Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Algebraische Zahlentheorie

Josias Brenner

January 19, 2019


Contents

1 Dedekind Ring 2
1.1 Some recollections . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Finite type A-modules and noetherians . . . . . . . . . . 3
1.2 Integral extension of rings . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Trace and norm if integral extensions . . . . . . . . . . . . . . . . 8
1.3.1 Recollection on trace and det over general rings . . . . . . 8
1.3.2 Recall on Traces, Norms and Galois theory . . . . . . . . 9

1
Chapter 1

Dedekind Ring

1.1 Some recollections


All the rings will be assumed to be commutative and unitary, if not otherwise
stated.
Corollary 1.1. Let A be such a ring. To say A  0 is equivalent to say 1  0
because 1  0 ñ @x P A : x  1  c  0  x  0.
Definition 1.2. Let α P A. We say that α is invertible, if Dβ P A : α  β  1.
A : tα P A | α invertibleu.

A is a group with neutral element 1, called the multiplicative group of A. For


instance: 1) A  0  1 ñ A  0. 2) A ring A is a field if A  0 ^ Azt0u  A .
If p is a prime, then Z{pZ is a filed, denoted Fp .

Definition 1.3 (integral domain). A commutative ring A is an integral domain


(ID), if A  0 and @x, y P A : xy  0 ñ x  0 _ y  0.
Definition 1.4 (field of fractions). Let A be an ID. We consider on A pAzt0uq
the relation px, y q  px1 , y 1 q ô xy 1  x1 y P A. This is an equivalence relation
1
and we denote by K the quotient set. A pAz0q Ñ K, pa, bq ÞÑ ab with ab ab1 
1 1 1 1
ab a b
bb1 and ab  ab1  aa
bb1 gives us the structure of a ring on K. K is a field, called
the fraction field of A and A Ñ K, a ÞÑ a1 is an injective ring homomorphism.
We will consider A € K.

Examples are Z € Q and

Remark 1.5. If a ring A is a subring of a field L (or more generally, if there


exists a ring homomorphism A Ñ L), then A is an integral domain and @a P
Azt0u with a P L the function A  pAzt0uq Ñ L, pa, bq ÞÑ ab  ab1 induces a
field homomorphism K Ñ L from the fraction field K of A into L.

For instance, let A, B be ID with fields of fractions K, L respectively and let

2
ϕ : A Ñ B be an injective ring homomorphism, then we have

ϕ : A ãÑ B
X X
ϕ: K Ñ L.
ã

Definition 1.6 (principal ideal ring). A principal ideal ring (PID) is a integral
domain A such that any ideal I € A is principal, i.e.

@I € A Ideal : Da P A : I  paq  tba | b P Au.


Recall that an ideal I € A is a subset which is a subgroup for and @pa, bq P
A  I : ab P I. Also recall that x  y ô Db P A : y  bx for x, y P A from which
paq  tx P A | a  xu follows.
Examples of PID include Z, K rX s if K is a field and Zris. But ZrX s is not a
PID, which can be seen by the fact, that p2, X q  t2P XQ | P, Q P ZrX su is
not principal.
Remark 1.7. Every euclidean ring A is a PID.

Proof. Let N : A Ñ N and I € A be an Ideal. If I  0 let a P I zt0u with N paq


minimal. If x P I, by euclidean division, Dq, r P A : x  aq r ^ N prq   N paq.
Now r P I ñ r  0 ñ I  paq
Theorem 1.8. Let A be a PID and M be an A-module of finite type. Then
À
there exists exactly one decreasing sequence of ideals A I1      Ir in A
where there exists an isomorphism of A-modules M  i1 A{Ii .
r

Remark 1.9. M is said to be torsion free if @pa, mq P A  M : am  0 in M


implies a  À
0 _ m  0. If M is a finite torsion free A-Module then Ii  0 for all
i and M  i1 A. If for instance M is a sub-A-module of Am then M  Ar
r

for some r ¤ m.
Remark 1.10. If A  Z then the A-modules are the abelian À groups. There
r s
exists exactly one Z-module structure on M . We have M  i1 Z{mi ` Z
s

with I1  pn1 q      Irs  pnrs q  0      0.


Example 1.11. K pX q, K  K ñ Jordan Form.
1.1.1 Finite type A-modules and noetherians
Definition 1.12. Let A be a ring and M ba an A-module. One says that M
is a finite type A-module if
¸
n
Dn ¥ 0 : Dm1 , . . . , mn P M : @m P M : Da1 , . . . , an P A : m  ai mi .

i 1
°
Equivalently Φm1 ,...,mn : An Ñ M, pa1 , . . . , an q ÞÑ i ai mi is onto M of finite
type means that there exists n ¥ 0 and an epimorphism An  M .
Definition 1.13. A is said to be noetherian if any sub-A-module N of a finite
type A-module M is also of finite type. ...

3
Remark 1.14. If A is a PID then A is noetherian.
Definition 1.15. Let A be an integral domain. An element π P A is said to be
irreducible if

π  0 ^ π P A ^ @a, b P A : π  ab ñ a P A _ b P A .

Sometimes if π is not 0 not in A and not irreducible one says π is reducible.


An element π P A is said to be a prime element

π  0 ^ π R A ^ p@a, b P A : π | ab ñ π | a _ π | bq .
Remark 1.16. A prime element is irreducible.
Example 1.17. In Z the prime elements are the irreducible elements which are
the primes and their negatives. In K rX s for a filed K the prime elements are
also the irreducible elements.
Definition 1.18. Two elements x, y P A are said to be associated if Du P
A : y  ux. Association is an equivalence relation.

4
Let A be an integral domain. An element π P A is said to be irreducible if
π  0, π
Definition 1.19. A unique factorial domain (UFD) is an integral domain A
such that any x P A, x  0, x R A can be expressed as a (finite) product of
prime elements.

This implies that the prime elements and the irreducible elements are the same.
The proof is easy: If π is an irreducible element, then π  p1  . . .  pr and the
property of being irreducible implies that r  1. Let Irr € A be the subset of
irreducible elements. Let P € Irr be a subset such that P Ñ Irr{A (= equiv-
alent classes of associated irreducible elements) is a bijection. In other words,
for any class of associated irreducible elements, we choose one representative. If
follows:
@x P A, x  0 : D!pVp pzqqpPP : @p P P : Vp pzq P N ^ V pzq  0 expect finitly many.
¹ ¹
Du P A : x  u pVp pxq u pVp pxq .
P
p P P pq
p P,Vp x 0

Example: In Z one takes P as the set of prime numbers. In K rX s for a field K


one takes P as the set of irreducible and monic polynomials.
Proposition: A principal ideal domain is a unique factorial domain.
Lemma: Let A be a PID and let π P A. We have
π irreducible ô A{pπq integral domain ô A{pπq field
Remark: To say that π is a prime element in A is equivalent to say that A{pπ q
is an integral domain, because
π  ab ô ab  0 ô a  b  0 ô a  0 _ b  0.
If follows from the lemma, that in a PID the prime elements are the irreducible
elements. Sketch of Proof: (i) ñ (iii). Let α P A, α P A{pπ q  t0u. α  0
means π  α. We consider the ideal in A generated by π and α. We have
pπq € pπ, αq € A. Now pπ, αq  pβ q for some β, thus π  βγ for some γ. It
follows, that either β or γ is a unit. If γ is a unit, pπ q  pβ q  pπ, αq ñ α P pπ q,
which is a contradiction. Thus β have to be a unit. But then pπ, αq  pβ q  A
which implies Da, b P A : 1  πa αb form which
1  ab  a  b P A{pπ q ñ α P pA{pπ qq
follows.
End of the proof PID ñ UDF: Let x P A, x  0, x R A . Either x P Irr or x R Irr
implies Da1 , b1 P A : x  a1 b1 with a1 , b1  0 and a1 , b1 R A . either a1 , b1 are
irreversible or not:  a2 b2 b1      an  bn  . . .  b1 . If the process doesn’t
stop @n : an  an1 , pan”1 q ˆ pan q and thereby pa1 q ˆ pa2 q ˆ . . . . This is not
possible because pβ q  :n pan q is an ideal, but β P pan0 q ˆ pan0 1 q € pβ q. It
follows, that the product of prime elements is finite.
Remark: In a UFD one says that a, b are prime together, if tπ P Irr | π  a ^ π 
bu  tu.
gcdpa, bq  1

5
in a PID:

gcdpa, bq  1 ô pa, bq  A ô Du, v P A : ua vb  1

An other general example of UFD. Theorem: If A is a UFD, then so is ArX s. Ex-


ample: ZrX s is a UFD, but also ZrX1 , . . . , Xn s is a UFD. Remark: A polynomial
p P ArX s is called primitive if the set tπ P IrrpAq | π divides each coefficient of pu
is empty.
Gauss Lemma: P, Q primitive ñ P Q is primitive. Proof: π P IrrpAq

π  P Q ô P Q  0 in A{pπqrX s ô P  0 _ Q  0.
Let K be the fraction field of A.

IrrpArX sq  IrrpAq Y tP P Ar X s | P primitive and irreducible in K rX su

With our conventions, a filed is a PID and a UFD. If A is PID but not a field,
then ArX s is UFD but not a PID.

1.2 Integral extension of rings


We are going to study extension of rings A € B such that any element of B is
integral over A.
Definition 1.20. An element x P B is said to be integral over A, if there exists
a monic polynomial P P ArX s with P pxq  0.

Remark: In general, if A € B is an extension of rings, AB  tx P B |


x is integral over Au is always subring.

6
Corollary 1.21. Let A € B be a ring extension. 1) Let x1 , . . . , xr be elements
of B, each integral over A, then Arx1 , . . . , xr s is integral over A. 2) The AB 
tx P B, integral over Au is a subring of B. (@x, y integral | A, Arx, ys is
integral over A by 1)). Proof of 1): Let @i P t1, . . . , ru, let Pi P ArX s, monic
| Pi pxi q  0. Now Arx1 , . . . , xr s is the image of:

evx1 ,...,xr : Ar x 1 , . . . , x r s  Ar x 1 , . . . , x r s € B
P ÞÑ P px1 , . . . , xr q.
By assumption this homomorphism factorizes through

Arx1 , . . . , xr s  Arx1 , . . . , xr s{pP1 px1 q, . . . , Pr pxr q Ñ B q


pArX1 s{pP1 qqrx2 , . . . , xr s{pP2 px2 q, . . . , Pr pxr qq      ppppArx1 s{pP1 qqrx2 s{pP2 qqrx3 s{pP3 qq . . . rxr s{pPr sq
Observation: If R is a commutative ring and P P Rrxs a monic polynomial of
n1
degree n. Rrxs{pP q is a free R-module with basis p1, X, . . . , X q. It follows
by induction that Arx1 , . . . , xr s{pP1 px1 q, . . . , Pr pxr qq is a free A-module of rk
n1 , . . . , nr with ni  degPi with basis the classes of xs1 , . . . , xsr with @i : si ¤
1 r

ni1 . It follows that the image of evx1 ,...,xr is a ft A-module generated by the
(finite) family: xs11 , . . . , xsrr P B with @i, si ¤ ni1 . This Arx1 , . . . , xr s € B is a
finite A-module, and is integral {A by theorem 1.
Remark 1.22. One may also prove directly that the x1 , . . . , xr , si ¤ ni1
generate Arx1 , . . . , xr s as an A-module.
Example 1.23. Let Q € K be a number field. Then ZK is a subring of K,
denoted by OK ; the ring of integers over K.
? ?
Example 1.24. Q € K  Qr ds, d P Z square free. α : d.
OK  Z ` Zpαq when d  2, 3r4s
 Z ` Zpβ q with β  1 2 α if d  1r4s. In the first case: OK  Zrxs{px2  dq. In
the second case the minimal polynom of β is x2  x  d 4 , O :K  Zpxq{px 
1 2

x  d1
4 q.

Corollary 1.25. Let A € B € C be ring extensions. Assume B is integral over


A and C is integral over B, then C is integral over A.
Remark 1.26. It follows that if A € B is a ring extension pAB qB  AB .
Indeed we have: A € AB „ pAB qB € B
Definition 1.27. Let A be an integral domain with fraction filed K. One says
that A is integrally closed if A  AK € K.
Example 1.28. Z is integrally closed. Let x  pq P Q, with p, q P Z, q 
0, gcdpp, q q  1. Let P  xn an1 xn1    a0 P Zrxs{pP pxqq  0 by
q n  pP pxq  0q : pn q  an1 pn1    q n a0  0 ñ q | pn ñ q P t1u.
Example 1.29. more generally if A is a UFD then A is integrally closed.
?
Example 1.30. Let d P Z square free. and α  d?P C. Z € Zrαs is integrally
closed. Q € Qrαs only if d  1r4s for instance Zr 5s  Zrxs{px2  5q is not
integrally closed.

7
Example 1.31. Let A be an integral domain, K be its fraction field, then AK
is integrally closed (in K). A € AK € K and we have pAK qK  AK . More
generally, if A € K € L is a filed extension, then AL is integrally closed. paper

1.3 Trace and norm if integral extensions


1.3.1 Recollection on trace and det over general rings
Let A be a commutative ring. Let A be an A-module free of finite type (ô
M A An ) and let f : M Ñ M be an endomorphism of A-module. One may
define Trpf q P A, detpf q P A, χf P Arxs as follows. Let B  e1 , . . . , en be a basis
of M . This means:
à
n
φB : An p Aq Ñ A

i 0
¸
n
pλ1 , . . . , λn q ÞÑ λ i ei

i 1

is an isomorphism. Le M  MB pF q P Mn pAq, @m P M with coordinates


x  φ 1
B pmq. Let y  φB pf pmqq, then y  M x.
1

Remark 1.32. n is independent of the choice of B. Because if A  0 there


exists a nontrivial ring homomorphism A Ñ K  A{µ with µ the maximal
ideal. M  A° n
ÞÑ A{µM  K n . °M {µM is±an K-vector space. We claim
that TrpM q  i1 ai,i , detpM q  oPOn poq i1 ai,opiq P A and χM pX q 
n
° ±n
detpXIdn  M q  oPOn poq i1 pXδi,opiq ai,opiq q P ArX s are independent of
the choise of B. Indeed let C be another Basis, φb : An Ñ M , TC,B : An Ñ An ,
φC : An Ñ M , N  MC pf q ñ N  T  M  T 1 it follows that TrpN q 
TrpM q P A and so on.
Definition 1.33. Let A € B be a ring extension with B free of finite type
A-module (for instance Z € OK ). @x P B : φx : B Ñ B, y ÞÑ xy is an
A pxq, detpφx q 
endomorphism of A-modul. Trpφx q  TrB
Let A € B be an extension of rings with B a free A-module of finite type (of
rang r). Morphism of A-module:
TrB
A :B ÑA
A rφx : B Ñ B, y ÞÑ xy s  TrA pxq
x ÞÑ TrB B

NAB :B  Ñ A
x ÞÑ detpφx q  NAB pxq
@x P B : χx  χφ  detpXIdB  φx q P ArX s
x

is monix of degree r.
Example 1.34. If K € L is a finite field extension this gives the ususal con-
structions:

K :L Ñ K, K  linear
TrL
L
NK :L Ñ K  , group homomorphism
χx P K rX s

8
For insance if A € K is a PID with fraction field K, K € L separable ñ B  AL
contains A, if a free A-module of rk: rL : K s (=finite).

ñ TrLK |B  TrBA : B Ñ A
L
NK |B  NAB : B  Ñ A
NKL
|Bt 0u : B  t0u Ñ A  t0u
Example 1.35. K € L / rL : K s  2 quadratic extension. We assume Dα P
LzK | α2  a P K.

L  K rαs  K rX s{pX 2  aq € K

L  K1 ` Kα, x  u αv, u, v P K, p1, αq is a basis of L{K, φx : L Ñ L.




Mpa,αq pφx q 
u va
v u

K pxq  2a, NK pxq  u  av


TrL L 2 2

1.3.2 Recall on Traces, Norms and Galois theory


Assume K € L is a finite separable extension of fields. x P L is separable ô P ,
the minimal polynomial of x{K, is separable which means that gcdpP, P 1 q  1
±d
which is equivalent to P P K, P  i1 pX  xi q with xi  xj
Theorem 1.36. Dx P L | L  K rxs. If P is the minimal polynomial of x{K

then K rxs{pP q Ñ L.

We choose K P K an algebraic closure of K. We consider


SK pLq : HomK extension pL, K q.

So for φ P SK pLq we have φ : L ãÑ K


Lemma 1.37. If K € L is finite and separable, |SK pLq|  rL : K s
Proof. Of course SK pLq  SK pK rxs{P qq. We choose x P L positive. Now with
P pxq  P pxq  0

SK pK rxs{pP qq Ñ NP pK q € K
pφ : K rxs{pP q Ñ K q ÞÑ φpxq  φpxq

NP pM q  tx P M | P pxq  0u


Claim: SK pK rxs{pP qq Ñ NP pK q is a bijection. proof...
 
P being separable ñ NP pK q  deg P  rL : K s

9
Theorem 1.38 (Theorem 2). Let K € L be a finite separable extension. n 
rL : K s, SK pLq  tσ1 , . . . , σn u. Then for all x P L
¹
n
χx  pX  σi pxqq

i 1
°n ±n
K pxq 
In particular TrL  σi pxq and NK pxq   σi pxq.
L
i 1 i 1

Warning: For a given x P L, the σi pX q are not necessarily distinct. In fact this
is equivalent to say x is primitive.

±n L  K rxs  K rxs{pP q with P the±


Proof. Let x be primitive. minimal polyno-
mial of x. Then O  i1 pX  σi pxqq in K rX s because P  yPNP pK q px  y q
and

SK pK rX s{φq Ñ NP pK q
σy ÞÑ σp xq  y
Now p1, x, . . . , xn1 q is a basis of L{K and

0 0 ... 0 a0
1 0 . . .
 0 a1 

M1,...,xn1 pφx q  0 1 . . .
 0 a2 

. . . .. .. 
 .. .. .. . .
0 0 ... 1 an1
P  Xn an1 X n1  a0 .

X 0 0 ... 0 a0
1 X 0 ... 0 a1 
 
 .. .. 
χx  χφ 
x

 0
..
.
..
.
..
. . . 

 .. .. .. .. .. .. 
 . . . . . .
0 ... ... 1 X an1

Let x be not primitive. –195–


Theorem 1.39 (Theorem 3). Let A € K be an integral domain with fraction
field K. Let K € L be a finite field extension. Let x P L be integral over A.
Then the coefficients of χx integral over A we have χx has each of its coefficient
in AK . In particular if A is integrally closed (in K), then χx P ArX s.
Remark 1.40. χx pxq  0 by Cayley-Hamthon.

Proof. ...
Remark 1.41.

RrΣ1 , . . . , Σn s ãÑ RrX1 , . . . , Xn sσn
Σi ÞÑ Σi pX1 , . . . , Xn q

10
Case 2: General case: K € Ksep € L with the subfield Ksep  tx P L |
x separable over K u, the separable closure of K in L. Set p  char k ¡ 0 Ksep €
L is a purly inseparable extension. It means that @x P L : Dm ¥ 1 : xp P Ksep .
m

In fact rL : Ksep s  pm0 and @c P L, xp P Ksep . (Recall that if P P K rX s,


m0

irreducible, D!m P N, D!Q P K rX s : P pX  QpX p qq and Q is separable and


m

irreducible)
Let M P Mn pK q and χM  detpX  Id  M q P K rX s

Xp
m
 Idn  M p  pX  Idn  M qp
m m

(In any not necessarily commutative ring R of characteristic p Z{p P R : ap bp 


pa bqp if ab  ba.)
pchiM qp  detpχp  Idn  M p q
m m m

pχM qp  χM pX p q m
pm
m

M  matrix of φx : L Ñ L ñ pχL xq  χLx pX p q Now if xp P Ksep € L


m m m
p
pm

we have χx pX q  pχx pX qqrL:K s because K € Ksep  M € L, y P M


K sep sep
pm pm
Àr 
y  pχy q
χL . Choose a basis s1 , . . . , sr of L{M we have i1 M Ñ L. With
M L:M

choose m  m0 now

pχLx qp  χLx pX p q  pχKx pX p qqp


m0
pm0
m0
sep
pm0
m0 m0

ñ χLx 0χKx pX p qpas K rX s Ñ K rX s, P ÞÑ P p is injectiveq


sep
pm0
m0
ã

Remark 1.42. Global fields:

1. number field: Q € K finite.


2. function field: Fp pX q € K finite.

Now let A be an integrally closed domain with fraction field K i.e. A is an


integral domain and AK  A. Let K € L be an finite filed extension. Let
B : AL . Of course we have A € B. If K, L are number fields we have

Q€ K€ L
Y Y Y
Z€ OK € OL  OK L

TrL
K : LÑ K K-linear map
Y Y
induces TrB
A : B Ñ A a homomorphism of A-modules

L
NK : L Ñ K
Y Y
induces NAB : B Ñ A group homomorphism

We don’t need and we don’t know that B is a free A-module of finite type.

11
Remark 1.43. If x P L, NK L
pxq  detpφx q we have NKL pxq  0 ô x P L . If
x P B zt0u we have NAB pxq P A ô x P B  .

Proof. If x P B zt0u, NAB px P A q. χx  X n  TrBA pxq  X n1 an2 X n2


   a1 X p1qn NAB pxq. χx pxq  0 ...
Remark 1.44 (Roots of unity). Let K be a field,

µpK q  tx P K | Dn ¥ : xn  1u € K  .
µpK q € K  is a subgroup, the (group of) roots of unity.

@n ¥ 1 : µn pK q  tx P K | xn  1u € µpK q.
Observe: If K € K is a finite extension
L
NK :L Ñ K
L
NK |µpLq :µpLq Ñ µpLq
If A € K, mt.? closed then any root of unity is integral over A, so µpK q € A.
Example: µn pCq  Z{nZ.
Definition 1.45 (Discriminant). Let A € B be and ring extension, with B
a free A-module of rank n ¥ 1. Let x1 , . . . , xn P B. The discriminant of
px1 , . . . , xn q, with respect to A € B, is:
B
DA px1 , . . . , xn q  detpTrBA pxi xj qi,j q P A
Lemma 1.46. Let A € B as above. Let x1 , . . . , xn , y1 , . . . , yn P B. Let M P
Mn pAq and y  M x. Then DA B
py1 , . . . , yn q  pdetpM qq2 DAB px1 , . . . , xn q. Now
if x, y are both basis, then D!M : yM x, M P Gln pAq, that is detpM q P A .
B
Definition 1.47. DA B
is the principal ideal of A generated by DA pxq for x a
basis of B {A.

12

You might also like