SlideShare uma empresa Scribd logo
1 de 94
Baixar para ler offline
UNIVERSITƀ DEGLI STUDI DI TORINO
DIPARTIMENTO DI MATEMATICA GIUSEPPE PEANO
SCUOLA DI SCIENZE DELLA NATURA
Corso di Laurea Magistrale in Matematica
Master thesis
Calabi-Yau manifolds
Supervisor: Prof.ssa Anna Fino
Cosupervisor: Prof. Joel Fine Candidate: Giulia Marcaccio
Academic Year 2015-2016
Contents
Introduction iii
1 Complex manifolds 1
1.1 Complex coordinates . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Tangent space . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Complex forms: the (p,q)-forms . . . . . . . . . . . . . . . . . 4
1.3.1 1-forms . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Higher degree forms . . . . . . . . . . . . . . . . . . . 5
1.4 Canonical bundle . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.1 Line bundles over M . . . . . . . . . . . . . . . . . . . 10
1.5 Cohomologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5.1 The de Rham cohomology . . . . . . . . . . . . . . . . 13
1.5.2 Dolbeault cohomology . . . . . . . . . . . . . . . . . . 15
2 Some Algebraic Geometry tools 17
2.1 Aļ¬ƒne and projective variety . . . . . . . . . . . . . . . . . . . 17
2.2 Regular and rational functions . . . . . . . . . . . . . . . . . 18
2.3 Divisors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Rational diļ¬€erential forms and canonical divisors. . . . . . . . 22
2.5 Dualizing sheaf . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3 A look to some complex constructions 25
3.1 Almost complex structure . . . . . . . . . . . . . . . . . . . . 25
3.2 Symplectic manifolds . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 (Real) Symplectic manifolds . . . . . . . . . . . . . . . 28
3.2.2 (Complex) Symplectic manifolds . . . . . . . . . . . . 29
3.3 Compatible almost complex structures . . . . . . . . . . . . . 29
4 Calaby-Yau manifolds 32
4.1 Symplectic Calabi-Yau manifolds . . . . . . . . . . . . . . . . 32
4.2 Complex Calabi-Yau manifolds . . . . . . . . . . . . . . . . . 33
4.2.1 1-dimensional Calabi-Yau . . . . . . . . . . . . . . . . 33
4.2.2 2-dimensional Calabi-Yau . . . . . . . . . . . . . . . . 43
i
CONTENTS
4.2.3 3-dimensional and higher-dimensional Calabi-Yau man-
ifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 KƤhler manifolds . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.1 KƤhler and Calabi-Yau manifolds . . . . . . . . . . . . 63
4.4 Toric geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.4.1 Cones and fans . . . . . . . . . . . . . . . . . . . . . . 66
4.4.2 Toric divisors . . . . . . . . . . . . . . . . . . . . . . . 69
4.4.3 Toric Calabiā€“Yau threefolds . . . . . . . . . . . . . . . 70
5 Counterexamples 74
5.1 The Kodaira-Thurston example . . . . . . . . . . . . . . . . . 74
5.2 Hopf surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Symplectic but not complex . . . . . . . . . . . . . . . . . . . 79
6 Calabiā€“Yau Manifolds and String Theory 81
6.1 Compactiļ¬cation . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.1.1 Dimensional reduction . . . . . . . . . . . . . . . . . 82
ii
Introduction
Superstring theory is an attempt to explain all of the particles and funda-
mental forces of nature in one theory by modelling them as vibrations of
tiny supersymmetric strings.
Our physical space is observed to have three large spatial dimensions and,
along with time, is a boundless four-dimensional continuum, known as space-
time.
The problem is that if we want to apply superstring theory to our spacetime
universe we need it to be ten-dimensional. The discrepancy between the
critical dimension d = 10, required by string theory, and the number of
observed dimensions d = 4 is resolved by the idea of compactiļ¬cation.
Looking for a ten-dimensional manifold consistent with the requirements
imposed by this theory, the simplest possibility is to have a space time that
takes the product form M4 Ɨ N6, where M4 is a four-dimensional Minkowski
space and N6 is some compact six-dimensional manifold. This compact man-
ifold is usually taken to have a suļ¬ƒciently small size as to be unobservable
with present technology, and thus we would only see the four-dimensional
manifold M4. Requiring that supersymmetry is preserved at the compactiļ¬-
cation scale restricts us to a special class of manifolds known as Calabi-Yau
manifolds.
Eugenio Calabi, an American mathematician of Italian origins, in 1953
proposed that certain speciļ¬c geometric structures are allowed under some
topological condition.
In particular he surmised that whether a certain kind of complex manifold,
namely the compact (ļ¬nite in extent) and "KƤhler" ones, could satisfy the
general topological conditions of vanishing ļ¬rst Chern class and could also
satisfy the geometrical condition of having a Ricci-ļ¬‚at metric (excluding the
ļ¬‚at torus).
In 1976 the Chinese professor of mathematics at Harvard, Shing-Tung Yau,
proved the existence of the geometric structure as surmised by the Calabiā€™s
conjecture. He was able to prove, by expressing the conjecture in terms of non
linear partial diļ¬€erential equations, the existence of many multi-dimensional
shapes that are Ricci-ļ¬‚at, i.e., satisfying the Einstein equation in 3 complex
dimensions and empty space.
iii
Chapter 0. Introduction
Such structure is now called Calabi-Yau manifold.
Its special properties are indispensable for compactiļ¬cation in Superstring
Theory.
This thesis provides an introduction to Calabi-Yau manifolds. We will
not go into details of the physical problem of superstring theory but we will
point out the mathematical aspect.
The outline is as follows. In Chapter 1 and 2 we prepare the ground to
understand and built the matter that will be later developed: respectively
we give a short review of complex geometry and we analyse some algebraic
geometry tools. More speciļ¬cally on the former chapter we describe the
analogue of the diļ¬€erential manifold on the complex world: we give the
holomorphic deļ¬nition of tangent and cotangent bundle leading us to the
one of canonical bundle; while on the latter, following [Smi], we study the
concept of divisor and sheaf that will be very useful in order to analyse some
examples of Calabi-Yau.
Chapter 3 follows up the topic developed in the ļ¬rst chapter, moreover it
tries to connect the symplectic real world with the complex one, up to the
deļ¬nition of symplectic complex manifold.
In Chapter 4 we come to the deļ¬nition of Calaby-Yau manifold and we discuss
several example focusing on some algebraic geometryā€™s ones. Following
[Shu12] and [Raa10] we begin showing how a complex one dimensional torus
(ļ¬rst example of one-dimensional Calabi-Yau) can be seen as an elliptic curve.
In order to present two-dimensional examples of Calabi-Yau we refer to
[Mor88] and [Kut10]. Concerning the higher dimension we set the problem
on weighted projective space Pn(w) and we give an algeabric condition on
a variety to be Calabi-Yau. Furthermore we give the deļ¬nition of KƤhler
manifold in order to see how they are linked with the Calabi-Yau structures.
Pursuing the aim of understanding the links between the complex world and
the symplectic one, some famous counterexamples are showed in Chapter 5.
Finally the dissertation ends with a view on compactiļ¬cations: the last
chapter is an adaptation of [FT] and provides a glance on the fascinating
dimension that we could not detect.
iv
Chapter 1
Complex manifolds
This chapter is concerned with the theory of complex manifolds. Apparently,
their deļ¬nition is identical to the one of smooth manifolds. It is only necessary
to replace open subsets of Rn by open subsets of Cn, and smooth functions
by holomorphic functions. Nevertheless we will illustrate how the complex
and real worlds are fundamentally diļ¬€erent.
We try to enter into this world because it provides the setting of our issue.
1.1 Complex coordinates
We will give the deļ¬nition of complex manifold following the deļ¬nition of
smooth manifold, with the appropriate changing.
Deļ¬nition 1.1. M is a complex manifold of complex dimension n if:
ā€¢ it is a topological space T2 and countable;
ā€¢ it is endowed with a complex atlas i.e. a collection of coordinate charts
F = {(Ui, Ļ•i)| i āˆˆ I},
where I is a set of index such that
ā€“ Ui āŠ‚ M are open sets and {(Ui)}iāˆˆI is a cover of M i.e.
M =
iāˆˆI
Ui;
ā€“ Ļ•i : Ui ā†’ Cn is a homeomorphism onto an open set in Cn,
ā€“ the change of coordinates
Ļ•i ā—¦ Ļ•āˆ’1
j : Ļ•j(Ui āˆ© Uj) ā†’ Ļ•i(Ui āˆ© Uj)
are biholomorphic, āˆ€ i, j s.t. Ui āˆŖ Uj = āˆ…;
1
Chapter 1. Complex manifolds
ā€“ the collection F is maximal towards the previous condition. This
means that if (U, Ļ•) is a system of charts such that Ļ•i ā—¦ Ļ• and
Ļ• ā—¦ Ļ•āˆ’1
i are biholomorphic āˆ€i āˆˆ I, then (U, Ļ•) āˆˆ F.
In other words M is a diļ¬€erentiable manifold endowed with a holomorphic
atlas.
Observation 1.2. Charts deļ¬ne coordinates. Suppose Ļ•: U ā†’ Cn is some chart
centered around a point p āˆˆ U (meaning p maps to the origin.) Then Ļ• can
be written as local coordinates (z1, . . . , zn), where each zi is a holomorphic
function on U. These are complex coordinates on M. Note that if M is a
complex nāˆ’dimensional manifold, it can be realized as a real 2nāˆ’dimensional
manifold with coordinates (xi, yi) coming from zj = xj + iyj.
Example 1.3 (Complex Torus). Trivially Euclidean space Cn is a comlplex
manifold. We can now consider the action of Zn on Cn by translation:
(m1, n1, . . . , mn, nn) Ā· (z1, . . . , zn) = (z1 + m1 + in1, . . . , zn + mn + inn),
where mi, ni āˆˆ Z, zi āˆˆ C, āˆ€i.
Since this action is properly discontinuous and holomorphic, the quotient
Cn/Zn inherits the structure of a complex manifold from the standard atlas
on Cn. Thus Tn := Cn/Zn is a manifold, called complex torus.
Example 1.4 (Complex projective space). CPn is the set of equivalence classes
{[z] = [z0 : Ā· Ā· Ā· : zn], | zi āˆˆ C},
where y āˆˆ Cn and w āˆˆ Cn, belong to the same class iļ¬€ there exist Ī» āˆˆ C{0}
such that y = Ī»w.
Let
Ui = {[z0 : Ā· Ā· Ā· : zn] | zi = 0} .
An open cover of CPn is given by
{Ui}i=0,...,n .
The maps Ļ•i : Ui ā†’ Cn given by
[z0 : Ā· Ā· Ā· : zn] ā†’ (z0/zi, . . . , zi/zi, . . . , zn/zi)
are bijective, so it remains to show that the transition functions are holo-
morphic.
These are given by Ļ•ij := Ļ•i ā—¦ Ļ•āˆ’1
j : Ļ•j(Ui āˆ© Uj) ā†’ Ļ•i(Ui āˆ© Uj).
Let wk be the coordinates on Uj and assume i < j. Then Ļ•ij is deļ¬ned by
(w0, . . , Ė†wj, . . , wn)
Ļ•āˆ’1
j
āˆ’āˆ’ā†’ [w0 : . . : 1 : . . : wn]
Ļ•i
āˆ’ā†’ (
w0
wi
, . . ,
Ė†wi
wi
, . . ,
1
wi
, . . ,
wn
wi
).
Since wi = 0 on Ļ•j(Ui āˆ© Uj), we see that the coordinate functions Ļ•ij are of
the form wk/wi or 1/wi, which are holomorphic on the domain of Ļ•ij. So
CPn is a complex manifold.
2
Chapter 1. Complex manifolds
1.2 Tangent space
Let M be a complex manifold of complex dimension n and x be a point of
M.
Let Cn be identiļ¬ed with R2n via the map (z1, . . . , zn) ā†’ (x1, y1 . . . , xn, yn).
Let (z1, . . . , zn) be a local coordinate system near x with zj = xj + iyj,
j = 1, . . . , n.
Then the real tangent space Tx(M) is spanned by
āˆ‚
āˆ‚x1 x
,
āˆ‚
āˆ‚y1 x
, . . . ,
āˆ‚
āˆ‚xn x
,
āˆ‚
āˆ‚yn x
.
Deļ¬ne an Rāˆ’linear map J from Tx(M) onto itself by
J
āˆ‚
āˆ‚xi x
=
āˆ‚
āˆ‚yi x
, J
āˆ‚
āˆ‚yi x
= āˆ’
āˆ‚
āˆ‚xi x
for all j = 1, . . . , n.
Obviously, we have J2 = āˆ’1, and J is called the complex structure on Tx(M).
We observe that the deļ¬nition of J is independent of the choice of the local
coordinates (z1, . . . , zn). The complex structure J induces a natural splitting
of the complexiļ¬ed tangent space
CTx(M) = Tx(M) āŠ—R C.
First we extend J to the whole complexiļ¬ed tangent space by
J(x āŠ— Ī±) = (Jx) āŠ— Ī±.
It follows that
J : CTx(M) ā†’ CTx(M)
is a Cāˆ’linear map with eigenvalues i and āˆ’i.
Denote by T1,0
x (M) and by T0,1
x (M) the eigenspaces of J corresponding to i
and āˆ’i respectively. It is easily veriļ¬ed that
ā€¢ T1,0
x (M) = T0,1
x (M),
ā€¢ T1,0
x (M) āˆ© T0,1
x (M) = {0},
ā€¢ T1,0
x (M) is spanned by
āˆ‚
āˆ‚z1 x
, . . . ,
āˆ‚
āˆ‚zn x
,
where āˆ‚
āˆ‚zi x
= 1
2
āˆ‚
āˆ‚xi
āˆ’ i āˆ‚
āˆ‚yi x
, for 1 ā‰¤ i ā‰¤ n.
Consequently T0,1
x (M) is spanned by
āˆ‚
āˆ‚ ĀÆz1 x
, . . . ,
āˆ‚
āˆ‚ ĀÆzn x
,
3
Chapter 1. Complex manifolds
Any vector v āˆˆ T1,0
x (M) is called a vector of type (1, 0), and we call v āˆˆ
T0,1
x (M) a vector of type (1, 0). The space T1,0
x (M) is called the holomorphic
tangent space at x.
Therefore
CTx(M) = Tx(M) āŠ—R C = T1,0
x (M) āŠ• T0,1
x (M).
Elements of CTx(M) can be realized as Cāˆ’linear derivations in the ring
of complex valued Cāˆž functions on M around x.
Indeed, given the real derivation v āˆˆ Tx(M), the elementary tensor v āŠ— z āˆˆ
Tx(M) āŠ—R C acts on f + ig in the following way
(v āŠ— z)(f + ig) = z Ā· (v(f) + iv(g)).
1.3 Complex forms: the (p,q)-forms
1.3.1 1-forms
Let CTāˆ—
x (M) be the dual space of CTx(M). By duality, J also induces a
splitting on
Cāˆ—
Tx(M) = T1,0
x
āˆ—
(M) āŠ• T0,1
x
āˆ—
(M) := Ī›1,0
x (M) āŠ• Ī›0,1
x (M), (1.1)
where Ī›1,0
x (M) and Ī›0,1
x (M) are eigenspaces corresponding to the eigenvalues
i and āˆ’i respectively. Let U be an open set containing x and let z1, . . . , zn
be local coordinates. It is easy to see that the vectors
(dz1
)x, . . . , (dzn
)x
span Ī›1,0
x (U) and that the space Ī›0,1
x (U) is spanned by
(dĀÆz1
)x, . . . , (dĀÆzn
)x.
This means that any diļ¬€erential 1-form with complex coeļ¬ƒcients can be
written uniquely as a sum
n
j=1
fjdzj
+ gjdĀÆzj
where fj, gj āˆˆ Cāˆž(U), i.e. fj, gj : U ā†’ C are holomorphic.
Because of this splitting the complex 1-forms are also called the (1, 1)-forms.
Example 1.5. Let f : M ā†’ C be a smooth function, the exterior derivative
d = āˆ‚ + ĀÆāˆ‚ is deļ¬ned by
df =
n
j=1
āˆ‚f
āˆ‚zj
dzj
+
āˆ‚f
āˆ‚ ĀÆzj
dĀÆzj
= āˆ‚f + ĀÆāˆ‚f.
4
Chapter 1. Complex manifolds
1.3.2 Higher degree forms
Let p and q be a pair of non-negative integers ā‰¤ n. One can deļ¬ne the wedge
product of complex diļ¬€erential forms in the same way as with real forms.
The space Ī›
(p,q)
x of (p, q)-forms is deļ¬ned by taking linear combinations of the
wedge products of p elements from Ī›
(1,0)
x (M) and q elements from Ī›
(0,1)
x (M),
i.e.,
Ī›(p,q)
x (M) = Ī›(1,0)
x (M) āˆ§ Ā· Ā· Ā· āˆ§ Ī›(1,0)
x (M)
pāˆ’times
āˆ§ Ī›(0,1)
x (M) āˆ§ Ā· Ā· Ā· āˆ§ Ī›(0,1)
x (M)
qāˆ’times
.
At this point the isomorphism (1.1) leads to the following splitting
Ī›k
x(M) := Ī›k
x(Cāˆ—
Tx(M)) = Ī›k
(T1,0
x
āˆ—
(M) āŠ• T0,1
x
āˆ—
(M)) (1.2)
=
p+q=k
Ī›p
x(T1,0
x
āˆ—
(M)) āˆ§ Ī›q
x(T0,1
x
āˆ—
(M)) (1.3)
=
p+q=k
Ī›(p,q)
x (M). (1.4)
This means that given u a (p, q)āˆ’form, it can be written in local coordinates
in an open set U as
u(z) =
|I|=p,|J|=q
uIJ (z)dzI
āˆ§ dĀÆzJ
where uIJ āˆˆ Cāˆž(U.)
If we consider the disjoint union of the Ī›k
x(M) over the points of the manifold,
we get the space of all complex diļ¬€erential forms of total degree k, i.e.,
Ī›k
(M) =
xāˆˆM
Ī›k
x(M).
Example 1.6. Let u be a (p, q)āˆ’form. We can deļ¬ne de following operator
āˆ‚ : Ī›(p,q)
(M) ā†’ Ī›(p+1,q)
(M) and ĀÆāˆ‚ : Ī›(p,q)
(M) ā†’ Ī›(p,q+1)
(M) (1.5)
and respectively in an open coordinate set their actions are given by
āˆ‚u =
I,J 1ā‰¤kā‰¤n
āˆ‚uI,J
āˆ‚zk
dzk
āˆ§ dzI
dzJ
, ĀÆāˆ‚u =
I,J 1ā‰¤kā‰¤n
āˆ‚uI,J
āˆ‚ĀÆzk
dĀÆzk
āˆ§ dzI
dzJ
.
According to the previous consideration we can consider the diļ¬€erential
operator
d = āˆ‚ + ĀÆāˆ‚. (1.6)
It satisfy
ā€¢ d: Ī›k(M) ā†’ Ī›k+1(M), āˆ€k
5
Chapter 1. Complex manifolds
ā€¢ d2 = 0
ā€¢ if Ļ‰ āˆˆ Ī›r(M) and Ī± āˆˆ Ī›s(M) then d(Ļ‰ āˆ§ Ī±) = dĻ‰ āˆ§ Ī±(āˆ’1)rĻ‰ āˆ§ dĪ±
ā€¢ if f āˆˆ Cāˆž(U) then the operator is the same as in the Example 1.5.
Remark. If M is a complex manifold, as in the case of the real manifolds,
Ī›k(M) deļ¬nes a complex vector bundle on M, of "complex" rank n
k . In the
following section we will recall the deļ¬nition in the real case and it will turn
out that it is a useful construction to study complex manifolds.
1.4 Canonical bundle
Let M be a real smooth manifold and let Tāˆ—M be its cotangent bundle.
As we saw in the complex case, the vector bundle of the rāˆ’forms on M is
deļ¬ned by
Ī›r
(M) =
xāˆˆM
Ī›r
x(Tāˆ—
x M) :=
xāˆˆM
Ī›r
x(M)
and it has rank n
r as real vector bundle.
Deļ¬nition 1.7 (Canonical bundle). Let r = n. The bundle of the nāˆ’forms
of a smooth manifold of dimension n is called the canonical bundle.
Since n
n = 1 the canonical bundle is a line bundle.
Deļ¬nition 1.8 (Holomorphic canonical bundle). Let M be a complex man-
ifold. The holomorphic line bundle
Ī›n
(M) := K
is called the canonical line bundle.
Our aim will be to investigate manifold with trivial canonical bundle. It
is known that in the real case this condition is equivalent to:
Ī›n
(M) is trivial ā‡ā‡’ āˆƒ a nowhere vanishing section
ā‡ā‡’ M is orientable.
The second condition means that there exist a diļ¬€erentiable n form Ļ‰ nowhere
vanishing, i.e., Ļ‰(p) = 0, for every p in M. While the ļ¬rst condition still
applies in the complex case, since it is a topological condition, the second
equivalence it is not still true, as the following theorem proves.
Theorem 1.9. Any complex manifold M of dimension n is orientable.
6
Chapter 1. Complex manifolds
Proof. Let (Ui, Ļ•i)iāˆˆI be a system of holomorphic coordinates around a point
p, and
Ļ•i : Ui ā†’ Ļ•(Ui), p ā†’ (z1(p), . . . , zn(p)),
Ļ•j : Uj ā†’ Ļ•(Uj), p ā†’ (Z1(p), . . . , Zn(p)).
By identifying Cn with R2n in the usual way we can write:
(z1, . . . , zn) = (x1, y1, . . . , xn, yn) and (Z1, . . . , Zn) = (X1, Y1, . . . , Xn, Yn)
and thus we obtain a real manifold structure on M.
We will now calculate the Jacobian matrices of the transition functions for
both structures.
In the complex case we have
JacĻ•i(p)(Ļ•j ā—¦ Ļ•āˆ’1
i ) =
āˆ‚(Ļ•j ā—¦ Ļ•āˆ’1
i )k
āˆ‚zl 1ā‰¤k,lā‰¤n
=
āˆ‚Zk(z1, . . . , zn)
āˆ‚zl 1ā‰¤k,lā‰¤n
= (ckl)1ā‰¤k,lā‰¤n āˆˆ GL(n, C).
Instead in the real case we will ļ¬nd
JacĻ•i(p)(Ļ•j ā—¦ Ļ•āˆ’1
i ) =
ļ£«
ļ£­
āˆ‚Xk(x1,y1...,xn,yn)
āˆ‚xl
āˆ‚Xk(x1,y1...,xn,yn)
āˆ‚yl
āˆ‚Yk(x1,y1...,xn,yn)
āˆ‚xl
āˆ‚Yk(x1,y1...,xn,yn)
āˆ‚yl
ļ£¶
ļ£ø
1ā‰¤k,lā‰¤n
.
Using the Cauchy-Riemann conditions, this coincides with
Re(ckl) āˆ’Im(ckl)
Im(ckl) Re(ckl)
1ā‰¤k,lā‰¤n
āˆˆ GL(2n, R).
We will now calculate the determinant of these matrices and we will see that
it is always positive, which is equivalent to state that M is orientable as a
real manifold.
Consider the following homomorphism
Ļ: Mn(C) ā†’ M2n(R),
deļ¬ned by
(ckl)1ā‰¤k,lā‰¤n ā†’
Re(ckl) āˆ’Im(ckl)
Im(ckl) Re(ckl)
1ā‰¤k,lā‰¤n
.
The map Ļ is continuous, since it is Rāˆ’linear and the spaces involved are
ļ¬nite dimensional. Also, being a Lie algebra homomorphism, we have
det(Ļ(Pāˆ’1
AP)) = det(Ļ(Pāˆ’1
)Ļ(A)Ļ(P)) = det(Ļ(A)).
7
Chapter 1. Complex manifolds
Finally, the diagonalizable matrices are dense in Mn(C), so we can restrict
our calculations to diagonal matrices in Mn(C). Therefore we obtain:
det(Ļ(Diag(c1, . . . , cn))) = det Diag
Re(c1) āˆ’Im(c1)
Im(c1) Re(c1)
, . . . ,
Re(cn) āˆ’Im(cn)
Im(cn) Re(cn)
=
n
i
det
Re(ci) āˆ’Im(ci)
Im(ci) Re(ci)
=
n
i
|ci|2
= | det (Diag(c1, . . . , cn)) |2
.
So we can conclude that,
det(Ļ(A)) = | det(A)|2
, āˆ€A āˆˆ Mn(C).
Finally, we get that the Jacobian matrices of the transition functions for the
chart Ļ•i for M have positive determinants, thus the real underlying manifold
M is orientable.
Synthetically if we want to verify if a holomorphic canonical bundle is
trivial, the following equivalence applies:
Ī›n
(M) is trivial ā‡ā‡’ āˆƒ a nowhere vanishing section of Ī›n
(M).
The nowhere vanishing section is also called a holomorphic volume form.
Example 1.10. We are going to examine the triviality of the canonical bundle
of some well-known manifolds.
ā€¢ The canonical bundle of a complex nāˆ’dimensional torus K ā†’ Tn is
trivial. In fact we can explicitly show a nowhere vanishing section of
its canonical bundle. With the same notation used in Example 1.3, we
can see that
dz1
āˆ§ Ā· Ā· Ā· āˆ§ dzn
is a non vanishing complex nāˆ’form belonging to Ī›n(M).
ā€¢ The canonical bundle of the projective line K ā†’ CP1 is not trivial.
One could prove that using the general following property for a closed
oriented Riemann surface M:
Ī›n
(M) is trivial ā‡” Ļ‡(M) = 0. (1.7)
Indeed Ļ‡(CP1) = 2, but we will not go into details. Furthermore asking
K to be trivial is equivalent to ask TCP1 to be trivial, and as we can
8
Chapter 1. Complex manifolds
Figure 1.1: Non vanishing section on torus.
Figure 1.2: As we can see from the picture it is not possible to ļ¬nd a holomorphic
non vanishing form on the sphere: the poles correspond to the two pots with zero
net ļ¬‚ow.
9
Chapter 1. Complex manifolds
see from ļ¬gure 1.2 it is not possible to ļ¬nd a non vanishing form on
this manifold1. In fact since the following proposition applies we have
that the tangent bundle of CP1 is diļ¬€eomorphic to the tangent bundle
of S2 thus it is not trivial.
Proposition 1.11. The complex projective line CP1 is diļ¬€eomorphic to
the 2-sphere S2.
Proof. Consider a point [z : w] with w = 0. Then we send this point to
z
w āˆˆ C. We use then the stereographic projection C ā†’ S2N, and we
send the point [1 : 0] to the north pole of S2. It is easy to check that
the composition
CP1
ā†’ Cāˆ—
ā†’ S2
is a diļ¬€eomorphism.
The inverse of the ļ¬rst map sends a point h āˆˆ C to
h
1 + |h|2
,
1
1 + |h|2
and inļ¬nity to [1, 0], so we can easily obtain the inverse
S2
ā†’ CP1
explicitly.
ā€¢ The canonical bundle of a Riemann surface Ī£, K ā†’ Ī£ is not trivial
if its genus, g(M), is greater than one. One could see this using (1.7)
together with the fact that for a closed Riemann surface its genus is
related to the Euler characteristic by the formula
2 āˆ’ 2g(M) = Ļ‡(M).
1.4.1 Line bundles over M
We should now open a parenthesis on line bundles in order to clarify some
ideas and to understand the potential of these structures.
Let M be a complex manifold and L1, L2, L3 be elements of the set
G = {isomorphism classes of all holomorphic line budle overM} .
They satisfy:
1
This topic is closely related to the problem of combing a hairy sphere, see "hairy ball
theorem" for further details.
10
Chapter 1. Complex manifolds
1. Closure: if L1 and L2 are holomorphic line bundles then the tensor
product L1 āŠ— L2 is a holomorphic line bundle.
In fact by deļ¬nition the tensor product of two bundles is the bundle
whose ļ¬ber is the tensor product of the ļ¬ber. Therefore, L1 āŠ— L2 has
ļ¬ber (p, C) āŠ— (p, C)), āˆ€p. That is because
C āŠ— C = C
since
C āŠ— C = C āŠ— Cāˆ—
= Hom(C, C).
Moreover if we consider the following isomorphism
Ī¦: C ā†’ Hom(C, C)
a ā†’ Ī¦(a): C ā†’ C
b ā†’ a Ā· b
we conclude that Hom(C, C) = C.
2. Associativity:
(L1 āŠ— L2) āŠ— L3 = L1 āŠ— (L2 āŠ— L3)
3. Identity element idL1 : it is the trivial bundle C Ɨ M since
L1 āŠ— C Ɨ M = L1.
4. Inverse element: Lāˆ—
1. Since
idL1 āˆˆ Hom(L1, L1) = idL1 .
idL1 is a nowhere vanishing section of idL1 and this means that
L1 āŠ— Lāˆ—
1 = C.
5. Commutativity:
L1 āŠ— L2 = L2 āŠ— L1,
even if they are diļ¬€erent bundles, they are isomorphic.
According to the consideration we have made, we can conclude that the space
of line bundles modulo equivalence forms a group under the tensor product.
Often instead of studying ļ¬ber bundles it is convenient to analyse their
smooth sections. In the case of the holomorphic line bundles, the holomorphic
sections correspond to the holomorphic functions f : M ā†’ C. But this is
quite useless since, as we will prove, holomorphic functions over a compact
connected manifold have to be constant. First we will examine the following
lemma that will let us to prove what we claimed.
11
Chapter 1. Complex manifolds
Lemma 1.12. Let B be an open set of Cn and f : B ā†’ C. If |f| has a
maximum in B, then f is constant.
Proof. Let x, y āˆˆ Rn such that z = x + iy āˆˆ Cn and f(x, y) = u(x, y) +
iv(x, y) āˆˆ Cn.
If |f| has a maximum, then |f|2 = u(x, y)2 + v(x, y)2 has a maximum. Since
|f|2 is a real valued function, to ļ¬nd its stationary point, we set the partial
derivative to zero, namely
ļ£±
ļ£²
ļ£³
āˆ‚|f|2
āˆ‚xi
= 2 āˆ‚u
āˆ‚xi
Ā· u + 2 āˆ‚v
āˆ‚xi
Ā· v = 0
āˆ‚|f|2
āˆ‚yi
= 2 āˆ‚u
āˆ‚yi
Ā· u + 2 āˆ‚v
āˆ‚yi
Ā· v = 0
, āˆ€i āˆˆ {1, . . . , n}. (1.8)
By linear algebra, solving the above system is equal to impose:
ļ£«
ļ£­
āˆ‚u
āˆ‚xi i
āˆ‚v
āˆ‚xi i
āˆ‚u
āˆ‚yi i
āˆ‚v
āˆ‚yi i
ļ£¶
ļ£ø u
v
= 0, āˆ€i āˆˆ {1, . . . , n}, (1.9)
and using the Cauhy-Riemann conditions this is equivalent to ask
ļ£«
ļ£­
āˆ‚u
āˆ‚xi i
āˆ’ āˆ‚u
āˆ‚yi i
āˆ‚u
āˆ‚yi i
āˆ‚u
āˆ‚xi i
ļ£¶
ļ£ø u
v
= 0 āˆ€i āˆˆ {1, . . . , n}. (1.10)
It is easily seen that the last matrix has rank 1, so the condition (1.10) is
equivalent to impose
āˆ‚u
āˆ‚xi
2
+
āˆ‚u
āˆ‚yj
2
= 0, āˆ€i, j. (1.11)
Hence
āˆ‚u
āˆ‚xi
=
āˆ‚u
āˆ‚yj
= 0, āˆ€i, j. (1.12)
Therefore we are asking f to be constant.
Theorem 1.13. Let M be a complex compact and connected manifold,
f : M ā†’ C be a holomorphic function. Then f is constant.
Proof. Since f is holomorphic, |f| is a continuous function deļ¬ned on a
compact set and for the Weirstrass theorem |f| has a maximum on M.
This means that if (UĪ±, Ļ•Ī±) is a ļ¬nite cover of M around the point p where
f reaches his maximum, the function
|f ā—¦ Ļ•āˆ’1
Ī± |: Ļ•Ī±(UĪ±) āŠ† Cn
ā†’ C
has a maximum. Using Lemma 1.12 we get that f is constant on UĪ±. Now
we can conclude considering that every holomorphic function is an analytic
function, more speciļ¬cally it is a continuous function on UĪ±. Since f is
constant on UĪ± āˆŖUĪ², āˆ€Ī±, Ī², because of the uniqueness of the Taylor expansion
f is constant on M.
12
Chapter 1. Complex manifolds
1.5 Cohomologies
Before turning to the Dolbeault cohomology of complex manifolds, we will
give a very brief summary of these concepts in the real situation which, of
course, also applies to complex manifolds if they are viewed as real analytic
manifolds.
Deļ¬nition 1.14. A sequence Cāˆ— = (Cn, āˆ‚n)nāˆˆZ of modules Cn over a ring
R and homomorphisms āˆ‚n : Cn ā†’ Cnāˆ’1 is called a chain complex, if for all
n āˆˆ Z we have that
āˆ‚nāˆ’1 ā—¦ āˆ‚n = 0
holds. The āˆ‚n functions are usually called the boundary operators or diļ¬€er-
entials.
A chain complex is usually visualised in a diagram as such,
Ā· Ā· Ā·
āˆ‚n+1
āˆ’ā†’ Cn
āˆ‚n
āˆ’ā†’ Cnāˆ’1
āˆ‚nāˆ’1
āˆ’ā†’ Ā· Ā· Ā·
Observation 1.15. Since āˆ‚nāˆ’1 ā—¦ āˆ‚n = 0 we immediately get that
Im(āˆ‚n) āŠ‚ ker(āˆ‚nāˆ’1).
Note that these are both submodules of Cn.
Deļ¬nition 1.16. The n-cycles of a chain complex Cāˆ— is
Zn(Cāˆ—) = ker(āˆ‚n).
The n-boundaries of a chain complex Cāˆ— is
Bn(Cāˆ—) = Im(āˆ‚n+1).
The n-th homology module of a chain complex Cāˆ— is
Hn(Cāˆ—) = Zn/Bn.
Similarly one can deļ¬ne a cochain to consist of the modules of Rāˆ’linear
maps from your modules to R, together with special boundary maps
dn : Cn ā†’ Cn+1.
1.5.1 The de Rham cohomology
In order to apply what we have seen to manifolds we will use for Rāˆ’modules in
this case real vector spaces, namely the vector spaces of diļ¬€erential kāˆ’forms
Ī›k(M), for all k = 1 . . . n, if n is the dimension of the manifold.
13
Chapter 1. Complex manifolds
Deļ¬nition 1.17. A diļ¬€erential form Īø is called closed if dĪø = 0. And a
diļ¬€erential kāˆ’form Ī± is exact if there exist a diļ¬€erential (k āˆ’ 1)āˆ’form Ī²
such that dĪ² = Ī±.
Observation 1.18. Note that because d ā—¦ d = 0 we have that all exact forms
are also closed.
Lemma 1.19. Let M be a smooth manifold, the following diagram is a
cochain
Ā· Ā· Ā·
d
āˆ’ā†’ Ī›nāˆ’1
(M)
d
āˆ’ā†’ Ī›n
(M)
d
āˆ’ā†’Ī›n+1
(M)
d
āˆ’ā†’ Ā· Ā· Ā·
Observation 1.20. The n-cycles are exactly the closed n-forms on M. And
the n-boundaries are the exact n-forms on M.
Observation 1.21. We also use the fact that Ī›n = 0 for n > dim(M).
Proof. Note that Ī›n(M) is a real vector space, thus an Rāˆ’module. And
d is a Rāˆ’linear map. Furthermore d ā—¦ d = 0 which is the same as saying
Im(d) āŠ‚ ker(d). Thus we are dealing with a cochain.
Deļ¬nition 1.22. The p āˆ’ th de Rham cohomology group is equal to the
p āˆ’ th cohomology groups of the cochain in Lemma 1.19. This is usually
denoted
Hp
deRham(M).
We will now analyse a property that can be applied to symplectic mani-
folds.
Proposition 1.23. Let M be a 2n dimensional oriented diļ¬€erentiable man-
ifold which is compact and without boundary.
For all 0 ā‰¤ p ā‰¤ n it exists an isomorphism
Hp
deRham(M) = Hnāˆ’p
deRham(M)
āˆ—
.
In particular
bk(M) = bnāˆ’k(M).
For a proof of the theorem and for a general clariļ¬cation the reader is
referred to [Huy06].
To be more precise we should write Hp
deRham(M, R) instead of Hp
deRham(M).
The symbol R is used here to stress that we are considering real valued
pāˆ’ forms; of course one can introduce a similar group
Hp
deRham(M, C)
for complex valued forms, i.e. forms with values in C āŠ— Ī›p(M). Then
Hp
deRham(M, C) = C āŠ— Ī›p
(M) (1.13)
is the complexiļ¬cation of the real De Rham cohomology group.
14
Chapter 1. Complex manifolds
1.5.2 Dolbeault cohomology
Most of the facts about homology and De Rham cohomology on real manifolds
are also valid on complex manifolds if one views them as real analytic
manifolds. However one can use the complex structure to deļ¬ne as we have
seen in (1.5) the āˆ‚-cohomology or Dolbeault cohomology. With the same
notation as those in Chapter 3
ĀÆāˆ‚ : Ī›(p,q)
(M) ā†’ Ī›(p,q+1)
(M).
In particular, we get diļ¬€erential cochain complexes
Ā· Ā· Ā·
ĀÆāˆ‚
āˆ’ā†’ Ī›(p,qāˆ’1)
(M)
ĀÆāˆ‚
āˆ’ā†’ Ī›(p,q)
(M)
ĀÆāˆ‚
āˆ’ā†’Ī›(p,q+1)
(M)
ĀÆāˆ‚
āˆ’ā†’ Ā· Ā· Ā·
Deļ¬nition 1.24. We say that a (p, q)āˆ’form Ī± is ĀÆāˆ‚-closed if ĀÆāˆ‚Ī± = 0. The
space of ĀÆāˆ‚-closed (p, q)āˆ’forms is denoted by
Z
(p,q)
ĀÆāˆ‚
(M).
A (p, q)āˆ’form Ī² is ĀÆāˆ‚-exact if it is of the form Ī² = ĀÆāˆ‚Ī³ for Ī³ āˆˆ Ī›(p,qāˆ’1)(M).
Observation 1.25. Since ĀÆāˆ‚2 = 0, ĀÆāˆ‚(Ī›(p,q)(M)) āŠ‚ Z
(p,q+1)
ĀÆāˆ‚
(M).
Deļ¬nition 1.26. Dolbeault cohomology groups are then deļ¬ned as
H
(p,q)
ĀÆāˆ‚
(M) =
Z
(p,q)
ĀÆāˆ‚
(M)
ĀÆāˆ‚(Ī›(p,qāˆ’1)(M))
. (1.14)
Deļ¬nition 1.27. The dimensions of the (p, q) cohomology groups are called
Hodge numbers
h(p,q)
(M) = dimC H
(p,q)
ĀÆāˆ‚
(M).
They are ļ¬nite for compact complex manifolds. The Hodge numbers of a
compact complex manifold are often arranged in the Hodge diamond:
h0,0
h1,0 h0,1
h2,0 h1,1 h0,2
h3,0 h2,1 h1,2 h0,3
h3,1 h2,2 h1,3
h3,2 h2,3
h3,3
which we have displayed here for a three complex dimensional manifold. The
general diagrams take the following form
15
Chapter 1. Complex manifolds
hn,n
hn,nāˆ’1 hnāˆ’1,n
hn,nāˆ’2 hnāˆ’1,nāˆ’1 hnāˆ’2,n
. . .
... . . .
h2,0 h1,1 h0,2
h1,0 h0,1
h0,0
The decomposition (1.4) does not carry over the cohomology group in fact
we have that
dim(Hk
deRham(M, C)) ā‰¤
p+q=k
h(p,q)(M)
. (1.15)
We will see in Chapter 4 that the equality (1.15) is for a special types of
complex manifolds, namely Calabi-Yau manifolds and KƤhler manifolds.
Deļ¬nition 1.28. dim(Hk
deRham(M, C)) is called the k āˆ’ th Betty number.
16
Chapter 2
Some Algebraic Geometry
tools
In this chapter we analyse some tools of algebraic geometry that will be used
in the following discussion in order to understand the algebraic aspect of
some Calaby-Yau manifolds.
2.1 Aļ¬ƒne and projective variety
Let k be an algebraically closed ļ¬eld and ļ¬x S āŠ† k[x1, . . . , xn]. Let An
k be
the n-dimensional aļ¬ƒne space over k.
Deļ¬nition 2.1. The set
V(S) = {p āˆˆ An
k | f(p) = 0 āˆ€f āˆˆ S}.
is called aļ¬ƒne algebraic set.
Let I(S) the ideal generated by S, i.e., the smallest ideal of k[x1, . . . , xn]
which contains S.
Observation 2.2. We have that V(S) = V(I(S)).
Moreover the ring k[x1, . . . , xn] is nƶetherian, thus every ideal is ļ¬nitely
generated. For every ideal I of k[x1, . . . , xn] there exist polynomials f1, . . . , fr āˆˆ
k[x1, . . . , xn] such that
V(I) = V(f1, . . . , fr)
= {(a1, . . . , an) āˆˆ An
k | fi(a1, . . . , an) = 0, 1 ā‰¤ i ā‰¤ r}.
Deļ¬nition 2.3. The Zariski topology on An
k is the topology whose closed
set are the algebraic subset of An
k .
If X āŠ‚ An
k the Zariski topology on X is the one induced by the Zariski
topology on An
k .
17
Chapter 2. Some Algebraic Geometry tools
We can now give the deļ¬nition of reducible and irreducible set which are
purely topological concepts.
Deļ¬nition 2.4. Let X be a topological space and Y āŠ† X such that Y = āˆ….
Y is irreducible if it is not the union of two closed proper subsets of Y.
A subset of X is reducible if it is not irreducible.
Thus every non empty set can be expressed as union of irreducible subsets
Y = Yi
where Yi are the irreducible components.
Deļ¬nition 2.5. An algebraic variety over the ļ¬eld k is an irreducible closed
subset of An
k , endowed with the Zariski topology. An open subset of an aļ¬ƒne
variety is called quasi-aļ¬ƒne variety.
The deļ¬nition we have previously given can be extended, in a natural
way, to the projective space.
The projective space Pn
k can be considered an extension of the aļ¬ƒne space
by the following immersion
An
k ā†’ Pn
k
(a1, . . . , an) ā†’ (1, a1, . . . , an)
Let k be an algebraically closed ļ¬eld and consider the projective space Pn
k . Let
R = k[x1, . . . , xn, xn+1] be the ring of polynomials in n + 1 indeterminates
and f be an homogeneous polynomial of R, then it makes sense to ask
whether or not f(p) = 0 for a point p āˆˆ Pn. As in the aļ¬ƒne case we have the
following deļ¬nition.
Deļ¬nition 2.6. A projective algebraic subset of Pn
k is the common zero set
of a collection of homogeneous polynomials in R.
We can deļ¬ne as well projective varieties and the Zariski topology in Pn
k .
2.2 Regular and rational functions
Fix X āŠ‚ An
k , algebraic set.
Deļ¬nition 2.7. A function
X ā†’ k
is regular if it agrees with the restriction to X of some polynomial function
on the ambient An
k .
18
Chapter 2. Some Algebraic Geometry tools
The set of all regular functions on X has a natural ring structure (where
addition and multiplication are the functional notions). This is the coordinate
ring of X, denoted k[X].
For all open set U āŠ‚ X, we will denote OX(U) (or simply O(U)) the set of
regular functions on U. Whereas sum and product of regular function is a
regular function, the set OX(U) is actually a ring, rather a kāˆ’algebra. We
call OX(U) the sheaf of regular functions on U.
Fix now an aļ¬ƒne algebraic set X and assume that X is irreducible. Even
though we will not prove the following fact we have to remark that
X is irreducible ā‡ā‡’ k[X] is a domain.
Deļ¬nition 2.8. The function ļ¬eld of X is the fraction ļ¬eld of k[X], denoted
k(X).
Deļ¬nition 2.9. A rational function on X is an element Ļ• āˆˆ k(X) i.e., Ļ• is
an element of the equivalence class f
g , where f, g āˆˆ k[X], g = 0. Here
f
g
āˆ¼
f
g
ā‡ā‡’ fg = gf
as elements of k[X]. A rational function Ļ• āˆˆ k(X) is regular at p āˆˆ X if it
admits a representation Ļ• = f
g where g(p) = 0.
The domain of deļ¬nition of Ļ• āˆˆ k(X) is the locus of all points p āˆˆ X where
Ļ• is regular.
The deļ¬nition or regular and rational function is a little diļ¬€erent from
the one of the aļ¬ƒne case.
Deļ¬nition 2.10. Let X āŠ‚ Pn
k an algebraic set and U an open subset of
X. A function f is regular around a point p āˆˆ U if there exist an open
neighbourhood V of p in U and two homogeneous polynomials g, h of the
same degree such that for all (a1, . . . , an+1) āˆˆ V,
h(a1, . . . , an+1) = 0
and
f|V =
g
h
.
The function f is regular in U if it is regular in every points of U.
As in the aļ¬ƒne case, for every open subset U of X we denote OX(U) (or
O(U)) the ring of the regular functions in U.
19
Chapter 2. Some Algebraic Geometry tools
2.3 Divisors
Let X be an irreducible variety.
Deļ¬nition 2.11. A prime divisor or irreducible divisor on X is a codimen-
sion 1 irreducible (closed) subvariety of X. A divisor D on X is a formal
Zāˆ’linear combination of prime divisors
D =
t
i=1
kiDi, ki āˆˆ Z.
In P2, C = V(xy āˆ’ z2), L1 = V(x) and L2 = V(y) are prime divisors,
while 2C, 2L1 āˆ’ L2 are divisors which are not prime.
We say a divisor D = t
i=1 kiDi is eļ¬€ective if each ki ā‰„ 0.
The support of D is the list of prime divisors occurring in D with non-zero
coeļ¬ƒcient.
The set of all divisors on X form a group Div(X), the free abelian group on
the set of prime divisors of X. The zero element is the trivial divisor
D = 0Di,
and Supp(0) = āˆ….
Example 2.12. Consider
Ļ• =
f
g
=
(t āˆ’ Ī»1)a1 Ā· Ā· Ā· (t āˆ’ Ī»n)an
(t āˆ’ Āµ1)b1 Ā· Ā· Ā· (t āˆ’ Āµm)bm
āˆˆ k(A1
) = k(t)
where f, g āˆˆ k[t].
The divisor of zeros and poles of Ļ• is
a1{Ī»1} + a2{Ī»2} + Ā· Ā· Ā· an{Ī»n}
Divisors of zeroes
āˆ’ b1{Āµ1} āˆ’ Ā· Ā· Ā· āˆ’ bm{Āµm}
Divisors of poles
.
Example 2.13. Let An = X. A prime divisor is D = V(h), where h āˆˆ
k[x1, . . . , xn] is irreducible. Write
Ļ• =
f
g
=
fa1
1 Ā· Ā· Ā· fan
n
gb1
1 Ā· Ā· Ā· gbm
m
āˆˆ k(An
) = k(x1, . . . , xn),
where f, g āˆˆ k[x1, . . . , xn] and fi, gi irreducible, ai āˆˆ N.
Denoting the divisor of zeros and poles of Ļ• by div(Ļ•), we have
div(Ļ•) = a1V(f1) + a2V(f2) + Ā· Ā· Ā· + anV(fn) āˆ’ b1V(g1) āˆ’ Ā· Ā· Ā· āˆ’ bmV(gm)
20
Chapter 2. Some Algebraic Geometry tools
On almost any X, we will associate to each Ļ• āˆˆ k(X){0} some divisor,
div(Ļ•), the divisor of zeros and poles, in such a way that the map
k(X)āˆ—
= k(X){0} ā†’ Div(X)
Ļ• ā†’ div(Ļ•) =
DāŠ†X
prime
Ī½D(Ļ•) Ā· D
preserves the group structure on k(X)āˆ—, i.e.,
(Ļ•1 ā—¦ Ļ•2) ā†’ div(Ļ•1) + div(Ļ•2).
The image of this map will be the group of principal divisors:
P(X) āŠ† Div(X).
We will write
div(Ļ•) =
DāŠ†X
prime
Ī½D(Ļ•) Ā· D
where Ī½D(Ļ•) = ord(Ļ•) which corresponds to the order of vanishing of Ļ•
along D and it is computed as follows: take u1, . . . , un local coordinates for
a point x āˆˆ D; write
Ļ• = fdu1 āˆ§ Ā· Ā· Ā· āˆ§ dun,
where f āˆˆ k(X). Then Ī½D(Ļ•) = Ī½D(f).
We should thus focus on the deļ¬nition of order of vanishing of Ļ• āˆˆ k(X){0}
along a prime divisor D, denoted Ī½D(Ļ•).
Assuming that X is non-singular in codimension 1, we distinguish two cases.
Case 1.Let X be aļ¬€ne, Ļ• āˆˆ k[X], D = V (Ļ€) is a hypersurface deļ¬ned by Ļ€ āˆˆ
k[X]. We say that Ļ• vanishes along D provided that D = V(Ļ€) āŠ† V(Ļ•).
So by the Nullstellensatz, (Ļ•) āŠ† (Ļ€).
Deļ¬nition 2.14. The order of vanishing of Ļ• along D, denoted Ī½D(Ļ•),
is the unique integer k ā‰„ 0 such that Ļ• āˆˆ (Ļ€k)(Ļ€k+1).
Observation 2.15. Ī½D(Ļ•) = 0 =ā‡’ Ļ• āˆˆ (Ļ€0)(Ļ€1) = k[X](Ļ€), i.e., Ļ•
does not vanish on all of D.
If Ļ• is rational and Ļ• = f
g , where f, g āˆˆ k[X], deļ¬ne
Ī½D(Ļ•) = Ī½D(f) āˆ’ Ī½D(g).
Case 2.General case: Ļ• āˆˆ k(X){0}, D āŠ† X arbitrary prime divisor. Choose
U āŠ† X open aļ¬ƒne such that
ā€¢ U is smooth;
21
Chapter 2. Some Algebraic Geometry tools
ā€¢ U āˆ© D = āˆ…;
ā€¢ D is a hypersurface: D = V(Ļ€) for some Ļ€ āˆˆ k[U] = OX(U)).
We have Ļ• āˆˆ k(X) = k(U). Deļ¬ne Ī½D(Ļ•) as in case 1.
Example 2.16. Let Ļ• = x
y āˆˆ k(x, y) = k(A2) we have that
div(Ļ•) =
DāŠ†A2
prime
Ī½D
x
y
D
where Ī½D
x
y is 0 for all divisors D except for L1 = V(x), where the order
of vanishing is 1, and L2 = V(y), where Ī½L2 (Ļ•) = āˆ’1.
2.4 Rational diļ¬€erential forms and canonical divi-
sors.
A rational diļ¬€erential form on X is intuitively f1dg1 + Ā· Ā· Ā· + frdgr, where fi
and gi are rational functions on X. Formally:
Deļ¬nition 2.17. A rational diļ¬€erential form on X is an equivalence class
of pairs (U, Ļ•) where U āŠ† X is open and Ļ• āˆˆ ā„¦X(U) and
(U, Ļ•) āˆ¼ (U , Ļ• ) ā‡ā‡’ Ļ•|Uāˆ©U = Ļ• Uāˆ©U
We can deļ¬ne the divisor of a rational diļ¬€erential form. If Ļ‰ is a rational
diļ¬€erential form on X, then div(Ļ‰) āˆˆ Div(X) is called a canonical divisor.
The canonical divisors form a linear equivalence class on X, denoted KX.
We are going to present another way to deļ¬ne the canonical divisor of a
compact complex manifold.
Deļ¬nition 2.18. Let X be a compact complex manifold and Yi āŠ‚ X
codimension 1 subvarieties then we deļ¬ne the canonical divisor of X
KX = niYi
where ni āˆˆ Z.
From this deļ¬nition we can glimpse that there exist a connection between
the canonical divisors and the canonical bundle of a complex manifold but
only later we will clarify this link.
A diļ¬€erential form Ļˆ on X is regular if āˆ€x āˆˆ X, there is an open
neighborhood U such that x āˆˆ U and Ļˆ|U agrees with t
i=1 gidfi, where
fi, gi āˆˆ OX(U). In other words, viewing Ļˆ as a section of the cotangent
bundle of X, the section map is regular.
22
Chapter 2. Some Algebraic Geometry tools
Example 2.19. The diļ¬€erential form
Ļˆ = 2xd(xy)
= 2x(xdy + ydx)
= 2x2
dy + 2xydy
is a regular diļ¬€erential form in A2.
Deļ¬nition 2.20. For U āŠ‚ X open, let ā„¦X(U) be the set of regular diļ¬€er-
ential forms on the variety U.
2.5 Dualizing sheaf
The following discussion is far from being a complete presentation about
sheaves. We will not enter in details and we will just give a sketchy presenta-
tion of them.
A sheaf of rings F on a topological space X is a functor from the category
of open subsets of X, where the morphisms are inclusions, to the category of
rings where the objects are rings and the morphisms are ring homomorphism,
satisfying the standard sheaf axioms. In particular, for an open subset U
we have F(U) is a ring and if U, V are both subsets of X such that U ā†’ V ,
then the induced morphism
F(U) ā†’ F(V )
is a ring homomorphism.
A ringed space is a pair (X, OX) where X is a topological space and OX is
a sheaf of unital rings. The sheaf OX is called the structure sheaf of the
ringed space (X, OX).
Deļ¬nition 2.21. Let (X, OX) be a ringed space. Let F be a sheaf of
OXāˆ’ modules. We say F is locally free if for every point x āˆˆ X there exists
a set I and an open neighbourhood x āˆˆ U āŠ‚ X such that F|U is isomorphic
to
i
āˆˆ OX|U
as an OX|U āˆ’module.
ā„¦X(U) is a module over OX(U). In fact, ā„¦X(U) is a sheaf of OX-modules.
On An, ā„¦X is the free OXāˆ’module generated by dx1, . . . , dxn.
Theorem 2.22. If X is smooth then the sheaf ā„¦(X) is a locally free
OXāˆ’module of rank d = dim X.
23
Chapter 2. Some Algebraic Geometry tools
We will not prove this fact but we have that the set of rational diļ¬€erential
forms forms a vector space over k(X). We recall the deļ¬nition of canonical
bundle that we have seen in Chapter 1 in order to analyse it in an algebraic
point of view.
For each p āˆˆ N, look at the sheaf Ī›pā„¦(X) of pāˆ’diļ¬€erentiable forms on
X, which assigns to open U āŠ† X the set of all regular pāˆ’forms, āˆ€x āˆˆ U
Ļ•(x): Ī›p
TxX ā†’ K
Locally these look like
fidgi1 āˆ§ Ā· Ā· Ā· āˆ§ dgip
Rational pāˆ’forms are deļ¬ned analogously.
Observation 2.23. The set of rational pāˆ’forms on X is a k(X)āˆ’vector space
of dimension n
p .
Deļ¬nition 2.24. Let X be a smooth nāˆ’dimensional, the canonical sheaf
(or dualizing sheaf ) of X is
Ļ‰X = Ī›n
ā„¦X.
Observation 2.25. The canonical sheaf satisfy the following properties:
ā€¢ Ļ‰X is locally free of rank 1.
ā€¢ The set of rational canonical nāˆ’forms is a vector space of dimension 1
over k(X).
Thus we have establish a connection between holomorphic sections and
divisors however we will better develop it in Chapter 4.
24
Chapter 3
A look to some complex
constructions
In this chapter we will equip smooth manifolds with a smooth linear complex
structure on each tangent space. The existence of this structure is a necessary,
but not suļ¬ƒcient, condition for a manifold to be a complex manifold. That
is, every complex manifold is an almost complex manifold, but not vice
versa. Almost complex structures have important applications in symplectic
geometry. Therefore we will also have a look to some aspect of symplectic
geometry in order to connect them to the complex world.
3.1 Almost complex structure
Deļ¬nition 3.1 (Almost complex structure). An almost complex structure
on a diļ¬€erentiable manifold M is a diļ¬€erentiable endomorphism of the tangent
bundle
J : T(M) ā†’ T(M)
such that
J2
x = āˆ’id, (3.1)
where Jx : Tx(M) ā†’ Tx(M).
A diļ¬€erentiable manifold with some ļ¬xed almost complex structure is called
an almost complex manifold.
Almost complex manifolds must be even dimensional. In fact the following
preposition applies.
Proposition 3.2. If M admits an almost complex structure, it must be
even-dimensional.
25
Chapter 3. A look to some complex constructions
Proof. This can be seen as follows. Suppose M is nāˆ’dimensional, and let
J : T(M) ā†’ T(M) be an almost complex structure. Since the determinant
det: GL(n, R) ā†’ Rāˆ—
A ā†’ det(A)
is a group homomorphism, we have that
det(J2
x) = det(Jx)2
.
Using (3.1) we obtain
det(Jx)2
= det(āˆ’id) = (āˆ’1)n
.
But if M is a real manifold, then det(J) is a real number, thus n must be
even if M has an almost complex structure.
For instance every a complex manifold has a natural structure of almost
complex manifold, while the vice versa is not always true.
Theorem 3.3. Every complex manifold has a canonical almost complex
structure.
Proof. Let M be a complex manifold of dimension n, and p āˆˆ M. Let
(U, (z1, . . . , zn)) be a holomorphic chart around p. In this case , if we set
xk := Rezk and yk := Imzk,
then (U, (x1, . . . , xn, y1, . . . , yn)) is a local chart of M, seen as a smooth
manifold.
Firstly set forall q āˆˆ U,
Jq
āˆ‚
āˆ‚xi q
=
āˆ‚
āˆ‚yi q
, Jq
āˆ‚
āˆ‚yi q
= āˆ’
āˆ‚
āˆ‚xi q
.
In order to conclude we will now show that J is globally well-deļ¬ned, in other
words we will prove that J does not depend on the choice of coordinates.
Let (U, (zk)k) and (V, (wk)k) holomorphic charts around p; set
zk := xk + iyk, wk = uk + ivk
for all k āˆˆ {1, . . . , n} and denote by J and J the almost complex struc-
ture on (U, (zk)k) and (V, (wk)k) respectively. If we think that xk =
xk(u1, . . . , un, v1, . . . , vn) and yk = yk(u1, . . . , un, v1, . . . , vn), in U āˆ© V we
have, ļ£±
ļ£²
ļ£³
āˆ‚
āˆ‚xk
= n
j=1
āˆ‚uj
āˆ‚xk
āˆ‚
āˆ‚uj
+
āˆ‚vj
āˆ‚xk
āˆ‚
āˆ‚vj
āˆ‚
āˆ‚yk
= n
j=1
āˆ‚uj
āˆ‚yk
āˆ‚
āˆ‚uj
+
āˆ‚vj
āˆ‚yk
āˆ‚
āˆ‚vj
. (3.2)
26
Chapter 3. A look to some complex constructions
The Cauchy-Riemann conditions are,
ļ£±
ļ£²
ļ£³
āˆ‚uj
āˆ‚xk
=
āˆ‚vj
āˆ‚yk
āˆ‚uj
āˆ‚yk
= āˆ’
āˆ‚vj
āˆ‚xk
(3.3)
Therefore,
J
āˆ‚
āˆ‚xk
= J
ļ£«
ļ£­
n
j=1
āˆ‚uj
āˆ‚xk
āˆ‚
āˆ‚uj
+
āˆ‚vj
āˆ‚xk
āˆ‚
āˆ‚vj
ļ£¶
ļ£ø (3.4)
=
n
j=1
āˆ‚uj
āˆ‚yk
āˆ‚
āˆ‚uj
+
āˆ‚vj
āˆ‚yk
āˆ‚
āˆ‚vj
(3.5)
=
āˆ‚
āˆ‚yk
(3.6)
= J
āˆ‚
āˆ‚xk
in U āˆ© V. (3.7)
Similarly,
J
āˆ‚
āˆ‚yk
= J
āˆ‚
āˆ‚yk
in U āˆ© V. (3.8)
Example 3.4. In order to make it clear here some simple examples of almost
complex manifolds.
ā€¢ Let (x, y) be the standard coordinates on R2. It is east to show that
J : R2 ā†’ R2, deļ¬ned by
J
āˆ‚
āˆ‚x
=
āˆ‚
āˆ‚y
, J
āˆ‚
āˆ‚y
= āˆ’
āˆ‚
āˆ‚x
,
satisļ¬es J2 = id. Therefore R2 admits an almost complex structure J.
Identifying R2 with C in the usual way, z = x + iy, we can see J as a
multiplication by i, i.e., a rotation of Ļ€
2 in the plane.
ā€¢ More in general, R2n admits an almost complex structure, for every inte-
ger n ā‰„ 0. In fact if we consider global coordinate (x1 . . . xn, y1, . . . , yn),
as we saw before,
J
āˆ‚
āˆ‚xi
=
āˆ‚
āˆ‚yi
, J
āˆ‚
āˆ‚yi
= āˆ’
āˆ‚
āˆ‚xi
,
is an almost complex structure.
27
Chapter 3. A look to some complex constructions
3.2 Symplectic manifolds
3.2.1 (Real) Symplectic manifolds
Deļ¬nition 3.5. Let M be a smooth 2nāˆ’dimensional manifold, a 2-form
Ļ‰ āˆˆ Ī›2(M) is said symplectic form if it is closed and non degenerate, i.e. if
satisļ¬es the two condition
1. dĻ‰ = 0;
2. Ļ‰p = 0 for every p āˆˆ M.
If M is a smooth manifold and Ļ‰ is a symplectic form on M, then the pair
(M, Ļ‰) is called symplectic manifold.
In other words a symplectic form is a section of the bundle of the 2-forms
on M such that:
1. dĻ‰ = 0 (analytic condition);
2. (TpM, Ļ‰P ) is a symplectic vector space for all p āˆˆ M (algebraic condi-
tion).
Example 3.6. Let x1, . . . , x2n be local coordinates around a point p āˆˆ R2n
and endow R2n with the 2-form
n
i=1
dxi
āˆ§ dxn+i
.
R2n, Ļ‰ is a symplectic manifold and the matrix of Ļ‰ in the base
āˆ‚
āˆ‚x1 p
, . . . ,
āˆ‚
āˆ‚x2n p
of TpM
is
0 In
āˆ’In 0
.
Example 3.7 (Cotangent bundles). Let N be a smooth manifold of dimension
n and let M = Tāˆ—N be its cotangent bundle. This has a natural symplectic
structure, which may be deļ¬ned locally as follows. Choose local coordinates
x1, . . . , xn on N. Then the 1āˆ’forms dx1, . . . , dxn provide a local trivialisation
of Tāˆ—N, so we obtain local coordinate functions Ī¾1, . . . , Ī¾n on the ļ¬bres of
Tāˆ—N. Thus M has local coordinates (x1, . . . , xn, Ī¾1, . . . , Ī¾n.) We may deļ¬ne
a 1-form
Īø =
n
i=1
Ī¾idxi
28
Chapter 3. A look to some complex constructions
locally on M and it turns out that this local deļ¬nition in fact deļ¬nes a global
one-form (the so-called ā€œLiouville formā€) on M. The exterior derivative
Ļ‰ = dĪø
is a natural symplectic form on M. Clearly it is closed (since it is exact), and
it is nondegenerate because in local coordinates has the following expression
Ļ‰ =
n
i=1
dĪ¾i
āˆ§ dxi
.
Example 3.8. The 2-sphere S2 endowed with the symplectic form (the volume
form)
Ļ‰ = sin ĪødĪø āˆ§ dĻ†.
In general it is not true that the 2nāˆ’sphere S2n, with n > 1 is a symplectic
manifold, for the proof of this fact see Corollary 5.11.
3.2.2 (Complex) Symplectic manifolds
A complex symplectic manifold is a pair (M, Ļ‰) consisting of a complex
manifold M and a holomorphic 2-form Ļ‰ (of type (2, 0)) such that:
1. Ļ‰ is closed, i.e., dĻ‰ = 0;
2. Ļ‰ is non degenerate, i.e., the associated linear map
TpM ā†’ Tāˆ—
p M
v ā†’ (Ļ‰)p(v, āˆ’)
from the holomorphic tangent space to the holomorphic cotangent
space, is an isomorphism at each point p āˆˆ M.
These are sometimes also referred to as holomorphic symplectic manifolds.
3.3 Compatible almost complex structures
Deļ¬nition 3.9. Let (M, Ļ‰) be a symplectic manifold and p a point of
M. An almost complex structure J on M is called compatible with Ļ‰ (or
Ļ‰āˆ’ compatible) if
ā€¢ Ļ‰(Ju, Jv) = Ļ‰(u, v) āˆ€u, v āˆˆ TpM;
ā€¢ Ļ‰(Ju, u) > 0, āˆ€u āˆˆ TpM such that u = 0.
29
Chapter 3. A look to some complex constructions
Remark. g: TpM Ɨ TpM ā†’ R deļ¬ned by
g(u, v) = Ļ‰(Ju, v), u, v āˆˆ TpM,
is a Riemannian metric. This bilinear form is simmetric since
g(w, v) = Ļ‰(w, Jv) = Ļ‰(Jw, J2
v) = Ļ‰(Jw, āˆ’v) = Ļ‰(v, Jw) = g(v, w),
where v, w āˆˆ TpM. By deļ¬nition of Ļ‰ it is positive deļ¬nite, i.e. given
u āˆˆ TpM, we have
g(v, v) = Ļ‰(Ju, u) > 0.
Remark. The triple (Ļ‰, J, g) is said a compatible triple. Moreover any one of
J, Ļ‰, g can be written in terms of the other two by the following formulas.
ā€¢ g(u, v) = Ļ‰(u, Jv);
ā€¢ Ļ‰(u, v) = g(Ju, v);
ā€¢ J(u) = Ėœgāˆ’1(ĖœĻ‰(u)),
where ĖœĻ‰: TM ā†’ Tāˆ—
M, Ėœg: TM ā†’ Tāˆ—
M
u ā†’ Ļ‰(u, Ā·) u ā†’ g(u, Ā·).
Theorem 3.10. On any symplectic manifold (M, Ļ‰), there exists almost
complex structures J that are compatible with Ļ‰.
Proof. First we show this is true for a symplectic vector space V. Let g be a
Riemannian metric on V and deļ¬ne A by
Ļ‰(u, v) = g(Au, v).
Since Ļ‰ is skew-symmetric and g is a metric, thus it is symmetric, we have
Ļ‰(u, v) = Ļ‰(āˆ’v, u) = g(āˆ’Av, u) = g(u, āˆ’Av). (3.9)
By deļ¬nition of adjoint matrix and by eqution (3.9) we have
g(u, Aāˆ—
v) = g(Au, v) = g(u, āˆ’Av),
then
Aāˆ—
= āˆ’A.
Furthermore AAāˆ— is
ā€¢ symmetric, i.e. (AAāˆ—)āˆ— = AAāˆ—;
ā€¢ positive deļ¬nite, i.e. g(AAāˆ—v, v) = g(Aāˆ—v, Aāˆ—v) > 0, āˆ€v = 0.
30
Chapter 3. A look to some complex constructions
Since every symmetric matrix B can be factored into B = Qāˆ†Qāˆ’1 where Q
is orthogonal matrix, i.e. Qāˆ’1 = QT and āˆ† = diag(Ī»1, . . . , Ī»2n), AAāˆ— can be
written as
AAāˆ—
= Qāˆ†Qāˆ’1
,
moreover Ī»i > 0, for every i = 1, . . . , 2n because AAāˆ— is positive deļ¬ned. Set
J := (
āˆš
AAāˆ—)āˆ’1
A. (3.10)
The factorization A =
āˆš
AAāˆ— is called the polar decomposition of A. Using
the diagonalization we can write J = Q
āˆš
āˆ†Qāˆ’1A, this implies that JJāˆ— = id
and Jāˆ— = āˆ’J, in fact
JJāˆ—
= Q(
āˆš
āˆ†)āˆ’1
Qāˆ’1
A Ā· Aāˆ—
(Qāˆ’1
)āˆ—
((
āˆš
āˆ†)āˆ’1
)āˆ—
Qāˆ—
,
since ((
āˆš
āˆ†)āˆ’1)āˆ— = (
āˆš
āˆ†)āˆ’1 and Q is orthogonal,
= Q(
āˆš
āˆ†)āˆ’1
Qāˆ’1
Ā· Qāˆ†Qāˆ’1
Ā· Q(
āˆš
āˆ†)āˆ’1
Qāˆ’1
= Q(
āˆš
āˆ†)āˆ’1
āˆ†(
āˆš
āˆ†)āˆ’1
Qāˆ’1
= id,
where the last equality applies since the diagonal matrices commute. Thus
J2 = āˆ’id is an almost complex structure and
ā€¢ Ļ‰(Ju, Jv) = g(AJu, Jv) = g(JAu, Jv) = g(Au, v) = Ļ‰(u, v)
where the second equality applies since A commutes with
āˆš
AAāˆ— and
thus J commutes with
āˆš
AAāˆ—;
ā€¢ Ļ‰(u, Ju) = g(āˆ’JAu, u) = g(
āˆš
AAāˆ—u, u) > 0,
so J is compatible.
Finally we can use this construction to each tangent space TpM of every point
of M. It turns out that J is globally well-deļ¬ned since the polar decomposition
is canonical after a choice of Riemannian metric, i.e. it does not depend on
the choice of Q nor of the ordering of the eigenvalues {Ī»1, . . . , Ī»2n}. Hence J
is smooth.
31
Chapter 4
Calaby-Yau manifolds
Currently, research on Calabi-Yau manifolds is a central focus in both
mathematics and mathematical physics. The spread of this study is partially
propelled by the prominent role of the Calabi-Yau in superstring theories.
Many beautiful properties of Calabi-Yau manifolds have been discovered and
nowadays this subject represents an extremely active research ļ¬eld.
In this chapter we will give the deļ¬nition of Calabi-Yau manifolds and we will
provide many examples, using the tools we have previously studied. Moreover
we will try to place this type of manifold in the complex geometry world and
we will investigate how they interact with others complex manifolds.
4.1 Symplectic Calabi-Yau manifolds
Following [JD08] we ļ¬rstly analyse Calabi-Yau manifolds from the symplectic
point of view.
Deļ¬nition 4.1. Let M be a symplectic real manifold of dimension 2n. M
is a symplectic Calabi-Yau if its canonical bundle
K ā†’ M
is trivial.
There are many examples of compact symplectic Calabi-Yau manifolds.
Firstly, many nilmanifolds are of this type.
Example 4.2 (Nilmanifolds). A Nilmanifold is a compact quotient space of a
nilpotent Lie group modulo a closed discrete subgroup, or (equivalently) a
homogeneous space with a nilpotent Lie group acting transitively on it.
The most famous example of a nilmanifold is the so-called Kodaira Thurston
surface, which will be discussed in Chapter 5.
32
Chapter 4. Calaby-Yau manifolds
4.2 Complex Calabi-Yau manifolds
We give now the deļ¬nition of a Calabi-Yau manifold from the complex point
of view.
Deļ¬nition 4.3. A complex Calabi-Yau is a complex compact manifold such
that its canonical bundle
K ā†’ M
is trivial.
Remark. Here the canonical bundle is a holomorphic line bundle while in
the case of the symplectic Calabi-Yau manifold it does not exist a notion of
holomorphic and the triviality is a topological condition.
In the following sections will now give some examples of complex Calabi-
Yau manifolds.
4.2.1 1-dimensional Calabi-Yau
In one complex dimension, the only compact examples are tori, see Example
1.10.
In order to expand the horizon of the 1-dimensional Calabi-Yau manifolds
we will show now how complex tori C/Ī› can also be viewed as cubic curves.
These cubic curves are called elliptic despite not being ellipses, due to a
connection between them and the arc length of an actual ellipse.
Elliptic curves
Deļ¬nition 4.4 (Elliptic curves over a ļ¬eld K). An elliptic curve is an
irreducible cubic in P2(K) which is non singular.
By an appropriate change of variables, a general elliptic curve over a ļ¬eld
K with ļ¬eld characteristic diļ¬€erent from 2, 3, can be written in the aļ¬ƒne
form
y2
= 4x3
āˆ’ ax + b,
with a3 āˆ’ 27b2 = 0 (condition of non singularity).
Let K = C we will now see how the points of an elliptic curve form a torus.
The meromorphic functions on a complex torus relate the manifold to a
cubic curve. Given a lattice Ī›, the meromorphic functions f : C/Ī› ā†’ C on
the torus, where C = C āˆŖ {āˆž}, are naturally identiļ¬ed with the Ī›-periodic
meromorphic functions f : C ā†’ C on the plane. Let Ī› be a lattice, a function
f such that
f(z + Ļ‰) = f(z),
for any z āˆˆ C and Ļ‰ āˆˆ Ī›, is called elliptic function.
It follows that if Ī› = Ļ‰1Z āŠ• Ļ‰2Z,
f(z + Ļ‰1) = f(z + Ļ‰2) = f(z),
33
Chapter 4. Calaby-Yau manifolds
and we call Ļ‰1, Ļ‰2 the periods of f.
A non-trival example of an elliptic function is the Weierstrass ā„˜-function.
Given a lattice Ī›, we deļ¬ne the Weierstrass ā„˜-function by
ā„˜(z) =
1
z2
+
Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)2
āˆ’
1
Ļ‰2
, (4.1)
where z āˆˆ C, z /āˆˆ Ī›.
Proposition 4.5. The ā„˜(z) function satisļ¬es the following properties:
1. ā„˜(z) is well deļ¬ned in the sense that the sum converges absolutely and
uniformly on compact sets ā„¦ such that ā„¦ āˆ© Ī› = āˆ….
2. ā„˜(āˆ’z) = ā„˜(z) for all z āˆˆ C.
3. The derivative
ā„˜ (z) = āˆ’2
Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)3
(4.2)
has period Ī›.
4. ā„˜(z) has period Ī›.
Proof. We will just sketch the proofs. For 1 it is convenient to state the
following lemma.
Lemma 4.6. If k > 2 then the following series
Ļ‰āˆˆĪ›{0}
1
|Ļ‰|k
(4.3)
converges over the entire lattice Ī›.
The convergence of the series is proven using an integral comparison test
and estimates on the diagonal of the fundamental parallelogram for Ī›,
Ī  = {x1Ļ‰1 + x2Ļ‰2 : x1, x2 āˆˆ [0, 1]} .
Then the absolute and uniform convergence is a consequence of another
estimate and the exclusion of ļ¬nitely many terms.
(2) If Ļ‰ āˆˆ Ī› then it is also true that āˆ’Ļ‰ āˆˆ Ī›, by multiplying by āˆ’1. Hence
ā„˜(āˆ’z) =
1
(āˆ’z)2
+
Ļ‰āˆˆĪ›{0}
1
(āˆ’z āˆ’ Ļ‰)2
āˆ’
1
Ļ‰2
=
1
(z)2
+
āˆ’Ļ‰āˆˆĪ›{0}
1
(āˆ’z + Ļ‰)2
āˆ’
1
(āˆ’Ļ‰)2
=
1
(z)2
+
āˆ’Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)2
āˆ’
1
Ļ‰2
= ā„˜(z).
34
Chapter 4. Calaby-Yau manifolds
(3) The sum ā„˜ (z) converges uniformly and absolutely by the lemma with
simple comparison to 1
|w|3 . Moreover if Ļ‰, Ļ āˆˆ Ī› then it is also true that
Ļ‰ āˆ’ Ļ āˆˆ Ī›; thus
ā„˜ (z + Ļ) = āˆ’2
Ļ‰āˆˆĪ›{0}
1
(z + Ļ āˆ’ Ļ‰)3
= āˆ’2
Ļ‰āˆ’ĻāˆˆĪ›{0}
1
(z āˆ’ (Ļ‰ āˆ’ Ļ))3
= ā„˜ (z).
(4) Using 3 we ļ¬nd out that the derivative of ā„˜(z + Ļ‰) āˆ’ ā„˜(z) is equal to
zero, if z /āˆˆ Ī›. Thus there exists a constant cĻ‰ such that
ā„˜(z + Ļ‰) āˆ’ ā„˜(z) = cĻ‰,
for all z /āˆˆ Ī›. Setting z = āˆ’Ļ‰
2 we obtain
cĻ‰ = ā„˜ āˆ’
Ļ‰
2
āˆ’ ā„˜ āˆ’
Ļ‰
2
.
By the fact that ā„˜(z) is even we can conclude that
cĻ‰ = 0.
Therefore
ā„˜(z + w) = ā„˜(z) (4.4)
for all w āˆˆ Ī›.
Remark. Let Ī› = Ļ‰1Z āŠ• Ļ‰2Z. Since ā„˜ satisļ¬es the condition (4.4), we say
that ā„˜ is a doubly periodic function, as it has two independent periods Ļ‰1
and Ļ‰2.
We will now see some functions that will appear in the Laurent expansion
of the Weierstrass ā„˜(z)-function for Ī›.
Deļ¬nition 4.7. Let Ī› be a lattice and k be an integer, the Eisenstein series
are functions deļ¬ned by
Gk(Ī›) =
Ļ‰āˆˆĪ›{0}
1
Ļ‰k
, (4.5)
for k > 2, even.
Remark. The Eisenstein series satisfy the homogeneity condition
Gk(mĪ›) = māˆ’k
Gk(Ī›)
for all m āˆˆ C{0}.
35
Chapter 4. Calaby-Yau manifolds
Theorem 4.8. Let ā„˜ be the Weierstrass function with respect to a lattice Ī›.
Then
(i) The Laurent expansion of ā„˜ is
ā„˜(z) =
1
z2
+
āˆž
n=1
(2n + 1)G2n+2(Ī›)z2n
, (4.6)
for all z such that 0 < |z| < inf{|Ļ‰| : Ļ‰ āˆˆ Ī›{0}}.
(ii) The functions ā„˜ and ā„˜ satisfy the relation
(ā„˜ (z))2
= 4(ā„˜(z))3
āˆ’ g2(Ī›)ā„˜(z) āˆ’ g3(Ī›) (4.7)
where g2(Ī›) = 60G4(Ī›) and g3(Ī›) = 140G6(Ī›).
(iii) Let Ī› = Ļ‰1Z āŠ• Ļ‰2Z and let Ļ‰3 = Ļ‰1 + Ļ‰2. Then the cubic equation
satisļ¬ed by ā„˜ and ā„˜ , y2 = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›) is
y2
= 4(x āˆ’ e1)(x āˆ’ e2)(x āˆ’ e3), (4.8)
where ei = ā„˜ Ļ‰i
2 for i = 1, 2, 3.
This equation is non singular, meaning its right side has distinct roots.
Proof. (i) Firstly we observe how the geometric series squares i.e.,
āˆž
n=0
qn
2
=
āˆž
n=0
qn
āˆž
k=0
qk
=
āˆž
k=0
qk
+ q
āˆž
k=0
qk
+ Ā· Ā· Ā· + qn
āˆž
k=0
qk
+ . . .
= 1 + q + q2
+ . . . + q + q2
+ q3
+ . . . + Ā· Ā· Ā· + qn
+ qn+1
+ qn+2
+ . . . + . . .
= 1 + 2q + 3q2
+ 4q3
+ Ā· Ā· Ā· + (n + 1)qn
+ . . .
=
āˆž
n=0
(n + 1)qn
,
where q āˆˆ C. As a result, if 0 < |z| < inf{|Ļ‰| : Ļ‰ āˆˆ Ī›{0}}, we obtain that
ā„˜(z) =
1
z2
+
Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)2
āˆ’
1
Ļ‰2
=
1
z2
+
Ļ‰āˆˆĪ›{0}
1
Ļ‰2
1
1 āˆ’ z
Ļ‰
2 āˆ’ 1
=
1
z2
+
Ļ‰āˆˆĪ›{0}
1
Ļ‰2
ļ£«
ļ£­
āˆž
n=0
z
Ļ‰
n 2
āˆ’ 1
ļ£¶
ļ£ø
36
Chapter 4. Calaby-Yau manifolds
and by using the result of the geometric square series we get
=
1
z2
+
Ļ‰āˆˆĪ›{0}
1
Ļ‰2
āˆž
n=0
(n + 1)
z
Ļ‰
n
āˆ’ 1
=
1
z2
+
Ļ‰āˆˆĪ›{0}
1
Ļ‰2
1 +
āˆž
n=1
(n + 1)
z
Ļ‰
n
āˆ’ 1
=
1
z2
+
Ļ‰āˆˆĪ›{0}
1
Ļ‰2
āˆž
n=1
(n + 1)
z
Ļ‰
n
=
1
z2
+
Ļ‰āˆˆĪ›{0}
āˆž
n=1
(n + 1)
zn
Ļ‰2+n
.
Convergence results allow the resulting double sum to be rearranged, and
then the inner sum cancels when n is odd. In fact the last expression is equal
to
1
z2
+
āˆž
n=1
(n + 1)
Ļ‰2
1
zn
Ļ‰n
1
+
āˆž
n=1
(n + 1)
(āˆ’Ļ‰1)2
zn
(āˆ’Ļ‰1)n + . . .
+
āˆž
n=1
(n + 1)
Ļ‰2
i
zn
Ļ‰n
i
+
āˆž
n=1
(n + 1)
(āˆ’Ļ‰i)2
zn
(āˆ’Ļ‰i)n + . . .
with Ļ‰i āˆˆ Ī›{0}. Therefore we can conclude that
ā„˜(z) =
1
z2
+
āˆž
n=1 Ļ‰āˆˆĪ›{0}
(2n + 1)
Ļ‰2n+2
z2n
and by defenition of Einstein series
=
1
z2
+
āˆž
n=1
(2n + 1)G2n+2(Ī›)z2n
.
(ii) The series expansions of ā„˜ and ā„˜ from the previous proposition leads to
ā„˜(z) =
1
z2
+ 3G4z2
+ 5G6z4
+ O(z5
)
ā„˜ (z) = āˆ’
2
z3
+ 6G4z + 20G6z3
+ O(z4
)
and taking the cube and the square we have
ā„˜(z)3
=
1
z6
+
9G4
z2
+ 15G6 + O(z2
)
ā„˜ (z)2
=
4
z6
āˆ’
24G4
z2
āˆ’ 80G6 + O(z2
).
37
Chapter 4. Calaby-Yau manifolds
Deļ¬ne a function f as follows,
f(z) = (ā„˜ (z))2
āˆ’ 4(ā„˜(z))3
+ g2(Ī›)ā„˜(z) + g3(Ī›).
Since f is an elliptic function, as a polynomial of ā„˜ and ā„˜ , with no poles,
for the First Liouville Theorem1 it is constant. If we substitute the value we
found above, we get
f(z) =
4
z6
āˆ’
24G4
z2
āˆ’ 80G6 āˆ’ 4
1
z6
+
9G4
z2
+ 15G6 + g2(Ī›)
1
z2
+ 3G4z2
+ 5G6z4
+ g3(Ī›) + O(z2
)
=
4
z6
āˆ’
24G4
z2
āˆ’ 80G6 āˆ’ 4
1
z6
+
9G4
z2
+ 15G6 + 60G4(Ī›)(
1
z2
+
+ 3G4z2
+ 5G6z4
) + 140G6(Ī›) + O(z2
)
= 180 G2
4z2
+ 300 G4G6z4
+ O(z2
)).
Noticing that the last right-hand side grows at the order of z2, and considering
that lim
zā†’0
= O(z2), we conclude that f(z) ā‰” 0.
(iii) We start observing that ā„˜(z) and ā„˜ (z) have a double pole at each
Ļ‰ āˆˆ Ī›. Moreover since ā„˜ (z) is doubly periodic, i.e. ā„˜ (z + Ļ‰) = ā„˜ (z), letting
Ī› = Ļ‰1Z āŠ• Ļ‰2Z, we see that if z = āˆ’Ļ‰i
2 , and Ļ‰ = Ļ‰i
2 ,
ā„˜ (
Ļ‰i
2
) = ā„˜ (āˆ’
Ļ‰i
2
) (4.9)
for i = 1, 2, 3. Secondly, since ā„˜ is odd,
ā„˜ (āˆ’
Ļ‰i
2
) = āˆ’ā„˜ (
Ļ‰i
2
). (4.10)
Equation (4.9) together with (4.10) allow to conclude that zi = Ļ‰i
2 are points
of order 2 with ā„˜ (zi) = 0, for i = 1, 2, 3. The relation between ā„˜(z) and
ā„˜ (z) from (ii) shows that the corresponding values xi = ā„˜(zi) for i = 1, 2, 3
are roots of the cubic polynomial
pĪ›(x) = 4x3
āˆ’ g2(Ī›)x āˆ’ g3(Ī›),
so it factors as claimed. Each xi is a double value of ā„˜ since, as we have seen,
ā„˜ (zi) = 0, furthermore ā„˜ has degree 2, meaning it takes each value twice
counting multiplicity, this makes the three xi distinct. That is, the cubic
polynomial pĪ› has distinct roots.
1
Theorem 4.9 (First Liouville Theorem). If an elliptic function has no poles, then it is
constant.
38
Chapter 4. Calaby-Yau manifolds
The isomorphism between complex tori and complex elliptic curves
Part (ii) of Theorem 4.8 shows that the map
C/Ī› ā†’ C2
z ā†’ ā„˜(z), ā„˜ (z)
takes nonlattice points of C to points (x, y) āˆˆ C2 satisfying the nonsingular
cubic equation of part (iii), y2 = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›). Moreover this
application extends to all z āˆˆ C by mapping lattice points to a suitably
deļ¬ned point at inļ¬nity, but we will see the details in the following theorem.
Theorem 4.10. Let Ī› be a lattice on C and E be the elliptic curve
E = {(x, y) āˆˆ C2
| y2
= 4x3
āˆ’ g2(Ī›)x āˆ’ g3(Ī›)}.
Then the map Ļ† deļ¬ned by
C/Ī› ā†’ E
z ā†’ ā„˜(z), ā„˜ (z) if z = 0
0 ā†’ {āˆž},
is a group isomorphism.
Before going into the proof of the theorem we need to state the following
result.
Theorem 4.11. Let f be an elliptic function for the lattice Ī› and let Ī  be
a fundamental parallelogram for Ī›. If f is not constant then
f : C ā†’ C āˆŖ {āˆž}
is surjective. If n is the sum of the orders of the poles2 of f in Ī  and z0 āˆˆ C,
then f(z) = z0 has n solutions (counting multiplicities).
We can now see the proof of Theorem 4.10, which allows us to understand
that complex tori (Riemann surfaces, complex analytic objects) are equivalent
to elliptic curves (solution sets of cubic polynomials, algebraic objects).
Proof. We need to show that Ļ† is:
1. injective. Suppose
ā„˜(z1), ā„˜ (z1) = ā„˜(z2), ā„˜ (z2) , (4.11)
where z1, z2 āˆˆ C/Ī› are such that z1 = z2 mod Ī›. The only poles for
ā„˜(z) belong to Ī›, thus we can distinguish to cases.
2
If f is an elliptic function, we can write it as a Laurent series expansion around Ļ‰ āˆˆ Ī›
as
f(z) = ar(z āˆ’ Ļ‰)r
+ ar+1(z āˆ’ Ļ‰)r
+ 1 + . . .
with ar = 0. We deļ¬ne the order of f at Ļ‰ as ordĻ‰f = r.
39
Chapter 4. Calaby-Yau manifolds
ā€¢ If z1 is a pole of ā„˜ then z1 āˆˆ Ī›. Since (4.11) applies, z2 is a pole
of ā„˜, i.e., z2 āˆˆ Ī›. This implies z1 = z2 mod Ī›.
ā€¢ Now suppose z1 is not a pole of ā„˜, i.e., z1 /āˆˆ Ī›. Then h(z) =
ā„˜(z) āˆ’ ā„˜(z1) has a double pole at z = 0 and no other poles in
Ī  = {x1Ļ‰1 + x2Ļ‰2 : x1, x2 āˆˆ [0, 1]} . By Theorem 4.11, h(z) has
exactly two zeros (counting multiplicities).
ā€“ Suppose z1 = Ļ‰i
2 for some i. From the proof of (iii) of Theorem
4.8 we know that ā„˜ (Ļ‰i
2 ) = 0, so z1 is a double root of h(z)
hence the only root. Thus z2 = z1.
ā€“ Suppose z1 is not of the form Ļ‰i
2 . Since h(āˆ’z1) = h(z1) = 0
(by evenness of ā„˜) and since z1 = z2 mod Ī›, two zeros of h
are z1 and z2 = āˆ’z1 mod Ī›. But
y = ā„˜ (z2) = ā„˜ (āˆ’z1) = āˆ’ā„˜ (z1) = āˆ’y.
Hence ā„˜ (z1) = y = 0. But ā„˜ (z) has only a triple pole, thus
has only three zeros in Ī . But from the proof of (iii) of
Theorem 4.8, we know that these zeros occur at Ļ‰i
2 , hence
a contradiction since z = Ļ‰i
2 . Thus z1 = z2 mod Ī› and Ļ† is
injective.
2. surjective. Let (x, y) āˆˆ E; we need to prove that there exists z āˆˆ C
such that Ļ†(z) = (x, y), i.e., that x = ā„˜(z) and y = ā„˜ (z).
Since ā„˜(z) āˆ’ x has a double pole, Theorem 4.11 implies it has zeros,
hence there exists z āˆˆ C such that ā„˜(z) = x. The elliptic equation in
the (ii) part of Theorem 4.8 implies that ā„˜ (z)2 = y2, so ā„˜ (z) = Ā±y.
ā€¢ If ā„˜ (z) = y we are done.
ā€¢ If ā„˜ (z) = āˆ’y, then by the evenness of the ā„˜ function, ā„˜ (āˆ’z) = y
and ā„˜(āˆ’z) = x, so āˆ’z ā†’ (x, y).
Hence Ļ† is onto.
3. a group homomorphism. We need to show that
Ļ†(z1 + z2) = Ļ†(z1) + Ļ†(z2),
where z1, z2 āˆˆ C. The map Ļ† transfers the group law from the complex
torus to the elliptic curve. The proof is not trivial and it is well devel-
oped in [Kna92], thus we will not focus on the demonstration and we
will try to understand the group law on the curve.
40
Chapter 4. Calaby-Yau manifolds
Let z1 + Ī› and z2 + Ī› be nonzero points of the torus. The image points
(ā„˜(z1), ā„˜ (z1)) and (ā„˜(z2), ā„˜ (z2)) on the curve determine a secant or tangent
line of the curve in C2, ax + by + c = 0. Consider the function
f(z) = aā„˜(z) + bā„˜ (z) + c.
This is meromorphic on C/Ī›. When b = 0 it becomes
f(z) = a
1
z2
+
Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)2
āˆ’
1
Ļ‰2
+ b āˆ’ 2
Ļ‰āˆˆĪ›{0}
1
(z āˆ’ Ļ‰)3
+ c. (4.12)
As we can see it has a triple pole at 0 + Ī› and zeros at z1 + Ī› and z2 + Ī›.
One could also prove that its third zero is at the point z3 + Ī› such that
z1 + z2 + z3 + Ī› = 0 + Ī›
in C/Ī›. When b = 0, f has a double pole at 0 + Ī› and zeros at z1 + Ī› and
z2 + Ī›, furthermore
z1 + z2 + Ī› = 0 + Ī›
in C/Ī›. In this case let z3 = 0+Ī› so that again z1 +z2 +z3 +Ī› = 0+Ī›, and
since the line is vertical view it as containing the inļ¬nite point (ā„˜(0), ā„˜ (0))
whose second coordinate arises from a pole of higher order than the ļ¬rst.
Therefore for any value of b the elliptic curve points on the line ax+by+c = 0
are the points
(xi, yi) = ā„˜(zi), ā„˜ (zi))
for i = 1, 2, 3.
Since z1 + z2 + z3 + Ī› = 0 + Ī› on the torus in all cases, the resulting group
law on the curve is:
ā€¢ The identity element of the curve is the inļ¬nite point;
ā€¢ collinear triples on the curve sum to zero.
Our aim now is to ļ¬nd an addition law and a duplication law for the
function ā„˜.
Consider the aļ¬ƒne equation of the elliptic curve and the aļ¬ƒne equation of a
line:
E : y2
= 4x3
āˆ’ g2x āˆ’ g3, L: y = mx + b.
If a point (x, y) āˆˆ C2 lies on E āˆŖ L then its x-coordinate satisļ¬es the cubic
polynomial obtained by substituting mx + b for y in the equation of E,
4x3
āˆ’ m2
x2
+ āˆ’b2
āˆ’ 2mbx āˆ’ g2x āˆ’ g3 = 0.
Thus, given three points collinear points (x1, y1), (x2, y2), and (x3, y3) on
the curve, necessarily
x1 + x2 + x3 =
m2
4
,
41
Chapter 4. Calaby-Yau manifolds
where m is the slope of their line,
m =
ļ£±
ļ£²
ļ£³
y1āˆ’y2
x1āˆ’x2
x1 = x2
12x2
1āˆ’g2
2y1
x1 = x2
(4.13)
A slight restatement is that
x3 =
m2
4
āˆ’ x1 āˆ’ x2, m as above.
(And also y3 = m(x3 āˆ’ x1) + y1.)
These results translate back to the desired addition law and duplication
law for the Weierstrass ā„˜-function. Since the three points on the curve are
collinear, we have for some z1 + Ī›, z2 + Ī› āˆˆ C/Ī›,
(x1, y1) = ā„˜(z1), ā„˜ (z1) ,
(x2, y2) = ā„˜(z2), ā„˜ (z2) ,
(x3, y3) = ā„˜(āˆ’z1 āˆ’ z2), ā„˜ (āˆ’z1 āˆ’ z2) .
But ā„˜ is even and ā„˜ is odd, so that in fact
(x3, y3) = ā„˜(z1 + z2), āˆ’ā„˜ (z1 + z2) .
That is,
ā„˜(z1 + z2) =
1
4
ā„˜ (z1) āˆ’ ā„˜ (z2)
ā„˜(z1) āˆ’ ā„˜(z2)
2
āˆ’ ā„˜(z1) āˆ’ ā„˜(z2) ifz1 + Ī› = Ā±z2 + Ī›,
ā„˜(2z) =
1
4
12ā„˜(z)2 āˆ’ g2
2ā„˜ (z)
2
āˆ’ 2ā„˜(z) ifz /āˆˆ
1
2
Ī› + Ī›.
In order to conclude this paragraph we outline that every elliptic curve
over C comes from a torus. That is, given an elliptic curve E, then we can
produce a lattice Ī› unique up to some homothetic equivalence. We point
out that not only does every complex torus C/Ī› lead via the Weierstrass
ā„˜-function to an elliptic curve
y2
= 4x3
āˆ’ a2x āˆ’ a3, a3
2 āˆ’ 27a2
3 = 0
with a2 = g2(Ī›) and a3 = g3(Ī›), but the converse holds as well.
Deļ¬nition 4.12 (Homothetic Lattices). Let Ī› = ZĻ‰1 + ZĻ‰2 be a lattice in
C. We deļ¬ne
Ļ„ :=
Ļ‰1
Ļ‰2
.
Since Ļ‰1 and Ļ‰2 are linearly independent over R, Ļ„ cannot be real. Hence,
by switching Ļ‰1 and Ļ‰2 if necessary, we can assume the imaginary part
42
Chapter 4. Calaby-Yau manifolds
Im(Ļ„) > 0, i.e., Ļ„ lies in the upper half plain H = {x + iy āˆˆ C | y > 0}. Now
if we let Ī›Ļ„ = ZĻ„ + Z, then Ī› is homothetic to Ī›Ļ„ , that is
Ī› = Ī»Ī›Ļ„
for some Ī» āˆˆ C. In this case Ī» = Ļ‰2.
Finally we can conclude thanks to the following result.
Theorem 4.13. Let y2 = 4x3 āˆ’ Ax + b deļ¬ne an elliptic curve E over C.
Then there exists a lattice Ī› such that g2(Ī›) = A and g3(Ī›) = B and there
is an isomorphism of groups C/Ī› = E.
Observation 4.14. The existence of such a lattice is a homothetic equivalence,
that is, if we ļ¬nd Ī› that works, then any Ī› = Ī»Ī› for Ī» āˆˆ C will suļ¬ƒce.
There are several approaches to proving the statement but we will not
go any further. The reader is referred to [Raa10] for more details.
We have decided to point out the connection between complex tori and
complex elliptic curves in order to give diļ¬€erent example of 1-dimensional
Calabi-Yau manifolds; moreover this turned out to be a good tools to analyse
the same object from the algebraic point of view.
4.2.2 2-dimensional Calabi-Yau
Calabiā€“Yau manifolds of dimension two are two-dimensional complex tori
and complex K3 surfaces, most of the latter are not algebraic. This means
that they cannot be embedded in any projective space as a surface deļ¬ned
by polynomial equations.
K3 surfaces
Deļ¬nition 4.15. A K3 surface is a (smooth) surface X which is simply
connected and has trivial canonical bundle.
We will now present a deļ¬nition of K3 surfaces which is slightly diļ¬€erent
from the standard one. Before going into the matter we need to understand
the following deļ¬nition.
Deļ¬nition 4.16 (Rational double points). A complex surface X has rational
double points if the dualizing sheaf Ļ‰X is locally free, and if there is a
resolution of singularities Ļ€: X ā†’ X such that
Ļ€āˆ—
Ļ‰X = Ļ‰X
= OX
KX
.
This means that for every P āˆˆ X there exists a neighborhood U of P
and a holomorphic 2-form
Ī± = Ī±(z1, z2)dz1 āˆ§ dz2
43
Chapter 4. Calaby-Yau manifolds
deļ¬ned on U āˆ’ {P} such that Ļ€āˆ—(Ī±) extends to a nowhere-vanishing holo-
morphic form on Ļ€āˆ’1(U).
The structure of rational double points (sometimes called simple singularities)
is well-known and each such point must be analytically isomorphic to one of
the following:
An(n ā‰„ 1) x2 + y2 + zn+1 = 0 Fig4.1
Dn(n ā‰„ 4) x2 + yz2 + znāˆ’1 = 0 Fig4.2
E6 x2 + y3 + z4 = 0 Fig4.3 a)
E7 x2 + y3 + yz3 = 0 Fig4.3 b)
E8 x2 + y3 + z5 = 0 Fig4.3 c)
and the resolution X ā†’ X replaces such a point with a collection of rational
curves of self-intersection -2 in the following conļ¬guration: An, Dn, E6, E7, E8.
Figure 4.1: From left to right in the ļ¬gure are represented A2, A4, A6, A8. Note that
every A2n+1 for 1 ā‰¤ n can not be plotted since all the points satisfying the related
equation are complex. Furthermore we can observe that as n increases, the top of
the surfaces becomes more pointed.
44
Chapter 4. Calaby-Yau manifolds
Figure 4.2: From left to right the image shows D4, D5, D6, D7. In the foreground
stand out the even Di 4 ā‰¤ i ā‰¤ 7 while in the background stand out the odd ones.
As we can see the conļ¬guration of the surfaces and the look of the singular points
change letting i even or odd.
45
Chapter 4. Calaby-Yau manifolds
(a) E6. (b) E7.
(c) E8.
Figure 4.3
Deļ¬nition 4.17 (K3 surfaces). A K3 surface is a compact complex analytic
surface X with only rational double points such that
h(0,1)
= dim(H1
(OX)) = 0
and Ļ‰X = OX.
If X is smooth, the dualizing sheaf Ļ‰X is the line bundle associated to
the canonical divisor KX, so this last condition implies that the canonical
divisor is trivial since KX= OX.
If X is a K3 surface and Ļ€: X ā†’ X is the minimal resolution of singularities
(i. e. the one which appeared in the deļ¬nition of rational double point) then
it turns out that the pull-back Ļ€āˆ— establishes an isomorphism
H1
(OX) = H1
(OX
),
and also we have Ļ‰X
= Ļ€āˆ—OX = OX
. Thus, the smooth surface X is also a
K3 surface.
We collect a number of construction methods for K3 surfaces. One should,
however, keep in mind that most K3 surfaces, especially of high degree,
do not admit explicit descriptions. Their existence is solely predicted by
deformation theory.
46
Chapter 4. Calaby-Yau manifolds
Proposition 4.18. The following properties apply for a K3 surface.
(i) All K3 surfaces are simply connected.
(ii) Every K3 surface over C is diļ¬€eomorphic to the Fermat quartic (see
Example 4.19).
(iii) The Hodge diamond
h0,0
h1,0 h0,1
h2,0 h1,1 h0,2
h2,1 h1,2
h2,2
=
1
0 0
1 22 1
0 0
1
is completely determined (even in positive characteristic).
Example 4.19. A non-singular degree 4 surface, such as the Fermat quartic,
[w, x, y, z] āˆˆ CP3
| w4
+ z4
+ y4
+ x4
= 0
is a K3 surface.
Example 4.20. The intersection of a quadric and a cubic in CP4
gives K3
surfaces.
Example 4.21. The intersection of three quadrics in CP5
gives K3 surfaces.
Example 4.22. A Kummer surface is a special type of quartic surface. As a
projective variety, a Kummer surface may be described as the vanishing set
of an ideal of polynomials. However, these surfaces may also be viewed more
abstractly, in terms of Jacobian varieties but we will not enter in details.
However we will devote the next section to the main feature of these surfaces.
See [End03] for more details.
Kummer surfaces
Deļ¬nition 4.23. The Kummer surface with parameter Āµ āˆˆ R is the projec-
tive variety given by
KĀµ : x2
y2
+ z2
āˆ’ Āµ2
w2
āˆ’ Ī»pqrs = 0 āŠ‚ RP3
,
where Āµ2 = 1
3, 1, 3,
Ī» =
3Āµ2 āˆ’ 1
3 āˆ’ Āµ2
and p, q, r, s are the ā€œtetrahedral coordinates,ā€ given by
p = w āˆ’ z āˆ’
āˆš
2x,
q = w āˆ’ z +
āˆš
2x,
r = w + z + 2y,
s = w + z āˆ’ 2y.
47
Chapter 4. Calaby-Yau manifolds
(a) Āµ2
= 1
3
(double sphere). (b) Āµ2
= 1 (Roman surface).
(c) Āµ2
= 3 (4 planes).
Figure 4.4: Real plots of the three exceptional cases. Pictures taken from [End03].
We exclude the values Āµ2 = 1
3, 1, 3 because these are exceptional cases,
for which most of the statements we will make about Kummer surfaces KĀµ
will not hold. These three cases correspond, repectively, to the double sphere,
the Roman surface, and 4 planes, and are shown in Figure 4.4 (in the plots
in the ļ¬gures the parameter w is set to w=1). As a side note, recall the
Veronese surface, which is given by the embedding RP2
ā†’ RP5
by
[x : y : z] ā†’ [x2
: y2
: z2
: xy : xz : yz].
The Roman surface referred to above is the projection of this surface into
RP3
.
Example 4.24. We will now give a non-algebraic example of Kummer surface.
Let
T2
= C2
/Ī“
48
Chapter 4. Calaby-Yau manifolds
be a complex torus of complex dimension 2. (Thus, Ī“ āŠ‚ C2 is an additive
subgroup such that there is an isomorphism of Rāˆ’vector spaces Ī“ āŠ— R = C2.
Let (z, w) be coordinates on C2, and deļ¬ne
i(z, w) = (āˆ’z, āˆ’w).
Since Ī“ is a subgroup under addition,
i(Ī“) = Ī“.
Thus, i descends to an automorphism Ėœi: T2 ā†’ T2. If we want to ļ¬nd the
ļ¬xed points of i, we need to study the solutions to
i(z, w) ā‰” (z, w) mod Ī“.
These solutions are
(z, w)|(2z, 2w) āˆˆ Ī“
and so Ėœi has as ļ¬xed points 1
2Ī“/Ī“. (There are 16 of these.)
If we deļ¬ne
X := T2
/Ėœi,
it turns out that X is a Kummer surface. This surface in fact has 16 singular
points at the images of the ļ¬xed points of Ėœi.
To see the structure of these singular points, consider the action of i on a small
neighborhood U of (0, 0) in C2. Then U/i is isomorphic to a neighborhood of
a singular point of X. To describe U/i, we note that the invariant functions
on U are generated by
z2
, zw, w2
.
Thus, if we let r = z2, s = zw and t = w2 we can write
U/i = {(r, s, t) near (0, 0, 0)|rt = s2
}.
This is a rational double point of type A1.
dz āˆ§ dw is a global holomorphic 2-form on C2, invariant under the action of
Ī“, and so descends to a form on T2. Since
d(āˆ’z) āˆ§ d(āˆ’w) = dz āˆ§ dw,
the form is also invariant under the action of i, so we get a form dz āˆ§ dw on
X{singular points}. In local coordinates, dr = 2zdz, dt = 2wdw so that
dz āˆ§ dw =
dr āˆ§ dt
4zw
=
dr āˆ§ dt
4s
.
It is easy to check that this form induces a global nowhere vanishing holo-
morphic 2-form on the minimal resolution X of X. To ļ¬nish checking that
X is a K3 surface, we use the fact that
H1
(OX
) = H1
(OX) = { elements of H1(OT2 ) invariant under Ėœi }.
49
Chapter 4. Calaby-Yau manifolds
Now H1(OT2 ) = H(0,1)(T2), the space of global diļ¬€erential forms of type
(0, 1). Since
Ėœiāˆ—
(dĀÆz) = āˆ’dĀÆz and Ėœiāˆ—
(d ĀÆw) = āˆ’d ĀÆw,
this space is generated by dĀÆz and d ĀÆw. It follows that
H1
(OX) = H1
(OX
) = (0),
and that X and X are both K3 surfaces.
Proposition 4.25. We have the following facts about the Kummer surface
KĀµ.
ā€¢ KĀµ is irreducible.
ā€¢ As suggested by the appearance of the tetrahedral coordinates in the
deļ¬ning equation, KĀµ has tetrahedral symmetry.
ā€¢ KĀµ has 16 singularities, each of which is an ordinary double point, a
three-dimensional analogue of a node singularity. It is interesting to
note that a complex surface may have at most ļ¬nitely many ordinary
double points. For quartic surfaces, the maximum number of ordinary
double points is 16, which is achieved by KĀµ.
ā€¢ Resolving the 16 singularities3 of KĀµ, we obtain a K3 surface. This
K3 surface, which is sometimes given as the deļ¬nition of a Kummer
surface, contains 16 disjoint rational curves.
Some Kummer surfaces are shown in Figure 4.5. Note that since these
plots are restricted to the real numbers, some of the ordinary double points
may not be visible. In fact, in the case 0 ā‰¤ Āµ2 ā‰¤ 1
3 (not plotted), KĀµ only
contains 4 real points (each of which turns out to be an ordinary double
point). In the following table are collected all the real and complex nodes of
the surfaces obtained by letting Āµ vary.
3
In algebraic geometry, the problem of resolution of singularities asks whether every
algebraic variety V has a resolution, a non-singular variety W with a proper birational map
W ā†’ V. For varieties over ļ¬elds of characteristic 0 this was proved in Hironaka (1964),
while for varieties over ļ¬elds of characteristic p it is an open problem in dimensions at
least 4.
50
Chapter 4. Calaby-Yau manifolds
(a) 1
3
ā‰¤ Āµ2
ā‰¤ 1. (b) 1 ā‰¤ Āµ2
ā‰¤ 3.
(c) 3 ā‰¤ Āµ2
. (d) Āµ2
= āˆž.
Figure 4.5: Some Kummer Surfaces. Pictures taken from [End03].
Āµ2 Real nodes Complex nodes Picture
0 ā‰¤ Āµ2 < 1
3 4 12 4 real points
Āµ2 = 1
3 / / Fig.4.4 a)
1
3 < Āµ2 < 1 4 12 Fig.4.5 a)
Āµ2 = 1 / / Fig.4.4 b)
1 < Āµ2 < 3 16 0 Fig.4.5 b)
Āµ2 = 3 / / Fig.4.4 c)
Āµ2 ā‰„ 3 16 0 Fig.4.5 c)
Āµ2 = āˆž 16 0 Fig.4.5 d)
51
Chapter 4. Calaby-Yau manifolds
4.2.3 3-dimensional and higher-dimensional Calabi-Yau man-
ifolds
While the number of Calabi-Yau manifolds with one or two complex dimen-
sions is known, the situation in three complex dimension is much diļ¬€erent.
Several thousand Calabi-Yau three-folds have been discovered; with one
exception T3, their metrics are not explicitly known, and it is not even
known (although it is strongly suspected) that the number of topologically
distinct Calabi-Yau three-folds is ļ¬nite. Therefore we will focus on complex
projective spaces and we will analyse some interesting result.
Before entering into details we need to recall some rudiments about intersec-
tion theory.
Let f1, . . . , fk be complex, homogeneous polynomials of degree d1, . . . , dk in
n + k + 1 complex variables z = (z1, . . . , zn+k+1), deļ¬ne
X(f1, . . . , fk) := {[z] āˆˆ CPn+k
| fi(z) = 0 for i = 1, . . . , k}.
The set X = X(f1, . . . , fk) is an algebraic variety of dimension n.
I(X) = {Fi(z) āˆˆ C[z] | Fi(a) = 0, āˆ€a āˆˆ X}.
Deļ¬nition 4.26 (Complete intersection). Let X = X(f1, . . . , fk) be an n
dimensional algebraic variety in CPn+k. X is a complete intersection if there
exists k homogeneous polynomials
Fi(z1, . . . , zn+k+1), for i = 1, . . . , k,
which generate all other homogeneous polynomials that vanish on X, i.e.,
I(X) =< F1, . . . , Fk > .
Proposition 4.27. The n dimensional complete intersection X of k smooth
hypersurfaces of degree d1, . . . , dk in CPn+k
{is a variety with trivial canonical bundle} ā‡ā‡’ d1+Ā· Ā· Ā·+dk = n+k+1.
When n = 3, for example, we have the following solutions of 4 + k =
d1 + Ā· Ā· Ā· + dk with di > 1 :
5 = 5, 6 = 4 + 2 or 6 = 3 + 3, and 7 = 3 + 2 + 2.
In looking for further three dimensional examples, we consider complete
intersections in weighted projective spaces.
52
Chapter 4. Calaby-Yau manifolds
Weighted Projective Spaces. A weighted projective space CPn
(w) is
a generalization of CPn
. Both are quotients of Cn+1{0} by an action of
Cāˆ— = C{0}. The weights w are sequences of natural numbers
w = (w0, . . . , wn) āˆˆ Nn+1
and the action of Ī» āˆˆ Cāˆ— on (z0, . . . , zn) āˆˆ Cn+1{0} is given by the formula
Ī» Ā· (z0, . . . , zn) = (Ī»w0
z0, . . . , Ī»wn
zn).
We assume that the greatest common divisor of the wi is 1.
Remark. From now on we will work on C thus we will denote CPn
(w) simply
with Pn(w).
For each subset S =āŠ‚ {w0, . . . , wn} we denote by q(S) the greatest
common divisor of the wi with i āˆˆ S. Let H(S) denote the subset of all
(zj) āˆˆ Pn(w) with zi = 0 for i /āˆˆ S. The points in H(S) are cyclic quotient
singularities for the group Zq(S)Z.
Furthermore we need the polynomials to be quasihomogeneous due to the
nature of the weighted projective space. A quasihomogeneous polynomial is
deļ¬ned as:
Deļ¬nition 4.28. A polynomial f is called quasihomogeneous of degree d if
the following relation holds:
f(Ī»w0
z0, . . . , Ī»wn
zn) = Ī»d
f(z0, . . . , zn). (4.14)
We will deļ¬ne hypersurfaces in Pn(w) by using transverse polynomials.
We deļ¬ne a transverse polynomial as follows:
Deļ¬nition 4.29. A polynomial f is transverse if f = 0 only at the origin.
This means that for a given set of weights not any polynomial is quasiho-
mogeneous, as can be seen in the following example:
Example 4.30. We consider the space P1(2, 3) and the polynomial:
f = x2
y + y2
= (Ī»2
x)2
(Ī»3
y) + (Ī»3
y)2
= Ī»2Ā·2+3Ā·1
x2
y + Ī»3Ā·2
y2
= Ī»d
f.
Clearly there is no degree d to satisfy the relations and hence this polynomial
is not quasihomogeneous.
53
Chapter 4. Calaby-Yau manifolds
Complete Intersections in Weighted Projective Spaces. The equa-
tions of hypersurfaces in the weighted projective space Pn(w) of degree d are
given by transverse quasihomogeneous polynomial equations f(z0, . . . , zn) = 0.
Now in order to ļ¬nd Calabi Yau manifold in Pn(w) the following proposition
can help.
Proposition 4.31. The complete intersections of multiple degree (d1, . . . , dk)
in the weighted projective space Pm+k(w) with trivial canonical bundle are
those satisfying the following condition
w0 + Ā· Ā· Ā· + wm+k = d1 + Ā· Ā· Ā· + dk.
For instance elliptic curves arise either as quartic curves in P2(1, 1, 2) and
sextic curves in P2(1, 2, 3). We can take for example
w4
+ x4
+ y2
= 0 and w6
+ x3
+ y2
= 0
respectively.
Example 4.32. Degree 8 hypersurface in P4(1, 1, 2, 2, 2) with equation
X : y8
0 + y8
1 + y4
2 + y4
3 + y4
4 = 0.
The degree 8 = 1 + 1 + 2 + 2 + 2 so the hypersurface in this degree is a
Calabiā€“Yau manifold.
Example 4.33. Degree 12 hypersurface in P4(1, 1, 2, 2, 6) with equation
y12
0 + y12
1 + y6
2 + y6
3 + y2
4 = 0.
The degree 12 = 1 + 1 + 2 + 2 + 6 so the hypersurface in this degree is a
Calabiā€“Yau manifold.
Example 4.34. Degree 18 hypersurface in P4(1, 1, 1, 6, 9) with equation
y18
0 + y18
1 + y18
2 + y3
3 + y2
4 = 0.
The degree 18 = 1 + 1 + 1 + 6 + 9 so that the hypersurface in this degree is a
Calabiā€“Yau manifold.
There is a complete classiļ¬cation of Calabiā€“Yau varieties arising from
transverse hypersurfaces in P4(w) but we will not go further.
4.3 KƤhler manifolds
Deļ¬nition 4.35. Let (M, Ļ‰) be a symplectic manifold. If there exist a
genuine complex structure J compatible with Ļ‰, then (M, Ļ‰, J) is a KƤhler
manifold. The symplectic form Ļ‰ is then called a KƤhler form.
54
Chapter 4. Calaby-Yau manifolds
Remark. Here genuine means coming from holomorphic coordinates making
M a complex manifold.
Hence a KƤhler manifold is a symplectic manifold endowed with a com-
patible complex structure, and as we saw in the previous chapters, g deļ¬ned
by
g(u, v) = Ļ‰(u, Jv),
is a Riemannian metric called KƤhler metric and Ļ‰ is called KƤhler form.
Locally on the almost complex manifold M, the holomorphic vector ļ¬elds
T1,0
x (M), as we have already seen, have a basis āˆ‚
āˆ‚z1
, . . . , āˆ‚
āˆ‚zn
and the holo-
morphic 1-forms Ī›
(1,0)
x have the related dual basis denoted dz1, . . . , dzn. The
coeļ¬ƒcients of the KƤhler form in these local coordinates are
gj,k(z, ĀÆz) = g
āˆ‚
āˆ‚zj
,
āˆ‚
āˆ‚ĀÆzk
,
and the diļ¬€erential form
Ļ‰ = āˆ’2i
n
j,k=1
gj,k(z, ĀÆz)dzj
āˆ§ ĀÆzk
.
Therefore a KƤhler manifold is a smooth manifold equipped with
ā€¢ complex structure;
ā€¢ Riemannian structure;
ā€¢ symplectic structure,
which are compatible (as seen in Section 3.3).
The existence of a KƤhler metric on a compact manifold constraints the
topology. In particular the following properties apply.
Proposition 4.36. If (M, J, Ļ‰) is a compact KƤhler manifold of real dimen-
sion 2n, then
i. the complex structure on M leads to the Hodge decomposition (which
is equivalent to having equality in (1.15))
Hk
deRham(M) =
p+q=k
Hp,q
deRham(M) =
p+q=k
Hp,q
ĀÆāˆ‚
(M);
ii. h(p,q)(M) = h(q,p)(M)
iii. the odd Betty numbers
b2r+1(M) := dim H2r+1
deRham(M) = dim
ker d
Im d
,
where d: Ī›2r+1(M) ā†’ Ī›2r+2(M), are even āˆ€i = 1, . . . , n;
55
Chapter 4. Calaby-Yau manifolds
iv. b2r(M) > 0.
Before entering into the details of the proof we will make a brief digression
concerning some operators and results of complex diļ¬€erential geometry.
Each tangent space V = TxM has a positive inner product (Ā·, Ā·), part of
the Riemannian metric in a compatible triple.
Let e1, . . . , en be a positively oriented orthonormal basis of V. The star
operator is a linear operator āˆ—: Ī›(V ) ā†’ Ī›(V ) deļ¬ned by
āˆ—(1) = e1 āˆ§ Ā· Ā· Ā· āˆ§ en
(e1 āˆ§ Ā· Ā· Ā· āˆ§ en) = 1
(e1 āˆ§ Ā· Ā· Ā· āˆ§ ek) = āˆ—(ek+1 āˆ§ Ā· Ā· Ā· āˆ§ en).
We see that āˆ—: Ī›k(V ) ā†’ Ī›nāˆ’k and satisļ¬es āˆ—āˆ— = (āˆ’1)k(nāˆ’k). The āˆ— operator
will be hereafter used to deļ¬ne an inner product on forms.
Let ĀÆāˆ‚ be deļ¬ned as in Deļ¬nition 1.5 and d as in Deļ¬nition 1.6. We now
deļ¬ne the adjoint operator of ĀÆāˆ‚
ĀÆāˆ‚āˆ—
: Ī›(p,q)
ā†’ Ī›(p,qāˆ’1)
(M) (4.15)
by requiring that
ĀÆāˆ‚āˆ—
Ļˆ, Ī· = Ļˆ, ĀÆāˆ‚Ī·
for all Ī· āˆˆ Ī›(p,qāˆ’1)(M).
The Dolbeault cohomology group Hp,q
ĀÆāˆ‚
(M) = Zp,q
ĀÆāˆ‚
(M)/ĀÆāˆ‚Ī›(p,qāˆ’1)(M) is rep-
resented by the solutions of the two ļ¬rst-order equations
ĀÆāˆ‚Ļˆ = 0, ĀÆāˆ‚āˆ—
Ļˆ = 0. (4.16)
These two may be replaced by the single second-order equation
āˆ†ĀÆāˆ‚Ļˆ = ĀÆāˆ‚ ĀÆāˆ‚āˆ—
+ ĀÆāˆ‚āˆ— ĀÆāˆ‚ Ļˆ = 0.
For a complete proof see [GH14].
The operator
āˆ†ĀÆāˆ‚ : Ī›(p,q)
(M) ā†’ Ī›(p,q)
(M)
is called the ĀÆāˆ‚āˆ’Laplacian or simply the Laplacian (written āˆ†) if no ambiguity
is likely. Diļ¬€erential forms satisfying the Laplace equation
āˆ†Ļˆ = 0
are called harmonic forms; the space of harmonic forms of type (p, q) is
denoted Hp,q(M) and called the harmonic space.
The isomorphism
Hp,q
(M) = Hp,q
ĀÆāˆ‚
(M)
56
Chapter 4. Calaby-Yau manifolds
can be proved (see for instance [GH14]).
Before stating the Hodge theorem, whose isomorphism (4.3) is a part, we
begin by giving an explicit formula for the adjoint operator ĀÆāˆ‚āˆ—. First we
deļ¬ne the star operator,
āˆ—: Ī›(p,q)
(M) ā†’ Ī›(pāˆ’1,qāˆ’1)
(M)
by requiring
(Ļˆ, Ī·) =
M
Ļˆ āˆ§ āˆ—Ī·
for all Ļˆ āˆˆ Ī›(p,q)(M). If we suppose that M is compact this deļ¬ne an inner
product on forms. Therefore if we write
Ī· =
I,J
Ī·IJ Ļ•I āˆ§ ĀÆĻ•J
then
āˆ—Ī· = 2p+qāˆ’n
I,J
ĪµIJ ĀÆĪ·IJ Ļ•I0 āˆ§ ĀÆĻ•J0 ,
where I0 = {1, . . . , n} āˆ’ I and we write ĪµIJ for the sign of the permutation
(1, . . . , n, 1 , . . . , n ) ā†’ (i1, . . . , ip, j1, . . . , jq, i0
1, . . . , i0
nāˆ’p, j0
1, . . . , j0
nāˆ’q).
The signs work out so that
āˆ— āˆ— Ī· = (āˆ’1)p+q
Ī·.
In terms of star the adjoint operator is
ĀÆāˆ‚āˆ—
= āˆ’ āˆ— ĀÆāˆ‚ āˆ— .
Observation 4.37. Note that ĀÆāˆ‚2 = 0 ā‡’ ĀÆāˆ‚āˆ—2 = 0.
We are now ready to state the following theorem.
Theorem 4.38 (Hoge). Let M be a compact complex manifold, then
1. dim Hp,q(M) < āˆž and
2. the orthogonal projection
H: Ī›(p,q)
(M) ā†’ Hp,q
(M) (4.17)
is well deļ¬ned and there exists a unique operator, the Greenā€™s operator,
G: Ī›(p,q)
(M) ā†’ Ī›(p,q)
(M),
with G(H(p,q)(M)) = 0, ĀÆāˆ‚G = GĀÆāˆ‚, ĀÆāˆ‚āˆ—G = GĀÆāˆ‚āˆ— and
I = H + āˆ†G (4.18)
on Ī›(p,q)(M).
57
Chapter 4. Calaby-Yau manifolds
The content of (4.18) is sometimes expressed by saying that, given Ī·, the
equation
āˆ†Ļˆ = Ī·
has a solution Ļˆ if and only if H(Ī·) = 0, and then
Ļˆ = G(Ī·)
is the unique solution satisfying H(Ļˆ) = 0. Thus we should try to solve the
Laplace equation on a compact manifold. The idea is to ļ¬nd a Ļˆ such that
(Ļˆ, āˆ†Ļ•) = (Ī·, Ļˆ)
for all Ļ• āˆˆ Ī›(p,q)(M) and to prove that this Ļˆ is in fact Cāˆž.
Remark. We remark that we may deļ¬ne the adjoint dāˆ— of d, form the
Laplacian āˆ†d = ddāˆ— + dāˆ—d, and arrive at the exact formalism as for ĀÆāˆ‚ on
complex manifolds. Moreover the Hodge theorem is also true.
Let M be a compact complex manifold with Hermitian metric ds2, and
suppose that in some open set U āŠ‚ M, ds2 is Euclidean; that is there exists
local holomorphic coordinates z = (z1, . . . , zn) such that
ds2
= dzi āŠ— dĀÆzi.
Theorem 4.39. With the same hypothesis as above, for a diļ¬€erential form
Ļ• = Ļ•IJ dzI āˆ§ dĀÆzJ
compactly supported in U,
āˆ†d = 2āˆ†ĀÆāˆ‚. (4.19)
Proof. Let zi = xi + iyi, then
āˆ†ĀÆāˆ‚(Ļ•) = āˆ’2
I,J,i
āˆ‚2
āˆ‚ziāˆ‚ĀÆzi
Ļ•IJ dzI āˆ§ dĀÆzJ
= āˆ’
1
2 I,J,i
āˆ‚2
āˆ‚x2
i
+
āˆ‚2
āˆ‚y2
i
Ļ•IJ dzI āˆ§ dĀÆzJ
=
1
2
āˆ†d(Ļ•),
i.e., the ĀÆāˆ‚āˆ’Laplacian is equal to the ordinary dāˆ’Laplacian in U, up to a
constant. Although very few compact complex manifolds have everywhere
Euclidean metrics, but as it turns out in order to insure Equation (4.19) on a
complex manifold it is suļ¬ƒcient that the metric approximate the Euclidean
metric to the second order at each point.
We can now start the demonstration of Theorem 4.36.
58
Chapter 4. Calaby-Yau manifolds
Proof. (i) and (ii). Set
Hp,q
d (M) = {Ī· āˆˆ Ī›p,q
(M): āˆ†dĪ· = 0},
Hr
d(M) = {Ī· āˆˆ Ī›r
(M): āˆ†dĪ· = 0}.
Note that the two groups depend on the particular metric whilst the group
Hp,q
deRham(M) is intrinsically deļ¬ned by the complex structure. By the com-
mutativity of āˆ†d and Ī p,q : Hr
deRham(M) ā†’ Hp,q
deRham(M) and the fact that
āˆ†d is real, the harmonic forms satisfy
Hr(M) = p+q=r Hp,q(M)
Hp,q(M) = Hp,q
d (M)
(4.20)
On the other hand, for Ī· a closed form of pure type (p, q),
Ī· = Ī· + ddāˆ—
G(Ī·),
where the harmonic part Ī· also has a pure type (p, q). Thus every de Rham
cohomology class on a compact oriented riemannian manifold M possesses a
unique harmonic representative, i.e.,
Hk
= Hk
DeRahm(M).
We also have the following orthogonal decomposition with respect to (Ā·, Ā·) :
Ī›k
= Hk
āŠ• āˆ†(Ī›k
(M))
= Hk
āŠ• d(Ī›kāˆ’1
) āŠ• dāˆ—
(Ī›k+1
).
The proof involves functional analysis, elliptic diļ¬€erential operators, pseu-
dodiļ¬€erential operators and Fourier analysis; see [GH14].
When M is KƤhler, the Laplacian satisļ¬es āˆ†d = 2āˆ†ĀÆāˆ‚, hence harmonic forms
are also bigraded
Hk
=
p+q=k
Hp,q
.
Combining this with 4.20 and Theorem 4.38 we obtain the Hodge decompo-
sition for the Laplacian āˆ†d. For a compact KƤhler manifold M, the complex
cohomology satisļ¬es
Hr
DeRahm(M, C) = p+q=r Hp,q
DeRahm(M)
Hp,q
DeRahm(M) = Hp,q
DeRahm(M).
(4.21)
Hence, we have the following isomorphisms:
Hk
DeRahm(M) = Hk
=
p+q=r
Hp,q
(M) =
p+q=r
Hp,q
ĀÆāˆ‚
(M).
59
Chapter 4. Calaby-Yau manifolds
(iii) For the point (ii) of the proposition it follows that
b2r+1(M) =
p+q=2r+1
h(p,q)
(M)
and considering that h(p,q)(M) = h(q,p)(M), we obtain
b2r+1(M) = 2
j
h(j,2r+1āˆ’j)
(M).
Therefore b2r+1(M) ā‰” 0 mod 2.
(iv) This fact directly follows from the fact that every KƤhler manifold is
also a symplectic manifold. In fact let Ļ‰ be the symplectic form ensured by
the KƤhler structure. If dĪ± = Ļ‰r with 0 ā‰¤ r ā‰¤ n by Stokesā€™ Theorem we
have that
M
Ļ‰n
=
M
d(Ī± āˆ§ Ļ‰nāˆ’r
) = 0.
Remark. Note that (i) is true because
dimC (Hr
deRham(M, C)) = dimR (Hr
deRham(M, R)) ,
since we previously deļ¬ned the Betti numbers for real de Rham cohomology
groups.
Actually KƤhler manifolds satisfy several other topological property but
here we have mentioned only the ones that will be used later. For more
details see for instance [DS01].
Deļ¬nition 4.40 (KƤhler potential). Let (X, J, Ļ‰) be a KƤhler manifold then
around every point x āˆˆ X there exists a neighbourhood U and a function
f āˆˆ Cāˆž(U, R) for which the KƤhler form Ļ‰ can be written as
Ļ‰|U = iāˆ‚ ĀÆāˆ‚f.
Here, the operators
āˆ‚ =
k
āˆ‚
āˆ‚zk
dzk
and
ĀÆāˆ‚ =
k
āˆ‚
āˆ‚ĀÆzk
dĀÆzk
are called the Dolbeault operators.
60
Chapter 4. Calaby-Yau manifolds
For instance, in Cn, the function f = |z|2
2 is a KƤhler potential for the
KƤhler form, because
iāˆ‚ ĀÆāˆ‚
1
2
|z|2
=
1
2
iāˆ‚ ĀÆāˆ‚
k
zk ĀÆzk
=
1
2
iāˆ‚
k
zkdĀÆzk
=
1
2
i
k
dzk āˆ§ dĀÆzk
= Ļ‰.
We have that the converse holds true in fact we can apply the following
property. Before going into the proposition we will see a necessary deļ¬nition.
Actually the metric we will talk about is the same as discussed in Section
4.3.
Deļ¬nition 4.41 (Compatible Riemannian Metric.). Let M be a complex
manifold with corresponding complex structure J. We say that a Riemannian
metric g is compatible with J if
g(JX, JY ) = g(X, Y ) (4.22)
for all vector ļ¬elds X, Y on M. A complex manifold together with a compat-
ible Riemannian metric is called a Hermitian manifold.
Proposition 4.42. Let M be a complex manifold with a compatible Rieman-
nian metric g, as in (4.22). Then the following assertions are equivalent:
1. g is a KƤhler metric.
2. For each point x āˆˆ M, there is a smooth real function f in a neigh-
bourhood U of x such that Ļ‰|U = iāˆ‚ ĀÆāˆ‚f.
3. dĻ‰ = 0.
Proof. By deļ¬nition, 1 and 3 are equivalent.
3 ā‡’ 2 Since Ļ‰ is real and dĻ‰ = 0, we have Ļ‰ = dĪ± locally, where Ī± is a real
1-form. Then Ī± = Ī² + ĀÆĪ², where Ī² is a form of type (1, 0). Since Ļ‰ is of type
(1, 1), we have
āˆ‚Ī² = 0, ĀÆāˆ‚ ĀÆĪ² = 0 and Ļ‰ = ĀÆāˆ‚Ī² + āˆ‚ ĀÆĪ²
Hence Ī² = āˆ‚Ļ† locally, where Ļ† is a smooth complex function. Then ĀÆĪ² = ĀÆāˆ‚ ĀÆĻ†
and hence
Ļ‰ = ĀÆāˆ‚āˆ‚Ļ† + āˆ‚ ĀÆāˆ‚ ĀÆĻ†
= āˆ‚ ĀÆāˆ‚(ĀÆĻ† āˆ’ Ļ†)
= iāˆ‚ ĀÆāˆ‚f,
61
Chapter 4. Calaby-Yau manifolds
with f = i(Ļ† āˆ’ ĀÆĻ†).
2 ā‡’ 3 Let d = āˆ‚ + ĀÆāˆ‚, then we simply observe that
āˆ‚Ļ‰ = iāˆ‚2 ĀÆāˆ‚f
= 0,
since āˆ‚2 = 0 and that
ĀÆāˆ‚Ļ‰ = iāˆ‚ ĀÆāˆ‚2
f
= 0,
since ĀÆāˆ‚2 = 0.
Thanks to the equivalence of the deļ¬nition of KƤhler metric we will show
in the following example how the complex projective space can be seen as a
KƤhler manifold.
Example 4.43 (Complex Projective space). Using the same notation as in
example 1.4, let CPn be the complex projective space. To see that CPn has
a natural KƤhler manifold structure, we introduce the following functions
which turn out to be KƤhler potentials. Let f be deļ¬ned in an open set Uj
by
fj = log
ļ£«
ļ£­1 +
k=j
zk
zj
2
ļ£¶
ļ£ø
= log
n
k=0
|zk|2
āˆ’ log |zj|2
.
On the intersection of two coordinate charts Uj āˆŖ Uk, the diļ¬€erence
fj āˆ’ fk = log |zk|2
āˆ’ log |zj|2
= log
zk
zj
āˆ’ log
ĀÆzk
ĀÆzj
.
satisļ¬es the equation
āˆ‚ ĀÆāˆ‚ (fj āˆ’ fk) = 0,
and hence there exists a global form Ļ‰ on CPn with Ļ‰|Uj
= iāˆ‚ ĀÆāˆ‚fj.
To see that this form Ļ‰ is the form associated with a KƤhler structure, we
consider its coeļ¬ƒcients gk,l in local coordinates where
Ļ‰|Uj
= iāˆ‚ ĀÆāˆ‚fj =
k,l
gk,ldwk āˆ§ d ĀÆwl
62
Chapter 4. Calaby-Yau manifolds
and wk = zk
zj
.
In order to see that the coeļ¬ƒcient matrix (hk,l) is positive deļ¬nite and
Hermitian symmetric we calculate the diļ¬€erentials using
fj = log
ļ£«
ļ£­1 +
k=j
zk
zj
2
ļ£¶
ļ£ø = log 1 +
n
k=1
|wk|2
.
Thus
ĀÆāˆ‚fj = 1 +
n
k=1
|wk|2
āˆ’1
Ā·
n
k=1
wkd ĀÆwk
and
āˆ‚ ĀÆāˆ‚fj = 1 +
n
k=1
|wk|2
āˆ’1
Ā·
n
k=1
dwk āˆ§ d ĀÆwk āˆ’ 1 +
n
k=1
|wk|2
āˆ’2
Ā·
n
k,l=1
ĀÆwkdwk āˆ§ wld ĀÆwl
= 1 +
n
k=1
|wk|2
āˆ’2
Ā·
n
k,l=1
Ī“k,l 1 +
n
k=1
|wk|2
āˆ’ ĀÆwkwl dwk āˆ§ d ĀÆwl.
For a complex vector Ī¾ = (Ī¾1, . . . , Ī¾n) āˆˆ Cn, we study the positivity properties
of the following expression using the Hermitian inner product
< w, Ī¾ >=
n
k=1
wk
ĀÆĪ¾k
to obtain the inequality
n
k,l=1
Ī“k,l 1 +
n
k=1
|wk|2
āˆ’ ĀÆwkwl Ī¾k
ĀÆĪ¾l
=< Ī¾, Ī¾ >2
1+ < w, w >2
āˆ’ | < w, Ī¾ > |2
for Ī¾ = 0, and the Schwarz inequality | < w, Ī¾ > |2 ā‰¤ < Ī¾, Ī¾ >2< w, w >2
leads to
ā‰„ < Ī¾, Ī¾ >2
1+ < w, w >2
āˆ’ < Ī¾, Ī¾ >2
< w, w >2
=< Ī¾, Ī¾ >2
ā‰„ 0.
4.3.1 KƤhler and Calabi-Yau manifolds
We have decided to introduce KƤhler manifold in our dissertation in order
to see their connection with Calabi-Yauā€™s. We want to clarify that Calabi-
Yau manifolds are a particular kind of KƤhler manifolds. Many diļ¬€erent
deļ¬nitions of Calabiā€“Yau manifolds exist in the literature; we list here some
63
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi
Marcaccio_Tesi

Mais conteĆŗdo relacionado

Destaque

Chinaā€™s latitude obligation
Chinaā€™s latitude obligationChinaā€™s latitude obligation
Chinaā€™s latitude obligationJohn Jeffery
Ā 
SK Engineering & Allied Works, Bahraich, Dal Plant
SK Engineering & Allied Works, Bahraich, Dal PlantSK Engineering & Allied Works, Bahraich, Dal Plant
SK Engineering & Allied Works, Bahraich, Dal PlantIndiaMART InterMESH Limited
Ā 
How to deal with your own negativity
How to deal with your own negativityHow to deal with your own negativity
How to deal with your own negativityStephanie Chan
Ā 
I learningframework
I learningframeworkI learningframework
I learningframeworkLia Yuliana
Ā 
NATURALEZA EXPERIMENTAL
NATURALEZA EXPERIMENTALNATURALEZA EXPERIMENTAL
NATURALEZA EXPERIMENTALPATRICIO YUCAILLA
Ā 
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del Defensor
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del DefensorENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del Defensor
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del DefensorENJ
Ā 
äø‰åˆ†é˜äŗ†č§£ęµē؋圖
äø‰åˆ†é˜äŗ†č§£ęµē؋圖äø‰åˆ†é˜äŗ†č§£ęµē؋圖
äø‰åˆ†é˜äŗ†č§£ęµē؋圖吉吉
Ā 
Vegetables A To Z
Vegetables A To ZVegetables A To Z
Vegetables A To ZMike Maddox
Ā 
04 medicinal natural products
04 medicinal natural products04 medicinal natural products
04 medicinal natural productsDr. Harish Kakrani
Ā 
S. pyogenes, its virulence, antibiotic, phytochemicals
S. pyogenes, its virulence, antibiotic, phytochemicalsS. pyogenes, its virulence, antibiotic, phytochemicals
S. pyogenes, its virulence, antibiotic, phytochemicalsUniversitƩ Laval
Ā 
Flavonoids: Benefits for Total Health [INFOGRAPHIC]
Flavonoids: Benefits for Total Health [INFOGRAPHIC]Flavonoids: Benefits for Total Health [INFOGRAPHIC]
Flavonoids: Benefits for Total Health [INFOGRAPHIC]Food Insight
Ā 
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...Indian dental academy
Ā 
Mutig sein
Mutig seinMutig sein
Mutig seingoforgabi
Ā 

Destaque (19)

Chinaā€™s latitude obligation
Chinaā€™s latitude obligationChinaā€™s latitude obligation
Chinaā€™s latitude obligation
Ā 
Ofimatica
Ofimatica Ofimatica
Ofimatica
Ā 
SK Engineering & Allied Works, Bahraich, Dal Plant
SK Engineering & Allied Works, Bahraich, Dal PlantSK Engineering & Allied Works, Bahraich, Dal Plant
SK Engineering & Allied Works, Bahraich, Dal Plant
Ā 
How to deal with your own negativity
How to deal with your own negativityHow to deal with your own negativity
How to deal with your own negativity
Ā 
I learningframework
I learningframeworkI learningframework
I learningframework
Ā 
NATURALEZA EXPERIMENTAL
NATURALEZA EXPERIMENTALNATURALEZA EXPERIMENTAL
NATURALEZA EXPERIMENTAL
Ā 
Ofimatica
OfimaticaOfimatica
Ofimatica
Ā 
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del Defensor
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del DefensorENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del Defensor
ENJ-100 CĆ³digo de Comportamiento Ɖtico - Curso Ɖtica del Defensor
Ā 
äø‰åˆ†é˜äŗ†č§£ęµē؋圖
äø‰åˆ†é˜äŗ†č§£ęµē؋圖äø‰åˆ†é˜äŗ†č§£ęµē؋圖
äø‰åˆ†é˜äŗ†č§£ęµē؋圖
Ā 
Vegetables A To Z
Vegetables A To ZVegetables A To Z
Vegetables A To Z
Ā 
Eintauchen in MVP mit GWT
Eintauchen in MVP mit GWT Eintauchen in MVP mit GWT
Eintauchen in MVP mit GWT
Ā 
04 medicinal natural products
04 medicinal natural products04 medicinal natural products
04 medicinal natural products
Ā 
S. pyogenes, its virulence, antibiotic, phytochemicals
S. pyogenes, its virulence, antibiotic, phytochemicalsS. pyogenes, its virulence, antibiotic, phytochemicals
S. pyogenes, its virulence, antibiotic, phytochemicals
Ā 
Flavonoids: Benefits for Total Health [INFOGRAPHIC]
Flavonoids: Benefits for Total Health [INFOGRAPHIC]Flavonoids: Benefits for Total Health [INFOGRAPHIC]
Flavonoids: Benefits for Total Health [INFOGRAPHIC]
Ā 
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...
Luting agents for fixed prosthodontics/ orthodontic course by indian dental a...
Ā 
Ofimatica
OfimaticaOfimatica
Ofimatica
Ā 
Interjections
InterjectionsInterjections
Interjections
Ā 
Nutricion enteral
Nutricion enteralNutricion enteral
Nutricion enteral
Ā 
Mutig sein
Mutig seinMutig sein
Mutig sein
Ā 

Semelhante a Marcaccio_Tesi

Math516 runde
Math516 rundeMath516 runde
Math516 rundesgcskyone
Ā 
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdf
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdfThe Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdf
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdfDanielsen9
Ā 
Geometer Toolkit For String Theory
Geometer Toolkit For String TheoryGeometer Toolkit For String Theory
Geometer Toolkit For String TheoryVijay Sharma
Ā 
A Research Study On Developing Writers Voices In A Standardized Testing Envir...
A Research Study On Developing Writers Voices In A Standardized Testing Envir...A Research Study On Developing Writers Voices In A Standardized Testing Envir...
A Research Study On Developing Writers Voices In A Standardized Testing Envir...Fiona Phillips
Ā 
Compiled Report
Compiled ReportCompiled Report
Compiled ReportSam McStay
Ā 
Algebraic topology of finite topological spaces and applications
Algebraic topology of finite topological spaces and applications Algebraic topology of finite topological spaces and applications
Algebraic topology of finite topological spaces and applications sisirose
Ā 
M2 Internship report rare-earth nickelates
M2 Internship report rare-earth nickelatesM2 Internship report rare-earth nickelates
M2 Internship report rare-earth nickelatesYiteng Dang
Ā 
Classical String Calculations in curved space
Classical String Calculations in curved spaceClassical String Calculations in curved space
Classical String Calculations in curved spaceIsmail Abdulaziz
Ā 
The Foundations of Geometry Hilbert the author
The Foundations of Geometry Hilbert the authorThe Foundations of Geometry Hilbert the author
The Foundations of Geometry Hilbert the authorijsonlin
Ā 
Duality in Physics
Duality in PhysicsDuality in Physics
Duality in PhysicsDominic Barker
Ā 
Teoria das supercordas
Teoria das supercordasTeoria das supercordas
Teoria das supercordasXequeMateShannon
Ā 
Algebraic Topology - Hatcher.pdf
Algebraic Topology - Hatcher.pdfAlgebraic Topology - Hatcher.pdf
Algebraic Topology - Hatcher.pdfFaith Brown
Ā 
AN INTRODUCTION TO LAGRANGIAN MECHANICS
AN INTRODUCTION TO LAGRANGIAN MECHANICSAN INTRODUCTION TO LAGRANGIAN MECHANICS
AN INTRODUCTION TO LAGRANGIAN MECHANICSJoshua Gorinson
Ā 
VECTOR_QUNTIZATION
VECTOR_QUNTIZATIONVECTOR_QUNTIZATION
VECTOR_QUNTIZATIONAniruddh Tyagi
Ā 
VECTOR_QUNTIZATION
VECTOR_QUNTIZATIONVECTOR_QUNTIZATION
VECTOR_QUNTIZATIONaniruddh Tyagi
Ā 
Senior_Thesis_Evan_Oman
Senior_Thesis_Evan_OmanSenior_Thesis_Evan_Oman
Senior_Thesis_Evan_OmanEvan Oman
Ā 
The mechanics of the bow and arrow
The mechanics of the bow and arrowThe mechanics of the bow and arrow
The mechanics of the bow and arrowToz Koparan
Ā 

Semelhante a Marcaccio_Tesi (20)

Math516 runde
Math516 rundeMath516 runde
Math516 runde
Ā 
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdf
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdfThe Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdf
The Mathematics any Physicist Should Know. Thomas Hjortgaard Danielsen.pdf
Ā 
Geometer Toolkit For String Theory
Geometer Toolkit For String TheoryGeometer Toolkit For String Theory
Geometer Toolkit For String Theory
Ā 
A Research Study On Developing Writers Voices In A Standardized Testing Envir...
A Research Study On Developing Writers Voices In A Standardized Testing Envir...A Research Study On Developing Writers Voices In A Standardized Testing Envir...
A Research Study On Developing Writers Voices In A Standardized Testing Envir...
Ā 
Compiled Report
Compiled ReportCompiled Report
Compiled Report
Ā 
main
mainmain
main
Ā 
Algebraic topology of finite topological spaces and applications
Algebraic topology of finite topological spaces and applications Algebraic topology of finite topological spaces and applications
Algebraic topology of finite topological spaces and applications
Ā 
M2 Internship report rare-earth nickelates
M2 Internship report rare-earth nickelatesM2 Internship report rare-earth nickelates
M2 Internship report rare-earth nickelates
Ā 
MAINPH
MAINPHMAINPH
MAINPH
Ā 
Classical String Calculations in curved space
Classical String Calculations in curved spaceClassical String Calculations in curved space
Classical String Calculations in curved space
Ā 
Applied Math
Applied MathApplied Math
Applied Math
Ā 
The Foundations of Geometry Hilbert the author
The Foundations of Geometry Hilbert the authorThe Foundations of Geometry Hilbert the author
The Foundations of Geometry Hilbert the author
Ā 
Duality in Physics
Duality in PhysicsDuality in Physics
Duality in Physics
Ā 
Teoria das supercordas
Teoria das supercordasTeoria das supercordas
Teoria das supercordas
Ā 
Algebraic Topology - Hatcher.pdf
Algebraic Topology - Hatcher.pdfAlgebraic Topology - Hatcher.pdf
Algebraic Topology - Hatcher.pdf
Ā 
AN INTRODUCTION TO LAGRANGIAN MECHANICS
AN INTRODUCTION TO LAGRANGIAN MECHANICSAN INTRODUCTION TO LAGRANGIAN MECHANICS
AN INTRODUCTION TO LAGRANGIAN MECHANICS
Ā 
VECTOR_QUNTIZATION
VECTOR_QUNTIZATIONVECTOR_QUNTIZATION
VECTOR_QUNTIZATION
Ā 
VECTOR_QUNTIZATION
VECTOR_QUNTIZATIONVECTOR_QUNTIZATION
VECTOR_QUNTIZATION
Ā 
Senior_Thesis_Evan_Oman
Senior_Thesis_Evan_OmanSenior_Thesis_Evan_Oman
Senior_Thesis_Evan_Oman
Ā 
The mechanics of the bow and arrow
The mechanics of the bow and arrowThe mechanics of the bow and arrow
The mechanics of the bow and arrow
Ā 

Marcaccio_Tesi

  • 1. UNIVERSITƀ DEGLI STUDI DI TORINO DIPARTIMENTO DI MATEMATICA GIUSEPPE PEANO SCUOLA DI SCIENZE DELLA NATURA Corso di Laurea Magistrale in Matematica Master thesis Calabi-Yau manifolds Supervisor: Prof.ssa Anna Fino Cosupervisor: Prof. Joel Fine Candidate: Giulia Marcaccio Academic Year 2015-2016
  • 2. Contents Introduction iii 1 Complex manifolds 1 1.1 Complex coordinates . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Tangent space . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Complex forms: the (p,q)-forms . . . . . . . . . . . . . . . . . 4 1.3.1 1-forms . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3.2 Higher degree forms . . . . . . . . . . . . . . . . . . . 5 1.4 Canonical bundle . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4.1 Line bundles over M . . . . . . . . . . . . . . . . . . . 10 1.5 Cohomologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.5.1 The de Rham cohomology . . . . . . . . . . . . . . . . 13 1.5.2 Dolbeault cohomology . . . . . . . . . . . . . . . . . . 15 2 Some Algebraic Geometry tools 17 2.1 Aļ¬ƒne and projective variety . . . . . . . . . . . . . . . . . . . 17 2.2 Regular and rational functions . . . . . . . . . . . . . . . . . 18 2.3 Divisors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4 Rational diļ¬€erential forms and canonical divisors. . . . . . . . 22 2.5 Dualizing sheaf . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3 A look to some complex constructions 25 3.1 Almost complex structure . . . . . . . . . . . . . . . . . . . . 25 3.2 Symplectic manifolds . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.1 (Real) Symplectic manifolds . . . . . . . . . . . . . . . 28 3.2.2 (Complex) Symplectic manifolds . . . . . . . . . . . . 29 3.3 Compatible almost complex structures . . . . . . . . . . . . . 29 4 Calaby-Yau manifolds 32 4.1 Symplectic Calabi-Yau manifolds . . . . . . . . . . . . . . . . 32 4.2 Complex Calabi-Yau manifolds . . . . . . . . . . . . . . . . . 33 4.2.1 1-dimensional Calabi-Yau . . . . . . . . . . . . . . . . 33 4.2.2 2-dimensional Calabi-Yau . . . . . . . . . . . . . . . . 43 i
  • 3. CONTENTS 4.2.3 3-dimensional and higher-dimensional Calabi-Yau man- ifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.3 KƤhler manifolds . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.3.1 KƤhler and Calabi-Yau manifolds . . . . . . . . . . . . 63 4.4 Toric geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.4.1 Cones and fans . . . . . . . . . . . . . . . . . . . . . . 66 4.4.2 Toric divisors . . . . . . . . . . . . . . . . . . . . . . . 69 4.4.3 Toric Calabiā€“Yau threefolds . . . . . . . . . . . . . . . 70 5 Counterexamples 74 5.1 The Kodaira-Thurston example . . . . . . . . . . . . . . . . . 74 5.2 Hopf surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 5.3 Symplectic but not complex . . . . . . . . . . . . . . . . . . . 79 6 Calabiā€“Yau Manifolds and String Theory 81 6.1 Compactiļ¬cation . . . . . . . . . . . . . . . . . . . . . . . . . 82 6.1.1 Dimensional reduction . . . . . . . . . . . . . . . . . 82 ii
  • 4. Introduction Superstring theory is an attempt to explain all of the particles and funda- mental forces of nature in one theory by modelling them as vibrations of tiny supersymmetric strings. Our physical space is observed to have three large spatial dimensions and, along with time, is a boundless four-dimensional continuum, known as space- time. The problem is that if we want to apply superstring theory to our spacetime universe we need it to be ten-dimensional. The discrepancy between the critical dimension d = 10, required by string theory, and the number of observed dimensions d = 4 is resolved by the idea of compactiļ¬cation. Looking for a ten-dimensional manifold consistent with the requirements imposed by this theory, the simplest possibility is to have a space time that takes the product form M4 Ɨ N6, where M4 is a four-dimensional Minkowski space and N6 is some compact six-dimensional manifold. This compact man- ifold is usually taken to have a suļ¬ƒciently small size as to be unobservable with present technology, and thus we would only see the four-dimensional manifold M4. Requiring that supersymmetry is preserved at the compactiļ¬- cation scale restricts us to a special class of manifolds known as Calabi-Yau manifolds. Eugenio Calabi, an American mathematician of Italian origins, in 1953 proposed that certain speciļ¬c geometric structures are allowed under some topological condition. In particular he surmised that whether a certain kind of complex manifold, namely the compact (ļ¬nite in extent) and "KƤhler" ones, could satisfy the general topological conditions of vanishing ļ¬rst Chern class and could also satisfy the geometrical condition of having a Ricci-ļ¬‚at metric (excluding the ļ¬‚at torus). In 1976 the Chinese professor of mathematics at Harvard, Shing-Tung Yau, proved the existence of the geometric structure as surmised by the Calabiā€™s conjecture. He was able to prove, by expressing the conjecture in terms of non linear partial diļ¬€erential equations, the existence of many multi-dimensional shapes that are Ricci-ļ¬‚at, i.e., satisfying the Einstein equation in 3 complex dimensions and empty space. iii
  • 5. Chapter 0. Introduction Such structure is now called Calabi-Yau manifold. Its special properties are indispensable for compactiļ¬cation in Superstring Theory. This thesis provides an introduction to Calabi-Yau manifolds. We will not go into details of the physical problem of superstring theory but we will point out the mathematical aspect. The outline is as follows. In Chapter 1 and 2 we prepare the ground to understand and built the matter that will be later developed: respectively we give a short review of complex geometry and we analyse some algebraic geometry tools. More speciļ¬cally on the former chapter we describe the analogue of the diļ¬€erential manifold on the complex world: we give the holomorphic deļ¬nition of tangent and cotangent bundle leading us to the one of canonical bundle; while on the latter, following [Smi], we study the concept of divisor and sheaf that will be very useful in order to analyse some examples of Calabi-Yau. Chapter 3 follows up the topic developed in the ļ¬rst chapter, moreover it tries to connect the symplectic real world with the complex one, up to the deļ¬nition of symplectic complex manifold. In Chapter 4 we come to the deļ¬nition of Calaby-Yau manifold and we discuss several example focusing on some algebraic geometryā€™s ones. Following [Shu12] and [Raa10] we begin showing how a complex one dimensional torus (ļ¬rst example of one-dimensional Calabi-Yau) can be seen as an elliptic curve. In order to present two-dimensional examples of Calabi-Yau we refer to [Mor88] and [Kut10]. Concerning the higher dimension we set the problem on weighted projective space Pn(w) and we give an algeabric condition on a variety to be Calabi-Yau. Furthermore we give the deļ¬nition of KƤhler manifold in order to see how they are linked with the Calabi-Yau structures. Pursuing the aim of understanding the links between the complex world and the symplectic one, some famous counterexamples are showed in Chapter 5. Finally the dissertation ends with a view on compactiļ¬cations: the last chapter is an adaptation of [FT] and provides a glance on the fascinating dimension that we could not detect. iv
  • 6. Chapter 1 Complex manifolds This chapter is concerned with the theory of complex manifolds. Apparently, their deļ¬nition is identical to the one of smooth manifolds. It is only necessary to replace open subsets of Rn by open subsets of Cn, and smooth functions by holomorphic functions. Nevertheless we will illustrate how the complex and real worlds are fundamentally diļ¬€erent. We try to enter into this world because it provides the setting of our issue. 1.1 Complex coordinates We will give the deļ¬nition of complex manifold following the deļ¬nition of smooth manifold, with the appropriate changing. Deļ¬nition 1.1. M is a complex manifold of complex dimension n if: ā€¢ it is a topological space T2 and countable; ā€¢ it is endowed with a complex atlas i.e. a collection of coordinate charts F = {(Ui, Ļ•i)| i āˆˆ I}, where I is a set of index such that ā€“ Ui āŠ‚ M are open sets and {(Ui)}iāˆˆI is a cover of M i.e. M = iāˆˆI Ui; ā€“ Ļ•i : Ui ā†’ Cn is a homeomorphism onto an open set in Cn, ā€“ the change of coordinates Ļ•i ā—¦ Ļ•āˆ’1 j : Ļ•j(Ui āˆ© Uj) ā†’ Ļ•i(Ui āˆ© Uj) are biholomorphic, āˆ€ i, j s.t. Ui āˆŖ Uj = āˆ…; 1
  • 7. Chapter 1. Complex manifolds ā€“ the collection F is maximal towards the previous condition. This means that if (U, Ļ•) is a system of charts such that Ļ•i ā—¦ Ļ• and Ļ• ā—¦ Ļ•āˆ’1 i are biholomorphic āˆ€i āˆˆ I, then (U, Ļ•) āˆˆ F. In other words M is a diļ¬€erentiable manifold endowed with a holomorphic atlas. Observation 1.2. Charts deļ¬ne coordinates. Suppose Ļ•: U ā†’ Cn is some chart centered around a point p āˆˆ U (meaning p maps to the origin.) Then Ļ• can be written as local coordinates (z1, . . . , zn), where each zi is a holomorphic function on U. These are complex coordinates on M. Note that if M is a complex nāˆ’dimensional manifold, it can be realized as a real 2nāˆ’dimensional manifold with coordinates (xi, yi) coming from zj = xj + iyj. Example 1.3 (Complex Torus). Trivially Euclidean space Cn is a comlplex manifold. We can now consider the action of Zn on Cn by translation: (m1, n1, . . . , mn, nn) Ā· (z1, . . . , zn) = (z1 + m1 + in1, . . . , zn + mn + inn), where mi, ni āˆˆ Z, zi āˆˆ C, āˆ€i. Since this action is properly discontinuous and holomorphic, the quotient Cn/Zn inherits the structure of a complex manifold from the standard atlas on Cn. Thus Tn := Cn/Zn is a manifold, called complex torus. Example 1.4 (Complex projective space). CPn is the set of equivalence classes {[z] = [z0 : Ā· Ā· Ā· : zn], | zi āˆˆ C}, where y āˆˆ Cn and w āˆˆ Cn, belong to the same class iļ¬€ there exist Ī» āˆˆ C{0} such that y = Ī»w. Let Ui = {[z0 : Ā· Ā· Ā· : zn] | zi = 0} . An open cover of CPn is given by {Ui}i=0,...,n . The maps Ļ•i : Ui ā†’ Cn given by [z0 : Ā· Ā· Ā· : zn] ā†’ (z0/zi, . . . , zi/zi, . . . , zn/zi) are bijective, so it remains to show that the transition functions are holo- morphic. These are given by Ļ•ij := Ļ•i ā—¦ Ļ•āˆ’1 j : Ļ•j(Ui āˆ© Uj) ā†’ Ļ•i(Ui āˆ© Uj). Let wk be the coordinates on Uj and assume i < j. Then Ļ•ij is deļ¬ned by (w0, . . , Ė†wj, . . , wn) Ļ•āˆ’1 j āˆ’āˆ’ā†’ [w0 : . . : 1 : . . : wn] Ļ•i āˆ’ā†’ ( w0 wi , . . , Ė†wi wi , . . , 1 wi , . . , wn wi ). Since wi = 0 on Ļ•j(Ui āˆ© Uj), we see that the coordinate functions Ļ•ij are of the form wk/wi or 1/wi, which are holomorphic on the domain of Ļ•ij. So CPn is a complex manifold. 2
  • 8. Chapter 1. Complex manifolds 1.2 Tangent space Let M be a complex manifold of complex dimension n and x be a point of M. Let Cn be identiļ¬ed with R2n via the map (z1, . . . , zn) ā†’ (x1, y1 . . . , xn, yn). Let (z1, . . . , zn) be a local coordinate system near x with zj = xj + iyj, j = 1, . . . , n. Then the real tangent space Tx(M) is spanned by āˆ‚ āˆ‚x1 x , āˆ‚ āˆ‚y1 x , . . . , āˆ‚ āˆ‚xn x , āˆ‚ āˆ‚yn x . Deļ¬ne an Rāˆ’linear map J from Tx(M) onto itself by J āˆ‚ āˆ‚xi x = āˆ‚ āˆ‚yi x , J āˆ‚ āˆ‚yi x = āˆ’ āˆ‚ āˆ‚xi x for all j = 1, . . . , n. Obviously, we have J2 = āˆ’1, and J is called the complex structure on Tx(M). We observe that the deļ¬nition of J is independent of the choice of the local coordinates (z1, . . . , zn). The complex structure J induces a natural splitting of the complexiļ¬ed tangent space CTx(M) = Tx(M) āŠ—R C. First we extend J to the whole complexiļ¬ed tangent space by J(x āŠ— Ī±) = (Jx) āŠ— Ī±. It follows that J : CTx(M) ā†’ CTx(M) is a Cāˆ’linear map with eigenvalues i and āˆ’i. Denote by T1,0 x (M) and by T0,1 x (M) the eigenspaces of J corresponding to i and āˆ’i respectively. It is easily veriļ¬ed that ā€¢ T1,0 x (M) = T0,1 x (M), ā€¢ T1,0 x (M) āˆ© T0,1 x (M) = {0}, ā€¢ T1,0 x (M) is spanned by āˆ‚ āˆ‚z1 x , . . . , āˆ‚ āˆ‚zn x , where āˆ‚ āˆ‚zi x = 1 2 āˆ‚ āˆ‚xi āˆ’ i āˆ‚ āˆ‚yi x , for 1 ā‰¤ i ā‰¤ n. Consequently T0,1 x (M) is spanned by āˆ‚ āˆ‚ ĀÆz1 x , . . . , āˆ‚ āˆ‚ ĀÆzn x , 3
  • 9. Chapter 1. Complex manifolds Any vector v āˆˆ T1,0 x (M) is called a vector of type (1, 0), and we call v āˆˆ T0,1 x (M) a vector of type (1, 0). The space T1,0 x (M) is called the holomorphic tangent space at x. Therefore CTx(M) = Tx(M) āŠ—R C = T1,0 x (M) āŠ• T0,1 x (M). Elements of CTx(M) can be realized as Cāˆ’linear derivations in the ring of complex valued Cāˆž functions on M around x. Indeed, given the real derivation v āˆˆ Tx(M), the elementary tensor v āŠ— z āˆˆ Tx(M) āŠ—R C acts on f + ig in the following way (v āŠ— z)(f + ig) = z Ā· (v(f) + iv(g)). 1.3 Complex forms: the (p,q)-forms 1.3.1 1-forms Let CTāˆ— x (M) be the dual space of CTx(M). By duality, J also induces a splitting on Cāˆ— Tx(M) = T1,0 x āˆ— (M) āŠ• T0,1 x āˆ— (M) := Ī›1,0 x (M) āŠ• Ī›0,1 x (M), (1.1) where Ī›1,0 x (M) and Ī›0,1 x (M) are eigenspaces corresponding to the eigenvalues i and āˆ’i respectively. Let U be an open set containing x and let z1, . . . , zn be local coordinates. It is easy to see that the vectors (dz1 )x, . . . , (dzn )x span Ī›1,0 x (U) and that the space Ī›0,1 x (U) is spanned by (dĀÆz1 )x, . . . , (dĀÆzn )x. This means that any diļ¬€erential 1-form with complex coeļ¬ƒcients can be written uniquely as a sum n j=1 fjdzj + gjdĀÆzj where fj, gj āˆˆ Cāˆž(U), i.e. fj, gj : U ā†’ C are holomorphic. Because of this splitting the complex 1-forms are also called the (1, 1)-forms. Example 1.5. Let f : M ā†’ C be a smooth function, the exterior derivative d = āˆ‚ + ĀÆāˆ‚ is deļ¬ned by df = n j=1 āˆ‚f āˆ‚zj dzj + āˆ‚f āˆ‚ ĀÆzj dĀÆzj = āˆ‚f + ĀÆāˆ‚f. 4
  • 10. Chapter 1. Complex manifolds 1.3.2 Higher degree forms Let p and q be a pair of non-negative integers ā‰¤ n. One can deļ¬ne the wedge product of complex diļ¬€erential forms in the same way as with real forms. The space Ī› (p,q) x of (p, q)-forms is deļ¬ned by taking linear combinations of the wedge products of p elements from Ī› (1,0) x (M) and q elements from Ī› (0,1) x (M), i.e., Ī›(p,q) x (M) = Ī›(1,0) x (M) āˆ§ Ā· Ā· Ā· āˆ§ Ī›(1,0) x (M) pāˆ’times āˆ§ Ī›(0,1) x (M) āˆ§ Ā· Ā· Ā· āˆ§ Ī›(0,1) x (M) qāˆ’times . At this point the isomorphism (1.1) leads to the following splitting Ī›k x(M) := Ī›k x(Cāˆ— Tx(M)) = Ī›k (T1,0 x āˆ— (M) āŠ• T0,1 x āˆ— (M)) (1.2) = p+q=k Ī›p x(T1,0 x āˆ— (M)) āˆ§ Ī›q x(T0,1 x āˆ— (M)) (1.3) = p+q=k Ī›(p,q) x (M). (1.4) This means that given u a (p, q)āˆ’form, it can be written in local coordinates in an open set U as u(z) = |I|=p,|J|=q uIJ (z)dzI āˆ§ dĀÆzJ where uIJ āˆˆ Cāˆž(U.) If we consider the disjoint union of the Ī›k x(M) over the points of the manifold, we get the space of all complex diļ¬€erential forms of total degree k, i.e., Ī›k (M) = xāˆˆM Ī›k x(M). Example 1.6. Let u be a (p, q)āˆ’form. We can deļ¬ne de following operator āˆ‚ : Ī›(p,q) (M) ā†’ Ī›(p+1,q) (M) and ĀÆāˆ‚ : Ī›(p,q) (M) ā†’ Ī›(p,q+1) (M) (1.5) and respectively in an open coordinate set their actions are given by āˆ‚u = I,J 1ā‰¤kā‰¤n āˆ‚uI,J āˆ‚zk dzk āˆ§ dzI dzJ , ĀÆāˆ‚u = I,J 1ā‰¤kā‰¤n āˆ‚uI,J āˆ‚ĀÆzk dĀÆzk āˆ§ dzI dzJ . According to the previous consideration we can consider the diļ¬€erential operator d = āˆ‚ + ĀÆāˆ‚. (1.6) It satisfy ā€¢ d: Ī›k(M) ā†’ Ī›k+1(M), āˆ€k 5
  • 11. Chapter 1. Complex manifolds ā€¢ d2 = 0 ā€¢ if Ļ‰ āˆˆ Ī›r(M) and Ī± āˆˆ Ī›s(M) then d(Ļ‰ āˆ§ Ī±) = dĻ‰ āˆ§ Ī±(āˆ’1)rĻ‰ āˆ§ dĪ± ā€¢ if f āˆˆ Cāˆž(U) then the operator is the same as in the Example 1.5. Remark. If M is a complex manifold, as in the case of the real manifolds, Ī›k(M) deļ¬nes a complex vector bundle on M, of "complex" rank n k . In the following section we will recall the deļ¬nition in the real case and it will turn out that it is a useful construction to study complex manifolds. 1.4 Canonical bundle Let M be a real smooth manifold and let Tāˆ—M be its cotangent bundle. As we saw in the complex case, the vector bundle of the rāˆ’forms on M is deļ¬ned by Ī›r (M) = xāˆˆM Ī›r x(Tāˆ— x M) := xāˆˆM Ī›r x(M) and it has rank n r as real vector bundle. Deļ¬nition 1.7 (Canonical bundle). Let r = n. The bundle of the nāˆ’forms of a smooth manifold of dimension n is called the canonical bundle. Since n n = 1 the canonical bundle is a line bundle. Deļ¬nition 1.8 (Holomorphic canonical bundle). Let M be a complex man- ifold. The holomorphic line bundle Ī›n (M) := K is called the canonical line bundle. Our aim will be to investigate manifold with trivial canonical bundle. It is known that in the real case this condition is equivalent to: Ī›n (M) is trivial ā‡ā‡’ āˆƒ a nowhere vanishing section ā‡ā‡’ M is orientable. The second condition means that there exist a diļ¬€erentiable n form Ļ‰ nowhere vanishing, i.e., Ļ‰(p) = 0, for every p in M. While the ļ¬rst condition still applies in the complex case, since it is a topological condition, the second equivalence it is not still true, as the following theorem proves. Theorem 1.9. Any complex manifold M of dimension n is orientable. 6
  • 12. Chapter 1. Complex manifolds Proof. Let (Ui, Ļ•i)iāˆˆI be a system of holomorphic coordinates around a point p, and Ļ•i : Ui ā†’ Ļ•(Ui), p ā†’ (z1(p), . . . , zn(p)), Ļ•j : Uj ā†’ Ļ•(Uj), p ā†’ (Z1(p), . . . , Zn(p)). By identifying Cn with R2n in the usual way we can write: (z1, . . . , zn) = (x1, y1, . . . , xn, yn) and (Z1, . . . , Zn) = (X1, Y1, . . . , Xn, Yn) and thus we obtain a real manifold structure on M. We will now calculate the Jacobian matrices of the transition functions for both structures. In the complex case we have JacĻ•i(p)(Ļ•j ā—¦ Ļ•āˆ’1 i ) = āˆ‚(Ļ•j ā—¦ Ļ•āˆ’1 i )k āˆ‚zl 1ā‰¤k,lā‰¤n = āˆ‚Zk(z1, . . . , zn) āˆ‚zl 1ā‰¤k,lā‰¤n = (ckl)1ā‰¤k,lā‰¤n āˆˆ GL(n, C). Instead in the real case we will ļ¬nd JacĻ•i(p)(Ļ•j ā—¦ Ļ•āˆ’1 i ) = ļ£« ļ£­ āˆ‚Xk(x1,y1...,xn,yn) āˆ‚xl āˆ‚Xk(x1,y1...,xn,yn) āˆ‚yl āˆ‚Yk(x1,y1...,xn,yn) āˆ‚xl āˆ‚Yk(x1,y1...,xn,yn) āˆ‚yl ļ£¶ ļ£ø 1ā‰¤k,lā‰¤n . Using the Cauchy-Riemann conditions, this coincides with Re(ckl) āˆ’Im(ckl) Im(ckl) Re(ckl) 1ā‰¤k,lā‰¤n āˆˆ GL(2n, R). We will now calculate the determinant of these matrices and we will see that it is always positive, which is equivalent to state that M is orientable as a real manifold. Consider the following homomorphism Ļ: Mn(C) ā†’ M2n(R), deļ¬ned by (ckl)1ā‰¤k,lā‰¤n ā†’ Re(ckl) āˆ’Im(ckl) Im(ckl) Re(ckl) 1ā‰¤k,lā‰¤n . The map Ļ is continuous, since it is Rāˆ’linear and the spaces involved are ļ¬nite dimensional. Also, being a Lie algebra homomorphism, we have det(Ļ(Pāˆ’1 AP)) = det(Ļ(Pāˆ’1 )Ļ(A)Ļ(P)) = det(Ļ(A)). 7
  • 13. Chapter 1. Complex manifolds Finally, the diagonalizable matrices are dense in Mn(C), so we can restrict our calculations to diagonal matrices in Mn(C). Therefore we obtain: det(Ļ(Diag(c1, . . . , cn))) = det Diag Re(c1) āˆ’Im(c1) Im(c1) Re(c1) , . . . , Re(cn) āˆ’Im(cn) Im(cn) Re(cn) = n i det Re(ci) āˆ’Im(ci) Im(ci) Re(ci) = n i |ci|2 = | det (Diag(c1, . . . , cn)) |2 . So we can conclude that, det(Ļ(A)) = | det(A)|2 , āˆ€A āˆˆ Mn(C). Finally, we get that the Jacobian matrices of the transition functions for the chart Ļ•i for M have positive determinants, thus the real underlying manifold M is orientable. Synthetically if we want to verify if a holomorphic canonical bundle is trivial, the following equivalence applies: Ī›n (M) is trivial ā‡ā‡’ āˆƒ a nowhere vanishing section of Ī›n (M). The nowhere vanishing section is also called a holomorphic volume form. Example 1.10. We are going to examine the triviality of the canonical bundle of some well-known manifolds. ā€¢ The canonical bundle of a complex nāˆ’dimensional torus K ā†’ Tn is trivial. In fact we can explicitly show a nowhere vanishing section of its canonical bundle. With the same notation used in Example 1.3, we can see that dz1 āˆ§ Ā· Ā· Ā· āˆ§ dzn is a non vanishing complex nāˆ’form belonging to Ī›n(M). ā€¢ The canonical bundle of the projective line K ā†’ CP1 is not trivial. One could prove that using the general following property for a closed oriented Riemann surface M: Ī›n (M) is trivial ā‡” Ļ‡(M) = 0. (1.7) Indeed Ļ‡(CP1) = 2, but we will not go into details. Furthermore asking K to be trivial is equivalent to ask TCP1 to be trivial, and as we can 8
  • 14. Chapter 1. Complex manifolds Figure 1.1: Non vanishing section on torus. Figure 1.2: As we can see from the picture it is not possible to ļ¬nd a holomorphic non vanishing form on the sphere: the poles correspond to the two pots with zero net ļ¬‚ow. 9
  • 15. Chapter 1. Complex manifolds see from ļ¬gure 1.2 it is not possible to ļ¬nd a non vanishing form on this manifold1. In fact since the following proposition applies we have that the tangent bundle of CP1 is diļ¬€eomorphic to the tangent bundle of S2 thus it is not trivial. Proposition 1.11. The complex projective line CP1 is diļ¬€eomorphic to the 2-sphere S2. Proof. Consider a point [z : w] with w = 0. Then we send this point to z w āˆˆ C. We use then the stereographic projection C ā†’ S2N, and we send the point [1 : 0] to the north pole of S2. It is easy to check that the composition CP1 ā†’ Cāˆ— ā†’ S2 is a diļ¬€eomorphism. The inverse of the ļ¬rst map sends a point h āˆˆ C to h 1 + |h|2 , 1 1 + |h|2 and inļ¬nity to [1, 0], so we can easily obtain the inverse S2 ā†’ CP1 explicitly. ā€¢ The canonical bundle of a Riemann surface Ī£, K ā†’ Ī£ is not trivial if its genus, g(M), is greater than one. One could see this using (1.7) together with the fact that for a closed Riemann surface its genus is related to the Euler characteristic by the formula 2 āˆ’ 2g(M) = Ļ‡(M). 1.4.1 Line bundles over M We should now open a parenthesis on line bundles in order to clarify some ideas and to understand the potential of these structures. Let M be a complex manifold and L1, L2, L3 be elements of the set G = {isomorphism classes of all holomorphic line budle overM} . They satisfy: 1 This topic is closely related to the problem of combing a hairy sphere, see "hairy ball theorem" for further details. 10
  • 16. Chapter 1. Complex manifolds 1. Closure: if L1 and L2 are holomorphic line bundles then the tensor product L1 āŠ— L2 is a holomorphic line bundle. In fact by deļ¬nition the tensor product of two bundles is the bundle whose ļ¬ber is the tensor product of the ļ¬ber. Therefore, L1 āŠ— L2 has ļ¬ber (p, C) āŠ— (p, C)), āˆ€p. That is because C āŠ— C = C since C āŠ— C = C āŠ— Cāˆ— = Hom(C, C). Moreover if we consider the following isomorphism Ī¦: C ā†’ Hom(C, C) a ā†’ Ī¦(a): C ā†’ C b ā†’ a Ā· b we conclude that Hom(C, C) = C. 2. Associativity: (L1 āŠ— L2) āŠ— L3 = L1 āŠ— (L2 āŠ— L3) 3. Identity element idL1 : it is the trivial bundle C Ɨ M since L1 āŠ— C Ɨ M = L1. 4. Inverse element: Lāˆ— 1. Since idL1 āˆˆ Hom(L1, L1) = idL1 . idL1 is a nowhere vanishing section of idL1 and this means that L1 āŠ— Lāˆ— 1 = C. 5. Commutativity: L1 āŠ— L2 = L2 āŠ— L1, even if they are diļ¬€erent bundles, they are isomorphic. According to the consideration we have made, we can conclude that the space of line bundles modulo equivalence forms a group under the tensor product. Often instead of studying ļ¬ber bundles it is convenient to analyse their smooth sections. In the case of the holomorphic line bundles, the holomorphic sections correspond to the holomorphic functions f : M ā†’ C. But this is quite useless since, as we will prove, holomorphic functions over a compact connected manifold have to be constant. First we will examine the following lemma that will let us to prove what we claimed. 11
  • 17. Chapter 1. Complex manifolds Lemma 1.12. Let B be an open set of Cn and f : B ā†’ C. If |f| has a maximum in B, then f is constant. Proof. Let x, y āˆˆ Rn such that z = x + iy āˆˆ Cn and f(x, y) = u(x, y) + iv(x, y) āˆˆ Cn. If |f| has a maximum, then |f|2 = u(x, y)2 + v(x, y)2 has a maximum. Since |f|2 is a real valued function, to ļ¬nd its stationary point, we set the partial derivative to zero, namely ļ£± ļ£² ļ£³ āˆ‚|f|2 āˆ‚xi = 2 āˆ‚u āˆ‚xi Ā· u + 2 āˆ‚v āˆ‚xi Ā· v = 0 āˆ‚|f|2 āˆ‚yi = 2 āˆ‚u āˆ‚yi Ā· u + 2 āˆ‚v āˆ‚yi Ā· v = 0 , āˆ€i āˆˆ {1, . . . , n}. (1.8) By linear algebra, solving the above system is equal to impose: ļ£« ļ£­ āˆ‚u āˆ‚xi i āˆ‚v āˆ‚xi i āˆ‚u āˆ‚yi i āˆ‚v āˆ‚yi i ļ£¶ ļ£ø u v = 0, āˆ€i āˆˆ {1, . . . , n}, (1.9) and using the Cauhy-Riemann conditions this is equivalent to ask ļ£« ļ£­ āˆ‚u āˆ‚xi i āˆ’ āˆ‚u āˆ‚yi i āˆ‚u āˆ‚yi i āˆ‚u āˆ‚xi i ļ£¶ ļ£ø u v = 0 āˆ€i āˆˆ {1, . . . , n}. (1.10) It is easily seen that the last matrix has rank 1, so the condition (1.10) is equivalent to impose āˆ‚u āˆ‚xi 2 + āˆ‚u āˆ‚yj 2 = 0, āˆ€i, j. (1.11) Hence āˆ‚u āˆ‚xi = āˆ‚u āˆ‚yj = 0, āˆ€i, j. (1.12) Therefore we are asking f to be constant. Theorem 1.13. Let M be a complex compact and connected manifold, f : M ā†’ C be a holomorphic function. Then f is constant. Proof. Since f is holomorphic, |f| is a continuous function deļ¬ned on a compact set and for the Weirstrass theorem |f| has a maximum on M. This means that if (UĪ±, Ļ•Ī±) is a ļ¬nite cover of M around the point p where f reaches his maximum, the function |f ā—¦ Ļ•āˆ’1 Ī± |: Ļ•Ī±(UĪ±) āŠ† Cn ā†’ C has a maximum. Using Lemma 1.12 we get that f is constant on UĪ±. Now we can conclude considering that every holomorphic function is an analytic function, more speciļ¬cally it is a continuous function on UĪ±. Since f is constant on UĪ± āˆŖUĪ², āˆ€Ī±, Ī², because of the uniqueness of the Taylor expansion f is constant on M. 12
  • 18. Chapter 1. Complex manifolds 1.5 Cohomologies Before turning to the Dolbeault cohomology of complex manifolds, we will give a very brief summary of these concepts in the real situation which, of course, also applies to complex manifolds if they are viewed as real analytic manifolds. Deļ¬nition 1.14. A sequence Cāˆ— = (Cn, āˆ‚n)nāˆˆZ of modules Cn over a ring R and homomorphisms āˆ‚n : Cn ā†’ Cnāˆ’1 is called a chain complex, if for all n āˆˆ Z we have that āˆ‚nāˆ’1 ā—¦ āˆ‚n = 0 holds. The āˆ‚n functions are usually called the boundary operators or diļ¬€er- entials. A chain complex is usually visualised in a diagram as such, Ā· Ā· Ā· āˆ‚n+1 āˆ’ā†’ Cn āˆ‚n āˆ’ā†’ Cnāˆ’1 āˆ‚nāˆ’1 āˆ’ā†’ Ā· Ā· Ā· Observation 1.15. Since āˆ‚nāˆ’1 ā—¦ āˆ‚n = 0 we immediately get that Im(āˆ‚n) āŠ‚ ker(āˆ‚nāˆ’1). Note that these are both submodules of Cn. Deļ¬nition 1.16. The n-cycles of a chain complex Cāˆ— is Zn(Cāˆ—) = ker(āˆ‚n). The n-boundaries of a chain complex Cāˆ— is Bn(Cāˆ—) = Im(āˆ‚n+1). The n-th homology module of a chain complex Cāˆ— is Hn(Cāˆ—) = Zn/Bn. Similarly one can deļ¬ne a cochain to consist of the modules of Rāˆ’linear maps from your modules to R, together with special boundary maps dn : Cn ā†’ Cn+1. 1.5.1 The de Rham cohomology In order to apply what we have seen to manifolds we will use for Rāˆ’modules in this case real vector spaces, namely the vector spaces of diļ¬€erential kāˆ’forms Ī›k(M), for all k = 1 . . . n, if n is the dimension of the manifold. 13
  • 19. Chapter 1. Complex manifolds Deļ¬nition 1.17. A diļ¬€erential form Īø is called closed if dĪø = 0. And a diļ¬€erential kāˆ’form Ī± is exact if there exist a diļ¬€erential (k āˆ’ 1)āˆ’form Ī² such that dĪ² = Ī±. Observation 1.18. Note that because d ā—¦ d = 0 we have that all exact forms are also closed. Lemma 1.19. Let M be a smooth manifold, the following diagram is a cochain Ā· Ā· Ā· d āˆ’ā†’ Ī›nāˆ’1 (M) d āˆ’ā†’ Ī›n (M) d āˆ’ā†’Ī›n+1 (M) d āˆ’ā†’ Ā· Ā· Ā· Observation 1.20. The n-cycles are exactly the closed n-forms on M. And the n-boundaries are the exact n-forms on M. Observation 1.21. We also use the fact that Ī›n = 0 for n > dim(M). Proof. Note that Ī›n(M) is a real vector space, thus an Rāˆ’module. And d is a Rāˆ’linear map. Furthermore d ā—¦ d = 0 which is the same as saying Im(d) āŠ‚ ker(d). Thus we are dealing with a cochain. Deļ¬nition 1.22. The p āˆ’ th de Rham cohomology group is equal to the p āˆ’ th cohomology groups of the cochain in Lemma 1.19. This is usually denoted Hp deRham(M). We will now analyse a property that can be applied to symplectic mani- folds. Proposition 1.23. Let M be a 2n dimensional oriented diļ¬€erentiable man- ifold which is compact and without boundary. For all 0 ā‰¤ p ā‰¤ n it exists an isomorphism Hp deRham(M) = Hnāˆ’p deRham(M) āˆ— . In particular bk(M) = bnāˆ’k(M). For a proof of the theorem and for a general clariļ¬cation the reader is referred to [Huy06]. To be more precise we should write Hp deRham(M, R) instead of Hp deRham(M). The symbol R is used here to stress that we are considering real valued pāˆ’ forms; of course one can introduce a similar group Hp deRham(M, C) for complex valued forms, i.e. forms with values in C āŠ— Ī›p(M). Then Hp deRham(M, C) = C āŠ— Ī›p (M) (1.13) is the complexiļ¬cation of the real De Rham cohomology group. 14
  • 20. Chapter 1. Complex manifolds 1.5.2 Dolbeault cohomology Most of the facts about homology and De Rham cohomology on real manifolds are also valid on complex manifolds if one views them as real analytic manifolds. However one can use the complex structure to deļ¬ne as we have seen in (1.5) the āˆ‚-cohomology or Dolbeault cohomology. With the same notation as those in Chapter 3 ĀÆāˆ‚ : Ī›(p,q) (M) ā†’ Ī›(p,q+1) (M). In particular, we get diļ¬€erential cochain complexes Ā· Ā· Ā· ĀÆāˆ‚ āˆ’ā†’ Ī›(p,qāˆ’1) (M) ĀÆāˆ‚ āˆ’ā†’ Ī›(p,q) (M) ĀÆāˆ‚ āˆ’ā†’Ī›(p,q+1) (M) ĀÆāˆ‚ āˆ’ā†’ Ā· Ā· Ā· Deļ¬nition 1.24. We say that a (p, q)āˆ’form Ī± is ĀÆāˆ‚-closed if ĀÆāˆ‚Ī± = 0. The space of ĀÆāˆ‚-closed (p, q)āˆ’forms is denoted by Z (p,q) ĀÆāˆ‚ (M). A (p, q)āˆ’form Ī² is ĀÆāˆ‚-exact if it is of the form Ī² = ĀÆāˆ‚Ī³ for Ī³ āˆˆ Ī›(p,qāˆ’1)(M). Observation 1.25. Since ĀÆāˆ‚2 = 0, ĀÆāˆ‚(Ī›(p,q)(M)) āŠ‚ Z (p,q+1) ĀÆāˆ‚ (M). Deļ¬nition 1.26. Dolbeault cohomology groups are then deļ¬ned as H (p,q) ĀÆāˆ‚ (M) = Z (p,q) ĀÆāˆ‚ (M) ĀÆāˆ‚(Ī›(p,qāˆ’1)(M)) . (1.14) Deļ¬nition 1.27. The dimensions of the (p, q) cohomology groups are called Hodge numbers h(p,q) (M) = dimC H (p,q) ĀÆāˆ‚ (M). They are ļ¬nite for compact complex manifolds. The Hodge numbers of a compact complex manifold are often arranged in the Hodge diamond: h0,0 h1,0 h0,1 h2,0 h1,1 h0,2 h3,0 h2,1 h1,2 h0,3 h3,1 h2,2 h1,3 h3,2 h2,3 h3,3 which we have displayed here for a three complex dimensional manifold. The general diagrams take the following form 15
  • 21. Chapter 1. Complex manifolds hn,n hn,nāˆ’1 hnāˆ’1,n hn,nāˆ’2 hnāˆ’1,nāˆ’1 hnāˆ’2,n . . . ... . . . h2,0 h1,1 h0,2 h1,0 h0,1 h0,0 The decomposition (1.4) does not carry over the cohomology group in fact we have that dim(Hk deRham(M, C)) ā‰¤ p+q=k h(p,q)(M) . (1.15) We will see in Chapter 4 that the equality (1.15) is for a special types of complex manifolds, namely Calabi-Yau manifolds and KƤhler manifolds. Deļ¬nition 1.28. dim(Hk deRham(M, C)) is called the k āˆ’ th Betty number. 16
  • 22. Chapter 2 Some Algebraic Geometry tools In this chapter we analyse some tools of algebraic geometry that will be used in the following discussion in order to understand the algebraic aspect of some Calaby-Yau manifolds. 2.1 Aļ¬ƒne and projective variety Let k be an algebraically closed ļ¬eld and ļ¬x S āŠ† k[x1, . . . , xn]. Let An k be the n-dimensional aļ¬ƒne space over k. Deļ¬nition 2.1. The set V(S) = {p āˆˆ An k | f(p) = 0 āˆ€f āˆˆ S}. is called aļ¬ƒne algebraic set. Let I(S) the ideal generated by S, i.e., the smallest ideal of k[x1, . . . , xn] which contains S. Observation 2.2. We have that V(S) = V(I(S)). Moreover the ring k[x1, . . . , xn] is nƶetherian, thus every ideal is ļ¬nitely generated. For every ideal I of k[x1, . . . , xn] there exist polynomials f1, . . . , fr āˆˆ k[x1, . . . , xn] such that V(I) = V(f1, . . . , fr) = {(a1, . . . , an) āˆˆ An k | fi(a1, . . . , an) = 0, 1 ā‰¤ i ā‰¤ r}. Deļ¬nition 2.3. The Zariski topology on An k is the topology whose closed set are the algebraic subset of An k . If X āŠ‚ An k the Zariski topology on X is the one induced by the Zariski topology on An k . 17
  • 23. Chapter 2. Some Algebraic Geometry tools We can now give the deļ¬nition of reducible and irreducible set which are purely topological concepts. Deļ¬nition 2.4. Let X be a topological space and Y āŠ† X such that Y = āˆ…. Y is irreducible if it is not the union of two closed proper subsets of Y. A subset of X is reducible if it is not irreducible. Thus every non empty set can be expressed as union of irreducible subsets Y = Yi where Yi are the irreducible components. Deļ¬nition 2.5. An algebraic variety over the ļ¬eld k is an irreducible closed subset of An k , endowed with the Zariski topology. An open subset of an aļ¬ƒne variety is called quasi-aļ¬ƒne variety. The deļ¬nition we have previously given can be extended, in a natural way, to the projective space. The projective space Pn k can be considered an extension of the aļ¬ƒne space by the following immersion An k ā†’ Pn k (a1, . . . , an) ā†’ (1, a1, . . . , an) Let k be an algebraically closed ļ¬eld and consider the projective space Pn k . Let R = k[x1, . . . , xn, xn+1] be the ring of polynomials in n + 1 indeterminates and f be an homogeneous polynomial of R, then it makes sense to ask whether or not f(p) = 0 for a point p āˆˆ Pn. As in the aļ¬ƒne case we have the following deļ¬nition. Deļ¬nition 2.6. A projective algebraic subset of Pn k is the common zero set of a collection of homogeneous polynomials in R. We can deļ¬ne as well projective varieties and the Zariski topology in Pn k . 2.2 Regular and rational functions Fix X āŠ‚ An k , algebraic set. Deļ¬nition 2.7. A function X ā†’ k is regular if it agrees with the restriction to X of some polynomial function on the ambient An k . 18
  • 24. Chapter 2. Some Algebraic Geometry tools The set of all regular functions on X has a natural ring structure (where addition and multiplication are the functional notions). This is the coordinate ring of X, denoted k[X]. For all open set U āŠ‚ X, we will denote OX(U) (or simply O(U)) the set of regular functions on U. Whereas sum and product of regular function is a regular function, the set OX(U) is actually a ring, rather a kāˆ’algebra. We call OX(U) the sheaf of regular functions on U. Fix now an aļ¬ƒne algebraic set X and assume that X is irreducible. Even though we will not prove the following fact we have to remark that X is irreducible ā‡ā‡’ k[X] is a domain. Deļ¬nition 2.8. The function ļ¬eld of X is the fraction ļ¬eld of k[X], denoted k(X). Deļ¬nition 2.9. A rational function on X is an element Ļ• āˆˆ k(X) i.e., Ļ• is an element of the equivalence class f g , where f, g āˆˆ k[X], g = 0. Here f g āˆ¼ f g ā‡ā‡’ fg = gf as elements of k[X]. A rational function Ļ• āˆˆ k(X) is regular at p āˆˆ X if it admits a representation Ļ• = f g where g(p) = 0. The domain of deļ¬nition of Ļ• āˆˆ k(X) is the locus of all points p āˆˆ X where Ļ• is regular. The deļ¬nition or regular and rational function is a little diļ¬€erent from the one of the aļ¬ƒne case. Deļ¬nition 2.10. Let X āŠ‚ Pn k an algebraic set and U an open subset of X. A function f is regular around a point p āˆˆ U if there exist an open neighbourhood V of p in U and two homogeneous polynomials g, h of the same degree such that for all (a1, . . . , an+1) āˆˆ V, h(a1, . . . , an+1) = 0 and f|V = g h . The function f is regular in U if it is regular in every points of U. As in the aļ¬ƒne case, for every open subset U of X we denote OX(U) (or O(U)) the ring of the regular functions in U. 19
  • 25. Chapter 2. Some Algebraic Geometry tools 2.3 Divisors Let X be an irreducible variety. Deļ¬nition 2.11. A prime divisor or irreducible divisor on X is a codimen- sion 1 irreducible (closed) subvariety of X. A divisor D on X is a formal Zāˆ’linear combination of prime divisors D = t i=1 kiDi, ki āˆˆ Z. In P2, C = V(xy āˆ’ z2), L1 = V(x) and L2 = V(y) are prime divisors, while 2C, 2L1 āˆ’ L2 are divisors which are not prime. We say a divisor D = t i=1 kiDi is eļ¬€ective if each ki ā‰„ 0. The support of D is the list of prime divisors occurring in D with non-zero coeļ¬ƒcient. The set of all divisors on X form a group Div(X), the free abelian group on the set of prime divisors of X. The zero element is the trivial divisor D = 0Di, and Supp(0) = āˆ…. Example 2.12. Consider Ļ• = f g = (t āˆ’ Ī»1)a1 Ā· Ā· Ā· (t āˆ’ Ī»n)an (t āˆ’ Āµ1)b1 Ā· Ā· Ā· (t āˆ’ Āµm)bm āˆˆ k(A1 ) = k(t) where f, g āˆˆ k[t]. The divisor of zeros and poles of Ļ• is a1{Ī»1} + a2{Ī»2} + Ā· Ā· Ā· an{Ī»n} Divisors of zeroes āˆ’ b1{Āµ1} āˆ’ Ā· Ā· Ā· āˆ’ bm{Āµm} Divisors of poles . Example 2.13. Let An = X. A prime divisor is D = V(h), where h āˆˆ k[x1, . . . , xn] is irreducible. Write Ļ• = f g = fa1 1 Ā· Ā· Ā· fan n gb1 1 Ā· Ā· Ā· gbm m āˆˆ k(An ) = k(x1, . . . , xn), where f, g āˆˆ k[x1, . . . , xn] and fi, gi irreducible, ai āˆˆ N. Denoting the divisor of zeros and poles of Ļ• by div(Ļ•), we have div(Ļ•) = a1V(f1) + a2V(f2) + Ā· Ā· Ā· + anV(fn) āˆ’ b1V(g1) āˆ’ Ā· Ā· Ā· āˆ’ bmV(gm) 20
  • 26. Chapter 2. Some Algebraic Geometry tools On almost any X, we will associate to each Ļ• āˆˆ k(X){0} some divisor, div(Ļ•), the divisor of zeros and poles, in such a way that the map k(X)āˆ— = k(X){0} ā†’ Div(X) Ļ• ā†’ div(Ļ•) = DāŠ†X prime Ī½D(Ļ•) Ā· D preserves the group structure on k(X)āˆ—, i.e., (Ļ•1 ā—¦ Ļ•2) ā†’ div(Ļ•1) + div(Ļ•2). The image of this map will be the group of principal divisors: P(X) āŠ† Div(X). We will write div(Ļ•) = DāŠ†X prime Ī½D(Ļ•) Ā· D where Ī½D(Ļ•) = ord(Ļ•) which corresponds to the order of vanishing of Ļ• along D and it is computed as follows: take u1, . . . , un local coordinates for a point x āˆˆ D; write Ļ• = fdu1 āˆ§ Ā· Ā· Ā· āˆ§ dun, where f āˆˆ k(X). Then Ī½D(Ļ•) = Ī½D(f). We should thus focus on the deļ¬nition of order of vanishing of Ļ• āˆˆ k(X){0} along a prime divisor D, denoted Ī½D(Ļ•). Assuming that X is non-singular in codimension 1, we distinguish two cases. Case 1.Let X be aļ¬€ne, Ļ• āˆˆ k[X], D = V (Ļ€) is a hypersurface deļ¬ned by Ļ€ āˆˆ k[X]. We say that Ļ• vanishes along D provided that D = V(Ļ€) āŠ† V(Ļ•). So by the Nullstellensatz, (Ļ•) āŠ† (Ļ€). Deļ¬nition 2.14. The order of vanishing of Ļ• along D, denoted Ī½D(Ļ•), is the unique integer k ā‰„ 0 such that Ļ• āˆˆ (Ļ€k)(Ļ€k+1). Observation 2.15. Ī½D(Ļ•) = 0 =ā‡’ Ļ• āˆˆ (Ļ€0)(Ļ€1) = k[X](Ļ€), i.e., Ļ• does not vanish on all of D. If Ļ• is rational and Ļ• = f g , where f, g āˆˆ k[X], deļ¬ne Ī½D(Ļ•) = Ī½D(f) āˆ’ Ī½D(g). Case 2.General case: Ļ• āˆˆ k(X){0}, D āŠ† X arbitrary prime divisor. Choose U āŠ† X open aļ¬ƒne such that ā€¢ U is smooth; 21
  • 27. Chapter 2. Some Algebraic Geometry tools ā€¢ U āˆ© D = āˆ…; ā€¢ D is a hypersurface: D = V(Ļ€) for some Ļ€ āˆˆ k[U] = OX(U)). We have Ļ• āˆˆ k(X) = k(U). Deļ¬ne Ī½D(Ļ•) as in case 1. Example 2.16. Let Ļ• = x y āˆˆ k(x, y) = k(A2) we have that div(Ļ•) = DāŠ†A2 prime Ī½D x y D where Ī½D x y is 0 for all divisors D except for L1 = V(x), where the order of vanishing is 1, and L2 = V(y), where Ī½L2 (Ļ•) = āˆ’1. 2.4 Rational diļ¬€erential forms and canonical divi- sors. A rational diļ¬€erential form on X is intuitively f1dg1 + Ā· Ā· Ā· + frdgr, where fi and gi are rational functions on X. Formally: Deļ¬nition 2.17. A rational diļ¬€erential form on X is an equivalence class of pairs (U, Ļ•) where U āŠ† X is open and Ļ• āˆˆ ā„¦X(U) and (U, Ļ•) āˆ¼ (U , Ļ• ) ā‡ā‡’ Ļ•|Uāˆ©U = Ļ• Uāˆ©U We can deļ¬ne the divisor of a rational diļ¬€erential form. If Ļ‰ is a rational diļ¬€erential form on X, then div(Ļ‰) āˆˆ Div(X) is called a canonical divisor. The canonical divisors form a linear equivalence class on X, denoted KX. We are going to present another way to deļ¬ne the canonical divisor of a compact complex manifold. Deļ¬nition 2.18. Let X be a compact complex manifold and Yi āŠ‚ X codimension 1 subvarieties then we deļ¬ne the canonical divisor of X KX = niYi where ni āˆˆ Z. From this deļ¬nition we can glimpse that there exist a connection between the canonical divisors and the canonical bundle of a complex manifold but only later we will clarify this link. A diļ¬€erential form Ļˆ on X is regular if āˆ€x āˆˆ X, there is an open neighborhood U such that x āˆˆ U and Ļˆ|U agrees with t i=1 gidfi, where fi, gi āˆˆ OX(U). In other words, viewing Ļˆ as a section of the cotangent bundle of X, the section map is regular. 22
  • 28. Chapter 2. Some Algebraic Geometry tools Example 2.19. The diļ¬€erential form Ļˆ = 2xd(xy) = 2x(xdy + ydx) = 2x2 dy + 2xydy is a regular diļ¬€erential form in A2. Deļ¬nition 2.20. For U āŠ‚ X open, let ā„¦X(U) be the set of regular diļ¬€er- ential forms on the variety U. 2.5 Dualizing sheaf The following discussion is far from being a complete presentation about sheaves. We will not enter in details and we will just give a sketchy presenta- tion of them. A sheaf of rings F on a topological space X is a functor from the category of open subsets of X, where the morphisms are inclusions, to the category of rings where the objects are rings and the morphisms are ring homomorphism, satisfying the standard sheaf axioms. In particular, for an open subset U we have F(U) is a ring and if U, V are both subsets of X such that U ā†’ V , then the induced morphism F(U) ā†’ F(V ) is a ring homomorphism. A ringed space is a pair (X, OX) where X is a topological space and OX is a sheaf of unital rings. The sheaf OX is called the structure sheaf of the ringed space (X, OX). Deļ¬nition 2.21. Let (X, OX) be a ringed space. Let F be a sheaf of OXāˆ’ modules. We say F is locally free if for every point x āˆˆ X there exists a set I and an open neighbourhood x āˆˆ U āŠ‚ X such that F|U is isomorphic to i āˆˆ OX|U as an OX|U āˆ’module. ā„¦X(U) is a module over OX(U). In fact, ā„¦X(U) is a sheaf of OX-modules. On An, ā„¦X is the free OXāˆ’module generated by dx1, . . . , dxn. Theorem 2.22. If X is smooth then the sheaf ā„¦(X) is a locally free OXāˆ’module of rank d = dim X. 23
  • 29. Chapter 2. Some Algebraic Geometry tools We will not prove this fact but we have that the set of rational diļ¬€erential forms forms a vector space over k(X). We recall the deļ¬nition of canonical bundle that we have seen in Chapter 1 in order to analyse it in an algebraic point of view. For each p āˆˆ N, look at the sheaf Ī›pā„¦(X) of pāˆ’diļ¬€erentiable forms on X, which assigns to open U āŠ† X the set of all regular pāˆ’forms, āˆ€x āˆˆ U Ļ•(x): Ī›p TxX ā†’ K Locally these look like fidgi1 āˆ§ Ā· Ā· Ā· āˆ§ dgip Rational pāˆ’forms are deļ¬ned analogously. Observation 2.23. The set of rational pāˆ’forms on X is a k(X)āˆ’vector space of dimension n p . Deļ¬nition 2.24. Let X be a smooth nāˆ’dimensional, the canonical sheaf (or dualizing sheaf ) of X is Ļ‰X = Ī›n ā„¦X. Observation 2.25. The canonical sheaf satisfy the following properties: ā€¢ Ļ‰X is locally free of rank 1. ā€¢ The set of rational canonical nāˆ’forms is a vector space of dimension 1 over k(X). Thus we have establish a connection between holomorphic sections and divisors however we will better develop it in Chapter 4. 24
  • 30. Chapter 3 A look to some complex constructions In this chapter we will equip smooth manifolds with a smooth linear complex structure on each tangent space. The existence of this structure is a necessary, but not suļ¬ƒcient, condition for a manifold to be a complex manifold. That is, every complex manifold is an almost complex manifold, but not vice versa. Almost complex structures have important applications in symplectic geometry. Therefore we will also have a look to some aspect of symplectic geometry in order to connect them to the complex world. 3.1 Almost complex structure Deļ¬nition 3.1 (Almost complex structure). An almost complex structure on a diļ¬€erentiable manifold M is a diļ¬€erentiable endomorphism of the tangent bundle J : T(M) ā†’ T(M) such that J2 x = āˆ’id, (3.1) where Jx : Tx(M) ā†’ Tx(M). A diļ¬€erentiable manifold with some ļ¬xed almost complex structure is called an almost complex manifold. Almost complex manifolds must be even dimensional. In fact the following preposition applies. Proposition 3.2. If M admits an almost complex structure, it must be even-dimensional. 25
  • 31. Chapter 3. A look to some complex constructions Proof. This can be seen as follows. Suppose M is nāˆ’dimensional, and let J : T(M) ā†’ T(M) be an almost complex structure. Since the determinant det: GL(n, R) ā†’ Rāˆ— A ā†’ det(A) is a group homomorphism, we have that det(J2 x) = det(Jx)2 . Using (3.1) we obtain det(Jx)2 = det(āˆ’id) = (āˆ’1)n . But if M is a real manifold, then det(J) is a real number, thus n must be even if M has an almost complex structure. For instance every a complex manifold has a natural structure of almost complex manifold, while the vice versa is not always true. Theorem 3.3. Every complex manifold has a canonical almost complex structure. Proof. Let M be a complex manifold of dimension n, and p āˆˆ M. Let (U, (z1, . . . , zn)) be a holomorphic chart around p. In this case , if we set xk := Rezk and yk := Imzk, then (U, (x1, . . . , xn, y1, . . . , yn)) is a local chart of M, seen as a smooth manifold. Firstly set forall q āˆˆ U, Jq āˆ‚ āˆ‚xi q = āˆ‚ āˆ‚yi q , Jq āˆ‚ āˆ‚yi q = āˆ’ āˆ‚ āˆ‚xi q . In order to conclude we will now show that J is globally well-deļ¬ned, in other words we will prove that J does not depend on the choice of coordinates. Let (U, (zk)k) and (V, (wk)k) holomorphic charts around p; set zk := xk + iyk, wk = uk + ivk for all k āˆˆ {1, . . . , n} and denote by J and J the almost complex struc- ture on (U, (zk)k) and (V, (wk)k) respectively. If we think that xk = xk(u1, . . . , un, v1, . . . , vn) and yk = yk(u1, . . . , un, v1, . . . , vn), in U āˆ© V we have, ļ£± ļ£² ļ£³ āˆ‚ āˆ‚xk = n j=1 āˆ‚uj āˆ‚xk āˆ‚ āˆ‚uj + āˆ‚vj āˆ‚xk āˆ‚ āˆ‚vj āˆ‚ āˆ‚yk = n j=1 āˆ‚uj āˆ‚yk āˆ‚ āˆ‚uj + āˆ‚vj āˆ‚yk āˆ‚ āˆ‚vj . (3.2) 26
  • 32. Chapter 3. A look to some complex constructions The Cauchy-Riemann conditions are, ļ£± ļ£² ļ£³ āˆ‚uj āˆ‚xk = āˆ‚vj āˆ‚yk āˆ‚uj āˆ‚yk = āˆ’ āˆ‚vj āˆ‚xk (3.3) Therefore, J āˆ‚ āˆ‚xk = J ļ£« ļ£­ n j=1 āˆ‚uj āˆ‚xk āˆ‚ āˆ‚uj + āˆ‚vj āˆ‚xk āˆ‚ āˆ‚vj ļ£¶ ļ£ø (3.4) = n j=1 āˆ‚uj āˆ‚yk āˆ‚ āˆ‚uj + āˆ‚vj āˆ‚yk āˆ‚ āˆ‚vj (3.5) = āˆ‚ āˆ‚yk (3.6) = J āˆ‚ āˆ‚xk in U āˆ© V. (3.7) Similarly, J āˆ‚ āˆ‚yk = J āˆ‚ āˆ‚yk in U āˆ© V. (3.8) Example 3.4. In order to make it clear here some simple examples of almost complex manifolds. ā€¢ Let (x, y) be the standard coordinates on R2. It is east to show that J : R2 ā†’ R2, deļ¬ned by J āˆ‚ āˆ‚x = āˆ‚ āˆ‚y , J āˆ‚ āˆ‚y = āˆ’ āˆ‚ āˆ‚x , satisļ¬es J2 = id. Therefore R2 admits an almost complex structure J. Identifying R2 with C in the usual way, z = x + iy, we can see J as a multiplication by i, i.e., a rotation of Ļ€ 2 in the plane. ā€¢ More in general, R2n admits an almost complex structure, for every inte- ger n ā‰„ 0. In fact if we consider global coordinate (x1 . . . xn, y1, . . . , yn), as we saw before, J āˆ‚ āˆ‚xi = āˆ‚ āˆ‚yi , J āˆ‚ āˆ‚yi = āˆ’ āˆ‚ āˆ‚xi , is an almost complex structure. 27
  • 33. Chapter 3. A look to some complex constructions 3.2 Symplectic manifolds 3.2.1 (Real) Symplectic manifolds Deļ¬nition 3.5. Let M be a smooth 2nāˆ’dimensional manifold, a 2-form Ļ‰ āˆˆ Ī›2(M) is said symplectic form if it is closed and non degenerate, i.e. if satisļ¬es the two condition 1. dĻ‰ = 0; 2. Ļ‰p = 0 for every p āˆˆ M. If M is a smooth manifold and Ļ‰ is a symplectic form on M, then the pair (M, Ļ‰) is called symplectic manifold. In other words a symplectic form is a section of the bundle of the 2-forms on M such that: 1. dĻ‰ = 0 (analytic condition); 2. (TpM, Ļ‰P ) is a symplectic vector space for all p āˆˆ M (algebraic condi- tion). Example 3.6. Let x1, . . . , x2n be local coordinates around a point p āˆˆ R2n and endow R2n with the 2-form n i=1 dxi āˆ§ dxn+i . R2n, Ļ‰ is a symplectic manifold and the matrix of Ļ‰ in the base āˆ‚ āˆ‚x1 p , . . . , āˆ‚ āˆ‚x2n p of TpM is 0 In āˆ’In 0 . Example 3.7 (Cotangent bundles). Let N be a smooth manifold of dimension n and let M = Tāˆ—N be its cotangent bundle. This has a natural symplectic structure, which may be deļ¬ned locally as follows. Choose local coordinates x1, . . . , xn on N. Then the 1āˆ’forms dx1, . . . , dxn provide a local trivialisation of Tāˆ—N, so we obtain local coordinate functions Ī¾1, . . . , Ī¾n on the ļ¬bres of Tāˆ—N. Thus M has local coordinates (x1, . . . , xn, Ī¾1, . . . , Ī¾n.) We may deļ¬ne a 1-form Īø = n i=1 Ī¾idxi 28
  • 34. Chapter 3. A look to some complex constructions locally on M and it turns out that this local deļ¬nition in fact deļ¬nes a global one-form (the so-called ā€œLiouville formā€) on M. The exterior derivative Ļ‰ = dĪø is a natural symplectic form on M. Clearly it is closed (since it is exact), and it is nondegenerate because in local coordinates has the following expression Ļ‰ = n i=1 dĪ¾i āˆ§ dxi . Example 3.8. The 2-sphere S2 endowed with the symplectic form (the volume form) Ļ‰ = sin ĪødĪø āˆ§ dĻ†. In general it is not true that the 2nāˆ’sphere S2n, with n > 1 is a symplectic manifold, for the proof of this fact see Corollary 5.11. 3.2.2 (Complex) Symplectic manifolds A complex symplectic manifold is a pair (M, Ļ‰) consisting of a complex manifold M and a holomorphic 2-form Ļ‰ (of type (2, 0)) such that: 1. Ļ‰ is closed, i.e., dĻ‰ = 0; 2. Ļ‰ is non degenerate, i.e., the associated linear map TpM ā†’ Tāˆ— p M v ā†’ (Ļ‰)p(v, āˆ’) from the holomorphic tangent space to the holomorphic cotangent space, is an isomorphism at each point p āˆˆ M. These are sometimes also referred to as holomorphic symplectic manifolds. 3.3 Compatible almost complex structures Deļ¬nition 3.9. Let (M, Ļ‰) be a symplectic manifold and p a point of M. An almost complex structure J on M is called compatible with Ļ‰ (or Ļ‰āˆ’ compatible) if ā€¢ Ļ‰(Ju, Jv) = Ļ‰(u, v) āˆ€u, v āˆˆ TpM; ā€¢ Ļ‰(Ju, u) > 0, āˆ€u āˆˆ TpM such that u = 0. 29
  • 35. Chapter 3. A look to some complex constructions Remark. g: TpM Ɨ TpM ā†’ R deļ¬ned by g(u, v) = Ļ‰(Ju, v), u, v āˆˆ TpM, is a Riemannian metric. This bilinear form is simmetric since g(w, v) = Ļ‰(w, Jv) = Ļ‰(Jw, J2 v) = Ļ‰(Jw, āˆ’v) = Ļ‰(v, Jw) = g(v, w), where v, w āˆˆ TpM. By deļ¬nition of Ļ‰ it is positive deļ¬nite, i.e. given u āˆˆ TpM, we have g(v, v) = Ļ‰(Ju, u) > 0. Remark. The triple (Ļ‰, J, g) is said a compatible triple. Moreover any one of J, Ļ‰, g can be written in terms of the other two by the following formulas. ā€¢ g(u, v) = Ļ‰(u, Jv); ā€¢ Ļ‰(u, v) = g(Ju, v); ā€¢ J(u) = Ėœgāˆ’1(ĖœĻ‰(u)), where ĖœĻ‰: TM ā†’ Tāˆ— M, Ėœg: TM ā†’ Tāˆ— M u ā†’ Ļ‰(u, Ā·) u ā†’ g(u, Ā·). Theorem 3.10. On any symplectic manifold (M, Ļ‰), there exists almost complex structures J that are compatible with Ļ‰. Proof. First we show this is true for a symplectic vector space V. Let g be a Riemannian metric on V and deļ¬ne A by Ļ‰(u, v) = g(Au, v). Since Ļ‰ is skew-symmetric and g is a metric, thus it is symmetric, we have Ļ‰(u, v) = Ļ‰(āˆ’v, u) = g(āˆ’Av, u) = g(u, āˆ’Av). (3.9) By deļ¬nition of adjoint matrix and by eqution (3.9) we have g(u, Aāˆ— v) = g(Au, v) = g(u, āˆ’Av), then Aāˆ— = āˆ’A. Furthermore AAāˆ— is ā€¢ symmetric, i.e. (AAāˆ—)āˆ— = AAāˆ—; ā€¢ positive deļ¬nite, i.e. g(AAāˆ—v, v) = g(Aāˆ—v, Aāˆ—v) > 0, āˆ€v = 0. 30
  • 36. Chapter 3. A look to some complex constructions Since every symmetric matrix B can be factored into B = Qāˆ†Qāˆ’1 where Q is orthogonal matrix, i.e. Qāˆ’1 = QT and āˆ† = diag(Ī»1, . . . , Ī»2n), AAāˆ— can be written as AAāˆ— = Qāˆ†Qāˆ’1 , moreover Ī»i > 0, for every i = 1, . . . , 2n because AAāˆ— is positive deļ¬ned. Set J := ( āˆš AAāˆ—)āˆ’1 A. (3.10) The factorization A = āˆš AAāˆ— is called the polar decomposition of A. Using the diagonalization we can write J = Q āˆš āˆ†Qāˆ’1A, this implies that JJāˆ— = id and Jāˆ— = āˆ’J, in fact JJāˆ— = Q( āˆš āˆ†)āˆ’1 Qāˆ’1 A Ā· Aāˆ— (Qāˆ’1 )āˆ— (( āˆš āˆ†)āˆ’1 )āˆ— Qāˆ— , since (( āˆš āˆ†)āˆ’1)āˆ— = ( āˆš āˆ†)āˆ’1 and Q is orthogonal, = Q( āˆš āˆ†)āˆ’1 Qāˆ’1 Ā· Qāˆ†Qāˆ’1 Ā· Q( āˆš āˆ†)āˆ’1 Qāˆ’1 = Q( āˆš āˆ†)āˆ’1 āˆ†( āˆš āˆ†)āˆ’1 Qāˆ’1 = id, where the last equality applies since the diagonal matrices commute. Thus J2 = āˆ’id is an almost complex structure and ā€¢ Ļ‰(Ju, Jv) = g(AJu, Jv) = g(JAu, Jv) = g(Au, v) = Ļ‰(u, v) where the second equality applies since A commutes with āˆš AAāˆ— and thus J commutes with āˆš AAāˆ—; ā€¢ Ļ‰(u, Ju) = g(āˆ’JAu, u) = g( āˆš AAāˆ—u, u) > 0, so J is compatible. Finally we can use this construction to each tangent space TpM of every point of M. It turns out that J is globally well-deļ¬ned since the polar decomposition is canonical after a choice of Riemannian metric, i.e. it does not depend on the choice of Q nor of the ordering of the eigenvalues {Ī»1, . . . , Ī»2n}. Hence J is smooth. 31
  • 37. Chapter 4 Calaby-Yau manifolds Currently, research on Calabi-Yau manifolds is a central focus in both mathematics and mathematical physics. The spread of this study is partially propelled by the prominent role of the Calabi-Yau in superstring theories. Many beautiful properties of Calabi-Yau manifolds have been discovered and nowadays this subject represents an extremely active research ļ¬eld. In this chapter we will give the deļ¬nition of Calabi-Yau manifolds and we will provide many examples, using the tools we have previously studied. Moreover we will try to place this type of manifold in the complex geometry world and we will investigate how they interact with others complex manifolds. 4.1 Symplectic Calabi-Yau manifolds Following [JD08] we ļ¬rstly analyse Calabi-Yau manifolds from the symplectic point of view. Deļ¬nition 4.1. Let M be a symplectic real manifold of dimension 2n. M is a symplectic Calabi-Yau if its canonical bundle K ā†’ M is trivial. There are many examples of compact symplectic Calabi-Yau manifolds. Firstly, many nilmanifolds are of this type. Example 4.2 (Nilmanifolds). A Nilmanifold is a compact quotient space of a nilpotent Lie group modulo a closed discrete subgroup, or (equivalently) a homogeneous space with a nilpotent Lie group acting transitively on it. The most famous example of a nilmanifold is the so-called Kodaira Thurston surface, which will be discussed in Chapter 5. 32
  • 38. Chapter 4. Calaby-Yau manifolds 4.2 Complex Calabi-Yau manifolds We give now the deļ¬nition of a Calabi-Yau manifold from the complex point of view. Deļ¬nition 4.3. A complex Calabi-Yau is a complex compact manifold such that its canonical bundle K ā†’ M is trivial. Remark. Here the canonical bundle is a holomorphic line bundle while in the case of the symplectic Calabi-Yau manifold it does not exist a notion of holomorphic and the triviality is a topological condition. In the following sections will now give some examples of complex Calabi- Yau manifolds. 4.2.1 1-dimensional Calabi-Yau In one complex dimension, the only compact examples are tori, see Example 1.10. In order to expand the horizon of the 1-dimensional Calabi-Yau manifolds we will show now how complex tori C/Ī› can also be viewed as cubic curves. These cubic curves are called elliptic despite not being ellipses, due to a connection between them and the arc length of an actual ellipse. Elliptic curves Deļ¬nition 4.4 (Elliptic curves over a ļ¬eld K). An elliptic curve is an irreducible cubic in P2(K) which is non singular. By an appropriate change of variables, a general elliptic curve over a ļ¬eld K with ļ¬eld characteristic diļ¬€erent from 2, 3, can be written in the aļ¬ƒne form y2 = 4x3 āˆ’ ax + b, with a3 āˆ’ 27b2 = 0 (condition of non singularity). Let K = C we will now see how the points of an elliptic curve form a torus. The meromorphic functions on a complex torus relate the manifold to a cubic curve. Given a lattice Ī›, the meromorphic functions f : C/Ī› ā†’ C on the torus, where C = C āˆŖ {āˆž}, are naturally identiļ¬ed with the Ī›-periodic meromorphic functions f : C ā†’ C on the plane. Let Ī› be a lattice, a function f such that f(z + Ļ‰) = f(z), for any z āˆˆ C and Ļ‰ āˆˆ Ī›, is called elliptic function. It follows that if Ī› = Ļ‰1Z āŠ• Ļ‰2Z, f(z + Ļ‰1) = f(z + Ļ‰2) = f(z), 33
  • 39. Chapter 4. Calaby-Yau manifolds and we call Ļ‰1, Ļ‰2 the periods of f. A non-trival example of an elliptic function is the Weierstrass ā„˜-function. Given a lattice Ī›, we deļ¬ne the Weierstrass ā„˜-function by ā„˜(z) = 1 z2 + Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)2 āˆ’ 1 Ļ‰2 , (4.1) where z āˆˆ C, z /āˆˆ Ī›. Proposition 4.5. The ā„˜(z) function satisļ¬es the following properties: 1. ā„˜(z) is well deļ¬ned in the sense that the sum converges absolutely and uniformly on compact sets ā„¦ such that ā„¦ āˆ© Ī› = āˆ…. 2. ā„˜(āˆ’z) = ā„˜(z) for all z āˆˆ C. 3. The derivative ā„˜ (z) = āˆ’2 Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)3 (4.2) has period Ī›. 4. ā„˜(z) has period Ī›. Proof. We will just sketch the proofs. For 1 it is convenient to state the following lemma. Lemma 4.6. If k > 2 then the following series Ļ‰āˆˆĪ›{0} 1 |Ļ‰|k (4.3) converges over the entire lattice Ī›. The convergence of the series is proven using an integral comparison test and estimates on the diagonal of the fundamental parallelogram for Ī›, Ī  = {x1Ļ‰1 + x2Ļ‰2 : x1, x2 āˆˆ [0, 1]} . Then the absolute and uniform convergence is a consequence of another estimate and the exclusion of ļ¬nitely many terms. (2) If Ļ‰ āˆˆ Ī› then it is also true that āˆ’Ļ‰ āˆˆ Ī›, by multiplying by āˆ’1. Hence ā„˜(āˆ’z) = 1 (āˆ’z)2 + Ļ‰āˆˆĪ›{0} 1 (āˆ’z āˆ’ Ļ‰)2 āˆ’ 1 Ļ‰2 = 1 (z)2 + āˆ’Ļ‰āˆˆĪ›{0} 1 (āˆ’z + Ļ‰)2 āˆ’ 1 (āˆ’Ļ‰)2 = 1 (z)2 + āˆ’Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)2 āˆ’ 1 Ļ‰2 = ā„˜(z). 34
  • 40. Chapter 4. Calaby-Yau manifolds (3) The sum ā„˜ (z) converges uniformly and absolutely by the lemma with simple comparison to 1 |w|3 . Moreover if Ļ‰, Ļ āˆˆ Ī› then it is also true that Ļ‰ āˆ’ Ļ āˆˆ Ī›; thus ā„˜ (z + Ļ) = āˆ’2 Ļ‰āˆˆĪ›{0} 1 (z + Ļ āˆ’ Ļ‰)3 = āˆ’2 Ļ‰āˆ’ĻāˆˆĪ›{0} 1 (z āˆ’ (Ļ‰ āˆ’ Ļ))3 = ā„˜ (z). (4) Using 3 we ļ¬nd out that the derivative of ā„˜(z + Ļ‰) āˆ’ ā„˜(z) is equal to zero, if z /āˆˆ Ī›. Thus there exists a constant cĻ‰ such that ā„˜(z + Ļ‰) āˆ’ ā„˜(z) = cĻ‰, for all z /āˆˆ Ī›. Setting z = āˆ’Ļ‰ 2 we obtain cĻ‰ = ā„˜ āˆ’ Ļ‰ 2 āˆ’ ā„˜ āˆ’ Ļ‰ 2 . By the fact that ā„˜(z) is even we can conclude that cĻ‰ = 0. Therefore ā„˜(z + w) = ā„˜(z) (4.4) for all w āˆˆ Ī›. Remark. Let Ī› = Ļ‰1Z āŠ• Ļ‰2Z. Since ā„˜ satisļ¬es the condition (4.4), we say that ā„˜ is a doubly periodic function, as it has two independent periods Ļ‰1 and Ļ‰2. We will now see some functions that will appear in the Laurent expansion of the Weierstrass ā„˜(z)-function for Ī›. Deļ¬nition 4.7. Let Ī› be a lattice and k be an integer, the Eisenstein series are functions deļ¬ned by Gk(Ī›) = Ļ‰āˆˆĪ›{0} 1 Ļ‰k , (4.5) for k > 2, even. Remark. The Eisenstein series satisfy the homogeneity condition Gk(mĪ›) = māˆ’k Gk(Ī›) for all m āˆˆ C{0}. 35
  • 41. Chapter 4. Calaby-Yau manifolds Theorem 4.8. Let ā„˜ be the Weierstrass function with respect to a lattice Ī›. Then (i) The Laurent expansion of ā„˜ is ā„˜(z) = 1 z2 + āˆž n=1 (2n + 1)G2n+2(Ī›)z2n , (4.6) for all z such that 0 < |z| < inf{|Ļ‰| : Ļ‰ āˆˆ Ī›{0}}. (ii) The functions ā„˜ and ā„˜ satisfy the relation (ā„˜ (z))2 = 4(ā„˜(z))3 āˆ’ g2(Ī›)ā„˜(z) āˆ’ g3(Ī›) (4.7) where g2(Ī›) = 60G4(Ī›) and g3(Ī›) = 140G6(Ī›). (iii) Let Ī› = Ļ‰1Z āŠ• Ļ‰2Z and let Ļ‰3 = Ļ‰1 + Ļ‰2. Then the cubic equation satisļ¬ed by ā„˜ and ā„˜ , y2 = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›) is y2 = 4(x āˆ’ e1)(x āˆ’ e2)(x āˆ’ e3), (4.8) where ei = ā„˜ Ļ‰i 2 for i = 1, 2, 3. This equation is non singular, meaning its right side has distinct roots. Proof. (i) Firstly we observe how the geometric series squares i.e., āˆž n=0 qn 2 = āˆž n=0 qn āˆž k=0 qk = āˆž k=0 qk + q āˆž k=0 qk + Ā· Ā· Ā· + qn āˆž k=0 qk + . . . = 1 + q + q2 + . . . + q + q2 + q3 + . . . + Ā· Ā· Ā· + qn + qn+1 + qn+2 + . . . + . . . = 1 + 2q + 3q2 + 4q3 + Ā· Ā· Ā· + (n + 1)qn + . . . = āˆž n=0 (n + 1)qn , where q āˆˆ C. As a result, if 0 < |z| < inf{|Ļ‰| : Ļ‰ āˆˆ Ī›{0}}, we obtain that ā„˜(z) = 1 z2 + Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)2 āˆ’ 1 Ļ‰2 = 1 z2 + Ļ‰āˆˆĪ›{0} 1 Ļ‰2 1 1 āˆ’ z Ļ‰ 2 āˆ’ 1 = 1 z2 + Ļ‰āˆˆĪ›{0} 1 Ļ‰2 ļ£« ļ£­ āˆž n=0 z Ļ‰ n 2 āˆ’ 1 ļ£¶ ļ£ø 36
  • 42. Chapter 4. Calaby-Yau manifolds and by using the result of the geometric square series we get = 1 z2 + Ļ‰āˆˆĪ›{0} 1 Ļ‰2 āˆž n=0 (n + 1) z Ļ‰ n āˆ’ 1 = 1 z2 + Ļ‰āˆˆĪ›{0} 1 Ļ‰2 1 + āˆž n=1 (n + 1) z Ļ‰ n āˆ’ 1 = 1 z2 + Ļ‰āˆˆĪ›{0} 1 Ļ‰2 āˆž n=1 (n + 1) z Ļ‰ n = 1 z2 + Ļ‰āˆˆĪ›{0} āˆž n=1 (n + 1) zn Ļ‰2+n . Convergence results allow the resulting double sum to be rearranged, and then the inner sum cancels when n is odd. In fact the last expression is equal to 1 z2 + āˆž n=1 (n + 1) Ļ‰2 1 zn Ļ‰n 1 + āˆž n=1 (n + 1) (āˆ’Ļ‰1)2 zn (āˆ’Ļ‰1)n + . . . + āˆž n=1 (n + 1) Ļ‰2 i zn Ļ‰n i + āˆž n=1 (n + 1) (āˆ’Ļ‰i)2 zn (āˆ’Ļ‰i)n + . . . with Ļ‰i āˆˆ Ī›{0}. Therefore we can conclude that ā„˜(z) = 1 z2 + āˆž n=1 Ļ‰āˆˆĪ›{0} (2n + 1) Ļ‰2n+2 z2n and by defenition of Einstein series = 1 z2 + āˆž n=1 (2n + 1)G2n+2(Ī›)z2n . (ii) The series expansions of ā„˜ and ā„˜ from the previous proposition leads to ā„˜(z) = 1 z2 + 3G4z2 + 5G6z4 + O(z5 ) ā„˜ (z) = āˆ’ 2 z3 + 6G4z + 20G6z3 + O(z4 ) and taking the cube and the square we have ā„˜(z)3 = 1 z6 + 9G4 z2 + 15G6 + O(z2 ) ā„˜ (z)2 = 4 z6 āˆ’ 24G4 z2 āˆ’ 80G6 + O(z2 ). 37
  • 43. Chapter 4. Calaby-Yau manifolds Deļ¬ne a function f as follows, f(z) = (ā„˜ (z))2 āˆ’ 4(ā„˜(z))3 + g2(Ī›)ā„˜(z) + g3(Ī›). Since f is an elliptic function, as a polynomial of ā„˜ and ā„˜ , with no poles, for the First Liouville Theorem1 it is constant. If we substitute the value we found above, we get f(z) = 4 z6 āˆ’ 24G4 z2 āˆ’ 80G6 āˆ’ 4 1 z6 + 9G4 z2 + 15G6 + g2(Ī›) 1 z2 + 3G4z2 + 5G6z4 + g3(Ī›) + O(z2 ) = 4 z6 āˆ’ 24G4 z2 āˆ’ 80G6 āˆ’ 4 1 z6 + 9G4 z2 + 15G6 + 60G4(Ī›)( 1 z2 + + 3G4z2 + 5G6z4 ) + 140G6(Ī›) + O(z2 ) = 180 G2 4z2 + 300 G4G6z4 + O(z2 )). Noticing that the last right-hand side grows at the order of z2, and considering that lim zā†’0 = O(z2), we conclude that f(z) ā‰” 0. (iii) We start observing that ā„˜(z) and ā„˜ (z) have a double pole at each Ļ‰ āˆˆ Ī›. Moreover since ā„˜ (z) is doubly periodic, i.e. ā„˜ (z + Ļ‰) = ā„˜ (z), letting Ī› = Ļ‰1Z āŠ• Ļ‰2Z, we see that if z = āˆ’Ļ‰i 2 , and Ļ‰ = Ļ‰i 2 , ā„˜ ( Ļ‰i 2 ) = ā„˜ (āˆ’ Ļ‰i 2 ) (4.9) for i = 1, 2, 3. Secondly, since ā„˜ is odd, ā„˜ (āˆ’ Ļ‰i 2 ) = āˆ’ā„˜ ( Ļ‰i 2 ). (4.10) Equation (4.9) together with (4.10) allow to conclude that zi = Ļ‰i 2 are points of order 2 with ā„˜ (zi) = 0, for i = 1, 2, 3. The relation between ā„˜(z) and ā„˜ (z) from (ii) shows that the corresponding values xi = ā„˜(zi) for i = 1, 2, 3 are roots of the cubic polynomial pĪ›(x) = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›), so it factors as claimed. Each xi is a double value of ā„˜ since, as we have seen, ā„˜ (zi) = 0, furthermore ā„˜ has degree 2, meaning it takes each value twice counting multiplicity, this makes the three xi distinct. That is, the cubic polynomial pĪ› has distinct roots. 1 Theorem 4.9 (First Liouville Theorem). If an elliptic function has no poles, then it is constant. 38
  • 44. Chapter 4. Calaby-Yau manifolds The isomorphism between complex tori and complex elliptic curves Part (ii) of Theorem 4.8 shows that the map C/Ī› ā†’ C2 z ā†’ ā„˜(z), ā„˜ (z) takes nonlattice points of C to points (x, y) āˆˆ C2 satisfying the nonsingular cubic equation of part (iii), y2 = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›). Moreover this application extends to all z āˆˆ C by mapping lattice points to a suitably deļ¬ned point at inļ¬nity, but we will see the details in the following theorem. Theorem 4.10. Let Ī› be a lattice on C and E be the elliptic curve E = {(x, y) āˆˆ C2 | y2 = 4x3 āˆ’ g2(Ī›)x āˆ’ g3(Ī›)}. Then the map Ļ† deļ¬ned by C/Ī› ā†’ E z ā†’ ā„˜(z), ā„˜ (z) if z = 0 0 ā†’ {āˆž}, is a group isomorphism. Before going into the proof of the theorem we need to state the following result. Theorem 4.11. Let f be an elliptic function for the lattice Ī› and let Ī  be a fundamental parallelogram for Ī›. If f is not constant then f : C ā†’ C āˆŖ {āˆž} is surjective. If n is the sum of the orders of the poles2 of f in Ī  and z0 āˆˆ C, then f(z) = z0 has n solutions (counting multiplicities). We can now see the proof of Theorem 4.10, which allows us to understand that complex tori (Riemann surfaces, complex analytic objects) are equivalent to elliptic curves (solution sets of cubic polynomials, algebraic objects). Proof. We need to show that Ļ† is: 1. injective. Suppose ā„˜(z1), ā„˜ (z1) = ā„˜(z2), ā„˜ (z2) , (4.11) where z1, z2 āˆˆ C/Ī› are such that z1 = z2 mod Ī›. The only poles for ā„˜(z) belong to Ī›, thus we can distinguish to cases. 2 If f is an elliptic function, we can write it as a Laurent series expansion around Ļ‰ āˆˆ Ī› as f(z) = ar(z āˆ’ Ļ‰)r + ar+1(z āˆ’ Ļ‰)r + 1 + . . . with ar = 0. We deļ¬ne the order of f at Ļ‰ as ordĻ‰f = r. 39
  • 45. Chapter 4. Calaby-Yau manifolds ā€¢ If z1 is a pole of ā„˜ then z1 āˆˆ Ī›. Since (4.11) applies, z2 is a pole of ā„˜, i.e., z2 āˆˆ Ī›. This implies z1 = z2 mod Ī›. ā€¢ Now suppose z1 is not a pole of ā„˜, i.e., z1 /āˆˆ Ī›. Then h(z) = ā„˜(z) āˆ’ ā„˜(z1) has a double pole at z = 0 and no other poles in Ī  = {x1Ļ‰1 + x2Ļ‰2 : x1, x2 āˆˆ [0, 1]} . By Theorem 4.11, h(z) has exactly two zeros (counting multiplicities). ā€“ Suppose z1 = Ļ‰i 2 for some i. From the proof of (iii) of Theorem 4.8 we know that ā„˜ (Ļ‰i 2 ) = 0, so z1 is a double root of h(z) hence the only root. Thus z2 = z1. ā€“ Suppose z1 is not of the form Ļ‰i 2 . Since h(āˆ’z1) = h(z1) = 0 (by evenness of ā„˜) and since z1 = z2 mod Ī›, two zeros of h are z1 and z2 = āˆ’z1 mod Ī›. But y = ā„˜ (z2) = ā„˜ (āˆ’z1) = āˆ’ā„˜ (z1) = āˆ’y. Hence ā„˜ (z1) = y = 0. But ā„˜ (z) has only a triple pole, thus has only three zeros in Ī . But from the proof of (iii) of Theorem 4.8, we know that these zeros occur at Ļ‰i 2 , hence a contradiction since z = Ļ‰i 2 . Thus z1 = z2 mod Ī› and Ļ† is injective. 2. surjective. Let (x, y) āˆˆ E; we need to prove that there exists z āˆˆ C such that Ļ†(z) = (x, y), i.e., that x = ā„˜(z) and y = ā„˜ (z). Since ā„˜(z) āˆ’ x has a double pole, Theorem 4.11 implies it has zeros, hence there exists z āˆˆ C such that ā„˜(z) = x. The elliptic equation in the (ii) part of Theorem 4.8 implies that ā„˜ (z)2 = y2, so ā„˜ (z) = Ā±y. ā€¢ If ā„˜ (z) = y we are done. ā€¢ If ā„˜ (z) = āˆ’y, then by the evenness of the ā„˜ function, ā„˜ (āˆ’z) = y and ā„˜(āˆ’z) = x, so āˆ’z ā†’ (x, y). Hence Ļ† is onto. 3. a group homomorphism. We need to show that Ļ†(z1 + z2) = Ļ†(z1) + Ļ†(z2), where z1, z2 āˆˆ C. The map Ļ† transfers the group law from the complex torus to the elliptic curve. The proof is not trivial and it is well devel- oped in [Kna92], thus we will not focus on the demonstration and we will try to understand the group law on the curve. 40
  • 46. Chapter 4. Calaby-Yau manifolds Let z1 + Ī› and z2 + Ī› be nonzero points of the torus. The image points (ā„˜(z1), ā„˜ (z1)) and (ā„˜(z2), ā„˜ (z2)) on the curve determine a secant or tangent line of the curve in C2, ax + by + c = 0. Consider the function f(z) = aā„˜(z) + bā„˜ (z) + c. This is meromorphic on C/Ī›. When b = 0 it becomes f(z) = a 1 z2 + Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)2 āˆ’ 1 Ļ‰2 + b āˆ’ 2 Ļ‰āˆˆĪ›{0} 1 (z āˆ’ Ļ‰)3 + c. (4.12) As we can see it has a triple pole at 0 + Ī› and zeros at z1 + Ī› and z2 + Ī›. One could also prove that its third zero is at the point z3 + Ī› such that z1 + z2 + z3 + Ī› = 0 + Ī› in C/Ī›. When b = 0, f has a double pole at 0 + Ī› and zeros at z1 + Ī› and z2 + Ī›, furthermore z1 + z2 + Ī› = 0 + Ī› in C/Ī›. In this case let z3 = 0+Ī› so that again z1 +z2 +z3 +Ī› = 0+Ī›, and since the line is vertical view it as containing the inļ¬nite point (ā„˜(0), ā„˜ (0)) whose second coordinate arises from a pole of higher order than the ļ¬rst. Therefore for any value of b the elliptic curve points on the line ax+by+c = 0 are the points (xi, yi) = ā„˜(zi), ā„˜ (zi)) for i = 1, 2, 3. Since z1 + z2 + z3 + Ī› = 0 + Ī› on the torus in all cases, the resulting group law on the curve is: ā€¢ The identity element of the curve is the inļ¬nite point; ā€¢ collinear triples on the curve sum to zero. Our aim now is to ļ¬nd an addition law and a duplication law for the function ā„˜. Consider the aļ¬ƒne equation of the elliptic curve and the aļ¬ƒne equation of a line: E : y2 = 4x3 āˆ’ g2x āˆ’ g3, L: y = mx + b. If a point (x, y) āˆˆ C2 lies on E āˆŖ L then its x-coordinate satisļ¬es the cubic polynomial obtained by substituting mx + b for y in the equation of E, 4x3 āˆ’ m2 x2 + āˆ’b2 āˆ’ 2mbx āˆ’ g2x āˆ’ g3 = 0. Thus, given three points collinear points (x1, y1), (x2, y2), and (x3, y3) on the curve, necessarily x1 + x2 + x3 = m2 4 , 41
  • 47. Chapter 4. Calaby-Yau manifolds where m is the slope of their line, m = ļ£± ļ£² ļ£³ y1āˆ’y2 x1āˆ’x2 x1 = x2 12x2 1āˆ’g2 2y1 x1 = x2 (4.13) A slight restatement is that x3 = m2 4 āˆ’ x1 āˆ’ x2, m as above. (And also y3 = m(x3 āˆ’ x1) + y1.) These results translate back to the desired addition law and duplication law for the Weierstrass ā„˜-function. Since the three points on the curve are collinear, we have for some z1 + Ī›, z2 + Ī› āˆˆ C/Ī›, (x1, y1) = ā„˜(z1), ā„˜ (z1) , (x2, y2) = ā„˜(z2), ā„˜ (z2) , (x3, y3) = ā„˜(āˆ’z1 āˆ’ z2), ā„˜ (āˆ’z1 āˆ’ z2) . But ā„˜ is even and ā„˜ is odd, so that in fact (x3, y3) = ā„˜(z1 + z2), āˆ’ā„˜ (z1 + z2) . That is, ā„˜(z1 + z2) = 1 4 ā„˜ (z1) āˆ’ ā„˜ (z2) ā„˜(z1) āˆ’ ā„˜(z2) 2 āˆ’ ā„˜(z1) āˆ’ ā„˜(z2) ifz1 + Ī› = Ā±z2 + Ī›, ā„˜(2z) = 1 4 12ā„˜(z)2 āˆ’ g2 2ā„˜ (z) 2 āˆ’ 2ā„˜(z) ifz /āˆˆ 1 2 Ī› + Ī›. In order to conclude this paragraph we outline that every elliptic curve over C comes from a torus. That is, given an elliptic curve E, then we can produce a lattice Ī› unique up to some homothetic equivalence. We point out that not only does every complex torus C/Ī› lead via the Weierstrass ā„˜-function to an elliptic curve y2 = 4x3 āˆ’ a2x āˆ’ a3, a3 2 āˆ’ 27a2 3 = 0 with a2 = g2(Ī›) and a3 = g3(Ī›), but the converse holds as well. Deļ¬nition 4.12 (Homothetic Lattices). Let Ī› = ZĻ‰1 + ZĻ‰2 be a lattice in C. We deļ¬ne Ļ„ := Ļ‰1 Ļ‰2 . Since Ļ‰1 and Ļ‰2 are linearly independent over R, Ļ„ cannot be real. Hence, by switching Ļ‰1 and Ļ‰2 if necessary, we can assume the imaginary part 42
  • 48. Chapter 4. Calaby-Yau manifolds Im(Ļ„) > 0, i.e., Ļ„ lies in the upper half plain H = {x + iy āˆˆ C | y > 0}. Now if we let Ī›Ļ„ = ZĻ„ + Z, then Ī› is homothetic to Ī›Ļ„ , that is Ī› = Ī»Ī›Ļ„ for some Ī» āˆˆ C. In this case Ī» = Ļ‰2. Finally we can conclude thanks to the following result. Theorem 4.13. Let y2 = 4x3 āˆ’ Ax + b deļ¬ne an elliptic curve E over C. Then there exists a lattice Ī› such that g2(Ī›) = A and g3(Ī›) = B and there is an isomorphism of groups C/Ī› = E. Observation 4.14. The existence of such a lattice is a homothetic equivalence, that is, if we ļ¬nd Ī› that works, then any Ī› = Ī»Ī› for Ī» āˆˆ C will suļ¬ƒce. There are several approaches to proving the statement but we will not go any further. The reader is referred to [Raa10] for more details. We have decided to point out the connection between complex tori and complex elliptic curves in order to give diļ¬€erent example of 1-dimensional Calabi-Yau manifolds; moreover this turned out to be a good tools to analyse the same object from the algebraic point of view. 4.2.2 2-dimensional Calabi-Yau Calabiā€“Yau manifolds of dimension two are two-dimensional complex tori and complex K3 surfaces, most of the latter are not algebraic. This means that they cannot be embedded in any projective space as a surface deļ¬ned by polynomial equations. K3 surfaces Deļ¬nition 4.15. A K3 surface is a (smooth) surface X which is simply connected and has trivial canonical bundle. We will now present a deļ¬nition of K3 surfaces which is slightly diļ¬€erent from the standard one. Before going into the matter we need to understand the following deļ¬nition. Deļ¬nition 4.16 (Rational double points). A complex surface X has rational double points if the dualizing sheaf Ļ‰X is locally free, and if there is a resolution of singularities Ļ€: X ā†’ X such that Ļ€āˆ— Ļ‰X = Ļ‰X = OX KX . This means that for every P āˆˆ X there exists a neighborhood U of P and a holomorphic 2-form Ī± = Ī±(z1, z2)dz1 āˆ§ dz2 43
  • 49. Chapter 4. Calaby-Yau manifolds deļ¬ned on U āˆ’ {P} such that Ļ€āˆ—(Ī±) extends to a nowhere-vanishing holo- morphic form on Ļ€āˆ’1(U). The structure of rational double points (sometimes called simple singularities) is well-known and each such point must be analytically isomorphic to one of the following: An(n ā‰„ 1) x2 + y2 + zn+1 = 0 Fig4.1 Dn(n ā‰„ 4) x2 + yz2 + znāˆ’1 = 0 Fig4.2 E6 x2 + y3 + z4 = 0 Fig4.3 a) E7 x2 + y3 + yz3 = 0 Fig4.3 b) E8 x2 + y3 + z5 = 0 Fig4.3 c) and the resolution X ā†’ X replaces such a point with a collection of rational curves of self-intersection -2 in the following conļ¬guration: An, Dn, E6, E7, E8. Figure 4.1: From left to right in the ļ¬gure are represented A2, A4, A6, A8. Note that every A2n+1 for 1 ā‰¤ n can not be plotted since all the points satisfying the related equation are complex. Furthermore we can observe that as n increases, the top of the surfaces becomes more pointed. 44
  • 50. Chapter 4. Calaby-Yau manifolds Figure 4.2: From left to right the image shows D4, D5, D6, D7. In the foreground stand out the even Di 4 ā‰¤ i ā‰¤ 7 while in the background stand out the odd ones. As we can see the conļ¬guration of the surfaces and the look of the singular points change letting i even or odd. 45
  • 51. Chapter 4. Calaby-Yau manifolds (a) E6. (b) E7. (c) E8. Figure 4.3 Deļ¬nition 4.17 (K3 surfaces). A K3 surface is a compact complex analytic surface X with only rational double points such that h(0,1) = dim(H1 (OX)) = 0 and Ļ‰X = OX. If X is smooth, the dualizing sheaf Ļ‰X is the line bundle associated to the canonical divisor KX, so this last condition implies that the canonical divisor is trivial since KX= OX. If X is a K3 surface and Ļ€: X ā†’ X is the minimal resolution of singularities (i. e. the one which appeared in the deļ¬nition of rational double point) then it turns out that the pull-back Ļ€āˆ— establishes an isomorphism H1 (OX) = H1 (OX ), and also we have Ļ‰X = Ļ€āˆ—OX = OX . Thus, the smooth surface X is also a K3 surface. We collect a number of construction methods for K3 surfaces. One should, however, keep in mind that most K3 surfaces, especially of high degree, do not admit explicit descriptions. Their existence is solely predicted by deformation theory. 46
  • 52. Chapter 4. Calaby-Yau manifolds Proposition 4.18. The following properties apply for a K3 surface. (i) All K3 surfaces are simply connected. (ii) Every K3 surface over C is diļ¬€eomorphic to the Fermat quartic (see Example 4.19). (iii) The Hodge diamond h0,0 h1,0 h0,1 h2,0 h1,1 h0,2 h2,1 h1,2 h2,2 = 1 0 0 1 22 1 0 0 1 is completely determined (even in positive characteristic). Example 4.19. A non-singular degree 4 surface, such as the Fermat quartic, [w, x, y, z] āˆˆ CP3 | w4 + z4 + y4 + x4 = 0 is a K3 surface. Example 4.20. The intersection of a quadric and a cubic in CP4 gives K3 surfaces. Example 4.21. The intersection of three quadrics in CP5 gives K3 surfaces. Example 4.22. A Kummer surface is a special type of quartic surface. As a projective variety, a Kummer surface may be described as the vanishing set of an ideal of polynomials. However, these surfaces may also be viewed more abstractly, in terms of Jacobian varieties but we will not enter in details. However we will devote the next section to the main feature of these surfaces. See [End03] for more details. Kummer surfaces Deļ¬nition 4.23. The Kummer surface with parameter Āµ āˆˆ R is the projec- tive variety given by KĀµ : x2 y2 + z2 āˆ’ Āµ2 w2 āˆ’ Ī»pqrs = 0 āŠ‚ RP3 , where Āµ2 = 1 3, 1, 3, Ī» = 3Āµ2 āˆ’ 1 3 āˆ’ Āµ2 and p, q, r, s are the ā€œtetrahedral coordinates,ā€ given by p = w āˆ’ z āˆ’ āˆš 2x, q = w āˆ’ z + āˆš 2x, r = w + z + 2y, s = w + z āˆ’ 2y. 47
  • 53. Chapter 4. Calaby-Yau manifolds (a) Āµ2 = 1 3 (double sphere). (b) Āµ2 = 1 (Roman surface). (c) Āµ2 = 3 (4 planes). Figure 4.4: Real plots of the three exceptional cases. Pictures taken from [End03]. We exclude the values Āµ2 = 1 3, 1, 3 because these are exceptional cases, for which most of the statements we will make about Kummer surfaces KĀµ will not hold. These three cases correspond, repectively, to the double sphere, the Roman surface, and 4 planes, and are shown in Figure 4.4 (in the plots in the ļ¬gures the parameter w is set to w=1). As a side note, recall the Veronese surface, which is given by the embedding RP2 ā†’ RP5 by [x : y : z] ā†’ [x2 : y2 : z2 : xy : xz : yz]. The Roman surface referred to above is the projection of this surface into RP3 . Example 4.24. We will now give a non-algebraic example of Kummer surface. Let T2 = C2 /Ī“ 48
  • 54. Chapter 4. Calaby-Yau manifolds be a complex torus of complex dimension 2. (Thus, Ī“ āŠ‚ C2 is an additive subgroup such that there is an isomorphism of Rāˆ’vector spaces Ī“ āŠ— R = C2. Let (z, w) be coordinates on C2, and deļ¬ne i(z, w) = (āˆ’z, āˆ’w). Since Ī“ is a subgroup under addition, i(Ī“) = Ī“. Thus, i descends to an automorphism Ėœi: T2 ā†’ T2. If we want to ļ¬nd the ļ¬xed points of i, we need to study the solutions to i(z, w) ā‰” (z, w) mod Ī“. These solutions are (z, w)|(2z, 2w) āˆˆ Ī“ and so Ėœi has as ļ¬xed points 1 2Ī“/Ī“. (There are 16 of these.) If we deļ¬ne X := T2 /Ėœi, it turns out that X is a Kummer surface. This surface in fact has 16 singular points at the images of the ļ¬xed points of Ėœi. To see the structure of these singular points, consider the action of i on a small neighborhood U of (0, 0) in C2. Then U/i is isomorphic to a neighborhood of a singular point of X. To describe U/i, we note that the invariant functions on U are generated by z2 , zw, w2 . Thus, if we let r = z2, s = zw and t = w2 we can write U/i = {(r, s, t) near (0, 0, 0)|rt = s2 }. This is a rational double point of type A1. dz āˆ§ dw is a global holomorphic 2-form on C2, invariant under the action of Ī“, and so descends to a form on T2. Since d(āˆ’z) āˆ§ d(āˆ’w) = dz āˆ§ dw, the form is also invariant under the action of i, so we get a form dz āˆ§ dw on X{singular points}. In local coordinates, dr = 2zdz, dt = 2wdw so that dz āˆ§ dw = dr āˆ§ dt 4zw = dr āˆ§ dt 4s . It is easy to check that this form induces a global nowhere vanishing holo- morphic 2-form on the minimal resolution X of X. To ļ¬nish checking that X is a K3 surface, we use the fact that H1 (OX ) = H1 (OX) = { elements of H1(OT2 ) invariant under Ėœi }. 49
  • 55. Chapter 4. Calaby-Yau manifolds Now H1(OT2 ) = H(0,1)(T2), the space of global diļ¬€erential forms of type (0, 1). Since Ėœiāˆ— (dĀÆz) = āˆ’dĀÆz and Ėœiāˆ— (d ĀÆw) = āˆ’d ĀÆw, this space is generated by dĀÆz and d ĀÆw. It follows that H1 (OX) = H1 (OX ) = (0), and that X and X are both K3 surfaces. Proposition 4.25. We have the following facts about the Kummer surface KĀµ. ā€¢ KĀµ is irreducible. ā€¢ As suggested by the appearance of the tetrahedral coordinates in the deļ¬ning equation, KĀµ has tetrahedral symmetry. ā€¢ KĀµ has 16 singularities, each of which is an ordinary double point, a three-dimensional analogue of a node singularity. It is interesting to note that a complex surface may have at most ļ¬nitely many ordinary double points. For quartic surfaces, the maximum number of ordinary double points is 16, which is achieved by KĀµ. ā€¢ Resolving the 16 singularities3 of KĀµ, we obtain a K3 surface. This K3 surface, which is sometimes given as the deļ¬nition of a Kummer surface, contains 16 disjoint rational curves. Some Kummer surfaces are shown in Figure 4.5. Note that since these plots are restricted to the real numbers, some of the ordinary double points may not be visible. In fact, in the case 0 ā‰¤ Āµ2 ā‰¤ 1 3 (not plotted), KĀµ only contains 4 real points (each of which turns out to be an ordinary double point). In the following table are collected all the real and complex nodes of the surfaces obtained by letting Āµ vary. 3 In algebraic geometry, the problem of resolution of singularities asks whether every algebraic variety V has a resolution, a non-singular variety W with a proper birational map W ā†’ V. For varieties over ļ¬elds of characteristic 0 this was proved in Hironaka (1964), while for varieties over ļ¬elds of characteristic p it is an open problem in dimensions at least 4. 50
  • 56. Chapter 4. Calaby-Yau manifolds (a) 1 3 ā‰¤ Āµ2 ā‰¤ 1. (b) 1 ā‰¤ Āµ2 ā‰¤ 3. (c) 3 ā‰¤ Āµ2 . (d) Āµ2 = āˆž. Figure 4.5: Some Kummer Surfaces. Pictures taken from [End03]. Āµ2 Real nodes Complex nodes Picture 0 ā‰¤ Āµ2 < 1 3 4 12 4 real points Āµ2 = 1 3 / / Fig.4.4 a) 1 3 < Āµ2 < 1 4 12 Fig.4.5 a) Āµ2 = 1 / / Fig.4.4 b) 1 < Āµ2 < 3 16 0 Fig.4.5 b) Āµ2 = 3 / / Fig.4.4 c) Āµ2 ā‰„ 3 16 0 Fig.4.5 c) Āµ2 = āˆž 16 0 Fig.4.5 d) 51
  • 57. Chapter 4. Calaby-Yau manifolds 4.2.3 3-dimensional and higher-dimensional Calabi-Yau man- ifolds While the number of Calabi-Yau manifolds with one or two complex dimen- sions is known, the situation in three complex dimension is much diļ¬€erent. Several thousand Calabi-Yau three-folds have been discovered; with one exception T3, their metrics are not explicitly known, and it is not even known (although it is strongly suspected) that the number of topologically distinct Calabi-Yau three-folds is ļ¬nite. Therefore we will focus on complex projective spaces and we will analyse some interesting result. Before entering into details we need to recall some rudiments about intersec- tion theory. Let f1, . . . , fk be complex, homogeneous polynomials of degree d1, . . . , dk in n + k + 1 complex variables z = (z1, . . . , zn+k+1), deļ¬ne X(f1, . . . , fk) := {[z] āˆˆ CPn+k | fi(z) = 0 for i = 1, . . . , k}. The set X = X(f1, . . . , fk) is an algebraic variety of dimension n. I(X) = {Fi(z) āˆˆ C[z] | Fi(a) = 0, āˆ€a āˆˆ X}. Deļ¬nition 4.26 (Complete intersection). Let X = X(f1, . . . , fk) be an n dimensional algebraic variety in CPn+k. X is a complete intersection if there exists k homogeneous polynomials Fi(z1, . . . , zn+k+1), for i = 1, . . . , k, which generate all other homogeneous polynomials that vanish on X, i.e., I(X) =< F1, . . . , Fk > . Proposition 4.27. The n dimensional complete intersection X of k smooth hypersurfaces of degree d1, . . . , dk in CPn+k {is a variety with trivial canonical bundle} ā‡ā‡’ d1+Ā· Ā· Ā·+dk = n+k+1. When n = 3, for example, we have the following solutions of 4 + k = d1 + Ā· Ā· Ā· + dk with di > 1 : 5 = 5, 6 = 4 + 2 or 6 = 3 + 3, and 7 = 3 + 2 + 2. In looking for further three dimensional examples, we consider complete intersections in weighted projective spaces. 52
  • 58. Chapter 4. Calaby-Yau manifolds Weighted Projective Spaces. A weighted projective space CPn (w) is a generalization of CPn . Both are quotients of Cn+1{0} by an action of Cāˆ— = C{0}. The weights w are sequences of natural numbers w = (w0, . . . , wn) āˆˆ Nn+1 and the action of Ī» āˆˆ Cāˆ— on (z0, . . . , zn) āˆˆ Cn+1{0} is given by the formula Ī» Ā· (z0, . . . , zn) = (Ī»w0 z0, . . . , Ī»wn zn). We assume that the greatest common divisor of the wi is 1. Remark. From now on we will work on C thus we will denote CPn (w) simply with Pn(w). For each subset S =āŠ‚ {w0, . . . , wn} we denote by q(S) the greatest common divisor of the wi with i āˆˆ S. Let H(S) denote the subset of all (zj) āˆˆ Pn(w) with zi = 0 for i /āˆˆ S. The points in H(S) are cyclic quotient singularities for the group Zq(S)Z. Furthermore we need the polynomials to be quasihomogeneous due to the nature of the weighted projective space. A quasihomogeneous polynomial is deļ¬ned as: Deļ¬nition 4.28. A polynomial f is called quasihomogeneous of degree d if the following relation holds: f(Ī»w0 z0, . . . , Ī»wn zn) = Ī»d f(z0, . . . , zn). (4.14) We will deļ¬ne hypersurfaces in Pn(w) by using transverse polynomials. We deļ¬ne a transverse polynomial as follows: Deļ¬nition 4.29. A polynomial f is transverse if f = 0 only at the origin. This means that for a given set of weights not any polynomial is quasiho- mogeneous, as can be seen in the following example: Example 4.30. We consider the space P1(2, 3) and the polynomial: f = x2 y + y2 = (Ī»2 x)2 (Ī»3 y) + (Ī»3 y)2 = Ī»2Ā·2+3Ā·1 x2 y + Ī»3Ā·2 y2 = Ī»d f. Clearly there is no degree d to satisfy the relations and hence this polynomial is not quasihomogeneous. 53
  • 59. Chapter 4. Calaby-Yau manifolds Complete Intersections in Weighted Projective Spaces. The equa- tions of hypersurfaces in the weighted projective space Pn(w) of degree d are given by transverse quasihomogeneous polynomial equations f(z0, . . . , zn) = 0. Now in order to ļ¬nd Calabi Yau manifold in Pn(w) the following proposition can help. Proposition 4.31. The complete intersections of multiple degree (d1, . . . , dk) in the weighted projective space Pm+k(w) with trivial canonical bundle are those satisfying the following condition w0 + Ā· Ā· Ā· + wm+k = d1 + Ā· Ā· Ā· + dk. For instance elliptic curves arise either as quartic curves in P2(1, 1, 2) and sextic curves in P2(1, 2, 3). We can take for example w4 + x4 + y2 = 0 and w6 + x3 + y2 = 0 respectively. Example 4.32. Degree 8 hypersurface in P4(1, 1, 2, 2, 2) with equation X : y8 0 + y8 1 + y4 2 + y4 3 + y4 4 = 0. The degree 8 = 1 + 1 + 2 + 2 + 2 so the hypersurface in this degree is a Calabiā€“Yau manifold. Example 4.33. Degree 12 hypersurface in P4(1, 1, 2, 2, 6) with equation y12 0 + y12 1 + y6 2 + y6 3 + y2 4 = 0. The degree 12 = 1 + 1 + 2 + 2 + 6 so the hypersurface in this degree is a Calabiā€“Yau manifold. Example 4.34. Degree 18 hypersurface in P4(1, 1, 1, 6, 9) with equation y18 0 + y18 1 + y18 2 + y3 3 + y2 4 = 0. The degree 18 = 1 + 1 + 1 + 6 + 9 so that the hypersurface in this degree is a Calabiā€“Yau manifold. There is a complete classiļ¬cation of Calabiā€“Yau varieties arising from transverse hypersurfaces in P4(w) but we will not go further. 4.3 KƤhler manifolds Deļ¬nition 4.35. Let (M, Ļ‰) be a symplectic manifold. If there exist a genuine complex structure J compatible with Ļ‰, then (M, Ļ‰, J) is a KƤhler manifold. The symplectic form Ļ‰ is then called a KƤhler form. 54
  • 60. Chapter 4. Calaby-Yau manifolds Remark. Here genuine means coming from holomorphic coordinates making M a complex manifold. Hence a KƤhler manifold is a symplectic manifold endowed with a com- patible complex structure, and as we saw in the previous chapters, g deļ¬ned by g(u, v) = Ļ‰(u, Jv), is a Riemannian metric called KƤhler metric and Ļ‰ is called KƤhler form. Locally on the almost complex manifold M, the holomorphic vector ļ¬elds T1,0 x (M), as we have already seen, have a basis āˆ‚ āˆ‚z1 , . . . , āˆ‚ āˆ‚zn and the holo- morphic 1-forms Ī› (1,0) x have the related dual basis denoted dz1, . . . , dzn. The coeļ¬ƒcients of the KƤhler form in these local coordinates are gj,k(z, ĀÆz) = g āˆ‚ āˆ‚zj , āˆ‚ āˆ‚ĀÆzk , and the diļ¬€erential form Ļ‰ = āˆ’2i n j,k=1 gj,k(z, ĀÆz)dzj āˆ§ ĀÆzk . Therefore a KƤhler manifold is a smooth manifold equipped with ā€¢ complex structure; ā€¢ Riemannian structure; ā€¢ symplectic structure, which are compatible (as seen in Section 3.3). The existence of a KƤhler metric on a compact manifold constraints the topology. In particular the following properties apply. Proposition 4.36. If (M, J, Ļ‰) is a compact KƤhler manifold of real dimen- sion 2n, then i. the complex structure on M leads to the Hodge decomposition (which is equivalent to having equality in (1.15)) Hk deRham(M) = p+q=k Hp,q deRham(M) = p+q=k Hp,q ĀÆāˆ‚ (M); ii. h(p,q)(M) = h(q,p)(M) iii. the odd Betty numbers b2r+1(M) := dim H2r+1 deRham(M) = dim ker d Im d , where d: Ī›2r+1(M) ā†’ Ī›2r+2(M), are even āˆ€i = 1, . . . , n; 55
  • 61. Chapter 4. Calaby-Yau manifolds iv. b2r(M) > 0. Before entering into the details of the proof we will make a brief digression concerning some operators and results of complex diļ¬€erential geometry. Each tangent space V = TxM has a positive inner product (Ā·, Ā·), part of the Riemannian metric in a compatible triple. Let e1, . . . , en be a positively oriented orthonormal basis of V. The star operator is a linear operator āˆ—: Ī›(V ) ā†’ Ī›(V ) deļ¬ned by āˆ—(1) = e1 āˆ§ Ā· Ā· Ā· āˆ§ en (e1 āˆ§ Ā· Ā· Ā· āˆ§ en) = 1 (e1 āˆ§ Ā· Ā· Ā· āˆ§ ek) = āˆ—(ek+1 āˆ§ Ā· Ā· Ā· āˆ§ en). We see that āˆ—: Ī›k(V ) ā†’ Ī›nāˆ’k and satisļ¬es āˆ—āˆ— = (āˆ’1)k(nāˆ’k). The āˆ— operator will be hereafter used to deļ¬ne an inner product on forms. Let ĀÆāˆ‚ be deļ¬ned as in Deļ¬nition 1.5 and d as in Deļ¬nition 1.6. We now deļ¬ne the adjoint operator of ĀÆāˆ‚ ĀÆāˆ‚āˆ— : Ī›(p,q) ā†’ Ī›(p,qāˆ’1) (M) (4.15) by requiring that ĀÆāˆ‚āˆ— Ļˆ, Ī· = Ļˆ, ĀÆāˆ‚Ī· for all Ī· āˆˆ Ī›(p,qāˆ’1)(M). The Dolbeault cohomology group Hp,q ĀÆāˆ‚ (M) = Zp,q ĀÆāˆ‚ (M)/ĀÆāˆ‚Ī›(p,qāˆ’1)(M) is rep- resented by the solutions of the two ļ¬rst-order equations ĀÆāˆ‚Ļˆ = 0, ĀÆāˆ‚āˆ— Ļˆ = 0. (4.16) These two may be replaced by the single second-order equation āˆ†ĀÆāˆ‚Ļˆ = ĀÆāˆ‚ ĀÆāˆ‚āˆ— + ĀÆāˆ‚āˆ— ĀÆāˆ‚ Ļˆ = 0. For a complete proof see [GH14]. The operator āˆ†ĀÆāˆ‚ : Ī›(p,q) (M) ā†’ Ī›(p,q) (M) is called the ĀÆāˆ‚āˆ’Laplacian or simply the Laplacian (written āˆ†) if no ambiguity is likely. Diļ¬€erential forms satisfying the Laplace equation āˆ†Ļˆ = 0 are called harmonic forms; the space of harmonic forms of type (p, q) is denoted Hp,q(M) and called the harmonic space. The isomorphism Hp,q (M) = Hp,q ĀÆāˆ‚ (M) 56
  • 62. Chapter 4. Calaby-Yau manifolds can be proved (see for instance [GH14]). Before stating the Hodge theorem, whose isomorphism (4.3) is a part, we begin by giving an explicit formula for the adjoint operator ĀÆāˆ‚āˆ—. First we deļ¬ne the star operator, āˆ—: Ī›(p,q) (M) ā†’ Ī›(pāˆ’1,qāˆ’1) (M) by requiring (Ļˆ, Ī·) = M Ļˆ āˆ§ āˆ—Ī· for all Ļˆ āˆˆ Ī›(p,q)(M). If we suppose that M is compact this deļ¬ne an inner product on forms. Therefore if we write Ī· = I,J Ī·IJ Ļ•I āˆ§ ĀÆĻ•J then āˆ—Ī· = 2p+qāˆ’n I,J ĪµIJ ĀÆĪ·IJ Ļ•I0 āˆ§ ĀÆĻ•J0 , where I0 = {1, . . . , n} āˆ’ I and we write ĪµIJ for the sign of the permutation (1, . . . , n, 1 , . . . , n ) ā†’ (i1, . . . , ip, j1, . . . , jq, i0 1, . . . , i0 nāˆ’p, j0 1, . . . , j0 nāˆ’q). The signs work out so that āˆ— āˆ— Ī· = (āˆ’1)p+q Ī·. In terms of star the adjoint operator is ĀÆāˆ‚āˆ— = āˆ’ āˆ— ĀÆāˆ‚ āˆ— . Observation 4.37. Note that ĀÆāˆ‚2 = 0 ā‡’ ĀÆāˆ‚āˆ—2 = 0. We are now ready to state the following theorem. Theorem 4.38 (Hoge). Let M be a compact complex manifold, then 1. dim Hp,q(M) < āˆž and 2. the orthogonal projection H: Ī›(p,q) (M) ā†’ Hp,q (M) (4.17) is well deļ¬ned and there exists a unique operator, the Greenā€™s operator, G: Ī›(p,q) (M) ā†’ Ī›(p,q) (M), with G(H(p,q)(M)) = 0, ĀÆāˆ‚G = GĀÆāˆ‚, ĀÆāˆ‚āˆ—G = GĀÆāˆ‚āˆ— and I = H + āˆ†G (4.18) on Ī›(p,q)(M). 57
  • 63. Chapter 4. Calaby-Yau manifolds The content of (4.18) is sometimes expressed by saying that, given Ī·, the equation āˆ†Ļˆ = Ī· has a solution Ļˆ if and only if H(Ī·) = 0, and then Ļˆ = G(Ī·) is the unique solution satisfying H(Ļˆ) = 0. Thus we should try to solve the Laplace equation on a compact manifold. The idea is to ļ¬nd a Ļˆ such that (Ļˆ, āˆ†Ļ•) = (Ī·, Ļˆ) for all Ļ• āˆˆ Ī›(p,q)(M) and to prove that this Ļˆ is in fact Cāˆž. Remark. We remark that we may deļ¬ne the adjoint dāˆ— of d, form the Laplacian āˆ†d = ddāˆ— + dāˆ—d, and arrive at the exact formalism as for ĀÆāˆ‚ on complex manifolds. Moreover the Hodge theorem is also true. Let M be a compact complex manifold with Hermitian metric ds2, and suppose that in some open set U āŠ‚ M, ds2 is Euclidean; that is there exists local holomorphic coordinates z = (z1, . . . , zn) such that ds2 = dzi āŠ— dĀÆzi. Theorem 4.39. With the same hypothesis as above, for a diļ¬€erential form Ļ• = Ļ•IJ dzI āˆ§ dĀÆzJ compactly supported in U, āˆ†d = 2āˆ†ĀÆāˆ‚. (4.19) Proof. Let zi = xi + iyi, then āˆ†ĀÆāˆ‚(Ļ•) = āˆ’2 I,J,i āˆ‚2 āˆ‚ziāˆ‚ĀÆzi Ļ•IJ dzI āˆ§ dĀÆzJ = āˆ’ 1 2 I,J,i āˆ‚2 āˆ‚x2 i + āˆ‚2 āˆ‚y2 i Ļ•IJ dzI āˆ§ dĀÆzJ = 1 2 āˆ†d(Ļ•), i.e., the ĀÆāˆ‚āˆ’Laplacian is equal to the ordinary dāˆ’Laplacian in U, up to a constant. Although very few compact complex manifolds have everywhere Euclidean metrics, but as it turns out in order to insure Equation (4.19) on a complex manifold it is suļ¬ƒcient that the metric approximate the Euclidean metric to the second order at each point. We can now start the demonstration of Theorem 4.36. 58
  • 64. Chapter 4. Calaby-Yau manifolds Proof. (i) and (ii). Set Hp,q d (M) = {Ī· āˆˆ Ī›p,q (M): āˆ†dĪ· = 0}, Hr d(M) = {Ī· āˆˆ Ī›r (M): āˆ†dĪ· = 0}. Note that the two groups depend on the particular metric whilst the group Hp,q deRham(M) is intrinsically deļ¬ned by the complex structure. By the com- mutativity of āˆ†d and Ī p,q : Hr deRham(M) ā†’ Hp,q deRham(M) and the fact that āˆ†d is real, the harmonic forms satisfy Hr(M) = p+q=r Hp,q(M) Hp,q(M) = Hp,q d (M) (4.20) On the other hand, for Ī· a closed form of pure type (p, q), Ī· = Ī· + ddāˆ— G(Ī·), where the harmonic part Ī· also has a pure type (p, q). Thus every de Rham cohomology class on a compact oriented riemannian manifold M possesses a unique harmonic representative, i.e., Hk = Hk DeRahm(M). We also have the following orthogonal decomposition with respect to (Ā·, Ā·) : Ī›k = Hk āŠ• āˆ†(Ī›k (M)) = Hk āŠ• d(Ī›kāˆ’1 ) āŠ• dāˆ— (Ī›k+1 ). The proof involves functional analysis, elliptic diļ¬€erential operators, pseu- dodiļ¬€erential operators and Fourier analysis; see [GH14]. When M is KƤhler, the Laplacian satisļ¬es āˆ†d = 2āˆ†ĀÆāˆ‚, hence harmonic forms are also bigraded Hk = p+q=k Hp,q . Combining this with 4.20 and Theorem 4.38 we obtain the Hodge decompo- sition for the Laplacian āˆ†d. For a compact KƤhler manifold M, the complex cohomology satisļ¬es Hr DeRahm(M, C) = p+q=r Hp,q DeRahm(M) Hp,q DeRahm(M) = Hp,q DeRahm(M). (4.21) Hence, we have the following isomorphisms: Hk DeRahm(M) = Hk = p+q=r Hp,q (M) = p+q=r Hp,q ĀÆāˆ‚ (M). 59
  • 65. Chapter 4. Calaby-Yau manifolds (iii) For the point (ii) of the proposition it follows that b2r+1(M) = p+q=2r+1 h(p,q) (M) and considering that h(p,q)(M) = h(q,p)(M), we obtain b2r+1(M) = 2 j h(j,2r+1āˆ’j) (M). Therefore b2r+1(M) ā‰” 0 mod 2. (iv) This fact directly follows from the fact that every KƤhler manifold is also a symplectic manifold. In fact let Ļ‰ be the symplectic form ensured by the KƤhler structure. If dĪ± = Ļ‰r with 0 ā‰¤ r ā‰¤ n by Stokesā€™ Theorem we have that M Ļ‰n = M d(Ī± āˆ§ Ļ‰nāˆ’r ) = 0. Remark. Note that (i) is true because dimC (Hr deRham(M, C)) = dimR (Hr deRham(M, R)) , since we previously deļ¬ned the Betti numbers for real de Rham cohomology groups. Actually KƤhler manifolds satisfy several other topological property but here we have mentioned only the ones that will be used later. For more details see for instance [DS01]. Deļ¬nition 4.40 (KƤhler potential). Let (X, J, Ļ‰) be a KƤhler manifold then around every point x āˆˆ X there exists a neighbourhood U and a function f āˆˆ Cāˆž(U, R) for which the KƤhler form Ļ‰ can be written as Ļ‰|U = iāˆ‚ ĀÆāˆ‚f. Here, the operators āˆ‚ = k āˆ‚ āˆ‚zk dzk and ĀÆāˆ‚ = k āˆ‚ āˆ‚ĀÆzk dĀÆzk are called the Dolbeault operators. 60
  • 66. Chapter 4. Calaby-Yau manifolds For instance, in Cn, the function f = |z|2 2 is a KƤhler potential for the KƤhler form, because iāˆ‚ ĀÆāˆ‚ 1 2 |z|2 = 1 2 iāˆ‚ ĀÆāˆ‚ k zk ĀÆzk = 1 2 iāˆ‚ k zkdĀÆzk = 1 2 i k dzk āˆ§ dĀÆzk = Ļ‰. We have that the converse holds true in fact we can apply the following property. Before going into the proposition we will see a necessary deļ¬nition. Actually the metric we will talk about is the same as discussed in Section 4.3. Deļ¬nition 4.41 (Compatible Riemannian Metric.). Let M be a complex manifold with corresponding complex structure J. We say that a Riemannian metric g is compatible with J if g(JX, JY ) = g(X, Y ) (4.22) for all vector ļ¬elds X, Y on M. A complex manifold together with a compat- ible Riemannian metric is called a Hermitian manifold. Proposition 4.42. Let M be a complex manifold with a compatible Rieman- nian metric g, as in (4.22). Then the following assertions are equivalent: 1. g is a KƤhler metric. 2. For each point x āˆˆ M, there is a smooth real function f in a neigh- bourhood U of x such that Ļ‰|U = iāˆ‚ ĀÆāˆ‚f. 3. dĻ‰ = 0. Proof. By deļ¬nition, 1 and 3 are equivalent. 3 ā‡’ 2 Since Ļ‰ is real and dĻ‰ = 0, we have Ļ‰ = dĪ± locally, where Ī± is a real 1-form. Then Ī± = Ī² + ĀÆĪ², where Ī² is a form of type (1, 0). Since Ļ‰ is of type (1, 1), we have āˆ‚Ī² = 0, ĀÆāˆ‚ ĀÆĪ² = 0 and Ļ‰ = ĀÆāˆ‚Ī² + āˆ‚ ĀÆĪ² Hence Ī² = āˆ‚Ļ† locally, where Ļ† is a smooth complex function. Then ĀÆĪ² = ĀÆāˆ‚ ĀÆĻ† and hence Ļ‰ = ĀÆāˆ‚āˆ‚Ļ† + āˆ‚ ĀÆāˆ‚ ĀÆĻ† = āˆ‚ ĀÆāˆ‚(ĀÆĻ† āˆ’ Ļ†) = iāˆ‚ ĀÆāˆ‚f, 61
  • 67. Chapter 4. Calaby-Yau manifolds with f = i(Ļ† āˆ’ ĀÆĻ†). 2 ā‡’ 3 Let d = āˆ‚ + ĀÆāˆ‚, then we simply observe that āˆ‚Ļ‰ = iāˆ‚2 ĀÆāˆ‚f = 0, since āˆ‚2 = 0 and that ĀÆāˆ‚Ļ‰ = iāˆ‚ ĀÆāˆ‚2 f = 0, since ĀÆāˆ‚2 = 0. Thanks to the equivalence of the deļ¬nition of KƤhler metric we will show in the following example how the complex projective space can be seen as a KƤhler manifold. Example 4.43 (Complex Projective space). Using the same notation as in example 1.4, let CPn be the complex projective space. To see that CPn has a natural KƤhler manifold structure, we introduce the following functions which turn out to be KƤhler potentials. Let f be deļ¬ned in an open set Uj by fj = log ļ£« ļ£­1 + k=j zk zj 2 ļ£¶ ļ£ø = log n k=0 |zk|2 āˆ’ log |zj|2 . On the intersection of two coordinate charts Uj āˆŖ Uk, the diļ¬€erence fj āˆ’ fk = log |zk|2 āˆ’ log |zj|2 = log zk zj āˆ’ log ĀÆzk ĀÆzj . satisļ¬es the equation āˆ‚ ĀÆāˆ‚ (fj āˆ’ fk) = 0, and hence there exists a global form Ļ‰ on CPn with Ļ‰|Uj = iāˆ‚ ĀÆāˆ‚fj. To see that this form Ļ‰ is the form associated with a KƤhler structure, we consider its coeļ¬ƒcients gk,l in local coordinates where Ļ‰|Uj = iāˆ‚ ĀÆāˆ‚fj = k,l gk,ldwk āˆ§ d ĀÆwl 62
  • 68. Chapter 4. Calaby-Yau manifolds and wk = zk zj . In order to see that the coeļ¬ƒcient matrix (hk,l) is positive deļ¬nite and Hermitian symmetric we calculate the diļ¬€erentials using fj = log ļ£« ļ£­1 + k=j zk zj 2 ļ£¶ ļ£ø = log 1 + n k=1 |wk|2 . Thus ĀÆāˆ‚fj = 1 + n k=1 |wk|2 āˆ’1 Ā· n k=1 wkd ĀÆwk and āˆ‚ ĀÆāˆ‚fj = 1 + n k=1 |wk|2 āˆ’1 Ā· n k=1 dwk āˆ§ d ĀÆwk āˆ’ 1 + n k=1 |wk|2 āˆ’2 Ā· n k,l=1 ĀÆwkdwk āˆ§ wld ĀÆwl = 1 + n k=1 |wk|2 āˆ’2 Ā· n k,l=1 Ī“k,l 1 + n k=1 |wk|2 āˆ’ ĀÆwkwl dwk āˆ§ d ĀÆwl. For a complex vector Ī¾ = (Ī¾1, . . . , Ī¾n) āˆˆ Cn, we study the positivity properties of the following expression using the Hermitian inner product < w, Ī¾ >= n k=1 wk ĀÆĪ¾k to obtain the inequality n k,l=1 Ī“k,l 1 + n k=1 |wk|2 āˆ’ ĀÆwkwl Ī¾k ĀÆĪ¾l =< Ī¾, Ī¾ >2 1+ < w, w >2 āˆ’ | < w, Ī¾ > |2 for Ī¾ = 0, and the Schwarz inequality | < w, Ī¾ > |2 ā‰¤ < Ī¾, Ī¾ >2< w, w >2 leads to ā‰„ < Ī¾, Ī¾ >2 1+ < w, w >2 āˆ’ < Ī¾, Ī¾ >2 < w, w >2 =< Ī¾, Ī¾ >2 ā‰„ 0. 4.3.1 KƤhler and Calabi-Yau manifolds We have decided to introduce KƤhler manifold in our dissertation in order to see their connection with Calabi-Yauā€™s. We want to clarify that Calabi- Yau manifolds are a particular kind of KƤhler manifolds. Many diļ¬€erent deļ¬nitions of Calabiā€“Yau manifolds exist in the literature; we list here some 63