%! TEX root = EC.tex % vim: tw=50 % 12/03/2025 11AM % \item Linearity in second argument: Let $T_1, T_2 \in E'[\hat{\phi}]$. \begin{align*} \div(f_1) &= n(T_1) - n(0) &\phi^* f &= g_1^n \\ \div(f_2) &= n(T_2) - n(0) &\phi^* f_2 &= g_2^n \end{align*} There exists $h \in \ol{K}(E')$ such that \[ \div(h) = (T_1) + (T_2) - (T_1 + T_2) - (0) .\] Then put $f = \frac{f_1 f_2}{h^n}$ and $g = \frac{g_1 g_2}{\phi^* h}$. Check: $\div(f) = n(T_1 + T_2) - n(0)$. Yes. \[ \phi^* f = \frac{\phi^* f_1 \phi^* f_2}{(\phi^* h)^n} = \left( \frac{g_1 g_2}{\phi^*(h)} \right)^n = g^n .\] Therefore \begin{align*} \ephi(S, T_1 + T_2) &= \frac{g(X + S)}{g(X)} \\ &= \frac{g_1(X + S)}{g_1(X)} \frac{g_2(X + S)}{g_2(X)} \ub{\frac{h(\phi(X))}{h(\phi(X + S))}}_{\substack{= 1 \\ \text{since $S \in E[\phi]$}}} \\ &= \ephi(S, T_1) \ephi(S, T_2) \end{align*} \item $\ephi$ is non-degenerate. Fix $T \in E'[\hat{\phi}]$. Suppose $\ephi(S, T) = 1$ for all $S \in E[\phi]$. Then $\tau_S^* g = g$ for all $S \in E[\phi]$. Have $\ol{K}(E) / \phi^* \ol{K}(E')$ is a Galois extension with Galois group $E[\phi]$ ($S \in E[\phi]$ acts on $\ol{K}(E)$ via $\tau_S^*$). Therefore $g = \phi^* h$ for some $h \in \ol{K}(E')$. So $\phi^* f = g^n = \phi^*(h^n)$. So $f = h^n$, and hence $\div(h) = (T) - (0)$. So $T = 0$. \qedhere \end{enumerate} \end{proof} We've shown $E'[\phi] \injto \Hom(E[\phi], \mu_n)$. It is an isomorphism since $\#E[\phi] = \#E'[\hat{\phi}] = n$. \begin{remark*} \phantom{} \begin{enumerate}[(i)] \item If $E, E', \phi$ are defined over $K$ then $\ephi$ is Galois equivariant, i.e. $\forall \sigma \in \Gal(\ol{K} / K)$, $\forall S \in E[\phi]$, $\forall T \in E'[\phi]$, \[ \ephi(\sigma S, \sigma T) = \sigma(\ephi(S, T)) .\] \item Taking $\phi = [n] : E \to E$ (so $\hat{\phi} = [n]$) gives \[ e_n : E\tors[n] \times E\tors[n] \to \cancel{\mu_{n^2}} \mu_n .\] \end{enumerate} \end{remark*} \begin{fccoro}[] \label{coro:14.6} Assuming: - $E\tors[n] \subset E(K)$ Then: $\mu_n \subset K$. \end{fccoro} \begin{proof} Let $T \in E\tors[n]$ have order $n$. Non degeneracy of $e_n$ implies that there exists $S \in E\tors[n]$ such that $e_n(S, T)$ is a primitive $n$-th root of unity, say $\zeta_n$. Then \[ \sigma(\zeta_n) = \sigma(e_n(S, T)) = e_n(\cancel{\sigma} S, \cancel{\sigma} T) = \zeta_n \] for all $\sigma \in \Gal(\ol{K} / K)$. So $\zeta_n \in K$. \end{proof} \begin{example*} There does not exist $E / \Qbb$ with $E(\Qbb)_{\text{tors}} \cong (\Zbb / 3\Zbb)^2$. \end{example*} \begin{remark*} In fact, $e_n$ is alternating, i.e. $e_n(T, T) = 1$ for all $T \in E\tors[n]$. (This implies $e_n(S, T) = e_n(T, S)^{-1}$). \end{remark*} \newpage \section{Galois Cohomology} Let $G$ be a group and $A$ be a $G$-module (an abelian group with an action of $G$ via group homomorphisms). $G$-module means exactly the same thing as $\Zbb[G]$-module. \begin{fcdefnstar}[$H^0$] Define \[ H^0(G, A) = A^G = \{a \in A \st \sigma(a) = a~\forall \sigma \in G\} .\] \end{fcdefnstar} \begin{fcdefnstar}[Cochains] \glsnoundefn{cochain}{cochain}{cochains}% Define \[ C^1(G, A) = \{\text{maps $G \to A$}\} \] (called ``cochains''). \end{fcdefnstar} \begin{fcdefnstar}[Cocycles] \glssymboldefn{Zone}% Define \[ Z^1(G, A) = \{(a_\sigma)_{\sigma \in G} \st a_{\sigma\tau} = \sigma(a_\tau) + a_\sigma ~\forall \sigma, \tau \in G\} \] (called ``cocycles''). \end{fcdefnstar} \begin{fcdefnstar}[Coboundaries] \glssymboldefn{Bone}% Define \[ B^1(G, A) = \{(\sigma b - b)_{\sigma \in G} \st b \in A\} \] (called ``coboundaries''). \end{fcdefnstar} \begin{note*} $C^1(G, A) \supset Z^1(G, A) \supset B^1(G, A)$. \end{note*} Then we can define: \begin{fcdefnstar}[$H^1$] \glssymboldefn{Hone}% Define \[ H^1(G, A) = \frac{Z^1(G, A)}{B^1(G, A)} .\] \end{fcdefnstar} \begin{remark*} If $G$ acts trivially on $A$, then \[ \Hone(G, A) = \Hom(G, A) .\] \end{remark*} \begin{fcthm}[] \label{thm:15.1} Assuming: - we have a short exact sequence of $G$-modules \[ 0 \to A \stackrel{\phi}{\to} B \stackrel{\psi}{\to} C \to 0 .\] Then: it gives rise to a long exact sequence of abelian groups: \[ 0 \to A^G \stackrel{\phi}{\to} B^G \stackrel{\psi}{\to} C^G \stackrel{\delta}{\to} \Hone(G, A) \stackrel{\phi^*}{\to} \Hone(G, B) \stackrel{\psi^*}{\to} \Hone(G, C) .\] \end{fcthm} \noproof \begin{proof} Omitted. \end{proof} \begin{fcdefnstar}[$delta$] Let $c \in C^G$. Then $b \in B$ such that $\psi(b) = c$. Then \[ \psi(\sigma b - b) = \sigma c - c = 0 \quad \forall \sigma \in G \] so $\sigma b - b = \phi(a_\sigma)$ for some $a_\sigma \in A$. Can check $(a_\sigma)_{\sigma \in G} \in \Zone(G, A)$. Then $\delta(c) =$ class of $(a_\sigma)_{\sigma \in G}$ in $\Hone(G, A)$. \end{fcdefnstar} \begin{fcthm}[] \label{thm:15.2} Assuming: - $A$ is a $G$-module - $H \normalsub G$ a normal subgroup Then: there is an inflation restriction exact sequence \[ 0 \to \Hone(G / H, A^H) \stackrel{\operatorname{inf}}{\to} \Hone(G, A) \stackrel{\operatorname{res}}{\to} \Hone(H, A) \] \end{fcthm} \noproof \begin{proof} Omitted. \end{proof} Let $K$ be a perfect field. Then $\Gal(\ol{K} / K)$ is a topological group with basis of open subgroups being the $\Gal(\ol{K} / L)$ for $[L : K] < \infty$. If $G = \Gal(\ol{K} / K)$ then we modify the definition of $\Hone(G, A)$ by insisting: \begin{enumerate}[(1)] \item The stabiliser of each $a \in A$ is an open subgroup of $G$. \item All \glspl{cochain} $G \to A$ are continuous, where $A$ is given the discrete topology. \end{enumerate} Then \[ \Hone(\Gal(\ol{K} / K), A) = \lim_{\substack{\longrightarrow \\ L \\ \text{$L / K$ finite Galois}}} \Hone(\Gal(L / K), A^{\Gal(\ol{K} / L)}) .\] (direct limit is with respect to inflation maps). \begin{fcthmstar}[Hilbert's Theorem 90] \glsnoundefn{hilb90}{Hilbert 90}{NA}% Assuming: - $L / K$ a finite Galois extension Then: \[ \Hone(\Gal(L / K), L^*) = 0 .\] \end{fcthmstar} \begin{proof} Let $G = \Gal(L / K)$. Let $(a_\sigma)_{\sigma \in G} \in \Zone(G, L^*)$. Distinct automorphisms are linearly independent, so there exists $y \in L$ such that \[ \ub{\sum_{\tau \in G} a_\tau^{-1} \tau(y)}_{= x} \neq 0 .\] Then \begin{align*} \sigma(x) &= \sum_{\tau \in G} \sigma(a_\tau)^{-1} \sigma \tau(y) \\ &= a_\sigma \ub{\sum_{\tau \in G} a_{\sigma \tau}^{-1} \sigma \tau(y)}_{=x} \end{align*} Then $a_\sigma = \frac{\sigma(x)}{x}$ for all $\sigma \in G$. ($a_\sigma \tau = \sigma(a_\tau) a_\sigma$ $\implies$ $\sigma(a_\tau)^{-1} = a_\sigma a_{\sigma \tau}^{-1}$). So $(a_\sigma)_{\sigma \in G} \in \Bone(G, L^*)$. Therefore $\Hone(G, L^*) = 0$. \end{proof} \begin{corollary*} $\Hone(\Gal(\ol{K} / K), \ol{K}^*) = 0$. \end{corollary*}