nearly "-Kähler spaces, which we discuss in detail
in the third section of this chapter. Since for these examples the characteristic
connection has complex reducible holonomy, it is a natural question to characterise
the class of nearly "-Kähler spaces having this property in Sect. 3.4. After deriving
some information in the first two subsections, we focus on the six-dimensional
case, which illustrates the necessary geometric assumption, called of twistorial
type, for the general case of the fourth subsection. The last subsection combines
the information to prove the inverse twistor construction in Theorem 3.4.18. The
next section is an excursion to our classification of flat pseudo-Riemannian Lie
groups (Theorem 3.5.1). Combined with our classification results for flat nearly "-
Kähler manifolds of the sixth subsection this yields compact such examples. These
results impose the question to find (non-flat, non-Kähler) Ricci-flat examples. The
positive answer is a correspondence between conical Ricci-flat para-Kähler and
conical Ricci-flat strict nearly para-Kähler manifolds in Theorems 3.7.3 and 3.7.5
of the present chapter. This construction has an interpretation as a cone construction
over para-Sasaki Einstein manifolds (Theorem 3.7.9). Remaining in the area of co-
dimension 1 constructions we generalise as another application of the above cited
exterior system the evolution of hypo and nearly hypo structures to the pseudo-
Riemannian setup in Sect. 3.8. As further application of the exterior system we
observe in Sect. 3.9.1, that the automorphisms of certain classes of SU" . p; q/-
structures .!; C /—including nearly Kähler structures—are completely described
by demanding, that these fix only one of the two stable forms. In Sect. 3.9.2, we
finally obtain the aforementioned structure results on SL.2; R/  SL.2; R/ and
we also extend the results on S3  S3 by proving the non-existence of nearly
(para-)Kähler structures of indefinite signature. The following two subsections
consider real reducible holonomy of the characteristic connection and the image
of 3-symmetric spaces under T-duality. The subject of the last section of the third
chapter is Lagrangian submanifolds. In the first subsection we derive some geo-
metric information on Lagrangian submanifolds of nearly Kähler spaces, which
are used in the second section to prove, that in dimension 6 such manifolds are
minimal, i.e. have vanishing mean curvature. In the subsequent section we show,
that a Lagrangian submanifold splits according to the decomposition of the ambient
nearly Kähler manifolds into its Kähler and strict nearly Kähler factor. The
fourth subsection considers ambient nearly Kähler spaces coming from the twistor
construction and the last subsection is devoted to deformations of Lagrangian
submanifolds of nearly Kähler spaces.
The fourth chapter studies the above cited extension of Hitchin’s flow equations.
In the first subsection we give the lift from half-flat structures to metrics with
1.5 Zusammenfassung 13

./
holonomy G2 (Theorem 4.1.3), discuss completeness of the corresponding met-
./
rics and extend co-calibrated G2 to parallel Spin.7/- and Spin0 .3; 4/-structures.
The second subsection then evolves nearly Kähler structures to (nearly) parallel
./ ./
G2 -structures. The sine-cones over these G2 -structures then yield metrics of
holonomy Spin.7/- and Spin0 .3; 4/: In Sect. 1.3 we give explicit examples on
nilmanifolds of the type n.H3  H3 /: In the first subsection we reduce the
evolution of !O to some 4-dimensional subspace and classify left-invariant half-flat
structures. One class is rather rigid, i.e. either flat or a product of a para-hyper-
Kähler symmetric space N 4 and a flat factor (cf. Proposition 4.3.7). After these
preparations we can solve the evolution equations for all the classes of invariant half-
flat structures and find examples of full holonomy G2 and G2 (Theorem 4.3.14). The
last subsection collects the above announced results on special Kähler geometry.

1.5 Zusammenfassung

Das zweite Kapitel der Arbeit stellt einige später benötigte Grundlagen zusammen.
Es beginnt mit einer Einführung in den Formalismus der stabilen Formen, wobei
wir im Detail dessen Anwendung im Falle der verschiedenen reellen Formen der
Lieschen Gruppen SL.3; C/; GC 2 und Spin.7; C/ ausarbeiten. Im zweiten Abschnitt
besprechen wir (para-)hermitesche Geometrien und U " . p; q/-Strukturen, während
der dritte Abschnitt etwas lineare Algebra von Differentialformen dritter Stufe
vorbereitet, welche im folgenden Verlauf verwandt wird, um die oben erwähnten
Zerlegungssätze zu erhalten und Ricci-flachen nearly para-Kähler-Beispiele zu
konstruieren. Bevor der letzte Abschnitt an einige Definitionen und Grundlagen zu
para-Sasaki-Mannigfaltigkeiten erinnert und die Konstruktion solcher Geometrien
mittels T-Dualität erläutert, wiederholt der nächste Abschnitt kurz etwas Geometrie
von pseudo-Riemannschen Submersionen.
Das dritte Kapitel befaßt sich mit nearly (para-)Kähler-Mannigfaltigkeiten.
Beginnend mit der Definition einer nearly "-Kähler-Mannigfaltigkeit (Def-
inition 3.1.1) verallgemeinern wir die Charakterisierung dieser Klasse von
fast-"-hermiteschen Geometrien in Termen des Nijenhuis-Tensors und von d!;
wobei ! die fundamentale 2-Form der unterliegenden fast-hermiteschen Struktur ist
(Proposition 3.1.2). Im Anschluß führen wir den charakteristischen Zusammenhang
rN einer fast-"-hermiteschen Mannigfaltigkeit ein (Theorem 3.1.5) und zeigen, dass
dieser parallele Torsion hat (vgl. Propostion 3.1.10). An dieser Stelle begegnen
wir zum ersten Mal der Tatsache, dass der Formalismus stabiler Formen nicht
in der Ricci-flachen Situation anwendbar ist (cf. Propostion 3.1.10), weil der
Support des Nijenhuis-Tensors hier isotrop ist. Wir beenden daher den ersten
Abschnitt mit der Herleitung nützlicher Informationen für Geometrien mit
isotropem Nijenhuis-Tensor. Im zweiten Abschnitt liefern wir nun den Beweis
des äußeren Systems, welches die zu nearly Kähler-Mannigfaltigkeiten gehörigen
SU" . p; q/-Strukturen .!; C ;  / charakterisiert (Theorem 3.1.19). Auf dem Weg
14 1 Introduction

zu diesem Resultat leiten wir Informationen zu allgemeineren äußeren Systemen

her, bevor wir im letzten Unterabschnitt verschiedene Kümmungsidentitäten für
nearly "-Kähler-Mannigfaltigkeiten sammeln. Im zweiten Abschnitt behandeln wir
die oben beschriebenen Strukturresultate. Zuerst splitten wir den Kähler-Faktor
ab (Theorem 3.2.1), stellen einen Bezug zwischen der Einstein-Bedingung und
reduzibler Holonomie des charakteristischen Zusammenhangs her (Theorem 3.2.5)
und wenden dies in den Dimensionen acht und zehn an (Theorem 3.2.4 und
Theorem 3.2.10). Eine der dabei erhaltenen Klassen motiviert das Studium
der Twistor-Konstruktion von nearly "-Kähler-Mannigfaltigkeiten, welche wir
im Detail im dritten Abschnitt dieses Kapitels besprechen. Da die Holonomie
des charakteristischen Zusammenhanges für diese Beispiele komplex reduzibel
ist, untersuchen wir die Klasse von nearly "-Kähler-Mannigfaltigkeiten mit
dieser zusätzlichen Eigenschaft im vierten Abschnitt. Nachdem wir einige
Information in den ersten zwei Teilabschnitten hergeleitet haben, konzentrieren
wir uns auf den 6-dimensionalen Fall, da dieser die notwendige geometrische
Annahme, welche wir “ twistorial type” nennen, für den allgemeinen im vierten
Teilabschnitt behandelten Fall illustriert. Im letzten Teilabschnitt kombinieren
wir die verschiedenen Informationen, um die inverse Twistor-Konstruktion in
Theorem 3.4.18 zu erhalten. Der folgende Abschnitt ist eine Exkursion in unsere
Klassifikation flacher pseudo-Riemannscher Liescher Gruppen (Theorem 3.5.1).
Im Zusammenspiel mit unseren Klassifikationsergebnissen zu flachen nearly "-
Kähler-Mannigfaltigkeiten im sechsten Abschnitt liefert dies im flachen Falle
kompakte Beispiele. Letztere werfen wiederum die Frage nach (nicht flachen, nicht
Kählerschen) Ricci-flachen Beispielen auf, die wir mittels einer in den Theore-
men 3.7.3 und 3.7.5 gegebenen Korrespondenz zwischen konischen Ricci-flachen
para-Kähler-Mannigfaltigkeiten und konischen Ricci-flachen strikten nearly para-
Kähler-Mannigfaltigkeiten beantworten. Eine Interpretation dieser Korrespondenz
als eine Kegelkonstruktion über para-Sasaki-Einstein-Mannigfaltigkeiten wird in
Theorem 3.7.9 bewiesen. Im Kontext von Konstruktionen in einer Kodimension
verbleibend verallgemeinern wir im Abschnitt 3.8 als eine Anwendung des obigen
äußeren Systems die Evolution von hypo- und nearly hypo-Strukturen in das
semi-Riemannsche Setup. Als eine weitere Anwendung des äußeren Systems
beobachten wir im Abschnitt 3.9.1, dass Automorphismen gewisser SU" . p; q/-
Strukturen bereits vollständig dadurch bestimmt sind, dass diese eine der beiden
stabilen Formen fixieren. Im Abschnitt 3.9.2 erhalten wir die Strukturresultate für
SL.2; R/  SL.2; R/ und erweitern die Ergebnisse für S3  S3 ; indem wir die nicht-
Existenz von (para-)Kähler-Strukturen in indefiniter Signatur zeigen. Die darauf
folgenden Unterabschnitte beschäftigen sich mit reell reduzibler Holonomie des
charakteristischen Zusammenhanges und dem Bild 3-symmetrischer Räume unter
T-Dualität. Im letzten Abschnitt des dritten Kapitels präsentieren wir Resultate zu
Lagrange-Untermannigfaltigkeiten in nearly Kähler-Mannigfaltigkeiten. Nachdem
wir im ersten Unterabschnitt einige geometrische Informationen zu Lagrange-
Untermannigfaltigkeiten in nearly Kähler-Mannigfaltigkeiten abgeleitet haben,
zeigen wir im zweiten, dass solche Untermannigfaltigkeiten verschwindende
mittlere Krümmung besitzen. Im darauf folgenden Unterabschnitt zeigen wir,
1.5 Zusammenfassung 15

dass eine Lagrange-Untermannigfaltigkeit in einer nearly Kähler-Mannigfaltigkeit

entsprechend der Zerlegung der nearly Kähler-Mannigfaltigkeit als ein Produkt aus
einem Kähler- und einem striktem nearly Kähler-Faktor in ein Produkt zerfällt.
Der vierte Unterabschnitt untersucht den Fall, dass der umgebende Raum von
der Twistor-Konstruktion herkommt und der letzte Unterabschnitt behandelt
Deformationen von Lagrange-Untermannigfaltigkeiten in nearly Kähler-Räumen.
Das vierte Kapitel widmet sich der oben beschriebenen Erweiterung des
Hitchin-Flusses. Im ersten Abschnitt geben wir den Lift halbflacher Strukturen
./
zu pseudo-Riemannschen Metriken mit Holonomie G2 ; besprechen die
Vollständigkeit der zugehörigen pseudo-Riemannschen Metriken und setzen
./
kokalibrierte G2 -Strukturen zu parallelen Spin.7/- und Spin0 .3; 4/-Strukturen fort.
./
Danach lassen wir nearly Kähler-Metriken zu (nearly) parallelen G2 -Strukturen
./
fließen und erhalten auf dem Sinuskegel über diesen Mannigfaltigkeiten mit G2 -
Struktur Metriken mit Holonomy Spin.7/- und Spin0 .3; 4/: Im dritten Abschnitt
geben wir explizite Beispiele auf Nilmannigfaltigkeiten vom Typ n.H3  H3 /:
Hierfür reduzieren wir die Evolution von !O auf einen 4-dimensionalen Unterraum
und klassifizieren linksinvariante halbflache Strukturen. Eine der erhaltenen Klassen
ist sehr rigide, d.h. entweder flach oder ein Produkt eines para-Hyperkähler
symmetrischen Raumes N 4 und eines flachen Faktors (siehe Proposition 4.3.7).
Nach diesen Vorbereitungen können wir die Evolutionsgleichungen für alle Klassen
invarianter halbflacher Strukturen lösen und Beispiele voller Holonomie G2 und
G2 finden (Theorem 4.3.14). Der letzte Abschnitt sammelt die angekündigten
Ergebnisse zu speziellen Kähler-Mannigfaltigkeiten.
Chapter 2
Preliminaries

2.1 Stable Forms

In this section we collect some basic facts about stable forms, their orbits and their
stabilisers.
Proposition 2.1.1 Let V be an n-dimensional real or complex vector
  space. The
general linear group GL.V/ has an open orbit in ƒk V  , 0  k  n2 , if and only if
k  2 or if k D 3 and n D 6; 7 or 8.
Proof The representation of GL.V/ on ƒk V  is irreducible. In the complex case
the result thus follows, for instance, from the classification of irreducible complex
prehomogeneous vector spaces, [85]. The result in the real case follows from the
complex case, since the complexification of the GL.n; R/-module ƒk Rn is an
irreducible GL.n; C/-module. t
u
Remark 2.1.2 An open orbit is unique in the complex case, since an orbit which is
open in the usual topology is also Zariski-open and Zariski-dense [84, Prop. 2.2].
Over the reals, the number of open orbits is finite by a well-known theorem of
Whitney.
Definition 2.1.3 A k-form  2 ƒk V  is called stable if its orbit under GL.V/ is
open.
Proposition 2.1.4 Let V be an oriented real vector space and assume that k 2
f2; n  2g and n even, or k 2 f3; n  3g and n D 6; 7 or 8. There is a GL.V/C -
equivariant mapping

 W ƒk V  ! ƒn V  ;

homogeneous of degree nk , which assigns a volume form to a stable k-form and

which vanishes on non-stable forms. Given a stable k-form , the derivative of  in

© Springer International Publishing AG 2017 17

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0_2
18 2 Preliminaries

 defines a dual .n  k/-form O 2 ƒnk V  by the property

d .˛/ D O ^ ˛ for all ˛ 2 ƒk V  : (2.1)

The dual form O is also stable and satisfies1

.StabGL.V/ .//0 D .StabGL.V/ .//

O 0:

A stable form, its volume form and its dual are related by the formula
n
O ^  D ./: (2.2)
k

Proof This result can be viewed as a consequence of the theory of prehomogeneous

vector spaces, [84], as follows. Replacing V and GL.V/ by the complexifications
VC and GL.VC /, the situations we are considering correspond to examples 2.3, 2.5,
2.6 and 2.7 of §2, [84]. In all cases, the complement of the open orbit under GL.VC /
is a hypersurface in ƒk .VC / defined by a complex irreducible non-degenerate
homogeneous polynomial f which is invariant under GL.VC / up to a non-trivial
character W GL.VC / ! C .
Due to Proposition 4.1, [84], the polynomial f restricted to ƒk V  is real-valued
and the character restricts to W GL.V/ ! R . Moreover, by Proposition 4.5,
[84], the complement of the zero set of f in ƒk V  has a finite number of connected
components which are open GL.V/-orbits. Since the only characters of GL.V/
are the powers of the determinant, there is an equivariant mapping from ƒk V  to
.ƒn V  /˝s for some positive integer s. Taking the s-th root, which depends on the
choice of an orientation if s is even, we obtain the GLC .V/-equivariant map . By
construction, a k-form  is stable if and only if ./ ¤ 0. The equivariance under
scalar matrices implies that the map  is homogeneous of degree nk .
The derivative
D
ƒk V  ! .ƒk V  / ˝ ƒn V  ! ƒnk V  ;  7! d  7! O

inherits the GLC .V/-equivariance from  and is an immersion since f is non-

degenerate, compare Theorem 2.16, [84]. Therefore, it maps stable forms to stable
forms such that the connected components of the stabilisers are identical. Finally,
formula (2.2) is just Euler’s formula for the homogeneous mapping . t
u
In the following, we discuss stable forms, their volume forms and their duals in
the cases which are relevant in this text. In each case, V is a real n-dimensional
vector space.
k D 2; n D 2m: The orbit of a non-degenerate two-form is open and there is only
one open orbit in ƒ2 V  . Thus, the stabiliser of a stable two-form ! is isomorphic to
Sp.2m; R/. The polynomial invariant is the Pfaffian determinant. We normalise the

1
Here G0 is the identity component of a Lie group G:
2.1 Stable Forms 19

associated equivariant volume form such that it corresponds to the Liouville volume
form
1 m
.!/ D ! :
mŠ

Differentiation of the homogeneous polynomial map ! 7! .!/ yields

1
!O D ! m1 :
.m  1/Š

k D .n  2/; n D 2m: As ƒn2 V  Š ƒ2 V ˝ ƒn V  , there is again only one open

orbit. More precisely, an .n  2/-form  is stable if and only if there is a stable
two-form ! with  D !O since the mapping ! 7! !O is an equivariant immersion.
If m is even, such an ! is unique and we define the volume form ./ D .!/.
If m is odd, we need an orientation on V to uniquely define an associated volume
form. We choose the (m-1)-th root ! with positively oriented ! m and define again
./ D .!/. In both cases, we find

1
O D ! (2.3)
m1

with the help of (2.2). The stabiliser of a stable .n  2/-form in GLC .V/ is again the
real symplectic group.
k D 3; n D 6: Let V be an oriented six-dimensional vector space and let  denote
the canonical isomorphism

 W ƒk V  Š ƒ6k V ˝ ƒ6 V  :

Given any three-form , we define K W V ! V ˝ ƒ6 V  by

K .v/ D ..vy / ^ /

and the quartic invariant

1
./ D tr .K2 / 2 .ƒ6 V  /˝2 : (2.4)
6

Recall that, for any one-dimensional vector space L, an element u 2 L˝2 is defined
to be positive, u > 0, if u D s ˝ s for some s 2 L and negative if u > 0. Therefore,
the norm of an element u 2 L˝2 is well-defined and we set
p
./ D j./j (2.5)
20 2 Preliminaries

for the positively oriented square root. If ./ ¤ 0, we furthermore define

1
J D K : (2.6)
./

Proposition 2.1.5 A three-form  on an oriented six-dimensional vector space V

with volume form is stable if and only ./ ¤ 0. There are two open orbits.
One orbit consists of all three-forms  satisfying one of the following equivalent
properties.
(a) ./ > 0:
(b) There are two uniquely defined real decomposable three-forms ˛ and ˇ such
that  D ˛ C ˇ and ˛ ^ ˇ > 0.
(c) The stabiliser of  in GLC .V/ is SL.3; R/  SL.3; R/.
(d) It holds ./ ¤ 0 and the endomorphism J is a para-complex structure on V,
i.e. J 2 D idV and the eigenspaces for the eigenvalues ˙1 are three-dimensional
(see also Sect. 2.2 of the present chapter).
(e) There is a basis fe1 ; ldots; e6 g of V such that D e123456 > 0 and

 D e123 C e456

where eijk is the standard abbreviation for ei ^ ej ^ ek . In this basis, it holds

./ D ˝2 , J ei D ei for i 2 f1; 2; 3g and J ei D ei for i 2 f4; 5; 6g.
The other orbit consists of all three-forms  satisfying one of the following
equivalent properties.
(a) ./ < 0:
(b) There is a unique complex decomposable three-form ˛ such that  D ˛ C ˛N and
i.˛N ^ ˛/ > 0.
(c) The stabiliser of  in GLC .V/ is SL.3; C/.
(d) It holds ./ ¤ 0 and the endomorphism J is a complex structure on V.
(e) There is a basis fe1 ; ldots; e6 g of V such that D e123456 > 0 and

 D e135  e146  e236  e245 :

In this basis, it holds ./ D 4 ˝2 , J ei D eiC1 and J eiC1 D ei for i 2

f1; 3; 5g.
Proof All properties are proved in Section 2 of [75]. The only fact we added is the
observation that J is a para-complex structure if ./ > 0 which is obvious in the
standard basis. t
u
It is also possible to introduce a basis describing both orbits simultaneously.
Indeed, given a generic stable three-form and an orientation, there is a basis
fe1 ; : : : ; e6 g of V and an " 2 f˙1g such that D e123456 > 0 and

" D e135 C ".e146 C e236 C e245 / (2.7)

2.1 Stable Forms 21

with ./ D 4" ˝2 . Furthermore, it holds J ei D "eiC1 , J eiC1 D ei for i 2 f1; 3; 5g

and

J" " D e246 C ".e235 C e145 C e136 /: (2.8)

Analogies between complex and para-complex structures are elaborated in a unified

language in Sect. 2.2 of this chapter. In this language, a stable three-form always
induces an "-complex structure J since J2 D "id for the normal form " .
Lemma 2.1.6 The dual of a stable three-form  2 ƒ3 V  on an oriented six-
dimensional vector space V is

O D J : (2.9)

Proof We already observed that the connected components of the stabilisers of 

and O have to be identical. Therefore, since the space of real three-forms invariant
under SL.3; C/ respectively SL.3; R/  SL.3; R/ is two-dimensional, we can make
the ansatz

O D c1  C c2 J 

with real constants c1 and c2 . Computing

6 (2.2) (2.7/;.2.8)
./ D O ^  D c2 J  ^  D 2c2 ./;
3
we find c2 D 1. By

d . J / D O ^ J  D c1  ^ J  D 2c1 ./;

the constant c1 vanishes if the derivative of  (recall (2.5)) in  in direction of J 

vanishes. However, using the normal form (2.7) again, we compute . C tJ / D
4"." C t2 /2 .e123456 /˝2 and the assertion follows. t
u
A convenient way to compute the dual of  without determining J is given by
the following corollary. In fact, the corollary explicitly shows the equivalence of the
two different definitions of  7! O given in [76] and [75].
Corollary 2.1.7 If ./ > 0 and  D ˛ C ˇ in terms of decomposables ordered
such that ˛ ^ ˇ > 0, the dual of  satisfies O D ˛  ˇ.
If ./ < 0 and  is the real part of a complex decomposable three-form ˛ such
that i.˛N ^ ˛/ > 0, the dual of  is the imaginary part of ˛. In particular, the complex
three-form ˛ is a .3; 0/-form w.r.t. J .
Proof The assertions are easily proved by comparing the claimed formulas for O
with formula (2.9) in the standard bases given in part (2.1.5) of Proposition 2.1.5.
t
u
Finally, we note that for a fixed orientation, it holds

OO D  and JO D "J : (2.10)

22 2 Preliminaries

k D 3; n D 7: Given any three-form ', we define a symmetric bilinear form with

values in ƒ7 V  by

1
b' .v; w/ D .vy '/ ^ .wy '/ ^ ': (2.11)
6

Since the determinant of a scalar-valued bilinear form is an element of .ƒ7 V  /˝2 ,

we have det b 2 .ƒ7 V  /˝9 . If and only if ' is stable, the seven-form
1
.'/ D .det b' / 9

defines a volume form, independent of an orientation on V, and the scalar-valued

symmetric bilinear form

1
g' D b'
.'/
p
is non-degenerate. Notice that .'/ D det g' is the metric volume form.
It is known [23, 72] that a stable three-form defines a multiplication “” and a
vector cross product “” on V by the formula

'.x; y; z/ D g' .x; y  z/ D g' .x; y  z/; (2.12)

such that .V; / is isomorphic either to the imaginary octonions Im O or to the

Q Thus, there are exactly two open orbits of stable
imaginary split-octonions Im O.
three-forms having isotropy groups
(
G2  SO.7/; if g' is positive definite,
StabGL.V/ .'/ Š (2.13)
G2  SO.3; 4/; if g' is of signature (3,4):

There is always a basis fe1 ; : : : ; e7 g of V such that

7
X
' D e124 C ei .iC1/ .iC3/ (2.14)
iD2

with  2 f˙1g and indices modulo 7. For  D 1, the induced metric g' is positive
definite and the basis is orthonormal such that this basis corresponds to the Cayley
basis of Im O. For  D 1, the metric is of signature (3,4) and the basis is pseudo-
orthonormal with e1 , e2 and e4 being the three spacelike basis vectors.
The only four-forms having the same stabiliser as ' are the multiples of the
Hodge dual g' ', [23, Propositions 2.1, 2.2]. Since the normal form satisfies
g' .'; '/ D 7, we have by definition of the Hodge dual ' ^ g' ' D 7 .'/ and
2.1 Stable Forms 23

therefore
1
'O D g '; (2.15)
3 '
by comparing with (2.2).
Lemma 2.1.8 Let ' be a stable three-form in a seven-dimensional vector space V.
Let ˇ be a one-form or a two-form. Then ˇ ^ ' D 0 if and only if ˇ D 0.
Proof For the compact case, see also [22]. If ˇ is a one-form, the proof is very easy.
If ˇ isPa two-form, we choose a basis such that ' is in the normal form (2.14). Let
ˇ D i<j bi;j eij and compute

ˇ ^ ' D .b2;3  b1;6 / e12356 C .b2;3  b4;7 / e23457 C .b1;6 C b4;7 / e14567
C.b5;7  C b1;2 / e12457 C .b3;6  b5;7 / e34567 C .b1;2  b3;6 / e12346
.b3;7  C b2;4 / e12347 C .b5;6  C b2;4 / e12456 C .b3;7 C b5;6 / e13567
C.b2;5  b4;6 / e23456 C .b4;6  b1;7 / e13467  .b2;5 C b1;7 / e12357
C.b4;5 C b2;6 / e24567  .b1;3 C b2;6 / e12367 C .b4;5 C b1;3 / e13457
C.b3;5 C b6;7 / e23567 C .b1;4  b3;5 / e12345 C .b6;7   b1;4 / e12467
C.b3;4 C b1;5 / e13456 C .b2;7  b1;5 / e12567 C .b3;4  b2;7 / e23467 :

This five-form is written as a linear combination of linearly independent forms and

each line contains exactly three different coefficients of ˇ. Inspecting the coefficient
equations line by line, it is easy to see that all coefficients of ˇ vanish if and only if
ˇ ^ ' D 0. t
u

2.1.1 Real Forms of SL.3; C/

By the following proposition, any real form of SL.3; C/ can be written as a

simultaneous stabiliser of a stable two-form and a stable three-form.
Proposition 2.1.9 Let V be a six-dimensional real vector space. Let ! 2 ƒ2 V  and
 2 ƒ3 V  be stable forms which are compatible in the sense that

! ^  D 0: (2.16)

Then, we have
(
SU. p; q/  SO.2p; 2q/ ; pCq D 3; if ./ < 0;
StabGL.V/ .; !/ Š
SL.3; R/  SO.3; 3/ ; if ./ > 0;
24 2 Preliminaries

where SL.3; R/ is embedded in SO.3; 3/ such that it acts by the standard represen-
tation and its dual, respectively, on the maximally isotropic ˙1-eigenspaces of the
para-complex structure J induced by .
Proof Let V be oriented by .!/ D 16 ! 3 and let J be the unique (para-) complex
structure (2.6) associated to the three-form  and this orientation. By " 2 f˙1g, we
denote the sign of ./, that is J2 D "idV . In the basis in which  is in the normal
form (2.7), it is easy to verify that ! ^  D 0 is equivalent to the skew-symmetry of
J with respect to !. Equivalently, the pseudo-Euclidean metric

g D g.!;/ D " !.; J /; (2.17)

induced by ! and , is compatible with J in the sense that g. J ; J / D "g.; /.

The stabiliser of the set of tensors .!; J ; g; ; J / satisfying this compatibility
condition is well-known to be SU. p; q/ respectively SL.3; R/. t
u
We call a compatible pair of stable forms .!; / 2 ƒ2 V   ƒ3 V  normalised if
2 3
./ D 2 .!/ ” J  ^  D ! : (2.18)
3
Moreover, a stable three-form  is non-degenerate in the sense that for v 2 V one
has

vy  D 0 ) v D 0: (2.19)

Remark 2.1.10 By our conventions, the metric (2.17) induced by a normalised,

compatible pair is either of signature .6; 0/ or .2; 4/ or .3; 3/, where the first number
denotes the number of spacelike directions. We emphasise that our conventions are
such that

! D g.; J /:

This sign choice turned out to be necessary in order to achieve that ./ is indeed a
positive multiple of .!/ in the positive definite case.
Sometimes it is convenient to have a unified adapted basis. For a compatible and
normalised pair .!; /, there is always a pseudo-orthonormal basis fe1 ; : : : ; e6 g of
V with dual basis fe1 ; : : : ; e6 g such that  D " is in the normal form (2.7) and

! D .e12 C e34 / C e56 (2.20)

for ."; / 2 f.1; 1/; .1; 1/; .1; 1/g. The signature of the induced metric with
respect to this basis is
8
ˆ
<.C; C; C; C; C; C/ for " D 1 and  D 1;
ˆ
.; "; ; "; 1; "/ D .; ; ; ; C; C/ for " D 1 and  D 1; (2.21)
ˆ
:̂.C; ; C; ; C; / for " D 1 and  D 1;
2.1 Stable Forms 25

and we have
8
ˆ
<SU.3/  SO.6/
ˆ for " D 1 and  D 1;
StabGL.6;R/ .!; / Š SU.1; 2/  SO.2; 4/ for " D 1 and  D 1;
ˆ
:̂SL.3; R/  SO.3; 3/ for " D 1:

For instance, the following observation is easily verified using the unified basis.
Lemma 2.1.11 Let .!; / be a compatible and normalised pair of stable forms on a
six-dimensional vector space. Then, the volume form .!/ is in fact a metric volume
form w.r.t. to the induced metric g D g.!;/ and the corresponding Hodge dual of !
and  is

g ! D "!O ; g  D O (2.22)

2.1.2 Relation Between Real Forms of SL.3; C/ and GC

2

The relation between stable forms in dimension 6 and 7 corresponding to the

embedding SU.3/  G2 is well-known. We extend this relation by including also
the embeddings SU.1; 2/  G2 and SL.3; R/  G2 as follows.
Proposition 2.1.12 Let V D W ˚ L be a seven-dimensional vector space decom-
posed as a direct sum of a six-dimensional subspace W and a line L. Let ˛ be
a non-trivial one-form in the annihilator W 0 of W and .!; / 2 ƒ2 L0  ƒ3 L0
a compatible and normalised pair of stable forms inducing the scalar product
h D h.!;/ given in (2.17). Then, the three-form ' 2 ƒ3 V  defined by

' D!^˛C (2.23)

is stable and induces the scalar product

g' D h  "˛  ˛; (2.24)

where " denotes the sign of ./ such that J2 D "id. The stabiliser of ' in GL.V/ is
(
G2 for " D 1 and positive definite h,
StabGL.V/ .'/ Š
G2 otherwise.

Proof We choose a basis fe1 ; : : : ; e6 g of L0 such that ! and  are in the generic
normal forms (2.7) and (2.20). With e7 D ˛, we have

' D .e127 C e347 / C e567 C e135 C ".e146 C e236 C e245 /: (2.25)

26 2 Preliminaries

The induced bilinear form (2.11) turns out to be

b' .v; w/ D ."v 1 w1 C v 2 w2  "v 3 w3 C v 4 w4  "v 5 w5 C v 6 w6 C v 7 w7 /e1234567

P P
for v D v i ei and w D wi ei . Hence, the three-form ' is stable for all signs of "
and  and its associated volume form is
1
.'/ D .det b' / 9 D " e1234567 :

The formula (2.24) for the metric g' induced by ' follows, since the basis
fe1 ; : : : ; e7 g of V is pseudo-orthonormal with respect to this metric of signature
8
ˆ
ˆ.C; C; C; C; C; C; C/ for " D 1 and  D 1;
<
.; "; ; "; 1; "; "/ D .; ; ; ; C; C; C/ for " D 1 and  D 1; (2.26)
ˆ
:̂
.C; ; C; ; C; ; / for " D 1 and  D 1:

The assertion on the stabilisers now follows from (2.13). t

u
Lemma 2.1.13 Under the assumptions of the previous proposition, the dual four-
form of the stable three-form ' is

3'O D ' ' D  " .˛ ^ O C !/

O D "˛ ^ h  C h !; (2.27)

where ' denotes the Hodge dual with respect to the metric g' and the orientation
induced by .'/.
Proof In the basis of the previous proof, the Hodge dual of ' is

' ' D ".e3456 C e1256 /  "e1234 C " e2467 C e2357 C e1457 C e1367 :

The second equality follows when comparing this expression with ".e7 ^ O C 12 ! 2 /
in this basis using (2.8) and (2.20). The first and the third equality are just the
formulas (2.15) and (2.22), respectively. t
u
The inverse process is given by the following construction.
Proposition 2.1.14 Let V be a seven-dimensional real vector space and ' 2 ƒ3 V 
a stable three-form which induces the metric g' on V. Moreover, let n 2 V be a
unit vector with g' .n; n/ D " 2 f˙1g and let W D n? denote the orthogonal
complement of Rn. Then, the pair .!; / 2 ƒ2 W   ƒ3 W  defined by

! D .n '/jW ;  D 'jW ; (2.28)

2.1 Stable Forms 27

is a pair of compatible normalised stable forms. The metric h D h.!;/ induced by

this pair on W satisfies h D .g' /jW and the stabiliser is
8
ˆ
ˆ
<SU.3/; if g' is positive definite,
StabGL.W/ .!; / Š SU.1; 2/; if g' is indefinite and " D 1;
ˆ
:̂SL.3; R/; if " D 1:

When .V; '/ is identified with the imaginary octonions, respectively, the imaginary
split-octonions, by (2.12), the "-complex structure induced by  is given by

J v D n  v D n  v for v 2 V: (2.29)

Proof Due to the stability of ', we can always choose a basis fe1 ; : : : ; e7 g of V
with n D e7 such that ' is given by (2.25) where " D g' .n; n/ and  2 f˙1g
depends on the signature of g' . As this basis is pseudo-orthonormal with signature
given by (2.26), the vector n has indeed the right scalar square and fe1 ; : : : ; e6 g is a
pseudo-orthonormal basis of the complement W D n? . Since the pair .!; / defined
by (2.28) is now exactly in the generic normal form given by (2.7) and (2.20), it
is stable, compatible and normalised and the induced endomorphism J is an "-
complex structure. The identity h D .g' /jW for the induced metric h.!;/ follows
from comparing the signatures (2.26) and (2.21) and the assertion for the stabilisers
is an immediate consequence. Finally, the formula for the induced "-complex
structure J is another consequence of g D .g' /jW since we have

(2.12)
g' .x; n  y/ D '.x; n; y/ D !.x; y/ D h.x; J y/

for all x; y 2 W. t
u
Notice that, for a fixed metric h of signature .2; 4/ or .3; 3/, the compatible and
normalised pairs .!; / of stable forms inducing this metric are parametrised by
the homogeneous spaces SO.2; 4/= SU.1; 2/ and SO.3; 3/= SL.3; R/, respectively.
Thus, the mapping .!; / 7! ' defined by formula (2.23) yields isomorphisms

SO.2; 4/ SO.3; 4/ SO.3; 3/ SO.3; 4/

Š ; Š ;
SU.1; 2/ G2 SL.3; R/ G2

since the metric h completely determines the metric g' by the formula (2.24).

2.1.3 Relation Between Real Forms of GC

2
and Spin.7; C/

It is possible to extend this construction to dimension 8 as follows. Starting with

a stable three-form ' on a seven-dimensional vector space V, we can consider the
28 2 Preliminaries

four-form

ˆ D e8 ^ ' C ' ': (2.30)

on the eight-dimensional space V ˚ Re8 . Although the four-form ˆ is not stable, it

is shown in [23] that it induces the metric

gˆ D g' C .e8 /2 (2.31)

on V ˚ Re8 and that its stabiliser is

(
Spin.7/  SO.8/; if g' is positive definite,
StabGL.V˚Re8 / .ˆ/ Š
Spin0 .3; 4/  SO.4; 4/; if g' is indefinite.

The index “0” denotes, as usual, the connected component of the identity. Starting
conversely with a four-form ˆ on V ˚ Re8 such that its stabiliser in GL.V ˚ Re8 /
is isomorphic to Spin.7/ or Spin0 .3; 4/, the process can be reversed by setting ' D
e8 ˆ. As before, the metric induced by ˆ on V˚Re8 is determined by the metric g'
induced by ' on V. Thus, the indefinite analogue of the well-known isomorphisms

SO.6/ SO.7/ SO.8/

RP7 Š Š Š
SU.3/ G2 Spin.7/

is given by

SO.2; 4/ SO.3; 3/ SO.3; 4/ SO.4; 4/

Š Š  Š : (2.32)
SU.1; 2/ SL.3; R/ G2 Spin0 .3; 4/

2.2 Almost Pseudo-Hermitian and Almost Para-Hermitian

Geometry

We recall that an almost para-complex structure on a 2m-dimensional manifold M

is an endomorphism field squaring to the identity such that both eigendistributions
(for the eigenvalues ˙1) are m-dimensional. An almost para-Hermitian structure
consists of a neutral metric and an antiorthogonal almost para-complex structure.
For a survey on para-complex geometry we refer to [9] or [47].
In the following, we introduce a unified language describing almost pseudo-
Hermitian and almost para-Hermitian geometry simultaneously. The philosophy is
to put an “"” in front of all notions which is to be replaced by “para” for " D 1 and
is to be replaced by “pseudo” or to be omitted for " D 1. From now on, we always
suppose " 2 f˙1g. Let us note, that there exist other names in the literature, like
for example double numbers or hypercomplex numbers. To begin with, we consider
2.2 Almost Pseudo-Hermitian and Almost Para-Hermitian Geometry 29

the "-complex numbers C" D fx C i" y ; x; y 2 Rg with i2" D ". For the para-
complex numbers, " D 1, there are obvious analogues of conjugation, real and
imaginary parts and the square of the (not necessarily positive) absolute value given
by jzj2 D zNz:
Moreover, let V be a real vector space of even dimension n D 2m. We call an
endomorphism J " an "-complex structure if J "2 D "idV and if additionally, for
" D 1, the ˙1-eigenspaces V ˙ are m-dimensional. If we consider only one of the
two cases, we write J for " D 1 and P or  for " D 1: An "-Hermitian structure is
an "-complex structure J " together with a pseudo-Euclidean scalar-product g which
is "-Hermitian in the sense that it holds

g.J " ; J " / D "g.; /:

We denote the stabiliser in GL.V/ of an "-Hermitian structure as the "-unitary group

U.J " ; g/ D U" . p; q/ D fL 2 GL.V/ j ŒL; J "  D 0; L g D gg

(
U. p; q/; p C q D m; for " D 1;
Š
GL.m; R/; for " D 1:

Here, the pair .2p; 2q/ is the signature2 of the metric for " D 1. For " D 1, the
group GL.m; R/ acts reducibly such that V D V C ˚ V  and the signature is always
.m; m/. In other words, there exists a basis called isotropic or null-coordinates of V
such that P and g are represented by the matrices
   
0 Id Id 0
gcan D g0 D and Pcan D P0 D (2.33)
Id 0 0 Id

and one computes that

 ˇ 
ˇ A 0
U. P; g/ Š A 2 GL.Rm / ˇ Š GL.Rm /:
ˇ 0 .At /1

Later we consider the case, where the para-Kähler manifold is Ricci-flat, then the
restricted holonomy group lies in the special para-unitary group SU.P; g/ which in
the above model is
 ˇ 
ˇ
m ˇ A 0
SU. P; g/ Š A 2 SL.R / ˇ Š SL.Rm /; (2.34)
0 .At /1

see also Proposition 2.1.9.

2
Please note that in our convention 2p refers to the negative directions.
30 2 Preliminaries

An almost "-Hermitian manifold is a manifold M of dimension n D 2m endowed

with a U" . p; q/-structure or, equivalently, with an almost "-Hermitian structure
which consists of an almost "-complex structure J " and an "-Hermitian metric g.
The non-degenerate two-form ! WD g.; J " / is called fundamental two-form.
Given an almost "-Hermitian structure .g; J " ; !/, there exist pseudo-orthonormal
local
P frames fe1 ; : : : ; e2m g such that Jei D eiCm for i D 1; : : : ; m and ! D
" m iD1 i e i.iCm/
, where i WD g.ei ; ei / for i D 1; : : : ; m. Upper indices always
denote dual (not metric dual) one-forms and eij stands for ei ^ ej . We call such a
frame "-unitary. If m  3, we can always achieve 1 D 2 by reordering the basis
vectors.
For both values of ", the "-complexification TM ˝ C" of the tangent bundle
decomposes into the ˙i" -eigenbundles TM 1;0 and TM 0;1 . This induces the well-
known bi-grading of C" -valued exterior forms

r;s D r;s .M/ D .ƒr;s / D .ƒr .TM 1;0 / ˝ ƒs .TM 0;1 / /:

If X is a vector field on M, we use the notation

1 1
X 1;0 D .X C i" "J " X/ 2 .TM 1;0 / ; X 0;1 D .X  i" "J " X/ 2 .TM 0;1 /;
2 2

for the real isomorphisms from TM to TM 1;0 respectively TM 0;1 . As usual in almost
Hermitian geometry [104], we define the bundles ƒr;s  for r ¤ s and Œƒr;r  by the
property

ƒr;s  ˝ C" D ƒr;s  ˚ i" ƒr;s  D ƒr;s ˚ ƒs;r ;

Œƒr;r  ˝ C" D Œƒr;r  ˚ i" Œƒr;r  D ƒr;r :

The sections of these bundles are denoted as real forms of type .r; s/ C .s; r/
respectively of type .r; r/ and the spaces of sections by r;s  respectively by Œr;r :
For instance, it holds

Œ1;1  D f˛ 2 2 M j ˛.X; Y/ D "˛.J " X; J " Y/g;

such that the fundamental form is of type .1; 1/ and similarly

3;0  D f˛ 2 3 M j ˛.X; Y; Z/ D "˛.X; J " Y; J " Z/g: (2.35)

Only in the para-complex case, " D 1, there is a decomposition of the real tangent
bundle TM D V ˚H into the ˙1-eigenbundles of J " which also induces a bi-grading
of real differential forms. It is also straightforward to show that

ƒ3;0  Š ƒ3 V  ˚ ƒ3 H ; (2.36)
2.2 Almost Pseudo-Hermitian and Almost Para-Hermitian Geometry 31

when considering the characterisation (2.35).

Returning to analogies, we recall that the Nijenhuis tensor of the almost "-
complex structure J " satisfies

N.X; Y/ D "ŒX; Y  ŒJ " X; J " Y C J " ŒJ " X; Y C J " ŒX; J " Y

D .rJ " X J " / Y C .rJ " Y J " / X C J " .rX J " / Y  J " .rY J " / X (2.37)

for real vector fields X; Y; Z and for any torsion-free connection r on M. For
both values of ", it is well-known that the Nijenhuis tensor is the obstruction to
the integrability of the almost "-complex structure. Since integrability for complex
structures is well-known, let us just note, that a para-complex structure on M is
called integrable if the distributions T ˙ M are integrable distributions.
In the following, let r always denote the Levi-Civita connection of the metric g
of an almost "-Hermitian manifold. Differentiating the almost "-complex structure,
its square and the fundamental two-form yields for both values of " the formulas

.rX J " / J " Y D J " .rX J " / Y;

g..rX J " /Y; Z/ D .rX !/. Y; Z/ (2.38)

for all vector fields X; Y; Z. Using these formulas, it is easy to show that for any
almost "-Hermitian manifold, the tensor A defined by

A.X; Y; Z/ D g..rX J " /Y; Z/ D .rX !/. Y; Z/ (2.39)

has the symmetries

A.X; Y; Z/ D A.X; Z; Y/; (2.40)

" "
A.X; Y; Z/ D "A.X; J Y; J Z/ (2.41)

for all vector fields X; Y; Z.

The decomposition of the U" . p; q/-representation space of tensors with the
same symmetries as A into irreducible components leads to a classification of
almost "-Hermitian manifolds which is classical for U.m/ [68] and is called the
Gray-Hervella list. The para-complex case for the group GL.m; R/ is completely
worked out in [59]. In [86], the Gray-Hervella classes are generalised to almost "-
Hermitian structures, which are denoted by generalised almost Hermitian or GAH
structures there. Analogues of all 16 Gray-Hervella classes are established. These
are invariant under the respective group action, but obviously not irreducible for the
para-Hermitian case when compared to the decomposition in [59].
Finally, we mention the useful formula

2.rX !/. Y; Z/ D d!.X; Y; Z/ C "d!.X; J " Y; J " Z/ C "g.N. Y; Z/; J " X/ (2.42)
32 2 Preliminaries

holding true for all vector fields X; Y; Z on any almost "-Hermitian manifold. Proofs
can be found in Chapter IX of [87] for " D 1 and [80] for " D 1.

2.3 Linear Algebra of Three-Forms in Dimension 8 and 10

In this subsection we shall freely identify the real vector space V WD Ck;l
1 D C D
k;l
2k;2l 2n 2n 
R D R and V WD C1 D R D R with its dual V by means of the pseudo-
n n;n

Euclidian scalar product g D gcan and we denote the canonical "-complex structure
"
on V by Jcan :
Definition 2.3.1 The support of  2 ƒ3 V is defined by

† WD spanfX Y j X; Y 2 Vg  V: (2.43)

The name support is motivated by the following Lemma of [41].

Lemma 2.3.2 Let  2 ƒ3 V; then one has  2 ƒ3 † :
Proof We take a complement W  V of L D † : The decomposition
M
ƒ3 V D ƒp L ^ ƒq W
pCqD3

induces a decomposition
X
D Πp;q :
pCqD3

Œ0;3 Œ1;2 Œ0;3

Taking X; Y 2 L0 yields3 L 3 X Y D X Y C X Y: Now since X Y 2 W
Œ1;2
and X Y 2 L, we get Œ0;3 D 0: Further the choice X 2 L0 and Y 2 W 0 yields
Œ1;2 D 0 and then the choice X; Y 2 W 0 yields Œ2;1 D 0: This shows  D Œ3;0 :
t
u
In Chap. 3 of the present text we are interested in the tensor field gp ..rX J/Y; Z/
for p 2 M and X; Y; Z 2 Tp M on a nearly "-Kähler manifold .M; J; g/; which is
shown to be a real form of type .3; 0/ C .0; 3/: The type condition implies that
† is a J " -invariant subspace. For example, it follows that the complex dimension
of the support of a non-zero such form is at least three. In Sect. 3.6 of Chap. 3 the
classification of Levi-Civita flat nearly "-Kähler manifolds is related to the existence
of real three-forms of type .3; 0/ C .0; 3/ with isotropic support, i.e. such that † is
an isotropic subspace. This is our reason for the interest in the subsequent Lemmata
[41, 43].

3
As above 0 denotes the annihilator.
2.3 Linear Algebra of Three-Forms in Dimension 8 and 10 33

Lemma 2.3.3 One has

3;0  D f 2 ƒ3 V j .; Jcan

" "
; Jcan / D ".; ; /g
D f 2 ƒ3 V j fX ; Jcan
"
g D 0; 8X 2 Vg: (2.44)

Later in this text we need to consider three-forms which satisfy the condition

X ı Y D 0; 8X; Y 2 V; (2.45)

where X for some fixed X 2 V is seen as an endomorphism of V: Therefore we

show the next Lemma.
Lemma 2.3.4 The support † of  2 ƒ3 V is an isotropic subspace if and only if
"
 satisfies the condition (2.45). If  is of real type .3; 0/ C .0; 3/; then † is Jcan -
invariant.
Proof First the isotropy of † is equivalent to g.X Y; Z W/ D 0 for all X; Y; Z;
W 2 V:
From

g.X Y; Z/ D .X; Y; Z/ D .X; Z; Y/ D g.X Z; Y/ D g. Y; X Z/; 8X; Y; Z

we get

g.X Y; Z W/ D g.Z X Y; W/:

This equation shows X ı Y D 0; for all X; Y 2 V; if and only if its support † is

isotropic.
The last assertion follows from

" " Eq: (2.44) "

Jcan † D spanfJcan X Y j X; Y 2 Vg D spanfX Jcan Y j X; Y 2 Vg D † ;

where we used that  is of real type .3; 0/ C .0; 3/; if and only if X anti-commutes
"
with Jcan : t
u
We define the kernel of a three-form  2 ƒ3 V  by K D K D ker.X 7! Xy/:
Lemma 2.3.5 (Lemma 3.1 of [108]) One has K D †? ?
 and † D K :

Lemma 2.3.6 (Lemma 3.2 of [108]) Let .V; J; h; i/ be a pseudo-Hermitian vector
space with dimR .V/ D 8 then a real three-form  of type .3; 0/ C .0; 3/ and of
non-vanishing length has a (complex) one dimensional kernel K ; which admits an
orthogonal complement .K /? : Moreover, one has † D .K /? :
Remark 2.3.7 Let us observe, that, if  has length zero, one can replace the
orthogonal complement by the null-space and obtain an analogous statement as in
the last proposition.
34 2 Preliminaries

Lemma 2.3.8 (Lemma 3.4 of [108]) Let .V; J; h; i/ be a pseudo-Hermitian vector
space with dimR .V/ D 10 then a real three-form  of type .3; 0/ C .0; 3/ and of non-
vanishing length is of the following possible types:
(i) There exists an orthonormal real basis f fi g10
iD1 D fe1 ; Je1 ; : : : ; e5 ; Je5 g and real
numbers ˛; ˇ such that

.e1 ; e2 ; e3 / D ˛ ¤ 0I .e4 ; e5 ; e1 / D ˇ (2.46)

and . fi ; fj ; fk / D 0 for the cases which are not obtained from (2.46) by skew-
symmetry and type relations.
(ii) There exists an orthonormal real basis f fi g10 iD1 D fe1 ; Je1 ; : : : ; e5 ; Je5 g and real
numbers ˛; ˇ such that

.e1 ; e2 ; e3 / D ˛ ¤ 0I .e4 ; e5 ; e1 C e3 / D ˇ with he1 ; e1 i D he3 ; e3 i

(2.47)

and . fi ; fj ; fk / D 0 for the cases which are not obtained from (2.47) by skew-
symmetry and type relations.

2.4 Structure Reduction of Almost "-Hermitian

Six-Manifolds

Let .V; g; J " ; !/ be a 2m-dimensional "-Hermitian vector space and ‰ D C C

i"  be an .m; 0/-form of non-zero length. We define the special "-unitary group
SU" . p; q/ as the stabiliser of ‰ in the "-unitary group U" . p; q/ such that
(
SU. p; q/ ; p C q D m; for " D 1;
SU" . p; q/ D StabGL.V/ .g; J " ; ‰/ Š
SL.m; R/; for " D 1;

where SL.m; R/ acts reducibly such that V D V C ˚ V  .

With this notation, an SU" . p; q/-structure on a manifold M 2m is an almost
"-Hermitian structure .g; J " ; !/ together with a global .m; 0/-form ‰ of non-
zero constant length. Locally, there exists an "-unitary frame fe1 ; : : : ; em ; emC1 D
Je1 ; : : : ; e2m D Jem g which is adapted to the SU" . p; q/-reduction ‰ in the sense that

‰ D a.e1 C i" e.mC1/ / ^ : : : ^ .em C i" e2m / (2.48)

for a constant a 2 R .
In dimension 6, there is a characterisation of SU" . p; q/-structures in terms of
stable forms. Given a six-dimensional real vector space V; we call a pair .!; / of a
2.4 Structure Reduction of Almost "-Hermitian Six-Manifolds 35

stable ! 2 ƒ2 V  and a stable  2 ƒ3 V  compatible if it holds

! ^  D 0: (2.49)

Let us recall, that the stabiliser in GL.V/ of a compatible pair is

StabGL.V/ .!; / D SU" . p; q/; p C q D 3;

where " 2 f˙1g is the sign of ./, that is J2 D "idV . This is seen as follows.
For the two-form !, stability is equivalent to non-degeneracy and we choose the
orientation on V such that ! 3 is positive. By the previous section, we can associate
an "-complex structure J to the stable three-form . For instance in an adequate
basis, it is easy to verify that ! ^  D 0 is equivalent to the skew-symmetry of J
with respect to !. Equivalently, the pseudo-Euclidean metric

g D " !.; J /; (2.50)

induced by ! and  is "-Hermitian with respect to J . Since ‰ D  C i" J  is a

.3; 0/-form and the stabiliser of ! and  also stabilises the tensors induced by them,
the claim follows.
We conclude that an SU" . p; q/-structure, p C q D 3, on a six-manifold is
characterised by a pair .!; C / 2 2 M3 M of everywhere stable and compatible
forms such that the induced .3; 0/-form ‰ D C C i" J  C C D C C i"  has
constant non-zero length with respect to the induced metric (2.50). In an "-unitary
frame which is adapted to ‰ in the sense of (2.48), the formula

 C C 21
^ Dk k !3 (2.51)
6

is easily verified. Thus, given a compatible pair .!; C / of stable forms, it can
be checked that the induced .3; 0/-form ‰ has constant non-zero length without
explicitly computing the induced metric.
We remark, that in the almost Hermitian case, the literature often requires ‰ to
be normalised such that k C k2 D 4, for instance in [32]. In the more general almost
"-Hermitian case, we have in an adapted local "-unitary frame (2.48) with 1 D 2
C
D a.e123 C ".e156 C e426 C e453 // and k C 2
k D 4a2 3 (2.52)

for a real constant a. Therefore we have to consider two different normalisations

k C k D ˙4 or we multiply the metric by 1 if necessary such that k C k2 is
always positive.
Finally, we remark that SU.3/-structures are classified in [32] and it is shown that
the intrinsic torsion is completely determined by the exterior derivatives d!, d C
and d  .
36 2 Preliminaries

2.5 Pseudo-Riemannian Submersions

Let us consider the setting of a pseudo-Riemannian submersion  W .M; g/ !

.N; h/: The tangent bundle TM of M splits orthogonally into the direct sum

TM D H ˚ V; (2.53)

where V is the vertical and H the horizontal distribution. Denote by H ; V the

canonical inclusions and by H ; V the canonical projections. We recall the defi-
nition [18, 100] of the fundamental tensorial invariants A and T of the submersion 

T D H ı rV ./ ı V C V ı rV ./ ı H ;
A D H ı rH ./ ı V C V ı rH ./ ı H ;

where  is a vector field on M:

The components of the Levi-Civita connection r are given in the next proposi-
tion (compare [100], [18] 9.24 and 9.25).
Proposition 2.5.1 Let  W .M; g/ ! .N; h/ be a pseudo-Riemannian submersion,
denote by r the Levi-Civita connection of g and define r V WD V ı r ı V : For
vector fields X; Y in H and U; V in V we have the following identities

rU V D rUV V C TU V; (2.54)
rU X D TU X C H .rU X/; (2.55)
rX U D V .rX U/ C AX U; (2.56)
rX Y D AX Y C H .rX Y/; (2.57)
V ŒX; Y D 2AX Y; (2.58)
g.AX Y; U/ D g.AX U; Y/; or more generally A is alternating. (2.59)

The canonical variation of the metric g for t 2 R  f0g is given by

8
ˆ
<g.X; Y/; for X; Y 2 H;
ˆ
gt WD tg.V; W/; for V; W 2 V;
ˆ
:̂g.V; X/ D 0; for V 2 V; X 2 H:

Lemma 2.5.2 Denote by X; Y vector fields in H and by U; V vector fields in V:

(1) Let At and T t be the tensorial invariants for gt and A and T those for g D g1 :
Then it holds

AtX Y D AX Y; AtX U D tAX U and (2.60)

TUt V D tTU V; TUt X D TU XI (2.61)
2.6 Para-Sasaki Manifolds 37

(2) rUt V V D rUV VI

(3)

H .rXt Y/ D H .rX Y/ and V .rXt V/ D V .rX V/I (2.62)

(4)

H .r t V X  rV X/ D .t  1/AX VI (2.63)
rUt V D rU V C .t  1/TU V: (2.64)

Proof The first part can be found in Lemma 9.69 of [18]. On the right hand-side of
the Koszul formulas one only needs the metric gt on V to determine r t V : This shows
r t V D r V : An analogous argument using the Koszul formulas shows H .r t X Y/ D
H .rX Y/ and V .rXt V/ D V .rX V/: The first part of the point (4) follows from
the identities (2.58) and (2.59) and the Koszul formulas. The last equation follows
from (1) and (2): rUt V D r t V V
U V C TU V D rU V C tTU V D rU V C .t  1/TU V: u
t
t

2.6 Para-Sasaki Manifolds

In Sect. 3.7 of Chap. 3 we give a construction of (non-flat) Ricci-flat (strict) nearly

para-Kähler manifolds using a cone construction over para-Sasaki manifolds, which
we are therefore going to discuss now. Let .M; h/ be a semi-Riemannian manifold.
A cone is a semi-Riemannian manifold

b D RC  M; gc WD cdr2 C r2 h/ with c 2 R ;
.M

where r > 0 is the coordinate of RC : One calls gc space-like for c > 0 and time-
like for c < 0: Since we are considering semi-Riemannian metrics, we may suppose
b the curvature
c D ˙1 by rescaling gc : For lifts of vector fields X; Y; Z; U on M to M
4b b
tensor R of M is
 
b 1
R.X; Y; Z; U/ D r2 R.X; Y; Z; U/  Œg. Y; Z/g.X; U/  g.X; Z/g. Y; U/ :
c

Moreover, it is

@r yb
R D 0:

Hence b R is flat if and only if R has constant curvature 1c : As in the complex

setting there are different characterisations of the para-complex analogue of Sasaki

4
See for example [100, pp. 209–211] for the curvature of warped products.
38 2 Preliminaries

manifolds and different names occur in the literature like para-Sasaki manifolds [8]
or time-like Sasaki manifolds [82].
Definition 2.6.1 (See Definition 6.3 of [82] or Proposition 8.2 of [8])
A para-Sasaki manifold is a triple .M; g; T/; where .M; g/ is a semi-Riemannian
manifold of signature5 .m; m C 1/ and T is a time-like unit Killing vector field such
that the endomorphism ˆ WD DT 2 .End.TM// satisfies

ˆ2 D Id C g.; T/T; (2.65)

.DX ˆ/Y D g.X; Y/T C g. Y; T/X; 8X; Y 2 .TM/: (2.66)

A para-Sasaki Einstein manifold .M; g; T/ is a para-Sasaki manifold such that g is

an Einstein metric.
Remarks 2.6.2
(i) For a semi-Riemannian manifold .M; g/ there is a one-to-one correspondence
(Theorem 8.1 of [8]) between para-Sasaki structures .M; g; T/ and para-Kähler
b g D g1 ; b
structures on the (space-like) cone .M;b P/ with

b T b
P.@r / D ; b
P.T/ D r@r and b
P.X/ D b
DX T D DX T; for X 2 T ? :
g
r
(ii) The Einstein equation for a para-Sasaki manifold .M; g; T/ of dimension 2mC1
implies as in the Riemannian case that the scalar curvature equals 2m.2m C 1/
or the Einstein constant equals 2m (see Proposition 6.5 of [82]).
In particular, the corresponding para-Kähler cone .M;bb g; b
P/ is Ricci-flat if
and only if .M; g; T/ is para-Sasaki Einstein. Therefore M is a para-Sasaki
b is
Einstein manifold if and only if the (restricted) holonomy group of its cone M
contained in SU.P0 ; g0 / (see Proposition 6.6 of [82]), which by Remark 3.1.12
(ii) of Chap. 3 stabilises an m-form that for m D 3 has isotropic support (in the
sense of Sect. 2.3). In consequence M b can be endowed with a parallel m-form
and in dimension 6, one case of our interest, M b admits a parallel three-form of
isotropic support.

2.6.1 The T-Dual Space

In order to explain the examples given below we recall the concept of the T-dual
space introduced in [81, 82].
Let g be a compact Lie algebra and h  g be a sub-algebra. One fixes an ad.h/-
invariant complement m of h in g and an involutive automorphism with
T W g ! g; T ¤ Id:

5
In our convention m C 1 is the number of negative directions.
2.6 Para-Sasaki Manifolds 39

Let

g˙ WD fX 2 g j TX D ˙Xg

be the ˙1-eigenspaces of T. This data gives rise to another (non-compact) real form

g0 WD gC ˚ ig

of the complexified Lie algebra gC : One additionally fixes an ad.h/-invariant scalar

product h; i on m: Further one assumes that it holds

T.h/ D h and T.m/ D m

and that T W m ! m is an isometry.

Then

h0 D h \ gC ˚ i.h \ g /  g0

is a sub-algebra and setting m0 WD m \ gC ˚ i.m \ g / one has

g0 D m0 ˚ h0 and Œh0 ; m0   m0 :

The restriction h; i0 of the sesquilinear extension on mC of h; i to m0  mC yields

an ad.h0 /-invariant pseudo-Euclidean scalar product h; i0 :
Definition 2.6.3 The quadruple .g0 ; h0 ; m0 ; h; i0 / constructed from .g; h; m; h; i/ in
the above way is called T-dual to .g; h; m; h; i/ :
Definition 2.6.4 Let G=H be a Riemannian homogeneous space, where G is a
compact Lie group. Denote by .g; h; m; h; i/ the associated data, then the semi-
Riemannian homogeneous space G0 =H 0 with associated data .g0 ; h0 ; m0 ; h; i0/ is
called T-dual if .g; h; m; h; i/ and .g0 ; h0 ; m0 ; h; i0/ are T-dual for some involutive
automorphism T of g:
Examples 2.6.5 As mentioned in the Introduction we obtain a correspondence
(Theorems 3.7.3 and 3.7.5 of Chap. 3) between conical Ricci-flat (strict) nearly para-
Kähler manifolds and conical Ricci-flat para-Kähler manifolds. Regular conical
Ricci-flat para-Kähler manifolds can be characterised as local cones over para-
Sasaki Einstein manifolds (Proposition 3.7.8 of Chap. 3). In particular, the cone over
a para-Sasaki Einstein manifold yields a conical Ricci-flat para-Kähler manifold and
via our construction a conical Ricci-flat (strict) nearly para-Kähler manifold. For this
reason we include the subsequent classes of examples.
(i) This class of examples is given in Example 6.2 of [82]: Denote by .B; h; P/
a 2m-dimensional para-Kähler Einstein manifold with scalar curvature
40 2 Preliminaries

4m.m C 1/ and consider the R -bundle6 M D PU.P0 ;g0 /  R ; where  is

the following representation of the para-unitary group
 
 A 0
 W U. P0 ; g0 / ! R ; 7! det.A/
0 .At /1

with the connection ' induced by the Levi-Civita connection. On M we define

a metric by

1
g.X; Y/ WD .  h/.X; Y/  '.X/'. Y/; for X; Y 2 .TM/;
.m C 1/2

where the Lie algebra of R is identified with R and where  W M ! B is the

1
projection. Moreover, let  WD mC1 '; T the vector field defined by  D g.T; /;
then .M; g; T/ is a para-Sasaki Einstein manifold.
(ii) Consider a compact homogeneous space M D G=H with left-invariant
Riemannian metric g associated to .g; h; m; h; i/: Let .; ; '/ be a Sasakian
structure on .M; g/; which is G-invariant. Moreover, let .G0 =H 0 ; g0 / be T-dual
to .G=H; g/; where H 0 is connected and denote by 0 the Ad H-invariant vector
in m corresponding to the G-invariant vector field : Provided that 0 2 m ;
then i0 defines a G0 -invariant vector field  0 on G0 =H 0 : Setting ˆ WD D 0
and 0 WD g. 0 ; / one obtains a para-Sasaki structure on .G0 =H 0 ; g0 /; cf.
Example 6.3 of [82]. If g is an Einstein metric, then by Corollary 4.2 of [81]
the T-dual metric g0 is an Einstein metric, too.
Let us recall that a compact homogeneous Sasaki-Einstein manifold M of
dimension 5 is a circle bundle over either CP2 or CP1  CP1 which means
that M is either S5 or S2  S3 ; see Corollary 11.1.14 of [20]. The Einstein
metric on S2  S3 was constructed in [79] (see also Section 10.2 of [56]) by
considering it as Stiefel manifold V4;2 D SO.4/=SO.2/ and the related T-duals
are discussed in Section 8.1 of [82]. The T-dual space of importance here is
SOC .2; 2/=SOC .1; 1/ which fibers over

SOC .2; 1/=SOC .1; 1/  SOC .2; 1/=SOC .1; 1/

and is para-Sasaki Einstein of non constant sectional curvature, since from this
manifold we get an (first) example for a non-flat Ricci-flat nearly para-Kähler
manifold. This structure has also recently been considered in Section 3 of [29]
as the natural generalisation of the well-known Sasaki-Einstein metric on the
tangent sphere bundle T1 S3 Š S3  S2 :
(iii) Examples of para-Sasaki Einstein manifolds also arise in projective special
para-Kähler geometry, see for instance Theorem 4.4.13 of Chap. 4 or [39].

6
Note, that the frame bundle of a para-Kähler manifold can be reduced to the bundle of para-unitary
frames PU.P0 ;g0 / .
Chapter 3
Nearly Pseudo-Kähler and Nearly Para-Kähler
Manifolds

3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler

Manifolds

3.1.1 General Properties

In this subsection we collect some information on almost "-Hermitian manifolds

with a special emphasis on the nearly "-Kähler case.
Definition 3.1.1 An almost "-Hermitian manifold .M 2m ; g; J " ; !/ is called nearly
"-Kähler manifold, provided that its Levi-Civita connection r satisfies the nearly
"-Kähler condition

.rX J " / X D 0; 8X 2 TM:

A nearly "-Kähler manifold is called strict if rX J " ¤ 0 for all non-trivial vector
fields X.
A tensor field B 2 ..TM  /˝2 ˝ TM/ on a pseudo-Riemannian manifold .M; g/
is called (totally) skew-symmetric if the tensor g.B.X; Y/; Z/ is a three-form. The
following characterisation of a nearly "-Kähler manifold is well-known in the
Riemannian context and we refer to Proposition 3.2 of [110] for the complete proof
in the pseudo-Riemannian setting.
Proposition 3.1.2 An almost "-Hermitian manifold .M 2m ; g; J " ; !/ satisfies the
nearly "-Kähler condition if and only if d! is of real type .3; 0/ C .0; 3/ and the
Nijenhuis tensor is totally skew-symmetric.
Remark 3.1.3 The notion of nearly "-Kähler manifold corresponds to the gener-
alised Gray-Hervella class W1 in [86]. However, in the para-Hermitian case, there

© Springer International Publishing AG 2017 41

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0_3
42 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

are two subclasses, see [59]. Indeed, we already observed that

(2.36)
A D r! 2 3;0  D .ƒ3 V  ˚ ƒ3 H /

for a nearly para-Kähler manifold.

Definition 3.1.4 A connection rN on an "-Hermitian manifold .M 2m ; g; J " ; !/ is
N D 0 and rJ
called "-Hermitian provided, that it satisfies rg N " D 0.
The Riemannian case of the next result is due to [57], the para-complex case is
shown in [78]. In fact, the sketched proof in [57] holds literally for the almost
pseudo-Hermitian case with indefinite signature as well. A direct and simultaneous
proof of all cases can be found in Proposition 3.4 of [110].
Theorem 3.1.5 An "-Hermitian manifold .M 2m ; g; J " ; !/ admits an "-Hermitian
connection with totally skew-symmetric torsion if and only if the Nijenhuis tensor
is totally skew-symmetric. If this is the case, the connection rN and its torsion T are
uniquely defined by

1
g.rN X Y; Z/ D g.rX Y; Z/ C g.T.X; Y/; Z/;
2
g.T.X; Y/; Z/ D "g.N.X; Y/; Z/  d!.J " X; J " Y; J " Z/;

and we call rN the characteristic "-Hermitian connection (with skew-symmetric

torsion).
This connection can be seen as a natural generalisation of the Chern- or Bismut-
connection. Another name for the characteristic connection is canonical connection.
Remark 3.1.6 An almost Hermitian manifold is said to be of type G1 if it admits a
Hermitian connection with skew-symmetric torsion. In terms of the Gray-Hervella
list [68], this means, that it is of type W1 ˚ W3 ˚ W4 ; i.e. the missing part is the
almost Kähler component W2 :
More generally, the proposition justifies to say that an almost "-Hermitian
manifold is of type G1 if it admits an "-Hermitian connection with skew-symmetric
torsion.
In particular, the proposition applies to nearly "-Kähler manifolds .M; g; J " ; !/:
In fact, comparing the identities (2.35) and (2.41), we see that the real three-form A
is of type .3; 0/ C .0; 3/. Since d! is the alternation of r!, we have

d! D 3r! D 3A 2 3;0 ; (3.1)

where A is defined in (2.39). Furthermore, if we apply the nearly "-Kähler condition

to the expression (2.37), the Nijenhuis tensor of a nearly "-Kähler structure
simplifies to

N.X; Y/ D 4 J " .rX J " /Y: (3.2)

3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds 43

We conclude that the Nijenhuis tensor is skew-symmetric since

(2.41)
g.N.X; Y/; Z/ D 4A.X; Y; J " Z/ D 4"J " A.X; Y; Z/: (3.3)

Explicitly the connection rN is then given by

1
rN X Y D rX Y C "J " .rX J " /Y; for X; Y 2 .TM/: (3.4)
2
In this case, the skew-symmetric torsion T of the characteristic "-Hermitian
connection simplifies to

1
T.X; Y/ D "J " .rX J " /Y D "N.X; Y/
4
due to the identities (3.1)–(3.3).
For a proof of the next result we may refer to Lemma 2.4 of [16] for nearly Kähler
manifolds, Theorem 5.3 of [78] for nearly para-Kähler manifolds and Proposition
3.2 of [108] for the remaining case. As the attentive reader observes, the proof relies
on the curvature identity (3.20), even though we list it already in this section as one
of the very useful properties of the characteristic connection.
Proposition 3.1.7 The characteristic "-Hermitian connection rN of a nearly "-
Kähler manifold .M 2m ; J " ; g; !/ satisfies

N
r.rJ "
/D0 N
and r.T/ D 0:

A direct consequence is the following Corollary.

Corollary 3.1.8 On a nearly "-Kähler manifold .M 2m ; J " ; g; !/ the tensors rJ " and
N D 4"T have constant length.
Remark 3.1.9 In dimension 6, the fact that rJ " has constant length is usually
expressed by the equivalent assertion that a nearly "-Kähler six-manifold is of
constant type, i. e. there is a constant ˛ 2 R such that

g..rX J " /Y; .rX J " /Y/ D ˛ f g.X; X/g. Y; Y/  g.X; Y/2 C "g.J " X; Y/2 g: (3.5)

In fact, the constant is ˛ D 14 krJ " k2 . Furthermore, it is well-known in the

Riemannian case that strict nearly Kähler six-manifolds are Einstein manifolds with
Einstein constant 5˛ [67]. The same is true in the para-Hermitian case [78] and in
the pseudo-Hermitian case [108] or Theorem 3.2.8 of this chapter. The sign of the
type constant depends on the signature .2p; 2q/ of g by sign. p  q/; see for example
[82]. In particular, in the Riemannian case it follows ˛ > 0 and as a consequence a
strict nearly Kähler manifold cannot be Ricci-flat.
44 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

The case krJ " k2 D 0 for a strict nearly "-Kähler six-manifold can only occur in the
para-complex world. We give different characterisations of such structures which
provide an obvious break in the analogy of nearly para-Kähler and nearly pseudo-
Kähler manifolds. To emphasise that we are only considering the nearly para-Kähler
case we write  for J " with " D 1:
Proposition 3.1.10 For a six-dimensional strict nearly para-Kähler manifold
.M 6 ; g; ; !/ the following properties are equivalent:
(i) krk2 D kAk2 D 0:
(ii) The three-form A D r! 2 3;0  is either in .ƒ3 V  / or in .ƒ3 H /.
(iii) The three-form A D r! 2 3;0  is not stable.
(iv) The metric g is Ricci-flat.
In consequence for a Ricci-flat nearly para-Kähler manifold the 3-forms D!.; ; /
and N.; ; / are not stable in the sense of Hitchin [75, 76], cf. Sect. 2.1 of Chap. 2
for details on stable forms and hence the powerful methods of stable forms are not
available. The following observation is used later in this text to construct examples
of non-flat Ricci-flat nearly para-Kähler six-manifolds.
Corollary 3.1.11
(a) On a Ricci-flat nearly para-Kähler six-manifold .M 6 ; ; g/ the 3-forms r! and
N have isotropic support.
(b) Let .M; ; g/ be a nearly para-Kähler manifold such that the Nijenhuis tensor
N has isotropic support, then one has NX ı NY D 0:
Proof The identity (3.5) combined with ˛ D 0 yields g..rX /Y; .rX /Y/ D 0 and
further

g..rX /Y; .rX /Y/ D 0: (3.6)

This shows that the two 3-forms g..rX /Y; Z/ and g..rX /Y; Z/ have isotropic
support. Finally we obtain after polarisation of (3.6), that one has

g.NX NY Z; W/ D 16g..rY /Z; .rX /W/ D 0

for all X; Y; Z; W 2 .TM/: This yields the last statement. t

u
Remark 3.1.12 Let us consider R2m with its standard para-Hermitian structure
.P0 ; g0 / and isotropic basis .e1 ; : : : ; em ; f1 ; : : : ; fm / with dual isotropic basis
.e1 ; : : : ; em ; f 1 ; : : : ; f m /; compare Eq. (2.33). Then the m-forms e1 ^ : : : ^ em and
f 1 ^ : : : ^ f m are invariant under SU.P0 ; g0 /; which follows from Eq. (2.34) and
have isotropic support in the above sense for m D 3:
As the (restricted) holonomy of a Ricci-flat para-Kähler six-manifold .M 6 ; P; g/
lies in SU.P0 ; g0 /; it follows that a Ricci-flat para-Kähler six-manifold admits a
family of (non-vanishing) parallel 3-forms with isotropic support.
3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds 45

Lemma 3.1.13 Let .M; ; g/ be a nearly para-Kähler manifold such that the
Nijenhuis tensor N.X; Y; Z/ has isotropic support, then it holds

.rX N/. Y; Z/ D 0; (3.7)

.rX N/. Y; Z; W/ D 0: (3.8)

In particular, these identities are satisfied for a Ricci-flat nearly para-Kähler six-
manifold.
Proof We directly compute Eq. (3.7) using Theorem 3.1.5 and Corollary 3.1.11 (b)

.rX N/. Y; Z/ D rX .N. Y; Z//  N.rX Y; Z/  N. Y; rX Z/

    
N 1 N 1
D rX  NX .N. Y; Z//  N rX  NX Y; Z
8 8
   
1
N Y; rN X  NX Z D .rN X N/. Y; Z/ D 0:
8

Combining Eq. (3.7) with rg D 0 and N.X; Y; Z/ D g.N.X; Y/; Z/ we obtain

Eq. (3.8). The last statement follows from Corollary 3.1.11 (a). t
u
Flat strict nearly para-Kähler manifolds .M; g; J; !/ are classified in work with
V. Cortés, see Sect. 3.6 of this chapter. It turns out that these always satisfy
krJ " k2 D 0. In [59], almost para-Hermitian structures on tangent bundles TN of
real three-dimensional manifolds N 3 are discussed. It is shown that the existence of
nearly para-Kähler manifolds satisfying the second condition of Proposition 3.1.10
is equivalent to the existence of a certain connection on N 3 without constructing an
example. However, to our best knowledge, there was no reference for an example of
a Ricci-flat non-flat strict nearly para-Kähler structure until the author’s paper [109]
discussed in Sect. 3.7 of this chapter.

3.1.2 Characterisations by Exterior Differential Systems

in Dimension 6

The following lemma explicitly relates the Nijenhuis tensor to the exterior differen-
tial. For " D 1, it gives a characterisation of Bryant’s notion of a quasi-integrable
U. p; q/-structure, p C q D 3, in dimension 6 [24].
Let .M 6 ; g; J " ; !/ be a six-dimensional almost "-Hermitian manifold. If
fe1 ; : : : ; e6 D Je3 g is a local "-unitary frame, we define a local frame fE1 ; E2 ; E3 g
of .TM 1;0 / by

Ei WD .ei C i" "J " ei / D .ei C i" eiCm /

46 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

for i D 1; 2; 3 and call it a local "-unitary frame of .1; 0/-forms. The dual vector
fields of the .1; 0/-forms are

1 1
Ei D e1;0
i D .ei C i" "J " ei / D .ei C i" "eiCm /;
2 2
such that the C" -bilinearly extended metric in this kind of frame satisfies

1
g.Ei ; EN j / D i ıij and g.Ei ; Ej / D 0:
2
Lemma 3.1.14 The Nijenhuis tensor of an almost "-Hermitian six-manifold
.M 6 ; g; J " ; !/ is totally skew-symmetric if and only if for every local "-unitary
frame of .1; 0/-forms, there exists a local C" -valued function  such that

.dEs.1//0;2 D  s.1/ Es.2/ s.3/ (3.9)

for all even permutations s of f1; 2; 3g.

Proof First of all, the identities

N W/
N.V; N D 4"ŒV;
N W
N 1;0 N D0
and N.V; W/

for any vector fields V D V 1;0 , W D W 1;0 in TM 1;0 follow immediately from the
definition of N. Using the first identity, we compute in an arbitrary local "-unitary
frame

dEi .EN j ; EN k / D Ei .ŒEN j ; EN k / D 2i g.ŒEN j ; EN k ; EN i /

1
D 2i g.ŒEN j ; EN k 1;0 ; EN i / D " i g.N.EN j ; EN k /; EN i /
2
for all possible indices 1  i; j; k  3. If the Nijenhuis tensor is totally skew-
symmetric, Eq. (3.9) follows by setting

1
D " g.N.EN 1 ; EN 2 /; EN 3 /: (3.10)
2
Conversely, the assumption (3.9) for every local "-unitary frame implies that the
Nijenhuis tensor is everywhere a three-form when considering the same computa-
N D 0.
tion and N.V; W/ t
u
From the last Lemma we get the following Corollary.
Corollary 3.1.15 For an almost "-Hermitian six-manifold .M 6 ; g; J " ; !/ with
totally skew-symmetric Nijenhuis tensor, there exists a function f 2 C1 .M/ such
that one has

gp .X; Y/ D f . p/ tr.NX ı NY /; p 2 M; X; Y 2 Tp M:
3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds 47

In particular, if this function f does not vanish, i.e. if the almost complex structure
is quasi-integrable, the almost "-complex structure fixes the conformal class of g:
If there is an SU" . p; q/-reduction (cf. Sect. 2.4 of Chap. 2) with closed real part,
this characterisation can be reformulated globally in the following sense.
Proposition 3.1.16 Let .!; C / be an SU" . p; q/-structure on a six-manifold M
such that C is closed. Then the Nijenhuis tensor is totally skew-symmetric if and
only if

d D ! ^! (3.11)

for a global real function .

Proof It suffices to proof this locally. Let fEi g be an "-unitary frame of .1; 0/-forms
with 1 D 2 which is adapted to the SU" . p; q/-reduction such that ‰ D C C
i"  D aE123 for a real constant a as in (2.48). The fundamental two-form is

1 X
m
N
! D  i" k Ekk
2 kD1

in such a frame. Furthermore, as C is closed, we have d‰ D i" d  D d‰; N

which implies that d‰ 2 ƒ2;2 . Considering this, we compute the real 4-form



d D "i" d‰ D "i" a .dE1 /0;2 ^ E23 C .dE2 /0;2 ^ E31 C .dE3 /0;2 ^ E12

and compare this expression with

1 N N N N N N
!^! D ".2 3 E2233 C 1 3 E1133 C 1 2 E1122 /
2
1 NN NN NN
D  "3 .1 E2323 C 2 E3131 C 3 E1212 /:
2
Hence, by Lemma 3.1.14, the Nijenhuis tensor is totally skew-symmetric if and only
if d  D ! ^ ! holds true for a real function . More precisely, the two functions
and  are related by the formula

D 23 i" a: (3.12)

t
u
An SU" . p; q/-structure .!; / is called half-flat if

d D 0; d! 2 D 0;

and nearly half-flat if

d D ! ^!
48 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

for a real constant . These notions are defined for the Riemannian signature in
[32] respectively [53] and extended to all signatures in our paper [46] presented in
Chap. 4 of this text.
Corollary 3.1.17 Let .!; C / be a half-flat SU" .p; q/-structure on a six-manifold
M. Then, the Nijenhuis tensor is totally skew-symmetric if and only if .!;  / is
nearly half-flat.
Proof If .!;  / is nearly half-flat, Eq. (3.11) is satisfied by definition and the
Nijenhuis tensor is skew-symmetric by the previous proposition. In particular one
has d! 2 D 0: Conversely, if the Nijenhuis tensor is skew, we know that (3.11) holds
true for a real function , since we have d C D 0. Differentiating this equation and
using d! 2 D 0, we obtain d ^ ! 2 D 0. The assertion follows as wedging by ! 2 is
injective on one-forms. t
u
Remark 3.1.18 An interesting property of SU" . p; q/-structures which are both half-
flat and nearly half-flat in the sense of the corollary is the fact that, given that the
manifold and the SU" . p; q/-structure are analytic, the structure can be evolved to
both a parallel G2 -structure and a nearly parallel G2 -structure via the Hitchin flow.
For details, we refer to [76] and [114] for the compact Riemannian case and Chap. 4
of this text or our paper [46] for the non-compact case and indefinite signatures.
In [33], six-dimensional nilmanifolds N admitting an invariant half-flat SU.3/-
structure .!; C / such that .!;  / is nearly half-flat are classified. As six
nilmanifolds admit such a structure, we conclude that these structures are not as
scarce as nearly Kähler manifolds. It is also shown in the same article, that these
structures induce invariant G2 -structures with torsion on N  S1 .
We give another example of a (normalised) left-invariant SU.3/-structure on S3 
S which satisfies d C D 0; d  D ! ^ ! such that d! neither vanishes nor
3

is of type (3,0) + (0,3). We choose a global frame of left-invariant vector fields

fe1 ; e2 ; e3 ; f1 ; f2 ; f3 g on S3  S3 such that

de1 D e23 ; de2 D e31 ; de3 D e12 I df 1 D f 23 ; df 2 D f 31 ; df 3 D f 12 ;

p
and set with x D 2 C 3

! D e1 f 1 C e2 f 2 C e3 f 3 ;
C 1
D  x2 e123 C 2xe12 f 3  2xe13 f 2  2xe1 f 23 C 2xe23 f 1
2
C 2xe2 f 13  2xe3 f 12 C .4x  8/f 123 ;
 1 123
D xe  2e1 f 23 C 2e2 f 13  2e3 f 12 C 4f 123 ;
2
g D x .e1 /2 C x .e2 /2 C x .e3 /2 C 4 . f 1 /2 C 4 . f 2 /2 C 4 . f 3 /2
 2x e1 f 1  2x e2 f 2  2xe3 f 3 :
3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds 49

Finally, we come to the characterisation of six-dimensional nearly "-Kähler

manifolds by an exterior differential system generalising the classical result of [103]
which holds for " D 1 and Riemannian metrics.
Theorem 3.1.19 Let .M; g; J " ; !/ be an almost "-Hermitian six-manifold. Then M
is a strict nearly "-Kähler manifold with krJ " k2 ¤ 0 if and only if there is a
reduction ‰ D C C i"  to SU" . p; q/ which satisfies
C
d! D 3 ; (3.13)

d D 2 ˛ ! ^ !; (3.14)

where ˛ D 14 krJ " k2 is constant and non-zero.

Remark 3.1.20 Due to our sign convention ! D g.:; J " :/, the constant ˛ is positive
in the Riemannian case and the second equation differs from that of other authors.
Furthermore, we will sometimes use the term nearly "-Kähler manifold of non-zero
type if krJ " k2 ¤ 0.
Proof By Proposition 3.1.2, the manifold M is nearly "-Kähler if and only if d! is
of type .3; 0/ C .0; 3/ and the Nijenhuis tensor is totally skew-symmetric.
Therefore, when .g; J " ; !/ is a strict nearly "-Kähler structure such that kAk2 D
krJ " k2 is constant (by Corollary 3.1.8) and not zero (by assumption), we can define
the reduction ‰ D C C i"  by C D 13 d! D A and  D J " C such that
the first equation is satisfied. Since ! is of type .1; 1/ and therefore d.! ^ !/ D
2d! ^ ! D 0, this reduction is half-flat. Thus, Corollary 3.1.17 and the skew-
symmetry of N imply that there is a constant 2 R such that d  D ! ^ !.
According to (2.48), we can choose an "-unitary local frame with 1 D 2 , such
that

‰ D A  i" J " A D aE123 ;

where a is constant and satisfies 4˛ D krJ " k2 D k C k2 D 4a2 3 by (2.52). Now,

the functions defined in Lemma 3.1.14 and Proposition 3.1.16 evaluate as

1
"g.N.EN 1 ; EN 2 /; EN 3 / D 2J  A.EN 1 ; EN 2 ; EN 3 / D " i" a;
(3.10) (3.3)
 D
2
(3.12)
D 23 i" a D 23 a2 D 2˛:

Conversely, if a given SU" . p; q/-structure satisfies the exterior system, the

real three-form C is obviously closed and the Nijenhuis tensor is totally skew-
symmetric by Corollary 3.1.17. Considering that d! D 3r! is of type .3; 0/C.0; 3/
by the first equation, the structure is nearly "-Kähler. Since A D  C is stable, the
structure is strict nearly "-Kähler and krJ " k D kAk ¤ 0 by Proposition 3.1.10.
Now, the computation of the constants in the adapted "-unitary frame shows that in
fact krJ " k D 4˛. t
u
50 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.1.3 Curvature Identities for Nearly "-Kähler Manifolds

Most of these identities are here only used for the almost complex case. If we are
only considering the complex case we write J and in case, that we consider the
para-complex case we write  for the "-complex structure J " : The starting point of
a series of curvature identities are

R.W; X; Y; Z/  R.W; X; JY; JZ/ D g..rW J/X; .rY J/Z/; (3.15)

R.W; X; W; Z/ C R.W; JX; W; JZ/ (3.16)
 R.W; JW; X; JZ/ D 2g..rW J/X; .rW J/Z/;
R.W; X; Y; Z/ D R. JW; JX; JY; JZ/; (3.17)

which were already proven for pseudo-Riemannian metrics by Gray [67]. In the
para-complex case the analogue of the first identity, i.e. the relation

R.W; X; Y; Z/ C R.W; X; Y; Z/ D g..rW /X; .rY /Z/; (3.18)

is shown in Proposition 5.2 of [78].

Let fei g2n
iD1 be a local orthonormal frame field, then the Ricci- and the Ricci*-
tensor are given by

2n
X 2n
1X
g.Ric X; Y/ D i R.X; ei ; Y; ei /; g.Ric X; Y/ D i R.X; JY; ei ; Jei /
iD1
2 iD1

with i D g.ei ; ei / D g. Jei ; Jei / and X; Y 2 TM: The frame fei g2n
iD1 is called adapted
if it holds Jei D eiCn for i D 1; : : : ; n: Then it follows using an adapted frame from
Eqs. (3.16) and (3.17) that

2n
X
g.rX; Y/ WD g..Ric  Ric /X; Y/ D i g..rX J/ei ; .rY J/ei /: (3.19)
iD1

Using the right hand-side we see

ΠJ; r D 0:

For the second derivative of the complex structure one has the identity
2
2g.rW;X . J/Y; Z/ D X;Y;Z g..rW J/X; .rY J/JZ/; (3.20)
3.1 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds 51

which was proven in [67] for Riemannian metrics and holds true in the pseudo-
Riemannian setting, cf. [82, Proposition 7.1]. This identity implies

2n
X
i re2i ;ei . J/Y D r. JY/: (3.21)
iD1

From Proposition 3.1.7 and the relation (3.4) of the connections r and rN one obtains
the following identities for the curvature tensor RN of rN and the curvature tensor R of
the Levi-Civita connection r

N 1
R.W; X; Y; Z/ D R.W; X; Y; Z/  g..rW J/X; .rY J/Z/
2
1
C Œg..rW J/Y; .rX J/Z/  g..rW J/Z; .rX J/Y/ (3.22)
4
1
D Œ3R.W; X; Y; Z/ C R.W; X; JY; JZ/
4
CXYZ R.W; X; JY; JZ/;

N 1
R.W; JW; Y; JZ/ D Œ5R.W; JW; Y; JZ/
4
R.W; Y; W; Z/  R.W; JY; W; JZ/: (3.23)

With the help of Eq. (3.22) it follows

N
R.W; N Y; Z; W; X/ D R.X;
X; Y; Z/ D R. N N
W; Y; Z/ D R.W; X; Z; Y/: (3.24)

N D 0 and rg
Using rJ N D 0 we obtain

N
R.W; N
X; Y; Z/ D R.W; X; JY; JZ/ (3.25)
N JW; JX; Y; Z/ D R.
D R. N JW; JX; JY; JZ/:

The general form of the first Bianchi identity (cf. Chapter III of [87]) for a
connection with torsion yields in the case of parallel torsion:

N
 R.W; X; Y; Z/ D   g..rW J/X; .rY J/Z/: (3.26)
XYZ XYZ

In a similar way we get from the second Bianchi identity (cf. Chapter III of [87])
for a connection with parallel torsion or from the second Bianchi identity for r

  rN V .R/.W;
N N
X; Y; Z/ D  R..rV J/JW; X; Y; Z/: (3.27)
VWX VWX
52 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

From deriving Eq. (3.22) and the second Bianchi identity of r one gets after a direct
computation

N 1
 rV .R/.W; X; Y; Z/ D g..rY J/Z;  .rX J/.rV J/JW/; (3.28)
VWX 2 VWX

which implies

N
 rV .R/.W; X; Y; JY/ D 0: (3.29)
VWX

Proposition 3.1.21 (Proposition 2.3 of [108]) The tensor r on a nearly pseudo-

N
Kähler manifold .M; J; g/ is parallel with respect to the characteristic connection r:
Theorem 3.1.22 (Theorem 2.4 of [108]) Let .M; J; g/ be a nearly pseudo-Kähler
manifold and let W; X be vector fields on M then it holds

2n
X  
i j g.rei ; ej / R.W; ei ; X; ej /  5R.W; ei ; JX; Jej / D 0: (3.30)
i;jD1

Let us remark, that the Riemannian case is done in [67] and the para-Kähler case in
[78].

3.2 Structure Results

As we have seen above, for a nearly pseudo-Kähler manifold r! is a differen-

tial form of type .3; 0/ C .0; 3/: In consequence real two- or four-dimensional
nearly pseudo-Kähler manifolds are automatically pseudo-Kähler. Six dimensional
nearly pseudo-Kähler manifolds are either pseudo-Kähler manifolds or strict nearly
pseudo-Kähler manifolds. Therefore we start this section in real dimension 8.1

3.2.1 Kähler Factors and the Structure in Dimension 8

The aim of this subsection is to split off the pseudo-Kähler factor of a nearly pseudo-
Kähler manifold. This will be done by means of the kernel of rJ and allows to
reduce the (real) dimension from 8 to 6.
For p 2 M we set

Kp D ker.X 2 Tp M 7! rX J/:

1
The reference for the section is the author’s paper [108].
3.2 Structure Results 53

Theorem 3.2.1 Let .M; J; g/ be a nearly pseudo-Kähler manifold. Suppose, that

the distribution K has constant dimension and admits an orthogonal complement,
(i) then M is locally a pseudo-Riemannian product M D K  M1 of a pseudo-
Kähler manifold K and a strict nearly pseudo-Kähler manifold M1 :
(ii) if M is complete and simply connected then it is a pseudo-Riemannian product
M D K  M1 of a pseudo-Kähler manifold K and a strict nearly pseudo-Kähler
manifold M1 :
N since rJ is
Proof The distribution K is parallel for the characteristic connection r;
r-parallel. By the formula (3.4) and the nearly Kähler condition it follows rN X K D
N
rX K for sections K in K and X in TM: This implies that K is parallel for the Levi-
Civita connection and in consequence its orthogonal complement .K/? is Levi-
Civita parallel. The proof of (i) finishes by the local version of the theorem of de
Rham and the proof of (ii) by the global version. t
u
Remark 3.2.2 There exist nearly pseudo-Kähler manifolds .M; J; g/ without
pseudo-Kähler de Rham factor, such that K ¤ f0g admits no orthogonal
complement. In fact, we construct Levi-Civita flat nearly pseudo-Kähler manifolds
in our paper [41], which is subject of Sect. 3.6 of this chapter, such that the three-
form p .X; Y; Z/ D gp . J.rX J/Y; Z/; for p 2 M; has a support †  Tp M
which is a maximally isotropic subspace (Here we identified Tp M and Tp M via
the metric g:). Obviously, J.rX J/Y and J.rU J/V are elements of the support of
 for arbitrary X; Y; U; V 2 Tp M: It then follows 0 D g. J.rX J/Y; J.rU J/V/ D
g. J.rJ.rX J/Y J/U; V/ for all V 2 Tp M: Hence it is †  K : Moreover for general
reasons we have shown before † D K? which shows K \ K? ¤ f0g for the
above examples. From these examples we learn, that the Theorem 3.2.1 does not
hold true, if there is no orthogonal complement.
Definition 3.2.3 A nearly pseudo-Kähler manifold .M; J; g/ is called nice if the
three-form g..r J " /; / has non-zero length in each point p 2 M:
Theorem 3.2.4 Let .M 8 ; J; g/ be a complete simply connected eight-dimensional
nice nearly pseudo-Kähler manifold. Then M D M1  M2 where M1 is a
two-dimensional Kähler manifold and M2 is a six-dimensional strict nearly pseudo-
Kähler manifold.
Proof Since .M; J; g/ is a nice nearly pseudo-Kähler manifold we can use
Lemma 2.3.6 of Chap. 2 to obtain an orthogonal splitting in the two-dimensional
distribution K and its orthogonal complement, which coincides with † : Therefore
we are in the situation of Theorem 3.2.1 (ii). t
u

3.2.2 Einstein Condition Versus Reducible Holonomy

N
In this part we study reducible r-holonomy and discuss the consequences in small
dimensions.
54 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Theorem 3.2.5 Let .M; J; g/ be a nearly pseudo-Kähler manifold.

(i) Suppose that r has more than one eigenvalue, then the characteristic Hermitian
connection has reduced holonomy.
(ii) If the tensor field r has exactly one eigenvalue then M is a pseudo-Riemannian
Einstein manifold.
Proof
(i) Let i for i D 1; : : : ; l be the eigenvalues of r: Then the decomposition in
N
the according eigenbundles Eig.i / is r-parallel and hence its holonomy is
reducible.
(ii) From the identity of Theorem 3.1.22 and r D IdTM we obtain

2n
X
0D i .R.W; ei ; X; ei /  5R.W; ei ; JX; Jei // D g..Ric  5Ric /W; X/;
iD1

where we used the Bianchi identity and an adapted frame to obtain the last
equality. This shows comparing with r D Ric  Ric that it holds Ric D 54 :
t
u
Let us recall, that in the pseudo-Riemannian setting the decomposition into the
eigenbundles is not automatically ensured to be an orthogonal direct decomposition.
Therefore we introduce the following notion.
Definition 3.2.6 A nearly pseudo-Kähler manifold .M; J; g/ is called decompos-
able if the above decomposition into the eigenbundles of the tensor r is orthogonal.
Lemma 3.2.7 Let .M; J; g/ be a decomposable nearly pseudo-Kähler manifold and
denote by i for i D 1; : : : ; l the eigenvalues of r and by Ei D Eig.i /; i D 1; : : : ; l;
the corresponding eigenbundles.
(i) For X 2 Ei and Y 2 Ej with i ¤ j one has Ric.X; Y/ D 0:
(ii) For X; Y 2 Ei it is

1 X
l
i
Ric.X; Y/ D g.X; Y/ C s g.rs X; Y/;
4 i sD1

where the tensors rs W TM ! TM; 1  s  l; are defined as

g.rs X; Y/ WD trEs ..rX J/ ı .rY J// :

Proof Let us first prove (i). We consider a basis of TM which gives a pseudo-
orthonormal basis for the Ei ; i D 1; : : : l: The Ricci curvature decomposes w.r.t.
the eigenbundles as

X
l X
Ric.X; Y/ D k R.X; ek ; Y; ek /:
sD1 ek 2Es
3.2 Structure Results 55

N
Using R.X; ek ; Y; ek / D 0 for s ¤ j one gets by Eq. (3.22)

1
R.X; ek ; Y; ek / D g..rX J/ek ; .rY J/ek /; for ek 2 Es :
4

Further for s D j one has s ¤ i and again it is

1
R.X; ek ; Y; ek / D R. Y; ek ; X; ek / D g..rX J/ek ; .rY J/ek /; for ek 2 Es :
4
In summary one obtains

X
l X
1X 1
Ric.X; Y/ D k R.X; ek ; Y; ek / D k g..rX J/ek ; .rY J/ek / D g.rX; Y/ D 0:
4 4
sD1 ek 2Es k

This shows (i). Next we show part (ii). From the identity of Theorem 3.1.22 we
conclude

X
l X
0D s k .R.W; ek ; X; ek /  5R.W; ek ; JX; Jek // :
sD1 ek 2Es

As in part (i) we get for s ¤ i with help of Eq. (3.22)

3
R.X; ek ; JY; Jek / D 3R. Y; ek ; X; ek / D  g..rX J/ek ; .rY J/ek /; for ek 2 Es :
4
It follows, that
0 1
X X
4 s g.rs X; Y/ C i @ k .R.W; ek ; X; ek /  5R.W; ek ; JX; Jek //A D 0
s¤i ek 2Ei

and another time using Eq. (3.22)

X
i g..Ric  5Ric /X; Y/ C 4 .s  i /g.rs X; Y/ D 0;
s¤i

which follows by

X
l X

g..Ric  5Ric /X; Y/ D k .R.X; ek ; Y; ek /  5R.X; ek ; JY; Jek // :
sD1 ek 2Es
56 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Pl
The identity (ii) follows now from Ric  Ric D r and sD1 rs D r: In fact, it is

4 X
g..Ric  5Ric /X; Y/ D 4g.Ric X; Y/ C 5g.rX; Y/ D .i  s /g.rs X; Y/
i
s¤i

and in consequence one obtains

4 X 4 X
l
4g.Ric X; Y/ D 5g.rX; Y/  .i  s /g.rs X; Y/ D g.rX; Y/ C s g.rs X; Y/;
i i
s¤i sD1

which finishes the proof. t

u
Theorem 3.2.8 A strict nearly pseudo-Kähler six-manifold .M 6 ; J; g/ of constant
type ˛ is a pseudo-Riemannian Einstein manifold with Einstein constant 5˛.
Proof In an adapted basis we obtain from the symmetries of rJ

3
X 3
X
g.rX; X/ D 2 i g..rX J/ei ; .rX J/ei / D 2 i g..rX J/2 ei ; ei /:
iD1 iD1

This is exactly minus the trace of the operator .rX J/2 which has a simple form in a
cyclic frame. It follows after polarising g.rX; Y/ D 4˛g.X; Y/: From Theorem 3.2.5
we compute the Einstein constant 5˛ where ˛ is the type constant of the strict nearly
pseudo-Kähler manifold M 6 : t
u
Proposition 3.2.9 Let .M 10 ; J; g/ be a nice nearly pseudo-Kähler ten-manifold.
(i) Then the tensor r in a frame of the first type in Lemma 2.3.8 of Chap. 2 is given
by

re1 D 4.˛ 2 C ˇ 2 /e1 ;

re2 D 4˛ 2 e2 ; re3 D 4˛ 2 e3 ;
re4 D 4ˇ 2 e4 ; re5 D 4ˇ 2 e5 ;
r. Jei / D Jr.ei /; i D 1; : : : ; 5;

where ˛; ˇ are constants.

(ii) For a frame of the second type in Lemma 2.3.8 of Chap. 2 the tensor r is given
by
2 3 2 2 32 3
e1 ˛ C ˇ 2 4 5 0 ˇ 2 4 5 e1
4 5
r e2 D 4 4 0 ˛ 2
0 5 4 e2 5
2 2 2
e3 ˇ 4 5 0 ˛ C ˇ 4 5 e3
3.2 Structure Results 57

re4 D 0;
re5 D 4 ˇ 2 .21 4  1/e5 ;
r. Jei / D Jr.ei /; i D 1; : : : ; 5:

The eigenvalues are f0I 4˛ 2 I 4 ˇ 2 .21 4  1/I 4.˛ 2 C 2ˇ 2 4 5 /g; where the
eigenbundles are given as

Ker.r/ D spanfe4 ; Je4 g;

Eig.r; 4˛ 2 / D spanfe1 C e3 ; e2 ; Je1 C Je3 ; Je2 g;
Eig.r; 4 ˇ 2 .21 4  1// D spanfe5 ; Je5 g;
Eig.r; 4.˛ 2 C 2ˇ 2 4 5 /// D spanfe1 C e3 ; Je1 C Je3 g;

where ˛; ˇ are constants. For ˇ 2 ¤ 0 the second case is not decomposable.

(iii) Suppose ˇ D 0 in the cases (i) and (ii). Then it follows

Eig.r; 4˛ 2 / D spanfe1 ; e2 ; e3 ; Je1 ; Je2 ; Je3 g;

Ker.r/ D spanfe4 ; e5 ; Je4 ; Je5 g:

Proof In an adapted frame we obtain from the symmetries of rJ

5
X 5
X
g.rX; Y/ D 2 i g..rX J/ei ; .rY J/ei / D 2 i g..rY J/.rX J/ei ; ei /:
iD1 iD1

This is exactly minus the trace of the operator .rY J/.rX J/: Using the form of
Lemma 2.3.8 of Chap. 2 one can calculate r by hand or using computer algebra
systems to obtain the claimed results t
u
Theorem 3.2.10 Let .M 10 ; J; g/ be a complete simply connected nice decompos-
able nearly pseudo-Kähler manifold of dimension 10. Then M 10 is of one of the
following types
(i) the tensor r has a kernel and M 10 D K  M 6 is a product of a four-dimensional
pseudo-Kähler manifold K and a strict nearly pseudo-Kähler six-manifold M 6 :
(ii) the tensor r has trivial kernel and r has eigenvalues 4.˛ 2 Cˇ 2 / with multiplicity
2; 4˛ 2 ; 4ˇ 2 with multiplicity 4 for some ˛; ˇ ¤ 0:
A nice nearly pseudo-Kähler manifold .M 10 ; J; g/ is decomposable if the dimension
of the kernel of r is not equal to two.
Proof Since we suppose, that .M 10 ; J; g/ is a nice and decomposable nearly pseudo-
Kähler manifold, Proposition 3.2.9 implies that one has the two different cases:
(i) the distribution K; which is the tangent space of the Kähler factor has dimension
4 and admits an orthogonal complement of dimension 6. This is part (iii) of
Proposition 3.2.9. Part (i) of the Theorem now follows from Theorem 3.2.1.
58 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

(ii) the tensor r has trivial kernel and we are in the situation of Proposition 3.2.9
part (i) with ˛; ˇ ¤ 0 and part (ii) follows.
t
u
Remark 3.2.11 Nearly pseudo-Kähler manifolds falling in the second case of the
last theorem are related to twistor spaces in Sect. 3.4.5 of this chapter.

3.3 Twistor Spaces over Quaternionic

and Para-Quaternionic Kähler Manifolds

In this section2 we consider pseudo-Riemannian submersions  W .M; g/ !

.N; h/ endowed with a complex structure J on M which is compatible with the
decomposition (2.53).
Lemma 3.3.1 (Lemma 5.1 of [108]) Let  W .M; g/ ! .N; h/ be a pseudo-
Riemannian submersion endowed with a complex structure J on M which is
compatible with the decomposition (2.53). Then .M; g; J/ is a pseudo-Kähler
manifold if and only if the following equations3 are satisfied

H ..rX J/Y/ D H ..rV J/X/ D 0; (3.31)

.rUV J/V D V ..rX J/V/ D 0; (3.32)
AX . JY/  JAX Y D 0; AX . JV/  JAX V D 0; (3.33)
TV . JX/  JTV X D 0; TU . JV/  JTU V D 0; (3.34)

where X; Y are vector fields in H and U; V are vector fields in V:

Further we define a second complex structure by
(
J on H;
JO WD
J on V:

We observe that JOO D J: This construction was made in [98] for the Riemannian
setting and imitates the construction on twistor spaces, which was first done in [50].
Proposition 3.3.2 Suppose, that the foliation induced by the pseudo-Riemannian
submersion  is totally geodesic and that .M; J; g/ is a pseudo-Kähler manifold and
O is a
J is compatible with the decomposition (2.53), then the manifold .M; gO D g 1 ; J/
2
nearly pseudo-Kähler manifold. The distributions H and V are parallel with respect

2
The reference still is the author’s paper [108].
3
Please note, the small difference between the torsion T.X; Y/ of r N and the second fundamental
form TU V; which is not dangerous, as later we are considering totally geodesic fibrations.
3.3 Twistor Spaces over Quaternionic and Para-Quaternionic Kähler Manifolds 59

to the characteristic Hermitian connection rN of .M; gO ; J/:

O In other words the nearly
N
O has reducible r-holonomy.
pseudo-Kähler manifold .M; gO ; J/
Proof Let U; V be vector fields in V and X; Y be vector fields in H W In the following
rO is the Levi-Civita connection of gO : Since the fibres are totally geodesic, i.e. T 0;
we obtain from Eq. (2.54), that rO U V D rO UV V C TO U V D rUV V CTU V D rU V; which
yields .rO U J/V
O D .rU J/V D 0:
In the sequel we denote the O’Neill tensors of the pseudo-Riemannian foliations
induced by V on .M; g/ and on .M; gO / by A and A; O respectively. From Lemma 2.5.2
of Chap. 2 it follows AX Y D AO X Y and consequently the same Lemma yields rX Y D
rO X Y:
Since .M; g/ is Kähler, Lemma 3.3.1 implies A ı J D J ı A and we compute

.rO X J/Y
O D O  JO rO X Y
rO X .JY/ (3.35)
D H ŒrO X .JY/ C V ŒrO X .JY/  J.
O H .rO X Y/ C V .rO X Y//

D H ŒrO X .JY/  J rO X Y C V ŒrO X .JY/ C J rO X Y

D H ..rO X J/Y/ C AO X .JY/ C J AO X Y
(2.60);(2.62);(3.33) (3.31)
D H ..rX J/Y/ C 2AX . JY/ D 2AX . JY/ D 2JAX Y:

With the identity AX V D 2AO X V of Lemma 2.5.2 of Chap. 2 we get

.rO X J/V
O D O  JO rO X V
rO X .JV/ (3.36)
D V .rO X .JV//  H .rO X .JV// C JV .rO X V/  JH .rO X V/
D V ..rO X J/V/  AO X JV  J AO X V
(2.60);(2.62);(3.33)
D V ..rX J/V/  JAX V D AX JV:

The vanishing of the second fundamental form T; Eq. (2.61) and a second time
AX V D 2AO X V show

.rO V J/X
O D V .rO V .JX// C V .J rO V X/ C H .rO V .JX/  J rO V X/ (3.37)
1
D TO V . JX/ C J.TO V X/ C H ..rV J/X/ C . JAX V  AJX V/ D JAX V;
(2.63)
2
where we used AJX V D JAX V which follows, since AX is alternating (compare
Eq. (2.59)) and commutes with J: The next Lemma finishes the proof. t
u
Lemma 3.3.3 (Lemma 5.3 of [108])
O gO / is a nearly pseudo-Kähler manifold and JO is compatible
1) Suppose, that .M; J;
with the decomposition (2.53), then the following statements are equivalent:
N
(i) the splitting (2.53) is r-parallel,
60 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

(ii) the fundamental tensors AO and TO satisfy:

TO V X D 0; JO TO V W D TO V JW
O , JL TO V W D TO V JW OO (3.38)
L for JL D J;
1O O O 1  
AO X V D J. rX J/V; AO X Y D V J. O rO X J/Y
O : (3.39)
2 2

2) If it holds .rO V J/W

O D 0 then rN V W 2 V for V; W 2 V is equivalent to TV W D 0:
Moreover it is .rO VV J/W
O D 0:
We apply Proposition 3.3.2 to twistor spaces and obtain.
Corollary 3.3.4 The twistor space Z of a quaternionic Kähler manifold of dimen-
sion 4k with negative scalar curvature admits a canonical nearly pseudo-Kähler
structure of reducible holonomy contained in U.1/  U.2k/:
Proof We remark, that in negative scalar curvature the twistor space of a quater-
nionic Kähler manifold is the total space of a pseudo-Riemannian submersion with
totally geodesic fibres. It admits a compatible pseudo-Kähler structure of signature
.2; 4k/; cf. Besse [18, 14.86 b)]. The assumption of positive scalar curvature is
often made to obtain a positive definite metric on Z: Here we focus on pseudo-
Riemannian metrics and consequently on negative scalar curvature. t
u
Proposition 3.3.5 The twistor spaces Z of non-compact duals of Wolf spaces and
of Alekseevskian spaces admit a nearly pseudo-Kähler structure.
Proof Non-compact duals of Wolf spaces are known [117] to be quaternionic
Kähler manifolds of negative scalar curvature. The same holds for Alekseevskian
spaces [3, 38]. t
u
Studying the lists given in [3, 38, 117] we find examples of six-dimensional
nearly pseudo-Kähler manifolds.
Corollary 3.3.6 The twistor spaces Z of

Q 1 D Sp.1; 1/=Sp.1/Sp.1/ and SU.1; 2/=S.U.1/U.2//

HP

provide six-dimensional nearly pseudo-Kähler manifolds.

Remark 3.3.7 The situation in negative scalar curvature is more flexible than in the
positive case. This is illustrated by the following results in this area: In the main
theorem of [89] it is shown that the moduli space of complete quaternionic Kähler
metrics on R4n is infinite dimensional. A construction of super-string theory, called
the c-map [54], yields continuous families of negatively curved quaternionic Kähler
manifolds. Let us mention, that the c-map enjoys very recent interest [10, 73, 93]
in differential geometry. These results show that Corollary 3.3.4 is a good source of
examples.
3.3 Twistor Spaces over Quaternionic and Para-Quaternionic Kähler Manifolds 61

Another source of examples is given by twistor spaces over para-quaternionic

Kähler manifolds. Since these manifolds are less classical than quaternionic Kähler
manifolds, we recall some definitions (cf. [5] and references therein).
Definition 3.3.8 Let .1 ; 2 ; 3 / D .1; 1; 1/ or some permutation thereof. An
almost para-quaternionic structure on a differentiable manifold M 4k is a rank 3
sub-bundle Q  End .TM/; which is locally generated by three anti-commuting
endomorphism-fields J1 ; J2 ; J3 D J1 J2 : These satisfy Ji2 D i Id for i D 1; : : : ; 3:
Such a triple is called standard local basis of Q: A linear torsion-free connection
preserving Q is called para-quaternionic connection. An almost para-quaternionic
structure is called a para-quaternionic structure if it admits a para-quaternionic
connection. An almost para-quaternionic Hermitian structure .M; Q; g/ is a
pseudo-Riemannian manifold endowed with a para-quaternionic structure such that
Q consists of skew-symmetric endomorphisms. For n > 1 .M 4k ; Q; g/ is a para-
quaternionic Kähler manifold if Q is preserved by the Levi-Civita connection of g:
In dimension 4 a para-quaternionic Kähler manifold M 4 is an anti-self-dual Einstein
manifold.
We use the same notions omitting the word para for the quaternionic case. The
condition that Q is preserved by the Levi-Civita connection is in a given standard
local basis f Ji g3iD1 of Q equivalent to the equations

rX Ji D k .X/j Jj C j .X/k Jk ; for X 2 TM; (3.40)

where i; j; k is a cyclic permutation of 1; 2; 3 and fi g3iD1 are local one-forms. In

the context of para-quaternionic manifolds one can define twistor spaces for s D
1; 0; 1

Z s WD fA 2 Q j A2 D sId; with A ¤ 0g:

The case of interest in this text is Z D Z 1 ; since this twistor space is a complex
manifold, such that the conditions of Proposition 3.3.2 hold true (cf. [5]). Therefore
we obtain the following examples of nearly pseudo-Kähler manifolds.
Corollary 3.3.9 The twistor space Z of a para-quaternionic Kähler manifold with
non-zero scalar curvature of dimension 4k admits a canonical nearly pseudo-Kähler
structure of reducible holonomy contained in U.k; k/  U.1/:
Example 3.3.10 The para-quaternions H e are the R-algebra generated by f1; i; j; kg
subject to the relations i2 D 1; j2 D k2 D 1; ij D ji D k: Like the quaternions,
the para-quaternions are a real Clifford algebra which in the convention of [88] is
e D Cl1;1 Š Cl0;2 Š R.2/: One defines the para-quaternionic projective space HP
H e n
by the obvious equivalence relation on the para-quaternionic right-module He nC1
of
e n is a para-quaternionic Kähler
.n C 1/-tuples of para-quaternions. The manifold HP
manifold [21] in analogue to the quaternionic projective space HPn : This yields
examples of the type described in the last Corollary.
62 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds

Motivation In this section we study the case of a nearly pseudo-Kähler manifold

.M 2n ; J; g/; such that the holonomy of the characteristic connection rN is reducible,
in the sense that the tangent bundle TM admits a splitting

TM D H ˚ V

N
into two r-parallel sub-bundles H; V; which are orthogonal and invariant with
respect to the almost complex structure J: We refer to this situation as complex
reducible. This is motivated by the examples on twistor spaces given in the last
section. In Sect. 3.9.4 we see, that real reducible nearly Kähler manifolds are locally
homogeneous.

3.4.1 General Properties

In this subsection we carefully check, generalising [99] to pseudo-Riemannian

foliations, the information which follows from the decomposition into the J-
invariant sub-bundles.
Lemma 3.4.1 (Lemma 6.1 of [108]) In the situation of this section and for a vector
field X in H; a vector field Y in TM and vector fields U; V in V it is

N
R.X; Y; U; V/ D g .ŒrU J; rV JX; Y/  g ..rX J/Y; .rU J/V/ : (3.41)

Corollary 3.4.2 For vector fields X; Y in H and V; W in V one has

(i) .rX J/.rV J/W D 0I .rV J/.rX J/Y D 0I
(ii) .rX J/.rY J/Z belongs to H for all Z 2 .H/I
(iii) .rV J/.rW J/X belongs to HI and .rX J/.rY J/V belongs to V:
Proof
N JX; JY; V; W/ D R.X;
(i) follows from the fact, that R. N Y; V; W/ and that the first
term of Eq. (3.41) has the same symmetry with respect to J: This yields on the
one hand

g ..rJX J/JY; .rV J/W/ D g ..rX J/Y; .rV J/W/

and on the other hand it is

g ..rJX J/JY; .rV J/W/ D g ..rX J/Y; .rV J/W/ :

3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds 63

Consequently one has g ..rX J/Y; .rV J/W/ D 0: Exchanging H and V finishes
part (i).
(ii) From (i) one gets the vanishing of

g..rV J/.rY J/Z; X/ D g.Z; .rY J/.rV J/X/

D g.Z; .rY J/.rX J/V/ D g..rX J/.rY J/Z; V/:

(iii) From (i) it follows 0 D R.X; N U; V; W/ D g.ŒrV J; rW JX; U/: This yields
ŒrV J; rW JX 2 H and by ŒrV J; rJW JJX D frV J; rW JgX 2 H we get the
first part. The second part follows by replacing H and V:
t
u

3.4.2 Co-dimension Two

Motivated by the above section on twistor spaces we suppose from now on that the
real dimension of V is two.
Lemma 3.4.3 (Lemma 6.2 of [108]) Let dimR .V/ D 2:
(i) Then the restriction of the metric g is either of signature .2; 0/ or .0; 2/:
(ii) a) T.V; W/ D 0 for all V; W 2 V:
b) T.X; U/ 2 H for all X 2 H and U 2 V:
c) In dimension 6 it is T.X; Y/ 2 V for all X; Y 2 H:
d) SpanfV .T.X; Y// j X; Y 2 Hg D V:
Corollary 3.4.4 Let dimR .V/ D 2: Then the foliation V has totally geodesic fibres
and the O’Neill tensor is given by AX Y D 12 V . J.rX J/Y/ and AX V D 12 J.rX J/V:
Moreover it is r V J D 0:
Proof From Lemma 3.4.3 (ii) a) we obtain .rV J/W D 0 with V; W 2 .V/: By
Lemma 3.3.3 part 2) it follows TV W D 0 and r V J D 0; since the decomposition
H ˚ V is rN parallel. Part 1) of Lemma 3.3.3 finishes the proof. t
u
Proposition 3.4.5 Let .M; J; g/ be a nearly pseudo-Kähler manifold such that the
property of Lemma 3.4.3 (ii) c) is satisfied and such that V has dimension 2, then
.M; JL D J;
O gL D g2 / is a pseudo-Kähler manifold.4
It is natural to suppose the property of Lemma 3.4.3 (ii) c), since this holds true in
the cases of twistorial type which are studied in the next sections.
Proof By the last Corollary the data of the submersion is TL D T 0;
AX Y D AL X Y D 12 V . J.rX J/Y/ and AL X V D 2AX V D J.rX J/V: Since A anti-
commutes with J it commutes with J:L This yields the conditions (3.33) and (3.34) of

4
Here we use L for the inverse construction of O:
64 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

L T;
Lemma 3.3.1 on the triple A; L J:
L Further it holds r V J D 0: From the reasoning of
Eq. (3.35) we obtain H ..rL X J/Y/
L D H ..rX J/Y/ which vanishes by the property
of Lemma 3.4.3 (ii) c). By an analogous argument we get from Eq. (3.36) the
identity V ..rL X J/V/
L D V ..rX J/V/: This vanishes by Lemma 3.4.3 (ii) b).
From Eq. (3.37) we derive H ..rX J/V/ D H ..rV J/X/ D H ..rL V J/X/ L
n:K:
C
2H . JAX V/: The definition of AX V yields H ..rL V J/X/
L D 0: These are all the
identities needed to apply Lemma 3.3.1. t
u
Proposition 3.4.6 Let X; Y be vector fields in H and V1 ; V2 ; V3 be vector fields in
V: Suppose that it holds T.V; W/ D 0 for all V; W 2 V then it is

N
R..rX J/JY; V1 ; V2 ; V3 / D g. JY; ŒrV1 J; ŒrV2 J; rV3 JX/: (3.42)

Moreover, one has rN U R.V

N 1 ; V2 ; V3 ; V4 / D 0:

Proof For V1 ; V2 ; V3 2 V and X 2 H the second Bianchi identity gives

  rN X .R/.
N Y; V1 ; V2 ; V3 / D  R..r
N X J/JY; V1 ; V2 ; V3 /:
XYV1 XYV1

N
As the decomposition H ˚V is r-parallel the terms on the left hand-side vanish due
N The right hand-side is determined
to the symmetries (3.24) of the curvature tensor R:
N
with the help of Lemma 3.4.1 and Corollary 3.4.2.
If we apply r to the for-
N N N
mula (3.42) we obtain by r.rJ/ D 0 the identity g rU .R/.V1 ; V2 ; V3 /; .rX J/Z D
0 with Z D JY: This yields the proposition using Lemma 3.4.3 (ii) part d). t
u

3.4.3 Six-Dimensional Nearly Pseudo-Kähler Manifolds

Before analysing the general case we first focus on dimension 6.

Lemma 3.4.7 (Lemma 6.7 of [108]) On a six-dimensional nearly pseudo-Kähler
manifold .M 6 ; J; g/ the integral manifolds of the foliation V have Gaussian curva-
ture 4˛ and constant curvature  D 4˛; where ˛ is the type constant.
Let us recall that the sign of ˛ is completely determined by the signature of the
metric g; cf. Remark 3.1.9.
Proposition 3.4.8 The manifold .M; J; g/ is the total space of a pseudo-
Riemannian submersion  W .M; g/ ! .N; h/ where .N; h/ is an almost
pseudo-Hermitian manifold and the fibres are totally geodesic Hermitian symmetric
spaces. In particular, the fibres are simply connected.
Proof The foliation which is induced by V is totally geodesic and each leaf is by
Proposition 3.4.6 a locally Hermitian symmetric space of complex dimension 1.
It is shown in Lemma 3.4.7 that each leaf has constant curvature : In the case
 > 0 the leaves are compact and we can apply a result of Kobayashi, cf. [18,
3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds 65

11.26], to obtain that the leaves are simply connected. Since the leaves are also
simply connected it follows, that the leaf holonomy is trivial and that the foliation
comes from a (smooth) submersion (cf. p. 90 of [113]). In the case  < 0 we
observe, that .M; J; g/ is a nearly pseudo-Kähler manifold of constant type ˛:
The same argument shows that the fibres are simply connected. t
u
Lemma 3.4.9 (Lemma 6.9 of [108]) Let .M 6 ; g; J/ be a strict nearly pseudo-
Kähler six-manifold of constant type ˛: For an arbitrary normalised5 local vector
field V 2 V; i.e. V D g.V; V/ 2 f˙1g; we consider the endomorphisms JQ1 WD JjH ;
p 1
JQ 2 W H 3 X 7! j˛j .rV J/X 2 H and JQ3 D JQ1 JQ2 : Then the triple .JQ1 ; JQ2 ; JQ3 /
defines an "-quaternionic triple on H with 1 D 1 and 2 D 3 D sign.˛V /
and it is

 H Œ.r JQ i /Y D k . / j JQ j Y C j . / k JQ k Y;

N
p perm. of i,j,k and with 1 . / Dp sign.˛/g. JV; r V/; 2 . / D
for a cyclic
sign.˛/ j˛jg.V; J / and 3 . / D sign.˛/ j˛jg.V; /: The sub-bundle of
endomorphisms spanned by .JQ1 ; JQ2 ; JQ3 / does not depend on the choice of V:
Lemma 3.4.10 Let .M 6 ; g; J/ be a strict nearly pseudo-Kähler six-manifold of
constant type ˛: Let s W U  N ! M be a (local) section6 of  on some open
set U: Define  by
 s
 D s ı  W Hs.n/ ! Tn N ! s .Tn N/  Ts.n/ M; for n 2 N

and set Jijn WD  ı JQijs.n/ ı .jH /1 for i D 1; : : : ; 3; where JQ i are defined in
Lemma 3.4.9. Then . J1 ; J2 ; J3 / defines a local -quaternionic basis preserved by the
Levi-Civita connection r N of N:
Proof We choose U such that the section s is a diffeomorphism onto W D s.U/ and
a vector field V in V defined on a subset containing W: As  is a pseudo-Riemannian
submersion we obtain from  ı s D Id that s is an isometry from U onto W:
Therefore it holds s .rXN Y/ D  s TN Œrs X s Y which yields rXN Y D  .rs X s Y/
and

.jH /1 .rXN Y/ D  H .rs X s Y/: (3.43)

For convenience let us identify U and W or in other words consider s as the inclusion
W  M: Then the projection on s TN is  D s  D  jH: Moreover, we need
the (tensorial) relation7

rXN . Z/    H .rXM Z/ D 0 or equivalently rXN ZQ    H .rXM  1 Z/

Q D 0;

5
Constant non-zero length suffices.
6
Local sections exist, since  is locally trivial [18, 9.3].
7
Here ZQ is the horizontal lift of Z:
66 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

which can be directly checked for basic vector fields. Using this identity we get for
i D 1; : : : ; 3

rXN . Ji Y/ D rXN . JQ i  1 Y/ D rXM .JQi  1 Y/ D  .rXM JQi /  1 Y C  JQi rXM . 1 Y/

D  .rXM JQi /  1 Y C  JQi  1 rXN Y D  .rXM JQ i /  1 Y C Ji rXN Y;

which reads .rXN Ji /Y D  .rXM JQ i /  1 Y: This finishes the proof, since the right
hand-side is completely determined by Lemma 3.4.9. Therefore we have checked
the condition (3.40), i.e. the manifold N is endowed with a parallel skew-symmetric
(para-)quaternionic structure, see also [18, 10.32 and 14.36]. t
u

3.4.4 General Dimension

In the last section we have seen that in dimension 6 the tensor rV J induces a (para-
)complex structure on H: This motivates the following definition.
Definition 3.4.11 The foliation induced by TM D H ˚ V is called of twistorial
type if for all p 2 M there exists a V 2 Vp such that the endomorphism

rV J W Hp ! Hp

is injective.
Obviously, if rV J defines a (para-)complex structure, then the foliation is of
twistorial type.
Proposition 3.4.12
(a) If the metric induced on H is definite, then the foliation is of twistorial type.
(b) If the foliation is of twistorial type, then for all p 2 M and all 0 ¤ U 2 Vp the
endomorphism

rU J W Hp ! Hp

is injective.
(c) It holds with A WD rV J for some vector field V in V of constant length and for
vector fields X 2 H and 2 TM

rN .A2 /X D 0: (3.44)

Further it holds ŒA2 ; .rU J/ D 0 for all U 2 V and

rU .A2 /X D 0 (3.45)

for vector fields U in V:

3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds 67

Proof Part (a) follows from .rV J/X 2 H for X 2 H and V 2 V; cf. Lemma 3.4.3
(i). For (b) we observe, that if rV J is injective so is rJV J D JrV J: As V is of
dimension 2 fV; JVg with V ¤ 0 is an orthogonal basis. With a; b 2 R it follows
g..arV J C brJV J/X; .arV J C brJV J/X/ D .a2 C b2 / g..rV J/X; .rV J/X/; which
yields, that raVCbJV J W Hp ! Hp is injective since a ¤ 0 or b ¤ 0: It remains to
prove part (c). We first observe, that, since V has constant length and since rN is a
metric connection and preserves V; it follows rN V D ˛. /JV for some one-form
N
˛: From r.rJ/ D 0 we obtain

.rN A/X D .rN .rV J//X D .rrN V J/X D ˛. /.rJV J/X D ˛. /JAX

and we compute using fA; Jg D 0

rN .A2 /X D A.rN A/X C .rN A/AX D ˛. /ŒA. J.AX// C JA2 X D 0:

The equation ŒA2 ; .rU J/ D 0 is tensorial in U and holds true for U D V: Therefore
we only need to compute ŒA2 ; .rJV J/ D ŒA2 ; J.rV J/ D JŒA2 ; .rV J/ D 0;
where we used that A2 commutes with J: This implies

1 1
rU .A2 /X D rN U .A2 /X C Œ J.rU J/; A2 X D  Œ.rJU J/; A2 X D 0
2 2
and proves part (c). t
u
In the following V is a local vector field of constant length V D g.V; V/ 2 f˙1g:
We denote by  the curvature form of the connection induced by rN on the
(complex) line bundle V; which is given by

N
R.X; Y/V D .X; Y/JV; for X; Y 2 TM; V 2 V:

Proposition 3.4.13 If the foliation is of twistorial type,

(i) then the endomorphism A WD rV JjH satisfies A2 D V IdH for some real
constant  ¤ 0 and

 D 2.2! V  ! H /;

where ! H .X; Y/ D g.X; JY/ is the restriction of the fundamental two-form !

to HI
(ii) for X; Y in H it is .rX J/Y 2 V:
The proof of this proposition is divided in several steps.
Lemma 3.4.14
N
(i) For X; Y in H and V in V it is R.X; Y; V; JV/ D 2g..rV J/2 X; JY/:
N
(ii) For a given X in H and V in V it follows R.X; V; V; JV/ D 0:
68 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proof
N
(i) Since H is r-parallel N
we obtain, that  R.X; N
Y; V; JV/ D R.X; Y; V; JV/: This
XYV
is the left hand-side of the first Bianchi identity (3.26) . The right hand-side
reads

  g..rX J/Y; .rV J/JV/ D g..rV J/X; .rY J/JV/  g..rY J/V; .rX J/JV/
XYV

D 2g..rV J/2 X; JY/:

(ii) From the symmetries (3.24) of the curvature tensor RN it follows R.X;
N V; V; JV/
N N
D R.V; JV; X; V/: This expression vanishes since H is r-parallel.
t
u
From the last lemma we derive the more explicit expression of the curvature form
.; /:

.; / D f ! V .; / C V ˛.; /; (3.46)

where f is a smooth function, ! V is the restriction of the fundamental two-form

! D g.; J/ to V and ˛.X; Y/ D 2g.A2 X; JY/:
Lemma 3.4.15 (Lemma 6.15 of [108]) It holds with U 2 V and X; Y 2 H W

d! V .X; U; JU/ D 0; (3.47)

d˛.X; U; JU/ D 0; (3.48)
V
d! .U; X; Y/ D g.rU J/X; Y/; (3.49)
d˛.U; X; Y/ D 4g.A2 .rU J/X; Y/: (3.50)

Proof of the Proposition 3.4.13

(i) Let X; Y be vector fields in H and V be a local vector field in V of constant
length. Since  as a curvature form of a (Hermitian) line bundle is closed,
we obtain from Eq. (3.46) V d˛.; ; / D fd! V .; ; / C df ^ ! V .; ; /:
Equations (3.47) and (3.48) imply dfjH D 0: This implies ŒX; Yf D 0 and using
N
that H is r-parallel we obtain .rN X Y/f D 0 D .rN Y X/f which yields finally
0 D T.X; Y/. f / D ΠJ.rX J/Y. f /: By Lemma 3.4.3 (ii) d) the last equation
shows dfjV D 0: Since M is connected, it follows f  for a constant :
Again using d.V; X; Y/ D 0 Eqs. (3.49) and (3.50) yield for arbitrary X; Y

g..rV J/X; Y/ C 4V g.A2 .rV J/X; Y/ D 0:

This implies .rV J/.IdH C 4V A2 / D 0: It follows


A2 D V IdH ;
4
3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds 69

since the foliation is of twistorial type. If we set 4˛ D  in analogue to

dimension 68 one gets A2 D V ˛IdH :
(ii) Since  is closed, it follows from part (i) and d! V .X; Y; Z/ D 0 for X; Y; Z 2 H
that it is d! H .X; Y; Z/ D 0: Using d! H .X; Y; Z/ D 3g..rX J/Y; Z/ yields part
(ii).
t
u
Proposition 3.4.16 Let .M 4kC2 ; g; J/ be a strict nearly pseudo-Kähler manifold of
twistorial type. Let s W U  N ! M be a (local) section of  on some open set U:
Define  by
 s
 D s ı  W Hs.n/ ! Tn N ! s .Tn N/  Ts.n/ M; for n 2 N

and set Ji jn WD  ı JQi js.n/ ı .jH /1 for i D 1; : : : ; 3; where JQ i are defined in
Lemma 3.4.9. Then . J1 ; J2 ; J3 / defines a local -quaternionic basis preserved by the
Levi-Civita connection r N of N:
Proof The proof of Lemma 3.4.9 only uses A2 D V 4 Id and .rX J/Y 2 V for
X; Y 2 H: Therefore we can generalise it by means of Proposition 3.4.13 to strict
nearly pseudo-Kähler manifolds of twistorial type. t
u
Corollary 3.4.17
(i) The tensor r has exactly two eigenvalues. More precisely, it has the eigenvalue
 on H and the eigenvalue V 2 .n  1/ on V with  D 4˛:
(ii) The Ricci-tensor has exactly two eigenvalues. More  precisely, it has the
eigenvalue 4 .V .n1/C3/ on H and the eigenvalue  V .n1/

8 C 1 on V: The
base manifold .N; h/ is an Einstein manifold with Einstein constant 4 V .n  1/:
Proof By definition we have

2n
X 2n
X
g.rX; Y/ D i g..rX J/ei ; .rY J/ei / D  i g..rY J/.rX J/ei ; ei /
iD1 iD1

for some pseudo-orthogonal basis with e1 ; : : : ; e2n2 2 H and e2n1 ; e2n 2 V. For
V 2 V with g.V; V/ D V we get

2n
X 2n2
X
g.rV; V/ D  i g..rV J/.rV J/ei ; ei / D  i g.A2 ei ; ei /
iD1 iD1
2n2
 X 
D V i g.ei ; ei / D V .n  1/:
4 iD1 2

8
Without risk of confusion we use the same latter for the constant ˛ as for the two-form ˛.; /.
70 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Let us now consider X 2 H and V as before and compute

2n2
X
g.rX; X/ D 2V g ..rV J/X; .rV J/X/ C i g..rX J/ei ; .rX J/ei /:
iD1

Since it is .rX J/ei 2 V; we get

i g..rX J/ei ; .rX J/ei / D i V .g..rX J/ei ; V/2 C g..rX J/ei ; JV/2 /

and for the sum this gives

2n2
X
g.rH X; X/ D i g..rX J/ei ; .rX J/ei /
iD1

2n2
X

D V i g..rX J/V; ei /2 C g..rX J/JV; ei /2
iD1

D 2V g ..rV J/X; .rV J/X/ D g.X; X/:
2

Summarizing it follows g.rX; X/ D 4V g ..rV J/X; .rV J/X/ D : This shows part
(i).
The statement (ii) follows from (i) using Lemma 3.2.7. Namely, for X; Y 2 H it
is
 1 
g.Ric.X/; Y/ D g.X; Y/CV .n1/ g.rV X; Y/ Cg.rH X; Y/ D .V .n1/C3/g.X; Y/;
4 2 „ ƒ‚ … 4

2 g.X;Y/

since it is using A2 D V 4 IdH

g.rV X; Y/ D trV ..rX J/ ı .rY J// D 2V g..rX J/.rY J/V; V/

D 2V g..rY J/V; .rX J/V/ D 2V g..rV J/Y; .rV J/X/

D 2V g.AY; AX/ D g.X; Y/:
2
Further, for U; V 2 V it is

.n  1/ 2
g.Ric.U/; V/ D V g.U; V/ C V  g.rH U; V/
8 .n  1/ „ ƒ‚ …
V 2 .n1/g.U;V/
 
.n  1/
D  V C 1 g.U; V/;
8
3.4 Complex Reducible Nearly Pseudo-Kähler Manifolds 71

where
2n2
X
g.rH V; V/ D trH ..rV J/ ı .rV J// D  i g..rV J/.rV J/ei ; ei /
iD1
 
D 2.n  1/ V D V .n  1/:
4 2
The last statement follows from O’Neill’s formula with the information, that the
O’Neill tensor is AX Y D 12 J.rX J/Y; c.f. Lemma 3.3.3. u
t

3.4.5 The Twistor Structure

In this subsection we finally characterise the nearly pseudo-Kähler structures, which

are related to the canonical nearly Kähler structure of twistor spaces.
Theorem 3.4.18 Suppose, that .M 2n ; J; g/ is a complex reducible nearly pseudo-
Kähler manifold of twistorial type, then one has:
(i) The manifold .M; J D J; L gL D g2 / is a twistor space of a quaternionic pseudo-
Kähler manifold, if it is V  > 0:
(ii) The manifold .M; J D J; L gL D g2 / is a twistor space of a para-quaternionic
Kähler manifold, if it is V  < 0:
Proof Denote by  Z W Z ! N the twistor space of the manifold N endowed
with the parallel skew-symmetric (para-)quaternionic structure constructed from the
foliation  W M ! N of twistorial type, cf. Proposition 3.4.9 for dimension 6 and
Proposition 3.4.16 for general dimension. We observe that the restriction of J to H
yields a (smooth) map

' W M ! Z; m 7! dm ı Jm jH ı .dmjH /1 DW j.m/ ;

which by construction satisfies  Z ı ' D  and as a consequence d Z ı d' D

d: Since  and  Z are pseudo-Riemannian submersions, the last equation implies
that d' induces an isometry of the according horizontal distributions and maps the
vertical spaces into each other. Let us determine the differential of ' on V:
Claim: For V 2 V one has

d'.V/ D 2 d ı .rV J/ ı .djH /1 ;

d'. JV/ D 2 d ı .rJV J/ ı .djH /1 D 2 d ı J.rV J/ ı .djH /1 :
72 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

To prove the claim we consider a (local) vector field V 2 V and a (local) integral
curve  of V on some interval I 3 0 with .0/ D m: Let X be a vector field in
N: Denote by XQ the horizontal lift of X: The Lie transport of XQ along the vertical
Q D X for all t 2 I and in consequence
curve  projects to X; i.e. it holds d.t/ .X/

1
d.t/ jH Q In other words d commutes with this Lie transport, which
X D X:
implies

Q
d'.V/X D d..LV J/X/;

as one directly checks using basic vector fields. Therefore we need to determine the
Lie-derivative L of J W

Q D  H .ŒV; J X
 H ..LV J/X/ Q  JŒV; X/
Q


Q  r Q V  JrV XQ C JrXQ V
D  H rV . J X/ JX
 
1 1

D  H .rV J/XQ  J.rJ XQ J/V C J J.rXQ J/ V D 2.rV J/X:
Q
2 2

This shows d'.V/ D 2 d ı .rV J/ ı .djH /1 ; which implies d'. JV/ D 2 d ı

.rJV J/ ı .djH /1 D 2 d ı J.rV J/ ı .djH /1 : Given a local section s W N ! M
and the associated adapted frame of the (para-)quaternionic structure it follows that
' ı s is J1 ; d'.V/ is related to J2 and d'. JV/ to J3 which span the tangent space
of the fibre F.m/ D S2 in '.m/: The complex structure of Z maps J2 to J3 : Hence
d' is complex linear for the opposite complex structure JL on M: Further one sees in
this local frame that ' maps horizontal part into horizontal part. Therefore ' is an
isometry for the metric gL D g2 ; where the parameter in the canonical variation of
the metric g is t D 2: This means that .M; J; L gL D g2 / is isometrically biholomorph
to Z: t
u
Combining Theorems 3.2.10 and 3.4.18 we obtain the following result.
Theorem 3.4.19 Let .M 10 ; J; g/ be a nice decomposable nearly pseudo-Kähler
manifold, then the universal cover of M is either the product of a pseudo-Kähler
surface and a (strict) nearly pseudo-Kähler manifold M 6 or a twistor space of an
eight-dimensional (para-)quaternionic Kähler manifold endowed with its canonical
nearly pseudo-Kähler structure.

3.5 A Class of Flat Pseudo-Riemannian Lie Groups

In this section we consider flat pseudo-Riemannian Lie groups which are closely
related to nearly Kähler geometry (cf. Sect. 3.6). These geometric objects are also
of independent interest [13]. Let V D .Rn ; h; i/ be the standard pseudo-Euclidian
vector space of signature .k; l/, n D k C l: Using the (pseudo-Euclidian) scalar
product we shall identify V Š V  and ƒ2 V Š so.V/. These identifications provide
3.5 A Class of Flat Pseudo-Riemannian Lie Groups 73

the inclusion ƒ3 V  V  ˝ so.V/. Using it we consider a three-vector  2 ƒ3 V as

an so.V/-valued one-form. Further we denote by X 2 so.V/ the evaluation of this
one-form on a vector X 2 V. Let us recall, that the support of  2 ƒ3 V is defined by

† WD spanfX Y j X; Y 2 Vg  V: (3.51)

Theorem 3.5.1 Each

[
 2 C.V/ WD f 2 ƒ3 V j† (totally) isotropicg D ƒ3 L
LV

defines a 2-step nilpotent simply transitive subgroup L./  Isom.V/; where the
union runs over all maximal isotropic subspaces. The subgroups L./, L.0 / 
Isom.V/ associated to ; 0 2 C.V/ are conjugated if and only if 0 D g   for some
element of g 2 O.V/:
2 any three-vector  2 ƒ3 V satisfies  2 ƒ3 † :
Proof Using Lemma 2.3.2 of Chap. S
This implies the equation C.V/ D ƒ3 L. Let an element  2 C.V/ be given. By
LV
Lemma 2.3.4 of Chap. 2 its support † is isotropic if and only if the endomorphisms
X 2 so.V/ satisfy X ı Y D 0 for all X; Y 2 V: The 2-step nilpotent group
    ˇ
 X Id C X X ˇ
L./ WD exp X D ˇX2V
0 0 0 1 ˇ

acts simply transitively on V Š V  f1g  V  R by isometries:

    
Id C X X 0 X
D :
0 1 1 1

Let us consider next ; 0 2 C.V/, g 2 O.V/. The computation

 ˇ  ˇ
Id C gX g1 gX ˇ Id C gg1 Y g1 Y ˇˇ
gL./g 1
D ˇX2V D Y 2V
0 1 ˇ 0 1 ˇ

shows that gL./g1 D L.0 / if and only if 0X D .g  /X D g g1 X g1 for all
X 2 V. t
u
Let L  Isom.V/ be a simply transitive group. Pulling back the scalar product on
V by the orbit map L 3 g 7! g0 2 V yields a left-invariant flat pseudo-Riemannian
metric h on L. A pair .L; h/ consisting of a Lie group L and a flat left-invariant
pseudo-Riemannian metric h on L is called a flat pseudo-Riemannian Lie group.
74 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Theorem 3.5.2
(i) The class of flat pseudo-Riemannian Lie groups .L./; h/ defined in Theo-
rem 3.5.1 exhausts all simply connected flat pseudo-Riemannian Lie groups
with bi-invariant metric.
(ii) A Lie group with bi-invariant metric is flat if and only if it is 2-step nilpotent.
Proof
(i) The group L./ associated to a three-vector  2 C.V/ is diffeomorphic to Rn by
the exponential map. We have to show that the flat pseudo-Riemannian metric
h on L./ is bi-invariant. The Lie algebra of L./ is identified with the vector
space V endowed with the Lie bracket

ŒX; Y WD X Y  Y X D 2X Y; X; Y 2 V:

The left-invariant metric h on L./ corresponds to the scalar product h; i on

V. Since  2 ƒ3 V, the endomorphisms X D 12 adX are skew-symmetric. This
shows that h is bi-invariant.
Conversely, let .V; Œ; / be the Lie algebra of a pseudo-Riemannian Lie group
of dimension n with bi-invariant metric h. We can assume that the bi-invariant
metric corresponds to the standard scalar product h; i of signature .k; l/ on V.
Let us denote by X 2 so.V/, X 2 V, the skew-symmetric endomorphism of
V which corresponds to the Levi-Civita covariant derivative DX acting on left-
invariant vector fields. From the bi-invariance and the Koszul formula we obtain
that X D 12 adX and, hence, R.X; Y/ D  14 adŒX;Y for the curvature. The last
formula shows that h is flat if and only if the Lie group is 2-step nilpotent. This
proves (ii). To finish the proof of (i) we have to show that, under this assumption,
 is completely skew-symmetric and has isotropic support. The complete skew-
symmetry follows from X D 12 adX and the bi-invariance. Similarly, using the
bi-invariance, we have

4hX Y; Z Wi D hŒX; Y; ŒZ; Wi D hY; ŒX; ŒZ; Wi D 0;

since the Lie algebra is 2-step nilpotent. This shows that † is isotropic.
t
u
Corollary 3.5.3 With the above notations, let L  V be a maximally isotropic
subspace. The correspondence  7! L./ defines a bijection between the points of
the orbit space ƒ3 L=GL.L/ and isomorphism classes of pairs .L; h/ consisting of a
simply connected Lie group L endowed with a flat bi-invariant pseudo-Riemannian
metric h of signature .k; l/.
Corollary 3.5.4 Any simply connected Lie group L with a flat bi-invariant metric h
of signature .k; l/ contains a normal subgroup of dimension  max.k; l/  12 dim V
which acts by translations on the pseudo-Riemannian manifold .L; h/ Š Rk;l .
3.5 A Class of Flat Pseudo-Riemannian Lie Groups 75

Proof Let a WD ker.X 7! X /  V be the kernel of . Then a D †?

 is co-isotropic
and defines an Abelian ideal a  l WD Lie L Š V Š R . The corresponding
k;l

normal subgroup A  L D L./ is precisely the subgroup of translations. So we

have shown that dim A  max.k; l/  12 dim V. t
u
Remarks 3.5.5
1) The number dim † is an isomorphism invariant of the groups L D L./, which
is independent of the metric. We will denote it by s.L/. Let L3  L4      L
be a filtration, where dim Lj D j runs from 3 to dim L. The invariant dim †
defines a decomposition of ƒ3 L=GL.L/ as a union

[L
dim
f0g [ ƒ3reg Lj =GL.Lj /;
jD3

where ƒ3reg Rj  ƒ3 Rj is the open subset of 3-vectors with j-dimensional

support. The points of the stratum ƒ3reg Lj =GL.Lj / Š ƒ3reg Rj =GL. j/ correspond
to isomorphism classes of pairs .L; h/ with s.L/ D j.
2) Since in the above classification † is isotropic, it is clear that a flat (or 2-step
nilpotent) bi-invariant metric on a Lie group is indefinite, unless  D 0 and the
group is Abelian. It follows from Milnor’s classification of Lie groups with a flat
left-invariant Riemannian metric [95] that a 2-step nilpotent Lie group with a flat
left-invariant Riemannian metric is necessarily Abelian.
Since a nilpotent Lie group with rational structure constants has a (co-compact)
lattice [94], we obtain.
Corollary 3.5.6 The groups .L./; h/ admit lattices   L./, provided that  has
rational coefficients with respect to some basis. M D M.; / WD  n L./ is a flat
compact homogeneous pseudo-Riemannian manifold. The connected component
of the identity in the isometry group of M is the image of the natural group
homomorphism  from L./ into the isometry group of M.
Proof First we remark that the bi-invariant metric h induces an L./-invariant
metric on the homogeneous space M D  nL./. Let G be the connected component
of the identity in the isometry group of .L./; h/ Š Rk;l . The connected component
of the identity in the isometry group of M is the image of the centraliser ZG ./ of
 in G under the natural homomorphism ZG ./ ! Isom.M/. Now the statement
about the isometry group follows from the fact that the centraliser of the left-action
of   L./ on L./ is precisely the right-action of L./ on L./, since   L./
is Zariski-dense, see Theorem 2.1 of [101] . t
u
76 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.6 Classification Results for Flat Nearly "-Kähler Manifolds

3.6.1 Classification Results for Flat Nearly Pseudo-Kähler

Manifolds

In this section we denote by Ck;l the complex vector space .Cn ; Jcan /, n D k C l,
endowed with the standard Jcan -invariant pseudo-Euclidean scalar product gcan of
signature .2k; 2l/.
Let .M; g; J/ be a flat nearly pseudo-Kähler manifold. Then there exists for each
point p 2 M an open set Up  M containing the point p, a connected open set U0 of
Ck;l containing the origin 0 2 Ck;l and an isometry

Q 0 ; gcan /;
ˆ W .Up ; g/!.U

such that at the point p we have:

ˆ Jp D Jcan ˆ :

In other words, we can suppose, that locally M is a connected open subset of Ck;l
containing the origin 0 and that g D gcan and J0 D Jcan :
Proposition 3.6.1 Let .M; g; J/ be a flat nearly pseudo-Kähler manifold. Then
1) X ı Y D 0 for all X; Y;
2) r D rN D 0:

Proof From the curvature identity (3.15) we have for X; Y; Z; W 2 TM

0 D R.W; X; Y; Z/  R.W; X; JY; JZ/ D g..rX J/Y; .rZ J/W/ D g..rZ J/.rX J/Y; W/
D g. J.rZ J/J.rX J/Y; W/ D 4g.Z X Y; W/:

This shows X ı Y D 0 for all X; Y 2 TM and finishes the proof of part 1). The
N D 0: In fact, one has
second part follows from 1) and r

.rX /Y D .rN X /Y C X Y C ŒX ; Y  D 0; for X; Y 2 TM;

which shows part 2). t

u
From Theorem 3.1.5 and Proposition 3.6.1 we obtain.
Corollary 3.6.2 Let M  Ck;l be an open neighborhood of the origin endowed with
a nearly pseudo-Kähler structure .g; J/ such that g D gcan and J0 D Jcan . Then the
.1; 2/-tensor

1
 WD JrJ
2
3.6 Classification Results for Flat Nearly "-Kähler Manifolds 77

defines a constant three-form on M  Ck;l D R2k;2l defined by

.X; Y; Z/ WD g.X Y; Z/

satisfying
(i) X Y D 0; 8X; Y 2 TM;
(ii) fX ; Jcan g D 0; 8X 2 TM:
Conversely, we have the next Lemma.
Lemma 3.6.3 (Lemma 5 of [41]) Let  be a constant three-form on an open
connected neighbourhood M  Ck;l of 0 satisfying (i) and (ii) of Corollary 3.6.2.
Then there exists a unique almost complex structure J on M such that
a) J0 D Jcan ;
b) fX ; Jg D 0; 8X 2 TM;
c) DJ D 2J;
where D stands for the Levi-Civita connection of the pseudo-Euclidian vector space
Ck;l . With rN WD D   and assuming b), the last equation is equivalent to
N D 0:
c’) rJ
More precisely, the almost complex structure is given by the formula

2n
!
X
J D exp 2 x @i Jcan
i

iD1
2n
!
.i/ X
D Id C 2 xi @i Jcan ; (3.52)
iD1

@
where xi are linear coordinates of Ck;l D R2k;2l D R2n and @i D @xi
:
Theorem 3.6.4 Let  be a constant three-form on a connected open set U  Ck;l
containing 0 which satisfies (i) and (ii) of Corollary 3.6.2. Then there exists a unique
almost complex structure given by Eq. (3.52) on U such that
a) J0 D Jcan ;
b) M.U; / WD .U; g D gcan ; J/ is a flat nearly pseudo-Kähler manifold.
Any flat nearly pseudo-Kähler manifold is locally isomorphic to a flat nearly
pseudo-Kähler manifold of the form M.U; /:
Now we discuss the general form of solutions of (i) and (ii) of Corollary 3.6.2. In
the following we shall freely identify the real vector space V WD Ck;l D R2k;2l D R2n
with its dual V  by means of the pseudo-Euclidian scalar product g D gcan .
Let us recall, that the support of  2 ƒ3 V is defined by

† WD spanfX Y j X; Y 2 Vg  V:
78 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proposition 3.6.5 A three-form  2 ƒ3 V  Š ƒ3 V satisfies (i) of Corollary 3.6.2 if

and only if there exists an isotropic subspace L  V such that  2 ƒ3 L  ƒ3 V. If
 satisfies (i) and (ii) of Corollary 3.6.2 then there exists a Jcan -invariant isotropic
subspace L  V with  2 ƒ3 L.
Proof The proposition follows from Lemmata 2.3.2 and 2.3.4 of Chap. 2 by taking
L D † . t
u
Remark 3.6.6 From the Proposition 3.6.5 we conclude that there are no strict flat
nearly pseudo-Kähler manifolds of dimension less than 8. We shall see later that the
dimension cannot be smaller than 12, see Corollary 3.6.8.
In the following we set ƒ W WD ƒ3;0 W C ƒ0;3 W; where W is a complex vector
space.
Theorem 3.6.7 A three-form  2 ƒ3 V  Š ƒ3 V satisfies (i) and (ii) of Corol-
lary 3.6.2 if and only if there exists an isotropic Jcan -invariant subspace L  V such
that  2 ƒ L  ƒ3 L  ƒ3 V. (The smallest such subspace L is † .)
Proof By Proposition 3.6.5, the conditions (i) and (ii) of Corollary 3.6.2 imply the
existence of an isotropic Jcan -invariant subspace L  V such that  2 ƒ3 L. By
Lemma 2.3.3 of Chap. 2 the condition (ii) is equivalent to  2 ƒ V: Therefore
 2 ƒ3 L \ ƒ V D ƒ L. The converse statement follows from the same argument.
t
u
Corollary 3.6.8 There are no strict flat nearly pseudo-Kähler manifolds of dimen-
sion less than 12.
Proof By Theorems 3.6.4 and 3.6.7 any flat nearly pseudo-Kähler manifold M is
locally of the form M.U; /, where  2 ƒ L for an isotropic Jcan -invariant subspace
L  V and U  V is an open subset. M.U; / is strict if and only if  ¤ 0, which is
possible only for dimC L  3, i.e. for dim M  12. t
u
Theorem 3.6.9 Any strict flat nearly pseudo-Kähler manifold is locally a pseudo-
Riemannian product M D M0  M.U; / of a flat pseudo-Kähler factor M0 of
maximal dimension and a strict flat nearly pseudo-Kähler manifold M.U; / of
(real) signature .2m; 2m/, 4m D dim M.U; /  12. The Jcan -invariant isotropic
support † has complex dimension m.
Proof By Theorems 3.6.4 and 3.6.7, M is locally isomorphic to an open subset of
a manifold of the form M.V; /, where  2 ƒ3 V has a Jcan -invariant and isotropic
support L D † . We choose a Jcan -invariant isotropic subspace L0  V such that
V 0 WD L C L0 is nondegenerate and L \ L0 D 0 and put V0 D .L C L0 /? . Then
 2 ƒ3 V 0  ƒ3 V and M.V; / D M.V0 ; 0/  M.V 0 ; /. Notice that M.V0 ; 0/ is
simply the flat pseudo-Kähler manifold V0 and that M.V 0 ; / is strict and of split
signature .2m; 2m/, where m D dimC L  3. t
u
Corollary 3.6.10 Let .M; g; J/ be a flat nearly Kähler manifold with a (positive or
negative) definite metric g then  D 0; rN D D and DJ D 0; i.e. .M; g; J/ is a Kähler
manifold.
3.6 Classification Results for Flat Nearly "-Kähler Manifolds 79

For the rest of this section we consider the case V Š Cm;m and denote a maximal
Jcan -invariant isotropic subspace by L. We will say that a complex three-form  2
ƒ3 .Cm / has maximal support if spanf.Z; W; /jZ; W 2 Cm g D .Cm / .
Corollary 3.6.11 Any non-zero complex three-form  2 ƒ3;0 L Š ƒ3 .Cm / defines
a complete flat simply connected strict nearly pseudo-Kähler manifold M./ WD
M.V; /,  D  C N 2 ƒ3 L  ƒ3 V, of split signature. M./ has no pseudo-Kähler
de Rham factor if and only if  has maximal support.
Conversely, any complete flat simply connected nearly pseudo-Kähler manifold
without pseudo-Kähler de Rham factor is of this form.
Proof This follows from the previous results observing that the support of  is
maximally isotropic if and only if  has maximal support. u
t
Corollary 3.6.12 The map  7! M. C / N induces a bijective correspondence
between GLm .C/-orbits on the open subset ƒ3reg .Cm /  ƒ3 .Cm / of three-forms
 with maximal support and isomorphism classes of complete flat simply connected
N of real dimension 4m  12 and without
nearly pseudo-Kähler manifolds M. C /
pseudo-Kähler de Rham factor.
Example 3.6.13
1) The case m  5.
For m D 3; 4; 5 the group GLm .C/ acts transitively on ƒ3reg .Cm / D
ƒ3 .Cm / n f0g. Therefore there exists precisely one complete flat simply
connected strict nearly pseudo-Kähler manifold of dimension 12, 16 and 20
respectively.
2) The case m D 6.
GL6 .C/ has precisely one open orbit in ƒ3reg .C6 / . This orbit consists of
the stable three-forms ƒ3stab .C6 / in the sense of Hitchin [75], cf. Sect. 2.1
in Chap. 2. We may recall, that a three-form  on C6 is stable if and only
if  D e1 ^ e2 ^ e3 C e4 ^ e5 ^ e6 for some basis .e1 ; e2 ; : : : ; e6 / of C6 .
ƒ3 .C6 / n ƒ3stab .C6 / is precisely the zero-set of the unique homogeneous
quartic SL6 .C/-invariant and we have the following strict inclusions:

ƒ3stab .C6 /  ƒ3reg .C6 /  ƒ3 .C6 / n f0g:

An example of an instable regular form is

e1 ^ e2 ^ e3 C e1 ^ e4 ^ e5 C e2 ^ e4 ^ e6 :

3.6.2 Classification of Flat Nearly Para-Kähler Manifolds

In this subsection we consider .Cn ; can / endowed with the standard can -anti-
invariant pseudo-Euclidian scalar product gcan of signature .n; n/:
80 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Let .M; g; / be a flat nearly para-Kähler manifold. Then there exists for each
point p 2 M an open set Up  M containing the point p; a connected open set U0
of Cn containing the origin 0 2 Cn and an isometry ˆ W .Up ; g/!.UQ 0 ; gcan /; such
that in p 2 M we have ˆ p D can ˆ : In other words, we can suppose, that locally
M is a connected open subset of Cn containing the origin 0 and that g D gcan and
0 D can :
Proposition 3.6.14 Let .M; g; / be a flat nearly para-Kähler manifold. Then
1) X ı Y D 0 for all X; Y 2 TM;
2) r D rN D 0:

Summarising Theorem 3.1.5 and Proposition 3.6.14 we obtain the next Corollary.
Corollary 3.6.15 Let M  Cn be an open neighbourhood of the origin endowed
with a nearly para-Kähler structure .g; / such that g D gcan and 0 D can . The
.1; 2/-tensor

1
 WD  D
2

defines a constant three-form on M  Cn D Rn;n given by .X; Y; Z/ D g.X Y; Z/

and satisfying
(i)  2 C.V/; i.e. X Y D 0; 8X; Y 2 TM;
(ii) fX ; can g D 0; 8X 2 TM:
The rest of this subsection is devoted to the local classification result. In Sect. 3.6.2
we study the structure of the subset of C.V/ given by the condition (ii) in more detail
and give global classification results. The converse statement of Corollary 3.6.15 is
given in the next lemma.
Lemma 3.6.16 (Lemma 2.10 of [43]) Let  be a constant three-form on an
open connected neighbourhood M  Cn of the origin 0 satisfying (i) and (ii) of
Corollary 3.6.15. Then there exists a unique para-complex structure  on M such
that
a) 0 D can ;
b) fX ; g D 0; 8X 2 TM;
c) D D 2;
where D is the Levi-Civita connection of the pseudo-Euclidian vector space Cn :
Let rN WD D   and assume b) then c) is equivalent to
N D 0:
c)’ r
Furthermore, this para-complex structure  is skew-symmetric with respect to gcan :
In fact, one shows, that the para-complex structure  is given by the following
3.6 Classification Results for Flat Nearly "-Kähler Manifolds 81

formula
2n
! 2n
!
X .i/ X
 D exp 2 xi @i can D Id C 2 xi @i can ; (3.53)
iD1 iD1

@
where xi are linear coordinates of Cn D Rn;n D R2n and @i D @xi
:
Theorem 3.6.17 Let  be a constant three-form on a connected open set U  Cn
containing the origin 0 which satisfies (i) and (ii) of Corollary 3.6.15. Then there
exists a unique almost para-complex structure

2n
!
X
 D exp 2 x @i can
i
(3.54)
iD1

on U such that a) 0 D can ; and b) M.U; / WD .U; g D gcan ; / is a flat nearly

para-Kähler manifold. Any flat nearly para-Kähler manifold is locally isomorphic
to a flat nearly para-Kähler manifold of the form M.U; /:

The Variety C .V/

Now we discuss the solution of (i) and (ii) of Corollary 3.6.15. In the following
we shall freely identify the real vector space V WD Cn D Rn;n D R2n with its
dual V  by means of the pseudo-Euclidian scalar product g D gcan . The geometric
interpretation is given in terms of an affine variety C .V/  ƒ3 V:
Proposition 3.6.18 A three-form  2 ƒ3 V  Š ƒ3 V satisfies (i) of Corol-
lary 3.6.15, i.e. X ı Y D 0; X; Y 2 V; if and only if there exists an isotropic
subspace L  V such that  2 ƒ3 L  ƒ3 V. If  satisfies (i) and (ii) of
Corollary 3.6.15 then there exists a can -invariant isotropic subspace L  V with
 2 ƒ3 L.
Proof The proposition follows from Lemmata 2.3.2 and 2.3.4 of Chap. 2 by taking
L D † . t
u
A three-form  on a para-complex vector space .W; can / decomposes with respect
to the grading induced by the decomposition W 1;0 ˚ W 0;1 into  D C C  : In the
remainder of this subsection we set for convenience C 2 ƒC W WD ƒ2;1 W Cƒ1;2 W
and  2 ƒ W WD ƒ3;0 W C ƒ0;3 W:
Theorem 3.6.19 A three-form  2 ƒ3 V  Š ƒ3 V satisfies (i) and (ii) of
Corollary 3.6.15 if and only if there exists an isotropic can -invariant subspace L
such that  2 ƒ L D ƒ3;0 L C ƒ0;3 L  ƒ3 L  ƒ3 V (The smallest such subspace
L is † .).
Proof By Proposition 3.6.18, the conditions (i) and (ii) of Corollary 3.6.15 imply
the existence of an isotropic can -invariant subspace L  V such that  2 ƒ3 L.
82 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

By Lemma 2.3.3 of Chap. 2 the condition (ii) is equivalent to  2 ƒ V: Therefore

 2 ƒ3 L \ ƒ V D ƒ L. The converse statement follows from the same argument.
t
u
Corollary 3.6.20
(i) The conical affine variety

C .V/ WD f j  satisfies .i/ and .ii/g  ƒ3 V

has the following description

[ [
C .V/ D ƒ L D .ƒ3 LC C ƒ3 L /;
LV LV

where the union is over all -invariant maximal isotropic subspaces.

(ii) If dim V < 12 then it holds C .V/ D ƒ3 V C [ ƒ3 V  :
(iii) Any flat nearly para-Kähler manifold M is locally of the form M.U; /, for
some  2 C .V/ and some open subset U  V:
(iv) There are no strict flat nearly para-Kähler manifolds of dimension less than 6.
Proof
(i) follows from Theorem 3.6.19.
(ii) Let L  V be a -invariant isotropic subspace. If dim V < 12; then dim L < 6
and, hence, either dim LC < 3 or dim L < 3: In the first case we have

ƒ L D ƒ3 LC C ƒ3 L D ƒ3 L  ƒ3 V  ;

in the second case it is ƒ L D ƒ3 LC C ƒ3 L D ƒ3 LC  ƒ3 V C :

(iii) is a consequence of (i), Theorems 3.6.17 and 3.6.19.
(iv) By (iii) the strict flat nearly para-Kähler manifold M is locally of the form
M.U; /; which is strict if and only if  ¤ 0. This is only possible for dim L 
3, i.e. for dim M  6.
t
u
Example 3.6.21 We have the following example which shows that part (ii) of
Corollary 3.6.20 fails in dimension  12:
Consider .V; / D .C6 ; i" / D R6 ˚ i" R6 ; for " D 1; with a basis given by
.e1 ; : : : ; eC
C   ˙
6 ; e1 ; : : : ; e6 /; such that ei form a basis of V
˙
with g.eC 
i ; ej / D ıij :
C C C   
Then the form  WD e1 ^ e2 ^ e3 C e4 ^ e5 ^ e6 lies in the variety C .V/:
Theorem 3.6.22 Any strict flat nearly para-Kähler manifold is locally a pseudo-
Riemannian product M D M0  M.U; / of a flat para-Kähler factor M0 of maximal
dimension and a flat nearly para-Kähler manifold M.U; /,  2 C .V/, of signature
.m; m/, 2m D dim M.U; /  6 such that † has dimension m.
3.6 Classification Results for Flat Nearly "-Kähler Manifolds 83

Proof By Theorems 3.6.17 and 3.6.19, M is locally isomorphic to an open subset of

a manifold of the form M.V; /, where  2 ƒ3 V has a can -invariant and isotropic
support L D † . We choose a can -invariant isotropic subspace L0  V such that
V 0 WD L C L0 is nondegenerate and L \ L0 D 0 and put V0 D .L C L0 /? . Then
 2 ƒ3 V 0  ƒ3 V and M.V; / D M.V0 ; 0/  M.V 0 ; /. Notice that M.V0 ; 0/ is
simply the flat para-Kähler manifold V0 and that M.V 0 ; / is strict of split signature
.m; m/, where m D dim L  3. t
u
Corollary 3.6.23 Any simply connected nearly para-Kähler manifold with a
(geodesically) complete flat metric is a pseudo-Riemannian product M D M0 M./
of a flat para-Kähler factor M0 D Rl;l of maximal dimension and a flat nearly para-
Kähler manifold M./ WD M.V; /,  2 C .V/, of signature .m; m/ such that †
has dimension m D 0; 3; 4; : : :.
Next we wish to describe the moduli space of (complete simply connected)
flat nearly para-Kähler manifolds M of dimension 2n up to isomorphism. Without
restriction of generality we will assume that M D M./ has no para-Kähler de Rham
factor, which means that  2 C .V/ has maximal support † , i.e. dim † D n. We
reg
denote by C .V/  C .V/ the open subset consisting of elements with maximal
support. The group

G WD Aut.V; gcan ; can / Š GL.n; R/

reg
acts on C .V/ and preserves C .V/. Two nearly para-Kähler manifolds M./ and
M.0 / are isomorphic if and only if  and 0 are related by an element of the group G.
For  2 C .V/ we denote by p, q the dimensions of the eigenspaces of  on †
for the eigenvalues 1; 1, respectively. We call the pair . p; q/ 2 N0  N0 the type
of . We will also say that the corresponding flat nearly para-Kähler manifold M./
p;q
has type . p; q/. We denote by C .V/ the subset of C .V/ consisting of elements of
reg
type . p; q/. Notice that p C q  n with equality if and only if  2 C .V/. We have
the following decomposition
[
Creg .V/ D Cp;q .V/;
. p;q/2…

where … WD f. p; q/j p; q 2 N0 n f1; 2g; p C q D ng. The group G D GL.n; R/ acts

p;q p;q
on the subsets C .V/ and we are interested in the orbit space C .V/=G.
Fix a -invariant maximally isotropic subspace L  V of type . p; q/ and put
ƒ  C 
reg p;q
reg L WD ƒ L \ C .V/  C .V/. The stabiliser GL Š GL.L /  GL.L / Š
C  
GL.p; R/  GL.q; R/ of L D L C L in G acts on ƒreg L.
Theorem 3.6.24 There is a natural one-to-one correspondence between complete
simply connected flat nearly para-Kähler manifolds of type . p; q/, p C q D n, and
the points of the following orbit space:

Cp;q .V/=G Š ƒ  3 C C 3  
reg L=GL  ƒ L=GL D ƒ L =GL.L /  ƒ L =GL.L /:
84 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proof Consider two complete simply connected flat nearly para-Kähler manifolds
M, M 0 . By the previous results we can assume that M D M./, M 0 D M.0 / are
associated with ; 0 2 C .V/. It is clear that M and M 0 are isomorphic if  and 0
p;q

are related by an element of G. To prove the converse we assume that ' W M ! M 0

is an isomorphism of nearly para-Kähler manifolds. By the results of Sect. 3.5 
defines a simply transitive group of isometries. This group preserves also the para-
complex structure , which is r-parallel and hence left-invariant. This shows that
M and M 0 admit a transitive group of automorphisms. Therefore, we can assume
that ' maps the origin in M D V to the origin in M 0 D V. Now ' is an isometry of
pseudo-Euclidian vector spaces preserving the origin. Thus ' is an element of O.V/
preserving also the para-complex structure  and hence ' 2 G.
The identification of orbit spaces can be easily checked using Lemma 2.3.2 and
the fact that any -invariant isotropic subspace † D †C C † can be mapped onto
L by an element of G. t
u

3.7 Conical Ricci-Flat Nearly Para-Kähler Manifolds

Definition 3.7.1 A conical semi-Riemannian manifold .M; g; / is a semi-

Riemannian manifold .M; g/ endowed with a vector field  such that

r D Id ; (3.55)

where r is the Levi-Civita connection of g: It is called regular, if the function

k WD g.; / has no zeros.
A conical nearly para-Kähler manifold .M; ; g; / is a nearly para-Kähler
manifold .M; ; g/ such that .M; g; / is conical and a conical para-Kähler manifold
.M; P; g; / is a para-Kähler manifold .M; P; g/ such that .M; g; / is conical.
For a proof of the following Proposition we refer to Proposition 6 of [39].
Proposition 3.7.2 Let .M; g; / be a regular conical semi-Riemannian manifold.
Then the level sets Mc WD fk D cg; c 2 R; are smooth hypersurfaces perpendicular
to  or empty. If Mc ¤ ;; then g induces a semi-Riemannian metric gc on Mc :
Theorem 3.7.3 Let .M; ; g; / be a Ricci-flat conical (strict) nearly para-Kähler
manifold and define an endomorphism field P by
 
1
P WD Id C N ı : (3.56)
4

(i) If N.X; Y; Z/ has isotropic support, then .M; P; g/ is a para-Kähler manifold.

(ii) If the real dimension of M is 6, then .M; P; g/ is a para-Kähler manifold.
3.7 Conical Ricci-Flat Nearly Para-Kähler Manifolds 85

We remark that the tuple .g; / remains the same and in consequence .M; g; /
is conical and Ricci-flat. Hence .M; ; g; / is a Ricci-flat conical para-Kähler
manifold.
Proof It suffices to show (i), since by Corollary 3.1.11 the Nijenhuis tensor
N.X; Y; Z/ has isotropic support in dimension 6. Using fN ; g D 0; where N was
defined as the endomorphism field given by N Y D N.; Y/; we compute
  2    
1 1 1
P2 D Id C N ı  D Id C N ı Id  N ı  2
4 4 4
1
D Id  N ı N D Id:
16
Moreover, the condition

g. PX; Y/ D g.X; PY/

follows, since N ı  is skew-symmetric. In fact, we have

g.N X; Y/ D g.N.; X/; Y/ D 4g..r /X; Y/ D 4g..r /X; Y/;

which is skew-symmetric in X; Y: Hence .M; P; g/ defines an almost para-Hermitian

structure. To show that it is para-Kähler we determine
  
1 1
.rX P/Y D rX Id C N ı  Y D .rX /Y C rX .N ı /Y
4 4
1 
D .rX /Y C .rX N /Y C N .rX /Y
4
1 1
D .rX /Y C NrX  .Y/ D .rX /Y  NX Y D 0:
4 4

In this computation we used that N.; ; / has isotropic support and that N.X; Y/ is
r-parallel (by Lemma 3.1.13). Namely, it is

.rX N /W D rX .N.; W//  N.; rX W/ D .rX N/.; W/ C N.rX ; W/ D NrX  W:

The statement that N.X; Y; Z/ is r-parallel is also shown in Lemma 3.1.13 and it
does not vanish if .M; ; g/ is strict nearly para-Kähler. t
u


Remark 3.7.4 As the attentive reader observes, the ansatz P D Id C 14 N ı yields
an almost para-complex structure, if N is of type .3; 0/ C .0; 3/ and has isotropic
support. This structure is para-Kähler if and only if it holds NrX  Y D 4.rX /Y;
i.e. NX Y D NrX  Y: If M is strict nearly para-Kähler, this implies rX  D X: This
means  needs to be conical.
86 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Theorem 3.7.5 Let .M; P; g; / be a Ricci-flat conical para-Kähler manifold of

(real) dimension 2m carrying a (non-vanishing) parallel 3-form '.X; Y; Z/ of type
.3; 0/ C .0; 3/ with isotropic support and define an endomorphism field  by
 
1
 D Id  ' ı P: (3.57)
4

Then .M; ; g/ is a (strict) nearly para-Kähler manifold.

As the pair .g; / is not changed, .M; g; / is conical and Ricci-flat. In consequence
.M; ; g; / is a (strict) Ricci-flat conical nearly para-Kähler manifold.
Proof Here the endomorphism field 'X is given by 'X Y WD g1 ı '.X; Y; /:9 Since
'.X; Y; Z/ has type .3; 0/ C .0; 3/ one has f' ; Pg D 0 (this follows from Eq. (2.35))
and we compute as before
  2  
2 1 1
 D Id  ' ı P D Id  ' ı ' ı P2 D Id:
4 16

The last step follows, since '.X; Y; Z/ has isotropic support (cf. the proof of
Corollary 3.1.11 (b)). By the type condition it is '.; PX; Y/ D '.P; X; Y/ D
'.P; Y; X/ which means that ' ı P is skew-symmetric. From this it follows
g.X; Y/ D g.X; Y/: It is left to check the nearly para-Kähler condition
  
1 1
.rX /Y D rX Id  ' ı P Y D .rX P/Y  rX .' ı P/Y
4 4
1  1 1
D .rX ' / ı PY C ' .rX P/Y D  'rX  . PY/ D P.'X Y/;
4 4 4

which is skew-symmetric, since '.X; Y; Z/ is a 3-form. Hence .M; ; g/ is a nearly

para-Kähler manifold. If '.X; Y; Z/ is non-vanishing, then .M; ; g/ is strict nearly
para-Kähler. t
u
Remark 3.7.6 One may choose '.; ; /; 0 ¤  2 R; in place of '.; ; /:
Geometrically this corresponds to rescaling the conical vector field  by the factor :
Remark 3.7.7 Let us make an observation concerning Theorems 3.7.3 and 3.7.5.
The Ansatz for  only gives a para-complex structure, if it is ' ı ' D 0:
This implies, that '.X; Y; Z/ has isotropic support and a para-Hermitian metric
has automatically split signature. Therefore we only can have these examples for
indefinite metrics (compare also Remark 3.7.11 for more comments).
In the following, we suppose that  is space-like, i.e. it is g.; / > 0: We can always
achieve this by replacing the metric g by g: Since ˙g have the same Levi-Civita
connection, this operation is compatible with the nearly para-Kähler condition (3.1)

9
Here g1 is the inverse of the map g W TM ! T  M; X 7! g.X; /.
3.7 Conical Ricci-Flat Nearly Para-Kähler Manifolds 87

and the conical condition (3.55). One observes that one always can assume M1 D
fg.; / D 1g ¤ ; after rescaling g by a positive constant without violating neither
the nearly para-Kähler nor the conical condition.
Proposition 3.7.8 Let .M; P; g; / be a regular conical para-Kähler manifold with
M1 ¤ ;; then M1 with the induced metric g1 and the Reeb vector field T D PjM1 is
a para-Sasaki manifold.
The manifold M is Ricci-flat if and only if M1 is an Einstein manifold with scalar
curvature 2m.2m C 1/:
Proof The conical vector field  is regular and Proposition 3.7.2 implies that
.M1 ; g1 / is a semi-Riemannian manifold. Denote by r 1 the Levi-Civita connection
of g1 : By construction T is time-like, i.e. g1 .T; T/ D 1 and tangential (see
Proposition 3.7.2). Moreover, T is a Killing vector field, since one has for vector
fields X; Y on M1

LT g1 .X; Y/ D g1 .rX1 T; Y/ C g1 .X; rY1 T/ D g.rX P; Y/ C g.X; rY P/

D g. PrX ; Y/ C g.X; PrY / D g. PX; Y/ C g.X; PY/ D 0:

Additionally T is geodesic, since for X 2 TM1 it is

g1 .rT1 T; X/ D g.rT T; X/ D g.rT P; X/ D g. PT; X/ D 0:

Since T is a Killing vector field ˆ WD r 1 T is skew-symmetric and we have

ˆX D rX1 T D .rX T/tan D . PX/tan D PX  g.; PX/ D PX C g1 .T; X/;

where tan is the projection on TM1 : This means ˆT D 0 and ˆX D PX for X 2 TM1
perpendicular to T: It follows

ˆ2 .X/ D PˆX C g1 .T; ˆX/ D PˆX D X C g1 .T; X/T:

We compute .rX1 ˆ/Y for Y D T

.rX1 ˆ/T D rX1 .ˆT/  ˆ.rX1 T/ D ˆ2 .X/ D X  g1 .T; X/T

and for Y perpendicular to T

.rX1 ˆ/Y D rX1 .ˆY/  ˆ.rX1 Y/

./
D P.rX1 Y/  g.; PrX1 Y/  ˆ.rX1 Y/  g1 .X; Y/T
./
D g1 .X; Y/T:
88 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

For ./ we used

rX1 .ˆY/ D rX1 . PY/ D .rX . PY//tan D . PrX Y/tan D . P.rX Y/tan /tan C g.rX Y; /T
D . P.rX1 Y//tan  g. Y; rX /T D P.rX1 Y/  g.; PrX1 Y/  g1 .X; Y/T

and ./ follows from

P.rX1 Y/  g.; PrX1 Y/  ˆ.rX1 Y/ D g1 . P; rX1 Y/  g1 .T; rX1 Y/ D 0:

Summarizing it holds

.rU1 ˆ/V D g1 .U; V/T C g1 .V; T/U:

Hence we have checked all the conditions of Definition 2.6.1 and conclude that M1
is a para-Sasaki manifold. In consequence the cone M c1 is a para-Kähler manifold,
which is Einstein if and only if .M1 ; g1 / is Einstein and the scalar curvature of g1
equals 2m.2m C 1/ (cf. Remark 2.6.2 (i) and (ii)). t
u
Theorem 3.7.9 Let .N 5 ; g; T/ be a para-Sasaki Einstein manifold of dimension 5
and denote by .M 6 D b N;bg; P; / the associated conical Ricci-flat para-Kähler
manifold on the cone M D b N over N; then the cone M can be endowed with the
structure of a conical Ricci-flat strict nearly para-Kähler six-manifold .M; ;b
g; /:
Moreover, M is flat if and only if N has constant curvature.
Proof By Remark 2.6.2 (ii) the cone b N is a conical Ricci-flat para-Kähler six-
manifold .bN; gO ; P; / and hence admits a non-vanishing parallel three-form ' with
isotropic support. From Theorem 3.7.5 we obtain a strict nearly para-Kähler
structure  on bN such that .bN; ; gO ; / is a conical Ricci-flat nearly para-Kähler six-
manifold. The last statement follows from the fact that b N is flat if and only if N has
constant curvature. t
u
In the following we call a nearly para-Kähler manifold M, which is the space-like
metric cone M D b N over some semi-Riemannian manifold N a nearly para-Kähler
cone. Summarising we have shown the following result.
Theorem 3.7.10 There is a one to one correspondence between Ricci-flat strict
nearly para-Kähler cones with isotropic Nijenhuis tensor and space-like cones over
para-Sasaki Einstein manifolds endowed with a parallel 3-form having isotropic
support.
Remarks 3.7.11
(a) An analogous Ansatz can be made in almost complex geometry.
(i) In this setting one still needs a form with isotropic support. Since non-trivial
three-forms with isotropic support do not exist for Riemannian metrics,
the Ansatz does only give something new, i.e. non-Kähler examples,
3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 89

for pseudo-Riemannian metrics and real dimension dim M  12; cf.

Theorem 3.6.7 and Corollary 3.6.8.
(ii) Further one would get a cone over a Sasaki Einstein manifold with
indefinite metric. Such manifolds can for instance be obtained as T-duals
of homogeneous Sasaki manifolds of real dimension at least 11. These
manifolds are only classified in dimensions  7 and the classification is
possibly extended to dimension 9 and 11 using [19] (see Section 11.1.1 of
[20]).
Details shall be postponed to future work.
(b) When we are not insisting on irreducible examples, one has the following
construction in the almost pseudo-Hermitian world: Denote by .M m ; gM ; JM /
and .N n ; gN ; JN / two nearly Kähler Einstein-manifolds with the same Einstein
constant, then the pseudo-Riemannian product .M  N; gM ˚ .gN /; JM ˚ JN /
is a nearly pseudo-Kähler manifold with vanishing Ricci curvature.

3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler

Six-Manifolds

3.8.1 Linear Algebra of Five-Dimensional Reductions

of SU.1; 2/-Structures

In this short section we prepare the linear algebra of dimensional reductions.

Lemma 3.8.1 Let V be a six-dimensional real vector space and .!; / 2 ƒ2 V  
ƒ3 V  a compatible normalised pair of stable forms. Denote by h D h.!;/ the
induced metric, let N 2 V be a unit vector with h.N; N/ D " 2 f˙1g and denote
by W D N ? the orthogonal complement of R N: Then the quadruple .; !1 ; !2 ; !3 /
defined by

 D ˇ .Ny!/; !1 D ˛ !jW ; !2 D NyJ ; !3 D Ny (3.58)

with ˛; ˇ 2 f˙1g defines an SU" . p; q/-structure with p C q D 2 on W:

Moreover, one has

! D ˛ !1 C ˇ n ^ ;
 D "ˇ  ^ !2  n ^ !3 ;
J  D "ˇ  ^ !3 C n ^ !2 ;

where n 2 V  is the dual of N and

 ^ !2 D "ˇ jW and  ^ !3 D "ˇ J jW :

90 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proof As above (see Eqs. (2.7) and (2.20) of Chap. 2) we may choose a basis
fe1 ; : : : ; e6 g of V; such that the stable forms !;  and J  are given in the normal
forms

! D e12  e34 C e56 ;

 D e135  .e146 C e236 C e245 /

with ./ D 4 ˝2 for D e123456 > 0. Furthermore, it holds J ei D eiC1 ,

J eiC1 D ei for i 2 f1; 3; 5g and

J  D e246  .e235 C e145 C e136 /:

1) In the case " D 1 we can suppose N D e1 and obtain

 D ˇ e2 ; ˛ !1 D e34 C e56 ; !2 D e36  e45 ; !3 D e35 C e46 :

One easily sees

!12 D !22 D !32 D 2e3456 and  ^ !j2 D 2ˇ e23456 ¤ 0

and !j ^ !k D 0 for 1  j < k  3:

Moreover, one gets

! D ˛ !1 C ˇ n ^ ;
 D ˇ  ^ !2  n ^ !3 and
J  D ˇ  ^ !3 C n ^ !2 ;

where n is the dual of N: Further, one has

 ^ !2 D ˇ jW and  ^ !3 D ˇ J jW :

2) For " D 1 we can choose N D e5 and get

 D ˇ e6 ; ˛ !1 D e12  e34 ; !2 D e14  e23 ; !3 D e13 C e24 :

One easily sees

!12 D !22 D !32 D 2e1234 and  ^ !j2 D 2ˇ e12346 ¤ 0

and !j ^ !k D 0 for 1  j < k  3:

3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 91

Moreover, one gets

! D ˛ !1 C ˇ n ^ ;
 D ˇ  ^ !2  n ^ !3 and
J  D ˇ  ^ !3 C n ^ !2 ;

where n is the dual of N: Finally, one has

 ^ !2 D ˇ jW and  ^ !3 D ˇ J jW :

t
u

3.8.2 Evolution of Hypo Structures

A five-manifold N 5 carries an SU" . p; q/-structure with p C q D 2 provided,

that its frame bundle admits a reduction to SU" . p; q/: For the group SU.2/ it is
shown in [36], that such a structure is determined by a quadruple of differential
forms .!1 ; !2 ; !3 ; /: We shortly derive the analogous statement for our setting. Let
f W N 5 ! M 6 be an oriented hypersurface in a six-manifold M 6 endowed with an
SU.1; 2/-structure given by a triple .!; C ;  / of compatible stable forms (cf.
Sect. 2.1).
This SU.1; 2/-structure induces an SU" . p; q/-structure with p C q D 2 on N 5 via
the definitions

!1 D ˛ f  !; b !2 D y 
; bC !3 D y C
;  D ˇ y!; (3.59)

where ˛; ˇ; bC ; b 2 f˙1g are real constants and denotes the unit normal vector
field of N 5 of length " D g. ; /:
In case, that the holonomy of M 6 is contained in SU.1; 2/ or in other words the
SU.1; 2/-structure is integrable, which is equivalent to the equations
C 
d! D 0; d D 0 and d D0 (3.60)

we obtain a hypo structure on N in the sense of the next Definition.

Definition 3.8.2 An SU" . p; q/-structure with p C q D 2 determined by
.; !1 ; !2 ; !3 / is called hypo provided, that it satisfies

d!1 D 0; d. ^ !2 / D 0 and d. ^ !3 / D 0:

For the Riemannian case the next lemma is shown in [36].

92 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Lemma 3.8.3 Let f W N 5 ! M 6 be an oriented hypersurface in a six-manifold

M 6 endowed with an integrable SU.1; 2/-structure, then the induced SU" . p; q/-
structure, given by (3.59) and with p C q D 2; on N 5 is a hypo-structure.
Proof From ˛ f  ! D !1 one has d!1 D ˛ d. f  !/ D ˛ f  .d!/ D 0: Moreover,
with help of the Lemma 3.8.1 one has

bC f  C
D "ˇ  ^ !2 and b f  
D "ˇ  ^ !3 ;

which implies using (3.60)

"ˇ d. ^ !3 / D b d. f  
/ D b f  d 
D0
C  C C  C
D b d. f /Db f d D "ˇ d. ^ !2 /:

This shows, that the induced SU" . p; q/-structure is a hypo structure on N 5 : t

u
Starting with an SU" . p; q/-structure with p C q D 2 on N 5 determined by
.; !1 ; !2 ; !3 / we define a two-form

! D ˛!1 C "ˇ dt ^  (3.61)

˙
and three-forms on N 5  R by
C
D aC  ^ !2 C bC dt ^ !3 ; 
D a  ^ !3 C b dt ^ !2 ; (3.62)

where t is the coordinate on R and ˛; ˇ; a˙ ; b˙ 2 f˙1g are non-zero real constants.

Note, that the a˙ are determined from ˛; ˇ and b˙ by Lemma 3.8.1. Then a partial
converse of the result of the last Lemma is given in the next Proposition.
Proposition 3.8.4 One can lift a hypo SU" . p; q/-structure with p C q D 2 to an
integrable SU.1; 2/-structure on N  R if it belongs to a one-parameter family
of SU" . p; q/-structures ..t/; !1 .t/; !2 .t/; !3 .t//; where t is the coordinate on R;
satisfying the Conti-Salamon type evolution equations

@t !1 D "ˇ˛ d5 ; (3.63)
 
@t . ^ !3 / D a b d!2 ; (3.64)
@t . ^ !2 / D aC bC d!3 : (3.65)

Proof In this proof we write d5 and d6 for the exterior differentials on N 5 and M 6 D
N  R: With (3.61) it follows, that

0 D d6 ! D ˛d5 !1 C .˛@t !1  "ˇ d5 / ^ dt:

3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 93

This is equivalent to d5 !1 D 0 and @t !1 D "ˇ˛ d5 ; i.e. Eq. (3.63). From the

definition of C one gets, that


0 D d6 C
D aC d5 .!2 ^ / C dt ^ aC @t .!2 ^ /  bC d5 !3

is equivalent to d5 .!2 ^ / D 0 and @t .!2 ^ / D aC bC d5 !3 ; i.e. Eq. (3.65). For


as given in (3.62) we obtain, that

0 D d6 
D a d5 .!3 ^ / C dt ^ .a @t .!3 ^ /  b d5 !2 /

itself is equivalent to d5 .!3 ^ / D 0 and Eq. (3.64). t

u
Examples of this type are given by the pseudo-Riemannian cousins of Sasaki-
Einstein manifolds, namely para-Sasaki-Einstein and Lorentzian-Sasaki-Einstein
manifolds. These can be characterised by the fact, that the space-like/time-like cone
is a Ricci-flat Kähler-Einstein manifold or equivalently this cone has an integrable
SU.1; 2/-structure. Here one considers the special solution of the above evolution
equations on N 5  R given by

! D t2 ˛!1 C t"ˇdt ^ ; (3.66)

C
D aC t3  ^ !2 C t2 bC dt ^ !3 ; (3.67)
  3 2 
D a t  ^ !3 C t b dt ^ !2 : (3.68)

The integrability conditions read

0 D d! D d.t2 ˛!1 C t"ˇdt ^ / D tdt ^ .2˛!1  "ˇd/;

C
0Dd D d.aC t3  ^ !2 C t2 bC dt ^ !3 /
D t2 dt ^ .3aC  ^ !2  bC d5 !3 / C t3 aC d5 . ^ !2 /;

0Dd D d.a t3  ^ !3 C t2 b dt ^ !2 /
D t2 dt ^ .3a  ^ !3  b d5 !2 / C t3 a d5 . ^ !3 /:

This is equivalent to the para-Sasaki-Einstein or Lorentz-Sasaki-Einstein equations

d D 2"˛ ˇ !1 ; d!2 D 3b a !3 ^ ; d!3 D 3bC aC !2 ^ : (3.69)

Obviously, Eq. (3.69) imply the hypo equations.

The next result has been discovered in the Riemannian case in [53].
Proposition 3.8.5 Let f W N 5 ! M 6 be a totally geodesic oriented hypersurface in a
nearly pseudo-Kähler manifold M 6 with unit normal vector field ; then the induced
SU" . p; q/-structure with p C q D 2 and " D g. ; / satisfies the hypo equations.
94 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proof From the first nearly Kähler equation one has

d!1 D ˛d. f  !/ D ˛f  d! D 3˛f  C

D 3.˛aC / ^ !2 : (3.70)

Moreover, we compute

ˇ 1 d D d. y!/ D .L !/  yd! D .L !/  3 y C
D .L !/  3bC !3 ;

where L is the Lie-derivative. Since N 5 is totally geodesic, it is .L !/ D r !:

N D 0 we get with rN D r C 1 T
Using r! 2

1
r !.X; Y/ D rN !.X; Y/ C Œ!.T. ; X/; Y/ C !.X; T. ; Y//
2
D !.T. ; X/; Y/ D bC !3 .X; Y/;

as with !.; / D g.; J/ it is

C
!.T. ; X/; Y/ D g. J.r J/X; JY/ D g.X; .r J/Y/ D . ; X; Y/ D bC !3 .X; Y/;

which shows

d D 2ˇbC !3 :


Finally, we compute with help of d D 2! ^ !, i.e. the second nearly Kähler
equation

b d!2 D d. y 
/ D L   yd  D r 
C 2 y.! ^ !/
1 X
D rN 
 y k .ek y  / ^ .ek y 
/ C 2 y.! ^ !/
4 k

./1 3
D  y.2! ^ !/ C 2 y.! ^ !/ D y.! ^ !/ D 3.˛ˇ/ ^ !1 ;
4 2

where fe1 ; : : : ; e6 D g is some adapted basis and where in ./ we used

6
X
 
k .ek y / ^ .ek y / D 2! ^ !;
kD1

which holds for a nearly pseudo-Kähler six-manifold. t

u
3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 95

Theorem 3.8.6 Any totally geodesic hypersurface N 5 of a nearly pseudo-Kähler

six-manifold M 6 carries a hypo structure given by the quadruplet .;
Q !Q 1 ; !Q 2 ; !Q 3 / D
.; "!3 ; !2 ; !1 /; and in consequence the Conti-Salamon type evolution equations
can be solved on N  R:
Proof By the last Lemma we obtain a hypo SU" . p; q/-structure .; Q !Q 1 ; !Q 2 ; !Q 3 /;
which is a solution of the first and the third para-Sasaki-Einstein or Lorentz-Sasaki-
Einstein equations (3.69). Using bC D 1 and aC D ˇ (by Lemma 3.8.1) and
setting ˛ D 1 yields

ŠŠ
d Q D d D 2.ˇbC / !3 D 2ˇ !3 D 2 ".ˇ˛/ !Q 1 ;
(3.70) ŠŠ
d!Q 3 D d!1 D 3.˛aC /  ^ !2 D 3.˛ˇ/ Q ^ !Q 2 D 3.bC aC /Q ^ !Q 2 ;

since again by Lemma 3.8.1,10 it is bC aC D ˇ D a b : The remaining para-

Sasaki-Einstein or Lorentz-Sasaki-Einstein equation

ŠŠ
d!Q 2 D d!2 D 3.b ˛ˇ/ ^ !1 D 3.˛ˇ/Q ^ !Q 3 D 3.b a /Q ^ !Q 3

holds true using b D 1 (by Lemma 3.8.1). t

u

3.8.3 Evolution of Nearly Hypo Structures

In this subsection we generalise results of [53] to construct examples of nearly

pseudo-Kähler manifolds via the nearly hypo evolution equations.
Definition 3.8.7 An SU" . p; q/-structure with p C q D 2 determined by
.; !1 ; !2 ; !3 / is called nearly hypo provided, that it satisfies the conditions

d!1 D 3˛aC  ^ !2 ; d. ^ !3 / D 2a !1 ^ !1 : (3.71)

Proposition 3.8.8 An SU" . p; q/-structure .; !1 ; !2 ; !3 / with p C q D 2 can be

lifted to a nearly pseudo-Kähler structure .!.t/; C .t/;  .t// on N 5  R defined
in (3.61) and (3.62) if and only if it is a nearly hypo structure which generates
a 1-parameter family of SU" . p; q/-structures ..t/; !k .t// satisfying the following
nearly hypo evolution equations

@t !1 D 3bC ˛ !3 C "ˇ˛ d;

@t . ^ !3 / D a b d!2  4"a ˛ˇ!1 ^ ; (3.72)
@t . ^ !2 / D aC bC d!3 :

10
Observe, that there is a relative factor " between dual and the metric dual of :
96 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Remark 3.8.9 Setting ˇ D 1 and with b D 1 D bC and bC aC D b a D

ˇ D 1 (by Lemma 3.8.1) we get

@t .˛!1 / D 3 !3 C " d;

@t . ^ !3 / D d!2 C 4" ^ .˛!1 /; (3.73)
@t . ^ !2 / D d!3 :

For .;
Q !Q 1 ; !Q 2 ; !Q 3 / D .; ˛!1 ; !2 ; !3 /; this yields

@t !Q 1 D 3 !Q 3  " d;
Q
@t .Q ^ !Q 3 / D d !Q 2  4"Q ^ !Q 1 ; (3.74)
@t .Q ^ !Q 2 / D d !Q 3 :

Hence the exterior differential system on the modified data .;

Q !Q 1 ; !Q 2 ; !Q 3 / looks
formally the same as the system found in the Riemannian case [53].
Proof From the definitions of !1 ; !2 and  we get, that the first nearly Kähler
equation is equivalent to

d! D d.˛!1 C "ˇdt ^ / D ˛d!1 C dt ^ .˛@t !1  "ˇd/

C
D3 D 3aC  ^ !2 C 3bC dt ^ !3 ;

i.e. d!1 D 3˛aC  ^ !2 and @t !1 D 3bC ˛ !3 C "ˇ˛ d: For the second nearly
Kähler equation we have

d D d.a  ^ !3 C b dt ^ !2 / D a d. ^ !3 / C dt ^ .a @t . ^ !3 /  b d!2 /
D 2.˛!1 C "ˇdt ^ /2 D 2 !1 ^ !1  4"˛ˇ dt ^ !1 ^ ;

which is equivalent to

d. ^ !3 / D 2a !1 ^ !1 and @t . ^ !3 / D a b d!2  4"a ˛ˇ!1 ^ :

These are the first two evolution equations and the nearly hypo equations. The third
equation is needed to show, that the nearly hypo property is conserved along the
evolution. Firstly, one has

@t .d!1  3˛aC  ^ !2 / D d.@t !1 /  3˛aC @t . ^ !2 /

D d.3bC ˛ !3 C "ˇ˛ d/  3˛aC @t . ^ !2 /
D 3bC ˛ d!3  3˛aC @t . ^ !2 /;
3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 97

which vanishes by the third evolution equation @t . ^ !2 / D aC bC d!3 : For the

other nearly hypo equation we compute

@t Œd. ^ !3 / C 2a !1 ^ !1  D dŒ@t . ^ !3 / C 2a @t .!1 ^ !1 /

D dŒa b d!2  4"a ˛ˇ!1 ^ 
C4a @t .!1 / ^ !1
D 4"a ˛ˇd.!1 ^ /
C4a !1 ^ .3bC ˛ !3 C "ˇ˛ d/
D 4"a ˛ˇ.d.!1 ^ /  !1 ^ d/
D 4"a ˛ˇ d!1 ^  D 0;

where we used, that by the already proven first hypo equation d!1 is (along the flow)
a multiple of  ^ !2 : Hence the nearly hypo condition is preserved along a solution
of the system (3.72). t
u
Proposition 3.8.10 Any SU" . p; q/-structure with p C q D 2 satisfying the para-
or pseudo-Sasaki equations (3.69) defines a nearly hypo structure .;
Q !Q 1 ; !Q 2 ; !Q 3 / D
.; !3 ; !2 ; !1 /:
Proof From (3.69) one has after setting ˛ D bC D 1 (by Lemma 3.8.1)

d!Q 1 D d!3 D 3bC aC !2 ^  D 3˛aC Q ^ !Q 2

and
ŠŠ
Q !Q 3 / D d.^!1 / D d^!1 D 2"˛ˇ !1 ^!1 D 2˛ˇ !3 ^!3 D 2a !Q 1 ^ !Q 1 ;
d.^

where we used !1 ^ !1 D " !3 ^ !3 : This yields the claim, since one has a D
ˇ D ˛ˇ (by Lemma 3.8.1). t
u
Proposition 3.8.11 Let f W N 5 ! M 6 be an immersion of an oriented 5-manifold
into a 6-dimensional nearly pseudo-Kähler manifold, then the induced SU" . p; q/-
structure .;
Q !Q 1 ; !Q 2 ; !Q 3 / D .; !1 ; !2 ; !3 / with pCq D 2 is a nearly hypo structure.
Proof Let us first observe, that one has

aC  ^ !2 D f  C
and a  ^ !3 D f  
;

which implies

d!Q 1 D d!1 D ˛ d. f  !/ D ˛f  .d!/

D 3˛ f C
D 3 ˛ aC  ^ !2 D 3˛aC Q ^ !Q 2 ;
98 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

a d.Q ^ !Q 3 / D a d. ^ !3 / D f  d 

D 2f  .! ^ !/ D 2 !1 ^ !1 D 2 !Q 1 ^ !Q 1 :

This proves the nearly hypo equations. t

u
Theorem 3.8.12 Let .N 5 ; ; !10 ; !20 ; !30 / be an "-Sasaki-Einstein SU " . p; q/-
structure, then


!1 D f"2 f" !10 C f"0 !30 ; (3.75)
!2 D f"2 !20 ; (3.76)


!3 D f"2 f"0 !10 C "f" !30 ; (3.77)
 D f" 0 (3.78)

with
( (
sinh.t/; for t 2 R and " D 1; R for " D 1;
f" .t/ WD sin" .t/ D and I" WD
sin.t/; for t 2 Œ0;  and " D 1 .0; / for " D 1

is a solution of the nearly hypo evolution equations and yields a nearly pseudo-
Kähler structure on N  I" Metrik angeben, anpassen with metric g D dt2 C f"2 gN
with conical singularities in f0g and f0; g respectively.
Proof Recall, that we have to solve (where we omit the Q) the following system

@t !1 D 3 !3  " d;
@t . ^ !3 / D d!2  4" ^ !1 ; (3.79)
@t . ^ !2 / D d!3 :

As Ansatz we consider the following family of SU" . p; q/-structures with p C q D 2


!1 D f 2 f !10 ˙ f 0 !30 ;
!2 D f 2 !20 ;


!3 D f 2 f 0 !10 ˙ "f !30 ;
 D f 0 ;
3.8 Evolution of Hypo Structures to Nearly Pseudo-Kähler Six-Manifolds 99

where we set f .t/ D sin" .t/; which yields f 00 D "f and f 0 2  "f 2 D 1: First we
compute ˛ˇ D 1
 
2
@t !1 D 3 f 2 f 0 !10 ˙ 2f f 0 C f 2 f 00 !30
./

D 3 f 2 f 0 !10 ˙ "f 3 !30 ˙ 2f !30
ŠŠ
D 3 !3 ˙ "fd0 D 3 !3 ˙ "d D 3 !3  "d;

where in ./ we used

2
2f . f 0 /2 C "f 3 D f .2f 0 C "f 2 / D 2f ."f 2 C 1/ C "f 3 D 3 " f 3 C 2f

and d0 D 2"!30 ; since N 5 is an "-Sasaki manifold. Hence we need to fix  D

1:
Next we calculate


@t . ^ !2 / D @t f 3 0 ^ !20 D 3f 2 f 0 0 ^ !20

and using d!30 D 0 and d!10 D 3aC bC 0 ^ !20 yields


bC aC d!3 D bC aC d5 f 2 f 0 !10 ˙ "f !30 D bC aC f 2 f 0 d5 !10 D 3 f 2 f 0 0 ^ !20 ;

which shows @t . ^ !2 / D bC aC d!3 D d!3 : It remains to determine the

evolution of  ^ !3


@t . ^ !3 / D @t ˙"f 4 0 ^ !30 C f 3 f 0 0 ^ !10


 
D  ˙4" f 3 f 0 0 ^ !30 C 3f 2 . f 0 /2 C f 3 f 00 0 ^ !10
./

D 4" ˙ f 3 f 0 0 ^ !30 C f 4 0 ^ !10 C 3f 2 0 ^ !10
b a D1
D 4"  ^ !1 C b a f 2 d!20 D 4"  ^ !1  f 2 d!20
D 4"  ^ !1  d!2 ;

where in ./ we used

2
3f 2 f 0 C " f 4 D 3f 2 ."f 2 C 1/ C "f 4 D 4 "f 4 C 3f 2 :

This yields

@t . ^ !3 / D 4"  ^ !1 C d!2

and finishes the proof of the Theorem. t

u
100 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.9 Results in the Homogeneous Case

3.9.1 Consequences for Automorphism Groups

An automorphism of an SU" . p; q/-structure on a six-manifold M is an automor-

phism of principal fibre bundles or equivalently, a diffeomorphism of M preserving
all tensors defining the SU" . p; q/-structure. By our discussion on stable forms in
Sect. 2.1 of Chap. 2, an SU" . p; q/-structure is characterised by a pair of compatible
stable forms .!; / 2 2 M  3 M. Since the construction of the remaining tensors
J;  and g is invariant, a diffeomorphism preserving the two stable forms is already
an automorphism of the SU" . p; q/-structure and in particular an isometry.
This easy observation has the following consequences when combined with the
exterior systems of the previous section and the naturality of the exterior derivative.
Proposition 3.9.1 Let .!; C
/ be an SU" . p; q/-structure on a six-manifold M.
(i) If the exterior differential equation
C
d! D 

is satisfied for a constant  ¤ 0, then a diffeomorphism ˆ of M preserving !

is an automorphism of the SU" . p; q/-structure and in particular an isometry.
(ii) If the exterior differential equation

d D !^!

is satisfied for a constant ¤ 0, then a diffeomorphism ˆ of M preserving

(a) the real volume form and C ;
(b) or the real volume form and  ;
C 
(c) or the "-complex volume form ‰ D C i" ;
is an automorphism of the SU" . p; q/-structure and in particular an isometry.
We like to emphasise that both parts of the Proposition apply to strict nearly
"-Kähler structures of non-zero type. The same holds true for the following
Proposition.
Proposition 3.9.2 Let .M 6 ; g; J " ; !/ be an almost "-Hermitian six-manifold with
totally skew-symmetric Nijenhuis tensor and ˆ be a diffeomorphism of M pre-
serving the almost "-complex structure J " . Suppose, that the structure J " is
quasi-integrable,
(i) then ˆ is a conformal map.
(ii) and additionally, assume, that one has d! 2 D 0; then ˆ is a homothety on
connected components of M: If moreover, ˆ preserves the volume, then it is an
isometry.
3.9 Results in the Homogeneous Case 101

Proof As ˆ preserves the "-complex structure, it also preserves the Nijenhuis

tensor. From Corollary 3.1.15 it follows g.X; Y/ D f tr.NX ı NY / for some function
f on M: This yields in p 2 M

gˆ. p/ .ˆ X; ˆ Y/ D f .ˆ. p//tr.Nˆ X ı Nˆ Y /


D f .ˆ. p// tr .ˆ N/X ı .ˆ N/Y D f .ˆ. p//f . p/ gp.X; Y/;

i.e. the conformal factor is c WD f ˆ f : Further let us assume, that one has d! 2 D 0:
From above we know ˆ .! ^ !/ D c2 .! ^ !/; which yields

0 D d.c2 ! ^ !// D d.c2 / ^ ! 2 C c2 d! 2 D d.c2 / ^ ! 2 :

Using, that the map  2 ƒ1 T  M 6 !  ^ ! 2 2 ƒ5 T  M 6 is an isomorphism, we

obtain d.c2 / D 0 and hence the function c is constant on connected components
of M: Recall, that the metric volume is a multiple of ! 3 : This implies, that one has
c D 1: t
u
Corollary 3.9.3 Let .M; J " ; g; !/ be a nearly "-Kähler six-manifold with krJ " k2 ¤
0; then a diffeomorphism ˆ of M preserving J " is an automorphism of the SU" . p; q/-
structure and in particular an isometry.
Proof The second nearly "-Kähler equation implies d! 2 D 0: Hence we obtain
from Proposition 3.9.2, that one has ˆ .!/ D c !; for some constant c (on each
connected component) and by the first nearly pseudo-Kähler equation

d.ˆ !/ c
ˆ . C
/D D d! D c C
:
3 3

As J " is preserved, this yields ˆ . 

/Dc 
and another time using the second
nearly pseudo-Kähler equation

cd D ˆ .d 
/ D ˆ .! 2 / D c2 ! 2 ;

forces c D 1: t
u
Conversely, it is known for complete Riemannian nearly Kähler manifolds,
that orientation-preserving isometries are automorphism of the almost Hermitian
structure except for the round sphere S6 , see for instance [26, Proposition 4.1].
However, this is not true if the metric is incomplete. In [53, Theorem 3.6], a nearly
Kähler structure is constructed on the incomplete sine-cone over a Sasaki-Einstein
five-manifold .N 5 ; ; !1 ; !2 ; !3 /. In fact, the Reeb vector field dual to the one-form
 is a Killing vector field which does not preserve !2 and !3 . Thus, by the formulae
given in [53], its lift to the nearly Kähler six-manifold is a Killing field for the sine-
cone metric which does neither preserve ‰ nor ! nor J.
102 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.9.2 Left-Invariant Nearly "-Kähler Structures

on SL.2 ; R/  SL.2 ; R/

The following lemma is the key to proving the forthcoming structure result, since
it considerably reduces the number of algebraic equations on the nearly "-Kähler
candidates.
Lemma 3.9.4 (Lemma 4.1 of [110]) Denote by .R1;2 ; h; i/ the vector space R3
endowed with its standard Minkowskian scalar-product and denote by SO0 .1; 2/ the
connected component of the identity of its group of isometries. Consider the action
of SO0 .1; 2/  SO0 .1; 2/ on the space of real 3  3 matrices Mat.3; R/ given by

ˆ W SO0 .1; 2/  Mat.3; R/  SO0 .1; 2/ ! Mat.3; R/

.A; C; B/ 7! At CB:

Then any invertible element C 2 Mat.3; R/ lies in the orbit of an element of the
form
0 1 0 1
˛ x y 0ˇ z
@0 ˇ zA or @˛ x y A
0 0 0 0

with ˛; ˇ; ; x; y; z 2 R and ˛ˇ ¤ 0:

Finally, we prove our main result of this subsection which is the following
theorem. By a homothety, we define the rescaling of the metric by a real number
which we do not demand to be positive since we are working with all possible
signatures.
Theorem 3.9.5 Let G be a Lie group with Lie algebra sl.2; R/. Up to homothety,
there is a unique left-invariant nearly "-Kähler structure with krJ " k2 ¤ 0 on G 
G. This is the nearly pseudo-Kähler structure of signature (4,2) constructed as 3-
symmetric space in Sect. 3.9.4. In particular, there is no left-invariant nearly para-
Kähler structure.
Remark 3.9.6 The proof also shows that there is a left-invariant nearly "-Kähler
structure of non-zero type on G  H with Lie.G/ D Lie.H/ D sl.2; R/ if G ¤ H
which is unique up to homothety and exchanging the orientation.
Proof More precisely, we will prove uniqueness up to equivalence of left-invariant
almost "-Hermitian structures and homothety. We will consider the algebraic
exterior system
C
d! D 3 ; (3.80)

d D 2! ^ ! (3.81)
3.9 Results in the Homogeneous Case 103

on the Lie algebra sl.2; R/ ˚ sl.2; R/. By Theorem 3.1.19, solutions of this system
are in one-to-one correspondence to left-invariant nearly "-Kähler structure on G 
G with krJ " k2 D 4. This normalisation can always be achieved by applying a
homothety. Furthermore, two solutions which are isomorphic under an inner Lie
algebra automorphism from

Inn.sl.2; R/ ˚ sl.2; R// D SO0 .1; 2/  SO0 .1; 2/

are equivalent under the corresponding Lie group isomorphism. Since both factors
are equal, we can also lift the outer Lie algebra automorphism exchanging the two
summands to the group level. In summary, it suffices to show the existence of a
solution of the algebraic exterior system (3.80), (3.81) on the Lie algebra which is
unique up to inner Lie algebra automorphisms and exchanging the summands.
A further significant simplification is the observation that all tensors defining a
nearly "-Kähler structure of non-zero type can be constructed out of the fundamental
two-form ! with the help of the first nearly Kähler equation (3.80) and the
stable form formalism described in Sect. 2.4 of Chap. 2. We break the main part
of the proof into three lemmas, step by step simplifying ! under Lie algebra
automorphisms in a fixed Lie bracket.
We call fe1 ; e2 ; e3 g a standard basis of so.1; 2/ if the Lie bracket satisfies

de1 D e23 ; de2 D e31 ; de3 D e12 :

In this basis, an inner automorphism in SO0 .1; 2/ acts by usual matrix multiplication
on so.1; 2/.
Lemma 3.9.7 Let g D h D so.1; 2/ and let ! be a non-degenerate two-form in

ƒ2 .g ˚ h/ D ƒ2 g ˚ .g ˝ h/ ˚ ƒ2 h :

Then we have

d! 2 D 0 , ! 2 g ˝ h: (3.82)

Proof By inspecting the standard basis, we observe that all two-forms on so.1; 2/
are closed whereas no non-trivial 1-form is closed. Thus, when separately taking
the exterior derivative of the components of ! 2 in ƒ4 D .ƒ3 g ˝ h / ˚ .ƒ2 g ˝
ƒ2 h / ˚ .g ˝ ƒ3 h ), the equivalence is easily deduced. t
u
Lemma 3.9.8 Let g D h D so.1; 2/ and let fe1 ; e2 ; e3 g be a basis of g and
fe4 ; e5 ; e6 g a basis of h such that the Lie brackets are given by

de1 D e23 ; de2 D e31 ; de3 D e12 and de4 D e56 ; de5 D e64 ; de6 D e45
(3.83)
104 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

for some  2 f˙1g. Then, every non-degenerate two-form ! on g ˚ h satisfying

d! 2 D 0 can be written

! D ˛ e14 C ˇ e25 C  e36 C x e15 C y e16 C z e26 (3.84)

for ˛; ˇ;  2 R  f0g and x; y; z 2 R modulo an automorphism in SO0 .1; 2/ 

SO0 .1; 2/.
Proof We choose standard bases fe1 ; e2 ; e3 g for g and fe4 ; e5 ; e6 g for
P h. Using the
previous lemma and the assumption d! 2 D 0, we may write ! D 3i;jD1 cij ei. jC3/


for an invertible matrix C D cij 2 Mat.3; R/. When a pair .A; B/ 2 SO0 .1; 2/ 
SO0 .1; 2/ acts on the two-form !, the matrix C is transformed to At CB. Applying
Lemma 3.9.4, we can achieve by an inner automorphism that C is in one of the
normal forms given in that lemma. However, an exchange of the base vectors e1 and
e2 corresponds exactly to exchanging the first and the second row of C. Therefore,
we can always write ! in the claimed normal form by adding the sign  in the Lie
bracket of the first summand g. t
u
Lemma 3.9.9 (Lemma 4.6 of [110]) Let fe1 ; : : : ; e6 g be a basis of so.1; 2/ 
so.1; 2/ such that

de1 D e23 ; de2 D e31 ; de3 D e12 and de4 D e56 ; de5 D e64 ; de6 D e45 :
(3.85)
Then the only SU" . p; q/-structure .!; C / modulo inner automorphisms and
modulo exchanging the summands, which solves the two nearly "-Kähler equa-
tions (3.80) and (3.81), is determined by
p
3 14
!D .e C e25 C e36 /: (3.86)
18
In fact, the uniqueness, existence and non-existence statements claimed in the
theorem follow directly from

this lemma and formula obtained for the quartic
invariant which implies  13 d! < 0.
As explained in Sect. 3.9.4, we know that there is a left-invariant nearly pseudo-
Kähler structure of indefinite signature on all the groups in question. After applying
a homothety, we can achieve krJ " k2 D 4 and this structure has to coincide with
the unique structure we just constructed. Therefore, the indefinite metric has to be
of signature (4,2) by our sign conventions.
We summarise the data of the unique nearly pseudo-Kähler structure in the
basis (3.85) and can easily double-check the signature of the metric explicitly:

1p
!D 3 .e14 C e25 C e36 /
18
C 1p
D 3 .e126  e135 C e156  e234 C e246  e345 /
54
3.9 Results in the Homogeneous Case 105

 1
D .2 e123 C e126  e135  e156  e234  e246 C e345 C 2 e456 /
54
1p 2p 2p 1p
J.e1 / D  3 e1  3 e4 ; J.e4 / D 3 e1 C 3 e4
3 3 3 3
1p 2p 2p 1p
J.e2 / D  3 e2 C 3 e5 ; J.e5 / D  3 e2 C 3 e5
3 3 3 3
1p 2p 2p 1p
J.e3 / D  3 e3 C 3 e6 ; J.e6 / D  3 e3 C 3 e6
3 3 3 3
1
gD . .e1 /2  .e2 /2  .e3 /2 C .e4 /2  .e5 /2
9
 .e6 /2  e1  e4  e2  e5  e3  e6 /:

t
u
Observing that in [25] very similar arguments have been applied to the Lie group
S3  S3 , we find the following non-existence result.
Proposition 3.9.10 On the Lie groups G  H with Lie.G/ D Lie.H/ D so.3/,
there is neither a left-invariant nearly para-Kähler structure of non-zero type nor a
left-invariant nearly pseudo-Kähler structure with an indefinite metric.
Proof The unicity of the left-invariant nearly Kähler structure S3  S3 is proved
in [25, Section 3], with a strategy analogous to the proof of Theorem 3.9.5. In the
following, we will refer to the English version [26]. There, it is shown in the proof
of Proposition 2.5, that for any solution of the exterior system
C
d! D 3
C
d D 2! 2

there is a basis of the Lie algebra of S3  S3 and a real constant ˛ such that

de1 D e23 ; de2 D e31 ; de3 D e12 and de4 D e56 ; de5 D e64 ; de6 D e45 ;
! D ˛.e14 C e25 C e36 /:

In this basis, a direct computation or formula (18) in [26] show that the quartic
invariant that we denote by  is

1 4
D ˛
27

with respect to the volume form e123456 . Therefore, a nearly para-Kähler structure
cannot exist on all the Lie groups with the same Lie algebra as S3  S3 by
Theorem 3.1.19. A nearly pseudo-Kähler structure with an indefinite metric cannot
exist either, since the induced metric is always definite as computed in the second
part of Lemma 2.3 in [26]. t
u
106 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.9.3 Real Reducible Holonomy

N
Nearly pseudo-Kähler manifolds admitting a J-invariant and r-parallel decompo-
sition of the tangent bundle TM are related to twistor spaces [108] (cf. [16, 98]
for Riemannian metrics) and Sects. 3.3 and 3.4 of this chapter. The next Proposition
considers the complementary situation, i.e. the case where TM decomposes into two
sub-bundles and J interchanges these sub-bundles and generalises a result of [99] to
pseudo-Riemannian metrics.
Proposition 3.9.11 Let .M; g; J/ be a complete, strict, simply connected nearly
N
pseudo-Kähler manifold. Suppose, that TM admits an orthogonal, r-parallel
0 0
decomposition TM D V ˚ V with V D JV; then .M; g/ is a homogeneous space.
Proof For a vector field X D JV1 in V 0 D JV; a vector field Y D JV2 in TM and
vector fields V3 ; V4 in V by the same argument as in Lemma 3.4.1 it is

N JV1 ; JV2 ; V3 ; V4 / D g .ŒrV3 J; rV4 JJV1 ; JV2 /  g ..rJV1 J/JV2 ; .rV3 J/V4 /
R.
D g .ŒrV3 J; rV4 JV1 ; V2 / C g ..rV1 J/V2 ; .rV3 J/V4 / (3.87)

By the symmetries (3.24) and (3.25) of the curvature tensor RN the last equation
determines R: N
N By Proposition 3.1.7 the torsion T and rJ are r-parallel. In
particular, we have

rN U ..rX J/Y/ D .rX J/rN U Y C .rrN U X J/Y:

Deriving (3.87) this implies rN RN D 0: Hence rN is an Ambrose-Singer connection

and as M is simply connected and complete it follows, that .M; J; g/ is a homoge-
neous space (see [116]). t
u

3.9.4 3-Symmetric Spaces

The idea of a three-symmetric space is to replace the symmetry of order two as in the
case of a symmetric space by a symmetry of order three. Nearly Kähler geometry
on such spaces was first studied in [66, 69].
Like symmetric spaces three-symmetric spaces have a homogenous model,
which we shortly resume: Let G be a connected Lie group and s an automorphism
of order 3 and let Gs0  H  Gs be a subgroup contained in the fix-point set Gs of
s: The differential s decomposes

g ˝ C D h ˝ C ˚ mC ˚ m
3.9 Results in the Homogeneous Case 107

p
into the eigenspaces of s with eigenvalues 1 and 12 .1 ˙ 3/: With the definition
m WD .mC ˚ m / \ g the decomposition

g D h˚m (3.88)

is reductive and G=H is a reductive homogenous space. The characteristic complex

structure is then defined by
p
1 3
sjm D  Id C J: (3.89)
2 2

The choice of an Ad.H/-invariant and s -invariant (pseudo-)Euclidean scalar-

product B on m endows G=H with the structure of a (pseudo-)Riemannian three-
symmetric space, such that B is almost Hermitian with respect to J: The next
Theorem locally relates homogeneous spaces to three-symmetric spaces.
Theorem 3.9.12 An almost pseudo-Hermitian manifold .M; J; g/ is a locally three-
symmetric space if and only if it is a quasi-Kähler manifold and the torsion T and
the curvature tensor RN of the characteristic Hermitian connection rN are parallel,
N D 0 and rN RN D 0:
i.e. rT
The proof of Theorem 3.9.12 is based on the following description of locally three-
symmetric spaces given in [66].
Theorem 3.9.13 Let .M; J; g/ be an almost pseudo-Hermitian manifold. Then there
exists a family of local cubic diffeomorphisms .sx /x2M such that J is the induced
complex structure and such that M is a three-symmetric space if and only if
(i) M is quasi-Kähler, i.e. one has .rX J/Y C .rJX J/JY D 0;
(ii)  D s preserves r 2 J;
(iii) for X; Y; Z; T 2 .TM/ one has

R.X; Y; Z; T/ D R. JX; JY; Z; T/ C R. JX; Y; JZ; T/ (3.90)

C R. JX; Y; Z; JT/;

(iv) for X; Y; Z; T 2 .TM/ one has

.rW R/.X; Y; Z; T/ C .rW R/. JX; JY; JZ; JT/ D 0:

Proof of Theorem 3.9.12 We claim that the conditions (i)-(iv) of Theorem 3.9.13
are equivalent to the following system of equations:

X Y C JX JY D 0; (3.91)
N D 0;
r (3.92)
rN RN D 0; (3.93)
108 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Let us recall the definition  WD 12 J.rJ/ which yields that the condition (3.91) is
equivalent to Theorem 3.9.13 part (i). From rX J D 2JX it follows


2
.rX;Y J/Z D 2J.2X Y Z  .rX /Y Z/ D 2J 2X Y Z  .rN X /Y Z  X Y Z C ŒY ; X Z :

By Eq. (3.92) this expression only depends on  and J; which are both preserved
by :
Conversely, we suppose that  preserves rJ and r 2 J: From Proposition 3.3 of
Gray [66] one obtains

N
r.X; N JX; JY; Z; T/ C r.
Y; Z; T/ D r. N JX; Y; JZ; T/
N JX; Y; Z; JT/ D 3r.
Cr. N JX; JY; Z; T/;

N D 0 and Eq. (3.91). This implies replacing

where the last equality follows from rJ
X by JX and Y by JY

N JX; JY; Z; T/ D r.X;

N 1N
3r. Y; Z; T/ D r. JX; JY; Z; T/
3

and finally we obtain r N D 0:

Moreover, the condition (3.92) implies Theorem 3.9.13 part (iii). In fact, we
claim that Theorem 3.9.13 part (iii) can be re-written as R 2 L2 ; where L2 is one
of the irreducible components of the space of curvature tensors considered as a
GL.n; C/ representation [52]. More precisely, in our case we identify u. p; q/ with
Œ1;1  (instead of u.n/) and u. p; q/? (rather than u.n/? ) with 2;0  and then apply
the results of [52] to obtain an analogous decomposition. In particular, it follows
in the quasi-Kähler case, that the complement of L2 only depends on r N and we
conclude that Eq. (3.92) implies Theorem 3.9.13 part (iii).
It remains to relate Theorem 3.9.13 part (iv) and Eq. (3.93). From r D rN C  it
follows

N
R.X; Y; Z; W/ D R.X; Y; Z; W/ C g.ŒX ; Y Z; W/  g.X YY X Z; W/;

where we use the condition (3.92). Using a second time the condition (3.92) and
N D 0 we get
rg

rN RN D rR
N D rR    R (3.94)

with

  R D R.; ; ; / C R.; ; ; / C R.; ; ; / C R.; ; ; /:

3.9 Results in the Homogeneous Case 109

Now Eq. (3.90) implies R. JX; JY; JZ; JW/ D R.X; Y; Z; W/; see for instance
Corollary 3.4 of [66]. As  and J anti-commute, it follows from (3.90)

.  R/. JX; JY; JZ; JW/ D .  R/.X; Y; Z; W/ (3.95)

and as rN is Hermitian we have

N
rR.X; N JX; JY; JZ; JW/:
Y; Z; W/ D rR. (3.96)

Equation (3.94) and (3.96) yield

2.rN W R/.X;
N Y; Z; T/ D .rN W R/.X;
N Y; Z; T/ C .rN W R/.
N JX; JY; JZ; JT/
Eqs: (3.94);(3.95)
D .rW R/.X; Y; Z; T/ C .rW R/. JX; JY; JZ; JT/;

which shows the equivalence of Theorem 3.9.13 part (iv) and Eq. (3.93). t
u
The following proposition relates the information coming from the Hermitian
structure to the data of the homogeneous space.
One may suppose G to be simply connected, since otherwise one considers its
universal cover  W G Q ! G and the isomorphic homogeneous space G= Q H
Q with
HQ D  1 .H/:

Proposition 3.9.14 Let .M D G=H; J; g/ be a (simply connected) reductive homo-

geneous almost pseudo-Hermitian manifold, then M D G=H is three-symmetric
if and only if it is quasi-Kähler and the connection rN coincides with the normal
connection r nor of the reductive homogeneous space G=H:
Proof Let .M D G=H; J; g/ be a reductive homogeneous space with adapted
reductive decomposition g D h ˚ m: The invariant almost complex structure J
induces a complex structure on m and an invariant decomposition

mC D m1;0 ˚ m0;1 : (3.97)

The invariance of m and J implies

Œh; m1;0   m1;0 and Œh; m0;1   m0;1 : (3.98)

The three-symmetry s is now obtained by the integration of the map 

jh D Idjh ; jm1;0 D jIdjm1;0 ; jm0;1 D j2 Idjm0;1 ;

p
where j D  12 Id C 23 i: The map  integrates (since G is supposed to be simply
connected) to s if and only if it is an automorphism of the Lie algebra g: By
Eq. (3.98) and the definition of  this is the case if and only if one has

Œm1;0 ; m1;0   m0;1 ; Œm0;1 ; m0;1   m1;0 and Œm1;0 ; m0;1   h: (3.99)
110 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Recall, that the torsion of the normal connection (see [87, Chapter X]) is given by
the invariant tensor

T nor .u; v/ D Œu; vm :

In terms of the torsion T nor the integrability conditions (3.99) are

0;1
T nor .u; v/ D Œu; vm ; for u; v 2 m1;0 ;
1;0
T nor .u; v/ D Œu; vm ; for u; v 2 m0;1 ;
T nor .u; v/ D 0; for u 2 m1;0 and v 2 m0;1 :

In other words T nor is a multiple of the Nijenhuis tensor.

Contraction

with
 the

metric yields a tensor g.T nor .; /; / which is of type ˝3 m1;0 ˚ ˝3 m0;1 w.r.t.
the complex structure induced by m1;0 ˚ m0;1 and skew-symmetric in the first
two entries. These symmetries exclude contributions of the pseudo-Riemannian
version of class W3 and W4 in the Gray-Hervella list [68] and hence .M; J; g/ is
of type W1 ˚ W2 ; i.e. .M; J; g/ is a quasi-Kähler manifold. Moreover, we have
T nor 2 2;0 ˝ 1;0  and we obtain r nor  r g 2 W1 ˚ W2  T  M ˝ u? p;q :
This means that r nor is the intrinsic connection which equals the characteristic
Hermitian connection. Summarizing (3.97) is the decomposition into eigenspaces
of an automorphism of order three if and only if .M D G=H; J; g/ is quasi-Kähler
and the normal connection coincides with the intrinsic connection. t
u
As a consequence the torsion and the curvature of the connection rN are given by

N v/ D Œu; vh with u; v 2 m:

T.u; v/ D Œu; vm and R.u; (3.100)

reductive, if it holds

B.ŒX; Ym ; Z/ D B.X; Œ Y; Zm / for X; Y; Z 2 m:

The next result was already shown in [66] Proposition 5.6 for pseudo-Riemannian
metrics. It is a consequence of r nor D rN for three-symmetric spaces.
Proposition 3.9.15 A three-symmetric space is a nearly pseudo-Kähler manifold if
and only if it is a naturally reductive homogeneous space.
In the sequel, we consider two homogeneous spaces G=H and G0 =H 0 which are T-
dual to each other in the sense of the construction given in Sect. 2.6.1 of Chap. 2 and
we are going to show that this construction is compatible with 3-symmetry.
As a preparation we recall the construction of the related complex structures. Let
us suppose, that g is a compact Lie algebra with a subalgebra h and that .M D
G=H; g; J/ is a Riemannian 3-symmetric space with a nearly Kähler structure of
3.9 Results in the Homogeneous Case 111

above discussed type. Moreover, denote by

g0 D gC ˚ ig ; m0 D mC ˚ im and h0 D hC ˚ ih

the associated decompositions of a fixed T-dual space M 0 D G0 =H 0 :

In this situation, there exists a natural almost complex structure J 0 on M 0 which
we shortly recall next, cf. Section 3.4 of [82]. Firstly, one decomposes gl.m/ into

gl.m/C WD fA 2 gl.m/ j A.mC /  mC ; A.m /  m g ;

gl.m/ WD fB 2 gl.m/ j B.mC /  m ; B.m /  mC g :

The Lie algebra gl.m/C ˚ igl.m/  gl.m/C is isomorphic to gl.m0 / via the
extension of the following definition

A.iX/ WD iA.X/; A. Y/ WD A. Y/; .iB/.iX/ WD B.X/; .iB/. Y/ WD iB. Y/;

where X 2 m and Y 2 mC and A 2 gl.m/C and B 2 gl.m/ : Further denote by

j 2 gl.m/ the linear map associated to the invariant complex structure J on G=H:
Assume on the one hand, that one even has j 2 gl.m/C ; then (as shown in
Proposition 3.5 of [82]) using the above identification of gl.m/C ˚ igl.m/ and
gl.m0 / the map j 2 gl.m/C  gl.m0 / induces an invariant almost complex structure
J 0 on G0 =H 0 ; such that if g is Hermitian for J then J 0 is pseudo-Hermitian for g0 : If
on the other hand one has j 2 gl.m/ ; then i j 2 gl.m/  gl.m0 / is a para-complex
structure on G0 =H 0 :
For the 3-symmetric case we recover the 3-symmetry using Eq. (3.89), i.e. for
j 2 gl.m/C one has
p
1 3
jm D sjm D  Id C j 2 gl.m/C ;
2 2

which induces as before an endomorphism of m0

p
1 3
jm0 D  Id C j 2 gl.m/C  gl.m0 /;
2 2

i.e. after extending  by the identity on h0 this yields a local 3-symmetry for G0 =H 0
and assuming G0 to be simply connected one may integrates  to a 3-symmetry of
G0 :
By construction it follows, that if .g; m; h; h; i/ is naturally reductive, the T-dual
.g0 ; m0 ; h0 ; h; i0 / is naturally reductive, too. Using Proposition 3.9.15 it follows,
that the T-dual of a nearly Kähler 3-symmetric space is nearly pseudo-Kähler.
Summarising our discussion we have shown.
Theorem 3.9.16 Let .G=H; J; g/ be a nearly Kähler 3-symmetric space (with
compact G) associated to .g; m; h; h; i/ with the above described nearly Kähler
112 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

structure and G0 =H 0 be a T-dual of G=H with data .g0 ; m0 ; h0 ; h; i0 / such that
the map j associated to the invariant complex structure J lies in gl.m/C ; then
.G0 =H 0 ; J 0 ; g0 / is a nearly pseudo-Kähler 3-symmetric space.
A natural question is starting with some homogeneous nearly Kähler manifold G=H
as above to give a classification of all T-dual spaces. Even though there is no general
answer to this question the cases of interest for the sequel are discussed in [82]. Let
us recall, that by [25] the list of homogenous strict nearly Kähler six-manifolds is

S6 D G2 =SU.3/;
CP3 D Sp.2/=.SU.2/  U.1//;
F.1; 2/ D SU.3/=.U.1/  U.1// and
S3  S3 D .SU.2/  SU.2/  SU.2//=
.SU.2//:

For these spaces all possible T-duals (given in [82]) are the following pruefen

S2;4 D G2 =SU.1; 2/; cf. Chap. 4 or [82];

Z.S2;2 / D SOC .2; 3/=U.1; 1/; c.f. Example 3.1 of [82];
Z.S4;0 =Z2 / D SOC .4; 1/=U.2/; c.f. Example 3.1 of [82];
Z.CP2;0 / D SU.2; 1/=.U.1/  U.1//; c.f. Example 3.2 of [82];
Z..SL.3; R/=GLC .2; R// D SLC .3; R/=R  SO.2/; c.f. Example 3.2 of [82];
SL.2; R/  SL.2; R/ D .SU.1; 1/  SU.1; 1/  SU.1; 1//=
.SU.1; 1//:

Let us recall, that the twistor spaces already appeared in Sects. 3.3 and 3.4 of the
present chapter. Moreover, one may wonders, if one obtains all nearly pseudo-
Kähler structures as T-duals of some homogeneous space G=H with a compact
Lie group G: The answer can be found in a recent preprint [11], where six-
dimensional homogeneous almost complex structures with semi-simple isotropy
have been classified. In this list an example of a left invariant nearly pseudo-Kähler
structure on a solvable Lie group is given (cf. Remark 3 of [11]), which does not
appear in the above list of T-dual spaces.

3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler

Manifolds

This section is based on results with Smoczyk and Schäfer [111] which are extended
to pseudo-Riemannian signature in Sects. 3.10.3 and 3.10.4.
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 113

3.10.1 Definitions and Geometric Identities

For the rest of this section let us assume that L  M is a Lagrangian submanifold11
of a nearly pseudo-Kähler manifold .M 2n ; J; g/ in the sense of the next definition.
Definition 3.10.1 Let .M 2n ; J; g; !/ be a nearly pseudo-Kähler manifold. A sub-
manifold W Ln ! M 2n is called Lagrangian submanifold, provided that one has
!jTL D  ! D 0; the dimension n of L is half the dimension 2n of M and that  g is
non-degenerate.
Since n D dim.L/ D 12 dim.M/ we have, that for a Lagrangian submanifold

g. JX; Y/ D 0 ; 8 X; Y 2 TL , J W TL ! T ? L is an isomorphism:

We observe that the (symmetric) signature of the metric g restricted to L is . p; q/ if

the signature of g is .2p; 2q/: From !jTL D 0 we deduce d!jTL D 0. On the other
hand (3.1) implies

d!.X; Y; Z/ D 3g..rX J/Y; Z/ :

From this and the symmetries of rJ the following Lemma easily follows (see also
[77]).
Lemma 3.10.2 Suppose L  M is a Lagrangian submanifold in a nearly Kähler
manifold .M; J; g/ with (possibly) indefinite metric. Then

.rX J/Y 2 T ? L ; 8 X; Y 2 TL ; (3.101)

.rX J/Y 2 T ? L ; 8 X; Y 2 T ? L ; (3.102)
? ?
.rX J/Y 2 TL ; if X 2 TL; Y 2 T L or if X 2 T L; Y 2 TL : (3.103)

Denote by II the second fundamental form of the Lagrangian immersion L  M 2n

into a nearly Kähler manifold M:
Proposition 3.10.3 For a Lagrangian submanifold L  M 2n in a nearly Kähler
manifold (with possibly indefinite metric) we have the following information.
(i) The second fundamental form is given by hII.X; Y/; Ui D hrN X Y; Ui for X; Y 2
.TL/ and U 2 .T ? L/:
(ii) The tensor C.X; Y; Z/ WD hII.X; Y/; JZi D !.II.X; Y/; Z/, 8 X; Y; Z 2 TL is
totally symmetric.

11
In order to compute expressions like for example rX Y one needs to extend the vector fields on L
to vector fields on M: It is common to use the same symbols for the extended vector fields, since
the induced objects do not depend on the choice of extension.
114 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

Proof From Lemma 3.10.2 we compute for X; Y 2 .TL/ and U 2 .T ? L/ the
second fundamental form II
1
hII.X; Y/; Ui D hrX Y; Ui D hrN X Y  J.rX J/Y; Ui D hrN X Y; Ui :
2

This yields part (i). Next we prove (ii): First we observe for X; Y; Z 2 .TL/

C.X; Y; Z/ D hII.X; Y/; JZi D hrN X Y; JZi D hY; rN X . JZ/i

D hY; J rN X Zi D hrN X Z; JYi D C.X; Z; Y/:

Since the second fundamental form is symmetric, it follows that C is totally

symmetric. t
u
Next we generalise an identity of [51] to nearly Kähler manifolds of arbitrary
dimension and signature of the metric. This and the next lemma will be crucial to
prove that Lagrangian submanifolds in strict nearly (pseudo-)Kähler six-manifolds
and in twistor spaces Z 4nC2 over quaternionic Kähler manifolds with their canonical
nearly Kähler structure are minimal. A six-dimensional version of the Lemma was
also proved in [71], see also Remark 3.10.7.
Lemma 3.10.4 The second fundamental form II of a Lagrangian immersion L 
M 2n into a nearly (pseudo-)Kähler manifold and the tensor rJ satisfy the following
identity

hII.X; J.rY J/Z/; Ui D hJ.rII.X;Y/ J/Z; Ui C hJ.rY J/II.X; Z/; Ui (3.104)

with X; Y 2 TL and U 2 T ? L:
N D 0; r.rJ/
Proof The proof of this identity uses rJ N D 0 and Lemma 3.10.2. With
X; Y; Z 2 .TL/ and U 2 .T ? L/ we obtain

N
rJD0
hII.X; J.rY J/Z/; Ui D hrN X . J.rY J/Z/; Ui D hJ rN X .rY J/Z/; Ui
N
r..rJ/D0  
D hJ .rrN X Y J/Z C .rY J/rN X Z ; Ui
D hJŒ.rZ J/rN X Y C .rY J/rN X Z; Ui
D hrN X Y; .rZ J/JUi C hrN X Z; .rY J/JUi
D hII.X; Y/; .rZ J/JUi C hII.X; Z/; .rY J/JUi
D hJ.rZ J/II.X; Y/; Ui C hJ.rY J/II.X; Z/; Ui
D hJ.rII.X;Y/ J/Z; Ui C hJ.rY J/II.X; Z/; Ui:

This is exactly the claim of the Lemma. t

u
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 115

Given the tensor C.X; Y; T.Z; V// we define the following traces

X
n
˛.X; Y/ WD i C.ei ; X; T.ei ; Y// ;
iD1

X
n
ˇ.X; Y; Z/ WD i C.T.ei ; X/; Y; T.ei ; Z// ;
iD1

where fe1 ; : : : ; en g is a local (pseudo-)orthonormal frame of TL and where we set

i D g.ei ; ei /:
Lemma 3.10.5 For a Lagrangian immersion in a nearly (pseudo-)Kähler manifold
and any X; Y; Z; V 2 TL holds:

C.X; Y; T.Z; V// C C.X; Z; T.V; Y// C C.X; V; T. Y; Z// D 0 ; (3.105)

!

˛.X; Y/  ˛. Y; X/ D h H ; JT.X; Y/i ; (3.106)
ˇ.X; Y; Z/ D ˇ.Z; Y; X/ D ˇ. Y; X; Z/ C ˛.T. Y; X/; Z/ ; (3.107)
˛.T.X; Y/; Z/ C ˛.T. Y; Z/; X/ C ˛.T.Z; X/; Y/ D 0 : (3.108)
!

Here H denotes the mean curvature vector of L.
Proof Let us first rewrite the identity in Lemma 3.10.4 in terms of the tensor C and
the torsion T.X; Y/ D J.rX J/Y of r: N Let X; Y; Z; V 2 .TL/ be arbitrary. Then
Lemma 3.10.4 gives

C.X; T.Z; Y/; V/ D hII.X; J.rY J/Z/; JVi

D hJ.rII.X;Y/ J/Z; JVi C hJ.rY J/II.X; Z/; JVi
D hJ.rZ J/. JV/; II.X; Y/i  hJ.rY J/. JV/; II.X; Z/i
D hJ 2 .rZ J/V; II.X; Y/i C hJ 2 .rY J/V; II.X; Z/i
D hJT.Z; V/; II.X; Y/i  hJT. Y; V/; II.X; Z/i
D C.X; Y; T.Z; V//  C.X; Z; T. Y; V// :

This is (3.105). Taking a trace gives

X
n
˛.X; Y/ D i C.ei ; X; T.ei ; Y//
iD1

(3.105) X
n X
n
D i C.ei ; ei ; T.X; Y// C i C.ei ; Y; T.ei ; X//
iD1 iD1
!

D h H ; JT.X; Y/i C ˛. Y; X/ ;
116 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

which is (3.106). The first identity in (3.107) is clear since C is fully symmetric. If
we apply (3.105) to ˇ.X; Y; Z/, then we get

X
n
ˇ.X; Y; Z/ D i C.T.ei ; X/; Y; T.ei ; Z//
iD1

X
n X
n
D i C.T.X; Y/; ei ; T.ei ; Z//  i C.T. Y; ei /; X; T.ei ; Z//
iD1 iD1

X
n X
n
D i C.ei ; T. Y; X/; T.ei ; Z// C i C.T.ei ; Y/; X; T.ei ; Z//
iD1 iD1

D ˛.T. Y; X/; Z/ C ˇ. Y; X; Z/ :

This is the second identity in (3.107). In view of this we also get

˛.T.X; Y/; Z/ C ˛.T. Y; Z/; X/ C ˛.T.Z; X/; Y/

D ˇ. Y; X; Z/  ˇ.X; Y; Z/ C ˇ.Z; Y; X/
ˇ. Y; Z; X/ C ˇ.X; Z; Y/  ˇ.Z; X; Y/
D0

and this is (3.108). t

u

3.10.2 Lagrangian Submanifolds in Nearly Kähler

Six-Manifolds

By a well known theorem of Ejiri [51] Lagrangian submanifolds of S6 are minimal.

In this section we will see that this is a special case of a much more general
theorem which is a consequence of Lemma 3.10.4 and was shown independently
for Riemannian metrics in Theorem 7 of [71].
Theorem 3.10.6 Let L3 be a Lagrangian immersion in a strict nearly (pseudo-
)Kähler six-manifold M 6 . Then we have

˛ D 0; (3.109)
!

H D 0: (3.110)

In particular, any Lagrangian immersion in a strict nearly (pseudo-)Kähler six-

manifold is orientable and minimal.
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 117

Proof Let fe1 ; e2 ; e3 g be an orthonormal basis of Tp L for a fixed point p 2 L. From

the skew-symmetry of hT.X; Y/; Zi we see that there exists a (nonzero) constant a
such that T.e1 ; e2 / D a3 e3 . Then we also have T.e2 ; e3 / D a1 e1 ; T.e3 ; e1 / D
a2 e2 . The symmetry of C implies

˛.e1 ; e1 / D 1 C.e1 ; e1 ; T.e1 ; e1 // C 2 C.e2 ; e1 ; T.e2 ; e1 // C 3 C.e3 ; e1 ; T.e3 ; e1 //

D 0  a2 3 C.e2 ; e1 ; e3 / C a3 2 C.e3 ; e1 ; e2 / D 0

and

˛.e1 ; e2 / D 1 C.e1 ; e1 ; T.e1 ; e2 // C 2 C.e2 ; e1 ; T.e2 ; e2 // C 3 C.e3 ; e1 ; T.e3 ; e2 //

D a1 3 C.e1 ; e1 ; e3 / C 0  a3 1 C.e3 ; e1 ; e1 / D 0 :

Similarly we prove that ˛.ei ; ej / D 0 for all i; j D 1; : : : ; 3. This shows ˛ D 0. But

!

then (3.106) also implies H D 0. The observation, that the frame fe1 ; e2 ; e3 g defines
an orientation on L; finishes the proof. The fact that a is a constant was not used in
the proof. t
u
Remark 3.10.7 The constant a in the formula T.e1 ; e2 / D ae3 from above is related
to the type constant ˛ of the nearly Kähler manifold M; cf. Sect. 3.1.1 of this chapter,
by the formula

a2 D ˛ :

A six-dimensional strict nearly Kähler manifold is of constant type and a nearly

Kähler manifold of constant type has dimension 6 [67]. The authors of [71] used the
Eq. (3.5) to obtain a six-dimensional version of Lemma 3.10.4 for arbitrary nearly
Kähler six-manifolds.
In the pseudo-Riemannian case we only have the relation a2 D j˛j : The sign
of the type constant depends on the signature .2p; 2q/ of g by sign. p  q/; see for
example [82], see also Sect. 3.1.1 of this chapter.
The connection induced on L by rN is intrinsic in the following sense.
Proposition 3.10.8 Let L be a Lagrangian submanifold in a strict nearly (pseudo-
)Kähler six-manifold. Then the connection rN L on L induced by the connection rN is
completely determined by the restriction of g to L:
Proof We observe that the torsion of rN L considered as a three-form is a constant
multiple of the volume form T L D c volLg : A metric connection D with prescribed
torsion T D is known to be unique. If the torsion is totally skew-symmetric we can
recover it from the formula

g 1
g.DX Y; Z/ D g.rX Y; Z/ C T.X; Y; Z/:
2

This finishes the proof, since T L D cvolLg is determined by g: t

u
118 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

3.10.3 The Splitting Theorem

The following example shows that Theorem 3.10.6 does not extend to eight
dimensions:
Example 3.10.9 Let L0  MSNK 6
be a (minimal) Lagrangian submanifold in a strict
6
nearly Kähler manifold MSNK and suppose   † is a curve on a Riemann surface
†. Then the Lagrangian submanifold L WD  L0  M in the nearly Kähler manifold
M WD †  MSNK is minimal, if and only if  is a geodesic in †.
In this section we will see that this is basically the only counterexample to
Theorem 3.10.6 that occurs in dimension 8. Nearly Kähler manifolds .M; J; g/
split locally into a Kähler factor and a strict nearly Kähler factor and under the
assumption, that M is complete and simply connected this splitting is global [98], cf.
Theorem 3.2.1 of this chapter for the pseudo-Riemannian case. The natural question
answered in the following theorem is in which way Lagrangian submanifolds lie in
this decomposition.
These facts motivate the next Theorem.
Theorem 3.10.10 Let M be a nearly Kähler manifold and L be a Lagrangian
submanifold. Then M and L decompose locally into products M D MK  MSNK ,
L D LK  LSNK , where MK is Kähler, MSNK is strict nearly Kähler and LK  MK ,
LSNK  MSNK are both Lagrangian. The dimension of LK is given by

1
dim LK D dim ker.r/
2
Moreover, if the splitting of M is global and L is simply connected, then L
decomposes globally as well.
Proof
i) We define

Kp WD fX 2 Tp M W rX D 0g ; Kp? WD f Y 2 Tp M W hX; Yi D 0 ; 8 X 2 Kp g :

N D 0; rg
Because of rr N D 0 this defines two orthogonal smooth distributions
[ [
DK WD Kp ; DSNK WD Kp?
p2M p2M

on M.
ii) The splitting theorem of de Rham can be applied, see Sect. 3.2.1 of this chapter,
to the distributions DK and DSNK and the nearly Kähler manifold .M; J; g/ splits
(locally) into a Riemannian product

.M; J; g/ D .MK ; JK ; gK /  .MSNK ; JSNK ; gSNK / ;

3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 119

where TMK D DK , TMSNK D DSNK : Here .MK ; JK ; gK / is Kähler and

.MSNK ; JSNK ; gSNK / is strict nearly Kähler.
iii) Now let L  M D MK MSNK be Lagrangian. We prove that r leaves tangent and
normal spaces of L invariant. To see this, we fix an adapted local orthonormal
frame field fe1 ; : : : ; e2n g of M such that e1 ; : : : ; en are tangent to L and enC1 D
Je1 ; : : : ; e2n D Jen are normal to L. Since for any three vectors X; Y; Z we have

h.rX J/JZ; .rY J/JZi D hJ.rX J/Z; J.rY J/Zi D h.rX J/Z; .rY J/Zi ;

we obtain
2n
X
hrX; Yi D h.rX J/ei ; .rY J/ei i
iD1

X
n X
n
D h.rX J/ei ; .rY J/ei i C h.rX J/Jei ; .rY J/Jei i
iD1 iD1

X
n
D2 h.rX J/ei ; .rY J/ei i:
iD1

Now, if X 2 TL, Y 2 T ? L, then by Lemma 3.10.2 we have

.rX J/ei 2 T ? L ; .rY J/ei 2 TL ;

so that

h.rX J/ei ; .rY J/ei i D 0 ; 8 i D 1; : : : ; n :

Further it follows

X
n
hrX; Yi D 2 h.rX J/ei ; .rY J/ei i D 0 :
iD1

Since this works for any X 2 TL; Y 2 T ? L and since r is selfadjoint we

conclude

r.TL/  TL ; r.T ? L/  T ? L :

At a given point p 2 L we may now choose an orthonormal basis f f1 ; : : : ; fn g of

Tp L that consists of eigenvectors of rjTL considered as an endomorphism of TL.
Since Œr; J D 0 and L is Lagrangian, the set f f1 ; : : : ; fn ; Jf1 ; : : : ; Jfn g then also
determines an orthonormal eigenbasis of r 2 End.TM/. In particular, since J
leaves the eigenspaces invariant, Kp D ker.r.p// and Tp L intersect in a subspace
120 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

KpL of dimension 12 dim.Kp / D 12 dim.MK /. For the same reason Kp? \ Tp L

gives an 12 dim.MSNK /-dimensional subspace. The corresponding distributions,
denoted by DKL and DSNK
L
are orthogonal and both integrable, since in view of

DKL D DK \ TL ; DSNK
L
D DSNK \ TL

they are given by intersections of integrable distributions. We may now apply

again the splitting theorem of de Rham to the Lagrangian submanifold. This
completes the proof.
t
u
Remark 3.10.11 A more detailed analysis of the proof of the last theorem shows,
that the result can be shown in the pseudo-Riemannian setting:
Let .M; J; g/ be a nearly pseudo-Kähler manifold. Suppose, that the distribution
K has constant dimension and admits an orthogonal complement, and the kernel of
rjL admits an orthogonal complement in TL, then M and L decompose locally into
products M D MK  MSNK , L D LK  LSNK , where MK is Kähler, MSNK is strict
nearly pseudo-Kähler and LK  MK , LSNK  MSNK are both Lagrangian. Moreover,
if the splitting of M is global and L is simply connected, then L decomposes globally
as well.
Corollary 3.10.12 If L  M is Lagrangian and p 2 L a fixed point, then to
each eigenvalue  of the operator r at p there exists a basis e1 ; : : : ; ek ; f1 ; : : : ; fk
of eigenvectors of Eig./ such that e1 ; : : : ; ek 2 Tp L, f1 ; : : : ; fk 2 Tp? L. Here, 2k
denotes the multiplicity of .
Proposition 3.10.13 Let .M; J; g/ be a nearly Kähler manifold and L  M be a
Lagrangian submanifold. Then TL and T ? L are invariant by the Ricci tensor. In
particular the spectrum of Ric is compatible with TL ˚ T ? L:
Proof Let us recall (cf. [98]) that the Ricci-tensor satisfies hRic X; Yi D 0 if X and Y
are vector fields in eigenbundles Eig.X / and Eig.Y / of the tensor r with different
eigenvalue X ¤ Y : If X; Y belong to the same eigenvalue  then hRic X; Yi is
given by the following formula

 1X
hRic X; Yi D hX; Yi C i hrEig.i / X; Yi; (3.111)
4  iD1

where  is the number of different eigenvalues of r and rEig.i / is defined by

hrEig.i / X; Yi D  trEig.i / Œ.rX J/ ı .rY J/:

Like in the proof of Theorem 3.10.10 we obtain using Corollary 3.10.12

rEig.i / .TL/  TL; rEig.i / .T ? L/  T ? L:

3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 121

Equation (3.111) implies that

Ric.TL/  TL; Ric.T ? L/  T ? L:

This further implies that the spectrum of Ric is compatible with the decomposition
TL ˚ T ? L and finishes the proof. t
u
Remark 3.10.14 The last proposition gives in the minimal case (only) a partial
information on the Ricci curvature of L  M: Recall the Gauss equation

hRL .V; W/X; Yi D hR.V; W/X; Yi C hII.V; X/; II.W; Y//i  hII.V; Y/; II.W; X/i

which implies using minimality

X
dimL X
dimL X
dimL
hRL .ei ; W/ei ; Yi D hR.ei ; W/ei ; Yi  hII.ei ; Y/; II.ei ; W/i:
iD1 iD1 iD1

It is straight-forward to show, that the second term on the right hand-side vanishes
if and only if II is zero, i.e. L is a totally geodesic manifold. In that case
Proposition 3.10.13 yields the Ricci tensor of L.
Let us recall the situation in dimension 8 and 10 [67, 98].
Proposition 3.10.15
(i) Let M 8 be a simply connected complete nearly Kähler manifold of dimension 8.
Then M 8 is a Riemannian product M 8 D †  MSNK 6
of a Riemannian surface †
6
and a six-dimensional strict nearly Kähler manifold MSNK :
10
(ii) Let M be a simply connected complete nearly Kähler manifold of dimension
10. Then M 10 is either the product MK4  MSNK
6
of a Kähler surface MK4 and
6
a six-dimensional strict nearly Kähler manifold MSNK or M is a twistor space
over a positive, eight dimensional quaternionic Kähler manifold.
Note, that any complete, simply connected eight dimensional positive quaternionic
Kähler manifold equals one of the following three spaces: HP2 ; Gr2 .C2 /; G2 =SO.4/:
In the next theorem, part (i) and (ii) collect the information on Lagrangian
submanifolds in nearly Kähler manifolds of dimension 8 and 10.
Theorem 3.10.16
(i) Let L be a Lagrangian submanifold in a simply connected nearly Kähler
manifold M 8 . Then M 8 D †  MSNK 6 6
, where † is a Riemann surface, MSNK
0
is strict nearly Kähler and L D   L is a product of a (real) curve   †
and a minimal Lagrangian submanifold L0  MSNK6
:
(ii) Let L be a Lagrangian submanifold in a simply connected complete nearly
Kähler manifold M 10 , then either
122 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

(a) M 10 D MK4  MSNK 6

and the manifold L D S  L0 is a product
of a Lagrangian (real) surface S  MK4 and a minimal Lagrangian
submanifold L0  MSNK6
or
(b) the manifold L is a Lagrangian submanifold in a twistor space over a
positive, eight dimensional quaternionic Kähler manifold.
(iii) Let M1 , M2 be two nearly Kähler manifolds. Denote the operator r on Mi by ri ,
i D 1; 2. If Spec.r1 / \ Spec.r2 / D ; and L  M1  M2 is Lagrangian, then L
splits (locally) into L D L1  L2 , where Li  Mi , i D 1; 2 are Lagrangian. If L
is simply connected, then the decomposition is global.
Proof This is a combination of the results of Theorem 3.10.10, Corollary 3.10.12
and Proposition 3.10.15. t
u
Remark 3.10.17 As in Remark 3.10.11 we may note, that using our results one can
generalise Theorem 3.10.16 to the case of indefinite metrics, where we omit part
(iii):
Theorem 3.10.18
(i) Let L be a Lagrangian submanifold in a simply connected nice nearly pseudo-
Kähler manifold M 8 . Then M 8 D †  MSNK
6
, where † is a Riemann surface,12
6 0
MSNK is strict nearly Kähler and L D  L is a product of a (real) curve   †
and a minimal Lagrangian submanifold L0  MSNK 6
:
(ii) Let L be a Lagrangian submanifold in a simply connected complete nice
decomposable nearly pseudo-Kähler manifold M 10 , such that the kernel13 of
rjTL admits an orthogonal complement, then either
(a) M 10 D MK4 MSNK6
and the manifold L D SL0 is a product of a Lagrangian
(real) surface S  MK4 and a minimal Lagrangian submanifold L0  MSNK 6

or
(b) the manifold L is a Lagrangian submanifold in a twistor space over
a negative, eight dimensional quaternionic Kähler manifold or a para-
quaternionic Kähler manifold.
Theorems 3.10.16 (3.10.16) and 3.10.18 (3.10.18), parts (b) motivate the discussion
of Lagrangian submanifolds in twistor spaces in the subsequent section. Indeed, the
results derived in the next section imply that Lagrangian submanifolds in twistor
spaces are, regardless their dimension, always minimal.

12
Remark, that the restriction of g to † is always definite.
13
Let us recall, that in the case (b) r has trivial kernel.
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 123

3.10.4 Lagrangian Submanifolds in Twistor Spaces

An important class of examples for nearly pseudo-Kähler manifolds is given

by twistor spaces Z 4nC2 over quaternionic Kähler or para-quaternionic Kähler
manifolds N 4n ; as we have seen in Sects. 3.3 and 3.4 of this chapter.
For the readers convenience, let us shortly recall that the twistor space is
the bundle of almost complex structures in the quaternionic bundle Q over the
(para-)quaternionic Kähler manifold N: It can be endowed with a Kähler structure
.Z; J Z ; gZ /; such that the projection  W Z ! N is a Riemannian submersion
with totally geodesic fibres S2 : Denote by H and V the horizontal and the vertical
distributions of the submersion : Then the direct sum decomposition

TZ D H ˚ V (3.112)

is orthogonal and compatible with the complex structure J Z : Let us consider now a
second almost Hermitian structure . J; g/ on Z which is defined by
8
ˆ
<g .X; Y/; for X; Y 2 H;
ˆ Z

g WD 12 gZ .V; W/; for V; W 2 V;

ˆ
:̂gZ .V; X/ D 0; for V 2 V; X 2 H

and
(
J Z on H;
J WD
J Z on V:

Note, that in view of (3.112), the decomposition TZ D H ˚ V is also compatible

w.r.t. J and orthogonal w.r.t. g.
The manifold .Z; J; g/ is a strict nearly pseudo-Kähler manifold and the distri-
butions V and H are parallel w.r.t. the connection r:N The projection  is also a
Riemannian submersion with totally geodesic fibres for the metric g:
Let us summarise some information which will be useful later in this section.
Lemma 3.10.19 In this situation we have the following information:
(a) The torsion T D JrJ of the characteristic connection satisfies (see
Lemma 3.4.3)

T.X; Y/ 2 V; for X; Y 2 H; (3.113)

T.X; V/ 2 H; for X 2 H; V 2 V; (3.114)
T.U; V/ D 0; for U; V 2 V: (3.115)
124 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

(b) The association (see Lemma 3.4.3)

H 3 X 7! T. Y; X/ 2 V (3.116)

is surjective for 0 ¤ Y 2 H and the map

ˆV W H 3 X 7! T.V; X/ (3.117)

with 0 ¤ V 2 V is invertible and squares to "k2 IdH for some k 2 R; " 2 f˙1g;
cf. Lemma 3.4.9.
(c) The operator r has eigenvalues H D 4k2 , V D n1 2 "H : If n > 1, then the
eigenbundle of H is H and V is the eigenbundle of V , cf. Corollary 3.4.17.
In the rest of this section we consider a nearly pseudo-Kähler manifold .M D
Z; J; g/ of twistor type and study Lagrangian submanifolds L  M:
Remark 3.10.20 As will be shown in the next theorem, for n > 1 we have

 H .TL/ D H \ TL ;  V .TL/ D V \ TL ;

where  H .TL/;  H .T ? L/ and  V .TL/;  V .T ? L/ are the orthogonal projections of

TL and T ? L w.r.t. H ˚ V.
Lemma 3.10.21 Let L2nC1  M 4nC2 with n > 1 be a Lagrangian submanifold in a
twistor space as described above. Then the second fundamental form II satisfies

II.X; Y/ 2  H .T ? L/; for X; Y 2  H .TL/; (3.118)

II.X; Y/ 2  V .T ? L/; for X; Y 2  V .TL/; (3.119)
H V
II.X; Y/ D 0; for X 2  .TL/; Y 2  .TL/: (3.120)

Proof The second fundamental form is given by C.X; Y; Z/ D hrN X Y; JZi for
N
X; Y; Z 2 .TL/: The lemma follows since the decomposition (3.112) is r-parallel,
orthogonal and J-invariant and as the tensor C is completely symmetric. t
u
With these preparations we prove the next result.
Theorem 3.10.22 Let L2nC1  M 4nC2 be a Lagrangian submanifold in a nearly
pseudo-Kähler manifold of the above type. Then L is minimal. If n > 1, then the
tangent space of L splits into a one-dimensional vertical part and a 2n-dimensional
horizontal part. Moreover, the second fundamental form II of the vertical normal
direction vanishes completely if n > 1.
Proof
i) By Theorem 3.10.6 it suffices to consider the case n > 1.
ii) Let L  M be a Lagrangian submanifold. Since n > 1, the two eigenvalues
H ; V of r are distinct and the eigenspace V of V is two-dimensional. By
Corollary 3.10.12 this induces a one-dimensional vertical tangential distribution
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 125

D on L in the Riemannian case. In the pseudo-Riemannian case this follows,

since the restriction of the metric to V is definite. In particular, D is not isotropic.
Then, by the Lagrangian condition, we get D WD  V .TL/.
iii) Denote by D? the orthogonal complement of D in TL: We claim, that the trace
of the second fundamental form II of L restricted to D? is zero.
Proof First we observe that by Lemma 3.10.21 we can restrict the second
fundamental form II to D? D  H .TL/: We fix U 2 D of unit length. Using
Lemmas 3.10.2 and 3.10.19 we observe, that ˆ.X/ WD 1k J.rU J/X defines an
(almost) (para-)complex structure on D? which is compatible with the metric.
With X 2 D? and ˆ2 D "Id we compute

1
II.X; X/ D "II.X; ˆ.ˆ.X/// D " II.X; J.rU J/ ˆ.X//
k
1  
D " J .rII.X;U/ J/ˆ.X/ C .rU J/II.X; ˆ.X//
k
1
D " J.rU J/ II.X; ˆ.X// D "ˆII.X; ˆ.X// :
k
After polarising we obtain

II.ˆX; ˆY/ D "II.X; Y/ ; 8 X; Y 2 D? : (3.121)

In particular, taking a trace over (3.121) we get

?
trD II D 0 ;

where we keep in mind, that it holds g.ˆ; ˆ/ D "g.; /:

iv) We have

˛.X; Y/ D 0 ; 8 X; Y 2 D? :

Proof By (ii) we may choose a pseudo-orthonormal frame fe1 ; : : : ; e2nC1 g of

TL such that e1 ; : : : ; e2n 2 D? and e2nC1 2 D. Since

2nC1
X
˛.X; Y/ D i C.ei ; X; T.ei ; Y//
iD1

and the tensor C is fully symmetric we see that by Lemma 3.10.21 all terms
on the RHS vanish since either ei 2 D D  V .TL/, X 2 D? D  H .TL/ or
ei ; X 2 D? and T.ei ; Y/ 2 D (cf. Lemma 3.10.19).
!
 !

v) By Lemma 3.10.21 and (iii) the mean curvature vector H satisfies J H 2 D :
From (3.106) and (iv) we get
!

hJ H ; T.X; Y/i D 0 ; 8 X; Y 2 D? : (3.122)
126 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

!

Since J maps V to itself, we also have J H 2 D  V. Now we choose X 2 D?
and YQ 2 H with

Q D J!
T.X; Y/

H:

This is possible since the map H 3 YQ 7! T.X; Y/Q 2 V is surjective by (3.116).

Q
Let Y D Y C Y be the orthogonal decomposition of YQ into the tangent and
?
Q Note that Y; Y ? are both horizontal. We claim
normal parts of Y.

T.X; Y ? / D 0:

This follows, since T.X; / maps tangent to tangent and normal to normal
vectors and one has

Q D T.X; Y/ C T.X; Y ? / D J!
T.X; Y/

H 2 TL :

Therefore there exist two tangent vectors X; Y 2 D? with

!

T.X; Y/ D J H :

This implies

!
 !
 .3.122/
j H j2 D hJ H ; T.X; Y/i D 0;

which proves that the mean curvature vector vanishes, as the metric restricted
to V is definite (even in the pseudo-Riemannian case). From this, the fact that
D is one-dimensional and from Lemma 3.10.21 it follows that II.V; / D 0 for
any V 2 D.
t
u
Corollary 3.10.23 Let L  M be a Lagrangian submanifold in a twistor space
M 4nC2 as above with n > 1. Then the integral manifolds c of the distribution D are
geodesics (hence locally great circles) in the totally geodesic fibres S2 :
Proof The last theorem implies that the geodesic curvature vanishes and that in
consequence an integral manifold c of D is totally geodesic in the fibres. u
t
Remark 3.10.24
(i) It is well-known, that the twistor space of HPn is CP2nC1 : Therefore the
above result applies to .CP2nC1 ; J; g/ endowed with its canonical nearly Kähler
structure.
(ii) Using Remark 3.10.14 (for Riemannian metrics) and Lemma 3.10.19 (c) we
observe that totally geodesic Lagrangian submanifolds in twistor spaces have
two different Ricci eigenvalues with multiplicities 2n and 1:
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 127

3.10.5 Deformations of Lagrangian Submanifolds in Nearly

Kähler Manifolds

Our aim in this section is to study the space of deformations of a given Lagrangian
(and hence minimal Lagrangian) submanifold L in a strict six-dimensional nearly
(pseudo-)Kähler manifold M 6 . In an article by Moroianu et al. [96] the deformation
space of nearly Kähler structures on six-dimensional nearly Kähler manifolds has
been related to the space of coclosed eigenforms of the Hodge-Laplacian. As we
will show below, a similar statement holds for the deformation of Lagrangian
submanifolds in strict nearly (pseudo-)Kähler six-manifolds. To this end we assume
that

F W L  .; / ! M

is a smooth variation of Lagrangian immersions Ft WD F.; t/ W L ! M, t 2 .; /

into a nearly (pseudo-)Kähler manifold M. Let

d
V WD Ft
dt
denote the variation vector field. Since tangential deformations correspond to
diffeomorphisms acting on L, we may assume w.l.o.g. that V 2 .T ? L/ is a normal
vector field. The Cartan formula and Ft ! D 0 for all t then imply that

0 D d.iV !/ C iV d!

holds everywhere on L. By the nearly Kähler condition this is equivalent to

d.Vy!/ C 3Vyr! D 0 (3.123)

on L. Let us define the variation 1-form  2 1 .L/ by

 WD Vy! :

This Theorem has recently been used in [97]. In this paper the authors relate
generalised Killing spinors on spheres to Lagrangian graphs in the nearly Kähler
manifold S3  S3 :
Theorem 3.10.25 Let Ft W L ! M be a variation of Lagrangian immersions in a
six-dimensional nearly (pseudo-)Kähler manifold M. Then the variation 1-form 
is a coclosed eigenform of the Hodge-Laplacian, where the eigenvalue  satisfies
 D 9˛ with the type constant ˛ of M. If the metric is positive definite this space is
finite dimensional.
128 3 Nearly Pseudo-Kähler and Nearly Para-Kähler Manifolds

In the case of Riemannian metrics a similar result was also shown in Theorem 7 of
[71]. For a more recent study of deformations of Lagrangian submanifolds we refer
to [91].
Proof For X; Y 2 TL and V 2 T ? L we compute

.Vyr!/.X; Y/ D r!.V; X; Y/
D h.rX J/Y; Vi
D hJ.rX J/Y; JVi
D hJV; T.X; Y/i :

Since T induces an orientation on the Lagrangian submanifold by the three-form

.X; Y; Z/ WD hT.X; Y/; Zi ;

we obtain a naturally defined -operator  W p .L/ ! 3p .L/ which for 1-forms
is given by

 WD p  ı T :
j˛j

Here, ˛ is the type constant of M (cf. Remark 3.10.7) and  2 f˙1g depends only
on the signature. This implies that equation (3.123) can be rewritten in the form
p
d D 3 j˛j  : (3.124)

Consequently, if the signature of the metric g restricted to L is . p; q/; we obtain

sign. p  q/ ı D d D 0

and
p
sign. p  q/ ıd D j˛j d
3
2
. DId/ p
D 3 j˛j d
(3.124)
D 9j˛j  
.2 DId/
D 9j˛j :
3.10 Lagrangian Submanifolds in Nearly Pseudo-Kähler Manifolds 129

In total as the sign of ˛ is also sign. p  q/ we get

Hodge  D .ıd C dı/ D 9˛ :

This proves the theorem. Since one has Ric D 5˛g this is equivalent to

3

Hodge  D scal  ;
10
where scal is the scalar curvature of M. t
u
Chapter 4
Hitchin’s Flow Equations

This chapter relies on work with Leistner et al. [46].

./
4.1 Half-Flat Structures and Parallel G2 -Structures

Now we want to put the algebraic structures considered in the first chapter onto
smooth manifolds. This is best done in terms of reductions of the bundle of frames
of the manifold. This bundle has GL.n; R/ as its structure group if n is the dimension
of the manifold. A subbundle whose structure group is a subgroup G of GL.n; R/
is a called a reduction of the frame bundle, or a G-structure. For example, if G 
O. p; q/, for p C q D n, the reduction determines a pseudo-Riemannian metric of
signature . p; q/ and the distinguished frames are orthonormal with respect to this
metric. If G  O. p; q/, again with p C q D n, then a G-structure is called parallel if
the G-subbundle is invariant under the parallel transport defined by the Levi-Civita
connection of the corresponding metric. This is equivalent to the property that the
holonomy group of the Levi-Civita is contained in G.
In the following we will consider G-structures that are given by the groups
described in the previous sections. According to the notations given there, we denote
by H "; a real form of SL.3; C/ and G"; the corresponding real form of GC 2 in which
H "; is embedded, i.e. H 1;1 D SU.3/  SO.6/, H 1;1 D SU.1; 2/  SO.2; 4/,
H 1;1 D SL.3; R/  SO.3; 3/, G1;1 D G2  SO.7/, and G1;1 D G1;1 D G2 
./
SO.3; 4/. We will also use the notation G2 as a shorthand for “G2 respectively G2 ”.
An H "; -structure is equivalent to a pair of everywhere stable forms ! 2 2 M
and  2 3 M on M, considered up to rescaling of  by a nonzero constant, that
satisfy the compatibility condition

^! D0 (4.1)

© Springer International Publishing AG 2017 131

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0_4
132 4 Hitchin’s Flow Equations

corresponding to (2.16) and in addition

1
./ D c .!/ ; i.e. J  ^  D c !3; (4.2)
3
for a positive real constant c. Indeed, if an H "; -structure is given, these forms are
obtained by applying the formulae (2.7) and (2.20) to one of the frames of the H "; -
structure. By construction, the stable forms then satisfy (4.2) with c D 2.
On the other hand, if ! 2 2 M and  2 3 M are everywhere stable and
satisfy (4.1) and (4.2), we can find a local frame, in which they are in normal form
after rescaling  by a constant. This frame then determines the H "; -structure.
Note that stable forms define an H "; -structure, even if they only satisfy (4.1) but
not the second compatibility condition (4.2). In this case  can always be rescaled by
a smooth function such that (4.2) holds. When we say that the pair of stable forms
defines an H "; -structure, we will always assume that both compatibility conditions
are satisfied. We will call the H "; -structure normalised if c D 2. This seems to be a
common normalisation for SU.3/-structures in the literature.
Furthermore, one can show that the H "; -structure is parallel if and only if
, ,
O and ! are closed. The proof of this fact given in [75, p. 567] generalises
to SU.1; 2/-structures and also to SL.3; R/-structures, in the latter case using
Frobenius’ Theorem instead of the Newlander-Nirenberg Theorem. In all cases the
parallel H "; -structure is equivalent to M being a Ricci-flat (para-)Kähler manifold.
Now we consider a weaker condition, that will turn out to be related to parallel
./
G2 -structures.
Definition 4.1.1 An H "; -structure .; !/ is called half-flat if

d D 0; (4.3)
d D 0; (4.4)

where 2 D ! 2 .
./
Similarly, a smooth seven-manifold admits a G2 -structure if and only if there
is a stable three-form '. Again, this structure is parallel if and only if ' is closed
and co-closed, i.e. d' D d  ' D 0, where  denotes the Hodge operator with
./
respect to the metric induced by the G2 -structure. For a proof in both cases see
[64, Theorem 4.1].
Note that any orientable hypersurface in a manifold with G2 - or G2 -
structure admits an H "; -structure by the algebraic construction described in
./
Proposition 2.1.14. If the G2 -structure ' is parallel, the induced H "; -structure
is half-flat due to Eqs. (2.23) and (2.27). For the various results on the SU.3/-
structures on hypersurfaces in G2 -structures, we refer to [27, 28] and references
therein.
On the other hand, certain one-parameter families of half-flat structures define
./
parallel G2 -structures.
 ./
4.1 Half-Flat Structures and Parallel G2 -Structures 133

Proposition 4.1.2 Let H "; be a real form of SL.3; C/, G"; the corresponding real
form of GC ";
2 and .; !/ a one-parameter family of H -structures on a six-manifold
M with a parameter t from an interval I. Then, the three-form

' D ! ^ dt C 

defines a parallel G"; -structure on M  I if and only if the H "; -structure .; !/ is
half-flat for all t and satisfies the following evolution equations

P D d!; (4.5)
P D d O (4.6)

with  D 12 ! 2 .
Proof Let .; !/ be an H "; -structure and ' D ! ^ dt C  a stable three-form on
ML WD M  I. By (2.27), the Hodge-dual of ' is given by

' D " .O ^ dt  / :

Denoting by dL the differential on ML and by d the differential on M we calculate

L D d! ^ dt C dt ^ P C d D .d!  /
d' P ^ dt C d; (4.7)
dL  ' D " .d O ^ dt  dt ^ P  d/ D ".d O  P / ^ dt  "d: (4.8)

Thus, ' defines a parallel G"; -structure if and only if the evolution equations (4.5)
and (4.6) and the half-flat equations are satisfied. t
u
The evolution equations (4.5) and (4.6) are the Hitchin flow equations, as found
in [76] for SU.3/-structures, applied to H "; -structures. Their solutions .; !/, called
Hitchin flow, have to satisfy possibly dependent conditions in order to yield a
./
parallel G2 -structure: the evolution equations and the compatibility equations for
the family of half-flat structures. The following theorem shows that the evolution
equations together with an initial condition already ensure that the family consists
of half-flat structures. A special version of this theorem was proved in [76] under
the assumption that M is compact and that H D SU.3/.
Theorem 4.1.3 Let .0 ; !0 / be a half-flat H "; -structure on a six-manifold M.
Furthermore, let .; !/ 2 3 M  2 M be a one-parameter family of stable forms
with parameters from an interval I satisfying the evolution equations (4.5) and (4.6).
If ..t0 /; !.t0 // D .0 ; !0 / for a t0 2 I, then .; !/ is a family of half-flat H "; -
structures. In particular, the three-form

' D ! ^ dt C  (4.9)
134 4 Hitchin’s Flow Equations

defines a parallel G"; -structure on M  I and the induced metric

g' D g.t/  "dt2 ; (4.10)

has holonomy contained in G"; , where g D g.t/ is the family of metrics on M

associated to .; !/.
Proof Differentiating the evolution equations (4.5) and (4.6) gives d P D d P D 0.
The initial condition for t0 was that .0 ; !0 / is half-flat. This implies

d D 0;
d D 0

for all t 2 I. Hence, in order to obtain a family of half-flat structures we have to

verify that the compatibility condition (4.1) holds for all t 2 I.
Lemma 4.1.4 Let M be a six-manifold with H "; -structure .; !/,  W 3 M !
6 M defined pointwise by the map  W ƒ3 Tp M ! ƒ6 Tp M given in Proposi-
tion 2.1.4 and O defined by d ./ D O ^  for all  2 3 M. If LX denotes the
Lie derivative, then

LX ..// D O ^ LX :

Proof First note that the GL.n; R/-equivariance of the map  W ƒ3 Tp M ! ƒ6 Tp M
implies that the corresponding map  W 3 M ! 6 M is equivariant under
diffeomorphisms. Indeed, if is a (local) diffeomorphism of M we get that
 
..// D . /:

Let t be the flow of the vector field X. Then the Lie derivative is given by

d

 d 
LX ..// D t ./ jtD0 D . t /jtD0 D d .LX /;
dt dt
implying the statement. t
u
3
Lemma 4.1.5 A stable three-form  2  M on a six-manifold satisfies for any
X 2 X.M/

OX ^  D O ^ X ; (4.11)

O X ^  D O ^ .d/X ;
.d / (4.12)

where X denotes the interior product of X with the form .

Proof In order to verify the first identity, we can assume that  D p is a stable
three-form on V D Tp M and X 2 V for a p 2 M. If ./ < 0, the stabiliser SL.3; C/
 ./
4.1 Half-Flat Structures and Parallel G2 -Structures 135

of  in GLC .V/ acts transitively on V n f0g. If ./ > 0, we can decompose V in

the ˙1-eigenspaces V ˙ of J . The stabiliser SL.3; R/  SL.3; R/ of  in GLC .V/
acts transitively on the dense open subset V C n f0g  V  n f0g  V and there is an
automorphism exchanging V C and V  which stabilises . Thus, it suffices to verify
the first identity for the normal form (2.7), (2.8) and X D e1 , which is easy.
For the second identity, using Lemma 4.1.4 in the second step, we compute

O X ^   O ^ .d/X D dO ^ X C O ^ d.X /  O ^ LX 

.d /
D d.O ^ X /  LX ..//
D d.O ^ X C ./X /
1
(2.2)
D  d.O ^ X C OX ^ /:
2
Hence, the first identity (4.11) implies (4.12). t
u
Using this lemma, we calculate the t-derivative of the six-form !X ^ ! ^  D X ^ 
for any vector field X:

@
.X ^ / D P X ^  C X ^ P
@t
(4.5 ),(4.6 )
D .d /
O X ^  C X ^ d!
(4.12)
D O ^ .d/X C !X ^ ! ^ d!
(4.3);(4.4)
D 0:
Together with the initial condition !0 ^ 0 D 0 this implies that X ^  D 0 for
all t 2 I and for all vector fields X. Since ! is nondegenerate, the product of any
one-form with ! ^  vanishes and thus, the compatibility condition ! ^  D 0 holds
for all t.
The preservation of the normalisation (4.2) in time is shown in [76], in the final
part of the proof of Theorem 8. The idea is to compute the second derivative of the
volume form assigned to a stable three-form. In fact, the proof holds literally for all
signatures since all it uses is the first compatibility condition we have just proved.
t
u
Corollary 4.1.6 Let M be a real analytic six-manifold with a half-flat H "; -structure
that is given by a pair of analytic stable forms .!0 ; 0 /.
(i) Then, there exists a unique maximal solution .!; / of the evolution equa-
tions (4.5), (4.6) with initial value .!0 ; 0 /, which is defined on an open
neighbourhood   R  M of f0g  M. In particular, there is a parallel
G"; -structure on .
(ii) Moreover, the evolution is natural in the sense that, given a diffeomorphism f
of M, the pullback . f  !; f  / of the solution with initial value .!0 ; 0 / is the
solution of the evolution equations for the initial value . f  !0 ; f  0 /.
136 4 Hitchin’s Flow Equations

In particular, if f is an automorphism of the initial structure .!0 ; 0 /, then,

for all t 2 R, f is an automorphism of the solution .!.t/; .t// defined on the
(possibly empty) open set Ut D f p 2 M j .t; p/ 2  and .t; f . p// 2 g.
(iii) Furthermore, assume that M is compact or a homogeneous space M D
G=K such that the H "; -structure is G-invariant. Then there is a unique
maximal interval I 3 0 and a unique solution .!; / of the evolution
equations (4.5), (4.6) with initial value .!0 ; 0 / on I  M. In particular, there
is a parallel G"; -structure on I  M.
Proof If the manifold and the initial structure .!0 ; 0 / are analytic, there exists a
unique maximal solution of the evolution equations on a neighbourhood  of M 
f0g in M  R by the Cauchy-Kovalevskaya theorem. The naturality of the solution
is an immediate consequence of the uniqueness due to the naturality of the exterior
derivative. If M is compact, there is a maximal interval I such that the solution is
defined on M  I. The same is true for a homogeneous half-flat structure .!0 ; 0 / as
it is determined by .!0 ; 0 /j p for any p 2 M. t
u
We remark that, for a homogeneous half-flat structure .!0 ; 0 /, the evolution
equations reduce to a system of ordinary differential equations due to the naturality
assertion of the corollary. This simplification will be used in Sect. 4.3.3 to construct
metrics with holonomy equal to G2 and G2 .

4.1.1 Remark on Completeness: Geodesically Complete

Conformal G2 -Metrics

./
The G2 -metrics arising from the Hitchin flow on a six-manifold N are of the form
.I N; dt2 Cgt / with an open interval I D .a; b/ and a family of Riemannian metrics
gt depending on t 2 I (formula (4.10) in Theorem 4.1.3). As curves of the form
t 7! .t; x/ are geodesics for this metric, they are obviously geodesically incomplete
if a or b 2 R.
For the Riemannian case and compact manifolds N, we shall explain how one
easily obtains complete metrics by a conformal change of the G2 -metric.
Lemma 4.1.7 Let N be a compact manifold with a family gr of Riemannian metrics.
Then the Riemannian metric on R  N defined by h D dr2 C gr is geodesically
complete.
Proof Denote by d the distance on R  N induced by the Riemannian metric h D
dr2 C gr and by dr the distance on N induced by gr . For a curve  in M D R  N
we have that the length of .t/ D .r.t/; x.t// satisfies
Z 1 q Z 1
`. / D rP .t/2 C gr.t/ .Px.t/; xP .t//dt  jPr.t/jdt  jr.1/  r.0/j:
0 0
 ./
4.1 Half-Flat Structures and Parallel G2 -Structures 137

As the distance of two points p D .r; x/ and q D .s; y/ is defined as the infimum of
the lengths of all curves joining them, this inequality implies that

d. p; q/  jr  sj: (4.13)

Note also that a curve .t/ D ..s  r/t C r; x/ joining p D .r; x/ and q D .s; x/ in
Rfxg has length `. / D jr sj and thus, for such p, q we get that d. p; q/ D jr sj.
On the other hand, for p D .r; x/ and q D .r; y/ with the same R-projection r we
only get that d. p; q/  dr .x; y/.
Since h has Riemannian signature we can use the Hopf-Rinow Theorem and
consider a Cauchy sequence pn D .rn ; xn / 2 R  N w.r.t. the distance d.
Equation (4.13) then implies that the sequence rn is a Cauchy sequence in R. Hence,
rn converges to r 2 R. Since N is compact, the sequence xn has a subsequence xnk
converging to x 2 N. For p D .r; x/ and qnk WD .r; xnk / the triangle inequality
implies that

d. p; pnk /  d . p; qnk / C d .qnk ; pnk /  dr .x; xnk /C d .qnk ; pnk / D dr .x; xnk /C jr  rnk j:

Hence, pnk converges to p. As pn was a Cauchy sequence, we have found p as a

limit for pn . By the Theorem of Hopf and Rinow, M is geodesically complete. u
t
The consequence of the lemma is
Proposition 4.1.8 Let .M D I  N; h D dt2 C gt / be a Riemannian metric on a
product of an open interval I and a compact manifold N. Then .M; h/ is globally
conformally equivalent to a metric on R  N that is geodesically complete. The
scaling factor depends only on t 2 I and is determined by a diffeomorphism ' W
R ! I.
Proof Let ' W R ! I be a diffeomorphism with inverse r D ' 1 . Changing the
coordinate t to r, the metric h on I  N can be written as
 

2 1
h D ' 0 .r/dr C g'.r/ D ' 0 .r/2 dr2 C 0 2 g'.r/ :
' .r/

1
Hence, h is globally conformally equivalent to the metric dr2 C ' 0 .r/ 2 g'.r/ on R  N.

By the lemma, this metric is geodesically complete. t

u
Regarding the solution of the Hitchin flow equations, using Theorem 4.1.3,
Corollary 4.1.6, and Proposition 4.1.8 we obtain the following consequence.
Corollary 4.1.9 Let M be a compact analytic six-manifold with half-flat SU.3/-
structure given by analytic stable forms .0 ; !0 /. Then there is a complete metric on
R  M that is globally conformal to the parallel G2 -metric obtained by the Hitchin
flow.
In Example 4.3.15 of Sect. 4.3.3 we will construct explicit examples of this type.
Finally, note that due to the Cheeger-Gromoll splitting Theorem, see for example
138 4 Hitchin’s Flow Equations

[18, Theorem 6.79], one cannot expect to obtain by the Hitchin flow irreducible
G2 -metrics that are complete without allowing degenerations of gt .

4.1.2 Nearly Half-Flat Structures and Nearly Parallel

./
G2 -Structures

./
A G2 -structure ' on a seven-manifold N is called nearly parallel if

d' D  ' ' (4.14)

for a constant  2 R . Nearly parallel G2 - and G2 -structures are also characterised
by the existence of a Killing spinor, refer [58] respectively [81].
./
By Proposition 2.1.14, a G2 -structure on a seven-manifold .N; '/ induces an
"; ./
H -structure .!; / on an oriented hypersurface in .N; '/. If the G2 -structure
";
is nearly parallel, the H -structure satisfies the equation d D "!O due to the
formulas (2.23) and (2.27). This observation motivates the following definition (see
also Sect. 3.1.2 of Chap. 3).
Definition 4.1.10 An H "; -structure .!; / on a six-manifold M is called nearly
half-flat if

 2
d D ! D  (4.15)
2

for some constant  2 R .

The notion of a nearly half-flat SU.3/-structure was introduced in [53], where also
evolution equations on six-manifolds leading to nearly parallel G2 -structures are
considered. For compact manifolds M, it is shown in [114] that a solution which is
a nearly half-flat SU.3/-structure for a time t D t0 already defines a nearly parallel
G2 -structure. In the following, we extend these evolution equations to all possible
signatures and give a simplified proof for the properties of the solutions which also
holds for non-compact manifolds.
Proposition 4.1.11 Let H "; be a real form of SL.3; C/, G"; the corresponding real
form of GC ";
2 and .; !/ a one-parameter family of H -structures on a six-manifold
M with a parameter t from an interval I. Then, the three-form

' D ! ^ dt C 

defines a nearly parallel G"; -structure for the constant  ¤ 0 on M  I if and only
if the H "; -structure .; !/ is nearly half-flat for the constant " for all t 2 I and
 ./
4.1 Half-Flat Structures and Parallel G2 -Structures 139

satisfies the evolution equation

P D d!  ":
O (4.16)

Proof The assertion follows directly from the following computation, analogously
to the proof of Proposition 4.1.2:

L D d! ^ dt C dt ^ P C d D .d!  /
d' P ^ dt C d;
  ' D " .O ^ dt  / :

t
u
The main theorem for the parallel case generalises as follows. Recall (2.3) that
for a stable four-form  D 12 ! 2 D !,
O the application of the operator  7! O yields
the stable two-form
1
!OO D O D !:
2

Theorem 4.1.12 Let .0 ; !0 / be a nearly half-flat H "; -structure for the constant
 ¤ 0 on a six-manifold M. Let M be oriented such that !03 > 0. Furthermore, let
 2 3 M be a one-parameter family of stable forms with parameters coming from
an interval I such that .t0 / D 0 and such that the evolution equation

2 b
P D d.d/ C  O (4.17)


b is a family of nearly half-flat H "; -

is satisfied for all t 2 I. Then .; ! D 2 d/
structures for the constant . In particular, the three-form

' D ! ^ dt C 

defines a nearly parallel G"; -structure for the constant " on M  I.

Proof First of all, we observe that d is stable in a neighbourhood of the stable
form d0 D 0 , since stability is an open condition. Furthermore, the operator
d 7! d b is uniquely defined by the orientation induced from !0 . Therefore, the
evolution equation is locally well-defined and we assume that  is a solution on an
interval I. The only possible candidate for a nearly half-flat structure for the constant
b since only this two-form ! satisfies the nearly half-flat equation
 is .; ! D 2 d/
 D !O D 1 d. Obviously, it holds

d D 0 D d! ^ !: (4.18)

By Proposition 4.1.11, it only remains to show that this pair of stable forms defines
an H "; -structure, or equivalently, that the compatibility conditions (4.1) and (4.2)
140 4 Hitchin’s Flow Equations

are preserved in time. By taking the exterior derivative of the evolution equation, we
find
1
P D d P D dO (4.19)

which is in fact the second evolution equation of the parallel case. Completely anal-
ogous to the parallel case, the following computation implies the first compatibility
condition:

@
.X ^ / D P X ^  C X ^ P
@t
(4.17 ),(4.19 )
D .d /
O X ^  C X ^ d! C  X ^ O
(4.12),(4.15)
D O ^ .d/X C !X ^ ! ^ d! C .d/X ^ O
(4.18)
D 0:

The proof of the second compatibility condition in [76] again holds literally since
the term O ^ P D O ^ d! is the same as in the case of the parallel evolution. t
u
The system (4.17) of second order in  can easily be reformulated into a system
of first order in .!; / to which we can apply the Cauchy-Kovalevskaya theorem.
Indeed, a solution .!; / of the system

P D d! C O ; P D d ;
O (4.20)

with nearly half-flat initial value .!.t0 /; .t0 // is nearly half-flat for all t and also
satisfies the system (4.17). Conversely, (4.17) implies (4.20) with  D !O D 1 d.
Therefore, for an initial nearly half-flat structure which satisfies assumptions
analogous to those of Corollary 4.1.6, we obtain existence, uniqueness and natu-
rality of a solution of the system (4.20), or, equivalently, of (4.17).

./
4.1.3 Cocalibrated G2 -Structures and Parallel Spin.7/-
and Spin0 .3; 4/-Structures

In [76], another evolution equation is introduced which relates cocalibrated G2 -

structures on compact seven-manifolds M to parallel Spin.7/-structures. As before,
we generalise the evolution equation to non-compact manifolds and indefinite
metrics.
As we have already seen in Sect. 2.1.3 of Chap. 2, the stabiliser in GL.V/ of
a four-form ˆ0 on an eight-dimensional vector space V is Spin.7/ or Spin0 .3; 4/
if and only if it can be written as in (2.30) for a stable three-form ' on a seven-
dimensional subspace with stabiliser G2 - or G2 , respectively. Thus, a Spin.7/- or
 ./
4.1 Half-Flat Structures and Parallel G2 -Structures 141

Spin0 .3; 4/-structure on an eight-manifold M is defined by a four-form ˆ 2 4 M

such that ˆp 2 ƒ4 Tp M has this property for all p. By formula (2.31) for the metric
gˆ induced by ˆ, an oriented hypersurface in .M; ˆ/ with spacelike unit normal
vector field n with respect to gˆ carries a natural G2 - or G2 -structure, respectively,
defined by ' D n ˆ.
A Spin.7/- or Spin0 .3; 4/-structure ˆ is parallel if and only if dˆ D 0. We
remark that the proof for the Riemannian case given in [105, Lemma 12.4] is not
hard to transfer to the indefinite case when considering [23, Proposition 2.5] and
using the complexification of the two spin groups.
./
Due to this fact, the induced G2 -structure ' on an oriented hypersurface in an
eight-manifold M with parallel Spin.7/- or Spin0 .3; 4/-structure ˆ is cocalibrated,
i.e. it satisfies

d ' ' D 0: (4.21)

./
Conversely, a cocalibrated G2 -structure can be embedded in an eight-manifold
with parallel Spin.7/- or Spin0 .3; 4/-structure as follows.
Theorem 4.1.13 Let M be a seven-manifold and ' 2 3 M be a one-parameter
family of stable three-forms with a parameter t in an interval I satisfying the
evolution equation

@
.' '/ D d': (4.22)
@t
If ' is cocalibrated at t D t0 2 I, then ' defines a family of cocalibrated G2 - or
./
G2 -structures for all t 2 I. Moreover, the four-form

ˆ D dt ^ ' C ' ' (4.23)

defines a parallel Spin.7/- or Spin0 .3; 4/-structure on M  I, respectively, which

induces the metric

gˆ D g' C dt2 : (4.24)

Proof Since the time derivative of d  ' vanishes when inserting the evolution
equation, the family stays cocalibrated if it is cocalibrated at an initial value. As
before, we denote by dL the exterior differential on ML WD M  I and differentiate the
four-form (4.23):

L D dt ^ d' C d.'/ C dt ^ @ .'/:

dˆ
@t
142 4 Hitchin’s Flow Equations

Obviously, this four-form is closed if and only the evolution equation is satisfied
and the family is cocalibrated. The formula for the induced metric corresponds to
formula (2.31). t
u
As before, the Cauchy-Kovalevskaya theorem guarantees existence and uniqueness
of solutions if assumptions analogous to those of Corollary 4.1.6 are satisfied.
Remark 4.1.14 We observe that nearly parallel G2 - and G2 -structures are in
particular cocalibrated such that analytic nearly half-flat structures in dimension
6 can be embedded in parallel Spin.7/- or Spin0 .3; 4/-structures in dimension 8 by
evolving them twice with the help of the Theorems 4.1.12 and 4.1.13.

4.2 Evolution of Nearly "-Kähler Manifolds

In this section, we consider the evolution of nearly pseudo-Kähler and nearly para-
Kähler six-manifolds which can be unified by the notion of a nearly "-Kähler
manifold. The explicit solution of the Hitchin flow yields a simple and unified proof
./
for the correspondence of nearly "-Kähler manifolds and parallel G2 -structures
on cones. We complete the picture by considering similarly the evolution of nearly
./
Kähler structures to nearly parallel G2 -structures on (hyperbolic) sine cones and
./
the evolution of nearly parallel G2 -structures to parallel Spin.7/- and Spin0 .3; 4/-
structures on cones. Our presentation in terms of differential forms unifies various
results in the literature, which were originally obtained using spinorial methods, and
applies to all possible real forms of the relevant groups.

4.2.1 Cones over Nearly "-Kähler Manifolds

For the readers convenience let us shortly recall the setting. In the language of [5]
and [110], an almost "-Hermitian manifold .M 2m ; g; J/ is defined by an almost "-
complex structure J which squares to "id and a pseudo-Riemannian metric g which
is "-Hermitian in the sense that g. J; J/ D "g.; /. Consequently, a nearly "-
Kähler manifold is defined as an almost "-Hermitian manifold such that rJ is skew-
symmetric. On a six-manifold M, a nearly "-Kähler structure .g; J; !/ with jrJj2 D
4 (i.e. of constant type 1 in the terminology of [67]) is equivalent to a normalised
H "; -structure .!; / which satisfies

d! D 3 ; (4.25)
dO D 4!:
O (4.26)

This result is well-known for Riemannian signature [103] and is generalised to

arbitrary signature in [110, Theorem 3.14], see Sect. 3.1.2 of Chap. 3. In particular,
4.2 Evolution of Nearly "-Kähler Manifolds 143

nearly "-Kähler structures .!; / in dimension 6 are half-flat and the structure .!; /
O
is nearly half-flat (for the constant  D 4).
Proposition 4.2.1 Let .M; h0 / be a pseudo-Riemannian six-manifold of signature
N D M  RC ; gN " D h0  "dt2 / be the timelike cone
.6; 0/, .4; 2/ or .3; 3/ and let .M
for " D 1 and the spacelike cone for " D 1. There is a one-to-one correspondence
between nearly "-Kähler structures .h0 ; J/ with jrJj2 D 4 on .M; h0 / and parallel
G2 - and G2 -structures ' on M N which induce the cone metric gN " .

Proof This well-known fact is usually proved using Killing spinors, see [12, 70] and
[82]. We give a proof relying exclusively on the framework of stable forms and the
Hitchin flow. For Riemannian signature, this point of view is also adopted in [32]
and [25].
The H "; -structures inducing the given metric h0 are the reductions of the bundle
of orthonormal frames of .M; h0 / to the respective group H "; . Given any H "; -
reduction .!0 ; 0 / of h0 , we consider for t 2 RC the one-parameter family

! D t2 !0 ;  D t3 0 ; (4.27)

which induces the family of metrics h D t2 h0 . By formula (4.10), the metric g' on
M N induced by the stable three-form ' D ! ^ dt C  is exactly the cone metric gN " .
It is easily verified that the family (4.27) consists of half-flat structures satisfying
the evolution equations if and only if the initial value .!.1/; .1// D .!0 ; 0 /
satisfies the exterior system (4.25), (4.26). Therefore, the stable three-form ' on
the cone .M; N gN " / is parallel if and only if the H "; -reduction .!0 ; 0 / of h0 is a nearly
"-Kähler structure with jrJj2 D 4.
Conversely, let ' be a stable three-form on M N which induces the cone metric
gN " . Since @t is a normal vector field for the hypersurface M D M  f1g satisfying
gN .@t ; @t / D ", we obtain an H "; -reduction .!0 ; 0 / of h0 defined by

!0 D @t '; 0 D 'jTM (4.28)

with the help of Proposition 2.1.14 of Chap. 2. Since the two constructions are
inverse to each other, the proposition follows. u
t
Example 4.2.2 Consider the flat .R.3;4/ n f0g; h:; :i/ which is isometric to the cone
.M "  RC ; t2 h"  "dt2 / over the pseudo-spheres M " WD f p 2 R.3;4/ j hp; pi D "g,
" D ˙1, with the standard metrics h" of constant sectional curvature " and
signature .2; 4/ for " D 1 and .3; 3/ for " D 1. Obviously, a stable three-form
' inducing the flat metric h:; :i is parallel if and only if it is constant. Thus, the
previous discussion and Proposition 2.1.14 of Chap. 2, in particular formula (2.29),
yield a bijection

SO.3; 4/= G2 ! f"-complex structures J on M " such that .h" ; J/ is nearly "-Kählerg
' 7! J with Jp .v/ D p  v; 8 p 2 M"
144 4 Hitchin’s Flow Equations

where the cross-product  induced by ' is defined by formula (2.12). In other words,
the pseudo-spheres (M " ; h" ) admit a nearly "-Kähler structure which is unique up to
conjugation by the isometry group O.3; 4/ of h" . In fact, these "-complex structures
on the pseudo-spheres are already considered in [92] and the nearly para-Kähler
property for " D 1 is for instance shown in [15].

4.2.2 Sine Cones over Nearly "-Kähler Manifolds

For Riemannian signature, it has been shown in [53] that the evolution of a
nearly Kähler SU.3/-structure to a nearly parallel G2 -structure induces the Einstein
sine cone metric. This result can be extended as follows. We prefer to consider
(hyperbolic) cosine cones since they are defined on all of R in the hyperbolic case.
Proposition 4.2.3 Let .M; h0 / be a pseudo-Riemannian six-manifold.
(i) If h0 is Riemannian, or has signature .2; 4/, respectively, there is a one-to-one
correspondence between nearly (pseudo-)Kähler structures .h0 ; J/ on M with
jrJj2 D 4 and nearly parallel G2 -structures, or G2 -structures, respectively, for
the constant  D 4 on the spacelike cosine cone
 
.M  . ; /; cos2 .t/h0 C dt2 /:
2 2

(ii) If h0 has signature .3; 3/, there is a one-to-one correspondence between nearly
para-Kähler structures .h0 ; J/ on M with jrJj2 D 4 and nearly parallel G2 -
structures for the constant  D 4 on the timelike hyperbolic cosine cone

.M  R;  cosh2 .t/h0  dt2 /:

Proof
(i) Starting with any SU.3/- or SU.1; 2/-reduction .!0 ; 0 / of h0 , the one-
parameter family

! D cos2 .t/!0 ;  D  cos3 .t/.sin.t/0 C cos.t/O0 /

with .!.0/; .0// D .!0 ; O0 / defines a stable three-form ' D ! ^ dt C  on

M  . 2 ; 2 /. Since z‰0 D z.0 C iO0 / is a .3; 0/-form w.r.t. the induced almost
complex structures JRe .z‰0 / for all z 2 C , the structure J D J0 is constant
in t. Thus, the metric g' induced by ' is the cosine cone metric. Moreover, it
holds O D  cos3 .t/.sin.t/O0  cos.t/0 / due to Corollary 2.1.7 of Chap. 2.
It takes a short calculation to verify that the one-parameter family is nearly
half-flat (for the constant  D 4) and satisfies the evolution equation (4.16)
if and only .!0 ; 0 / satisfies the exterior system (4.25), (4.26). Thus, applying
Proposition 4.1.11, the three-form ' D ! ^ dt C  defines a nearly parallel
4.2 Evolution of Nearly "-Kähler Manifolds 145

G"; -structure on M  . 2 ; 2 / (for the constant  D 4) if and only if .h0 ; J0 /
is nearly "-Kähler with jrJj2 D 4.
The inverse construction is given by (4.28) in analogy to the case of the
ordinary cone.
(ii) The proof in the para-complex case is completely analogous if we consider the
one-parameter family

! D cosh2 .t/!0 ;  D  cosh3 .t/.sinh.t/0 C cosh.t/O0 /

which is defined for all t 2 R. We note the following subtleties regarding

signs. By Proposition 2.1.4 of Chap. 2, we know that the mapping  7! O is
homogeneous of degree 1, but not linear. Indeed, by applying Corollary 2.1.7

1
of Chap. 2, we find

sinh.t/0 C cosh.t/O0 D  sinh.t/O0  cosh.t/0 :

Using this formula, one can check that J D JO0 D J0 is constant in t such
that the metric induced by .!; / is in fact h D  cosh2 .t/h0 . t
u
The fact that the (hyperbolic) cosine cone over a six-manifold carrying a Killing
spinor carries again a Killing spinor was proven in [81]. By relating spinors to
./
differential forms, these results also imply the existence of a nearly parallel G2 -
structures on the (hyperbolic) cosine cone over a nearly "-Kähler manifold.
Example 4.2.4 The (hyperbolic) cosine cone of the pseudo-spheres .M " ; h" / of
Example 4.2.2 has constant sectional curvature 1, for instance due to [8, Corol-
lary 2.3], and is thus (locally) isometric to the pseudo-sphere S3;4 D f p 2
R.4;4/ j hp; pi D 1g D Spin0 .3; 4/= G2 .

./
4.2.3 Cones over Nearly Parallel G2 -Structures

By Lemma 9 in [12], there is a one-to-one correspondence on a Riemannian

seven-manifold .M; g0 / between nearly parallel G2 -structures and parallel Spin.7/-
structures on the Riemannian cone. In order to illustrate the evolution equations for
nearly parallel G2 -structures, we extend this result to the indefinite case by applying
Theorem 4.1.13. This is possible since nearly parallel G2 -structures are in particular
cocalibrated. Again, the fact that the cone over a nearly parallel G2 -manifold admits
a parallel spinor can be derived from the connection to Killing spinors as observed
in [81].
Proposition 4.2.5 Let .M; g0 / be a pseudo-Riemannian seven-manifold of sig-
nature .3; 4/. There is a one-to-one correspondence between nearly parallel
146 4 Hitchin’s Flow Equations

G2 -structures for the constant 4 which induce the given metric g0 and parallel
Spin0 .3; 4/-structures on M  RC inducing the cone metric gN D t2 g0 C dt2 .
Proof Let '0 be any cocalibrated G2 -structure on M inducing the metric g0 . The
one-parameter family of three-forms defined by ' D t3 '0 for t 2 RC induces the
family of metrics g D t2 g0 such that the Hodge duals are ' ' D t4 '0 '0 . By (4.24),
the Spin0 .3; 4/-structure ‰ D dt ^ ' C ' ' on M  RC induces the cone metric gN .
Conversely, given a Spin0 .3; 4/-structure ‰ on the cone .M  RC ; gN /, we have the
cocalibrated G2 -structure '0 D @t ‰ on M, which also induces the given metric
g0 . Since the evolution equation (4.22) is satisfied if and only if the initial value '0
is nearly parallel for the constant 4 and since the two constructions are inverse to
each other, the assertion follows from Theorem 4.1.13. t
u
Example 4.2.6 We consider again the easiest example, i.e. the flat R.4;4/ nf0g which
is isometric to the cone over the pseudo-sphere S3;4 . Analogous to Example 4.2.2,
the proposition just proved yields a proof of the fact that the nearly parallel G2 -
structures for the constant 4 on S3;4 are parametrised by SO.4; 4/= Spin0 .3; 4/, i.e.
by the four homogeneous spaces (2.32). In particular, these structures are conjugated
by the isometry group O.4; 4/ of S3;4 .
Summarising the application of the three Propositions 4.2.1, 4.2.3 and 4.2.5 to
pseudo-spheres, we find a mutual one-to-one correspondence between
(1) nearly pseudo-Kähler structures with jrJj2 ¤ 0 on .S2;4 ; gcan /,
(2) nearly para-Kähler structures with jrJj2 ¤ 0 on .S3;3 ; gcan /,
(3) parallel G2 -structures on .R.3;4/ ; gcan /,
(4) nearly parallel G2 -structures on the spacelike cosine cone over .S2;4 ; gcan /,
(5) nearly parallel G2 -structures on the timelike hyperbolic cosine cone over
.S3;3 ; gcan /,
(6) nearly parallel G2 -structures on .S3;4 ; gcan / and
(7) parallel Spin0 .3; 4/-structures on .R.4;4/ ; gcan /.
This geometric correspondence is reflected in the algebraic fact that the four
homogeneous spaces (2.32) are isomorphic.

4.3 The Evolution Equations on Nilmanifolds  n H3 H3

Let H3 be the three-dimensional real Heisenberg group with Lie algebra h3 . In this
./
section, we will develop a method to explicitly determine the parallel G2 -structure
induced by an arbitrary invariant half-flat structure on a nilmanifold  n H3  H3
without integrating. In particular, this method is applied to construct three explicit
large families of metrics with holonomy equal to G2 or G2 , respectively.
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 147

4.3.1 Evolution of Invariant Half-Flat Structures

on Nilmanifolds

Left-invariant half-flat structures .!0 ; 0 / on a Lie group G are in one-to-one

correspondence with normalised pairs .!; / of compatible stable forms on the Lie
algebra g of G which satisfy d D 0 and d! 2 D 0. To shorten the notation, we will
speak of a half-flat structure on a Lie algebra.
Given as initial value a half-flat structure on a Lie algebra, the evolution equations

P D d! ; P D dO ; (4.29)

reduce to a system of ordinary differential equations and a unique solution exists

on a maximal interval I. Due to the structure of the equation, the solution differs
from the initial values by adding exact forms to 0 and 0 . In other words, an initial
value .0 ; 0 / evolves within the product Œ0   Œ0  of their respective Lie algebra
cohomology classes.
Every nilpotent Lie group N with rational structure constants admits a co-
compact lattice  and the resulting compact quotients nN are called nilmanifolds.
Recall that a geometric structure on a nilmanifold nN is called invariant if is
induced by a left-invariant geometric structure on N.
Explicit solutions of the Hitchin flow equations on several nilpotent Lie algebras
can be found for instance in [31] and [2]. In both cases, a metric with holonomy
contained in G2 has been constructed before by a different method and this
information is used to obtain the solution. For a symplectic half-flat initial value,
another explicit solution on one of these Lie algebras is given in [37]. In all cases,
the solution depends only on one variable.
At least for four nilpotent Lie algebras including h3 ˚ h3 , a reason for the simple
structure of the solutions has been observed in [2]. Indeed, the following lemma
shows that the evolution of  takes place in a one-dimensional space. As usual, we
define a nilpotent Lie algebra by giving the image of a basis of one-forms under the
exterior derivative, see for instance [105]. The same reference also contains a list of
all six-dimensional nilpotent Lie algebras.
Lemma 4.3.1 Let  be a closed stable three-form with dual three-form O on a six-
dimensional nilpotent Lie algebra g.
(i) If g is one of the three Lie algebras

.0; 0; 0; 0; e12; e34 / ; .0; 0; 0; 0; e13Ce42 ; e14 Ce23 /; .0; 0; 0; 0; e12; e14 Ce23 /;

then dO 2 ƒ4 U for the four-dimensional kernel U of d W ƒ1 g ! ƒ2 g .

(ii) If g is the Lie algebra

.0; 0; 0; 0; 0; e12 C e34 /;

148 4 Hitchin’s Flow Equations

then dO 2 ƒ4 U for the four-dimensional subspace U D spanfe1 ; e2 ; e3 ; e4 g of

ker d.
Remark 4.3.2 The assertion of the lemma is not true for the remaining six-
dimensional nilpotent Lie algebras with b1 D dim.ker d/ D 4 or b1 D 5. In
each case, we have constructed a closed stable  such that dO is not contained in
ƒ4 .ker d/.
In fact, this lemma can also be viewed as a corollary of the following lemma which
we will prove first.
Lemma 4.3.3 Let  be a closed stable three-form on one of the four Lie algebras
of Lemma 4.3.1 and let U be the four-dimensional subspace of ker d defined there.
In all four cases, the space U is J -invariant where J denotes the almost (para-
)complex structure induced by .
Proof For ./ < 0, the assertion is similar to that of [2, Lemma 2]. However,
since the only proof seems to be given for the Iwasawa algebra for integrable J in
[83, Theorem 1.1], we give a complete proof.
Let g be one of the three Lie algebras given in part (i) of Lemma 4.3.1 and
U D ker d. Obviously, the two-dimensional image of d lies within ƒ2 U in all three
cases. By J D J we denote the almost (para-)complex structure associated to the
closed stable three-form . As before, we denote by " 2 f˙1g the sign of ./ such
that J D "id. Let the symbol i" be defined by the property i2" D " such that the para-
complex numbers and the complex numbers can be unified by C" D RŒi" . Thus, a
.1; 0/-form can be defined for both values of " as an eigenform of J in ƒ1 g ˝ C"
for the eigenvalue i" .
We define the J-invariant subspace W WD U \J  U of g such that 2  dim W  4.
In fact, dim W D 4 is equivalent to the assertion. The other two cases are not
possible, which can be seen as follows. To begin with, assume that W is two-
dimensional. When choosing a complement W 0 of W in U, we have by definition of
W that

V D W ˚ W 0 ˚ JW 0:

We observe that, for " D 1, the ˙1-eigenspaces of J restricted to W 0 ˚ J  W 0

are both two-dimensional. Therefore, we can choose for both values of " a basis
fe1 ; e2 ; e3 ; e4 D J  e1 ; e5 D J  e2 ; e6 D J  e3 g of V such that e1 ; e2 ; e3 and e4 are
closed and de5 ; de6 2 ƒ2 U. Since  C i" J  is a (3,0)-form in both cases, it is
possible to change the basis vectors e1 ; e4 within W  ker d such that

 C i" J  D .e1 C i" e4 /^.e2 C i" e5 /^.e3 C i" e6 /

and thus

 D e123 C "e156  "e246 C "e345 :

4.3 The Evolution Equations on Nilmanifolds  n H3 H3 149

By construction of the basis, we have that

0 D d D "e1 ^ de5 ^ e6 C "e1 ^ e5 ^ de6 C ˛

with ˛ 2 ƒ4 U. As the first two summands are linearly independent and not in ƒ4 U,
we conclude that both e1 ^ de5 and e1 ^ de6 vanish. Thus, the closed one-form e1
has the property that the wedge product of e1 with any exact two-form vanishes.
However, an inspection of the standard basis of each of the three Lie algebras in
question reveals that such a one-form does not exist on these Lie algebras and we
have a contradiction to dim W D 2.
Since a J-invariant space cannot be three-dimensional for " D 1, the proof is
finished for this case. However, if " D 1, the case dim W D 3 cannot be excluded
that easy. Assuming that it is in fact dim W D 3, we choose again a complement W 0
of W in U and find a decomposition

V D W ˚ W 0 ˚ J  W 0 ˚ W 00

with J  W 00 D W 00 . Without restricting generality, we can assume that J acts trivially

on W 00 . Then, we find a basis for V such that the C1-eigenspace of J is spanned by
fe1 ; e4 C e5 ; e6 g and the 1-eigenspace by fe2 ; e3 ; e4  e5 g, where e1 ; e2 ; e3 and e4
are closed and e5 D J  e4 . Since the given closed three-form  generates this J, it
has to be of the form

 D ae1 ^.e4 C e5 /^e6 C be23 ^ .e4  e5 /

for two real constants a; b. The vanishing exterior derivative

d D ae1 ^ d.e56 / mod ƒ4 U

leads to the same contradiction as in the first case and part (i) is shown.
In fact, the same arguments apply to the Lie algebra of part (ii). The four-
dimensional space U  ker d spanned by fe1 ; : : : ; e4 g also satisfies im d  ƒ2 U.
Going through the above arguments, the only difference is that e5 or e6 may be
closed. However, at least one of them is not closed and its image under d generates
the exact two-forms. Again, there is no one-form ˇ 2 U such that ˇ ^  D 0 for
all exact two-forms  and the arguments given in part (i) lead to contradictions for
both dim W D 2 and dim W D 3. t
u
Proof of Lemma 4.3.1 Let  be a closed stable three-form on one of the four
nilpotent Lie algebras and U  ker d as defined in the lemma. For both values of ",
we can apply Lemma 4.3.3 and choose two linearly independent closed .1; 0/-forms
E1 and E2 within the J -invariant space U ˝ C" . Considering that  C i" O is a .3; 0/-
form for both values of ", there is a third .1; 0/-form E3 such that  C i" O D E123 .
150 4 Hitchin’s Flow Equations

Since d D 0 and im d  ƒ2 U, it follows that the exterior derivative

d O D "i" d.E123 / D "i" E12 ^ dE3

is an element of ƒ4 U. t
u

4.3.2 Left-Invariant Half-Flat Structures on H3  H3

From now on, we focus on the Lie algebra g D h3 ˚ h3 . Apart from describing
all half-flat structures on this Lie algebra, i.e. all initial values for the evolution
equations, we give various explicit examples and prove a strong rigidity result
concerning the induced metric.
Obviously, pairs of compatible stable forms on a Lie algebra which are iso-
morphic by a Lie algebra automorphism induce equivalent H "; -structures on the
corresponding simply connected Lie group. Thus, we derive, to begin with, a normal
form modulo Lie algebra automorphisms for stable two-forms ! 2 ƒ2 g which
satisfy d! 2 D 0.
A basis fe1 ; e2 ; e3 ; f1 ; f2 ; f3 g for h3 ˚ h3 such that the only non-vanishing Lie
brackets are given by

de3 D e12 ; df 3 D f 12 ;

will be called a standard basis. The connected component of the automorphism

group of the Lie algebra h3 ˚ h3 in the standard basis is
80 1 9
ˆ
ˆ A 0 0 0 >
>
<B t =
B a det.A/ ct 0 C C 2
Aut0 .h3 ˚ h3 / D @ ; A; B 2 GL.2; R/; a; b; c; d 2 R :
ˆ 0 0 B 0 A >
>
:̂ t ;
d 0 bt det.B/
(4.30)

We denote by gi , i D 1; 2, the two summands, by zi their centres and by z the centre

of g. The annihilator of the centre is z0 D ker d and similarly for the summands by
restricting d. We have the decompositions

g1 g
g Š z01 ˚ z02 ˚ 0
˚ 02 ;
z1 z2
g g g g g g
ƒ2 g Š ƒ2 .z0 / ˚ . 01 ^ 02 / ˚ .z01 ^ 02 / ˚ .z02 ^ 01 / ˚ .z01 ^ 01 / ˚ .z02 ^ 02 / :
z z z z z1 z2
„ 1 ƒ‚ 2… „ ƒ‚ 2… „ ƒ‚ 1… „ ƒ‚ …
k1 k2 k3 k4
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 151

By ! ki we denote the projection of a two-form ! onto one of the spaces ki , i D


1; 2; 3; 4, defined as indicated in the decomposition. We observe that k1 D ƒ2 . gz0 /
and ! k1 D 0 if and only if !.z; z/ D 0.
Lemma 4.3.4 Consider the action of Aut.h3 ˚ h3 / on the set of non-degenerate
two-forms ! on g with d! 2 D 0. The orbits modulo rescaling are represented in a
standard basis by the following two-forms:

!1 D e1 f 1 C e2 f 2 C e3 f 3 ; if ! k1 ¤ 0;
!2 D e2 f 2 C e13 C f 13 ; if d! D 0 ” ! k1 D 0; ! k2 D 0; ! k3 D 0;
!3 D e1 f 3 C e2 f 2 C e3 f 1 ; if ! k1 D 0; ! k2 ¤ 0; ! k3 ¤ 0; ! k4 D 0;
!4 D e1 f 3 C e2 f 2 C e3 f 1 C e13 C ˇf 13 ; if ! k1 D 0; ! k2 ¤ 0; ! k3 ¤ 0; ! k4 ¤ 0;
!5 D e1 f 3 C e2 f 2 C e13 C f 13 otherwise,

where ˇ 2 R and ˇ ¤ 1.

Proof Let
X X X
!D ˛i e.iC1/.iC2/ C ˇi f .iC1/.iC2/ C i;j ei f j

be an arbitrary non-degenerate two-form expressed in a standard basis. We will

give in each case explicitly a change of standard basis by an automorphism of the
form (4.30) with the notation
! !
a1 a2 b1 b2
AD ; a D .a5 ; a6 /; B D
t
; bt D .b5 ; b6 /; ct D .c1 ; c2 /; dt D .d1 ; d2 /:
a3 a4 b3 b4

First of all, if ! k1 ¤ 0, the term 3;3 e3 f 3 is different from zero and we rescale
such that 3;3 D 1. Then, the application of the change of basis

a1 D 1; a2 D 0; a3 D 0; a4 D 1; a5 D 1;3 ; a6 D 2;3 ;
b1 D 2;2  2;3 3;2  ˛1 ˇ1 ; b2 D 1;2  ˇ1 ˛2 C 3;2 1;3 ;
b3 D 2;1 C 3;1 2;3  ˇ2 ˛1 ;
b4 D 1;1  ˛2 ˇ2  1;3 3;1 ; b5 D 3;1 2;2 C 3;1 ˛1 ˇ1 C 3;2 2;1 C 3;2 ˇ2 ˛1 ;
b6 D 3;1 1;2 C 3;1 ˇ1 ˛2  3;2 1;1 C 3;2 ˛2 ˇ2 ;
c1 D ˇ2 2;2  ˇ2 2;3 3;2 C ˇ1 2;1  ˇ1 3;1 2;3 ; d1 D ˛2 ;
c2 D ˇ2 1;2 C ˇ2 3;2 1;3  ˇ1 1;1 C ˇ1 1;3 3;1 ; d2 D ˛1 ;
152 4 Hitchin’s Flow Equations

transforms ! into !Q D Q1;1 .e1 f 1 C e2 f 2 C e3 f 3 / C ˛Q 3 e12 C ˇQ3 f 12 , Q1;1 ¤ 0. This

two-form satisfies d!Q 2 D 0 if and only if ˛Q 3 D 0; ˇQ3 D 0 and the normal form !1
is achieved by rescaling.
Secondly, the vanishing of d! corresponds to ! k1 D 0, ! k2 D 0, ! k3 D 0 or
3;3 D 1;3 D 2;3 D 3;1 D 3;2 D 0 in a standard basis. By non-degeneracy,
at least one of ˛1 and ˛2 is not zero and we can always achieve ˛1 D 0; ˛2 ¤ 0.
Indeed, if ˛1 ¤ 0, we apply the transformation (4.30) with a1 D 1, a2 D 1, a4 D ˛˛21 ,
B D 1 and all remaining entries zero. With an analogous argument, we can assume
that ˇ1 D 0, ˇ2 ¤ 0. Since 2;2 ¤ 0 by non-degeneracy, we can rescale ! such that
2;2 D 1. Now, the transformation of the form (4.30) given by

a1 D 1; a2 D 0; a3 D 0; a4 D ˇ2 ; b1 D 1; b2 D 0; b3 D 0; b4 D ˛2 ; a5 D 0;
˛3 ˇ2 ˛2 ˇ3 1;1
a6 D  ; b5 D 0; b6 D  ; c1 D ; c2 D 1;2 ; d1 D 0; d2 D 2;1 ;
˛2 ˇ2 ˛2

maps ! to a multiple of the normal form !2 .

Thirdly, we assume that ! is non-degenerate with ! k1 D 0, i.e. 3;3 D 0 and both
! ¤ 0, i.e. 1;3 or 2;3 ¤ 0, and ! k3 ¤ 0, i.e. 3;1 or 3;2 ¤ 0. Similar as before,
k2

we can achieve 2;3 D 0, 1;3 ¤ 0 by applying, if 2;3 ¤ 0, the transformation (4.30)

1;3
with a1 D 1, a2 D 1, a4 D  2;3 , B D 1 and all remaining entries zero.
Analogously, we can assume 3;2 D 0, 3;1 ¤ 0 and rescaling yields 2;2 D 1, which
is non-zero by non-degeneracy. After this simplification, the condition d! 2 D 0
implies that ˛1 D ˇ1 D 0 and the transformation

˛2 ˇ3  3;1 1;2
a1 D 1; a2 D 0; a3 D ; a4 D 1;3 ; a5 D 0; a6 D 0; b1 D 1; b2 D 0;
3;1
ˇ2 ˛3  1;3 2;1 1;2 1;3 2;1 3;1  1;1 1;3 3;1  ˛2 ˛3 ˇ2 ˇ3
b3 D ; b4 D 3;1 ; b5 D 2
;
1;3 1;3 3;1
b6 D 0;
c1 D 0; c2 D ˇ3 ; d1 D 0; d2 D ˛3 ;

maps ! to !Q D e1 f 3 C e2 f 2 C e3 f 1 C ˛Q 2 e31 C ˇQ2 f 31 . The condition ! k4 D 0

corresponds to ˛Q 2 D 0, ˇQ2 D 0, i.e. normal form !3 . If ! k4 ¤ 0, we can achieve
˛Q 2 ¤ 0 by possibly changing the summands. Now, the transformation

1
a1 D 1; a2 D 0; a3 D 0; a4 D  ; a5 D 0; a6 D 0; c1 D 0; c2 D 0;
˛Q 2
1 1
b1 D  ; b2 D 0; b3 D 0; b4 D  ; b5 D 0; b6 D 0; d1 D 0; d2 D 0;
˛Q 2 ˛Q 2

maps !Q to the fourth normal form !4 .

The cases that remain are ! k1 D 0 and either ! k2 ¤ 0; ! k3 D 0 or ! k3 D
0; ! k2 ¤ 0. After changing the summands if necessary, we can assume ! k3 D 0 and
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 153

! k2 ¤ 0, i.e. 3;1 D 3;2 D 3;3 D 0 and at least one of 1;3 or 2;3 non-zero. As
1;3
before, we can achieve 2;3 D 0 by the transformation a1 D 1; a2 D 1; a4 D  2;3 .
2
Evaluating d! D 0 yields ˛1 D 0. Now, non-degeneracy enforces that ˇ1 ¤ 0 or
ˇ2 ¤ 0, and after another similar transformation ˇ1 D 0. Finally, the simplified ! is
non-degenerate if and only if 2;2 ˛2 ˇ2 ¤ 0 and, after rescaling such that 2;2 D 1,
the transformation
2 2
1;3 1;3 .1;3 2;1  ˛3 ˇ2 /
a1 D 1; a2 D 0; a3 D 0; a4 D  ; a5 D 0; a6 D ;
ˇ2 ˛2 ˇ22
1;3 ˇ3 ˛2
b1 D  ; b2 D 0; b3 D 0; b4 D ˛2 ; b5 D 0; b6 D  ;
ˇ2 ˇ2
2
1;2 ˇ2 C 1;3 ˇ3 1;1 1;3 2;1
c1 D 0; c2 D  ; d1 D  ; d2 D ;
ˇ2 ˇ2 ˇ22

maps ! to a multiple of the fifth normal form !5 . t

u
Using this lemma, it is possible to describe all half-flat structures .!; / on h3 ˚h3
as follows. In a fixed standard basis such that ! is in one of the normal forms, the
equations d D 0 and ! ^  D 0 are linear in the coefficients of an arbitrary three-
form . Thus, it is straightforward to write down all compatible closed three-forms
for each normal form which depend on nine parameters in each case. The stable
forms in this nine-dimensional space are parametrised by the complement of the
zero-set of the polynomial ./ of order four. One parameter is eliminated when
we require a stable  to be normalised in the sense of (2.18). We remark that the
computation of the induced tensors J , O and g.!;/ may require computer support,
in particular, the signature of the metric is not obvious. However, stability is an open
condition: If a single half-flat structure .!0 ; 0 / is explicitly given such that !0 is
one of the normal forms, then the eight-parameter family of normalised compatible
closed forms defines a deformation of the given half-flat structure .!0 ; 0 / in some
neighbourhood of .!0 ; 0 /.
For instance, the closed three-forms which are compatible with the first normal
form

! D e1 f 1 C e2 f 2 C e3 f 3 (4.31)

in a standard basis can be parametrised as follows:

 D .a1 ; : : : ; a9 / D a1 e123 C a2 f 123 C a3 e1 f 23 C a4 e2 f 13

Ca5 e23 f 1 C a6 e13 f 2 C a7 .e2 f 23  e1 f 13 / C a8 .e12 f 3  e3 f 12 /
Ca9 .e23 f 2  e13 f 1 /: (4.32)
154 4 Hitchin’s Flow Equations

The quartic invariant ./ depending on the nine parameters is

./ D .2a6 a4 a28 C 2a1 a2 a28 C 2a28 a3 a5  4a5 a27 a6  4a29 a4 a3  4a29 a2 a8 C 4a27 a8 a1
C4a7 a28 a9 C a21 a22 C a26 a24 C a23 a25 C a48  2a6 a4 a3 a5 C 4a5 a7 a9 a3 C 4a9 a4 a6 a7
4a5 a2 a6 a8 C 4a4 a8 a1 a3  4a9 a2 a1 a7  2a1 a2 a6 a4  2a1 a2 a3 a5 / .e123 f 123 /˝2 :

Example 4.3.5 For each possible signature, we give an explicit normalised half-flat
structure with fundamental two-form (4.31). The first and the third example appear
in [112]. To begin with, the closed three-form

1
 D p .e123  f 123  e1 f 23 C e23 f 1  e2 f 31 C e31 f 2  e3 f 12 C e12 f 3 / (4.33)
2

induces a half-flat SU.3/-structure .!; / such that the standard basis is orthonor-
mal. Similarly, the closed three-form

1
 D p .e123  f 123  e1 f 23 C e23 f 1 C e2 f 31  e31 f 2 C e3 f 12  e12 f 3 / (4.34)
2

induces a half-flat SU.1; 2/-structure .!; / such that the standard basis is pseudo-
orthonormal with e1 and e4 being spacelike. Finally, the closed three-form
p 123
D 2 .e C f 123 /; (4.35)

induces a half-flat SL.3; R/-structure .!; / such that the two h3 -summands are the
eigenspaces of the para-complex structure J , which is integrable since also d O D 0.
The induced metric is


g D 2 e1  e4 C e2  e5 C e3  e6 :

In fact, half-flat structures with Riemannian metrics are only possible if !

belongs to the orbit of the first normal form.
Lemma 4.3.6 Let .!; / be a half-flat SU.3/-structure on h3 ˚ h3 . Then it holds
! k1 ¤ 0. In particular, there is a standard basis such that ! D !1 D e1 f 1 C e2 f 2 C
e3 f 3 .
Proof Suppose that .!; / is a half-flat SU.3/-structure on h3 ˚ h3 with ! k1 D 0.
Thus, we can choose a standard basis such that ! is in one of the normal forms
!2 ; : : : ; !5 of Lemma 4.3.4 and  belongs to the corresponding nine-parameter
family of compatible closed three-forms. We claim that the basis one-form e1 is
isotropic in all four cases which yields a contradiction since the metric of an SU.3/-
structure is positive definite. The quickest way to verify the claim is the direct
computation of the induced metric, which depends on nine parameters, with the
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 155

help of a computer. In order to verify the assertion by hand, the following formulas
shorten the calculation considerably. For all one-forms ˛, ˇ and all vectors v, the "-
complex structure J and the metric g induced by a compatible pair .!; / of stable
forms satisfy

1
˛ ^ J ˇ ^ ! 2 D g.˛; ˇ/ ! 3 ;
3
J ˛.v/./ D ˛ ^  ^ .v /;

which is straightforward to verify in the standard basis (2.7), (2.20), cf. also [112,
Lemmas 2.1,2.2]. For instance, for the second normal form !2 , it holds e1 ^ !22 D
2e12 f 123 . Thus, by the first formula, it suffices to show that J e1 .e3 / D e1 . J e3 / D
0 which is in turn satisfied if e1 ^^.e3 / D 0 due to the second formula. A similar
simplification applies to the other normal forms and we omit the straightforward
calculations. t
u
Moreover, the geometry turns out to be very rigid if ! k1 D 0. We recall that
simply connected para-hyper-Kähler symmetric spaces with abelian holonomy are
classified in [7, 45]. In particular, there exists a unique simply connected four-
dimensional para-hyper-Kähler symmetric space with one-dimensional holonomy
group, which is defined in [7, Section 4]. We denote the underlying pseudo-
Riemannian manifold as .N 4 ; gPHK /.
Proposition 4.3.7 Let .!; / be a left-invariant half-flat structure with ! k1 D 0 on
H3  H3 and let g be the pseudo-Riemannian metric induced by .!; /. Then, the
pseudo-Riemannian manifold .H3 H3 ; g/ is either flat or isometric to the product of
.N 4 ; gPHK / and a two-dimensional flat factor. In particular, the metric g is Ricci-flat.
Proof Due to the assumption ! k1 D 0, we can choose a standard basis such
that ! is in one of the normal forms !2 ; : : : ; !5 . In each case separately, we
do the following. We write down all compatible closed three-forms  depending
on nine parameters. With computer support, we calculate the induced metric g.
For the curvature considerations, it suffices to work up to a constant such that
we can ignore the rescaling by ./ which is different from zero by assumption.
Now, we transform the left-invariant co-frame fe1 ; : : : ; f 3 g to a coordinate co-frame
fdx1 ; : : : ; dy3 g by applying the transformation defined by

e1 D dx1 ; e2 D dx2 ; e3 D dx3 C x1 dx2 ; f 1 D dy1 ; f 2 D dy2 ; f 3 D dy3 C y1 dy2 ;

(4.36)
such that the metric is accessible for any of the numerous packages computing
curvature. The resulting curvature tensor R 2 .End ƒ2 TM/, M D H 3  H 3 , has in
each case only one non-trivial component

R.@x1 ^ @y1 / D c @x3 ^ @y3 (4.37)

156 4 Hitchin’s Flow Equations

for a constant c 2 R and R is always parallel. Thus, the metric is flat if c D 0 and
symmetric with one-dimensional holonomy group if c ¤ 0, for H3  H3 is simply
connected and a naturally reductive homogeneous metric is complete.
Furthermore, it turns out that the metric restricted to TN WD spanf@x1 ; @x3 ; @y1 ; @y3 g
is non-degenerate and of signature .2; 2/ for all parameter values. Thus, the mani-
fold splits in a four-dimensional symmetric factor with neutral metric and curvature
tensor (4.37) and the two-dimensional orthogonal complement which is flat. Since a
simply connected symmetric space is completely determined by its curvature tensor
and the four-dimensional para-hyper-Kähler symmetric space .N 4 ; gPHK / has the
same signature and curvature tensor, the four-dimensional factor is isometric to
.N 4 ; gPHK /. Finally, the metric g is Ricci-flat since gPHK is Ricci-flat. t
u
Example 4.3.8 The following examples define half-flat normalised SU.1; 2/-
structures with ! k1 D 0 in a standard basis. None of the examples is flat. Thus, the
four structures are equivalent as SO.2; 4/-structures due to Proposition 4.3.7, but
the examples show that the geometry of the reduction to SU.1; 2/ is not as rigid.
p p
! D !2 ;  D e12 f 3 C 2e13 f 2 C e1 f 23 C e23 f 1  e3 f 12 C 2 f 123 ;
p p
g D  .e2 /2  . f 2 /2 C 2 e1 e3  2 2 e1 f 3 C 2 2 e3 f 1  2 f 1 f 3 ;
(Ricci-flat pseudo-Kähler since d! D 0; dO D 0/I
! D !3 ;  D e123 C e12 f 3 C e13 f 2 C e1 f 12  2e1 f 23 C e2 f 13  e3 f 12 ;
g D  .e2 /2  2 . f 2 /2 C 2 e1 f 1 C 2 e1 f 3 C 2 e2 f 2  2 e3 f 1  2 f 1 f 3 ;
.d! ¤ 0; J integrable since dO D 0/I
ˇ C 1 23 1 ˇ 4  ˇ  1 2 13
! D !4 ;  D ˇ e12 f 3  ˇ e13 f 2 C ˇ e1 f 23 C e f C ef
ˇ3 ˇ3
ˇ e3 f 12  .ˇ 2 C 2ˇ/ f 123 ; .d! ¤ 0; dO ¤ 0/;
1 2 2
gD .e /  ˇ 2 . f 2 /2 C 2ˇ 2 e1 f 3
ˇ2
2 2.ˇ 4 C ˇ C 1/ 1 3
 e3 f 1  f f I
ˇ 2 .ˇC 1/ ˇ2
! D !5 ;  D e12 f 3 C e13 f 2  e1 f 23 C e23 f 1  e3 f 12 C f 123 ; .d! ¤ 0; dO ¤ 0/;
g D  .e2 /2  2 . f 2 /2 C 2 e1 e3 C 2 e2 f 2 C 2 f 1 f 3 :

Example 4.3.9 Moreover, we give examples of half-flat normalised SL.3; R/-

structures with ! k1 D 0. Again, none of the structures is flat.
p
! D !2 ; D 2 . e1 f 23 C e23 f 1 /; .d! D 0; dO D 0/;
1 3 2 2 1 3
g D 2 e e  2 e f  2 f f I
p
! D !3 ;  D 2 . e12 f 3 C e13 f 2 C e1 f 12  e3 f 12 /; .d! ¤ 0; dO ¤ 0/;
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 157

g D 2 .e1 /2 C 2 e1 e3  2 e1 f 3 C 2 e2 f 2  2 f 1 f 3 I
p
! D !4 ;  D  2ˇ C 2 . e12 f 3  e1 f 23 C e2 f 13  e3 f 12 /; .d! ¤ 0; dO ¤ 0/;
g D 2 . f 2 /2 C 2 e1 e3 C 2 e1 f 3 C 2 e2 f 2  2 e3 f 1  .2ˇ C 4/ f 1 f 3 I
p
! D !5 ;  D 2 . e123 C f 123 /; .d! ¤ 0; dO D 0/;
g D 2 e1 f 3 C 2 e2 f 2 C 2 e3 f 1 :

4.3.3 Solving the Evolution Equations on H3  H3

Due to the preparatory work of the Lemmas 4.3.1 and 4.3.4, it turns out to be
possible to explicitly evolve every half-flat structure on h3 ˚ h3 without integrating.
Proposition 4.3.10 Let .!0 ; 0 / be any half-flat H "; -structure on h3 ˚ h3 with
!0k1 D 0. Then, the solution of the evolution equations (4.29) is affine linear in
the sense that

.t/ D 0 C t dO0 ; .t/ D 0 C t d!0 (4.38)

and is well-defined for all t 2 R.

Proof Let fe1 ; : : : ; f3 g be a standard basis such that !0 is in one of the normal forms
!2 ; : : : ; !5 of Lemma 4.3.4 which satisfy !0k1 D 0. By Lemma 4.3.1 and the second
evolution equation, we know that there is a function y.t/ with y.0/ D 0 such that

1 2
.t/ D 0 C y.t/e12 f 12 D ! C y.t/e12 f 12 :
2 0

For each of the four normal forms, the unique two-form !.t/ with 12 !.t/2 D .t/
and !.0/ D !0 is

!.t/ D !0  y.t/e1 f 1 :

However, the two-form e1 f 1 is closed such that the exterior derivative d!.t/ D d!0
is constant. Therefore, we have .t/ D 0 C t d!0 by the first evolution equation.
Moreover, the two-form !.t/ is stable for all t 2 R since it holds .!.t// D .!0 /
for each of the normal forms and for all t 2 R. It remains to show that d .t/ O is
constant in all four cases which implies that the function y.t/ is linear by the second
evolution equation.
As explained in Sect. 4.3.2, it is easy to write down, for each normal form !0
separately, all compatible, closed three-forms 0 , which depend on nine parameters.
For .t/ D 0 C t d!0 , we verify with the help of a computer that ..t// D .0 /
is constant such that .t/ is stable for all t 2 R since 0 is stable. When we also
158 4 Hitchin’s Flow Equations


calculate J.t/ and .t/
O D J.t/ .t/, it turns out in all four cases that d .t/
O is constant.
This finishes the proof. t
u
We cannot expect that this affine linear evolution of spaces which have one-
dimensional holonomy, due to Proposition 4.3.7, yields metrics with full holonomy
G2 . Indeed, due to the following result the geometry does not change significantly
compared to the six-manifold.
Corollary 4.3.11 Let .!0 ; 0 / be a half-flat H "; -structure on h3 ˚ h3 with !0k1 D 0
and let g' be the Ricci-flat metric induced by the parallel stable three-form ' on M
R defined by the solution (4.38) of the evolution equations with initial value .!0 ; 0 /.
Then, the pseudo-Riemannian manifold .M  R; g' / is either flat or isometric to the
product of the four-dimensional para-hyper-Kähler symmetric space .N 4 ; gPHK / and
a three-dimensional flat factor.
Proof By formula (2.24), the metric g' is determined by the time-dependent metric
g.t/ induced by .!.t/; .t//. All assertions follow from the analysis of the curvature
of g' completely analogous to the proof of Proposition 4.3.7. t
u
The situation changes completely when we consider the first normal form !1 of
Lemma 4.3.4.
Proposition 4.3.12 Let .!0 ; 0 / be any normalised half-flat H "; -structure on h3 ˚
h3 with !0k1 ¤ 0. There is always a standard basis fe1 ; : : : ; f3 g such that !0 D
e1 f 1 C e2 f 2 C e3 f 3 . In such a basis, we define .!.x/; .x// by

.x/ D 0 C x.e12 f 3  e3 f 12 /;
 
1 1 1
!.x/ D 2 .".x// 2 ".x/ e1 f 1 C ".x/ e2 f 2 C e3 f 3 ;
4 4

where .x/ .e123 f 123 /˝2 D ..x//. Furthermore, let I be the maximal interval
containing zero such that the polynomial .x/ of order four does not vanish for any
x 2 I. The parallel stable three-form (4.9) on M  I obtained by evolving .!0 ; 0 /
along the Hitchin flow (4.29) is

1p
'D ".x/ !.x/ ^ dx C .x/:
2
The metric induced by ', which has holonomy contained in G"; , is by (4.10) given
as
1
g' D g.x/  .x/dx2 ; (4.39)
4

where g.x/ denotes the metric associated to .!.x/; .x// via (2.6) and (2.17). The
variable x is related to the parameter t of the Hitchin flow by the ordinary differential
equation (4.42).
4.3 The Evolution Equations on Nilmanifolds  n H3 H3 159

Proof Since !0k1 ¤ 0, we can always choose a standard basis such that !0 D e1 f 1 C
e2 f 2 Ce3 f 3 is in the first normal form of Lemma 4.3.4. Then 0 is of the form (4.32).
Moreover, by Lemma 4.3.1, there is a function y.t/ which is defined on an interval
containing zero and satisfies y.0/ D 0 such that the solution of the second evolution
equation can be written

.t/ D 0 C y.t/e12 f 12 :

The unique !.t/ that satisfies !.0/ D !0 and 12 !.t/2 D .t/ for all t is

p p 1
!.t/ D 1  y.t/ e1 f 1 C 1  y.t/ e2 f 2 C p e3 f 3 :
1  y.t/

Since
1
d!.t/ D p .e12 f 3  e3 f 12 /; (4.40)
1  y.t/

there is another function x.t/ with x.0/ D 0 such that the solution of the first
evolution equation can be written

.t/ D 0 C x.t/.e12 f 3  e3 f 12 /: (4.41)

This three-form is compatible with !.t/ for all t, as one can easily see from (4.32).
Furthermore, the solution is normalised by Theorem 4.1.3, which implies
p p
"..t// D ..t// D 2.!.t// D 2 1  y.t/ e123 f 123 :

Hence, we can eliminate y.t/ by

1
y.t/ D 1  ".x.t//:
4

We remark that the normalisation of 0 D .0/ corresponds to .0/ D 4".

Comparing (4.40) and (4.41), the evolution equations are equivalent to the single
ordinary differential equation

2
xP D p (4.42)
".x.t//

for the only remaining parameter x.t/. In fact, we do not need to solve this equation
./
in order to compute the parallel G2 -form when we substitute
p the coordinate t by
x via the local diffeomorphism x.t/ satisfying dt D 12 ".x.t// dx. Inserting all
substitutions into the formulas (4.9) and (4.10) for the stable three-form ' on M  I
160 4 Hitchin’s Flow Equations

and the induced metric g' , all assertions of the proposition follow immediately from
Theorem 4.1.3. t
u
Example 4.3.13 The invariant .x/ and the induced metric g.x/ for the three explicit
half-flat structures of Example 4.3.5 are the following.
If .!0 ; 0 / is the SU.3/-structure (4.33), it holds
p 3 p p p
.x/ D .x  2/ .x C 2/; I D . 2; 2/;
1p

g.x/ D .1  2 x/ .e1 /2 C .e2 /2 4.x/1 .e3 /2 C .e4 /2 C .e5 /2 4.x/1 .e6 /2
2
p 1p

C 2 x.1  2 x/ e1 e4 C e2 e5 C 4.x/1 e3 e6 :
2

If .!0 ; 0 / is the SU.1; 2/-structure (4.34), we have

p p p p
.x/ D .x  2/.x C 2/3 ; I D . 2; 2/;
1p

g.x/ D .1 C 2 x/ .e1 /2  .e2 /2 C 4.x/1 .e3 /2 C .e4 /2  .e5 /2 C 4.x/1 .e6 /2
2
p 1p

 2 x.1 C 2 x/ e1 e4 C e2 e5 C 4.x/1 e3 e6 :
2

And for the SL.3; R/-structure (4.35), it holds

.x/ D .2 C x2 /2 ; I D R;
2

1 4 2 5

g.x/ D .2 C x / e e C e e
p

C4.2  x2 /.x/1 e3 e6 C 4 2 x.x/1 .e3 /2  .e6 /2 :

Theorem 4.3.14 Let .!.x/; .x// be the solution of the Hitchin flow with one of
the three half-flat structures .!0 ; 0 / of Example 4.3.5 as initial value (see Proposi-
tion 4.3.12 for the explicit solution and Example 4.3.13 for the corresponding metric
g.x/, defined for x 2 I).
Then, the holonomy of the metric g' on M  I defined by formula (4.39) equals
G2 for the SU.3/-structure .!0 ; 0 / and G2 for the other two structures.
Moreover, restricting the eight-parameter family of half-flat structures given
by (4.32) to a small neighbourhood of the initial value .0 ; !0 / yields in each case
an eight-parameter family of metrics of holonomy equal to G2 or G2 .
Proof For all three cases, we can apply the transformation (4.36) and calculate the
curvature R of the metric g' defined by (4.39). Carrying this out with the package
“tensor” contained in Maple 10, we obtained that the rank of the curvature viewed
as endomorphism on two-vectors is 14. This implies that the holonomy of g' in fact
equals G2 or G2 .
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 161

The assertion for the eight-parameter family is an immediate consequence.

Indeed, by construction, the rank of the curvature endomorphism is bounded from
above by 14 and being of maximal rank is an open condition. t
u
To conclude this section we address the issue of completeness and use the
Riemannian family in Example 4.3.13 and Corollary 4.1.9 to construct a complete
conformally parallel G2 -metric on R  .nH3  H3 /.
Example 4.3.15 Let H3 be the Heisenberg group and N D nH p 3 pH3 be a compact
nilmanifold given by a lattice . Let us denote by x W I ! . 2; 2/ the maximal
solution to the equation

2
xP .t/ D q p p ;
. 2  x.t//3 .x.t/ C 2/

with initial condition x.0/ D 0, defining the t-dependent family of Riemannian

metrics
p p 

2  x.t/
1 2  
gt D p .e / C .e2 /2 C .e4 /2 C .e5 /2 C x.t/ 2  x.t/ e1  e4 C e2  e5
2
p
2 2
3 2  4x.t/
C p p .e / C .e6 /2  p p e3  e6 :
2
. 2  x.t// .x.t/ C 2/ . 2  x.t//2 .x.t/ C 2/

If ' W R ! I is a diffeomorphism, then the metric

1
dr2 C g'.r/
' 0 .r/2

is globally conformally parallel G2 and geodesically complete.

4.4 Special Geometry of Real Forms of the Symplectic

SL.6; C/-Module ^3 C6

Homogeneous projective special Kähler manifolds of semisimple groups with

possibly indefinite metric and compact stabiliser were classified in [4]. This includes
the case of manifolds with (positive or negative) definite metrics, for which
the stabiliser is automatically compact. Projective special Kähler manifolds with
negative definite metric play an important role in supergravity and string theory. The
space of local deformations of the complex structure of a Calabi Yau three-fold, for
instance, is an example of a projective special Kähler manifold with negative definite
metric. As a particular result of the classification [4], there is an interesting one-to-
one correspondence between complex simple Lie algebras l of type A, B, D, E, F
and G and homogeneous projective special Kähler manifolds of semisimple groups
162 4 Hitchin’s Flow Equations

with negative definite metric. The resulting spaces are certain Hermitian symmetric
spaces of non-compact type. The homogeneous projective special Kähler manifold
associated to the complex simple Lie algebra of type E6 , for instance, is precisely
the Hermitian symmetric space SU.3; 3/= S.U.3/  U.3//.
Under the above assumptions, the homogeneous projective special Kähler man-
ifold G=K is realised as an open orbit of a real semisimple group G acting on a
smooth projective algebraic variety X  P.V/, where V is the complexification of
a real symplectic module V0 of G and the cone C.X/ WD fv 2 V j .v/ 2 Xg  V
over X is Lagrangian. Here  W V n f0g ! P.V/ denotes the canonical projection.
In fact, C.X/ is the orbit of the highest weight vector of the GC -module V under
the complexified group GC . In the case G=K D SU.3; 3/= S.U.3/  U.3// the
symplectic module is given by V D ^3 C6 . It was shown in [14] that the real
symplectic G-module V0 always admits a homogeneous quartic invariant , which is
related to the hyper-Kähler part of the curvature tensor of a symmetric quaternionic
Kähler manifold associated to the given complex simple Lie algebra l. Moreover,
the level sets f D cg are proper affine hyperspheres for c ¤ 0 and the affine special
Kähler manifold M underlying the homogeneous projective special Kähler manifold
MN D G=K can be realised as one of the open orbits of R  G on V0 [14].
In the following we shall describe all real forms .G; V0 / of the SL.6; C/-module
V D ^3 C6 and study the affine special geometry of the corresponding open orbits
of R  G. As a consequence, we obtain a list of projective special Kähler manifolds,
which admit a transitive action of a real form of SL.6; C/ by automorphisms of the
special Kähler structure. Besides the unique stationary compact example

SU.3; 3/= S.U.3/  U.3//;

we obtain the homogeneous projective special Kähler manifolds

SU.3; 3/= S.U.2; 1/  U.1; 2//; SU.5; 1/= S.U.3/  U.2; 1//
and SL.6; R/= .U.1/  SL.3; C//;

which are symmetric spaces with indefinite metrics and non-compact stabiliser.
The Hermitian signature of the metric is .4; 5/, .6; 3/ and .3; 6/, respectively. The
latter result .3; 6/ corrects Proposition 7 in [76], according to which the Hermitian
signature of the underlying affine special Kähler manifold GLC .6; R/= SL.3; C/
is .1; 9/. The correct Hermitian signature of the affine special Kähler manifold is
.4; 6/.
Finally, we find that one of the two open orbits of SL.6; R/ on ^3 R6 carries affine
special para-Kähler geometry, the geometry of N D 2 vector multiplets on Euclidian
rather than Minkowskian space-time [44]. The corresponding homogeneous projec-
tive special para-Kähler manifold is the symmetric space

SL.6; R/= S.GL.3; R/  GL.3; R//:

4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 163

4.4.1 The Symplectic SL.6; C/-Module V D ^3 C6

and Its Lagrangian Cone C.X/
of Highest Weight Vectors

We consider the 20-dimensional irreducible SL.6; C/-module V D ^3 C6 equipped

with a generator of the line ^6 C6 . The choice of determines an SL.6; C/-
invariant symplectic form , which given by

.v; w/ D v ^ w; v; w 2 V: (4.43)

The highest weight vectors in V are precisely the non-zero decomposable three-
vectors. They form a cone C.X/  V over a smooth projective variety X  P.V/,
namely the Grassmannian Gr3 .C6 / of complex three-planes in C6 . The group
SL.6; C/ acts transitively on the cone C.X/ and, hence, on the compact variety

X Š SL.6; C/=P Š SU.6/= S.U.3/  U.3//;

where P D SL.6; C/x  SL.6; C/ is the stabiliser of a point x 2 X (a parabolic

subgroup).
Proposition 4.4.1 The cone C.X/ D fv 2 ^3 C6 n f0g j v is decomposableg  V is
Lagrangian.
Proof Let .e1 ; : : : ; e6 / be a basis of C6 and put p D e123 . Then

Tp C.X/ D spanfeijk j #f i; j; kg \ f1; 2; 3g  2g

is ten-dimensional and is clearly totally isotropic with respect to . t

u

4.4.2 Real Forms .G; V0 / of the Complex Module

.SL.6; C/; V/

Let G be a real form of the complex Lie group SL.6; C/. There exists a G-invariant
real structure  on V D ^3 C6 if and only if G D SL.6; R/, SU.3; 3/, SU.5; 1/.
In the first case  is simply complex conjugation with respect to V0 D ^3 R6 . In
order to describe the real structure in the other two cases, we first endow C6 with
the standard SU. p; q/-invariant pseudo-Hermitian form h; i. The pseudo-Hermitian
form h; i on C6 induces an SU. p; q/-invariant pseudo-Hermitian form  on V such
that

.v1 ^ v2 ^ v3 ; w1 ^ w2 ^ w3 / D det.hvi ; wj i/; (4.44)

164 4 Hitchin’s Flow Equations

for all v1 ; : : : ; w3 2 C6 . Then we define an SU. p; q/-invariant anti-linear map  W

V ! V by the equation
p
 WD 1 1 ı :

Notice that  W V ! V  ; v 7! .; v/ is linear, whereas  W V ! V  ; v 7! .; v/

and  1 W V  ! V are anti-linear.
Proposition 4.4.2 The anti-linear map  is an SU. p; q/-invariant real structure on
V D ^3 C6 if and only if p  pq 0 .mod 4/. In that case, the SU. p; q/-invariant
pseudo-Hermitian form  D 1.; / on V has signature .10; 10/. Otherwise,
 is an SU. p; q/-invariant quaternionic structure on V.
Proof We present the calculations in the relevant cases . p; q/ D .3; 3/ and . p; q/ D
.5; 1/. The calculations in the other cases are similar.
Case . p; q/ D .3; 3/. Let .e1 ; e2 ; e3 ; f1 ; f2 ; f3 / be a unitary basis of .C6 ; h; i/ D
3;3
C , such that hei ; ei i D h fi ; fi i D 1. We consider the following basis of V:

.e123 ; e1 ^ f12 ; e1 ^ f13 ; e1 ^ f23 ; e2 ^ f12 ; e2 ^ f13 ; e2 ^ f23 ; e3 ^ f12 ; e3 ^ f13 ; e3 ^ f23 ;

f123 ; e23 ^ f3 ; e23 ^ f2 ; e23 ^ f1 ; e13 ^ f3 ; e13 ^ f2 ; e13 ^ f1 ; e12 ^ f3 ; e12 ^ f2 ; e12 ^ f1 /:

With respect to that basis and D e123 ^ f123 we have

! ! !
110 0 0 110 p p 0 110
D ; D ;  D 1 1 ı  D 1 :
0 110 110 0 110 0
(4.45)

This implies  2 D Id, since  is anti-linear.

Case . p; q/ D .5; 1/. Let .e1 ; : : : ; e5 ; f / be a unitary basis of .C6 ; h; i/ D C5;1 ,
such that hei ; ei i D h f ; f i D 1. With respect to the basis

.e123 ; e124 ; e125 ; e134 ; e135 ; e145 ; e234 ; e235 ; e245 ; e345 ; (4.46)
e45 ^ f ; e35 ^ f ; e34 ^ f ; e25 ^ f ; e24 ^ f ; e23 ^ f ; e15 ^ f ; e14 ^ f ; e13 ^ f ; e12 ^ f /

of V and D e12345 ^ f we have again the formulas (4.45) and  2 D Id. t

u

4.4.3 Classification of Open G-Orbits on the Grassmannian X

and Corresponding Special Kähler Manifolds

For each of the real forms .G; V0 / obtained in the previous section, we will now
describe all open orbits of the real simple Lie group G on the Grassmannian
X D Gr3 .C6 / D fE  C6 a three-dimensional subspaceg ,! P.V/, E 7! ^3 E. We
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 165

Table 4.1 Homogeneous G=H Hermitian signature

projective special Kähler
manifolds M N D G=H of real SU.3; 3/= S.U.3/  U.3// (0,9)
simple groups G of type A5 . SU.3; 3/= S.U.2; 1/  U.1; 2// (4,5)
Notice that dimC M N D9 SU.5; 1/= S.U.3/  U.2; 1// (6,3)
SL.6; R/= .U.1/  SL.3; C// (3,6)

N  P.V/
will also describe the projective special Kähler structure of these orbits M
and the (affine) special Kähler structure of the corresponding cones M D C.M/ N 
V. The resulting homogeneous projective special Kähler manifolds are listed in
Table 4.1. Let us first recall some definitions and constructions from special Kähler
geometry.

Basic Facts About Special Kähler Manifolds

Definition 4.4.3 A special Kähler manifold .M; J; g; r/ is a (pseudo-)Kähler

manifold .M; J; g/ endowed with a flat torsion-free connection r such that r! D 0
and dr J D 0, where dr J is the exterior covariant derivative of the vector valued
one-form J.
A conical special Kähler manifold .M; J; g; r; / is a special Kähler manifold
.M; J; g; r/ endowed with a timelike or a spacelike vector field  such that r D
D D Id, where D is the Levi-Civita connection. The vector field  is called Euler
vector field.
The vector fields  and J generate a free holomorphic action of a two-dimensional
Abelian Lie algebra. If the action can be integrated to a free holomorphic C -action
such that the quotient map M ! M N WD M=C is a holomorphic submersion, then
MN is called a projective special Kähler manifold. We will see now that M N carries a
canonical (pseudo-)Kähler metric gN compatible with the induced complex structure
J on M N D M=C . Multiplying the metric g with 1 if necessary, we can assume
that g.; / > 0. Then S D f p 2 M j g.. p/; . p// D 1g is a smooth hypersurface
invariant under the isometric S1 -action generated by the Killing vector field J and
we can recover M N as the base of the circle bundle S ! M N D S=S1. Then M N carries a
unique pseudo-Riemannian metric gN such that S ! M N is a Riemannian submersion.
N J; gN / is in fact the Kähler quotient of .M; J; g/ by the S1 -action generated by the
.M;
Hamiltonian Killing vector field J.
Next we explain the extrinsic construction of special Kähler manifolds from [6].
Let .V; / be a complex symplectic vector space of dimension 2n endowed with a
real structure  such that

.v; w/ D .v; w/; for all v; w 2 V: (4.47)

166 4 Hitchin’s Flow Equations

Then the pseudo-Hermitian form

p
 WD 1.; / (4.48)

has signature .n; n/.

Definition 4.4.4 A holomorphic immersion  W M ! V from an n-dimensional
complex manifold .M; J/ into V is called
(i) nondegenerate if    is nondegenerate,
(ii) Lagrangian if    D 0 and
(iii) conical if . p/ 2 d.Tp M/ and .. p/; . p// ¤ 0 for all p 2 M.
Theorem 4.4.5 ([6])
(i) Any nondegenerate Lagrangian immersion  W M ! V induces on the complex
manifold .M; J/ the structure of a special Kähler manifold .M; J; g; r/, where
g D Re    and r is determined by the condition r  ˛ D 0 for all ˛ 2 V 
which are real valued on V  .
(ii) Any conical nondegenerate Lagrangian immersion  W M ! V induces on
.M; J/ the structure of a conical special Kähler manifold .M; J; g; r; /. The
vector field  is determined by the condition d . p/ D . p/.

The Case G D SL.6; R/

Using the complex conjugation  W v 7! vN on C6 we can decompose X into G-

N D kg 
invariant real algebraic subvarieties X.k/ D X.k/ ./ WD fE 2 X j dim.E \ E/
6
X, where E  C runs through all three-dimensional subspaces and k 2 f0; 1; 2; 3g.
Notice that only X.0/  X is open.
Proposition 4.4.6 The group SL.6; R/ acts transitively on the open real subvariety
X.0/ D fE 2 X j E \ EN D 0g  X.
Proof Given bases .e1 ; e2 ; e3 /, .e01 ; e02 ; e03 / of E; E0 2 X.0/ , respectively, let ' be the
linear transformation, which maps the basis .e1 ; e2 ; e3 ; eN 1 ; eN 2 ; eN 3 / of C6 to the basis
.e01 ; e02 ; e03 ; eN 01 ; eN 02 ; eN 03 / of C6 . Then ' 2 SL.6; R/ and 'E D E0 . t
u
Theorem 4.4.7 The group SL.6; R/ has a unique open orbit X.0/ Š SL.6; R/=
.U.1/  SL.3; C// on the highest weight variety P.V/  X Š Gr3 .C6 / of V D ^3 C6 .
The cone V  M D C.X.0/ / Š GLC .6; R/= SL.3; C/ carries an SL.6; R/-invariant
special Kähler structure of Hermitian signature .4; 6/, which induces on MN D X.0/
the structure of a homogeneous projective special Kähler manifold MN of Hermitian
signature .3; 6/.
Proof Let e1 ; e2 ; e3 2 C6 be three vectors which span a three-dimensional subspace
E  C6 such that E 2 X.0/ . Then C6 D E ˚ EN and the tangent space of M D C.X.0/ /
at p D e123 is given by Tp M D ^3 E ˚ ^2 E ^ E. N We choose the real generator
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 167

p p
D 1e123 ^ eN 123 2 ^3 R6 and compute  D 1.; / on Tp M using the
formula (4.43). The matrix of  jTp M with respect to the basis .e123 ; e12 ^ eN 3 ; e13 ^
eN 2 ; e23 ^ eN 1 ; e12 ^ eN 1 ; e13 ^ eN 1 ; e12 ^ eN 2 ; e23 ^ eN 3 ; e23 ^ eN 2 ; e13 ^ eN 3 / is given by
0 1
1 0 0 0
B 0 13 0 0 C
B C
@ 0 0 0 13 A : (4.49)
0 0 13 0

This shows that  has signature .4; 6/ on Tp M. Since GLC .6; R/ acts transitively
on M and preserves the pseudo-Hermitian form  up to a positive factor (SL.6; R/
acts isometrically), the signature of the restriction of  to M does not depend on
the base point. This shows that the inclusion M  V is a holomorphic conical
nondegenerate Lagrangian immersion. By Theorem 4.4.5, it induces a conical
special (pseudo-)Kähler structure . J; g; r; / on M. It follows that the image
MN D .M/ D X.0/ Š SL.6; R/= .U.1/  SL.3; C// of M under the projection
 W V n f0g ! P.V/ is a homogeneous projective special Kähler manifold.
The induced pseudo-Hermitian form N on T. p/ M N has signature .3; 6/. The latter
statement follows from formula N .dp X; dp Y/ D .X;Y/
. p;p/
for X; Y 2 Tp M \ p?  V
(see [4]), since . p; p/ D 1 for p D e123 . t
u

The Case G D SU.3; 3/

Using the pseudo-Hermitian form h D h; i on C6 invariant under G D SU.3; 3/ we

can decompose the Grassmannian X D Gr3 .C6 / into the G-invariant real algebraic
subvarieties X.k/ D X.k/ .h/ WD fE 2 X j rk.hjE / D kg, k 2 f0; 1; 2; 3g. Notice
that only X.3/  X is open and that it can be decomposed further according to the
possible signatures of hjE :

X.s;t/ WD fE 2 X j E has signature .s; t/g;

where .s; t/ 2 f.3; 0/; .2; 1/; .1; 2/; .0; 3/g.
Theorem 4.4.8 The group SU.3; 3/ has precisely four open orbits on the highest
weight variety P.V/  X Š Gr3 .C6 / of V D ^3 C6 , namely X.3;0/ , X.2;1/ , X.1;2/ and
X.0;3/ . In all four cases the cone M D C.X.s;t/ /  V carries an SU.3; 3/-invariant
special Kähler structure.

C.X.3;0/ / Š R  SU.3; 3/= SU.3/  SU.3/

has Hermitian signature .1; 9/. C.X.0;3/ / Š R  SU.3; 3/= SU.3/  SU.3/ has
Hermitian signature .9; 1/. For fs; tg D f2; 1g,

C.X.s;t/ / Š R  SU.3; 3/= SU.2; 1/  SU.1; 2/

168 4 Hitchin’s Flow Equations

has Hermitian signature .5; 5/. In all cases, the conical special Kähler manifold
M D C.X.s;t/ / induces on MN D X.s;t/ the structure of a homogeneous projective
N
special Kähler manifold M. For fs; tg D f3; 0g,

N D X.s;t/ Š X.3;0/ Š SU.3; 3/= S.U.3/  U.3//

M

has Hermitian signature .0; 9/, for fs; tg D f2; 1g,

N D X.s;t/ Š X.2;1/ Š SU.3; 3/= S.U.2; 1/  U.1; 2//

M

has Hermitian signature .4; 5/. The special Kähler manifolds C.X.s;t/ / and C.X.t;s/ /
are equivalent. In fact, they are related by a holomorphic r-affine anti-isometry,
which induces a holomorphic isometry between the corresponding projective special
Kähler manifolds.
Proof X.3/  X is Zariski open and is decomposed into the four open (in the
standard topology) orbits X.s;t/ of G D SU.3; 3/. Let .e1 ; e2 ; e3 ; f1 ; f2 ; f3 / be a unitary
basis of .C6 ; h/, such that hei ; ei i D h fi ; fi i D 1. Then X.3;0/ , X.2;1/ , X.1;2/ , X.0;3/ are
the G-orbits of the lines generated by the elements e123 ; e1 ^ f12 ; e12 ^ f1 , f123 2 V,
respectively.
For p D e123 and M D C.X.3;0/ /, we calculate

Tp M D spanfe123 ; fi ^ ejk g:

From (4.44) we see that the matrix of  jTp M with respect to the basis .e123 ; fi ^ ejk /
is diag.1; 19 /. Therefore  has signature .1; 9/ on Tp M. Since .e123 ; e123 / D 1,
the signature of the induced pseudo-Hermitian form N on Tp M N is .0; 9/.
For p D e12 ^ f1 and M D C.X.2;1/ /, we obtain

Tp M D spanfe1 ^ f12 ; e2 ^ f12 ; e2 ^ f13 ; e1 ^ f13 ; e123 ; eij ^ f1 ; e12 ^ fi g:

The matrix of  jTp M with respect to the above basis is diag.15 ; 15 /. Therefore 
has signature .5; 5/ on Tp M. We have .e12 ^ f1 ; e12 ^ f1 / D 1 and the signature
of the induced pseudo-Hermitian form N on Tp M N is .4; 5/.
The linear transformation which sends the vectors ei to fi and fi to ei induces a
linear map ' W V ! V, which interchanges the cone C.X.s;t/ / with C.X.t;s/ / and
maps  to  . This shows that  has signature .9; 1/ and .5; 5/ on C.X.0;3/ / and
C.X.1;2/ /, respectively. As a consequence, the induced pseudo-Hermitian form N on
MN D X.0;3/ , X.1;2/ , has still signature .0; 9/ and .4; 5/, respectively.
It follows from these calculations that the inclusion M D C.X.s;t/ /  V is a
holomorphic conical nondegenerate Lagrangian immersion. By Theorem 4.4.5, it
induces a conical special (pseudo-)Kähler structure . J; g; r; / on M and M N D
X.s;t/  P.V/ is a projective special Kähler manifold.
The above linear anti-isometry ' W .V; p  / ! .V;  / maps  to , and,
hence, preserves the real structure  D 1 1 ı . As a result, it maps the
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 169

special Kähler structure . J; g; r/ of C.X.s;t/ / to . J 0 ; g0 ; r 0 /, where . J 0 ; g0 ; r 0 /

is the special Kähler structure of C.X.t;s/ /. In particular, it induces a holomorphic
isometry X.s;t/ Š X.t;s/ . t
u

The Case G D SU.5; 1/

Let h D h; i be the standard pseudo-Hermitian form of signature .5; 1/ on C6 ,

which is invariant under G D SU.5; 1/. Let us fix a unitary basis .e1 ; : : : ; e5 ; f /
of .C6 ; h/, such that hei ; ei i D h f ; f i D 1. As in the previous subsection,
X D Gr3 .C6 / is decomposed into the G-invariant real algebraic subvarieties
X.k/ D X.k/ .h/, of which X.3/  X is Zariski open. X.3/ is now the union of the
two open G-orbits X.3;0/ and X.2;1/ . X.3;0/ is the orbit of the line Ce123 2 P.V/ and
X.2;1/ is the orbit of Ce45 ^ f 2 P.V/.
Theorem 4.4.9 The group SU.5; 1/ has precisely two open orbits on the highest
weight variety P.V/  X Š Gr3 .C6 / of V D ^3 C6 , namely X.3;0/ and X.2;1/ . In both
cases the cone M D C.X.s;t/ /  V carries an SU.5; 1/-invariant special Kähler
structure.

C.X.3;0/ / Š R  SU.5; 1/= SU.3/  SU.2; 1/

has Hermitian signature .7; 3/.

C.X.2;1/ / Š R  SU.5; 1/= SU.3/  SU.2; 1/

has Hermitian signature .3; 7/. In both cases, the conical special Kähler manifold
M D C.X.s;t/ / induces on MN D X.s;t/ the structure of a homogeneous projective
N
special Kähler manifold M.

N D X.3;0/ Š SU.5; 1/= S.U.3/  U.2; 1//

M

and

N D X.2;1/ Š SU.5; 1/= S.U.3/  U.2; 1//

M

have both Hermitian signature .6; 3/. The special Kähler manifolds C.X.3;0/ /
and C.X.2;1/ / are equivalent. In fact, they are related by a holomorphic r-affine
anti-isometry, which induces a holomorphic isometry between the corresponding
projective special Kähler manifolds.
Proof For p D e123 and M D C.X.3;0/ /, we have

Tp M D spanfe123 ; e234 ; e235 ; e134 ; e135 ; e124 ; e125 ; e12 ^ f ; e13 ^ f ; e23 ^ f g
170 4 Hitchin’s Flow Equations

and the restriction of  to Tp M is represented by the matrix diag.17 ; 13 / with

respect to the above basis. This shows that the inclusion M D C.X.3;0/ /  V is
a holomorphic conical nondegenerate Lagrangian immersion. By Theorem 4.4.5,
it induces a conical special (pseudo-)Kähler structure . J; g; r; / of Hermitian
N D X.3;0/  P.V/ is a projective special Kähler manifold
signature .7; 3/ on M and M
of Hermitian signature .6; 3/. The anti-isometry relating C.X.3;0/ / and C.X.2;1/ / is
induced by the linear map ' W V ! V which has the matrix
 
0 110
;
110 0

with respect to the basis (4.46). t

u

4.4.4 The Homogeneous Projective Special Para-Kähler

Manifold SL.6; R/= S.GL.3; R/  GL.3; R//

Let us first briefly recall the necessary definitions and constructions from special
para-Kähler geometry, see [44] for more details.

Basic Facts About Special Para-Kähler Manifolds

Definition 4.4.10 A special para-Kähler manifold .M; P; g; r/ is a para-Kähler

manifold .M; P; g/ endowed with a flat torsion-free connection r such that r! D 0
and dr P D 0, where ! D g.; P/.
A conical special para-Kähler manifold .M; P; g; r; / is a special para-Kähler
manifold
.M; P; g; r/ endowed with a timelike or a spacelike vector field  such that r D
D D Id, where D is the Levi-Civita connection.
It follows from the definition of a para-Kähler manifold that the eigenspaces of P
are of the same dimension and involutive. An endomorphism field P 2 .End TM/
with these properties is called a para-complex structure on M. The pair .M; P/
is then called a para-complex manifold. A smooth map f W .M; PM / ! .N; PN /
between para-complex manifolds is called para-holomorphic if df ı PM D PN ı df .
The skew-symmetry of P in the definition of a para-Kähler manifold implies that
the eigenspaces of P are totally isotropic of dimension n D 12 dim M. In particular,
M is of even dimension 2n and g is of signature .n; n/.
On any conical special para-Kähler manifold, the vector fields  and P generate
a free para-holomorphic action of a two-dimensional Abelian Lie algebra. If the
action can be integrated to a free para-holomorphic action of a Lie group A such
that the quotient map M ! M N WD M=A is a para-holomorphic submersion, then
MN is called a projective special para-Kähler manifold. The quotient M N carries a
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 171

canonical para-Kähler metric gN compatible with the induced para-complex structure

P on MN D M=A.
Next we explain the extrinsic construction of special para-Kähler manifolds.
Recall that a para-complex vector space V of dimension n is simply a free module
V Š Cn over the ring C D RŒe, e2 D 1, of para-complex numbers. Notice that
Cn is a para-complex manifold with the para-complex structure v 7! ev and any
para-complex manifold of real dimension 2n is locally isomorphic to Cn . An R-
linear map  W V ! V on a para-complex vector space is called anti-linear if
.ev/ D e.v/ for all v 2 V. An example is the para-complex conjugation
Cn ! Cn , z D xCey 7! zN WD xey. Let .V; / be a para-complex symplectic vector
space of dimension 2n endowed with a real structure (i.e. an anti-linear involution)
 such that (4.47) holds true. Then

 WD e.; / (4.50)

is a para-Hermitian form and gV WD Re  is a flat para-Kähler metric on V.

Definition 4.4.11 A para-holomorphic immersion  W M ! V from a para-
complex manifold .M; P/ of real dimension 2n into V is called
(i) nondegenerate if    is nondegenerate,
(ii) Lagrangian if    D 0 and
(iii) conical if . p/ 2 d.Tp M/ and .. p/; . p// ¤ 0 for all p 2 M.

Theorem 4.4.12 ([44])

(i) Any nondegenerate para-holomorphic Lagrangian immersion  W M ! V
induces on the para-complex manifold .M; P/ the structure of a special para-
Kähler manifold .M; P; g; r/, where g D Re    and r is determined by the
condition r  ˛ D 0 for all ˛ 2 V  which are real valued on V  .
(ii) Any conical nondegenerate para-holomorphic Lagrangian immersion  W M !
V induces on .M; P/ the structure of a conical special para-Kähler manifold
.M; P; g; r; /. The vector field  is determined by the condition d . p/ D
. p/.

The (Affine) Special Para-Kähler Manifold as a Para-Complex Lagrangian

Cone

Now we consider the real symplectic module V0 D ^3 R6 of G D SL.6; R/. For

convenience, the standard basis of R6 is denoted by .e1 ; e2 ; e3 ; f1 ; f2 ; f3 /. The para-
complexification V WD V0 ˝ C D ^3 C6 Š C20 of V0 is a para-complex symplectic
vector space endowed with a real structure  such that V  D V0 and (4.47). We put
ui WD ei C efi and consider the orbit

V  M D GLC .6; R/p Š GLC .6; R/= SL.3; R/  SL.3; R/

of the element p D u1 ^ u2 ^ u3 .
172 4 Hitchin’s Flow Equations

Theorem 4.4.13 M D GLC .6; R/p  V is a nondegenerate para-complex

Lagrangian cone. The inclusion M  V induces on M an SL.6; R/-invariant special
N D .M/ Š SL.6; R/= S.GL.3; R/GL.3; R//
para-Kähler structure. The image M
under the projection  W V 0 ! P.V 0 / is a homogeneous projective special para-
Kähler manifold of real dimension 18. Here V 0  V stands for the subset of
nonisotropic vectors.
Proof Using the formulas (4.43) and (4.50) with D eu123 ^ uN 123 D 8e123 ^ f123
we compute: . p; p/ D 1. This shows that M D GLC .6; R/p  V 0 consists of
spacelike vectors. The tangent Tp M  V has the following basis:

.u123 ; uN 1 ^u23 ; uN 2 ^u31 ; uN 3 ^u12 ; uN 2 ^u23 ; uN 2 ^u12 ; uN 3 ^u23 ; uN 1 ^u13 ; uN 3 ^u13 ; Nu1 ^u12 /:

The restriction of  to Tp M is zero in view of (4.43). The para-Hermitian form

 jTp M is represented by the matrix (4.49). This shows that  jTp M is nondegenerate.
Hence, the inclusion M  V is a conical para-holomorphic nondegenerate
Lagrangian immersion. In virtue of Theorem 4.4.12 it induces an SL.6; R/-invariant
conical special para-Kähler structure .P; g; r; / on M, which in turn induces a
homogeneous projective special para-Kähler structure on M N D .M/  P.V 0 /. u t

The Special Para-Kähler Manifold as an Open Orbit of GLC .6; R/ on ^3 R6

The conical special Kähler manifold M D GLC .6; R/= SL.3; C/ described in
Theorem 4.4.7 as a complex Lagrangian cone M  V0 ˝ C can be identified with
the open GLC .6; R/-orbit f < 0g  V0 D ^3 R6 , where  stands for the quartic
SL.6; R/-invariant, see Eq. (2.4):
Proposition 4.4.14 The projection  W V0 ˝ C ! V0 , v 7! Re v, induces
a GLC .6; R/-equivariant diffeomorphism from the Lagrangian cone C.X.0/ / 
V0 ˝ C described in Theorem 4.4.7 onto f < 0g  V0 :

C.X.0/ / Š f < 0g Š GLC .6; R/= SL.3; C/:

Proof This follows from the fact that  is negative on the real part of a non-zero
decomposable .3; 0/-vector, since f < 0g Š GLC .6; R/= SL.3; C/ is connected,
see Proposition 2.1.5. t
u
In that picture the complex structure is less obvious than in the complex
Lagrangian picture but the flat connection and symplectic (Kähler) form are simply
the given structures of the symplectic vector space V0 . The complex structure
p is
then obtained from the metric, which is the Hessian of the function f D jj. (We
consider  as a scalar invariant by choosing a generator of ^6 R6 .) This route was
followed by Hitchin in [76].
The other open GLC .6; R/-orbit f > 0g  V0 cannot be obtained as the real
image of a GLC .6; R/-orbit on the complex Lagrangian cone C.X/  V0 ˝ C over
4.4 Special Geometry of Real Forms of the Symplectic SL.6; C/-Module ^3 C6 173

the highest weight variety X  P.V0 ˝ C/. In fact, GLC .6; R/ has only one open
orbit on X, see Theorem 4.4.7, and that orbit maps to f < 0g  V0 under the
projection V0 ˝ C ! V0 . Instead we have:
Proposition 4.4.15 The projection  W V0 ˝ C ! V0 , v 7! Re v D vC2 vN , induces
a GLC .6; R/-equivariant diffeomorphism from the para-complex Lagrangian cone
M D GLC .6; R/p  V0 ˝ C, p D u123 , described in Theorem 4.4.13 onto the open
GLC .6; R/-orbit f > 0g  V0 :

M Š f > 0g Š GLC .6; R/= .SL.3; R/  SL.3; R//:

Proof It suffices to check that Re u123 2 f > 0g. This follows from the expression

2Re u123 D ..e1 C ef1 / ^ .e2 C ef2 / ^ .e3 C ef3 /

C.e1  ef1 / ^ .e2  ef2 / ^ .e3  ef3 //
D ..e1 C f1 / ^ .e2 C f2 / ^ .e3 C f3 /
C.e1  f1 / ^ .e2  f2 / ^ .e3  f3 //;

since a three-vector belongs to f > 0g if and only if it can be written as the sum
of two decomposable three-vectors which have a non-trivial wedge product, see
Proposition 2.1.5. t
u
Let us denote by r the standard flat connection of the vector space V0 , by 
the position vector field, by ! its SL.6; R/-invariant symplectic
p form and by Xf the
Hamiltonian vector field associated to the function f D . Then we have:
Theorem 4.4.16 The data (P D rXf , g D !ıP, r, ) define on U D f > 0g  V0
an SL.6; R/-invariant conical special para-Kähler structure.
Proof Any three-vector 2 U can be written uniquely as C C  with
decomposable three-vectors ˙ such that C ^  D f . / , cf. Eq. (2.2) and
Corollary 2.1.7. Differentiation at in direction of a vector  2 V0 yields
C  C 
.df / D .  / ^  D !.  ; / ;

that is
C 
Xf . / D  : (4.51)

Using this equation, we can calculate P D rXf by ordinary differentiation in the

vector space V0 . The result is that P acts as identity on the subspace ^3 EC ˚^2 EC ^
E  V0 D ^3 R6 and as minus identity on the subspace ^3 E ˚ ^2 E ^ EC where
˙
E˙ D spanf˛ j ˛ 2 ^2 .R6 / g denotes the support of the three-vectors C

and . This shows that P2 D Id and that P is skew-symmetric with respect to !. To
prove that the data (P, g D ! ı P, r, ) define on U D f > 0g  V0 an SL.6; R/-
174 4 Hitchin’s Flow Equations

invariant conical special para-Kähler structure, it suffices to show that under the map
 W V0 ˝ C ! V0 these data correspond to the conical special para-Kähler structure
on M D GLC .6; R/p  V0 ˝ C, p D u123 , described in Theorem 4.4.13. It follows
from Proposition 4.4.15 and the definition of the structures on M that the data (!,
r, ) on U correspond to the symplectic structure, flat connection and Euler vector
field of the conical special para-Kähler manifold M. One can check by a simple
direct calculation that the endomorphism P on T. p/ U corresponds to multiplication
by e 2 C on Tp M  V0 ˝ C. This proves the theorem.
Alternatively, we give now a direct argument which avoids the use of Theo-
rem 4.4.13. The structure P on U satisfies

dr P D dr rXf D .dr /2 Xf D 0;

since r is flat. This easily implies the integrability of P by expanding the brackets in
the Nijenhuis tensor using that r has zero torsion. In view of the fact that P is skew-
symmetric for !, we conclude that .U; P; g D ! ı P/ is para-Kähler. Finally, the flat
torsion-free connection r satisfies not only dr P D 0 but also r! D 0, since the
two-form ! on V0 is constant. This proves that .U; P; g; r/ is special para-Kähler.
Now we check that .U; P; g; r; / is a conical special para-Kähler manifold, that is
r D D D Id. It is clear that r D Id, since  is the position vector field in V0 . To
prove the second equation, we first remark that the Levi-Civita connection is given
by
1
D D r C PrP:
2
(It suffices to check that D is metric and torsion-free.). Therefore, the equation D D
Id is reduced to r P D 0. Let us first prove that  is para-holomorphic, that is
L P D 0. By homogeneity of f and !, we have the Lie derivatives

L f D 2f ; L df D 2df ; L ! D 2!; L ! 1 D 2! 1

and, hence,

L Xf D 0:

The latter equation implies

L P D L .rXf / D 0;

since  is an affine (and even linear) vector field. Using r  L D r D Id we get

that

r P D L P C ŒId; P D 0:

t
u
References

1. I. Agricola, J. Höll, Cones of G manifolds and Killing spinors with skew torsion. Ann. Mat.
Pura Appl. 194, 673–718 (2015)
2. V. Apostolov, S. Salamon, Kähler reduction of metrics with holonomy G2 . Commun. Math.
Phys. 246(1), 43–61 (2004). doi:10.1007/s00220-003-1014-2. MR2044890 (2005e:53070)
3. D.V. Alekseevsky, Classification of quaternionic spaces with transitive solvable group of
motions. Izv. Akad. Nauk SSSR Ser. Mat. 39(2), 315–362 (1975). 472 [Russian]. MR0402649
(53 #6465)
4. D.V. Alekseevsky, V. Cortés, Classification of stationary compact homogeneous special
pseudo-Kähler manifolds of semisimple groups. Proc. Lond. Math. Soc. (3) 81(1), 211–230
(2000). doi: 10.1112/S0024611500012363. MR1758493 (2001h:53063)
5. D. Alekseevsky, V. Cortés, The twistor spaces of a para-quaternionic Kähler manifold. Osaka
J. Math. 45(1), 215–251 (2008). MR2416658 (2009g:53072)
6. D.V. Alekseevsky, V. Cortés, C. Devchand, Special complex manifolds. J. Geom. Phys. 42(1–
2), 85–105 (2002). doi:10.1016/S0393-0440(01)00078-X. MR1894078 (2003i:53064)
7. D.V. Alekseevsky, N. Blaz̆ić, V. Cortés, S. Vukmirović, A class of Osserman spaces.
J. Geom. Phys. 53(3), 345–353 (2005) doi:10.1016/j.geomphys.2004.07.004. MR2108535
(2005h:53067)
8. D.V. Alekseevsky, V. Cortés, A.S. Galaev, T. Leistner, Cones over pseudo-
Riemannian manifolds and their holonomy. J. Reine Angew. Math. 635, 23–69 (2009).
doi:10.1515/CRELLE.2009.075. MR2572254 (2011b:53115)
9. D.V. Alekseevsky, K. Medori, A. Tomassini, Homogeneous para-
Kählerian Einstein manifolds. Uspekhi Mat. Nauk 64(1)(385), 3–50 (2009).
doi:10.1070/RM2009v064n01ABEH004591 (Russian, with Russian summary) English
translation, Russian Math. Surv. 64(1), 1–43 (2009). MR2503094 (2010k:53068)
10. D.V. Alekseevsky, V. Cortés, T. Mohaupt, Conification of Kähler and hyper-Kähler manifolds.
Commun. Math. Phys. 3242, 637–655 (2013). doi:10.1007/s00220-013-1812-0. MR3117523
11. D.V. Alekseevsky, B.S. Kruglikov, H. Winther, Homogeneous almost complex structures in
dimension 6 with semi- simple isotropy. Ann. Global Anal. Geom. 46(4), 361–387 (2014).
doi:10.1007/s10455-014-9428-y. MR3276441
12. C. Bär, Real Killing spinors and holonomy. Commun. Math. Phys. 154(3), 509–521 (1993).
MR1224089 (94i:53042)
13. O. Baues, Prehomogeneous affine representations and flat pseudo-Riemannian manifolds,
in Handbook of Pseudo- Riemannian Geometry and Supersymmetry. IRMA Lectures in
Mathematics and Theoritical Physics, vol. 16 (European Mathematical Society, Zürich, 2010),
pp. 731–817. doi:10.4171/079-1/22. MR2681607 (2012a:53082)

© Springer International Publishing AG 2017 175

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0
176 References

14. O. Baues, V. Cortés, Proper affine hyperspheres which fiber over projective special Kähler
manifolds. Asian J. Math. 7(1), 115–132 (2003). MR2015244 (2005a:53076)
15. C.-L. Bejan, Some examples of manifolds with hyperbolic structures. Rend. Mat. Appl.
(7) 14(4), 557–565 (1994) [English, with English and Italian summaries]. MR1312817
(95k:53042)
16. F. Belgun, A. Moroianu, Nearly Kähler 6-manifolds with reduced holonomy. Ann.
Global Anal. Geom. 19(4), 307–319 (2001). doi:10.1023/A:1010799215310. MR1842572
(2002f:53083)
17. L. Bérard Bergery, A. Ikemakhen, Sur l’holonomie des variétés pseudo-riemanniennes de
signature (n n/. Bull. Soc. Math. France 125(1), 93–114 (1997) [French, with English and
French summaries]. MR1459299 (98m:53087)
18. A.L. Besse, Einstein Manifolds. Ergebnisse der Mathematik und ihrer Grenzgebiete (3)
[Results in Mathematics and Related Areas (3)], vol. 10 (Springer, Berlin, 1987). MR867684
(88f:53087)
19. C. Böhm, M.M. Kerr, Low-dimensional homogeneous Einstein manifolds. Trans. Am.
Math. Soc. 358(4), 1455–1468 (2006). doi:10.1090/S0002-9947-05-04096-1. MR2186982
(2006g:53056)
20. C. Boyer, K. Galicki, Sasakian Geometry. Oxford Mathematical Monographs (Oxford
University Press, Oxford, 2008). MR2382957 (2009c:53058)
21. N. Blažić, Paraquaternionic projective space and pseudo-Riemannian geometry. Publications
de l’Institut Mathématique Nouvelle série 60, 101–107 (1996)
22. E. Bonan, Sur des variétés riemanniennes à groupe d’holonomie G2 ou spin (7). C. R. Acad.
Sci. Paris Sér. A-B 262, A127–A129 (1966) [French]. MR0196668 (33 #4855)
23. R.L. Bryant, Metrics with exceptional holonomy. Ann. Math. (2) 126(3), 525–576 (1987).
doi:10.2307/1971360. MR916718 (89b:53084)
24. R.L. Bryant, On the geometry of almost complex 6-manifolds. Asian J. Math. 10(3), 561–605
(2006). doi:10.4310/AJM.2006.v10.n3.a4. MR2253159 (2007g:53029)
25. J.-B. Butruille, Classification des variétés approximativement kähleriennes homogénes. Ann.
Global Anal. Geom. 27(3), 201–225 (2005). doi: 10.1007/s10455-005-1581-x (French, with
English summary). MR2158165 (2006f:53060)
26. J.-B. Butruille, Homogeneous nearly Kähler manifolds, in Handbook of Pseudo-Riemannian
Geometry and Supersymmetry. IRMA Lectures in Mathematics and Theoretical Physics, vol.
16 (European Mathematical Society, Zürich, 2010), pp. 399–423, doi: 10.4171/079-1/11.
MR2681596 (2011m:53083)
27. F.M. Cabrera, SU(3)-structures on hypersurfaces of manifolds with G2 -structure. Monatsh.
Math. 148(1), 29–50 (2006). doi:10.1007/s00605-005-0343-y. MR2229065 (2007b:53059)
28. E. Calabi, Construction and properties of some 6-dimensional almost complex manifolds.
Trans. Am. Math. Soc. 87, 407–438 (1958). MR0130698 (24 #A558)
29. B. Cappelletti-Montano, A. Carriazo, V. Martín-Molina, Sasaki-Einstein and paraSasaki-
Einstein metrics from (g g-structures. J. Geom. Phys. 73, 20–36 (2013).
doi:10.1016/j.geomphys.2013.05.001. MR3090100
30. S. Cecotti, C. Vafa, Topological-anti-topological fusion. Nucl. Phys. B 367(2), 359–461
(1991). doi:10.1016/0550-3213(91)90021-O. MR1139739 (93a:81168)
31. S.G. Chiossi, A. Fino, Conformally parallel G2 structures on a class of solvmanifolds. Math.
Z. 252(4), 825–848 (2006). doi: 10.1007/s00209-005-0885-7. MR2206629 (2007a:53098)
32. S. Chiossi, S. Salamon, The intrinsic torsion of SU(3) and G2 structures, in Differential
Geometry, Valencia, 2001 (World Scientific Publisher, River Edge, NJ, 2002), pp. 115–133.
MR1922042 (2003g:53030)
33. S.G. Chiossi, A. Swann, G2 -structures with torsion from half-integrable nilmanifolds. J.
Geom. Phys. 54(3), 262–285 (2005). doi:10.1016/j.geomphys.2004.09.009. MR2139083
(2006a:53054)
34. R. Cleyton, A. Swann, Einstein metrics via intrinsic or parallel torsion. Math. Z. 247(3), 513–
528 (2004). doi:10.1007/s00209-003-0616-x. MR2114426 (2005i:53054)
References 177

35. D. Conti, T.B. Madsen, The odd side of torsion geometry. Ann. Mat. Pura Appl. (4) 193(4),
1041–1067 (2014). doi:10.1007/s10231-012-0314-6. MR3237915
36. D. Conti, S. Salamon, Generalized Killing spinors in dimension 5. Trans. Am. Math.
Soc. 359(11), 5319–5343 (2007). doi: 10.1090/S0002-9947-07-04307-3. MR2327032
(2008h:53077)
37. D. Conti, A. Tomassini, Special symplectic six-manifolds. Q. J. Math. 58(3), 297–311 (2007).
doi:10.1093/qmath/ham013. MR2354920 (2008m:53118)
38. V. Cortés, Alekseevskian spaces. Differ. Geom. Appl. 6(2), 129–168 (1996). doi:
10.1016/0926- 2245(96)89146-7. MR1395026 (97m:53079)
39. V. Cortés, Odd Riemannian symmetric spaces associated to four-forms. Math. Scand. 98(2),
201–216 (2006). MR2243702 (2007f:53057)
40. V. Cortés, L. Schäfer, Topological-antitopological fusion equations, pluriharmonic maps and
special Kähler mani- folds. Progress in Mathematics, vol. 234 (Birkhäuser Boston, Boston,
MA, 2005), pp. 59–74
41. V. Cortés, L. Schäfer, Flat nearly Kähler manifolds. Ann. Global Anal. Geom. 32(4), 379–389
(2007). doi:10.1007/s10455- 007-9068-6. MR2346224 (2009a:53048)
42. V. Cortés, L. Schäfer, Differential geometric aspects of the tt -equations, in From Hodge
Theory to Integrability and TQFT tt*-Geometry. Proceedings of Symposia in Pure Math-
ematics, vol. 78 (American Mathematical Society, Providence, RI, 2008), pp. 75–86. doi:
10.1090/pspum/078/2483749. MR2483749 (2010f:53103)
43. V. Cortés, L. Schäfer, Geometric structures on Lie groups with flat bi-invariant metric. J. Lie
Theory 19(2), 423–437 (2009). MR2572138 (2010m:53104)
44. V. Cortés, C. Mayer, T. Mohaupt, F. Saueressig, Special geometry of Euclidean super-
symmetry. I. Vector multi- plets. J. High Energy Phys. 3(028), 73pp. (2004) [electronic]
doi:10.1088/1126-6708/2004/03/028. MR2061551 (2005c:53055)
45. V. Cortés, M.-A. Lawn, L. Schäfer, Affine hyperspheres associated to special para-
Kähler manifolds. Int. J. Geom. Methods Mod. Phys. 3(5–6), 995–1009 (2006).
doi:10.1142/S0219887806001569. MR2264401 (2007j:53077)
46. V. Cortés, T. Leistner, L. Schäfer, F. Schulte-Hengesbach, Half-flat structures and special
holonomy. Proc. Lond. Math. Soc. (3) 102(1), 113–158 (2011). doi:10.1112/plms/pdq012.
MR2747725 (2012d:53075)
47. V. Cruceanu, P. Fortuny, P.M. Gadea, A survey on paracomplex geometry. Rocky Mt. J. Math.
26(1), 83–115 (1996). doi:10.1216/rmjm/1181072105. MR1386154 (97c:53112)
48. B. Dubrovin, Geometry and integrability of topological-antitopological fusion. Commun.
Math. Phys. 152(3), 539–564 (1993). MR1213301 (95a:81227)
49. M.J. Duff, B.E.W. Nilsson, C.N. Pope, Kaluza-Klein supergravity. Phys. Rep. 130(1–2), 1–
142 (1986). doi:10.1016/0370-1573(86)90163-8. MR822171 (87f:83061)
50. J. Eells, S. Salamon, Twistorial construction of harmonic maps of surfaces into four-
manifolds. Ann. Scuola Norm. Sup. Pisa Cl. Sci. (4) 12(4), 589–640 (1985/1986). MR848842
(87i:58042)
51. N. Ejiri, Totally real submanifolds in a 6-sphere. Proc. Am. Math. Soc. 83(4), 759–763 (1981).
doi:10.2307/2044249. MR630028 (83a:53033)
52. M. Falcitelli, A. Farinola, S. Salamon, Almost-Hermitian geometry. Differ Geom. Appl. 4(3),
259–282 (1994). doi: 10.1016/0926-2245(94)00016-6. MR1299398 (95i:53047)
53. M. Fernández, S. Ivanov, V. Muñoz, L. Ugarte, Nearly hypo structures and compact nearly
Kähler 6-manifolds with conical singularities. J. Lond. Math. Soc. (2) 78(3), 580–604 (2008).
doi:10.1112/jlms/jdn044. MR2456893 (2009m:53061)
54. S. Ferrara, S. Sabharwal, Quaternionic manifolds for type II superstring vacua of Calabi-
Yau spaces. Nucl. Phys. B 332(2), 317–332 (1990). doi:10.1016/0550-3213(90)90097-W.
MR1046353 (91g:53051)
55. L. Foscolo, M. Haskins, New G2 holonomy cones and exotic nearly Kaehler structures on the
6-sphere and the product of a pair of 3-spheres. arXiv:1501.07838. https://spiral.imperial.ac.
uk:8443/handle/10044/1/49672
178 References

56. T. Friedrich, Der erste Eigenwert des Dirac-Operators einer kompakten, Riemannschen Man-
nigfaltigkeit nichtnega- tiver Skalarkrümmung. Math. Nachr. 97, 117–146 (1980) [German]
MR600828 (82g:58088)
57. T. Friedrich, S. Ivanov, Parallel spinors and connections with skew-symmetric torsion in string
theory. Asian J. Math. 6(2), 303–335 (2002). MR1928632 (2003m:53070)
58. T. Friedrich, I. Kath, A. Moroianu, U. Semmelmann, On nearly parallel G2 -structures. J.
Geom. Phys. 23(3–4), 259–286 (1997). doi:10.1016/S0393-0440(97)80004-6. MR1484591
(98j:53053)
59. P.M. Gadea, J.M. Masque, Classification of almost para-Hermitian manifolds. Rend. Mat.
Appl. (7) 11(2), 377–396 (1991) [English, with Italian summary]. MR1122346 (92i:53028)
60. S. Gallot, Équations différentielles caractéristiques de la sphére. Ann. Sci. École Norm. Sup.
(4) 12(2), 235–267 (1979). (French). MR543217 (80h:58051)
61. R. Gover, R. Panai, T. Willse, Nearly Kähler geometry and (2 3 5)-distributions via projective
holonomy (2014). arxiv:1403.1959
62. A. Gray, Minimal varieties and almost Hermitian submanifolds. Mich. Math. J. 12, 273–287
(1965). MR0184185 (32 #1658)
63. A. Gray, Vector cross products on manifolds. Trans. Am. Math. Soc. 141, 465–504 (1969).
MR0243469 (39 #4790)
64. A. Gray, Almost complex submanifolds of the six sphere. Proc. Am. Math. Soc. 20, 277–279
(1969). MR0246332 (39 #7636)
65. A. Gray, Nearly Kähler manifolds. J. Differ. Geom. 4, 283–309 (1970). MR0267502 (42
#2404)
66. A. Gray, Riemannian manifolds with geodesic symmetries of order 3. J. Differ. Geom. 7,
343–369 (1972). MR0331281 (48 #9615)
67. A. Gray, The structure of nearly Kähler manifolds. Math. Ann. 223(3), 233–248 (1976).
MR0417965 (54 #6010)
68. A. Gray, L.M. Hervella, The sixteen classes of almost Hermitian manifolds and their
linear invariants. Ann. Mat. Pura Appl. (4) 123, 35–58 (1980). doi: 10.1007/BF01796539.
MR581924 (81m:53045)
69. A. Gray, J.A. Wolf, Homogeneous spaces defined by Lie group automorphisms. II. J. Differ.
Geom. 2, 115–159 (1968). MR0236329 (38 #4625b)
70. R. Grunewald, Six-dimensional Riemannian manifolds with a real Killing spinor. Ann. Global
Anal. Geom. 8(1), 43–59 (1990). doi: 10.1007/BF00055017. MR1075238 (92a:58146)
71. J. Gutowski, S. Ivanov, G. Papadopoulos, Deformations of generalized calibrations and
compact non-Kähler man- ifolds with vanishing first Chern class. Asian J. Math. 7(1), 39–
79 (2003). MR2015241 (2004j:53069)
72. F.R. Harvey, Spinors and Calibrations. Perspectives in Mathematics, vol. 9 (Academic,
Boston, MA, 1990). MR1045637 (91e:53056)
73. A. Haydys, HyperKähler and quaternionic Kähler manifolds with S1 -symmetries. J.
Geom. Phys. 58(3), 293–306 (2008). doi: 10.1016/j.geomphys.2007.11.004. MR2394039
(2008m:53113)
74. C. Hertling, tt geometry, Frobenius manifolds, their connections, and the construction
for singularities. J. Reine Angew. Math. 555, 77–161 (2003). doi: 10.1515/crll.2003.015.
MR1956595 (2005f:32049)
75. N. Hitchin, The geometry of three-forms in six dimensions. J. Differ. Geom. 55(3), 547–576
(2000). MR1863733 (2002m:53070)
76. N. Hitchin, Stable forms and special metrics, (Bilbao, 2000). Contemporary Mathe-
matics, vol. 288 (American Mathematical Society, Providence, RI, 2001), pp. 70–89.
doi:10.1090/conm/288/04818. MR1871001 (2003f:53065)
77. Z.H. Hou, On totally real submanifolds in a nearly Kähler manifold. Port. Math. (N.S.) 58(2),
219–231 (2001). MR1836264 (2002b:53097)
References 179

78. S. Ivanov, S. Zamkovoy, Parahermitian and paraquaternionic manifolds. Differ. Geom. Appl.
23(2), 205–234 (2005). doi: 10.1016/j.difgeo.2005.06.002. MR2158044 (2006d:53025)
79. G.R. Jensen, Imbeddings of Stiefel manifolds into Grassmannians. Duke Math. J. 42(3), 397–
407 (1975). MR0375164 (51 #11360)
80. S. Kaneyuki, M. Kozai, Paracomplex structures and affine symmetric spaces. Tokyo J. Math.
8(1), 81–98 (1985). doi:10.3836/tjm/1270151571. MR800077 (87c:53078)
81. I. Kath, G -structures on pseudo-Riemannian manifolds. J. Geom. Phys. 27(3–4), 155–177
(1998). doi:10.1016/S0393-0440(97)00073-9. MR1645016 (99i:53023)
82. I. Kath, Killing Spinors on Pseudo-Riemannian Manifolds. Habilitationsschrift an der
Humboldt-Universität zu Berlin (1999)
83. G. Ketsetzis, S. Salamon, Complex structures on the Iwasawa manifold. Adv. Geom. 4(2),
165–179 (2004). doi: 10.1515/advg.2004.012. MR2055676 (2005g:22008)
84. T. Kimura, Introduction to Prehomogeneous Vector Spaces. Translations of Mathematical
Monographs, vol. 215 (American Mathematical Society, Providence, RI, 2003). Translated
from the 1998 Japanese original by Makoto Nagura and Tsuyoshi Niitani and revised by the
author. MR1944442 (2003k:11180)
85. T. Kimura, M. Sato, A classification of irreducible prehomogeneous vector spaces and their
relative invariants. Nagoya Math. J. 65, 1–155 (1977). MR0430336 (55 #3341)
86. V.F. Kirichenko, Generalized Gray-Hervella classes and holomorphically projective transfor-
mations of gen- eralized almost Hermitian structures. Izv. Ross. Akad. Nauk Ser. Mat. 69(5),
107–132 (2005). doi: 10.1070/IM2005v069n05ABEH002283 (Russian, with Russian sum-
mary); English translation: Izv. Math. 69(5), 963–987 (2005). MR2179416 (2006g:53028)
87. S. Kobayashi, K. Nomizu, Foundations of Differential Geometry. Volume II. Interscience
Tracts in Pure and Applied Mathematics, No. 15, vols. I/II (Interscience Publishers Wiley,
New York/London/Sydney, 1963/1969)
88. H.B. Lawson Jr., M.-L. Michelsohn, Spin Geometry. Princeton Mathematical Series, vol. 38
(Princeton University Press, Princeton, NJ, 1989). MR1031992 (91g:53001)
89. C. LeBrun, On complete quaternionic-Kähler manifolds. Duke Math. J. 63(3), 723–743
(1991). doi: 10.1215/S0012-7094-91-06331-3. MR1121153 (92i:53042)
90. A.J. Ledger, M. Obata, Affine and Riemannian s-manifolds. J. Differ. Geom. 2, 451–459
(1968). MR0244893 (39 #6206)
91. H.V. Le, L. Schwachhöfer, Lagrangian submanifolds in strictly nearly K”ahler 6-manifolds.
arXiv:1408.6433v1
92. P. Libermann, Sur le probléme d’équivalence de certaines structures infinitésimales. Ann.
Mat. Pura Appl. (4) 36, 27–120 (1954) [French]. MR0066020 (16,520c)
93. O. Macia, A. Swann, Twist geometry of the c-map. Commun. Math. Phys. 336(3), 1329–1357
(2015). doi:10.1007/s00220-015-2314-z. MR3324146
94. A.I. Mal0 cev, On a class of homogeneous spaces. Izvestiya Akad. Nauk. SSSR. Ser. Mat. 13,
9–32 (1949) [Russian]. MR0028842 (10,507d)
95. J. Milnor, Curvatures of left invariant metrics on Lie groups. Adv. Math. 21(3), 293–329
(1976). MR0425012 (54 #12970)
96. A. Moroianu, P.-A. Nagy, U. Semmelmann, Deformations of nearly Kähler structures. Pac. J.
Math. 235(1), 57–72 (2008). doi: 10.2140/pjm.2008.235.57. MR2379770 (2009b:53038)
97. A. Moroianu, U. Semmelmann, Generalized Killing spinors and Lagrangian graphs. Differ.
Geom. Appl. 37, 141–151 (2014). doi:10.1016/j.difgeo.2014.09.005. MR3277524
98. P.-A. Nagy, On nearly-Kähler geometry. Ann. Global Anal. Geom. 22(2), 167–178 (2002).
doi:10.1023/A:1019506730571. MR1923275 (2003g:53073)
99. P.-A. Nagy, Nearly Kähler geometry and Riemannian foliations. Asian J. Math. 6(3), 481–504
(2002). MR1946344 (2003m:53043)
100. B. O’Neill, Semi-Riemannian Geometry. Pure and Applied Mathematics, vol. 103 (Academic
[Harcourt Brace Jovanovich Publishers], New York, 1983). With applications to relativity.
MR719023 (85f:53002)
101. M.S. Raghunathan, Discrete Subgroups of Lie Groups (Springer, New York, 1972). Ergeb-
nisse der Mathematik und ihrer Grenzgebiete, Band 68. MR0507234 (58 #22394a)
180 References

102. F. Raymond, A.T. Vasquez, 3-manifolds whose universal coverings are Lie groups. Topol.
Appl. 12(2), 161–179 (1981). doi: 10.1016/0166-8641(81)90018-3. MR612013 (82i:57011)
103. R. Reyes Carrión, Some special geometries defined by Lie groups. PhD-thesis, Oxford (1993)
104. S. Salamon, Riemannian Geometry and Holonomy Groups. Pitman Research Notes in
Mathematics Series, vol. 201 (Longman Scientific & Technical, Harlow, 1989); copublished
in the United States with Wiley, New York, 1989. MR1004008 (90g:53058)
105. S. Salamon, Complex structures on nilpotent Lie algebras. J. Pure Appl. Algebra 157(2–3),
311–333 (2001). doi:10.1016/S0022-4049(00)00033-5. MR1812058 (2002g:53089)
106. L. Schäfer, tt -bundles in para-complex geometry, special para-Kähler manifolds
and para-pluriharmonic maps. Differ. Geom. Appl. 24(1), 60–89 (2006).
doi:10.1016/j.difgeo.2005.07.001. MR2193748 (2007c:53080)
107. L. Schäfer, tt -geometry on the tangent bundle of an almost complex manifold. J.
Geom. Phys. 57(3), 999–1014 (2007). doi:10.1016/j.geomphys.2006.08.004. MR2275206
(2007m:53023)
108. L. Schäfer, On the structure of nearly pseudo-Kähler manifolds. Monatsh. Math. 163(3), 339–
371 (2011). doi:10.1007/s00605-009-0184-1. MR2805878 (2012m:53106)
109. L. Schäfer, Conical Ricci-flat nearly para-Kähler manifolds. Ann. Global Anal. Geom. 45(1),
11–24 (2014). doi:10.1007/s10455-013-9385-x. MR3152085
110. L. Schäfer, F. Schulte-Hengesbach, Nearly pseudo-Kähler and nearly para-Kähler six-
manifolds, in Handbook of Pseudo-Riemannian Geometry and Supersymmetry. IRMA
Lectures in Mathematics and Theoritical Physics, vol. 16 (European Mathematical Society,
Zürich, 2010), pp. 425–453. doi: 10.4171/079-1/12. MR2681597 (2011g:53055)
111. L. Schäfer, K. Smoczyk, Decomposition and minimality of Lagrangian submani-
folds in nearly Kähler manifolds. Ann. Global Anal. Geom. 37(3), 221–240 (2010).
doi:10.1007/s10455-009-9181-9. MR2595677 (2011g:53131)
112. F. Schulte-Hengesbach, Half-flat structures on products of three-dimensional Lie groups. J.
Geom. Phys. 60(11), 1726–1740 (2010). doi: 10.1016/j.geomphys.2010.06.012. MR2679416
(2011j:53074)
113. R.W. Sharpe, Differential Geometry. Graduate Texts in Mathematics, vol. 166 (Springer, New
York, 1997). Cartan’s generalization of Klein’s Erlangen program; With a foreword by S. S.
Chern. MR1453120 (98m:53033)
114. S. Stock, Lifting SU(3)-structures to nearly parallel G2 -structures, J. Geom. Phys. 59(1), 1–7
(2009). doi: 10.1016/j.geomphys.2008.08.003. MR2479256 (2010b:53049)
115. A. Strominger, Superstrings with torsion. Nucl. Phys. B 274(2), 253–284 (1986). doi:
10.1016/0550- 3213(86)90286-5. MR851702 (87m:81177)
116. F. Tricerri, L. Vanhecke, Homogeneous Structures on Riemannian Manifolds. London
Mathematical Society Lecture Note Series, vol. 83 (Cambridge University Press, Cambridge,
1983). MR712664 (85b:53052)
117. J.A. Wolf, Complex homogeneous contact manifolds and quaternionic symmetric spaces. J.
Math. Mech. 14, 1033–1047 (1965). MR0185554 (32 #3020)
Index

3-symmetric space nearly Kähler manifold

S3  S3 , 8 automorphisms, 100
SL.2; R/  SL.2; R/, 9 left-invariant, 103
of constant type, 43
nearly para-Kähler manifold, 41
almost para-complex structure, 28 defined by exterior system, 49
almost para-Hermitian manifold, 28 Ricci-flat, 45
almost pseudo-Hermitian manifold, 28 nearly pseudo-Kähler manifold, 41
SL.2; R/  SL.2; R/, 9, 102
characteristic para-Hermitian connection, 42 defined by exterior system, 49
characteristic pseudo-Hermitian connection, 42 Nijenhuis tensor, 31
totally skew-symmetric, 41, 42, 46–48

G1 -manifold, 42
Gray-Hervella classification, 31 SL.3; R/-structure
half-flat, 48
nearly half-flat, 48
half-flat structure, 48
SU.p; q/-structure
half-flat, 48
isometries nearly half-flat, 48
of nearly Kähler structures, 100

torsion
Killing vector fields parallel, 43
of nearly Kähler structures, 100 totally skew-symmetric, 42

nearly half-flat structure, 48

nearly Kähler manifold U.p; q/-structure, 30
S3  S3 , 8 quasi-integrable, 45

© Springer International Publishing AG 2017 181

L. Schäfer, Nearly Pseudo-Kähler Manifolds and Related Special Holonomies,
Lecture Notes in Mathematics 2201, DOI 10.1007/978-3-319-65807-0
LECTURE NOTES IN MATHEMATICS 123
Editors in Chief: J.-M. Morel, B. Teissier;

Editorial Policy
1. Lecture Notes aim to report new developments in all areas of mathematics and their
applications – quickly, informally and at a high level. Mathematical texts analysing new
developments in modelling and numerical simulation are welcome.
Manuscripts should be reasonably self-contained and rounded off. Thus they may, and
often will, present not only results of the author but also related work by other people. They
may be based on specialised lecture courses. Furthermore, the manuscripts should provide
sufficient motivation, examples and applications. This clearly distinguishes Lecture Notes
from journal articles or technical reports which normally are very concise. Articles
intended for a journal but too long to be accepted by most journals, usually do not have
this “lecture notes” character. For similar reasons it is unusual for doctoral theses to be
accepted for the Lecture Notes series, though habilitation theses may be appropriate.

2. Besides monographs, multi-author manuscripts resulting from SUMMER SCHOOLS or

similar INTENSIVE COURSES are welcome, provided their objective was held to present
an active mathematical topic to an audience at the beginning or intermediate graduate level
(a list of participants should be provided).
The resulting manuscript should not be just a collection of course notes, but should require
advance planning and coordination among the main lecturers. The subject matter should
dictate the structure of the book. This structure should be motivated and explained in
a scientific introduction, and the notation, references, index and formulation of results
should be, if possible, unified by the editors. Each contribution should have an abstract
and an introduction referring to the other contributions. In other words, more preparatory
work must go into a multi-authored volume than simply assembling a disparate collection
of papers, communicated at the event.

3. Manuscripts should be submitted either online at www.editorialmanager.com/lnm to

Springer’s mathematics editorial in Heidelberg, or electronically to one of the series edi-
tors. Authors should be aware that incomplete or insufficiently close-to-final manuscripts
almost always result in longer refereeing times and nevertheless unclear referees’ rec-
ommendations, making further refereeing of a final draft necessary. The strict minimum
amount of material that will be considered should include a detailed outline describing
the planned contents of each chapter, a bibliography and several sample chapters. Parallel
submission of a manuscript to another publisher while under consideration for LNM is not
acceptable and can lead to rejection.

4. In general, monographs will be sent out to at least 2 external referees for evaluation.
A final decision to publish can be made only on the basis of the complete manuscript,
however a refereeing process leading to a preliminary decision can be based on a pre-final
or incomplete manuscript.
Volume Editors of multi-author works are expected to arrange for the refereeing, to the
usual scientific standards, of the individual contributions. If the resulting reports can be
 forwarded to the LNM Editorial Board, this is very helpful. If no reports are forwarded
or if other questions remain unclear in respect of homogeneity etc, the series editors may
wish to consult external referees for an overall evaluation of the volume.

5. Manuscripts should in general be submitted in English. Final manuscripts should contain

at least 100 pages of mathematical text and should always include
– a table of contents;
– an informative introduction, with adequate motivation and perhaps some historical
remarks: it should be accessible to a reader not intimately familiar with the topic
treated;
– a subject index: as a rule this is genuinely helpful for the reader.
– For evaluation purposes, manuscripts should be submitted as pdf files.

6. Careful preparation of the manuscripts will help keep production time short besides
ensuring satisfactory appearance of the finished book in print and online. After accep-
tance of the manuscript authors will be asked to prepare the final LaTeX source files
(see LaTeX templates online: https://www.springer.com/gb/authors-editors/book-authors-
editors/manuscriptpreparation/5636) plus the corresponding pdf- or zipped ps-file. The
LaTeX source files are essential for producing the full-text online version of the book,
see http://link.springer.com/bookseries/304 for the existing online volumes of LNM). The
technical production of a Lecture Notes volume takes approximately 12 weeks. Additional
instructions, if necessary, are available on request from lnm@springer.com.

7. Authors receive a total of 30 free copies of their volume and free access to their book on
SpringerLink, but no royalties. They are entitled to a discount of 33.3 % on the price of
Springer books purchased for their personal use, if ordering directly from Springer.

8. Commitment to publish is made by a Publishing Agreement; contributing authors of

multiauthor books are requested to sign a Consent to Publish form. Springer-Verlag
registers the copyright for each volume. Authors are free to reuse material contained in
their LNM volumes in later publications: a brief written (or e-mail) request for formal
permission is sufficient.

Addresses:
Professor Jean-Michel Morel, CMLA, École Normale Supérieure de Cachan, France
E-mail: moreljeanmichel@gmail.com

Professor Bernard Teissier, Equipe Géométrie et Dynamique,

Institut de Mathématiques de Jussieu – Paris Rive Gauche, Paris, France
E-mail: bernard.teissier@imj-prg.fr

Springer: Ute McCrory, Mathematics, Heidelberg, Germany,

E-mail: lnm@springer.com