I hope this question is good to be here.

Let $J$ be the Jacobian of a hyperelliptic curve $H$ of genus $2$ given by $y^2 = x^5 + h$.

I was calculating an explicit formula in Mumford coordinates of $[\sqrt{5}]\in \text{End}(J)$. When I say Mumford coordinates I mean divisors $D:=(x_1+x_2, x_1x_2, \tfrac{y_1-y_2}{x_1-x_2},\tfrac{x_2y_1-x_1y_2}{x_1-x_2})=:(A,B,C,D)\in J$ where $\text{Supp}(D)=\{(x_1,y_1), (x_2,y_2)\}$ and $(x_i,y_i)\in H$.

I used these coordinates in order to work with MAGMA.

The thing is, I got the formulae, the problem is that I do not know how to handle an special situation.

What I did is to calculate $[\zeta_5+\zeta_5^4] \in \text{Aut}(J)$ where the action of $\zeta_5^i$ on the curve $H$ is given by $\zeta_5^i(x,y)=(\zeta_5^i x,y)$ and then extend to the whole Jacobian.

After this, I calculated the formula for $2[\zeta_5+\zeta_5^4]+1=[\sqrt{5}]$ using Mumford coordinates and then if $D=(A,B,C,D)$ then I got $\sigma_i\in k(J)$ where $[\sqrt{5}](D)=(\sigma_1(A,B,C),\sigma_2(A,B,C), \sigma_3(A,B,C), \sigma_4(A,B,C))\in J$

Also I have formulae when the input divisor is of the form $[(x,y)-\infty]\in \Theta\subset J$

My problem is when the input is not in $\Theta$ (an ordinary, generic Divisor) but the output is in $\Theta$.

My question is, if $D:=(A,B,C,D)$ how can I compute if $[\sqrt{5}](D)\in \Theta$, and more... how can I compute the resulting divisor?

Detecting it I think is easy, I just need to calculate a determinant of a $4\times 4$ matrix in terms of $A,B,C$. But I do not know how to compute the divisor using my coefficients $(A,B,C,D)$. Is there a way to get explicit formulae for this "exceptional case" when is already known that the image will be in $\Theta$?



1 Answer 1


Here is a trick that might help: Compute $\sqrt{5}D$ as $\sqrt{5}(D+R)-\sqrt{5}R$ where $R$ is in the large set you know how to compute.

  • $\begingroup$ Thanks for this suggestion @Watson, but the problem if I understood correctly your suggestion, is the same, since the resulting divisors in the substraction will be fine, but calculating the difference, even if I fix $R$ wont give me information about $\sqrt{5}D$ since its support may be different from the support of $R$ $\endgroup$ Aug 28, 2017 at 9:36
  • $\begingroup$ So there is an explicit algorithm for doing computations in the Jacobian of a hyperelliptic curve which does normalize the divisors. $\endgroup$ Aug 28, 2017 at 16:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.