algebra-13-dummit-1

Solutions to
Abstract Algebra
Chapter 13 - Field Theory
David S. Dummit & Richard M. Foote
Solutions by positrón0802
https://positron0802.wordpress.com
1 January 2021
Contents
13 Field Theory
13.1 Basic Theory and Field Extensions . . . . . . . . . .
13.2 Algebraic Extensions . . . . . . . . . . . . . . . . .
13.3 Classical Straightedge and Compass Constructions
13.4 Splitting Fields and Algebraic Closures . . . . . . .
13.5 Separable and Inseparable Extension . . . . . . . .
13.6 Cyclotomic Polynomials and Extensions . . . . . .
13
13.1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
5
12
13
15
19
Field Theory
Basic Theory and Field Extensions
Exercise 13.1.1. The polynomial 𝑝 (𝑥) = 𝑥 3 + 9𝑥 + 6 is irreducible in Z[𝑥] by Eisenstein Criterion
with 𝑝 = 3. By Gauss Lemma, it is thien rreducible in Q[𝑥]. To find (1 + 𝜃 ) −1, we apply the
Euclidean algorithm to 𝑝 (𝑥) and 1 + 𝑥 to find
𝑥 3 + 9𝑥 + 6 = (1 + 𝑥) (𝑥 2 − 𝑥 + 10) − 4.
Evaluating at 𝜃, we have (1 + 𝜃 ) (𝜃 2 − 𝜃 + 10) = 4. Therefore
(1 + 𝜃 ) −1 =
𝜃 2 − 𝜃 + 10
.
4
1
13.1
Basic Theory and Field Extensions
Exercise 13.1.2. Let 𝑓 (𝑥) = 𝑥 3 − 2𝑥 − 2. The polynomial 𝑓 is irreducible over Z by Eisenstein
Criterion with 𝑝 = 2, hence over Q by Gauss Lemma. Now, if 𝜃 is a root of 𝑓 , then 𝜃 3 = 2𝜃 + 2, so
that
(1 + 𝜃 ) (1 + 𝜃 + 𝜃 2 ) = 1 + 2𝜃 + 2𝜃 2 + 𝜃 3 = 3 + 4𝜃 + 2𝜃 2 .
1+𝜃
For computing
, first we compute (1 + 𝜃 + 𝜃 2 ) −1 . Applying the Euclidean algorithm, we
1 + 𝜃 + 𝜃2
obtain
𝑥 3 − 2𝑥 − 2 = (𝑥 2 + 𝑥 + 1) (𝑥 − 1) − 2𝑥 − 1
and
𝑥2 𝑥 7
9
𝑥 − 2𝑥 − 2 = (2𝑥 + 1)
− −
− .
2
4 8
8
3
Evaluating at 𝜃, from these equalities it follows that
2
(𝜃 + 𝜃 + 1) (𝜃 − 1) = 2𝜃 + 1
so that
and
(2𝜃 + 1)
−1
8 𝜃2 𝜃 7
=
,
− −
9 2
4 8
𝜃2 𝜃 7
8 2
(𝜃 + 𝜃 + 1) (𝜃 − 1) ( − − ) = 1.
9
2
4 8
Then
2
(𝜃 + 𝜃 + 1)
−1
2
8
2𝜃 2 𝜃 5
𝜃
𝜃 7
= (𝜃 − 1)
=−
− −
+ + ,
9
2
4 8
3
3 3
where we used 𝜃 3 = 2𝜃 + 2 again. It follows that
2𝜃 2 𝜃 5
𝜃 2 2𝜃 1
1+𝜃
=
(1
+
𝜃
)
−
=− +
+
+
+ .
2
1+𝜃 +𝜃
3
3 3
3
3 3
Exercise 13.1.3. Since 03 + 0 + 1 = 1 and 11 + 1 + 1 = 1 in F2, it follows that 𝑥 3 + 𝑥 + 1 is irreducible
over F2 . Since 𝜃 is root of 𝑥 3 + 𝑥 + 1, we have 𝜃 3 = −𝜃 − 1 = 𝜃 + 1. Hence, the powers of 𝜃 in F2 (𝜃 )
are
𝜃, 𝜃 2, 𝜃 3 = 𝜃 + 1, 𝜃 4 = 𝜃 2 + 𝜃, 𝜃 5 = 𝜃 2 + 𝜃 + 1, 𝜃 6 = 𝜃 2 + 1, and 𝜃 7 = 1.
Exercise 13.1.4. Denote this map by 𝜑. Then
√
√
√
√
√
√
𝜑 (𝑎 + 𝑏 2 + 𝑐 + 𝑑 2) = 𝑎 + 𝑐 − 𝑏 2 − 𝑑 2 = 𝜑 (𝑎 + 𝑏 2) + 𝜑 (𝑐 + 𝑑 2)
and
√
√
√
𝜑 ((𝑎 + 𝑏 2) · (𝑐 + 𝑑 2)) = 𝜑 (𝑎𝑐 + 2𝑏𝑑 + (𝑎𝑑 + 𝑏𝑐) 2)
√
= 𝑎𝑐 + 2𝑏𝑑 − (𝑎𝑑 + 𝑏𝑐) 2
√
√
= (𝑎 − 𝑏 2) (𝑐 − 𝑑 2)
√
√
= 𝜑 (𝑎 + 𝑏 2)𝜑 (𝑐 + 𝑑 2),
2
Solutions by positrón0802
13.1
Basic Theory and Field Extensions
√
√
√
√
√
so 𝜑 is an homomorphism. If 𝜑 (𝑎 + 𝑏 2) = 𝜑 (𝑐 + 𝑑 2), then 𝑎 − 𝑏 2 =√𝑐 − 𝑑 2;√ as 2 ∉ Q,
this implies
∈ Q( 2), we have
√ 𝑐 = 𝑑. Thus 𝜑 is injective. Furthermore, given 𝑎 + 𝑏 2 √
√ 𝑎 = 𝑏 and
𝜑 (𝑎 − 𝑏 2) = 𝑎 + 𝑏 2, so 𝜑 is surjective. Therefore 𝜑 is an isomorphism of Q( 2) with itself.
Exercise 13.1.5. Let 𝛼 = 𝑝/𝑞 be a root of a monic polynomial 𝑝 (𝑥) = 𝑥 𝑛 + · · · + 𝑎 1𝑥 + 𝑎 0 over Z,
with gcd(𝑝, 𝑞) = 1. Then
𝑛−1
𝑛
𝑝
𝑝
𝑝
+ 𝑎𝑛−1
+ · · · + 𝑎 1 + 𝑎 0 = 0,
𝑞
𝑞
𝑞
so that
𝑞(𝑎𝑛−1𝑝 𝑛−1 + · · · + 𝑎 1𝑝𝑞𝑛−2 + 𝑎 0𝑞𝑛−1 ) = −𝑝 𝑛 .
Thus, every prime that divides 𝑞 must divide 𝑝 𝑛 as well, so divides 𝑝. Since gcd(𝑝, 𝑞) = 1, there is
no prime dividing 𝑞, hence 𝑞 = ±1. The result follows.
Exercise 13.1.6. This is straightforward. If
𝑎𝑛 𝛼 𝑛 + 𝑎𝑛−1𝛼 𝑛−1 + · · · + 𝑎 1𝛼 + 𝑎 0 = 0,
then
𝑛−1
(𝑎𝑛 𝛼)𝑛 + 𝑎𝑛−1 (𝑎𝑛 𝛼)𝑛−1 + 𝑎𝑛 𝑎𝑛−2 (𝑎𝑛 𝛼)𝑛−2 + · · · + 𝑎𝑛−2
𝑛 𝑎 1 (𝑎𝑛 𝛼) + 𝑎𝑛 𝑎 0
𝑛−1
𝑛−2
𝑛−1
= 𝑎𝑛𝑛 𝛼 𝑛 + 𝑎𝑛−1
+ 𝑎𝑛−1
+ · · · + 𝑎𝑛−1
𝑛 𝑎𝑛−1𝛼
𝑛 𝑎𝑛−2𝛼
𝑛 𝑎 1 𝛼 + 𝑎𝑛 𝑎 0
= 𝑎𝑛𝑛−1 (𝑎𝑛 𝛼 𝑛 + 𝑎𝑛−1𝛼 𝑛−1 + 𝑎𝑛−2𝛼 𝑛−2 + · · · + 𝑎 1𝛼 + 𝑎 0 ) = 0.
Exercise 13.1.7. Suppose 𝑥 3 − 𝑛𝑥 + 2 is reducible; then it must have a linear factor, hence a root.
By the Rational Root Theorem, if 𝛼 is a root of 𝑥 3 −𝑛𝑥 + 2, then 𝛼 must divide 2, so that 𝛼 = ±1, ±2.
If 𝛼 = −1 or 2, then 𝑛 = 3; if 𝛼 = −1, then 𝑛 = 1; and if 𝛼 = 2, then 𝑛 = 5. Therefore 𝑥 3 − 𝑛𝑥 + 2 is
irreducible for 𝑛 ≠ −1, 3, 5.
Exercise 13.1.8. We subdivide this exercise in cases and subcases.
If 𝑥 5 − 𝑎𝑥 − 1 is reducible then it has a root (linear factor) or is a product of two irreducible
polynomials of degrees 2 and 3 respectively.
Case 1. If 𝑥 5 − 𝑎𝑥 − 1 has a root, then, by the Rational Root Theorem, it must be 𝛼 = ±1. If
𝛼 = 1 is a root, then 𝑎 = 0. If 𝛼 = −1 is a root, then 𝑎 = 2.
Case 2. Now, assume that there exist 𝑓 (𝑥) and 𝑔(𝑥) irreducible monic polynomials over Z of
degrees 2 and 3 respectively, such that 𝑥 5 − 𝑎𝑥 − 1 = 𝑓 (𝑥)𝑔(𝑥). Write 𝑓 (𝑥) = 𝑥 2 + 𝑏𝑥 + 𝑐 and
𝑔(𝑥) = 𝑥 3 + 𝑟𝑥 2 + 𝑠𝑥 + 𝑡, where 𝑏, 𝑐, 𝑟, 𝑠, 𝑡 ∈ Z. Then
𝑥 5 − 𝑎𝑥 − 1 = (𝑥 2 + 𝑏𝑥 + 𝑐) (𝑥 3 + 𝑟𝑥 2 + 𝑠𝑥 + 𝑡)
= 𝑥 5 + (𝑏 + 𝑟 )𝑥 4 + (𝑏𝑟 + 𝑐 + 𝑠)𝑥 3 + (𝑏𝑠 + 𝑐𝑟 + 𝑡)𝑥 2 + (𝑏𝑡 + 𝑐𝑠) + 𝑡𝑐.
3
Solutions by positrón0802
13.1
Basic Theory and Field Extensions
Equating coefficients leads to
𝑏 + 𝑟 = 0,
𝑏𝑟 + 𝑐 + 𝑠 = 0,
𝑏𝑠 + 𝑐𝑟 + 𝑡 = 0,
𝑏𝑡 + 𝑐𝑠 = −𝑎,
𝑐𝑡 = −1.
From 𝑐𝑡 = −1 we deduce (𝑐, 𝑡) = (−1, 1) or (𝑐, 𝑡) = (1, −1), which gives us two cases.
Case 2.1. First suppose (𝑐, 𝑡) = (−1, 1). Then the system of equations reduces to
𝑏 + 𝑟 = 0,
𝑏𝑟 − 1 + 𝑠 = 0,
𝑏𝑠 − 𝑟 + 1 = 0,
𝑏 − 𝑠 = −𝑎.
Put 𝑏 = −𝑟 into the second and third equations to obtain −𝑟 2 − 1 + 𝑠 = 0 and −𝑟𝑠 − 𝑟 + 1 = 0, that
is, 𝑟 2 + 1 − 𝑠 = 0 and 𝑟𝑠 + 𝑟 − 1 = 0. Adding these last two equations we obtain 𝑟 2 + 𝑟𝑠 + 𝑟 − 𝑠 = 0.
Thus 𝑟 2 + 𝑟𝑠 + 𝑟 + 𝑠 = 2𝑠, so that (𝑟 + 1) (𝑟 + 𝑠) = 2𝑠. Now, from 𝑟 2 + 1 − 𝑠 = 0 we have 𝑟 2 = 𝑠 − 1, so
𝑟 2 + 𝑟𝑠 + 𝑟 − 𝑠 = 0 becomes 𝑟𝑠 + 𝑟 = 1, that is, 𝑟 (𝑠 + 1) = 1. Hence, 𝑟 = 1 and 𝑠 = 0, or 𝑟 = −1 and
𝑠 = −2. If 𝑟 = 1 and 𝑠 = 0, then (𝑟 + 1) (𝑟 + 𝑠) = 2𝑠 leads to 2 = 0, a contradiction. If 𝑟 = −1 and
𝑠 = −2, it leads to 0 = −4, another contradiction.
We deduce that (𝑐, 𝑡) = (−1, 1) is impossible.
Case 2.2. Suppose that (𝑐, 𝑡) = (1, −1). The system of equations reduces to
𝑏 + 𝑟 = 0,
𝑏𝑟 + 1 + 𝑠 = 0,
𝑏𝑠 + 𝑟 − 1 = 0,
−𝑏 + 𝑠 = −𝑎.
Adding the second and third equation we obtain 𝑏 (𝑟 + 𝑠) + 𝑟 + 𝑠 = 0, so that (𝑏 + 1) (𝑟 + 𝑠) = 0.
Then 𝑏 = −1 or 𝑟 = −𝑠, so one more time we have two cases. If 𝑟 = −𝑠, then 𝑏𝑟 + 1 + 𝑠 = 0 becomes
𝑏𝑟 + 1 − 𝑟 = 0. Hence, 𝑏 = −𝑟 and 𝑏𝑟 + 1 − 𝑟 = 0 gives 𝑟 2 + 𝑟 − 1 = 0. By the Rational Root Theorem,
this equation has no roots in Z. Since 𝑟 ∈ Z, we have a contradiction. Now suppose 𝑏 = −1. From
𝑏 = −𝑟 we obtain 𝑟 = 1; thus, from 𝑏𝑟 + 1 + 𝑠 = 0 we obtain 𝑠 = 0. Finally, from −𝑏 + 𝑠 = −𝑎 it
follows that 𝑎 = −1. Therefore, we find the consistent solution (𝑏, 𝑐, 𝑟, 𝑠, 𝑡) = (−1, 1, 1, 0, −1) and
the factorisation
𝑥 5 − 𝑎𝑥 − 1 = (𝑥 2 + 𝑏𝑥 + 𝑐) (𝑥 3 + 𝑟𝑥 2 + 𝑠𝑥 + 𝑡) = (𝑥 2 − 𝑥 + 1) (𝑥 3 + 𝑥 2 − 1).
4
Solutions by positrón0802
13.2
13.2
Algebraic Extensions
Algebraic Extensions
Exercise 13.2.1. Since the characteristic of F is 𝑝, its prime subfield is (isomorphic to) F𝑝 = Z/𝑝Z.
Then F is a vector space over F𝑝 . Since F is finite, we have [F : F𝑝 ] = 𝑛 for some 𝑛 ∈ Z+ . It follows
that
|F| = |F𝑝 | [F:F𝑝 ] = 𝑝 𝑛 .
Exercise 13.2.2. Note that 𝑔 and ℎ are irreducible over both F2 and F3 . If 𝜃 is a root of 𝑔, then
F2 (𝜃 ) F2 /(𝑔(𝑥)) has 4 elements and F3 (𝜃 ) F3 /(𝑔(𝑥)) has 9 elements. Furthermore, is 𝜃 2 if a
root of ℎ, then F2 (𝜃 2 ) F2 /(ℎ(𝑥)) has 8 elements and F3 (𝜃 2 ) F3 /(ℎ(𝑥)) has 27 elements.
The multiplication table for F2 /(𝑔(𝑥)) is
·
0
1
𝑥
𝑥 +1
0
0
0
0
0
1
0
1
𝑥
𝑥 +1
𝑥 +1
0
𝑥 +1
𝑥
𝑥
𝑥
0
𝑥
𝑥 +1
𝑥
The multiplication table for F3 /(𝑔(𝑥)) is
·
0
1
2
𝑥
𝑥 +1
𝑥 +2
2𝑥
2𝑥+1
2𝑥+2
0
0
0
0
0
0
0
0
0
0
1
0
1
2
𝑥
𝑥 +1
𝑥 +2
2𝑥
2𝑥+1
2𝑥+2
2
0
2
1
2𝑥
2𝑥+2
2𝑥+1
𝑥
𝑥 +2
𝑥 +1
𝑥 +1
0
𝑥 +1
2𝑥+2
1
𝑥 +2
2𝑥
2
𝑥
2𝑥+1
𝑥
0
𝑥
2𝑥
2𝑥+1
1
𝑥 +1
𝑥 +2
2𝑥+2
2
𝑥 +2
0
𝑥 +2
2𝑥+1
𝑥 +1
2𝑥
2
2𝑥+2
1
𝑥
2𝑥
0
2𝑥
𝑥
𝑥 +2
2
2𝑥+2
2𝑥+1
𝑥 +1
1
2𝑥+1
0
2𝑥+1
𝑥 +2
2𝑥+2
𝑥
1
𝑥 +1
2
2𝑥
2𝑥+2
0
2𝑥+2
𝑥 +1
2
2𝑥+1
𝑥
1
2𝑥
𝑥 +2
In both cases, 𝑥 is a generator of the cyclic group of non-zero elements.
Exercise 13.2.3. Since 1+𝑖 ∉ Q, its minimal polynomial is of degree at least 2. We try conjugation,
and obtain
(𝑥 − (1 + 𝑖)) (𝑥 − (1 − 𝑖)) = 𝑥 2 − 2𝑥 + 2,
which is irreducible by Eisenstein with 𝑝 = 2. Therefore, the minimal polynomial of 1 + 𝑖 over Q
is 𝑥 2 − 2𝑥 + 2.
√
√
√
√
Exercise 13.2.4.
First, note
that (2 + 3) 2 = 4 + 4 3 + 3 = 7 + 4 3. Let 𝜃 = 2 + 3. Then
√
√
2
𝜃 2 − 4𝜃 = 7 + 4 3 − 8 − 4 3 = −1, so 𝜃 is a root of 𝑥 2 − 4𝑥 + 1. Moreover,
√ 𝑥 − 4𝑥 + 1√is irreducible
2
over Q (because 𝜃 ∉ Q), so 𝑥 − 4𝑥 + 1 is the minimal polynomial of 2 + 3. Thus 2 + 3 has degree
2 over Q.
5
Solutions by positrón0802
13.2
Algebraic Extensions
√
Now let 𝛼 = 3 2 and 𝛽 = 1 + 𝛼 + 𝛼 2 . Then 𝛽 ∈ Q(𝛼), so Q ⊂ Q(𝛽) ⊂ Q(𝛼). We have
[Q(𝛼) : Q] = [Q(𝛼) : Q(𝛽)] [Q(𝛽) : Q]. Note that [Q(𝛼) : Q] = 3 since 𝛼 has minimal polynomial
𝑥 3 −2 over Q, so [Q(𝛽) : Q] is either 1 or 3. For the sake of a contradiction suppose [Q(𝛽) : Q] = 1,
so that 𝛽 ∈ Q. Then
𝛽 2 = (1 + 𝛼 + 𝛼 2 ) 2 = 1 + 2𝛼 + 3𝛼 2 + 2𝛼 3 + 𝛼 4 = 5 + 4𝛼 + 3𝛼 2,
where we used 𝛼 3 = 2, and therefore
𝛽 2 − 3𝛽 = 5 + 4𝛼 + 3𝛼 2 − 3(1 + 𝛼 + 𝛼 2 ) = 2 + 𝛼 .
But then 𝛼 = −𝛽 2 + 3𝛽 − 2 ∈ Q(𝛽) = Q, a contradiction. It follows that [Q(𝛽) : Q] = 3.
Exercise 13.2.5. Since the polynomials have degree 3, if they were reducible they would have a
linear factor, hence a root√in 𝐹√. Note that√every element of 𝐹 has the form 𝑎 + 𝑏𝑖, where 𝑎, 𝑏 ∈ Q.
The roots of 𝑥 3 − 2 are 3 2, 𝜁 3 2 and 𝜁 2 3 2, where
𝜁 is the primitive 3rd root of unity, i.e., 𝜁 =
√
√
3
1
exp(2𝜋𝑖/3) = cos(2𝜋/3) + 𝑖 sin(2𝜋/3) = − 2 + 2 . Since 3 ∉ Q, none of these elements belong to
√ √
√
𝐹, so 𝑥 3 − 2 is irreducible over 𝐹 . Similarly, the roots of 𝑥 3 − 3 are 3 3, 𝜁 3 3 and 𝜁 2 3 3, and by the
same argument none of these elements belong to 𝐹 . Hence 𝑥 3 − 3 is irreducible over 𝐹 .
Exercise 13.2.6. We have to prove that 𝐹 (𝛼 1, . . . , 𝛼𝑛 ) is the smallest field containing 𝐹 (𝛼 1 ), . . . , 𝐹 (𝛼𝑛 ).
Clearly 𝐹 (𝛼𝑖 ) ⊂ 𝐹 (𝛼 1, . . . , 𝛼𝑛 ) for all 1 ≤ 𝑖 ≤ 𝑛. Now let 𝐾 be a field such that 𝐹 (𝛼𝑖 ) ⊂ 𝐾 for all
𝑖. If 𝜃 is an element of 𝐹 (𝛼 1, . . . , 𝛼𝑛 ), it has the form 𝜃 = 𝑎 1𝛼 1 + · · · + 𝑎𝑛 𝛼𝑛 , where 𝑎 1, . . . , 𝑎𝑛 ∈ 𝐹 .
As every 𝑎𝑖 𝛼𝑖 belongs to 𝐾, we have 𝜃 ∈ 𝐾 . Thus 𝐹 (𝛼 1, . . . , 𝛼𝑛 ) ⊂ 𝐾 . It follows that 𝐹 (𝛼 1, . . . , 𝛼𝑛 )
contains all of the 𝐹 (𝛼𝑖 ) and is contained in every field containing all of the 𝐹 (𝛼𝑖 ), so 𝐹 (𝛼 1, . . . , 𝛼𝑛 )
is the composite of the fields 𝐹 (𝛼 1 ), 𝐹 (𝛼 2 ), . . . , 𝐹 (𝛼𝑛 ).
√ √
√ √
√ √
√ √
Exercise 13.2.7. Since 2 + √3 is an
element
of
Q(
2,
3),
clearly
Q(
2
+
3)
⊂
Q(
2, 3). On
√
√
√
√
2
3
the other hand, consider 𝜃 = 2 + 3. Then 𝜃 = 5 + 2 6, and 𝜃 = 11 2 + 9 3, so
√
√
1
1
2 = (𝜃 3 − 9𝜃 ) and
3 = (11𝜃 − 𝜃 3 ).
2
2
√
√ √
√ √
√ √
√
Therefore
2 ∈ Q(𝜃 ) and
√ √
√ 3√∈ Q(𝜃 ), so Q(√ 2,√ 3) ⊂ Q( 2 + 3). It follows that Q( 2 + 3) =
Q( 2, 3), so that [Q( 2 + 3) : Q] = [Q( 2, 3) : Q] = 4.
We also have
√
√
𝜃 4 − 10𝜃 2 = (49 + 20 6) − 10(5 + 2 6) = −1, so 𝜃 4 − 10𝜃 2 + 1 = 0.
√ √
Since
[Q(
2 + 3) : Q] = 4, the polynomial 𝑥 4 − 10𝑥 2 + 1 is irreducible over Q, and is satisfied by
√ √
2 + 3.
√ √
Exercise 13.2.8. The elements of 𝐹 ( 𝐷 1, 𝐷 2 ) can be written as
p
p
p
𝑎 + 𝑏 𝐷 1 + 𝑐 𝐷 2 + 𝑑 𝐷 1 𝐷 2, where 𝑎, 𝑏, 𝑐, 𝑑 ∈ 𝐹 .
6
Solutions by positrón0802
13.2
Algebraic Extensions
We have
p p
p p
p
p
[𝐹 ( 𝐷 1, 𝐷 2 ) : 𝐹 ] = [𝐹 ( 𝐷 1, 𝐷 2 ) : 𝐹 ( 𝐷 1 )] [𝐹 ( 𝐷 1 ) : 𝐹 ].
√
√ √
√ √
Since [𝐹 ( 𝐷 1 )√: 𝐹 ] √
= 2, [𝐹 ( √𝐷 1, 𝐷 2 ) : 𝐹 ] can be either 2 or 4. Now [𝐹 ( 𝐷 1, 𝐷 2 ) : 𝐹 ] =√ 2 if
𝐷 1, 𝐷 2 ) : 𝐹 ( 𝐷 1 )] = 1, and that occurs exactly if 𝑥 2 − 𝐷 2 is reducible in 𝐹 ( 𝐷 1 )
and only
√ if [𝐹 ( √
(i.e., if 𝐷 2 ∈ 𝐹 ( 𝐷 1 )), that is, if there exists 𝑎, 𝑏 ∈ 𝐹 such that
p
p
(𝑎 + 𝑏 𝐷 1 ) 2 = 𝐷 2, so that 𝑎 2 + 2𝑎𝑏 𝐷 1 + 𝑏 2 𝐷 12 = 𝐷 2 .
√
Note that 𝑎𝑏 = 0, as 𝑎𝑏 ≠ 0 implies 𝐷 1 ∈ 𝐹, contrary to the assumption. Then 𝑎 = 0 or 𝑏 = 0.
If 𝑏 = 0, then 𝐷 2 is a square in 𝐹, contrary to the assumption. If 𝑎 = 0, then
𝑏 2 𝐷 1 = 𝐷 2, and thus
√
𝐷2 2
𝐷 1 𝐷 2 = ( 𝑏 ) , so 𝐷 1 𝐷 2 is a square in 𝐹 . Thus 𝑥 2 − 𝐷 2 is reducible in 𝐹 ( 𝐷 1 ) if and only if 𝐷 1 𝐷 2
is a square in 𝐹 . The result follows.
p √
√
√
√
√
Exercise 13.2.9. Suppose 𝑎 + 𝑏 = 𝑚 + 𝑛 for some 𝑚, 𝑛 ∈ 𝐹, so
that 𝑎 + 𝑏 = 𝑚 + 𝑛 + 2 𝑚𝑛.
p
√
√
√
√
√
Since 𝑏 is not a square in 𝐹, this means 𝑏 = 2 𝑚𝑛. We also have 𝑎 + 𝑏 − 𝑛 = 𝑚, so
q
√
√
√
√
𝑏 = 2 𝑛( 𝑎 + 𝑏 − 𝑛).
Hence,
q
√
√
𝑏 = 2 𝑛(𝑎 + 𝑏) − 2𝑛
√
√
⇒
( 𝑏 + 2𝑛) 2 = 4𝑛(𝑎 + 𝑏)
√
√
⇒ 𝑏 + 4𝑛 𝑏 + 4𝑛 2 = 4𝑛(𝑎 + 𝑏)
⇒
𝑏 + 4𝑛 2 − 4𝑛𝑎 = 0
√
4𝑎 ± 16𝑎 2 − 16𝑏
⇒
𝑛=
8
√
2𝑛
𝑎2 − 𝑏 = ± .
⇒
𝑎
√
2
Therefore, since 𝑎 and 𝑛 belong to 𝐹, so does 𝑎 − 𝑏.
√
2 − 𝑏 is a square in 𝐹, so that 𝑎 2 − 𝑏 ∈ 𝐹 . We prove that there exist
Conversely, assume
that
𝑎
p √
√
√
𝑚, 𝑛 ∈ 𝐹 such that 𝑎 + 𝑏 = 𝑚 + 𝑛. Consider
√
√
𝑎 + 𝑎2 − 𝑏
𝑎 − 𝑎2 − 𝑏
𝑚=
and 𝑛 =
.
2
2
p √
√
√
Note that 𝑚 and 𝑛 belong to 𝐹 as char(𝐹 ) ≠ 2. We claim 𝑎 + 𝑏 = 𝑚 + 𝑛. Indeed, we have
p √ p
√
√
√ !2
√
(𝑎 + 𝑏) + 2 𝑎 2 − 𝑏 + (𝑎 − 𝑏)
𝑎+ 𝑏+ 𝑎− 𝑏
𝑚=
=
,
4
2
7
Solutions by positrón0802
13.2
and
Thus
Algebraic Extensions
p √
p
√
√
√
√ !2
(𝑎 + 𝑏) − 2 𝑎 2 − 𝑏 + (𝑎 − 𝑏)
𝑎+ 𝑏− 𝑎− 𝑏
𝑛=
=
.
4
2
p √ p
p √
p
√
√
√
𝑎+ 𝑏+ 𝑎− 𝑏
𝑎+ 𝑏− 𝑎− 𝑏
and
𝑛=
,
2
2
p
p √ p
p √
√
√
q
√
√
√
𝑎+ 𝑏+ 𝑎− 𝑏
𝑎+ 𝑏− 𝑎− 𝑏
𝑚+ 𝑛=
+
= 𝑎 + 𝑏,
2
2
√
𝑚=
so
as claimed.
p √
Finally, we use this to determine when is the field
Q(
𝑎, 𝑏 ∈ Q, biquadratic over Q.
p √ 𝑎 + 𝑏),
√
√
√ √
2
If 𝑎 − 𝑏 is a squarepin Q and 𝑏 is not, we have Q( 𝑎 + 𝑏) = Q( 𝑚 + 𝑛) = Q( 𝑚, 𝑛), so by
√
Exercise 13.2.8, Q( 𝑎 + 𝑏) is biquadratic over Q when 𝑎 2 − 𝑏 is a square in Q and none of 𝑏, 𝑚,
𝑛 or 𝑚𝑛 is a square in Q. Since
√
√
𝑎 + 𝑎2 − 𝑏 𝑎 − 𝑎2 − 𝑏 𝑏
𝑚𝑛 =
= ,
2
2
4
p √
𝑚𝑛 is never a square when 𝑏 is not. Thus, Q( 𝑎 + 𝑏) is biquadratic over Q exactly when 𝑎 2 − 𝑏
is a square in Q and none of 𝑏, 𝑚 or 𝑛 is a square in Q.
p
p √
√
Exercise 13.2.10. Note that 3 + 2 2 = 3 + 8. Recalling Exercise 13.2.9 with 𝑎 = 3 and 𝑏 =
8,we have that 𝑎 2 − 𝑏 = 9 −p8 = 1 is a square in Q and 𝑏 = 8pis not. Hence, we find (𝑚 = 2 and
√
√
√
√
𝑛 = 1 from Exercisep13.2.9) 3 + 8 = 2 + 1. Therefore, Q( 3 + 2 2) = Q( 2) and the degree
√
of the extension Q( 3 + 2 2) over Q is 2.
Exercise 13.2.11. (a) First, note that the conjugation map 𝑎 + 𝑏𝑖 → 𝑎 − 𝑏𝑖 is an automorphism of
C, so it takes squares roots√to square roots. Furthermore, it maps the first quadrant onto the fourth
(and reciprocally). Since 3 + 4𝑖 is the square root of 3 + √
4𝑖 in the first quadrant,
its conjugate
is
√
√
the square of root of 3 − 4𝑖 in the fourth quadrant, so is 3 − 4𝑖. Hence p3 + 4𝑖 and 3 − 4𝑖 are
√
√
conjugates to each other. Now we use Exercise 13.2.9. Note that 3 + 4𝑖 = 3 + −16. With 𝑎 = 3
2
and 𝑏 = −16, we have
√ is a square in Q and 𝑏 = −16 is not.
√ 𝑎 − 𝑏 = 25
√ Hence, we find 𝑚 = 1 and
𝑛√ = −4 and
thus
3
+
4𝑖
=
1
+
−4
=
1
+
2𝑖.
Furthermore,
we
find
3 − 4𝑖 = 1 − 2𝑖. Therefore,
√
√
√
3 + 4𝑖 + 3 − 4𝑖 = 4, i.e., 3 + 4𝑖 + 3 − 4𝑖 ∈ Q.
p √
p
√
(b) Let 𝜃 = 1 + −3 + 1 − −3. Then
q
q
√
√
√
√
√
2
𝜃 = ( 1 + −3 + 1 − −3) 2 = (1 + −3) + (2 1 + 3) + (1 − −3) = 6.
Since 𝑥 2 − 6 is irreducible over Q (Eisenstein with 𝑝 = 2), it follows that 𝜃 has degree 2 over Q.
8
Solutions by positrón0802
13.2
Algebraic Extensions
Exercise 13.2.12. Let 𝐸 be a subfield of 𝐾 containing 𝐹 . Then
[𝐾 : 𝐹 ] = [𝐾 : 𝐸] [𝐸 : 𝐹 ] = 𝑝.
Since 𝑝 is prime, either [𝐾 : 𝐸] = 1 or [𝐸 : 𝐹 ] = 1. The result follows.
Exercise 13.2.13. Note that, for all 1 ≤ 𝑘 ≤ 𝑛, √[Q(𝛼 1, . . . , 𝛼𝑘 ) : Q(𝛼 1, .√. . , 𝛼𝑘−1 )] is either
√3 1 or 2.
3
3
𝑚
Then [𝐹 : Q] = 2 for some 𝑚 ∈ N. Suppose 2 ∈ 𝐹 . Then
√ Q ⊂ Q( 2) ⊂ 𝐹, so [Q( 2) : Q]
divides [𝐹 : Q], that is, 3 divides 2𝑚 , a contradiction. Thus 3 2 ∉ 𝐹 .
Exercise 13.2.14. Since 𝛼 2 ∈ 𝐹 (𝛼), clearly 𝐹 (𝛼 2 ) ⊂ 𝐹 (𝛼). Thus we have to prove 𝛼 ∈ 𝐹 (𝛼 2 ). For
this purpose, consider the polynomial 𝑝 (𝑥) = 𝑥 2 − 𝛼 2, so that 𝑝 (𝛼) = 0. Note that 𝛼 ∈ 𝐹 (𝛼 2 ) if
and only if 𝑝 (𝑥) is reducible in 𝐹 (𝛼 2 ). For the sake of a contradiction, suppose 𝑝 (𝑥) is irreducible
in 𝐹 (𝛼 2 ), so that [𝐹 (𝛼) : 𝐹 (𝛼 2 )] = 2. Then
[𝐹 (𝛼) : 𝐹 ] = [𝐹 (𝛼) : 𝐹 (𝛼 2 )] [𝐹 (𝛼 2 ) : 𝐹 ] = 2[𝐹 (𝛼 2 ) : 𝐹 ],
so [𝐹 (𝛼) : 𝐹 ] is even, a contradiction. Therefore, 𝑝 (𝑥) is reducible in 𝐹 (𝛼 2 ) and 𝛼 ∈ 𝐹 (𝛼 2 ).
Exercise 13.2.15. We follow the hint. Suppose there exists a counterexample. Let 𝛼 be of minimal
degree such that 𝐹 (𝛼) is not formally real and 𝛼 having minimal polynomial 𝑓 of odd degree, say
deg 𝑓 = 2𝑘 + 1 for some 𝑘 ∈ N. Since 𝐹 (𝛼) is not formally real, −1 can be expressed as a sum of
squares in 𝐹 (𝛼) 𝐹 [𝑥]/(𝑓 (𝑥)). Then, there exist polynomials 𝑝 1 (𝑥), . . . , 𝑝𝑚 (𝑥), 𝑔(𝑥) such that
−1 + 𝑓 (𝑥)𝑔(𝑥) = (𝑝 1 (𝑥)) 2 + · · · + (𝑝𝑚 (𝑥)) 2 .
As every element in 𝐹 [𝑥]/((𝑓 (𝑥)) can be written as a polynomial in 𝛼 with degree less than deg 𝑓 ,
we have deg 𝑝𝑖 < 2𝑘 + 1 for all 𝑖. Thus, the degree in the right-hand side of the equation is less
than 4𝑘 +1, so deg 𝑔 < 2𝑘 +1 as well. From the equation −1+ 𝑓 (𝑥)𝑔(𝑥) = (𝑝 1 (𝑥)) 2 + · · · + (𝑝𝑚 (𝑥)) 2,
for proving that the degree of 𝑔 is odd it suffices to prove that degree of (𝑝 1 (𝑥)) 2 + · · · + (𝑝𝑚 (𝑥)) 2
is even. Let 𝑑 be the maximal degree over all 𝑝𝑖 , we prove that 𝑥 2𝑑 is the leading term of (𝑝 1 (𝑥)) 2 +
· · · + (𝑝𝑚 (𝑥)) 2 . Note that 𝑥 2𝑑 is a sum of squares (of the leading coefficients of the 𝑝𝑖 ’s of maximal
degree). Now, since 𝐹 is formally real, 0 cannot be expressed as a sum of squares in 𝐹 . (Indeed, if
Í𝑙
Í𝑙−1
2
2
2𝑑 ≠ 0, so the degree of (𝑝 (𝑥)) 2 + · · · + (𝑝 (𝑥)) 2 is
1
𝑚
𝑖=1 𝑎𝑖 = 0, then 𝑖=1 (𝑎𝑖 /𝑎𝑙 ) = −1.) Thus 𝑥
2𝑑, and therefore the degree of 𝑔 is odd. Then 𝑔 must contain an irreducible factor of odd degree,
say ℎ(𝑥). Since deg 𝑔 < deg 𝑓 , we have deg ℎ < deg 𝑓 as well. Let 𝛽 be a root of ℎ(𝑥), hence a
root of 𝑔(𝑥). Then
𝑓 (𝑥)𝑔(𝑥)
−1 + ℎ(𝑥)
= (𝑝 1 (𝑥)) 2 + · · · + (𝑝𝑚 (𝑥)) 2,
ℎ(𝑥)
so −1 is a square in 𝐹 [𝑥]/((ℎ(𝑥)) 𝐹 (𝛽), which means that 𝐹 (𝛽) is not formally real. It follows
that 𝛽 is a root of the odd degree polynomial ℎ such that 𝐹 (𝛽) is not formally real. Since deg ℎ <
deg 𝑓 , this contradicts the minimality of 𝛼 . The result follows.
9
Solutions by positrón0802
13.2
Algebraic Extensions
Exercise 13.2.16. Let 𝑟 ∈ 𝑅 be non-zero. Since 𝑟 is algebraic over 𝐹, there exists an irreducible
polynomial 𝑝 (𝑥) = 𝑎 0 + 𝑎 1𝑥 + · · · + 𝑥 𝑛 ∈ 𝐹 [𝑥] such that 𝑝 (𝑟 ) = 0. Note that 𝑎 0 ≠ 0 since 𝑝 is
𝑛−1 + · · · + 𝑎 ). Since 𝑎 ∈ 𝐹 ⊂ 𝑅 and 𝑟 ∈ 𝑅, we have 𝑟 −1 ∈ 𝑅.
irreducible. Then 𝑟 −1 = −𝑎 −1
1
𝑖
0 (𝑟
Exercise 13.2.17. Let 𝑝 (𝑥) be an irreducible factor of 𝑓 (𝑔(𝑥)) of degree 𝑚. Let 𝛼 be a root of
𝑝 (𝑥). Since 𝑝 is irreducible, it follows that [𝐹 (𝛼) : 𝐹 ] = deg 𝑝 (𝑥) = 𝑚. Now, since 𝑝 (𝑥) divides
𝑓 (𝑔(𝑥)), we have 𝑓 (𝑔(𝛼)) = 0 and thus 𝑔(𝛼) is a root of 𝑓 (𝑥). Since 𝑓 is irreducible, this means
𝑛 = [𝐹 (𝑔(𝛼)) : 𝐹 ]. Note that 𝐹 (𝑔(𝛼)) ⊂ 𝐹 (𝛼). Then
𝑚 = [𝐹 (𝛼) : 𝐹 ] = [𝐹 (𝛼) : 𝐹 (𝑔(𝛼))] [𝐹 (𝑔(𝛼)) : 𝐹 ] = [𝐹 (𝛼) : 𝐹 (𝑔(𝛼))] · 𝑛,
so 𝑛 divides 𝑚, that is, deg 𝑓 divides deg 𝑝.
Exercise 13.2.18. (a) We follow the hint. Since 𝑘 [𝑡] is an unique factorisation domain and 𝑘 (𝑡)
is its field of fractions, it follows from the Gauss Lemma that 𝑃 (𝑋 ) − 𝑡𝑄 (𝑋 ) is irreducible in
𝑘 ((𝑡)) [𝑋 ] if and only if it is irreducible in (𝑘 [𝑡]) [𝑋 ]. Note that (𝑘 [𝑡]) [𝑋 ] = (𝑘 [𝑋 ]) [𝑡]. Since
𝑃 (𝑋 ) −𝑡𝑄 (𝑋 ) is linear in (𝑘 [𝑋 ]) [𝑡], it is clearly irreducible in (𝑘 [𝑋 ]) [𝑡] (i.e., in (𝑘 [𝑡]) [𝑋 ]), hence
in (𝑘 (𝑡)) [𝑋 ]. Thus 𝑃 (𝑋 ) −𝑡𝑄 (𝑋 ) is irreducible in 𝑘 (𝑡). Finally, 𝑥 is clearly a root of 𝑃 (𝑋 ) −𝑡𝑄 (𝑋 )
𝑃 (𝑥)
since 𝑃 (𝑥) − 𝑡𝑄 (𝑥) = 𝑃 (𝑥) −
𝑄 (𝑥) = 𝑃 (𝑥) − 𝑃 (𝑥) = 0.
𝑄 (𝑥)
(b) Let 𝑛 = max{deg 𝑃 (𝑥), deg 𝑄 (𝑥)}. Write
𝑃 (𝑥) = 𝑎𝑛 𝑥 𝑛 + · · · + 𝑎 1𝑥 + 𝑎 0
and 𝑄 (𝑥) = 𝑏𝑛 𝑥 𝑛 + · · · + 𝑏 1𝑥 + 𝑏 0,
where 𝑎𝑖 , 𝑏𝑖 ∈ 𝑘 for all 𝑖, so at least one of 𝑎𝑛 or 𝑏𝑛 is non-zero. The degree of 𝑃 (𝑋 )−𝑡𝑄 (𝑋 ) is clearly
≤ 𝑛, we shall prove it is precisely 𝑛. If either 𝑎𝑛 or 𝑏𝑛 is zero then clearly deg (𝑃 (𝑋 )−𝑡𝑄 (𝑋 )) = 𝑛, so
assume 𝑎𝑛 , 𝑏𝑛 ≠ 0. Then 𝑎𝑛 , 𝑏𝑛 ∈ 𝑘, but 𝑡 ∉ 𝑘, so it cannot be that 𝑎𝑛 = 𝑡𝑏𝑛 . Thus (𝑎𝑛 − 𝑡𝑏𝑛 )𝑋 𝑛 ≠ 0
and the degree of 𝑃 (𝑋 ) − 𝑡𝑄 (𝑋 ) is precisely 𝑛.
(c) Since 𝑃 (𝑋 ) − 𝑡𝑄 (𝑋 ) is irreducible over 𝑘 (𝑡) and 𝑥 is a root by part (a), it follows that
[𝑘 (𝑥) : 𝑘 (𝑡)] = deg 𝑃 (𝑋 ) − 𝑡𝑄 (𝑋 ), and this degree equals max{deg 𝑃 (𝑥), deg 𝑄 (𝑥)} by part (b).
Exercise 13.2.19. (a) Fix 𝛼 in 𝐾 . Since 𝐾 is (in particular) a commutative ring, we have 𝛼 (𝑎 +𝑏) =
𝛼𝑎 + 𝛼𝑏 and 𝛼 (𝜆𝑎) = 𝜆(𝛼𝑎) for all 𝑎, 𝑏, 𝜆 ∈ 𝐾 . If, in particular, 𝜆 ∈ 𝐹, we have the result.
(b) Fix a basis for 𝐾 as a vector space over 𝐹 . By part (a), for every 𝛼 ∈ 𝐾 we can associate
an 𝐹 -linear transformation 𝑇𝛼 of 𝐾 . Denote by 𝑇𝛼 the matrix
of 𝑇𝛼 with respect to the basis
previously fixed. Then define 𝜑 : 𝐾 → 𝑀𝑛 (𝐹 ) by 𝜑 (𝛼) = 𝑇𝛼 . We claim that 𝜑 is an isomorphism
onto its image. Indeed, if 𝛼, 𝛽 ∈ 𝐾, then 𝑇 (𝛼+𝛽) (𝑘) = (𝛼 + 𝛽) (𝑘) = 𝛼𝑘 + 𝛽𝑘 = 𝑇𝛼 (𝑘) + 𝑇𝛽 (𝑘) for
every 𝑘 ∈ 𝐾, so 𝑇 (𝛼+𝛽) = 𝑇𝛼 + 𝑇𝛽 . We also have 𝑇 (𝛼 𝛽) (𝑘) = (𝛼𝛽) (𝑘) = 𝛼 (𝛽𝑘) = 𝑇𝛼 𝑇𝛽 (𝑘) for every
𝑘 ∈ 𝐾, so 𝑇 (𝛼 𝛽) = 𝑇𝛼 𝑇𝛽 . Thus 𝜑 (𝛼 + 𝛽) = 𝜑 (𝛼) + 𝜑 (𝛽) and 𝜑 (𝛼𝛽) = 𝜑 (𝛼)𝜑 (𝛽) (since the basis is
fixed), so 𝜑 is an homomorphism. Now, if 𝜑 (𝛼) = 𝜑 (𝛽), then 𝛼𝑘 = 𝛽𝑘 for every 𝑘 ∈ 𝐾, so letting
𝑘 = 1 we find that 𝜑 is injective. Therefore, 𝜑 (𝐾) is isomorphic to a subfield of 𝑀𝑛 (𝐹 ), so the ring
𝑀𝑛 (𝐹 ) contains an isomorphic copy of every extension of 𝐹 of degree ≤ 𝑛.
10
Solutions by positrón0802
13.2
Algebraic Extensions
Exercise 13.2.20. The characteristic polynomial of 𝐴 is 𝑝 (𝑥) = det(𝐼𝑥 − 𝐴). For every 𝑘 ∈ 𝐾, we
have (𝐼𝛼 − 𝐴)𝑘 = 𝛼𝑘 − 𝐴𝑘 = 𝛼𝑘
𝐴) = 0 in 𝐾 . Therefore 𝑝 (𝛼) = 0.
√3 − 𝛼𝑘 = 0, so det(𝐼𝛼
√3 √−
3
Now, consider
the
field
Q(
2)
with
basis
{1,
2,
4}
Q. Denote the elements of this basis
√3
√3
√3
√over
√
3
by 𝑒 1 = 1, 𝑒 2 = 2 and 𝑒 3 = 4. Let 𝛼 = 2 and 𝛽 = 1 + 2 + 3 4. Then 𝛼 (𝑒 1 ) = 𝑒 2, 𝛼 (𝑒 2 ) = 𝑒 3 and
𝛼 (𝑒 3 ) = 2𝑒 1 . We also have 𝛽 (𝑒 1 ) = 𝑒 1 + 𝑒 2 + 𝑒 3, 𝛽 (𝑒 2 ) = 2𝑒 1 + 𝑒 2 + 𝑒 3 and 𝛽 (𝑒 3 ) = 2𝑒 1 + 2𝑒 2 + 𝑒 3 . Thus,
the associated matrices of the their linear transformations are, respectively,
0 0 2
©
ª
𝐴𝛼 = ­1 0 0®
«0 1 0¬
and
1 2 2
©
ª
𝐴 𝛽 = ­1 1 2 ® .
«1 1 1 ¬
The characteristic
polynomial of 𝐴𝛼 is 𝑥 3 − 2, hence is the monic polynomial of degree 3 satisfied
√3
by 𝛼 = 2. Furthermore, the characteristic polynomial of 𝐴𝛽 is 𝑥 3 − 3𝑥 2 − 3𝑥 − 1, hence is the
√ √
monic polynomial of degree 3 satisfied by 𝛽 = 1 + 3 2 + 3 4.
Exercise 13.2.21. The matrix
“multiplication
√ of the linear transformation √
√ by 𝛼”
√ on 𝐾 is found
by acting of 𝛼 in the basis 1, 𝐷. We have 𝛼 (1) = 𝛼 = 𝑎 + 𝑏 𝐷 and 𝛼 ( 𝐷) = 𝑎 𝐷 + 𝑏𝐷. Hence
√
𝑎 𝑏𝐷
𝑎 𝑏𝐷
. Now let 𝜑 : 𝐾 → 𝑀2 (Q) be defined by 𝜑 (𝑎 + 𝑏 𝐷) =
.
the matrix is
𝑏 𝑎
𝑏 𝑎
We have
√
√
√
√
𝑐 𝑑𝐷
𝑎 𝑏𝐷
𝑎 + 𝑐 (𝑏 + 𝑑)𝐷
= 𝜑 (𝑎 + 𝑏 𝐷) + 𝜑 (𝑐 + 𝑑 𝐷),
+
=
𝜑 (𝑎 + 𝑏 𝐷 + 𝑐 + 𝑑 𝐷) =
𝑑 𝑐
𝑏 𝑎
𝑏 +𝑑
𝑎 +𝑐
and
√
√
𝑎𝑐 + 𝑏𝑑𝐷
𝜑 ((𝑎 + 𝑏 𝐷) · (𝑐 + 𝑑 𝐷)) =
𝑎𝑑 + 𝑏𝑐
(𝑎𝑑 + 𝑏𝑐)𝐷
𝑎 𝑏𝐷
𝑐 𝑑𝐷
=
·
𝑎𝑐 + 𝑏𝑑𝐷
𝑏 𝑎
𝑑 𝑐
√
√
= 𝜑 (𝑎 + 𝑏 𝐷)𝜑 (𝑐 + 𝑑 𝐷),
so 𝜑 is an homomorphism. Since 𝐾 is a field, its only ideals are {0} and 𝐾, so ker(𝜑) is either
trivial or all of 𝐾 . But 𝜑 (𝐾) is clearly non-zero, so ker(𝜑) ≠ 𝐾 and thus ker(𝜑) = {0}. Hence 𝜑 is
injective. It follows that 𝜑 is an isomorphism of 𝐾 with a subfield of 𝑀2 (Q).
Exercise 13.2.22. Define 𝜑 : 𝐾1 × 𝐾2 → 𝐾1𝐾2 by 𝜑 (𝑎, 𝑏) = 𝑎𝑏. We prove that 𝜑 is 𝐹 -bilinear. Let
𝑎, 𝑎 1, 𝑎 2 ∈ 𝐾 and 𝑏, 𝑏 1, 𝑏 2 ∈ 𝐾2 . Then
𝜑 ((𝑎 1, 𝑏) + (𝑎 2, 𝑏)) = 𝜑 (𝑎 1 + 𝑎 2, 𝑏) = (𝑎 1 + 𝑎 2 )𝑏 = 𝑎 1𝑏 + 𝑎 2𝑏 = 𝜑 (𝑎 1, 𝑏) + 𝜑 (𝑎 2, 𝑏),
and
𝜑 ((𝑎,1 𝑏) + (𝑎, 𝑏 2 )) = 𝜑 (𝑎, 𝑏 1 + 𝑏 2 ) = 𝑎(𝑏 1 + 𝑏 2 ) = 𝑎𝑏 1 + 𝑎𝑏 1 = 𝜑 (𝑎, 𝑏 1 ) + 𝜑 (𝑎, 𝑏 2 ).
We also have, for 𝑟 ∈ 𝐹, 𝜑 (𝑎𝑟, 𝑏) = (𝑎𝑟 )𝑏 = 𝑎(𝑟𝑏) = 𝜑 (𝑟𝑏). Thus 𝜑 is a 𝐹 -bilinear map, so it induces
an 𝐹 -algebra homomorphism Φ : 𝐾1 ⊗𝐹 𝐾2 → 𝐾1𝐾2 . We shall use Φ to prove both directions. Note
that 𝐾1 ⊗𝐹 𝐾2 have dimension [𝐾1 : 𝐹 ] [𝐾2 : 𝐹 ] as a vector space over 𝐹 .
11
Solutions by positrón0802
13.3
Classical Straightedge and Compass Constructions
First, we assume [𝐾1𝐾2 : 𝐹 ] = [𝐾1 : 𝐹 ] [𝐾2 : 𝐹 ] and shall prove that 𝐾1 ⊗𝐹 𝐾2 is a field. In this
case 𝐾1 ⊗𝐹 𝐾2 and 𝐾1𝐾2 have the same dimension over 𝐹 . Let 𝐿 = Φ(𝐾1 ⊗𝐹 𝐾2 ). We claim 𝐿 = 𝐾1𝐾2,
i.e. Φ is surjective. Note that 𝐿 contains 𝐾1 and 𝐾2 . Since 𝐿 is a subring of 𝐾1𝐾2 containing 𝐾1
(or 𝐾2 ), it follows that 𝐿 is a field (Exercise 13.2.16). Hence 𝐿 is a field containing both 𝐾1 and 𝐾2,
and since 𝐾1𝐾2 is the smallest such field (by definition), we have 𝐿 = 𝐾1𝐾2 . So Φ is surjective, as
claimed. It follows that Φ is an 𝐹 -algebra surjective homomorphism between 𝐹 -algebras of the
same dimension, hence is an isomorphism. Thus, 𝐾1 ⊗𝐹 𝐾2 is a field.
Now assume that 𝐾1 ⊗𝐹 𝐾2 is a field. In this case Φ is a field homomorphism, so it either
injective or trivial. It is clearly non-trivial since Φ(1 ⊗ 1) = 1, so it must be injective. Hence,
[𝐾1 : 𝐹 ] [𝐾2 : 𝐹 ] ≤ [𝐾1𝐾2 : 𝐹 ]. As we already have [𝐾1𝐾2 : 𝐹 ] ≤ [𝐾1 : 𝐹 ] [𝐾2 : 𝐹 ] (by Proposition
21 in the book), the equality follows.
13.3
Classical Straightedge and Compass Constructions
Exercise 13.3.1. Suppose that the 9-gon is constructible. It has angles of 40◦ . Since we can bisect
an angle by straightedge and compass, the angle of 20◦ would be constructible. But then cos 20◦
and sin 20◦ would be constructible too, a contradiction (see proof of Theorem 24).
Exercise 13.3.2. Let 𝑂, 𝑃, 𝑄 and 𝑅 be the points marked in the figure below.
Then 𝛼 = ∠𝑄𝑃𝑂, 𝛽 = ∠𝑅𝑄𝑂, 𝛾 = ∠𝑄𝑅𝑂, and 𝜃 is an exterior angle of 4𝑃𝑅𝑂. Since 4𝑃𝑄𝑂
is isosceles, 𝛼 = ∠𝑄𝑃𝑂 = ∠𝑄𝑂𝑃 . Since 𝛽 is an exterior angle of 4𝑃𝑄𝑂, it equals the sum of the
two remote interior angles, i.e., equals ∠𝑄𝑃𝑂 + ∠𝑄𝑂𝑃 . These two angles equal 𝛼, so 𝛽 = 2𝛼 . Now,
4𝑄𝑅𝑂 is isosceles, so 𝛽 = 𝛾 . Finally, since 𝜃 is an exterior angle of 4𝑃𝑅𝑂, it equals the sum of the
two remote interior angles, which are 𝛼 and 𝛾 . It follows that 𝜃 = 𝛼 + 𝛾 = 𝛼 + 𝛽 = 3𝛼 .
Exercise 13.3.3. We follow the hint. The distances 𝑎, 𝑏, 𝑥, 𝑦 and 𝑥 − 𝑘 are marked in the figure
below.
12
Solutions by positrón0802
13.4
Splitting Fields and Algebraic Closures
From the figure, using similar triangles for (a), (b) and (c), and Pythagoras Theorem for (d),
the 4 relations are clear, that is,
√
√
𝑦
1 − 𝑘2
𝑏 +𝑘
1 − 𝑘2
𝑦=
, 𝑥 =𝑎
,
=
and (1 − 𝑘 2 ) + (𝑏 + 𝑘) 2 = (1 + 𝑎) 2 .
1+𝑎
1+𝑎
𝑥 −𝑘
3𝑘
√
3𝑘𝑦
Thus 𝑦 (1 + 𝑎) = 1 − 𝑘 2 = 𝑥−𝑘 , which implies 3𝑘 = (𝑥 − 𝑘) (1 + 𝑎). From the equation for 𝑥 above,
4𝑘+𝑘𝑎
we find 3𝑘 = ( 𝑎 (𝑏+𝑎)
1+𝑎 − 𝑘) (1 + 𝑎) = 𝑎(𝑏 + 𝑘) − 𝑘 (1 + 𝑎), so 𝑏 + 𝑘 =
𝑎 . Using this in the last
equation and reducing, we obtain
⇒
⇒
(1 − 𝑘 2 ) + (𝑏 + 𝑘) 2 = (1 + 𝑎) 2
2
4𝑘 + 𝑘𝑎
2
(1 − 𝑘 ) +
= (1 + 𝑎) 2
𝑎
𝑎 2 (1 − 𝑘 2 ) + (4𝑘 + 𝑘𝑎) 2 = 𝑎 2 (1 + 𝑎) 2
⇒ 𝑎 2 − (𝑘𝑎) 2 + (4𝑘) 2 + 8𝑘 2𝑎 + (𝑘𝑎) 2 = 𝑎 2 + 2𝑎 3 + 𝑎 4
⇒
𝑎 4 + 2𝑎 3 − 8𝑘 2𝑎 − 16𝑘 2 = 0.
We let 𝑎 = 2ℎ to obtain
ℎ 4 + ℎ 3 − 𝑘 2ℎ − 𝑘 2 = 0.
We find ℎ = 𝑘 2/3, so that 𝑎 = 2𝑘 2/3 . Finally, from 𝑏 = 4𝑘+𝑘𝑎
− 𝑘 we find 𝑏 = 2𝑘 1/3 . It follows that
𝑎
1/3
2/3
we can construct 2𝑘 and 2𝑘 using Conway’s construction.
Exercise 13.3.4. Let 𝑝 (𝑥) = 𝑥 3 + 𝑥 2 − 2𝑥 − 1 and 𝛼 = 2 cos(2𝜋/7). By the Rational Root Theorem,
if 𝑝 has a root in Q, it must be ±1 since it must divide its constant term. But 𝑝 (1) = −1 and
𝑝 (−1) = 1, so 𝑝 is irreducible over Q. Therefore, 𝛼 is of degree 3 over Q, so [Q(𝛼) : 𝑄] cannot be
a power of 2. Since we cannot construct 𝛼, it follows that the regular 7-gon is not constructible
by straightedge and compass.
Exercise 13.3.5. Let 𝑝 (𝑥) = 𝑥 2 + 𝑥 − 1 = 0 and 𝛼 = 2 cos(2𝜋/5). By the Rational Root Theorem, if 𝑝 has a root in Q, it must be ±1. Since 𝑝 (1) = 1 and 𝑝 (−1) = −1, we deduce that 𝑝
is irreducible over Q. Hence 𝛼 is of degree 2 over Q, so it is constructible. We can bisect an
angle by straightedge and compass, so 𝛽 = cos(2𝜋/5) is also constructible. Finally, as sin(2𝜋/5) =
p
1 − cos2 (2𝜋/5), sin(2𝜋/5) is also constructible. We conclude that the regular 5-gon is constructible by straightedge and compass.
13.4
Splitting Fields and Algebraic Closures
√ √ √
√
Exercise 13.4.1.√Let 𝑓 (𝑥) = 𝑥 4 −2. The roots of 𝑓√are 4 2, − 4 2, 𝑖 4 2√and −𝑖 4√2. Hence,√the splitting
field of 𝑓 √
is Q(𝑖, 4 2). This field has degree [Q(𝑖, 4 2) : Q] = [Q(𝑖, 4 2) : Q( 4 2)] [Q(√ 4 2) : Q] over
Q. Since 4 2 is a root√of the irreducible polynomial 𝑥 4 − √
2 over Q, we have [Q( 4 2) : Q] √= 4.
Furthermore,
𝑖 ∉ Q( 4 2), so 𝑥 2 √
+ 1 is irreducible over Q( 4 2) having 𝑖 as a root. So [Q(𝑖, 4 2) :
√4
Q( 2)] = 2 and therefore [Q(𝑖, 4 2) : Q] = 8.
13
Solutions by positrón0802
13.4
Splitting Fields and Algebraic Closures
Exercise 13.4.2. Let 𝑓 (𝑥)√= 𝑥 4 + 2. Let 𝐾 be the splitting field of 𝑓 and let 𝐿 be the splitting field
of 𝑥 4 − 2, that is, 𝐿 = Q(𝑖, 4 2) (Exercise 13.4.1). We claim 𝐾 = 𝐿, so that [𝐾 : Q] = 8 by Exercise
13.4.1.
√
√
√
Let 𝜁 = 22 + 𝑖 22 . First we prove 𝜁 ∈ 𝐿 and 𝜁 ∈ 𝐾 . That 𝜁 ∈ 𝐿 is easy: let 𝜃 = 4 2. Since
√
√
𝜃 ∈ 𝐿, we have 𝜃 2 = 2 ∈ 𝐿. We
√ also have 𝑖 ∈ 𝐿, so 2, 𝑖 4 ∈ 𝐿 implies 𝜁 ∈ 𝐿. We 4now prove
𝜁 ∈ 𝐾 . We shall prove 𝑖 ∈ 𝐾 and 2 ∈ 𝐾 . Let 𝛼 be a root of 𝑥 + 2 and 𝛽 be a root of 𝑥 − 1. Then
(𝛼𝛽) 4 = 𝛼 4 𝛽 4 = −2, so 𝛼𝛽 is also a root of 𝑥 4 + 2. Since the roots of 𝑥 4 − 1 are ±1, ±𝑖, the roots of
𝑥 4 + 2 are ±𝛼 and ±𝑖𝛼 . Since 𝐾 is generated over Q by these roots, it follows that 𝑖𝛼/𝛼 = 𝑖 ∈ 𝐾 .
2
Now√let 𝛾 = 𝛼 2√∈ 𝐾 . As 𝛾 2 = 𝛼 4 = −2, we have that 𝛾 is a root of 𝑥 2 + 2. Since the roots
√ of 𝑥 + 2
are 𝑖 2 and −𝑖 2, 𝛾 must be one of this roots. In either case 𝛾/𝑖 ∈ 𝐾, which implies 2 ∈ 𝐾, and
therefore 𝜁 ∈ 𝐾 .
Now we prove 𝐿 = 𝐾 . On the one hand, let 𝛼 be a root of 𝑥 4 + 2 and 𝜃 be a root of 𝑥 4 − 2. Then
4
𝛼 = −2 and 𝜃 4 = 2. Note that 𝜁 2 = 𝑖, so 𝜁 4 = −1. Hence (𝜁 𝜃 ) 4 = 𝜁 4𝜃 4 = −2, so 𝜁 𝜃 is a root of 𝑥 4 +2.
Then, as we proved earlier, the roots of 𝑥 4 + 2 are ±𝜁 𝜃 and ±𝑖𝜁 𝜃 . We also have (𝜁 𝛼) 4 = 𝜁 4𝛼 4 = 2,
so 𝜁 𝛼 is a root of 𝑥 4 − 2. Then, by Exercise 13.4.1, the roots of 𝑥 4 − 2 are ±𝜁 𝛼 and ±𝑖𝜁 𝛼 . Now, since
𝜁 and 𝛼 are in 𝐾, we have 𝜁 𝛼 ∈ 𝐾 . We also have 𝑖 ∈ 𝐾, so all the roots of 𝑥 4 − 2 are in 𝐾 . Since 𝐿
is generated by these roots, it follows that 𝐿 ⊂ 𝐾 . Similarly, 𝜁 and 𝜃 belong to 𝐿, so 𝜁 𝜃 ∈ 𝐿; since
𝑖 ∈ 𝐿, all the roots of 𝑥 4 + 2 are in 𝐿. Since 𝐾 is generated by these roots, it follows that 𝐾 ⊂ 𝐿.
We conclude that 𝐾 = 𝐿; in particular, [𝐾 : Q] = 8.
Exercise 13.4.3.
Let 𝑓 (𝑥) = 𝑥√4 + 𝑥 2 + 1. Note that 𝑓 (𝑥) = (𝑥 2 + 𝑥 + 1) (𝑥 2 − 𝑥 + 1), so the roots of 𝑓
√
are ± 12 ± 𝑖 23 . Write 𝑤 = 21 − 𝑖 23 , so that these roots are precisely 𝑤, −𝑤, 𝑤, −𝑤, where 𝑤 denotes
the complex conjugate of 𝑤 . Hence, the splitting field of 𝑓 is Q(𝑤, 𝑤). Since 𝑤 + 𝑤 = 1, we have
Q(𝑤, 𝑤) = Q(𝑤). Furthermore, 𝑤 is a root of 𝑥 2 − 𝑥 + 1, which is irreducible over Q since 𝑤 ∉ Q.
Therefore, the degree of the splitting field of 𝑓 is [Q(𝑤) : Q] = 2.
√
Exercise
13.4.4.
Let 𝑓 (𝑥) = 𝑥 6 − 4. Note that 𝑓 (𝑥) = (𝑥 3 − 2) (𝑥 3 + 2). The roots of 𝑥 3 − 2 are 3 2,
√3
√
𝜁 2 and 𝜁 2 3 2, where
𝜁 denotes the primitive 3rd root of unity, i.e., 𝜁 = exp(2𝜋𝑖/3) = cos(2𝜋/3) +
√
√
√
√
3
1
𝑖 sin(2𝜋/3) = − 2 + 2 . Furthermore, the roots of 𝑥 3 + 2 are − 3 2, −𝜁 3 2 and −𝜁 2 3 2. Therefore, the
√
√
√
√
√
√
splitting field of 𝑓 is Q(𝜁 , 3 2). We have [Q(𝜁 , 3 2) : Q]
= [Q(𝜁 , 3 2) : Q( 3 2)] [Q( 3 2) : Q]. As 3 2
√
Furthermore, 𝜁 is a
is a root of the irreducible polynomial 𝑥 3 − 2 over
Q, 3 2 has degree 3 over Q. √
√
3
3
2
root of 𝑥 + 𝑥 + 1, which is irreducible over Q(
√3 2), so 𝜁 has degree 2 over Q( 2). It follows that
the degree of the splitting field of 𝑓 is [Q(𝜁 , 2) : Q] = 6.
Exercise 13.4.5. We follow the hint. First assume that 𝐾 is a splitting field over 𝐹 . Then there
exists 𝑓 (𝑥) ∈ 𝐹 [𝑥] such that 𝐾 is the splitting field of 𝑓 . Let 𝑔(𝑥) be an irreducible polynomial
in 𝐹 [𝑥] with a root 𝛼 ∈ 𝐾 . Let 𝛽 be any root of 𝑔. We prove 𝛽 ∈ 𝐾, so that 𝑔 splits completely in
∼
𝐾 [𝑥]. By Theorem 8, there is an isomorphism 𝜑 : 𝐹 (𝛼) −
→ 𝐹 (𝛽) such that 𝜑 (𝛼) = 𝛽. Furthermore,
𝐾 (𝛼) is the splitting field of 𝑓 over 𝐹 (𝑎), and 𝐾 (𝛽) is the splitting field of 𝑓 over 𝐹 (𝛽). Therefore,
∼
by Theorem 28, 𝜑 extends to an isomorphism 𝜎 : 𝐾 (𝛼) −
→ 𝐾 (𝛽). Since 𝐾 = 𝐾 (𝛼), we have
[𝐾 : 𝐹 ] = [𝐾 (𝛼) : 𝐹 ] = [𝐾 (𝛽) : 𝐹 ], and therefore 𝐾 = 𝐾 (𝛽). Thus 𝛽 ∈ 𝐾 .
14
Solutions by positrón0802
13.5
Separable and Inseparable Extension
Conversely, assume that every irreducible polynomial in 𝐹 [𝑥] that has a root in 𝐾 splits completely in 𝐾 [𝑥]. Since [𝐾 : 𝐹 ] is finite, we have 𝐾 = 𝐹 (𝛼 1, . . . , 𝛼𝑛 ) for some 𝛼 1, . . . , 𝛼𝑛 . For every
1 ≤ 𝑖 ≤ 𝑛, let 𝑝𝑖 be the minimal polynomial of 𝛼𝑖 over 𝐹, and let 𝑓 = 𝑝 1𝑝 2 · · · 𝑝𝑛 . Since every 𝛼𝑖
is in 𝐾, every 𝑝𝑖 has a root in 𝐾, hence splits completely in 𝐾 . Therefore, 𝑓 splits completely in 𝐾
and 𝐾 is generated over 𝐹 by its roots, so 𝐾 is the splitting field of 𝑓 (𝑥) ∈ 𝐹 [𝑥].
Exercise 13.4.6. (a) Let 𝐾1 be the splitting field of 𝑓1 (𝑥) ∈ 𝐹 [𝑥] over 𝐹 and 𝐾2 be the splitting
field of 𝑓2 (𝑥) ∈ 𝐹 [𝑥] over 𝐹 . Thus 𝐾1 is generated over 𝐹 by the roots of 𝑓1, and 𝐾2 is generated
over 𝐹 by the roots of 𝑓2 . Then 𝑓1 𝑓2 splits completely in 𝐾1𝐾2 and 𝐾1𝐾2 is generated over 𝐹 by its
roots, hence is the splitting field of 𝑓1 𝑓2 (𝑥) ∈ 𝐹 [𝑥].
(b) We follow the hint. By Exercise 13.4.5, we shall prove that every irreducible polynomial in
𝐹 [𝑥] that has a root in 𝐾1 ∩ 𝐾2 splits completely in (𝐾1 ∩ 𝐾2 ) [𝑥]. Thus, let 𝑓 (𝑥) be an irreducible
polynomial in 𝐹 [𝑥] that has a root, say 𝛼, in 𝐾1 ∩ 𝐾2 . By Exercise 13.4.5, 𝑓 splits completely in
𝐾1 and splits completely in 𝐾2 . Since 𝐾1 and 𝐾2 are contained in 𝐾, by the uniqueness of the
factorisation of 𝑓 in 𝐾, the roots of 𝑓 in 𝐾1 must coincide with its roots in 𝐾2 . It follows that 𝑓
splits completely in (𝐾1 ∩ 𝐾2 ) [𝑥].
13.5
Separable and Inseparable Extension
Exercise 13.5.1. Let 𝑓 (𝑥) = 𝑎𝑛 𝑥 𝑛 + · · · + 𝑎 1𝑥 + 𝑎 0 and 𝑔(𝑥) = 𝑏𝑚 𝑥 𝑚 + · · · + 𝑏 1𝑥 + 𝑏 0 be two
polynomials. We may assume, without any loss of generality, that 𝑛 ≥ 𝑚. Thus, we can write
𝑔(𝑥) = 𝑏𝑛 𝑥 𝑛 + · · · + 𝑏 1𝑥 + 𝑏 0, where some of the last coefficients 𝑏𝑖 could be zero. We have
𝑓 (𝑥) + 𝑔(𝑥) = (𝑎𝑛 + 𝑏𝑛 )𝑥 𝑛 + · · · + (𝑎 1 + 𝑏 1 )𝑥 + (𝑎 0 + 𝑏 0 ), so
𝐷𝑥 (𝑓 (𝑥) + 𝑔(𝑥)) = 𝑛(𝑎𝑛 + 𝑏𝑛 )𝑥 𝑛−1 + · · · + 2(𝑎 2 + 𝑏 2 )𝑥 + (𝑎 1 + 𝑏 1 ) = 𝐷𝑥 (𝑓 (𝑥)) + 𝐷𝑥 (𝑔(𝑥)).
Í
Now, for ℓ = 1, . . . , 2𝑛, set 𝑐𝑛 = 𝑛𝑘=0 𝑎𝑘 𝑏𝑛−𝑘 , where we also set 𝑎𝑖 = 0 and 𝑏𝑖 = 0 if 𝑖 > 𝑛. Then
!
! 𝑛
2𝑛 Õ
ℓ
2𝑛
𝑛
Õ
Õ
Õ
Õ
𝑗
𝑖
𝑏𝑗𝑥 =
( 𝑎𝑘 𝑏 ℓ−𝑘 )𝑥 ℓ =
𝑐ℓ 𝑥 ℓ ,
𝑎𝑖 𝑥
𝑓 (𝑥)𝑔(𝑥) =
𝑗=0
𝑖=0
so that
𝐷𝑥 (𝑓 (𝑥)𝑔(𝑥)) = 𝐷𝑥
ℓ=0 𝑘=0
2𝑛
Õ
!
𝑐ℓ 𝑥
ℓ
=
ℓ=0
2𝑛−1
Õ
ℓ=0
(ℓ + 1)𝑐 ℓ+1𝑥 ℓ .
ℓ=0
Thus, the coefficient of 𝑥 ℓ in 𝐷𝑥 (𝑓 (𝑥)𝑔(𝑥)) is (ℓ + 1)𝑐 ℓ+1 . On the other hand, we have
! 𝑛
! 2𝑛−1 ℓ
!
𝑛−1
Õ
Õ Õ
Õ
𝑘
𝑘
(𝑘 + 1)𝑎𝑘+1𝑏 ℓ−𝑘 𝑥 ℓ
(𝑘 + 1)𝑎𝑘+1𝑥
𝑏𝑘 𝑥 =
𝐷𝑥 (𝑓 (𝑥)) 𝑔(𝑥) =
𝑘=0
ℓ=0
𝑘=0
𝑘=0
and
𝑓 (𝑥)𝐷𝑥 (𝑔(𝑥)) =
𝑛
Õ
𝑘=0
!
𝑎𝑘 𝑥 𝑘
𝑛−1
Õ
!
(𝑘 + 1)𝑏𝑘+1𝑥 𝑘 =
2𝑛−1
Õ
(
ℓ
Õ
𝑎𝑘 (ℓ − 𝑘 + 1)𝑏 ℓ−𝑘+1 )𝑥 ℓ ,
ℓ=0 𝑘=0
𝑘=0
15
Solutions by positrón0802
13.5
Separable and Inseparable Extension
so the coefficient of 𝑥 ℓ in 𝐷𝑥 (𝑓 (𝑥))𝑔(𝑥) + 𝐷𝑥 (𝑔(𝑥)) 𝑓 (𝑥) is
!
!
ℓ
ℓ
Õ
Õ
(𝑘 + 1)𝑎𝑘+1𝑏 ℓ−𝑘 +
(ℓ − 𝑘 + 1)𝑎𝑘 𝑏 ℓ−𝑘+1
𝑘=0
𝑘=0
= (ℓ + 1)𝑎 ℓ+1𝑏 0 +
= (ℓ + 1)𝑎 ℓ+1𝑏 0 +
= (ℓ + 1)𝑎 ℓ+1𝑏 0 +
ℓ−1
Õ
𝑘=0
ℓ
Õ
𝑘=1
ℓ
Õ
!
(𝑘 + 1)𝑎𝑘+1𝑏 ℓ−𝑘 +
ℓ
Õ
!
(ℓ − 𝑘 + 1)𝑎𝑘 𝑏 ℓ−𝑘+1 + (ℓ + 1)𝑎 0𝑏 ℓ+1
𝑘=1
!
𝑘𝑎𝑘 𝑏 ℓ−𝑘+1 +
ℓ
Õ
!
(ℓ − 𝑘 + 1)𝑎𝑘 𝑏 ℓ−𝑘+1 + (ℓ + 1)𝑎 0𝑏 ℓ+1
𝑘=1
!
(ℓ + 1)𝑎𝑘 𝑏 ℓ−𝑘+1 + (ℓ + 1)𝑎 0𝑏 ℓ+1
𝑘=1
= (ℓ + 1)
ℓ+1
Õ
!
𝑎𝑘 𝑏 ℓ−𝑘+1 = (ℓ + 1)𝑐 ℓ+1 .
𝑘=0
We deduce that 𝐷𝑥 (𝑓 (𝑥)𝑔(𝑥)) = 𝐷𝑥 (𝑓 (𝑥))𝑔(𝑥) + 𝐷𝑥 (𝑔(𝑥))𝑓 (𝑥).
Exercise 13.5.2. The polynomials 𝑥 and 𝑥 + 1 are the only (non-constant, i.e. ≠ 0, 1) polynomials
of degree 1 over F2 ; they are clearly irreducible. A polynomial 𝑓 (𝑥) ∈ F2 [𝑥] of degree 2 is
irreducible over F2 if and only if it does not have a root in F2, that is, exactly if 𝑓 (0) = 𝑓 (1) = 1.
Hence, the only irreducible polynomial of degree 2 over F2 is 𝑥 2 + 𝑥 + 1. Now, for a polynomial
𝑓 (𝑥) ∈ F2 [𝑥] of degree 4 to be irreducible, it must have no linear or quadratic factors. We can
also apply the condition 𝑓 (1) = 𝑓 (0) = 1 to discard the ones with linear factors. Furthermore,
𝑓 must have an odd number of terms (otherwise it would be 0), and must have constant term 1
(otherwise 𝑥 would be a factor). We are left with
𝑥 4 + 𝑥 3 + 𝑥 2 + 𝑥 + 1,
𝑥 4 + 𝑥 2 + 1,
𝑥 4 + 𝑥 3 + 1,
𝑥 4 + 𝑥 + 1.
For any of these polynomials to be irreducible, it cannot be factorised as a product of two quadratic
irreducible factors. Since 𝑥 2 + 𝑥 + 1 is the only irreducible polynomial of degree 2 over F2, only
(𝑥 2 + 𝑥 + 1) 2 = 𝑥 4 + 𝑥 2 + 1 of these four is not irreducible. Hence, the irreducible polynomials of
degree 4 over F2 are precisely 𝑥 4 + 𝑥 3 + 𝑥 2 + 𝑥 + 1, 𝑥 4 + 𝑥 3 + 1 and 𝑥 4 + 𝑥 + 1.
Now, since 𝑥 + 1 = 𝑥 − 1 in F2, we have (𝑥 + 1) (𝑥 4 + 𝑥 3 + 𝑥 2 + 𝑥 + 1) = 𝑥 5 − 1. We also have
2
(𝑥 + 𝑥 + 1) (𝑥 4 + 𝑥 + 1) (𝑥 4 + 𝑥 3 + 1) = 𝑥 10 + 𝑥 5 + 1. It follows that the product of all these irreducible
polynomials is
𝑥 (𝑥 + 1) (𝑥 2 + 𝑥 + 1) (𝑥 4 + 𝑥 + 1) (𝑥 4 + 𝑥 3 + 1) (𝑥 4 + 𝑥 3 + 𝑥 2 + 𝑥 + 1)
= 𝑥 (𝑥 5 − 1) (𝑥 10 + 𝑥 5 + 1) = 𝑥 16 − 𝑥 .
Exercise 13.5.3. We follow the hint. First assume that 𝑑 divides 𝑛, so that 𝑛 = 𝑞𝑑 for some 𝑞 ∈ Z+ .
Then 𝑥 𝑛 − 1 = 𝑥 𝑞𝑑 − 1 = (𝑥 𝑑 − 1) (𝑥 𝑞𝑑−𝑑 + 𝑥 𝑞𝑑−2𝑑 + . . . + 𝑥 𝑑 + 1), so 𝑥 𝑑 − 1 divides 𝑥 𝑛 − 1.
16
Solutions by positrón0802
13.5
Separable and Inseparable Extension
Conversely, assume that 𝑑 does not divide 𝑛. Then 𝑛 = 𝑞𝑑 + 𝑟 for some 𝑞 ∈ Z ≥0 and 0 < 𝑟 < 𝑑,
so that
𝑥 𝑛 −1 = (𝑥 𝑞𝑑+𝑟 −𝑥 𝑟 )+(𝑥 𝑟 −1) = 𝑥 𝑟 (𝑥 𝑞𝑑 −1)+(𝑥 𝑟 −1) = 𝑥 𝑟 (𝑥 𝑑 −1) (𝑥 𝑞𝑑−𝑑 +𝑥 𝑞𝑑−2𝑑 +· · ·+𝑥 𝑑 +1)+(𝑥 𝑟 −1).
Since 𝑥 𝑑 − 1 divides the first term, but does nt divide 𝑥 𝑟 − 1 (as 𝑟 < 𝑑), it follows that 𝑥 𝑑 − 1 does
not divide 𝑥 𝑛 − 1.
Exercise 13.5.4. The first assertion follows analogously as in Exercise 13.5.3. Now, F𝑝𝑑 is defined
𝑑
as the field whose 𝑝 𝑑 elements are the roots of 𝑥 𝑝 − 𝑥 over F𝑝 , and similarly F𝑝 𝑛 . Take 𝑎 = 𝑝.
𝑑
Thus, 𝑑 divides 𝑛 if and only if 𝑝 𝑑 − 1 divides 𝑝 𝑛 − 1, and that occurs exactly when 𝑥 𝑝 −1 − 1 divides
𝑑
𝑛
𝑑
𝑥 𝑝 −1 − 1 (by Exercise 13.5.3). Thus, if 𝑑 divides 𝑛, any root of 𝑥 𝑝 − 𝑥 = 𝑥 (𝑥 𝑝 −1 − 1) must be a
𝑛
𝑑
𝑛
root of 𝑥 𝑝 − 𝑥 = 𝑥 (𝑥 𝑝 −1 − 1). Hence F𝑝𝑑 ⊆ F𝑝 𝑛 . Conversely, if F𝑝𝑑 ⊆ F𝑝 𝑛 , then 𝑥 𝑝 −1 − 1 divides
𝑛
𝑥 𝑝 −1 − 1, so 𝑑 divides 𝑛.
Exercise 13.5.5. Let 𝑓 (𝑥) = 𝑥 𝑝 − 𝑥 + 𝑎. Let 𝛼 be a root of 𝑓 (𝑥). First we prove 𝑓 is separable.
Since (𝛼 + 1) 𝑝 − (𝛼 + 1) + 𝑎 = 𝛼 𝑝 + 1 − 𝛼 − 1 + 𝑎 = 0, it follows that 𝛼 + 1 is also a root of 𝑓 (𝑥).
This gives 𝑝 distinct roots of 𝑓 (𝑥) given by 𝛼 + 𝑘 with 𝑘 ∈ F𝑝 , so 𝑓 is separable.
Now we prove 𝑓 is irreducible. Let 𝑓 = 𝑓1 𝑓2 · · · 𝑓𝑛 where 𝑓𝑖 (𝑥) ∈ F𝑝 [𝑥] is irreducible for all
1 ≤ 𝑖 ≤ 𝑛. Let 1 ≤ 𝑖 < 𝑗 ≤ 𝑛, let 𝛼𝑖 be a root of 𝑓𝑖 and 𝛼 𝑗 be a root of 𝑓 𝑗 , so that 𝑓𝑖 is the minimal
polynomial of 𝛼𝑖 and 𝑓 𝑗 the minimal polynomial of 𝛼 𝑗 . We prove deg 𝑓𝑖 = deg 𝑓 𝑗 . Since 𝛼𝑖 is a
root of 𝑓𝑖 , it is a root of 𝑓 , hence there exists 𝑘 1 ∈ F𝑝 such that 𝛼𝑖 = 𝛼 + 𝑘 1 . Similarly, there exists
𝑘 2 ∈ F𝑝 such that 𝛼 𝑗 = 𝛼 + 𝑘 2 . Thus 𝛼𝑖 = 𝛼 𝑗 + 𝑘 1 − 𝑘 2, so 𝑓𝑖 (𝑥 + 𝑘 1 − 𝑘 2 ) is irreducible having
𝛼 𝑗 as a root, so it must be its minimal polynomial. It follows that 𝑓𝑖 (𝑥 + 𝑘 1 − 𝑘 2 ) = 𝑓 𝑗 (𝑥), so
deg 𝑓𝑖 = deg 𝑓 𝑗 , as claimed. Since 𝑖 and 𝑗 were arbitrary, all the 𝑓𝑖 are of the same degree, say 𝑞.
Then 𝑝 = deg 𝑓 = 𝑛𝑞, so we must have either 𝑛 = 1 or 𝑛 = 𝑝 (as 𝑝 is prime). If 𝑛 = 𝑝, then all
roots of 𝑓 are in F𝑝 , so 𝛼 ∈ F𝑝 and thus 0 = 𝛼 𝑝 − 𝛼 + 𝑎 = 𝑎, contrary to the assumption. Therefore
𝑛 = 1, so 𝑓 is irreducible.
𝑛
Exercise 13.5.6. By definition, F𝑝 𝑛 is the field whose 𝑝 𝑛 elements are the roots of 𝑥 𝑝 − 𝑥 over
𝑛
𝑛
F𝑝 . Since 𝑥 𝑝 − 1 = 𝑥 (𝑥 𝑝 −1 − 1), clearly
Ö
𝑛
𝑥 𝑝 −1 − 1 =
(𝑥 − 𝛼).
𝛼 ∈F𝑝×𝑛
Setting 𝑥 = 0, we have
−1 =
Ö
(−𝛼) = (−1) 𝑝
𝑛 −1
𝛼 ∈F𝑝×𝑛
⇒
(−1) 𝑝
𝑛 −1
Ö
𝛼
𝛼 ∈F𝑝×𝑛
(−1) = (−1) 𝑝
𝑛 −1
(−1) 𝑝
𝑛 −1
Ö
𝛼
𝛼 ∈F𝑝×𝑛
⇒
𝑛
(−1) 𝑝 =
Ö
𝛼.
𝛼 ∈F𝑝×𝑛
17
Solutions by positrón0802
13.5
Separable and Inseparable Extension
It follows that the product of the non-zero elements of F𝑝 𝑛 is +1 if 𝑝 = 2 and −1 is 𝑝 is odd.
For 𝑝 odd and 𝑛 = 1 we have
Ö
𝛼,
−1 =
𝛼 ∈F𝑝×
so taking module 𝑝 we find [1] [2] · · · [𝑝 − 1] = [−1], i.e., (𝑝 − 1)! ≡ −1 (mod 𝑝).
Exercise 13.5.7. Let 𝑎 ∈ 𝐾 such that 𝑎 ≠ 𝑏 𝑝 for every 𝑏 ∈ 𝐾 . Let 𝑓 (𝑥) = 𝑥 𝑝 − 𝑎. We prove that 𝑓
is irreducible and inseparable. If 𝛼 is a root of 𝑥 𝑝 − 𝑎, then 𝑥 𝑝 − 𝑎 = (𝑥 − 𝛼) 𝑝 , so 𝛼 is a multiple
root of 𝑓 (with multiplicity 𝑝) and hence 𝑓 is inseparable. Now let 𝑔(𝑥) be an irreducible factor
of 𝑓 (𝑥). Note that 𝛼 ∉ 𝐾, otherwise 𝑎 = 𝛼 𝑝 , contrary to the assumption. Then 𝑔(𝑥) = (𝑥 − 𝛼)𝑘 for
some 𝑘 ≤ 𝑝. Using the binomial theorem, we have
𝑔(𝑥) = (𝑥 − 𝛼)𝑘 = 𝑥 𝑘 − 𝑘𝛼𝑥 𝑘−1 + · · · + (−𝛼)𝑘 ,
so that 𝑘𝛼 ∈ 𝐾 . Since 𝛼 ∉ 𝐾, it follows that 𝑘 = 𝑝, so 𝑔 = 𝑓 . Hence 𝑓 is irreducible. We conclude
that 𝐾 (𝛼) is an inseparable finite extension of 𝐾 .
Exercise 13.5.8. Let 𝑓 (𝑥) = 𝑎𝑛 𝑥 𝑛 + . . . + 𝑎 1𝑥 + 𝑎 0 ∈ F𝑝 [𝑥]. Since F𝑝 has characteristic 𝑝, we have
(𝑎 + 𝑏) 𝑝 = 𝑎𝑝 + 𝑏 𝑝 for any 𝑎, 𝑏 ∈ F𝑝 . We can easily generalise this to a finite number of terms,
𝑝
𝑝
so that (𝑐 1 + · · · + 𝑐𝑛 ) 𝑝 = 𝑐 1 + · · · + 𝑐𝑛 for any 𝑐 1, . . . , 𝑐𝑛 ∈ F𝑝 . Furthermore, by Fermat’s Little
𝑝
Theorem, 𝑎 = 𝑎 for every 𝑎 ∈ F𝑝 . Thus, over F𝑝 , we have
𝑝
𝑝
𝑝
𝑓 (𝑥) 𝑝 = (𝑎𝑛 𝑥 𝑛 + · · · + 𝑎 1𝑥 + 𝑎 0 ) 𝑝 = 𝑎𝑛 𝑥 𝑛𝑝 + · · · + 𝑎 1 𝑥 𝑝 + 𝑎 0 = 𝑎𝑛 𝑥 𝑛𝑝 + · · · + 𝑎 1𝑥 𝑝 + 𝑎 0 = 𝑓 (𝑥 𝑝 ).
Exercise 13.5.9. From the binomial theorem we have
𝑝𝑛 Õ
𝑝𝑛 𝑖
(1 + 𝑥) 𝑝𝑛 =
𝑥,
𝑖
𝑖=0
so the coefficient of 𝑥 𝑝𝑖 in the expansion of (1 +𝑥) 𝑝𝑛 is 𝑝𝑛
𝑝𝑖 . Since F𝑝 has characteristic 𝑝, we have
𝑝 𝑖
𝑝 𝑛
(1 + 𝑥) 𝑝𝑛 = 1 + 𝑥 𝑝𝑛 = (1 + 𝑥 𝑝 )𝑛 , so over F𝑝 we have that 𝑝𝑛
𝑝𝑖 is the coefficient of (𝑥 ) in (1 + 𝑥 ) .
𝑝𝑛
𝑝
𝑛
Furthermore, (1 + 𝑥) = (1 + 𝑥 ) implies
𝑝 𝑛
(1 + 𝑥 ) =
𝑛 Õ
𝑛
𝑝 𝑖
(𝑥 ) =
𝑝𝑛 Õ
𝑝𝑛
𝑖
𝑖=0
𝑖=0
over F𝑝 , so that
𝑝𝑛 𝑝𝑖
≡
𝑛
𝑖
𝑥 𝑖 = (1 + 𝑥) 𝑝𝑛
𝑘
(mod 𝑝).
Exercise 13.5.10. This is equivalent to proving that for any prime number 𝑝, we have 𝑓 (𝑥 1, 𝑥 2, . . . , 𝑥𝑛 ) 𝑝 =
𝑝 𝑝
𝑝
𝑓 (𝑥 1 , 𝑥 2 , . . . , 𝑥𝑛 ) in F𝑝 [𝑥 1, 𝑥 2, . . . , 𝑥𝑛 ]. Let
Õ
𝛾
𝛾
𝑓 (𝑥 1, 𝑥 2, . . . , 𝑥𝑛 ) =
𝑎𝛾1,...,𝛾𝑛 𝑥 1 1 . . . 𝑥𝑛𝑛
𝛾 1 ,...,𝛾𝑛 =0
18
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
be an arbitrary element of F𝑝 [𝑥 1, 𝑥 2, . . . , 𝑥𝑛 ]. Since F𝑝 has characteristic 𝑝, we have (𝑐 1 +· · ·+𝑐𝑛 ) 𝑝 =
𝑝
𝑝
𝑐 1 + · · · + 𝑐𝑛 for any 𝑐 1, · · · , 𝑐𝑛 ∈ F𝑝 . Furthermore, by Fermat’s Little Theorem, 𝑎𝑝 = 𝑎 for every
𝑎 ∈ F𝑝 . Hence, over F𝑝 ,
Õ
Õ
Õ
𝛾
𝛾 𝑝
𝛾
𝛾
𝑝𝛾
𝑝𝛾
𝑓 (𝑥 1, 𝑥 2, . . . , 𝑥𝑛 ) 𝑝 =
𝑎𝛾1,...,𝛾𝑛 𝑥 1 1 . . . 𝑥𝑛𝑛 =
(𝑎𝛾1,...,𝛾𝑛 𝑥 1 1 . . . 𝑥𝑛𝑛 ) 𝑝 =
𝑎𝛾1,...,𝛾𝑛 (𝑥 1 1 . . . 𝑥𝑛 𝑛 )
𝑝
𝑝
𝑝
= 𝑓 (𝑥 1 , 𝑥 2 , . . . , 𝑥𝑛 ).
Exercise 13.5.11. Let 𝑓 (𝑥) ∈ 𝐹 [𝑥] have no repeated irreducible factors in 𝐹 [𝑥]. We may assume
that 𝑓 is monic. Then 𝑓 = 𝑓1 𝑓2 · · · 𝑓𝑛 for some monic irreducible polynomials 𝑓𝑖 (𝑥) ∈ 𝐹 [𝑥]. Since 𝐹
is perfect, 𝑓 is separable, hence all the 𝑓𝑖 have distinct roots. Thus, 𝑓 splits in linear factors in the
closure of 𝐹, hence splits in linear factors in the closure of 𝐾 . It follows that 𝑓 (𝑥) has no repeated
irreducible factors in 𝐾 [𝑥].
13.6
Cyclotomic Polynomials and Extensions
Exercise 13.6.1. Since (𝜁𝑚 𝜁𝑛 )𝑚𝑛 = 1, it follows that 𝜁𝑚 𝜁𝑛 is an 𝑚𝑛 th root of unity. Now assume
that for some positive integer 𝑘 we have (𝜁𝑚 𝜁𝑛 )𝑘 = 1. Then (𝜁𝑚 )𝑘𝑛 = (𝜁𝑚 )𝑘𝑛 (𝜁𝑛 )𝑘𝑛 = 𝜁 𝑘𝑛 = 1,
so that 𝑚 divides 𝑘𝑛. Since 𝑚 and 𝑛 are relatively prime, it follows that 𝑚 divides 𝑘. Similarly, 𝑛
divides 𝑘. Thus 𝑘 is a common multiple of 𝑚 and 𝑛; since they are coprime, it follows that 𝑘 is a
multiple of 𝑚𝑛. We conclude that 𝜁𝑚 𝜁𝑛 is a primitive 𝑚𝑛 th root of unity.
Exercise 13.6.2. Since (𝜁𝑛𝑑 ) (𝑛/𝑑) = 𝜁𝑛𝑛 = 1, it follows that 𝜁𝑛𝑑 is an (𝑛/𝑑) th root of unity. Now let
1 ≤ 𝑘 < (𝑛/𝑑). Then (𝜁𝑛𝑑 )𝑘 = 𝜁𝑛𝑘𝑑 . Since 1 ≤ 𝑘𝑑 < 𝑛, we have 𝜁𝑛𝑘𝑑 ≠ 1, so (𝜁𝑛𝑑 )𝑘 ≠ 1. Hence, the
order of 𝜁𝑛𝑑 is precisely (𝑛/𝑑), so it generates the cyclic group of all (𝑛/𝑑) th roots of unity, that is,
𝜁𝑛𝑑 is a primitive (𝑛/𝑑) th root of unity.
Exercise 13.6.3. Let 𝐹 be a field containing the 𝑛 th roots of unity for 𝑛 odd and let 𝜁 be a 2𝑛 th
root of unity. If 𝜁 𝑛 = 1, then 𝜁 ∈ 𝐹, so assume 𝜁 𝑛 ≠ 1. Since 𝜁 2𝑛 = 1, we have that 𝜁 𝑛 is
a root of 𝑥 2 − 1. The roots of this polynomial are 1 and −1, and 𝜁 𝑛 ≠ 1, so 𝜁 𝑛 = −1. Hence,
(−𝜁 )𝑛 = (−1)𝑛 (𝜁 )𝑛 = (−1)𝑛+1 = 1 (since 𝑛 is odd), so −𝜁 ∈ 𝐹 . Since 𝐹 is a field, we deduce that
𝜁 ∈ 𝐹.
Exercise 13.6.4. Let 𝐹 be a field with char 𝐹 = 𝑝. The roots of unity over 𝐹 are the roots of
𝑘
𝑘
𝑥 𝑛 − 1 = 𝑥 𝑝 𝑚 − 1 = (𝑥 𝑚 − 1) 𝑝 , so are the roots of 𝑥 𝑚 − 1. Now, since 𝑚 is relatively prime to
𝑝, so is 𝑥 𝑚 − 1 and its derivative 𝑚𝑥 𝑚−1, so 𝑥 𝑚 − 1 has no multiple roots. Hence, the 𝑚 different
roots of 𝑥 𝑚 − 1 are precisely the 𝑚 distinct 𝑛 th roots of unity over 𝐹 .
√
Exercise 13.6.5. We use the inequality 𝜑 (𝑛) ≥ 𝑛/2 for all 𝑛 ≥ 1, where 𝜑 denotes the Euler’s
phi-function. Let 𝐾 be an extension of Q with infinitely many roots of unity. Let 𝑁 ∈ N. Then
there exists 𝑛 ∈ N such that 𝑛 > 4𝑁 2 and there exists some 𝑛 th root of unity 𝜁 ∈ 𝐾 . Thus
√
[𝐾 : Q] ≥ [Q(𝜁 ) : Q] = 𝜑 (𝑛) ≥ 𝑛/2 > 𝑁 . Since 𝑁 was arbitrary, we deduce that [𝐾 : Q] > 𝑁
for all 𝑁 ∈ N, so [𝐾 : Q] is infinite. It follows that in any finite extension of Q there are only a
finite number of roots of unity.
19
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
Exercise 13.6.6. Since Φ2𝑛 (𝑥) and Φ𝑛 (−𝑥) are irreducible, they are the minimal polynomial of
any of its roots. Thus, it suffices to find a common root of both. Let 𝜁 2 = −1 be the primitive 2th
root of unity and let 𝜁𝑛 be a primitive 𝑛 th root of unity, so that 𝜁 2𝜁𝑛 = −𝜁𝑛 . Since 𝑛 is odd, 2 and
𝑛 are relatively prime. Thus, by Exercise 13.6.1, 𝜁 2𝜁𝑛 is a primitive 2𝑛 th root of unity, i.e, a root of
Φ2𝑛 (𝑥). Furthermore, −𝜁𝑛 is clearly a root of Φ𝑛 (−𝑥). Thus −𝜁𝑛 is a common root of both Φ2𝑛 (𝑥)
and Φ𝑛 (−𝑥), so Φ2𝑛 (𝑥) = Φ𝑛 (−𝑥).
Exercise 13.6.7. The Möbius Inversion Formula states that if 𝑓 (𝑛) is defined for all non-negative
Í
Í
integers and 𝐹 (𝑛) = 𝑑 |𝑛 𝑓 (𝑑), then 𝑓 (𝑛) = 𝑑 |𝑛 𝜇 (𝑑)𝐹 ( 𝑑𝑛 ). So let us start with the formula
Ö
𝑥𝑛 − 1 =
Φ𝑑 (𝑥).
𝑑 |𝑛
We take natural logarithm in both sides and obtain
©Ö
ª Õ
ln Φ𝑑 (𝑥).
Φ𝑑 (𝑥) ® =
ln(𝑥 𝑛 − 1) = ln ­
𝑑
|𝑛
𝑑
|𝑛
«
¬
Thus, we use the Möbius Inversion Formula for 𝑓 (𝑛) = ln Φ𝑛 (𝑥) and 𝐹 (𝑛) = ln(𝑥 𝑛 − 1) to obtain
Õ
Õ
ln Φ𝑛 (𝑥) =
𝜇 (𝑑) ln(𝑥 𝑛/𝑑 − 1) =
ln(𝑥 𝑛/𝑑 − 1) 𝜇 (𝑑) .
𝑑 |𝑛
𝑑 |𝑛
Taking exponentials we deduce
Ö
©Õ
ª Ö 𝑛/𝑑
Φ𝑛 (𝑥) = exp ­ ln(𝑥 𝑛/𝑑 − 1) 𝜇 (𝑑) ® =
(𝑥
− 1) 𝜇 (𝑑) =
(𝑥 𝑑 − 1) 𝜇 (𝑛/𝑑) .
𝑑 |𝑛
« 𝑑 |𝑛
¬ 𝑑 |𝑛
Exercise 13.6.8. (a) Since 𝑝 is prime, in F𝑝 [𝑥] we have (𝑥 − 1) 𝑝 = 𝑥 𝑝 − 1, so
Φℓ (𝑥) =
𝑥 ℓ−1
(𝑥 − 1) ℓ
=
= (𝑥 − 1) ℓ−1 .
𝑥 −1
𝑥 −1
(b) Note that 𝜁 has order ℓ, being a primitive ℓ th root of unity. Since 𝑝 𝑓 ≡ 1 mod ℓ, we have
𝑓
− 1 = 𝑞ℓ for some integer 𝑞, so that 𝜁 𝑝 −1 = 𝜁 𝑞ℓ = 1 and hence 𝜁 ∈ F𝑝 𝑓 . Now we prove that 𝑓 is
𝑛
the smallest integer with this property. Suppose 𝜁 ∈ F𝑝 𝑛 for some 𝑛. Then 𝜁 is a root of 𝑥 𝑝 −1 − 1,
so ℓ divides 𝑝 𝑛 − 1 (see Exercise 13.5.3). Since 𝑓 is the smallest power of 𝑝 such that 𝑝 𝑓 ≡ 1 mod
ℓ, it is the smallest integer such that ℓ divides 𝑝 𝑓 − 1, so 𝑛 ≥ 𝑙, as desired. This in fact proves that
F𝑝 (𝜁 ) = F𝑝 𝑓 , so the minimal polynomial of 𝜁 over F𝑝 has degree 𝑓 .
𝑝𝑓
(c) Since 𝜁 𝑎 ∈ F𝑝 (𝜁 ), clearly F𝑝 (𝜁 𝑎 ) ⊂ F𝑝 (𝜁 ). For the other direction we follow the hint. Let
𝑏 be the multiplicative inverse of 𝑎 mod ℓ, i.e 𝑎𝑏 ≡ 1 mod ℓ. Then (𝜁 𝑎 )𝑏 = 𝜁 , so 𝜁 ∈ F𝑝 (𝜁 𝑎 ) and
thus F𝑝 (𝜁 ) ⊂ F𝑝 (𝜁 𝑎 ). It follows that F𝑝 (𝜁 𝑎 ) = F𝑝 (𝜁 ).
20
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
Now, consider Φℓ (𝑥) as a polynomial over F𝑝 [𝑥]. Let 𝜁𝑖 , for 1 ≤ 𝑖 ≤ ℓ, be ℓ distinct primitive ℓ th
roots of unity. The minimal polynomial of each 𝜁𝑖 has degree 𝑓 by part (b). Hence, the irreducible
factors of Φℓ (𝑥) have degree 𝑓 . Since Φℓ have degree ℓ − 1, there must be ℓ−1
𝑓 factors, and all of
them are different since Φℓ (𝑥) is separable.
(d) If 𝑝 = 7, then Φ7 (𝑥) = (𝑥 − 1) 6 by part (a). If 𝑝 ≡ 1 mod 7, then 𝑓 = 1 in (b) and all
roots have degree 1, so Φ7 (𝑥) splits in distinct linear factors. If 𝑝 ≡ 6 mod 7, then 𝑓 = 2 is the
smallest integer such that 𝑝 𝑓 = 𝑝 2 ≡ 36 ≡ 1 mod 7, so we have 3 irreducible quadratics. If
𝑝 ≡ 2, 4 mod 7, then 𝑓 = 3 is the smallest integer such that 𝑝 3 ≡ 23, 43 ≡ 8, 64 ≡ 1 mod 7, so we
have 2 irreducible cubics. Finally, if 𝑝 ≡ 3, 5 mod 7, then 𝑓 = 6 is the smallest integer such that
𝑝 6 ≡ 36, 56 ≡ 729, 15626 ≡ 1 mod 7, hence we have an irreducible factor of degree 6.
Exercise 13.6.9. Let 𝐴 be an 𝑛 × 𝑛 matrix over C for which 𝐴𝑘 = 𝐼 for some integer 𝑘 ≥ 1. Then
the minimal polynomial of 𝐴 divides 𝑥 𝑘 − 1. Since we are working over C, there are 𝑘 distinct
roots of this polynomial, so the minimal polynomial of 𝐴 splits into linear factors. Thus 𝐴 is
diagonalisable.
1 𝛼
, where 𝛼 is an element of a field of characteristic 𝑝.
Now consider 𝐴 =
0 1
1 𝑛𝛼
𝑛
for every positive integer
Computing powers of 𝐴, inductively it follows that 𝐴 =
0 1
𝑛. Since 𝑝𝛼 = 0, we have 𝐴𝑝 = 𝐼 . Now, if 𝐴 is diagonalizable, then there exists some non-singular
matrix 𝑃 such that 𝐴 = 𝑃𝐷𝑃 −1, where 𝐷 is a diagonal matrix whose diagonal entries are the
eigenvalues of 𝐴. Since 𝐴 has 1 as its only, 𝐷 must be the identity and therefore also 𝐴. That is, if
𝐴 is diagonalisable, we must have 𝛼 = 0.
Exercise 13.6.10. For 𝑎, 𝑏 ∈ F𝑝 𝑛 we have 𝜑 (𝑎 + 𝑏) = (𝑎 + 𝑏) 𝑝 = 𝑎𝑝 + 𝑏 𝑝 = 𝜑 (𝑎) + 𝜑 (𝑏), and
𝜑 (𝑎𝑏) = (𝑎𝑏) 𝑝 = 𝑎𝑝 𝑏 𝑝 = 𝜑 (𝑎)𝜑 (𝑏), so 𝜑 is a homomorphism. Moreover, if 𝜑 (𝑎) = 0, then 𝑎𝑝 = 0
implies 𝑎 = 0, so 𝜑 is injective. Since F𝑝 𝑛 is finite, 𝜑 is also surjective and hence an isomorphism.
𝑛
𝑛
Furthermore, since every element of F𝑝 𝑛 is a root of 𝑥 𝑝 − 𝑥, we have 𝜑 𝑛 (𝑎) = 𝑎𝑝 = 𝑎 for all
𝑎 ∈ F𝑝 𝑛 , so 𝜑 𝑛 is the identity map. Now, let 𝑚 be a positive integer such that 𝜑 𝑚 is the identity
𝑚
𝑚
map. Then 𝑎𝑝 = 𝑎 for all 𝑎 ∈ F𝑝 𝑛 , so every element of F𝑝 𝑛 must be a root of 𝑥 𝑝 − 𝑥 . Hence,
F𝑝 𝑛 ⊂ F𝑝𝑚 and thus 𝑛 divides 𝑚 (Exercise 13.5.4), so 𝑛 ≤ 𝑚.
Exercise 13.6.11. Note that the minimal polynomial of 𝜑 is 𝑥 𝑛 − 1, for if 𝜑 satisfies some polyno𝑛−1
mial 𝑥 𝑛−1 + · · · + 𝑎 1𝑥 + 𝑎 0 of degree 𝑛 − 1 (or less) with coefficients in F𝑝 , then 𝑥 𝑝 + · · · + 𝑎 1𝑥 𝑝 + 𝑎 0
for all 𝑥 ∈ F𝑝 𝑛 , which is impossible. Since F𝑝 𝑛 has degree 𝑛 as a vector space over F𝑝 , it follows
that 𝑥 𝑛 − 1 is also the characteristic polynomial of 𝜑, hence is the only invariant factor. Therefore,
the rational canonical form of 𝜑 over F𝑝 is the companion matrix of 𝑥 𝑛 − 1, which is
0
©
­1
­
­0
­.
­.
­.
«0
0 ···
0 ···
1 ···
.. . .
.
.
0 ···
21
0 1
ª
0 0®
®
0 0® .
.. .. ®®
. .®
1 0¬
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
Exercise 13.6.12. We will work over the algebraic closure of F𝑝 𝑛 , to ensure the field contains all
eigenvalues. In Exercise 13.6.11 we proved that the minimal and characteristic polynomial of 𝜑 is
𝑥 𝑛 − 1. Moreover, the eigenvalues of 𝜑 are the 𝑛 th roots of unity. We use Exercise 13.6.4 and write
𝑘
𝑛 = 𝑝 𝑘 𝑚 for some prime 𝑝 and some 𝑚 relatively prime to 𝑝, so that 𝑥 𝑛 − 1 = (𝑥 𝑚 − 1) 𝑝 and
we get exactly 𝑚 distinct 𝑛 th roots of unity, each one of multiplicity 𝑝 𝑘 . Since all the eigenvalues
are zeros of both the minimal and characteristic polynomial of multiplicity 𝑝 𝑘 , we get 𝑚 Jordan
blocks of size 𝑝 𝑘 . Now, fix a primitive 𝑚 th root of unity, say 𝜁 . Then each Jordan block has the
form
𝜁𝑖 1 0 · · · 0 0
©
ª
­ 0 𝜁𝑖 1 · · · 0 0 ®
­
®
­ 0 0 𝜁𝑖 · · · 0 0 ®
­
𝐽𝑖 = ­ .. .. .. . .
.. .. ®®
.
. .®
­. . .
­
®
­ 0 0 0 · · · 𝜁𝑖 1 ®
𝑖
«0 0 0 ··· 0 𝜁 ¬
for some 0 ≤ 𝑖 ≤ 𝑚 − 1. Finally, we already know the Jordan canonical form is given by
𝐽0
©
­0
­
­0
­.
­.
­.
«0
0 ···
0
ª
𝐽1 · · ·
0 ®
®
0 ···
0 ®.
.. . .
.. ®®
.
.
. ®
0 · · · 𝐽𝑚−1 ¬
Exercise 13.6.13. (a) 𝑍 is a division subring of 𝐷, and it is commutative by definition of the centre,
so 𝑍 is a field. Since it is finite, its prime subfield is F𝑝 for some prime 𝑝, so 𝑍 is isomorphic to
F𝑝 𝑛 for some integer 𝑛. Let 𝑞 = 𝑝 𝑛 . Since 𝐷 is a vector space over 𝑍, we have |𝐷 | = 𝑞𝑛 for some
integer 𝑛.
(b) Let 𝑥 ∈ 𝐷 × and let 𝐶𝐷 (𝑥) be the set of the elements in 𝐷 that commutes with 𝑥 . Clearly
𝑍 ⊂ 𝐶𝐷 (𝑥). We prove that every element 𝑎 ∈ 𝐶𝐷 (𝑥) has an inverse in 𝐶𝐷 (𝑥). Since 𝑎 ∈ 𝐶𝐷 (𝑥),
we have 𝑎𝑥 = 𝑥𝑎, and since 𝐷 is a division ring, we also have 𝑎 −1 ∈ 𝐷. Moreover, 𝑎 −1𝑎𝑥 = 𝑎 −1𝑥𝑎
and thus 𝑥 = 𝑎 −1𝑥𝑎, so 𝑥𝑎 −1 = 𝑎 −1𝑥 and 𝑎 −1 ∈ 𝐶𝐷 (𝑥). Therefore 𝐶𝐷 (𝑥) is a division ring. As
𝑍 ⊂ 𝐶𝐷 (𝑥), it follows that 𝐶𝐷 (𝑥) is a 𝑍 -vector space, so |𝐶𝐷 (𝑥)| = 𝑞𝑚 for some integer 𝑚. If
𝑥 ∉ 𝑍, then 𝐶𝐷 (𝑥) is a proper subset of 𝐷 and hence 𝑚 < 𝑛.
(c) The class equation for the group 𝐷 × is
|𝐷 × | = |𝑍 (𝐷 × )| +
𝑟
Õ
|𝐷 × : 𝐶𝐷 × (𝑥𝑖 )|,
𝑖=1
where the 𝑥𝑖 are representatives of the distinct conjugacy classes in 𝐷 × not contained in the
centre of 𝐷 × . By (a) we have |𝐷 × | = 𝑞𝑛 − 1, |𝑍 (𝐷 × )| = 𝑞 − 1 and |𝐶𝐷 × (𝑥𝑖 )| = 𝑞𝑚𝑖 − 1. Then
22
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
|𝐷 × : 𝐶𝐷 × (𝑥𝑖 )| =
𝑞𝑛 − 1
𝑞𝑛 − 1
. Plugging these values in the class equation we obtain
= 𝑚
|𝐶𝐷 × (𝑥𝑖 )| 𝑞 𝑖 − 1
𝑞𝑛 − 1 = (𝑞 − 1) +
𝑟
𝑟
Õ
Õ
𝑞𝑛 − 1
𝑞𝑛 − 1
= (𝑞 − 1) +
.
|𝐶𝐷 × (𝑥𝑖 )|
𝑞𝑚𝑖 − 1
𝑖=1
𝑖=1
𝑞𝑛 − 1
is also an integer. Hence
𝑞𝑚𝑖 − 1
𝑞𝑚𝑖 − 1 divides 𝑞𝑛 − 1, so (Exercise 13.5.4) 𝑚𝑖 divides 𝑛. Since 𝑚𝑖 < 𝑛 (no 𝑥𝑖 is in 𝑍 ), no 𝑚𝑖th root of
unity is a 𝑛 th root of unity. Therefore, as Φ𝑛 (𝑥) divides 𝑥 𝑛 − 1, it must divide (𝑥 𝑛 − 1)/(𝑥 𝑚𝑖 − 1)
for 𝑖 = 1, 2, . . . , 𝑟 .. Letting 𝑥 = 𝑞 we deduce that Φ𝑛 (𝑞) divides (𝑞𝑛 − 1)/(𝑞𝑚𝑖 − 1) for 𝑖 = 1, 2, . . . , 𝑟 .
(d) Since |𝐷 × : 𝐶𝐷 × (𝑥𝑖 )| is an integer, |𝐷 × : 𝐶𝐷 × (𝑥𝑖 )| =
(e) From (d), Φ𝑛 (𝑞) divides (𝑞𝑛 − 1)/(𝑞𝑚𝑖 − 1) for 𝑖 = 1, 2, . . . , 𝑟, so the class equation in (c)
implies that Φ𝑛 (𝑞) divides 𝑞 − 1. Now, let 𝜁 ≠ 1 be a 𝑛 th root of unity. In the complex plane 𝑞 is
Î
closer to 1 than 𝜁 is, so |𝑞 − 𝜁 | > |𝑞 − 1| = 𝑞 − 1. Since Φ𝑛 (𝑞) = 𝜁 primitive (𝑞 − 𝜁 ) divides 𝑞 − 1,
this is impossible unless 𝑛 = 1. Hence, 𝐷 = 𝑍 and 𝐷 is a field.
Exercise 13.6.14. We follow the hint. Let 𝑃 (𝑥) = 𝑥 𝑛 + · · · + 𝑎 1𝑥 + 𝑎 0 be a monic polynomial
over Z of degree 𝑛 ≥ 1. For the sake of a contradiction, suppose there are only finitely many
primes dividing the values 𝑃 (𝑛), 𝑛 = 1, 2, . . . , say 𝑝 1, 𝑝 2, . . . , 𝑝𝑘 . Let 𝑁 be an integer such that
𝑃 (𝑁 ) = 𝑎 ≠ 0. Let 𝑄 (𝑥) = 𝑎 −1 𝑃 (𝑁 + 𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑥). Then, using the binomial theorem, we have
𝑄 (𝑥) = 𝑎 −1 𝑃 (𝑁 + 𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑥)
= 𝑎 −1 ((𝑁 + 𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑥)𝑛 + · · · + 𝑎 1 (𝑁 + 𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑥) + 𝑎 0 )
= 𝑎 −1 (𝑁 𝑛 + 𝑎𝑛−1 𝑁 𝑛−1 + · · · + 𝑎 1 𝑁 + 𝑎 0 + 𝑅(𝑥))
= 𝑎 −1 (𝑃 (𝑁 ) + 𝑅(𝑥))
= 1 + 𝑎 −1𝑅(𝑥)
for some polynomial 𝑅(𝑥) ∈ Z[𝑥] divisible by 𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 . Thus 𝑄 (𝑥) ∈ Z[𝑥]. Moreover, for all
𝑛 ∈ Z+ we have 𝑃 (𝑁 +𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑛) ≡ 𝑎 (mod 𝑝 1, 𝑝 2, . . . , 𝑝𝑘 ), so 𝑄 (𝑛) = 𝑎 −1 𝑃 (𝑁 +𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑛) ≡
𝑎 −1𝑎 = 1 (mod 𝑝 1, 𝑝 2, . . . , 𝑝𝑘 ). Now let 𝑚 be a positive integer such that |𝑄 (𝑚)| > 1, so that
𝑄 (𝑚) ≡ 1 (mod 𝑝𝑖 ) for all 𝑖. Therefore, none of the 𝑝𝑖 ’s divide 𝑄 (𝑚). Since |𝑄 (𝑚)| > 1, there
exists a prime 𝑞 such that 𝑞 ≠ 𝑝𝑖 for all 𝑖 and such that 𝑞 divides 𝑄 (𝑚). Then 𝑞 divides 𝑎𝑄 (𝑚) =
𝑃 (𝑁 +𝑎𝑝 1𝑝 2 . . . 𝑝𝑘 𝑚), contradicting the fact that only the primes 𝑝 1, 𝑝 2, . . . , 𝑝𝑘 divide the numbers
𝑃 (1), 𝑃 (2), . . . .
Exercise 13.6.15. We follow the hint. Since Φ𝑚 (𝑎) ≡ 0 (mod 𝑝), we have 𝑎𝑚 ≡ 1 (mod 𝑝). Then
there exists 𝑏 such that 𝑏𝑎 ≡ 1 mod 𝑝 (indeed, 𝑏 = 𝑎𝑚−1 works), so 𝑎 is relatively prime to 𝑝. We
prove that the order of 𝑎 is precisely 𝑚. For the sake of a contradiction, suppose 𝑎𝑑 ≡ 1 (mod 𝑝)
for some 𝑑 dividing 𝑚, so that Φ𝑑 (𝑎) ≡ 0 (mod 𝑝) for some 𝑑 < 𝑚. Then 𝑎 is a multiple root of
𝑥 𝑚 − 1, so it is also a root of its derivative 𝑚𝑎𝑚−1 . But then 𝑚𝑎𝑚−1 ≡ 0 mod 𝑝, impossible since 𝑝
does not divide 𝑚 nor 𝑎. Therefore, the order of 𝑎 in (Z/𝑝Z) × is precisely 𝑚.
23
Solutions by positrón0802
13.6
Cyclotomic Polynomials and Extensions
Exercise 13.6.16. Let 𝑝 be an odd prime dividing Φ𝑚 (𝑎). If 𝑝 does not divide 𝑚, then, by (c), 𝑎 is
relatively prime to 𝑝 and the order of 𝑎 in F𝑝× is 𝑚. Since |F𝑝× | = 𝑝 − 1, this implies that 𝑚 divides
𝑝 − 1, that is, 𝑝 ≡ 1 (mod 𝑚).
Exercise 13.6.17. By Exercise 13.6.14, there are infinitely many primes dividing Φ𝑚 (1), Φ𝑚 (2), Φ𝑚 (3), . . . .
Since only finitely many of them can divide 𝑚, it follows from by Exercise 13.6.16 that there must
exist infinitely many primes 𝑝 with 𝑝 ≡ 1 (mod 𝑚).
Please send comments, suggestions and corrections by e-mail, or at website.
https://positron0802.wordpress.com
[email protected]
24
Solutions by positrón0802