Math 256B Notes

James McIvor
April 16, 2012
Abstract
These are my notes for Martin Olssons Math 256B: Algebraic Ge-
ometry course, taught in the Spring of 2012, to be updated continually
throughout the semester. In a few places, I have added an example or
eshed out a few details myself. I apologize for any inaccuracies this may
have introduced. Please email any corrections, questions, or suggestions
to mcivor at math.berkeley.edu.
Contents
0 Plan for the course 4
1 Wednesday, January 18: Dierentials 4
1.1 An Algebraic Description of the Cotangent Space . . . . . . 4
1.2 Relative Spec . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Dual Numbers . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Friday, January 20th: More on Dierentials 7
2.1 Relationship with Leibniz Rule . . . . . . . . . . . . . . . . 7
2.2 Relation with the Diagonal . . . . . . . . . . . . . . . . . . 8
2.3 K ahler Dierentials . . . . . . . . . . . . . . . . . . . . . . . 10
3 Monday, Jan. 23rd: Globalizing Dierentials 12
3.1 Denition of the Sheaf of Dierentials . . . . . . . . . . . . 12
3.2 Relating the Global and Ane Constructions . . . . . . . . 13
3.3 Relation with Innitesimal Thickenings . . . . . . . . . . . 15
4 Wednesday, Jan. 25th: Exact Sequences Involving the
Cotangent Sheaf 16
4.1 Functoriality of
1
X/Y
. . . . . . . . . . . . . . . . . . . . . 16
4.2 First Exact Sequence . . . . . . . . . . . . . . . . . . . . . . 18
4.3 Second Exact Sequence . . . . . . . . . . . . . . . . . . . . 20
5 Friday, Jan. 27th: Dierentials on P
n
22
5.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 A Concrete Example . . . . . . . . . . . . . . . . . . . . . . 22
5.3 Functorial Characterization of
1
X/Y
. . . . . . . . . . . . . 23
1
6 Monday, January 30th: Proof of the Computation of
1
P
n
S
/S
24
7 Wednesday, February 1st: Conclusion of the Computation
of
1
P
n
S
/S
28
7.1 Review of Last Lectures Construction . . . . . . . . . . . . 28
7.2 Conclusion of the Proof . . . . . . . . . . . . . . . . . . . . 29
7.3 An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 31
8 Friday, February 3rd: Curves 32
8.1 Separating Points and Tangent Vectors . . . . . . . . . . . . 32
8.2 Motivating the Terminology . . . . . . . . . . . . . . . . . . 32
8.3 A Preparatory Lemma . . . . . . . . . . . . . . . . . . . . . 33
9 Monday, February 6th: Proof of the Closed Embedding
Criteria 34
9.1 Proof of Proposition 8.1 . . . . . . . . . . . . . . . . . . . . 34
9.2 Towards Cohomology . . . . . . . . . . . . . . . . . . . . . . 36
10 Wednesday, Feb. 8th: Cohomology of Coherent Sheaves
on a Curve 37
10.1 Black Box: Cohomology of Coherent Sheaves on Curves . . 37
10.2 Further Comments; Euler Characteristic . . . . . . . . . . . 39
10.3 Statement of Riemann-Roch; Reduction to the More Fa-
miliar Version . . . . . . . . . . . . . . . . . . . . . . . . . . 40
11 Friday, Feb. 10th: Proof and Consequences of Riemann-
Roch 41
11.1 Preparatory Remarks . . . . . . . . . . . . . . . . . . . . . 41
11.2 Proof of the Riemann-Roch Theorem . . . . . . . . . . . . . 42
11.3 Consequences . . . . . . . . . . . . . . . . . . . . . . . . . . 43
12 Monday, Feb. 13th: Applying Riemann-Roch to the Study
of Curves 44
12.1 Using Riemann-Roch to Detect Closed Immersions to P
n
. . 44
12.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
13 Wednesday, Feb. 15th: Where is the Equation? 46
14 Friday, Feb. 17th: Ramication and Degree of a Mor-
phism 49
15 Wednesday, Feb. 22nd: Hyperelliptic Curves 51
15.1 Canonical Embeddings . . . . . . . . . . . . . . . . . . . . . 51
15.2 Hyperelliptic Curves . . . . . . . . . . . . . . . . . . . . . . 52
16 Friday, Feb. 24th: Riemann-Hurwitz Formula 53
16.1 Preparatory Lemmata . . . . . . . . . . . . . . . . . . . . . 53
16.2 Riemann-Hurwitz Formula . . . . . . . . . . . . . . . . . . . 55
2
17 Monday, Feb. 27th: Homological Algebra I 56
17.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
17.1.1 Algebraic topology . . . . . . . . . . . . . . . . . . . 56
17.1.2 Algebraic geometry a.k.a. the Black Box . . . . . 56
17.2 Complexes, cohomology and snakes . . . . . . . . . . . . . . 57
17.3 Homotopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
18 Wednesday, Feb. 29th: Homological Algebra II 61
18.1 Additive and abelian categories . . . . . . . . . . . . . . . . 61
18.2 -functors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
18.3 Left- and right-exact functors . . . . . . . . . . . . . . . . . 65
18.4 Universality of cohomology . . . . . . . . . . . . . . . . . . 66
19 Friday, March 2nd: Homological Algebra III 67
19.1 Derived functors a construction proposal . . . . . . . . . . 67
19.1.1 Derived functors resolution of technical issues . . . 69
19.2 Derived functors conclusion . . . . . . . . . . . . . . . . . 71
19.3 Two useful lemmas . . . . . . . . . . . . . . . . . . . . . . . 72
20 Monday, March 5th: Categories of Sheaves Have Enough
Injectives 72
21 Wednesday, March 7th: Flasque Sheaves 75
22 Friday, March 9th: Grothendiecks Vanishing Theorem 77
22.1 Statement and Preliminary Comments . . . . . . . . . . . . 77
22.2 Idea of the Proof; Some lemmas . . . . . . . . . . . . . . . . 78
23 Monday, March 12th: More Lemmas 79
24 Wednesday, March 14th: Proof of the Vanishing Theorem 80
24.1 Base step . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
24.2 Reduction to X irreducible . . . . . . . . . . . . . . . . . . 81
24.3 Reduction to the case where T = ZU. . . . . . . . . . . . . 81
24.4 Conclusion of the Proof . . . . . . . . . . . . . . . . . . . . 83
25 Friday, March 16th: Spectral Sequences - Introduction 83
25.1 Motivating Examples . . . . . . . . . . . . . . . . . . . . . . 83
25.2 Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
25.3 Spectral Sequence of a Filtered Complex . . . . . . . . . . . 85
26 Monday, March 19th: Spectral Sequence of a Double Com-
plex 86
26.1 Double Complex and Total Complex . . . . . . . . . . . . . 86
26.2 Application: Derived Functors on the Category of Complexes 88
27 Wednesday, March 21st: Spectral Sequence of an Open
Cover 89
28 Friday, March 23rd:

Cech Cohomology Agrees with De-
rived Functor Cohomology in Good Cases 92
3
29 Monday, April 2nd: Serres Aneness Criterion 96
30 Wednesday, April 4th: Cohomology of Projective Space 99
31 Friday, April 6th: Conclusion of Proof and Applications 101
32 Monday, April 9th: Finite Generation of Cohomology 103
33 Wednesday, April 11th: The Hilbert Polynomial 106
0 Plan for the course
1. Dierentials
2. Curve Theory (Assuming Cohomology)
3. Build Cohomology Theory
4. Surfaces (Time Permitting)
Professor Olssons oce hours are Wednesdays 9-11AM in 879 Evans
Hall.
1 Wednesday, January 18: Dierentials
1.1 An Algebraic Description of the Cotangent Space
We want to construct the relative cotangent sheaf associated to a mor-
phism f : X Y . The motivation is as follows. A dierential, or du-
ally, a tangent vector, should be something like the data of a point in a
scheme, together with an innitesimal direction vector at that point.
Algebraically, this can be described by taking the morphism Spec k
Spec k[]/(
2
) and mapping it into our scheme. These two schemes have
the same topological space, and the nilpotency in k[]/(
2
) is meant to
capture our intuition of the innitesimal. More generally, we can take
any innitesimal thickening, i.e., a closed immersion T0 T dened
by a square-zero ideal sheaf J OT and map that into X. Of course,
we want this notion to be relative, so we should really map this thicken-
ing into the morphism f, which amounts to lling in the dotted arrow in
following diagram
T0


_

Y
which should be thought of as representing the following picture in the
case when Y = Spec k
picture omitted/pending
Example 1.1. Take Y = Spec k, T = Spec k and T0 = Spec k[]/(
2
).
Consider the set of morphisms lling in following diagram:
4
Spec k
xX

Spec k[]/(
2
)

Spec k
or, equivalently, k
OX,x

k[]/(
2
)
0

Commutativity of the square forces the composite k k to be the

identity, so the maximal ideal m OX,x maps to zero in k, hence along
the dotted arrow, m gets mapped into (), giving a new diagram
k OX,x/m
2
a

k[]/(
2
)
0

But k m/m
2

= OX,x/m
2
via the map (, ) b() + . More-
over, k m/m
2
can be given the structure of a k-algebra by dening
the multiplication (, ) (

) = (

). Also, as k-algebras,
k[]/(
2
)

= k k , so in fact we are looking for dotted arrows lling in
the following diagram of k-algebras:
k k m/m
2
a

k[]/(
2
)
0

Such arrows are determined by the map m/m

2
k. Thus the set of
arrows lling in diagram 1 is in bijection with (m/m
2
)

. This justies the

denition of the tangent space of X over k at x as (m/m
2
)

.
Greater generality can be achieved by working with X as a scheme over
some arbitrary scheme Y , and also by replacing the ring of dual numbers
k[]/(
2
) by a more general construction. Before doing so, we need the
notion of relative Spec.
1.2 Relative Spec
Given a scheme X and a quasicoherent sheaf of algebras / on X, we show
how to construct a scheme Spec
X
(/). Loosely speaking, the given sheaf
/ includes both the data of a ring over each ane open of X, and the
gluing data for these rings. We take Spec of each ring /(U) for opens U
of X, and glue them according to the specications of the sheaf /.
Alternatively, Spec
X
(/) is the representing object for the functor
F : (Schemes/X)
op
Set
which sends T
f
X to the set of OT -algebra morphisms f

/ OT . Two
things need to be checked, namely that this functor is representable at all,
5
and that the scheme obtained by gluing described above actually is the
representing object. Heres why its locally representable:
If X and T are ane, say X = Spec B, and T = Spec R, with / =

A
for some B-algebra A, then f

/ = (ABR)

, by denition of the pullback

sheaf. By the equivalence of categories between quasicoherent sheaves of
OT -modules and R-modules, the set of OT -algebra maps f

/ OT is in
natural bijection with the set of R-algebra maps AB R R, so we are
looking for dotted maps in the following diagram
AB R

R
R

w
w
w
w
w
w
w
w
w
w
But the vertical map on the left ts into a bered square as follows
A

AB R

R
B

w
w
w
w
w
w
w
w
w
w
Moreover, to give a map of R-algebras AB R R is equivalent, by the
universal property of the bered square, to giving a map A R along
the top:
A

AB R

R
B

w
w
w
w
w
w
w
w
w
w
This shows that in this case, the set of maps of OT -algebras f

/ OT
is in bijection with the maps of R-algebras A R, and hence that Spec A
represents the functor F. The remaining details of representability are
omitted.
In fact, the relative Spec functor denes an equivalence of categories
|ane morphisms Y X

|qcoh OX-algebras
Spec
X

where the map in the other direction sends the ane morphism Y
f
X
to fOY .
1.3 Dual Numbers
In the example of 1.1 we took a scheme X over k and mapped the in-
nitesimal thickening Spec k Spec k[]/(
2
) into X Spec k. In fact
this thickening can be globalized as follows. If X is a scheme and T a
quasicoherent sheaf of OX-algebras, dene a quasicoherent sheaf of OX-
algebras by
OX[T] = OX T
6
with the multiplicative structure given open-by-open by the rule
(a, b) (a

, b

) = (aa

, ab

+a

b)
just like before. We then have a diagram of OX-algebras
OX OX T
pr
1

OX

u
u
u
u
u
u
u
u
u
Id

where pr
1
denotes projection onto the rst factor and Id is the identity
map. We apply the relative Spec functor to this diagram, and dene the
scheme X[T] to be Spec
X
OX[T], which by the construction comes with
a diagram
X
Id

X[T]
z
z
z
z
z
z
z
z
X
Example 1.2. Consider the ane case, where B
f
A is a ring map, and
M an A-module. Here we are thinking of X in the above discussion as
the scheme Spec A over the scheme Spec B, and the dual numbers are just
A[M]

= A M, with multiplication dened as above. Thus the diagram
of OX-algebras above, when working over B, looks like
A A[M]

A
Id

z
z
z
z
z
z
z
z
B

A natural question to ask is: what maps along the diagonal, A A[M],
make this diagram commute? We take up this question in the next lecture.
2 Friday, January 20th: More on Dierentials
Given a ring map B A and A-module M, we form A[M] as last time,
with the dual multiplication rule. In the case A = M = k, we recover
the initial example: A[M]

= k[]/(
2
).
2.1 Relationship with Leibniz Rule
The module A[M] comes with maps of A-algebras
A

A[M]

A
whose composition is the identity. We would like to study sections of the
map A[M] A. First we need a denition.
7
Denition 2.1. If B A is a ring map and M is an A-module, a B-
linear derivation of A into M is a map : A M satisfying (b) = 0 for
all b in B, and (aa

) = a(a

) +a

(a) for al a, a

in A.
The second property is referred to as the Leibniz rule. Note also that
by (b), we really mean map b into A using the given ring map, and
then apply . The set of B-linear derivations of A into M is denoted
DerB(A, M), or often just Der(M) if the ring map B A is clear from
context. Note also that Der(M) itself has the structure of an A-module
(even though its elements are not A-linear maps!): if a is in A and is in
Der(M), dene (a )(x) = a (x), the dot on the right being the action
of A on M.
For an example, let B = R, A = C

(R), and M =
1
(R), the A-
module of smooth 1-forms on R. Then take = d, the usual dierential
operator. The conditions in this case say simply that d kills the constants
and satises the usual product rule for dierentiation.
Proposition 2.1. The set of A-algebra maps : A A[M] such that
= Id is in bijection with the set DerB(A, M).
Proof. The bijection is dened by sending a derivation to the map A
A[M] given by a a +(a). The resulting map is a section of since it
is the identity on the rst component; it is A-linear since it is additive in
both factors A and M, and for a, a

in A,
aa

aa

+(aa

) = aa

+a(a

) +a

(a) = (a +(a))(a

+(a

)),
the last equality as a result of the denition of the multiplication in A[M].
The map in the other direction sends an A-linear map : A A[M]
to (a) = (a) a. Note that the image of actually lies in M since
= Id.
Moreover, is a derivation since, writing (a) = a +m for some m in
M, we have
(aa

) = (aa

) aa

= (a)(a

) aa

= (a +m)(a

+m

) aa

= am

+a

m = a((a

) a

) +a

((a) a)
= a(a

) +a

(a)
2.2 Relation with the Diagonal
Now, to study these maps from a slightly dierent point of view, we
think of M above as varying over all A-modules, and construct a universal
derivation. To begin with, take Spec of the maps above, giving a diagram
Spec A

Id

L
L
L
L
L
L
L
L
L
L
Spec A[M]

Spec A
Now, putting the diagram over B, we are looking for such that the
following diagram commutes:
8
Spec A

Id

L
L
L
L
L
L
L
L
L
L
Id

Spec A[M]

Spec A

Spec A

Spec B
Let : Spec A Spec A B A denote the diagonal morphism, cor-
responding to multiplication A B A A. Then to give a morphism
above is equivalent to giving a map such that
Spec A

M
M
M
M
M
M
M
M
M
M
M
Spec A[M]

M
M
M
M
M
M
M
M
M
M
Spec AB A
pr
1

pr
2

Spec A

Spec A

Spec B
commutes, by the universal property of the ber product.
If the appearance of the diagonal is surprising, you should think of it
in the following way. The composite
X X Y X X
can be interpretted as expressing X as a retract of the diagonal, sort of
like a tubular neighborhood of X in X Y X. We think of the image of
a point x in X in X Y X as two nearby points, which then map back
down along the second map to the original point in X.
In any case, the above situation leads to the following algebraic prob-
lem. We are interested in : ABA A[M] (note that this is not exactly
the same as above) such that the following commutes:
A[M]

z
z
z
z
z
z
z
z
A AB A

A
aa1

0

Let I be the kernel of the multiplication A B A A. Then the

composite
I AB A

A[M]

A
is zero, by commutativity of the triangle above; hence the image of I lies
in M A[M], and there is an induced dotted map
9
A[M]

z
z
z
z
z
z
z
z
M

A AB A

J
J
J
J
J
J
J
J
J
I

.
A
aa1

0

Now since M
2
= 0 in A[M], I
2
A B A maps to 0 in A[M] under
any such , hence the desired must factor through A B A/I
2
. Thus
AB A/I
2
is the universal object of the form A[M], as M varies. In fact,
AB A

= A[I/I
2
], since the map A[I/I
2
] AB A/I
2
sending a + to
(a 1) + ts into the following map of exact sequences
0

I/I
2

A[I/I
2
]

0
0

I/I
2

AA/I
2
mult

A

0
so it is an isomorphism by the ve lemma.
Summarizing our results, we have established
Theorem 2.1. The following four sets are in natural bijection:
1. DerB(A, M)
2. |A-algebra maps : A A[M] [ = Id
3. |A-algebra maps [ A[I/I
2
]

K
K
K
A[M]
w
w
A

4. HomA(I/I
2
, M)
The bijection between 3 and 4 is given by sending to its restriction
to the second factor, I/I
2
. We can express the equivalence of 1 and 4 as
saying that the functor
Der: ModA ModA
as described above is represented by I/I
2
.
2.3 Kahler Dierentials
We have seen that every A-linear map A A[M] corresponds to a deriva-
tion : A M; the map A A[I/I
2
] is associated to the universal
derivation d: A I/I
2
, dened by
d(a) = a 1 1 a
The fact that this lies in I is implied by the following:
Lemma 2.1. The kernel I of the multiplication map A B A A is
generated by elements of the form 1 a a 1.
10
Proof. Clearly these elements are in I. Conversely, let x y be any
element of I. Then xy = 0, so (xy 1) = 0, giving x y = x(1
y y 1).
Theorem 2.2. The map d dened above is a derivation, and is universal
in the following sense: for any B-linear derivation : A M, there is a
unique A-linear map : I/I
2
M such that the diagram
A
d

B
B
B
B
B
B
B
B
I/I
2

M
commutes.
Proof. The universal property follows from the preceding discussion of
I/I
2
. It remains only to check that d is actually a derivation. For b in B,
we have
d(b) = 1 b b 1 = b 1 b 1 = 0
using the fact that the tensor product is taken over B. For the Leibniz
rule, rst note that I is generated by elements of the form 1 a a 1.
Then let a and a

be elements of A, and compute

d(aa

) a d(a

) a

d(a) = aa

1 1 aa

a(a

1 1 a

) a

(a 1 1 a)
= aa

1 1 aa

aa

1 +a a

a 1 +a

a
= 1 aa

aa

1 +a a

+a

a
= (a 1 1 a)(a

1 1 a

),
and this is zero in I/I
2
.
Remark. The above computation used the A-module structure on ABA
given by a (x y) = ax y. But there is also the action given by
a (x y) = x ay. Both of these actions, of course, restrict to I, but in
fact when working modulo I
2
, these two restrictions agree. This follows
from Lemma 1: if a is an element of A, then the dierence of the two
actions of a on I can be described as multiplication by (1 a a 1)
(using the usual multiplication for the tensor product of rings). But the
lemma shows that this is in I, and hence its action on I is zero modulo I
2
.
Professor Olsson used this in his computation that d satises the Leibniz
rule.
Denition 2.2. The A-module I/I
2
is called the module of (relative)
K ahler dierentials of B A, and denoted
A/B
. Note that it comes
together with the derivation d: A
A/B
.
In light of the previous discussion, one thinks of
A/B
as being the
algebraic analogue of the cotangent bundle. Here is an example to further
support this terminology.
11
Example 2.1. With notation as above, let B = k, and A = k[x, y]/(y2
x
3
5). Then

A/B

=
Adx Ady
2y dy 3x
2
dx
,
as you might expect from calculus. The universal derivaton d: A
A/B
is given by the usual dierentiation formula: df =
f
x
dx +
f
y
dy; the
quotient ensures that this is well dened when we work with polynomials
modulo y
2
x
3
5. The universal property is expressed by the following
diagram:
A

@
@
@
@
@
@
@
@
d

A/B

M
The map :
A/B
M guaranteed by the theorem sends dx and dy to
(x) and (y), respectively.
3 Monday, Jan. 23rd: Globalizing Dierentials
Today we will globalize the construuctions of last week, obtaining the
sheaf
1
X/Y
of relative dierentials associated to a morphism of schemes
X Y . We could do this by gluing together the modules
A/BB
of
last time, but instead we give a global denition, and then check that it
reduces to the original denition in the ane case.
Recall that for a ring map B I, letting I be the kernel of AB A
A, we dened
A/B
= I/I
2
. To relate this to todays construction, it will
be helpful to bear in mind the isomorphism
I/I
2

= I A
B
A A
which emphasizes the fact that this object is a pullback along the diagonal.
3.1 Denition of the Sheaf of Dierentials
Let f : X Y be a morphism of schemes. It induces a map
X/Y
: X
X Y X, from the following diagram:
X

X/Y

Id

Id

X Y X

X
f

X
f

Y
The map
X/Y
comes with a map of sheaves
#
X/Y
: OX
Y
X
(
X/Y
)OX on X Y X, which we will just abbreviate by . Although
really a map of sheaves of rings, we will regard it as a map of sheaves of
OXX-modules. Let be its kernel, again regarded as an OXX-module.
Denition 3.1. The sheaf of relative dierentials of X Y is the OX-
module
1
X/Y
=

.
12
Remark. Note that
X/Y
is not necessarily a closed immersion, so a
priori is just a sheaf of OX
Y
X-modules.
Now from the diagram
X

X/Y

Id

H
H
H
H
H
H
H
H
H
H
X Y X
pr
1

pr
2

X
we get maps of sheaves i : pr
1
i
OX OX
Y
X for i = 1, 2. Applying
the functor
1
X/Y
to these morphisms gives maps

1
pr
1
i
OX

i

1
OX
Y
X,
but by the commutativity of the diagram of schemes, we have
1
pr
1
i
OX

=
OX. Thus we have maps of sheaves on X,
1
i : OX
1
OX
Y
X.
Both maps (for i = 1, 2) agree when followed by the map
1
OX
Y
X
OX, so their dierence maps into the kernel of
1
OX
Y
X OX, giving

1
1
1
2 : OX ker(
1
OX
Y
X OX)
Recall that was dened as the kernel of OX
Y
X OX. Applying

1
, we have

1
=
1
ker(OX
Y
X OX)
Using the left-exactness of
1
, we can push it across the kernel and then
apply the adjunction between
1
and to get

1
= ker(
1
OX
Y
X OX)
Putting this together we now have a map

1
1
1
2 : OX
1

Now follow this with the map

1

=
1
X/Y
. We have now, at
great length, constructed the universal derivation d =
1
1
1
2 : OX

1
X/Y
, which is the global version of the map dA
1
A/B
in the ane
case.
3.2 Relating the Global and Ane Constructions
We now verify that our new denition of the cotangent sheaf agrees with
the ane denition locally, in the process showing further that
1
X/Y
is
quasicoherent.
Proposition 3.1. 1.
1
X/Y
is quasicoherent.
2. If f : X Y is locally of nite type and Y is locally Noetherian,
then
1
X/Y
is coherent.
13
Proof. Pick x in X, let y = f(x), take an open ane neighborhood Spec B
of y, and choose an open Spec A of X which lies in the preimage of Spec B
and contains x. This gives a commutative diagram
Spec A
open

X
f

Spec B
open

Y
mapping the inclusion Spec A X into the ber product gives a commu-
tative diagram
Spec A

A/B

open

Spec ASpec B Spec A

X/Y

X Y X
Let J = ker(AB A A). Our goal is to show that

1
X/Y

Spec A

= (J/J
2
)

,
which will prove simultaneously that
1
X/Y
is quasicoherent, and that it
agrees with our old denition in the ane case. To do so we make the
following computation

1
X/Y

Spec A

=
_

1
X/Y

1
O
XX
OX
_

Spec A
(1)

=
_

Spec A

(
1
O
XX
)

Spec A
OX

Spec A
(2)

= ker(
1
OX
Y
X OX)

Spec A

1
(A
B
A)
A

(3)

=
1
A/B
_
ker
_
(AB A)

__

1
(A
B
A)
A

(4)

=
1
J

1
(A
B
A)
A

(5)

(J

) (6)

=
_
J A
B
A A
_

(7)

=
_
J/J
2
_

(8)
The justications for the steps are as follows:
(1) Denition of
1
X/Y
(and denition of pullback sheaf)
(2) Tensor product commutes with restriction
(3) Denition of
(4) Left-exactness of
1
and commutativity of the square immediately
above this computation, and for the subscript on the , the fact
that
1
commutes with restriction
14
(5) Most importantly, the fact that ker
_
(ABA)

_
is quasicoher-
ent, which follows from the fact that in the ane case, the diagonal
is a closed immersion (note that , on the other hand, may not be
quasicoherent, precisely because in general, the diagonal map need
not be a closed immersion) Thanks to Chang-Yeon for pointing this
out to me
(6) Denition of pullback
(7) Pullback corresponds to tensor product over anes
(8) An observation made at the very beginning of this lecture
This proves part 1 of the theorm. Part 2 was an exercise on HW 1.
Note: At this point Prof. Olsson made some remark about the key
fact here is a certain property of the diagonal map ... I didnt write this
down, but if anyone remembers what he said here, please let me know.
Corollary 3.1. If B A is a ring map and f is an element of A, then
the unique map

1
A/B
A A
f

1
A

/B
is an isomorphism.
Proof. This follows immediately from quasicoherence, but heres an al-
ternate proof. Check that the localization of d: A
A/B
gives an
A
f
-linear derivation d
f
: A
f

A
f
/B
. Precomposing d
f
with the local-
ization map A A
f
gives a derivation A
A
f
/B
so by the universal
property of
A
f
/B
there is a unique A-linear map
A/B

A
f
/B
which
commutes with the localiztion map, d, and d
f
. Now one checks that this
map
A/B

A
f
/B
saties the universal property of the localization

A/B

A/B
A
f
. This also was assigned on HW 1, so I will not ll in
the details.
3.3 Relation with Innitesimal Thickenings
We motivated our discussion of dierentials with talk of mapping innites-
imal thickenings into a morphism X Y . Having now made our explicit
denition of the cotangent sheaf, its time we related it back to that initial
discussion. The following theorem says loosely that if there is a way to
do this at all, the number of ways to do so is controlled by the cotangent
sheaf.
First x some notation. Let T0 T be a closed immersion dened by
a square-zero quasicoherent (follows from closed immersion) ideal sheaf
. Although is a sheaf on T, the fact that
2
= 0 means the action
of OT on factors through that of OT
0
, so in fact we may regard as a
sheaf on T0 (recall that their topologial spaces are the same). Let x0 be
a point of X, and let S be the set of morphisms lling in the following
diagram
T0
i

x
0

X
f

T
y

Y
15
Theorem 3.1. Either S = or else there is a simply transitive action of
HomO
T
0
(x

1
X/Y
, ) on S.
Intuitively, this says that if we look at a point x in our scheme X,
although there is no preferred choice of tangent direction at X, once one
is chosen, we can obtain all the other by a rotation. In other words, the
dierence of two tangent directions is given by a vector in the cotangent
space.
We illustrate the theorem with a concrete example.
Example 3.1. Let k be an algebraically closed eld and X/k a nite-type
k-scheme, with X regular, connected, and of dimension 1. Thus if x is a
closed point, the local ring OX,x is a DVR.
Proposition 3.2. In the situation above,
1
X/k
is locally free of rank 1.
Proof. We know that
1
X/Y
is coherent by part 2 of Proposition 2. We
rst show that for any closed point x of X, the k-vector space
1
X/k
(x) =

1
X/k,x
_
mx
1
X/k,x
(equality since k is algebraically closed) is one-dimensional.
For this we calculate the dimension of the dual vector space Hom
k
(
1
X/Y
(x), k),
which by the pevious theorem is equal to the set of dotted arrows
Spec k
x

Spec k[]/
2

Spec k
and we calculated in Example 1 of Lecture 1 that this is equal to Hom
k
(mx/m
2
x
, k),
which is one-dimensional since OX,x is a DVR. Now, applying geometric
Nakayama (see Vakil 14.7D), we can lift a generator of
1
X/k
to some
ane neighborhood of x. Thus
1
X,k
is locally free of rank 1.
4 Wednesday, Jan. 25th: Exact Sequences In-
volving the Cotangent Sheaf
Before proving two sequences which give a further geometric perspective
on all weve done so far, it would be nice if the construction of
1
X/Y
were
functorial in some appropriate sense. Since this sheaf is associated to a
morphism X Y , we should consider a morphism of morphisms, i.e.,
a commutative diagram.
4.1 Functoriality of
1
X/Y
Theorem 4.1. Given a commutative diagram
X

Y

Y
there is an associated morphism g

1
X/Y

1
X

/Y
of sheaves on X

.
16
Proof. In the following diagram, the two outer trapezoids are the com-
mutative square given to us, and the upper left square is the usual ber
square
X

Y
X

A
A
A
A
A
A
A
A
X

Y
We get a map g g : X

Y
X

X Y X by the universal property of

X Y X, which also commutes with the relevant diagonal maps, giving
us the square
X

/Y

Y
X

gg

X/Y

X Y X
From the denition of we have a sequence

X/Y
OX
Y
X (
X/Y
)OX
of sheaves on XY X, and an analogous sequence of sheaves on X

Y
X

.
Applying (g g)

to the rst sequence gives

(g g)

X/Y
(g g)

O
X

Y
X
(g g)

(
X/Y
)OX
on X

Y
X

. Now commutativity of the square above shows that (gg)

/Y
=
X/Y
g, so (g g)

(
X/Y
)OX = (
X

/Y
) g

OX. But the

pullback of the structure sheaf along any morphism is always the structure
sheaf, so g

OX

= O
X
, and also (g g)

OX
Y
X

= O
X

Y
X
. This gives
us a diagram of sheaves on X

Y
X

:
(g g)

X/Y

O
X

Y
X

(g g)

(
X/Y
)OX

/Y

O
X

Y
X

(
X

/Y
)O
X

Even after applying (gg)

, the composite along the top is still zero, hence

the map (g g)

X/Y
O
X

Y
X

=
O
X

Y
X
(
X

/Y
)O
X
is
zero, so it factors through
X

/Y
, giving us a map (g g)

X/Y

/Y
. Now if we apply

/Y
to this morphism we get a map
g

1
X/Y
= g

X/Y

X/Y
=

/Y
(gg)

X/Y

/Y

X

/Y
=
1
X

/Y

which is what we were after.

17
In the ane case, this construction goes as follows. We are given a
diagram
Spec A

Spec A

Spec B

Spec B
which yields maps of rings
I

A

B
A

A

AB A

Here is just the restriction of to I; commutativity of the square on

the right ensures that it factors through I

.
First we form the A

B
A

-module I A
B
A
_
A

B
A

_
, which is
the analogue of the pullback (g g)

. Now we have an A B A-linear

map : I I

, and by the change of base adjunction

Hom
A

B
A

_
I A
B
A
_
A

B
A

_
, I

= HomA
B
A(I, I

)
this gives a map I A
B
A
_
A

B
A

_
I

of A

B
A

-modules, which
corresponds to the map (g g)

X/Y

X

/Y
in the above. Finally we
apply the functor
A

B
A
A

to this map, which gives a map

_
I A
B
A
_
A

B
A

_
_

B
A
A

B
A
A

But I

B
A
A

=
1
A

/B
by denition, and by changing base all over
the place,
_
I A
B
A
_
A

B
A

_
_

B
A
A

= I A
B
AA

=
_
I A
B
AA
_
AA

and this is
1
A/B
A A

, so we have produced the desired map. If you

work through the construction, you nd that this map is just da a

d(g(a)), where g is the given map A A

. In particular, this map is

surjective if g is.
4.2 First Exact Sequence
The following construction is what Vakil calls the relative cotangent exact
sequence (Thm 23.2.24).
Theorem 4.2. Given morphisms of schemes X
f
Y
g
Z, there is an
exact sequence of sheaves on X:
f

1
X/Y

1
X/Z

1
Y/Z
18
Proof. The result essentially follows from the following diagram
X

X/Y

X/Z

H
H
H
H
H
H
H
H
H
H
X Y X

X Z X

Y/Z

Y Z Y
in which the square is cartesian - it is a magic square (see Vakil 2.3R and
take X1 = X2 = X). Lets see what this says in the ane case. Then the
given morphisms correspond to ring maps A B C, and the diagram
above looks like
A AB A

AC A

B

G
G
G
G
G
G
G
G
G
G
B C B

We have a right exact sequence

ker(B C B B) B C B B 0
to which we apply the right exact functor B
C
B AC A, giving us
_
ACA
_
B
C
B
_
ker(BCB B)
_

_
ACA
_

_
ACA
_
B
C
BB 0
Here we have changed base at the middle term:
_
AC A
_
B
C
B
_
B C
B
_

= A C A. This is where the magic square comes in: it says that
A B A, which sits at the top left of that square, satises the universal
property of the tensor product of B and A C A over the base B C B,
so our sequence becomes
_
AC A
_
B
C
B
_
ker(B C B B)
_

_
AC A
_
AB A 0
Now the term on the left is already what we want; the other two terms
sit in their own sequences
A A
_
AC A
_

_
ker(B B B)
_

AC A

AB A

0
ker
_
AC A A
_

ker
_
AB A A
_

There is an induced map ker

_
A C A A
_
ker
_
A B A A
_
.
We would like to also nd a map
_
A C A
_

_
ker(B B B)
_

ker
_
AC A A
_
. For this we have to show that the composite
_
AC A
_

_
ker(B B B)
_
AC A A
19
is zero. But this is clear if you remember that we just changed the
base on the middle term. So the map on the left is just Id i, where
i :
_
ker(BB B)
_
BB is the inclusion. Then we follow this with
multiplcation, mapping us into A. But the magic square above show that
anything in the kernel of B B B maps to 0 in A. So the leftmost
map factors through the kernel of the center column, and we obtain our
sequence
_
ACA
_
B
C
B
_
ker(BCB B)
_
ker
_
ACA A
_
ker
_
ABA A
_
0
which by denition is
f

1
C/B

1
C/A

1
B/A
0
4.3 Second Exact Sequence
The next Theorem describes how dierentials behave under a closed im-
mersion. Of course, since the construction is relative, we should look at a
closed immersion X Y over some base scheme Z.
Theorem 4.3. Let
X
i

@
@
@
@
@
@
@
@
Y

Z
be a diagram of schemes, where i is a closed immersion dened by the ideal
sheaf 1 OY . Then the map 1 OY
d

1
Y/Z
induces an OX-linear
map i : i

1 i

1
Y/Z
which ts into an exact sequence
i

1
d
i

1
Y/Z

1
X/Z
0
Proof. (This is the corrected version, actually given in the following lec-
ture) Well prove it only in the ane case. In this setting, we have a
diagram
Spec A/I
i

K
K
K
K
K
K
K
K
K
K
Spec A
f

Spec B
and we wish to exhibit an exact sequence
I/I
2 d

1
A/B
A A/I
1
A/I/B
0
We claim that the sequence can be found buried in the belly of this
beast:
20
I A
B
A A
afaffa

ker(AB A A)

V
V
V
V
V
V
V
V
V
V
V
V
I AAI

U
U
U
U
U
U
U
U
U
U
U
U
U
U
ker(AB A A/I)

ker(A/I B A/I A/I)
I

0
Here is the map dened by a f + a

af + a

, extended by
linearity.
I havent worked through this yet. Ill come back to it. In case I forget,
the argument in Matsumura is nice and short. Read that instead.
Example 4.1 (Jacobian Matrix Calculation). Let B A be a nite type
ring map, with A and B Noetherian. We compute
1
A/B
as follows. First
write
A = B[x1, . . . , xn]/(f1, . . . , fs)
Then we have morphisms
Spec A

A
n
B

Spec B
We have computed already that
1
B[x
i
]/B
=

= Bdx1 Bdxn, and

pulling this back to Spec A is the same as tensoring up with A, giving

1
B[x
i
]/B

= Adx1 Adxn
Now by the previous proposition, this surjects onto
1
A/B
with kernel
I/I
2
, where I = (f1, . . . , fs). Furthermore, there is an obvious surjection
A
s
I/I
2
, sending the i-th basis element of A
s
to fi. Since precomposing
with a surjection does not change the cokernel, the sequence
A
s
Adx1 Adxn
1
A/B
expresses
1
A/B
as the cokernel of the map A
s
Adx1 Adxn.
But consider the image of a basis element ei under this map. It maps
rst to fi mod I
2
. Recall that the map I/I
2
sends the class of f to
df =

f
x
i
dxi. Thus in terms of the bases |ei for A
s
and | dxj for
Adx1 Adxn, this map is represented by the matrix
_
f
i
x
j
_
i,j
, the
familiar Jacobian matrix. This is the key to computing
1
A/B
in many
applications.
21
5 Friday, Jan. 27th: Dierentials on P
n
Remark. The rst problem on HW 1 should have convinced you that at
least in algebraic geometry, inseparable eld extensions are not patholo-
gies, despite what your Galois Theory course may have led you to believe.
One common situation in which they arise is the following. Say we have a
nite type k-scheme X, where k is algebraically closed, and a morphism
X Y whose bers are curves. Then one is led to consider the map of
elds k(x) k(y), where k(y) is typically not a perfect eld. It may be,
for example, of the form k(t1, . . . , tn), such as appeared on the assignment.
5.1 Preliminaries
Anyway, today will will compute the cotangent sheaf of P
n
S
. There are
many more concrete proofs (e.g., in Hartshorne), but we will take the
functorial approach. Recall that P
n
S
represents the functor
_
Schemes/S
_
op
Set
sending
_
T S
_
to the set of all surjections O
n+1
T
/, where / is a
locally free sheaf of rank one on T, up to isomorphism. In particular,
plugging P
n
S
itself in to this functor, we get an isomorphism
h
P
n
S
(P
n
S
)

= F(P
n
S
),
and the image of Id
P
n
S
under this isomorphism is the surjection
O
n+1
P
n
S
O
P
n
S
(1)
where O
P
n
S
(1) is the so-called twisting sheaf. A HW problem from last
term computed the abelian group structure of Pic(P
n
S
) under tensor prod-
uct. Applying O
P
n
S
O
P
n
S
(1) (which is right exact) to this surjection
gives another surjection
O
P
n
S
(1)
n+1
O
P
n
S
The following theorem asserts that the cotangent sheaf of P
n
S
is the kernel
of this surjection.
Theorem 5.1.
1
P
n
S
/S
is locally free of rank n, and there is an exact
sequence
0
1
P
n
S
/S
O
P
n
S
(1)
n+1
O
P
n
S
0
5.2 A Concrete Example
To warm up, lets take S = Spec k, with k algebraically closed. We have
seen that cotangent vectors can be regarded as dotted maps lling in the
following diagram
Spec k

P
n
k

Spec k[]/
2

Spec k
22
Since k is algebraically closed, the map along the top is given by choos-
ing a (closed) point [a0 : . . . : an] of P
n
k
.
To give a map from k[]/
2
into P
n
k
is to give a point [a0 + b0 : . . . :
an + bn] which agrees with the rst point modulo (since the vertical
map on the left is reduction mod . So since the ais have already been
chosen, lling in the diagram entail choosing the bis. But we could rescale
our coordinates by an element of k[]/
2
and possibly still get the same
map. Since (ai + bi)(x + y) = aix + (bix + aiy), we are forced to take
x = 1 in order to get the diagram to commute. So the map is determined
by choosing the bis, and the ambiguity in doing so can be expressed as
the corkernel of the map sending y (a0y, . . . , any). Thus we conclude,
loosely speaking, that the (co)tangent space is the cokernel of a map
k k
n+1
.
To relate this to the theorem, take the dual of the sequence which
appears there:
0 O
P
n
S
O
P
n
S
(1)
n+1

1
P
n
S
_

0
where the nal term is by denition the tangent sheaf of P
n
S
over S. This
is the coordinate-free expression of our computation above.
Compare this with the topological desription of complex projective
space. There we have a C

-bundle
C
n+1
\ |0
f

CP
n
and if p is a nonzero vector in C
n+1
, the tangent space to C
n+1
at p is just
C
n+1
, whereas the tangent space of CP
n
at f(p) is isomorphic to C
n
, and
we think of this as being the tangent space of C
n+1
module the tangent
space of the ber over f(p), which is just a line:
0 T
f
1
f(p)
(p) T
C
n+1
\{0}
(p) T
CP
n(f(p)) 0
0 C C
n+1
C
n
0
5.3 Functorial Characterization of
1
X/Y
Before we tackle the proof, lets ask what functor
1
X/Y
represents. Weve
basically already done this. Given X Y , a morphism of schemes, dene
T
X/Y
: QCoh(X) Set
by sending T to HomO
X
(
1
X/Y
, T). Given a quasicoherent sheaf T, we
saw the other day how to construct the following diagram
X
Id

X[T]
s
0

z
z
z
z
z
z
z
z

Y
23
Recall that, by analogy with the usual dual numbers, the scheme X[T]
comes equipped with morphisms of sheaves OX OX T OX. In
particular, there is a distinguished arrow s0 lling in the diagram, though
of course there may be others. Since we have a distinguished choice of
arrow, Theorem 3.1 says that the set of all such arrows is isomorphic
to the set of morphisms
1
X/Y
, T). In other words, the content of that
theorem is really that
1
X/Y
represents this functor T
X/Y
.
6 Monday, January 30th: Proof of the Computa-
tion of
1
P
n
S
/S
Today we aim to prove Theorem 5.1. We will do this in three steps, the
second of which is the most involved. Remember that we have to show
that
1
P
n
S
/S
ts into an exact sequence
0
1
P
n
S
/S
O
P
n
S
(1)
n+1
O
P
n
S
/S
0.
Following the discussion of last time, all that remains is to check that

1
P
n
S
/S
is the kernel of the map on the right. Let / be this kernel. The
following is an outline of the argument:
1. For any quasicoherent sheaf T on P
n
S
,
coker
_
om(O
P
n
S
/S
, T) om(O
P
n
S
/S
(1)
n+1
, T)
_

= om(/, T)
2. For any quasicoherent sheaf T on P
n
S
,
coker
_
om(O
P
n
S
/S
, T) om(O
P
n
S
/S
(1)
n+1
, T)
_

= om(
1
P
n
S
/S
, T)
3. By 1 and 2, om(/, T)

= om(
1
P
n
S
/S
, T), and therefore /

=
1
P
n
S
/S
Step 3 is basically just an application of Yoneda: once weve shown
that om(/, T)

= om(
1
P
n
S
/S
, T), then taking global sections gives
HomO
P
n
_
/, T
_

= HomO
P
n
_

1
P
n
S
/S
, T
_
, which implies
1
P
n
S
/S

= /, n-
ishing the proof.
Now we prove the rst two steps.
Proof of Step 1. We already have an exact sequence
0 / O
P
n
S
(1)
n+1
O
P
n
S
/S
0
We claim that application of the functor om gives another exact
sequence of sheaves on P
n
S
:
0 om(O
P
n
S
, T) om(O
P
n
S
(1)
n+1
, T) om(/, T) 0
The exactness follows from the following lemma
Lemma 6.1. Let
0 / c c

0
be an exact sequence of quasicoherent sheaves on X, where c

is locally
free of nite rank. Then for all T in QCoh(X), the sequence of sheaves
on X
0 om(c

, T) om(c, T) om(/, T) 0
is exact in the category of OX-modules.
24
Proof of Lemma. Exactness can be checked on stalks. Since c

is locally
free of rank r for some r, every point of X has a neighborhood U on
which c

= O
r
U
. Therefore replacing X by U we may assume c

= O
r
X
,
generated by basis elements ei in (X, c

), i = 1, . . . , r. Then dene
a section by lifting each ei to one of its preimages in (X, c), giving a
splitting
c

= / c

Then
om(c, T)

= om(/ c

, T)

= om(/, T) om(c

, T)
which implies the exactness of the sequence in the statement of the lemma.
Now, taking c = O
P
n
S
(1)
n+1
and c

= O
P
n
S
(which is locally free of
rank one), we have proved step 1.
Step 2 is the heart of the proof, and we will not nish it today. Before
we begin, here are a few facts which we will be useful to bear in mind.
Our approach involves the fact that P
n
S
represents the functor
F :
_
Schemes/S
_
op
Set
which sends an S-scheme T S to the set of surjections of OT -modules
O
n+1
T
/ where / is a line bundle on T, up to some equivalence relation
that we wont worry about here. But recall the way in which this F is
actually a (contravariant) functor: given a morphism
T

@
@
@
@
@
@
@
T
,






S
of S-schemes, the induced morphism F(f) sends the surjection O
n+1
T

/
in F(T) to the surjection O
n+1
T

/ in F(T

).
Why is g

still a surjection? Because g

is right exact, being a left

adjoint to the pushforward g. Why is g

/ a line bundle? Because

pullback sends vector bundles to vector bundles: this can be checked
locally, and if T

= Spec A, T = Spec B, and /

= (B

)
n
, we have
g

/

= (B

)
n

(B

)
A

= (B
n
B A)

= (A

)
n
as B

-modules. Now
we beign the proof.
Proof of Step 2. Let U be any open of P
n
S
, and T any quasicoherent sheaf
on P
n
S
. Then om
_

1
P
n
S
, T
_
(U) = HomO
U
_

1
P
n
S

U
, T

U
_
, and by Theo-
rem 3.1, this set is equal to the set of dotted maps lling in the following
diagram:
U

P
n
S

U[T]

S
25
Notice that in applying the theorem, we have used the fact that this set is
non-empty: we have a distinguished retract r : U[T] U, which when
composed with the inclusion U P
n
S
, lls in the diagram.
Now we consider such diagrams from the point of view of the functor F
which is represented by P
n
S
. As noted above, a map T P
n
S
corresponds
to a surjection of OT -modules O
n+1
T
/, with / a line bundle on T. So
the inclusion U P
n
S
corresponds to a section O
n+1
U
/U, and since the
inclusion commutes with the identity map P
n
S
P
n
S
, this line bundle /U
is actually just O
P
n
S
(1)

U
, but well continue to denote it by /U for now.
We also have our distinguished dotted arrow, the retract r (followed
by inclusion into P
n
S
), which corresponds to a surjection O
n+1
U[F]
/

.
Notice that O
U[F]

= r

OU since pullback sends the structure sheaf to the

structure sheaf. So our second surjection really looks like r

O
n+1
U
/

.
But the fact that the square above commutes imposes some constraints on
these two surjections. Namely, in light of our review of the functoriality
of F, we must have a commuting square
r

O
n+1
U
r

/U

O
n+1
U

/U
Now we are asking whether there are other maps which ll in the square of
schemes above, or equivalently, ll in this square of sheaves, and the func-
toriality therefore requires us to look for not just any surjection r

O
n+1

, but specically a surjection onto r

/U. So what can be said about

this r

/U? Firstly, it ts into an exact sequence of OU-modules

0 /U T r

/U /U 0
The reason is that r

/U can be rewritten as follows:

r

/U = O
U[F]
O
U
/U = /U O
U
(OU T

U
) = /U (/U O
U
T),
and this ts into the (split) exact sequence above. But the map on the
right is just the right hand vertical arrow in our square of sheaves two
paragraphs up, so we can t the two diagrams together, giving
0

/U T

r

/U

/U

0
O
n+1
U
T
n+1

= r

O
n+1
U
r

O
n+1
U

All the maps here are OU-linear, and in addition, r

is O
U[F]
-linear.
We ask: what other O
U[F]
-linear maps are there that ll in this diagram?
Observe rst that by commutativity of the square any two such must
agree when composed with the map r

/U /U killing T. Thus by
the exactness of the horizontal sequence their dierence factors through
/U O
U
T, i.e., is a map O
n+1
U[F]
/U T. So we can produce the desired
26
maps by adding to r

an O
U[F]
-linear map : O
n+1
U[F]
/U T. This
associates to each element of
HomO
U[F]
_
O
n+1
U[F]
, /U O
U
T
_
a dotted arrow lling in our diagram. But by changing base to the category
of OU-modules, this Hom set is in natural bijection with
HomO
U
_
O
n+1
U
, /U O
U
T
_
,
where /U O
U
T is regarded simply as an OU-module. Thus given any
element of this set, we have produced a surjection O
n+1
U[F]
r

/U of
sheaves on U[T].
Now recall that the line bundle /U is actually OU(1), and that maps
lling in the diagram correspond to elements of HomO
U
_

1
P
n
S

U
, T
_
, so
we have produced a map
HomO
U
_
O
n+1
U
, OU(1) O
U
T
_
HomO
U
_

1
P
n
S

U
, T
_
But with a little calculation, we can rewrite the domain as follows:
HomO
U
_
O
n+1
U
, OU(1) O
U
T
_

= HomO
U
_
OU(1)
n+1
O
U
OU(1), OU(1) O
U
T
_

= HomO
U
_
OU(1)
n+1
, HomO
U
_
OU(1), OU(1) O
U
T
__

= HomO
U
_
OU(1)
n+1
, T
_
The nal isomorphism being because OU(1) is locally isomorphic to OU.
For more detail, see the lemma below.
Summarizing, we now have a map
HomO
U
_
OU(1)
n+1
, T
_
HomO
U
_

1
P
n
S

U
, T
_
for every open U P
n
S
, which is compatible with restriction maps by
construction. Thus it determines a morphism of sheaf-Homs
omO
P
n
_
O
P
n(1)
n+1
, T
_
omO
P
n
_

1
P
n
S
, T
_
The rest of the proof of Part 2 will be nished in the next lecture.
Heres a lemma which was referred to above, and which gets used in the
next lecture also, but was neither explicitly stated nor proved in lecture.
Lemma 6.2. Let / be a line bundle on X, and T a quasicoherent sheaf
on X. Then there is an isomorphism of sheaves
: om(/, / O
X
T)

T
Proof. Choose a covering of X on which / is trivial and T is the sheaf
associated to some module, i.e., consisting of open anes U = Spec A, on
which we have isomorphisms
/

= OU

=

A and T

=

F,
27
for some A-module F. Then we have, by the equivalence of categories
QCoh(U)

= ModA,
om(/, / O
X
T)(U)

= HomA(A, AA F)
using which is given explicitly by the following: if HomA(A, AAF),
and (1) = 1f (we can always write it like this since the tensor product
is taken over A), we dene () = f. We need to check that this denition
of agrees on the intersection W of two anes U and V . But in this case,
we may restrict further if necessary to ensure that (T

U
)

= (T

V
)

W
.
Moreover, the isomorphisms /

= OU and /

= OV , when restricted
to W, dier by multiplication by a unit in OW. Since this unit appears
in both sides of the equation dening (specically, we multiply the 1 in
both sides by this unit), it may be cancelled, so that the equations dening
on U and on V are the same when restricted down to W. Thus the
denition of on open anes glues to give a morphism of sheaves, which
is an isomorphism of sheaves since it is so locally: HomA(A, A A F)

=
HomA(A, F)

= F.
7 Wednesday, February 1st: Conclusion of the
Computation of
1
P
n
S
/S
7.1 Review of Last Lectures Construction
By the end of the last lecture, we claimed to have produced, for each T
in QCoh(P
n
S
), a morphism of sheaves
: om(O
P
n
S
(1)
n+1
, T) om(
1
P
n
S
/S
, T).
Lets rst clarify the construction of this map . Consider an ane
open Spec R P
n
S
, with a sheaf T

=

F, for an R-module F. We are
interested in dotted arrows
Spec R

Id

P
n
S

Spec R[F]

r

S
Spec R
The inclusion Spec R P
n
S
corresponds to a surjection O
n+1
R

/,
where / is an invertible sheaf on Spec R.
1
We observed that in fact
/ = O(1), and that r

/

= / (T O
R
/) as OR-modules. This implies
that there is an exact sequence
0 T O
R
/ r

/ / 0,
1
In the following, we will abbreviate O
Spec R
by O
R
, and similarly for R[F].
28
in which the map r

/

= / (T /) / is projection onto the rst
factor; we t this into the following diagram involving and a potential
lifting :
0

T O
R
/

r

/

/

0
O
n+1
R[F]

O
n+1
R

If ei are basis elements of O

n+1
R
, then these lift to basis elements of
O
R[F]
[Reason: elements of F R[F] are nilpotent, hence contained in
the nilradical, hence the radical of R[F]. Thus Nakayama says we can lift
generators of R[F]/F = R to generators of R[F]]. Abusing notation and
denoting these lifts also by ei, the commutativity of the square means that
for any lling in the diagram, we must have (ei) = ((ei), i) for some
i in / F. Thus is determined by these i, so the set of maps lling
in the diagram is in bijection with n-tuples of sections of / T, which
is in turn in bijection with Hom
_
(/
1
)
n+1
, T
_
,according to the following
computation:
(/ T)
n+1

= Hom(O
n+1
, / T)

= Hom
_
(/
1
)
n+1
/, / T
_

= Hom
_
(/
1
)
n+1
, T
_
,
where we used Lemma 6.2 at the last step.
This construction is compatible with restriction, so this construction
of gives a map of sheaves
om(O
P
n(1)
n+1
, T) |maps U P
n
lling in the diagram above.
(We used /
1

= O
P
n(1) here). Now the sheaf on the right is a priori just
a sheaf of sets, but one can check that there is an O
P
n module structure
on it, in such a way that it is isomorphic as sheaves to om(
1
P
n, T).
Now we consider this map
: om(O
P
n
S
(1)
n+1
, T) om(
1
P
n
S
/S
, T).
7.2 Conclusion of the Proof
To nish the proof of Step 2, it suces to check that is surjective and
that its kernel is om(O
P
n, T). This shows that om(
1
P
n
S
/S
, T) is the
cokernel of om(O
P
n, T) om(O
P
n
S
(1)
n+1
, T) because sheaves of
O
P
n-modules form an abelian category, so every surjection (here, ) is the
cokernel of its kernel.
For surjectivity, we start with an open U P
n
, and consider a surjec-
tion

in the following diagram:

O
U[F]
n+1

O
n+1
U

/
29
Remember that our construction began with the canonical surjection
O
n+1
U
OU(1) and lifted it to a surjection O
n+1
U[F]
r

/. Thus a map

as above comes from our construction if and only if there exists an

isomorphism : /

/ over /, i.e., a diagram

/

@
@
@
@
@
@
@
r

|
|
|
|
|
|
|
|
/
which must furthermore t into the following:
O
n+1
U[F]

/ = / (/ T)
_q
q
q
q
q
q
q
q
q
q
q
q
q
O
n+1
U

/
We exhibit this locally: rst nd an open ane on which all three of
/, /

, and r

/ are trivial. Choose a basis for /, lift it via the surjections

/ / and r

/ in such a way that the triangle above commutes (this can

always be done, since locally, all three are isomorphic to some ring R).
Then our choice of bases for /

and r

/ determines locally. This proves

surjectivity.
To determine the kernel, we ask: when do two maps 1, 2 : O
P
n(1)
n+1

T determine the same : O

n+1
U[F]
r

/?. In other words, for which 1, 2

is there an isomorphism : r

/ r

/ lling in the following diagram

r

K
K
K
K
K
K
O
n+1
U[F]
+
2

o
o
o
o
o
o
+
1

O
O
O
O
O
O
/
r

s
s
s
s
s
s
Any isomorphism is given by multiplication by some unit in O

U[F]
; the
requirement that it be an isomorphism over / forces this unit to map to
1 in OU after collapsing T. In other words, such units have the form
1 + x O
U[F]
, where x T; in more otherer words, the sections of T

U
inject into the kernel of the map of sheaves of groups O

U[F]
O

U
.
Now for any section x of T

U
, we consider the maps
O
n+1
U[F]
+
2

+
1

L
L
L
L
L
L
L
L
L
L
/U (/ T)
(1+x)

/ (/ T)
30
Commutativity requires that for each basis element ei of O
U[F]
,
((ei), 1(ei)) = ((ei), (ei) x +2(ei)).
Thus the two maps agree if 1(ei) 2(ei) = (ei) x. So the kernel of
consists of these sections x of T; that is, ker

= T

= om(OU, T).
We regard this latter as a subsheaf of om(Ou(1)
n+1
, T) by applying
the functor om(, T) to the surjection OU(1)
n+1
OU, giving an
inclusion om(O, T) om(OU(1)
n+1
, T). Since our choice of U
was arbitrary and the constructions are compatible with restriction, the
conclusions hold with U replaced by P
n
S
, concluding the proof.
7.3 An Example
Wed like to compute
1
P
1
k
/k
for a eld k. We know its a line bundle, and
we know what the line bundles on P
1
are. But which line bundle is it?
Theorem 5.1 gives us the exact sequence
0
1
P
1
k
/k
O
P
1
k
(1)
2
O
P
1
k
/k
0.
Now since O is free, the sequence splits, so the middle term is isomor-
phic to the direct sum of the outer two. Hence the second exterior power
of the middle term is isomorphic to the tensor product of the outer two
terms. This means that

2
O(1)
2

=
1
P
1
k
/k
O

=
1
P
1
k
/k
,
but
2
O(1)
2

= O(2), so we have calculated

1
P
1
k
/k

= O(2)
More concretely, if we look on the ane open A
1
P
1
, we have
1
P
1

=

1
A
1

= O
A
1, generated by, say, dx. Similarly on the other chart A
1

=
D(y): there
1
P
1 is generated by dy. How are the two related? At
D(y) \ D(x), the local ring O
P
1
,
is a DVR with uniformizer x = 1/y,
giving dx = 1/y
2
dy. Thus the transition map for
1
is multiplication
by x
2
= 1/y
2
. The transition map for O(1) is multiplication by y. Thus

1
P
n

= O(2).
7.4 Concluding Remarks
Before moving on to the study of curves, lets review the techniques we
have available for calculating dierentials on a scheme:
1. Try to use the calculation of
1
P
n
2. Use the closed immersion exact sequence.
3. Use the relative cotangent exact sequence.
These ideas lead naturally into curve theory, where one of our main
tasks will be to understand embeddings of curves into P
n
using
1
.
31
8 Friday, February 3rd: Curves
Now we begin our study of curves by embedding them into projective
space. An equivalent approach which we exploit is to study line bundles
(and sections thereof) on a curve (especially
1
).
8.1 Separating Points and Tangent Vectors
Let k be an algebraically closed eld, and X a proper k-scheme. Given a
surjection O
n+1
X
/, when is the corresponding map : X P
n
k
a closed
immersion? First we x some notation. Let V = k
n+1
, with a xed choice
of basis |ei. We write O
n+1
X

= V
k
OX, where in the tensor product
we are really tensoring with the constant sheaf associated to V .
Let / be a line bundle on X, and O
n+1
X
/ a surjection. For p a
point of X, recall our notation /(p) = /p/mp/p. For any vector s in V ,
we obtain an element sp of /(p) by the composite map
V
vv1
(X, V
k
OX) (X, /) /p /(p)
Recall also that by the Nullstellensatz, the residue eld k(p) at any
closed point p of X is isomorphic to k (remember we assume k is alge-
braically closed).
With these observations in place, we answer the question above, viz.
when do surjections onto line bundles give embeddings into projective
space, by the following.
Proposition 8.1. If X/k is proper over an algebraically closed eld, and
: X P
n
the map associated to the surjection O
n+1
X

/, then is a
closed immersion if and only if the following two conditions hold
1. (Separating Points) For every pair of distinct point p, q in X, there
is an element s of V such that sp (as dened above) is nonzero in
/(p), whilst sq is zero in /(q).
2. (Separating Tangent Vectors) For every closed point p of X, the map
ker
_
V /(p)
_
mp/p/mp/
2
p
is surjective.
8.2 Motivating the Terminology
Well prove this in the next lecture. First we motivate the conditions. As
a warmup, write V = kx0 . . . kxn, and consider the xi as coordinates
on P
n
. If we take, for example, s = x0, then what is the set of closed
points p in X such that s goes to zero in /(p)?
First recall that we can write the element (p) P
n
k
as [a0 : . . . : an] as
follows. Choose a basis element for /(p), i.e., an isomorphism /(p)

k.
Then we get a map
k
n+1
= V /p

k
which sends each basis element xi to some ai in k. Of course, it is not
well-dened, since we had to choose a basis for /(p); but any other basis
diers from our chosen one by a nonzero element of k, so our coordinates
32
ai are well-dened up to nonzero scalars, giving a usual point of projective
space (in the classical sense). Thus we see that the set of closed points
of X for which x0 maps to zero is just the preimage under of the set
|[0 : a1 : . . . : an] P
n
k
.
More generally, if s is any linear form in the xi (i.e., an element of V ),
then the points of X for which s maps to zero in /(p) are the preimage
of the zero locus of s in P
n
k
. The zero locus of such a linear form is
usually called a hyperplane in P
n
k
. So what condition 1 says is that for
any p, q X, there exists a hyperplane passing through p but not q;
succinctly, the map is injective on points.
Of course, this does not guarantee that is a closed immersion. For
an example where this fails, let Y be the cusp Spec k[x, y]/(y
2
x
3
) and
X its normalization Spec k[t]. Then there is a morphism : X Y
corresponding to the ring map x t
2
, y t
3
. This is a homeomorphism
of topological spaces but not a closed immersion: the tangent space at the
origin on Y is too big.
To see why, look at the map on tangent spaces at the origin. On Y ,
we have mO/m
2
O

= k x k y, and on X, we have mO/m
2
O

= k t. But
the induced map between them is zero, since the images of x and y are
both in m
2
O
k[t]. In order for this map to be a closed immersion, this
map would need to be surjective.
Condition two says exactly this: the map T
P
n
_
(p)
_

TX(p)

is
surjective, or dually, the map on tangent spaces is injective.
8.3 A Preparatory Lemma
Well need the following in our proof next time.
Lemma 8.1. A map : X P
n
k
is a closed immersion if and only if, for
every i = 0, . . . , n,
1. The subscheme Xi = |p X

xi , 0 in /(p) is ane, and

2. the map k[y0, . . . , yn] (Xi, OX
i
) sending yj to (xj)/(xi) is
surjective.
We explain the second condition: is our surjection O
n+1
X
= V
k
OX /, which we think of as determining the homogeneous coordinates
of a point p X. Restricting to Xi, we get a map

X
i
: (Xi, O
n+1
X
i
)
(Xi, /

X
i
). For each p Xi, if we restrict to the stalk at p, we nd
that

X
i
(xi) is nonzero (its nonzero in /(p), hence nonzero in /p by
Nakayama). But this holds for any p Xi, therefore (xi) must have
been a unit in (Xi, /

X
i
) (reason: its a unit in every stalk, hence not
contained in any prime ideal of (Xi, /

X
i
)).
Now since / is a line bundle, we are guaranteed an open cover of X
on which it trivializes. The above remarks show that in fact the Xi form
such a cover, with the trivialization
OX
i
/

X
i
given by 1 (xi), since we just saw that (xi) generates / over Xi.
Therefore each global section (xj) of / over Xi (some or all of which may
33
be zero), maps back under this isomorphism to an element (xj)/(xi).
Then of course since k[y0, . . . , yn] is free, we may dene a map to (Xi, OX
i
)
by sending each yj to (xj)/(xi). This is the map which appears in the
second condition of Lemma 8.1, which we now prove.
Proof of Lemma 8.1. Since the condition of being a closed immersion is
ane-local on the target, the map : X P
n
k
is a closed immersion if
and only if, for each i = 0, . . . , n, the restriction

1
_
P
n
k
\ V (xi)
_
P
n
k
\ V (xi)
is a closed immersion. By denition, this morphism is just
Xi Spec k[
y0
yi
, . . . ,
yn
yi
].
Moreover, since the target now is ane, this morphism is a closed immer-
sion if and only if Xi also is ane (a closed immersion must be an ane
morphism) and the corresponding map of sheaves, which when Xi is ane
corresponds to a ring map k[
y
0
y
i
, . . . ,
yn
y
i
] (Xi, OX
i
), is surjective.
9 Monday, February 6th: Proof of the Closed
Embedding Criteria
9.1 Proof of Proposition 8.1
We need yet another lemma:
Lemma 9.1. Let f : A B be a local homomorphism of local Noetherian
rings, such that the induced map on residue elds A/mA B/mB is an
isomorphism, and the map on cotangent spaces
mA/m
2
A
mB/m
2
B
is surjective. Then f itself is surjective.
The intuition behind the lemma is that if, say, A = k[[x1, . . . , xn]]
and B = k[[x1, . . . , xn]]/(f1, . . . , fr), then surjectivity on cotangent spaces
means that in each equation fi = 0, we can solve for xi in terms of elements
of A.
Proof. The ideal mAB B is contained in mB since f is a local homomor-
phism. Since mA/m
2
A
mB/m
2
B
is surjective, the images of generators of
mA/m
2
A
generate mB/m
2
B
, and we can lift these by Nakayama to genera-
tors of mB. This implies that the images of generators of mA generate mB,
so we have mAB = mB. Thus by our rst assumption, A/mA

= B/mAB.
In particular we have a surjection A B/mAB, so the image of 1 gener-
ates B/mAB as an A-module. Applying Nakayama again, this time to B
as an A-module, we can lift the image of 1 in B/mAB to a generator of
B as an A-module, which means f is surjective.
34
Proof of Proposition 8.1. For the if direction, rst observe that since
X is proper over k, X P
n
k
is also proper. Therefore the image Z [P
n
k
[
of [X[ under is a closed set. The same holds with [X[ replaced by any
closed subset of X and Z replaced by (V ) (note closed immersions are
proper and compositions of proper morphisms are still proper), so is a
closed map.
Next, condition (1) implies that the map of topological spaces is in-
jective on closed points. We claim that this implies is actually injective.
Ill come back to this at some point.
Now a closed map that is also injective is a homeomorphism onto its
image. Therefore (check!) is quasi-nite. But a proper and quasinite
map is in fact nite (check!) and a nite map is always ane, so Xi =

1
(Ui) is ane.
It remains to show that if p X(k), then the map
m
P
n
,(p)
/m
2
P
n
,(p)
mX,p/m
2
X,p
is surjective. [Recall our notation: X(k) is the set of k-valued points,
i.e., maps Spec k X, which correspond to closed points of X since k is
algebraically closed. So were saying the above should be surjective for all
closed points p]. This will suce, because the condition that O
P
n OX
be surjective can be checked locally, since the global sections functor is
exact on anes.
For notational convenience, we make a linear change of coordinates so
that (p) = [0 : . . . : 0 : 1] P
n
. Then
m
P
n
,(p)
/m
2
P
n
,(p)

= k
x0
xn
. . . k
xn1
xn
,
the linear forms on the open ane Un = Spec k[
x
0
xn
, . . . ,
x
n1
xn
]. Since
p Xn, the global section sn = xn maps to a basis of /p, i.e., gives a
trivialization of /p (the section sn canonically trivializes / over Xn). So
mX,p/m
2
X,p

= mp/p/m
2
p
/p, and this isomorphism of OX-modules ts into
the following diagram
m
P
n
,(p)
/m
2
P
n
,(p)

k
x
0
xn
. . . k
x
n1
xn

mX,p/m
2
X,p

mp/p/m
2
p
/p
The map on the left is the one we want to show is surjective, and the map
on the right is the map of condition (2). This proves surjectivity of the
map of sheaves locally at closed points. It is left as an exercise to check
at the non-closed points (possible hint: use ber products somehow).
Note: we hand-waved the other implication (closed immersion implies
separates points and tangent vectors) in the statement of the Proposition
in class last time. Ill try to come back and write it up clearly at some
point.
35
9.2 Towards Cohomology
As before, consider a scheme X, proper over an algebraically closed eld,
of dimension one, normal, and connected. Such a curve has only two types
of points: a generic point, and some closed points X(k), at each of which
the local ring is a DVR.
Remark. If P is a closed point of X, then 1P , the (quasicoherent) ideal
sheaf associated to the embedding P X, is invertible. In general, if /
is an invertible sheaf on X, we may write /(P) = / O
X
1
1
P
.
Now, when we are given a rank one surjection O
n+1
X
/, how do we
check whether conditions (1) and (2) of Proposition 8.1 hold? First we
consider condition (1). Suppose given two distinct (closed) points P and
Q of X. We hope to produce a section s V such that s(P) = 0 but
s(Q) ,= 0.
Consider the case when V = (X, /). Then the s we desire is,
in particular, an element of the kernel of the map of k-vector spaces
(X, /) = V /P O
X,P
k(P), and this kernel is just (X, /(P)).
To see this, note that the map in queston is obtained by tensoring the
exact sequence
0 1P OX k(P) 0
with /, and then taking global sections (which is left-exact on the left, so
respects kernels). Then note that (X, / O
X
k(P))

= /P O
X,P
k(P)
since k(P) is a skyscraper sheaf
2
.
In order to determine whether we can nd an s such that, also, s(Q) ,=
0, we ask whether the map
(X, /(P)) /Q k(Q)
is surjective. Here we regard /Q k(Q)

= /Q/mQ/Q simply as a one-
dimensional vector space; but we may also regard /Qk(Q) as a skyscraper
sheaf at Q, which ts into the following exact sequence of sheaves of OX-
modules:
0 /(P Q) /(P) /Q k(Q) 0
Now the global sections functor is not right-exact, so the map
(X, /(P)) (X, /Q k(Q)) = /Q k(Q)
is not necessarily surjective. But since /Q k(Q) is a one-dimensional
k-vector space, there are only two possibilities: the map is zero, or else its
surjective. Condition (1) holds exactly when the map is not zero. Similar
considerations apply to condition (2) by taking P = Q in the above. So we
are led to the following general problem. Given the short exact sequence
0 /(P Q) /(P) /Q k(Q) 0,
when is the sequence
0 (X, /(P Q)) (X, /(P)) (X, /Q k(Q))
2
This means that the sections over any open containing P are all the same, so taking the
limit does nothing.
36
obtained by applying the global sections functor also exact? This is the
question of cohomology. Once we have dened cohomology, we will see
that there is a long exact sequence
0

(X, /(P Q))

(X, /(P))

(X, /Q k(Q))
dddddddddddddddddddd
H
1
(X, /(P Q))

H
1
(X, /(P))
. . .
and thus the vanishing of H
1
(X, /(P Q)) for all P and Q (not neces-
sarily distinct) is sucient to ensure that the associated map X P
n
is
a closed immersion.
10 Wednesday, Feb. 8th: Cohomology of Coher-
ent Sheaves on a Curve
Before proving Riemann-Roch, we introduce, without proof of its exis-
tence, the cohomology theory we need, together with its relevant prop-
erties. The rst section is a reproduction of a handout Professor Olsson
distributed in class. The second and third contain my notes on his com-
ments about the handout.
10.1 Black Box: Cohomology of Coherent Sheaves on Curves
For the results on curves that we will prove in this class, we will assume
the existence of a cohomology theory satisfying the following properties.
The existence of such a cohomology theory will be a consequence of the
general development of cohomology which we will study later.
Fix an algebraically closed eld k. In the following, a curve means a
proper normal scheme X/k of dimension 1.
If X is a curve and F is a coherent sheaf on X, we write H
0
(X, F) for
(X, F). As part of our black box we will include the following result:
(0) For every coherent sheaf F on a curve X, H
0
(X, F) is nite di-
mensional over k.
We assume given for every curve X a functor
H
1
(X, ) : (coherent sheaves on X) (nite dimensional k-vector spaces)
and for every short exact sequence of coherent sheaves on X
E : 0 F

F F

0
a map

E
: (X, F

) H
1
(X, F

)
such that the following conditions hold.
(i) The functor H
1
(X, ) is k-linear in the following sense. If a k and
F is a coherent sheaf, and if ma : F F denotes the multiplication by a
map on F (which is an endomorphism in the category of coherent sheaves
on X) then the induced map
H
1
(ma) : H
1
(X, F) H
1
(X, F)
37
is multiplication by a on the vector space H
1
(X, F).
(ii) For every short exact sequence E as above, the sequence
0

H
0
(X, F

)

H
0
(X, F)

H
0
(X, F

E
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
H
1
(X, F

)

H
1
(X, F)

H
1
(X, F

)

0
is exact.
(iii) The maps
E
is functorial in the exact sequence E in the following
sense. If
0

F

1

a

F1
b

1
c

0
0

F

2

F2

F

2

0
(9)
is a commutative diagram of coherent sheaves on a curve X with exact
rows, then the diagram
H
0
(X, F

1
)
H
0
(c)

E
1

H
1
(X, F

1
)
H
1
(a)

H
0
(X, F

2
)

E
2

H
1
(X, F

2
)
commutes, where we write E1 (resp. E2) for the top (resp. bottom) short
exact sequence in (9).
(iv) If F is a coherent sheaf on a curve X whose support is a nite set of
points, then H
1
(X, F) = 0.
(v) If X is a curve then there is an isomorphism tr : H
1
(X,
1
X
) k such
that for any locally free sheaf E on X with dual E

the pairing
H
0
(X, E) H
1
(X, E

1
X
)

H
1
(X,
1
X
)
tr

k
is perfect.
Remark. Axiom (i) implies that if F and G are coherent sheaves on
X and if z : F G is the zero map then the induced map H
1
(z) :
H
1
(X, F) H
1
(X, G) is also the zero map. Indeed z = m0 z, where
m0 : G G is multiplication by 0, and therefore H
1
(z) = H
1
(m0)H
1
(z).
Since H
1
(m0) is multiplication by 0 on H
1
(X, G) by (i) this implies that
H
1
(z) = 0.
Remark. Axiom (ii) implies that H
1
(X, ) commutes with nite direct
sums. Indeed to show this it suces by induction to consider two coherent
sheaves F and G. In this case we have a split exact sequence
0 F F G G 0.
Since the projection map F G G admits a section, then induced
map H
0
(X, F G) H
0
(X, G) is surjective, whence (ii) gives an exact
sequence
0 H
1
(X, F) H
1
(X, F G) H
1
(X, G) 0
38
split by the standard inclusion G F G. Since nite direct sums
are isomorphic to nite direct products in both the category of coherent
sheaves on X as well as the category of nite dimensional vector spaces, it
follows that H
1
(X, ) also commutes with nite products. This has many
consequences. For example, if F is a coherent sheaves and : F F F
is the summation map, then the induced map
H
1
(X, F) H
1
(X, F) H
1
(X, F F)
H
1
()

H
1
(X, F)
is the summation map on the vector space H
1
(X, F). Similarly if f, g :
F G are two maps of coherent sheaves, then the two maps
H
1
(f +g), H
1
(f) +H
1
(g) : H
1
(X, F) H
1
(X, G)
are equal.
10.2 Further Comments; Euler Characteristic
Axiom (v) is known as Serre duality. We explain how to obtain the rst
map, leaving the existence of the trace map tr as part of the axiom. For
a perfect pairing, we would like to associate to each e H
0
(X, E) an
operator e, ) on H
1
(X, E

1
X
), in such a way that this isomorphism
identies H
0
(X, E) with the dual space of H
1
(X, E

1
X
). First observe
that for each e H
0
(X, E) = (X, E), we get a map OX E, which we
tensor with E

to obtain maps of sheaves

E

OX O
X
E

e1

E E

Ox
where the vertical map is just evaluation. The dotted composite we now
tensor with
1
to give a map E

1

1
, to which we apply the
functor H
1
(X). So for each such global section e, we have a map
H
1
(X, E

1
X
) H
1
(X,
1
), and this denes the pairing. Note also
that the existence of the trace map in the axiom should be expected if
you are familiar with Hodge theory on complex manifolds.
We will state Riemann-Roch using Euler characteristic, and then red-
erive the more familiar form. The theorem is about calculating the di-
mensions of the k-vector spaces H
0
(X, /) for various line bundles /. This
can be quite hard, and it turns out to be easier to look instead at the
Euler characteristic dened by, for T Coh(X),
(X, T) = dim
k
H
0
(X, T) dim
k
H
1
(X, T)
One useful fact is that is additive on short exact sequences. More
precisely, we have the following:
Lemma 10.1. If the sequence 0 T

T T

0 is exact, then
(X, T) = (X, T

) +(X, T

).
39
Proof. From the given sequence we get a long exact sequence
0

L
L
L
L
L 0
I1

q
q
q
q
q
I3

O
O
O
O
0

H
0
(T

)

H
0
(T)

o
o
o
o
o
H
0
(T

K
K
K
H
1
(T

H
1
(T)

H
1
(T

)

0
I3

t
t
t

L
L
L
L
L
0

q
q
q
q
q
0
Here the I
k
are the kernels and images of the appropriate maps. This
gives us four short exact sequences involving the cohomology groups and
the I
k
. Each short exact sequence relates the dimensions of three of the
vector spaces. If you combine the four equations appropriately, you get
the result. Note: This diagram diers slightly from the one used in class,
though the argument is essentially the same. This more symmetric form
of the diagram was pointed out to me by Alex Kruckman.
Remark. Lets briey recall the relationship between divisors and line
bundles. On a curve X, we can dene a divisor as a formal sum of closed
points, D =

nP P. To such an object, we associate a line bundle, de-

noted by either OX(D) or /(D), as follows. Each closed point P may
be regarded as a closed subscheme of X, and thus has an associated
ideal sheaf 1P [Recall further that a closed point is the same as a map
i : Spec k X, which induces a map i
#
: OX iO
Spec k
, where the
latter is most usefully regarded as the skyscraper sheaf of k(P) = k at the
point P. The ideal sheaf 1P is dened to be the kernel of i
#
]. Then we
set
/(D) = /(

nP P) =

P
1
n
P
P
where here the superscript means tensor power.
Conversely, every invertible sheaf on X is of this form. Suppose given
a line bundle / on X. Take an open subset U X on which we have
a trivialization : OU

U
. Let s = (1), a section of / over U.
The complement of U is then a nite set of points |P1, . . . , Pr (possibly
empty, if / itself happened to be trivial). Since X is a curve, the local ring
at each Pi is a DVR. We want to use the valuations in each of these DVRs
to produce the integers nP
i
. Now we proved last semester that there is
an inclusion / K , where K is the constant sheaf associated to the
function eld k(X). So we take s, and map it into K (U)

= k(X). Then
at each Pi we have a valuation P
i
on k(X), and we dene nP
i
= P
i
(s).
These nP
i
, for each Pi, thus dene a divisor D =

nP
i
Pi on X. Of
course, we still have to check that /

= /(D). For this, you can refer to
the notes from last semester.
10.3 Statement of Riemann-Roch; Reduction to the More Fa-
miliar Version
From now on, we will use the notation h
i
(X, /) = dim
k
H
i
(X, /).
40
Theorem 10.1 (Riemann-Roch Theorem). If X is a curve, and / a line
bundle on X, then
(X, /)

= deg / +(X, OX).
We will prove this next lecture. First notice that
H
1
(X, /) = H
1
_
X, (/

1
X
)

1
X
_
,
and by Serre Duality, this is dual to H
0
(X, /

1
X
). Hence h
1
(X, /) =
h
0
(X, /

1
X
). This allows us to rewrite the left hand side of the equation
in the theorem as h
0
(X, /) h
0
(X, /

1
X
). Moreover, applying the
same trick to H
1
(X, OX), we get h
1
(X, OX) = h
0
(X,
1
X
), which is by
denition the genus, g, of X. Finally, we have h
0
(X, OX) = 1, by either
of two arguments presented at the beginning of the next lecture.
Thus we can rewrite Riemann-Roch as
h
0
(X, /) h
0
(X, /

1
X
) = deg / + 1 g.
Remark. If H
0
(X, /) ,= 0, then deg / 0. For if s is a nonzero global
section, then as an element of the fraction eld, its valuation at each point
P is non-negative (i.e., its in OX,P ). Loosely, its everywhere regular, so
has no poles. In fact, you can calculate the degree of / explicitly by
looking at the exact sequence
0 OX / //OX 0,
where the term on the right is supported on a nite set of points.
11 Friday, Feb. 10th: Proof and Consequences of
Riemann-Roch
11.1 Preparatory Remarks
Remark. Since H
0
(X, OX) is a k-algebra, h
0
(X, OX) is at least one.
Here are two dierent reasons why its equal to one.
Proof 1. Let f H
0
(X, OX). Then f denes a map X A
1
k
P
1
k
.
Since P
1
k
is proper, f(X) is a closed subset of P
1
k
, hence a nite set of
points in A
1
k
. Since X is connected, so is f(X), which must therefore
consist of a single point A
1
k
.
3
So the map to A
1
k
dened by f agrees
with the map dened by k H
0
(X, OX). Therefore the global
sections f and dier by nilpotence, and since X is reduced, they are
the same global section. Hence H
0
(X, OX) = k.
3
Recall that the map to A
1
k
dened by f sends x to the residue of f in k(x) for each x X.
For closed points, k(x) = k, so in our case, we have that f maps to the same k for
every closed point, and we think of this as living in A
1
k
(i.e., as corresponding to the ideal
(t ) k[t].
41
Proof 2. Since H
0
(X, OX) is nite-dimensional over k, its a nitely gen-
erated integral (commutative) algebra over k. This implies its a nitely
generated eld extension of k: to nd the inverse to a nonzero , use the
monic polynomial cancelling . That is, if
n
+bn1
n1
+. . . +b0 = 0,
then (
n1
+ bn1
n1
+ . . . + b1) = b0, which is nonzero since its a
domain, so is a unit.
Then since k is algebraically closed, there are no nontrivial nitely
generated extensions, hence H
0
(X, OX) = k.
Remark. Recall from last time that if / is an invertible sheaf on X,
and there is a nonzero global section s H
0
(X, /), then deg / 0.
Also, if deg / = 0, then /

= OX (so s k). To see this, note that the
section s denes a map OX /. For over each open U, we can dene
O(U) /

U
by sending 1 (i.e., the restriction of 1 H
0
(X, OX)) to s

U
.
This obviously commutes with restriction maps, so it denes a morphism
of sheaves. Of course, theres no reason for this to be an isomorphism,
but it is injective since s is nonzero over every open U (using that X is
integral).
We are interested in the cokernel of this map s. To measure the failure
of s to be an isomorphism, we begin with an open U on which / is trivial;
s itself furnishes a trivialization. The complement of U is a nite point
set |P1, . . . , Pr. The cokernel is trivial over U; looking at germs near
Pi, /P
i
is a free rank one OX,P
i
-module, and the cokernel of s locally has
the form /P
i
/m
n
i
P
i
/P
i
. Each ni is non-negative since s is a global section.
Thus the cokernel of s is the sheaf
coker(s) =

{P
1
,...,Pr}
/P
i
/m
n
i
P
i
/P
i
where each factor in the tensor product is a skyscraper sheaf at the ap-
propriate Pi. The global sections of this sheaf forms a nite-dimensional
k-vector space, whose dimension is the degree of /, hence deg / =

ni.
It follows, then, that if deg / = 0, each ni is zero, and the cokernel is
trivial, so s denes an isomorphism /

= OX.
11.2 Proof of the Riemann-Roch Theorem
Proof. We will prove the relation
(X, /) = deg / +(X, OX)
by induction on the degree of /. For deg / = 0, its a consequence of the
above remarks, since we just saw that in this case, /

= OX.
For the inductive step, we prove that the equation holds for / if and
only if it holds for /(P) = / 1
1
P
. To see this, tensor the sequence
0 1P OX k(P) 0
with /(P), giving
0 / /(P) k(P) 0
42
(note that the term on the right is a skyscraper sheaf, so tensoring with
/(P) does nothing). Since the Euler characteristic is additive on exact
sequences, this yields (X, /(P) = (X, /) +1 (note that h
1
(X, k(P) = 0
by one of our axioms for cohomology). Since deg /(P) = deg / + 1, this
proves the equivalence.
This proves the theorem, for we have seen above that any line bundle
can be obtained from OX by adding or subtracting a nite number of
points
11.3 Consequences
Corollary 11.1. deg
1
k
= 2g 2.
This comes from setting / =
1
k
in the statement of Theorem 10.1. It
suggests that the study of curves should divide into three cases: g = 0,
g = 1, or g 2, according as the degree of
1
k
is negative, zero, or positive,
respectively.
Corollary 11.2. If deg / > 2g 2,
h
0
(X, /) = deg / + 1 g
Corollary 11.3. If deg / > 2g1, then / is generated by global sections.
Remark. We could also express this by saying that the map H
0
(X, /)
k
OX / is surjective. What map is this? Let f : X Spec k be the
structure morphism. Since f

is a left adjoint to f, there is a natural

map f

f/ /. But f is just H
0
(X, /) sitting over the unique point
of Spec k. This pulls back to f
1
H
0
(X, /)
f
1
O
Spec k
OX, and since f
is actually an open map here, and H
0
(X, /) and O
Spec k
are constant
sheaves f
1
just pulls them back to the corresponding constant sheaves
on X, i.e., f
1
H
0
(X, /)
f
1
O
Spec k
OX

= H
0
(X, /)
k
OX.
Proof of Corollary 11.3. We prove that the cokernel of the map H
0
(X, /)
/ described in the previous remark is zero. Since the support of the coker-
nel (and indeed any sheaf) is closed, its enough to check that the cokernel
is zero at closed points. Thus we have to show that for any closed point
P, there is a global section f whose image in / k(P)

= /P /mP /P
is nonzero (we are using Nakayama here). This is equivalent to nding a
global section f which is in (X, /) but not in the subspace (X, /(P)).
This we can verify simply by comparing dimensions. Our degree assump-
tion ensures that the h
1
term in Riemann-Roch vanishes for both sheaves
(Corollary 11.2), so we have
h
0
(X, /) = deg /+1g and h
0
(X, /(P)) = deg /1+1g = deg /g
This (X, /(P)) is a proper subspace of (X, /), so we can nd the
desired global section f.
Remark. (In response to a question) Recall that /(P) has stalks equal
to / away from P, and equal to mP / at P.
43
Of course, it is not really necessary to require that deg / be greater
than 2g 1, for if / has any positive degree, then we can take a high
tensor power to obtain a sheaf whose degree is at least 2g 1.
Denition 11.1. A line bundle / on X is very ample if the map : H
0
(X, /)
k
/ / denes a closed immersion into P
n
k
. / is ample if there exists an
n > 0 such that : H
0
(X, /
n
)
k
/ /
n
denes a closed immersion.
We will prove next time that if deg / > 0, then / is ample.
12 Monday, Feb. 13th: Applying Riemann-Roch
to the Study of Curves
First we introduce some new terminology. Recall that we have constructed
a map H
0
(X, /)
k
OX / of sheaves, for a line budle / on a curve
X. We call the set of points at whose stalks this map is not surjective
the base locus of /, and if it surjective everywhere (i.e., / is generated by
global sections), we say / is base point-free.
In this language, we saw last time that if the genus of X is g and
deg / > 2g1, then / is base point-free, in which case we get a morphism
of schemes X PH
0
(X, /).
12.1 Using Riemann-Roch to Detect Closed Immersions to P
n
Any rank one quotient of O
n+1
X
corresponds to some morphism X P
n
k
.
When / is generated by global sections, we get a rank one quotient of
the form H
0
(X, /)
k
OX /, with corresponding morphism X
PH
0
(X, /). When is this a closed immersion?
By our earlier criteria, / must separate points and tangent vectors.
We now rephrase these conditions in such terms as allow us to apply
Riemann-Roch.
The condition of separating points is equivalent to requiring that for
any two distinct closed poitns P and Q in X, there is a section s in
H
0
(X, /(P)) \ H
0
(X, /(P Q)). The condition of separating tangent
vectors is equivalent to requiring that for any closed point P, there is
a section s in H
0
(X, /(P)) \ H
0
(X, /(2P)). We can guarantee the
satisfaction of both conditions at once by requiring that, for any P and
Q in X(k)
4
, not necessarily distinct, there is a section in H
0
(X, /(P)) \
H
0
(X, /(P Q)). This can be checked with a simple dimension count,
and Riemann-Roch gives the following criterion:
Proposition 12.1. If deg / > 2g, then / is very ample.
Proof. For any closed points P, Q (not necessarily distinct), our assump-
tion on the degree of / ensures that h
1
is zero for both /(P) and
/(P Q), so Riemann-Roch computes the following:
h
0
(X, /(P)) = (deg / 1) + 1 g
h
0
(X, /(P Q)) = (deg / 2) + 1 g,
4
Recall that this is the set of k-valued points of X, namely the closed points.
44
This means the latter is a proper subspace of the former, and we can nd
the section we desire, as described in the preceding discussion.
We also have the following:
Corollary 12.1. If deg / > 0, / is ample.
12.2 Examples
Genus zero
Let X be a curve with g = 0. We will show that X is isomorphic to
P
1
k
. For any P X(k), let, as usual, OX(P) = 1
1
P
, where 1P is the ideal
sheaf of P X. It has degree 1 > 2g 2 = 2, so
h
0
(X, OX(P)) = deg / + 1 g = 2
By choosing an isomorphism H
0
(X, OX(P))

k
2
, we get a closed
immersion (by Proposition 12.1, since deg / = 1 > 2g = 0) i : X
P
1
k
. We claim this is actually an isomorphism of schemes. Since its
a closed immersion, i is a closed map. Therefore its image is a closed
set, hence all of P
1
. Since its an immersion, its injective, and since
proper also, its a homeomorphism (cf. a related statement in point-set
topology - a continuous bijection from a compact space to a Hausdor
space is automatically a homeomorphism.) Now we check the map of
sheaves is an isomorphism. Since i is a closed immersion, it induces a
surjection of sheaves, and hence a surjection of stalks O
P
1
,P
OX,P
(here we identify X with its image, namely P
1
, and thus speak of the
same P on both sides). The map i is a surjective morphism of schemes,
hence in particular a dominant rational map, and so it induces a map of
funtion elds k(P
1
) k(X). This map is not the zero map, so it must be
injective. Thus we obtain the following diagram
O
P
1
,P

OX,P

k(P
1
)

k(X)
where the top arrow is surjective and the other three are injective. This
forces the top arrow to be an isomorphism, like we wanted. This shows
that X is isomorphic to P
1
k
.
Genus one
Now let X be a genus one curve
5
over k. Since 2g = 2, we need a line
bundle of degree at least three to embed in projective space. Pick a point
P X(k) and consider OX(3P). After choosing a basis for the three-
dimensional space H
0
(X, OX(3P)), we get a closed immersion i : X P
2
k
.
We will show next time that i identies X with a cubic curve in P
2
k
, which
can be given in characteristics dierent from 2 or 3 by an equation of the
form y
2
= x
3
+Ax +B.
5
This is not the same as saying that X is an elliptic curve - these are given by a genus one
curve together with a chosen point.
45
Remark. We chose a degree 3 line bundle to satisfy the hypothesis of the
theorem, but this is not necessarily the best we can do. For an example,
take X = P
1
k
, and the line bundle OX(2). Then h
0
(P
1
x
, O(2)), and we get
an embedding P
1
k
P
2
k
. Exercise: nd the equation of this embedding
(hint: its the Veronese map).
13 Wednesday, Feb. 15th: Where is the Equa-
tion?
Today we prove:
Theorem 13.1. Any curve of genus one (over k) is isomorphic to a cubic
in P
2
k
.
Recall that a homogeneous polynomial F of degree d in the variables,
say, x0, . . . , xn, denes a closed subscheme of P
n
k
, which we denote V (F).
We ask, given a morphism g : X P
n
k
corresponding to a rank one quo-
tient O
n+1
X
/, how can we detect whether g factors through V (F)?
To answer this, let the generating global sections from the surjection
O
n+1
X
/ be s0, . . . , sn (X, /). Recall that the identity map P
n
P
n
gives a surjection O
n+1
P
n O
P
n(1), with corresponding global sections
x0, . . . , xn, which we think of as the homogeneous coordinates on P
n
k
. Now
if we have a morphism g : O
n+1
P
n O
P
n(1), we can pull back the surjection
O
n+1
P
n O
P
n(1) over P
n
k
to the surjection O
n+1
X
g

O
P
n(1) (recall that
the pullback of the structure sheaf is again the structure sheaf). We will
abuse notation slightly and refer to the global sections of this pulled-back
surjection also as x0, . . . , xn. The surjections t into a diagram
O
n+1
X
(s
0
,...,sn)

(x
0
,...,xn)

P
P
P
P
P
P
P
P
P
P
P
P
P
/

O
P
n(1)
Here the vertical map is an isomorphism because both surjections corre-
spond to the same morphism X P
n
k
.
Proposition 13.1. The map g factors through V (F) if and only if F(s0, . . . , sn)
is zero in (X, /
d
).
Coarse Proof. The expression of a closed point P of P
n
in terms of homo-
geneous coordinates [x0 : . . . : xn] is obtained by taking each of the global
sections x0, . . . , xn of O
P
n(1) and evaluating them in O(1)P /mP O(1)P

=
k. A point P is in V (F) if and only if these evaluations satisfy F = 0,
or put another way, if and only if F(x0, . . . , xn), regarded as an element
of the sheaf O(1)
d
, maps to zero at P. This makes sense, since locally,
O(1)
d

= O(1) via the multiplication map a0 . . . an a1 an (of

course, theyre both isomorphic to O
P
n locally). So the evaluation of F
at P is obtained by taking F, restricting it to some open containing P on
which O(1) is trivial, multiplying the sections occuring in F
6
, and then
6
It doesnt make sense to multiply, say, x
0
x
1
over an arbitrary open since O(1) may not
be a ring over any open - it only is once we trivialize the line bundle.
46
passing to O(1)P /mP O(1)P . This gives an element of k, which we require
to be zero in order for P to be in V (F).
To ask whether a point of X maps into V (F), note that the image of
a point P X under g is given by evaluating the sections s0, . . . , sn at
P. So the discussion in the previous paragraph applies to (the images of)
points of X, simply replacing xi by si.
Lets look at some examples.
Example 13.1. Let X have genus zero, and pick a point P X(k). The
space H
0
(X, OX) is a proper subspace of H
0
(X, OX(P)). Weve already
computed dimH
0
(X, OX(P)) = 2, so we can write H
0
(X, OX(P)) as the
span of two global sections 1 and x. Here 1 spans the constant sections,
namely H
0
(X, OX), while x necessarily has a pole at P, or else it would
be in H
0
(X, OX).
Weve seen last time that a basis of global sections of OX(P) deter-
mines a map X P
1
(an isomorphism, in fact). Identifying X with P
1
via this isomorphism, one thinks of the global sections x and 1 as de-
homogenizations of the global sections (i.e., homogeneous coordinates) x
and y on P
1
.
Example 13.2. Let X have genus one, and x a point P X(k). The in-
vertible sheaf / = O(3P) denes an embedding X P
2
k
, since h
0
(X, /) =
3 > 2g = 2. We have a chain of subspaces (writing H
0
(X, OX) =
H
0
(OX), etc., for short)
H
0
(O) H
0
(O(P)) H
0
(O(2P)) H
0
(O(3P)) H
0
(O(4P))
H
0
(O(5P)) H
0
(O(6P)) . . .
Riemann-Roch allows us to compute the dimensions: 1,1,2,3,4,5,6,. . . .
Pick a basis 1, x for H
0
(O(2P)). Then x has a pole of order two at P, or
else the previous space would have been two-dimensional. Similarly, we
have a basis 1, x, y for H
0
(O(3P)), where y has a triple pole at P. Then
x
2
extends this to a basis for H
0
(O(4P)), and xy extends it further to
a basis for H
0
(O(5P)). But in H
0
(O(6P)), we have the seven elements
1, x, x
2
, x
3
, xy, y, y
2
, all of which have poles or order at most 6 at P. Since
this space is six-dimensional, there must be a dependence relation amongst
them, and this relations must involve both x
3
and y
2
, or else one of these
two would live in a previous space, but they both have poles of order
exactly six. Thus, after rescaling the coordinates if necessary, we obtain
a relation of the form
q(x, y) = y
2
+bxy +cy x
3
ex
2
fx g = 0
This implies that X Y = V (q(x, y)) P
2
. Now one argues that
since X is a curve, q(x, y) is irreducible, and Y is one-dimensional, X must
be homeomorphic to Y . To check the map X Y is an isomorphism, its
47
enough to look at the stalks. For the stalk at P, we have a diagram
OY,P

OX,P

k(Y )

k(X)
where the vertical maps are inclusions, and the bottom arrow is an iso-
morphism since Y is integral (we mentioned above its irreducible. Check
its reduced). Thus the map on stalks is an isomorphism at P.
In general, given X and /, with corresponding embedding X
PH
0
(X, /), it possible to express
X

= Proj

n0
(X, /
n
).
We also have
P
n
k
= Proj

n0
S
n
(X, /),
where S
n
V denotes the nth symmetric power. Then the embedding X
P
n
k
is given by a map of graded rings given in degree n by S
n
(X, /)
(X, /
n
). The equations dening X come from the kernel of this map
of graded rings.
Going back to Example 2, the equation q(x, y) = 0 denes an open
subset U of X inside the ane chart A
2
x,y
of P
2
k
(the point P is in the
complement of U). We can regard this inclusion Y A
2
as being given by
the coordinate functions x and y (this is basically the Spec adjunction
for morphisms to ane targets). But the coordinate function x also denes
a map A
2
A
1
. Precomposing this with the inclusion U A
2
and then
following with the inclusion A
1
P
1
gives a map U P
1
. We have seen
that such a map extends to a morphism X P
1
.
Now we ask: what is the line bundle and corresponding global sections
associated to this map X P
1
? Since we mapped A
2
to A
1
by forget-
ting the y-coordinate, it is perhaps not surprising that the map to P
1
comes from the space H
0
(X, O(2P)), which had as basis only 1 and x.
Remark. It is no accident that the degree of the line bundle O
P
1(1),
when pulled back to X, is two. This is also the degree of
k(x)[y]/(y
2
(x )(x )(x )).
Note also that, letting f be the map X P
1
, we have
deg f

O
P
1(1) = deg 1
f
1
()
= deg
_

P

f(P)=
1
e
P
_
=

f
1
()

eP
Here eP denotes the ramication index of f at P. We will discuss
degree and ramication in the next lecture.
48
14 Friday, Feb. 17th: Ramication and Degree
of a Morphism
Denition 14.1. Let f : X Y be a nonconstant morphism of curves.
The degree of f is the degree of the eld extension [k(Y ) : k(X)]. If P
is a closed point of X, with f(P) = Q, the ramication index eP of f
at P is dened to be the ramication index of the induced map of DVRs
OY,Q OX,P .
This isnt much of a denition until we have dened ramication in-
dices for maps of DVRs. Let P , Q uniformizers of OX,P and OY,Q,
respectively. Then Q maps to a unit times
e
P
P
, where eP 1 since this
is a local homomorphism. Then eP is the degree of this map of DVRs.
We say f is unramied at P if eP = 1, and ramied at P if eP > 1,
in which case we also call Q a branch point of f.
The degree of the morphism f is, intuitively, the cardinality of the
preimage of any point Q Y , but we have to be careful at the branch
points.
Proposition 14.1. For any closed point Q of Y ,

Pf
1
(Q)
eP = deg f.
Proof. Let A = fOX. Then X = Spec
Y
A. The idea of the proof is to
rst prove that A is a locally free sheaf of nite rank on Y . Then we can
compute the rank at any point of Y . Over the generic point, we show the
rank is [k(X) : k(Y )] = deg f, while over a closed point Q, we will show
the rank is

Pf
1
(Q)
eP , giving the desired equality.
First lets prove A is locally free of nite rank. Since f is proper and
quasinite, f is nite, so we will have nite rank if we can show A is
locally free. For this, it is enough to check at closed points Q Y (k),
since the generic point is a localization of any closed point. First we
observe that AQ is a at OY,Q-module. Flatness is equivalent to the map
I AQ A being injective for any ideal I of OY,Q. Since OY,Q is a
DVR, its enough to check that the action of the uniformizer Q on AQ
is nonzero. But over an open U, A(U) = OX(f
1
(U)), and the action
of Q is given by mapping Q into OX(f
1
(U)) and multiplying. But
OX(f
1
(U)) is a domain, so this action is nonzero. Now the stalks of
A (which are stalks of OX), are nitely generated over the stalks of OY
since f is nite. But at and nitely generated impies free, so A has free
stalks. We conclude that A is locally free by the following.
Lemma 14.1. If T is a coherent sheaf on Y such that TQ is a free OQ-
module for all Q, then T is locally free.
Proof. At each Q we have an isomorphism O
r
Y,Q
TQ. Let f

1
, . . . , f

r

TQ be the image of the generators under this isomorphism, and f1, . . . , fr
their representatives on some neighborhood U containing Q. Then these
fi dene a map of sheaves on U (in fact, of OU-modules) U : O
r
U
T

U
which is an isomorphism at the stalk at Q. Let / and ( be the kernel
and cokernel, respectively, of U.
49
Since O
r
U
and T are coherent, so are / and (. Thus their supports
are closed subsets of U not containing Q. In particular, there is an open
neighborhood V U containing Q such that /P and (P are both zero
for all P V . Since isomorphisms can be checked at stalks, this means
that U

V
is an isomorphism over V , and hence T

V
is free.
Returning to the proof, we have now that A is locally free, and want
to compute its rank at both a closed point and the generic point of Y .
First, letting be the generic point of X, we note that
A = lim
U
OX(f
1
(U)) = lim
V
OX(V ) = k(X),
the equality in the middle coming from the fact that for any open V
of X, we can nd U Y such that f
1
(U) V (preimages of opens
in Y are conal amongst opens in X). Namely, if V = X \ |Pi, set
U = Y \ |f(Pi). We use this to compute the rank of A at : its
[k(X) : k(Y )], i.e., deg f.
Now let Q be a closed point of Y . We will compute the rank of A at
Q. This is by denition the rank of AQ as an OY,Q-module, which is also
equal to the dimension of AQ O
Y,Q
k(Q) as a k(Q)-vector space. First
we claim that
AQ O
Y,Q
k(Q)

= (X Y Spec k(Q), O
XSpec k(Q)
)
To see this, we can replace X and Y by anes Spec R and Spec S, respec-
tively, where Q is contained in Spec S and f(Spec R) = Spec S since f is
ane. The ber product in question corresponds to the diagram
R S k(Q) k(Q)

which we can expand into a double bered diagram

R S k(Q) k(Q)

R S OY,Q

OY,Q

S
f
#

Thus the claim follows from the observation that in fact AQ

= RS OY,Q,
since
R S OY,Q = lim
QD(g)
R S Sg = lim
QD(g)
R
f
#
g
= lim
QD(g)
OX(f
1
(D(g))),
which is (fOX)Q = AQ.
We must therefore compute the dimension over k(Q) of the global
sections of X Y k(Q). Since f is nite, X Y Spec k(Q) is a disjoint
50
union of points: f
1
(Q) = |P1, . . . , Ps, and therefore it is an ane
scheme:
XY Spec k(Q)

=
s

1=1
Spec OX,P
i
O
Y,Q
k(Q)

= Spec
s

i=1
OX,P
i
O
Y,Q
k(Q).
Now, OX,P
i
O
Y,Q
k(Q)

= OX,P
i
/
e
i
i
, since the OY,Q-algebra structure
on OX,P
i
is given by sending the uniformizer of OY,Q to
e
i
i
, where i is
the uniformizer of OX,P
i
, and ei the ramication index of f at Pi. Thus
we conclude that
AQ O
Y,Q
k(Q)

=
s

i=1
OX,P
i
/
e
i
i
,
which has dimension
s

i=1
ei. This is therefore the rank of A at Q, and
equating it with the rank at the generic point (deg f) gives the result.
Example 14.1. If X is a curve of genus 1, and P a closed point, then
h
0
(X, OX(2P)) = 2, so we get a map f : X P
1
. On the complement of
P, we have a map to A
1
k
= Spec k[t] given by f
#
: k[t] (X \ P, OX).
The section f
#
(t) extends to a global section of Ox(2P) with a pole of
order 1 or 2 at P. Thus deg f 2. But the degree cannot be one, or else
f would be an isomorphism, which is impossible since X has genus 1. So
the divisor 2P denes a degree two map to P
1
.
15 Wednesday, Feb. 22nd: Hyperelliptic Curves
So far in our study of curves, we have seen that genus one curves are
isomorphic to P
0
while genus one curves are cut out by cubic equations
in P
2
. Today we study curves of genus greater than one. We will see that
they split into two camps: those that are canonically embedded in some
P
n
, and hyperelliptic curves.
15.1 Canonical Embeddings
For a curve X of genus g 2, we have deg
1
X
> 0, so we can ask whether
the line bundle
1
X
denes a morphism, or better yet, an embedding, into
P
n
. Since h
0
(X,
1
X
) = g, we will get a morphism to P
g1
; in the case
that its an embedding, we will call this the canonical embedding.
To get a morphism at all, we need
1
X
to be base point free, which
you recall means that there is no point of X at which all global sections
of
1
X
vanish. Since the dimension h
0
(X,
1
X
(P)) is always one less
than or equal to h
0
(X,
1
X
), it is equivalent to require that that for all
P X(k), h
0
(X,
1
X
(P)) < h
0
(X,
1
X
), i.e., there is some global section
not vanishing at P.
If furthermore we ask that this morphism be an embedding,
1
X
must
separate points and tangent vectors, which we have seen earlier is equiva-
lent to requiring that for all P, Q X(k), not necessarily distinct h
0
(X,
1
X
(P
Q)) < h
0
(X,
1
X
(P)). Thus we obtain easily the following criterion.
51
Lemma 15.1.
1
X
denes a closed embedding X P
g1
if and only if,
for all closed points P, Q of X (not necessarily distinct), h
0
(X, O
1
X
(P
Q)) = g 2.
Proof. This follows from the above discussion: the dimension of H
0
(X, O
1
X
)
is g. Being base point free is equivalent to requiring that the dimension
drop by one when we subtract a point from O
1
X
; separating points and
tangent vectors happens exactly when the dimension drops one further
after subtraction of another point.
When is the condition of the lemma satised? We apply Riemann-
Roch to the line bundle
1
X
(P Q):
h
0
(X,
1
X
(P Q)) h
0
(X, OX(P +Q)) = g 3.
So we get a closed embedding exactly when, for all P, Qin X, h
0
(X, OX(P+
Q)) = 1. Of course, there is always an inclusion k H
0
(X, OX(P +Q)),
so we are really asking that there be no nonconstant functions with simple
poles at one or both of P and Q, for any choice of P and Q. But the case
when there is such a function is also interesting, for such functions give
degree two morphisms to P
1
. We now turn our attention to this latter
situation.
15.2 Hyperelliptic Curves
Denition 15.1. A curve X is hyperelliptic if there exists a degree two
map X P
1
.
With this terminology, we summarize our results thus far by saying
that any curve of genus g 2 can either be (canonically) embedded in
P
g1
or else is hyperelliptic. We now ask for a classication of hyperelliptic
curves. This will be accomplished by explicitly writing down an equation
for such curves.
So suppose X P
1
is a degree two map. Then we have a separated
degree two, hence Galois, eld extension k(P
1
) k(X). This gives a Z/2
action on k(X), which lifts to an action of Z/2 on X over P
1
. For example,
if X were dened by an equation of the form y
2
= (x a1) (x an),
then the action would be given by sending y to y.
How can we obtain the Z/2 action on X from that on k(X)? In general,
since the map f : X P
1
is ane, A = fOX is a locally free sheaf of
O
P
1-algebras of rank two. For an open ane U of P
1
, we have maps
k(X) A(U)

k(P
1
)

O
P
1(U)

The maps O
P
1(U) k(P
1
) k(X) are inclusions, and hence identify
O
P
1(U) with a subring of k(X). We claim that A(U) is the integral
closure of O
P
1(U) in k(X). Since the map f is nite, A(U) is a nite
O
P
1(U)-module, and hence A(U) is integral over O
P
1(U). This shows
52
that we can recover X over P
1
from the inclusion of elds k(P
1
) k(X)
(since X = Spec A). By the functoriality of this description of A, we get
a Z/2-action on A over O
P
1. This breaks A into eigenspaces; we write
A = A+ A, where the generator of Z/2 acts trivially on A+ and
by 1 on A. Both eigensheaves are locally free of rank one, for this
can be checked locally, say at the generic point, where its true because
k(X)/k(P
1
) is Galois.
Also, O
P
1 injects into A+, and in fact this is an isomorphism (check), so
there is some line bundle / on P
1
(so /

= O
P
1(d)) such that A

= O
P
1 /
as O
P
1-modules. How does the algebra structure of A look under this
isomorphism? We already know how to multiply two elements of O
P
1,
or an element of O
P
1 and an element of /, so the algebra structure on
O
P
1 / is determined by a map : /
2
O
P
1.
Pick an open Spec k[t] = A
1
P
1
; then O
A
1

=

k[t]. Since Pic A
1
is
trivial, /

A
1

= O
A
1

=

k[y]. The map

A
1
: O
2
P
1

=

k[y
2
]

k[t]

= O
A
1
sends y
2
to an element g(t) k[t]. Thus
A

A
1

= O
A
1 /

A
1

= k[t, y]/(y
2
g(t)),
with the Z/2 action given by y y. In conclusion, hyperelliptic curves
X are given (in an open ane A
2
P
2
) by equations of the form y
2
= g(t).
We will see next lecture how to (almost) obtain the degree of g(t) solely
from the genus of X.
16 Friday, Feb. 24th: Riemann-Hurwitz Formula
Today we show how to use a morphism f : X Y of curves to relate
the genus gX of X to the genus gY of Y . The key ingredient in this
computation is the familiar exact sequence
f

1
Y/k

1
X/k

1
X/Y
0.
We will use the following terminology: the morphism f is separable if
the corresponding eld extension k(Y ) k(X) is separable.
16.1 Preparatory Lemmata
Lemma 16.1. The map f

1
Y/k

1
X/k
is injective if and only if f is
separable.
Proof. Injectivity of can be checked at stalks. Both X and Y are integral
schemes, so the vertical maps in the following diagram are injective:
_
f

1
Y
_
x

1
X,x

_
f

1
Y
_

1
X,
where is the generic point of X. Thus the top arrow is injective if and
only if the bottom one is, so it suces to check injectivity at the generic
53
point. The inclusions into X and Y of the respective generic points gives
a bered diagram
Spec k(X)

Spec k(Y )

Y
This implies that
_

1
X/Y
_

=
1
k(X)/k(Y )
, and this is zero if and only if f
is separable, as we checked on a previous HW assignment.
Lemma 16.2. The diagram
Pic(X)
deg
X

Z
Pic(Y )
f

deg
Y

Z
deg f

commutes.
Proof. Recall that any line bundle on Y has the form OY (

nP P), and
the degree of such a line bundle is just

nP . For the proof, it is enough

to check it on the generators of Pic(Y ), namely those line bundles of the
form OY (P) for P Y . Since such a line bundle has degree one, we
need to show that deg f = deg
X
(f

OY (P)). Let 1P be the ideal sheaf of

the inclusion P Y . Then f

1P is an ideal sheaf on X, isomorphic to

f(Q)=P
1
e
Q
Q
. Therefore
deg
X
(f

OY (P)) = deg
X
(f

1P ) = deg
X
_

f(Q)=P
1
e
Q
Q
_
=
_

f(Q)=P
eQ
_
= deg f,
where the last equality is from Proposition 14.1.
Our nal lemma concerns the local structure of the module of dier-
entials.
Lemma 16.3. Let X be a uniformizer of the local ring OX,P . Then

1
X,P

= OX,P dX.
Proof.
1
X,P
is generated by elements of the form df, for f OX,P . But
df = d(f f(P)), since d(f(P)) = 0. Since f f(P) is in the maximal
ideal m of OX,P , we can write it as gX. Then
df = d(gX) = g dX +X dg.
Thus the image of dX generates
1
X,P
/m
1
X,P
, and therefore dX gen-
erates
1
X,P
by Nakayama.
54
16.2 Riemann-Hurwitz Formula
We now assume that f is separable, so by Lemma 16.1 the sequence
0 f

1
Y/k

1
X/k

1
X/Y
0.
is exact. In particular, is injective; we want to understand its image.
For this we look at the stalk, where the sequence looks like
0
1
Y,f(P)
O
Y,f(P)
OX,P

P

1
X,P

1
X/Y,P
0.
Let Y be a uniformizer for O
Y,f(P)
and X a uniformizer for OX,P .
We can write the image of Y in OX,P as u
e
P
X
for some unit u and
integer eP 1. Since
1
Y,f(P)
is generated as an OY,P -module by dY ,
f

1
Y,f(P)

= dY OX,P , so using this isomorphism, P sends dY to

d(u
e
P
X
) =
e
P
X
du +epu
e
P
1
X
dX.
If it happens that the characteristic of k divides eP , the situation is
considerably more complicated, and gets a name, but we will not treat
this case.
Denition 16.1. The morphism f has wild ramication at P if the
characteristic of k divides eP .
So assume that the characteristic of k does not divide eP . Then fol-
lowing P with the quotient map to
1
X,P
/m
e
P

1
X,P
, the term
e
P
X
du
in the above vanishes, so the image is generated by
e
P
1
X
dX. By
Nakayama, then, the image of P itself is generated by
e
P
1
X
dX. Thus
f

1
Y,f(P)

=
e
P
1
X

1
X,P
. Allowing P to vary, we obtain the main theorem
of the lecture.
Theorem 16.1. If f is separable and has no wild ramication, then
f

1
Y
_

PX
(eP 1) P
_

=
1
X
Theorem 16.2 (Riemann-Hurwitz). If f is separable and has no wild
ramication, then
2gX 2 = deg f(2gY 2) +

PX
(eP 1)
Proof. Take degrees in Theorem 16.1.
Example 16.1. Let f : X P
1
be a degree two map, i.e., X be a
hyperelliptic curve. By the previous lecture, we can write the equation
for X as y
2
= (x a1) (x ar), so f is ramied at the points a
k
and
possibly at . For those points at which f is ramied, its ramication
index is two, so using 16.2 we nd
2gX 2 = 2(2) +r +,
where comes from the ramication at : it is 1 if f is ramied there,
and 0 if not. By parity, we see that if r is odd, = 1, so f is ramied at
and gX = (r 1)/2. If r is even, f is unramied at and gX = (r 2)/2.
This makes good on the promise made at the end of the previous lecture.
Example 16.2. As a reality check, consider the case where r = 3, i.e., X
is an elliptic curve. Its ramied at innity, and the computation above
agrees with our knowledge that gX = 1. Phew!
55
17 Monday, Feb. 27th: Homological Algebra I
Lecture and typesetting by Piotr Achinger
17.1 Motivation
17.1.1 Algebraic topology
was the main motivation for the concepts of homological algebra. The
term homology itself comes from Poincares rst attempts on algebraic
topology. Given a nice topological space
7
X and a commutative ring R,
one can associate to it the following sequence of maps between R-modules:
0 C
0
(X)
d
C
1
(X)
d
C
2
(X)
d
. . .
where C
i
(X) is the dual of the free R-module spanned by the set of all
continuous maps f : [0, 1]
n
X (,,singular cubes in X) and the maps d
are dual to the maps taking such ,,singular cubes to their ,,boundaries.
It is intuitively clear that the boundary of the boundary is zero, i.e., dd =
0. We can therefore look at the cohomology groups
8
:
H
i
(X, R) :=
ker(d : C
i
(X) C
i+1
(X))
im(d : C
i1
(X) C
i
(X))
.
These encode important topological information of X, and have many
useful properties, to note a few:
(a) H
0
(X, R) is a free R-module spanned by the (path-)connected com-
ponents of X.
(b) The H
i
(X, R) are functorial in both X and R contravariantly in
X and covariantly in R (the map H
i
(f) induced by f : X Y is
usually denoted by f

),
(c) given a nice inclusion
9
i : Y X, one gets a long cohomology exact
sequence
H
i1
(Y, R)

H
i
(Z, R)
p

H
i
(X, R)
i

H
i
(Y, R)

H
i+1
(Z, R) . . .
(10)
where Z = X/Y is X with Y contracted to a point, and p : X Z
is the projection.
17.1.2 Algebraic geometry a.k.a. the Black Box
Our main goal is to dene and study the sheaf cohomology functors H
i
(X, F),
where X is a topological space and F a sheaf of abelian groups on X.
We already know some of their properties, having used them as a ,,black
box to study algebraic curves:
(a) H
0
(X, F) = (X, F),
7
e.g. a CW-complex
8
If we didnt take duals here, we would get homology groups but cohomology serves a
better motivation for our purposes.
9
e.g. a cobration
56
(b) The H
0
(X, F) are functorial in both X and F contravariantly in
X and covariantly in F (contravariance in X is to be understood as
follows: if f : X Y is a continuous map and G is a sheaf on Y ,
we get a map H
i
(Y, G) H
i
(X, f
1
G)).
(c) Given a short exact sequence 0 F G H 0 of sheaves on
X, one gets a long cohomology exact sequence
H
i1
(X, H )

H
i
(X, F) H
i
(X, G) H
i
(X, H )

H
i+1
(X, F)
(11)
Propaganda 17.1. Our goal, motivated by algebraic topology, is to de-
velop a general notion of a ,,cohomology theory, encompassing (17.1.1)
and (17.1.2) above
10
and many other examples in a uniform fashion. Our
second goal is to develop a theory which would handle the non-exactness
of the functor (X, ) (and other non-exact functors).
17.2 Complexes, cohomology and snakes
We start with a rather lengthy denition. I hope most readers will be
familiar with most of the notions presented here.
Denition 17.2. Let R be a commutative ring.
1. A complex of R-modules is a pair C

= (C
n
, d
n
) where C
n
(n Z)
are R-modules and d
n
: C
n
C
n+1
are maps (called dierentials)
satisfying the condition d
n+1
d
n
= 0. (In practice, one usually
writes d instead of d
n
, so our condition can be written succintly as
d
2
= 0).
2. A morphism of complexes f

: C

is a family of maps f
n
:
C
n
D
n
commuting with the dierentials, i.e., d
n
D
f
n
= f
n+1

d
n
C
. One therefore gets a category Ch(R mod) of complexes of
R-modules.
3. The cohomology groups of a complex C

are the groups H

i
(C

) :=
ker d
n
/ imd
n1
. One easily checks that they are covariant functors
H
i
: Ch(Rmod) Rmod. Elements of ker d
n
are called cycles
and denoted by Z
n
(C) and elements of imd
n1
are called boundaries
and denoted by B
i
(C). Therefore H
i
(C) = Z
i
(C)/B
i
(C).
4. The kernel, image and cokernel of a morphism of complexes are
dened in the usual way. One can then talk about subcomplexes,
quotient complexes, exact sequences of complexes, complexes of com-
plexes, cohomology complexes of complexes of complexes and so on.
Propaganda 17.3. Here is one principle of our approach to homological
algebra: to study an object A, associate to it a complex C

(A). This
complex will depend on the ,,cohomology theory. In other words to
construct a cohomology theory H

, one rst describes a way to attach a

complex C

(A) to A and dene H

i
(A) as H
i
(C

(A)). Usually C

(A) is
not functorial in A (but its cohomology is).
10
In fact (again for ,,nice spaces), (1) is a special case of (2): H
i
(X, R) = H
i
(X, R) where
R is the constant sheaf associated to R.
57
How do we get long exact sequences? We extend our principle as
follows: if an object X ts into a ,,short exact sequence Y X Z,
then there should be an exact sequence of complexes 0 C

(Y )
C

(X) C

(Z) 0. Then we get a long exact sequence from the Snake

Lemma (below). This encompasses both long exact sequences (10) and
(11).
Lemma 17.1 (Snake Lemma
11
). (a) Given a commutative diagram of
R-modules with exact rows
. C D E 0
0 C

.
f g
h
there exists an arrow : ker g coker f such that the sequence
ker f ker g ker h

coker f coker g coker h
is exact. Here is the whole picture:
ker f ker g ker h
C D E 0
0 C

coker f coker g coker h

f
g
h

(b) The arrow is functorial, that is, given two diagrams as above and a
commutative system of arrows between them, the two -arrows keep
the system commutative.
(c) Given a short exact sequence of complexes 0 C

i
D

p
E

0,
there is a long exact sequence of cohomology
. . . H
i1
(D

)

H
i
(C

)
i
H
i
(D

)
p
H
i
(E

)

H
i+1
(C

) . . .
58
(d) The arrows above are functorial, that is, given two short exact se-
quences as above and a commutative system of arrows between them,
one gets a commutative ladder between the two long cohomology ex-
act sequences.
Proof. You should prove (a) and (b) yourself. For (c), apply (a) to the
diagram
. C
i
/B
i
(C) D
i
/B
i
(D) E
i
/B
i
(E) 0
0 Z
i+1
(C) Z
i+1
(D) Z
i+1
(E) .
d
i
C
d
i
D
d
i
E
(it is straightforward to check that the rows are exact) obtaining
H
i
(C) H
i
(D) H
i
(E)
C
i
/B
i
(C) D
i
/B
i
(D) E
i
/B
i
(E) 0
0 Z
i+1
(C) Z
i+1
(D) Z
i+1
(E)
H
i+1
(C) H
i+1
(D) H
i+1
(E)
d
i
C
d
i
D
d
i
E

Then (d) follows from (b) and (c).

17.3 Homotopy
In Section 1, we discussed cohomology groups H
i
(X, R) of a topological
space X. In fact, these are homotopy invariants of X: that two maps
f, g : X Y are homotopic if there exists a continuous map H : X
[0, 1] Y such that H(x, 0) = f(x) and H(x, 1) = g(x). One proves that
homotopic maps induce the same maps on cohomology groups. To prove
this, one rst constructs a family of maps s
i
: C
i
(Y ) C
i1
(X) such
59
that ds +sd = f

, as in the diagram below.

. . . C
i1
(Y ) C
i
(Y ) C
i+1
(Y ) . . .
. . . C
i1
(X) C
i
(X) C
i+1
(X) . . .
d
i1
Y
d
i
Y
d
i1
X
d
i
X
g

s
i+1
s
i+1
Then, the statement follows formally from the equation ds+sd = f

g
12
.
Denition 17.4. Let C

, D

be complexes and let f, g : C

be
two maps of complexes. We say that f and g are homotopic (denoted
f g) if there exists a family of maps s
i
: C
i
D
i1
such that
f
i
g
i
= d
i1
D
s
i
+s
i+1
d
i
C
.
Here is the diagram for your convenience:
. . . C
i1
C
i
C
i+1
. . .
. . . D
i1
D
i
D
i+1
. . .
d
d
g f s s
Lemma 17.2. If f g, then H
i
(f) = H
i
(g) for all i.
Proof. Let z Z
i
(C). We want to prove that there is an x D
i1
such
that f(z) g(z) = dx. But
f(z) g(z) = d(s(z)) +s(d(z)) = d(s(z)) + 0
since dz = 0 by assumption. Therefore we can take x = s(z).
Some more vocabulary:
Denition 17.5. The family of maps s
i
is called a homotopy between
f and g. A map which is homotopic to zero is called nullhomotopic. We
call an f : C

a homotopy equivalence if there exists a g : D

such that f g is homotopic to idD and g f is homotopic to idC.

Remark. By the Lemma, a homotopy equivalence induces an isomor-
phism on cohomology groups (such a map is called a quasi-isomorphism).
The problem with quasi-isomorphisms is that they are not preserved
under additive functors, for example a short exact sequence of sheaves
0 F

F

F

0 can be seen as a quasi-isomorphism as in the

diagram below:
. . . 0 F

0 0 . . .
. . . 0 F F

0 . . .

12
For the construction of s
i
, see e.g. A. Hatcher Algebraic Topology, Theorem 2.10
60
This map of complexes will not however remain a quasi-isomorphism af-
ter applying (X, ). On the other hand, homotopy equivalences are
preserved by additive functors, by the nature of their denition. This
means that homotopy equivalence is a much better behaved notion. In
fact, they will play a key role in our construction of ,,cohomology theories
(i.e., derived functors).
18 Wednesday, Feb. 29th: Homological Algebra
II
Lecture and typesetting by Piotr Achinger
Here is what we know after Lecture 1:
we know what is a complex (of abelian groups or R-modules), what
are its cohomology groups, what it means for it to be exact,
we know the Snake Lemma and that a short exact sequence of com-
plexes gives us a long cohomology exact sequence,
we know the notion of homotopy of maps between complexes and
that homotopic maps induce the same map on cohomology.
18.1 Additive and abelian categories
A model example of a category in which one can make homological con-
siderations is the category R- Mod of (left) modules over a ring R.
Observation 18.1. In the category R- Mod there exist:
(a) a zero (initial and terminal) object 0.
(b) a structure of an abelian group on the sets of morphisms Hom(M, N)
satisfying the distributivity laws of left and right composition with
repect to addition (i.e. for f : M N the functions
x x f : Hom(N, N

) Hom(M, N

),
x f x : Hom(M

, M) Hom(M

, N)
are group homomorphisms).
(c) direct sums MN which are simultaneously products (i.e. we have
both the projections pM : M N M, pN : M N N and
the inclusions iM : M M N, iN : N M N satisfying
pM iM = idM, pN iN = idN, iM pM +iN pN = idMN).
Denition 18.2. A category satisfying (a)(c) above is called additive
13
.
An additive functor is a functor F : C D between additive categories
taking 0 to 0, direct sums to direct sums and such that the induced maps
Hom(M, N) Hom(FM, FN) are group homomorphisms.
In an additive category, it makes sense to talk about complexes, but
not cohomology. Therefore we need additional properties.
13
One may think that ,,additive is an extra structure on a category. In fact, whenever
a category is additive, the group structure on the sets Hom(M, N) is uniquely determined.
Therefore ,,additive is a property of a category
61
Observation 18.3. In the category R- Mod there also exist:
(d) kernels, i.e. equalizers of any morphisms with the zero morphism
(that is, for any f : M N there is an object ker f = K together
with a map i
f
= i : K M with the following universal property:
given g : M

N such that f g = 0, there is a unique g : M

K
satisfying g = i g),
(e) cokernels, i.e. coequalizers of any morphisms with the zero morphism
(that is, for any f : M N there is an object coker f = C together
with a map p
f
= p : N C with the following universal property:
given g : N N

such that g f = 0, there is a unique g : C N

satisfying g = g p),
(f) canonical isomorphims ker(p
f
) = coker(i
f
), that is ,,the cokernel of
the kernel is the kernel of the cokernel (by the universal property of
kernel and cokernel, there is a canonical map ker(p
f
) coker(i
f
)
we require it to be an isomorphism).
Denition 18.4. An additive category satisfying additionally (d)(f) is
called abelian. In an abelian category, the usual notions of cohomology of a
complex or an exact sequence make sense. An additive functor F : C D
between two abelian categories is called exact if it takes kernels to kernels
and cokernels to cokernels (or equivalently takes exact sequences to
exact sequences).
Observation 18.5. In the category R- Mod we have the Snake Lemma,
which gives us long cohomology exact sequences for short exact sequences
of complexes.
At the rst glance, it is not obvious how the proof of the Snake Lemma
should go in any abelian category recall that our proof used ,,elements
of the objects in the diagram. However, the Snake Lemma and its con-
clusions hold in any abelian category, thanks to the following result:
Theorem 18.1 (Freyd-Mitchell, 1964). Let A be a small
14
abelian cat-
egory. Then A is a full abelian subcategory of the category R- Mod of
modules over a certain ring R. More precisely, there exists a ring R and
a fully faithful exact additive functor F : A S Mod.
Remark. In fact, the Freyd-Mitchell theorem allows us to perform dia-
gram chasing in any abelian category given a diagram (whose objects
form a set) we can pass to the smalles abelian subcategory containing the
objects of the diagram (which should be small). I will ignore set theoretic
issues of this type from now on.
Remark. The category Ch(A) of complexes of objects of an abelian
category A is abelian (we take sums, kernels and cokernels ,,levelwise).
The functors
H
i
: Ch(A) A, H
i
(A

) = ker(d
i
)/im(d
i1
)
are additive, but obviously not exact.
Our considerations concerning R- Mod and the Freyd-Mitchell theorem
show that
14
A category is small if its objects form a set.
62
Theorem 18.2. If A is an abelian category and : 0 A

f
B

0 is a short exact sequence of complexes in A, then there exists the

long cohomology exact sequence
. . . H
i1
C

i1

H
i
A

H
i
f
H
i
B

H
i
g
H
i
C

. . .
which is natural in the following sense: if

: 0 A

f
B

g
C

0
is another such sequence and we are given a morphism of short exact
sequences
15
:

, then the diagrams

H
i
C

H
i+1
A

H
i
C

H
i+1
A

commute.
Examples 18.6. The following additive categories are abelian:
abelian groups, R-modules,
sheaves and presheaves of abelian groups on a topological space X,
sheaves of OX-modules on a ringed space (X, OX),
coherent and quasi-coherent OX-modules on a scheme X,
complexes in an abelian category A.
and the following are not:
nitely generated free abelian groups,
vector bundles (i.e., locally free OX-modules of nite rank) on a
scheme X.
18.2 -functors
Propaganda 18.7. Recall that our slogan was ,,replace the objects we
study by complexes. One can nd manifestations of this idea on every
level of homological algebra, the most rened of them being the notion of
a derived category, developed by J.-L. Verdier in 1960s.
Trying to realize our motto in a naive way, imagine that to a given
object A of an abelian category A there is assigned a complex X

in
some other abelian category B. For simplicity (and better analogy with
our motivational examples coming from topology) we will assume that
X
i
= 0 for i < 0.
With this picture there should be an associated cohomology theory,
that is, a sequence of functors H
i
(A) = H
i
(X

) and for every short exact

sequence 0 A B C 0 a sequence of arrows
i
: H
i
(C)
H
i+1
(A) (as if 0 A B C 0 gave a short exact sequence of
complexes), giving a long cohomology exact sequence.
The notion realizing the above picture is called a -functor.
15
This is simply a commutative ,,ladder between and

.
63
Denition 18.8. Let A and B be abelian categories. By a (covariant,
cohomological) -functor T

from A to B we mean the following data

a sequence of additive functors T
i
: A B (i 0),
a sequence of morphisms
i

: T
i
C T
i+1
A (i 0) for any short
exact sequence : 0 A B C 0 in A,
satisfying the following conditions
1. for any exact sequence : 0 A
f
B
g
C 0 in A the sequence
0 T
0
A
T
0
f
T
0
B
T
0
g
T
0
C

. . . T
i
A
T
i
f
T
i
B
T
i
g
T
i
C

T
i+1
A . . .
is exact in B,
2. for any two exact sequences : 0 A B C 0 and

: 0
A

0 and a morphism :

the diagram
T
i
C T
i+1
A
T
i
C

T
i+1
A

commutes
Remark. There also exist notions of a contravariant -functor and of a
homological -functor.
Propaganda 18.9. What we would also like to have (this condition may
not sound very well motivated to a person without background from say
algebraic topology) is for this ,,cohomology theory to be ,,determined by
its coecients where by ,,coecients we mean the functor T
0
. Indeed,
the cohomology functors H
i
(X, R) satisfy
1. H
0
(X, R) = R for X connected,
2. if H
i
(X) is an ordinary cohomology theory, then H
0
(pt) determines
the whole theory (at least for CW-complexes),
3. a morphism R R

of coecient rings extends uniquely to a ,,mor-

phism of theories, i.e., a system of natural transformations H
i
(, R)
H
i
(, R

), compatible with the connecting homomorphisms in the

long exact sequences.
Another motivation comes from our Black Box we want universal func-
tors H
i
satisfying the properties we need.
To understand, what 3. above should mean in the context of -
functors, we should introduce the notion of a morphism of -functors
(,,natural transformation of cohomology theories in fact the motivating
example for Eilenberg and MacLanes denition of a natural transforma-
tion of functors).
64
Denition 18.10. Let T

, T

be two -functors from an abelian category

A to an abelian category B. A morphism of -functors : T

is
a sequence of natural transformations
i
: T
i
T
i
which commute with

i
,
i
, that is, for every short exact sequence : 0 A B C 0
the diagram
T
i
C T
i+1
A
T
i
C T
i+1
A

i
C

i+1
A

commutes.
We can now say what it means that a -functor T

is determined by
its coecients T
0
.
Denition 18.11. Let T

be a -functor from an abelian category A

to an abelian category B. We say that T

is universal if for every other

-functor S

from A to B and a natural transformation : T

0
S
0
there is a unique morphism of -functors : T

with
0
= .
As usual, the universal property implies that whenever a universal
-functor with a xed T
0
exists, it is unique.
18.3 Left- and right-exact functors
What additive functors F : A B does there exist a (not necessarily
universal) -functor T

with T
0
= F? From the ,,long exact sequence
property it follows that for every short exact sequence 0 A B
C 0 the sequence
0 FA FB FC
(with no 0 on the right!) is exact.
Denition 18.12. Let A, B be abelian categories and let F : A B
be an additive functor. We call F left-exact if for any short exact sequence
0 A B C 0 in A, the sequence
0 FA FB FC
is exact; F is right-exact if for any short exact sequence 0 A B
C 0 in A, the sequence
FA FB FC 0
is exact
16
.
Remark. There also is a notion of a contravariant left- or right-exact
functor. We just consider such a functor as a covariant functor A
op
B.
16
That is, F is left/right exact i it preserves kernels/cokernels or equalizers/coequalizers
or (because F is additive) nite limits/colimits.
65
Examples 18.13. For any object A A, the functors Hom(A, ) and
Hom(, A) are left-exact. If A R- Mod, then AR is right-exact. If
f : X Y is a map of topological spaces, then f : Sh(X) Sh(Y ) is left-
exact and f
1
: Sh(Y ) Sh(X) is right exact (where Sh are the categories
of sheaves of abelian groups) and similarly for f and f

for sheaves of
modules if X and Y are ringed spaces. In particular, : Sh(X) Ab is
left-exact. More generally, if F : A B is left adjoint to G : B A
then F is right exact and G is left exact (Problem 4).
A few observations are in order. First, a functor is exact if and only if
it is both left- and right-exact. Secondly, a functor extends to a -functor
only if it is left-exact. Finally, if F : A B is exact, then T
0
= F,
T
i
= 0 for i > 0 denes a universal -functor. Therefore F can give an
interesting universal -functor only if it is left-, but not right-exact.
Propaganda 18.14. We now change our paradigm a little: we consider
(universal) -functors as a tool to study not (or not only) objects, but the
non-exactness of left-exact functors. That is, we focus on the second goal
mentioned in the rst lecture: to develop a theory which would handle the
non-exactness of the functor (X, ) (and other non-exact functors).
Our main goal is now the construction of a universal -functor extend-
ing a given left-exact functor (the construction for right-exact functors is
analogous) we shall call them (left) derived functors.
18.4 Universality of cohomology
To make sure that our approach may be right and imagine how the general
construction should work, we solve the following ,,toy problem: since the
main prototype of the notion of a -functor are the cohomology functors
H
i
: Ch

(A) A (Ch

(A) is the category of complexes concentrated

in degrees 0), do they form a universal -functor?
Proposition 18.1. The functors H
i
: Ch

(A) A together with the

morphisms from Theorem 2 form a universal -functor.
Proof (sketch). Let there be given a -functor S

from Ch

(A) to A
and a natural transformation
0
: H
0
S
0
. We shall construct the

i
: H
i
S
i
inductively.
Suppose that i1 has already been constructed. From the Lemma
below it follows that for any complex A

there is an injection A

into an exact complex I

. This gives us a short exact sequence of

complexes 0 A

0. Looking at the long cohomology

exact sequences gives us the diagram
H
i1
A

H
i1
I

H
i1
B

H
i
A

H
i
I

= 0
S
i1
A

S
i1
I

S
i1
B

S
i
A

S
i
I

i1

i1

i1

i
By some easy diagram chasing we convince ourselves that this determines

i
: H
i
A S
i
A. We then have to check that does not depend on the
choice of A

, that it is a natural transformation etc.

66
Lemma 18.1 (Problem 3). For any A

Ch(A) there is an injection

A

where I

Ch(A) and H
i
(I

) = 0 for all i.
Remark for algebraic topologists. We can look at the above proof
in the following way: we embed a given space A into the cone I = CA,
which is contractible. Then B = I/A = A is the suspension of A. By the
suspension axiom we get H
i
(A) = H
i1
(B) for any cohomology theory
H

. But on H
i1
everything is already dened, hence the inductive step.
An important observation here is that the proof of Proposition 18.1
essentially shows the following
Lemma 18.2 (Eaceable -functors are universal. See Problem 5). Let
T

: A B be a -functor which is eaceable, that is, for every A A

there exists an injection A J into a T

-acyclic object J (J is called

acyclic for T

if T
i
(J) = 0 for i > 0). Then T

is universal.
19 Friday, March 2nd: Homological Algebra III
Lecture and Typesetting by Piotr Achinger
Here is what we know after Lecture 2:
what is an additive or abelian category, what is an additive or exact
functor,
what is a -functor and what it means for it to be universal,
what is a left- or right-exact functor,
that an eaceable
17
-functor is universal.
Our goal is to extend a given left-exact functor to a universal -functor.
19.1 Derived functors a construction proposal
Let us think about what was crucial in our proof of ,,universality of co-
homology. The key idea was that any object A could be embedded into
an object I on which the given functor was ,,exact (that is, the ,,coordi-
nates of the functor for i > 0 vanished on I).
Lemma 19.1 (Eaceable -functors are universal). Let T

: A B be
a -functor which is eaceable, that is, for every A A there exists an
injection A J into a T

-acyclic object J (J is called acyclic for T

if
T
i
(J) = 0 for i > 0). Then T

is universal.
Let A and B be abelian categories and let F : A B be a left-
exact functor. We are looking for a universal -functor T

with T
0
= F.
Motivated by the above Lemma, we want to nd a good class I of objects
of A, such that
0. every object A A injects into an object I(A) I,
1. objects of I are acyclic for F.
17
A -functor T

is eaceable if any object injects into an object on which T

1
, T
2
, . . . vanish.
67
It is not clear what it should mean for an I I to be acyclic, since we
dont have a -functor yet, but only its zero component T
0
= F. But in
any case, the following should hold: whenever 0 I M N 0 is a
short exact sequence, 0 FI FM FN 0 is also exact (because
the next term after FN should be T
1
I = 0). This of course holds when
0 I M N 0 splits, since a split short exact sequence stays split
(therefore exact) after applying F. Let us be greedy here and ask for the
class of objects I for which all such sequences split.
Denition 19.1. An object I A is called injective if any short exact
sequence 0 I M N 0 splits. An object P A is called
projective if any short exact sequence 0 M N P 0 splits (that
is, if P is injective in A
op
).
Problem 6. What are the injective and projective objects in the category
of abelian groups?
The class of projective objects plays the same role as injective objects
when one wants to study right-exact functors (or contravariant left-exact
functors). Then one has to consider surjections P A and the whole
program carries over.
With these assumptions in place, set I to be the class of injective ob-
jects in A and let us try to construct T
i
using the principle of ,,induction
by dimension shifting as in the proof of Lemma 19.1. Given A A, nd
A I, with I I and let B be the cokernel:
0 A

I

B 0.
After applying F, we get
0 FA
F
FI
F
FB
which we want to continue to the right
0 FA
F
FI
F
FB T
1
A T
1
I . . . ,
and since we wish that T
i
I = 0 for i > 0, we should put T
1
A = coker F
and T
i+1
A = T
i
B for i 1.
Thus to compute say T
2
A, we should nd an short exact sequence
0 B J

C 0 and compute coker F. This seems already
cumbersome, but we can make a long exact complex 0 A I
J . . . (an ,,injective resolution of A) by ,,splicing all the short exact
sequences.
Construction. Given A A, construct an injective resolution A
68
I

of I as follows
0 0
B
2
0 A I
0
I
1
I
2
. . .
B
1
B
3
0 0 0 0
i
A
i
B
1
i
B
2
i
B
3
(where I
0
= I(A), B
1
= coker iA, I
i
= I(B
i
) for i > 0, B
i+1
=
coker(B
i
I
i
)). Then apply F to I

:
F(I

) : F(I
0
) F(I
1
) F(I
2
) . . .
and dene
T
i
A = H
i
(F(I

)).
By left-exactness of F, since 0 A I
0
I
1
is exact, 0 FA
FI
0
FI
1
is exact, hence T
0
A = FA. Obviously this agrees with our
previous approach.
Remark. It is convenient to view 0 A I

. . . as a quasi-
isomorphism A I

(where A is identied with the complex . . .

0 A 0 . . .). This means that in the derived category of A (the
category of complexes where quasi-isomorphisms are made to be actual
isomorphisms), A and I

are the same.

There are several obvious problems here:
2. Why should T
i
A be well-dened (i.e., independent of the choice of
A I

)?
3. Why should it be functorial in A?
4. And how to dene the connecting homomorphisms to get a -
functor?
Suppose that (2)-(4) were resolved. Then (1) follows, since we can
choose 0 I I 0 to be the injective resolution of I, which gives us
T
i
I = 0 for i > 0. If in additions (0) would hold, then by Lemma 19.1 we
have constructed a universal -functor.
19.1.1 Derived functors resolution of technical issues
Regarding (2) T
i
A are well-dened
Note that T
i
constructed as above will not depend on the choice of 0
A I

if any two such resolutions will be homotopic (since homotopic

maps
1. remain homotopic after applying any additive functor,
69
2. induce the same map on cohomology).
What we need is the following
Theorem 19.1. 1. for any two resolutions 0 A I

amd 0
A J

there exist , morphisms between 0 A I

i
0 A J

(i.e. d = d oraz d = d) giving the identity on A:

0 A I
0
I
1
I
2
. . .
0 A J
0
J
1
J
2
. . .

2. any two such , be the homotopy inverses of each other i.e. there
exist s
i
: I
i
I
i1
and t
i
: J
i
J
i1
(i 0, we put I
1
= A =
J
1
) such that id = ds +sd and id = dt +td:
0 A I
0
I
1
I
2
. . .
0 A I
0
I
1
I
2
. . .
id id id s s s s
0 A J
0
J
1
J
2
. . .
0 A J
0
J
1
J
2
. . .
id id id t t t t
Let us try to construct inductively the rst step looks as follows:
0 A I
0
I
1
I
2
. . .
0 A J
0
J
1
J
2
. . .
We would like to extend the dotted arrow A J
0
to I
0
. This follows
from
Problem 7. An object I A is injective i any morphism A I
dened on A B can be extended to B (or, equivalently, i the functor
Hom(, I) is exact).
Regarding (3) Functoriality
In fact, a more general version of Theorem 19.1 holds:
Theorem 19.2. Let f : A

A be a map in A and let A I

, A

be resolutions of A and A

, respectively, with I
i
injective (the objects I
i
do not have to be injective). Then there exists a chain map

f : I

,
inducing f on H
0
, which is unique up to homotopy.
70
Problem 8. Prove Theorem 19.2 and deduce Theorem 19.1.
Thus, the exercise above resolves both our issues with (2) and (3).
Regarding (4) Construction of
i
This is called the Horseshoe Lemma:
Lemma 19.2. Given a short exact sequence 0 A B C 0 and
already chosen injective resolutions A I

A
and C I

C
one can nd an
injective resolution B I

B
so that one obtains a short exact sequence of
complexes 0 I

A
I

B
I

C
0 which is 0 A B C 0 on
H
0
.
Problem 9. Prove the above statement. Hint: set I
0
B
= I
0
A
I
0
C
and
construct an arrow B I
0
B
. Use the Snake Lemma, obtaining a short
exact sequence 0 I
0
A
/A I
0
B
/B I
0
C
/C 0. Proceed by induction.
Thus, given a short exact sequence 0 A B C 0, we
obtain the morphisms
i
: T
i
C T
i+1
A from the long cohomology exact
sequence associated to the short exact (why?) sequence of complexes 0
F(I

A
) F(I

B
) F(I

C
) 0 where I

A
, I

B
, I

C
are as in the Horseshoe
Lemma.
We need to check that the morphisms are well dened and natural,
which is also left to the reader.
Regarding (0) Suciently many injective objects
The only problem may happen with (0). For example, if A is the category
of all nitely generated abelian groups, then A has no nonzero injective
objects!
Denition 19.2. An abelian category A is said to have suciently many
injective objects if for any object there exists an injection into an injective
one. We say that A has suciently many projective objects if for any
object there is a surjection from a projective one.
Theorem 19.3. For any ring R, the category R- Mod has suciently
many injective and projective objects.
19.2 Derived functors conclusion
Theorem 19.4. Let A and B be abelian categories and let T
0
: A B
be a left-exact functor. If A has suciently many injective objects, then
T
0
extends to a universal -functor T

with T
0
= F.
Denition 19.3. We denote the functors T
i
by R
i
T
0
and call the right
derived functors of T
0
.
A similar Theorem holds for right-exact functors and projective objects
(then one has to use homological -functor). The left-derived functors of
a right-exact functor T0 are denoted by LiT0.
Examples 19.4. Let A A, then Ext
i
(A, ) := R
i
Hom(A, ) and
Ext
i
(, A) = R
i
Hom(, A) (these two coincide, so one can compute
Ext
i
(A, B) from an injective resolution of B or a projective resolution
of A). For A R Mod, ri(A, B) := Li(A R B) (here one proves
71
that ri(A, B) ri(B, A)). For a topological space X, H
i
(X, ) :=
R
i
: Sh(X) Ab are called the sheaf cohomology functors (one checks
rst that the category of abelian sheaves on X has suciently many in-
jective objects). For a continuous map f : X Y , the derived functors
R
i
f : Sh(X) Sh(Y ) are called the higher direct image functors. For a
group G, the functor of G-invariants ()
G
: GMod Ab is left-exact
and we call its derived functors H
i
(G, ) = R
i
()
G
the group cohomology
functors. One checks easily that M
G
= Hom(Z, G) where Z has the triv-
ial Z-action. Therefore H
i
(G, M) = Ext
i
(Z, M) can be computed from a
projective resolution of Z, which is useful since we do not have to pick a
new resolution whenever M changes, and also because projective objects
are generally easier to work with than injective objects.
19.3 Two useful lemmas
Lemma 19.3 (Acyclic resolutions suce). Let F : A B be left-exact,
where A has suciently many injective objects. Let A A and let 0
A J
0
J
1
. . . be an exact sequence with all J
n
acyclic for F
(that is, R
i
F(J
n
) = 0 for i > 0 and all n 0). Then there are natural
isomorphisms H
i
(F(I

)) R
i
F(A).
Lemma 19.4. Let A, B and C be abelian categories, F : A B and
G : B C be left-exact functors. Suppose that A and B have enough
injectives (so that the derived functors R
i
F, R
i
G and R
i
(G F) exist)
and that F maps injective objects of A to G-acyclic objects of B (that is,
if I A is injective, then (R
i
G)(F(A)) = 0 for i > 0). Let A A. Then
(a) if (R
i
F)(A) = 0 for i > 0, then R
i
(G F)(A) (R
i
G)(F(A)),
(b) if (R
i
G)(F(A)) = 0 for i > 0, then R
i
(G F)(A) G(R
i
F(A)).
In general, under the assumptions of the lemma, there is a spectral
sequence
E
pq
2
= (R
p
G)(R
q
F(A)) R
p+q
(G F)(A).
20 Monday, March 5th: Categories of Sheaves
Have Enough Injectives
We have seen last week how to construct the right derived functors of
an additive left-exact functor, so long as the categories we are interested
in have enough injectives. Today we verify that this is the case for the
categories we are interested in, namely sheaves of OX-modules ModO
X
,
or perhaps just sheaves of abelian groups Sh(X). The functor whose right
derived functors we construct is f, where f : X Y is a morphism of
schemes. We will be especially interested in the case where Y = Spec k; in
this case f is the just the global sections functor (X, ): Sh(X) Ab.
Remark. In some sense, it wont really matter what categories between
which we regard f as a functor. In fact this is reected in the notation:
R
i
f does not specify Sh(X) or ModO
X
, etc. Of course, some care needs
to be taken. For instance, we will not consider categories of coherent
sheaves, because over Spec Z, there arent any (except for the zero sheaf):
72
the reason is that injective Z-modules are things like Q and Q/Z, which
are not nitely-generated.
The goal of todays lecture is to establish the following, which holds
even for ringed spaces with noncommutative rings:
Theorem 20.1. If (X, OX) is a ringed space, then ModO
X
has enough
injectives.
Remark. Note that by taking OX = Z, we obtain as a corollary that
Sh(X) has enough injectives.
Proof. We rst show that categories of modules over a ring have enough
injectives, and then globalize. The following simple lemma is at the heart
of both steps of the proof.
Lemma 20.1. If F : A B is an additive functor which has an exact
18
left adjoint L: B A, then F takes injectives to injectives.
Proof. Let I A be an injective object. To show that F(I) is an injective
object of B we must show there exists a dotted arrow in the following
commutative diagram in B
0

M

FI
By the adjunction, this is equivalent to lling in the dotted arrow in
0

LM

LN

I
which can be done because I is injective. Note that the map LM LN
is still injective because L is left exact.
Now let R be a ring (commutative for simplicity). Then ModR has
enough injectives. This follows easily from the following lemma.
Lemma 20.2. Let R be a ring. Then the forgetful functor L: ModR
Ab has a right adjoint F : Ab ModR, and if M ModR, and LM I
is a monomorphism in Ab, then the composite M FL(M) FI is a
monomorphism.
Proof. We dene, for A any abelian group, FA = Hom
Ab
(R, A) which
has an R-module structure given by (r f)(x) = f(rx). To see this gives
an adjunction we need to establish a natural bijection, for M ModR,
A Ab,
Hom
Ab
(LM, A) HomR(M, Hom
Ab
(R, A))
18
We actually only use left-exactness of L, but right exactness comes for free because left-
adjoints preserve colimits
73
The map going to the right sends f : M A to the map m (r
f(rm)). The map going to the left sends : M Hom
Ab
(R, A) to the
map m m(1), where m is the image of m under . We omit the
verication that these constructions are inverse and functorial in M and
A.
Now we must check that given an inclusion LM I, the map M
FL(M) FI is also an inclusion. But this map is exactly the map
coming from the adjunction Hom
Ab
(LM, I) HomR(M, FI). We need
to check that given such that
K

M

FL(M)
then actually is zero. Applying L and using the adjunction gives
LK
L

LM
Id

LM
Thus L = 0, hence = 0, since L is faithful (L is the same map as ,
it just ignores the R-module structure). This nishes the proof.
To complete the proof of the theorem, we now consider OX-modules.
Fix x X, i.e., a morphism j : x X, and for ease of notation, let
A = ModO
X
, the category of OX-modules, and B = ModO
X,x
, the
category of modules over the local ring OX,x at x. Then we have an
adjoint pair of functors
A
j
1

B
j
.
Now j is a right adjoint, so if I B is injective, then so is jI, by the
lemma
19
.
For the remainder of the discussion, x T A, and since Bhas enough
injectives, pick for each x an inclusion Tx I
(x)
for some injective OX,x-
module I
(x)
. Then we have a map
T

xX
jxI
(x)
,
where each map jx is the inclusion of the point x. A product of injectives
is still injective, coarsely speaking, because both products and injectives
are dened by maps in. Moreover, is a monomorphism. We check this
at the stalk at an arbitrary point z X:
Tz
z

xX
jxI
(x)
_
z
jzI
(z)

= I
(z)
,
19
Note that j
1
is exact: in general, for f : X Y , and F a sheaf on Y , the stalk of f
1
F
at x X is the same as the stalk of F at f(x) Y .
74
where the isomorphism on the right is because jz and j
1
z
are an adjoint
pair and j
1
is exact. The composite map here is just the chosen inclusion
Tz I
(z)
. Since this is injective, so is z. Thus we have take an arbitrary
sheaf T of OX-modules, and embedded it into an injective object of A,
which shows that A has enough injectives.
21 Wednesday, March 7th: Flasque Sheaves
Having shown that ModO
X
has enough injectives, we can dene , for T
an OX-module
H
i
(X, T) = H
i
_
(X, I
0
) (X, I
1
) (X, I
2
)
_
,
where T I

is an injective resolution of T. However, in practice it

is easier to work with asque sheaves, which we now dene, rather than
injectives.
Denition 21.1. A sheaf T on a topological space is asque if for every
inclusion, U V , T(V ) T(U) is surjective.
This notion is useful because, in consideration of whether a sheaf is
asque, it does not matter whether we regard it as a sheaf of O-modules
or abelian groups. Moreover, every injective sheaf is asque. In the proof,
we will use the following.
Denition 21.2. Let j : U X be an open subset, and T be a sheaf of
OU-modules. Then the extension by zero of T to X is dened to be
the sheacation of the presheaf
W
_
T(W) if W U
0 if not.
It is denoted j
!
T.
Note that j
!
T has stalks equal to those of T inside U, and zero else-
where. Also, j
!
T is a left adjoint to j
120
.
Lemma 21.1. Let (X, OX) be a ringed space. If I ModO
X
is injective,
then I is asque.
Proof. Let V U be two opens in X, with inclusions i : U X, j : V
U. By the adjunction, we have for any sheaf T of OX-modules,
HomO
X
(i
!
OU, T) HomO
U
(OU, i
1
T) T(U)
Now there is an inclusion of OU-modules j
!
j
1
OV OU, which gives a
map
HomO
X
(i
!
OU, T) HomO
X
(i
!
(j
!
j
1
OV ), T)
by restricting along the above inclusion. But HomO
X
(i
!
OU, T) T(U),
and similarly HomO
U
(OU, i
1
T) T(V ). So we have a natural map
T(U) T(V ). This is for any T, but now take T = 1 an injective
20
Remember that j
1
is itself a left adjoint to j.
75
OX-module. Then since i
!
is exact
21
, the inclusion j
!
j
1
OV OU gives
an inclusion i
!
j
!
j
1
OV i
!
OU, and then injectivity implies the map
HomO
X
(i
!
OU, 1) HomO
X
(i
!
(j
!
j
1
OV ), 1) is surjective, i.e., 1(V )
1(U) is surjective.
Our next goal is to show that asque sheaves are acyclic. For this we
need a(nother) lemma.
Lemma 21.2. Let T be a asque sheaf of OX-modules, and embed T 1
into an injective sheaf 1, with cokernel (. Then for all opens U of X,
1(U) ((U) is surjective.
Proof. We have an exact sequence 0 T 1 ( 0. Fix a section
g ((U). By exactness at the stalk, for each point of U, there is a
neighborhood V over which we can lift g

V
to some gV 1(V ). Suppose
we have also some other neighborhood W, and a lifting gW of g

W
to
1(W). We would like to glue these to a lift in 1(V W). Since gV

WV
and gW

WV
both map to g

WV
in ((W V ), the exactness of the
above sequence implies that their dierence gV

WV
gW

WV
lies in
T(W V )
22
. Since T is asque, T(W) surjects onto T(W V ), so there
is some T(W) which maps to gV

WV
gW

WV
in T(WV ). Since
maps to zero in ((W), gW + 1(W) is a lift of g

W
to 1(W). Moreover,
gV

WV
( gW + )[WV = 0 in 1(W V ), so they glue to give a lifting
of g over W V .
Thus we have shown that we can lift sections of ( one open at a
time. We now apply Zorns lemma to the set
|(V, gV [ gV lifts g over V ,
partially ordered by (V, gV ) (V

, g
V
) if V V

and g
V

V
= gV . Any
increasing chain of such objects has an upper bound by taking the union
of the V . Thus we have a maximal such pair, which must be a lifting of
g over all of U: if it were not, we could extend further by the above and
violate maximality.
23
We can now prove
Proposition 21.1. Let T be a asque sheaf of OX-modules. Then H
i
(X, T) =
0 for i > 0.
Proof. First we pick an embedding T 1 of T into an injective 1. Then
1 is asque by Lemma 21.1. The quotient ( is also asque, because if
V U is an inclusion, then we have a diagram
1(U)

((U)

1(V )

((V )
21
This follows from the description of the stalks.
22
We are really using the left-exactness of the functor (W V, ) here.
23
Cf. also Hartshorne Ex II.1.16
76
in which the horizontal arrows are surjective by Lemma 21.2. This implies
that the vertical arrow on the right is surjective also, hence ( is asque.
Now H
1
(X, T) = 0 since it is the cokernel of the map (X, 1) (X, (),
which is surjective by the previous lemma. The short exact sequence
0 T 1 ( 0 gives rise to a long exact sequence in cohomology,
in which the middle terms H
r
(X, 1) are zero because 1 is injective. Thus
we obtain isomorphisms H
i
(X, () H
i+1
(X, T). Since ( is asque, the
result follows by induction.
22 Friday, March 9th: Grothendiecks Vanishing
Theorem
22.1 Statement and Preliminary Comments
Ther next few lectures will be devoted to the proof of the following
Theorem 22.1. If X is a Noetherian topological space of dimension n,
and T a sheaf of abelian groups on X, then H
i
(X, T) = 0 for i > 0.
Our primary application of this theorem will be when (X, OX) is a
ringed space and T an OX-module. In order to do so, we will need
the fact that computations of cohomology in the categories of abelian
groups give the same result. More precisely, we have a universal -functor
|H
i
M
(X, ) from ModO
X
Ab (the subscript M is for module), and
also a -functor |H
i
A
(X, ) from Sh(X) Ab (A is for abelian group),
which is dened by regarding a sheaf of abelian groups T Sh(X) as a
sheaf of Z-modules, where Z is the constant sheaf
24
There is also the
forgetful functor : ModO
X
Sh(X). Since is exact, the composition
|H
i
A
(X, ) is a -functor. We ask whether it is isomorphic, as -
functors, to |H
i
M
(X, ).
Theorem 22.2. If (X, OX) is a ringed space and T an OX-module, then
H
i
A
(X, T) H
i
M
(X, T).
Proof. We have an isomorphism
H
0
M
(X, ) H
0
A
(X, ())
which, by the universality of H
i
M
, induces a morphism of -functors
|H
i
M
(X, ) |H
i
A
(X, ) .
This is an isomorphism because |H
i
A
(X, ) is also universal. This
follows because it is eaceable: for T ModO
X
, pick an embedding
into an injective OX-module 1. This injective is also asque, and so 1 is
asque in Sh(X). Flasque sheaves in Sh(X) are acyclic for H
i
A
by applying
Lemma 21.1 with OX = Z.
24
Note that the same argument as in the case of Mod
O
X
shows that Sh(X) has enough
injectives, by taking O
X
= Z.
77
22.2 Idea of the Proof; Some lemmas
The proof of the vanishing theorem begins by reducing rst to the case
when X is irreducible, and then reducing to the case where T = j
!
Z for
j : U X and open. Then we use induction on the dimension of X. To
make the reduction on T, we will need to know how asque sheaves and
cohomology interact with direct limits.
Lemma 22.1. On a Noetherian topological space, a direct limit of asque
sheaves is asque.
Proof. HW exercise.
The next lemma says that cohomology commutes with direct limits.
More precisely, if (T) is a directed system in Sh(X), then for each
we have a map T lim

T. Applying H
i
we get maps H
i
(X, T)
H
i
(X, lim

T) for each . By the universal property of direct limit, these

induce a map
lim

H
i
(X, T) H
i
(X, lim

T)
Lemma 22.2. If X is a Noetherian topological space, the map lim

H
i
(X, T)
H
i
(X, lim

T) is an isomorphism.
Proof. Let A be a directed set, which we can regard as a category. We
denote by Sh(X)
A
the category of functors A Sh(X). Note that it is
an abelian category; kernels, cokernels, etc., are dened index by index.
It also has arbitrary products, and enough injectives. From the following
diagram
Sh(X)
A
lim

Sh(X)

Ab
A
lim

Ab
we obtain two -functors Sh(X)
A
Ab, namely the right-derived functors
of the two dierent compositions in the square. More precisely, the two
-functors in question are
1. (T) H
i
(X, lim

T)
2. (T) lim

H
i
(X, T)
We will be done if we can show that these are both universal, for then the
universality gives a canonical isomorphism between them. We show that
they are both eaceable. The reason is that for any sheaf T, there is a
functorial embedding T

T of T into a asque sheaf, namely take

T to
be the so-called sheaf of discontinuous sections of T, given by

T(U) = |s: U
_
pU
TP

s(p) Fp for each p.

Then the rst -functor is eaceable since this is functorial, i.e., given an
object (T) Sh(X)
A
, the construction produces a system

T of asque
sheaves, whose direct limit is asque by Lemma 22.1, hence H
i
(X, lim

T) =
78
0 for i > 0. The second is eaceable since each H
i
(X,

T) = 0 for i > 0,
so lim

H
i
(X, T) = 0.
23 Monday, March 12th: More Lemmas
We stated without proof last time that the functor lim

: Sh(X) Ab is
exact. Lets prove that now.
Proposition 23.1. For A a directed set, the functor lim

: Sh(X)
A

Sh(X) is exact.
Proof. Let
0 (T

) (T) (T

) 0
be a short exact sequence in Sh(X)
A
. For each x X, and each A,
the sequence
0 T

,x
T,x T

,x
0
is exact, and the functor lim

: Mod
A
O
X,x
ModO
X,x
is exact, so we have
short exact sequences
0 lim

,x
lim

T,x lim

,x
0
for each x X. Hence we will be done if we can show that lim

commutes
with stalks, i.e.,
lim

(T)x = lim

T,x
But stalk is a special case of pullback, so it suces to prove more generally
that for f : Y X, the functor f
1
: Sh(X) Sh(Y ) commutes with
direct limits. For this we use Yoneda: let (T) Sh(X) and ( Sh(Y ).
Then
HomY (f
1
lim

T, () = HomX(lim

T, f()
= lim

HomX(T, f()
= lim

HomY (f
1
T, ()
= HomY (lim

T, (),
so f
1
lim

T = lim

T.
25
The next lemma says that if we have a sheaf T on a closed subset
Z X, we can compute the cohomology of T directly on Z, or push it
forward to X and compute there, and we will get the same result.
Lemma 23.1. Let i : Z X be a closed subset; then for any T Sh(Z),
H
i
(Z, T) H
i
(X, iT)
25
The argument is a common one: colimits commute with left adjoints. Also note that
this isomorphism holds for O
X
-modules, although our proof only works in Sh(X), because
f
1
is not well-dened for O
X
-modules.
79
Proof. First note that since i is a closed immersion, i is exact.
26
Also,
i has an exact left-adjoint, so by Lemma 20.1, it takes injectives to in-
jectives. Thus taking an injective resolution T 1

of T in Sh(Z), we
get an injective resolution iT i1

of iT in Sh(X). But for any sheaf

( on Z, (X, i() = (Z, (), so applying the global sections functor to
these two injective resolutions gives the same sequence of abelian groups,
hence the result.
Remarks. 1. To see that i is exact when i is a closed immersion, note
that the stalk of iT is
(iT)x =
_
Tx if x Z
0 if not
.
Since exactness can be checked at stalks, it is easy to see in this case
that i is exact. However, this fails when i is not a closed immersion:
consider the open immersion
i : Y = A
2
k
\ |(0, 0) A
2
k
= X
Then the stalk (iOY )
(0,0)
is just O
X,(0,0)
. Intuitively, an open
neighborhood U of the origin, and the punctured neighborhood U \
|(0, 0) have the same sections, because the point has codimension
two.
2. In applying the lemma, we will frequently use the following fact. If
T is a sheaf on a space X, whose support is contained in a closed
subset i : Z X, then T = ii
1
T.
At this point we began the proof of Theorem 22.1, but Ill consolidate
the whole argument into the notes for the next lecture.
24 Wednesday, March 14th: Proof of the Vanish-
ing Theorem
We recall rst the statement: If X is a Noetherian topological space of
dimension n, and T a sheaf of abelian groups on X, then H
i
(X, T) = 0
for i > 0.
Proof. 24.1 Base step
We proceed by induction on n = dimX. For the case n = 0, the theorem
says exactly that the global sections functor is exact. But here X is a
disjoint union of points, nite by the Noetherian hypothesis. A sheaf on
a point is just an abelian group, so a sheaf on X is just a nite product
of abelian groups. Thus (X, ) is exact - it sends a nite product of
abelian groups to ... that same nite product of abelian groups.
Now we x n > 0, and assume the theorem holds for any space of
dimension less than n. Fix a sheaf T of abelian groups on X.
26
Its a right adjoint, so its always left exact (even if i is not a closed immersion). For
right-exactness in the case of a closed immersion, see the remark following the proof.
80
24.2 Reduction to X irreducible
Here we argue that it is enough, under our inductive hypothesis, to assume
that X is irreducible. Suppose it were not. Then pick an irreducible closed
subset X

of X, let U be the complement in X

of the other irreducible

components of X, and let Y = X \ U. Then we have inclusions
U
j
X
i
Y
with j open and i closed. Let TU = j
!
(T

U
) and TY = ii
1
T. We
obtain
27
an exact sequence of sheaves on X:
0 TU T TY 0,
which gives rise to a long exact sequence of cohomology
H
0
(X, TU) H
0
(X, T) H
0
(X, TY ) H
1
(X, TU) H
1
(X, T)
By Lemma 23.1, we have H
i
(X, TY ) = H
i
(Y, i
1
T). Also, since TU
is supported on X

, we may regard it as a sheaf on X

, and we have
H
i
(X, TU) = H
i
(X

, TU), again by Lemma 23.1

28
. So the long exact
sequence looks like
H
i1
(Y, i
1
T) H
i
(X

, TU) H
i
(X, T) H
i
(Y, i
1
T)
Since X

is irreducible, and Y

is a closed subset of X with less irreducible

components than X, this sequence will prove that H
i
(X, T) = 0 for i > n
by induction on the number of irreducible components, so long as we can
establish this in the case where X is irreducible (i.e., the base step for the
induction on the irreducible components).
24.3 Reduction to the case where T = ZU.
Now we assume X is irreducible. For j : U X any open subset, we let
as above Z
U
= j
!
Z, where Z is the constant sheaf Z on U. We will simply
write ZU from now on. Dene
S =

UX
T(U),
and let A be the set of nite subsets of S, partially ordered by inclusion.
A typical element of A looks like
= |(U1, f1), . . . , (Ur, fr), f
k
T(U
k
)
Such a thing denes a map
r

i=1
ZU
i
(f
i
)
T
27
Cf. Hartshorne, Ex II.1.19(c).
28
Abusing notation slightly: should be H
i
(X, F
U
) = H
i
(X

, ss
1
F
U
), where s: X

X
is the inclusion. See remark 2 following Lemma 23.1.
81
For each as above, we let T be the image of this map. Then the
inclusions T T induce a map lim

T T which is an isomorphism.
Its injective since the T are subsheaves of T. Its surjective on each
open, since if f T(U), then take = (U, f); the image of 1 under
ZU(U) T(U) (lim

T)(U) is a preimage for f over U.

Since H
i
(X, lim

T) lim

H
i
(X, T), it is now enough to prove that
when i > n, H
i
(X, T) = 0. With = |(U1, f1), . . . , (Ur, fr) as above,
and setting

= |(U1, f1), . . . , (Ur1, fr1), we have a diagram

0

T

T

Q

0
0

r1
i=1
ZU
i

r
i=1
ZU
i

ZUr

0
where the horizontal rows are exact by construction (here Q is just the
cokernel of T

T), and the left two vertical arrows are surjective by

denition. It follows that the rightmost vertical arrow is also surjective.
Now from the long exact sequence in cohomology induced by the top
sequence here, and using induction on the cardinality of , it suces to
show that H
i
(X, Q) = 0 for i > n, where Q is any quotient of a sheaf of
the form ZU.
For such a sheaf Q, we have an exact sequence
0 K ZU Q 0
and again using the long exact sequence, we see it will be enough to show
that H
i
(X, K ) = 0 for i > n, where K is any subsheaf of ZU (including
ZU itself). So x some subsheaf K of ZU; then the stalk Kx at x X
is a principal ideal (dx) Z, where we may assume that dx 0. If all
integers dx are zero, then K is the zero sheaf, whose higher cohomology
vanishes, so theres nothing to show. Otherwise, set d = inf|dx, where
the inmum is taken over all x such that dx ,= 0. Then we will have dx = d
for at least one point of X. Take some neighborhood V U containing
this x, and a section s K (V ) representing d on this neighborhood. The
section s denes an injective map ZV K sending 1 to d. Composing
with the inclusion K ZU gives a map
ZV K ZU
of sheaves on U, which is multiplication by d at stalks of V , and the zero
map elsewhere. Now we claim that in fact KV is isomorphic to ZV . We
check this at the stalk at a point y of V . There the inclusions above look
like
ZV,y Ky ZU,y,
where ZV,y ZU,y Z and the composite map is multiplication by d. But
this multiplication must factor through the image of Ky ZU,y, which
is just the subgroup generated by dy. So (d) (dy); but d is minimal
amongst the dx as x varies, therefore dy = d and Ky (d) ZV,y. In
particular, Z

V
K

V
.
82
This isomorphism extends to an inclusion ZV KV K . Let
the cokernel of ZV K be D. Then since ZV K over V , D is
supported on U \V , and hence on U \ V , which has lower dimension than
X. For if Z0 Zr is a chain of irreducible closed subsets in U \ V ,
then the inclusion Zr X extends this chain, since X has been assumed
irreducible. Thus the long exact sequence in cohomology coming from
0 ZV K D 0
shows that H
i
(X, ZV ) H
i
(X, K ) for i > n, so it will be enough to
prove H
i
(X, ZV ) = 0 for i > n (note that the long exact sequence even
shows H
n+1
(X, ZV ) H
n+1
(X, K ), since the cohomology of D vanishes
in degrees greater than or equal to n).
24.4 Conclusion of the Proof
Now we have nished our reduction. To show that H
i
(X, ZV ) = 0 in
degrees greater than n, consider the inclusion ZV Z (the constant sheaf
on X), and let Q be the cokernel. Since X is irreducible, Z is asque,
so its cohomology vanishes in positive degree. The long exact sequence
coming from
0 ZV Z Q 0
thus shows that H
i
(X, Q) H
i+1
(X, ZV ) for all i > 0. But Q is sup-
ported on X \ V , which as above has lower dimension than X since
X is irreducible
29
. So H
i
(X, Q) = 0 for i n by induction
30
. Thus
H
i
(X, ZV ) = 0 for all i n + 1, nishing the proof.
25 Friday, March 16th: Spectral Sequences - In-
troduction
In the following few lectures, we will state without proof the results about
spectral sequences which we will use, along with some common applica-
tions. Our aim is to obtain the following two results
1. If X is ane and T a quasicoherent sheaf on X, then the higher co-
homology of T vanishes (we saw the vanishing of H
1
last semester).
2. If f : X Y is an ane morphism and T QCoh(X), then the
higher direct image sheaves R
i
fT are zero for i > 0. This general-
izes the previous result by taking Y =point.
25.1 Motivating Examples
Examples 25.1. 1. Let X be a topological space and U = |Ui an
open cover. If T Sh(X), we ask: how can we compute H
s
(X, T)
29
The V from above was non-empty by construction. If we were proving this for general V ,
the case when V is empty implies that Z
V
is the zero sheaf, so theres nothing to show.
30
Were using Lemma 23.1 and Remark 2 following it again.
83
in terms of the H
s
(Ui, T

U
i
)? Consider rst the case when s = 0.
Then by the sheaf property, H
0
(X, T) is the kernel of the map

i
H
0
(Ui, T)

i,j
H
0
(Uij, T).
31
For the case s = 1, let
K = ker
_

i
H
1
(Ui, T)

i,j
H
1
(Uij, T)
_
.
We also dene groups (the

Cech cohomology groups)

H
1
(U, T) =
ker
_
i,j
H
0
(Uij, T)

i,j,k
H
0
(U
ijk
, T)
_
im
_
i
H
0
(Ui, T)

i,j
H
0
(Uij, T)
_

H
2
(U, T) =
ker
_
i,j,k
H
0
(U
ijk
, T)

i,j,k,l
H
1
(U
ijkl
, T)
_
im
_
ij
H
0
(Uij, T)

i,j,k
H
0
(U
ijk
, T)
_
At this point you should ask what actually are all those maps ?
Well come to that later. We will see that these groups t into an
exact sequence
0

H
1
(U, T) H
1
(X, T)

H
2
(U, T)
In the case mentioned above, namely when X is ane and T quasi-
coherent, the

i
H
1
(Ui, T) = 0, so K = 0, and we get

H
1
(U, T)
H
1
(X, T).
2. Consider continuous maps X
f
Y
g
Z, and a sheaf T on X. We
would like to relate the composite right derived functors R
q
g(R
p
fT)
with the right derived functors of the composite R
n
(fg)T. This is
accomplished by the Leray spectral sequence, which we will see soon.
25.2 Denition
Denition 25.2. Let A be an abelian category, and a N. A spectral
sequence starting at page a consists of the following data:
1. Objects E
pq
r
of A, for p, q Z and r a some integer. The collection
E
pq
r
for xed r is called the rth page.
2. Morphisms d
pq
r
: E
pq
r
E
p+r,qr+1
r
such that d
2
= 0 (whenever d
2
makes sense). The maps go r to the right and r 1 down.
3. Isomorphisms E
pq
r+1
H

(E
pq
r
).
Usually we will consider situations where p, q 0, in which case for
each p, q there exists a page r beyond which the dierentials d are all zero
(this r will depend on p, q), so E
pq
r
= E
pq
r+1
= . . .. We denote this common
value by E
pq

. We say then that the spectral sequence converges.

31
As usual, U
ij
= U
i
U
j
, and in the following we will denote triple intersections by
U
ijk
= U
i
U
j
U
k
, etc. Also, we will carelessly write F even where we should actually be
writing F

U
i
, etc.
84
Denition 25.3. Let H

= |H
n

nZ
be a collection of objects of A. We
say that E
pq
r
converges to H

if a) it converges in the sense described

above, and b) each H
n
has a ltration
32
F

such that F
p
/F
p+1
E
n,np

.
We use the notation E
pq
r
H

in this case.
In other words, E
pq
r
H

means that the nth antidiagonal of E

gives the graded pieces of a ltration on H
n
.
Remark. Note that this information is not always enough to determine
H
n
. It is, for example when A is the category of k-vector spaces. On the
other hand, the two complexes of abelian groups
0 Z/2
2
Z/4 Z/2 0 and
0 Z/2 Z/2 Z/2 Z/2 0
Both have ltrations with quotients Z/2, but are not isomorphic com-
plexes.
25.3 Spectral Sequence of a Filtered Complex
Let
K

. . . 0 K
r
K
r+1
. . .
be a complex in A, with a ltration by subcomplexes F
i
, so that
0 = F
s
F
s1
. . . F
1
F
0
= K

.
We say that the ltration is nite because the F
i
are eventually zero,
and exhaustive because F
0
= K

. We will use frequently the following

result:
Theorem 25.1. In the situation above, there is a spectral sequence
E
pq
1
= H
p+q
(gr
p
K

) H
p+q
(K).
A word on notation. When we write E
pq
1
= H
p+q
(gr
p
K

) H
p+q
(K),
we mean that the spectral sequence begins at page 1, with the rst page
being H
p+q
(gr
p
K

). Here gr
p
K

means the pth graded piece of the given

ltration on K

, namely F
p
/F
p+1
(note that this, too, is a complex, so
we can take cohomology). The notation for the spectral sequence gives
no indication of what the dierentials are, not even on the rst page. To
understand what they are, you must study the proof of the theorem. See
Weibels book, section 5.4, for instance.
One thing we can say is that the map H
p+q
(gr
p
K

) H
p+1+q
(gr
p+1
K

)
is the boundary map coming from application of the Snake Lemma to the
short exact sequence
0 F
p+1
/F
p+2
F
p
/F
p+2
F
p
/F
p+1
0.
32
For us a ltration will always mean a decreasing ltration, i.e., a sequence of inclusions
. . . F
i+1
F
i
. . . H
n
.
85
Example 25.1. The ltration bete on a complex
K

= K
0
K
1
K
2
. . .
is as follows:
F
2
: 0

K
2

. . .
F
1
: 0

K
1

K
2

. . .
K

: K
0
K
1
K
2 . . .
In this case gr
p
K

= F
p
/F
p+1
is the complex with the object K
p
in
degree p and zeroes elsewhere, which we denote K
p
[p]
33
Therefore
H
p+q
(gr
p
K

) =
_
K
p
if q = 0
0 if q ,= 0
Thus the rst page of our spectral sequence is
0 0 0 0
. . .
0 0 0 0
. . .
K
0
K
1
K
2
K
3 . . .
which converges by the second page since the dierentials are zero there.
But on the second page the objects are E
p0
2
= H
p
(K

). So the Theorem
tells us that the objects H
p+q
(K

) have a ltration whose pth graded

piece is 0 unless q = 0, in which case it is H
p
(K

). In other words, H
n
is
ltered by H
n
. Bete, indeed.
26 Monday, March 19th: Spectral Sequence of a
Double Complex
26.1 Double Complex and Total Complex
Today we discuss another important example of a spectral sequence. Sup-
pose we have a double complex I
,
; that is, a collection of objects I
pq
33
In general, whenever we write an object as standing for a complex, we mean the complex
containing that object in degree zero with zeroes elsewhere. Here we have shifted the degree
to make the object sit in degree p.
86
and two dierentials d and , such that the following diagram commutes
I
p,q+1

I
p+1,q+1,
I
p,q
d

I
p+1,q
d

Then we can form a new complex, the total complex tot(I) of I

,
,
dened as follows:
(tot I)
n
=

i+j=n
I
i,j
The dierential of this complex is
D

I
ij
= d + (1)
j

There are other sign conventions in the literature - other authors take
D = d + but insist that the squares in the original double complex
anticommute; the important point is that in order to get a complex, the
square in the following diagram must anticommute
I
i,j+2
I
i,j+1

I
i+1,j+1
I
i,j

I
i+1,j

I
i+2,j
D
2
is the sum of all the paths from the lower left to the three objects on
the upper right antidiagonal - these should sum to zero.
There is a ltration on tot(I
,
) whose pth term is
(F
p
)
n
=

i+j=n
ip
I
ij
In other words, the pth term F
p
of the ltration is that part of the nth
antidiagonal of I
,
which lies to the right of the p 1 column. From
this description, we see that the quotient F
p
/F
p+1
is a complex whose
objects are precisely the pth colum of I
,
. But I
pq
sits in degree n = p+q
according to the above denition, so the complex is graded by q, but with
a shift down in degree by p. I.e.,
F
p
/F
p+1
= I
p,
[p]
34
Theorem 25.1 in this situation can be stated as:
Theorem 26.1. In the situation above, there is a spectral sequence
E
pq
1
= H
p+q
(I
p,
[p]) H
p+q
(tot I
,
)
34
Recall our convention about grading shifts: (C[r])
n
= C
r+n
.
87
Note that H
p+q
(I
p,
[p]) = H
q
(I
p,
).
Corollary 26.1. Let f : I
,
J
,
be a morphism of double complexes
such that for each p, I
p,
J
p,
is a quasi-isomorphism. Then tot I
,

tot J
,
is a quasi-isomorphism.
Proof. The morphism f induces a morphism of spectral sequences, as
follows. By assumption, the maps H
q
(I
p,
) H
q
(J
p,
) induced by f
are isomorphisms between the rst pages of the two spectral sequences.
Thus when we take cohomology we get isomorphisms between the 2nd
pages, and so on. So isomorphisms between the rst pages give rise to
isomorphisms
E
pq

(I) E
pq

(J)
Now it is a general fact that a map of ltered complexes which induces
isomorphisms on each graded piece must be an isomorphism of complexes.
So the above implies that we get an isomorphism
H
n
(tot I
,
)

H
n
(tot J
,
).
26.2 Application: Derived Functors on the Category of Com-
plexes
The above will be useful for us in the following situation. Let F : A B
be a left exact functor between abelian categories, where A has enough
injectives. Let K

be a complex in A with K
p
= 0 for p << 0. We want
to compute R
i
FK

B.
First resolve the complex K

by injectives:
. . .
K
p1

K
p

K
p+1

. . .
. . .
I
p1,
I
p,
I
p+1, . . .
Here the rows and columns are complexes, and the columns are also res-
olutions (i.e. exact). In particular, its a double complex I
,
. We can
thus dene
R
i
FK

= H
i
(tot FI
,
)
Then of course we have to check that this does not depend on the reso-
lution I
,
. Suppose given two injective resolutions I
,
and J
,
of the
complex K

. Then one argues as in the HW a few weeks ago that it is

enough to consider the case when we have a morphism I
,
J
,
which
commutes with the inclusions of K

into each double complex. Fixing p,

we have injective resolutions K
p
I
p,
and K
p
J
p,
. We know that
the derived functors of F on objects do not depend on the resolution, i.e.,
the maps I
,
J
,
induce isomorphisms H
n
(FI
p,
) H
n
(FJ
p,
), and
hence the conditions of the corollary are met, showing that the derived
functors R
n
F are well-dened on the complex K

.
88
27 Wednesday, March 21st: Spectral Sequence of
an Open Cover
Let (X, OX) be a ringed space and T an OX-module. Fix an open cover
U = |UiiI of X, and assume I is totally ordered. We use the notation
i = (i0, . . . , i
k
), i0 < i1 < . . . < i
k
for a multi-index in I, with [i[ = k denoting the length (length minus
one, really, since our multi-indices start with i0). For such a multi-index,
dene U
i
= Ui
0
. . . Ui
k
, and let
j
i
: U
i
X
be the inclusion.
We dene a complex C

(U, T) of sheaves of abelian groups on X, the

Cech complex of T relative to the open cover U, as follows. For

each k 0, set
C
k
(U, T) =

|i|=k
j
i
T

U
i
.
The rst few terms are thus
C
0
(U, T) =

iI
jiT

U
i
C
1
(U, T) =

i
1
<i
2
j
(i
1
,i
2
)
T

U
i
1
U
i
2
The dierential on this complex is obtained by rst dening, for each
0 j k + 1, a map
dj : C
k
(U, T) C
k+1
(U, T)
which sends an element
(a
i
)
i=i
0
i
1
i
k
(a
s
0
s
1
s
j
s
k
)s
0
s
1
s
k
That is, the component of dj((a
i
)) on the s0s1 s
k
factor is obtained by
deleting the jth index sj and taking the appropriate entry of (a
i
).
Then we dene the dierential on C
k
(U, T) by
d =
k+1

j=0
dj
One checks that this denes a complex of sheaves.
Example 27.1. Take (X, OX) to be P
1
k
with its structure sheaf, and
U = |U0, U1 the standard open cover of P
1
by anes. Then
C
0
(U, O
P
1) = j0OU
0
j1OU
1
C
1
(U, O
P
1) = j01OU
01
C
k
(U, O
P
1) = 0 for k > 1
Then if (a0, a1) is a local section of C
1
, the two maps dj for j = 0, 1 send
it to a0 and a1, respectively, so the dierential d: C
0
C
1
maps (a0, a1)
to a1 a0.
89
The reason we care about this

Cech complex is the following:
Proposition 27.1. There is a quasi-isomorphism of complexes T
C

(U, T).
Remark. Recall that when we speak of a sheaf T as a complex, we mean
the complex which has T in degree zero and zeroes elsewhere, and with
dierential zero. Then to say that this complex is quasi-isomorphic to
some other complex K

(with K
p
= 0 for p 0) means that K

must
be exact except in degree zero, since the cohomology of the complex
T is zero in higher degrees. Moreover, the kernel of K
0
K
1
must be
isomorphic to T because of the isomorphism of cohomology in degree zero.
In other words, T quasi-isomorphic to K

means K

is a resolution of
T.
Proof. First note that we have a map
T

iI
jiT

U
i
Given by the product of the restriction maps. This map is actually an
isomorphism onto the kernel of d
0
because of the sheaf axiom for T. It
remains to show that the higher cohomology sheaves H
q
(C

(U, T)) are

zero for q > 0. The question is local since were working with sheaves,
so we can assume X is one of the Uis. In fact, after some fussing around
with indices and signs, we may assume that X = U0 for simplicity.
Now we dene a homotopy operator
h
k
: C
k+1
(U, T) C
k
(U, T)
which sends a local section (ai
0
i
k+1
) to (a0i
0
i
k
)i
0
i
k
. One checks that
these maps satisfy
dh
k
+h
k+1
d = Id
C
,
that is, they dene a homotopy between the identity map on C

(U, T)
and the zero map. This shows that the identity map and the zero map
induce the same map on cohomology, so it must be that (C

(U, T), d) is
exact (except, of course, in degree zero).
Next we want to apply our results on spectral sequences to

Cech com-
plexes. Take an injective resolution T I

of T. Then the

Cech sheaves
C(U, I) are all injective sheaves, and we obtain a double complex
.
.
.
.
.
.
C
0
(U, I
1
)

C
1
(U, I
1
)

C
0
(U, I
0
)

C
1
(U, I
0
)

90
The rows are (injective) resolutions of the sheaves I
k
, by the proposition.
We can regard the complex I

as a double complex I
,
with only one
column. Since the maps I
k
C

(U, I
k
) are resolutions (i.e., quasi-
isomorphisms), Corollary 26.1 from the previous lecture applies, and says
that the map of total complexes tot I
,
tot C

(U, I

) is a quasi-
isomorphism. But the total complex of I
,
is just I

. Thus we have a
chain of quasi-isomorphisms
T I

tot(C
p
(U, I
q
))
In other words, the total complex of the double complex above is a reso-
lution of T, also.
Now we take global sections all over the place. Let
C
q
(U, T) = (X, C
q
(U, T)) =

|i|=q
T(U
i
).
Then the quasi-isomorphisms I
k
C

(U, I
k
) give rise to quasi-isomorphisms
(X, I
k
) C

(U, I
k
)
of complexes of abelian groups
35
. Then we have a double complex
.
.
.
.
.
.
C
0
(U, I
1
)

C
1
(U, I
1
)

C
0
(U, I
0
)

C
1
(U, I
0
)

where here the rows are resolutions of (X, I
k
). By the same argument
as above, we nd that the total complex tot C

(U, I

) is a resolution of
(X, I

), so we can use this total complex to compute the cohomology

groups of T on X:
H

(tot C

(U, I

)) (H

(X, T)).
On the other hand, our discussion of the spectral sequence of a double
complex showed that there is a spectral sequence
E
pq
1
= H
q
(C
p
(U, I

)) H
p+q
(tot C

(U, I

))
But
H
q
(C
p
(U, I

)) = H
q
_

|i|=p
I

(U
i
)
_
=
36

|i|=p
H
q
(U
i
, I

) =

|i|=p
H
q
(U
i
, T)
35
This is only true because we began with resolutions of injective objects I
k
. In general,
if K

and K

are quasi-isomorphic complexes of injectives, then application of a left exact

functor F gives quasi-isomorphic complexes FK

and FK

, but the fact that both K

and
K

are injective is essential.

36
Not sure under what circumstances we can interchange cohomology and products like
this.
91
In conclusion, we have constructed a spectral sequence
E
pq
1
=

|i|=p
H
q
(U
i
, T) H
p+q
(X, T)
This is the spectral sequence of the open cover U.
28 Friday, March 23rd:

Cech Cohomology Agrees
with Derived Functor Cohomology in Good Cases
Last time, given an open cover U = |UiiI of a topological space X, and
a sheaf T of abelian groups on X, we constructed a spectral sequence
E
pq
1
=

|i|=p
H
q
(U
i
, T) H
p+q
(X, T)
Recall that the global

Cech complex consisted of groups
C
p
(U, T) =

|i|=p
T(U
i
),
with dierentials dened in the same way as for the sheafy version C

last
time. We call the cohomology groups
H
p
(C

(U, T))
the

Cech cohomology of T relative to the open cover U, and denote
them by

H
p
(U, T). Today we will use the above spectral sequence to prove
that in good circmstances, the

Cech cohomology groups agree with the
cohomology groups dened by derived functors of (X, ).
Let us write out the rst few pages of the global spectral sequence
from last time. The E1 page looks like
.
.
.
.
.
.

i
H
1
(Ui, T)

i<j
H
1
(Uij, T)


C
0
(U, T)

C
1
(U, T)


0 0
The second page looks as follows, letting K = ker .
.
.
.
.
.
.
K

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V

H
0
(U, T)

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

H
1
(U, T)

W
W
W
W
W
W
W
W
W
W
W
W
W
W

H
2
(U, T)
0 0 0
92
The map in to

H
0
(U, T) is also zero, so this shows that

H
0
(U, T) is the
only term in our ltration of H
0
(X, T), so

H
0
(U, T) H
0
(X, T)
To study the ltrations on the higher cohomology groups H
i
(X, T),
we look now at the third page:
.
.
.
.
.
.
ker

R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R

H
0
(U, T)

H
1
(U, T)
0 0 0 0
Note that now the rst antidiagonal p +q = 1 stabilizes, as the maps
into and out of both ker and

H
1
(U, T) are zero hereafter. So we know
that there is a ltration on H
1
(X, T):
0 = F
2
F
1
F
0
= H
1
(X, T)
with
F
0
/F
1
E
0,1

= E
0,1
3
ker
F
1
/F
2
E
1,0

= E
1,0
3


H
1
(U, T)
Since F
2
= 0, we have
ker F
0
/F
1

_
F
0
/F
2
_
/
_
F
1
/F
2
_
H
1
(X, T)/

H
1
(U, T)
So we get an exact sequence
0

H
1
(U, T) H
1
(X, T) ker 0.
It turns out that the map on the right, H
1
(X, T) ker , is compatible
with the inclusions
ker K

i
H
1
(Ui, T)
Thus we obtain a commutative diagram
0

H
1
(U, T)

H
1
(X, T)

G
G
G
G
G
G
G
G
G
G
ker

0
K

i
H
1
(Ui, T)
This is good because it relates H
1
(X, T) to

H
1
(U, T), but at the expense
of the smaller groups

i
H
1
(Ui, T). Of course, what we want is a way to
93
get a handle on these H
1
(Ui, T). More generally, we would like to handle
the groups H
i
(U
i
, T), where i > 0 and the multi-index i has arbitrary
length. A rst step in this direction is to take the cover U to consist of
open anes. The next theorem shows why this is useful. It is also the rst
of the two desired results mentioned at the beginning of our discussion of
spectral sequences.
Theorem 28.1. Let X be an ane scheme and T QCoh(X). Then
for all q > 0, H
q
(X, T) = 0.
Proof. We proceed by induction on q. The case when q = 1 was estab-
lished last semester: the global sections functor is exact on anes. So
now assume that for any ane scheme X and any quasicoherent sheaf T
on X, H
i
(X, T) = 0 for 1 i q. We x an ane scheme X, with a
cover
37
by open anes X = U0 . . . Ur. We can arrange it so that all
possible intersections of the Ui are also ane (by taking them to be basic
opens, say). We consider the spectral sequence from before, where by our
inductive assumption, all H
i
(U
i
, T) are zero, where i is a multi-index and
0 < i q. The rst page is thus

i
H
q+1
(Ui, T)

i<j
H
q+1
(Uij, T)


0 0 0
0 0 0
C
0
(X, T)

C
1
(X, T)


We include only two rows of zeroes for simplicity, but of course there will
be more for larger q. These rows are zero by our inductive assumption.
We now proceed to page two. The rst row will consist of the groups
H
p
(C

(U, T)), which are the right derived functors of (X, ), applied to
the

Cech sheaves C

(U, T). These sheaves are quasicoherent, so since

X is ane, (X, ) is exact and the higher right derived functors vanish.
So the second page is
K

0 0 0
0 0 0
H
0
(X, T) 0 0
Here K is just some subgroup of

i
H
q+1
(Ui, T). Since this spectral
sequence converges to H
p+q
(X, T), we have that H
q+1
(X, T) K

i
H
q+1
(Ui, T). It turns out that, by the construction of the spectral
sequence (which we did not go into in any detail), this map
H
q+1
(X, T)

i
H
q+1
(Ui, T)
37
Recall that an ane scheme is quasicompact, so we may take a nite cover here, although
we will not use the niteness.
94
is induced by the restriction maps along Ui X. We want to use this to
show that H
q+1
(X, T) = 0, for which we need the following lemma.
Lemma 28.1. IF X is any scheme, and H
i
(X, T) for i > 0. Then
there is an open cover X =

j
Uj such that the image of in each
H
i
(X, Uj) is zero.
Proof. First choose an injective resolution T I

. Our element
H
i
(X, T) can be represented by an element ker((X, I
i
) (X, I
i+1
)).
Since the copmlex of sheaves I

is exact, we can nd a neighborhood U

of any point in X such that

U
is in the image of I
i1
(U) I
i
(U).
Now choose an open cover

j
Uj of X by such U. Then we have [ ] = 0
in H
i
(Uj, T), for each j, as desired.
Using the Lemma, we may replace, i.e., rene, our original cover to get
some new Uj such that H
q+1
(X, T) restricts to 0 in each H
q+1
(Uj, T).
But the map H
q+1
(X, T)

i
H
q+1
(Ui, T) is injective, so is zero,
hence H
q+1
(X, T) = 0.
Corollary 28.1. If f : X Y is an ane morphism, and T a quasico-
herent sheaf on X, then R
i
fT = 0 for all i > 0.
The proof follows immediately from the theorem and the following
lemma.
Lemma 28.2. If f : X Y is a morphism of schemes, and T a sheaf
on X, then the higher pushforward R
i
fT is the sheaf associated to the
presheaf on Y
T
i
F
: U H
i
(f
1
(U), T).
Proof. My notes for Professor Olssons argument are unclear here. They
include only the cryptic phrase sheacation commutes with quotients.
See Hartshorne III.8.1 for a dierent argument.
Now we make good on the title of todays lecture
Theorem 28.2. If X is a scheme and U = |Ui
r
i=1
a cover by open
anes with all possible intersections ane
38
, then for any quasicoherent
sheaf T on X, we have
H
n
(X, T)

H
n
(U, T)
Proof. Here the spectral sequence of the open cover, namely
E
pq
1
=

|i|=p
H
q
(U
i
, T) H
p+q
(X, T),
is especially simple. On the rst page, all the rows except the rst are
zero, by Theorem 28.1. On the second page, the sequence has already
converged, and the nonzero stable groups are the

H
n
(U, T). By the the-
orem on this spectral sequence, these groups form a one term ltration of
H
n
(X, T), which proves it.
38
This happens, for instance, when X is separated.
95
Example 28.1. Well use the above to compute H

(P
1
k
, O
P
1
k
)
39
. We have
a cover of P
1
by two anes Spec k[x] and Spec k[y], whose intersection is
Spec k[x

], where y corresponds to 1/x on this intersection. The

Cech
complex is
C
0
(U, O) = k[x] k[y] k[x

] = C
1
(U, O)
where this map is given by (p(x), q(y)) p(x)q(1/x). The kernel of this
map consists of the constants k |(c, c) k[x] k[y]. It is surjective,
so the cokernel is zero. Therefore
H
n
(P
1
k
, O)
_
k i = 0
0 else
29 Monday, April 2nd: Serres Aneness Crite-
rion
Today we prove a useful result of Serre that uses cohomology of ideal
sheaves to test whether a Noetherian scheme is ane. The idea of the
proof introduces the point of view that cohomology groups should be
thought of as containing obstructions to various things.
We begin with a few more examples to review the concepts from last
time. Recall that the hypotheses of Theorem 28.2 are automatically sat-
ised if X is a scheme over R which is separated. The reason is that for
two open anes U, V X, we have a diagram
U V

U R V

X R X
and when X/R is separated, is a closed immersion, hence so is the top
arrow. But a closed immersion is an ane morphism and URV is ane,
hence so is U V .
Example 29.1. Let X be the ane line with doubled origin, A
1
k

Gm
A
1
k
.
This scheme is not separated, but nonetheless we have a cover by two
anes, with ane intersection, so the theorem applies.
Example 29.2. If X is ane, then we can take the open cover to consist
of just X itself, so Theorem 28.2 applies and

H
i
= H
i
. Moreover, the

Cech groups are zero in positive degree since our open cover has only one
element. This isnt really an application of the theorem, however, since
we used this fact in the proof; but we see another point of view on why
ane schemes have no higher cohomology.
Now to the main course:
Theorem 29.1 (Serre). If X is a Noetherian scheme, the following are
equivalent:
1. X is ane.
39
Hereafter well drop the subscript P
1
k
from the notation of the structure sheaf.
96
2. H
i
(X, T) = 0 for all quasicoherent sheaves T on X and all i > 0.
3. H
1
(X, I) = 0 for all quasicoherent ideal sheaves I on X.
40
Proof. All that needs to be shown is that 3 implies 1. For this we use the
following two facts
1. (Hartshorne II.2.17) X is ane if and only if there are global sections
f1, . . . , fr of OX such that for each i, X
f
i
is ane and the ideal
(f1, . . . , fr) = (X, OX).
2. If X is a Noetherian scheme, and U an open subset which contains
all the closed points of X, then U = X.
Now we produce the sections fi one point at a time. Specically,
pick a closed point P X, and an ane neighborhood U of P. Let
Y = X \ U, and give it a scheme structure (the reduced one, say). We
have the skyscraper sheaf k(P) at P, and it ts into a short exact sequence
0 IY P IY k(P) 0.
This comes from the fact that Y is a closed subscheme of Y P, so we
have an inclusion IY P IY . This is an isomorphism away from P; the
cokernel is then just the sheaf k(P). Taking cohomology gives the long
exact sequence
0 (X, IY P ) (X, IY ) k(P) 0
where the rightmost term is zero because of our assumption (3) that H
1
vanishes for quasicoherent ideal sheaves. Thus the map (X, IY ) k(P)
is surjective, so we can nd a global section f of IY which does not vanish
at P (namely a preimage of any nonzero element of k(P)). This means
that P X
f
, and moreover, since f vanishes on Y , we have that X
f
U.
Therefore the points where f vanishes are exactly the points where the
image of f in OX/IY OU vanishes, which is just U
f
. So X
f
is ane.
So we can cover all the closed points of X by these ane sets X
f
.
Therefore their union is all of X, by fact 2. Again using the Noetherian
hypothesis, we see that nitely many such X
f
suce to cover X. Thus
we have produced global sections f1, . . . , fr such that X =

X
f
i
.
It remains to check that (f1, . . . , fr) = (X, OX). The global sections
fi dene a map
O
r
X
OX
by sending (a0, . . . , ar) to

aifi. This map is surjective since we can

check at stalks, and at each point of X, not all fi vanish since the X
f
i
cover X. So we obtain an exact sequence
0 K O
r
X
OX 0
for some subsheaf K of O
r
X
. Looking at the induced long exact sequence
in cohomology, we see that if we can show that H
1
(X, K ) = 0, then we
will have a surjection (X, OX)
r
(X, OX), which will nish the proof.
40
(In reply to a question) An example of an ideal sheaf which is not quasicoherent: j
!
I
O
X
, where j : U X is any proper open subset, and I O
U
an ideal sheaf on U.
97
The idea is to lter K in such a way that we can apply our assumption
3 to the quotients of the ltration. The ltation
OX . . . O
r2
X
O
r1
X
O
r
X
,
where O
rj
has local sections of the form (, . . . , , 0, . . . , 0). This gives
rise to a ltration of K :
Kr . . . K2 K1 K
where Ki = O
ri
X
K . The local sections of Ki/Ki+1 have zeroes in all
but one slot, so they are isomorphic to a subsheaf of OX O
ri
X
/O
ri1
X
.
More precisely, we have the following diagram
0

ker

0

Ki+1

Ki

Ki/Ki+1

0
0

O
ri1
X

O
r1
X

OX

0
0

O
ri1
X
/Ki+1

O
r1
X
/Ki

0 0
The map on the bottom is injective since the upper left square is
cartesian. From the snake lemma, we obtain an exact sequence
0 ker O
ri1
X
/Ki+1 O
r1
X
/Ki . . .
and since the map O
ri1
X
/Ki+1 O
r1
X
/Ki is injective, ker = 0. This
shows that Ki/Ki+1 is a subsheaf of OX. We now use this to show that
H
1
(X, Ki) = 0. The long exact sequence induced by 0 Ki+1 Ki
Ki/Ki+1 0 gives, in degree 1,
. . . H
1
(X, Ki+1) H
1
(X, Ki) H
1
(X, Ki/Ki+1) . . .
Now since Ki/Ki+1 is an ideal sheaf, H
1
(X, Ki/Ki+1) = 0 by our assump-
tion of (3). Assume inductively that H
1
(X, Ki+1) = 0; then it follows that
H
1
(X, Ki) = 0. So we get that H
1
(X, K ) = 0 by decreasing induction
on i. As we mentioned above, this suces to nish the proof that X is
ane.
98
30 Wednesday, April 4th: Cohomology of Pro-
jective Space
We begin with a quick warm-up, using the criterion of last time to show
that X = A
2
k
\ |(0, 0) is not ane. By 29.1 it will be enough to nd
a quasicoherent sheaf T on X for which H
1
(X, T) ,= 0. Using the open
cover
U1 = Spec k[x

, y]
U2 = Spec k[x, y

]
U1 U2 = Spec k[x

, y

],
we have
H
1
(X, T) = coker
_
T(U1) T(U2)
(a,b)ab
T(U1 U2)
_
Now we can take T = OX. Then x
1
y
1
OX(U1 U2) is not in the
image of the dierence map, so H
1
(X, OX) is non-zero (its not even nite-
dimensional).
Now we turn to the calculation of the cohomology of the sheaves
O
P
r (n). Let A be a ring, S = A[x0, . . . , xr], and set X = P
r
A
= Proj S.
We use the notation (OX) for the graded ring

n0
(X, OX(n)).
Theorem 30.1. (a) There is an isomorphism S

(OX) induced by
the universal sections s0, . . . , sr (X, OX(1)).
(b) H
i
(X, OX(n)) = 0 for 0 < i < r and i > r.
(c) H
r
(X, OX(r 1)) A
41
.
(d) For each n, the map
H
0
(X, OX(n)) H
r
(X, OX(nr 1)) H
r
(X, OX(r 1)) A
is a perfect pairing
42
Remarks. 1. The theorem allows us to compute all cohomology groups
of OX(n) for any n.
2. The pairing of (d) is dened as follows. If s H
0
(X, OX(n)), it
denes a map of sheaves OX OX(n), or equivalently, a map
OX(n) OX. Tensoring with OX(r 1) gives a map OX(n
r 1) OX(r 1). Taking rth cohomology gives our map
H
r
(OX(n r 1)) H
r
(X, OX(r 1)).
Proof. Let T =

nZ
OX(n)
43
. We will compute H
i
(X, T). The stan-
dard open cover U of P
r
A
satises the hypotheses of Theorem 28.2, so we
can use the

Cech complex to compute this. Moreover, formation of the

Cech complex commutes with direct sums, so we have

H
i
(X, T) =

nZ
H
i
(X, OX(n)),
41
Not canonically.
42
This means it induces an isomorphism H
0
(X, O
X
(n)) H
r
(X, O
X
(n r 1))

.
43
This sheaf is actually the pushforward of the structure sheaf of X = A
r+1
A
\ {O} along
the map X P
r
A
.
99
so we will be able to recover the cohomology groups H
i
(X, OX(n)) for
each n from our calculation. More precisely, the

Cech complex looks like
r

i=1
T(Ui)

i<j
T(Uij) . . .
Note that for each n, the sections of OX(n) over the open set Ui = D+(xi)
are jsut the degree n part of Sx
i
. Summing over all n gives T(Ui) Sx
i
,
T(Uij) Sx
i
x
j
, etc., so we are interested in the following graded complex
of graded S-modules
r

i=1
Sx
i

i<j
Sx
i
x
j
. . .
Now the cohomology groups of this complex are by denition H
i
(C

(X, T))).
But because the dierential in the complex is just localization, it is com-
patible with the grading of the individual groups so the degree n piece of
H
i
(C

(X, T))) is just H

i
(X, OX(n)).
We now use the observations to prove the result, in the following order:
(a),(c),(d),(b).
Weve basically already seen (a) earlier. Heres an example in the case
r = 1. then H
0
is the kernel of the map
A[x

, y] A[x, y

] A[x

, y

],
which is isomorphic to A[x, y]. For (c), since the

Cech complex vanishes
in degree r + 1 and larger, we have
H
r
(X, T) = coker
_
r

k=0
S
x
0
x
k
xr
Sx
0
xr
_
.
But Sx
0
xr
is a free A-module with basis |x
l
0
0
x
lr
r
[ li Z, so this
cokernel is free with basis |x
l
0
0
x
lr
r
[ li < 0. Looking at the degree n
piece, we see that since each li < 0,

li (r + 1), so
H
r
(X, OX(n)) = 0
whenever n > r 1, and for n = r 1, the only monomial in the basis
is x
1
0
x
1
r
, so H
r
(X, OX(r 1)) is a free A-module of rank one. This
proves (c).
Now for (d). First, if n < 0, then OX(n) has no global sections, and
since n r 1 > r 1, H
r
(X, OX(n r 1)) = 0, as was observed
in the previous paragraph. So there is nothing to show. When n = 0,
H
0
(X, OX) A and H
r
(X, OX(r 1)) A by (c). For each global
section s A we get a map
A H
r
(X, OX(r 1)) H
r
(X, OX(r 1)) A
and this association denes an isomorphism of H
0
(X, OX) with H
r
(X, OX(r
1))

since H
r
(X, OX(r 1)) is free of rank one.
Now we consider the case when n > 0. Here H
0
(X, OX(n)) is free with
basis |x
m
0
0
x
mr
r
[ mi 0,

mi = n, while H
r
(X, OX(nr 1)) has
100
basis |x
l
0
0
x
lr
r
[ li < 0,

li = n r 1. The pairing acts on these

bases by
x
m
0
0
x
mr
r
, x
l
0
0
x
lr
r
) = x
m
0
+l
0
0
x
mr+lr
r
But in H
r
(X, OX(r 1)), x
m
0
+l
0
0
x
mr+lr
r
is zero unless each mi + li
is -1, in which case it is 1. So these form dual bases with respect to the
pairing, which is therefore perfect. This proves (d). We will prove (b)
next time.
31 Friday, April 6th: Conclusion of Proof and
Applications
We will prove (b) by induction on r. Note that the statement for i > r
is true, regardless of r, simply because the

Cech complex is zero after the
rth degree. If r = 1, the claim is vacuous.
Note that C

(U, T)x
i
is the

Cech complex of T

U
i
with respect to the
open cover Ui = |Ui Uj
r
j=1
. Therefore since Ui is ane, it is an exact
complex except in degree 0. This means that for each j, an for all i > 0,
H
i
(X, T)x
j
= 0. But to say that this localization is zero is to say that
the S-module H
i
(X, T) is annihilated by xj, so to prove (b), it will be
enough to show that
xj : H
i
(X, T) H
i
(X, T)
is injective for each 0 < i < r, 0 j r. Now xj (X, OX(1)) gives a
map OX(n 1) OX for each n, so it denes a short exact sequence of
graded S-modules
0 S(1)
x
j
S S/(xj) 0.
This gives rise to a short exact sequence of sheaves on P
r
0 T(1) T TH 0,
where H = V+(xj) P
r1
A
P
r
A
. Taking cohomology gives the long exact
sequence
H
0
(X, T)
x
j

H
0
(X, T)

H
0
(H, TH)
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
H
1
(X, T)

H
1
(X, T)

H
1
(H, TH)
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
h
.
.
.
.
.
.
.
.
.
H
r1
(X, T)
x
j

H
r1
(X, T)
101
First, the map H
0
(X, T) S H
0
(H, TH) S/(xj) is surjective, so
the map
H
1
(X, T)
x
j
H
1
(X, T)
is injective. Next, we may assume by our induction hypothesis that
H
i
(H, TH) = 0 for 0 < i < r 1, so the right hand column consists
of zeroes except in the top row. Therefore each map
H
i
(X, T)
x
j
H
i
(X, T)
for i = 2, 3, . . . , r 1 is injective. This nishes the proof.
Example 31.1. Let F k[x0, . . . , xr] be a homogeneous polynomial of
degree d, and consider the hypersurface
X = V+(F)
i
P
r
k
.
We want to calculate H
i
(X, OX). The global section F (P
r
, O
P
r )
denes an exact sequence
0 O
P
r (d)
F
O
P
r iOX 0
Recall that since i is a closed immersion we have H
i
(X, OX) = H
i
(P
r
, iOX).
So the long exact sequence can be written as
H
0
(P
r
, O
P
r (d))

H
0
(P
r
, O
P
r )

H
0
(X, OX)
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
H
1
(P
r
, O
P
r (d))

H
1
(P
r
, O
P
r )

H
1
(X, OX)
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
g
.
.
.
.
.
.
.
.
.
H
r
(P
r
, O
P
r (d))

H
r
(P
r
, O
P
r )

H
r
(X, OX)
We have already computed the rst two columns; the left column
is zero except in degree r, the middle column is zero except in degrees
0 and r. Moreover, in degree zero, H
0
(P
r
, O
P
r ) k. Also, we know
H
i
(X, OX) = 0 in degrees greater than r1 for dimension reasons. Hence
we have
H
0
(X, OX) k
H
i
(X, OX) 0 for r ,= 0, r 1
The only tricky part is to determine H
r1
(X, OX), which is the kernel
of the map : H
r
(P
r
, O
P
r (d)) H
r
(P
r
, O
P
r ). Recall that the global
section F denes a map
F : O
P
r O
P
r (d).
102
Twisting by r 1 and taking global sections we have
H
0
(P
r
, O
P
r (r 1))
F
H
0
(P
r
, O
P
r (d r 1))
Taking duals gives a map
H
0
(P
r
, O
P
r (d r 1))
F
H
0
(P
r
, O
P
r (r 1))

.
Applying Serre duality, we get a map H
r
(P
r
, O
P
r (d)) H
r
(P
r
, O
P
r ).
Unwinding the denitions, it turns out that this map is in fact . So
the kernel of is isomorphic to the kernel of H
0
(P
r
, O
P
r (r 1))
F

H
0
(P
r
, O
P
r (d r 1)).
Example 31.2. Specializing the previous example, let F be the polyno-
mial zy
2
(x
3
axz
2
+ bz
3
); this denes an elliptic curve E P
2
k
. We
have
h
0
(E, OE) = 1
h
1
(E, OE) =
_
2
2
_
= 1
Example 31.3. Recall that the canonical sheaf X is the top exterior
power of
1
X
; its a line bundle on X. In the case of P
r
, weve calculated
that
P
r is the line bundle O
P
r (r 1). For P
1
, we have
P
1 =
1
P
1 =
O
P
2(2); so H
1
(P
1
,
P
1) = H
1
(P
1
, O
P
1(2)).
32 Monday, April 9th: Finite Generation of Co-
homology
Today we discuss the question of whether the cohomology groups of a
coherent sheaf are nitely generated modules over the base ring. The
answer is yes in good circumstances.
Theorem 32.1. If A is a Noetherian ring, X/A projective, L is a very
ample sheaf on X, and T is any coherent sheaf on X, then
(a) For all i > 0, H
i
(X, T) is a nitely generated A-module.
(b) There is an integer n0 such that for all i > 0 and n n0,
H
i
(X, T L
n
) = 0.
Remark. Recall that L is very ample if there is an immersion i : X P
r
A
such that i

O
P
r (1) L. In the following we will write L as OX(1), in
order to use the simpler notation T(n) = T L
n
.
The theorem also admits the following generalization.
Theorem 32.2. 1. If A is a Noetherian ring and X is proper over A,
then for any coherent sheaf T on X and all i > 0, H
i
(X, T) is a
nitely generated A-module.
2. If f : X Y is a proper morphism of locally Noetherian schemes,
then for any coherent sheaf T on X and all i > 0, R
i
fT is a
coherent sheaf on Y .
103
Statement 2 is the most general; this is proved by reducing it to state-
ment 1, and then reducing that to the proof of the previous theorem,
which as we will now see, involves a reduction to the calculation of the
cohomology of projective space.
We will need the following technical lemma
Lemma 32.1. Let T be a coherent sheaf on P
r
and let s (D+(xi)).
Then there is an integer m such that sx
m
i
extends to a global section of
(P
r
, T(m)), i.e. there is a section s (P
r
, T(m)) such that s

D
+
(x
i
)
=
sx
m
i
(D+(xi), T(m)).
Proof. Heres a sketch of the idea. Given coherent T, set ( =

n0
T(n).
By the sheaf property, we have the following two equalizer diagrams, where
the vertical arrows are the localizations maps
((X)

j
((D+(xj))

j<j

((D+(xjx
j
))

((D+(xi))

j
((D+(xixj))

j<j

((D+(xixjx
j
))
One can lift sections up along the vertical maps by multiplying by a
suciently high power of xi. Use this and a diagram chase to lift s
((D+(xi)) to a section of ((X).
Corollary 32.1. If T is a coherent sheaf on P
r
A
, then T admits a sur-
jection E T, where E is a nite direct sum of sheaves of the form
O
P
r (q).
Proof. If we have generators s1, . . . , st for T over D+(xi), multiply each
one by an appropriate power of xi to get a section of T(m). These sections
dene a map O
t
P
r T(m), where this m is the maximum of the twists
involved in lifting the various sj. This map is surjective over D+(xi).
Tensoring with O
P
r (m) gives a map of sheaves O
P
r T which is sur-
jective over D+(xi). Do this oover each D+(xi), and then take the direct
sum to prove the corollary.
Proof of Theorem 32.1. First we reduce to the case X = P
r
A
. Let i : X
P
r
A
be the closed immersion. Since i is exact and has an exact left adjoint,
it doesnt change the cohomology. Also, it preserves coherence, since this
can be checked locally. If on some open ane, T corresponds to the B/I-
module M, then iT corresponds to the module MB obtained by base
change along the ring map B B/I, which is still nitely generated
since this is a surjective ring map. Thus H
i
(X, T) H
i
(P
r
, iT), so we
may assume X = P
r
.
The result is true for i > r: in that case H
i
= 0 since P
r
can be
covered by r + 1 anes
44
. We also know the result in the case that T
is OX(n), since we have explicitly calculated H
i
(P
r
, OX(n)); we also get
44
Note - we cannot argue that H
i
= 0 for i > r because of dimension reasons, since A is
arbitrary, so it is dicult to get our hands on the dimension of P
r
.
104
the result when T is a nite direct sum of copies of these since H
i
is an
additive functor.
Now for the inductive step: assume that for some xed i0, and for all
coherent (, H
i
(P
r
, () is nitely generated whenever i > i0. We will use
this to show that H
i
0
(P
r
, T) is nitely generated. Now use the lemma
above to nd an E =

i
O
P
r (qi) and a morphism E T, and let K be
the kernel. Then we have an exact sequence
0 K E T 0,
which gives a long exact sequence in cohomology in which the following
terms appear:
. . . H
i
0
(P
r
, E )

H
i
0
(P
r
, T)

H
i
0
+1
(P
r
, K ) . . . .
Note that H
i
0
(P
r
, E ) is nitely generated, as we observed above, and that
H
i
0
+1
(P
r
, K ) is nitely generated by induction. Since this sequence is
exact, we can break o the following short exact sequence
0 im H
i
(P
r
, T) ker 0,
and both im and ker are nitely generated, since kernels and images
of nitely generated modules over a Noetherian ring are also nitely gen-
erated. This proves (a).
For (b), we only provide a sketch; the crux of the argument is the same
as (a). We will need the following result, which is useful enough to isolate
as a separate lemma.
Lemma 32.2 (Projection Formula). If i : (X, OX) (Y, OY ) is a mor-
phism of ringed spaces, T an OX-module, and E a locally free OY -module
of nite rank, then
i(T
O
X
i

E ) (iT)
O
Y
E
A similar argument was made for this statement in the special case
when E is a line bundle in an earlier lecture. In our case, taking Y = P
r
A
,
and E = O
P
r (1), we have
iT(n) = i(T
O
X
OX(1)
n
) = (iT)
O
P
r
O
P
r (1)
n
= (iT)(n),
which shows that we can assume that X = P
r
A
. Now, the statement
of (b) is true when i > r, as in (a). We also know that for T =
OX(n), H
i
(X, OX(n) = 0 for 0 < i < r and H
r
(X, OX(n)) is dual to
H
0
(X, OX(n r 1)), which shows that the statement holds when
T = OX(n); it holds for nite direct sums of these again by additivity of
H
i
.
For the inductive step x i0, and assume that for any ( and any i > i0,
there is an integer n0 such that H
i
(X, ((n)) = 0 if n n0. Then one
shows that there is an m0 such that H
i
0
(X, T(m)) = 0 for m m0, by
arguments similar to those of (a).
Note in conclusion that the integer n0 of (b) in principle depends on all
of X, A, L, and T, but the essential dependence is on T. In particular,
we can always twist T down to force larger values of n0.
105
33 Wednesday, April 11th: The Hilbert Polyno-
mial
Fix a projective scheme X over a Noetherian ring A. Let OX(1) be a
very ample sheaf. For a coherent sheaf T on X, we dene the Euler
characteristic of T to be the integer
(T) =

i
(1)
i
h
i
(X, T).
Note that by the results of the previous lecture, each h
i
is nite, and there
are only nitely many nonzero terms, so the sum makes sense.
For such T, we also dene the Hilbert polynomial of T to be the
function pF : Z Z given by
pF(n) = (T(n))
Note that for suciently large n, we have pF(n) = h
0
(X, T(n)), again
using the results of the previous lecture. To justify calling this function a
polynomial, we prove the following
Theorem 33.1. pF is a polynomial function.
Proof. First we reduce to the case where X = P
r
A
. Pick a closed immersion
i : X P
r
A
such that i

O
P
r OX(1). This can be done since OX(1) is
very ample. Then as we saw last time, the projection formula implies
(iT)(n) = i(T(n)), so piF = pF, and we may replace X by P
r
A
.
Now pick a hyperplane H X = P
r
; this gives us an exact sequence
0 OX(1) OX OH 0
where we are omitting the pushforward of OH from the notation. Since
tensoring with T is right exact we get another exact sequence
0 K T(1) T T

H
0,
where K is dened as the kernel of T(1) T. Let I denote the image
of the map T(1) T. We can break this sequence into two short exact
sequences
0 K T(1) I 0
0 I T T

H
0
Since is additive on short exact sequences (we proved this for curves a
few weeks ago), we get two equations
p
K
+p
I
= p
F(1)
p
I
+p
F|
H
= pF.
The dierence of these two equations is the equality
p
F|
H
p
K
= pF p
F(1)
.
By induction, we may assume p
F|
H
is a polynomial. Also, the sheaf K is
supported on H, and a coherent sheaf supported on a closed subscheme
106
can be realized as the pushforward of a coherent sheaf on that closed sub-
scheme (see lemma below for details). Thus we may also assume by induc-
tion that p
K
is a polynomial. Finally, note that p
F(1)
(n) = pF(n 1).
Thus the equation above shows that pF(n) pF(n 1) is a polynomial
function. In fact (check) this implies that pF itself is a polynomial func-
tion.
To explain our treatment of T

H
in the above, we have the following
lemma and its corollary.
Lemma 33.1. Let X be a Noetherian scheme, T a coherent sheaf on X
with set-theoretic support Z X. Give Z the reduced scheme structure,
and let I OX be the ideal sheaf of Z. Then there exists r such that I
r
annihilates T.
Corollary 33.1. T is isomorphic to the pushforward of a coherent sheaf
on V (I
r
).
Proof. The question of whether, for given r, I
r
annihilates T can be
checked locally. Having produced such r locally, we may take a nite
cover of X by open anes and use the maximum of all the r obtained
in our local constructions. Thus we may as well let X = Spec A. For
I A an ideal and F a nitely generated A-module, the condition that

F
is supported on V (I) means that, if g1, . . . , gs are generators for I, then

F is 0 on D(g1) . . . D(gs), i.e., Fg

i
= 0 for each i. Let m1, . . . , mt be
generators for F. Since Fg
i
= 0, there is an ni such that g
n
i
i
mj = 0 for
all j. Let n be the maximum of these ni as i varies from 1 to s. Thus
g
n
i
mj = 0 for all i, j.
We can choose r big enough so that, for any monomial g
a
1
1
g
as
s
of
degree r, at least one of the ai is at least n. Then for this r, any monomial
of degree r annihilates F, so I
r
F = 0. This proves the lemma.
To see the corollary, observe that if I
r
F = 0, then F has the struc-
ture of an A/I
r
-module, so the sheaf

F on Spec A can be viewed as the
pushforward of a sheaf

F

on Spec A/I
r
along the ring map A A/I
r
.
Globalizing, we have a closd immersion i : V (I
r
) X. The action of
i
1
OX on i
1
T factors through O
V (I
r
)
, and i
1
T is still coherent as
an O
V (I
r
)
-module. Moreover, the map T ii
1
T is an isomorphism
of OX modules, so we can view T as the pushforward of the coherent
O
V (I
r
)
-module i
1
T.
107