3
$\begingroup$

Let $K \subset L=K(\alpha)$ be a number field extension with rings of integers $\mathcal{O}_K$ and $\mathcal{O}_L=\mathcal{O}_K[\alpha]$ respectively. Let $\pi$ be a prime ideal in $O_K$, and let $F = \mathcal O_K/\pi$.

Finally, let $p$ be the minimum monic polynomial of $\alpha$ over $\mathcal O_K$, and let $p(x) = \prod_i(p_i(x))^{f_i}$ be the factorization of $p$ in $F[x]$.

Then we can show that $$\mathcal{O}_L/(\pi \mathcal{O}_L) = \prod_i (\mathcal{O}_L/(\pi,p_i(\alpha)))^{f_i} .$$ I have seen multiple proofs of this theorem but none of them show that $$\underbrace{\pi \mathcal{O}_{L_\strut}}_{P} = \underbrace{\prod_i (\pi,p_i(\alpha))^{f_i}}_{Q}.$$

It is easy to show that $Q \subset P$ by simply multiplying everything out but can we show that $P \subset Q$ (or even that $P = Q$) directly? By this I mean mostly avoiding the use of the norm function(size of the ideals) to show that $Q|P \Rightarrow P = Q$

$\endgroup$
6
  • $\begingroup$ I think this is Theorem 8 (p. 6) of these notes by Keith Conrad. $\endgroup$ Commented Jul 21, 2015 at 3:27
  • 1
    $\begingroup$ Another good source would be Neukirch's Algebraic Number Theory, p.47-48 (image link). $\endgroup$ Commented Jul 21, 2015 at 3:32
  • $\begingroup$ Neukirch uses that $P|Q$ and that they have the same norm to show they are equal. I want to show directly without the use of the norm. @ZevChonoles $\endgroup$
    – Asvin
    Commented Jul 21, 2015 at 4:44
  • $\begingroup$ @SpamIAm I don't think conrad shows it directly. I haven't read it properly and might be wrong but I think he uses the fact that the norm is equal too. $\endgroup$
    – Asvin
    Commented Jul 21, 2015 at 4:46
  • 1
    $\begingroup$ I had a vague recollection that Stichtenoth does this in the function field case, so I checked my copy. He first proves that the inertia degrees are always $\ge \deg p_i$ and that there is equality, if either A) the factors of $p(x)$ are all simple, or B) the monomials basis is an integral basis. Having that addendum triples the length of the proof. Also, in practice it may be difficult to find monomial integral bases (IIRC they don't always exist). These may explain the relative "unpopularity". $\endgroup$ Commented Jul 21, 2015 at 10:43

2 Answers 2

2
$\begingroup$

I don't know if this is what you mean by norms (you mean the absolute norm?). If this is what you already knew I apologize in advance!

$$\begin{aligned}\mathcal{O}_L/\pi\mathcal{O}_L &\cong \mathcal{O}_K[T]/(p(T))/(\pi\mathcal{O}_K[T]/(p(T))\\ &\cong \mathcal{O}_K[T]/(p(t),\pi)\\ &\cong \left(\mathcal{O}_K/\pi\right)[T]/(p(T))\\ &= \prod_i \left(\mathcal{O}_K/\pi\right)[T]/(p_i(T)^{f_i})\end{aligned}$$

But, just by counting dimensions:

$$\#\left(\mathcal{O}_K/\pi)\right)[T]/(p_i(T)^{f_i})=\left(\#(\mathcal{O}_K/\pi)[T]/(p_i(T))\right)^{f_i}$$

So, $\mathcal{O}_L/P$ and $\mathcal{O}_L/Q$ have the same cardinality. So, containment implies equality.

$\endgroup$
5
  • $\begingroup$ Alex, I think the question was about the prime decomposition of the ideal $\pi\mathcal{O}_L$ (see the second displayed formula in the question). Not about this isomorphism of rings. $\endgroup$ Commented Jul 21, 2015 at 10:29
  • $\begingroup$ @JyrkiLahtonen He has containment of ideals in one direction, and I just showed that their quotients are isomorphic (so have the same cardinality) which implies that the ideals are actually equal. $\endgroup$ Commented Jul 21, 2015 at 10:40
  • $\begingroup$ Sorry. I missed the boat. $\endgroup$ Commented Jul 21, 2015 at 10:46
  • $\begingroup$ @JyrkiLahtonen No problem :) I fear that this is precisely the answer the OP didn't want though. So, we'll see. $\endgroup$ Commented Jul 21, 2015 at 10:47
  • $\begingroup$ I am sorry, your proof is indeed what I wanted to avoid. I should have been clearer. $\endgroup$
    – Asvin
    Commented Jul 21, 2015 at 14:21
1
$\begingroup$

There is a slight generalization of what you call Dedekind's theorem, where one only assumes that p does not divide the index of O_K [a] in O_L, p denoting the prime of Z lying under Pi). See Daniel A. Marcus, "Number Fields", Springer Universitex , chapter 3, theorem 27. The proof is clear and easy to follow, and - as far as I remember - doesn't use norms.

$\endgroup$
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .