I am trying to understand a proof of the following statement:

Let $R$ be a Dedekind ring and let $I$ be an ideal of $R$. $I$ is contained in only a finite number of prime ideals $\mathfrak{p}_1,...,\mathfrak{p}_n$ of $R$ and $I=\mathfrak{p}_1^{a_1}\cdots \mathfrak{p}_n^{a_n}$ for some $a_1,\cdots,a_n \in\mathbb{N}^+$.

The proof, which was taken from Janusz's Algebraic Number Fields and modified by me, goes as follows:

Consider the ring $S=R/I$. Note that since $R$ is of Krull dimension 1, $S$ is a commutative ring in which every prime ideal is maximal. Therefore every ideal of $S$ contains a product of prime ideals. The prime ideals of $S$ correspond bijectively with the prime ideals of $R$ which contain $I$. Note that there can be only finitely many prime ideals of $S$ and hence only finitely many prime ideals $\mathfrak{p}_1,\cdots,\mathfrak{p}_n$ of $R$ which contain $I$. In particular, there exists $a_1,\cdots, a_n \in \mathbb{N}$ such that $({\mathfrak{p}_1/I})^{a_1}\cdots ({\mathfrak{p}_n/I})^{a_n}=0$. This implies $\mathfrak{p}_1^{a_1}\cdots \mathfrak{p}_n^{a_n}\subseteq I$.

Let $\widetilde{R}=R/\mathfrak{p}_1^{a_1}\cdots \mathfrak{p}_n^{a_n}$. Since the ideals $\mathfrak{p}_i$ and $\mathfrak{p}_j$ are maximal for all $i$ and $j$, they are comaximal for all $i\neq j$. Thus by the Chinese remainder theorem we have $$ \widetilde{R}\cong R/{\mathfrak{p}_1}^{a_1}\times \cdots \times R/{\mathfrak{p}_n}^{a_n}. $$ The ideals of $\widetilde{R}$ are of the form $I_1\times \cdots \times I_n$, where $I_k$ is an ideal of $R/{\mathfrak{p}_k}^{a_k}$. Since the ideals of $R/\mathfrak{p}_k^{a_k}$ are all powers of $\mathfrak{p}_k/\mathfrak{p}_k^{a_k}$, the ideals of $\widetilde{R}$ are the image of some product $\mathfrak{p}_1^{b_1}\cdots \mathfrak{p}_n^{b_n}$ for some $b_i \leq a_i$. In particular, this means $I$ has the same image as some $\mathfrak{p}_1^{c_1}\cdots \mathfrak{p}_n^{c_n}$, where $c_i \leq a_i$ for all $i$. Since both $I$ and $\mathfrak{p}_1^{c_1}\cdots \mathfrak{p}_n^{c_n}$ contain $\mathfrak{p}_1^{a_1}\cdots \mathfrak{p}_n^{a_n}$ and map to the same element ideal of $\widetilde{R}$, we must have $I=\mathfrak{p}_1^{c_1}\cdots \mathfrak{p}_n^{c_n}$. $\square$

The part which I do not understand is the final step:

Since both $I$ and $\mathfrak{p}_1^{c_1}\cdots \mathfrak{p}_n^{c_n}$ contain $\mathfrak{p}_1^{a_1}\cdots \mathfrak{p}_n^{a_n}$ and map to the same element ideal of $\widetilde{R}$, we must have $I=\mathfrak{p}_1^{c_1}\cdots \mathfrak{p}_n^{c_n}$.

I'm sure the reason is staring me right in the face, but I cannot see it.


What is the reasoning behind the final step?

  • $\begingroup$ I have to say that the proof given in Janusz (at least from what I see above in your post) is shorter than either what I've seen in Neukirch of Marcus' Number Fields. $\endgroup$ – user38268 Jan 2 '13 at 12:37
  • $\begingroup$ @benjalim well, he builds it up with a series of lemmas which are used implicitly in the proof. So his exposition is a bit longer than this. $\endgroup$ – Holdsworth88 Jan 2 '13 at 12:44
  • $\begingroup$ Sorry, but who are referring to as "he" in your comment above? Are you referring to Janusz, Marcus or Neukirch? $\endgroup$ – user38268 Jan 2 '13 at 12:49

I believe this comes from the correspondence theorem: If you have two ideals $P$ and $Q$ of $R$ that contain $J$ such that under the canonical homomorphism $\pi : R \to R/J$ we have $\pi(P) = \pi(Q)$ then in fact $P = Q$.

In your case above we have $P = I$, $Q = \mathfrak{p}_1^{c_1}\ldots\mathfrak{p}_n^{c_n}$ and $J = \mathfrak{p}_1^{a_1}\ldots \mathfrak{p}_n^{a_n}$.

A lecturer I had once phrased this to me as follows: "The two ideals are equal upstairs iff they are equal downstairs."


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.