Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Is GL(n,F) dense in Zariski topology

98 views
Skip to first unread message

Gary McGuire

unread,
Jul 3, 2000, 3:00:00 AM7/3/00
to
Is GL(n,F) dense in M_n(F) in the Zariski topology
over an arbitrary field F?

Here,
M_n(F) = algebra of all n-by-n matrices over F.
GL(n,F)= invertible elements of M_n(F).
Let m=n^2 and identify M_n(F) with F^m.
The Zariski topology is defined by the closed sets,
which are the sets of common zeros of finite
sets of polynomials in F[x_1,...,x_m]

I think the following is a proof for infinite fields,
but what happens for finite fields?

GL(n,F) is dense if it meets every open set.
Let U be an open set in F^m and suppose

U intersect GL(n,F) = empty set.

This is equivalent to

U' union GL(n,F)' = F^m (*)

where ' denotes set complementation.
As U' is closed it is the set of common
zeros of polynomials g_1,...,g_r.
(Assume U is not M_n(F) so r>=1.)
Also, GL(n,F)' is the set of zeros of
the single polynomial det.
Therefore, (*) says that F^m is the set of
common zeros of the products
det*g_1, ... ,det*g_r .

Now we invoke the fact that over an infinite field,
the only polynomial in F[x_1,...,x_m]
that vanishes on all of F^m
is the zero polynomial,
to obtain a contradiction.

But over finite fields, this fact is not true
(x^q-x vanishes on the field of q elements).

Does anyone know what happens here?

-Gary McGuire

Daniel Giaimo

unread,
Jul 3, 2000, 3:00:00 AM7/3/00
to

"Gary McGuire" <g...@maths.may.ie> wrote in message
news:rgk85t...@forum.mathforum.com...

> Is GL(n,F) dense in M_n(F) in the Zariski topology
> over an arbitrary field F?

This is clearly false for finite fields by the following argument:
Point-sets are clearly closed in the Zariski topology as the point
(a_1, ..., a_m) is the one and only one simultaneous solution to the
equations x_1 - a_1, ..., x_m - a_m. Therefore, since the set of
matrices is finite, the topology is the discrete topology. Hence
the only dense set is the whole space. Hence GL(n,F) is not dense as
0 is not invertible.

--Daniel Giaimo

Clarence Wilkerson

unread,
Jul 4, 2000, 3:00:00 AM7/4/00
to
At the risk of egg on my face, try this argument. Let
G = GL(N,F), M= etc. If the points of G are
dense in M in the Zariski topology, then each polynomial f in
the variables (x_{ij} ) which vanishes on all of G must also
vanish on all of M.

But, if F is a
finite field with p^n elements, then each element g in GL(N,F)
satifies the equation det(g)^{p^n -1} = 1, which no
non-invertible element in M satifies. In particular, it fails
for m = the zero matrix.

Hence the points of G are not dense in the points of M....
--
Clarence Wilkerson \ HomePage: http://www.math.purdue.edu/~wilker
Prof. of Math. \ Internet: wil...@NOspam.math.purdue.edu
Dept. of Mathematics \ Messages: (765) 494-1903, FAX 494-0548
Purdue University, \
W. Lafayette, IN 47907-1395 \

0 new messages