Let $R \subset S$ be an integral extension of domains and $\mathfrak p \subset R$ a prime ideal. Can it be the case that there are infinitely many distinct primes ${\cal P} \subset S$ such that ${\cal P} \cap R=\mathfrak p$?

Certainly this is impossible if $S$ is a Dedekind domain, because the primes lying over $\mathfrak p$ are the primes of $S$ occurring in the factorization of $\mathfrak p$ over $S$. I don't have much of an intuition for integral ring extensions that aren't number fields, so past this I'm not particularly sure.

  • 1
    $\begingroup$ Related: en.wikipedia.org/wiki/Quasi-finite_morphism $\endgroup$ – Martin Brandenburg Feb 6 '13 at 18:48
  • $\begingroup$ If $S$ is Noetherian, then there are only finitely many primes lying over a given prime. (This is claimed here and can be proved by using similar arguments of those given here.) $\endgroup$ – user89712 Oct 28 '13 at 17:01

I think the answer is yes.

Consider $\mathbb{Z} \subset S$, where $S$ is the ring of all the algebraic integers in $\mathbb{C}$. This is by definition integral, and for a fixed prime $p\in\mathbb Z$, one can find a sequence of Galois extensions $L_i/\mathbb{Q}$ such that $g_i(p) \to \infty$, where $g_i(p)$ is the number of primes of $L_i$ lying above $p$. (Using cyclotomic extensions suffice.) As each of these $g_i(p)$ primes can be further lifted up to distinct primes in $S$, this shows that there are infinitely many primes in $S$ lying above $p$.

Why cyclotomic extensions have arbitrarily large number of primes lying above a fixed $p$?

Consider $L = \mathbb{Q}(\zeta_n)$. Take $n$ that is not divisible by $p$, then $e(p) = 1$ in this extension. $f(p)$ is the order of Frobenius, which in this case is the smallest positive integer $f$ such that $n \mid p^f - 1$, and $g(p) = \phi(n)/f$. So everything would work if $n$ is large and $f$ is small. So take any $f$, and let $n = p^f - 1$, then $\frac{\phi(p^f-1)}{f} \to \infty$ as $f \to \infty$.

  • $\begingroup$ I was thinking about $\mathbb Z$ into the algebraic integers. Is there an easy reason, or do you have a reference for why Cyclotomic extensions work? So for instance we could replace the algebraic integers, by integral closure of $\mathbb Z$ in $\mathbb Q^{ab}$? (Also I was asking if it was possible for an infinite number of primes to lie over, so the answer is yes). $\endgroup$ – JSchlather Feb 6 '13 at 18:35
  • $\begingroup$ @JacobSchlather Oh, sorry for mixing up the yes's and no's. I will add some details about cyclotomic extensions part. $\endgroup$ – user27126 Feb 6 '13 at 18:36

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.