Chapter 5

Varieties

This is a chapter from version 2.0 of the book “Mathematics of Public Key Cryptogra-
phy” by Steven Galbraith, available from http://www.math.auckland.ac.nz/˜sgal018/crypto-
book/crypto-book.html The copyright for this chapter is held by Steven Galbraith.
This book was published by Cambridge University Press in early 2012. This is the
extended and corrected version. Some of the Theorem/Lemma/Exercise numbers may be
different in the published version.
Please send an email to S.Galbraith@math.auckland.ac.nz if you find any mistakes.

The purpose of this chapter is to state some basic definitions and results from algebraic
geometry that are required for the main part of the book. In particular, we define algebraic
sets, irreducibility, function fields, rational maps and dimension. The chapter is not
intended as a self-contained introduction to algebraic geometry. We make the following
recommendations to the reader:

1. Readers who want a very elementary introduction to elliptic curves are advised
to consult one or more of Koblitz [348], Silverman-Tate [567], Washington [626],
Smart [572] or Stinson [592].

2. Readers who wish to learn algebraic geometry properly should first read a basic
text such as Reid [497] or Fulton [216]. They can then skim this chapter and
consult Stichtenoth [589], Moreno [439], Hartshorne [278], Lorenzini [394] or Sha-
farevich [543] for detailed proofs and discussion.

5.1 Affine algebraic sets

Let k be a perfect field contained in a fixed algebraic closure k. All algebraic extensions
k′ /k are implicitly assumed to be subfields of k. We use the notation k[x] = k[x1 , . . . , xn ]
(in later sections we also use k[x] = k[x0 , . . . , xn ]). When n = 2 or 3 we often write k[x, y]
or k[x, y, z].
Define affine n-space over k as An (k) = kn . We call A1 (k) the affine line and A2 (k)
the affine plane over k. If k ⊆ k′ then we have the natural inclusion An (k) ⊆ An (k′ ).
We write An for An (k) and so An (k) ⊆ An .

Definition 5.1.1. Let S ⊆ k[x]. Define

V (S) = {P ∈ An (k) : f (P ) = 0 for all f ∈ S}.

87
88 CHAPTER 5. VARIETIES

If S = {f1 , . . . , fm } then we write V (f1 , . . . , fm ) for V (S). An affine algebraic set is a

set X = V (S) ⊂ An where S ⊂ k[x].
Let k′ /k be an algebraic extension. The k′ -rational points of X = V (S) are

X(k′ ) = X ∩ An (k′ ) = {P ∈ An (k′ ) : f (P ) = 0 for all f ∈ S}.

An algebraic set V (f ), where f ∈ k[x], is a hypersurface. If f (x) is a polynomial of

total degree 1 then V (f ) is a hyperplane.
Informally we often write “the algebraic set f = 0” instead of V (f ). For example,
y 2 = x3 instead of V (y 2 − x3 ). We stress that, as is standard, V (S) is the set of solutions
over an algebraically closed field.
When an algebraic set is defined as the vanishing of a set of polynomials with coeffi-
cients in k then it is called a k-algebraic set. The phrase “defined over k” has a different
meaning and the relation between them will be explained in Remark 5.3.7.
Example 5.1.2. If X = V (x21 + x22 + 1) ⊆ A2 over Q then X(Q) = ∅. Let k = F2 and
let X = V (y 8 + x6 y + x3 + 1) ⊆ A2 . Then X(F2 ) = {(0, 1), (1, 0), (1, 1)}.
√
Exercise 5.1.3. Let k be a field. Show that {(t, t2 ) : t ∈ k} ⊆ A2 , {(t, ± t) : t ∈ k} ⊆ A2
and {(t2 + 1, t3 ) : t ∈ k} ⊆ A2 are affine algebraic sets.
Exercise 5.1.4. Let f (x, y) ∈ k[x, y] be a non-constant polynomial. Prove that V (f (x, y)) ⊂
A2 is an infinite set.
Example 5.1.5. Let k be a field. There is a one-to-one correspondence between the
set k∗ and the k-rational points X(k) of the affine algebraic set X = V (xy − 1) ⊂ A2 .
Multiplication in the field k corresponds to the function mult : X × X → X given
by mult((x1 , y1 ), (x2 , y2 )) = (x1 x2 , y1 y2 ). Similarly, inversion in k∗ corresponds to the
function inverse(x, y) = (y, x). Hence we have represented k∗ as an algebraic group,
which we call Gm (k).
Example 5.1.6. Another elementary example of an algebraic group is the affine algebraic
set X = V (x2 + y 2 − 1) ⊂ A2 with the group operation mult((x1 , y1 ), (x2 , y2 )) = (x1 x2 −
y1 y2 , x1 y2 + x2 y1 ). (These formulae are analogous to the angle addition rules for sine and
cosine as, over R, one can identify (x, y) with (cos(θ), sin(θ)).) The reader should verify
that the image of mult is contained in X. The identity element is (1, 0) and the inverse
of (x, y) is (x, −y). One can verify that the axioms of a group are satisfied. This group is
sometimes called the circle group.
Exercise 5.1.7. Let p ≡ 3 (mod 4) be prime and define Fp2 = Fp (i) where i2 = −1.
Show that the group X(Fp ), where X is the circle group from Example 5.1.6, is isomorphic
as a group to the subgroup G ⊆ F∗p2 of order p + 1.

Proposition 5.1.8. Let S ⊆ k[x1 , . . . , xn ].

1. V (S) = V ((S)) where (S) is the k[x]-ideal generated by S.
2. V (k[x]) = ∅ and V ({0}) = An where ∅ denotes the empty set.
3. If S1 ⊆ S2 then V (S2 ) ⊆ V (S1 ).
4. V (f g) = V (f ) ∪ V (g).
5. V (f ) ∩ V (g) = V (f, g).
Exercise 5.1.9. Prove Proposition 5.1.8.
5.1. AFFINE ALGEBRAIC SETS 89

Exercise 5.1.10. Show that V (S)(k) = An (k) does not necessarily imply that S = {0}.
The following result assumes a knowledge of Galois theory. See Section A.7 for back-
ground.
Lemma 5.1.11. Let X = V (S) be an algebraic set with S ⊆ k[x] (i.e., X is a k-algebraic
set). Let k′ be an algebraic extension of k. Let σ ∈ Gal(k/k′ ). For P = (P1 , . . . , Pn )
define σ(P ) = (σ(P1 ), . . . , σ(Pn )).
1. If P ∈ X(k) then σ(P ) ∈ X(k).
2. X(k′ ) = {P ∈ X(k) : σ(P ) = P for all σ ∈ Gal(k/k′ )}.
Exercise 5.1.12. Prove Lemma 5.1.11.
Definition 5.1.13. The ideal over k of a set X ⊆ An (k) is

Ik (X) = {f ∈ k[x] : f (P ) = 0 for all P ∈ X(k)}.

We define I(X) = Ik (X).1

An algebraic set X is defined over k (sometimes abbreviated to “X over k”) if I(X)
can be generated by elements of k[x].
Note that if X is an algebraic set defined over k then X is a k-algebraic set. Perhaps
surprisingly, it is not necessarily true that an algebraic set described by polynomials
defined over k is an algebraic set defined over k. In Remark 5.3.7 we will explain that
these concepts are equivalent for the objects of interest in this book.
Exercise 5.1.14. Show that Ik (X) = I(X) ∩ k[x].
The following example shows that Ik (X) is not necessarily the same as Ik (X(k)).
Example 5.1.15. Let X = V (x2 + y 2 ) ⊂ A2 be an algebraic set over R. Then X(R) =
{(0, 0)} while X(C) = {(x, ±ix) : x ∈ C}. One has IR (X) = (x2 + y 2 ) where this
denotes an R[x, y]-ideal. Similarly, IC (X) is the C[x, y]-ideal (x2 + y 2 ). Indeed, IC (X) =
IR (X) ⊗R C. On the other hand, IR (X(R)) is the R[x, y]-ideal (x, y).
Proposition 5.1.16. Let X, Y ⊆ An be sets and J a k[x]-ideal. Then
1. Ik (X) is a k[x]-ideal.
2. X ⊆ V (Ik (X)).
3. If X ⊆ Y then Ik (Y ) ⊆ Ik (X).
4. Ik (X ∪ Y ) = Ik (X) ∩ Ik (Y ).
5. If X is an algebraic set defined over k then V (Ik (X)) = X.
6. If X and Y are algebraic sets defined over k and Ik (X) = Ik (Y ) then X = Y .
7. J ⊆ Ik (V (J)).
8. Ik (∅) = k[x].
Exercise 5.1.17. Prove Proposition 5.1.16.
Definition 5.1.18. The affine coordinate ring over k of an affine algebraic set X ⊆ An
defined over k is
k[X] = k[x1 , . . . , xn ]/Ik (X).
1 The notation I (X) is not standard (Silverman [564] calls it I(X/k)), but the notation I(X) agrees
k
with many elementary books on algebraic geometry, since they work over an algebraically closed field.
90 CHAPTER 5. VARIETIES

Warning: Here k[X] does not denote polynomials in the variable X. Hartshorne and
Fulton write A(X) and Γ(X) respectively for the affine coordinate ring.
Exercise 5.1.19. Prove that k[X] is a commutative ring with an identity.
Note that k[X] is isomorphic to the ring of all functions f : X → k given by polyno-
mials defined over k.
Hilbert’s Nullstellensatz is a powerful tool for understanding Ik (X) and it has several
other applications (for example, we use it in Section 7.5). We follow the presentation of
Fulton [216]. Note that it is necessary to work over k.
Theorem 5.1.20. (Weak Nullstellensatz) Let X ⊆ An be an affine algebraic set defined
over k and let m be a maximal ideal of the affine coordinate ring k[X]. Then V (m) = {P }
for some P = (P1 , . . . , Pn ) ∈ X(k) and m = (x1 − P1 , . . . , xn − Pn ).
Proof: Consider the field F = k[X]/m, which contains k. Note that F is finitely
generated as a ring over k by the images of x1 , . . . , xn . By Theorem A.6.2, F is an
algebraic extension of k and so F = k.
It follows that, for 1 ≤ i ≤ n, there is some Pi ∈ k such that xi − Pi ∈ m. Hence,
n = (x1 − P1 , . . . , xn − Pn ) ⊆ m and, since k[X]/n = k it follows that m = n.
Finally, it is clear that P ∈ V (m) and if Q = (Q1 , . . . , Qn ) ∈ X(k) − {P } then there
is some 1 ≤ i ≤ n such that Qi 6= Pi and so (x − Pi )(Qi ) 6= 0. Hence V (m) = {P }. 
Corollary 5.1.21. If I is a proper ideal in k[x1 , . . . , xn ] then V (I) 6= ∅.
Proof: There is some maximal ideal m such that I ⊆ m. By Theorem 5.1.20, m =
(x1 − P1 , . . . , xn − Pn ) for some P = (P1 , . . . , Pn ) ∈ An (k) and so P ∈ V (I). 
Indeed, Corollary 5.1.21 is true when one starts with I a proper ideal in k[x1 , . . . , xn ];
see Lemma VIII.7.3 of [301].
We can now state the Hilbert Nullstellensatz. This form of the theorem (which
applies to Ik (V (I)) where k is not necessarily algebraically closed), appears as Proposition
VIII.7.4 of [301].
Theorem 5.1.22. Let I be an ideal in R = k[x1 , . . . , xn ]. Then Ik (V (I)) = radR (I) (see
Section A.9 for the definition of the radical ideal).
Proof: One has radR (I) ⊆ Ik (V (I)) since f n ∈ I implies f n (P ) = 0 for all P ∈ V (I)
and so f (P ) = 0 for all P ∈ V (I). For the converse suppose I = (F1 (x1 , . . . , xn ), . . . , Fm (x1 , . . . , xn ))
and G(x1 , . . . , xn ) ∈ Ik (V (I)). Define the k[x1 , . . . , xn+1 ]-ideal
J = (F1 (x1 , . . . , xn ), . . . , Fm (x1 , . . . , xn ), xn+1 G(x1 , . . . , xn ) − 1).
Then V (J) = ∅ since if P = (P1 , . . . , Pn+1 ) ∈ An+1 (k) is such that Fi (P1 , . . . , Pn ) = 0
for all 1 ≤ i ≤ m then G(P1 , . . . , Pn ) = 0 too and so one does not have Pn+1 G(P ) = 1. It
follows from (the stronger form of) Corollary 5.1.21 that J = (1) and so 1 ∈ J. In other
words, there are polynomials ai (x1 , . . . , xn+1 ) ∈ k[x1 , . . . , xn+1 ] for 1 ≤ i ≤ m + 1 such
that m
X
1 = am+1 (xn+1 G − 1) + ai Fi .
i=1
Write y = 1/xn+1 and let N = 1 + max1≤i≤m+1 {degxn+1 (ai )}. One has
m
X
y N = bm+1 (x1 , . . . , xn , y)(G − y) + bi (x1 , . . . , xn , y)Fi (x1 , . . . , xn )
i=1

for some polynomials bi ∈ k[x1 , . . . , xn , y]. Setting y = G proves that GN ∈ I and so

G ∈ radR (I). 
5.2. PROJECTIVE ALGEBRAIC SETS 91

Corollary 5.1.23. Let f (x, y) ∈ k[x, y] be irreducible over k and let X = V (f (x, y)) ⊂
A2 (k). Then I(X) = (f (x, y)), i.e., the ideal over k[x, y] generated by f (x, y).

Proof: By Theorem 5.1.22 we have I(X) = radk ((f (x, y))). Since k[x, y] is a unique
factorisation domain and f (x, y) is irreducible, then f (x, y) is prime. So g(x, y) ∈
radk ((f (x, y))) implies g(x, y)n = f (x, y)h(x, y) for some h(x, y) ∈ k[x, y] which implies
f (x, y) | g(x, y) and g(x, y) ∈ (f (x, y)). 

Theorem 5.1.24. Every affine algebraic set X is the intersection of a finite number of
hypersurfaces.

Proof: By Hilbert’s basis theorem (Corollary A.9.4) k[x] is Noetherian. Hence Ik (X) =
(f1 , . . . , fm ) and X = V (f1 ) ∩ · · · ∩ V (fm ). 

5.2 Projective Algebraic Sets

Studying affine algebraic sets is not sufficient for our applications. In particular, the set
of affine points of the Weierstrass equation of an elliptic curve (see Section 7.2) does not
form a group. Projective geometry is a way to “complete” the picture by adding certain
“points at infinity”.
For example, consider the hyperbola xy = 1 in A2 (R). Projective geometry allows an
interpretation of the behaviour of the curve at x = 0 or y = 0; see Example 5.2.7.

Definition 5.2.1. Projective space over k of dimension n is

Pn (k) = {lines through (0, . . . , 0) in An+1 (k)}.

A convenient way to represent points of Pn (k) is using homogeneous coordinates: Let

a0 , a1 , . . . , an ∈ k with not all aj = 0 and define (a0 : a1 : · · · : an ) to be the equivalence
class of (n + 1)-tuples under the equivalence relation

(a0 , a1 , · · · , an ) ≡ (λa0 , λa1 , · · · , λan )

for any λ ∈ k∗ . Thus Pn (k) = {(a0 : · · · : an ) : ai ∈ k for 0 ≤ i ≤ n and ai 6=

0 for some 0 ≤ i ≤ n}. Write Pn = Pn (k).

In other words, the equivalence class (a0 : · · · : an ) is the set of points on the line
between (0, . . . , 0) and (a0 , . . . , an ) with the point (0, . . . , 0) removed.
There is a map ϕ : An → Pn given by ϕ(x1 , . . . , xn ) = (x1 : · · · : xn : 1). Hence An is
identified with a subset of Pn .

Example 5.2.2. The projective line P1 (k) is in one-to-one correspondence with A1 (k)∪
{∞} since P1 (k) = {(a0 : 1) : a0 ∈ k} ∪ {(1 : 0)}. The projective plane P2 (k) is in
one-to-one correspondence with A2 (k) ∪ P1 (k).

Definition 5.2.3. A point P = (P0 : P1 : · · · : Pn ) ∈ Pn (k) is defined over k if there

∗
is some λ ∈ k such that λPj ∈ k for all 0 ≤ j ≤ n. If P ∈ Pn and σ ∈ Gal(k/k) then
σ(P ) = (σ(P0 ) : · · · : σ(Pn )).

Exercise 5.2.4. Show that P is defined over k if and only if there is some 0 ≤ i ≤ n such
that Pi 6= 0 and Pj /Pi ∈ k for all 0 ≤ j ≤ n. Show that Pn (k) is equal to the set of points
P ∈ Pn (k) that are defined over k. Show that σ(P ) in Definition 5.2.3 is well-defined
(i.e., if P = (P0 , . . . , Pn ) ≡ P ′ = (P0′ , . . . , Pn′ ) then σ(P ) ≡ σ(P ′ )).
92 CHAPTER 5. VARIETIES

Lemma 5.2.5. A point P ∈ Pn (k) is defined over k if and only if σ(P ) = P for all
σ ∈ Gal(k/k).
Proof: Let P = (P0 : · · · : Pn ) ∈ Pn (k) and suppose σ(P ) ≡ P for all σ ∈ Gal(k/k).
∗
Then there is some ξ : Gal(k/k) → k such that σ(Pi ) = ξ(σ)Pi for all 0 ≤ i ≤ n.
∗
One can verify2 that ξ is a 1-cocycle in k . It follows by Theorem A.7.2 (Hilbert 90)
∗
that ξ(σ) = σ(γ)/γ for some γ ∈ k . Hence, σ(Pi /γ) = Pi /γ for all 0 ≤ i ≤ n and all
σ ∈ Gal(k/k). Hence Pi /γ ∈ k for all 0 ≤ i ≤ n and the proof is complete. 
Recall that if f is a homogeneous polynomial of degree d then f (λx0 , . . . , λxn ) =
λd f (x0 , . . . , xn ) for all λ ∈ k and all (x0 , . . . , xn ) ∈ An+1 (k).
Definition 5.2.6. Let f ∈ k[x0 , . . . , xn ] be a homogeneous polynomial. A point P =
(x0 : · · · : xn ) ∈ Pn (k) is a zero of f if f (x0 , . . . , xn ) = 0 for some (hence, every) point
(x0 , . . . , xn ) in the equivalence class (x0 : · · · : xn ). We therefore write f (P ) = 0. Let S
be a set of polynomials and define

V (S) = {P ∈ Pn (k) : P is a zero of f (x) for all homogeneous f (x) ∈ S}.

A projective algebraic set is a set X = V (S) ⊆ Pn (k) for some S ⊆ k[x]. Such a set
is also called a projective k-algebraic set. For X = V (S) and k′ an algebraic extension of
k define

X(k′ ) = {P ∈ Pn (k′ ) : f (P ) = 0 for all homogeneous f (x) ∈ S}.

Example 5.2.7. The hyperbola y = 1/x can be described as the affine algebraic set
X = V (xy − 1) ⊂ A2 over R. One can consider the corresponding projective algebraic set
V (xy − z 2 ) ⊆ P2 over R whose points consist the points of X together with the points
(1 : 0 : 0) and (0 : 1 : 0). These two points correspond to the asymptotes x = 0 and y = 0
of the hyperbola and they essentially “tie together” the disconnected components of the
affine curve to make a single closed curve in projective space.
Exercise 5.2.8. Describe the sets V (x2 +y 2 −z 2 )(R) ⊂ P2 (R) and V (yz−x2 )(R) ⊆ P2 (R).
Exercise 5.2.9. What is V (xz + y 2 , xyz) ⊆ P2 (C)?
Exercise 5.2.10. What is V (y 2 + x2 , x2 z − y 3 ) ⊆ P2 (C)?
A set of homogeneous polynomials does not in general form an ideal as one cannot
simultaneously have closure under multiplication and addition. Hence it is necessary to
introduce the following definition.
Definition 5.2.11. A k[x0 , . . . , xn ]-ideal I ⊆ k[x0 , . . .P
, xn ] is a homogeneous ideal if
for every f ∈ I with homogeneous decomposition f = i fi we have fi ∈ I.
Exercise 5.2.12. Let S ⊂ k[x] be a set of homogeneous polynomials. Pn Define (S) to
be the k[x]-ideal generated by S in the usual way, i.e., (S) = { j=1 fj (x)sj (x) : n ∈
N, fj (x) ∈ k[x0 , . . . , xn ], sj (x) ∈ S}. Prove that (S) is a homogeneous ideal. Prove that
if I is a homogeneous ideal then I = (S) for some set of homogeneous polynomials S.
Definition 5.2.13. For any set X ⊆ Pn (k) define


Ik (X) = {f ∈ k[x0 , . . . , xn ] : f is homogeneous and f (P ) = 0 for all P ∈ X} .
2 At least, one can verify the formula ξ(στ ) = σ(ξ(τ ))ξ(σ). The topological condition also holds, but

we do not discuss this.

5.2. PROJECTIVE ALGEBRAIC SETS 93

We stress that Ik (X) is not the stated set of homogeneous polynomials but the ideal
generated by them. We write I(X) = Ik (X).
An algebraic set X ⊆ Pn is defined over k if I(X) can be generated by homogeneous
polynomials in k[x].
Proposition 5.2.14. Let k be a field.
1. If S1 ⊆ S2 ⊆ k[x0 , . . . , xn ] then V (S2 ) ⊆ V (S1 ) ⊆ Pn (k).
2. If f g is a homogeneous polynomial then V (f g) = V (f )∪V (g) (recall from Lemma A.5.4
that f and g are both homogeneous).
3. V (f ) ∩ V (g) = V (f, g).
4. If X1 ⊆ X2 ⊆ Pn (k) then Ik (X2 ) ⊆ Ik (X1 ) ⊆ k[x0 , . . . , xn ].
5. Ik (X1 ∪ X2 ) = Ik (X1 ) ∩ Ik (X2 ).
6. If J is a homogeneous ideal then J ⊆ Ik (V (J)).
7. If X is a projective algebraic set defined over k then V (Ik (X)) = X. If Y is another
projective algebraic set defined over k and Ik (Y ) = Ik (X) then Y = X.
Exercise 5.2.15. Prove Proposition 5.2.14.
Definition 5.2.16. If X is a projective algebraic set defined over k then the homoge-
nous coordinate ring of X over k is k[X] = k[x0 , . . . , xn ]/Ik (X).
Note that elements of k[X] are not necessarily homogeneous polynomials.
Definition 5.2.17. Let X be an algebraic set in An (respectively, Pn ) The Zariski
topology is the topology on X defined as follows: The closed sets are X ∩ Y for every
algebraic set Y ⊆ An (respectively, Y ⊆ Pn ).
Exercise 5.2.18. Show that the Zariski topology satisfies the axioms of a topology.
Definition 5.2.19. For 0 ≤ i ≤ n define Ui = {(x0 : · · · : xn ) ∈ Pn : xi 6= 0} =
Pn − V (xi ). (These are open sets in the Zariski topology.)
Exercise 5.2.20. Show that Pn = ∪ni=0 Ui (not a disjoint union).
Exercise 5.2.21. What points of P2 (k) do not lie in two of the three sets U0 (k), U1 (k), U2 (k)?
Definition. Let L ∈ GLn+1 (k) (i.e., L is an (n + 1) × (n + 1) matrix over k that is
invertible). The map L : Pn → Pn given by

L(x0 : · · · : xn ) = (L0,0 x0 + · · · + L0,n xn : · · · : Ln,0 x0 + · · · + Ln,n xn )

is called a linear change of variables on Pn over k. The inverse change of variables is

given by L−1 .
Example 5.2.22. The matrix
 
1 −1 0
L= 0 1 0 
0 0 1

gives a linear change of variables L : P2 → P2 of the form L(x0 : x1 : x2 ) = (y0 : y1 :

y2 ) = (x0 − x1 : x1 : x2 ). This maps the algebraic set X = V (x20 − x21 + x1 x2 ) to
Y = V (y02 + 2y0 y1 + y1 y2 ). In other words, if P ∈ X(k) then L(P ) ∈ Y (k).
94 CHAPTER 5. VARIETIES

A linear change of variable does not change the underlying geometry of an algebraic
set, but can be useful for practical computation. For instance, sometimes we will use
Exercise 5.2.23 to reduce any pair of points to affine space without changing the “shape”
of the algebraic set.

Exercise 5.2.23. Show that if P, Q ∈ Pn (k) then there is always a linear change of
variables L on Pn over k such that L(P ), L(Q) ∈ Un .

We already mentioned the map ϕ : An → Pn given by ϕ(x1 , . . . , xn ) = (x1 : · · · : xn :

1), which has image equal to Un . A useful way to study a projective algebraic set X is
to consider X ∩ Ui for 0 ≤ i ≤ n and interpret X ∩ Ui as an affine algebraic set. We now
introduce the notation for this.

Definition 5.2.24. Let ϕi : An (k) → Ui be the one-to-one correspondence

ϕi (y1 , . . . , yn ) = (y1 : · · · : yi : 1 : yi+1 : · · · : yn ).

We write ϕ for ϕn . Let

ϕ−1
i (x0 : · · · : xn ) = (x0 /xi , . . . , xi−1 /xi , xi+1 /xi , . . . , xn /xi ).

be the map ϕ−1 i : Pn (k) → An (k), which is defined only on Ui (i.e., ϕ−1 −1
i (X) = ϕi (X ∩
3
Ui )).
We write X ∩ An as an abbreviation for ϕ−1 n (X ∩ Un ).
Let ϕ∗i : k[x0 , . . . , xn ] → k[y1 , . . . , yn ] be the de-homogenisation map4

ϕ∗i (f )(y1 , . . . , yn ) = f ◦ ϕi (y1 , . . . , yn ) = f (y1 , . . . , yi , 1, yi+1 , . . . , yn ).

We write ϕ∗ for ϕ∗n .

Let ϕ−1∗
i : k[y1 , . . . , yn ] → k[x0 , . . . , xn ] be the homogenisation
deg(f )
ϕ−1∗
i (f )(x0 , . . . , xn ) = xi f (x0 /xi , . . . , xi−1 /xi , xi+1 /xi , . . . , xn /xi )

where deg(f ) is the total degree.

We write f as an abbreviation for ϕ−1∗
n (f ). For notational simplicity we often consider
polynomials f (x, y); in this case we define f = z deg(f ) f (x/z, y/z).
We now state some elementary relations between projective algebraic sets X and their
affine parts X ∩ Ui .
Lemma 5.2.25. Let the notation be as above.
1. ϕ∗i : k[x0 , . . . , xn ] → k[y1 , . . . , yn ] is a k-algebra homomorphism. The map ϕ−1∗ i :
k[y1 , . . . , yn ] → k[x0 , . . . , xn ] satisfies most of the properties of a k-algebra homo-
morphism, except that ϕ−1∗ i (f + g) = ϕ−1∗i (f ) + ϕ−1∗
i (g) if and only if f and g have
the same total degree.
2. Let P = (P0 : · · · : Pn ) ∈ Pn (k) with Pi 6= 0 and let f ∈ k[x0 , . . . , xn ] be homoge-
neous. Then f (P ) = 0 implies ϕ∗i (f )(ϕ−1
i (P )) = 0.
3. Let f ∈ k[x0 , . . . , xn ] be homogeneous. Then ϕ−1 ∗
i (V (f )) = V (ϕi (f )). In particular,
n
V (f ) ∩ A = V (f ◦ ϕ).
3 This notation does not seem to be standard. Our notation agrees with Silverman [564], but
Hartshorne [278] has ϕi and ϕ−1
i the other way around.
4 The upper star notation is extended in Definition 5.5.23.
5.2. PROJECTIVE ALGEBRAIC SETS 95

4. Let X ⊆ Pn (k). Then f ∈ Ik (X) implies ϕ∗i (f ) ∈ Ik (ϕ−1

i (X)). In particular,
f ∈ Ik (X) implies f ◦ ϕ ∈ Ik (X ∩ An ).
5. If P ∈ An (k) and f ∈ k[y1 , . . . , yn ] then f (P ) = 0 implies ϕ−1∗
i (f )(ϕi (P )) = 0. In
particular, f (P ) = 0 implies f (ϕ(P )) = 0.
6. For homogeneous f ∈ k[x0 , . . . , xn ] then ϕ−1∗
i (ϕ∗i (f )) | f . Furthermore, if f has a
−1∗
monomial that does not include xi then ϕi (ϕ∗i (f )) = f (in particular, f ◦ ϕ = f ).
Exercise 5.2.26. Prove Lemma 5.2.25.
Definition 5.2.27. Let I ⊆ k[y1 , . . . , yn ]. Define the homogenisation I to be the
k[x0 , . . . , xn ]-ideal generated by the set {f (x0 , . . . , xn ) : f ∈ I}.
Exercise 5.2.28. Let I ⊆ k[y1 , . . . , yn ]. Show that I is a homogeneous ideal.
n
 5.2.29. Let X ⊆ A (k). Define the projective closure of X to be X =
Definition

V I(X) ⊆ Pn .

Lemma 5.2.30. Let the notation be as above.

1. Let X ⊆ An , then ϕ(X) ⊆ X and X ∩ An = X.


2. Let X ⊆ An (k) be non-empty. Then Ik X = Ik (X).
Proof: Part 1 follows directly from the definitions.
Part 2 is essentially that the homogenisation of a radical ideal is a radical ideal, we
give a direct proof. Let f ∈ k[x0 , . . . , xn ] be such that f is homogeneous and f (X) = 0.
Write f = xd0 g where g ∈ k[x0 , . . . , xn ] has a monomial that does not include x0 . By
part 1 and X 6= ∅, g is not constant. Then g ◦ ϕ ∈ Ik (X) and so g = g ◦ ϕ ∈ Ik (X).
It follows from part 6 of Lemma 5.2.25 that f ∈ Ik (X). Hence, Ik (X) ⊆ Ik (X) and the
result follows. 
Theorem 5.2.31. Let f (x0 , x1 , x2 ) ∈ k[x0 , x1 , x2 ] be a k-irreducible homogeneous poly-
nomial. Let
X = V (f (x0 , x1 , x2 )) ⊆ P2 .
Then Ik (X) = (f (x0 , x1 , x2 )).
Proof: Let 0 ≤ i ≤ 2 be such that f (x0 , x1 , x2 ) has a monomial that does not feature xi
(such an i must exist since f is irreducible). Without loss of generality, suppose i = 2.
Write g(y1 , y2 ) = ϕ∗ (f ) = f (y1 , y2 , 1). By part 6 of Lemma 5.2.25 the homogenisation of
g is f .
Let Y = X ∩A2 = V (g). Note that g is k-irreducible (since g = g1 g2 implies, by taking
homogenisation, f = g1 g2 ). Let h ∈ Ik (X) then h◦ϕ ∈ Ik (Y ) and so, by Corollary 5.1.23,
h ◦ ϕ ∈ (g). In other words, there is some h1 (y1 , y2 ) such that h ◦ ϕ = gh1 . Taking
homogenisations gives f h1 | h and so h ∈ (f ). 
Corollary 5.2.32. Let f (x, y) ∈ k[x, y] be a k-irreducible polynomial and let X = V (f ) ⊆
A2 . Then X = V (f ) ⊆ P2 .
Exercise 5.2.33. Prove Corollary 5.2.32.
Example 5.2.34. The projective closure of V (y 2 = x3 + Ax + B) ⊆ An is V (y 2 z =
x3 + Axz 2 + Bz 3 ).
Exercise 5.2.35. Let X = V (f (x0 , x1 )) ⊆ A2 and let X ⊆ P2 be the projective closure
of X. Show that X − X is finite (in other words, there are only finitely many points at
infinity).
96 CHAPTER 5. VARIETIES

A generalisation of projective space, called weighted projective space, is defined as

follows: For i0 , . . . , in ∈ N denote by (a0 : a1 : · · · : an ) the equivalence class of elements
in kn+1 under the equivalence relation

(a0 , a1 , · · · , an ) ≡ (λi0 a0 , λi1 a1 , · · · , λin an )

for any λ ∈ k∗ . The set of equivalence classes is denoted P(i0 , . . . , in )(k). For example, it
makes sense to consider the curve y 2 = x4 + ax2 z 2 + z 4 as lying in P(1, 2, 1). We will not
discuss this topic further in the book (we refer to Reid [498] for details), but it should
be noted that certain coordinate systems used for efficient elliptic curve cryptography
naturally live in weighted projective space.

5.3 Irreducibility
We have seen that V (f g) decomposes as V (f ) ∪ V (g) and it is natural to consider V (f )
and V (g) as being ‘components’ of V (f g). It is easier to deal with algebraic sets that
cannot be decomposed in this way. This concept is most useful when working over an
algebraically closed field, but we give some of the theory in greater generality.
Definition 5.3.1. An affine k-algebraic set X ⊆ An is k-reducible if X = X1 ∪ X2
with X1 and X2 being k-algebraic sets and Xi 6= X for i = 1, 2. An affine algebraic set is
k-irreducible if there is no such decomposition. An affine algebraic set is geometrically
irreducible if X is k-irreducible. An affine variety over k is a geometrically irreducible
k-algebraic set defined over k.
A projective k-algebraic set X ⊆ Pn is k-irreducible (resp. geometrically irre-
ducible) if X is not the union X1 ∪ X2 of projective k-algebraic sets X1 , X2 ⊆ Pn
(respectively, projective k-algebraic sets) such that Xi 6= X for i = 1, 2. A projective
variety over k is a geometrically irreducible projective k-algebraic set defined over k.
Let X be a variety (affine or projective). A subvariety of X over k is a subset Y ⊆ X
that is a variety (affine or projective) defined over k.
This definition matches the usual topological definition of a set being irreducible if it
is not a union of proper closed subsets.
Example 5.3.2. The algebraic set X = V (x2 + y 2 ) ⊆ A2 over R is R-irreducible. How-
ever, over C we have X = V (x + iy) ∪ V (x − iy) and so X is C-reducible.
Exercise 5.3.3. Show that X = V (wx − yz, x2 − yz) ⊆ P3 is not irreducible.
It is often easy to determine that a reducible algebraic set is reducible, just by exhibit-
ing the non-trivial union. However, it is not necessarily easy to show that an irreducible
algebraic set is irreducible. We now give an algebraic criterion for irreducibility and some
applications of this result.
Theorem 5.3.4. Let X be an algebraic set (affine or projective). Then X is k-irreducible
if and only if Ik (X) is a prime ideal.
Proof: (⇒): Suppose X = V (S) where S ⊆ k[x] is k-irreducible and that there are
elements f, g ∈ k[x] such that f g ∈ Ik (X). Then X ⊆ V (f g) = V (f ) ∪ V (g), so X =
(X ∩V (f ))∪(X ∩V (g)). Since X ∩V (f ) = V (S, f ) and X ∩V (g) = V (S, g) are k-algebraic
sets it follows that either X = X ∩V (f ) or X = X ∩V (g), and so f ∈ Ik (X) or g ∈ Ik (X).
(⇐): Suppose I = Ik (X) is a prime ideal and that X = X1 ∪ X2 where X1 and X2 are
k-algebraic sets. Let I1 = Ik (X1 ) and I2 = Ik (X2 ). By parts 3 and 4 of Proposition 5.1.16
5.3. IRREDUCIBILITY 97

or parts 4 and 5 of Proposition 5.2.14 we have I ⊆ I1 , I ⊆ I2 and I = I1 ∩ I2 . Since

I1 I2 ⊆ I1 ∩ I2 = I and I is a prime ideal, it follows that either I1 ⊆ I or I2 ⊆ I.
Hence either I = I1 or I = I2 and so, by part 6 of Proposition 5.1.16 or part 7 of
Proposition 5.2.14, X = X1 or X = X2 . 
Exercise 5.3.5. Show that V (y − x2 ) is irreducible in A2 (k).
Exercise 5.3.6. Let X ⊂ An be an algebraic set over k. Suppose there exist polynomials
f1 , . . . , fn ∈ k[t] such that X = {(f1 (t), f2 (t), . . . , fn (t)) : t ∈ k}. Prove that X is
geometrically irreducible.
Remark 5.3.7. A k-algebraic set X is the vanishing of polynomials in k[x1 , . . . , xn ].
However, we say X is defined over k if Ik (X) is generated by polynomials in k[x1 , . . . , xn ].
Hence, it is clear that an algebraic set defined over k is a k-algebraic set. The converse
does not hold in general. However, if X is absolutely irreducible and k is a perfect field
then these notions are equivalent (see Corollary 10.2.2 of Fried and Jarden [214] and use
the fact that when X is absolutely irreducible then the algebraic closure of k in k(X) is
k). Note that Corollary 5.1.23 proves a special case of this result.
The next few results use the notation of Definitions 5.2.24, 5.2.27 and 5.2.29.
Corollary 5.3.8. Let X ⊆ An be a variety. Then X is geometrically irreducible. Let
X ⊆ Pn be a variety. Then X ∩ An is geometrically irreducible.

The case where X is empty is trivial so suppose X 6= ∅.
By Lemma 5.2.30,

Proof:
I X = I(X). Hence, if g, h ∈ k[x0 , . . . , xn ] are such that gh ∈ I X then (g ◦ϕ)(h◦ϕ) =
(gh) ◦ ϕ ∈ I(X) by part 4 of Lemma 5.2.25. Theorem 5.3.4 implies I(X)
is a prime ideal
and so either g ◦ ϕ or h ◦ ϕ is in I(X). Hence either g or h is in I X .
For the converse, suppose X ∩An 6= ∅. If gh ∈ I(X ∩An ) then gh = gh ∈ I(X ∩ An ) ⊆
I(X). Hence g or h is in I(X) and so, by part 4 of Lemma 5.2.25, g or h is in I(X ∩ An ).

Theorem 5.3.9. Let X ⊆ Pn be an algebraic set such that X ∩ An 6= ∅. Then X ∩ An ⊆
X. If X is a variety then X ∩ An = X.
Proof: If f ∈ I(X) then f ◦ ϕ ∈ I(X ∩ An )
and so f ◦ ϕ ∈ I(X ∩ An ). Hence, f ∈
I(X ∩ An ). In other words, I(X) ⊆ I X ∩ An and so X ∩ An ⊆ X.
Let X1 = X ∩ An ⊆ X and X2 = X ∩ V (x0 ). Then X = X1 ∪ X2 and so, if X is
irreducible and X ∩ An 6= ∅ then X = X1 . 
Theorem 5.3.10. Let k be a field and let f (x, y) ∈ k[x, y] (or f (x, y, z) ∈ k[x, y, z]
homogeneous) have no repeated factors over k. Let X = V (f (x, y)) ⊂ A2 (k) or X =
V (f (x, y, z)) ⊆ P2 (k). Then X is geometrically irreducible if and only if f is irreducible
over k.
Proof: Suppose X = V (f ) is geometrically irreducible but that f = gh is a factorization
in k[x, y] or k[x, y, z] with both g and h having degree ≥ 1. Since f has no repeated
factors we have that g and h have no irreducible factors in common. Now, V (f ) =
V (gh) = V (g) ∪ V (h). Since V (f ) is irreducible either V (g) = V (f ) or V (h) = V (f ).
Without loss of generality we may assume V (g) = V (f ). By Hilbert’s Nullstellensatz
(Theorem 5.1.22) it follows that g m ∈ hf i for some integer m, which means that f | g m .
Now, let q be an irreducible factor of h. Then q | f and so q | g m and so q | g. But g and
h are supposed to have no common irreducible factors, so this is a contradiction. Hence,
if X is geometrically irreducible then f is k-irreducible.
Conversely, by Corollary 5.1.23 (respectively, Theorem 5.2.31) we have Ik (V (f )) = (f ).
Since f is irreducible it follows that (f ) is a prime ideal and so X is irreducible. 
98 CHAPTER 5. VARIETIES

Example 5.3.11. It is necessary to work over k for Theorem 5.3.10. For example, let
f (x, y) = y 2 + x2 (x − 1)2 . Then V (f (x, y)) ⊆ A2 (R) consists of two points and so is
reducible, even though f (x, y) is R-irreducible.

Lemma 5.3.12. Let X be a variety and U ⊂ X a non-empty set. If U is open (in the
Zariski topology) in X then U is dense in X (i.e., the topological closure of U in X in
the Zariski topology is X).

Proof: Let X1 be the closure of U in X and X2 = X − U . Then X = X1 ∪ X2 and

X1 , X2 are closed sets. Since X is irreducible and X2 6= X it follows that X1 = X. 

Lemma 5.3.13. Let X be a variety and U a non-empty open subset of X. Then Ik (U ) =

Ik (X).

Proof: Since U ⊆ X we have Ik (X) ⊆ Ik (U ). Now let f ∈ Ik (U ). Then U ⊆ V (f ) ∩ X.

Write X1 = V (f ) ∩ X, which is an algebraic set, and X2 = X − U , which is also an
algebraic set. Then X = X1 ∪ X2 and, since X is irreducible and X2 6= X, X = X1 . In
other words, f ∈ Ik (X). 

Exercise 5.3.14. Let X be an irreducible variety. Prove that if U1 , U2 ⊆ X are non-

empty open sets then U1 ∩ U2 6= ∅.

5.4 Function Fields

If X is a variety defined over k then Ik (X) is a prime ideal and so the affine or homogeneous
coordinate ring is an integral domain. One can therefore consider its field of fractions. If
X is affine then the field of fractions has a natural interpretation as a set of maps X → k.
When X is projective then a ratio f /g of polynomials does not give a well-defined function
on X unless f and g are homogeneous of the same degree.

Definition 5.4.1. Let X be an affine variety defined over k. The function field k(X)
is the set
k(X) = {f1 /f2 : f1 , f2 ∈ k[X], f2 6∈ Ik (X)}
of classes under the equivalence relation f1 /f2 ≡ f3 /f4 if and only if f1 f4 − f2 f3 ∈ Ik (X).
In other words, k(X) is the field of fractions of the affine coordinate ring k[X] over k.
Let X be a projective variety. The function field is

k(X) = {f1 /f2 : f1 , f2 ∈ k[X] homogeneous of the same degree, f2 6∈ Ik (X)}

with the equivalence relation f1 /f2 ≡ f3 /f4 if and only if f1 f4 − f2 f3 ∈ Ik (X).

Elements of k(X) are called rational functions. For a ∈ k the rational function
f : X → k given by f (P ) = a is called a constant function.

Exercise 5.4.2. Prove that the field of fractions of an integral domain is a field. Hence,
deduce that if X is an affine variety then k(X) is a field. Prove also that if X is a
projective variety then k(X) is a field.

We stress that, when X is projective, k(X) is not the field of fractions of k[X] and
that k[X] 6⊆ k(X). Also note that elements of the function field are not functions X → k
but maps X → k (i.e., they are not necessarily defined everywhere).

Example 5.4.3. One has k(A2 ) ∼

= k(x, y) and k(P2 ) ∼
= k(x, y).
5.4. FUNCTION FIELDS 99

Definition 5.4.4. Let X be a variety and let f1 , f2 ∈ k[X]. Then f1 /f2 is defined or
regular at P if f2 (P ) 6= 0. An equivalence class f ∈ k(X) is regular at P if it contains
some f1 /f2 with f1 , f2 ∈ k[X] (if X is projective then necessarily deg(f1 ) = deg(f2 )) such
that f1 /f2 is regular at P .
Note that there may be many choices of representative for the equivalence class of f ,
and only some of them may be defined at P .
Example 5.4.5. Let k be a field of characteristic not equal to 2. Let X be the algebraic
set V (y 2 − x(x − 1)(x + 1)) ⊂ A2 (k). Consider the functions

x(x − 1) y
f1 = and f2 = .
y x+1

One can check that f1 is equivalent to f2 . Note that f1 is not defined at (0, 0), (1, 0) or
(−1, 0) while f2 is defined at (0, 0) and (1, 0) but not at (−1, 0). The equivalence class of
f1 is therefore regular at (0, 0) and (1, 0). Section 7.3 gives techniques to deal with these
issues for curves, from which one can deduce that no function in the equivalence class of
f1 is defined at (−1, 0).
Exercise 5.4.6. Let X be a variety over k. Suppose f1 /f2 and f3 /f4 are equivalent
functions on X that are both defined at P ∈ X(k). Show that (f1 /f2 )(P ) = (f3 /f4 )(P ).
Hence, if f is a function that is defined at a point P then it makes sense to speak of
the value of the function at P . If the value of f at P is zero then P is called a zero of
f .5
Exercise 5.4.7. Let X = V (w2 x2 − w2 z 2 − y 2 z 2 + x2 z 2 ) ⊆ P3 (k). Show that (x2 +
yz)/(x2 − w2 ) ≡ (x2 − z 2 )/(x2 − yz) in k(X). Hence find the value of (x2 − z 2 )/(x2 − yz)
at the point (w : x : y : z) = (0 : 1 : 1 : 1). Show that both representations of the function
have the same value on the point (w : x : y : z) = (2 : 1 : −1 : 1).
Theorem 5.4.8. Let X be a variety and let f be a rational function. Then there is
a non-empty open set U ⊂ X such that f is regular on U . Conversely, if U ⊂ X is
non-empty and open and f : U → k is a function given by a ratio f1 /f2 of polynomials
(homogeneous polynomials of the same degree if X is projective) that is defined for all
P ∈ U then f extends uniquely to a rational function f : X → k.
Proof: Let f = f1 /f2 where f1 , f2 ∈ k[X]. Define U = X − V (f2 ). Since f2 6= 0 in k[X]
we have U a non-empty open set, and f is regular on U .
For the converse, Let f = f1 /f2 be a function on U given as a ratio of polynomials.
Then one can consider f1 and f2 as elements of k[X] and f2 non-zero on U implies f2 6= 0
in k[X]. Hence f1 /f2 corresponds to an element of k(X). Finally, suppose f1 /f2 and
f3 /f4 are functions on X (where f1 , f2 , f3 , f4 are polynomials) such that the restrictions
(f1 /f2 )|U and (f3 /f4 )|U are equal. Then f1 f4 − f2 f3 is zero on U and, by Lemma 5.3.13,
(f1 f4 − f2 f3 ) ∈ Ik (X) and f1 /f2 ≡ f3 /f4 on X. 
Corollary 5.4.9. If X is a projective variety and X ∩ An 6= ∅ then k(X) ∼
= k(X ∩ An ).
If X is non-empty affine variety then k(X) ∼ = k(X).
Proof: The result follows since X ∩ An = X − V (xn ) is open in X and X is open in X.

5 For curves we will later define the notion of a function f having a pole at a point P . This notion

does not make sense for general varieties, as shown by the function x/y on A2 at (0, 0) for example.
100 CHAPTER 5. VARIETIES

Definition 5.4.10. Let X be a variety and U ⊆ X. Define O(U ) to be the elements of

k(X) that are regular on all P ∈ U (k).
Lemma 5.4.11. If X is an affine variety over k then O(X) = k[X].
Proof: (Sketch) Clearly k[X] ⊆ O(X). The converse follows since O(X) is the inter-
section of the local rings (see Definition 7.1.1) at all P ∈ X(k). We refer to Proposition
2 on page 43, Chapter 2 of [216] or Theorem I.3.2(a) of [278] for the details. 
Definition 5.4.12. Let X be a variety over k and f ∈ k(X). Let σ ∈ Gal(k/k). If
f = f1 /f2 where f1 , f2 ∈ k[x] define σ(f ) = σ(f1P
)/σ(f2 ) where
P σ(f1 ) and σ(f2 ) denote
the natural Galois action on polynomials (i.e., σ( i ai xi ) = i σ(ai )xi ). Some authors
write this as f σ .
Exercise 5.4.13. Prove that σ(f ) is well-defined (i.e., if f ≡ f ′ then σ(f ) ≡ σ(f ′ )). Let
P ∈ X(k). Prove that f (P ) = 0 if and only if σ(f )(σ(P )) = 0.
Remark 5.4.14. Having defined an action of G = Gal(k/k) on k(X) it is natural to
ask whether k(X)G = {f ∈ k(X) : σ(f ) = f ∀σ ∈ Gal(k/k)} is the same as k(X). The
issue is whether a function being “defined over k” is the same as “can be written with
coefficients in k”. Indeed, this is true but not completely trivial.
A sketch of the argument is given in Exercise 1.12 of Silverman [564] and we give a
few extra hints here. Let X be a projective variety. One first shows that if X is defined
over k and if k′ is a finite Galois extension of k then Ik′ (X) is an induced Galois module
(see page 110 of Serre [542]) for Gal(k′ /k). It follows from Section VII.1 of [542] that the
Galois cohomology group H 1 (Gal(k′ /k), Ik′ (X)) is trivial and hence, by Section X.3 of
[542], that H 1 (G, Ik′ (X)) = 0. One can therefore deduce, as in Exercise 1.12(a) of [564],
that k[X]G = k[X].
To show that k(X)G = k(X) let (f0 : f1 ) : X → P1 and let σ ∈ G. Then σ(f0 ) =
∗
λσ f0 + G0,σ and σ(f1 ) = λσ f1 + G1,σ where λσ ∈ k and G0,σ , G1,σ ∈ Ik (X). One shows
∗
first that λσ ∈ H 1 (G, k ), which is trivial by Hilbert 90, and so λσ = σ(α)/α for some
α ∈ k. Replacing f0 by αf0 and f1 by αf1 gives λσ = 1 and one can proceed to showing
that G0,σ , G1,σ ∈ H 1 (G, Ik (X)) = 0 as above. The result follows.
For a different approach see Theorem 7.8.3 and Remark 8.4.11 below, or Corollary 2
of Section VI.5 (page 178) of Lang [364].

5.5 Rational Maps and Morphisms

Definition 5.5.1. Let X be an affine or projective variety over a field k and Y an affine
variety in An over k. Let φ1 , . . . , φn ∈ k(X). A map φ : X → An of the form
φ(P ) = (φ1 (P ), . . . , φn (P )) (5.1)
is regular at a point P ∈ X(k) if all φi , for 1 ≤ i ≤ n, are regular at P . A rational
map φ : X → Y defined over k is a map of the form (5.1) such that, for all P ∈ X(k) for
which φ is regular at P then φ(P ) ∈ Y (k).
Let X be an affine or projective variety over a field k and Y a projective variety in Pn
over k. Let φ0 , . . . , φn ∈ k(X). A map φ : X → Pn of the form
φ(P ) = (φ0 (P ) : · · · : φn (P )) (5.2)
is regular at a point P ∈ X(k) if there is some function g ∈ k(X) such that all gφi , for
0 ≤ i ≤ n, are regular at P and, for some 0 ≤ i ≤ n, one has (gφi )(P ) 6= 0.6 A rational
6 This last condition is to prevent φ mapping to (0 : · · · : 0), which is not a point in Pn .
5.5. RATIONAL MAPS AND MORPHISMS 101

map φ : X → Y defined over k is a map of the form (5.2) such that, for all P ∈ X(k) for
which φ is regular at P , then φ(P ) ∈ Y (k).
We stress that a rational map is not necessarily defined at every point of the domain.
In other words, it is not necessarily a function.
Exercise 5.5.2. Let X and Y be projective varieties. Show that one can write a rational
map in the form φ(P ) = (φ0 (P ) : · · · : φn (P )) where the φi (x) ∈ k[x] are all homogeneous
polynomials of the same degree, not all φi (x) ∈ Ik (X), and for every f ∈ Ik (Y ) we have
f (φ0 (x), . . . , φn (x)) ∈ Ik (X).
Example 5.5.3. Let X = V (x − y) ⊆ A2 and Y = V (x − z) ⊆ P2 . Then

φ(x, y) = (x : xy : y)

is a rational map from X to Y . Note that this formula for φ is not defined at (0, 0).
However, φ is regular at (0, 0) since taking g = x−1 gives the equivalent form φ(x, y) =
(x−1 x : x−1 xy : x−1 y) = (1 : y : y/x) and y/x ≡ 1 in k(X). Also note that the image of
φ is not equal to Y (k) as it misses the point (0 : 1 : 0).
Similarly, ψ(x : y : z) = (x/y, z/y) is a rational map from Y to X. This map is not
regular at (1 : 0 : 1) but it is surjective to X. The composition ψ ◦ φ maps (x, y) to
(1/y, 1/x).
Example 5.5.4. Let X = V (y 2 z − (x3 + Axz 2 )) ⊆ P2 and Y = P1 . Consider the rational
map
φ(x : y : z) = (x/z : 1).
Note that this formula for φ is defined at all points of X except P0 = (0 : 1 : 0). Let
g(x : y : z) = (x2 + Az 2 )/y 2 ∈ k(X). Then the map (x : y : z) 7→ (gx/z : g) can be
written as (x : y : z) 7→ (1 : g) and this is defined at (0 : 1 : 0). It follows that φ is regular
at P0 and that φ(P0 ) = (1 : 0).
Lemma 5.5.5. Let X and Y be varieties over k and let φ : X → Y be a rational map.
Then there is an open set U ⊆ X such that φ is regular on U .
Proof: Write φ as (φ1 , . . . , φn ) if Y is affine or (φ0 : · · · : φn ) if Y is projective. By
Theorem 5.4.8 for each φi for 1 ≤ φi ≤ n (respectively, 0 ≤ φi ≤ n) there is a non-empty
open set Ui ⊂ X such that φi is regular. Taking U = ∩i Ui gives the result. 
It immediately follows that Theorem 5.4.8 generalises to rational maps.
Theorem 5.5.6. Let X and Y be varieties. Suppose φ1 , φ2 : X → Y are rational maps
that are regular on non-empty open sets U1 , U2 ⊆ X. Suppose further that φ1 |U1 ∩U2 =
φ2 |U1 ∩U2 . Then φ1 = φ2 .
Exercise 5.5.7. Prove Theorem 5.5.6.
Definition 5.5.8. Let X and Y be algebraic varieties over k. A rational map φ : X → Y
defined over k is a birational equivalence over k if there exists a rational map ψ : Y →
X over k such that:
1. ψ ◦ φ(P ) = P for all points P ∈ X(k) such that ψ ◦ φ(P ) is defined;
2. φ ◦ ψ(Q) = Q for all points Q ∈ Y (k) such that φ ◦ ψ(Q) = Q is defined.
Varieties X and Y are birationally equivalent if there is a birational equivalence φ :
X → Y between them.
102 CHAPTER 5. VARIETIES

Exercise 5.5.9. Show that X = V (xy−1) ⊆ A2 and Y = V (x1 −x2 ) ⊆ P2 are birationally
equivalent.
Exercise 5.5.10. Verify that birational equivalence is an equivalence relation.
Example 5.5.11. The maps ϕi : An → Pn and ϕ−1 i : Pn → An from Definition 5.2.24
n n
are rational maps. Hence A and P are birationally equivalent.
Definition 5.5.12. Let X and Y be varieties over k and let U ⊆ X be open. A rational
map φ : U → Y over k which is regular at every point P ∈ U (k) is called a morphism
over k.
Let U ⊆ X and V ⊆ Y be open. If φ : U → Y is a morphism over k and ψ : V → X
is a morphism over k such that φ ◦ ψ and ψ ◦ φ are the identity on V and U respectively
then we say that U and V are isomorphic over k. If U and V are isomorphic we write
U∼ =V.
Example 5.5.13. Let X be V (xy − z 2 ) ⊆ P2 and let φ : X → P1 be given by φ(x : y :
z) = (x/z : 1). Then φ is a morphism (for (1 : 0 : 0) replace φ by the equivalent form
φ(x : y : z) = (1 : z/x) and for (0 : 1 : 0) use φ(x : y : z) = (z/y : 1)). Indeed, φ is an
isomorphism with inverse ψ(x : z) = (x/z : z/x : 1).
Lemma 5.5.5 shows that every rational map φ : X → Y restricts to a morphism
φ : U → Y on some open set U ⊆ X.
Exercise 5.5.14. Let X = V (x2 + y 2 − 1) ⊂ A2 over k. By taking a line of slope t ∈ k
through (−1, 0) give a formula for a rational map φ : A1 → X. Explain how to extend
this to a morphism from P1 to V (x2 + y 2 − 1). Show that this is an isomorphism.
We now give the notion of a dominant rational map (or morphism). This is the
appropriate analogue of surjectivity for maps between varieties. Essentially, a rational
map from X to Y is dominant if its image is not contained in a proper subvariety of Y .
For example, the map ϕi : An → Pn is dominant. Birational maps are also dominant.
Definition 5.5.15. Let X and Y be varieties over k. A set U ⊆ Y (k) is dense if its
closure in the Zariski topology in Y (k) is equal to Y (k). A rational map φ : X → Y is
dominant if φ(X(k)) is dense in Y (k).
Example 5.5.16. Let φ : A2 → A2 be given by φ(x, y) = (x, x). Then φ is not dominant
(though it is dominant to V (x − y) ⊆ A2 ). Let φ : A2 → A2 be given by φ(x, y) = (x, xy).
Then φ is dominant, even though it is not surjective.
We now show that a morphism is a continuous map for the Zariski topology.
Lemma 5.5.17. Let X and Y be varieties over k. Let φ : X → Y be a morphism. Let
V ⊆ Y be an open set such that φ(X) ∩ V 6= ∅. Then φ−1 (V ) is an open set in X.
Similarly, let Z ⊆ Y be a closed set such that φ(X) ∩ Z 6= ∅. Then φ−1 (Z) is closed in
X.
Exercise 5.5.18.⋆ Prove Lemma 5.5.17.
Exercise 5.5.19. Let X and Y be varieties and let φ : X → Y be a morphism. Show
that the Zariski closure of φ(X) in Y is irreducible.
Lemma 5.5.20. Let X and Y be affine varieties over k, let U ⊆ X be open, and let
φ : U → Y be a morphism. Then the composition f ◦ φ induces a well-defined ring
homomorphism from k[Y ] to O(U ).
5.5. RATIONAL MAPS AND MORPHISMS 103

Proof: If f ∈ k[Y ] then f is regular on Y and, since φ is regular on U , f ◦ φ is regular

on U . Hence, f ◦ φ ∈ O(U ).
We now show that the map is well-defined. Suppose f ≡ 0 in k[Y ]. Since f vanishes on
Y it follows that, for all P ∈ U (k), f (φ(P )) = 0. Write f ◦ φ = f1 /f2 where f1 and f2 are
polynomials. It follows that f1 ∈ I(U ) = I(X) (using Lemma 5.3.13). Hence, f ◦ φ ≡ 0
in O(U ). Finally, the map is a ring homomorphism since (f1 + f2 ) ◦ φ = (f1 ◦ φ) + (f2 ◦ φ)
and similarly for multiplication. 

Definition 5.5.21. Let X and Y be affine varieties over k, let U ⊆ X be open, and let
φ : U → Y be a morphism. The pullback7 is the ring homomorphism φ∗ : k[Y ] → O(U )
defined by φ∗ (f ) = f ◦ φ.

Lemma 5.5.22. Let X and Y be affine varieties over k, let U ⊆ X be open, and let
φ : U → Y be a morphism. Then φ is dominant if and only if φ∗ is injective.

Proof: Let f ∈ k[Y ] be in the kernel of φ∗ . Now φ∗ (f ) = 0 is the same as f ◦ φ = 0 on U ,

which implies φ(U ) ⊆ V (f ) ∩ Y . If φ(U ) is dense in Y then Y ⊆ V (f ) and so f ∈ I(Y )
and φ∗ is injective. Conversely, if φ(U ) is not dense in Y then there is some polynomial
f 6∈ I(Y ) such that φ(U ) ⊆ Y ∩ V (f ). It follows that φ∗ (f ) = 0 and φ∗ is not injective.

Note that if X and φ are defined over k then φ∗ : k[Y ] → O(U ) restricts to φ∗ :
k[Y ] → k(X). If φ∗ is injective then one can extend it to get a homomorphism of the
field of fractions of k[Y ] to k(X).

Definition 5.5.23. Let X and Y be varieties over k and let φ : X → Y be a dominant

rational map defined over k. Define the pullback φ∗ : k(Y ) → k(X) by φ∗ (f ) = f ◦ φ.

We will now sketch a proof that φ∗ is a k-algebra homomorphism. Recall that a

k-algebra homomorphism of fields is a field homomorphism that is the identity map on k.

Theorem 5.5.24. Let X and Y be varieties over k and let φ : X → Y be a dominant

rational map defined over k. Then the pullback φ∗ : k(Y ) → k(X) is an injective k-algebra
homomorphism.

Proof: Without loss of generality we may assume that X and Y are affine. The rational
map φ is therefore given by φ(x) = (φ1 (x), . . . , φn (x)). Let U ⊆ X be an open set on
which φ is regular. Then φ : U → Y is a morphism and we know φ∗ : k[Y ] → O(U ) is
a ring homomorphism by Lemma 5.5.20. The field of fractions of k[Y ] is k(Y ) and the
field of fractions of O(U ) is k(X). The natural extension of φ∗ to φ∗ : k(Y ) → k(X) is
well-defined.
It immediately follows that φ∗ is a ring homomorphism and that φ∗ is the identity on k.
Hence, φ∗ is a k-algebra homomorphism. Furthermore, φ∗ is injective by Lemma 5.5.22.
Finally, since φ is defined over k it restricts to an injective homomorphism from k(Y ) to
k(X). 

Example 5.5.25. Consider the rational maps from Example 5.5.16. The map φ(x, y) =
(x, x) is not dominant and does not induce a well-defined function from k(x, y) to k(x, y)
since, for example, φ∗ (1/(x − y)) = 1/(x − x) = 1/0.
The map φ(x, y) = (x, xy) is dominant and φ∗ (f (x, y)) = f (x, xy) is a field isomor-
phism.
7 Pullback is just a fancy name for “composition”; but we think of it as “pulling” a structure from the

image of φ back to the domain.

104 CHAPTER 5. VARIETIES

Exercise 5.5.26. Let K1 , K2 be fields containing a field k. Let θ : K1 → K2 be a

k-algebra homomorphism. Show that θ is injective.
Theorem 5.5.27. Let X and Y be varieties over k and let θ : k(Y ) → k(X) be a k-
algebra homomorphism. Then θ induces a dominant rational map φ : X → Y defined
over k.
Proof: If Y is projective it suffices to construct a rational map to an affine part, say
Y ∩An . Hence, we assume that Y ⊆ An is affine and described by coordinates (y1 , . . . , yn ).
The homomorphism θ maps each yi to some φi (x) ∈ k(X) for 1 ≤ i ≤ n. Define
φ : X → An by
φ(P ) = (φ1 (P ), . . . , φn (P )).
We now show that if P ∈ X(k) and if φ is regular at P then φ(P ) ∈ Y (k). Let f ∈ I(Y ).
Then
f (φ(P )) = f (φ1 (P ), . . . , φn (P )) = f (θ(y1 )(P ), . . . , θ(yn )(P )).
Now, θ is a k-algebra homomorphism and f is a polynomial in k[y1 , . . . , yn ]. Hence

f (θ(y1 ), . . . , θ(yn )) = θ(f (y1 , . . . , yn )).

Since f (y1 , . . . , yn ) ∈ I(Y ) it follows that f (y1 , . . . , yn ) = 0 in k(Y ) and so θ(f ) = 0. It

follows that f (φ(P )) = θ(f )(P ) = θ(0)(P ) = 0 for all f ∈ I(Y ) and so P ∈ Y (k) by part
5 of Proposition 5.1.16.
Finally, by Exercise 5.5.26, θ is injective. Also, φ∗ equals θ and so φ∗ is injective.
Hence, Lemma 5.5.22 implies that φ is dominant. 
Theorem 5.5.28. Let X and Y be varieties over k. Then X and Y are birationally
equivalent over k if and only if k(X) ∼
= k(Y ) (isomorphic as fields).
Proof: Let φ : X → Y and ψ : Y → X be the birational equivalence. First we must
deduce that φ and ψ are dominating. There are subsets U ⊆ X and V ⊆ Y such that φ is
regular on U , ψ is regular on V and ψ ◦ φ is the identity on U (in other words, φ : U → V
is an isomorphism). The maps φ∗ : k[V ] → O(U ) and ψ ∗ : k[X] → O(V ) therefore satisfy
φ∗ ψ ∗ (f ) = f ◦ (ψ ◦ ψ) = f (at least, they are equal on U ∩ φ−1 (V ), which can be shown to
be open) and so are injective. It follows from Lemma 5.5.22 that φ and ψ are dominant.
Hence, φ induces a k-algebra homomorphism φ∗ : k(Y ) → k(X) and ψ induces a
k-algebra homomorphism φ∗ : k(X) → k(Y ). Finally, ψ ◦ φ induces a k-algebra homo-
morphism φ∗ ψ ∗ : k(X) → k(X) that is the identity (since it is the identity on a dense
open set). It follows that ψ ∗ and φ∗ are isomorphisms.
For the converse, if θ : k(Y ) → k(X) is an isomorphism then we associate a dominant
rational map φ : X → Y to θ and ψ : Y → X to θ−1 . Since θ−1 θ is the identity it follows
that ψ ◦ φ is the identity whenever it is regular. 
Some authors prefer to study function fields rather than varieties, especially in the case
of dimension 1 (there are notable classical texts that take this point of view by Chevalley
and Deuring; see Stichtenoth [589] for a more recent version). By Theorem 5.5.28 (and
other results) the study of function fields up to isomorphism is the study of varieties up to
birational equivalence. A specific set of equations to describe a variety is called a model.
Definition 5.5.29. Let X and Y be varieties over k and let φ : X → Y be a ratio-
nal map over k given by φ(P ) = (φ1 (P ), . . . , φn (P )) if Y is affine and (φ0 (P ) : · · · :
φn (P )) if Y is projective. Let σ ∈ Gal(k/k). Define σ(φ) : X → Y by σ(φ)(P ) =
(σ(φ1 )(P ), . . . , σ(φn (P ))) if Y is affine and σ(φ)(P ) = (σ(φ0 )(P ) : · · · : σ(φn (P ))) if Y is
projective. Many authors act by Galois on the right and so write the action as φσ .
5.6. DIMENSION 105

Lemma 5.5.30. Let X and Y be varieties over k and let φ : X → Y be a rational map
over k. If σ(φ) = φ for all σ ∈ Gal(k/k) then φ is defined over k.

Proof: If Y is affine then φ(P ) = (φ1 (P ), . . . , φn (P )) where φi ∈ k(X). If σ(φ) = φ then

σ(φi ) = φi for all 1 ≤ i ≤ n. Remark 5.4.14 therefore implies that φi ∈ k(X) for all i and
so φ is defined over k.
If Y is projective then φ(P ) = (φ0 (P ) : · · · : φn (P )) where φi ∈ k(X) for 0 ≤ i ≤ n.
∗
If φ(P ) = σ(φ)(P ) then, for all σ ∈ Gal(k/k), there is some xi(σ) ∈ k such that
∗
σ(φi ) = ξ(σ)φi for all 0 ≤ i ≤ n. As in Lemma 5.2.5, ξ : Gal(k/k) → k is a 1-cocycle and
so by Hilbert 90 is a co-boundary. It follows that one can choose the φi so that σ(φi ) = φi
and hence, by Remark 5.4.14, φi ∈ k(X) for 0 ≤ i ≤ n. 

5.6 Dimension
The natural notion of dimension (a point has dimension 0, a line has dimension 1, a
plane has dimension 2, etc) generalises to algebraic varieties. There are algebraic and
topological ways to define dimension. We use an algebraic approach.8
We stress that the notion of dimension only applies to irreducible algebraic sets. For
example X = V (x, y) ∪ V (x − 1) = V (x(x − 1), y(x − 1)) ⊆ A2 is the union of a point and
a line so has components of different dimension.
Recall the notion of transcendence degree of an extension k(X) over k from Defini-
tion A.6.3.

Definition 5.6.1. Let X be a variety over k. The dimension of X, denoted dim(X), is

the transcendence degree of k(X) over k.

Example 5.6.2. The dimension of An is n. The dimension of Pn is n.

Theorem 5.6.3. Let X and Y be varieties. If X and Y are birationally equivalent then
dim(X) = dim(Y ).

Proof: Immediate from Theorem 5.5.28. 

Corollary 5.6.4. Let X be a projective variety such that X ∩ An is non-empty.

Then
dim(X) = dim(X ∩ An ). Let X be an affine variety. Then dim(X) = dim X .

Exercise 5.6.5. Let f be a non-constant polynomial and let X = V (f ) be a variety in

An . Show that dim(X) = n − 1.

Exercise 5.6.6. Show that if X is a non-empty variety of dimension zero then X = {P }

is a single point.

An useful alternative formulation of dimension is as follows.

Definition 5.6.7. Let R be a ring. The Krull dimension of R is the supremum of

n ∈ Z≥0 such that there exists a chain I0 ⊂ I1 ⊂ · · · ⊂ In of prime R-ideals such that
Ij−1 6= Ij for 1 ≤ j ≤ n.

Theorem 5.6.8. Let X be an affine variety over k. Then dim(X) is equal to the Krull
dimension of the affine coordinate ring k[X].

Proof: See Proposition I.1.7 and Theorem I.1.8A of [278]. 

8 See Chapter 8 of Eisenbud [191] for a clear criticism of this approach.
106 CHAPTER 5. VARIETIES

Corollary 5.6.9. Let X and Y be affine varieties over k such that Y is a proper subset
of X. Then dim(Y ) < dim(X).
Proof: Since Y 6= X we have Ik (X) ( Ik (Y ) and both ideals are prime since X and Y
are irreducible. It follows that the Krull dimension of k[X] is at least one more than the
Krull dimension of k[Y ]. 
Exercise 5.6.10. Show that a proper closed subset of a variety of dimension 1 is finite.

5.7 Weil Restriction of Scalars

Weil restriction of scalars is simply the process of re-writing a system of polynomial
equations over a finite algebraic extension k′ /k as a system of equations in more variables
over k. The canonical example is identifying the complex numbers A1 (C) with A2 (R) via
z = x + iy ∈ A1 (C) 7→ (x, y) ∈ A2 (R). We only need to introduce this concept in the
special case of affine algebraic sets over finite fields.
Lemma 5.7.1. Let q be a prime power, m ∈ N and fix a vector space basis {θ1 , . . . , θm } for
Fqm over Fq . Let x1 , . . . , xn be coordinates for An and let y1,1 , . . . , y1,m , . . . , yn,1 , . . . , yn,m
be coordinates for Anm . The map φ : Anm → An given by

φ(y1,1 , . . . , yn,m ) = (y1,1 θ1 + · · · + y1,m θm , y2,1 θ1 + · · · + y2,m θm , . . . , yn,1 θ1 + · · · + yn,m θm )

gives a bijection between Anm (Fq ) and An (Fqm ).

Exercise 5.7.2. Prove Lemma 5.7.1.
Definition 5.7.3. Let X = V (S) ⊆ An be an affine algebraic set over Fqm . Let φ be as
in Lemma 5.7.1. For each polynomial f (x1 , . . . , xn ) ∈ S ⊆ Fqm [x1 , . . . , xn ] write

φ∗ (f ) = f ◦ φ = f (y1,1 θ1 + · · · + y1,m θm , y2,1 θ1 + · · · + y2,m θm , . . . , yn,1 θ1 + · · · + yn,m θm )

(5.3)
as
f1 (y1,1 , . . . , yn,m )θ1 + f2 (y1,1 , . . . , yn,m )θ2 + · · · + fm (y1,1 , . . . , yn,m )θm (5.4)
where each fj ∈ Fq [y1,1 , . . . , yn,m ]. Define S ′ ⊆ Fq [y1,1 , . . . , yn,m ] to be the set of all such
polynomials fj over all f ∈ S. The Weil restriction of scalars of X with respect to
Fqm /Fq is the affine algebraic set Y ⊆ Amn defined by

Y = V (S ′ ).

Example 5.7.4. Let p ≡ 3 (mod 4) and define Fp2 = Fp (i) where i2 = −1. Consider the
algebraic set X = V (x1 x2 − 1) ⊆ A2 . The Weil restriction of scalars of X with respect to
Fp2 /Fp with basis {1, i} is

Y = V (y1,1 y2,1 − y1,2 y2,2 − 1, y1,1 y2,2 + y1,2 y2,1 ) ⊆ A4 .

Recall from Example 5.1.5 that X is an algebraic group. The multiplication operation
mult((x1 , x2 ), (x′1 , x′2 )) = (x1 x′1 , x2 x′2 ) on X corresponds to the operation
′ ′ ′ ′
mult((y1,1 , y1,2 , y2,1 , y2,2 ), (y1,1 , y1,2 , y2,1 , y2,2 ))
′ ′ ′ ′ ′ ′ ′ ′
= (y1,1 y1,1 − y1,2 y1,2 , y1,1 y1,2 + y1,2 y1,1 , y2,1 y2,1 − y2,2 y2,2 , y2,1 y2,2 + y2,2 y2,1 )

on Y .
5.7. WEIL RESTRICTION OF SCALARS 107

Exercise 5.7.5. Let p ≡ 3 (mod 4). Write down the Weil restriction of scalars of
X = V (x2 − 2i) ⊂ A1 with respect to Fp2 /Fp .
Exercise 5.7.6. Let p ≡ 3 (mod 4). Write down the Weil restriction of scalars of
V (x21 + x22 − (1 + 2i)) ⊂ A2 with respect to Fp2 /Fp .
Theorem 5.7.7. Let X ⊆ An be an affine algebraic set over Fqm . Let Y ⊆ Amn be the
Weil restriction of X. Let k ∈ N be coprime to m. Then there is a bijection between
X(Fqmk ) and Y (Fqk ).
Proof: When gcd(k, m) = 1 it is easily checked that the map φ of Lemma 5.7.1 gives a
a one-to-one correspondence between Anm (Fqk ) and An (Fqmk ).
Now, let P = (x1 , . . . , xn ) ∈ X and write Q = (y1,1 . . . , yn,m ) for the corresponding
point in Amn . For any f ∈ S we have f (P ) = 0. Writing f1 , . . . , fm for the polynomials
in equation (5.4) we have

f1 (Q)θ1 + f2 (Q)θ2 + · · · + fm (Q)θm = 0.

Since {θ1 , . . . , θm } is also a vector space basis for Fqmk over Fqk we have

f1 (Q) = f2 (Q) = · · · = fm (Q) = 0.

Hence f (Q) = 0 for all f ∈ S ′ and so Q ∈ Y . Similarly, if Q ∈ Y then fj (Q) = 0 for all
such fj and so f (P ) = 0 for all f ∈ S. 
Note that, as the following example indicates, when k is not coprime to m then
X(Fqmk ) is not usually in one-to-one correspondence with Y (Fqk ).
Exercise 5.7.8. Consider the algebraic set X from Exercise 5.7.5. Show that X(Fp4 ) =
{1 + i, −1 − i}. Let Y be the Weil restriction of X with respect to Fp2 /Fp . Show that
Y (Fp2 ) = {(1, 1), (−1, −1), (i, −i), (−i, i)}.
Note that the Weil restriction of Pn with respect to Fqm /Fq is not the projective
closure of Amn . For example, considering the case n = 1, P1 has one point not contained
in A1 , whereas the projective closure of Am has an (m − 1)-dimensional algebraic set of
points at infinity.
Exercise 5.7.9. Recall from Exercise 5.5.14 that there is a morphism from P1 to Y =
V (x2 + y 2 − 1) ⊆ A2 . Determine the Weil restriction of scalars of Y with respect to
Fp2 /Fp . It makes sense to call this algebraic set the Weil restriction of P1 with respect to
Fp2 /Fp .