Professional Documents
Culture Documents
1. God created the integers and all else was the work
of man
I remember that this famous quote by Kronecker made no sense to
me when I saw it on the pages of my Math 112 notes. It seemed out
of place in a world of limits, complex numbers and power series, not
to mention how atheistic it sounded. For me it took several years to
get a coherent picture of what Kronecker probably had in mind. In
this handout I attempt to explain a tiny bit of ”the work of man” by
exhibiting a special case of the famous Kronecker-Weber Theorem.
Of course, we have to start with what God has given us, namely the
integers. That seems fairly easy, but soon one realizes that, as Jerry
Shurman once put it, ”The integers are simply too small.” Even worse,
the field of fractions of the integers, the rationals, is essentially as small
as the integers. Even simple algebraic relations like x2 − 2 = 0 are
not satisfied by any rational number. But man, being the inquisitive
creature that its nature entails, found a way around this by imagining
that there were rational number with unusual properties. Moreover
these new ”rationals” had ”integers” of their own. Not surprisingly
man’s imagination was outrunning his means and he soon realized that
there was no use of all that. He had created something too big, in
fact immeasurably bigger than the numbers that God had given him,
so he had to rethink his approach and focus on what was manageable
and finite. So he created Algebraic Number Theory. But, as it often
happens, the new numbers that man created had questions of their
own.
When any of the above conditions holds, we will say that P lies over
p and p lies under P .
THEOREM: Every prime P of LZ lies over a unique prime p of
FZ ; every prime p of FZ lies under some prime P of LZ .
The following definitions and theorem are crucial (note that they
make sense because of the previous theorem):
DEFINITION: Suppose that p = gi=1 Piei . Then
Q
In the case of L/F being normal (and therefore Galois, since we are
working over Q and consequently separability is not an issue) we can
say even more.
TRANSITIVITY OF THE GALOIS ACTION ON IDEALS:
If P and P 0 are any prime ideals of LZ such that P ∩ FZ = P 0 ∩ FZ 6= ∅
then there exist σ ∈ G such that σ(P ) = P 0 .
Proof: Let G = {σ1 , . . . , σn } = Gal(L/F ). Assume that P 0 6=
σi (P ) for every σi ∈ G. Then, since σi (P ) is a prime ideal, we can use
the Chinese Remainder Theorem to find an x ∈ LZ , such that
x≡0 mod P 0
x≡1 mod σi (P ), i = 1, . . . , n.
Let
n
Y
a= σi (x),
i=1
After seeing Galois Theory do all the work for us we need to define
an important invariant of a number field called the discriminant.
Let σ1 , . . . , σn be the automorphisms of L. For any n-tuple of ele-
ments a1 , . . . , an ∈ L define the discriminant (δ) of a1 , . . . , an to be the
square of the determinant of the matrix with ij th entry σi (aj ), i.e.
δ(a1 , . . . , an ) = |σi (aj )|2 .
THEOREM: For any number field L there exist element a1 , . . . , an ∈
LZ that form both a basis for L over Q and a basis for LZ over Z.
THEOREM: For any number field L the discriminant of an integral
basis is invariant under a change of integral basis and is called the
discriminant of L and denoted by δL .
It is not too difficult to see that δL ∈ Z. Finally, we come to an
important theorem, which will be used repeatedly:
THEOREM: Let p be a prime in Z then p ramifies in a number
ring LZ ⇐⇒ p | δL .
With this our summary of facts will end, but the reader who wants
to learn more about other questions and developments in Algebraic
Number Theory can read about them in [Mar] and [Rib] .
where |Hi | is a power of a prime (not necessarily the same one) for all
i’s. For any i, such that 1 ≤ i ≤ s define
M
Li = Hj .
j6=i
thus L = L1 · · · Ls .
Now assuming that the theorem is true for all the Li ’s we may write
Li ⊆ Q(ξi ),
where ξi is a primitive root of unity. Let ζ be a primitive root of unity
of order equal to the least common multiple of the orders of the ξi ’s.
Then
L = L1 · · · Ls ⊆ Q(ξ1 , . . . , ξs ) ⊆ Q(ζ).
And this finishes the proof of the reduction step.
We will show an easy special case of the theorem when the degree
of the extension and the discriminant are a power of the same odd
rational prime. Even in this case we will make some assumptions,
which are theorems in their own right. As usual we will refer the
reader to [Mar] of [Rib] for the proofs. From now on p will denote an
odd prime number of Z.
CASE 1: If [L : Q] = pm , δL = pk , where m, k ≥ 1, then the
theorem holds.
KRONECKER-WEBER FOR KIDS 7
0
Lc ;R
L ∩O R0
Q
It can be shown that LR0 /Q is a cyclic extension, so assume that, then
the subgroup of Gal(LR0 /Q), Gal(LR0 /L ∩ R0 ) is also cyclic. But by
Galois Theory
Gal(LR0 /L ∩ R0 ) ∼
= Gal(L/L ∩ R0 ) × Gal(R0 /L ∩ R0 ).
Thus one of the groups on the right is trivial, i.e. L = L ∩ R0 or
R0 = L ∩ R0 .
If L = L ∩ R0 then L ⊆ R0 , if R0 = L ∩ R0 then R0 ⊆ L, but both
have the same degree over Q and thus in either case
L ⊆ R0 ⊆ Q(ζ).
for some odd primes pi and m, k ≥ 1 then there exists a primitive root
of unity ζ, such that L ⊆ Q(ζ).
8 KRONECKER-WEBER FOR KIDS
while if 1 + t = 0 then
X
ξpx(1+t) = 1 + 1 + · · · + 1 = (p − 1).
x
The last thing to notice is that since there are an equal number of
P that have (∗/p) equal to 1 as there are with (∗/p) = −1, we
elements
have t (t/p) = 0.
Returning to the calculation of τ 2 this gives
X t
2 −1
τ = (p − 1) + (−1)
p −16=t
p
−1 X t
= p−
p t
p
−1
= p−0
p
= ±p.
√ √ √
Thus ±p ∈ Q(ξp ). Which immediately gives p ∈ Q(ξp ) or p ∈
√
Q( −1, ξp ) ⊆ Q(ξ8 , ξp ).
Combining everything that we have found out so far we get
√
L = Q( d) ⊆ Q(ξ8 , ξp1 , . . . , ξpr ) ⊆ Q(ζ),
where ζ is a primitive root of unity of degree m = 8p1 · · · pr .
We have shown an improved version of our mini-Kronecker-Weber:
10 KRONECKER-WEBER FOR KIDS