%! TEX root = Galois.tex % vim: tw=50 % 10/10/2023 12AM \begin{flashcard}[transcendental-defn] \begin{definition}[Transcendental number] \glsdefn{transc_el}{transcendental}{N/A} \cloze{ Let $L \extendover K$ be a \gls{field_ex}. $\alpha \in L$ is \emph{transcendental} over $K$ if it is not \gls{alg_el}. In this case $\ringadj{K}{\alpha} \cong \ringadj{K}{X}$ and $\fieldadj{K}{\alpha} \cong \fieldadj{K}{X}$. } \end{definition} \end{flashcard} \begin{remark*} Since $\ringadj{K}{X}$ is not a \gls{field} and $1, X, X^2, \ldots$ are linearly independent over $K$, we have $\fieldadj{K}{\alpha} \neq \ringadj{K}{\alpha}$ and $\degover{\fieldadj{K}{\alpha} }{K} = \infty$. \end{remark*} \begin{flashcard}[algebraic-defn] \begin{definition*}[Algebraic Field] \glsdefn{alg_ex}{algebraic}{N/A} \cloze{ We say a \gls{field_ex} $L \extendover K$ is \emph{algebraic} if every $\alpha \in L$ is \gls{alg_el} over $K$. } \end{definition*} \end{flashcard} \begin{remark*} \phantom{} \begin{enumerate}[(i)] \item We have $\degover{\fieldadj{K}{\alpha}}{K} < \infty$ if and only if $\alpha$ is \gls{alg_el} over $K$. \item If $\degover{L}{K} < \infty$ then for any $\alpha \in L$ we certainly have $\degover{\fieldadj{K}{\alpha}}{K} < \infty$. So any \gls{finite_ex} \gls{ex} is \gls{alg_ex}, but the inverse is not true. \begin{example} \vspace{0.5em} $K = \QQ$, $L = \bigcup_{n = 1}^\infty \ringadj{\QQ}{\sqrt[2^n]{2}} \subset \RR$. This is a union of a nested sequence of \glspl{field} \[ \fieldadj{\QQ}{\sqrt{2}} \subset \fieldadj{\QQ}{\sqrt[4]{2}} \subset \fieldadj{\QQ}{\sqrt[8]{2}} \subset \cdots \] and so this is a \gls{field}. $\degover{L}{K} = \infty = \infty$ since $\degover{\fieldadj{\QQ}{\sqrt[2^n]{n}}}{\QQ} = 2^n$ is unbounded. But every $\alpha \in L$ belongs to a \gls{finite_ex} \gls{ex} of $\QQ$ and so is \gls{alg_el} over $\QQ$. Therefore $L \extendover \QQ$ is \gls{alg_ex}. \end{example} \end{enumerate} \end{remark*} \begin{remark} \glsdefn{transc_Q}{transcendental}{N/A} \glsdefn{alg_Q}{algebraic}{N/A} Classically $\alpha \in \CC$ is called algebraic / transcendental if it is \gls{alg_el} / \gls{transc_el} over $\QQ$. A countability argument (see IA Numbers and Sets) shows that \gls{transc_Q} numbers exist. Liouville showed that $\sum_{n = 1}^\infty \frac{1}{10^{n!}}$ is \gls{transc_Q}. It was proved in the 19th century that $e$ and $\pi$ are \gls{transc_Q} (Hermite \& Lindeman). \end{remark} \begin{example} \label{lec3_last_example} \phantom{} \begin{enumerate}[(i)] \item Let $f(X) = X^d - n$ ($d, n \in \ZZ$, $d \ge 2$, $n \neq 0$). Suppose there exists a prime $p$ such that when we write $n = p^e m$ with $p \nmid m$, then $(d, e) = 1$. We claim that $f$ is irreducible in $\ringadj{\QQ}{X}$. Equivalently we show that $\degover{\fieldadj{\QQ}{\alpha}}{ \QQ} = d$ where $\alpha = \sqrt[d]{n}$. By Euclid's algorithm there exist $r, s \in \ZZ$ such that $rd + se = 1$ (we may arrange $s > 0$). Then $p^{dr} n^s = p^{dr} (p^e m)^s = pm^s$. Let $\beta = p^r \alpha^s$ so that $\beta^d = p m^s$. Then $\beta$ is a root of $g(X) = X^d - p m^s$, which is irreducible in $\ringadj{\ZZ}{X}$ by Eisenstein's criterion, so irreducible in $\ringadj{\QQ}{X}$ by Gauss' Lemma, therefore $\degover{\fieldadj{\QQ}{\beta}}{\QQ} = d$. But $\fieldadj{\QQ}{\beta} \subset \fieldadj{\QQ}{\alpha}$ and $\degover{\fieldadj{\QQ}{\alpha}}{\QQ} \le \deg f = d$. This gives $\degover{\fieldadj{\QQ}{\alpha}}{\QQ} = d$ as required. \item Let $p$ be an odd prime, $\zeta_p = e^{2\pi i / p}$ and \[ \alpha = 2\cos \left( \frac{2\pi}{p} \right) = \zeta_p + \zeta_p^{-1} .\] Let's compute $\degover{\fieldadj{\QQ}{\alpha}}{\QQ}$. $\zeta_p$ is a root of \[ f(X) = \frac{X^p - 1}{X - 1} = X^{p - 1} + \cdots + X^2 + X + 1 .\] which is irreducible by Eisenstein's criterion applied to $f(X + 1)$. Therefore $\degover{\fieldadj{\zeta_p}}{\QQ} = \deg f = p - 1$. Note that $\fieldadj{\QQ}{\zeta_p} \extendover \fieldadj{\QQ}{\alpha} \extendover \QQ$, so we will try to use \nameref{tower_law} to find the \gls{ex_deg} of the \gls{ex} $\fieldadj{\QQ}{\alpha} \extendover \QQ$. $\zeta_p$ is a root of \[ g(X) = (X - \zeta_p)(X - \zeta_p^{-1}) = X^2 - \alpha X + 1 \in \ringadj{\fieldadj{\QQ}{\alpha}}{X} \] Therefore $\degover{\fieldadj{\QQ}{\zeta_p}}{\fieldadj{\QQ}{\alpha}} \le \deg g = 2$. But $\alpha \in \RR$ and $\zeta_p \not\in \RR$ so $\zeta_p \not\in \fieldadj{\QQ}{\alpha} \subset \RR$. So $\degover{\fieldadj{\QQ}{\zeta_p}} {\fieldadj{\QQ}{\alpha}} = 2$, so by \nameref{tower_law} $\degover{\fieldadj{\QQ}{\alpha}}{\QQ} = \frac{p - 1}{2}$. \item Suppose $m, n$ and $mn$ are not perfect squares. Let $\alpha = \sqrt{m} + \sqrt{n}$. Let's compute $\degover{\fieldadj{\QQ}{\alpha}}{\QQ}$. Clearly $\fieldadjoin\QQ(\alpha) \subset \fieldadjoin\QQ(\sqrt{m}, \sqrt{n})$. Conversely we have \[ m = (\alpha - \sqrt{n})^2 = \alpha^2 - 2\alpha \sqrt{n} + n \] \[ \implies \sqrt{n} = \frac{\alpha^2 - m + n}{2\alpha} \] so $\sqrt{n} \in \fieldadjoin\QQ(\alpha)$, and similarly $\sqrt{m} \in \fieldadjoin\QQ(\alpha)$. So therefore $\fieldadjoin\QQ(\alpha) = \fieldadjoin\QQ(\sqrt{m}, \sqrt{n})$. Note that $\fieldadjoin\QQ(\sqrt{m}, \sqrt{n}) \extendover \fieldadjoin\QQ(\sqrt{n}) \extendover \QQ$, with $\exdeg[\fieldadjoin\QQ(\sqrt{n}) : \QQ] = 2$. We also know $\exdeg[\fieldadj{\fieldadjoin\QQ(\sqrt{n})}{\sqrt{m}} : \fieldadjoin\QQ(\sqrt{n})] \le 2$. Suppose $\sqrt{m} \in \fieldadjoin\QQ(\sqrt{n})$. Then $\sqrt{m} = r + s\sqrt{n}$ for some $r, s \in \QQ$. Therefore $m = r^2 + 2rs\sqrt{n} + s^2n$. Since $\sqrt{n} \not\in \QQ$ we must have $rs = 0$. If $r = 0$ then $mn$ will be a square, and if $s = 0$ then $m$ is a square. Either way we get a contradiction. So $\exdeg[\fieldadjoin\QQ(\sqrt{m}, \sqrt{n}) : \fieldadjoin\QQ(\sqrt{n})] = 2$. So by \nameref{tower_law}, $\exdeg[\fieldadjoin\QQ(\alpha) : \QQ] = 4$. \end{enumerate} \end{example}