%! TEX root = EC.tex % vim: tw=50 % 10/02/2025 11AM Let $0 \neq \omega \in \Omega_C$. Let $P \in C$ be a smooth point, and $t \in K(C)$ a uniformiser at $P$. Then $\omega = f \dd t$ for some $f \in K(C)^*$. We define $\ord_p(\omega) = \ord_p(f)$ (independent of choice of $t$). We assume $C$ is a smooth projective curve. \begin{definition*}[] $\div(\omega) = \sum_{P \in C} \ord_P(\omega) P \in \Div(C)$. \end{definition*} \begin{note*} This is a divisor, i.e. $\ord_p(\omega) = 0$ for all but finitely many $P \in C$. \end{note*} \begin{fcdefnstar}[Regular differential] A differential $\omega \in \Omega_C$ is \emph{regular} if $\div(\omega) \ge 0$, i.e. it has no poles. \end{fcdefnstar} \[ g(C) = \dim_K \{\omega \in \Omega_C : \div(\omega) \ge 0\} .\] As a consequence of Riemann-Roch we have: \begin{quote} If $0 \neg \omega \in \Omega_C$, then $\deg(\div \omega) = 2g - 2$. \end{quote} \textbf{Fact:} Suppose $f \in K(C)^*$, $\ord_p(f) = n \neq 0$. If $\char K \nmid n$, then $\ord_p(\dd f) = n - 1$. \begin{fclemma}[] \label{lemma:6.1} Assuming: - $\char K \neq 2$ - $E$: $y^2 = (x - e_1)(x - e_2)(x - e_3)$, $e_1, e_2, e_3 \in K$ distinct Then: $\omega = \frac{\dd x}{y}$ is a differential on $E$ with no zeroes or poles $\implies g(E) = 1$. In particular, the $K$-vector space of regular differentials on $E$ is 1-dimensional, spanned by $\omega$. \end{fclemma} \begin{proof} Let $T_i = (e_i, 0)$, $E\tors[2] = \{0, T_1, T_2, T_3\}$. \[ \div(y) = (T_1) + (T_2) + (T_3) - 3(0) \tag{1} \label{lec8eq1} .\] For $P \in E \setminus \{0\}$, \[ \div(x - x_p) = (P) + (-P) - 2(0) .\] If $P \in E \setminus E\tors[2]$, then $\ord_p(x - x_p) = 1 \implies \ord_p(\dd x) = 0$. If $P \in E \setminus E\tors[2]$ then $\ord_p(x - x_p) = 1$ $\implies \ord_p(\dd x) = 0$. If $P = T_i$ then $\ord_p(x - x_p) = 2$ $\implies \ord_p(\dd x) = 1$. If $P = 0$ then $\ord_p(x) = -2$ $\implies \ord_p(\dd x) = -3$. Therefore \[ \div(\dd x) = (T_1) + (T_2) + (T_3) - 3(0) \tag{2} \label{lec8eq2} .\] \eqref{lec8eq1} and \eqref{lec8eq2} implies $\div \left( \frac{\dd x}{y} \right) = 0$. \end{proof} \begin{definition*}[] For $\phi : C_1 \to C_2$ a non-constant morphism we define \begin{align*} \phi^* : \Omega_{C_2} &\to \Omega_{C_1} \\ f \dd g &\mapsto (\phi^* f) \dd (\phi^* g) \end{align*} \end{definition*} \begin{fclemma}[] \label{lemma:6.2} Assuming: - $P \in E$, \begin{align*} \tau_P : E &\to E \\ X &\mapsto P + X \end{align*} - $\omega = \frac{\dd x}{y}$ as above Then: $\tau_P^* \omega = \omega$ (we say $\omega$ is the \emph{invariant differential}). \end{fclemma} \begin{proof} $\tau_P^* \omega$ is a regular differential on $E$. So $\tau_P^* \omega = \lambda_p \omega$. The map $E \to \Pbb^1$, $P \mapsto \lambda_P$ is a morphism of smooth projective curves but \emph{not} surjective (misses $0$ and $\infty$). Therefore it is constant (by \cref{thm:2.8}), i.e. there exists $\lambda \in K^*$ such that $\tau_P^* \omega = \lambda \omega$ for all $P \in E$. Taking $P = 0_E$ shows $\lambda = 1$. \end{proof} \begin{remark*} If $K = \Cbb$, $\Cbb / \Lambda \stackrel{\sim}{\to} E(\Cbb)$, $z \mapsto (\wp(z), \wp'(z))$, \[ \frac{\dd x}{y} = \frac{\wp'(z) \dd z}{\wp'(z)} = \dd z .\] (invariant under $z \mapsto z + (\text{const})$). \end{remark*} \begin{fclemma}[] \label{lemma:6.3} Assuming: - $\phi, \psi \in \Hom(E_1, E_2)$ - $\omega$ an invariant differential on $E_2$ Then: $(\phi + \psi)^* \omega = \phi^* \omega + \psi^* \omega$. \end{fclemma} \begin{proof} Write $E = E_2$ \begin{align*} E \times E &\to E \\ \mu : (P, Q) &\mapsto P + Q \\ \mathrm{pr}_1 : (P, Q) &\mapsto P \\ \mathrm{pr}_2 : (P, Q) &\mapsto Q \end{align*} \textbf{Fact:} $\Omega_{E \times E}$ is a 2-dimensional $K(E \times E)$-vector space with basis $\mathrm{pr}_1^* \omega$ and $\mathrm{pr}_2^* \omega$. Therefore \[ \mu^* \omega = f \mathrm{pr}_1^* \omega + g \mathrm{pr}_2^* \omega \tag{1} \label{lec8eq3} \] for some $f, g \in K(E \times E)$. For fixed $Q \in E$, let \begin{align*} \iota_Q : E &\to E \times E \\ P &\mapsto (P, Q) \end{align*} Applying $\iota_Q^*$ to \eqref{lec8eq3} gives \begin{align*} (\mu \iota_Q)^* \omega &= (\iota_Q^* f)(\mathrm{pr}_1 \iota_Q)^* \omega + (\iota_Q^* g)(\mathrm{pr}_2 \iota_Q)^* \omega \\ \ub{\tau_Q^* \omega}_{= \omega} &= \ub{(\iota_Q^* f)}_{\text{by \cref{lemma:6.2}}} \omega + 0 \end{align*} Therefore $\iota_Q^* f = 1$ for all $Q \in E$, so $f(P, Q) = 1$ for all $P, Q \in E$. Similarly $g(P, Q) = 1$ for all $P, Q \in E$. \[ \eqref{lec8eq3} \implies \mu^* \omega = \mathrm{pr}_1^* \omega + \mathrm{pr}_2^* \omega .\] Now pull back by \begin{align*} E_1 &\to E \times E \\ P &\mapsto (\phi(P), \psi(P)) \end{align*} to get \[ (\phi + \psi)^* \omega = \phi^* \omega + \psi^* \omega . \qedhere \] \end{proof} \begin{fclemma}[] \label{lemma:6.4} Assuming: - $\phi : C_1 \to C_2$ a non-constant morphism Then: \begin{iffc} \lhs $\phi$ is separable \rhs $\phi^* : \Omega_{C_1} \to \Omega_{C_2}$ is non-zero \end{iffc} \end{fclemma} \noproof \begin{proof} Omitted. \end{proof} \begin{example*} $\mathbb{G}_m = \Abb^1 \setminus \{0\}$ (multiplicative group). \begin{align*} \phi : \mathbb{G}_m &\to \mathbb{G}_m \\ x &\mapsto x^n \end{align*} ($n \ge 2$ integer). $\phi^*(\dd x) = \dd (x^n) = nx^{n - 1} \dd x$. So if $\char K \nmid n$ then $\phi$ is separable. \cref{thm:2.8} implies $\#\phi^{-1}(Q) = \deg \phi$ for all but finitely many $Q \in \mathbb{G}_m$. But $\phi$ is a group homomorphism, so \[ \#\phi^{-1}(Q) = \#\ker(Q) \] for all $Q \in \mathbb{G}_m$. Thus $\#\ker \phi = \deg \phi = n$ and hence $K (= \ol{K})$ contains exactly $n$ $n$-th roots of unity. \end{example*} \begin{fcthm}[] \label{thm:6.5} Assuming: - $\char K \nmid n$ Then: $E\tors[n] \cong (\Zbb / n\Zbb)^2$. \end{fcthm}