Professional Documents
Culture Documents
II
1
I give a free paraphrase of Galois’s original statement. See [4, Prop. I,
p. 51].
Proof.
Suppose that g(x1 , . . . , xn ) is a function of the roots. We have
proved that all the roots are rational functions of V1 , then
g(x1 , . . . , xn ) = ḡ(V1 ).
ḡ(V ) − k = 0.
It follows that this equation admits all the other roots of F (t)
(an irreducible polynomial) and its value is left unchanged by
the substitutions of G.
Remark
Note that this proof does not use the fact that the substitutions
identified by the roots of F (t) constitute a group. In effect, what
is used is the symmetry of the roots themselves.
Remark
The proof given by Jordan of the fact that the roots of F (t)
identify the Galois Group is the following.2
To give it in the simplest way, it is convenient to change a little
the notations: V1 is still the value that the identity substitution
1 gives to V ; and Vσ , Vτ , . . . will denote the values that V
assumes under the effect of the permutations σ, τ, . . . that
identify the other roots of F (t). Hence
This is possible if and only if the roots are still the same, which
means that
σ, σ 2 , στ, . . .
are the original permutations
1, σ, τ. . . .
x4 − 10x2 + 1. (1)
1 1
α α −α −
α α
1 1
−α −α α −
α α
1 1 1
− α −α
α α α
1 1 1
− − −α α
α α α
Note that this presentation is obtained without any need of
complicated calculations.
The first example given by Galois
He states as obvious that the general equation of degree m has
as its group Sm because “in this case the symmetric functions
are the only ones that can be determined rationally.”4 The
result is quite evident if the general equation is considered as an
equations having the roots x1 , x2 , . . . , xm to be conceived as
indeterminates. But it is not so obvious if the general equation
of degree m is an equation of the form
7
See [4, pp. 89-91].
8
See the example given in the previous Remark which shows a simple
calculation of the Galois Group of an equation in the case in which all the
roots are functions of one of them.
We need two preliminary facts.
Theorem
An equation f (x) = 0 is irreducible if and only if its Galois
group G is transitive.
Proof.
Suppose that G be not transitive and that the root x1 may be
sent by the substitution of G only into x1 , x2 , . . . , xm , with
m < ∂f . Then the substitutions of G do not modify the
symmetric functions of x1 , x2 , . . . , xm , and the polynomial
(x − x1 )(x − x2 ) · · · (x − xm ) has its coefficients in the ground
field and is a divisor of f (x). Then f (x) must be reducible.
Conversely, suppose that G be transitive and f (x) be divisible
by (x − x1 ) · · · (x − xm ) with m < ∂f . A substitution of G
exists that sends x1 into xm+1 . By the effect of this substitution
the polynomial (x − x1 ) · · · (x − xm ) cannot remain unchanged,
because it acquires the new root xm+1 . Hence it cannot have all
its coefficients in the ground field.9
9
[5, p. 259].
The second fact we need is that the order of a transitive group
over n letters is divisible by n, as it is easily seen.10
Theorem
Let p be a prime number and consider the equation
xp − 1
= xp−1 + xp−2 + · · · + x + 1 = 0. (3)
x−1
The group of this equation is cyclic of order p − 1.
Proof.
The polynomial at the left side of (3) is irreducible, as was
proved by Gauss. Gauss also proved that Z∗p has a primitive
root (a generator of the cyclic group of the elements different
from 0). Let γ be such a primitive root and, having singled out
a root r of the equation (3), denote all its roots by
2 p−2
x1 = r, x2 = rγ , x3 = rγ , . . . , xp−1 = rγ .
10
[5, p. 29].
Proof (continued).
The Galois group has been defined by a Galois Resolvent
2 p−1
V1 = a1 r + a2 rγ + a3 rγ + · · · + ap−1 rγ = w(r). (4)
xp−1 + xp−2 + · · · + x + 1
j
It follows that all the values rγ are possible candidates to
produce the substitutions of the Galois group.
Proof (continued).
The substitution of rγ for r produces the arrangement
xp−1 , x1 , x2 , . . . , xp−2
11
See also [5, pp. 291-297].
Let us consider the equation x5 − 1 = 0. It has the root 1 which
is fixed by the substitutions of its Galois Group, and the roots
2π i
r=e 5 , r2 , r3 , r4
r r2 r4 r3
r2 r4 r3 r
r4 r3 r r2
r3 r r2 r4