## With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

No credit card required We now jump from speciﬁc examples of Galois repre-
sentations to an important general m ethod for ﬁnding
them. This method, using
´
etale cohomology, sets up a
direct relationship between systems of Z-equations and
representations of the absolute Galois group G.Thisre-
lationship comes from some very advanced mathematical
constructions whose nature we can describe only in a
very sketchy way. The discovery (or invention) of
´
etale
cohomology is one of the major successes of twentieth-
century mathematics.
Vector Spaces and Linear Actions of Groups
We have seen in preceding chapters examples of one- and two-
dimensional Galois representations. But it is very unlikely that
they include all the information to be found in the absolute Galois
group of Q. Indeed they do not. We seek a large supply of Galois
representations of all dimensions.
We need a general method to derive linear Galois representations
from Z-varieties. There are several of these methods, but we discuss
just one: ´etale cohomology. It is too advanced to explain in detail 226 CHAPTER 20
here, but we can give a very rough idea of what is going on. First, a
deﬁnition:
DEFINITION: If k is a ﬁeld and n 1, then an n-dimensional
vector over k is an n-by-1 matrix with entries from k.Theset
of all n-dimensional vectors over k is denoted k
n
, and is called
a vector space.
For example, C
3
contains the three-dimensional vector
3
i
2
.
The vector
2
4
3
10
is contained in Q
4
, and is also in R
4
.
DEFINITION: A line in k
n
is a subset of k
n
consisting of all
vectors that are proportional to some given vector.
An example of a line in C
3
is the set of all the three-dimensional
vectors
3c
ic
2c
,
where c takes on all possible values in C (including 0—the 0-vector
is in every line by this deﬁnition.)
A linear representation, you will remember, is a morphism from
agroupG to a group of square matrices GL(n, k), where k is some
number system. We call this a “linear” representation because if M
is a matrix and L is a line in the n-dimensional space of vectors k
n
,
then ML is always a line again. A MACHINE FOR MAKING GALOIS REPRESENTATIONS 227
For example, if M is the matrix
12
45
in GL(2, Q)andL is the
line of all vectors
a
b
such that b = 2a (if you graphed this, it would
be the line of slope 2 through the origin) then ML is the line of all
vectors
c
d
,whered = (14/5)c.
Why? When we multiply the matrix M by the vector
a
b
we get
a + 2b
4a + 5b
.Nowif
a
b
were a point on the line L to begin with, that
would mean that b = 2a. If we substitute this in, we get
a + 2b
4a + 5b
=
a + 4a
4a + 10a
=
5a
14a
. This last vector is of the form
c
d
where d =
14
5
c.
Thus M sends the line of slope 2 to the line of slope
14
5
. Simi-
larly, you can check that M sends any line through the origin to
another line through the origin. And this is why we call v Mv a
linear map.
Now suppose M is any n-by-n matrix in GL(n, k)andv and w are
any vectors in k
n
,anda is any number in k.ThenM(v + w) = Mv +
Mw and M(av) = aMv. These formulas follow from the distributive
and associative laws of matrix multiplication.
Conversely, if you have any permutation s of k
n
that always
satisﬁes the formulas s(v + w) = s(v) + s(w)ands(av) = as(v), then
there is some matrix M so that s is the same as multiplication by
M: s(v) = Mv for every v in k
n
.Ofcourse,M depends on s.
So one way of constructing linear representations of a group G is
by ﬁnding a vector space k
n
and a linear action of G on it. This is
by deﬁnition a morphism f from G to the permutation group of k
n
that satisﬁes the formulas s(v + w) = s(v) + s(w)ands(av) = as(v)
for every permutation s = f (g) coming from any group element
g in G.
For example, think about the group A
4
of symmetries of a
tetrahedron which we discussed in chapter 12. There we deﬁned
amorphismr with the property that for any element σ in A
4
,
r(σ ) is the rotation of space that causes the permutation σ on the
vertices of the tetrahedron. You can think of a rotation of space as
a permutation of R
3
. As such, it is a linear action on R
3
because
its effect as a permutation of R
3
can be given as multiplication
by a certain matrix in GL(3, R). (The particular matrix depends, of

## With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

No credit card required