Learning Notes for Algebraic

Geometry
Typed by
Bowen Liu1

[email protected]
 1

C ONTENTS

Part 1. Preliminaries 3
1. Category theory 3
1.1. Category 3
1.2. Abelian category 4
2. Sheaf and cohomology 6
2.1. Sheaves 6
2.2. Derived functor formulation of sheaf cohomology 12
2.3. Acyclic resolution 14
2.4. Examples about acyclic sheaf 15
2.5. Proof of de Rham theorem using sheaf cohomology 20
2.6. Hypercohomology 21

Part 2. Schemes 23
3. Schemes and Morphisms 23
3.1. Schemes 23
3.2. Proj construction 25
4. Properties of schemes 27
4.1. Quasi-compact, irreducible, Noetherian topological space 27
4.2. Reduced, irreducible and integral scheme 28
4.3. Affine criterion 28
4.4. Noetherian scheme 29
5. Properties of morphisms 31
5.1. Quasi-compact, affine, finite type and finite 31
5.2. Birational morphism 33
5.3. Open immersion and closed immersion 34
5.4. Fibred product 35
5.5. Separated morphisms 36
5.6. Proper and projective morphisms 38
6. Coherent sheaves 40
6.1. O X -modules 40
6.2. Coherent sheaves 41
6.3. Sheaf of ideals 43
6.4. Coherent sheaves on Proj 44

Part 3. Homework 47
7. Homework 47
7.1. Homework-1 47
7.2. Homework-2 52
7.3. Homework-3 58
7.4. Homework-4 60
7.5. Homework-5 61
2

7.6. Homework-6 65
7.7. Homework-7 67
7.8. Homework-8 69
7.9. Homework-9 71
7.10. Homework-10 79
7.11. Homework-11 81
7.12. Homework-12 84
7.13. Homework-13 87
References 90
 3

Part 1. Preliminaries
1. C ATEGORY THEORY

1.1. Category.

1.1.1. Category and Functors.

1.1.2. Morphisms.

Definition 1.1.1 (monomorphism). A morphism f : A → B in C is called a monomor-

phism (or injective) if for any two morphisms α, β : C → A satisfying f ◦ α = f ◦ β,
we have α = β.

Definition 1.1.2 (epimorphism). A morphism f : A → B in C is called a epimor-

phism (or surjective) if for any two morphisms α, β : A → C satisfying α ◦ f = β ◦ f ,
we have α = β.

Definition 1.1.3 (bijective). A morphism is called bijective if it’s both monomor-

phism and epimorphism.

Definition 1.1.4 (isomorphism). A morphism is called an isomorphism if it admits

two-sided inverse.

Remark 1.1.1. Any isomorphism is bijective, but in general a bijective morphism

may not be an isomorphism. For example, in the category of topological spaces, it’s
easy to construct a morphism (continuous map) which is a bijective map, but it’s
not an isomorphism.

1.1.3. Categorical objects.

Definition 1.1.5 (direct product). Let { A i } i∈ I be a family of objects in category

Q Q
C . The direct product of A i is tuple ( i∈ I A i , p i ), where i∈ I A i is an object in
Q
C , and p i : i∈ I A i → A i is a family of morphisms called projections, such that
the following universal property: For any object C and any family of morphisms
Q
f i : C → A i , there exists a unique morphism f : C → i∈ I A i such that p i ◦ f = f i
for all i ∈ I .
Q f
I ∈I A i C
pi
fi
Ai

Definition 1.1.6 (direct sum). Let { A i } i∈ I be a family of objects in category C .

L L
The direct sum of A i is tuple ( i∈ I A i , k i ), where i∈ I A i is an object in C , and
L
k i : A i → i∈ I A i is a family of morphisms called projections, such that the follow-
ing universal property: For any object C and any family of morphisms f i : A → C ,
L
there exists a unique morphism f : i∈ I A i → C such that f ◦ k i = f i for all i ∈ I .
4

L f
i∈ I Ai C
ki
fi
Ai
1.2. Abelian category.
1.2.1. Additive category.
Definition 1.2.1 (additive category). A category C is called an additive category
if for any objects A, B, C in C ,
(1) the direct product of A and B exists;
(2) Hom( A, B) is an abelian group, and 0 ∈ Hom( A, B) is called zero morphism;
(3) the map
Hom( A, B) × Hom(B, C ) → Hom( A, C )
( f , g) 7→ g ◦ f
is bilinear.
Definition 1.2.2. Let C be an additive category and f : A → B be a morphism in
C.
(1) A morphism K → A is the kernel of f if the composite K → A → B is 0, and for
any morphism K 0 → A such that the composite K 0 → A → B is 0, there exists a
unique morphism K 0 → K such that the diagram
K0

K A
commutes. For convenience we often denote K by ker f and call it the kernel
of f .
(2) A morphism B → C is the cokernel of f if the composite A → B → C is 0, and
for any morphism B → C 0 such that the composite A → B → C is 0, there exists
a unique morphism C → C 0 such that the diagram
B C

C0
commutes. For convenience we often denote C by coker f and call it the coker-
nel of f .
(3) The image of f is defined to be the kernel of the cokernel of f , and the coimage
of f is defined to be the cokernel of the kernel of f .
Remark 1.2.1. A kernel is neccessarily a monomorphism, and a cokernel is nec-
cessarily an epimorphism.
Remark 1.2.2. There is a natural morphism coim f → im f induced by universal
property
 5

f
ker f A B coker f

coim f im f

Definition 1.2.3 (zero object). Let C be an additive category. A zero object 0 in C

is an object such that Hom(0, 0) = {0}.
1.2.2. Abelian category.
Definition 1.2.4 (abelian category). An abelian category C is an additive category
with zero objects such that for every morphism f in C , the kernel and the cokernel
of f exist, and the canonical morphism coim f → im f is an isomorphism.
Proposition 1.2.1. In abelian category, a bijective morphism is an isomorphism.
Definition 1.2.5 (exact). In an abelian category, a sequence of morphisms
u v
A −→ B −→ C
is called exact if v ◦ u = 0 and the canonical morphism from coim u → ker v is an
isomorphism.
Definition 1.2.6 (short exact sequence). An exact sequence of the form
0→ A→B→C→0
is called a short exact sequence.
Definition 1.2.7 (split). A short exact sequence 0 → A → B → C → 0 is called split
if it’s isomorphic to
0 → A → A ⊕ C → C → 0,
where A → A ⊕ C and A ⊕ C → C are the canonical morphisms.
6

2. S HEAF AND COHOMOLOGY

2.1. Sheaves. Along this section, X denotes a topological space.

2.1.1. Definitions and Examples.

Definition 2.1.1 (sheaf). A presheaf of abelian group F on X consisting of the

following data:
(1) For any open subset U of X , F (U ) is an abelian group.
(2) If U ⊆ V are two open subsets of X , then there is a group homomorphism
r VU : F (V ) → F (U ). Moreover, above data satisfy
I F (∅) = 0.
II rUU = id.
III If W ⊆ U ⊆ V are open subsets of X , then r V W = rUW ◦ r VU .
Moreover, F is called a sheaf if it satisfies the following extra conditions
IV Let {Vi } i∈ I be an open covering of open subset U ⊆ X and s ∈ F (U ). If
s|Vi := rUVi ( s) = 0 for all i ∈ I , then s = 0.
V Let {Vi } i∈ I be an open covering of open subset U ⊆ X and s i ∈ F (Vi ). If
s i |Vi ∩V j = s j |Vi ∩V j for all i, j ∈ I , then there exists s ∈ F (U ) such that s|Vi =
s i for all i ∈ I .

Example 2.1.1 (constant presheaf). For an abelian group G , the constant presheaf
assign each open subset U the group G itself, but in general it’s not a sheaf.

Definition 2.1.2 (morphism of presheaves). A morphism φ : F → G between presheaves

consisting of the following data:
(1) For any open subset U of X , there is a group homomorphism φ(U ) : F (U ) →
G (U ).
(2) If U ⊆ V are two open subsets of X , then the following diagram commutes
φ(U )
F (V ) G (V )
r VU r VU
φ(V )
F (U ) G (U )

Notation 2.1.1. For convenience, for s ∈ F (U ), we often write φ( s) instead of

φ(U )( s).

Remark 2.1.1. The morphisms between sheaves are defined as morphisms of presheaves.

Definition 2.1.3 (isomorphism). A morphism of presheaves φ : F → G is called

an isomorphism if it has two-sided inverse, that is, there exists a morphism of
presheaves ψ : G → F such that ψ ◦ φ = idF and φ ◦ ψ = idG .

Remark 2.1.2. A morphism of presheaves φ : F → G is an isomorphism if and only

if for every open subset U ⊆ X , φ(U ) → G (U ) is an isomorphism of abelian groups.
 7

2.1.2. Stalks.

Definition 2.1.4 (stalks). For a presheaf F and p ∈ X , the stalk at p is defined as

F p = lim F (U )
−−→
p∈U

Remark 2.1.3 (alternative definition). In order to avoid language of direct limit, we

give a more useful but equivalent description of stalk: For p ∈ U ∩ V , sU ∈ F (U )
and s V ∈ F (V ) are equivalent if there exists x ∈ W ⊆ U ∩ V such that sU |W = s V |W .
An element s p ∈ F p , which is called a germ, is an equivalence class [ sU ], and for
s ∈ F (U ), the germ given by s is denoted by s| p .

Notation 2.1.2.
(1) For s ∈ F (U ) and p ∈ U , s| p denotes the equivalent class it gives.
(2) For s p ∈ F p , s ∈ F (U ) denotes the section such that s| p = s p .

Definition 2.1.5 (morphisms on stalks). Given a morphism of sheaves φ : F → G ,

it induces a morphism of abelian groups φ p : F p → G p as follows:
φp : Fp → Gp
s p 7→ φ( s)| p .

Remark 2.1.4. It’s neccessary to check the φ p is well-defined since there are dif-
ferent choices s such that s| p = s p .

Proposition 2.1.1. Let φ : F → G be a morphism between sheaves. Then φ is an

isomorphism if and only if the induced map φ p : F p → G p is an isomorphism for
every p ∈ X .

Proof. It’s clear if φ is an isomorphism between sheaves, then it induces an iso-

morphism between stalks. Conversely, it suffices to show φ(U ) : F (U ) → G (U ) is
an isomorphism for every open subset U ⊆ X .
(1) Injectivity: For s, s0 ∈ F (U ) such that φ( s) = φ( s0 ), by passing to stalks one has
φ p ( s| p ) = φ p ( s0 | p ) for every p ∈ U , and thus s| p = s0 | p since φ p is an isomor-
phism. By definition of stalks there exists an open subset Vp ⊆ U containing p
such that s agrees with s0 on Vp . Then it gives an open covering {Vp } of U , and
by axiom (IV) one has s = s0 on U .
(2) Surjectivity: For t ∈ G (U ), by passing to stalks there exists s p ∈ F p such that
φ p ( s p ) = t| p for every p ∈ U since φ p is surjective. By definition of stalks there
exists an open subset Vp ⊆ U containing p and s ∈ F (Vp ) such that φ( s) = t
on Vp . This gives a collection of sections defined on an open covering {Vp } of
U , and by injectivity we proved above one has these sections agree with each
other on the intersections. Then by axiom (V) there exists a section s ∈ F (U )
such that φ( s) = t.
□
8

2.1.3. Sheafification. In Example 2.1.1, we come across a presheaf that is not a

sheaf. To obtain a sheaf from a presheaf, we require a process known as sheafifi-
cation. One approach to defining sheafification is through its universal property.

Definition 2.1.6 (sheafification). Given a presheaf F there is a sheaf F + and

a morphism θ : F → F + with the property that for any sheaf G and any mor-
phism φ : F → G there is a unique morphism φ : F + → G such that the following
diagram commutes:
φ
F G
φ
θ

F+
The universal property shows that if the sheafification exists, then it’s unique
up to a unique isomorphism. One way to give an explicit construction of sheafifi-
cation is to glue stalks together in a suitable way. Let F + (U ) be a set of functions
a
f:U→ Fp
p∈U

such that f ( p) ∈ F p and for every p ∈ U there is an open subset Vp ⊆ U containing

p and t ∈ F (Vp ) such that t| q = f ( q) for all q ∈ Vp .

Proposition 2.1.2. F + is the sheafication of F .

Proof. Firstly let’s show F + is a sheaf: It’s clear F + is a presheaf, so it suffices to

check conditions (IV) and (V) in the definition. Let U ⊆ X be an open subset and
{Vi } be an open covering of U .
(1) If s ∈ F + (U ) such that s|Vi = 0 for all i , then s must be zero: It suffices to
show s( p) = 0 for all p ∈ U . For any p ∈ U , then there exists an open subset Vi
contains p, hence s( p) = s|Vi ( p) = 0.
(2) Suppose there exists a collection of sections { s i ∈ F + (Vi )} i∈ I such that
s i |Vi ∩V j = s j |Vi ∩V j
holds for all i, j ∈ I . Now we construct s ∈ F + (U ) as follows: For p ∈ U and Vi
containing p, we define s( p) = s i ( p). This is well-defined since s i agree on the
intersections, so it remains to show s ∈ F + (U ). It’s clear s( p) ∈ F p . For p ∈ U ,
there exists Vi containing p, and thus there exists Wi ⊆ Vi containing p and
t ∈ F (Wi ) such that t| q = s i ( q) = s( q) for all q ∈ Vp .
There is a canonical morphism θ : F → F + as follows: For open subset U ⊆ X ,
and s ∈ F (U ), θ ( s) is defined by
a
θ ( s) : U → Fp
p∈U
p 7→ s| p .
Note that if F is a sheaf, the canonical morphism θ : F → F + is an isomorphism.
 9

(1) Injectivity: If s ∈ F (U ) such that s| p = 0 for all p ∈ U , then there exists an open
covering {Vi } i∈ I of U such that s|Vi = 0, by axiom (IV) of sheaf one has s = 0.
(2) Surjectivity: For f ∈ F + (U ) and p ∈ U , there exists p ∈ Vp ⊆ U and t ∈ F (Vp )
such that f ( p) = t| p by construction of F + . Then glue these sections together
to get our desired s such that θ ( s) = f .
Finally let’s show F + statisfies the universal property of sheafification. A mor-
phism of presheaves φ : F → G induces a map on stalks
φp : Fp → Gp.
For f ∈ F + (U ), the composite of f with the map
a a a
φp : Fp → Gp
p∈U p∈U p∈U
`
gives a map φ e( f ) : U → p∈U G p , and in fact φ e( f ) ∈ G + (U ): For p ∈ U , φ
e( f )( p) ∈ G p
since f ( p) ∈ F p and φ p : F p → G p . If for all q ∈ Vp we have t| q = f ( q), then
e( f )( q) = φ q ( f ( q)) = φ q ( t| q ) = φ( t)| q .
φ
Since G is a sheaf, the canonical morphism θ 0 : G → G + is an isomorphism, so we
can define φ : = θ 0−1 ◦ φ
e. Now let’s show φ = φ ◦ θ = θ 0−1 ◦ φ e ◦ θ . It’s easy to show
they coincide on each stalk since φ p = θ 0−
p
1
◦ e
φ p ◦ θ p , and thus φ = φ ◦ θ by Propo-
sition 2.1.1. Furthermore, uniqueness follows from the fact that φ p is uniquely
determined by φ p . □
Remark 2.1.5. From the construction, one can see the stalk of F + at p is exactly
Fp.
Remark 2.1.6. The sheafification can be described in a more fancy language: Since
we have sheaf of abelian groups on X as a category, denote it by Ab X , and presheaf
is a full subcategory of Ab X , there is a natural inclusion functor ι from category of
sheaf to category of presheaf. The sheafification is the adjoint functor of ι.
Example 2.1.2 (constant sheaf). For an abelian group G , the associated constant
sheaf G is the sheafication of the constant presheaf. By the construction of sheafifi-
cation, G can be explicitly expressed as
G (U ) = {locally constant function f : U → G }
2.1.4. Exact sequence of sheaf. Given a morphism φ : F → G between sheaves of
abelian groups, there are the following presheaves
U 7→ ker φ(U )
U 7→ im φ(U )
U 7→ coker φ(U ),
since φ(U ) : F (U ) → G (U ) is a group homomorphism.
Proposition 2.1.3. Kernel of a morphism between sheaves is a sheaf.
Proof. Let {Vi } i∈ I be an open covering of U .
10

(1) For s ∈ ker φ(U ), if s|Vi = 0, then s = 0 since s is also in F (U ).

(2) If there exists s i ∈ ker φ(Vi ) such that s i |Vi ∩V j = s j |Vi ∩V j , then they glue to-
gether to get s ∈ F (U ). Note that
φ(U )( s)|Vi = φ(Vi )( s|Vi ) = φ(Vi )( s i ) = 0
Then s ∈ ker φ(U ).
□
But the image of morphism may not be a sheaf. Although we can prove the first
requirement in the same way, the proof for the second requirement fails: If there
exists s i ∈ im φ(Vi ), and we can glue them together to get a s ∈ G (U ), but s may
not be the image of some t ∈ F (U ). The cokernel fails to be a sheaf for the same
reason.
Definition 2.1.7 (image and cokernel). Let φ : F → G be a morphism between
sheaves of abelian groups. Then the image and cokernel of φ is defined to be the
sheafification of the following presheaves
U 7→ im φ(U )
U 7→ coker φ(U )
respectively.
Definition 2.1.8 (exact). For a sequence of sheaves:
φ i−1 φi
· · · → F i−1 −→ F i −→ F i+1 → . . .
It’s called exact at F i , if ker φ i = im φ i−1 . If a sequence is exact at everywhere,
then it’s an exact sequence of sheaves.
Definition 2.1.9 (short exact sequence). An exact sequence of sheaves is called a
short exact sequence if it looks like
φ ψ
0 → F −→ G −→ H → 0
Proposition 2.1.4. Let φ : F → G be a morphism between sheaves of abelian
groups. Then for any p ∈ X , one has
(ker φ) p = ker φ p
(im φ) p = im φ p .
Proof. For (1). It’s clear (ker φ) p ⊆ ker φ p . Conversely, if s p ∈ ker φ p , then φ p ( s p ) =
0 ∈ G p . In other words, there exists an open subset U containing p and s ∈ F (U )
such that s| p = s p and φ( s)| p = 0, which implies there is another open subset V
containing p such that φ( s)|V = 0. Hence φ( s|V ) = 0, that is, s|V ∈ ker φ(V ). Thus
s p = ( s|V )| p ∈ (ker φ) p .
For (2). It’s clear (im φ) p ⊆ im φ p since the sheafication doesn’t change stalk.
Conversely, if s p ∈ im φ p , then there exists t p ∈ F p such that φ p ( t p ) = s p . Suppose
t ∈ F (U ) is a section of some open subset U containing p such that t| p = t p . Then
 11

φ( t)| p = φ p ( t p ) = s p . In other words, s p is in the stalk of the image presheaf at p,

but the sheafication doesn’t change stalk, so we have s p ∈ (im φ) p . □
Corollary 2.1.1. The sequence of sheaves
φ i−1 φi
· · · → F i−1 −→ F i −→ F i+1 → . . .
is exact if and only if the sequence of abelian groups are exact
φ ip−1 φ ip
· · · → F pi−1 −→ F pi −→ F pi+1 → . . .
for all p ∈ X .

Corollary 2.1.2. The the sequence of sheaves

0→F →G
is exact if and only if for any open subset U , the following sequence of abelian
groups is exact
0 → F (U ) → G (U ).

Method one. For any open subset U ⊆ X , one has

φ(U ) : F (U ) → G (U )
is injective, since by definition we have for any open subset U ⊆ X , ker φ(U ) = 0,
that is injectivity. □
Method two. Or from another point of view, for each p ∈ U , we have
φp : Fp → Gp
is injective. That is ker φ p = 0. So we obtain (ker φ(U )) p = 0 for all p ∈ U . But
for a section s ∈ F (U ) if we have s| p = 0, then we must have s = 0. So we obtain
ker φ(U ) = 0. □
Example 2.1.3 (exponential sequence). Let X be a complex manifold and O X be
its holomorphic function sheaf. Then
p exp ∗
0 → 2π −1Z → O X −→ O X →0
is an exact sequence of sheaves, called exponential sequence.

Proof. The difficulty is to show exp is surjective on stalks at p ∈ X . That is we

∗
need to construct logarithms of functions g ∈ O X (U ) for U , a neighborhood of p.
We may choose U is simply-connected, then define
ˆ
dg
log g( q) = log g( p) +
γq g
for q ∈ U , where γ q is a path from p to q in U , and the definition is independent of
the choice of γ q since U is simply-connected. □
12

Remark 2.1.7. In fact, U is simply-connected is crucial for constructing logarithm.

If we consider X = C and U = C \{0}, then
∗
exp : O X (U ) → O X (U )
cannot be surjective.
2.2. Derived functor formulation of sheaf cohomology. The category Ab X :
sheaves of abelian groups on X . In this section we will introduce sheaf cohomology
by considering it as a derived functor.
Given an exact sequence of sheaf as follows
ϕ ψ
0 → F 0 −→ F −→ F 00 .
By taking its section over open subset U , we obtain a sequence of abelian groups
ϕ(U ) ψ(U )
0 → F 0 (U ) −→ F (U ) −→ F 00 (U ).
Above sequence is not only exact at F 0 (U ), but also is exact at F (U ). In other
words, the functor given by taking section over open subset is a left exact functor.
(1) Firstly let’s show ker ψ(U ) ⊇ im ϕ(U ). For s ∈ F 0 (U ), if we want to show ψ ◦
ϕ( s) = 0, it suffices to show (ψ ◦ ϕ( s))| p = 0 for all p ∈ U since F 00 is a sheaf. For
any p ∈ U , by considering stalk at p we obtain an exact sequence of abelian
groups
ϕp ψp
0 → F p0 −→ F p −→ F p00 .
Then we obtain ψ p ◦ ϕ p ( s| p ) = 0, which implies (ψ ◦ ϕ( s))| p = 0.
(2) Conversely, Given s ∈ ker ψ(U ), we have s| p ∈ ker ψ p for any p ∈ U . By exact-
ness of stalks, there exists t p ∈ F p0 such that ϕ p ( t p ) = s| p . Thus there exists
an open subset Vi containing p and t i ∈ F 0 (Vi ) such that ϕ( t i ) = s|Vi . Now it
suffices to show these t i can be glued together to obtain t ∈ F (U ), and since F
is a sheaf, it suffices to check these t i agree on intersections Vi ∩ V j . Note that
ϕ( t i − t j |Vi ∩Vi ) = s|Vi ∩V j − s|Vi ∩V j = 0, then these t i agree on intersections since
ϕ is injective.

Remark 2.2.1. From above argument, we can see that

ϕ ψ
0 → F 0 −→ F −→ F 00
is exact if and only if for any open subset U ⊆ X
ϕ(U ) ψ(U )
0 → F 0 (U ) −→ F (U ) −→ F 00 (U )
is exact.
In homological algebra, we always consider the derived functor of a left or right-
exact functor. In particular, the functor of taking global section is a left exact func-
tor, and its right derived functor defines the cohomology of a sheaf. Before we come
into the definition of derived functor, firstly let’s define the injective resolution of
a sheaf.
 13

Definition 2.2.1 (injective). A sheaf I is injective if Hom(−, I ) is an exact functor.

Definition 2.2.2 (injective resolution). Let F be a sheaf. An injective resolution
of F is an exact sequence
0 → F → I 0 → I 1 → I 2 → ...
where I i are injective for all i .
Theorem 2.2.1. Every sheaf admits an injective resolution.
Theorem 2.2.2. Let F → I • and G → G • are two resolutions and ϕ : F → G be
e : I • → G • which lifts ϕ,
a morphism of sheaves. Then there exists a morphism ϕ
which is unique up to homotopy.
Definition 2.2.3 (sheaf cohomology). Let F be a sheaf of abelian groups. Then
H p ( X , F ) := H p (I • ( X )).
Remark 2.2.2. The Theorem 2.2.2 shows that the definition of sheaf cohomology is
independent of the choice of injective resolution.
Example 2.2.1. By definition, the 0-th cohomology is exact the global section
© ª
H 0 ( X , F ) := ker I 0 ( X ) → I 1 ( X ) .
Thus H 0 ( X , F ) = F ( X ), the global sections of sheaf.
Example 2.2.2. If F is a injective sheaf, then H i ( X , F ) = 0 for all i > 0, since the
sheaf cohomology of injective sheaf can be computed by using the following special
injective resolution
id
0 → F −→ F → 0 → 0 → . . .
Theorem 2.2.3 (zig-zag). If
0→F →G →H →0
is a short sequence of sheaves, then there is an induced long exact sequence of
abelian groups
0 → H0( X , F ) → H0( X , G ) → H0( X , H ) → H1( X , F ) → H1( X , G ) → . . .
Definition 2.2.4 (direct image). Let f : X → Y be continuous map between topo-
logical spaces and F be a sheaf of abelian groups on X . The direct image of F ,
denoted by f ∗ F , is a sheaf on Y defined by
f ∗ F (U ) := F ( f −1 (U )).
Proposition 2.2.1. f ∗ : Ab X → AbY is a left exact functor.
Proof. Given an exact sequence of sheaves on X
0 → F 0 → F → F 00 .
Then we need to show
0 → f ∗ F 0 → f ∗ F → f ∗ F 00
14

is also an exact sequence of sheaves on Y . By Remark 2.2.1 it suffices to show that

for any open subset V ⊆ Y , we have the following exact sequence
0 → f ∗ F 0 (V ) → f ∗ F (V ) → f ∗ F 00 (V ),
and that’s exactly
0 → F 0 ( f −1 (V )) → F ( f −1 (V )) → F 00 ( f −1 (V )).
Since f is continuous, then f −1 (V ) is an open subset in X , and thus above sequence
of abelian is exact since 0 → F 0 → F → F 00 is exact. □

2.3. Acyclic resolution. In practice it may be difficult for us to choose an injec-

tive resolution, so we usual other resolutions to compute sheaf cohomology.

Definition 2.3.1 (acyclic sheaf). A sheaf F is acyclic if H i ( X , F ) = 0 for all i > 0.

Example 2.3.1. Injective sheaf is acyclic.

Definition 2.3.2 (acyclic resolution). Let F be a sheaf. An acyclic resolution of

F is an exact sequence
0 → F → A 0 → A 1 → A 2 → ...
where A i is acyclic for all i .

Proposition 2.3.1. The cohomology of sheaf F can be computed using acyclic

resolution.

In fact, it’s a quite homological techniques, called dimension shifting, so we will

state this technique in language of homological algebra. Let’s see a baby version
of it.

Example 2.3.2. Let F be a left exact functor and 0 → A → M1 → B → 0 be an exact

sequence with M1 is F -acyclic. Then R i+1 F ( A ) ∼
= R i F (B) for i > 0, and R 1 F ( A ) is
the cokernel of F ( M1 ) → F (B).

Proof. By considering the long exact sequence induced by 0 → A → M 1 → B → 0,

one has
R i F ( M 1 ) → R i F (B) → R i+1 F ( A ) → R i+1 F ( M 1 )
(1) If i > 0, then R i F ( M 1 ) = R i+1 F ( M 1 ) = 0 since M 1 is F -acyclic, and thus
R i+1 F ( A ) ∼
= R i F (B) for i > 0.
(2) If i = 0, then
0 → F ( M 1 ) → F (B ) → R 1 F ( A ) → 0
implies R 1 F ( A ) = coker{F ( M 1 ) → F (B)}.
□

Now let’s prove dimension shifting in a general setting.

 15

Lemma 2.3.1 (dimension shifting). If

0 → A → M1 → M2 → · · · → M m → B → 0

is exact with M i is F -acyclic, then R i+m F ( A ) ∼

= R i F (B) for i > 0, and R m F ( A ) is
the cokernel of F ( M m ) → F (B).

Proof. Prove it by induction on m. For m = 1, we already see it in Example 2.3.2.

Assume it holds for m < k, then for m = k, let’s split 0 → A → M 1 → M 2 → · · · →
dk
M k −→ B → 0 into two exact sequences

0 → A → M 1 → M 2 → · · · → M k−1 → ker dk → 0
dk
0 → ker dk → M k −→ B → 0.

Then by induction hypothesis, for i > 0 we have

R i+k−1 F ( A ) ∼
= R i F (ker dk )
R i+1 F (ker dk ) ∼
= R i F (B).

Combine these two isomorphisms together we obtain R i+k F ( A ) ∼ = R i F (B) for i >
0, as desired. For i = 0, it suffices to let i = 1 in R i+k−1 F ( A ) ∼
= R i F (ker dk ), then
we obtain
R k F ( A ) = R 1 F (ker dk ) = coker{F ( M k ) → F (B)}.

This completes the proof. □

Corollary 2.3.1. If 0 → A → M • is a F -acyclic resolution, then R i F ( A ) = H i (F ( M • )).

Proof. Truncate the resolution as

0 → A → M 0 → M 1 → . . . M i−1 → B → 0
0 → B → M i → M i+1 → . . .

Since we already have R i F ( A ) = coker{F ( M i−1 ) → F (B)}, and F is left exact, one
has
F (B) = ker{F ( M i ) → F ( M i+1 )}.

Thus we obtain

R i F ( A ) = coker{F ( M i−1 ) → ker{F ( M i ) → F ( M i+1 )}} = H i (F ( M • )).

2.4. Examples about acyclic sheaf.

16

2.4.1. Flabby sheaf. First kind of acyclic sheaf is flabby2 sheaf.

Definition 2.4.1 (flabby). A sheaf F is flabby if for all open U ⊆ V , the restriction
map F (V ) → F (U ) is surjective.
Now let’s see some examples about flabby sheaves.
Example 2.4.1. A constant sheaf on an irreducible topological space is flabby.
Proof. Note that the constant presheaf on a irreducible topological space is a sheaf
in fact, and it’s easy to see this constant presheaf is flabby. □
In particular, we have
Example 2.4.2. Let X be an algebraic variety. Then constant sheaf Z X is flabby.
Example 2.4.3. If F is a flabby sheaf on X , and f : X → Y is a continuous map,
then f ∗ F is a flabby sheaf on Y .
Proof. For V ⊆ W in Y , it suffices to show f ∗ F (W ) → f ∗ F (V ) is surjective, and
that’s
F ( f −1 W ) → F ( f −1 V )
it’s surjective since F is flabby. □
Example 2.4.4. An injective sheaf is flabby.
Proof. Let I be an injective sheaf and V ⊆ U be open subsets. Now we define sheaf
ZU on X by
(
Z(W ) W ⊆ U
ZU :=
0 otherwise
where Z is constant sheaf valued in Z, and similarly we define sheaf ZV . By
construction one has ZU (W ) = ZV (W ) unless W ⊆ U and W 6⊆ V . Thus we obtain an
exact sequence
0 → ZV → ZU .
Applying the functor Hom(−, I ), which is exact, we obtain an exact sequence
Hom(ZU , I ) → Hom(ZV , I ) → 0.
Now let’s explain why we need such a weird sheaf ZU . In fact, we will prove
Hom(ZU , I ) = I (U ). Indeed since φ : ZU → I is a sheaf morphism. Then if W 6⊆ U ,
then φ(U ) must be zero. If W = U , then the group of sections of ZU (U ) over any
connected component is simply Z and hence φ(U ) on this connected component is
determined by the image of 1 ∈ Z. Thus φ(U ) can be thought of an element of I (U ).
Now on any proper open subset of U , φ is determined by restriction maps. Hence
Hom(ZU , I ) = I (U ), as desired. The same argument shows Hom(ZU , I ) = I (V ),
and thus we obtain an exact sequence
I (U ) → I (V ) → 0,
which shows I is flabby. □
2Some authors also call this flasque.
 17

Our goal is to prove a flabby sheaf is acyclic, but we still need some property of
flabby sheaves.
ϕ ψ
Proposition 2.4.1. If 0 → F 0 −→ F −→ F 00 → 0 is an exact sequence of sheaves,
and F 0 is flabby, then for any open subset U , the sequence
ϕ(U ) ψ(U )
0 → F 0 (U ) −→ F (U ) −→ F 00 (U ) → 0
is exact.
Proof. It suffices to show F (U ) → F 00 (U ) → 0 is exact. Here we only gives a sketch
of the proof. Since we have exact sequence on stalks for each p ∈ U as follows
ϕp ψp
0 → F p0 −→ F p −→ F p00 → 0
Then for each s ∈ F 00 (U ), there exists t p ∈ F p such that ψ p ( t p ) = s| p , so there
exists open subset Vp ⊆ U containing p and t ∈ F (Vp ) such that ψ( t) = s|Vp . If we
can glue these t together then we get a section in F (U ) and is mapped to s, which
completes the proof. However, they may not equal on the intersection. But things
are not too bad, consider another point q and t0 ∈ F (Vq ) such that ψ( t0 ) = s|Vq ,
( t0 − t)|Vp ∩Vq ∈ ker ψ(Vp ∩ Vq ) = im ϕ(Vp ∩ Vq ). So there exists t00 ∈ F 0 (Vp ∩ Vq ) such
that
ϕ( t00 ) = ( t0 − t)|Vp ∩Vq
Now since F 0 is flabby, then there exists t000 ∈ F (Vp ) such that t000 |Vp ∩Vq = t00 . And
consider t + ϕ( t000 ) ∈ F (Vp ), which will coincide with t0 on Vp ∩ Vq . After above
corrections, we can glue t after correction together. □
Proposition 2.4.2. If 0 → F 0 → F → F 00 → 0 is an exact sequence of sheaves,
and if F 0 and F are flabby, then F 00 is flabby.
Proof. Take V ⊆ U and consider the following diagram

0 F 0 (U ) F (U ) F 00 (U ) 0

0 F 0 (V ) F (V ) F 00 (V ) 0
Then the desired result follows from five lemma. □
Proposition 2.4.3. A flabby sheaf is acyclic.
Proof. Let F be a flabby sheaf. Since there are enough injective objects in the
category of sheaf of abelian groups, there is an exact sequence
0→F →I →Q→0
with I is injective. By Example 2.4.4 we have I is flabby, and thus by Proposition
2.4.2 we have Q is flabby. Consider the long exact sequence induced from above
short exact sequence
F ( X ) → I ( X ) → Q( X ) → H 1 ( X , F ) → H 1 ( X , I ) → . . .
18

Note that H 1 ( X , I ) = 0 since I is injective, and thus acyclic. Then H 1 ( X , F ) =

coker{I ( X ) → Q ( X )}. But Proposition 2.4.1 shows that I ( X ) → Q ( X ) is surjective
since F is flabby, so H 1 ( X , F ) = 0.
Now let’s prove H k ( X , F ) = 0 for k > 0 by induction on k, and above argument
shows it’s true for k = 1. Assume this holds for k < n, and consider
· · · → H n−1 ( X , Q ) → H n ( X , F ) → H n ( X , I ) → H n ( X , Q ) → . . .

By induction hypothesis, we can reduce above sequence to

· · · → 0 → H n ( X , F ) → 0 → H n ( X , Q) → . . .

which implies H n ( X , F ) = 0. This completes the proof. □

2.4.2. Soft sheaf. The second kind of acyclic sheaves is called soft sheaves, which
is quit similar to flabby.

Definition 2.4.2 (soft). A sheaf F over X is soft if for any closed subset S ⊆ X the
restriction map F ( X ) → F (S ) is surjective.

Remark 2.4.1. For closed subset S , the section over it is defined by

F (S ) := lim F (U )
−−→
S ⊆U

Parallel to Proposition 2.4.1 and Proposition 2.4.2, soft sheaf has the following
properties:
ϕ ψ
Proposition 2.4.4. If 0 → F 0 −→ F −→ F 00 → 0 is an exact sequence of sheaves,
and F 0 is soft, then the following sequence
ϕ( X ) ψ( X )
0 → F 0 ( X ) −→ F ( X ) −→ F 00 ( X ) → 0
is exact.

Proposition 2.4.5. If 0 → F 0 → F → F 00 → 0 is an exact sequence of sheaves,

and if F 0 and F are soft, then F 00 is soft.

Proposition 2.4.6. A soft sheaf is acyclic.

So you may wonder, what’s the difference between flabby and soft since the
definitions are quite similar, and both of them are acyclic. Clearly by definition of
sections over a closed subset, we know that every flabby sheaf is soft, but converse
fails

Example 2.4.5. The sheaf of smooth functions on a smooth manifold is soft but
not flabby.

Lemma 2.4.1. If M is a sheaf of modules over a soft sheaf of rings R , then M is

a soft sheaf.
 19

Proof. Let s ∈ M (K ) for some closed subset K ⊆ X . Then s extends to some open
neighborhood U of K . Let ρ ∈ R (K ∪ ( X \ U )) be defined by
(
1, on K
ρ=
0, on X \ U
Since R is soft, then ρ extends to a section over X , then ρ ◦ s is the desired exten-
sion of s. □
2.4.3. Fine sheaf. Another important kind of acyclic sheaves, which behaves like
sheaf of differential forms ΩkX is called fine sheaf. Recall what is a partition of
unity: Let U = {U i } i∈ I be a locally finite open covering of topological space X . A
partition of unity subordinate to U is a collection of continuous functions f i : U i →
[0, 1] for each i ∈ I such that its support lies in U i , and for any x ∈ X
X
f i ( x ) = 1.
i∈ I

Definition 2.4.3 (fine sheaf). A fine sheaf F on X is a sheaf of A -modules, where

A is a sheaf of rings such that for every locally finite open covering {U i } i∈ I of X ,
there is a partition of unity X
ρi = 1
i∈ I
where ρ i ∈ A ( X ) and supp(ρ i ) ⊆ U i .
Remark 2.4.2. For a sheaf F on X and a section s ∈ F ( X ), its support is defined
as
supp( s) := { x ∈ X : s| x 6= 0}.
Proposition 2.4.7. A fine sheaf is acyclic.
Proof. Let F be a sheaf of A -modules and a fine sheaf. And choose a injective
resolution
d d d d
0 → F −→ I 0 −→ I 0 −→ I 1 −→ . . .
such that I i are injective sheaves of A -modules. Let s ∈ I p ( X ) such that d s = 0.
Then by exactness of injective resolution we have X is covered by open subsets
U i such that for each i there is an element t i ∈ I p−1 (U i ) such that d t i = s|Ui . By
passing to a refinement we may assume that the cover {U i } is locally finite. Let
P
{ρ i } be a partition of unity subordinate to {U i }. Then we have t = ρ i t i ∈ I p−1 ( X )
such that d t = s. This completes the proof. □
Example 2.4.6. Let M be a smooth manifold and π : E → M be a vector bundle.
Then the sheaf of smooth sections of E is a C ∞ ( M )-module sheaf, which is a fine
sheaf. In particular, the sheaf of tangent bundle, sheaf of differential forms Ω M and
k-forms ΩkM are fine sheaves.
Remark 2.4.3. As a consequence, it’s meaningless to compute cohomology of sheaf
of differential k-forms, or any other vector bundle over a smooth manifold. But in
20

complex version, something interesting happens: Let ( X , O X ) be a complex mani-

fold and π : E → X be a holomorphic vector bundle. Then the sheaf of holomorphic
sections of E is not a fine sheaf since there is no partition of unity may not be
holomorphic, so the cohomology of holomorphic vector bundle is not trivial, and
that’s what Dolbeault cohomology computes.
For fine sheaf and soft sheaf, we have
Lemma 2.4.2. Fine sheaf is soft.
Proof. Let F be a fine sheaf, S ⊆ X closed and s ∈ F (S ). Let {U i } be an open
covering of S and s i ∈ F (U i ) such that
s i |S ∩Ui = s|S ∩Ui .
`
Let U0 = X − S , and s 0 = 0. Then {U i } {U0 } is an open covering of X . Without lose
of generality, we assume this open covering is locally finite and choose a partition
of unity {ρ i } subordinate to it. Then
X
s := ρ i ( s i )
i
is a section in F ( X ) which extends s. □
Remark 2.4.4. Until now, we have shown that soft, fine and flabby sheaves are
acyclic. Lemma 2.4.2 shows fine sheaf is soft, and by definition a flabby sheaf is
soft. The Example 2.4.5 shows that soft sheaf may not be flabby, and constant
sheaf on an irreducible space is flabby but not fine. In a summary, we have the
following relations:
Acyclic

Soft

Fine

Flabby

2.5. Proof of de Rham theorem using sheaf cohomology. As we already

know, for constant sheaf R over a smooth manifold M , we have the following fine
resolution
i d d d
0 → R −→ Ω0M −→ Ω1M −→ Ω2M −→ . . .
And de Rham cohomology computes the sheaf cohomology of R. de Rham theorem
implies that de Rham cohomology equals to the singular cohomology with real
 21

coefficient. So if we can give constant sheaf another resolution using singular

cochains, we may derive the de Rham cohomology.
We state this in a general setting: Let X be a topological manifold, and a con-
stant sheaf G over X , where G is an abelian group. Let S p (U,G ) be the group
of singular cochains in U with coefficients in G , and let δ denote the coboundary
operator.
Let S p (G ) be the sheaf over X generated by the presheaf U 7→ S p (U,G ), with
δ
induced differential mapping S p (G ) −→ S p+1 (G ).
Similar to Poincaré lemma, we have for a unit ball U in Euclidean space, we
have the following sequence
δ δ
· · · → S p−1 (U,G ) −→ S p (U,G ) −→ S p+1 (U,G ) → . . .
is exact. So we have the following resolution of the constant sheaf G
δ δ
0 → G → S 0 (G ) −→ S 1 (G ) −→ S 2 (G ) → . . .
Remark 2.5.1. If M is a smooth manifold, then we can consider smooth chains,
that is f : ∆ p → U , where f is a smooth function. The corresponding results above
still hold, and we have a resolution by smooth cochains with coefficients in G :
•
0 → G → S∞ (G )
•
So if we choose G = R, then it suffices to show 0 → R → S∞ (R) is an acyclic
resolution, then we obtain de Rham theorem.
p
First, note that S∞ is a S∞ 0
-module, given by cup product on open subsets.
Then by Lemma 2.4.1 and the fact S∞ 0
is soft we know that it’s a soft resolution.
This completes the proof.
2.6. Hypercohomology. In homological algebra, the hypercohomology is a gen-
eralization of cohomology functor which takes as input not objects in abelian cat-
egory but instead chain complexes of objects.
One of the motivations for hypercohomology is to generalize the zig-zag lemma,
that is, the short exact sequence of sheaves induces a long exact sequence of coho-
mology groups. It turns out hypercohomology gives techniques for constructing a
similar cohomological associated long exact sequence from an arbitrary long exact
sequence
0 → F1 → F2 → · · · → F k → 0
Now let’s give the definition of hypercohomology: Let F • : · · · → F i−1 → F i →
i +1
F → · · · be a complex of sheaves of abelian groups, which is bounded from below,
that is, F n = 0 for n ¿ 0. Then F • admits an injective resolution F • → I • . In
other words,
... F i−1 Fi F i+1 ...

... I i−1 Ii I i+1 ...

22

such that
(1) All I i are injective sheaves.
(2) The induced homomorphism H i (F • ) → H i (I • ) is an isomorphism.
The hypercohomology of F • is defined by
H i ( X , F • ) := H i (Γ( X , I • ))
Definition 2.6.1. For a sheaf F , F • [ n] is a sheaf of complex defined by
(
• F i=n
(F [ n]) =
i
0 otherwise.

Example 2.6.1. Let F be a sheaf and 0 → F → I 0 → I 1 → . . . be an injective

resolution of F . Then
0 F 0 0 ...

0 I0 I1 I2 ...
•
is an injective resolution of F [0]. Indeed, I are injective for all i ≥ 0, and
i
(
• F, n =0
i
H (I ) = = H i (F • [0])
0, otherwise
So by definition of hypercohomology, we have H i ( X , F • [0]) = H i (Γ( X , I • )) = H i ( X , F • ).
In general, one has
H i ( X , F • [ n]) ∼
= H i+n ( X , F ).
Theorem 2.6.1 (zig-zag). Let 0 → F • → G • → H • → 0 be a short exact sequence
of complexes of sheaves which are bounded from below. Then there is an induced
long exact sequence
· · · → H i−1 ( X , H • ) → H i ( X , F • ) → H i ( X , G • ) → H i ( X , H • ) → H i+1 ( X , F • ) → . . .
 23

Part 2. Schemes
3. S CHEMES AND M ORPHISMS
3.1. Schemes. Throughout this lecture, all rings are assumed to be commutative
with identity element, and all homomorphisms of rings are assumed to map 1 to
1.

3.1.1. Ringed space.

Definition 3.1.1 (ringed space). A ringed space is a pair ( X , O X ) consisting of a
topological space X and a sheaf of rings O X on X .
Definition 3.1.2 (locally ringed space). A ringed space ( X , O X ) is a locally ringed
space if for every p ∈ X , the stalk O X ,p is a local ring.
Definition 3.1.3 (morphisms between ringed space). Let ( X , O X ) and (Y , O Y ) be
two ringed space. A morphism from ( X , O X ) to (Y , O Y ) is a pair ( f , f ] ) consisting of
a continuous map f : X → Y and a morphism of sheaves f ] : O Y → f ∗ O X .
Remark 3.1.1. Let ( f , f ] ) be a morphism between ringed spaces ( X , O X ) and (Y , O Y ).
For every point p ∈ X , there is a homomorphism ( f ∗ O X ) f ( p) → O X ,p defined by
( f ∗ O X ) f ( p) = lim ( f ∗ O X )(V ) = lim O X ( f −1 (V )) → lim O X (U ) = O X ,p .
−−→ −−→ −−→
f ( p)∈V p∈ f−1 (V ) p∈U

On the other hand, the morphism of sheaves f ] : O Y → f ∗ O X induces a homeomor-

phism between stalks
( f ] ) p : O Y , f ( p) → ( f ∗ O X ) f ( p) .
By composing above two homomorphisms, there is a homomorphism
]
f p : O Y , f ( p) → O X ,p .

Definition 3.1.4 (morphisms between locally ringed space). A morphism ( f , f ] )

between locally ringed spaces ( X , O X ) and (Y , O Y ) is a morphism between ringed
spaces, and for each p ∈ X , the morphism
]
f p : O Y , f ( p) → O X ,p
is a local homomorphism of local rings.
Definition 3.1.5 (isomorphism). A isomorphism of locally ringed space is a mor-
phism with a two-side inverse.
3.1.2. Affine schemes.
Definition 3.1.6 (prime spectrum). Let A be a ring. The spectrum of A is a locally
ringed space, consisting of the following data:
(1) A topological space Spec A , which is the set of all prime ideals of A , equipped
with Zariski topology, that is, all closed subsets of Spec A are of the form V (a) =
{p ∈ Spec A | a ⊆ p}.
24

(2) A structure sheaf O Spec A , which is defined as follows: For every open subset
`
U of Spec A , O Spec A (U ) consists of mappings s : U → p∈Spec A A p satisfying the
following two conditions:
(a) For every p ∈ U , one has s(p) ∈ A p .
( b) For every p ∈ U , there exists a neighborhood Up of p contained in U and
a, f ∈ A such that for every q ∈ Up , one has f 6∈ q and s(q) = a/ f in A q .
Definition 3.1.7 (affine scheme). A locally ringed space that is isomorphic to
(Spec A, O Spec A ) for some ring A is called an affine scheme.
Definition 3.1.8 (distinguished open subset). Let A be a ring and f ∈ A . The
distinguished open subset D ( f ) is defined by Spec A \ V ( f ).
Proposition 3.1.1. The distinguished open subsets form a topology basis of Zariski
topology.
Corollary 3.1.1. The affine scheme is quasi-compact.
Proposition 3.1.2. Let A be a ring.
(1) For every p ∈ Spec A , there is a canonical isomorphism O p ∼ = Ap.
(2) For every f ∈ A , there is a canonical isomorphism O (D ( f )) ∼
= A f . In particular,
O (Spec A ) ∼
= A.
Proof. See Proposition 1.2.4 in [Fu06]. □
Proposition 3.1.3.
(1) Let ϕ : A → B be a homomorphism of rings. Then ϕ induces a canonical mor-
phism of locally ringed spaces
( f , f ] ) : (Spec B, O Spec B ) → (Spec A, O Spec A ).
(2) Any morphism ( f , f ] ) between (Spec B, O Spec B ) and (Spec A, O Spec A ) is obtained
this way.
Proof. See Proposition 1.2.5 in [Fu06]. □
Proposition 3.1.4. For any f ∈ A , there is a canonical isomorphism of locally
ringed spaces
(D ( f ), O Spec A |D ( f ) ) ∼
= (Spec A f , O Spec A f ).
Proof. See Proposition 1.2.6 in [Fu06]. □
3.1.3. Schemes.
Definition 3.1.9 (scheme). A scheme ( X , O X ) is a locally ringed space for which
there exists an open covering {U i } i∈ I of X such that each (U i , O X |Ui ) is an affine
scheme.
Notation 3.1.1. For convenience, if there is no ambiguity of the underlying topo-
logical space, we will simply write O instead of O X .
 25

Proposition 3.1.5. Let ( X , O X ) be a scheme and U be an open subset of X . Then

(U, O X |U ) is a scheme.
Proof. It follows from Proposition 3.1.4. □
Proposition 3.1.6. Let X be a scheme and A be a ring. Then there is a one-to-
one correspondence between the set of morphisms of schemes from X to Spec A
and the set of homomorphisms of rings from A to O X ( X ).
Proof. See Proposition 1.2.8 in [Fu06]. □
3.2. Proj construction. In this section we fix a graded ring S with decompostion
L
S= ∞ d =0 S d .

Proposition 3.2.1. An ideal of S is called a homogeneous ideal if it satisfies one

of the following equivalent conditions:
L
(1) a = d (a ∩ S d ).
P
(2) If a ∈ a and a = d a d with a d ∈ S d , then a d ∈ a.
(3) a is generated by homogeneous elements as an additive subgroup of S .
(4) a is generated by homogeneous elements an an ideal of S .
Proposition 3.2.2. Let a be a homogeneous ideal of S . If for any homogeneous
elements f and g in S such that f g ∈ a, one has either f ∈ a or g ∈ a, then a is a
prime ideal.
Proposition 3.2.3. Let a, b be homogeneous ideals of S . Then
(1) a + b, ab, a ∩ b are homogeneous ideals.
p
(2) a is a homogeneous ideal.
L
Proposition 3.2.4 (Proj). Let S + = ∞ d =1 S d and Proj S be the set of all homo-
geneous prime ideals of S not containing S + . For any homogeneous ideal a of S ,
define
V+ (a) = {p ∈ Proj S | a ⊆ p}.
Then
(1) V+ (0) = Proj S and V+ (S ) = ∅.
T P
(2) i∈ I V+ (a i ) = V+ ( i∈ I a i ) for any family of homogeneous ideals {a i } i∈ I of S .
(3) V+ (a) ∩ V+ (b) = V+ (ab) = V+ (a ∩ b) for any homogeneous ideals a and b of S .
In particular, Proj S is a topological space so that closed subsets are of the form
V+ (a) for homogeneous ideals a of S . This topology is called Zariski topology of
Proj S .
Now let’s define the structure sheaf on Proj S . For any homogeneous prime ideal
p ∈ Proj S , consider the ring
a
S (p ) = {∈ S p | a and t are homogeneous of the same degree}.
t
For open subset U ⊆ Proj S , O Proj S (U ) is defined to be the set of functions s : U →
`
p∈U S (p) such that
26

(1) For every p ∈ U , one has s(p) ∈ S (p) .

(2) For every p ∈ U , there exists a neighborhood Up of p contained in U and homo-
geneous elements a, f ∈ S of the same degree such that for every q ∈ Up , one
has f ∉ q and s(q) = a/ f in S (q) .
Proposition 3.2.5.
(1) For every p ∈ Proj S , there is a canonical isomorphism O Proj S,p ∼
= S (p) .
(2) For every homomorphism element f ∈ S + , let
D + ( f ) = Proj S \ V+ (( f )) = {p ∈ Proj S | f ∉ p}.
Then D + ( f ) is open in Proj S , and open subsets of this type form a basis for the
topology of Proj S . Moreover, there is an isomorphism of locally ringed space
(D + ( f ), O Proj S |D ( f ) ) ∼
+ = (Spec S ( f ) , O Spec S ).
(f )

In particular, (Proj S, O Proj S ) is a scheme.

Proof. See Proposition 1.2.10 in [Fu06]. □
 27

4. P ROPERTIES OF SCHEMES

4.1. Quasi-compact, irreducible, Noetherian topological space.

4.1.1. Quasi-compact scheme.
Definition 4.1.1 (quasi-compact). A scheme ( X , O X ) is called quasi-compact if any
open covering of X admits a finite subcovering.
Remark 4.1.1. A scheme is quasi-compact if and only if it’s a finite union of affine
schemes.
4.1.2. Irreducible topological space.
Definition 4.1.2 (irreducible topological space). A topological space X is called
irreducible if X is not the union of two proper closed subsets. A subset Y ⊆ X
is called irreducible if Y is a irreducible topological space equipped with induced
topological.
Proposition 4.1.1. Let X be a topological space and Y ⊆ X be a subset equipped
with induced topological. If Y is irreducible, then Y is also irreducible.
Proof. Suppose Y is not irreducible, so we have closed subsets S, T in X such that
Y is not contained in either S or T but Y ⊂ T ∪ S . If Y ⊆ S , then Y ⊆ S = S , a
contradiction, so Y is not contained in S . Similarly Y is not contained in T . But
Y ⊆ Y ⊆ S ∪ T , thus Y is not irreducible, a contradiction. □
Proposition 4.1.2. Let A be a ring. A closed subset of Spec A is irreducible if and
only if it’s of the form V (p) for some prime ideal p of A .
Proof. See Proposition 1.3.2 in [Fu06]. □
Proposition 4.1.3. Let X be a scheme. For any irreducible closed subset Y of X ,
there exists a unique point y ∈ Y such that Y = { y}
Proof. See Proposition 1.3.3 in [Fu06]. □
4.1.3. Noetherian topological space.
Definition 4.1.3 (Noetherian topological space). A topological space X is called
a Noetherian topological space if the family of closed subsets of X satisfies the
descending chain conditions.
Example 4.1.1. If A is a Noetherian ring, then Spec A is a Noetherian topological
space.
Proposition 4.1.4. Suppose X is a Noetherian topological space.
(1) For every closed subset Y of X , there is a decompostion Y = Y1 ∪ · · · ∪ Yn into
closed irreducible subsets Yi such that Yi 6⊂ Y j whenever i 6= j , where Yi are
called irreducible component of Y .
(2) An irreducible closed subset Y is an irreducible component of X if and only if
Y is maximal among the family of irreducible closed subset of X .
28

Proof. See Proposition 1.3.4 in [Fu06]. □

Definition 4.1.4 (minimal prime ideal). Let A be a ring. A prime ideal of A is
called a minimal prime ideal if it contains no prime ideal other than itself.
Corollary 4.1.1. Let A be a Noetherian ring. Then there is a one to one corre-
spondence between the family of irreducible components of Spec A and the family
of minimal prime ideals of A .
4.2. Reduced, irreducible and integral scheme.
Definition 4.2.1. Let ( X , O X ) be a scheme. Then it’s
(1) connected if X is connected.
(2) irreducibe if X is irreducible.
(3) reduced if for every open subset U of X , O X (U ) is reduced.
(4) integral if for every open subset U of X , O X (U ) is an integral domain.
(5) locally integral if O X ,p is an integral domain for every p ∈ X .
Proposition 4.2.1. A scheme ( X , O X ) is integral if and only if it’s irreducibe and
reduced.
Proof. See Proposition 1.3.6 in [Fu06]. □
Proposition 4.2.2. Let ( X , O X ) be an integral scheme and ξ be its generic point.
Then O X ,ξ is a field.
Proposition 4.2.3. A scheme ( X , O X ) is reduced if and only if O X ,p is reduced for
every p ∈ X .
Proof. Suppose ( X , O X ) is reduced and s ∈ O X ,p such that s n = 0. Then there exists
an open neighborhood U of p such that s n = 0 in O X (U ), and thus s = 0 since O X (U )
is reduced. Conversely, suppose O X ,p is reduced for every p ∈ X and s ∈ O X (U ) such
that s n = 0. Then by passing to the stalks one has s p = 0 for every p ∈ U , and thus
s = 0. □
Proposition 4.2.4. Let ( X , O X ) be a scheme such that X is a Noetherian topo-
logical space. Then ( X , O X ) is locally integral if and only if it’s reduced and its
irreducibe component are disjoint.
Proof. See Proposition 1.3.8 in [Fu06]. □
4.3. Affine criterion.
Definition 4.3.1. Let ( X , O X ) be a scheme. For any section f ∈ O X ( X ), X f is
defined to be the subset of X consisting of thoes p ∈ X such that the germ of f at
p is a unit in O X ,p .
Proposition 4.3.1. Let ( X , O X ) be a scheme.
(1) For every f ∈ O X ( X ), X f is open. It’s empty if and only if there exists an open
covering {U i } i∈ I of X such that each f |Ui is nilpotent.
 29

(2) For any f , g ∈ O X ( X ), we have X f ∩ X g = X f g .

(3) Let (φ, φ] ) : ( X , O X ) → (Y , O Y ) be a morphism of schemes and f ∈ O Y (Y ). Then
φ−1 (Y f ) = X φ] ( f ) .
(4) Suppose X can be covered by finitely many affine open subschemes {U i } i∈ I
such that U i ∩ U j can be covered by finitely many affine open subschemes for
all i, j ∈ I . Let A = O X . Then for any f ∈ A , we have O X ( X f ) = A f .
Proof. Here we only give the proof of (1), and for others one can refer to the Propo-
sition 1.3.9 in [Fu06]. Suppose {U i = Spec A i } i∈ I is an affine covering of X and
S
define f i = f |Ui . Then it’s clear that X f = i∈ I D ( f i ), and thus it’s open.
Now let’s prove the half part:
(a) Suppose there exists an open covering {U i } i∈ I such that each f i := f |Ui is nilpo-
tent. Then without lose of generality we may assume this open covering con-
sists of affine open subsets (by taking refinement when neccessary). By previ-
S
ous result one has X f = i∈ I D ( f i ) = ∅ since D ( f i ) = ∅ if f i is nilpotent.
( b) Conversely, if X f is empty, then there exists an affine open covering {U i =
Spec A i } i∈ I such that D ( f i ) = ∅ for each i ∈ I , and thus f i is nilpotent.
□
Proposition 4.3.2. A scheme ( X , O X ) is affine if and only if there exist finitely
many sections f 1 , . . . , f n ∈ O X ( X ) generating the unit ideal of O X ( X ) such that each
open subscheme ( X f i , O X | X f i ) is affine.
Proof. See Proposition 1.3.10 in [Fu06]. □
4.4. Noetherian scheme.
Definition 4.4.1 (Noetherian scheme). A scheme ( X , O X ) is called locally Noe-
therian if it can be covered by affine open subschemes {U i = Spec A } i∈ I such that
each A i is Noetherian, and it’s called Noetherian if it’s quasi-compact and locally
Noetherian.
Proposition 4.4.1. Let ( X , O X ) be a Noetherian scheme. Then every open subset
of X is quasi-compact.
Proof. If X is Noetherian, then it admits a finite affine open covering {U i = Spec A } i∈ I
such that each A i is Noetherian. In particular, X is a Noetherian topological space,
and thus every open subset of X is quasi-compact. □
Proposition 4.4.2. Let ( X , O X ) be a locally Noetherian scheme. Then for any
affine open subscheme U = Spec A of X , A is Noetherian. In particular, an affine
scheme (Spec A, O Spec A ) is locally Noetherian (and thus Noetherian) if and only if
A is Noetherian.
Proof. See Proposition 1.3.11 in [Fu06]. □
Proposition 4.4.3. If ( X , O X ) is a Noetherian scheme, then X is a Noetherian
topological space, but the converse is not true.
30

Proof. Recall that if a ring A is Noetherian, then Spec A is a Noetherian topolog-

ical space. Now since ( X , O X ) is a Noetherian scheme, then there exists a finite
affine open covering {U i = Spec A i }ni=1 such that each A i is a Noetherian ring, and
it’s clear that a finite union of Noetherian topological space is also Noetherian.
Conversely, consider the ring A = k[ x1 , x2 , . . . ]/( x12 , x22 , . . . ), where k is a field.
Since each variable xn is nilpotent, every prime must contain I = ( x1 , x2 , . . . ). But
A / I is just k, so I is already a maximal ideal, and thus I is the only prime ideal.
This shows Spec A has only one point, and it’s obviously Noetherian. But I is not
finitely generated, so A is not Noetherian. In particular, Spec A is not a Noether-
ian scheme. □
 31

5. P ROPERTIES OF MORPHISMS

5.1. Quasi-compact, affine, finite type and finite.

Definition 5.1.1. Let π : X → Y be a morphism of schemes. It’s called

(1) quasi-compact if there exists a covering of Y by affine open subschemes
{Vi } i∈ I such that each π−1 (Vi ) is quasi-compact.
(2) affine if there exists a covering of Y by affine open subschemes {Vi } i∈ I such
that each π−1 (Vi ) is affine.
(3) locally of finite type if if there exists a covering of Y by affine open sub-
schemes {Vi = Spec B i } i∈ I such that each π−1 (Vi ) can be covered by affine open
subschemes {U i j = Spec A i j } j∈ Ji for some finitely generated B i -algebra A i j .
(4) finite type if it’s quasi-compact and locally of finite type.
(5) finite if there exists a covering of Y by affine open subschemes {Vi = Spec B i } i∈ I
such that each π−1 (Vi ) = Spec A i for some finitely generated B i -module A i .

Definition 5.1.2 (affine-local). Let ( X , O X ) be a scheme and {Spec A i } i∈ I be an

open covering of X . A property P is called affine-local, if every Spec A i has prop-
erty P implies any affine open subset of X has property P .

Lemma 5.1.1. Suppose Spec A and Spec B are affine open subschemes of a scheme
X . Then Spec A ∩ Spec B is the union of open subsets that are simultaneously
distinguished open subschemes of Spec A and Spec B.

Proof given in [Vak17]. It suffices to show for any p ∈ Spec A ∩ Spec B, there exists
an open neighborhood of p in Spec A ∩ Spec B that is simultaneously distinguished
in both Spec A and Spec B. Let Spec A f be a distinguished open subset of Spec A
contained in Spec A ∩ Spec B and containing p and Spec B g be a distinguished open
subset of Spec B contained in Spec A f and containing p. Then g ∈ O X (Spec B) = B
restricts to an element g0 ∈ O X (Spec A f ) = A f . Note that
Spec B g = Spec A f \ {q ∈ Spec A f | g0 ∈ q} = (Spec A f ) g0 .
If g0 = g00 / f n in A f , where g00 ∈ A , then (Spec A f ) g0 = Spec A f g00 , and this completes
the proof. □
Lemma 5.1.2 (affine communication lemma). Let P be a property enjoyed by
some affine open subsets of a scheme X , such that
(1) if an affine open subset Spec A ,→ X has property P , then for any f ∈ A ,
Spec A f ,→ X does too.
(2) If ( f 1 , . . . , f n ) = A , and Spec A f i ,→ X has property P for each i , then so does
Spec A ,→ X .
Then the property P is affine-local.

Proof given in [Vak17]. Suppose {Spec A i } be an affine open covering of the scheme
X enjoying the property P . For any affine open subscheme Spec A of X , by Lemma
5.1.1 we may cover Spec A with a finite number of distinguished open subsets
32

Spec A g i , each of which is distinguished in some Spec A i . By (1) one has each
Spec A g i has property P , and thus Spec A has the property P by (2). □
Proposition 5.1.1. Let π : X → Y be a morphism of schemes.
(1) π is quasi-compact if and only if for every affine open subscheme V of Y ,
f −1 (V ) is quasi-compact.
(2) π is affine if and only if for every affine open subscheme V of Y , f −1 (V ) is
affine.
(3) π is of locally finite type if and only if for every affine open subscheme V =
Spec B of Y , f −1 (V ) can be covered by many affine open subschemes {U j =
Spec A j } j∈ J such that each A j is a finitely generated B-algebra.
(4) π is of finite type if and only if for every affine open subscheme V = Spec B
of Y , f −1 (V ) can be covered by finitely many affine open subschemes {U j =
Spec A j } j∈ J such that each A j is a finitely generated B-algebra.
(5) π is finite if and only if for every affine open subscheme V = Spec B of Y ,
f −1 (V ) = Spec A for some finitely generated B-module A .
Proof. Here we only give the proof of (1), and proofs for others are similar. By
affine communication lemma, it suffices to show the following two statements
(a) If Spec A ⊆ Y is an affine open subset such that π−1 (Spec A ) is quasi-compact,
then for any f ∈ A one has π−1 (Spec A f ) is quasi-compact.
( b) If ( f 1 , . . . , f n ) = A and π−1 (Spec A f i ) is quasi-compact for each i , then π−1 (Spec A )
is quasi-compact.
The statement ( b) is easy since the finite union of quasi-compact subset is still
quasi-compact, and now let’s prove statement (a). Suppose Spec A ⊆ Y is an affine
open subset such that π−1 (Spec A ) is quasi-compact. Then for any f ∈ A , one has
π−1 (Spec A f ) = (π−1 (Spec A ))π] ( f ) ,
where π] ( f ) ∈ O X (π−1 (Spec A )). On the other hand, since π−1 (Spec A ) is quasi-
compact, there exists a finite affine open covering U of it, and by (1) of Proposition
4.3.1 one has the intersection of (π−1 (Spec A ))π] ( f ) with any affine open subset in
U is still affine. This shows (π−1 (Spec A ))π] ( f ) is a quasi-compact since it’s a finite
union of affine open subsets. □
Corollary 5.1.1. Any affine morphism is quasi-compact.
Corollary 5.1.2. If the source of a morphism is a Noetherian scheme, then the
morphism is quasi-compact.
Proof. It follows from Proposition 4.4.1. □
Proposition 5.1.2. Suppose f : X → Y is a morphism of schemes. Then f is locally
of finite type if and only if for every affine open subscheme V = Spec B of Y and
every affine open subscheme U = Spec A of X such that f (U ) ⊆ V , the B-algebra A
is finitely generated.
Proof. See Exercise 7.7.1. □
 33

5.2. Birational morphism.

Proposition 5.2.1. Let X → S and Y → S be two morphisms and assume Y → S
is locally of finite type.
(1) Let f , g : X → Y be two morphisms making the following diagram commutes
X Y

S
] ]
Let x ∈ X such that f ( x) = g( x) = y and such that f x = g x . Then there exists an
open neighborhood U of x in X such that f |U = g|U .
(2) Suppose S is locally Noetherian. Let x ∈ X and y ∈ Y be two points such
that their images in S are the same point s ∈ S , and let ϕ : O Y ,y → O X ,x be
a homomorphism making the following diagram commutes
ϕ
O X ,x O Y ,y

O S,s
Then there exists an open neighborhood U of x in X and a morphism f : U → Y
]
such that f ( x) = y, f x = ϕ and the following diagram commutes
f
U Y

S
(3) Suppose S is locally Noetherian, X → S is also locally of finite type, and f : X →
Y is a morphism making the diagram in (1) commutes. Assume f ( x) = y and
]
f x : O Y ,y → O X ,x is an isomorphism. Then there exist open neighborhoods U of
x in X and V of y in Y such that f induces an isomorphism from U to V .
Proof. See Proposition 1.3.13 of [Fu06]. □
Definition 5.2.1 (dominant and birational morphism). Let X and Y be integral
schemes.
(1) A morphism f : X → Y is dominant if f ( X ) = Y .
]
(2) A dominant morphism f : X → Y is called birational if f ξ : O Y ,η → O X ,ξ is an
isomorphism, where ξ, η are generic points of X and Y respectively.
Corollary 5.2.1. Let S be a locally Noetherian scheme and let X and Y be two
integral schemes. Suppose we have the following commutative diagram
f
X Y

S
34

such that X → S and Y → S are locally of finite type and f is a birational mor-
phism. Then there exists non-empty open subsets U ⊆ X and V ⊆ Y such that f
induces an isomorphism from U to V .
5.3. Open immersion and closed immersion.
Definition 5.3.1 (open immersion). A morphism ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is called
an open immersion if it induces an isomorphism of ( Z, O Z ) with an open subscheme
of ( X , O X ).
Definition 5.3.2 (closed immersion). A morphism ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is called
a closed immersion if it induces a homeomorphism of Z with a closed subset of X ,
and f ] : O X → f ∗ O Z is surjective.
Definition 5.3.3 (immersion). A morphism Z → X is called an immersion if it can
be written as a composite Z → U → X such that U → X is an open immersion and
Z → U is a closed immersion.
Definition 5.3.4 (locally closed subset). A subset Z of X is called locally closed if
it’s the intersection of an open subset with a closed subset.
Remark 5.3.1. In other words, a subset Z is locally closed if and only if Z is open
in Z .
Proposition 5.3.1. Let ( f , f ] ) : ( Z, O Z ) → ( X , O X ) be a morphism of schemes.
(1) ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is an open immersion if and only if f induces a home-
]
omorphism of Z with an open subset of X and f p : O X , f ( p) → O Z,p is an isomor-
phism for every p ∈ Z .
(2) ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is an immersion if and only if f induces a homeo-
]
morphism of Z with a locally closed subset of X and f p : O X , f ( p) → O Z,p is an
epimorphism.
(3) The immersions are monomorphisms in the category of schemes. Moreover, the
composite of immersions is an immersion, so are open immersion and closed
immersion.
Proof. See Exercise 7.8.1. □
Proposition 5.3.2. Let A be a ring.
(1) For every ideal a of A , the morphism φ : Spec A /a → Spec A induced by the
canonical homomorphism ϕ : A → A /a is a closed immersion.
(2) Every closed immersion into Spec A is isomorphic to Spec A /a → Spec A for
some ideal a of A .
Proof. See Proposition 1.3.16 of [Fu06]. □
Corollary 5.3.1. A closed immersion is affine, and thus quasi-compact.
Proposition 5.3.3 (reduced closed subscheme structure). Let ( X , O X ) be a scheme
and Y be a closed subset of X . Then there exists a unique reduced scheme struc-
ture (Y , O Y ) on Y which makes Y a closed subscheme of X . If ( Z, O Z ) → ( X , O X ) is
 35

a closed immersion such that its image in X contains Y , then there exists a unique
morphism (Y , O Y ) → ( Z, O Z ) such that the following diagram commutes
Y X

Z
Proof. See Proposition 1.3.18 of [Fu06]. □
5.4. Fibred product. In this section S always is a scheme.
Definition 5.4.1.
(1) An S -scheme is a scheme X together with a morphism X → S .
(2) An S -morphism from an S -scheme X to an S -scheme Y is a morphism X → Y
such that the diagram
X Y

S
commutes.
Remark 5.4.1. For any scheme X , there is a unique morphism X → Spec Z, so the
category of schemes coincides with the category of Spec Z-schemes.
Definition 5.4.2 (fibred product). Let X and Y be S -schemes. The product in the
category of S -schemes is called the fibred product of X and Y over S , which is a
S -scheme denoted by X ×S Y .
Proposition 5.4.1. For S -schemes X and Y , their fibred product over S exists
and unique up to unique isomorphism.
Proof. See Proposition 1.3.20 of [Fu06]. □
Proposition 5.4.2. Let f : X → Y be a morphism of schemes and y be a point in
Y with residue field k( y). The projection
X ×Y Spec k( y) → X
induces a homeomorphism from X ×Y Spec k( y) with f −1 ( y) on the underlying topo-
logical spaces.
Proof. See Proposition 1.3.21 of [Fu06]. □
Proposition 5.4.3. Let
X Y

S
36

be morphisms of schemes. Then the diagram

X ×T Y X ×S Y

T T ×S T
is Cartesian, where the top horizontal arrow is the unique morphism whose com-
posites with the two projections of X ×S Y to its factors are the two projections of
X ×T Y to its factors, and the right vertical arrow is the unique morphism whose
compositions with the two projections of T ×S T to its factors are the morphisms
X → T and Y → T respectively.
Proof. See Exercise 7.9.4. □

5.4.1. Base change.

Definition 5.4.3 (stable under base change). Let f : X → Y be a morphism of
schemes. A property P of f is called stable under base change, if for any morphism
Y 0 → Y , the base change f 0 : X ×Y Y 0 → Y 0 also has property P .
Proposition 5.4.4.
(1) The quasi-compact morphism is stable under base change.
(2) The finite type morphism is stable under base change.
(3) The finite morphism is stable under base change.
(4) The immersion (closed immersion, open immersion) is stable under base.
Proof. Here we only give the proof of (1), and the proofs for others are similar.
Let f : X → Y be a quasi-compact morphism and g : Y 0 → Y be a morphism.
For any affine open subset U = Spec A of Y , we choose an affine open covering
{Vi = Spec B i } of g−1 (U ), where B i ’s are A -algebras. Then
f 0−1 (Vi ) = X ×Y Vi = f −1 (U ) ⊗U Vi .
On the other hand, since f is quasi-compact, there exists a finite affine open cov-
ering {W j = Spec C j }nj=1 of f −1 (U ), where C j ’s are A -algebras. Then
[
n
f 0−1 (Vi ) = f −1 (U ) ⊗U Vi = Spec C j ⊗ A B i .
j =1

This shows f 0−1 (Vi ) is quasi-compact. By varing U and thus g−1 (U ), we obtain an
open covering {Vi } i∈ I of Y 0 such that f 0−1 (Vi ) is quasi-compact for each i ∈ I . This
shows f 0 is quasi-compact. □

5.5. Separated morphisms.

Definition 5.5.1 (diagonal morphism). Let f : X → Y be a morphism of schemes.
The diagonal morphism ∆ X /Y : X → X ×Y X to be the unique morphism satisfying
p ◦ ∆ X /Y = q ◦ ∆ X /Y = id X .
 37

Definition 5.5.2 (separated morphism). Let f : X → Y be a morphism of schemes.

It’s called separated if ∆ X /Y is a closed immersion.
Definition 5.5.3 (separated scheme). A scheme X is called separated if the canon-
ical morphism X → Spec Z is separated.
Definition 5.5.4 (quasi-separated).
(1) A morphism f : X → Y of schemes is called quasi-separated if the diagonal
morphism is quasi-compact.
(2) A scheme X is quasi-separated if the canonical morphism X → Spec Z is quasi-
separated.
Proposition 5.5.1. Let f : Spec B → Spec A be a morphism of affine schemes.
Then f is separated.
Proof. See Proposition 1.3.22 of [Fu06]. □
Proposition 5.5.2. Let f : X → Y be a morphism of schemes.
(1) The diagonal morphism ∆ : X → X ×Y X is an immersion.
(2) f : X → Y is separated if and only if ∆ X /Y ( X ) is a closed subset of X ×Y X .
Proof. See Proposition 1.3.23 of [Fu06]. □
Proposition 5.5.3.
(1) A morphism f : X → Y of schemes is separated if and only if there exists an
open covering {Vi } i∈ I of Y such that f −1 (Vi ) → Vi is separated.
(2) The immersion is separated.
(3) The composite of two separated morphisms is separated.
(4) The separated morphism is stable under base change.
(5) Let f : X → Y and g : Y → Z be morphisms of schemes. If g ◦ f is separated,
then f is separated.
Proof. Here we only give the proof of (1), and for others one can refer to the Propo-
sition 1.3.26 in [Fu06].
Let {Vi } i∈ I be an open covering of Y and U i = f −1 (Vi ) such that U i → Vi is sep-
arated. By (2) of Proposition 5.5.2 it suffices to show ∆ X /Y ( X ) is a closed subset
of X ×Y X , and it suffices to find an open covering U such that the preimage of
each open subset in U under ∆ X /Y is closed. Suppose g : X ×Y X → Y is the natural
morphism. Then { g−1 (Vi )} gives an open covering of X ×Y X and by the property
of fibred product one has g−1 (Vi ) = U i ×Vi U i . It’s clear ∆−X1/Y (U i ×Vi U i ) = U i , and
U i → U i ×Vi U i is closed since U i → Vi is separated.
Conversely, suppose f : X → Y is separated. Then for any open covering {Vi } i∈ I
of Y , f −1 (Vi ) → Vi is the base change of f under Vi ,→ V , and thus it’s separated
since the separated morphism is stable under base change. □
Proposition 5.5.4. Let X be a S -scheme, where S is an affine scheme.
(1) If X is separated over S , then the intersection of any two affine open sub-
schemes is affine, and this fails in general if X is not separated.
38

(2) If X is quasi-separated over S , then the intersection of any two affine open
subschemes is a finite union of affine subschemes,

Proof. For (1). Let U = Spec A and V = Spec B be two affine open subschemes of
X . Then by the property of fibred product one has U ×S V is affine. On the other
hand, by Proposition 5.4.3 one has the following diagram is Cartesian

U ×X V U ×S V

X X ×S X

Thus U ∩ V = U × X V → U ×S V is a closed immersion since X → X ×S X is a

closed immersion, and the closed immersion is stable under base change. As a
consequence, U ∩ V is affine since every closed immersion into an affine scheme is
affine, and the proof of (2) is similar.
In general, consider the affine plane with double origin, that is, two copies U
and V of affine plane A2k over a field k glued along the identity morphism on
the open subscheme A2k \{0}. Then the intersection U ∩ V ∼= A2k \{0}, which is not
affine. □

5.6. Proper and projective morphisms.

5.6.1. Proper morphisms.

Definition 5.6.1 (universally closed). A morphism f : X → Y of schemes is called

universally closed, if for any morphism Y 0 → Y , the base change f 0 : X ×Y Y 0 → Y 0
of f is a closed map on the underlying topological spaces.

Definition 5.6.2 (proper). A morphism f : X → Y of schemes is proper if f is finite

type, separated and universally closed.

Proposition 5.6.1.
(1) A morphism f : X → Y of schemes is proper if and only if there exists an open
covering {Vi } i∈ I of Y such that f −1 (Vi ) → Vi is proper.
(2) The closed immersion is proper.
(3) The composite of two proper morphisms is proper.
(4) The proper morphism is stable under base change.
(5) Let f : X → Y and g : Y → Z be morphisms of schemes. If g ◦ f is proper and g
is separated, then f is proper.

Proposition 5.6.2. The finite morphism is proper.

5.6.2. Projective morphisms.

n
Definition 5.6.3 (projective space). The projective space PZ is defined by Proj Z[ x0 , . . . , xn ],
n
and for any scheme Y , the projective space over Y is the Y -scheme PYn := PZ ×Z Y .
 39

Definition 5.6.4 (projective morphisms). A morphism f : X → Y of schemes is

projective if f can factorized as a composite
X → PYn → Y
such that X → PYn is a closed immersion and PYn → Y is the projection. It’s called
quasi-projective if it can be factorized as above with X → PYn being an immersion.
Definition 5.6.5 (projective). A S -scheme X is called projective over S , if X → S
is a projective morphism.
Lemma 5.6.1. For any finitely generated A -module, one has
{p ∈ Spec A | Mp = 0} = Spec A \ V (ann A ( M )).
Proposition 5.6.3. The projective morphism is proper.
Proposition 5.6.4.
(1) The closed immersion is projective.
(2) The composite of projective morphisms is projective.
(3) The projective morphism is stable under base change.
(4) Let f : X → Y and f 0 : X 0 → Y 0 be projective S -morphisms between S -schemes.
Then f × f 0 : X ×S X 0 → Y ×S Y 0 is projective.
(5) Let f : X → Y and g : Y → Z be morphism of schemes. If g ◦ f is projective and
g is separated, then f is projective.
Proposition 5.6.5 (Segre embedding). There exists a closed immersion
m
PS n
×S PS → P(Sm+1)(n+1)−1 ,
which is an S -morphism.
40

6. C OHERENT SHEAVES

6.1. O X -modules. Let ( X , O X ) be a ringed space.

Definition 6.1.1 (O X -module). A sheaf of O X -module (or O X -module) is a sheaf F
such that
(1) every open subset U ⊆ X , F (U ) is an O X (U )-module;
(2) for every inclusion of open subsets V ⊆ U , the restriction F (U ) → F (V ) is
compatible with the module structure via the ring homomorphism O X (U ) →
O X (V ).
Definition 6.1.2 (morphism of O X -module). Let F and G be O X -modules. A mor-
phism of O X -modules is a morphism of sheaves ϕ : F → G such that for each open
subset U ⊆ X , ϕ(U ) : F (U ) → G (U ) is a homomorphism of O X (U )-modules.
Notation 6.1.1. The set of morphisms between O X -modules F and G is denoted
by HomO X (F , G ), which is a O X ( X )-module.
Example 6.1.1. Let F and G be O X -modules.
(1) The sheaf hom H O X (F , G ) is the O X -module
om
U 7→ HomO X |U (F |U , G |U ).
(2) The tensor F ⊗O X G is the O X -module associated to the presheaf
U 7→ F (U ) ⊗O X (U ) G (U ).
Remark 6.1.1. For any p ∈ X , one has
(F ⊗ G ) p = F p ⊗ G p .
Example 6.1.2. Let {F i } i∈ I be a family of O X -modules.
(1) The direct sum in the category of O X -module is the sheaf associated to the
presheaf sheaf M
U 7→ F i (U ).
i∈ I
(2) The direct product in the category of O X -module is the sheaf associated to the
presheaf sheaf Y
U 7→ F i (U ).
i∈ I

Example 6.1.3. Let I be a direct set.

(1) For a direct system (F i , ϕ i j ) of O X -modules, its direct limit in the category of
O X -modules is the sheaf associated to the presheaf
U → lim F i (U ).
−−→
i
(2) For a inverse system (F i , ϕ i j ) of O X -modules, its direct limit in the category of
O X -modules is the sheaf associated to the presheaf
U → lim F i (U ).
←−−
i
 41

Definition 6.1.3 (direct image and inverse image). Let ( f , f ] ) : ( X , O X ) → (Y , O Y )

be a morphism of ringed spaces.
(1) Let F be an O X -module. The direct image f ∗ F is an f ∗ O X -module, and it
becomes a O Y -module via the morphism f ] : O Y → f ∗ O X .
(2) Let G be an O Y -module. The inverse image of G is defined to be O X -module
f ∗ G = O X ⊗ f −1 OY f −1 G .
Definition 6.1.4 (finite representation). An O X -module F is called of finite pre-
sentation if there exists an open covering {U i } i∈ I of X such that on each U i , there
is an exact sequence of the form
⊕m i ⊕n i
OU → OU → F |Ui → 0.
i i

Definition 6.1.5 (finite type). An O X -module F is called of finite type if there ex-
ists an open covering {U i } i∈ I of X such that on each U i , there is an exact sequence
of the form
⊕n
OU i → F |Ui → 0.
i

Proposition 6.1.1. Let ( X , O X ) be a ringed space and F , G be O X -modules. If F

is of finite representation, then for every p ∈ X , one has
(H O (F , G )) p =
om ∼ HomO (F p , G p ).
X X ,p

Proof. See Proposition 1.4.1 in [Fu06]. □

6.2. Coherent sheaves.
6.2.1. Sheaf associated to a module over affine space.
Definition 6.2.1. Let A be a ring and M be a A -module. The O Spec A -module
associated to M , denoted by M ∼ is defined as follows: For every open subset
`
U ⊆ Spec A , M ∼ (U ) consists of those mappings s : U → p∈Spec A Mp satisfying the
following two conditions
(1) For every p ∈ U , one has s(p) ∈ Mp .
(2) For every p ∈ U , there exists an open neighborhood Up of p, m ∈ M and f ∈ A ,
such that for every q ∈ Up , one has f 6∈ q and s(q) = m/ f in Mq .
For every inclusion of open subsets V ⊆ U , M ∼ (U ) → M ∼ (V ) is defined to be the
restriction of mappings.
Example 6.2.1. Let A be a ring. Then A ∼ = O Spec A .
Proposition 6.2.1. Let A be a ring and M be a A -module.
(1) ( M ∼ )p ∼
= Mp for every p ∈ Spec A .
(2) ( M ∼ )(D ( f )) ∼
= M f for every f ∈ A .
(3) A sequence of A -modules
M 0 → M → M 00
is exact if and only if the sequence of O Spec A -modules
M 0∼ → M ∼ → M 00∼
42

is exact.
(4) For any A -modules M and N , one has
HomOSpec A ( M ∼ , N ∼ ) ∼
= Hom A ( M, N )
∼ ∼∼
M ⊗O N = ( M ⊗ A N )∼ .
Spec A

If M is an A -module with finite presentation, then

H O om (M∼, N ∼) ∼
Spec A = (Hom A ( M, N ))∼ .
(5) For a family { M i } i∈ I of A -modules, one has
M ∼ M
M = ∼ ( M i )∼ .
i
i i
(6) For a direct system ( M i , ϕ i j ) i∈ I of A -modules, one has
lim M ∼ ∼
= (lim M i )∼ .
−−→ i −−→
i i

Proof. Here we only give the proofs of (3), (5) and (6), since (1),(2) can be proved
by the similar argument as shown in structure sheaf case, and the proof of (4) is
shown in Proposition 1.4.2 in [Fu06].
For (3). Note that a sequence of A -modules M 0 → M → M 00 is exact if and only if
Mp0 → Mp → Mp00
is exact for every p ∈ Spec A , and thus it’s equivalent to the sequence of sheaves
M 0∼ → M ∼ → M 00∼ is exact by (1).
For (5). It suffices to note that for each p ∈ Spec A , one has
M M ∼ M M
( M ∼ )p ∼
i = ( M )p ∼
= M i,p ∼
i = ( M i )∼ , p
i i
since taking stalk commutes with direct sum, and by the same argument one can
prove (6), since both taking stalk and direct limit are colimits, and colimit com-
mutes colimit. □
Proposition 6.2.2. Let ϕ : A → B be a homomorphism of rings and f : Spec B →
Spec A be the corresponding morphism.
(1) For every B-module N , one has f ∗ N ∼ ∼ = N ∼ , where on the right hand N is
regarded as an A -module.
(2) For every A -module M , one has f ∗ M ∼ ∼
= ( B ⊗ A M )∼ .
Proof. See Proposition 1.4.3 in [Fu06]. □
6.2.2. Coherent sheaves on Noetherian scheme.
Definition 6.2.2 (quasi-coherent). Let ( X , O X ) be a scheme. A sheaf of O X -module
F is called quasi-coherent if X can be covered with affine open subschemes U i =
Spec A i such that F |Ui ∼
= M∼
i
for some A i -module M i .
Definition 6.2.3 (coherent). Let ( X , O X ) be a Noetherian scheme. A sheaf of O X -
module F is called quasi-coherent if X can be covered with affine open subschemes
U i = Spec A i such that F |Ui ∼
= M∼
i
for some finitely generated A i -module M i .
 43

Proposition 6.2.3. Let X be a scheme and F be a quasi-coherent sheaf on X .

Then for any affine open subscheme U = Spec A of X , there exists an A -module
M such that F |U = M ∼ . If X is Noetherian and F is coherent, then M is finitely
generated.
Proof. See Proposition 1.4.5 in [Fu06]. □
Corollary 6.2.1. Let X be a scheme and F , G be quasi-coherent O X -modules.
(1) The tensor product F ⊗O X G is quasi-coherent, and if F is of finite presentation,
then H O X (F , G ) is quasi-coherent.
om
(2) Let ϕ : F → G be a morphism. Then ker ϕ, coker ϕ and im ϕ are quasi-coherent.
If X is Noetherian, then the same statements hold for coherent sheaf.
Proof. See Corollary 1.4.6 in [Fu06]. □
Proposition 6.2.4. Let f : X → Y be a morphism of schemes.
(1) If G is a quasi-coherent O Y -module, then f ∗ G is a quasi-coherent O X -module.
(2) If X and Y are Noetherian and G is coherent, then f ∗ G is coherent.
(3) If f is quasi-compact and quasi-separated, and F is a quasi-coherent O X -
module, then f ∗ F is a quasi-coherent O Y -module.
Proof. See Proposition 1.4.9 in [Fu06]. □
6.2.3. Coherent sheaves on general scheme.
Definition 6.2.4 (coherent sheaf on ringed space). Let ( X , O X ) be a ringed space.
An O X -module F is called coherent if F is of finite type and for every open subset
⊕n
U of X and every homomorphism u : OU → F |U , the kernel of u is of finite type.
Proposition 6.2.5. Let ( X , O X ) be a ringed space.
(1) Suppose O X is coherent. Prove that an O X -module F is coherent if and only if
F is of finite presentation.
(2) Prove that Definition 6.2.4 of coherence coincides with the one in Definition
6.2.3 for Noetherian schemes.
Proof. See Exercise 7.11.2. □
Lemma 6.2.1 (Oka). Let X be a complex manifold and O X be the sheaf of holo-
morphic functions (and thus ( X , O X ) gives a locally ringed space). Then O X is
coherent.
6.3. Sheaf of ideals.
Definition 6.3.1 (sheaf of ideals). Let ( X , O X ) be a ringed space. A sheaf of ideals
I of O X is an O X -submodule of O X .
Definition 6.3.2 (support). Let ( X , O X ) be a ringed space and F be an O X -module.
The support of F is defined to be the set
supp F = { p ∈ X | F p 6= 0}.
44

Proposition 6.3.1. Let X be a scheme and F be a quasi-coherent O X − module

of finite type. Then supp F is a closed subset of X . Suppose furthermore that X
is Noetherian. Then for every coherent sheaf of ideals I of O X such that supp
F ⊆ supp(O X /I ), there exists n ∈ N such that I n F = 0.

Proof. See Proposition 1.4.14 of [Fu06]. □

Definition 6.3.3 (scheme theoretic image). Let f : X → Y be a morphism between
schemes. A closed immersion i : Z → Y such that f = i ◦ g for a morphism g : X → Z
is called a scheme theoretic image, if i statisfies the following universal property:
For every factorization f = i 0 ◦ g0 such that i 0 : Z 0 → Y is a closed immersion and
g0 : Z 0 → Y is a morphism, there exists a unique morphism j : Z → Z 0 such that
i 0 ◦ j = i and j ◦ g = g0 .
f
X Y
g i

Z
g0 i0

Z0

Proposition 6.3.2. Let f : X → Y be a quasi-compact and quasi-separated mor-

phism between schemes. Then f ( X ) is its scheme theoretic image.

Corollary 6.3.1. Let f : Z → X be an immersion. If f is quasi-compact, then f

can be factorized as Z → F → X such that the first arrow is an open immersion
and the second arrow is a closed immersion.

Proof. See Corollary 1.4.17 of [Fu06]. □

Proposition 6.3.3 (Chow’s lemma). Let S be a Noetherian scheme and f : X → S
be a proper morphism. Then there exists a projective morphism g : X 0 → X such
that f ◦ g is projective, and g induces an isomorphism g−1 (U ) → U for some dense
open subset U of X .
g
X0 X
f

Proof. See Proposition 1.4.18 of [Fu06]. □

L∞
6.4. Coherent sheaves on Proj. Let S = d =0 S d be a graded ring and M is a
graded S -module.

Definition 6.4.1 (shifted graded module). M ( n) is the graded S -module whose

underlying S -module is M and whose grading is defined by ( M ( n))d = M d +n .
 45

Definition 6.4.2. For every homogeneous prime ideal p of S , define M(p) to be the
S (p) -module
m
M(p) = { | m ∈ M and t ∈ S \ p are homogeneous of the same degree}.
t
Definition 6.4.3. For every homogeneous element f in S , define M( f ) to be the
S ( f ) -module
m
M( f ) = { | k ∈ N and m ∈ M is homogeneous of degree k deg f }.
fk
Definition 6.4.4. The O Proj S -module associated to M , denoted by M ∼ is defined
as follows: For every open subset U ⊆ Spec A, M ∼ (U ) consists of those mappings
s : U → q p∈Spec A M(p) satisfying the following two conditions
(1) For every p ∈ U , one has s(p) ∈ M(p) .
(2) For every p ∈ U , there exists an open neighborhood Up of p, m ∈ M and f ∈ S
of the same degree, such that for every q ∈ Up , one has f ∉ q and s(q) = m/ f in
M (q ) .
For every inclusion of open subsets V ⊆ U, M ∼ (U ) → M ∼ (V ) is defined to be the
restriction of mappings.

Proposition 6.4.1.
(1) For every p ∈ Proj S , one has ( M ∼ )p ∼
= Mp .
(2) For every homogeneous element f in S of positive degree, one has M ∼ |D ( f ) ∼
=
( M ( f ) )∼ .
(3) M ∼ is quasi-coherent. If S is Noetherian and M is finitely generated S -module,
then Proj S is Noetherian and M ∼ is coherent.

Definition 6.4.5 (twisting). For any integer n, O Proj S ( n) is defined by S ( n)∼ and
O Proj S (1) is called the twisting sheaf. Moreover, for any O Proj S − module F , define
F ( n) = F ⊗OProj S O Proj S ( n).

Proposition 6.4.2. Let S be a graded ring which is generated by S 1 as a S 0 -

algebra.
(1) O Proj S ( n) is an invertible sheaf for every n.
(2) Let M and N be S -modules. Then
( M ⊗S N )∼ ∼
= M ∼ ⊗OProjS N ∼ .

Proposition 6.4.3. Let S be a graded rings which are generated by S 1 as S 0 -

algebra and so is T . Let ϕ : T → S be a homomorphism with the property ϕ(S d ) ⊆
T d for all d , and
U = {q ∈ Proj T | S + ⊊ ϕ−1 (q)}.
Then U is an open subset of Proj T and ϕ induces a canonical morphism f : U →
Proj S .
46

(1) For every graded S -module M , one has

f ∗ M∼ ∼
= (T ⊗S M )∼ |U .
In particular, one has
f ∗ O Proj S ( n) ∼
= O Proj T ( n)|U .
(2) For every graded T -module N , one has
f ∗ ( N ∼ |U ) ∼
= N ∼,
where on the right hand side, N is regarded as an S -module.
6.4.1. Graded S -module associated to O Proj S -module. Let S be a graded ring which
is generated by S 1 as S 0 -algebra and X = Proj S .
Definition 6.4.6 (graded S -module associated to O Proj S -module). For any O X -
module F , the graded S -module associated to F is defined by
M∞
Γ∗ (F ) = F ( n)( X ),
n=−∞
where the S acts on Γ∗ (F ) by the following composite of canonical homomorphisms
S d ⊗ F ( n)( X ) → O Proj S ( d )( X ) ⊗ F ( n)( X ) → F ( d + n)( X ).
Proposition 6.4.4. Let F be a quasi-coherent O Proj S -module. Then Γ∗ (F )∼ ∼
= F.
Corollary 6.4.1.
(1) Let X be a closed subscheme of Pm A
= Proj A [ x0 , . . . , xm ], where A is a ring.
Then X is isomorphic to Proj A [ x0 , . . . , xm ]/a for some homogeneous ideal a of
A [ x0 , . . . , xm ].
(2) A scheme X over Spec A is projective if and only if it’s isomorphic to Proj S
for some graded ring S such that S 0 = A and S is finitely generated by S 1 as
S 0 -algebra.
6.4.2. Ampleness and globally generated.
Definition 6.4.7 (very ample). Let X be a S -scheme. An invertible O X -module L
is called very ample over S if there exists an immersion i : X → PSm which is an
S -morphism such that
L∼ = i ∗ (O PSm (1)).
Definition 6.4.8 (generated by global sections). An O X -module F is generated by
global sections if there exists a family of sections s i ∈ F ( X ) such that for every
p ∈ X , F p is generated by the germs ( s i ) p as an O X ,p -module.
Theorem 6.4.1 (Serre). Let X be a scheme proper over Spec A for a Noetherian
ring A , let O X (1) be an invertible O X -module very ample over Spec A , and let F
be a coherent O X -module. There exists an integer N such that for every n ≥ N , the
O X -module F ( n) = F ⊗ O X O X (1)⊗n is generated by finitely many global sections.
 47

Part 3. Homework
7. H OMEWORK
7.1. Homework-1.
Exercise 7.1. 1. A filtered abelian group is a pair ( A, F • A ) such that A is an
abelian group and
· · · ⊃ F i A ⊃ F i+1 A ⊃ · · ·
is a decreasing family of subgroups of A with indices i ∈ Z. A homomorphism
f : ( A, F • A ) → (B, F • B) of filtered abelian groups is a homomorphism f : A → B of
abelian groups such that f (F i A ) ⊂ F i B for all i ∈ Z.
(1) Prove that filtered abelian groups form an additive category with zero objects
and every morphism has kernel and cokernel.
(2) Given an example of a morphism f such that the canonical morphism coim f →
im f is not an isomorphism.
Proof. For (1). Suppose ( A, F • A ) and (B, F • B) are filtered abelian groups. The
direct product of ( A, F • A ) and (B, F • B) is given by ( A ⊕, F • ( A ⊕ B)), where the fil-
tration of A ⊕ B is given by F i ( A ⊕ B) = F i A ⊕ F i B, and it’s clear morphisms between
( A, F • A ) and (B, F • B) form an abelian group such that the composition is bilinear.
This shows the category of filtered abelian groups is additive, and the zero object
in this category is zero group with trivial filtration.
Suppose f : ( A, F • A ) → (B, F • B) is a morphism between filtered abelian groups.
Since f is also a group homomorphism between abelian groups, it has kernel
and cokernel in the category of abelian groups. More precisely, ker f ⊂ A and
coker f = B/ im f . Then the filtrations on A and B induce filtrations on ker f and
coker f respectively, and thus it gives kernel and cokernel in the category of fil-
tered abelian groups.
For (2). Suppose A = Z ⊕ Z with filtration Z ⊕ Z ⊃ Z ⊃ {0} and B = Z ⊕ Z ⊕ Z with
filtration Z ⊕ Z ⊕ Z ⊃ Z ⊕ Z ⊃ {0}. For homomorphism given by
A→B
(a, b) 7→ (a, b, 0),
the coimage is exactly A with filtration Z ⊕ Z ⊃ Z ⊃ {0}, but the image is Z ⊕ Z with
filtration Z ⊕ Z ⊃ Z ⊕ Z ⊃ {0}. □
Exercise 7.2. Let 0 → A → B → C → 0 be a sequence of morphisms in an abelian
category. Prove the following statements are equivalent:
(1) The sequence is a short exact sequence.
(2) B → C is an epimorphism and A → B is its kernel.
(3) A → B is a monomorphism and B → C is its cokernel.
Proof. Firstly let’s show the following lemma:
v
Lemma 7.1.1. Suppose B → C → 0 is a sequence of morphisms in an abelian
category. Then the following statements are equivalent:
48

(a) B → C → 0 is exact.
( b) the cokernel of v is C → 0.
( c) v is an epimorphism.
Proof.
(a) to ( b): If B → C → 0 is exact, then coim v = im v is the kernel of C → 0, that is
the im v = C → C . On the other hand, im v is the kernel of cokernel v. Thus the
cokernel of v is C → 0.
( b) to (a): If the cokernel of v is C → 0, then coim v = im v = ker(coker v) = ker{C →
0}, that is B → C → 0 is exact.
( b) to ( c): If the cokernel of v is C → 0 and α, β : C → D are morphisms such that
α ◦ v = β ◦ v. Then (α − β) ◦ v = 0, and thus by universal property of cokernel there
exists the following commutative diagram
v
B C 0
α−β

D
This shows α = β, that is, v is an epimorphism.
( c) to ( b): If v is an epimorphism and f : C → D is a morphism such that f ◦ v = 0,
then f = 0 since v is an epimorphism, and thus every morphism f such that
f ◦ v = 0 factors through C → 0, that is, the cokernel of v is C → 0.
□
Remark 7.1.1. By the same argument one can see a sequence of morphisms 0 →
u
A → B in abelian category is exact if and only if u is a monomorphism, also if and
only if 0 → A is the kernel of u.
u v
Now suppose 0 → A → B → C → 0 is an exact sequence in abelian category. Then
we claim u is the kernel of v: Since v ◦ u = 0, by the universal property of kernel
there exists the following diagram
u
0 A B
u

coim{0 → A } ker v
Note that u is an epimorphism, since A → coim{0 → A } is an epimorphism and
coim{0 → A } → ker v is an isomorphism. Moreover, u is a monomorphism since u
is a monomorphism: If α, β : D → A such that u ◦ α = u ◦ β, then we compose them
with ker v → B, one has u ◦ α = u ◦ β, and thus α = β. Then u is both monomorphism
and epimorphism, and since the category is abelian, one has u is an isomorphism,
and thus u is the kernel of v. By the same argument, it’s easy to see v is the
cokernel of u.
In a summary, above arguments show that (1) implies (2) and (3). To see (2)
implies (1), it suffices to show 0 → A → B → C is exact, since v : B → C is epimor-
phism already implies B → C → 0 is exact. Firstly, since u is the kernel of v, then
 49

it’s monomorphism, and thus 0 → A → B is exact. By previous lemma one has

0 → A is the kernel of u, and thus coim u = A → A . On the other hand, kernel of v
is u. This shows the coimage of u is exactly the kernel of v, that is A → B → C is
exact.
u v
0 A B C
∼
= u

coim u = A ker v = A
□
Exercise 7.3. Let A and B be objects in an abelian category. Prove that the
canonical sequence
i1 p2
0→ A → A⊕B → B →0
is exact.
Proof. By Exercise 7.2 it suffices to show i 1 is a monomorphism and cokernel of i 1
is p 2 . By definition there exists p 1 : A ⊕ B → A such that p 1 ◦ i 1 = id A and i 2 : B →
A ⊕B such that p 2 ◦ i 2 = idB . Moreover, p 2 ◦ i 1 = p 1 ◦ i 2 = 0 and i 1 ◦ p 1 + i 2 ◦ p 2 = id A ⊕B .
(1) Suppose α, β : C → A are morphisms such that i 1 ◦ α = i 1 ◦ β. Then p 1 ◦ i 1 ◦ α =
p 1 ◦ i 1 ◦ β implies α = β, and thus i 1 is a monomorphism.
(2) Suppose α : C → A ⊕ B is a morphism such that p 2 ◦ α = 0. Then
i 1 ◦ p 1 ◦ α = ( i 1 ◦ p 1 + i 2 ◦ p 2 ) ◦ α = α.
Thus we have the following commutative diagram
i1 p2
0 A A⊕B B 0
p 1 ◦α α

C
This shows i 1 : A → A ⊕ B satisfies the universal property of kernel.
□
Exercise 7.4. Let I be a category whose objects form a set, and let F be a covariant
functor from I to the category of Abelian groups. For each i ∈ I , let k i : F ( i ) →
L L
i ∈ I F ( i ) be the canonical monomorphism. Let H be the subgroup of i∈ I F ( i )
generated by
k i ( x i ) − k j (F ( i → j )( x i )),
where i → j goes over all morphisms in I , and x i goes over all elements F ( i ). Set
à !
M
lim F ( i ) = F ( i ) / H.
−−→ i∈ I
i∈ I
L
Let ϕ i : F ( i ) → lim F ( i ) be the composite of k i with the projection i∈ I F ( i ) →
L −−
→ i ∈ I
( i∈ I F ( i )) / H . Then we have ϕ j ◦ F ( i → j ) = ϕ i for every morphism i → j in I . If
A is an abelian group and ψ i : F ( i ) → A ( i ∈ I ) is a family of homomorphisms such
50

that ψ j ◦ F ( i → j ) = ψ i for all morphisms i → j in I , then there exists one and only
one homomorphism ψ : lim F ( i ) → A such that ψ ◦ ϕ i = ψ i for all i .
−−→ i∈ I
Proof. Firstly let’s show the existence: Note that by universal property of di-
L
rect sum, there exists a morphism ϕ : i F ( i ) → A , such that ψ i = ϕ ◦ k i , where
L
k i : F ( i ) → i F ( i ) is canonical inclusion. Moreover, for any element k i ( x i )− k j (F ( i →
j )( x i )) ∈ H , one has
ϕ( k i ( x i ) − k j (F ( i → j )( x i ))) = ψ i ( x i ) − ψ j ◦ F ( i → j )( x i ) = 0.
This shows H ⊆ ker ϕ, and thus we obtain a morphism ψ : lim F ( i ) → A induced
−−→ i∈ I
by ϕ, and it’s clear ψ i = ψ ◦ ϕ i .

F ( i)
ψi

L ϕ
ϕi i F ( i) A
ψ

lim F ( i )
−−→ i∈ I
Before we begin to prove the uniqueness, we claim any element of lim F ( i ) can
−−→ i∈ I
be written in the form ϕ i ( x i ) for some i ∈ I and some x i ∈ F ( i ): For any element
L
x ∈ lim F ( i ) = i∈ I F ( i )/ H , we write it as
−−→ i∈ I
Xn
x= ϕ i ( x j ), x j ∈ F ( j ).
j =1

It suffices to check the case of n = 2: Since I is a directed set, there exists k ∈ I

such that k ≥ 1, k ≥ 2. Then
ϕ1 ( x1 ) + ϕ2 ( x2 ) = ϕk ◦ F (1 → k)( x1 ) + ϕk ◦ F (2 → k)( x2 ).
Then x can be written as ϕk (F (1 → k)( x1 ) + F (1 → k)( x2 )) as desired.
Let’s show the uniqueness: If ψ0 : lim F ( i ) → A is another morphism such
−−→ i∈ I
that ψ i = ψ0 ◦ ϕ i for all i ∈ I . By above claim, we know each element can be written
as ϕ i ( x i ) for x i ∈ F ( i ). So it suffices to check ψ(ϕ i ( x i )) = ψ0 (ϕ i ( x i )), which is clear
ψ(ϕ i ( x i )) = ψ i ( x i ) = ψ0 (ϕ i ( x i )).
□
Exercise 7.5. Let ( A i , ϕ ji ) i∈N be an inverse system of abelian groups over the
direct set (N, ≤) of natural numbers. Consider the homomorphism
Y Y
f: Ai → A i , (a i ) 7→ (a i − ϕ i+1,i (a i+1 )).
i ∈N i ∈N

Define lim1 A i = coker f . Let u : ( A 0i , ϕ ji ) i∈N → ( A i , ϕ ji ) i∈N and v : ( A i , ϕ ji ) i∈N →

←−− i
( A 00i , ϕ ji ) i∈N be morphisms of inverse systems of abelian groups such that the se-
quences
ui vi
0 → A 0i −→ A i −→ A 00i → 0
 51

are exact for all i .

Prove that we have an exact sequence 0 → lim A 0i → lim A i → lim A 00i → lim1 A 0i →
←−− i ←−− i ←−− i ←−− i
lim1 A i → lim1 A 00i → 0.
←−− i ←−− i
Proof. Consider the following commutative diagram consisting of exact sequences

ker f 0 ker f ker f 00

Q u Q v Q
0 i∈ I A 0i i∈ I Ai i∈ I A 00i 0
f0 f f 00
Q u Q v Q
0 i∈ I A 0i i∈ I Ai i∈ I A 00i 0

coker f 0 coker f coker f 00

Since ker f ∼
= lim i A i , the snake lemma yields the desired result. □
←−−
52

7.2. Homework-2.
7.2.1. Part I. In the following, we work with morphisms in an abelian category C .
Exercise 7.6. Let f : A → B and g : B → C be morphisms.
(1) Suppose f and g are monomorphisms. Prove g ◦ f is a monomorphism.
(2) Suppose g ◦ f is a monomorphism. Prove f is a monomorphism.
Proof. For (1). Suppose α, β : D → A are arbitrary morphisms such that g ◦ f ◦ α =
g ◦ f ◦ β. Then f ◦ α = f ◦ β since g is a monomorphism, and thus α = β since f is
also a monomorphism.
For (2). Suppose α, β : D → A are arbitrary morphisms such that f ◦ α = f ◦ β. By
composing g one has
g ◦ f ◦ α = g ◦ f ◦ β,
and thus α = β since g ◦ f is a monomorphism. □
Exercise 7.7. Let f : A → B be a morphism in C . Recall that we have a commu-
tative diagram
f
A B

∼
=
coim f im f
Moreover A → coim f is an epimorphism and im f ,→ B is a monomorphism. Sup-
pose we have a commutative diagram
f
A B
ϕ ψ
∼
=
C D
such that ϕ : A → C is an epimorphism, ψ : D ,→ B is a monomorphism, and C ∼
=D
∼
= ∼
=
is an isomorphism. Prove that there exist isomorphisms coim f → C and D → im f
such that the following diagram commutes:
ϕ ∼
= ψ
A C D B
∼
= ∼
=
∼
=
coim f im f
Thus ϕ : A → C can be identified with ϕ : A → coim f , and ψ : D ,→ B can be identi-
fied with im f ,→ B.
Proof. For convenience we denote the kernel of f by τ : ker f → A , denote the iso-
morphism between C and D by g, and denote canonical morphism from A to coim f
by u.
Note that ψ ◦ g ◦ ϕ ◦ τ = f ◦ τ = 0. Then ϕ ◦ τ = 0 since ψ is a monomorphism and
g is an isomorphism. By universal property of cokernel there is a morphism from
coim f → C , denoted by α. Since α ◦ u = ϕ and both ϕ and u are epimorphisms,
 53

one has α is an epimorphism. By the same argument one can see there exists a
morphism β : im f → D which is a monomorphism. Since C is an abelian category,
there is canonical isomorphism between coim f and im f , and thus α is a monomor-
phism and β is an epimorphism. This shows both α and β are isomorphisms in C ,
since C is an abelian category.

τ f
0 ker f A B coker f 0
u v
ϕ ψ
α g β
coim f C D im f

∼
=

□
Exercise 7.8. Define the opposite category C ◦ of C as follows:
(a) C ◦ has the same objects as C . For any object A in C , we denote the corre-
sponding object in C ◦ by A ◦ .
( b) For any objects A and B in C , we define
¡ ¢
HomC ◦ A ◦ , B◦ = HomC (B, A ).

For any morphism ϕ : A → B in C , we denote by ϕ◦ : B◦ → A ◦ the corresponding

morphism in C ◦ .
Then
(1) Prove that C ◦ is an abelian category.
(2) Suppose
ϕ ψ
A→B→C
is an exact sequence in C . Prove that
ψ◦ ϕ◦
C ◦ → B◦ → A ◦

is an exact sequence in C ◦ .

Proof. For (1). Firstly, let’s see C is an additive category. For objects A ◦ , B◦ and
C ◦ of C ◦ , by definition HomC◦ ( A ◦ , B◦ ) = HomC ( A, B) is an abelian group, and the
composition
HomC ◦ ( A ◦ , B◦ ) × HomC ◦ (B◦ , C ◦ ) → HomC ◦ ( A ◦ , C ◦ )

is bilinear. Moreover, the direct sum of A ◦ , B◦ in C ◦ is the product of A, B in C ,

which also exists. Secondly, let’s show C ◦ is an abelian category. For morphism
f ◦ : B◦ → A ◦ in C ◦ corresponding to f : A → B in C , we’re going to show the kernel
of f ◦ is the cokernel of f . For arbitrary morphism α◦ : C ◦ → B◦ such that f ◦ ◦ α = 0,
by universal property of kernel, there exists the following commutative diagram
54

f◦
ker f ◦ B◦ A◦
◦
α

C◦
This corresponds to the following commutative diagram in category C
f
ker f ◦ B A
α

C
Then by uniqueness of cokernel, one has ker f ◦ is exactly the cokernel of f . Simi-
larly one can show the cokernel of f ◦ is exactly the kernel of f . This shows for any
morphism f ◦ : B◦ → A ◦ , it has kernel and cokernel since C is an abelian category.
Moreover, by the same argument it’s easy to see coim f ◦ is isomorphic to im f , and
im f ◦ is isomorphic to coim f , and thus
coim f ◦ ∼
= im f ◦ ,
since C is an abelian category.
ϕ ψ
For (2). Note that A → B → C is exact if and only if ker ψ = coim ϕ, and since C
is an abelian category, it’s equivalent to ker ψ = im ϕ. By arguments in the proof of
(1) it’s equivalent to coker ψ◦ = coim ϕ◦ . □
7.2.2. Part II.
Exercise 7.9. Let X be a topological space, A an abelian group endowed with the
discrete topology, and F the sheaf so that F (U ) is the group of continuous maps
from U to A for every open subset U of X . Prove that F is isomorphic to the sheaf
associated to the constant presheaf U 7→ A .
Proof. Firstly note that if A is equipped with discrete topology, then continuous
map f from U to A is locally constant since every point a ∈ A is an open subset,
and thus its preimage f −1 (a) is an open subset in U . On the other hand, by the
construction of constant sheaf associated to the constant presheaf, the sections of
it over U are also locally constant maps from U to A . This shows F is exactly the
sheafication of constant presheaf. □
Exercise 7.10. For every open subset U of the complex plane C, let O (U ) be the
ring of holomorphic functions on U , and let O ∗ (U ) be the group of units in O (U ).
Prove that the morphism O → O ∗ defined by
O (U ) → O ∗ (U )
p
−1 f
f 7→ e2π
is an epimorphism in the category of sheaves of abelian groups, but not an epi-
morphism in the category of presheaves. Here we regard O as a sheaf of abelian
 55

groups with respect to addition of functions. Prove that the kernel of this mor-
phism is isomorphic to the sheaf associated to the constant presheaf U 7→ Z.
exp
Proof. For the first part, if we want to show O −→ O ∗ → 0 is an exact sequence in
the category of sheaves of abelian groups, it suffices to check for each x ∈ C, the
following sequence of stalks is exact
exp
O x −→ O x∗ → 0.
It holds since for any non-vanishing holomorphic function f , log f is well-defined
on a sufficiently small neighborhood of x, which proves the surjectivity. On the
exp
other hand, O −→ O ∗ → 0 is not an exact sequence in the category of presheaves of
abelian groups, since
exp
O (C∗ ) −→ O ∗ (C∗ ) → 0
fails to be an exact sequence.
For the half part, we need to prove
p
2π −1 exp
0 → Z −→ O −→ O ∗
is an exact sequence in the category of sheaves of abelian groups. It suffices to
show for any open subset U ⊆ C, the following sequence of abelian groups is exact
p
2π −1 exp
0 → Z(U ) −→ O (U ) −→ O ∗ (U ).
p
If u : U → Z is a locally constant function, then it’s clear exp(2π −1 u) = 0. Con-
versely, if v : U →
p C is a holomorphic function such that exp v = 0. Then for each
x ∈ U , v( x) = 2π −1 u( x), where
p u : U → Z is a continuous function since v is con-
tinuous, and thus v ∈ 2π −1Z(U ), since continuous integral-valued function is
locally constant. □
e : C → (Sets) be
Exercise 7.11. Let C be a category. For any object X ∈ ob C , let X
the contravariant functor from C to the category of sets defined by
e (Y ) = Hom(Y , X ).
X
A functor from C to the category of sets is called representable by X if it is iso-
morphic to Xe . For any contravariant functor G : C → (Sets), prove that we have a
one-to-one correspondence
Hom( Xe ,G ) → G ( X )
α → α X (id X ),
where Hom( X e ,G ) is the set of natural transformations from the functor X
e to the
functor G . Prove the same result for covariant functors.
Proof. Let us first check this correspondence is surjective: For an object s ∈ G ( X ),
e → G as follows: For X 0 ∈ C , let α X 0 : X
we define α = α( s) : X e ( X 0 ) → G ( X 0 ) be the
0
morphism of set which sends f : X → C to G ( f )( s). Now let’s show α : X e → G is a
natural transformation: For any morphism g : X 00 → X 0 in C , it suffices to show
the following diagram commutes
56

αX 0
e ( X 0)
X G( X 0)
e ( g)
X G ( g)
α X 00
e ( X 00 )
X G ( X 00 )

e ( X 0 ), that is, a morphism f : X 0 → X , one has

For any element f ∈ X
G ( f ◦ g)( s) = G ( g) ◦ G ( f )( s).
This shows above diagram commutes by the construction of α. Moreover, it’s clear
αC (id X ) = G (id X )( s) = s
as desired.
To see above correspondence is injective: If there are two natural transforma-
e → G such that α X (id X ) = η X (id X ), we need to show α = η. In other
tion α, η : X
words, it suffices to show for any X 0 ∈ C , we have α X 0 = η X 0 . For any morphism
g : X 0 → X , as α is a natural transformation, we have the following commutative
diagram
αX
e (X )
X G(X )
e ( g)
X G ( g)
αX 0
e ( X 0)
X G( X 0)

It follows that
e ( g)(id X ) = α X 0 ( g).
G ( g) ◦ α X (id X ) = α X 0 ◦ X
Similarly as η is a natural transformation, one has (G ( g)◦η X )(id X ) = η X 0 ( g). Hence
α X 0 ( g) = G ( g) ◦ α X (id X ) = G ( g) ◦ η X (id X ) = η X 0 ( g).
By considering the opposite category, it’s clear the same result holds for covariant
functors. □
Exercise 7.12. Let u : C → D be a functor. Suppose that for each object D ∈ ob D ,
the functor
C → (Sets)
C 7→ Hom( u(C ), D )
is representable by an object v(D ) ∈ ob C . Then v : D → C is a functor right adjoint
to u.

Proof. In other words, for any objects C ∈ C , D ∈ D , there is an one-to-one corre-

spondence
Hom( u(C ), D ) ∼
= Hom(C, v(D )).
Thus by definition v is a right adjoint to u. □
Exercise 7.13. Let u : C → D be a functor.
 57

(1) We say u is faithful (resp. fully faithful) if for any objects C 1 , C 2 ∈ ob C , the
map
Hom (C 1 , C 2 ) → Hom ( u (C 1 ) , u (C 2 ))
is injective (resp. bijective).
(2) We say u is essentially surjective if for any object D in D , there exists an object
C in C such that we have an isomorphism u(C ) ∼ = D.
(3) We say u is an equivalence of categories if u is both fully faithful and essen-
tially surjective.
Suppose u is an equivalence of categories. For any D ∈ ob D , choose an object
v(D ) ∈ ob C such that u ◦ v(D ) ∼= D . Prove that v is a functor that is both left and
right adjoint to u : D → C . It is called a quasi-inverse of u.
Proof. Firstly let’s show v is a functor: If f : D 1 → D 2 is a morphism in D , then
consider the following commutative diagram
v u
D1 v( D 1 ) D1
f v( f ) f
v u
D2 v( D 2 ) D2
Since u is an equivalence of categories, and thus it’s fully faithfully, so there exists
a morphism v( f ) : v(D 1 ) → v(D 2 ) still making above diagram commutes, which
shows v is a functor.
Now let’s show v is the right adjoint of u, that is to show for any C ∈ C and
D ∈ D , there is a one-to-one correspondence Hom( u(C ), D ) = Hom(C, v(D )). Note
that u is essentially surjective, so there exists C 0 such that u(C 0 ) = D , and thus
Hom( u(C ), D ) = Hom( u(C ), u(C 0 )) = Hom(C, C 0 ).
On the other hand, one has
Hom(C, v(D )) = Hom(C, v◦ u(C 0 )) = Hom( u(C ), u◦v◦ u(C 0 )) = Hom( u(C ), u(C 0 )) = Hom(C, C 0 ).
This shows v is the right adjoint of v, and by the same argument one can see v is
the left adjoint of u. □
58

7.3. Homework-3.
Exercise 7.14. Let A be a ring. For every open subset U of Spec A , let SU be the
T −1
multiplicative subset SU = p∈U ( A − p), and let P (U ) = SU A . For every inclusion
V ⊆ U of open subsets, we have SU ⊆ S V and hence we have a canonical homo-
morphism P (U ) → P (V ). This makes P a presheaf of rings on Spec A . Prove that
O Spec A ∼
= P +.
Proof. It suffices to show for each point p ∈ Spec A , one has
O Spec A,p ∼
= P +. p
= A p , so it suffices to show P p+ ∼
Note that O Spec A,p ∼ = A p . But by the construction of
P , one has
P p+ = P p = lim P (U ) = lim SU
−1
A.
−−→ −−→
p∈U p∈U
Now it suffices to show that A p satisfies the universal property of inverse limit
lim S −1 A , which follows from the universal property of localization. □
−−→p∈U U
Exercise 7.15. Let S be a multiplicative subset of a ring A . Prove that the canon-
ical morphism Spec S −1 A → Spec A induces an embedding on the underlying topo-
logical spaces.
Proof. Recall that the prime ideals in S −1 A are in one to one correspondence with
prime ideals in A which do not intersect with S , and the correspondence is given by
pullback. This shows the canonical morphism ϕ : Spec S −1 A → Spec A is bijective,
and it’s clear that ϕ is continuous, so it suffices to show that ϕ is closed.
Note that every ideal in S −1 A is an extended ideal, that is, it’s of the form S −1 a,
where a ⊆ A is an ideal. Then
ϕ(V (S −1 a)) = ϕ({S −1 p | S −1 a ⊆ S −1 p, p is prime})
= {p | a ⊆ p, p is prime}
= V (a).
This completes the proof. □
Exercise 7.16. Let x be a point in scheme X , and let k( x) = O X ,x /m x be the residue
field at x. Construct a natural morphism i : Spec k( x) → X with image x so that
the homomorphism O X ,x → k( x) induced by i ] is the canonical homomorphism.
Proof. If we want to construct a morphism from scheme Spec k( x) → X , it suf-
fices to construct a continuous map between topological spaces and a morphism
between structure sheaves.
(1) For the continuous map between topological spaces Spec k( x) and X , we simply
send Spec k( x) to the point x ∈ X since Spec k( x) is just a single point.
(2) For the morphism i ] between structure sheaves, it’s defined as follows: For
open subset U ⊆ X , i ] : O X (U ) → O Spec k( x) ( i −1 (U )) = k( x) is defined by
α
O X (U ) −→ O X ,x → k( x),
 59

where α is given by taking limit if x ∈ U , otherwise α is zero map.

Then above data gives a morphism between schemes Spec k( x) and X , and by def-
inition the homomorphism O X ,x → k( x) induced by i ] is canonical morphism. □
60

7.4. Homework-4.
Exercise 7.17. Let X be a topological space and C X be the sheaf of complex-
valued functions on X . Prove that ( X , C X ) is a locally ringed space. Moreover, for
each p ∈ X , one has
m p = { f ∈ C X ,p | f ( p) = 0},
and the residue field k( p) = C.
Proof. To show ( X , C X ) is a locally ringed space, it suffices to show that for each
p ∈ X , every element in C X ,p \ m p is a unit. Then m p is the unique maximal ideal
and thus C X ,p is the local ring.
For f ∈ C X ,p \ m p , since f ( p) 6= 0, we may construct a continuous function g
defined on an open neighborhood U of p such that g( p) = 1/ f ( p). Then g is an
inverse of f in C X ,p .
To see the residue field, it suffices to note that
0 → m p → C X ,p → C → 0
is an exact sequence. □
Exercise 7.18. Let f : U → X be an embedding of topological spaces. Then for any
sheaf F defined on U and p ∈ U , prove that
( f ∗ F ) f ( p) ∼
= Fp.
Proof. Since f is a topological embedding, without lose of generality we may as-
sume U ⊆ X equipped with subspace topology and f is the inclusion map i : U ,→
X . By definition one has
( i ∗ F ) p = lim i ∗ F (V )
−−→
p∈V ⊆ X

= lim F ( i −1 (V ))
−−→
p∈V ⊆ X
= lim F (V ∩ U ).
−−→
p∈V ⊆ X

On the other hand, since U is equipped with subspace topology, every open subset
of U containing p is exactly of the form V ∩ U , where V ⊆ X is an open subset
containing p. This shows
lim F (V ∩ U ) = lim F (U ) = F p
−−→ −−→
p∈V ⊆ X p∈U

□
 61

7.5. Homework-5.
7.5.1. Part I.
Exercise 7.5.1. Let S be a graded ring.
(1) Let p be a prime ideal of S , and let
M
p0 = (p ∩ S d ).
d
0
Prove that p is a homogeneous prime ideal of S .
p
(2) Let a be a homogeneous ideal of S . Prove that a is the intersection of homoge-
neous prime ideals containing a.
Proof. For (1). Note that for any degree d , one has
p0 ∩ S d = p ∩ S d .
This shows M M
p0 = (p ∩ S d ) = (p0 ∩ S d ),
d d
and thus p is a homogeneous ideal.
To see p0 is prime, it suffices to show if a, b are two homogeneous elements such
that ab ∈ p0 , then either a or b in p. Since both a and b are homogeneous, then
ab is also homogeneous. If ab ∈ p ∩ S d , then either a or b in p since p is prime,
and thus either a or b in some p ∩ S d 0 since both a and b are homogeneous. This
completes the proof of p0 is a prime homogeneous ideal.
For (2). Suppose I is the set of all homogeneous prime ideals of S containing
p T p
a. Firstly one has a ⊆ p0 ∈ I p0 since a equals the intersection of all prime ideals
containing a. On the other hand, for any prime ideal p containing a, one has the
L
homogeneous prime p0 = d (p ∩ S d ) ⊆ p also contains a, since
M M
a = (a ∩ S d ) ⊆ (a ∩ S d ).
d p
p L 0
Thus a= p0 ∈ I p as desired. □
Exercise 7.5.2. Let ϕ : S → T be a homomorphism of graded rings. Suppose
there exists an integer m ≥ 1 such that ϕ(S d ) ⊂ T md for all d . Let I be the ho-
mogeneous ideal of T generated by ϕ(S + ), and let U = Proj T − V (T ). Construct a
morphism f : U → Proj S of schemes so that f (q) = ϕ−1 (q) for any q ∈ U and that
]
f q : O Proj S, f (q) → O Proj T,q can be identified with the homomorphism ϕq : S (ϕ−1 (q)) →
T (q ) .
Proof. Firstly let’s construct the continuous map between the base topological
spaces U and Proj S . Since ϕ : S → T is a homomorphism of graded rings, one
has the pullback of a homogeneous ideal q ⊆ T still is a homogeneous ideal of S .
As a consequence, if q ∈ U = Proj T \ V (T ), one has ϕ−1 (q) is a homogeneous prime
ideal, and ϕ−1 (q) doesn’t contain S + , otherwise q will contain T , which is a con-
tradiction. This shows f (q) := ϕ−1 (q) gives a well-defined map from U to Spec S ,
which is also continuous.
62

Now let’s construct the morphism between structure sheaves O Proj S and f ∗ OU ,
where OU is O Spec T |U in fact. For each open subset V ⊆ Spec S , it suffices to con-
struct a homomorphism of rings

O Proj S (V ) → OU ( f −1 (V )) = O Proj T |U ( f −1 (V )),

which is compatible with the restriction map between different open subsets. Given
q ∈ Proj T \ V (T ), we will construct a homomorphism ϕq : Sϕ−1 (q) → T(q) as follows:
For any element s/ t ∈ S ϕ−1 (q) , ϕ( s/ t) is defined by ϕ( s)/ϕ( t). Since t 6∈ ϕ−1 (q), one has
ϕ( t) 6∈ q, and thus ϕ( s/ t) ∈ Tq . On the other hand, since ϕ is a ring homomorphism
between graded rings, it maps elements of degree zero to the one of degree zero,
and thus ϕ( s/ t) ∈ T(q) as desired.
`
For any element s ∈ O Proj S (V ), it’s a map s : V → p S p satisfying some proper-
ties. Given q ∈ f −1 (V ) with ϕ−1 (q) = p, that is, f (q) = p ∈ V . By composing
f sp ϕq
q −→ p −→ S(p) −→ T(q) ,
`
one can construct a map t : f −1 (V ) → q∈ f −1 (V ) T(q) . A routine check shows that
t gives an element of OU ( f −1 (V )), and this correspondence gives a morphism be-
tween sheaves O Proj S → f ∗ OU such that the induced morphism on stalks is exactly
ϕq : S (ϕ−1 (q)) → T(q) . □
L
Exercise 7.5.3. Let A be a ring, I an ideal of A , and S = ∞ d
d =0 I . Then S is a
graded ring. We call Proj S the blowing-up of Spec A along the ideal I . Prove that
the inclusion A = S 0 ,→ S induces a morphism of schemes f : Proj S → Spec A such
∼
=
that over the open subset U = Spec A − V ( I ), f induces an isomorphism f −1 (U ) −→
U.

Proof. For convenience we denote the inclusion i : A ,→ S . Firstly let’s show the
inclusion i gives a continuous map f between topological spaces between Proj S
and Spec A with im f ⊆ U . Given p ∈ Proj S , one has i −1 (p) is a prime ideal of A ,
and i −1 (q) cannot contain the ideal I , otherwise q contains all power of I , and thus
it contains S + , a contradiction.
(1) Note that the continuous map f : Proj S → U is surjective, since for any prime
ideal p ⊆ A , automatically it’s a homogeneous prime ideal in S with f (p) = p.
(2) On the other hand, if homogeneous prime ideals p, q ∈ S such that i −1 (p) =
i −1 (q), then p = i −1 (p)S + = i −1 (q)S + = q. This shows f is injective.
(3) Finally let’s show f is a closed map. Suppose V+ (a) is a closed subset of Proj S ,
where a ⊆ S is a homogeneous prime ideal. Then
f (V+ (a)) = f ({p ∈ Proj S | a ⊆ p})
= { i −1 (p) | p ∈ Proj S, a ⊆ p}
= {p ∈ Spec A | i −1 (a) ⊆ p}
= V ( i −1 (a)).
 63

∼
=
Above arguments shows that f −1 (U ) −→ U as topological spaces.
To see there exists an isomorphism between structure sheaves OU and f ∗ O f −1 (U ) ,
it suffices to show OU, f (p) ∼
= O f −1 (U ),p holds for each p ∈ f −1 (U ). On one hand, one
has
OU, f (p) = A i−1 (p) .
On the other hand, O −1 ∼
= S (p) . Note that
f (U ),p
s
S (p) = { | s ∈ S, t ∈ p, s, t are homogeneous and of the same degree.}
t
It’s clear that there exists an inclusion A i−1 (p) ,→ S (p) . Conversely, for any s/ t ∈ S (p) ,
it suffices to construct an element a/ b ∈ A i−1 (p) such that a/ b = s/ t in S (p) . □
Exercise 7.5.4. Let A = R [ x1 , . . . , xn ] be a polynomial ring, let I = ( x1 , . . . , xn ) be the
L
ideal of A generated by x1 , . . . , xn , let S = ∞ d
d =0 I , let T = A [ y1 , . . . , yn ] be the graded
ring so that T d consists of homogeneous polynomials of degree d in the variables
y1 , . . . , yn with coefficients in A , and let J be the homogeneous ideal of T generated
by x i y j − x j yi ( i, j = 1, . . . , n). Consider the epimorphism of graded rings ϕ : T → S
so that ϕ(a) = a ∈ S 0 for any a ∈ A and ϕ( yi ) = x i ∈ S 1 . Prove that ϕ induces an
∼
=
isomorphism of schemes Proj S −→ Proj A [ y1 , . . . , yn ]/ J .
Proof. Here we only give a proof for n = 2, and cases for more variables are similar.
Firstly note that the kernel of the epimorphism of graded rings
ϕ: T → S
contains the ideal J , since ϕ( x1 y2 − x2 y1 ) = 0. Conversely, if α = a y1 + b y2 is
mapped to 0, then it must be in the ideal generated by x1 y2 − x2 y1 since x1 , x2
are algebraically independent. This shows ϕ : T → S induces an isomorphism be-
tween graded rings S and A [ y1 , . . . , yn ]/ J , that is, a ring isomorphism which pre-
serves the degree. Thus it induces an isomorphism between schemes Proj S and
Proj A [ y1 , . . . , yn ]/ J . □

7.5.2. Part II.

Exercise 7.5.5. Let S be a graded ring and p be a homogeneous prime ideal. Prove
that
(1) If f ∈ S \ p, then S ( f ) → S (p) is injective.
(2) S (p) is a local ring.
Proof. For (1). Suppose a/ f n = b/ f m in S (p) . Then there exists a homogeneous
element s ∈ S \ p such that
s ( a f m − b f n ) = 0.
Then a f m − b f n = 0 since p is a prime ideal, and thus a/ f n = b/ f m in S ( f ) .
For (2). Consider
s
m = { | s, t ∈ p, s, t are homogeneous and of the same degree}.
t
64

Note that any element outside of m is invertible, and thus m is the only maximal
ideal of the local ring S (p) . □
Exercise 7.5.6. Prove that if ∅ 6= U ⊆ Spec A , then 0 6= 1 ∈ O Spec A (U ).
Proof. Since ∅ 6= U , we may assume there exists a non-zero ideal p ∈ U . If 0 = 1 ∈
O Spec A , that is, O Spec A (U ) is a zero ring, then for any open subset V ⊆ U , one has
O Spec A (V ) is also zero ring. In particular, lim O (U ) = A p is a zero ring, a
−−→p∈U Spec A
contradiction. □
 65

7.6. Homework-6.

Exercise 7.6.1. Let X be a Noetherian topological space and let F be a presheaf.

Suppose that for every open subset U and every finite open covering (U i ) i∈ I , the
following conditions hold:
(1) For any sections s, t ∈ F (U ) such that s|Ui = t|Ui for every i ∈ I , then s = t.
(2) Let s i ∈ F (U i ) be sections such that s i |Ui ∩U j = s j |Ui ∩U j for every pair i, j ∈ I .
Then there exists a section s ∈ F (U ) such that s|Ui = s i for every i ∈ I .
Prove that F is a sheaf.

Proof. For a Noetherian topological space X , any open subset U ⊆ X is quasi-

compact, that is, any open covering of U admits a finite subcovering. Thus it
suffices to chech above two conditions for every finite open covering. □

Exercise 7.6.2. Let (F i , ϕ i j ) i∈ I be a direct system of sheaves of abelian groups on

X , and let lim F i be the sheaf associated to the presheaf U 7→ lim F i (U ). Prove
−−→ i −−→ i
that lim F i is the direct limit of (F i , ϕ i j ) i∈ I in the category of sheaves, and for
−−→ i
every P ∈ X , we have

(lim F i ) p ∼
= lim F i,p .
−−→ −−→
i i

Suppose furthermore that X is a Noetherian topological space. Prove that the

presheaf U 7→ lim F i (U ) is a sheaf.
−−→ i
Proof. Firstly let’s show lim F i statisfies the universal property of the direct limit
−−→ i
of (F i , ϕ i j ) i∈ I . Suppose C is a sheaf and ψ i : F i → C are morphisms such that
ψ j ϕ i j = ψ i . Then For any open subset U ⊆ X , by the universal property of lim F i (U ),
−−→ i
one has

F i (U )
ψ i (U )

ϕ i j (U ) lim F i (U ) C (U )
−−→ i
ψ j (U )
F j (U )

For convenience we denote ψ(U ) : lim F (U ) → C (U ). Since all of ϕ i j , ψ i , ψ j are

−−→ i
morphisms between (pre)sheaves, it’s clear that the collection of group homomor-
phisms {ψ(U )}U ⊆ X gives a morphism of presheaves U 7→ lim F i (U ) and C , and
−−→ i
thus gives a morphism of sheaves lim F i and C .
−−→ i
Secondly, note that
66

F i (U ) lim F i (U )
−−→ i

F i,p F j (U )

lim F i,p F j,p

−−→ i
The morphism lim F i (U ) → lim F i,p statisfies the universal property of taking
−−→ i −−→ i
stalk, and thus by the uniqueness one has
(lim F i ) p ∼
= lim F i,p .
−−→ −−→
i i
Finally let’s suppose X is a Noetherian topological space and prove that U 7→
lim F i (U ) is a sheaf. For every finite open covering {Uα }α∈A of U . If s, t ∈ lim F i (U )
−−→ i −−→ i
such that s|Uα = t|Uα for all α ∈ A , then for each α ∈ A , there exists Nα such that
for all i > Nα , one has s|Uα = t|Uα in F i (Uα ). Since A is a finite index set, we may
take N > max{ Nα | α ∈ A }. Then for all i > N and α ∈ A , one has s|Uα = t|Uα in
F i (Uα ), and since F i is a sheaf, one has s = t in F i (U ) for all i > N , and thus s = t
in lim F i (U ). Similarly, one can check the other condition for lim F i to be a sheaf
−−→ i −−→ i
by the same argument. □
Exercise 7.6.3. Let (F i , ϕ i j ) i∈ I be an inverse system of sheaves of abelian groups
on X . Prove that the presheaf U 7→ lim F i (U ) is a sheaf and it is the inverse limit
←−− i
of (F i , ϕ i j ) i∈ I in the category of sheaves.
Proof. By the same argument as above exercise one can show that lim F i is the
←−− i
inverse limit in the category of sheaves, so here we only prove that U 7→ lim F i (U )
←−− i
is a sheaf.
Recall that a presheaf F is a sheaf if and only if for an open subset U and open
covering {Uα } of U , the following sequence is exact
Y Y
0 → F (U ) → F (Uα ) → F (Uαβ ).
α α,β

Since the inverse limit is left exact, one has

Y Y
0 → lim F i (U ) → lim F i (Uα ) → lim F i (Uαβ ).
←−− α ← −− α,β
←−−
i i i
This completes the proof. □
 67

7.7. Homework-7.

Exercise 7.7.1. Let f : X → Y be a morphism of schemes which is locally of finite

type. Let V = Spec B be an affine open subscheme of Y and U = Spec A be an affine
open subscheme of X such that f (U ) ⊆ V .
(1) Prove that f −1 (V ) can be covered by affine open subschemes {Uλ = Spec A λ }λ∈Λ
such that A λ are finitely generated B-algebras.
(2) Prove that there exist finitely many a 1 , . . . , a n ∈ A such that D (a 1 ), . . . , D (a n )
cover U , and each A a i is a finitely generated B-algebra.
(3) Prove that A is a finitely generated B-algebra.

Proof. For (1). Since V ∩ Vi is an open subset of Vi = Spec B i , then V ∩ Vi is a

union of distinguished open subsets D ( f ik ) = Spec(B i ) f ik , and for each f ik ∈ B i ,
for convenience we still use f ik to denote the image of f ik under the morphism
B i → A i j . Then ( A i j ) f ik is a finitely generated (B i ) f ik -algebra. After relabelling
the index, in fact we have shown that V is covered by affine schemes Spec C i such
that each f −1 (Spec C i ) is a union of affine schemes Spec D i j , where D i j is a finitely
generated C i -algebra.
For each point p ∈ V , suppose it lies in the affine scheme Spec C i . Then there
exists a distinguished open subset Spec B f p ⊆ Spec C i which contains p. For conve-
nience we still use f p to denote the image of f p under the morphism B → C i → D i j .
Then each (D i j ) f p is a finitely generated B f p -algebra, and thus a finitely generated
B-algebra. This completes the proof.
For (2). Since f −1 (V ) is covered by affine schemes Spec A λ , then U = Spec A is
also covered by the intersection of Spec A ∩ Spec A λ . Moreover, since both A and
A λ are affine, then by Lemma 5.1.1 one can pick a collection of open subset Uλ i
such that Uλ i are simultaneously the distinguished open subsets of Spec A and
Spec A λ . For convenience we write

Uλ i ∼
= Spec A f λi ∼
= Spec( A λ ) g λi .

Since A λ are finitely generated B-algebra, so is ( A λ ) g λi , and thus each A f λi can be

realized as a finitely generated B-algebra, which completes the proof.
For (3). Note that Spec A is covered by distinguished open subsets D (a 1 ), . . . , D (a n )
if and only if (a 1 , . . . , a n ) = A . Thus it reduces to the following lemma of commuta-
tive algebra.

Lemma 7.7.1. Let A be a B-algebra and (a 1 , . . . , a n ) = A . If A a i is a finitely gener-

ated B-algebra for each i , then A is also a finitely generated B-algebra.

Proof. For each x ∈ A , since A a i is finitely generated B-algebra, its image in A a i is

equal to some
x1i x ij
i i
F ( ,..., ),
ki k ij
ai 1 ai i
68

ki
where x ij /a i j are generators of A a i over B, and F i is some polynomial with coeffi-
cients in B. After multiplying by a large power of a i , there are n equations in A
which are of the form
aN ei i i
i x = F ( x i , . . . , x j i , a i ).
On the other hand, since (a 1 , . . . , a n ) = 1, there exists m 1 , . . . , m n ∈ A such that
m 1 a 1 + · · · + m n a n = 1.
Exponentiate above equation to the nN -th power and multiply by x, one has
x = G (a 1 , . . . , a n , m 1 , . . . , m n , x ij ),
where G is a polynomial with coefficients in B, since each monimial of m i , there
exists some a j in the coefficients such that the power of a j is ≥ N , and thus a N i
x
can be replaced by F e i . This shows any x ∈ A can be expressed as a polynomial of
m i , a i , x ij with coefficients in B, and thue A is a finitely generated B-algebra. □
□
 69

7.8. Homework-8.

Exercise 7.8.1. Let ( f , f ] ) : ( Z, O Z ) → ( X , O X ) be a morphism of schemes.

(1) ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is an open immersion if and only if f induces a home-
]
omorphism of Z with an open subset of X and f p : O X , f ( p) → O Z,p is an isomor-
phism for every p ∈ Z .
(2) ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is an immersion if and only if f induces a homeo-
]
morphism of Z with a locally closed subset of X and f p : O X , f ( p) → O Z,p is an
epimorphism for every p ∈ Z .
(3) The immersions are monomorphisms in the category of schemes. Moreover, the
composite of immersions is an immersion, so are open immersions and closed
immersions.

Proof. For (1). Note that by definition one has ( f , f ] ) : ( Z, O Z ) → ( X , O X ) is an open

immersion if and only if it induces an isomorphism between ( Z, O Z ) and an open
subscheme of ( X , O X ). Since f has already induced a homeomorphism of Z with
]
an open subset of X , it suffices to show for every p ∈ Z , f p : O X , f ( p) → O Z,p is an
isomorphism if and only if ( f ∗ O Z ) f ( p) ∼
= O X , f ( p) . In general it fails, but since f
induces a homeomorphism, one has

( f ∗ O Z ) f ( p) = lim O Z ( f −1 (V )) = lim O Z (U ) = O Z,p

−−→ −−→
p∈ f−1 (V ) p∈U

is an isomorphism. On the other hand, one has the following commutative dia-
gram

]
f p : O X , f ( p) O Z,p

( f ∗ O Z ) f ( p)

]
Thus f p is an isomorphism if and only if ( f ∗ O Z ) f ( p) ∼
= O X , f ( p) , and by the same
argument one can show (2).
For (3). Since the composite of epimorphisms is an epimorphism, one has the
composite of immersions is an immersion, so are open immersions and closed im-
mersions.
Now let’s show immersions are monomorphisms in the category of schemes.
e → Z are two morphisms between schemes such that f ◦ α = f ◦ β,
Suppose α, β : Z
where f is an immersion. Firstly α = β as morphisms between topological spaces
since f induces homeomorphism between underlying topological spaces. Moreover,
on each stalk one has
α] ◦ f ] = β] ◦ f ] ,
and thus one has α] = β] on each stalk since f ] is an epimorphism. □
70

Exercise 7.8.2. Let S be a graded ring and let a be a homogeneous ideal of

S . Prove that the canonical homomorphism S → S /a induces a closed immersion
Proj S /a → Proj S .
Proof. Firstly the canonical homomorphism S → S /a is a homomorphism of graded
rings, and thus the pullback of homogeneous prime ideals are still homogeneous
prime ideals. For any p/a ∈ Proj S /a, the pullback of p/a is p, which is a homo-
geneous prime ideal which contains a and cannot contain S + , otherwise p/a con-
tains (S /a)+ , a contradiction. On the other hand, any homogeneous prime ideals
p ∈ V+ (a) gives an element in Proj S /a, and these two constructions are inverse
to each other. Thus one has Proj S /p ∼ = V+ (a) as sets. Moreover, it’s also easy to
show the canonical homomorphism Proj S /a → V+ (a) is a closed map and thus it’s
a homeomorphism with respect to Zariski topology.
Now it suffices to show for each p/a ∈ Proj S /a, the canonical homomorphism
O Proj S,p → O Proj S /a,p/a is surjective. Note that
O Proj S,p ∼
= S (p )
O Proj S /a,p/a ∼
= (S /a)(p/a) ,
and the canonical homomorphism is given by projection, which is surjective. This
completes the proof. □
 71

7.9. Homework-9.

7.9.1. Part I.

Exercise 7.9.1. Finish step 3 of the proof of Proposition 1.3.20 as follows.

(1) Let
qi
U i ×S Y Y
pi

Ui S
be the fibred product of U i and Y . Prove that there exists one and only one
isomorphism ϕ i j : p−i 1 (U i ∩U j ) → p−j 1 (U i ∩U j ) such that the following diagrams
commute:
ϕi j ϕi j
p−i 1 (U i ∩ U j ) p−j 1 (U i ∩ U j ) p−i 1 (U i ∩ U j ) p−j 1 (U i ∩ U j )
pj qj
pi qi
Ui ∩ U j Y

(2) Prove that ϕ i j = ϕ−ji1 and ϕ jk ◦ ϕ i j = ϕ ik when restricted to p−i 1 (U i ∩ U j ∩ Uk ). So

we can glue the schemes U i ×S Y together to get a scheme Z .
(3) Suppose we have a commutative diagram
q
Z Y
p

X S
S
and suppose X has an open covering X = i Ui such that
q
p−1 (U i ) Y
p

Ui S
are fibred product for all i . Prove that Z in the first diagram is the fibred
product of X and Y over S .

Proof. For (1). By step 2 of the proof of Proposition 1.3.20, one has p−i 1 (U i ∩ U j ) is
the fibred product of U i ∩ U j and Y over S , so is p−j 1 (U i ∩ U j ). Thus there exists
one and only one isomorphism ϕ i j : p−i 1 (U i ∩ U j ) → p−j 1 (U i ∩ U j ) by the universal
property of fibred product.
For (2). Note that ϕ i j ◦ ϕ ji is an isomorphism such that the following diagram
commutes
72

ϕ i j ◦ϕ ji
p−i 1 (U i ∩ U j ) p−i 1 (U i ∩ U j )
pi
pi
Ui ∩ U j ,

so is the identity map. Then by the fact that the fibred product is unique up to a
unique isomorphism, one has ϕ i j ◦ ϕ ji = id, that is, ϕ i j = ϕ−ji1 as desired. The same
argument shows that ϕ jk ◦ ϕ i j = ϕ ik .
For (3). In order to prove that Z satisfies the universal property of fibred prod-
uct, we need to show for any commutative diagram
β
W Y
α

X S,

there exists a unique morphism W → Z such that the following diagram commutes
β
W

q
Z Y
α
p

X S.
S
Since X = i U i , an observation is that the morphism α : W → X is equivalent to a
collection of morphisms {α i : X → U i } which are compatible with each other. Thus
for each i there exists a unique morphism W → p−1 (U i ) such that the following
commutative diagram
β
W

q
p−1 (U i ) Y
αi
p

Ui S

since p−1 (U i ) is the fibred product of U i and Y over S . By uniqueness the collection
of morphisms {W → π−1 (U i )} can be glued to a unique morphism from W → Z , as
desired. □
Exercise 7.9.2. Prove the isomorphism X i ×S Y ∼
= X i ×S i Yi in step 6 in the proof
of Proposition 1.3.20.
Proof. It suffices to note that given morphisms f : Z → X and g : Z → Y over S ,
the image of g must land inside Yi . □
 73

7.9.2. Part II.

Exercise 7.9.3. Let C be a category in which fibred product exists. Consider com-
mutative diagrams
f0 g0 g0 ◦ f 0
X 00 X0 X X 00 X

f g g◦ f
S 00 S0 S S 00 S
Suppose the second square is Cartesian. Prove that the first square is Cartesian if
and only if the third one is Cartesian.
Proof. Suppose the first square is Cartesian and consider the following commuta-
tive diagram
β
W

g0 ◦ f 0
X 00 X
α

S 00 S
g◦ f

By composing α and f , one obtains the following commutative diagram

β
W
ϕ

g0
X0 X
f ◦α

S0 g S

where the morphism ϕ : W → X 0 is induced by the assumption that the second

square is Cartesian. On the other hand, consider the following commutative dia-
gram
ϕ
W
ψ

f0
X 00 X0
α

S 00 S0,
f

where the morphism ψ : W → X 00 is induced by the assumption that the first square
is Cartesian, and that’s exactly the morphism making the third square to be Carte-
sian.
74

Conversely, suppose the third square is Cartesian, and consider the following
commutative diagram

β
W

f0
X 00 X0
α

S 00 S0.
f

Then by the assumption the third square is Cartesian, one has the following com-
mutative diagram

g 0 ◦β
W
ϕ

g0 ◦ f 0
X 00 X
α

S 00 S,
g◦ f

where the induced morphism W → X 00 is denoted by ϕ. In order to show the first

square is Cartesian, it suffices to show the following diagram commutes

β
W
ϕ

f0
X 00 X0
α

S 00 S,
f

which follows from the second square is Cartesian. □

Exercise 7.9.4.
(1) Let f : X → S and g : Y → S be maps of sets. Prove that their fibred product is
( X ×S Y , p, q), where
X ×S Y = {( x, y) | x ∈ X , y ∈ Y , f ( x) = g( y)},
and p : X ×S Y → X , q : X ×S Y → Y are the projections p( x, y) = x and q( x, y) = y
respectively.
(2) Use the description of the fibred product in (1) to prove Proposition 5.4.3 for the
category of sets.
(3) Let C be a category and let
 75

X0 X

S0 S
be a commutative diagram in C . For every object Z ∈ ob C , it induces a com-
mutative diagram

Hom( Z, X 0 ) Hom( Z, X )

Hom( Z, S 0 ) Hom( Z, S )
in the category of sets. Prove that the first diagram is Cartesian in C if and
only if the second diagram is Cartesian in the category of sets for every object
Z ∈ ob C .
(4) Use (2) and (3) to prove Proposition 1.3.24 for every category C in which fibred
product exists.

Proof. For (1). Suppose there exist morphisms (between sets) α : W → X and
β : W → Y such that the following diagram commutes

β
W Y
α g

X S.
f

Then one can construct

φ : W → X ×S Y
w 7→ (α(w), β(w))

such that the following diagram commutes

β
W
φ

α
X ×S Y Y
g

X S.
f

Moreover, any morphism from W → X ×S Y such that above diagram commutes

must be of this form. This shows X ×S Y statisfies the universal property of fibred
product.
For (2). Given the morphisms between sets as follows
76

X Y
f g

T
ϕ

S.

By using the description of the fibred product in (1), the following diagram com-
mutes

X ×T Y X ×S Y

T T ×S T,

since the morphisms in above diagram can be described as follows:

( a) X ×T Y → X ×S Y is given by ( x, y) 7→ ( x, y).
( b) X ×S Y → T ×S T is given by ( x, y) 7→ ( f ( x), g( y)).
( c) T 7→ T ×S T is given by t 7→ ( t, t).
(d ) X ×T Y → T is given by ( x, y) 7→ f ( x) or ( x, y) 7→ g( y), since in this case f ( x) =
g( y).
Moreover, given morphisms α : W → T and β : W → X ×S Y such that the following
diagram commutes
β
W

X ×T Y X ×S Y
α

T T ×S T,

an observation is that the image of β lies in X ×T Y , and thus there is a unique

morphism from W to X ×T Y such that above diagram commutes, which is exactly
β itself. This shows X ×T Y statisfies the universal property of fibred product.
For (3). Suppose

X0 X
f

S0 g S

is Cartesian. For any Z ∈ ob C and morphisms α : Z → S 0 and β : Z → X such

that f ◦ β = g ◦ α, by the universal property there exists a unique mo ϕ : Z → X 0 .
Conversely given morphisms Z → X 0 , it’s easy to construct morphisms α : Z → X
 77

and β : Z → S 0 such that f ◦ β = g ◦ α. This shows Hom( Z, X 0 ) ∼

= Hom( Z, S 0 ) ×Hom( Z,S )
Hom( Z, X ) by the description of fibred product in (1). This shows

Hom( Z, X 0 ) Hom( Z, X )

Hom( Z, S 0 ) Hom( Z, S )

is Cartesian, and by the same argument one can prove the converse statement.
For (4). By (3), it suffices to show that for each Z ∈ ob C , the following diagram
is Cartesian
Hom( Z, X ×T Y ) Hom( Z, X ×S Y )

Hom( Z, T ) Hom( Z, T ×S T ).

Note that by the proof of (3), one can see

Hom( Z, T ×S T ) = Hom( Z, T ) ×Hom( Z,S ) Hom( Z, T )
Hom( Z, X ×S Y ) = Hom( Z, X ) ×Hom( Z,S ) Hom( Z, Y )
Hom( Z, X ×T Y ) = Hom( Z, X ) ×Hom( Z,T ) Hom( Z, Y ),
and thus the desired result follows from (2). □
7.9.3. Part III.
Exercise 7.9.5. Let X , Y be S -schemes and f , g : X → Y be S -morphisms. Suppose
( f , g) : X → Y ×S Y is the morphism such that p ◦ ( f , g) = f , q ◦ ( f , g) = g and K is
the fibred product of X and Y over Y ×S Y , which can be seen as follows
b
K Y
0
h ι ∆
h ( f ,g) p
Z X Y ×S Y Y
q

Y S.
(1) Prove that ι : K → X is an immersion and f ◦ ι = g ◦ ι.
(2) Let h : Z → X be a morphism such that f ◦ h = g ◦ h. Prove that there is a unique
morphism h0 : Z → K such that ι ◦ h0 = h.
Proof. For (1). If α, β : Z → K are morphisms such that ι ◦ α = ι ◦ β, then one also
has ∆ ◦ b ◦ α = ∆ ◦ b ◦ β. Then by the universal property of K one can see morphism
from Z → K with such property must be unique, and thus α = β. This shows ι is a
monomorphism. For the half part, note that
p ◦ ( f , g) ◦ ι = f ◦ ι
q ◦ ( f , g ) ◦ ι = g ◦ ι.
78

On the other hand, ( f , g) ◦ ι = ∆ ◦ b and p ◦ ∆ = q ◦ ∆.

For (2). Since h : Z → X statisfies f ◦ h = g ◦ h, one has p ◦ ( f , g) ◦ h = q ◦ ( f , g) ◦ g,
which gives a morphism Z → Y satisfying desired commutative property, and thus
by the universal property of K as a fibred product, there exists a morphism h0 : Z →
K such that ι ◦ h0 = h. □
 79

7.10. Homework-10.

7.10.1. Part I.
Exercise 7.10.1. Let X be an S -scheme, S 0 → S a morphism, X 0 = X ×S S 0 the base
change of X , ∆ : X → X ×S X and ∆0 : X 0 → X 0 ×S 0 X 0 the diagonal morphisms. Use
the result in Exercise 7 on page 58 of the textbook to prove the following diagram is
Cartesian:
∆0
X0 X 0 ×S 0 X 0

∆
X X ×S X

Proof. By (3) of Exercise 7 on page 58, it suffices to show for every scheme Z , the
following diagram is Cartesian

Hom( Z, X 0 ) Hom( Z, X 0 ×S 0 X )

Hom( Z, X ) Hom( Z, X ×S X ).

On the other hand, one has

Hom( Z, X ×S X ) = Hom( Z, X ) ×Hom( Z,S ) Hom( Z, X )
Hom( Z, X 0 ×S 0 X 0 ) = Hom( Z, X 0 ) ×Hom( Z,S 0 ) Hom( Z, X 0 ).
Thus it suffices to show the following diagram is Cartesian

Hom( Z, X 0 ) Hom( Z, X 0 ) ×Hom( Z,S 0 ) Hom( Z, X 0 )

Hom( Z, X ) Hom( Z, X ) ×Hom( Z,S ) Hom( Z, X ),

which is clear by the description of fibred product in the category of sets. □

Exercise 7.10.2. Let X and S be locally compact topological spaces, S is Haus-
dorff, and let f : X → S be a continuous map.
(1) Prove that a proper map is a closed map.
(2) Let S 0 be a locally compact topological space and let g : S 0 → S be a continuous
map. For any proper map f : X → S , prove the base change f 0 : X ×S S 0 → S 0 of
f is proper.
Proof. For (1). Let V ⊆ X be a closed subset. It suffices to show S \ f (V ) is open. For
s ∈ S \ f (V ), there exists an open neighborhood of U of s with compact closure since
S is locally compact. Then f −1 (U ) is compact since f is proper. Let E = V ∩ f −1 (V ).
Then E is compact since it’s a closed subset of a compact set, and hence f (E ).
Again by S is Hausdorff, one has f (E ) is closed in S . Then U \ f (E ) is an open
neighborhood of s which is disjoint from f (V ). This shows S \ f (V ) is open.
80

For (2). Note that the fibred product in the category of topological spaces can be
described as X ×S S 0 = {( x, y) ∈ X × S 0 | f ( x) = g( y)} and f 0 : X ×S S 0 → S 0 is given by
( x, y) 7→ y. Then for any compact subset K ⊆ S 0 , one has
( f 0 )−1 (K ) ⊆ f −1 g(K ) × K,
which is a compact subset since f is proper. □
7.10.2. Part II.
Exercise 7.10.3. The set
I = {( i 0 , . . . , i n ) | i 0 + · · · + i n = d, i 0 , . . . , i n ∈ Z≥0 }
¡ d + n¢
has n solutions. Prove that the homomorphism
φ : Z[ yi 0 ...i n ]( i 0 ,...,i n )∈ I → Z[ x0 , . . . , xn ]
i i
yi 0 ...i n 7→ x00 · · · xnn

n (d+n)−1
induces a morphism PZ → PZ n .
L
Proof. For convenience, we use A = n∈Z≥0 A n to denote the graded ring Z[ yi 0 ...i n ]( i 0 ,...,i n )∈ I .
Note that for each degree n, one has φ( A n ) ⊆ Z[ x0 , . . . , xn ]nd . This shows ring
homomorphism φ preserves the grade, and thus it induces a morphism between
Proj. □
 81

7.11. Homework-11.

7.11.1. Part I.
Exercise 7.11.1. Let X be a topological space, B a basis of topology for X , F and G
sheaves of abelian groups on X . Suppose for every member U in B , we are provided
with a homomorphism ϕ(U ) : F (U ) → G (U ) such that for every pair V ⊆ U in B ,
the following diagram commutes,
ϕ(U )
F (U ) G (U )

ϕ (V )
F (V ) G (V )
Prove that there exists a unique morphism of sheaves ϕ : F → G such that for any
member U in B , one has ϕ(U ) : F (U ) → G (U ) coincides with ϕ(U ).
Proof. For convenience we denote B = {Uα }α∈A and use ϕα to denote ϕ(Uα ). For
S
any open subset U ⊆ X , we write it as U = α Uα , since B is a basis for topology.
Then for any x ∈ U , if x ∈ Uα , then we define ϕ(U )( x) = ϕα ( x). This is well-defined
since if x ∈ Uα ∩ Uβ , then
ϕα ( x) = ϕ(Uα ∩ Uβ )( x) = ϕβ ( x).
This gives a unique morphism of sheaves ϕ : F → G such that for any U ∈ B , one
has ϕ(U ) : F (U ) → G (U ) coincides with ϕ(U ). □

7.11.2. Part II.

Exercise 7.11.2 (coherent sheaf on ringed space). Let ( X , O X ) be a ringed space.
An O X -module F is called coherent if F is of finite type and for every open subset
⊕n
U of X and every homomorphism u : OU → F |U , the kernel of u is of finite type.
(1) Suppose O X is coherent. Prove that an O X -module F is coherent if and only if
F is of finite presentation.
(2) Prove that this definition of coherence coincides with the one in Definition 6.2.3
for Noetherian schemes.
Proof. For (1). Firstly let’s show if F is coherent, then F is of finite presentation
(To prove this, we don’t need the assumption O X is coherent). Since F is of finite
type, there exists an open covering {U i } of X such that
⊕n i u i
OU −→ F |Ui → 0
i

is exact, and the ker u i is of finite type implies that there exists a refinement of
{U i }, denoted by {U i j } such that the following sequence is exact
⊕m i j ⊕n i
OU → OU → F |Ui j → 0.
ij ij

This shows F is of finite presentation. For the converse statement, firstly let’s
prove the following lemma.
82

Lemma 7.1. An O X -module F is coherent if and only if there exists an open

covering {U i } of X such that F |Ui is coherent.

Proof. It’s clear that if F is coherent, then every F |Ui is coherent. Conversely,
suppose there exists an open covering {U i } of X such that F |Ui is coherent. Firstly,
since each F |Ui is of finite type, there exists a refinement {U i j } of {U i } such that
the following sequences are exact
⊕n i j
O |U → F |Ui j → 0,
ij

⊕n
and thus F is of finite type. For every homomorphism u : OU → F |U , its restric-
tion gives the following homomorphism
⊕n
u i : OU ∩U i → F |U ∩U i ,

and the ker u i is of finite type since F |U ∩Ui is coherent, and thus by the previous
argument one has ker u is of finite type. □

Now suppose O X is coherent and F is of finite representation. Then there exists

an open covering {U i } such that the following sequence
⊕m i ⊕n i
OU → OU → F |Ui → 0
i i

is exact, and thus F |Ui is coherent, since the quotient of coherent sheaves is co-
herent3. Then by Lemma 7.1 one has F is coherent.
For (2). Suppose X is a Noetherian scheme. In this case, O X is coherent, and
thus is suffices to show the definition of coherence for Noetherian schemes in Def-
inition 6.2.3 is equivalent to finite presentation.
(a) If F is of finite presentation, then without lose of generality we may assume
there exists an affine open covering {U i = Spec A i } such that
⊕m i ⊕n i
OU → OU → F |Ui → 0
i i

⊕n ⊕n
is exact. Note that OU i itself is of the form M ∼
i
, where M i = A i i . This shows
i
F is coherent (in the sense of Definition 6.2.3).
( b) If F is coherent (in the sense of Definition 6.2.3), then there exists an affine
open covering {U i = Spec A i } such that
ui
M∼
i −→ F |U i → 0

is exact, where M i is a finitely generated A i -module. Since X is Noetherian,

one has A i is a Noetherian ring, and thus M i is also a Noetherian module since
it’s finitely generated. In particular, any submodule of M i is finitely generated,
and thus ker u i is of finite type.
□
3However, it’s also highly non-trivial with the definition of coherence given here, but I don’t want
to give a proof here. See Lemma 17.12.4 in [Sta23].
 83

Exercise 7.11.3. Let X and Y be S -schemes, let f , g : X → Y be S -morphisms, and

let U be a dense open subset of X such that f |U = g|U as morphisms of schemes.
Suppose Y is separated over S . Then f = g as maps on topological spaces. Suppose
furthermore that X is reduced. Then f = g as morphisms of schemes.
Proof. Consider the following diagram

X ×Y ×S Y Y Y
∆Y /S
h= f × S g
X Y ×S Y .

If Y is separated, then ∆Y /S (Y ) ⊆ Y ×S Y is a closed subset, and thus h−1 (∆Y /S (Y ))

is also a closed subset since h is continuous. On the other hand, suppose U ⊆ X is
the dense open subset such that f |U = g|U . Then U ⊆ h−1 (∆Y /S (Y )), and thus one
has h−1 (∆Y /S (Y )) = X by taking closure of U . As a consequence, one has f = g on
X as maps on topological spaces.
Now it suffices to show f ] , g] induce the same morphisms between sheaves, and
thus without lose of generality we may assume both X = Spec A and Y = Spec B
are affine schemes. For convenience we write f ] : O Spec A → f −1 O Spec B by using
adjoint between f ∗ and f −1 . If ker( f ] − g] ) ⊆ Spec A is of the form V (b), then one
has U ⊆ V (b), and thus V (b) = Spec A , since U is dense. As a consequence, b is the
radical of A , and thus b = 0 since A is reduced. □
84

7.12. Homework-12.
7.12.1. Part I.
Exercise 7.12.1. 1. Let f : X → Y be a continuous map between topological spaces,
let F and G be sheaves of abelian groups on X and Y , respectively, let δ : G → f ∗ F
be a morphism, and let ψ : f −1 G → F be the morphism induced by δ by adjunction.
For any x ∈ X , and δ x : G f ( x) → F x be the composite
δ f ( x)
G f ( x) −→ ( f ∗ F ) f ( x) = lim ( f ∗ F )(V )
−−→
f ( x)∈V

= lim F ( f −1 (V )) → lim F (U ) = F x ,
−−−→ −−→
x∈ f 1 (V ) x∈U

and let ψ x : G f ( x) → F x be the composite

ψx
G f ( x) ∼
= ( f −1 G ) x −→ F x .
Prove that δ x = ψ x .
Proof. For open subset U ⊆ X , recall that the adjoint morphism ψ(U ) is given by
the following composite
f −1 δ(U )
f −1 G (U ) −→ f −1 f ∗ F (U ) → F (U ).
By passing to stalks one has the following commutative diagram
δ f ( x)
G f ( x) ( f ∗ F ) f ( x) Fx
∼
= ∼
= ∼
=
−1
(f δ) x
( f −1 G ) x ( f −1 f ∗ F ) x Fx

This completes the proof. □

Exercise 7.12.2. Let ( f , f ] ) : ( X , O X ) → (Y , O Y ) be a morphism of locally ringed
spaces, F an O X -module, G an O Y -module, δ : G → f ∗ F a homomorphism of O Y -
modules, and ψ : f ∗ G → F the morphism induced by δ by adjunction. For any
x ∈ X , and δ x : G f ( x) → F x be defined as in above exercise, and let ψ x : O X ,x ⊗OY , f (x)
G f ( x) → F x be the composite
ψx
O X ,x ⊗OY , f (x) G f ( x) ∼
= ( f ∗ G ) x −→ F x .
Prove that
ψ x ( r ⊗ s) = r δ x ( s)
for any r ∈ O X ,x and s ∈ G f ( x) .
Proof. For open subset U ⊆ X , recall that the adjoint morphism ψ(U ) is given by
the following composite
id ⊗ f −1 δ(U )
f ∗ G (U ) = (O X ⊗ f −1 OY f −1 G )(U ) −→ (O X ⊗ f −1 OY f −1 f ∗ F )(U ) → F (U ).
 85

Similarly, by passing to stalks one has the following commutative diagram

id ⊗δ f ( x)
O X ,x ⊗OY , f (x) G f ( x) O X ,x ⊗OY , f (x) ( f ∗ F ) f ( x) Fx
∼
= ∼
= ∼
=
∗
( f δ) x
( f ∗ G )x ( f ∗ f ∗ F )x Fx

This completes the proof.

□
Exercise 7.12.3. Let ( X , O X ) be a scheme and F a quasi-coherent O X -module.
Suppose F is of finite presentation. Prove that for any affine open subset U ∼ =
Spec A , we have F |U ∼
= M ∼ for some A -module M with finite presentation.
Proof. Firstly, for any affine open subset U ∼ = Spec A , we may assume F |U ∼ = M∼
for some A -module M , so it suffices to show M is of finite presentation. Since F is
of finite presentation, there exists an open covering of U , without lose of generality
we may assume it’s given by {U i = Spec A f i }ni=1 with ( f 1 , . . . , f n ) = A , such that on
each U f one has the following exact sequence
⊕m i ⊕n i
OU → OU → F |Ui → 0,
i i

where m i , n i ∈ Z>0 . In other words, the localization M f i is of finite presentation

for each i = 1, 2, . . . , n, and thus M is of finite presentation. □
7.12.2. Part II.
Exercise 7.12.4. Let X be a scheme, L be an invertible O X -module with f ∈ L ( X )
and F be a quasi-coherent O X -module.
(1) Define X f to be the subset of X consisting of those points x ∈ X such that the
germ of f at x does not lie in m x L x , where m x is the maximal ideal of the local
ring O X ,x . Then X f is open.
(2) Suppose X is quasi-compact. If s ∈ F ( X ) is a section whose restriction to X f
vanishes, then there exists a natural number n such that the section s ⊗ f ⊗n ∈
(F ⊗O X L ⊗n )( X ) vanishes.
(3) Suppose X is quasi-compact and quasi-separated. Given a section t in F ( X f ),
there exists a natural number n such that the section t ⊗ f ⊗n in (F ⊗O X L ⊗n )( X f )
can be extended to a section in (F ⊗O X L ⊗n )( X ).
Proof. For (1). Since L is an invertible O X -module, there exists an open covering
{U i } of X such that L |Ui ∼ = OUi . If we use f i to denote f |Ui and use X f i to denote
the subset of U i consisting of those points x ∈ U i such that the germ of f i at x does
S
not lie in m x L x , then X f = i X f i , and by (1) of Proposition 4.3.1 one has each X f i
is open.
For (2). Since X is quasi-compact, we may assume {U i = Spec A i }ni=1 is a finite
affine open covering of X , and thus X f ∩ Spec A i = D ( f i ), where f i := f |Ui . Since
the restriction of s to X f vanishes, one has the restriction of s to each D ( f i ) =
86

Spec( A i ) f i vanishes for each i = 1, . . . , n, and thus there exists some n i ∈ Z>0 such
that f n i s = 0 in A i . Now it suffices to take n = max i { n i }, one has s ⊗ f ⊗n ∈ (F ⊗O X
L n )( X ) vanishes.
For (3). Since X is quasi-compact and quasi-separated, by (2) of Proposition
5.5.4 we may assume {U i = Spec A i }ni=1 is a finite affine open covering of X such
that each U i ∩ U j is a union of finitely many affine open subschemes. For conve-
nience we denote t i := t|Ui . Note that t i − t j vanishes on U i ∩ U j , and since U i ∩ U j
is a union of finite affine open subschemes, then by (2) there exists some n i j ∈ N
n
such that ( t i − t j ) ⊗ f i ji j ∈ (F ⊗O X L ⊗n i j )(U i ∩ U j ) vanishes. On the other hand,
for each t i ∈ F ( X f ∩ U i ), there exists some m i ∈ N such that t i ⊗ f m i extends to
(F ⊗O X L ⊗m i )( X f ∩ U i ). Since everything is finite, there exists a sufficiently large
n such that t ⊗ f n extends to a section in (F ⊗O X L ⊗n )( X ). □
 87

7.13. Homework-13.

Exercise 7.13.1. Let S be a graded ring which is generated by S 1 as an S 0 -algebra.

(1) Prove that for every f ∈ S 1 and every graded S -module M , we have an S ( f ) -
module isomorphism
M∞
M ( n )( f ) ∼
= Mf .
n=−∞

In particular, taking M = S , we get an S ( f ) -module isomorphism

M
∞ ∼
=
S ( n)( f ) −→ S f .
n=−∞

Prove that we have an S f -module isomorphism

M( f ) ⊗S ( f ) S f ∼
= Mf

(2) Let M and N be graded S -modules. Construct an isomorphism

∼
=
M( f ) ⊗S( f ) N( f ) −→ ( M ⊗S N )( f ) ,

so that after taking the tensor product ⊗S〈 f S f and composing with the isomor-
phisms in (1), we get the canonical isomorphism

M f ⊗S f N f ∼
= ( M ⊗S N ) f .

(3) Let p ∈ Proj S , and let T be the set of homogeneous elements in S \ p. Prove that
we have an S (p) -module isomorphism
M
∞ ∼
=
M ( n)(p) −→ T −1 M.
n=−∞

In particular, taking M = S , we get an S (p) -module isomorphism

M
∞ ∼
=
S ( n)(p) −→ T −1 S.
n=−∞

Prove that we have a T −1 S -module isomorphism

∼
=
M(p) ⊗S(p) T −1 S −→ T −1 M.

(4) Let M and N be graded S -modules. Construct an isomorphism

∼
=
M(p) ⊗S{ p} N(p) −→ ( M ⊗S N )(p)

so that after taking the tensor product ⊗S(p) T −1 S and composing with the iso-
morphisms in (3), we get the canonical isomorphism
∼
=
T −1 M ⊗T −1 S T −1 N −→ T −1 ( M ⊗S N ).
88

Proof. For (1). Consider the following homomorphism

M∞
ϕ: M ( n )( f ) → M f
n=−∞
mn X
∞ m
n
( ) 7→ .
f kn n=−∞ f kn
It’s well-defined since there are only finitely many m n / f k n 6= 0.
(a) Firstly let’s show ϕ surjective: Since for any m/ f k ∈ M f , we may write m =
P
i m i , where m i are homogeneous elements of degree i . If i = k + n, then
m i / f k ∈ M ( n)( f ) , and thus
mi X mi m
ϕ(( k )) = k
= k.
f i f f
P∞
( b) Now let’s prove ϕ is injective: If n=−∞ m n / f k n = 0 in M f , then there exists
some N ∈ N such that
X∞ m
n
fN k
= 0.
n=−∞ f n
Since deg m n = k n + n, then by degree reason f N m n / f k n = 0 for every n ∈ Z. On the
other hand, multiplying by f N gives an isomorphism from M ( n)( f ) to M ( n + N )( f )
for each n ∈ Z since f ∈ S 1 . This shows m n / f k n = 0 for each n ∈ Z, and thus ϕ is
injective. In particular, one has
M
∞ ∼
=
S ( n)( f ) −→ S f .
n=−∞
Moreover, since M ( n)( f ) ∼
= M( f ) ⊗S( f ) S ( n)( f ) as S ( f ) -modules, one has
M
∞
Mf ∼
= M ( n)( f )
n=−∞
M∞
∼
= ( M ( f ) ⊗ S ( f ) S ( n )( f ) )
n=−∞
∼
= M( f ) ⊗S ( f ) S f .
For (2). Consider the following homomorphism
ϕ : M( f ) ⊗S( f ) N( f ) → ( M ⊗S N )( f )
X mi ni X mi ⊗ ni
ki
⊗ →
7 k i +l i
.
i f f li i f
It’s clear that ϕ is surjective, so it suffices to check it’s injective. If
X mi ⊗ ni 1 X αi
k i +l i
= f mi ⊗ ni = 0
i f fN i
P
in ( M ⊗S N )( f ) , where N = max i { k i + l i } and k i + l i + α i = N for all i , then i f αi m i ⊗
n i = 0 in M ⊗S N , and thus
X mi ni
k
⊗ l =0
i f f i
i
 89

in M( f ) ⊗S( f ) N( f ) . This shows the injectivity. As a consequence, after taking the

tensor product ⊗S( f ) S f and composing with the isomorphisms in (1), we get the
canonical isomorphism
M f ⊗S f N f ∼ = ( M ⊗S N ) f .
For (3). Note that
M
∞ ( a) M∞
M ( n)(p) ∼= lim ( M ( n)( f ) )∼
n=∞ n=−∞ − −→
p∈ D + ( f )
M
∞
∼
= lim ( M ( n)( f ) )∼
−−→
p∈D + ( f ) n=−∞
µ ¶∼
M
∞
∼
= lim M ( n )( f )
−−→
p∈D + ( f ) n=−∞
∼
= lim ( M f )∼
−−→
p∈ D + ( f )
( b)
∼
= T −1 M,
where (a) and ( b) hold from S is generated by S 1 as an S 0 -algebra, and by the
same argument as (1) one can show
∼
=
M(p) ⊗S(p) T −1 S −→ T −1 M.
For (4). It follows from (2) by using the same modification as (3).
□
Exercise 7.13.2. Let X be a scheme, and Y be a closed subscheme of X with ideal
sheaf I . Define the blowing-up X e of X along Y to be the scheme Proj(L∞ I n )
L∞ L n=0
over X , where Proj( n=0 I n ) is obtained by gluing Proj ( ∞ n=0 I n
(U )) over open
affine subschemes U of X .
(1) Let π : Xe → X be the canonical morphism. Prove that π induces an isomorphism
∼
=
−1
π ( X \ Y ) −→ X \ Y .
(2) Let f : X 0 → X be a closed immersion, and let Y 0 = X 0 × X Y be the closed sub-
scheme of X 0 obtained from Y → X by base change. Prove that the blowing-up
Xe 0 of X 0 along Y 0 is isomorphic to the scheme theoretic image of the composite
−1
f π
e
X 0 \ Y 0 −→ X \ Y −→ π−1 ( X \ Y ) ,→ X
Proof. For (1). Suppose {Uα = Spec A α } is an affine open covering of X . If we write
e → X on the affine pieace Uα is induced by the natural
I α = I (Uα ), then π : X
inclusion M d
A α ,→ S α = Iα .
d ≥0
By previous homework (Exercise 7.5.3), one has π|π−1 (Uα ) : π−1 (Uα \ V ( I α )) → Uα \
V ( I α ) is an isomorphism. This shows π : π−1 ( X \ Y ) → X \ Y is an isomorphism. □
90

R EFERENCES
[Fu06] L. Fu. Algebraic Geometry. Mathematics series for graduate students. Tsinghua University
Press, 2006.
[Sta23] The Stacks project authors. The stacks project. https://stacks.math.columbia.edu,
2023.
[Vak17] Ravi Vakil. The rising sea: Foundations of algebraic geometry. preprint, 2017.