Everything2
Near Matches
Ignore Exact
Full Text
Everything2

Proof of Wedderburn's theorem about finite division rings

created by Noether

(idea) by Noether (2.9 y) (print)   ?   (I like it!) Tue Aug 08 2000 at 13:38:58

See Wedderburn's theorem about finite division rings for the statement of the theorem. The rather beautiful proof here is due to Witt (from 1931).

Let F be our finite division ring. Let k be its centre. Note that k is a finite field, as such is has q elements, for some power of a prime number q. Consider F as a vector space over k, say it has dimension n. It follows that F has qn elements. Consider now the multiplicative group U(F) of units in F. Because F is a division ring, U(F) has qn-1 elements.

Next we consider the conjugacy classes of the elements of U(F). An element of U(F) in the centre has a singleton conjugacy class. So consider a in U(F) that is not central. The centralizer C of a in F is a subring of F and obviously it contains k. Further, since a is not central, C is not equal to F. It follows from the argument in the first paragraph that C has qd elements. Because C is itself a division ring we have that qd-1=|U(C)| is a divisor of qn-1. It follows quickly that d divides n.

Now C(a) the centralizer of a in U(F) consists of the nonzero elements in C (and so has qd elements) and we know that the conjugacy class of a in U(F) has |U(F)|/|C(a)| elements. Since the conjugacy classes of U(F) give a partition of U(F) we get a formula

qn-1 = q-1 + Sumd (qn-1)/(qd-1) (*) where the sum is over various proper divisors d of n and it could be that some ds occur more than once. (Note that the q-1 comes from the central elements in U(F).)

Now we will make use of cycn(x) the nth cyclotomic polynomial. It is a factor of any (xn-1)/(xd-1) for a proper divisor d of n. So if we evaluate at q we get that cycn(q) divides each (qn-1)/(qd-1). For the same reason it also divides qn-1. So examining (*) we have that cycn(q) | q-1.

By the way the cyclotomic polynomial is defined cycn(q) is the product of (q-e) where e varies over the primitive nth roots of unity in the complex numbers.

Suppose that n>1. We are going to obtain a contradiction from this. This will show that n=1 and so F=k is a field. Since n>1 then |q-e| > q-1 (draw a picture to see this) and so we see that |cycn(q)| cannot possibly divide q-1. This contradiction establishes the proof of the theorem.


printable version
chaos

Wedderburn's theorem about finite division rings cyclotomic polynomial conjugacy class proof of the third isomorphism theorem
ring theory primitive field centralizer
Centre division ring finite set 1
mathematics complex number Polynomial divides
Singleton unit multiplicative group dimension
finite field proof theorem
Y'know, if you log in, you can write something here, or contact authors directly on the site. Create a New User if you don't already have an account.
  Epicenter
Login
Password

password reminder
register

Everything2 Help

Cool Staff Picks
Little presents from the Node Fairy:
Jazz Pianists
How to love someone who is mentally ill
On finding a wife
Koala
Zazen
Guidebook of Colored Ribbons
The Michigan PBB disaster
Till Eulenspiegel
throat singing
ginger tea
Noah Webster
Chopstick vs. Fork
Dead people are not sleeping. They are dead.
New Writeups
octillion369
Frost wyrm(person)
kalen
Three "T"s(idea)
octillion369
Undead(idea)
archiewood
Ico(fiction)
Heisenberg
Why I love Everything2(log)
octillion369
Death Knight(person)
XWiz
Are you hoping for a miracle?(review)
santo
The Host(review)
LostPsion
"Shut the Fuck Up" Theaters(idea)
Vanish
The line between normal and not(place)
Vanish
insanity(thing)
beatrice
You've been slowly taking me over for nearly a year, do you know that?(idea)
Berek
YouTube(thing)
shaogo
How to Pretend to Have a Job(idea)
hapax
Les Provinciales(review)
This page courtesy of The Everything Development Company