Computing cohomology groups of elliptic curves

853 Views Asked by At

I'm skimming through Silverman's text to recall some theory of elliptic curves that I've learned in undergrad. In practice however, I'm having trouble actually computing the cohomology groups. For instance, I know intuitively that $H^1(\mathbb{F}_q,E)$ will vanish, but how does one show this? Analogously, how does one compute $H^1(\mathbb{R},E)$?

I know that for a field $k$ and $n\in\mathbb{Z}$, the corresponding map $n:E(\bar k)\to E(\bar k)$ is surjective. This gives an exact sequence \begin{equation} 0\to E(\bar k)[n]\to E(\bar k)\to^n E(\bar k)\to 0, \end{equation} which further gives the exact sequence \begin{equation} 0\to E(k)[n]\to E(k)\to^n E(k)\to H^1(k,E(\bar k)[n])\\ \hspace{5em}\to H^1(k,E(\bar k))\to^n H^1(k,E(\bar k)). \end{equation} Hence from the long exact sequence we have \begin{equation} 0\to E(k)/nE(k)\to H^1(k,E(\bar k)[n])\to H^1(k,E(\bar k))[n]\to 0. \end{equation} How can one use this to compute cohomology over a finite field or real numbers?

1

There are 1 best solutions below

2
On BEST ANSWER

Both calculations you ask about are done in this lecture note of mine on WC-groups: see pages 5-7. Mariano's remark is used in the former computation; it is also what I have in mind in "Exercise 6", although if you keep reading you'll see the computation redone several more times in fancier ways.

(A proof of Theorem 5 of the above note appears in $\S 3$ of this eternally not quite finished paper of mine.)

$\newcommand{\F}{\mathbb{F}}$ Added: Another slick proof that $H^1(\F_q,E) = 0$ is to note that a nonzero element would yield a genus one curve $C_{/\mathbb{F}_q}$ without a $\F_q$-rational point. But Weil proved that for any curve $C_{/\mathbb{F}_q}$ of genus $g$, $|\# C(\F_q) - (q+1)| \leq 2g \sqrt{q}$. When $g = 1$ this implies $\# C(\F_q) > 0$(!!).