CALABI-YAU SUBMANIFOLDS OF JOYCE MANIFOLDS OF THE FIRST
KIND
By
Barı¸ Efe
s

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Mathematics
2011

ABSTRACT
CALABI-YAU SUBMANIFOLDS OF JOYCE MANIFOLDS OF THE
FIRST KIND
By
Barı¸ Efe
s
Akbulut and Salur suggested the study of Calabi-Yau submanifolds of G2 manifolds that come from a certain process. The author, in a joint paper with Akbulut and
Salur, applied this process to a Joyce manifold, more specifically, to the Joyce manifold J(1/2, 0, 0, 1/2, 1/2), and obtained a pair of Borcea-Voisin 3-folds with Hodge
numbers h1,1 = h2,1 = 19.
In this thesis, we first list all possible Joyce manifolds of the first kind. Then we
describe the Calabi-Yau submanifolds of these manifolds that come from the process,
we mentioned above, using the coordinate directions. This way we obtain two different
Borcea-Voisin manifolds, as well as T2 × K3 and T6 .

ii

ACKNOWLEDGMENT
I would like to thank my supervisor Dr. Selman Akbulut for his excellent guidance, patience and constant encouragement. Many thanks to my committee members
Dr. Ronald Finthushel, Dr. Rajesh Kulkarni, Dr. Benjamin Schmidt and Dr. Xiaodong Wang. I would also like to thank Dr. Sema Salur for her advices and fruitful
discussions.
Finally, special thanks to my family, my wife Melike and my mother Fatma, for
their constant support and encouragement.

iii

TABLE OF CONTENTS

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vi

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 Background Material .
2.1 Holonomy Groups . . .
2.2 Calabi-Yau Manifolds .
2.3 Orbifolds . . . . . . . .
2.4 Borcea-Voisin 3-folds .

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

.
.
.
.
.

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

.
.
.
.
.

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

.
.
.
.
.

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

.
.
.
.
.

3
3
4
6
9

3 G2 Manifolds . . . . . . . . .
3.1 Holonomy Group G2 . . . . . .
3.2 Akbulut-Salur Construction . .
3.3 MirrorDuality . . . . . . . . . .
3.4 Joyce Manifolds . . . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.

. .
. .
. .
. .
. .

.
.
.
.
.

. .
. . .
. . .
. . .
. . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.

. .
. .
. .
. .
. .

.
.
.
.
.

.
.
.
.
.

.
.
.
.

. .
. .
. .
. .
. .

.
.
.
.
.

.
.
.
.
.

11
11
13
16
18

4 Calabi-Yau Submanifolds of Joyce Manifolds . .
4.1 Joyce Manifolds of First Kind: J(b1 , b2 , c1 , c2 , c3 )
4.2 Calabi-Yau Submanifolds . . . . . . . . . . . . . .
4.2.1 x1 direction . . . . . . . . . . . . . . . . .
4.2.2 x2 direction . . . . . . . . . . . . . . . . .
4.2.3 x3 direction . . . . . . . . . . . . . . . . .
If c1 = 0 . . . . . . . . . . . . . . . . . . .
If c1 = 1/2 . . . . . . . . . . . . . . . . . .
4.2.4 x4 direction . . . . . . . . . . . . . . . . .
4.2.5 x5 direction . . . . . . . . . . . . . . . . .
If c2 = 0 . . . . . . . . . . . . . . . . . . .
If c2 = 1/2 . . . . . . . . . . . . . . . . . .
4.2.6 x6 direction . . . . . . . . . . . . . . . . .
If b1 = 0 . . . . . . . . . . . . . . . . . . .
If b1 = 1/2 . . . . . . . . . . . . . . . . . .
4.2.7 x7 direction . . . . . . . . . . . . . . . . .
4.3 Summary of results . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

21
21
25
26
28
29
29
30
32
32
32
33
33
33
35
35
36

. . . . . . . . . . . . . . . . . . . . . . . . . .

39

Bibliography

.
.
.
.
.

.
.
.
.
.

iv

LIST OF TABLES

4.1

The action of generators of Γ on T7 . . . . . . . . . . . . . . . . . . .

21

4.2

The action of mixed terms of Γ . . . . . . . . . . . . . . . . . . . . .

22

4.3

Fixed point sets for all possible 5-tuples . . . . . . . . . . . . . . . . .

25

4.4

The action of < α, β > on T6 . . . . . . . . . . . . . . . . . . . . . .

27

4.5

The action of < α, γ > on T6 . . . . . . . . . . . . . . . . . . . . . .

28

4.6

The action of < α, βγ > on T6 . . . . . . . . . . . . . . . . . . . . . .

30

v

LIST OF FIGURES

4.1

Base S 1 has x1 coordinate . . . . . . . . . . . . . . . . . . . . . . . .

26

4.2

The action of α on T4 . . . . . . . . . . . . . . . . . . . . . . . . . .

27

4.3

Base S 1 has x2 coordinate . . . . . . . . . . . . . . . . . . . . . . . .

29

4.4

Base S 1 has x3 coordinate, c1 = 0 . . . . . . . . . . . . . . . . . . . .

30

4.5

Base S 1 has x3 coordinate, c1 = 1/2 . . . . . . . . . . . . . . . . . .

31

4.6

Base S 1 has x4 coordinate . . . . . . . . . . . . . . . . . . . . . . . .

32

4.7

Base S 1 has x5 coordinate, c2 = 0 . . . . . . . . . . . . . . . . . . . .

33

4.8

Base S 1 has x5 coordinate, c2 = 1/2 . . . . . . . . . . . . . . . . . .

34

4.9

Base S 1 has x6 coordinate, b1 = 0 . . . . . . . . . . . . . . . . . . . .

34

4.10 Base S 1 has x6 coordinate, b1 = 1/2 . . . . . . . . . . . . . . . . . .

35

4.11 Base S 1 has x7 coordinate, b2 = 0 . . . . . . . . . . . . . . . . . . . .

36

4.12 Base S 1 has x7 coordinate, b2 = 1/2 . . . . . . . . . . . . . . . . . .

36

vi

Chapter 1
Introduction
In this thesis we study Calabi-Yau submanifolds of a family of G2 manifolds constructed by Joyce in [J1, J2]. The method we use to find these Calabi-Yau 3-folds
is given by Akbulut and Salur in [AS2]. We first find a (locally) non-vanishing vector field on the G2 manifold, then the submanifold normal to this vector field has a
Calabi-Yau structure induced from the G2 structure. Akbulut and Salur defines 2
such submanifolds to be a mirror pair if they come from the same G2 structure. They
call this as a mirror duality inside a G2 manifold. This is an interesting concept as we
will see in our examples. As we will note later in the text, in some cases the concept
of mirror duality and mirror symmetry coincides, in other words, some mirror pairs
are actually mirror symmetric.
Another motivation comes from string theory. As far as the author understands,
in string theory electrons and quarks are considered to be 1-dimensional strings rather
than 0-dimensional objects. Physicists study the motion of these strings. In different
theories, they consider the space time to have not only 4 dimensions (space plus time)
but as many as 26 (bosonic case), 10 (superstring theory) or 11 (M-theory) dimensions. To model the extra 6 dimensions in superstring theory they use Calabi-Yau
manifolds, and to model the extra 7 dimensions in M-theory they use G2 manifolds.
1

Therefore, considering Calabi-Yau manifolds as submanifolds of G2 manifolds might
have interesting meanings from the point of view of a physicist. Again, these are only
interpretations of the author who does not know physics.
The outline of this thesis is as follows: In the second chapter we give basic definitions and some well known facts on holonomy and Calabi-Yau manifolds. Also we
introduce a famous type of Calabi-Yau manifold constructed by Borcea and Voisin.
This 3-folds will appear in our examples on the last chapter. In chapter 3 we give the
definition and examples of G2 manifolds, and we explain the construction of Akbulut
and Salur. In the last chapter, we give our results. We consider Joyce manifold of the
first kind and study their Calabi-Yau submanifolds obtained by the method given by
Akbulut and Salur.

2

Chapter 2
Background Material
In this chapter we will recall basic definitions and some examples. In the first section
we introduce the notion of holonomy for Riemannian manifolds and give a classifying
theorem of Berger. For further reading on holonomy groups we refer the reader to
[Bes],[J3] and [P]. In the second section we define Calabi-Yau manifolds. In the
third section we define orbifolds and give some facts on their resolutions. Finally,
on the last section we give an important example to Calabi-Yau manifolds, namely,
Borcea-Voisin 3-folds.

2.1

Holonomy Groups

Let (M, g) be a Riemannian manifold of dimension n and let

be the Levi-Civita

connection. Suppose that γ : [0, 1] −→ M is a smooth curve with γ(0) = x and
γ(1) = y, where x, y ∈ M . Then for each u ∈ Tx M there exists a unique section
s of γ ∗ (T M ) satisfying

γ(t) s(t)
˙

= 0 for each t ∈ [0, 1], with s(0) = u. Define

Pγ : Tx M −→ Ty M by Pγ (u) = s(1). Pγ is a well-defined linear map, called the
parallel transport map. One can generalize this definition to the case when γ is
continuous and piecewise smooth by requiring s to be continuous, and differentiable
3

whenever γ is differentiable.
Definition 2.1.1. The holonomy group Holx (g) based at x is defined to be

Holx (g) = Pγ : γ is a loop based at x ⊆ GL(n)

If M is connected, the holonomy group is independent of the base point because if
−1 =
x, y ∈ M can be connected by a piecewise smooth curve γ in M then Pγ Holx (g)Pγ ∼

Holy (g). Therefore in this case, we can drop the subscripts x and write the holonomy
group as Hol(g). Under certain assumptions on M and g, Berger [Ber] gave a list of
all possible holonomy groups:
Theorem 2.1.2. (Berger) Suppose that (M, g) is a simply-connected Riemannian
manifold of dimension n, and that g is irreducible and non-symmetric, then exactly
one of the following seven cases holds.
i) Hol(g) = SO(n),
ii) n = 2m with m ≥ 2, and Hol(g) = U (m) in SO(2m),
iii) n = 2m with m ≥ 2, and Hol(g) = SU (m) in SO(2m),
iv) n = 4m with m ≥ 2, and Hol(g) = Sp(m) in SO(4m),
v) n = 4m with m ≥ 2, and Hol(g) = Sp(m)Sp(1) in SO(4m),
vi) n = 7 and Hol(g) = G2 in SO(7),
vii) n = 8 and Hol(g) = Spin(7) in SO(8).

2.2

Calabi-Yau Manifolds

There are several inequivalent definitions of Calabi-Yau manifolds in use in the literature. We will use the following as our definition which is equivalent to saying that
4

Hol(g) ⊆ SU (n), which is also equivalent to saying that the canonical bundle of the
manifold is trivial (see [J3]).
Definition 2.2.1. A compact n dimensional K¨hler manifold (N, J, ω) is called a
a
Calabi-Yau manifold if N has a holomorphic n-form Ω that vanishes nowhere.
Example 2.2.2. In dimension one the only examples are tori, T 2 . In dimension two
all Calabi-Yau manifolds are either T 4 or K3 surfaces (compact, complex surface with
h1,0 = 0 and trivial canonical bundle).
Example 2.2.3. Let N be a hypersurface of degree n + 1 in CPn , so
N = {[z0 , z1 , ..., zn ] ∈ CPn : f (z0 , z1 , ..., zn ) = 0}. One can show that (for example
[J3],section 6.7) N is a Calabi-Yau manifold of complex dimension n − 1. This is
perhaps the simplest known method of finding Calabi-Yau manifolds. But all nonsingular hypersurfaces in CPn of degree n + 1 are diffeomorphic, and thus, this method
provides only one smooth manifold in each dimension.
Some of the well-known properties of Calabi-Yau manifolds are given by the following two propositions. For proofs, we refer the reader to [J3].
Proposition 2.2.4. If (N, J, g) is a compact K¨hler manifold of dimension n and
a
Hol(g) is SU (n) or Sp(n/2), then g is Ricci flat and irreducible and N has finite
fundamental group.
Proposition 2.2.5. Let (N, J, g) be a Calabi-Yau manifold of dimension n with
Hol(g) = SU (n) and let hp,q be its Hodge numbers. Then h0,0 = hn,0 = 1 and
hp,0 = 0 for p = 0, n.
Therefore, for Calabi-Yau manifolds of complex dimension 3 with holonomy SU (3),
the Hodge diamond is given by:
5

1
0

0
h1,1

0
h2,1

1

0
h2,1

h1,1

0
0

1
0

0
1

2.3

Orbifolds

In this section we will give definition and some facts on orbifolds. For further reading
refer to Satake [Sat], who calls them V-manifolds, and Joyce [J3].
Definition 2.3.1. An orbifold is a singular real manifold X of dimension n where
singularities are locally isomorphic to quotient singularities Rn /G for finite subgroups
G ⊂ GL(n), such that if 1 = γ ∈ G then the subspace Vγ of Rn fixed by γ has
dim Vγ ≤ n − 2.
Definition 2.3.2. For each singular point x ∈ X in an orbifold X, there is a finite
group Gx ⊂ GL(n), unique up to conjugation, such that an open neighborhood of
x ∈ X is homeomorphic to an open neighborhood of 0 ∈ Rn /Gx . We call x an
orbifold point of X and Gx the orbifold group of x.
Example 2.3.3. If M is a manifold and G is a finite group that acts smoothly on
M , with non-identity fixed point sets of codimension at least two, then M/G is an
orbifold.
The following proposition, taken from [J3], describes the singular set of M/G.
Proposition 2.3.4. Let M be a smooth manifold and G be a finite group acting
smoothly and faithfully on M preserving orientation. Then M/G is an orbifold. For
6

each x ∈ M define the stabilizer subgroup of x to be Stab(x) = {γ ∈ G : γ · x = x}.
If Stab(x) = {1} then xG is a non-singular point of M/G. If Stab(x) = {1} then xG
is a singular point of M/G and has orbifold group Stab(x). Thus the singular set of
M/G is
S = {xG ∈ M/G : x ∈ M and γ · x = x f or some γ ∈ G}
Definition 2.3.5. A complex orbifold is a singular complex manifold of dimension n
whose singularities are locally isomorphic to Cn /G, where G is a finite subgroup of
GL(n, C). The orbifold points and orbifold groups are defined as above.
Recall that a metric g on a complex manifold (M, J) is called a Hermitian metric
if g(u, v) = g(Ju, Jv) for all vectors u, v on M . The corresponding Hermitian form
ω is the 2-form defined by ω(u, v) = g(Ju, v). This form ω is called a K¨hler form if
a
it is closed (dω = 0), and in this case g is called a K¨hler metric and M is called a
a
K¨hler manifold.
a
Definition 2.3.6. We say that g is a K¨hler metric on a complex orbifold (X, J),
a
if g is K¨hler in the usual sense on the non-singular part of X, and wherever X is
a
locally isomorphic to Cn /G, we can identify g with the quotient of a G−invariant
K¨hler metric defined near 0 ∈ Cn . In this case (X, J, g) is called a K¨hler orbifold.
a
a
Many definitions and results about manifolds can be generalized to orbifolds, such
as the definition of K¨hler metrics above. In particular, the ideas of smooth k-forms,
a
(p,q)-forms makes sense, De Rham and Dolbeault cohomology are well defined and
have nearly all their usual properties. If all the orbifold groups of X lie in SL(n, C),
then the canonical bundle KX (nth exterior power of the cotangent bundle) is a
genuine line bundle over X. The singularities of orbifolds may be resolved to obtain
non-singular manifolds. To understand these resolutions we need to understand them
locally first.
7

Definition 2.3.7. A resolution (X, π) of Cn /G is a nonsingular complex manifold
X of dimension n with a proper biholomorphic map π : X → Cn /G that induces
a biholomorphism between dense open sets. We call X a crepant resolution if the
canonical bundles are isomorphic, KX ∼ π ∗ (KCn /G ).
=
Each singularity C2 /G for G a finite subgroup of SU (2) admits a unique crepant
resolution ([M]). For G ⊂ SL(n, C) any finite subgroup, C3 /G admits a crepant
resolution ([R]). For n ≥ 4, C4 /G may or may not admit a crepant resolution.
There is a conjecture from [IR], usually called the McKay Correspondence, which
aims to describe the topology and geometry of crepant resolutions (X, π) of Cn /G in
terms of the group G. We need the following definition to state the conjecture, and
after the conjecture we will give the cases that have been proved.
Definition 2.3.8. Let G ⊂ SL(n, C) be a finite subgroup. Then each γ ∈ G has n
eigenvalues e2πia1 , ..., e2πian , where a1 , ..., an ∈ [0, 1) are uniquely defined up to order.
Define the age of γ to be age(γ) = a1 +...+an . Since det(γ) = 1 = e2πi age(γ) , age(γ)
is an integer between 0 and n − 1.
Conjecture 2.3.9. Let G be a finite subgroup of SL(n, C), and (X, π) a crepant
resolution of Cn /G. Then there exists a basis of H ∗ (X, Q) consisting of algebraic
cycles in 1 − 1 correspondence with conjugacy classes of G, such that conjugacy
classes with age k correspond to basis elements of H 2k (X, Q). In particular, b2k (X)
is the number of conjugacy classes of G with age k, and b2k+1 (X) = 0, so the Euler
characteristic χ(X) is the number of conjugacy classes in G.
The case n = 2 is already known to be true by McKay [M]. Ito and Reid [IR]
proved that the conjecture is true for n = 3, Batyrev and Dais [BD] proved for
arbitrary n when G is abelian, using toric geometry, and also gave their own proof
for n = 3 case.
8

2.4

Borcea-Voisin 3-folds

In this section we will give an important example to Calabi-Yau 3-folds, constructed
by Borcea [Bo] and Voisin [V].
The proofs of the following Lemma and Theorem can be found in [V].
Lemma 2.4.1. If S is a K3 surface with and involution j such that j induces a nontrivial automorphism on H 2,0 (S), then the fixed points of j have several possibilities:
i) no fixed points,
ii) a finite number of rational curves and at most one curve with genus > 0, or
iii) two elliptic curves.
Theorem 2.4.2. Let E be an elliptic curve C/Λ and i the involution induced by
the involution on C, z → −z. Let S be a K3 surface with involution j inducing a
non-trivial automorphism on H 2,0 (S), and let k(e, s) := (i(e), j(s)) be the product
automorphism on E ×S. Then X = (E × S)/k, the minimal resolution of the orbifold
(E × S)/k, is a Calabi-Yau manifold.
Voisin gives a formula for the Hodge numbers of X of this theorem: Let n be the
number of fixed curves of j on S as in the above Lemma (n is possibly 0), and let n
be the total genus of these n fixed curves. Then the Hodge numbers of X are given
by
h1,1 = 11 + 5n − n and h2,1 = 11 + 5n − n.

(2.4.1)

The importance of this construction is that Nikulin’s classification [N] implies that
if (S, j) is a K3 surface with an involution that has n fixed curves with total genus n ,
then there exists a complementary pair (S , j ) with n fixed curves with total genus
n. If we let X to be the manifold obtained by Borcea-Voisin construction on S , then
9

X will have Hodge numbers h1,1 = 11 + 5n − n and h2,1 = 11 + 5n − n . Therefore
manifolds constructed using this method always come in mirror pairs.

10

Chapter 3
G2 Manifolds
3.1

Holonomy Group G2

For the proofs of the facts listed in this section we refer to [AS2], [Br1], [J3].
Definition 3.1.1. Let (x1 , ..., x7 ) be coordinates on R7 . Write dxij...l for the exterior
form dxi ∧ dxj ∧ ... ∧ dxl on R7 . Define a 3-form ϕ0 on R7 by
ϕ0 = dx123 + dx145 + dx167 + dx246 − dx257 − dx347 − dx356

(3.1.1)

The subgroup of GL(7) preserving ϕ0 is the exceptional Lie group G2 .
Theorem 3.1.2. The subgroup G2 ⊆ GL(7) is compact, connected, simply-connected
and of dimension 14. Moreover, G2 acts irreducibly on R7 and transitively on the
spaces of lines in R7 and 2-planes in R7 . Finally, G2 is isomorphic to the group of
algebra automorphisms of octanions.
Definition 3.1.3. A smooth 7-manifold M has a G2 structure if there is a 3-form
ϕ ∈ Ω3 (M ) such that at each x ∈ M the pair (Tx M, ϕ(x)) is isomorphic to (T0 R7 , ϕ0 ).
We call (M, ϕ) a manifold with G2 structure.
11

A G2 structure ϕ on M gives an orientation µ ∈ Ω7 (M ) on M , and µ determines
a metric gϕ =<, > on M by:

< u, v >= (u ϕ ∧ v ϕ ∧ ϕ)/6µ

(3.1.2)

Definition 3.1.4. A manifold with G2 structure (M, ϕ) is called a G2 manifold if
Hol(gϕ ) ⊆ G2 .
Equivalent definitions can be given by the following proposition which follows from
[Sal](section 11.5).
Proposition 3.1.5. Let (M, ϕ) be a 7-manifold with a G2 structure. Then the following are equivalent:
i) Hol(gϕ ) ⊆ G2
ii)

ϕ = 0, where

is the Levi-Civita connection of gϕ

iii) dϕ = d∗ ϕ = 0
iv) dϕ = d(∗ϕ ϕ) = 0
Example 3.1.6. Let (N, ω, Ω) be a Calabi-Yau 3-fold, then N × S 1 has holonomy
SU (3) ⊂ G2 . In this case ϕ = ReΩ + ω ∧ dt. Similarly, N × R is a non-compact G2
manifold.
Example 3.1.7. Let Y be a Riemannian 3-manifold with constant sectional curvature
+1. Bryant and Salamon [BS] gave an explicit metric on the spinor bundle S ∼
=
Y × R4 .
Example 3.1.8. Joyce [J1, J2] gave the first examples of compact 7-manifolds with
holonomy G2 . We will give details to this construction in section 3.4.
12

3.2

Akbulut-Salur Construction

In this section we will describe the method of finding Calabi-Yau submanifolds of G2
manifolds given by Akbulut and Salur [AS2]. Let (M, ϕ) be a 7-manifold with a G2
structure, and let gϕ =<, > be the corresponding metric given by equation (3.1.2).
Definition 3.2.1. Define a cross product structure × on the tangent bundle of M
as follows
ϕ(u, v, w) =< u × v, w >

(3.2.1)

We can also view cross product as a tangent bundle valued 2-form ψ ∈ Ω2 (M, T M )
defined by ψ(u, v) = u × v
Definition 3.2.2. Define the tangent bundle valued 3-form χ ∈ Ω3 (M, T M ) by

< χ(u, v, w), z >= ∗ϕ(u, v, w, z)

(3.2.2)

Definition 3.2.3. Let ξ be a nonvanishing vector field of M . We can define a
symplectic ωξ and a complex structure Jξ on the 6-plane bundle Vξ = ξ ⊥ by

ωξ =< ψ, ξ >= ξ ϕ and Jξ (X) = X × ξ

(3.2.3)

And define a complex valued (3,0) form Ωξ = ReΩξ + i ImΩξ by

ReΩξ = ϕ|V

ξ

and ImΩξ =< χ, ξ >= ξ ∗ ϕ

(3.2.4)

Example 3.2.4. (example of section 3.1 in [AS2]) Consider M = T 7 as G2 manifold
with calibration 3-form ϕ = e123 + e145 + e167 + e246 − e257 − e347 − e356 where
{e1 , ..., e7 } is the basis of T M . If we choose ξ = e7 , then Vξ =< e1 , ..., e6 >, the
symplectic form is ωξ = e16 −e25 −e34 , the complex structure Jξ is e1 → −e6 , e2 → e5 ,
13

e3 → e4 , and the complex valued (3,0) form is Ωξ = (e1 +i e6 )∧(e2 −i e5 )∧(e3 −i e4 );
note that this is just Ωξ = (e1 − i Jξ (e1 )) ∧ (e2 − i Jξ (e2 )) ∧ (e3 − i Jξ (e3 )).

Example 3.2.5. (continued from previous example) If we choose ξ = e3 , then
ˆ
Vξ =< e1 , ..., e3 , ..., e7 >, the symplectic form is ωξ = e12 − e47 − e56 , the complex structure Jξ is e1 → −e2 , e4 → e7 , e5 → e6 , and Ωξ = (e1 + i e2 ) ∧ (e4 − i e7 ) ∧
(e5 − i e6 ) which is just Ωξ = (e1 − i Jξ (e1 )) ∧ (e2 − i Jξ (e2 )) ∧ (e3 − i Jξ (e3 )).

Now let ξ ∈ Ω0 (M, T M ) be a non-vanishing unit vector field, which gives a
codimension one distribution Vξ = ξ ⊥ on M , which is equipped with the structures
Vξ , ωξ , Ωξ , Jξ

as in the definitions above. Let ξ be the dual 1-form of ξ. Let

eξ and ξ denote the exterior and interior product operations respectively. Since
eξ ◦ ξ + ξ ◦ eξ = id we have

ϕ = eξ ◦ ξ (ϕ) + ξ ◦ eξ (ϕ) = ωξ ∧ ξ + ReΩξ

(3.2.5)

Recall that the condition that the distribution Vξ be integrable (the involutive
condition which implies ξ ⊥ comes from a foliation) is given by

dξ ∧ ξ = 0

(3.2.6)

By Thomas [T], even if Vξ is not integrable, it is homotopic to a foliation. Let Xξ
be a page of this foliation, and for simplicity assume that this 6-manifold is smooth.
Now we will give the following two Lemmas with proofs from [AS2].

Lemma 3.2.6. Jξ is compatible with ωξ , and it is metric invariant.
14

Proof. Let u, v ∈ Vξ , then
ωξ Jξ (u) , v

= ωξ (u × ξ, v) = ϕ (u × ξ, v, ξ)
= −ϕ (ξ, ξ × u, v) = − < ξ × (ξ × u) , v >
= − < − |ξ|2 u+ < ξ, u > ξ, v >
= |ξ|2 < u, v > − < ξ, u >< ξ, v >
= < u, v >

The first line comes from the definitions of ωξ and Jξ , the second line comes from the
definition of cross product, and the identity ξ × (ξ × u) = − |ξ|2 u+ < ξ, u > ξ for
the third line is from [Br2]. Hence < Jξ (u) , Jξ (v) >= −ωξ u, Jξ (v) =< u, v >
Lemma 3.2.7. Ωξ is a nonvanishing (3,0) form.
Proof.
¯
(−i/2)Ωξ ∧ Ωξ = ImΩξ ∧ ReΩξ = (ξ ∗ ϕ) ∧ ξ (ξ ∧ ϕ)
= −ξ
= ξ
=

(ξ ∗ ϕ) ∧ (ξ ∧ ϕ)
∗(ξ ∧ ϕ) ∧ (ξ ∧ ϕ)
2

ξ ∧ ϕ ξ vol(M )

= 4 ξ

2

(∗ξ ) = 4 vol(Xξ )

Third line is from the identity ∗(ξ α) = (−1)k+1 (ξ ∧ ∗α) for α ∈ Ωk (M ). And
the identity |ϕ ∧ β|2 = 4 |β|2 for β ∈ Ω1 (M ) is from [Br2].
The following observations is again from [AS2]. One can show that ∗ReΩξ =
−ImΩξ ∧ ξ and ∗ImΩξ = ReΩξ ∧ ξ . And if

is the star operator of Xξ , then

ReΩξ = ImΩξ . Note that ωξ is a symplectic structure on Xξ whenever dϕ = 0
and Lξ (ϕ)|V = 0, where L is the Lie derivative. This comes from ωξ = ξ ϕ and
ξ

dωξ = Lξ (ϕ) − ξ dϕ = Lξ (ϕ). We also have d∗ ϕ = 0 ⇒ d ωξ = 0, since ∗ϕ =
15

ωξ − ImΩξ ∧ ξ , and hence d( ωξ ) = d(∗ϕ|X ) = 0. Also dϕ = 0 ⇒ d(ReΩξ ) =
ξ
d(ϕ|X ) = 0. Furthermore, if d∗ ϕ = 0 and Lξ (∗ϕ)|V = 0 then d(ImΩξ ) = 0, which
ξ
ξ
follows from ImΩξ = ξ (∗ϕ). Also, Jξ is integrable when dΩ = 0 ([H]). Using the
following definition, all these observations sum up to the theorem below.
Definition 3.2.8. (X 6 , ω, Ω, J) is called an almost Calabi-Yau manifold, if X is a
Riemannian manifold with a non-degenerate 2-form ω which is co-closed, and J is
a metric invariant almost complex structure which is compatible with ω, and Ω is a
non-vanishing (3,0) form with ReΩ closed. Furthermore, when ω and ImΩ are closed,
we call this a Calabi-Yau manifold.
Theorem 3.2.9. Let (M, ϕ) be a G2 manifold, and ξ be a unit vector field which
comes from a codimension one foliation on M , then (Xξ , ωξ , Ωξ , Jξ ) is an almost
Calabi-Yau manifold with ϕ|X = ReΩξ and ∗ϕ|X = ωξ . Furthermore, if Lξ (ϕ)|X =
ξ

ξ

ξ

0 then dωξ = 0, and if Lξ (∗ϕ)|X = 0 then Jξ is integrable; when both of these conξ
ditions are satisfied, (Xξ , ωξ , Ωξ , Jξ ) is a Calabi-Yau manifold.

3.3

MirrorDuality

In this section, we will define the concept of mirror duality of Calabi-Yau manifolds
inside a G2 manifold, introduced by Akbulut and Salur in [AS1, AS2].
Definition 3.3.1. Let (M, ϕ) be a manifold with a G2 structure. A 4-dimensional
submanifold X ⊂ M is called a co-associative if ϕ|X = 0. A 3-dimensional submanifold Y ⊂ M is called an associative if ϕ|Y = vol(Y ).
By a theorem of Thomas, all orientable 7-manifolds admit non-vanishing 2-frame
fields [T]. Using this, one obtains an additional structure on the tangent bundle of
G2 manifolds [AS1].
16

Lemma 3.3.2. A non-vanishing oriented 2-plane field Λ on a manifold with G2
structure (M, ϕ) induces a splitting of T M = E⊕V, where E is a bundle of associative
3-planes, and V = E⊥ is a bundle of coassociative 4-planes. The unit sections ξ of
the bundle E → M give complex structures Jξ on V.
Proof. Let Λ = span {u, v} be the 2-plane spanned by the basis vectors of an orthonormal 2-frame {u, v} in M . Then we define E = span {u, v, u × v}, and V = E⊥ . We
can define the complex structure on V by Jξ (x) = x × ξ.
Note that, this complex structure Jξ extends naturally to a complex structure on
Vξ as in section 3.2.
Definition 3.3.3. Two Calabi-Yau manifolds are mirror pairs of each other, if their
complex structures are induced from the same calibration 3-form in a G2 manifold.
Furthermore, we call them strong mirror pairs if their normal vector fields ξ and ξ
are homotopic to each other through non-vanishing vector fields.
Example 3.3.4. If we reconsider the examples 3.2.4 and 3.2.5 with the notions of
this section, we have Λ = span {e1 , e2 } and E = span {e1 , e2 , e3 = e1 × e2 }, V =
span {e4 , e5 , e6 , e7 }. For ξ = e3 and ξ = e7 we obtained two different complex
structures on T6 , which are by definition mirror pairs of each other. In this case the
notion of mirror duality in a G2 manifold and the famous notion of mirror symmetry
(h1,1 (Vξ ) = h2,1 (Vξ ) and h1,1 (Vξ ) = h2,1 (Vξ )) coincides.
Example 3.3.5. In [AES], we considered a Joyce manifold of the first kind, more
specifically J(1/2, 0, 0, 1/2, 1/2), and we found a pair of Borcea-Voisin 3-folds with
Hodge numbers h1,1 = h2,1 = 19 as mirror pairs, which are again mirror symmetric.
This example will be considered in a more general setting in chapter 4 in which we
will see that mirror pairs are not always mirror symmetric.
17

3.4

Joyce Manifolds

Joyce’s construction of 7-manifolds with holonomy G2 is similar to the Kummer
construction for K3 surfaces, and he calls his construction as the generalized Kummer
construction. For more details reader is referred to [J1, J2, J3]. He starts with a flat
Riemannian 7-torus T7 , divide it by the action of a finite group, Γ, of automorphisms
of T7 (not an arbitrary group but one that gives nice singularities), then resolves the
singularities to obtain the 7-manifold M . Then he shows the existence of a G2 metric
on this manifold. Let’s give a little more detail on this construction.
Definition 3.4.1. Let Tn be an n-torus with a flat Riemannian metric. Let Γ be
a finite group of isometries of Tn . Let S be the singular set of Tn /Γ. Let M be a
compact, smooth n-manifold and Φ : M → Tn /Γ be a surjective continuous map that
is smooth except at S. The quadraple (Tn , Γ, M, Φ) is called a generalized Kummer
construction if it has the following properties:
i) Φ is injective on Φ−1 (M − S),
ii) Φ−1 (S) is a finite union of compact submanifolds of M , and
iii) for each s ∈ S, Φ−1 (s) is a connected, simply-connected, finite union of compact
submanifolds of M .
To avoid “bad” types of singularities where two or more singular submanifolds of
S intersect, we put a condition on Γ:
Condition*. Let Γ be a finite group of orientation preserving isometries of Tn .
Suppose that whenever γ1 , γ2 are non-identity elements of Γ that have fixed points
k
in Tn , then either γ1 γ2 has no fixed points in Tn or γ1 = γ2 for some k ∈ Z.

This condition on Γ guarantees the following (Lemma 2.1.3 of [J2]):
18

Lemma 3.4.2. S is a disjoint union of connected components, and each connected
component has a neighborhood isometric to a neighborhood of the singular set in
(Tn−2l × (R2l /Zp ))/F . Here l is a positive integer with 2l ≤ n, Z2 is a nontrivial cyclic subgroup of SO(2l) acting freely on R2l − {0}, and F is a finite group of
isometries of Tn−2l × (R2l /Zp ) that acts freely on Tn−2l .
Definition 3.4.3. Let G be a finite subgroup of SO(n) that acts freely on Rn −{0} (so
n is even, otherwise G has fixed points). An ALE (asymptotically locally Euclidean)
space X is a complete Riemannian manifold with one end modeled on the end of
Rn /G, such that the metric g on X is asymptotic to the Euclidean metric h on Rn /G
in the sense that,
There exists a surjective continuous map φ : X → Rn /G that is smooth in the
appropriate sense, such that φ−1 (0) is a connected, simply-connected, finite union
of compact submanifolds of X and φ induces a diffeomorphism from X − φ−1 (0) to
(Rn − {0})/G such that
φ∗ (g) − h = O(r−4 ), ∂φ∗ (g) = O(r−5 ), ∂ 2 φ∗ (g) = O(r−6 ) for large r,
where r is the distance from the origin in Rn /G and ∂ is the flat connection on Rn /G.
To construct M , Joyce uses Tn−2l × X to desingularize the singular component
modeled on Tn−2l × (R2l /Zp ) where X is an ALE space for the group Zp ⊂ SO(2l).
In his paper, he considers only the cases n = 7 and l = 2 or 3. And he proves the
following (Theorem 2.2.2 of [J2]):
Theorem 3.4.4. Let ϕ be a flat G2 structure on T7 , and let Γ be a finite group of
diffeomorphisms of T7 preserving ϕ. Let S1 , ..., Sk be the connected components of
the singular set S of T7 /Γ. Suppose that for each j = 1, ..., k either
(i) Sj has a neighborhood isometric to a neighborhood of the singular set of (T3 ×
C2 /Gj )/Fj , where T3 is a flat Riemannian torus, Gj a finite subgroup of SU (2), and
Fj is a group of isometries of T3 × C2 /Gj acting freely on T3 . There is an ALE space
19

Xj with holonomy SU (2) asymptotic to C2 /Gj , and an action of Fj on Xj such that
(T3 × Xj )/Fj is asymptotic to (T3 × C2 /Gj )/Fj , or
(ii) Sj has a neighborhood isometric to a neighborhood of the singular set of (S1 ×
C3 /Gj )/Fj , where Gj a finite subgroup of SU (3) acting freely except at 0, and Fj
is a group of isometries of S1 × C3 /Gj acting freely on S1 . There is an ALE space
Xj with holonomy SU (3) asymptotic to C3 /Gj , and an action of Fj on Xj such that
(S1 × Xj )/Fj is asymptotic to (S1 × C3 /Gj )/Fj .
Then there exists a compact 7-manifold M constructed from T7 /Γ and X1 , ..., Xk , a
positive constant θ, and a family {ϕt : t ∈ (0, θ]} of smooth, closed sections of Λ3 M .
+
Let gt be the metric on M associated to ϕt . There exists a family {ψt : t ∈ (0, θ]} of
smooth 3-forms on M with d∗ ψt = d∗ ϕt , where d∗ is defined using gt . There exist
positive constants D1 , ..., D5 independent of t, such that the following five conditions
hold for each t ∈ (0, θ], where all norms are calculated using gt :
i) ψt 2 ≤ D1 t4 and ψt C 2 ≤ D1 t4
ii) the injectivity radius δ(gt ) satisfies δ(gt ) ≥ D2 t
iii) the Riemannian curvature R(gt ) of gt satisfies R(gt ) C 0 ≤ D3 t−2
iv) the volume vol(M ) satisfies vol(M ) ≥ D4 and
v) the diameter diam(M ) satisfies diam(M ) ≤ D5
Then Joyce shows that on this manifold M , for small t one can deform ϕt (by
adding dηt where ηt is a small 2-form) to obtain a family of torsion free G2 structures
ϕt , and he proves the following (Theorem 2.2.3 of [J2]):
˜
Theorem 3.4.5. M of the previous theorem admits a smooth family of torsion free
G2 structures of dimension b3 (M ).

20

Chapter 4
Calabi-Yau Submanifolds of Joyce
Manifolds
4.1

Joyce Manifolds of First Kind: J(b1, b2, c1, c2, c3)

In this section we will consider a special family of Joyce manifolds, which is usually
referred as Joyce manifolds of the first kind. Let (x1 , ..., x7 ) be coordinates on T7 =
R7 /Z7 where xi ∈ R/Z. Define a section ϕ of Λ3 T7 by:
+
ϕ = dx1 ∧ dx2 ∧ dx3 + dx1 ∧ dx4 ∧ dx5 + dx1 ∧ dx6 ∧ dx7 + dx2 ∧ dx4 ∧ dx6
−dx2 ∧ dx5 ∧ dx7 − dx3 ∧ dx4 ∧ dx7 − dx3 ∧ dx5 ∧ dx6
Let Γ =< α, β, γ >∼ Z3 defined by:
= 2
x1

x2

x3

x4

x5

x6

x7

α

x1

x2

x3

−x4

−x5

−x6

−x7

β

x1

−x2

−x3

x4

x5

b1 − x 6

b2 − x 7

γ

−x1

x2

c1 − x 3

x4

c2 − x 5

x6

c3 − x 7

Table 4.1: The action of generators of Γ on T7
21

where bi , ci ∈ {0, 1/2}.
Clearly α, β, γ preserves ϕ and α2 = β 2 = γ 2 = 1, αβ = βα, αγ = γα, βγ = γβ.
For any choice of bi ’s and ci ’s, each of α, β, γ fixes 16 copies of T3 :

F ix(α) = {(x1 , ..., x7 ) ∈ T7 : x4 , x5 , x6 , x7 ∈ {0, 1/2}}
F ix(β) = {(x1 , ..., x7 ) ∈ T7 : x2 , x3 ∈ {0, 1/2}, x6 ∈ {b1 /2, (1 + b1 )/2},
x7 ∈ {b2 /2, (1 + b2 )/2}}
F ix(γ) = {(x1 , ..., x7 ) ∈ T7 : x1 ∈ {0, 1/2}, x3 ∈ {c1 /2, (1 + c1 )/2},
x5 ∈ {c2 /2, (1 + c2 )/2}, x7 ∈ {c3 /2, (1 + c3 )/2}}
We need to put extra conditions on bi ’s and ci ’s so that Γ satisfies condition *. In
other words, we don’t want αβ, βγ, γα, αβγ to have any fixed points.
There are only 13 possible 5-tuples (b1 , b2 , c1 , c2 , c3 ) satisfying this condition as
explained below.
x1

x2

x3

x4

x5

x6

x7

αβ

x1

−x2

−x3

−x4

−x5

b1 + x 6

b 2 + x7

βγ

−x1

−x2

c1 + x 3

x4

c2 − x 5

b1 − x 6

b2 + c 3 + x 7

γα

−x1

x2

c1 − x 3

−x4

c2 + x 5

−x6

c3 + x 7

αβγ

−x1

−x2

c1 + x 3

−x4

c2 + x 5

b1 + x 6

b2 − x 7

Table 4.2: The action of mixed terms of Γ

i) If b1 = b2 = 0 then αβ will have fixed points,
ii) If c1 = 0 and b2 = c3 then βγ will have fixed points,
iii) If c2 = c3 = 0 then γα will have fixed points,
iv) If c1 = c2 = b1 = 0 then αβγ will have fixed points.
22

We need to choose bi ’s and ci ’s so that they don’t satisfy any of the above conditions i),...,iv). Condition iv) is unnecessary since if c1 = c2 = b1 = 0 then b2 = 1/2
from condition i) and hence c3 = 0 from condition ii) which is impossible as long as
we take care of condition iii). Now we will find all possible 5-tuples:
b1 = b2 = 1/2

=⇒

c1 = 1/2 or c3 = 0 (by ii)

=⇒

(c1 , c2 , c3 ) = (1/2, 1/2, 0) or (1/2, 0, 1/2)
or (1/2, 1/2, 1/2) or (0, 1/2, 0) (by iii)

b1 = 0, b2 = 1/2

=⇒

c1 = 1/2 or c3 = 0 (by ii)

=⇒

(c1 , c2 , c3 ) = (1/2, 1/2, 0) or (1/2, 0, 1/2)
or (1/2, 1/2, 1/2) or (0, 1/2, 0) (by iii)

b1 = 1/2, b2 = 0

=⇒

c1 = 1/2 or c3 = 1/2 (by ii)

=⇒

(c1 , c2 , c3 ) = (1/2, 1/2, 0) or (1/2, 0, 1/2)
or (1/2, 1/2, 1/2) or (0, 0, 1/2) or (0, 1/2, 1/2) (by iii)

Therefore the only posibilities are:
(1/2, 1/2, 1/2, 1/2, 0), (1/2, 1/2, 1/2, 0, 1/2), (1/2, 1/2, 1/2, 1/2, 1/2)
(1/2, 1/2, 0, 1/2, 0),

(0, 1/2, 1/2, 1/2, 0),

(0, 1/2, 1/2, 1/2, 1/2), (0, 1/2, 0, 1/2, 0),
(1/2, 0, 1/2, 0, 1/2),

(0, 1/2, 1/2, 0, 1/2)
(1/2, 0, 1/2, 1/2, 0)

(1/2, 0, 1/2, 1/2, 1/2), (1/2, 0, 0, 0, 1/2)

(1/2, 0, 0, 1/2, 1/2)

For the 13 cases that satisfy condition *, let’s see how (the neighborhood of) the
singular set looks like. Since Γ is abelian, α will preserve (setwise) F ix(β) and F ix(γ).

x ∈ F ix(β) =⇒ βα(x) = αβ(x) = α(x) =⇒ α(x) ∈ F ix(β)
x ∈ F ix(γ) =⇒ γα(x) = αγ(x) = α(x) =⇒ α(x) ∈ F ix(γ)
23

Similarly β will preserve F ix(α) and F ix(γ), and γ will preserve F ix(α) and
F ix(β).
Consider the 16 T3 fixed by α. Their neighborhoods in T7 / < α > look like
T3 × C2 /{±1}. As b1 and b2 are not 0 at the same time, the action of β on at least
one of x6 or x7 is x ↔ 1/2 − x. And the fixed T3 ’s of α have x6 , x7 ∈ {0, 1/2}. Hence
β pairs 8 of these T3 ’s with the remaining 8. As c2 and c3 are not 0 at the same time,
the action of γ on at least one of x5 or x7 is x ↔ 1/2 − x. And the fixed T3 ’s of α
have x5 , x7 ∈ {0, 1/2}. Hence γ pairs 8 of these T3 ’s with the remaining 8. So the
contribution to the singular set of T7 /Γ coming from F ix(α) (with neighborhoods) is
either 4 copies of T3 × C2 /{±1} or 8 copies of (T3 × C2 /{±1})/ < βγ >. The latter
is the case b1 = 0, b2 = 1/2, c2 = 0, c3 = 1/2, when β and γ both changes only x7
on F ix(α). Note that in this case since b2 = c3 we must have c1 = 1/2.
Similar observations can be made on F ix(β) and F ix(γ). Consider the neighborhoods T3 × C2 /{±1} of the 16 fixed T3 ’s in T7 / < β >, the C2 part coming from
coordinates x2 , x3 , x6 , x7 . The action of α on x6 and x7 is x ↔ −x. Since b1 and b2
are not 0 at the same time, α will change at least one of x6 or x7 on F ix(β). The
action of γ on x3 is x ↔ c1 − x and on x7 it is x ↔ c3 − x. So γ changes x3 on
F ix(β) if and only if c1 = 1/2 and changes x7 on F ix(β) if and only if b2 + c3 = 1/2.
Hence the contribution from F ix(β) is either 4 copies of T3 × C2 /{±1} or 8 copies
of (T3 × C2 /{±1})/ < γα >. The latter is the case b1 = 0, b2 = 1/2, c1 = 0, c3 = 0,
when α and γ both changes only x7 on F ix(β). Note that in this case since c3 = 0
we must have c2 = 1/2.
Finally, consider the neighborhoods T3 × C2 /{±1} of the 16 fixed T3 ’s in T7 / <
γ >, the C2 part coming from coordinates x1 , x3 , x5 , x7 . The action of α on x5 and
x7 is x ↔ −x. Since c2 and c3 are not 0 at the same time, α will change at least one
of x5 or x7 on F ix(γ). The action of β on x3 is x ↔ −x and on x7 it is x ↔ b2 − x.
So β changes x3 on F ix(γ) if and only if c1 = 1/2 and changes x7 on F ix(γ) if
24

and only if b2 + c3 = 1/2. Hence the contribution from F ix(γ) is either 4 copies of
T3 × C2 /{±1} or 8 copies of (T3 × C2 /{±1})/ < γα >. The latter is the case c2 = 0,
c3 = 1/2, c1 = 0, b2 = 0, when α and β both changes only x7 on F ix(γ). Note that
in this case since b2 = 0 we must have b1 = 1/2.
Therefore, the contributions to singular set (with neighborhoods) for each case is
given by:
case

from F ix(α)

from F ix(β)

from F ix(γ)

(0, 1/2, 0, 1/2, 0)

4T3 × C2 /±

8(T3 × C2 /±)/Z2

4T3 × C2 /±

4T3 × C2 /±

4T3 × C2 /±

(0, 1/2, 1/2, 0, 1/2) 8(T3 × C2 /±)/Z2
(1/2, 0, 0, 0, 1/2)

4T3 × C2 /±

4T3 × C2 /±

8(T3 × C2 /±)/Z2

other 10

4T3 × C2 /±

4T3 × C2 /±

4T3 × C2 /±

Table 4.3: Fixed point sets for all possible 5-tuples
To obtain the smooth 7-manifold underlying the Joyce manifold, Joyce uses EguchiHanson space (we will call it X) which is a complete hyperk¨hler metric on T ∗ CP 1 .
a
So he replaces the neighborhoods (of the singular sets) of the form T3 × C2 /{±1}
by T3 × X and (T3 × C2 /{±1})/Z2 by (T3 × X)/Z2 . The former resolution is the
unique crepant resolution, yet the latter can be made in topologically two different
ways: depending on the induced action of Z2 on [CP 1 ] ∈ H 2 (T ∗ CP 1 ), which can be
chosen as either id or −id.

4.2

Calabi-Yau Submanifolds

In this section we describe Calabi-Yau submanifolds of Joyce manifolds of the first
kind that are obtained by the construction of Akbulut-Salur that we mentioned in the
previous chapter, by choosing ξ to be the direction corresponding to each coordinate
of T7 separately.
In the following diagrams, for the left diagram T7 = S 1 × T6 where S 1 is the base
25

T6 is the fiber, empty circles are fixed T3 ’s of α, β or γ, filled circles are T2 ’s (parts of
fixed T3 ’s on fibers). The right hand diagram is T7 /Γ as a fibration over the interval
[0, 1/2] or [0, 1/4] depending on what we obtain as S 1 /Γ (image of the base of left
hand diagram) with generic fiber shown. In each case, the singular set of this generic
fiber is a number of copies of T2 × C2 /{±1} which will be resolved by replacing it
with T2 × X where X is the Eguchi Hanson space. Fixed sets in each graph from up
to down comes from α, β and γ in this order. Numbers near the fixed sets gives the
number of fixed T3 ’s or T2 ’s.

4.2.1

x1 direction

16

8

16

8

8

8
3/4
1/2

0

0

1/2

1/4
Figure 4.1: Base S 1 has x1 coordinate
In this case (figure 4.1), the base is S 1 corresponding to x1 coordinate. Fixed T3 ’s
of α and β have x1 component, therefore they are drawn as dotted circles on the left
hand graph where these circles meet each fiber at a T2 . Fixed T3 ’s of γ lies on the
two fibers x1 = 0 and x1 = 1/2. The action of γ on x1 is x1 ↔ −x1 so it will pair the
fibers as shown in the graph, fixing the fibers over x1 = 0 and x1 = 1/2. Therefore,
26

after taking the quotient we obtain a fibration over the interval [0, 1/2] (right hand
diagram in figure 4.1) with fibers over (0, 1/2) equal to T6 / < α, β >, whose fixed set
is now 16 copies of T2 , 8 from α and 8 from β, since α pairs the fixed set of β and β
pairs the fixed set of α.
x2

x3

x4

x5

x6

x7

α

x2

x3

−x4

−x5

−x6

−x7

β

−x2

−x3

x4

x5

b1 − x 6

b 2 − x7

Table 4.4: The action of < α, β > on T6
Let N be the corresponding resolution of T6 / < α, β > in M . We can think of the
resolution that gives N in two steps. First, resolve T6 / < α >= T2 × T4 /Z2 , where
Z2 action on T4 (with coordinates x4 , x5 , x6 , x7 ) is {±1}.

(0, 0)

( 1 , 0) ( 1 , 1 )
2
2 2

(0, 1 )
2

Figure 4.2: The action of α on T4
Figure 4.2 explains the Z2 action of α on T4 . We think T4 as T2 × T2 , base T2
27

corresponding to the coordinates x6 , x7 and fiber T2 corresponding to the coordinates
x4 , x5 (as in the first diagram of figure 4.2). α will fold the base making it a pillow
(sphere with 4 singular points), fibers over non-singular points will be T2 ’s and fibers
over singular points will be T2 /Z2 , which are again pillows (as in the diagrams on the
right in figure 4.2). 16 fixed points (4 corners of pillow fibers above 4 corners of the
base pillow) of this action on T4 will be replaced by Eguchi-Hanson spaces X (blow
up of C2 at the origin), which gives a K3 surface. Hence after resolving T6 / < α >
we obtain T2 × K3.
The action of β can be extended trivially on T2 × K3. β will act by {±1} on T2
and by a holomorphic involution that acts by −1 on its holomorphic two form lifted
from dz1 ∧ dz2 on T4 (z1 = x4 + ix5 , z2 = x6 + ix7 ). The fixed set of the action of β
on K3 consists of two copies of T2 ’s. These tori, in the last diagram of figure 4.2, are
two regular fibers at two points on one of the belt circles (dotted circle on the base
pillow). For example, if (b1 , b2 ) = (1/2, 1/2) then the fibers are at the points that are
the images of (x6 , x7 ) = (1/4, 1/4), (1/4, 3/4).
Therefore, N is a Borcea-Voisin 3-fold as we described in section 2.4. Fixed points
of the involution on K3 is two copies of T2 ’s as in part iii) of lemma 2.4.1, so the
formula 2.4.1 with n = n = 2 gives the Hodge numbers of N as h1,1 = h2,1 = 19.

4.2.2

x2 direction

In this case (figure 4.3), the base is S 1 corresponding to x2 coordinate. The generic
fiber of T7 /Γ is T6 / < α, γ > with the following action:
x1

x3

x4

x5

x6

x7

α

x1

x3

−x4

−x5

−x6

−x7

γ

−x1

c1 − x 3

x4

c2 − x 5

x6

c3 − x 7

Table 4.5: The action of < α, γ > on T6
28

16

8
8

8
16

8
3/4
1/2

0

0

1/2

1/4
Figure 4.3: Base S 1 has x2 coordinate
Note that, all of the analysis of the previous section can be done here, by changing
the roles of β and γ. And we obtain the same manifold N , a Borcea-Voisin 3-fold
with Hodge numbers h1,1 = h2,1 = 19.

4.2.3

x3 direction

We will divide this case into two subcases depending on the choice of the number c1 ,
because it will decide the position of the fixed T3 ’s of γ.
If c1 = 0
In this case (figure 4.4), the base is S 1 corresponding to x3 coordinate. Both β and γ
changes x3 coordinate but βγ fixes it. Therefore, the generic fiber is T6 / < α, βγ >,
with fixed point set being 8 copies of T2 ’s coming from the fixed points of α. The
action is given by the following table (recall that from section 4.1 if c1 = 0 then
b2 = c3 ):
Let N be the resulting resolution of T6 / < α, βγ >. As in section 4.2.1, we will
29

16

8

8

8

8

8
3/4
1/2

0

0

1/2

1/4
Figure 4.4: Base S 1 has x3 coordinate, c1 = 0
x1

x2

x4

x5

x6

x7

α

x1

x2

−x4

−x5

−x6

−x7

βγ

−x1

−x2

x4

c2 − x 5

b1 − x 6

1/2 + x7

Table 4.6: The action of < α, βγ > on T6
consider this resolution in two steps. First resolve T2 × T4 / < α > as before to obtain
T2 × K3. The action of βγ on T2 is {±1} and on K3 (after we lift the action to K3)
is an involution which acts by −1 on its holomorphic two form lifted from dz1 ∧ dz2
on T4 (this time z1 = x5 + ix6 , z2 = x4 + ix7 ). But the difference in this case is, the
action of βγ on K3 has no fixed points, since βγ(x7 ) = 1/2 + x7 .
Therefore, N is a Borcea-Voisin 3-fold with Hodge numbers h1,1 = h2,1 = 11,
which are obtained from the formula 2.4.1 with n = n = 0.
If c1 = 1/2
In this case (figure 4.5), fixed T3 ’s of β will be in fibers at x3 = 0, 1/2 and fixed T3 ’s
of γ will be in fibers at x3 = 1/4, 3/4. β will fold the base S 1 fixing the points 0
30

16

16
8

8
8
3/4

8
1/2

0

0

1/4

1/4
Figure 4.5: Base S 1 has x3 coordinate, c1 = 1/2

and 1/2, while γ will fold the base in the perpendicular direction, fixing 1/4 and 3/4.
The resulting orbifold T7 /Γ is then will be a fibration over the interval [0,1/4] with
generic fiber T6 / < α >, which is T2 × T4 / {±1}. The resolution N of T6 / < α > is
therefore, N = T2 × K3. Using the Hodge diamonds of T2 and K3:

1
1
1

0
1 and 1

1

0
20

0

1
0

1

and using the K¨nneth formula, we obtain the Hodge numbers of N as h1,1 =
u
h2,1 = 21.
31

8

8

8

16
16

8
3/4
1/2

0

0

1/2

1/4
Figure 4.6: Base S 1 has x4 coordinate

4.2.4

x4 direction

In this case (figure 4.6), the base is S 1 corresponding to x4 coordinate. This case is
similar to x1 direction. The generic fiber is T6 / < β, γ >, with fixed point set being
8 copies of T2 ’s coming from the fixed points of β and 8 copies coming from the fixed
points of γ. The resolution N of T6 / < β, γ > is again a Borcea-Voisin 3-fold with
Hodge numbers h1,1 = h2,1 = 19.

4.2.5

x5 direction

We have two subcases depending on the choice of c2 .

If c2 = 0
In this case (figure 4.7), generic fiber of T7 /Γ is T6 / < β, αγ > similar to the c1 = 0
case of the x3 direction. T6 / < β >= T2 × T4 / {±1} where T4 has coordinates
x2 , x3 , x6 , x7 . After we obtain T2 × K3, the action of αγ on K3 will have no fixed
32

8

8

8

16
8

8
3/4
1/2

0

0

1/2

1/4
Figure 4.7: Base S 1 has x5 coordinate, c2 = 0
points, because from section 4.1; c2 and c3 can not be 0 at the same time, therefore
αγ will change x7 coordinate. If we call N to be the resolution of T6 / < β, αγ >,
then N is a Borcea-Voisin 3-fold with Hodge numbers h1,1 = h2,1 = 11.
If c2 = 1/2
This case (figure 4.8) is similar to c1 = 1/2 case of x3 direction. Generic fiber before
resolution is T6 / < β >= T2 ×T4 / {±1}. And after resolution we obtain N = T2 ×K3
with Hodge numbers h1,1 = h2,1 = 21.

4.2.6

x6 direction

We have two subcases depending on the choice of b1 .
If b1 = 0
The generic fiber before resolution is T6 / < γ, αβ > (figure 4.9). Similar to c1 = 0
case of x3 direction, we first divide by the action of γ and resolve the resulting orbifold
33

8

8

16

16
8
3/4

8

0

1/2

1/4

0

1/4
Figure 4.8: Base S 1 has x5 coordinate, c2 = 1/2

8

8

8

8
8

16
3/4
1/2

0

0

1/2

1/4
Figure 4.9: Base S 1 has x6 coordinate, b1 = 0
to obtain T2 ×K3. The action of αβ on K3 has no fixed points since αβ(x7 ) = 1/2+x7
(b1 and b2 can not be 0 at the same time). Therefore we obtain a Borcea-Voisin 3-fold
N with Hodge numbers h1,1 = h2,1 = 11.
34

If b1 = 1/2

8

8
8
8
16

16
3/4
1/2

0

0

1/4

1/4
Figure 4.10: Base S 1 has x6 coordinate, b1 = 1/2
In this case (figure 4.10) generic fiber before resolution is T6 / < γ >. Similar to
the case c1 = 1/2 in of x3 direction, we obtain N = T2 × K3 with Hodge numbers
h1,1 = h2,1 = 21.

4.2.7

x7 direction

We have two cases: either (b2 , c3 ) = (0, 1/2) or (b2 , c3 ) = (1/2, 0). In both cases
(see figures 4.11 and 4.12) The fixed T3 ’s of α, β and γ lie on the fibers at x7 =
0, 1/4, 1/2, 3/4. Therefore the generic fiber of T7 /Γ will be T6 / < αβ >= T6 and
T6 / < αγ >= T6 respectively. So N = T6 and has Hodge numbers h1,1 = h2,1 = 9.

35

8

8

8

8
8
3/4
0

8

1/2

0

1/4

1/4
Figure 4.11: Base S 1 has x7 coordinate, b2 = 0

8

8

8
8

8
3/4
0

8

1/2

0

1/4

1/4
Figure 4.12: Base S 1 has x7 coordinate, b2 = 1/2

4.3

Summary of results

To sum up the results so far, as we mentioned in example 3.3.5 this work has covered
the example of our joint work with Akbulut and Salur [AES], in which we obtained
36

a pair of Borcea-Voisin manifolds with Hodge numbers h1,1 = h2,1 = 19, as special
cases of sections 4.2.1 and 4.2.4. In addition, we have obtained another Borcea-Voisin
manifold with Hodge numbers h1,1 = h2,1 = 11, as well as Calabi-Yau manifolds
T2 ×K3 and T6 as submanifolds of Joyce manifolds of the first kind. For example, if we
consider J(1/2, 0, 0, 1/2, 1/2), x1 , x2 , and x4 directions give Borcea-Voisin manifolds
with Hodge numbers (19, 19), x3 direction gives a Borcea-Voisin manifold with Hodge
numbers (11, 11), x5 and x6 directions give T2 × K3, and x7 direction gives T6 . So
all of these Calabi-Yau 3-folds are mirror pairs as in the definition 3.3.3, not all pairs
are mirror symmetric, but each 3-fold is self-mirror.

37

BIBLIOGRAPHY

38

BIBLIOGRAPHY

[AES]

S. Akbulut, B. Efe, S. Salur, Mirror Duality in a Joyce Manifold, Advances in Math., vol. 223, Issue 2 (2010), 444-453

[AS1]

S. Akbulut, S. Salur, Deformations in G2 manifolds, Advances in
Math., vol 217, Issue 5 (2008) 2130-2140.

[AS2]

S. Akbulut, S. Salur, Mirror duality via G2 and Spin(7) manifolds,
Arithmetic and Geometry Around Quantization, Progress in Math.,
vol 279 (2010) 1-21.

[BD]

V.V. Batyrev and D.I. Dais, Strong McKay correspondence, stringtheoretic Hodge numbers and mirror symmetry, Topology, 35:901929,1996.

[Ber]

M. Berger, Sur les groupes d’holonomie homog`ne des vari´t´s a cone
ee `
nexion affine et des vari´t´s riemanniennes, Bull. Soc. Math. France,
ee
83 (1955), 225-238.

[Bes]

A.L. Besse, Einstein Manifolds, Springer-Verlag, New York, 1987.

[Bo]

C. Borcea, K3 surfaces with involution and Mirror pairs of Calabi-Yau
manifolds, AMS/IP Studies in Advances Math 1 (1997), 717-743.

[Br1]

R.L. Bryant, Metrics with exceptional holonomy, Ann. of Math. 126
(1987), 525-576.

[Br2]

R.L. Bryant, Some remarks on G2 structures, Proceedings of G¨kova
o
Geometry-Topology Conference (GGT) (2005), 75109.

[BS]

R.L. Bryant and S.M. Salamon, On the construction of some complete
metrics with exceptional holonomy, Duke Math. J., 58 (1989), 829-850.

[H]

N.J. Hitchin, The moduli space of special Lagrangian submanifolds,
Ann. Scuola Norm. Sup. Pisa Cl. Sci. (4) 25 (1997), no. 3-4, 503515
(1998).
39

[IR]

Y. Ito and M. Reid, The McKay correspondence for finite subgroups of
SL(n, C), in Higher Dimensional Complex Varieties (Trento, 1994), M.
Andreatta et al., editors, 221-240, de Gruyter (1996).

[J1]

D.D. Joyce, Compact Riemannian 7-manifolds with holonomy G2 , I,
Journal of Differential Geometry 43 (1996), 291-328.

[J2]

D.D. Joyce, Compact Riemannian 7-manifolds with holonomy G2 , II,
Journal of Differential Geometry 43 (1996), 329-375.

[J3]

D.D. Joyce, Compact manifolds with Special Holonomy, OUP, Oxford,
2000.

[M]

J. McKay, Graphs, singularities and finite groups, Proc. Symp. Pure
Math. 37 (1980), 183-186.

[N]

V.V. Nikulin, Discrete Reflection groups in Lobachevsky Spaces and
Algebraic Surfaces, Proc. Int. Congr. Math., Berkeley, (1986), 654.

[P]

P. Petersen, Riemannian Geometry, Second Edition, Springer, 2006.

[R]

S.S. Roan, Minimal Resolution of Gorenstein Orbifolds, Topology, 35
(1996), 489-508.

[Sal]

S.M. Salamon, Riemannian Geometry and Holonomy groups, Pitman
Research Notes in Math., vol 201, Longman, Harlow, 1989.

[Sat]

I. Satake, On a generalization of notion of manifold, Proceeding of the
National Academy of Sciences of the U.S.A. 42 (1956), 359-363

[T]

E. Thomas, Postnikov invariants and higher order cohomology operations, Ann. of Math. vol 85 (1967), 184-217.

[V]

C. Voisin, Miroirs et involutions sur les surfaces K3, Asterisque 218
(1993), 273-323.

40