Browse Source

galois-notes.tex: detour, isomorphism theorems

master
arnaucube 16 hours ago
parent
commit
6593857e91
2 changed files with 222 additions and 10 deletions
  1. BIN
      galois-theory-notes.pdf
  2. +222
    -10
      galois-theory-notes.tex

BIN
galois-theory-notes.pdf


+ 222
- 10
galois-theory-notes.tex

@ -37,6 +37,21 @@
note = {\url{https://scipp.ucsc.edu/~haber/ph251/Dn_subgroups.pdf}},
url = {https://scipp.ucsc.edu/~haber/ph251/Dn_subgroups.pdf}
}
@misc{judson,
author={Thomas W. Judson},
title={Abstract Algebra: theory and applications},
year={1994},
note={Available at \url{http://abstract.ups.edu/download.html} },
pages={438}
}
@misc{dummitfoote,
author={David S. Dummit and Richard M. Foote},
title={Abstract Algebra (Third Edition)},
year={2004},
pages={945}
}
\end{filecontents}
\nocite{*}
@ -87,7 +102,8 @@
\tableofcontents
\section{Recap on the degree of field extensions}
\section{Galois Theory notes}
\subsection{Chapters 4-6}
(Definitions, theorems, lemmas, corollaries and examples enumeration follows from Ian Stewart's book \cite{ianstewart}).
\begin{defn}{4.10}
@ -176,6 +192,201 @@
[...] TODO: pending to add key parts up to Chapter 15.
\subsection{Detour: Isomorphism Theorems}
\begin{thm}{}(\emph{First Isomorphism Theorem}) \label{1stisothm}
\begin{minipage}{0.75\textwidth}
If $\psi: G \longrightarrow H$ a group homomorphism, then $ker(\psi) \triangleleft G$.\\
Let $\phi: G \longrightarrow G/ker(\psi)$ be the canonical homomorphism.\\
Then $\exists$ unique isomorphism $\eta: G/ker(\psi) \longrightarrow \psi(G)$ such that $\psi = \eta \phi$.\\
$\Longleftrightarrow$ ie. $G/ker(\psi) \cong \psi(G)$.
\end{minipage}
\hfill
\begin{minipage}{0.2\textwidth}
\begin{tikzpicture}[node distance=2.5cm, auto]
\node (G) {$G$};
\node (H) [right of=G] {$H$};
\node (GmodK) [below of=G, xshift=1.25cm] {$G/\ker(\psi)$};
\draw[->] (G) to node {$\psi$} (H);
\draw[->] (G) to node [swap] {$\phi$} (GmodK);
\draw[->] (GmodK) to node [swap] {$\eta$} (H);
\end{tikzpicture}
\end{minipage}
\end{thm}
\begin{proof}
\emph{(proof from Thomas W. Judson book "Abstract Algebra" \cite{judson})}\\
Let $K=ker(\psi)$.
Since
$$\eta: G/K \longrightarrow \psi(G)$$
let
$$\eta: gK \longrightarrow \psi(g)$$
ie. $\eta(gK)=\psi(g)$.
\begin{enumerate}[i.]
\item show that $\eta$ is a \emph{well defined} map:
if $g_1 K=g_2 K$, then for some $k \in K$, $g_1 k =g_2$, so
$$\eta(g_1K)=\psi(g_1) = \psi(g_1)\psi(k) = \psi(g_1 k) = \psi(g_2) = \eta(g_2 k)$$
Thus, $\eta$ does not depend on the choice of coset representatives, and
the map $\eta: G/ker(\psi) \longrightarrow \psi(G)$ is uniquely defined
since $\psi=\eta\phi$.
\item show that $\eta$ is a homomorphism:
Observe:
$$\eta(g_1 K g_2 K) = \eta(g_1 g_2 K) = \psi(g_1 g_2) = \psi(g_1) \psi(g_2) = \eta(g_1 K) \eta(g_2 K)$$
$\Longrightarrow$ so $\eta$ is a homomorphism.
\item show that $\eta$ is an isomorphism:
Since each element of $H=\psi(G)$ has at least a preimage, then $\eta$ is \emph{surjective} (onto $\psi(G)$).
Show that it is also \emph{injective} (onet-to-one):
Suppose 2 different preimatges lead to the same image in $\psi(G)$, ie.
$\eta(g_1 K) = \eta(g_2 K)$
then,
$$\psi(g_1) = \psi(g_2)$$
which implies $\psi(g_1^{-1} g_2) = e$, ie. $g_1^{-1} g_2 \in ker(\psi)$,
hence
$$g_1^{-1} g_2 K = K$$
$$g_1 K = g_2 K$$
so $\eta$ is injective.
\end{enumerate}
Since $\eta$ is injective and surjective $\Longrightarrow$ $\eta$ is a bijective homomorphism,\\
ie. $\eta$ is an \emph{isomorphism}.
\end{proof}
\begin{thm}{}(\emph{Second Isomorphism Theorem}) \label{2ndisothm}
Let $H \subseteq G$, $N \triangleleft G$. Then
\begin{enumerate}[i.]
\item $HN \subseteq G$
\item $H \cap N \triangleleft H$
\item $\frac{H}{H \cap N} \cong \frac{HN}{N}$
\end{enumerate}
\end{thm}
\begin{proof}
\emph{(proof from Thomas W. Judson book "Abstract Algebra" \cite{judson})}\\
\begin{enumerate}[i.]
\item show $HN \subseteq G$:\\
Note that $HN = \{ hn : h\in H, n\in N \}$. Let $h_1 n_1, h_2 n_2 \in HN$.
Since $N$ normal $\Longrightarrow~ h_2^{-1} n_1 h_2 \in N$, so
$$(h_1 n_1)(h_2 n_2) = h_1 h_2 (h_2^{-1} n_1 h_2) \in HN$$
[Recall: since $N \triangleleft G$, $gN=Ng ~\forall g \in G$ $\Longrightarrow gn=n'g$ for some $n' \in N$.]
To see that $(hn)^{-1} \in HN$:\\
since $(hn)^{-1} = n^{-1} h^{-1} = h^{-1} (h n^{-1} h^{-1})$, thus $(hn)^{-1} \in HN$.
Thus $HN \subseteq G$.
In fact, $$HN = \bigcup_{h \in H} hN$$
(TODO: diagram)
\item show that $H \cap N \triangleleft H$:\\
Let $h \in H,~ n \in H \cap N$ (recall: $H \cap N \subseteq H$).\\
Then $h^{-1}nh \in H$ $\longleftarrow$ since $h^{-1}, n, h \in H$.\\
Since $N \triangleleft G$, $h^{-1} n h \in N$.\\
Therefore, $h^{-1}nh \in H \cap N$ $\Longrightarrow~ H \cap N \triangleleft H$
\item show that $\frac{H}{H \cap N} \cong \frac{HN}{N}$:\\
Define a map
\begin{align*}
\phi: &H \longrightarrow \frac{HN}{N}\\
\text{by}~ \phi: &h \longmapsto hN
\end{align*}
$\phi$ is surjective (onto), since any coset $hnN=hN$ is the image of $h \in H$, ie. $\phi(h)$
$\phi$ is a homomorphism, since
$$\phi(h h') = h h' N = hN h'N = \phi(h)\phi(h')$$
By the First Isomorphism Theorem \ref{1stisothm},
$$\frac{HN}{N} \cong \frac{H}{ker(\phi)}$$
and since
\begin{align*}
&ker(\phi) = \{ h \in H : h \in N\}\\
&\text{then}~~ker(\phi) = H \cap N
\end{align*}
so then,
$$\frac{HN}{N} = \phi(H) \cong \frac{H}{ker(\phi)} = \frac{H}{H \cap N}$$
thus
$$\frac{HN}{N} \cong \frac{H}{H \cap N}$$
\end{enumerate}
\end{proof}
\begin{thm}{}(\emph{Third Isomorphism Theorem}) \label{2ndisothm}\\
Let $H \subseteq K$ and $K \triangleleft G,~ H \triangleleft G$.\\
Then
$\frac{K}{H} \triangleleft \frac{G}{H}$
and
$$\frac{ G/H }{ K/H } \cong \frac{G}{K}$$
\end{thm}
\begin{proof}
\emph{(proof from Dummit and Foote book ``Abstract Algebra" \cite{dummitfoote})}\\
Easy to see that $\frac{K}{H} \triangleleft \frac{G}{H}$.
Define
\begin{align*}
\psi: &\frac{G}{H} \longrightarrow \frac{G}{K}\\
\text{by}~ \psi: &gH \longmapsto gK
\end{align*}
To show that $\psi$ is \emph{well defined}:\\
suppose $g_1 H = g_2 H$, then $g_1 = g_2 h$ for some $h \in H$.\\
Since $H \subseteq K \Longrightarrow~ h \in K$, hence $g_1 K = g_2 K$,\\
ie. $\psi(g_1 H) = \psi(g_2 H)$, which shows that $\psi$ is well defined.
Since $g \in G$ may be chosen arbitrarily in $G$, $\psi$ is a surjective homomorphism.
Finally,
\begin{align*}
ker(\psi) &= \{ gH \in \frac{G}{H} \mid \psi(gH)=1K \}\\
&= \{ gH \in \frac{G}{H} \mid gK =1K \}\\
&= \{ gH \in \frac{G}{H} \mid g \in K \}\\
&= \frac{K}{H}
\end{align*}
By the First Isomorphism Theorem (\ref{1stisothm}),
\begin{tikzpicture}[node distance=2.5cm, auto]
\node (GmodH) {$\frac{G}{H}$};
\node (GmodK) [right of=GmodH] {$\frac{G}{K}$};
\node (GmodHmodKer) [below of=G, xshift=1.25cm] {$\frac{G/H}{\ker(\psi)} = \frac{G/H}{K/H}$};
\draw[->] (GmodH) to node {$\psi$} (GmodK);
\draw[->] (GmodH) to node [swap] {$\phi$} (GmodHmodKer);
\draw[->] (GmodHmodKer) to node [swap] {$\eta$} (GmodK);
\end{tikzpicture}
So, by
$$\eta: \frac{G/H}{K/H} \longrightarrow \frac{G}{K}$$
since $\eta$ is
bijective (we know it by the First Isomorphism Theorem), $\eta$ it is the isomorphism:
$$\frac{ G/H }{ K/H } \cong \frac{G}{K}$$
\end{proof}
\subsection{Chapter 14}
\newpage
@ -296,7 +507,7 @@ $$t^{12}-1 = \Phi_1 \Phi_2 \Phi_3 \Phi_4 \Phi_6 \Phi_{12}$$
\begin{defn}{21.5}
The polynomial $\Phi_d(t)$ defined by
$$\Phi_n(t) = \prod_{a\in \mathbb{Z}_n,(a,n)=1} (t- \zeta^a)$$
is the $n$-th \emph{cyclotomic polynomial} over \mathbb{C}.
is the $n$-th \emph{cyclotomic polynomial} over $\mathbb{C}$.
\end{defn}
\begin{cor}{21.6}
@ -305,14 +516,15 @@ $$t^{12}-1 = \Phi_1 \Phi_2 \Phi_3 \Phi_4 \Phi_6 \Phi_{12}$$
\begin{thm}{21.9}
\begin{enumerate}
\item The Galois group $\Gamma(\mathbb{Q}(\zeta):\mathbb{Q})$ consists of the
$\mathbb{Q}$-automorphisms $\psi_j$ defined by
$$\psi_j(\zeta)=\zeta^j$$
where $0 \leq j \leq n-1$ and $j$ is prime to $n$.
\item $\Gamma(\mathbb{Q}(\zeta):\mathbb{Q}) \stackrel{iso}{\cong} \mathbb{Z}_n^*$, and is an abelian group.
\item its order is $\phi(n)$
\item if $n$ is prime, $\mathbb{Z}_n^*$ is cyclic
\item The Galois group $\Gamma(\mathbb{Q}(\zeta):\mathbb{Q})$ consists of the
$\mathbb{Q}$-automorphisms $\psi_j$ defined by
$$\psi_j(\zeta)=\zeta^j$$
where $0 \leq j \leq n-1$ and $j$ is prime to $n$.
\item $\Gamma(\mathbb{Q}(\zeta):\mathbb{Q}) \stackrel{iso}{\cong} \mathbb{Z}_n^*$, and is an abelian group.
\item its order is $\phi(n)$
\item if $n$ is prime, $\mathbb{Z}_n^*$ is cyclic
\end{enumerate}
\end{thm}

Loading…
Cancel
Save