Skip to content

Latest commit

 

History

History
980 lines (732 loc) · 30.4 KB

File metadata and controls

980 lines (732 loc) · 30.4 KB
title author header-includes abstract
Abstract Algebra by Pinter, Chapter 31
Amir Taaki
- \usepackage{mathrsfs} - \usepackage{mathtools} - \usepackage{extpfeil} - \DeclareMathOperator\ker{ker} - \DeclareMathOperator\ord{ord} - \DeclareMathOperator\gcd{gcd} - \DeclareMathOperator\lcm{lcm} - \DeclareMathOperator\char{char} - \DeclareMathOperator\max{max} - \DeclareMathOperator\ran{ran} - \DeclareMathOperator\deg{deg} - \DeclareMathOperator\dim{dim} - \newcommand{\mod}[1]{\ (\mathrm{mod}\ #1)} - \newcommand{\repr}[1]{\overline{#1}} - \newcommand{\leg}[2]{\left(\frac{#1}{#2}\right)} - \let\vec\mathbf
Chapter 31 on Galois Theory Preamble

Note that root field is called splitting field in other texts.

A. Examples of Root Fields over $\mathbb{Q}$

Q1

sage: solve(x^2 - 2*x - 2, x)
[x == -sqrt(3) + 1, x == sqrt(3) + 1]
sage: solve(x^2 + 1, x)
[x == -I, x == I]

Q2

Possible roots for $x^2 - 3$ are $\pm 1, \pm 3$.

sage: p = lambda x: x^2 - 3
sage: p(-1), p(1), p(-3), p(3)
(-2, -2, 6, 6)

So it is irreducible.

For $x^2 - 2x - 2$ we can use Eisenstein's irreducibility criteria.

sage: solve(x^2 - 3, x)
[x == -sqrt(3), x == sqrt(3)]
sage: solve(x^2 - 2*x - 2, x)
[x == -sqrt(3) + 1, x == sqrt(3) + 1]

Root field is $\mathbb{Q}(\sqrt{3})$.

Q3

sage: solve(x^4 - 2, x)
[x == I*2^(1/4), x == -2^(1/4), x == -I*2^(1/4), x == 2^(1/4)]

Therefore the root field is $\mathbb{Q}(2^{\frac{1}{4}}, i)$

Since $\mathbb{Q}(2^{\frac{1}{4}}) \subseteq \mathbb{R}$, then the root field of $x^4 - 2$ over $\mathbb{R}$ is $\mathbb{R}(i)$.

Q4

sage: solve(x^4 - 2*x^2 + 9, x)
[x == -sqrt(2*I*sqrt(2) + 1), x == sqrt(2*I*sqrt(2) + 1), x == -sqrt(-2*I*sqrt(2) + 1), x == sqrt(-2*I*sqrt(2) + 1)]

Root field is $\mathbb{Q}(i, \sqrt{2})$.

sage: solve(x^2 - 2*sqrt(2)*x + 3, x)
[x == sqrt(2) - I, x == sqrt(2) + I]

Root field is $\mathbb{Q}(i, \sqrt{2})$.

Q5

sage: c
sqrt(3) + I
sage: ((c^2 - 2)^2).expand()
-12
sage: ((x^2 - 2)^2
....: ).expand()
x^4 - 4*x^2 + 4
sage: x^4 - 4*x^2 + 4 + 12
x^4 - 4*x^2 + 16
sage: solve(x^4 - 4*x^2 + 4 + 12, x)
[x == -sqrt(2*I*sqrt(3) + 2), x == sqrt(2*I*sqrt(3) + 2), x == -sqrt(-2*I*sqrt(3) + 2), x == sqrt(-2*I*sqrt(3) + 2)]

So there are roots $-2\sqrt{3}i - 2$ and $2\sqrt{3}i - 2$.

And $x^2 - 3$ for $b(x)$.

sage: c = sqrt(2) + sqrt(3)
sage: (c^2).expand()
2*sqrt(3)*sqrt(2) + 5
sage: (((c^2).expand() - 5)^2).expand()
24
sage: ((x^2 - 5)^2).expand()
x^4 - 10*x^2 + 25
sage: solve(x^4 - 10*x^2 + 1, x)
[x == -sqrt(2*sqrt(6) + 5), x == sqrt(2*sqrt(6) + 5), x == -sqrt(-2*sqrt(6) + 5), x == sqrt(-2*sqrt(6) + 5)]

Remembering that $\sqrt{6} = \sqrt{2} \sqrt{3}$.

Q6

All of them are valid root fields except the cube root one.

B. Examples of Root Fields over $\mathbb{Z}_p$

Q1

sage: R.<x> = IntegerModRing(3)[]
sage: f = x^3 + 2*x + 1
sage: S.<u> = R.extension(f)
sage: (x - u)*(x - (u + 1))*(x - (u + 2))
x^3 + 2*x + 1

List the elements:

sage: S = R.quotient(x^3 + 2*x + 1, 'u')
sage: len(S)
27
sage: list(S)
[0,
 1,
 2,
 u,
 u + 1,
 u + 2,
 2*u,
 2*u + 1,
 2*u + 2,
 u^2,
 u^2 + 1,
 u^2 + 2,
 u^2 + u,
 u^2 + u + 1,
 u^2 + u + 2,
 u^2 + 2*u,
 u^2 + 2*u + 1,
 u^2 + 2*u + 2,
 2*u^2,
 2*u^2 + 1,
 2*u^2 + 2,
 2*u^2 + u,
 2*u^2 + u + 1,
 2*u^2 + u + 2,
 2*u^2 + 2*u,
 2*u^2 + 2*u + 1,
 2*u^2 + 2*u + 2]

Roots of $a(x)$ are $u, u + 1, u + 2$.

Root field is therefore $\mathbb{Z}_3(u)$.

Q2

sage: S = R.quotient(x^2 + x + 2, 'u')
sage: list(S)
[0, 1, 2, u, u + 1, u + 2, 2*u, 2*u + 1, 2*u + 2]

Utilize the fact that $u^2 = - u - 2 = 2u + 1$, we can make the addition and multiplication tables.

sage: from sage.matrix.operation_table import OperationTable
sage: OperationTable(S, operation=operator.add)
+  a b c d e f g h i
 +------------------
a| a b c d e f g h i
b| b c a e f d h i g
c| c a b f d e i g h
d| d e f g h i a b c
e| e f d h i g b c a
f| f d e i g h c a b
g| g h i a b c d e f
h| h i g b c a e f d
i| i g h c a b f d e

sage: OperationTable(S, operation=operator.mul)
*  a b c d e f g h i
 +------------------
a| a a a a a a a a a
b| a b c d e f g h i
c| a c b g i h d f e
d| a d g h b e f i c
e| a e i b f g c d h
f| a f h e g c i b d
g| a g d f c i h e b
h| a h f i d b e c g
i| a i e c h d b g f

Roots are: $u, 2u + 2$ (figured by substitution)

Root field: $\mathbb{Z}_3(u)$

Q3

Root field will be all combos of polynomials over $\mathbb{Z}_2$ of degree 2 polynomials.

sage: R.<x> = IntegerModRing(2)[]
sage: S = R.quotient(x^3 + x^2 + 1, 'u')
sage: list(S)
[0, 1, u, u + 1, u^2, u^2 + 1, u^2 + u, u^2 + u + 1]

In $\mathbb{Z}_2$, $-1 = 1$ so $u^3 = u^2 + 1$.

sage: OperationTable(S, operation=operator.add)
+  a b c d e f g h
 +----------------
a| a b c d e f g h
b| b a d c f e h g
c| c d a b g h e f
d| d c b a h g f e
e| e f g h a b c d
f| f e h g b a d c
g| g h e f c d a b
h| h g f e d c b a

sage: OperationTable(S, operation=operator.mul)
*  a b c d e f g h
 +----------------
a| a a a a a a a a
b| a b c d e f g h
c| a c e g f h b d
d| a d g f b c h e
e| a e f b h d c g
f| a f h c d g e b
g| a g b h c e d f
h| a h d e g b f c
sage: R.<x> = IntegerModRing(2)[]
sage: f = x^3 + x^2 + 1
sage: S.<u> = R.extension(f)
...
sage: f(u)
0
sage: f(u^2)
0
sage: f(u^2 + u + 1)
0
sage: (x - u)*(x - u^2)*(x - (u^2 + u + 1))
x^3 + x^2 + 1

Root field: $\mathbb{Z}_3(u)$

Q4

sage: f = x^3 + x + 1
sage: S.<u> = R.extension(f)
sage: f(u)
0
sage: f(u + 1)
u^2 + u
sage: f(u^2)
0
sage: f(u^2 + 1)
u
sage: f(u^2 + u)
0
sage: f(u^2 + u + 1)
u^2

Roots are $u, u^2, u^2 + u$.

Q5

sage: R.<x> = IntegerModRing(3)[]
sage: f = x^3 + x^2 + x + 2
sage: S.<u> = R.extension(f)
sage: f(u)
0
sage: f(u + 1)
2*u
sage: f(u + 2)
u + 2
sage: f(u^2)
u^2
sage: f(u^2 + 1)
0
sage: f(u^2 + 2)
2*u^2 + 2
sage: f(u^2 + u)
2*u^2 + u
sage: f(u^2 + u + 1)
u^2
sage: f(u^2 + 2*u)
2*u^2 + 2*u
sage: f(u^2 + 2*u + 1)
u^2
sage: f(u^2 + u + 2)
2*u + 2
sage: f(u^2 + 2*u + 2)
u + 2
sage: f(2*u)
2*u^2 + 1
sage: f(2*u + 1)
2*u^2 + u + 1
sage: f(2*u + 2)
2*u^2 + 2*u
sage: f(2*u^2)
2*u^2 + u + 2
sage: f(2*u^2 + 1)
u + 2
sage: f(2*u^2 + 2)
u^2 + u + 1
sage: f(2*u^2 + u)
u^2 + 1
sage: f(2*u^2 + 2*u)
2*u^2 + 2*u
sage: f(2*u^2 + u + 1)
2*u^2 + 2*u + 1
sage: f(2*u^2 + u + 2)
u
sage: f(2*u^2 + 2*u + 1)
0

I think the answer in the book is wrong since

sage: R.<x> = IntegerModRing(3)[]
sage: f = x^3 + x^2 + x + 2
sage: S.<u> = R.extension(f)
sage: u^3 + u^2 + u + 2
0
sage: f(u^2 + 1)
0
sage: f(2*u^2 + 2*u + 1)
0
sage: (x - u)*(x - (u^2 + 1))*(x - (2*u^2 + 2*u + 1))
x^3 + x^2 + x + 2

We can even substitute these roots into $q(x)$ which is claimed to be irreducible

sage: a = lambda x: (x^2 + (u + 1)*x + (u^2 + u + 1))
sage: a(u^2 + 1)
0
sage: a(2*u^2 + 2*u + 1)
0

Roots are: $u, u^2 + 1, 2u^2 + 2u + 1$.

Basis is ${ u^2, u, 1 }$.

C. Short Questions Relating to Root Field

Q1

The basis for a degree 2 extension $\mathbb{F}(c)$ is ${c, 1}$ with $c^2 \in \mathbb{F}$. This includes $-c$ and so $\mathbb{F}(c)$ is the root field for $(x + c)(x - c) = x^2 - c^2$.

The question I assume is asking to prove every degree 2 field extension is normal. So since $c \in F(c)$, then we can divide the polynomial by $(x - c)$, leaving a degree 1 polynomial $(x - \alpha) \in F(c)[x]$ or that $\alpha \in F(c)$.

Q2

$a(x)$ as a polynomial of degree $n$ over a finite field will have $n$ distinct roots. All roots $c_1, \dots, c_n \in K$ and form a root field over $F$. Assume some roots are in $I$, then the roots of $a(x)$ are still the same, and so $K$ forms a root field over $I$ as well.

Q3

Any polynomial can be factored into a linear combination of complex roots. This means $\mathbb{C}$ is the root field of every polynomial. Since polynomials can also have root fields contained in $\mathbb{R}$, and $\mathbb{R} \subset \mathbb{C}$, this means the root field is either $\mathbb{R}$ or $\mathbb{C}$ of every polynomial.

Q4

This question is impossible. See the answer here.

Q5

As per the back of the book, we just need to show that $F(d_1, d_2) = F(c)$ and that it's the splitting field.

\begin{align*} d_1^2 &= \frac{a^2}{4} - b \ d_2^2 &= \frac{a}{2} + d_1 \ &= \frac{a}{2} + \sqrt{\frac{a^2}{4} - b} \end{align*}

Roots of $p(x)$ are $\pm \sqrt{\frac{a}{2} \pm d_1}$. We know that $\frac{a}{2} \pm d_1 \in F(d_1)$.

$[F(d_1): F] = 2$. Every extension of degree 2 is a root field from 31C1 above.

Q6

$$\sigma_c(a(x)) = a(c)$$ $$\ker \sigma_c = { a(x): a(c) = 0 }$$ Every ideal of a field is principal, so $J = \langle p(x) \rangle$. For some $a(x) \in J$, $a(x)$ is a multiple of $p(x)$. $$\sigma_c : F[x] \rightarrow F$$ so $F(c)$ contains $p(x)$ since the ideal $J$ contains all polynomials with $c$ as their root, and all polys are a multiple of $p(x)$.

Q7

From the previous argument $p(x)$ is the lowest degree polynomial in $J$ and hence irreducible. Otherwise $p(c) = f(c)g(c) = 0 \implies f(c) = 0 \textrm{ or } g(c) = 0$ which would be a contradiction since it implies there would be a lower degree polynomial in $J$. Every root field is a simple extension $F(c)$, we can convert a polynomial with roots $a, b$ to one with $c$ from theorem 2 in this chapter.

Q8

Give an isomorphism $h(x)$ which fixes F, and a polynomial $$a(x) = a_0 + a_1 x + \cdots + a_n x^n$$ We can see that where $c$ is an algebraic root in $K$, then \begin{align*} a(h(c)) &= a_0 + a_1 h(c) + \cdots + a_n h(c^n) \ &= h(a(c)) = h(0) = 0 \end{align*} So also $h(c)$ is a root of $a(x)$ and the isomorphism simply permutes roots since it sends unique elements of $K$ to $K'$. We also observe that the mapping is one to one, fixing $F$, permuting roots, and that $h: K \rightarrow K$. $$h({c_1, \dots, c_n}) = { c_1, \dots, c_n }$$

Since the polynomial $a(x)$ is irreducible over $F$, and $K$ being a finite extension can be reduced to a simple extension $F(c) = K$, where $c$ is a root of $a(x)$, which means that $$F(c) \cong F / \langle a(x) \rangle$$ which forms a vector space of $\deg a(x) = n \implies [K:F] = n$.

Given another polynomial $b(x)$ where $b(c) = 0 : c \in K \implies b(h(c)) = 0 \implies$ K is the splitting field for $b(x)$. This means $b(x)$ is split completely by $K$ into linear factors.

Note, this means every polynomial of degree $n$ which has a root in $K$ makes $K$ its splitting field. The converse does not hold. An irreducible polynomial of degree $n$ does not necessarily have a splitting field of degree $n$. See this answer.

Q9

First we prove this for an irreducible polynomial $p(x)$ with $n = \deg p(x)$ roots of the form $c_1, \dots, c_n$. Inductively adjoining $c_1$ to $F$ forms a field $F(c_1)$ such that $[F(c_1):F] = n$ with a basis ${ 1, c_1, \dots, c_1^{n - 1} }$. Dividing $p(x)$ by $(x - c_1)$ leaves a polynomial $q(x)$ with degree $n - 1$ and adjoining the second root to $F(c_1)$ forms a field extension with degree $[F(c_1, c_2):F(c_1)] = n - 1$. Proceeding in this way, we obtain a maximum order of $n!$.

The polynomial $a(x)$ is reducible to $n$ irreducible factors $a(x) = p_1(x) p_2(x) \cdots p_n(x)$ where $\deg p_i(x) = k_i$. Each of these fields forms a simple extension $F(c_1, \cdots, c_n)$, where $[F(c_1, \cdots, c_n):F] = [F(c_1, \cdots, c_n):F(c_1, \cdots, c_{n - 1})]\cdots[F(c_1):F] = k_n \cdot k_{n - 1} \cdots k_1$.

Since $\deg a(x) = N$ and $k_1 + \cdots k_n$, so $[K:F] \mid k_1! \cdots k_n!$.

From the binomial formula, given $n = k + l$, then there's an integer $z$ such that $$z = \frac{n!}{k!l!} \implies n! = z k! l!$$
So therefore given $n = k_1 + \cdots + k_n$, we can see that $x \mid k_1! \cdots k_n! \implies x \mid n!$

D. Reducing Iterated Extensions to Simple Extensions

Q1

a

$$\mathbb{Q}(\sqrt{2}, i\sqrt{3}) = \mathbb{Q}(\sqrt{2} + i\sqrt{3})$$ $$c = \sqrt{2} + i\sqrt{3}$$ $$[\mathbb{Q}(c):\mathbb{Q}] = 4$$ $$\implies \sqrt{2} = a_0 + a_1 c + a_2 c^2 + a_3 c^3$$ $$c^2 = 2i \sqrt{6} - 1 \implies i\sqrt{6} \in \mathbb{Q}(c)$$ $$c^3 = 4i\sqrt{3} - \sqrt{2} - 6\sqrt{2} - i\sqrt{3} = 3i\sqrt{3} - 7\sqrt{2}$$ $$c^3 + 7c = 10i\sqrt{3} \implies i\sqrt{3} \in \mathbb{Q}(c)$$ Same can be shown for $\sqrt{2}$.

b

$$(\sqrt[6]{2})^3 = \sqrt{2}, (\sqrt[6]{2})^2 = \sqrt[3]{2}$$ $$\implies \mathbb{Q}(\sqrt{2}, \sqrt[3]{2}) = \mathbb{Q}(\sqrt[6]{2})$$

Q2

The roots of $x^2 - 2x - 1$ by completing the square are $\pm \sqrt{2} + 1$.

For a cubic with real coefficients in a field, it either has all real roots or 2 complex roots. By differentiating and sketching the curve where it's increasing or decreasing, we see that this cubic has two complex roots.

According to the Complex Conjugate Theorem, if $x = a + ib$ is a solution to a polynomial with real coefficients, then so is $x = a - ib$.

Thus we conclude that $\mathbb{Q}(a, b) = \mathbb{Q}(a + b)$ where $a$ is a complex root of $x^3 - x - 1$.

Q3

These factors are all linearly independent so $c = \sqrt{2} + \sqrt{3} + \sqrt{-5}$.

Q4

From C6, because $\sqrt{2}$ and $\sqrt{3}$ are independent, the basis is ${ 1, \sqrt{2}, \sqrt{3}, \sqrt{6} }$ and so the minimum polynomial has degree 4.

sage: # We let x = sqrt(2) + sqrt(3), so now we square it on both sides
sage: (sqrt(2) + sqrt(3))^2
(sqrt(3) + sqrt(2))^2
sage: ((sqrt(2) + sqrt(3))^2).expand()
2*sqrt(3)*sqrt(2) + 5
sage: # so now (x^2 - 5) = 2*sqrt(3)*sqrt(2)
sage: # lets square again both sides
sage: (2*sqrt(3)*sqrt(2))**2
24
sage: ((x^2 - 5)^2).expand()
x^4 - 10*x^2 + 25
sage: ((x^2 - 5)^2).expand() - 24
x^4 - 10*x^2 + 1

Q5

The minimum polynomial will now have degree 8.

sage: x - sqrt(2) == sqrt(3) + I*sqrt(5)
x - sqrt(2) == I*sqrt(5) + sqrt(3)
sage: (x - sqrt(2) == sqrt(3) + I*sqrt(5))^2
(x - sqrt(2))^2 == (I*sqrt(5) + sqrt(3))^2
sage: ((x - sqrt(2) == sqrt(3) + I*sqrt(5))^2).expand()
x^2 - 2*sqrt(2)*x + 2 == 2*I*sqrt(5)*sqrt(3) - 2
sage: ((x - sqrt(2) == sqrt(3) + I*sqrt(5))^2).expand() + 2*sqrt(2)*x
x^2 + 2 == 2*sqrt(2)*x + 2*I*sqrt(5)*sqrt(3) - 2
sage: p = ((x - sqrt(2) == sqrt(3) + I*sqrt(5))^2).expand() + 2*sqrt(2)*x
sage: p
x^2 + 2 == 2*sqrt(2)*x + 2*I*sqrt(5)*sqrt(3) - 2
sage: p += 2
sage: p
x^2 + 4 == 2*sqrt(2)*x + 2*I*sqrt(5)*sqrt(3)
sage: (p^2).expand()
x^4 + 8*x^2 + 16 == 8*I*sqrt(5)*sqrt(3)*sqrt(2)*x + 8*x^2 - 60
sage: (p^2).expand() + 60 - 8*x^2
x^4 + 76 == 8*I*sqrt(5)*sqrt(3)*sqrt(2)*x
sage: p = (p^2).expand() + 60 - 8*x^2
sage: (p^2).expand()
x^8 + 152*x^4 + 5776 == -1920*x^2
sage: (p^2).expand() + 1920*x^2
x^8 + 152*x^4 + 1920*x^2 + 5776 == 0
sage: p = x^8 + 152*x^4 + 1920*x^2 + 5776
sage: p(x = sqrt(2) + sqrt(3) + I*sqrt(5)).expand()
0

E. Roots of Unity and Radical Extensions

Q1

The roots of $x^n - 1$ are $1, \omega, \dots, \omega^{n - 1}$ which is the basis for $\mathbb{Q}(\omega)$ generated by $\omega$ since it is primitive.

Q2

Define a substitution function $\sigma_\omega$ $$\sigma_\omega(a(x)) = a(\omega)$$ $\sigma_\omega$ is a homomorphism because \begin{align*} \sigma_\omega(a(x)b(x)) &= a(\omega)b(\omega) \ &= \sigma_\omega(a(x))\sigma_\omega(b(x)) \end{align*} Which has a kernel of \begin{align*} \ker \sigma_\omega &= { a(x) : \sigma_\omega(a(x)) = a(\omega) = 0 } \ &= J \end{align*} The kernel of any homomorphism is an ideal. In $F[x]$ every ideal is a principal ideal so $J = \langle p(x) \rangle$. So $p(x)$ is a polynomial of lowest degree among all nonzero polynomials in $J$. Hence it is irreducible.

When $n$ is prime, then $$x^{n - 1} + x^{n - 2} + \cdots + x + 1$$ is irreducible. Therefore $p(x) = x^{n - 1} + \cdots + 1$ and $p(\omega) = 0$. Since $$\mathbb{Q}(\omega) \cong \mathbb{Q}[x] / \langle p(x) \rangle$$ Then $$[\mathbb{Q}(\omega):\mathbb{Q}] = \deg p(x) = n - 1$$

Q3

$$p(\omega) = 0 \implies \omega^{n - 1} = -(\omega^{n - 2} + \cdots + \omega + 1)$$

Q4

n = 6

$$x^6 - 1 = (x^3 - 1)(x^3 + 1)$$

The roots are $1, s, s^2$ and $-1, -s, -s^2$ respectively where $s$ is the third root of unity.

But from above we know that $s^2 + s + 1 = 0 \implies s^2 = -(s + 1)$ and $s^2 \in \mathbb{Q}(s)$, so $\mathbb{Q}(\omega) = \mathbb{Q}(s)$ with basis ${ 1, s }$.

\begin{align*} [\mathbb{Q}(\omega):\mathbb{Q}] &= [\mathbb{Q}(s): \mathbb{Q}] \ &= 2 \end{align*}

n = 7

$n = 7$ is prime so $[\mathbb{Q}(\omega): \mathbb{Q}] = 6$.

n = 8

\begin{align*} x^8 - 1 &= (x^4 - 1)(x^4 + 1) \ &= (x^2 - 1)(x^2 + 1)(x^4 + 1) \end{align*}

With roots $-1, 1$ and $i, -i$ for $(x^2 - 1)$ and $(x^2 + 1)$ respectively.

The 4th roots of $-1$ are $\pm \frac{1}{\sqrt{2}} \pm \frac{i}{\sqrt{2}}$.

$\mathbb{Q}(\omega) = \mathbb{Q}(\sqrt{2}, i)$ with a basis ${ 1, \sqrt{2}, i, \sqrt{2}i }$ and $$[\mathbb{Q}(\omega):\mathbb{Q}] = [\mathbb{Q}(\sqrt{2}, i) : \mathbb{Q}] = 4$$

Q5

$$\forall r \in { 1, 2, \dots, n - 1 }, (\sqrt[n]{a} \omega^r)^n = 1$$

$$| { \sqrt[n]{a} \omega^r : r \in { 0, 1, \dots, n - 1 } } | = n$$

Q6

The basis of $\mathbb{Q}(\omega, \sqrt[n]{a})$ is the set ${ \omega^i (\sqrt[n]{a})^j }$ where $i, j \in { 0, 1, \dots, n - 1 }$, which contains the ideal for $\sigma(c) = x^n - 1$, which is $J = { \sqrt[n]{a}, \sqrt[n]{a} \omega, \dots, \sqrt[n]{a} \omega^{n - 1} }$.

Q7

The degree of $[\mathbb{Q}(\omega, \sqrt[3]{2}):\mathbb{Q}] = [\mathbb{Q}(\omega, \sqrt[3]{2}):\mathbb{Q}(\sqrt[3]{2})][\mathbb{Q}(\sqrt[3]{2}:\mathbb{Q}] = 2 \times 3 = 6$.

You can calculate $\cos (\pi / 3)$ and $\sin (\pi / 3)$ by splitting an equilateral triangle with unit sides in half. The sum of a triangle's angles will always be $\pi$, and so each corner of the equilateral triangle will be $(\pi / 3)$. Use pythagoreas theorem to calculate the midline as $o^2 = 1^2 - (\frac{1}{2})^2$.

Using wolfram alpha, we see that the cube roots of 1 are $1, -\frac{1}{2} + \frac{\sqrt{3}}{2} i, -\frac{1}{2} - \frac{\sqrt{3}}{2} i$.

$$\mathbb{Q}(\omega, \sqrt[3]{2}) = \mathbb{Q}(i\sqrt{3}, \sqrt[3]{2})$$

Q8

Let $s$ be the nth root of $a$ in $K$, then it is the root of $x^n - a$. Then $\sqrt[n]{a} \omega^i$ is also a root of this polynomial. For every $n$ there is an irreducible cyclotomic polynomial $p(x)$ with roots that consist of $\phi(n)$ nth roots of $a$ that have a multiplicative order of $n$. By theorem 7, since $p(x)$ has one root in $K$ and is irreducible, therefore all its roots are in $K$.

For $n$ is not prime, $p(x) \mid x^n - a$ so all $\phi(n)$ primitive roots of $p(x)$ are also roots of $x^n - a$ and so generates all roots. We use the irreducible polynomial $p(x)$ to prove there's an isomorphism permuting roots of $p(x)$, therefore showing both fields are equivalent and contain all roots.

Fixing $\mathbb{Q}$, there is an isomorphism $h: \mathbb{Q}(s) \rightarrow \mathbb{Q}(\sqrt[n]{a} \omega^i)$. Note that $p(h(c)) = (h(c))^n - a = h(c^n - a) = 0$ as the function is homomorphic so $h(c)^n = h(c^n)$ and fixes $\mathbb{Q}$ leaving $a$ unchanged. This function is an isomorphism mapping one to one and onto, we can say that $h(x)$ permutes the roots $c$ of $x^n - a$. So we can see that $\mathbb{Q}(s) = \mathbb{Q}(\sqrt[n]{a} \omega^i)$, and that $K$ contains all roots of $x^n - a$.

Since the spltting field contains all powers of $\sqrt[n]{a} \omega^i$ by the isomorphism $h(x)$ as well as $\sqrt[n]{a} \omega^0 = \sqrt[n]{a}$ and also its inverse $(\sqrt[n]{a})^{-1}$ (because it's a field), so it also contains all nth roots of unity since $(\sqrt[n]{a})^{-1} (\sqrt[n]{a} \omega^i) = \omega^i$.

F. Separable and Inseparable Polynomials

Q1

From theorem 1, $F$ has characteristic 0 $\implies$ irreducible polynomials never have multiple roots.

Q2

Each power of $x$ in $a(x)$ is independent, so $a_m x^m + a_n x^n \neq 0$ when $m \neq 0$ unless $a_m = a_n = 0$.

Therefore $a'(x) = \sum m_i a_i x^{m_i - 1} = 0 \implies p \mid m_i$ for all $m_i$ and all nonzero terms of $a(x)$ are of the form $a_{mp} x^{mp}$.

Q3

$$a(x) = (x - c)^2 q(x)$$ $$a'(x) = 2(x - c)q(x) + (x - c)^2 q'(x)$$ Both $a(x)$ and $a'(x)$ have $c$ as a root, but $a(x)$ is irreducible, so $a(x) \mid a'(x)$, but this cannot be true since $\deg a'(x) &lt; \deg a(x)$ unless $a'(x) = 0$.

From the previous answer $a'(x) = 0$ means the nonzero terms of $a(x)$ are of the form $a_{mp} x^{mp}$, and $a(x)$ is a polynomial in powers of $x^p$.

Q4

$a(x)$ is a polynomial in powers of $x^p \implies a'(x) = 0 \implies a(x) \mid a'(x) \implies$ share common factor of $a(x) \implies a(x)$ has a multiple root.

Q5

This follows from $[a(x) + b(x)]^p = a(x)^p + b(x)^p$ in any field of characteristic $p$ and is proved in 24D6.

Q6

$$a(x) = a_0 + a_1 x + \cdots + a_n x^n$$ $$a(x^p) = a_0 + a_1 x^p + \cdots + a_n x^{np}$$

The frobenius automorphism for a finite field of characteristic $p$ is bijective since the function is injective, and any injective function from a finite set to itself is also surjective.

This implies the coefficients of $a(x^p)$ all have $p$th roots, and so \begin{align*} a(x^p) &= c_0^p + c_1^p x^p + \cdots + c_n^p x^{np} \ &= (c_0 + c_1 x + \cdots + c_n x^n)^p \end{align*} Thus $b(x) = c_0 + c_1 x + \cdots + c_n x^n$ and $a(x^p) = [b(x)]^p$.

Q7

Assume there is an irreducible polynomial $a(x)$ that is inseperable. Then it is a polynomial in powers of $x^p$.

Then there is a polynomial $[b(x)]^p = a(x)$. Thus $a(x)$ is reducible, which is a contradiction.

Thus every irreducible polynomial is separable.

G. Multiple Roots over Infinite Fields of Nonzero Characteristic

Q1

There are infinite powers of $y$, so $\mathbb{Z}_p[y]$ is an infinite ring with characteristic $p$.

Thus $$\mathbb{Z}_p(y) = { a(y) / b(y) : a(y), b(y) \in \mathbb{Z}_p[y] }$$ is an infinite field, and so is $\mathbb{Z}_p(y^p)$

Q2

By binomial theorem, all coefficients for the terms of $(x - y)^p$ are a factor of $p$. $E[x]$ has characteristic $p$ so $$x^p - y^p = (x - y)^p$$ However $y \notin K[x]$ so $x^p - y^p$ is irreducible in $K[x]$.

Q3

$$\repr{i}(a_0 + \cdots + a_n x^n) = i(a_0) + \cdots + i(a_n) x^n$$ but for all $a_i \in F, i(a_i) = a_i$ so $\repr{i}(p(x)) = p(x)$.

Q4

Expanding out $(x - a)^m$, the coefficients in $F$ and $x$ remain fixed, but $a$ is mapped to $b$, so $$\repr{i}((x - a)^m) = (x - b)^m$$

Q5

Since $\repr{i}$ leaves $p(x)$ fixed, so \begin{align*} \repr{i}(p(x)) &= \repr{i}((x - a)^m s(x)) \ &= (x - b)^m \repr{i}(s(x)) \ &= p(x) \end{align*} so $a$ and $b$ have the same multiplicity in $p(x)$.

H. An Isomorphism Extension Theorem (Proof of Theorem 3)

Q1

$$F_1(a) \cong F_1[x] / \langle p(x) \rangle$$

Where $p(x)$ is the minimum polynomial with $a$ as a root. The homomorphism $\phi : F_1[x] \rightarrow F_1(a)$ by $\phi_c(a(x)) = a(c)$ has the kernel $J = \langle p(x) \rangle$ since in $F[x]$ every ideal is a principal ideal.

Thus since $s(x) = c(x) - d(x)$ has a root $a$ since $s(a) = 0$, so $s(x) \in J$ and it is a multiple of $p(x)$.

Observing that $h(c(x) - d(x)) = h(p(x)q(x))$, we easily see that $h(p(x)q(x)) = hp(x) hq(x)$ and also that $h(c(x) - d(x)) = hc(x) - hd(x)$ since

\begin{align*} hc(x) - hd(x) &= (h(c_0) - h(d_0)) + (h(c_1) - h(d_1))x + \cdots + (h(c_n) - h(d_n)) x^n \ &= h(c_0 - d_0) + h(c_1 - d_1) x + \cdots + h(c_n - d_n) x^n \ &= h(c(x) - d(x)) \end{align*}

Q2

\begin{align*} h(c(a)) &= h(c_0) + h(c_1) b + \cdots + h(c_n) b^n \ h(d(a)) &= h(d_0) + h(d_1) b + \cdots + h(d_n) b^n \end{align*} \begin{align*} h(c(a)) - h(d(a)) &= h(c_0 - d_0) + h(c_1 - d_1) b + \cdots + h(c_n - d_n) b^n \ &= h(c(x) - d(x))(b) \ &= [hp(x)(b)][hq(x)(b)] \end{align*} But $hp(x)(b) = 0$ so $h(c(a)) - h(d(a)) = 0 \implies h(c(a)) = h(d(a))$.

Q3

$$hc(x) = hd(x)$$ \begin{align*} \implies hc(x) - hd(x) &= 0 \ &= h(c(x) - d(x)) \ &= h(c_0 - d_0) + h(c_1 - d_1) x + \cdots + h(c_n - d_n) x^n \end{align*} But $h$ is isomorphic on $F_1 \rightarrow F_2$ so $c_i = d_i \implies c(x) = d(x)$.

Q4

$h(a) = b$, there is no other value that produces $b$.

All the coefficients for a polynomial $c(x)$ are reversible.

$$c(x) = h^{-1}(h(c_0)) + h^{-1}(h(c_1)) x + \cdots + h^{-1}(h(c_n)) x^n$$

Q5

\begin{align*} h(c(x) + d(x)) &= h(c_0 + d_0) + h(c_1 + d_1) x + \cdots + h(c_n + d_n) x^n \ &= (h(c_0) + h(d_0)) + (h(c_1) + h(d_1)) x + \cdots + (h(c_n) + h(d_n)) x^n \ &= hc(x) + hd(x) \end{align*} \begin{align*} h(c(x)d(x)) &= h(c_0 d_0) + h(\sum_{i + j = 1} c_i d_j x) + h(\sum_{i + j = 2} c_i d_j x^2) + \cdots + h(\sum_{i + j = n} c_i d_j x^n) \ &= h(c_0) h(d_0) + \sum_{i + j = 1} h(c_i) h(d_j) x + \sum_{i + j = 2} h(c_i) h(d_j) x^2 + \cdots + \sum_{i + j = n} h(c_i) h(d_j) x^n \ &= hc(x) hd(x) \end{align*}

I. Uniqueness of the Root Field

Q1

First note that $$F_1(u) \cong F_1[x] / \langle p(x) \rangle$$ Let $f : F_1[x] \rightarrow F_2(v)$ defined by $$f(a(x)) = h(a(x))(v)$$ Then the ideal is $J = \langle p(x) \rangle$ \begin{align*} f(a(x)) = f(b(x)) &\iff f(a(x)) - f(b(x)) = 0 \ &\iff f(a(x) - b(x)) = 0 \ &\iff a(x) - b(x) \in J \ &\iff J + a(x) = J + b(x) \end{align*} Let $\phi : F_1[x] / \langle p(x) \rangle \rightarrow F_2(v)$ by $$\phi(J + a(x)) = f(a(x)) = h(a(x))(v)$$ We can see that this function is an ismomorphism:

  • injective: $\phi(J + a(x)) = \phi(J + b(x)) \implies f(a(x)) = f(b(x))$ $$\implies J + a(x) = J + b(x)$$
  • surjective: $h(a(x))(v) = h(a(u))$ which is onto $F_2(v)$ and surjective, so $f(a(x)) = \phi(J + a(x))$ is surjective.
  • Finally, \begin{align*} \phi(J + a(x)) + \phi(J + b(x)) &= f(a(x)) + f(b(x)) \ &= ha(x) + hb(x) \ &= h(a(x) + b(x))(v) \ &= \phi(J + a(x) + b(x)) \end{align*}

$$\implies F_2(v) \cong F_1[x] / \langle p(x) \rangle$$ $$F_1(u) \cong F_1[x] / \langle p(x) \rangle$$ $$F_1(u) \cong F_2(v)$$

Q2

We start with $F_1(u) = K_1$ and want to prove that this means $F_2(v) = K_2$. This will also automatically prove the converse statement if shown to be true.

Let $p(x)$ be the minimum polynomial $p(x)$ for $u$ such that $p(u) = 0$.

$h: F_1(u) \rightarrow F_2(v)$ but $F_1(u) = K_1$ so $h: K_1 \rightarrow F_2(v)$, but $h$ is surjective and so $\deg p(x) = \deg hp(x)$ since both $p(x)$ and $hp(x)$ are irreducible.

$$\forall u_i \in K_1 : p(u_i) = 0, \exists v_i = h(u_i) : hp(v_i) = 0$$ $$\implies F_2(v) = K_2$$

Because there are $\deg hp(x) = \deg p(x)$ such roots $v_i$ which correspond to $u_i$ roots of $p(x)$.

Q3

$$a(x) = p(x) q(x)$$ $$p(u) = 0$$ \begin{align*} h(p(u)) &= h(p(x))(v) = 0 \ &= hp(x) = hp(v) \end{align*} Since both are equivalent.

We see that $v$ is a root of $hp(x)$.

$F_1(u) = K_1$

If $F_1(u) = K_1$, then $F_1(u)$ contains all roots of $p(x)$ and then $$F_1(u) = K_1 \iff F_2(v) = K_2$$ Recalling that $p(x)$ is an irreducible factor of $a(x)$, $$u \in K_1$$ $$p(u) = 0$$ $$v \in K_2$$ $$hp(v) = 0$$ $$\implies F_1(u) \cong F_2(v)$$ Putting both together $$K_1 \cong K_2$$

$F_1(u) \neq K_1$

See that we can extend $h$ fixing the base field. $$h(u) = v$$ $$h : F_1(u) \rightarrow F_2(v)$$ In $F_1(u)[x]$, $a(x) = (x - u)a_1(x)$.

In $F_2(v)[x], ha(x) = (x - v)ha_1(x)$.

And $\deg a_1(x) = \deg ha_1(x) = n - 1$

Now let there be a new $u' \in K_1, u' \notin F_1(u'), p(u') = 0$ and likewise for $hp(x)$ and $v$.

$\deg a(x) = 1$

Lastly when $n = \deg a(x) = 1$, then $K_1 = F_1$ and $K_2 = F_2$, since the basis are simply scalars and $a(x)$ is of the form $(x - a)$. The root of $a(x)$ is in $F_1$ itself, and $h$ is an ismomorphism from $F_1 \rightarrow F_2$ so $$K_1 \cong K_2$$

Q4

$$u \in K_1, v \in K_2$$ But $F_1 = F_2$ $$h : F[x] \rightarrow F[x]$$ $$\forall a \in F, h(a) = a$$ $$h = \textrm{id}_F$$ $$h(u) = v$$ $$\implies K_1 \cong K_2$$

J. Extending Isomorphism

Q1

$$h : \mathbb{Q}(\omega) \rightarrow \mathbb{C}, \forall a \in \mathbb{Q}, h(a) = a$$ Let $b \in \mathbb{Q}(\omega)$ $$b = s_0 + s_1 \omega + s_2 \omega^2 + \cdots + s_{p - 1} \omega^{p - 1}$$ $$h(b) = s_0 + s_1 h(\omega) + s_2 h(\omega)^2 + \cdots + s_{p - 1} \omega^{p - 1}$$ So $h$ is determined by $h(\omega)$.

Since isomorphisms preserve roots, we deduce they permute roots. There are $p - 1$ roots for the minimum polynomial of $\omega$ which has degree $p - 1$.

Q2

$p(x)$ is irreducible in $F[x]$. $c \in \mathbb{C} : p(c) = 0$.

Let $h : F \rightarrow \mathbb{C}$ be a monomorphism (injective homomorphism), then $h : F \rightarrow h(F)$ is an isomorphism and $$F \cong h(F)$$ The minimum polynomial are $p(x)$ and $hp(x)$ respectively with $\deg p(x) = \deg hp(x) = n$. Since $h$ permutes roots and by theorem 7 contains all roots, there are $n$ possible monomorphisms.

Q3

$$h : F \rightarrow h(F)$$ $$\phi : K \rightarrow \mathbb{C}$$ $$[K : F] = n$$ So $K$ forms a splitting field over $F$ for a minimum polynomial $p(x) \in F[x]$ of degree $n$.

From the previous question we see that there are $n$ monomorphisms $F(c) = K \rightarrow \mathbb{C}$.

Q4

$$h : \mathbb{Q} \rightarrow \mathbb{C}$$ $$h(x) = x$$ $$h(x) = h(y) \implies x = y$$ $$h(1) = 1_{\mathbb{C}}$$ \begin{align*} h\left(\frac{n}{n}\right) &= h\left(\frac{1}{n} + \cdots + \frac{1}{n}\right) \ 1 &= nh\left(\frac{1}{n}\right) \ \implies h\left(\frac{1}{n}\right) &= \frac{1}{n} \end{align*} \begin{align*} h\left(\frac{p}{q}\right) &= h\left(\frac{1}{q} + \cdots + \frac{1}{q}\right) \ &= \underbrace{h\left(\frac{1}{q}\right) + \cdots + h\left(\frac{1}{q}\right)}_p \ &= \frac{p}{q} \end{align*}

Thus all monomorphisms $h : \mathbb{Q}(a) \rightarrow \mathbb{C}$ fix $\mathbb{Q}$.

Q5

$$\mathbb{Q}(\sqrt[3]{2}) \rightarrow \mathbb{C}$$ Minimum polynomial for $c = \sqrt[3]{2}$ is $p(x) = x^3 - 2$

Three roots of $p(x)$ are $\sqrt[3]{2}, \sqrt[3]{2} \omega, \sqrt[3]{2} \omega^2$.

$\mathbb{Q}$ remains fixed. Roots are permuted (see above questions).

Three monomorphisms are $$\sqrt[3]{2} \rightarrow \sqrt[3]{2}$$ $$\sqrt[3]{2} \rightarrow \sqrt[3]{2} \omega$$ $$\sqrt[3]{2} \rightarrow \sqrt[3]{2} \omega^2$$

K. Normal Extensions

Q1

$K$ is a finite extension of $F \implies K$ is a simple extension and so that $K = F(c)$.

There is a minimum polynomial $p(x)$ for $c$ in $F$.

So by the question $K$ is a normal extension.

Q2

$K$ is a finite extension of $F \implies K = F(c)$.

Let $p(x)$ be the minimum polynomial for $c$, so $p(c) = 0$.

Let $h$ be an isomorphism fixing $F$ $$h: K \rightarrow h(K)$$ Then by the question $h(K) \subseteq K$ and since $h$ is an isomorphism where $K \cong h(K)$, so $h(K) = K$, and $h : K \rightarrow K$ is an automorphism fixing $F$ and permuting roots of $p(x)$.