I am sure this is well known, but I don't know what to search for:

Consider $M_{1,n}$, the moduli space of genus 1 curves with $n$ marked points. The symmetric group on $n$ letters acts on this space by acting on the marked points, and we can quotient out by this action, which corresponds to forgetting the ordering of the n points.

What is known about this space? For instance, for what $n$ (and in what characteristics) is it a rational variety? Does this space have a name? In short, how can I find literature on this?

  • 1
    $\begingroup$ Do you want the marked points to be distinct? I think the terminology "marked points" usually implies distinct points, while "degree n divisors" in the title usually would not have this restriction. $\endgroup$ – Harry Richman Nov 30 '20 at 8:29
  • $\begingroup$ I do want them to be distinct and the divisors to be squarefree. Sorry for the confusion. $\endgroup$ – Asvin Nov 30 '20 at 15:59
  • $\begingroup$ I want to say that this is related to the universal Jacobian $\mathscr{Jac}^n \to M_1$, but that should be $2$-dimensional while your space should be $(n+1)$-dimensional, so I've clearly missed something. Nevertheless, the phrase "universal Jacobian" could be useful. $\endgroup$ – Tabes Bridges Nov 30 '20 at 16:33

It doesn't have a more standard name than "$M_{1,n}/S_n$".

Belorousski's PhD thesis contains very explicit rational parametrizations of $M_{1,n}$ for $n<11$. I would expect (but haven't checked) that a modification of his constructions would work to show rationality of $M_{1,n}/S_n$, too.

The irrationality of $M_{1,11}$ can be seen from the nonvanishing Hodge number $H^0(\overline M_{1,11},\Omega^{11})$, which comes from modular forms; it is part of the two-dimensional "motive" associated to the Ramanujan cusp form $\Delta$. This motive transforms according to the sign representation of $S_n$ and in particular it dies in the cohomology of $M_{1,n}/S_n$. So I do not see an immediate cohomological obstruction for rationality. The same is true more generally that in the middle cohomology of $\overline M_{1,n}$ one finds the motive of cusp forms for $\mathrm{SL}(2,\mathbf Z)$ of weight $n+1$ and it transforms according to the sign representation. In fact according to a paper of Consani and Faber this is the only part of the rational cohomology of $\overline M_{1,n}$ transforming according to the sign representation of $S_n$.

The Kodaira dimension of $M_{1,n}$ is $0$ when $n = 11$ and $1$ for $n \geq 12$ by a paper of Bini-Fontanari. So this is at least an upper bound for the Kodaira dimension of $M_{1,n}/S_n$.

I'm not sure which (if any) parts of the above change in positive characteristic, but you could look at a paper of Will Sawin on the non-uniruledness of $\overline M_{1,n}$ in characteristic $p$.

  • $\begingroup$ Somewhere I saw the notation $\mathcal{M}_{1, \, [n]}$ $\endgroup$ – Francesco Polizzi Nov 30 '20 at 9:39
  • 1
    $\begingroup$ What do you mean by "the quotient of a rational variety by a reflection group is rational"? The cubic threefold is a quotient of a rational variety by $\Bbb{Z}/2$. $\endgroup$ – abx Nov 30 '20 at 10:07
  • 1
    $\begingroup$ @abx You're right, sorry! $\endgroup$ – Dan Petersen Nov 30 '20 at 10:41
  • $\begingroup$ Thanks, this is very helpful. Do you think it's possible that $M_{1,n}/S_n$ is always just $\mathbb P^n$ or do you see any easy obstruction to this? (I have some computations that seem to indicate this...) $\endgroup$ – Asvin Dec 3 '20 at 4:56
  • $\begingroup$ If you mean that $\overline M_{1,n}/S_n \cong \mathbb P^n$ that's certainly false (e.g. the Picard group of $\overline M_{1,2}/S_2$ has rank two). If you mean that $M_{1,n}/S_n$ is always a rational variety I highly doubt it but I don't have an argument. $\endgroup$ – Dan Petersen Dec 4 '20 at 9:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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