18.

785 Number theory I Fall 2016

Lecture #1 09/08/2016

1 Absolute values and discrete valuations

1.1 Introduction
At its core, number theory is the study of the integer ring Z. By the fundamental arithmetic,
every element of Z can be written uniquely as a product of primes (up to a unit ±1), so it is
natural to focus on the prime elements of Z. If p is a prime, the ideal (p) := pZ it generates
is a maximal ideal (Z is a ring of dimension one), and the residue field Z/pZ is the finite
field Fp with p elements (which is unique up to isomorphism). The fraction field of Z is the
field Q of rational numbers, and together Q and the finite fields Fp of prime order make up
the prime fields; every field k contains exactly one of them, according to its characteristic:
zero if k contains Q, and p if k contains Fp .
The structure of the ring Z and the distribution of its prime elements is intimately
related to the Reimann zeta function
X Y
ζ(s) = n−s = (1 − p−s )−1 .
p

This is a function of the complex variable s that is holomorphic and nonvanishing for
Re(s) > 1, and as we shall see it has an analytic continuation to the entire complex plane.
It has a simple pole at s = 1, which implies that there are infinitely many primes (otherwise
the product over primes on the RHS would be finite and converge). The location of its
zeros in the critical strip 0 < s < 1 is directly related to the distribution of primes (via the
explicit formula, which we will see later in the course), and as you are probably aware it is
conjectured that they all lie on the critical line Re(s) = 12 , this is the Riemann Hypothesis;
this conjecture remains open.
One can also consider finite extensions of Q, such as the field Q(i) := Q[x]/(x2 +1). These
are called number fields, and each can be constructed as the quotient of the polynomial ring
Q[x] by one of its maximal ideal; the ring Q[x] is a principal ideal domain and its maximal
ideals can all be written as (f ) for some monic irreducible f ∈ Z[x]. Associated to each
number field K is a zeta function ζK (s), and each of these has an associated conjecture
regarding the location of its zeros (these conjectures all remain open).
Number fields are one of two types of global fields that we will spend much of the first
part of the course studying; the other type are known as global function fields. Let Fq denote
the field with q elements, where q is any prime power. The polynomial ring Fq [t] has much
in common with the integer ring Z. Like Z, it is a principal ideal domain of dimension one,
and the residue fields Fq [t]/(f ) one obtains by taking the quotient by a maximal ideal (f ),
where f ∈ Fq [t] is any irreducible polynomial, is a finite field Fqd , where d is the degree of f .
In contrast to the situation with Z, the residue fields of Fq [t] all have the same characteristic
as its fraction field Fq (t), which plays a role analogous to Q. Global function fields are finite
extensions of Fq (t) (this includes Fq (t) itself, an extension of degree 1).
Associated to each global field k is an infinite collection of local fields corresponding to
the completions of k with respect to its absolute values; for the field of rational numbers Q,
these are the field of real numbers R and the p-adic fields Qp (as you will prove on Problem
Set 1).
The ring Z is a principal ideal domain (PID), as is Fq [t]. These rings have dimension
one, which means that every nonzero prime ideal is maximal; thus each nonzero prime ideal
has an associated residue field, and for both Z and Fq [t] these residue fields are finite. In the

Lecture by Andrew Sutherland

case of Z we have residue fields Z/pZ ' Fp for each prime p, and for Fq [t] we have residue
fields Fqd associated to each irreducible polynomial of degree d.
We will spend the first part of this course fleshing out this picture, in which we are
particularly interested in understanding the integral closure of the rings Z and Fq [t] in in
finite extensions of their fraction fields, known as rings of integers, and the prime ideals in
these rings of integers. Where possible we will treat number fields and function fields on an
equal footing, but we will also note some key differences. Perhaps somewhat surprisingly,
the function field setting often turns out to be simpler than the number field setting, and
considering the analogies between the two can provide insight into both.
While the topics above are typically classified under the heading of algebraic number
theory, a key tool for understanding global fields are their associated zeta functions, which
have a more analytic flavor (at least in the number field setting). The most famous of these
is the Riemann zeta function
X Y
ζ(s) := n−s = (1 − p−s )−1 ,
n≥1 p

which can be viewed both as a sum over integers and a product over primes (an Euler
product). As you are no doubt aware, the Riemann hypothesis is concerned with the location
of the complex zeros of the function ζ(s) and is one of the major open problems in number
theory. It is worth noting that the analog of the Riemann hypothesis in the function field
setting, the Riemann hypothesis for curves, is not an open problem. It was proved by André
Weil in the 1940s [5]; a further generalization to varieties of arbitrary dimension was proved
by Pierre Deligne in the 1970s [3].
Zeta functions provide the tool we need to understand the distribution of primes, both
in general, and within particular residue classes; the proofs of the prime number theorem
and Dirichlet’s theorem on primes in arithmetic progressions both use zeta functions in an
essential way. Dirichlet’s theorem states that for each integer m > 1 and each integer a
coprime to m, there are infinitely many primes p ≡ a mod m. In fact, more is true: the
Chebotarev density theorem tells us that for each modulus m the primes are equidistributed
among the residue classes of the integers a coprime to m. We will see this and several other
applications of the Chebotarev density theorem in the later part of the course.
Before we begin, let us note the following.

Remark 1.1. Our rings always have a multiplicative identity that is preserved by ring
homomorphisms (so the zero ring in which 1 = 0 is not an initial object in the category
of rings, but it is the terminal object in this category). Except where noted otherwise, the
rings we consider are commutative.

1.2 Absolute values

We begin with the general notion of an absolute value on a field; a reference for much of
this material is [4, Chapter 1].

Definition 1.2. An absolute value on a field k is a map | | : k → R≥0 such that for all
x, y ∈ k the following hold:

1. |x| = 0 if and only if x = 0;

2. |xy| = |x||y|;

18.785 Fall 2016, Lecture #1, Page 2

 3. |x + y| ≤ |x| + |y|.
If in addition the stronger condition
4. |x + y| ≤ max(|x|, |y|)
holds, then the absolute value is nonarchimedean; otherwise it is archimedean.
Example 1.3. The map | | : k → R≥0 defined by
(
1 if x 6= 0,
|x| =
0 if x = 0,
is the trivial absolute value on k. It is nonarchimedean.
Lemma 1.4. An absolute value | | on a field k is nonarchimedean if and only if
|1 + ··· + 1| ≤ 1
| {z }
n

for all n ≥ 1.
Proof. See problem set 1.

Corollary 1.5. In a field of positive characteristic every absolute value is nonarchimedean,

and the only absolute value on a finite field is the trivial one.
Definition 1.6. Two absolute values | | and | |0 on the same field k are equivalent if there
exists an α ∈ R>0 for which |x|0 = |x|α for all x ∈ k.

1.3 Absolute values on Q

To avoid confusion we will denote the usual absolute value on Q (inherited from R) by | |∞ ;
it is an archimedean absolute value. ButQthere are are infinitely many others. Recall that
any element of Q× may be written as ± q q eq , where the product ranges over primes and
the exponents eq ∈ Z are uniquely determined (as is the sign).
Definition 1.7. For a prime p the p-adic valuation vp : Q → Z is defined by
!
Y
vp ± q eq := ep ,
q

and we define vp (0) := ∞. The p-adic absolute value on Q is defined by

|x|p := p−vp (x) ,
where |0|p = p−∞ is understood to be 0.
Theorem 1.8 (Ostrowski’s Theorem). Every nontrivial absolute value on Q is equiva-
lent to | |p for some p ≤ ∞.
Proof. See Problem Set 1.

Theorem 1.9 (Product Formula). For every x ∈ Q× we have

Y
|x|p = 1.
p≤∞

Proof. See Problem Set 1.

18.785 Fall 2016, Lecture #1, Page 3

1.4 Discrete valuations
Definition 1.10. A valuation on a field k is a group homomorphism k × → R such that for
all x, y ∈ k we have
v(x + y) ≥ min(v(x), v(y)).
We may extend v to a map k → R ∪ {∞} by defining v(0) := ∞. For any any 0 < c < 1,
defining |x|v := cv(x) yields a nonarchimedean absolute value. The image of v in R is the
value group. We say that v is a discrete valuation if the value group is equal to Z (every
discrete subgroup of R is isomorphic to Z, so we can always rescale a valuation with a
discrete value group so that this holds). The set
A = {x ∈ k : v(x) ≥ 0},
is the valuation ring of k (with respect to v). The unit group of A is
A× = {x ∈ k : v(x) = 0},
since v(1/x) = v(1) − v(x) = −v(x) implies x is invertible if and only if v(x) = 0.
It is easy to verify that the valuation ring A is a (commutative) ring, and even an
integral domain (if x and y are nonzero then v(xy) = v(x) + v(y) 6= ∞, so xy 6= 0), and k
is its fraction field. Any integral domain A which is the valuation ring of its fraction field
with respect to some discrete valuation is called a discrete valuation ring (DVR).
Let us now assume that A is a discrete valuation ring. Any element π ∈ A for which
v(π) = 1 is called a uniformizer. Such a uniformizer necessarily exists, since v maps A
surjectively onto Z≥0 . If we fix a uniformizer π, every x ∈ k × can be written uniquely as
x = uπ n
where n = v(x) and u = x/π n ∈ A× and uniquely determined. Thus A is a unique
factorization domain (UFD), and in fact a principal ideal domain (PID). Indeed, every
nonzero ideal of A is equal to
(π n ) = {x ∈ A : v(x) ≥ n},
for some integer n ≥ 0. Moreover, the ideal (π n ) depends only on n, not the choice of
uniformizer π: if π 0 is any other uniformizer its unique representation π 0 = uπ 1 differs
from π only by a unit. It follows that the ideals of A are totally ordered (with the same
order type as Z≥0 ), and the ideal
m = (π) = {x ∈ A : v(x) > 0}
is the unique maximal ideal and only nonzero prime ideal of A. Rings with a unique maximal
ideal are called local rings.
Definition 1.11. The residue field of a discrete valuation ring A with unique maximal
ideal m is the field A/m.
We can now see how to determine the valuation v corresponding to a discrete valuation
ring A. Given a discrete valuation ring A with unique maximal ideal m = (π), for any
nonzero x ∈ A we must have v(x) = n, where n is the least integer for which x ∈ (π n )
(note that (π 0 ) = (1) = A, so such an n exists and is nonnegative); the integer n does
not depend on the choice of the uniformizer π. Defining v(0) = ∞ and extending v to
the fraction field of A via v(a/b) = v(a) − v(b) gives a discrete valuation v on k for which
A = {x ∈ k : v(x) ≥ 0} is the corresponding valuation ring.

18.785 Fall 2016, Lecture #1, Page 4

Example 1.12. For the p-adic valuation vp : Q → Z ∪ {∞} we have the valuation ring
na o
Z(p) := : a, b ∈ Z, p6 | b ,
b
with maximal ideal m = (p); this is the localization of the ring Z at the prime ideal (p).
The residue field is Z(p) /pZ(p) ' Z/pZ ' Fp .

Example 1.13. For any field k, the valuation v : k((t)) → Z ∪ {∞} on the field of Laurent
series over k defined by  
X
v an tn  = n0 ,
n≥n0

where an0 6= 0, has valuation ring k[[t]], the power series ring over k. For f ∈ k((t))× , the
valuation v(f ) ∈ Z is the order of vanishing of f at zero. For every α ∈ k one can similarly
define a valuation vα on k as the order of vanishing of f at α by taking the Laurent series
expansion of f about α.

1.5 Discrete Valuation Rings

Discrete valuation rings are in many respects the nicest rings that are not fields (a DVR
cannot be a field because its maximal ideal m = (π) is not the zero ideal: v(π) = 1 6= ∞).
In addition to being an integral domain, every discrete valuation ring A enjoys the following
properties:

• noetherian: every increasing sequence I1 ⊆ I2 ⊆ · · · of ideals in A eventually stabilizes;

equivalently, every ideal is finitely generated.
• principal ideal domain: not only is every ideal finitely generated, every ideal can be
generated by a single element.
• local : A has a unique maximal ideal m.
• dimension one: the height of a prime ideal p is the supremum of the lengths n The
(Krull) dimension of a ring R is the supremum of the lengths n of all chains of prime
ideals p0 ( p1 ( · · · ( pn (which need not be finite, in general). If A is a DVR then
dim A = 1, since (0) ⊆ m is the longest chain of prime ideals in A.
• regular : The dimension of the A/m-vector space m/m2 is the dimension of A. When
A is not already local, this applies to its localizations at prime ideals.
• integrally closed (or normal ): A contains every element of its fraction field that is
integral over A (the root of a monic polynomial in A[x]).
• maximal : There are no intermediate rings strictly between A and its fraction field.

There are several different combinations of these properties that uniquely characterize
discrete valuation rings (and hence may be taken as alternative definitions).

Theorem 1.14. For an integral domain A, the following are equivalent:

1. A is a DVR.
2. A is a PID with a unique nonzero prime ideal.
3. A is an integrally closed noetherian local ring of dimension one.

18.785 Fall 2016, Lecture #1, Page 5

 4. A is a regular noetherian local ring of dimension one.
5. A is a noetherian local ring whose maximal ideal is nonzero and principal.
6. A is a maximal noetherian ring of dimension one.
Proof. See [1, §23] or [2, §9].

1.6 Integral extensions

Integrality plays a key role in number theory, so it is worth discussing it in more detail.
Definition 1.15. Given a ring extension A ⊆ B, an element b ∈ B is integral over A if is
a root of a monic polynomial in A[x]. The ring B is integral over A if all its elements are.
Proposition 1.16. Let α, β ∈ B be integral over A ⊆ B. Then α + β and αβ are integral
over A.
Proof. Let f ∈ A[x] and g ∈ A[y] be such that f (a) = g(b) = 0, where

f (x) = a0 + a1 x + · · · + am−1 x + xm ,
g(y) = b0 + b1 y + · · · + bn−1 y + y n .

It suffices to consider the case

A[x, y]
A = Z[a0 , . . . , am−1 , b0 , . . . , bn−1 ], and B= ,
(f (x), g(y))
with α and β equal to the images of x and y in B, respectively, since given any A0 ⊆ B 0
we have homomorphisms A → A0 defined by ai → ai and bi → bi and B → B 0 defined by
x 7→ α and y 7→ β, and if x + y, xy ∈ B are integral in A then α + β, αβ ∈ B 0 must be
integral in A0 .
Let k be the algebraic closure of the fraction field of B, and let α1 , . . . , αm be the roots
of f in k and let β1 , . . . , βn be the roots of g in k. The polynomial
Y

h(z) = z − (αi + βj )
i,j

has coefficients that may be expressed as polynomials in the symmetric functions of the αi
h ∈ A[z], and
and βj , equivalently, the coefficients ai and bj of f and g, respectively. Thus Q
h(x+y) = 0, so x+y is integral over A. Applying the same argument to h(z) = i,j (z−αi βj )
shows that xy is also integral over A.

Definition 1.17. Given a ring extension B/A, the ring à = {b ∈ B : b is integral over A}
is the integral closure of A in B. When à = A we say that A is integrally closed in B. For
a domain A, its integral closure (or normalization) is its integral closure in its fraction field,
and A is integrally closed (or normal ) if it is integrally closed in its fraction field.
Proposition 1.18. If C/B/A is a tower of ring extensions in which B is integral over A
and C is integral over B then C is integral over A.
Proof. See [1, Thm. 10.27] or [2, Cor. 5.4].

Corollary 1.19. If B/A is a ring extension, then the integral closure of A in B is integrally
closed in B.

18.785 Fall 2016, Lecture #1, Page 6

Proposition 1.20. The ring Z is integrally closed.

Proof. We apply the rational root test: suppose r/s ∈ Q is integral over Z, where r and s
are coprime integers. Then
 r n  r n−1 r
+ an−1 + · · · a1 + a0 = 0
s s s
for some a0 , . . . , an−1 ∈ Z. Clearing denominators yields

rn + an−1 srn−1 + · · · a1 sn−1 r + a0 sn = 0,

thus rn = −s(an−1 rn−1 + · · · a1 sn−2 r + a0 sn−1 ) is a multiple of s. But r and s are coprime,
so s = ±1 and therefore r/s ∈ Z.

Corollary 1.21. Every unique factorization domain is integrally closed.

Proof. The proof of Proposition 1.20 works for any UFD.

√
Example √ 1.22. The ring Z[√ 5] is not a UFD because it is not integrally closed:
√ consider
φ = (1 + 5)/2 ∈ Frac Z[√ 5], which √ is integral over Z (and hence over Z[ 5]), since
2
φ − φ − 1 = 0. But φ 6∈ Z[ 5], so Z[ 5] is not integrally closed.

Definition 1.23. A number field K is a finite extension of Q. The ring of integers OK is

the integral closure of Z in K.

Remark 1.24. The notation ZK is also sometimes used to denote the ring of integers of K.
The symbol O emphasizes the fact that OK is an order in K; in any Q-algebra K of finite
dimension r, an order is a subring of K that is also a free Z-module of rank r, equivalently,
a Z-lattice in K that is also a ring. In fact, OK is the maximal order of K: it contains
every order in K.

Proposition 1.25. Let A be an integrally closed domain with fraction field K. Let α be
an element of a finite extension L/K, and let f ∈ K[x] be its minimal polynomial over K.
Then α is integral over A if and only if f ∈ A[x].

Proof. The reverse implication is immediate: if f ∈ A[x] then certainly α is integral over A.
For the forward implication, suppose α is integral over A and let g ∈ A[x] be a monic
polynomial for which g(α) = 0. In K[x] we may factor f (x) as
Y
f (x) = (x − αi ).
i

For each αi we have a field embedding K(α) → K that sends α to αi and fixes K. As
elements of K we have g(αi ) = 0, so each αi ∈ K is integral over A and lies in the integral
closure Ā of A in K. Each coefficient of f ∈ K[x] can be expressed as a sum of products
of the αi , and is therefore an element of the ring Ā that also lies in K. But A = Ā ∩ K,
since A is integrally closed in its fraction field K.

18.785 Fall 2016, Lecture #1, Page 7

References
[1] Allen Altman and Steven Kleiman, A term of commutative algebra, Worldwide Center
of Mathematics, 2013.

[2] Michael Atiyah and Ian MacDonald, Introduction to commutative algebra, Addison–
Wesley, 1969.

[3] Pierre Deligne, La conjecture de Weil: I , Publications Mathématiques l‘I.H.É.S. 43

(1974), 273–307.

[4] Jean-Pierre Serre, Local fields, Springer, 1979.

[5] André Weil, Numbers of solutions of equations in finite fields 55 (1949), 497–508.

18.785 Fall 2016, Lecture #1, Page 8