9
$\begingroup$

In every degree $d$, the Galois closure of the typical number field has the maximal possible Galois group $S_d$. Denote by $f(d)$ the least absolute value of a discriminant of an $S_d$-field of degree $d$. My question is what is provably known about the asymptotic behavior of this function $f(d)$, and in particular, whether or not it is known to be exponentially bounded in the degree $d$.

Regarding heuristics (I would be interested in good ones, too), let me only mention that the global function fields model, when appropriately formulated, does have an exponentially growing $f(d)$, but that this is probably a poor guide in this type of question. (For instance, since the function field model admits a fully explicit and monogenic construction, which is conjectured to not exist in the case of number fields: the root discriminant of an integer irreducible polynomial is widely believed to approach infinity as the degree grows.)

As everyone knows, the Golod-Shafarevich towers give plenty of examples of number fields with an exponentially bounded discriminant, but those by their construction have solvable Galois groups, and are irrelevant in my question.

Added example. It is easily seen that $f(d) < d^d$, but this crude bound, I suppose, would be far from the truth. To see this bound, recall that Selmer proved that the trinomial $t^d - t - 1$ is irreducible with maximal Galois group $S_d$, and of discriminant $\pm( d^d - (1-d)^{d-1})$, which gives the bound, at least for $d$ odd (and the slightly weaker bound $f(d) < 2d^d$ in general).

$\endgroup$
6
  • 1
    $\begingroup$ Not sure this is in the direction you want, but I think Bhargava-Shankar-Wang's paper "Squarefree values of polynomial discriminants I", arxiv.org/abs/1611.09806 proves a lower bound for the number of $S_d$ number fields of discriminant $<X$ that they conjecture to be essentially sharp. (And they discuss various related results.) $\endgroup$ Commented Feb 13, 2019 at 8:37
  • $\begingroup$ @TimDokchitser: Thank you, in these results though they assume that the degree $d$ is fixed, and don't bother with writing down implied constants depending on $d$. Presumably the growing degrees and small discriminate case is not as well understood. On the other hand, the question would be solved if there existed an $\mathrm{Spec} \, O_{K,S}$ whose etale fundamental group (group of the maximal unramified outside $S$ extension of $K$) is complicated enough to contain an $S_d$-quotient for each $d = 1, 2, \ldots$. $\endgroup$ Commented Feb 13, 2019 at 10:43
  • $\begingroup$ @VesselinDimitrov Why is this enough? There exist extensions of $\mathbb{Q}_p$ with unbounded root discriminant. $\endgroup$
    – Aurel
    Commented Feb 13, 2019 at 16:34
  • $\begingroup$ @Aurel: Sorry, due to the word count limitation of the comment window I omitted to state that $S$ is a finite set of places of some number field $K$, and that $O_{K,S}$ is here the ring of $S$-integers of $K$ (not a localization). You are right that my notation was ambiguous at best. $\endgroup$ Commented Feb 13, 2019 at 18:19
  • $\begingroup$ @VesselinDimitrov That was my understanding of your notation, but I don't see why a family of $S_d$-extensions of $K$ unramified outside a fixed $S$ should have bounded root discriminant. $\endgroup$
    – Aurel
    Commented Feb 13, 2019 at 18:46

0

You must log in to answer this question.