ALGEBRAIC GEOMETRY

CAUCHER BIRKAR
Contents
1. Introduction 1
2. Ane varieties 3
Exercises 10
3. Quasi-projective varieties. 12
Exercises 20
4. Dimension 21
5. Exercises 24
References 25
1. Introduction
These notes are for a rst graduate course on algebraic geometry. It
is assumed that the students are not familiar with algebraic geometry
so we have started from scratch. I have taken a moderate approach
emphasising both geometrical and algebraic thinking. We have bor-
rowed few main theorems of commutative algebra but rigorous proofs
are given for the rest, except the sections on sheaves and cohomology.
Algebraic geometry is about studying a system of polynomial equa-
tions. To be more precise, we rst specify a eld k and consider the
polynomial ring k[t
1
, . . . , t
n
] consisting of polynomials with coecients
in k. Next, we pick f
1
, . . . , f
r
in k[t
1
, . . . , t
n
] and consider the set of
solutions
(x
1
, . . . , x
n
) [ f
1
(x
1
, . . . , x
n
) = 0, . . . , f
r
(x
1
, . . . , x
n
) = 0
where x
i
k. There are two main questions:
Is the set of solutions non-empty?
Date: October 2011.
These are lecture notes for the rst 8 lectures of the part III algebraic geometry
course, Fall 2011. The other 16 lectures are not included. Send any comments to
c.birkar@dpmms.cam.ac.uk.
1
2 CAUCHER BIRKAR
If not empty, how does this set look like?
The most primitive form of algebraic geometry is linear algebra, that
is, if we choose the f
i
with deg f
i
1. Here we can use matrices
to investigate the solutions of the system. In particular, the set of
solutions always looks like a collection of vector spaces over k.
Now assume that k = Q. Then, we enter the eld of Diophantine
geometry which is generally considered to be in the realm of number
theory. By taking f
1
= t
2
1
+ t
2
2
+ 1 we realise that the set of solutions
can be empty and by taking f
1
= t
m
1
+t
m
2
t
m
3
you realise that we are
talking about serious stu (Fermats last theorem). In fact, roughly
speaking, for every statement in mathematics there is a Diophantine
equation so that the statement is provable if and only if the equation
has a solution. Therefore, there cannot be any easy way of determining
whether a system has solutions or not.
In this course we always assume that k is an algebraically closed eld
(unless stated otherwise) so the rst question is not very interesting.
We will mainly deal with the second question. Evidently there are
two sides to this story. On the one hand we have geometry (the set
of solutions) which has to do with intuition, and on the other hand
we have algebra (the polynomial ring) which has to do with precise
arguments and technicalities.
Remember that topology deals with continuous functions, analytic
geometry deals with analytic (holomorphic) functions, and dieren-
tial geometry deals with dierentiable functions. In contrast, algebraic
geometry deals with polynomial maps. Since a polynomial is dieren-
tiable, holomorphic, and continuous (at least when k = C), algebraic
geometers often use tools and ideas from topology, analysis, and dier-
ential geometry. However, since polynomials have rather special prop-
erties, algebraic geometry has its own unique techniques and methods,
and these often help other subjects when their problems are formulated
in terms of polynomials.
We assume good knowledge of commutative algebra, that is, general
properties of rings, ideals, modules, elds, etc.
References. The main reference for this course is the book of Sha-
farevich [5]. We will also make use of the books of Harris [2] and
Hartshorne [3]. For the commutative algebra background the main
source is Atiyah-Macdonald [1] but Matsumura [4] and Zariski-Samuel
[6] are also occasionally used.
ALGEBRAIC GEOMETRY 3
2. Affine varieties
Throughout this document, we assume k to be an algebraically closed
eld unless stated otherwise. We denote k[t
1
, . . . , t
n
] for the ring of
polynomials over k with the variables t
1
, . . . , t
n
. It is well-known that
this is a unique factorisation domain (UFD).
Corresponding to the algebraic object k[t
1
, , t
n
] we have the fol-
lowing geometric object.
Denition 2.1 (Ane space) The n-dimensional ane space A
n
k
is
dened as
A
n
k
= x = (x
1
, . . . , x
n
) [ x
i
k
We call A
1
k
the ane line and A
2
k
the ane plane. For the moment
the word n-dimensional is just a name but later on we will see that A
n
k
really has dimension = n.
We are interested in some special subsets of the n-dimensional ane
space A
n
k
.
Denition 2.2 (Ane algebraic set) An ane algebraic set X is a
set
X = V (f
1
, . . . , f
m
) = x A
n
k
[ f
i
(x) = 0, i
where f
i
k[t
1
, . . . , t
n
]. That is, an ane algebraic set is the set of
common solutions of nitely many polynomials.
One can also dene algebraic sets using ideals rather than polyno-
mials.
Denition 2.3 For an ideal I of k[t
1
, . . . , t
n
] dene its ane algebraic
set as
V (I) = x A
n
k
[ f(x) = 0, f I
The algebraic set of an ideal is not a new object by the following
theorem.
Theorem 2.4 (Hilbert basis theorem). Every ideal I of k[t
1
, . . . , t
n
]
is nitely generated, that is, there are f
1
, . . . , f
m
I such that I =
f
1
, . . . , f
m
. In other words, the ring k[t
1
, . . . , t
n
] is noetherian.
It is easy to see that if I = f
1
, . . . , f
m
, then V (I) = V (f
1
, . . . , f
m
).
Example 2.5 The ane algebraic subsets of A
1
k
are , A
1
k
and its
nite subsets.
Example 2.6 Let f k[t
1
, t
2
] be of degree 1, that is, f = at
1
+
bt
2
+c where a, b, c k. If b ,= 0, then we have
X = (x
1
,
ax
1
+ c
b
) A
2
k
[ x
1
k
4 CAUCHER BIRKAR
We get a similar description of X if a ,= 0. On the other hand, if
a = b = 0, then X = A
2
k
if c = 0 but X = if c ,= 0.
Example 2.7 When n = 2, V (t
1
t
2
) is a line, V (t
1
t
2
) is the union
of two lines that is the two axis, and V (t
1
, t
2
) is the origin (0, 0).
The reason we use ideals rather than polynomials is that it is more
convenient to relate algebraic sets and ideals.
Theorem 2.8. Let I, J be ideals of k[t
1
, . . . , t
n
]. Then,
V (IJ) = V (I J) = V (I) V (J) and V (I +J) = V (I) V (J).
Proof. Do as exercise.
If we are given an ane algebraic set we can try to reconstruct its
dening ideal. It is usually not possible to recover the ideal but we can
recover the biggest possible ideal which denes the given algebraic set.
Denition 2.9 Let X A
n
k
be an ane algebraic set. Dene its
ideal I
X
as
I
X
= f k[t
1
, . . . , t
n
] [ f(x) = 0, x X
Example 2.10 Let X = V (I) A
1
k
where I = t
2
. Then, I
X
= t.
In particular, this shows that some times I ,= I
X
.
Denition 2.11 (Zariski topology) We dene a topology on A
n
k
by
taking the ane algebraic sets as closed subsets. An open subset of
A
n
k
is then the complement of an ane algebraic set, that is, U =
A
n
k
X where X is an ane algebraic set. This induces a topology on
every subset of the ane space and we always refer to it as the Zariski
topology.
Recall that a topological space T is called irreducible if T = T
1
T
2
implies T = T
1
or T = T
2
where T
i
are closed subsets.
Denition 2.12 (Ane algebraic variety) An irreducible ane al-
gebraic set (with respect to the Zariski topology) is called an ane
algebraic variety or just an ane variety.
Theorem 2.13. Let X be an ane algebraic set. Then, X is an ane
algebraic variety i I
X
is a prime ideal.
Proof. Suppose that X A
n
k
is an ane algebraic variety. Assume
that fg I
X
for f, g k[t
1
, . . . , t
n
]. Then,
X = x X [ f(x) = 0 x X [ g(x) = 0
where the two latter sets are closed subsets of X. So, X should be
equal to one of them and so f I
X
or g I
X
.
ALGEBRAIC GEOMETRY 5
Conversely, suppose that I
X
is prime. If X = X
1
X
2
where X
i
are closed subsets then I
X
1
I
X
2
I
X
which implies that I
X
1
I
X
or
I
X
2
I
X
since I
X
is prime. Therefore, X X
1
or X X
2
.
The ring k[t
1
, . . . , t
n
] considered as an ideal of itself is not dened
to be prime, so in the theorem or other places we can exclude the case
X = .
Theorem 2.14 (Decomposition into irreducible components). Every
ane algebraic set is a union of nitely many ane algebraic varieties.
Proof. If an ane algebraic set is not irreducible then X = X
1
X
2
where X
1
, X
2
are closed subsets of X. If X
2
is not irreducible, then
X = X
1
X
3
X
4
and so on. This stops otherwise we nd a decreasing
sequence X

X of closed subsets. Therefore, we nd an

increasing sequence of ideals I
X
I
X
which contradicts the fact
that k[t
1
, . . . , t
n
] is a noetherian ring.
To see the relation between ane algebraic sets and ideals more
precisely we need to borrow the following important theorem.
Theorem 2.15 (Hilbert Nullstellensatz). Let I be an ideal of k[t
1
, . . . , t
n
].
Then,
(i) if X = V (I), then I
X
=

I,
(ii) any maximal ideal of k[t
1
, . . . , t
n
] is of the form t
1
x
1
, . . . , t
n
x
n
,
(iii) V (I) = i I = k[t
1
, , t
n
].
See [1] for a proof.
Theorem 2.16. An ane algebraic set X is a single point i I
X
is a
maximal ideal.
Proof. Use the Nullstellensatz theorem.
Remark 2.17 From the above theorems we can see that we have a
one-to-one correspondence between the prime ideals of k[t
1
, . . . , t
n
] and
the ane algebraic varieties X A
n
k
. We have a one-to-one correspon-
dence between the maximal ideals of k[t
1
, . . . , t
n
] and the points of A
n
k
.
We also have a one-to-one correspondence between the radical ideals
of k[t
1
, . . . , t
n
] and the ane algebraic sets in A
n
k
.
Example 2.18 (Hypersurface) An irreducible non-constant polyno-
mial f k[t
1
, , t
n
] denes a prime ideal I = f and so denes an
ane algebraic variety X = V (f) A
n
k
which is called a hypersurface.
If n = 2, this is also called a curve.
6 CAUCHER BIRKAR
Example 2.19 Any non-constant f k[t
1
, , t
n
] can be decom-
posed into irreducible factors f = f
1
f
m
. Therefore,
V (f) = V (f
1
) V (f
m
)
is the decomposition into irreducible components.
Example 2.20 (Subvarieties of A
2
k
) From commutative algebra, we
know that a prime ideal in k[t
1
, t
2
] is either 0, f for an irreducible
f, or a maximal ideal t
1
x
1
, t
2
x
2
. Therefore, an ane algebraic
variety in A
2
k
is either , A
2
k
, a curve or a point.
Example 2.21 (Product of ane algebraic sets) The set-theoretic
product A
n
k
A
m
k
may be identied with A
n+m
k
by identifying the point
(x
1
, . . . , x
n
), (y
1
, . . . , y
m
) with (x
1
, . . . , x
n
, y
1
, . . . , y
m
). Now if X A
n
k
and Y A
m
k
are ane algebraic sets, then X Y A
n+m
k
is an ane
algebraic set in the obvious way.
Denition 2.22 (Coordinate ring) For an ane algebraic set X
A
n
k
, its coordinate ring is dened as k[X] = k[t
1
, . . . , t
n
]/I
X
. This ring
is a nitely generated k-algebra. Elements of this ring are called regular
functions on X, they can be viewed as functions X k and each one
can be represented by a polynomial but not in a unique way.
Theorem 2.23. An ane algebraic set X A
n
k
is an ane algebraic
variety i k[X] is an integral domain. Moreover, X is a point i k[X] =
k.
Proof. Clear by Theorems 2.13 and 2.16.
Denition 2.24 (Regular map) Let X A
n
k
and Y A
m
k
be ane
algebraic sets. A regular map : X Y is a map given by =
(f
1
, . . . , f
m
) where f
i
are regular functions on X. is an isomorphism
if it has an inverse which is also a regular map.
Every regular map : X Y of ane algebraic sets gives rise to
a homomorphism of k-algebras

: k[Y ] k[X] by combining regular

functions on Y with .
For an ane algebraic set X and an ideal I of k[X] we dene
V
X
(I) = x X [ f(x) = 0, f I
Theorem 2.25. Let : X Y be a regular map of ane algebraic
sets. Then,
(i) is a continuous map with respect to the Zariski topology,
(ii)

is injective i (X) is dense in Y ,

(iii) is an isomorphism i

: k[Y ] k[X] is an isomorphism of

k-algebras.
ALGEBRAIC GEOMETRY 7
(iv)

is surjective i induces an isomorphism of X with a closed

subset of Y .
Proof. (i) This is the case because for an ideal I of k[Y ], we have

1
V
Y
(I) = V
X
(Ik[X]). In other words, inverse of closed subsets are
closed and so is continuous.
(ii) Suppose that (X) is dense in Y and

(f) = 0. Then, (X)

V
Y
(f) which is possible only if f = 0. Conversely, suppose that

is injective and assume that (X) V

Y
(I) for some ideal I of k[Y ].
Obviously, I ker

= 0, so (X) must be dense in Y .

(iii) Exercise.
(iv) Assume that there is a closed subset Z Y such that (X) Z
and that the map : X Z induced by is an isomorphism. If
: Z Y is the inclusion map, then = . By (iii),

: k[Z]
k[X] is an isomorphism. Since Z is a closed subset of Y , k[Z] = k[Y ]/I
for some ideal I and

is just the natural surjection k[Y ] k[Z]. Now

the claim follows from

.
Conversely, if

is surjective, then we get an isomorphism k[Y ]/I

k[X] for some ideal I and another application of (iii) shows that
induces an isomorphism of X with the closed subset of Y dened by
I.
The theorem suggests that ane algebraic sets are uniquely deter-
mined by their coordinate rings. So, one should be able to discover all
the properties of an ane algebraic set from its coordinate ring. This
led to the revolution by Grothendieck which completely transformed
algebraic geometry and related subjects. The idea is that why not look
at any commutative ring and dene some space for it and study its
geometry. Such spaces are called schemes.
Example 2.26 Regular functions on an ane algebraic set X are
regular maps from X to k = A
1
k
.
Example 2.27 The map : A
n
k
A
m
k
dened by = (t
1
, . . . , t
m
),
where m n, is a regular map which we refer to as a projection. If
X A
n
k
is an ane algebraic set, the restriction map [
X
: X A
m
k
is
also a regular map.
Example 2.28 Let X = V (t
1
t
2
1) A
2
k
. The map to A
1
k
given by
= t
1
is regular and injective but not surjective so not an isomorphism.
However, (X) = A
1
k
0 which is dence in A
1
k
.
Example 2.29 The regular map : A
1
k
Y given by = (t
2
, t
3
)
where Y = V (s
2
2
s
3
1
) A
2
k
(cusp singularity) is a 1-1 regular map
8 CAUCHER BIRKAR
however, it is not an isomorphism. In fact, the corresponding ho-
momorphism

: k[Y ] = k[s
1
, s
2
]/s
2
2
s
3
1
k[t] is determined by

(s
1
) = t
2
and

(s
2
) = t
3
. So t cannot be in the image of

hence
it is not an isomorphism. If one tries to get an inverse it should be as
= s
2
/s
1
which is not a regular function. Soon we will see that is a
rational function.
Denition 2.30 (Rational function) Let X A
n
k
be an ane alge-
braic variety. Since k[X] is an integral domain, its eld of fractions
k(X) can be constructed which is called the eld of rational functions
of X or the function eld of X. Each element of k(X) can be rep-
resented by f/g such that g ,= 0 and f, g k[X]. Equivalently, we
can write it as f/g where f, g k[t
1
, . . . , t
n
] such that g / I
X
. It can
be considered as a function : X k which may not be dened
everywhere. We say that = f/g is dened or regular at x X if we
can nd h, e such that = f/g = h/e and e(x) ,= 0.
Example 2.31 (i) Regular functions are also rational functions. (ii)
When X = A
n
k
, any element of the eld k(X) = k(t
1
, . . . , t
n
) is a
rational function on X. (iii) The function dened in Example 2.29
is a rational function on Y = V (s
2
2
s
3
1
) A
2
k
.
Theorem 2.32. Let X be an ane algebraic variety. A rational func-
tion : X k is regular everywhere (i.e. at every point) i it is a
regular function.
Proof. Suppose that is a rational function which is regular every-
where. Then, for each x X we can write = f
x
/g
x
such that g
x
(x) ,=
0. Now if I is the ideal in k[X] generated by all g
x
, then V
X
(I) = and
so I = k[X] by Hilbert Nullstellensatz theorem. Thus,

m
i=1
h
i
g
x
i
= 1
for some h
i
k[X] and nitely many points x
1
, . . . , x
m
X. Now
=

h
i
g
x
i
=

h
i
g
x
i
=

h
i
f
x
i
where = f
x
i
/g
x
i
hence k[X].
Let X be an ane algebraic variety and k(X) a rational function.
The domain of is the set of points of X at which is regular.
Theorem 2.33. Let X be an ane algebraic variety. Then, the do-
main of a rational function : X k is a nonempty open subset of
X.
Proof. Since we can write = f/g such that g ,= 0, the domain U is
not empty. Now if I is the ideal in k[X] generated by all g of dierent
ways of writing = f/g, then U = X V
X
(I).
ALGEBRAIC GEOMETRY 9
Denition 2.34 (Rational map) Let X A
n
k
be an ane algebraic
variety and Y A
m
k
an ane algebraic set. A rational map : X Y
is given by = (
1
, . . . ,
m
) where
i
k(X) and such that
(x) = (
1
(x), . . . ,
m
(x)) Y
if all the
i
are dened at x, and we say that is dened or regular at
x. In particular, (X) = (x) [ regular at x.
If (X) is dense in Y , then we get a homomorphism of the function
elds

: k(Y ) k(X). is called a birational isomorphism if Y is

also a variety and such that there is
1
which is the inverse of where
they are dened. We also call X and Y birational.
Example 2.35 Let : A
1
k
Y be the regular map given by =
(t
2
, t
3
) where Y = V (s
2
2
s
3
1
) A
2
k
. We proved that is not an
isomorphism. Now = s
2
/s
1
is the inverse of which is not regular
but rational. So, is a birational isomorphism.
Example 2.36 (Conics in the plane) Let Y = V (f) A
2
k
be a variety
dened by an irreducible polynomial f of degree 2 and after a linear
change of variables we can assume that (0, 0) Y . Let L
s
= V (t
2
st
1
)
where s k. Now L
s
intersects Y at (0, 0). Since f has degree 2, it
intersects Y at another point y
s
except for nitely many values of s.
To see this one needs to solve the equation g(t
1
) = f(t
1
, st
1
) = 0.
Except for nitely many s k, g(t
1
) is of degree 2 so g(t
1
) = t
1
h(t
1
)
for some polynomial h(t
1
) = A(s)t
1
+ B(s). Thus, the other root of
g is given by a rational function
1
= B(s)/A(s) which determines
the rst coordinate of y
s
. The other coordinate of y
s
is given by the
rational function
2
= s
1
. In short, we have dened a rational map
: A
1
k
Y given by = (
1
,
2
). Now dene : Y A
1
k
by
= t
2
/t
1
. We see that and are inverse of each other so Y and A
1
k
are birationally isomorphic.
Denition 2.37 (Quasi-ane algebraic set) A quasi-ane algebraic
set X A
n
k
is an open subset of an ane algebraic set. A regular
function on X is a function : X k such that for every x X, there
is a neighborhood U of x, and f, g k[t
1
, . . . , t
n
] such that on U, and
f/g are equal, in particular, g has no zero on U. The set of regular
functions on X is denoted by k[X] which is a k-algebra.
If X is irreducible, a rational function : X k on X is the
equivalence class of a regular function on some open subset of X in the
sense that if
U
and
V
are regular functions on the open subsets U
and V respectively, then
U
is equivalent to
V
if
U
[
UV
=
V
[
UV
.
The set of rational functions on X is denoted by k(X) which is a eld
and is called the function eld of X.
10 CAUCHER BIRKAR
When X A
n
k
is an ane algebraic set, we dened regular and
rational functions on X in Denitions 2.22 and 2.30. It is not too
dicult to show that those notions coincide with those in the previous
denition (see exercises below).
Exercises
Exercise 2.38 For an ane algebraic set X = V (I), prove that
I I
X
.
Exercise 2.39 Let X

, X

A
n
k
be ane algebraic sets. Prove that
if X

then I
X
I
X
. Prove that I
X

X
= I
X
I
X
. Show that
it may happen that I
X

X
,= I
X
+I
X
.
Exercise 2.40 Let X = V (t
2
1
+ t
2
2
1, t
1
1) A
2
k
. What is I
X
? Is
I
X
a prime ideal?
Exercise 2.41 Let X = V (t
2
1
+ t
2
2
+ t
2
3
) A
3
k
. Determine I
X
when
the characteristic of k is 2. Determine I
X
when the characteristic of k
is not 2.
Exercise 2.42 Decompose X = V (t
2
1
t
2
t
3
, t
1
t
3
t
1
) A
3
k
into its
irreducible components.
Exercise 2.43 * Let k be any eld not necessarily algebraically
closed, and let f, g k[t
1
, t
2
] be non-constant polynomials without
common factor in their factorisations into irreducible factors. Prove
directly that V (f) V (g) A
2
k
is nite. (Consider f, g as polynomials
over the ring k(x)[y] or similarly k(y)[x])
Exercise 2.44 Let X = P
1
, . . . , P
m
A
n
k
be nitely many distinct
points. Show that
k[X] = k[t
1
, . . . , t
n
]/I
X

m

j=1
k[t
1
, . . . , t
n
]/I
{P
j
}

m

j=1
k
Exercise 2.45 Let : A
1
k
A
1
k
be an isomorphism. Prove that is
given by a polynomial of degree 1.
Exercise 2.46 Let : A
2
k
A
2
k
be given by = (t
1
, t
1
t
2
). Is (A
2
k
)
open, close or dense?
Exercise 2.47 Let M
nm
be the set of nm matrices over k. Prove
that there is a bijection from M
nm
to A
nm
. Prove that the set of
non-invertible matrices in M
nn
corresponds to a hypersurface in A
n
2
.
Exercise 2.48 Let X = V (t
2
1
+t
2
2
1) A
2
k
and let be the rational
function of X dened by = (1 t
2
)/t
1
. Determine the domain of .
ALGEBRAIC GEOMETRY 11
Exercise 2.49 Let X be an ane algebraic variety. Prove that
(i) the intersection of nitely many nonempty open subsets of X is
open and nonempty,
(ii) if U is a nonempty open subset of X, then it is dense in X,
(iii) the domain of nitely many rational functions is also open and
nonempty.
Exercise 2.50 Prove that if two rational functions of an ane al-
gebraic variety are equal on a nonempty open subset, they should be
equal.
Exercise 2.51 Prove that a rational map : X Y of ane vari-
eties is a birational isomorphism i

is an isomorphism of elds.
Exercise 2.52 For an ane algebraic variety X and an open subset
U X we dene
k[U] := k(X) [ is regular at every point of U
Now assuming X = A
1
k
and U = X 0 determine k[U]. Do the same
with X = A
2
k
and U = X (0, 0).
Exercise 2.53 Let X be an ane algebraic variety, 0 ,= f k[X],
and U
f
:= X V
X
(f). Show that k[U
f
] is a k-algebra isomorphic
to k[X]
f
(the latter object is the localisation of the ring k[X] at the
element f).
Exercise 2.54 Let Y = V (t
2
2
t
2
1
t
3
1
) A
2
k
and let : A
1
k
Y be
the regular map dened by = (s
2
1, s(s
2
1)). Describe

and
prove that is not an isomorphism but it is a birational isomorphism.
Exercise 2.55 Assume characteristic of k is not 2. Let Y = V (t
2
1
+
(t
2
1)
2
1) A
2
k
. Show that Y is birational to A
1
k
by constructing
rational maps : A
1
k
Y and : Y A
1
k
explicitly so that and
are inverse of each other. Do the same for Y = V (t
2
1
+(t
2
1)
2
1) A
2
k
.
Exercise 2.56 * Show that the regular (and rational) functions on
an ane algebraic set X are the same as the regular (and rational)
functions on X considered as a quasi-ane algebraic set. (Hint: show
that there is a natural injective homomorphism k[X]
f
k[U
f
] for each
0 ,= f k[X] where U
f
= X V
X
(f); next use appropriate coverings
of X by such U
f
.)
12 CAUCHER BIRKAR
3. Quasi-projective varieties.
A (commutative) graded ring is a commutative ring S =

+
d=0
S
d
such that S
d
are abelian groups and S
d
S
d
S
d+d
. Elements of S
d
are
called homogeneous of degree d. An ideal I of S is called homogeneous
if I =

+
d=0
(S
d
I). This is equivalent to I being generated by homo-
geneous elements. In particular, S
+
=

+
d=1
S
d
is a homogeneous ideal.
Example 3.1 The most important example for us is the polynomial
ring S = k[s
0
, . . . , s
n
] which has a natural graded structure: S
d
is the
vector space over k generated by all the monomials of degree d. By
Hilbert basis theorem, every homogeneous ideal of this ring is nitely
generated.
We say that (x
0
, . . . , x
n
), (y
0
, . . . , y
n
) A
n+1
k
are equivalent if there
is a nonzero a k such that (y
0
, . . . , y
n
) = (ax
0
, . . . , ax
n
). The equiv-
alence class of (x
0
, . . . , x
n
) is denoted by (x
0
: : x
n
).
Denition 3.2 (Projective space) The n-dimensional projective space
is dened as
P
n
k
= x = (x
0
: : x
n
) [ x
i
k and some x
i
,= 0
We call P
1
k
the projective line and call P
2
k
the projective plane.
There are good reasons to introduce the projective space. Projec-
tive algebraic sets (that will be introduced soon) are the analogue of
compact topological spaces. Compact spaces behave better than the
non-compact ones. For example later on we will see that any two pro-
jective curves in P
2
k
intersect. This is clearly not true in A
2
k
. One can
think of the projective space as a compactied version of the ane
space. For example,
P
1
k
= (1 : a) [ a k (0 : 1)
and we can identify (1 : a) [ a k with A
1
k
which is like saying that
P
1
k
is obtained by adding the point at innity to A
1
k
.
Now let F k[s
0
, . . . , s
n
] and x P
n
k
. In general, if
x = (x
0
: : x
n
) = (x

0
: : x

n
)
then F(x
0
: : x
n
) and F(x

0
: : x

n
) are not equal so it does not
make sense to talk about the value F(x). However, if F is homogeneous,
then F(x
0
: : x
n
) = 0 i F(x

0
: : x

n
) = 0 so in this case at least
the vanishing statement F(x) = 0 makes sense (but still F(x) does not
make sense).
ALGEBRAIC GEOMETRY 13
Denition 3.3 (Projective algebraic set) Let I be a homogeneous
ideal of k[s
0
, . . . , s
n
]. Dene
V (I) = x P
n
k
[ F(x) = 0, F I homogeneous
A projective algebraic set X in P
n
k
is dened to be X = V (I) for some
homogeneous ideal I of k[s
0
, . . . , s
n
]. If F
1
, . . . , F
m
are homogeneous
polynomials generating I, then V (I) = V (F
1
, . . . , F
m
).
Denition 3.4 (Zariski topology) Similar to the ane case P
n
k
is a
topological space by taking its closed subsets to be projective algebraic
sets. This induces a topology on each subset of P
n
k
which we again call
the Zariski topology.
For a projective algebraic set X P
n
k
, dene its ideal I
X
as the ideal
generated by
F k[s
0
, . . . , s
n
] [ F is homogeneous, F(x) = 0 for all x X
Obviously, I
X
is a homogeneous ideal. Now dene the coordinate ring
of X as S[X] = k[s
0
, . . . , s
n
]/I
X
which is a graded ring.
Theorem 3.5. Let I be a homogeneous ideal of k[s
0
, . . . , s
n
]. Then,
(i) X = V (I) = i S
+


I,
(ii) If X = V (I) ,= , then I
X
=

I.
Proof. (i) Let Y = V (I) A
n+1
k
. Then, the map : Y 0 X
given by (x
0
, . . . , x
n
) = (x
0
: : x
n
) is surjective. Thus, X = i
Y 0 i S
+
=< s
0
, . . . , s
n
>

I.
(ii) Since I is homogeneous, I
Y
=

I is also homogeneous. Hence
I
Y
I
X
. On the other hand, since X ,= , I
X
,= S and since I
X
is ho-
mogeneous, every element of I
X
vanishes at every point of Y (including
the origin) hence I
X
I
Y
.
Denition 3.6 (Quasi-projective algebraic set) A quasi-projective
algebraic set is an open subset of a projective algebraic set. A projective
variety is an irreducible projective algebraic set. A quasi-projective
variety is an open subset of a projective variety.
As mentioned earlier, a homogeneous polynomial F k[s
0
, . . . , s
n
]
does not necessarily dene a function on P
n
k
or a subset of it. However,
if we take two homogeneous polynomials F, G k[s
0
, . . . , s
n
] of the
same degree, by putting
(F/G)(x
0
: : x
n
) = F(x
0
, . . . , x
n
)/G(x
0
, . . . , x
n
)
if G(x
0
, . . . , x
n
) ,= 0, we get a well-dened function F/G: P
n
k
V (G)
k.
14 CAUCHER BIRKAR
Denition 3.7 (Regular and rational functions) Let X be a quasi-
projective algebraic set. A function : X k is called a regular
function if for every x X, there is a neighborhood U of x, and
homogeneous polynomials F, G of the same degree such that on U,
and F/G are equal, in particular, G has no zero on U. The set of
regular functions on X is denoted by k[X] which is a k-algebra.
If X is irreducible, a rational function : X k on X is the
equivalence class of a regular function on some open subset of X in the
sense that if
U
and
V
are regular functions on the open subsets U
and V respectively, then
U
is equivalent to
V
if
U
[
UV
=
V
[
UV
.
A rational function then is uniquely determined by some F/G where
G is not identically zero on X. The set of rational functions on X is
denoted by k(X) which is a eld and is called the function eld of X.
Denition 3.8 (Regular and rational maps) Let X and Y be quasi-
ane or quasi-projective algebraic sets. A regular map : X Y is
a continuous map (with respect to the Zariski topologies) such that it
sends a regular function on a open subset V Y to a regular function
on
1
V , that is, if f k[V ] then the composition f k[
1
V ]. The
regular map : X Y is called an isomorphism if it has an inverse
which is also a regular map. When we say a quasi-ane or a quasi-
projective algebraic set is ane, we mean that it is isomorphic to an
ane algebraic set in some ane space.
When X is irreducible, a rational map : X Y is the equivalence
class of a regular map
U
: U Y for some open subset U X. We
say that is a birational isomorphism if Y is also a variety and has a
rational inverse . In this case, we also say that X and Y are birational.
Let X P
n
k
be a quasi-projective algebraic set and F
0
: : F
m
homogeneous polynomials of the same degree such that at least one of
them is not identically zero on X. Then, we can dene a well-dened
function
(F
0
: : F
m
): X V (F
0
, . . . , F
m
) P
m
k
by putting
(F
0
: : F
m
)(x
0
: : x
n
) = (F
0
(x
0
: : x
n
) : : F
m
(x
0
: : x
n
))
Theorem 3.9. Let X P
n
k
and Y P
m
k
be quasi-projective algebraic
sets and : X Y a map (set theoretic). Then, is a regular map
i for every x X, there is a neighborhood x U, and homogeneous
polynomials F
i
of the same degree such that = (F
0
: : F
m
) on U.
ALGEBRAIC GEOMETRY 15
Proof. We take the rings k[s
0
, . . . , s
n
] and k[r
0
, . . . , r
m
] corresponding
to P
n
k
and P
m
k
, respectively. First assume that is a regular map. Let
x X. Without loss of generality, we may assume (x) = (y
0
: : y
m
)
with y
0
,= 0. Let V = Y V
Y
(r
0
). Now r
i
/r
0
is a regular function
on V for any i, and by denition it gives a regular function
i
on

1
V . Moreover, the map
1
V V given by (1 :
1
: :
m
)
is identical to on
1
V . Each
i
is expressed as F
i
/G
i
on some
neighborhood of x. If we take U to be the intersection of all these
neighborhoods, then is given by (1 : F
1
/G
1
: : F
m
/G
m
) on U.
Now (G
1
...G
m
: F
1
G
2
. . . G
m
: ) is the desired map.
Now assume that is a map which is locally given as in the statement
of the theorem. First we prove that it is continuous. For each x X
there is a neighborhood x U on which is given by = (F
0
: :
F
m
). It is enough to prove that = [
U
: U Y is continuous. A
closed subset Z of Y is as Z = V
Y
(H
1
, . . . , H
l
). Now the composition
of H
1
, . . . , H
l
with give the dening equations of
1
Z which is a
closed subset of U. Hence, is continuous.
Now let V be any open subset of Y and a regular function on V .
By denition, is locally given as F/G. Hence the composition is
locally given by some F

/G

which means that is a regular function

on
1
V .
Corollary 3.10. Let X P
n
k
be a quasi-projective algebraic variety
and Y P
m
k
a quasi-projective algebraic set. A rational map : X
Y is uniquely determined by some = (F
0
: : F
m
) on some non-
empty open subset U X.
The following theorem claries the relation between the ane and
the projective worlds.
Theorem 3.11. Algebraic sets dened in the ane and projective
spaces are related as follows:
(i) Quasi-ane algebraic sets are quasi-projective.
(ii) Each quasi-projective algebraic set is covered by ane algebraic
sets.
Proof. Let U
i
= (x
0
: : x
n
) P
n
k
[ x
i
,= 0. Then, we have a
bijection
i
: A
n
k
U
i
given by
i
(a
1
, . . . , a
n
) = (a
1
: : 1 : : a
n
)
where 1 appears as the i-th coordinate. Note that the inverse of
i
is
given by the function that takes (x
0
: : x
n
) to
(
x
0
x
i
, . . . ,
x
i1
x
i
,
x
i+1
x
i
, . . . ,
x
n
x
i
)
Clearly, P
n
k
is covered by the sets U
0
, . . . , U
n
.
16 CAUCHER BIRKAR
For f k[t
1
, . . . , t
n
] of degree d, dene the homogeneous polyno-
mial F associated to f by s
d
i
f(
s
0
s
i
, . . . ,
sn
s
i
) which is in k[s
0
, . . . , s
n
] (note
that here we are ignoring
s
i
s
i
). Conversely, for a homogeneous G
k[s
0
, . . . , s
n
], we can associate g = G(t
1
, . . . , 1, . . . , t
n
) k[t
1
, . . . , t
n
]
where 1 appears in the i-th coordinate.
(i) Let X A
n
k
be a quasi-ane algebraic set and let X

=
i
(X).
We will see that
i
[
X
: X X

is an isomorphism. It is easy to see

that
i
is a homeomorphism which implies that
i
[
X
is also a homeo-
morphism. Let V X be an open subset and V

=
i
[
X
(V ). Then,
a regular function on V is a function V k which is locally given
by a fraction f/g. Such a function corresponds to a function V

k
which is locally of the form s
d
i
F/G for some d Z where F, G are the
homogeneous polynomials associated to f, g respectively.
Conversely, a regular function V

k is locally of the form F/G

for some homogeneous polynomials F, G of the same degree. Such a
function corresponds to a function V k which is locally of the form
f/g where f, g are associated to F, G.
So we have a 1-1 correspondence between regular functions on V and
V

. That is, we have an isomorphism.

(ii) Let X be a quasi-projective algebraic set. We will show that for
every x X, there is a neighborhood x U which is ane, that is, it is
isomorphic to an ane algebraic set. Since x is in one of the ane pieces
U
i
covering the projective space, x has a neighborhood U = U
i
X
which is quasi-ane. Now by replacing X with U we can assume that
X A
n
k
, that is, it is quasi-ane. Let Y = X = V (I) be the closure
in A
n
k
. Further shrinking X, we can assume that X = Y V
Y
(f) for
some polynomial f. Now let W = A
n
k
V (f) and let Z be the ane
algebraic set in A
n+1
k
= A
n
k
A
1
k
dened by Z = V (ft
n+1
1). The
map : Z W given by = (t
1
, . . . , t
n
) is a regular map. Moreover,
the map : W Z given by = (t
1
, . . . , t
n
, 1/f) is the inverse of .
So, Z and W are isomorphic and this induces an isomorphism between
X = W Y and a closed subset of Z.
We will now give a series of examples of quasi-projective algebraic
sets and their maps.
Example 3.12 Let M
22
be the set of 2 2 matrices over k. Any
nonzero element of M
22
determines a point of P
3
k
. Now all such ma-
trices which are not invertible dene a projective algebraic set in P
3
k
.
Example 3.13 (Projectivisation of an ane algebraic set) Let X =
V (I) A
n
k
be an ane algebraic set. By identifying A
n
k
with U
0
as in
the proof of Theorem 3.11 we can take the closure X in P
n
k
which is by
ALGEBRAIC GEOMETRY 17
denition the intersection of all projective algebraic sets in P
n
k
contain-
ing X. Let J be the homogeneous ideal of k[s
0
, . . . , s
n
] generated by all
the F associated to f I. We prove that X = V (J) P
n
k
. Let x X
and G J homogeneous. Then, G =

P
i
F
i
where F
i
is associated to
f
i
I, and so G(x) = 0 because F
i
(x) = 0 since f
i
(x) = 0. In other
words, X X V (J).
If X ,= V (J), then X V (G + J) V (J) for some homogeneous
polynomial G. Then X = V (g +I) where g is associated to G. Thus,
g

I and so G
l
J for some l N. Therefore, V (G +J) = V (J),
a contradiction.
Example 3.14 Let X = P
n
k
. Since A
n
k
is a dense open subset of X,
k(X) = k(A
n
k
) hence k(X) k(t
1
, . . . , t
n
) where we could identify t
i
with s
i
/s
0
.
Example 3.15 Let X = P
n
k
, we prove that k[X] = k. Let : X
k be a regular function. Then, for x X there is a neighborhood
x U and F, G homogeneous of the same degree in k[s
0
, . . . , s
n
] such
that = F/G on U, and G has no zero on U. We can assume that
F, G have no common factor. Suppose that G is not a constant and
let x

V (G) V (F). Then, there is a neighborhood x

and
F

, G

homogeneous of the same degree in k[s

0
, . . . , s
n
] such that =
F

/G

on U

, and G

has no zero on U

, and F

, G

have no common
factor. So, F/G and F

/G

give the same values on U U

,= . Since
X is irreducible, this is possible only if FG

G = 0. This is a
contradiction because F(x

)G

(x

) ,= 0 but F

(x

)G(x

) = 0. So, F, G
are constant and so is .
Example 3.16 Let X = V (s
1
s
2
s
2
0
) P
2
k
. Then, : X P
1
k
given
by = (s
0
: s
1
) = (s
2
: s
0
) is a regular map. Now is actually an
isomorphism with the inverse : P
1
k
X given by = (r
0
r
1
: r
2
1
: r
2
0
).
This example also shows that X
1
= X (0 : 1 : 0), (0 : 0 : 1) is
isomorphic to P
1
k
(1 : 0), (0 : 1). The rst one is a closed ane set
in A
2
k
but the second one is an open subset of A
1
k
.
Example 3.17 : P
n
k
P
n
k
given by = (s
d
0
, . . . , s
d
n
) is a regular
map. If d > 1, it is not an isomorphism.
Example 3.18 (Cremona transformation) : P
2
k
P
2
k
given by
= (s
1
s
2
: s
0
s
2
: s
0
s
1
) = (1/s
0
: 1/s
1
: 1/s
2
)
is a rational map. It is not regular at the three points (0 : 0 : 1),
(0 : 1 : 0), and (1 : 0 : 0). However, it is birational its inverse being
= (r
1
r
2
: r
0
r
2
: r
0
r
1
). This is called a Cremona transformation.
18 CAUCHER BIRKAR
Example 3.19 (Veronese embedding) Let d be a natural number and
m = (
n +d
d
) 1. Dene : P
n
k
P
m
k
by the m monomials s
d
0
0
s
dn
n
where

d
i
= d and d
i
N 0. Then, is a regular map which is
called the Veronese embedding.
Example 3.20 (Rational normal curve) The map : P
1
k
P
d
k
given
by
= (s
d
0
: s
d1
0
s
1
: s
d2
0
s
2
1
: : s
d
1
)
is a regular map and its image, say X, is called the rational normal
curve of degree d. We can also dene X as the set given by the equations
r
i
r
j
r
i1
r
j+1
for 1 i j d 1. Indeed, every point in X satises
those equations. Actually, : P
1
k
X is an isomorphism and its inverse
is given by the map
= (r
0
: r
1
) = (r
1
: r
2
) = = (r
d1
: r
d
)
Example 3.21 (Segre map) We would like to turn the set-theoretic
product P
m
k
P
n
k
into a projective algebraic set. We dene a map
: P
m
k
P
n
k
P
(m+1)(n+1)1
k
which is given by s
i
t
j
at the (i, j)-th coordinate where s
i
are the vari-
ables corresponding to P
m
k
and t
j
are variables corresponding to P
n
k
.
The image of , say X, turns out to be a closed subset of P
(m+1)(n+1)1
k
given by the equations r
i,j
r
p,q
r
i,q
r
p,j
. Moreover, is a bijection onto
X so we can identify P
m
k
P
n
k
with X and carry the topology on X
onto P
m
k
P
n
k
and again call it the Zariski topology. In addition, we
can think of P
m
k
P
n
k
as a projective algebraic set via .
Example 3.22 (Products) Let be the Segre map. Let
U
i
= (x
0
: : x
m
) P
m
k
[ x
i
,= 0
and
V
j
= (y
0
: : y
n
) P
n
k
[ y
j
,= 0
From the denition of we can see that (U
i
V
j
) is an open subset of
(P
m
k
P
n
k
) and that (U
i
V
j
) A
m+n
k
. Now if X P
m
k
and Y P
n
k
are closed subsets, then X Y is also a closed subset in the Zariski
topology of P
m
k
P
n
k
induced by : to see this it is enough to observe
that
((X Y ) (U
i
V
j
))
is a closed subset of (U
i
V
j
). Similar reasoning shows that if X
and Y are only assumed to be quasi-projective, then (X Y ) is also
naturally quasi-projective hence it induces a quasi-projective structure
ALGEBRAIC GEOMETRY 19
on X Y . Moreover, the topology on X Y is ner than the product
topology.
An important property of the product XY is that it comes with two
natural regular maps XY X and XY Y called the rst and
second projections (not to be confused with other kinds of projections)
where the rst map sends a point (x, y) to x and the second map sends
(x, y) to y.
Example 3.23 (Projection from a point) We can identify P
n1
k
with
the closed subset of P
n
k
consisting of points (x
0
: : x
n
) with x
n
= 0.
Pick a point y P
n
k
outside P
n1
k
. We can dene a map
: P
n
k
y P
n1
k
by sending a point x to the intersection L P
n1
k
where L is the line
passing through x and y. If we x the coordinates x = (x
0
: : x
n
)
and y = (y
0
: : y
n
), then by denition L is the set ax + by P
n
k
[
a, b k (obviously, a ,= 0 or b ,= 0).
We refer to as projection from the point y. If X P
n
k
is quasi-
projective, the restriction of gives a regular map X y P
n1
k
which we again call projection from y. If we take y = (0 : 0 : : 0 : 1),
then can be simply described as = (s
0
: : s
n1
).
Example 3.24 (Graph of a map) Let : X Y be a regular map.
We dene the graph of to be the closed subset

XY consisting
of those points (x, y) such that y = (x). Now assume that is just a
rational map and let U be its domain of denition, that is, the set of
points where is regular. We take the graph
|
U
and dene the graph

to be the closure of
|
U
inside X Y .
Example 3.25 (Blow up) Let : A
n
k
P
n1
k
be the rational map
which sends a point (x
1
, . . . , x
n
) to (x
1
: : x
n
). The map is dened
everywhere except at the origin 0 = (0 : : 0). We denote the graph

by B
0
A
n
k
and call it the blow up of A
n
k
at the origin 0. The rst
projection A
n
k
P
n1
k
A
n
k
gives us a regular map : B
0
A
n
k
A
n
k
which is also referred to as the blow up of A
n
k
at the origin 0.
Put V = A
n
k
0. Then the restricton of gives an isomorphism

1
V V : the inverse of this map is the map which sends a point v
to (v, (v)). On the other hand,
1
0 P
n1
k
which we can verify
as follows. Pick a point 0 ,= x A
n
and let L be the line in A
n
which
passes through x and the origin, and let L

= L 0 (unlike in the
projective case the line through v, v

A
n
k
is dened as av +(1a)v

[
a k). Every point of L

is mapped to the same point of P

n1
k
by ,
say y. Thus
L

(L y)
20 CAUCHER BIRKAR
and since

is a closed subset of A
n
k
P
n1
k
, the intersection

(L y) is a closed subset of L y hence it is equal to L y
which in particular means that (0, y)
1
0. Therefore, we have

1
0 = 0 P
n1
k
P
n1
k
.
Exercises
Exercise 3.26 Prove that intersection, sum, product and radical of
homogeneous ideals in a graded ring are again homogeneous.
Exercise 3.27 For homogeneous ideals of k[s
0
, . . . , s
n
], prove that
V (IJ) = V (I J) = V (I) V (J) and V (

I
l
) =

V (I
l
).
Exercise 3.28 Let X

, X

P
n
k
be projective algebraic sets. Prove
that if X

then I
X
I
X
. Prove that I
X

X
= I
X
I
X
.
Exercise 3.29 Let X be a quasi-projective algebraic set and X =

U
j
a cover of X by open subsets U
j
. Prove that only nitely many
of U
j
would be enough to cover X.
Exercise 3.30 Show that regular maps of ane algebraic sets are the
same as their regular maps as quasi-projective algebraic sets. Prove the
same for rational maps of ane varieties.
Exercise 3.31 Let U ,= be an open subset of a quasi-projective
variety X. Prove that k(U) = k(X).
Exercise 3.32 Let X be a closed subset of P
m
k
P
n
k
and assume that
s
i
and t
j
are the variables corresponding to P
m
k
and P
n
k
respectively.
Show that there are polynomials F
1
, . . . , F
l
k[s
0
, . . . , s
m
, t
0
, . . . , s
n
]
that are homogenuous in the s
i
and also homogenuous in the t
j
such
that X is the set of zeros of F
1
, . . . , F
l
.
Exercise 3.33 Find the equations which determine the blow up B
0
A
n
k
as a closed subset of A
n
k
P
n1
k
.
Exercise 3.34 Let X and Y be quasi-projective varieties. Show that
the following are equivalent:
X and Y are birational,
there are open subsets U X and V Y such that U and V are
isomorphic,
there is an isomorphism of elds k(X) k(Y ) over k.
Exercise 3.35 The variety X = V (s
0
s
3
s
1
s
2
) P
3
k
is called the
quadric surface. Show that
X P
1
k
P
1
k
,
X is birational to P
2
k
.
ALGEBRAIC GEOMETRY 21
4. Dimension
Let X be a topological space. The dimension of X, that is dimX,
is the sup of l Z such that there is a sequence Z
0
Z
1
Z
l
of
nonempty closed irreducible subsets. Dimension of a quasi-projective
algebraic set is dened with respect to its Zariski topology. Obviously,
dimension of a single point is zero and dimension of A
1
k
is 1. Dimension
of by convention is 1.
For any commutative ring R one can also dene the (Krull) dimen-
sion as the sup of length of sequences of prime ideals P
l
P
0
.
When X is an ane algebraic set, then dimX = dimk[X] because
irreducible closed subsets of X correspond to prime ideals of k[X]. We
borrow the following theorem from commutative algebra and develop
the dimension theory upon it.
Theorem 4.1. Let R be a nitely generated k-algebra integral domain.
Then, dimR < +. Moreover, if 0 ,= a R is not a unit, then
dimR/a = dimR 1.
Theorem 4.2. Let X be a quasi-projective algebraic variety. Then,
(i) if X is ane and f k[X] such that ,= V
X
(f) ,= X, then
dimV
X
(f) = dimX 1;
(ii) if U X is a non-empty open subset, then dimX = dimU;
(iii) if X and f are as in (i), then every irreducible component of
V
X
(f) has dimension dimX 1.
Proof. (i) From the previous theorem we get
dimV
X
(f) = dimk[X]/

f = dimk[X]/f = dimk[X] 1
(ii) Clearly, dimU dimX. Cover X by open ane subsets U
i
.
Then, dimX = maxdimU
i
so by replacing X with some U
i
and
replacing U with U U
i
we can assume that X is ane.
We may assume that Y := X U ,= . Since X is irreducible,
dimY < dimX. Now let f be a polynomial such that V
X
(f) U ,=
and such that V
X
(f) does not contain any irreducible component of Y .
Hence,
dimV
X
(f) = dimX 1 = dimZ
for some irreducible component Z of V
X
(f) (see Exercise 5.2). This
implies that Z is not inside Y , otherwise Y should have dimension
more than dimX 1. Now by induction
dimU > dimZ U = dimZ = dimX 1
therefore dimU = dimX.
22 CAUCHER BIRKAR
(iii) Let Z be an irreducible component of V
X
(f). Let U be an ane
open subset of X which intersects Z but not the other components of
V
X
(f). Now f is also a regular function on U and V
U
(f) = Z U.
So, dimension of Z U is dimU 1 = dimX 1. Thus, dimZ =
dimX 1.
Example 4.3 We show that dimA
n
k
= dimP
n
k
= n. Let H A
n
k
be the ane algebraic variety dened by t
n
= 0. Then, dimH =
dimA
n
k
1. On the other hand, H A
n1
k
, so by induction dimA
n
k
= n.
Since A
n
k
is an open subset of P
n
k
, dimP
n
k
= n.
Theorem 4.4. Let F
1
, F
2
R = k[s
0
, s
1
, s
2
] be two non-constant ho-
mogeneous polynomials and let X
i
= V (F
i
) P
2
k
. Then, X
1
X
2
,= ,
i.e. any two projective curves in P
2
k
intersect.
Proof. We can assume that the F
i
are irreducible and that F
1
, = F
2
.
Let Z = V (F
1
, F
2
) A
3
k
. Note that since F
1
and F
2
are homogeneous,
Z contains the origin and so it is not empty. We prove that dimZ = 1.
Let R

= R/F
1
and R

= R

/F
2
. By previous theorems, dimR =
3, dimR

= 2 and nally dimR

= 1. On the other hand, R

R/F
1
, F
2
and dimR

= dimZ, so dimZ = 1.
Now any point of Z which is not the origin gives a point of X
1
X
2
,
therefore X
1
X
2
,= .
Theorem 4.5. Let X be a quasi-projective variety. Then, dimX is
the transcendence degree of k(X) over k.
Proof. If U ,= is an open subset, then k(U) = k(X) so we can assume
that X A
n
k
is an ane variety. Let d be the transcendence degree
of k(X) over k. Then, there are elements z
1
, . . . , z
d
k(X) which are
algebraically independent over k, and they generate a subeld L =
k(z
1
, . . . , z
d
) k(X) such that k(X) is a nite extension of L.
First suppose that L = k(X). Then, using the z
i
, we get a rational
map : X A
d
k
given by = (z
1
, . . . , z
d
). Since we can recover all
the variables t
i
on X, is birational. On the other hand, A
d
k
can be
considered as a hypersurface Y in A
d+1
k
.
Now assume that L ,= k(X) for any set z
1
, . . . , z
d
of algebraically
independent elements. Then by [5, Appendix 5, Proposition 1], we
can nd algebraically independent elements z
1
, . . . , z
d
and an element
z
d+1
k(X) L such that k(X) = L(z
d+1
) and k(X) is a separable ex-
tension of L. There is an irreducible polynomial f k[s
1
, . . . , s
d+1
] such
that f(z
1
, . . . , z
d+1
) = 0. Dene Y = V (f) in the ane space A
d+1
k
.
We can dene a rational map : X Y by = (z
1
, . . . , z
d+1
) which
induces an isomorphism k(Y ) k(X) over k hence is a birational
ALGEBRAIC GEOMETRY 23
isomorphism (see exercises to Section 3). So, again X is birational to
a hypersurface in A
d+1
k
.
Since X and Y are birational, they have isomorphic open subsets U
and V . Therefore, dimX = dimY . Finally, since Y is a hypersurface
in A
d+1
k
, its dimension is d.
In particular, we also proved the following
Corollary 4.6. Any quasi-projective algebraic variety of dimension d
is birational to an irreducible hypersurface in A
d+1
k
.
For a regular map : X Y , we dene the bre of over y Y to
be X
y
:=
1
y.
Theorem 4.7. Let : X Y be a regular map of quasi-projective
algebraic varieties which is dominant, i.e. (X) is dense in Y . Let
dimX = d and dimY = d

. Then,
(i) for any y (X), dimX
y
d d

,
(ii) there is a non-empty open set U Y such that if y U (X),
then dimX
y
= d d

,
(iii) if is surjective, then for each l,
P
l
= y Y [ dimX
y
l
is a closed subset of Y .
Proof. (i) Let y (X). By replacing Y and X with appropriate open
ane subsets, we can assume that X, Y are ane. Pick f k[Y ]
so that y V
Y
(f) ,= Y . By previous theorems, every irreducible
component of
1
V
Y
(f) = V
X
(f) is of dimension d 1 and X
y

V
X
(f). Choose an irreducible component S of V
X
(f) which contains
an irreducible component of X
y
with maximal dimension. Let T be the
closure of (S). By induction applied to S T, we see that
dimX
y
dimS dimT d 1 (d

1) = d d

(ii) As in (i), we can assume that X, Y are ane. Since is dom-

inant, the induced map

: k[Y ] k[X] is injective by Theorem ??.

Moreover, the induced map k(Y ) k(X) on the function elds is also
injective and dd

is the transcendence degree of k(X) over k(Y ). Let

f
1
, . . . , f
r
k[X] so that they generate k[X] as a k-algebra. We can
assume that f
1
, . . . , f
dd
are algebraically independent over k(Y ) and
each of the f
dd

+1
, . . . , f
r
is algebraically dependent over the eld gen-
erated by k(Y ) and f
1
, . . . , f
dd
. Thus, for each dd

< i r, there is
a polynomial
i
k[Y ][u
1
, . . . , u
dd
, u] such that
i
(f
1
, . . . , f
dd
, f
i
) =
0.
24 CAUCHER BIRKAR
Now

i
(f
1
, . . . , f
dd
, u) k[Y ][f
1
, . . . , f
dd
][u]
and we can assume that this is not zero. Let
i
be the set of coecients
of
i
(f
1
, . . . , f
dd
, u) (so by construction
i
k[Y ][f
1
, . . . , f
dd
]
k[X]).
Applying induction to the irreducible components of V
X
(
i
) shows
that there is an open non-empty set U Y such that if y U (X)
and if M is a component of X
y
having maximal dimension (among
the irreducible components of X
y
) then M is not inside V
X
(
i
) for
any i. Therefore, if g
i
= f
i
[
M
, then k[M] = k[g
1
, . . . , g
r
] and from
the equations
i
(g
1
, . . . , g
dd
, g
i
) = 0 we deduce that each of the
g
dd

+1
, . . . , g
r
is algebraically dependent over the eld generated by
k and the g
1
, . . . , g
dd
. Therefore, the transcendence degree of k(M)
over k is at most d d

hence dimX
y
= dimM d d

and from (i)

we get dimX
y
= d d

for any y U (X).

(iii) Assume surjective. From (i) and (ii) we see that P
dd
=
Y . Let U the maximal open set which satises (ii) and let T be an
irreducible component of Y U. Pick an irreducible component S of

1
T such that S T is dominant. Applying induction to all such
S T shows that dimX
y
dd

+1 for any y T. So, P

dd

+1
= Y U
is a closed subset. Continuing this type of argument and by applying
induction on dimension and induction on l one proves that each P
l
is
closed.
5. Exercises
Exercise 5.1 Let X be a non-empty Hausdor topological space (eg,
a metric space). Show that dimX = 0. So, our denition of dimension
does not suit other types of geometry.
5.1
Exercise 5.2 Let X be a quasi-projective algebraic set. Prove that if
Y X is a closed subset of X, then dimY dimX. Prove that if X =
X
1
X
l
where X
i
are closed subsets, then dimX = maxdimX
i
.
Exercise 5.3 Let X be a quasi-projective algebraic set and X =

U
i
an open cover. Show that dimX = maxdimU
i
.
Exercise 5.4 Let : P
n
k
A
m
k
be a regular map. Prove that is
constant, that is, (P
n
k
) is a single point.
Exercise 5.5 Let : P
n
k
P
m
k
be a regular map where n > m. Prove
that is constant.
5.1
This was provided by Tom Sutton.
ALGEBRAIC GEOMETRY 25
References
[1] Atiyah, Macdonald; Introduction to commutative algebra.
[2] Harris; Algebraic geometry, a rst course. Springer 1992.
[3] Hartshorne; Algebraic geometry.
[4] Matsumura; Commutative ring theory.
[5] Shafarevich; Basic algebraic geometry I.
[6] Zariski-Samuel; Commutative algebra. Volume 1.