I am trying to understand the outline of the strategy for proving (via primary decomposition) that every non-zero ideal of a Dedekind domain can be expressed uniquely (up to the order of the factors) as a product of powers of distinct prime ideals.

I am using Atiyah & MacDonald and I have tried to outline their approach below. I would really appreciate it if someone would read it through and alert me of any slips in my understanding.


An integral domain (assumed not to be a field) is a Dedekind domain if

  1. it is Noetherian;

  2. it is integrally closed;

  3. every non-zero prime ideal is maximal (i.e. it is dimension one).

First, it is noted that properties (1) and (3) are enough to ensure a unique decomposition into a product of primary ideals with distinct radicals. Property (2) is then used to show that every primary ideal in a Dedekind domain is a power of a prime ideal (not true in general: $(X,Y^2)$ is $(X,Y)$-primary in $K[X,Y]$ however we have $(X,Y)^2\subsetneq(X,Y^2)\subsetneq (X,Y)$ and so it is not a power of $(X,Y)$.) This last part is broken into two steps:

  1. The localization $A_\mathfrak{p}$ of Dedekind domain $A$ at any non-zero prime ideal $\mathfrak{p}$ is a DVR;

  2. The only primary ideals of a DVR are powers of its unique non-zero prime ideal $\mathfrak{P}$.

Hence the $\mathfrak{p}$-primary ideals of $A$ correspond (under localization) to powers of $\mathfrak{P}=\mathfrak{p}^e$ in $A_\mathfrak{p}$ and so, as powers of ideals behave well under localization, any $\mathfrak{p}$-primary ideal of $A$ is just a power of $\mathfrak{p}$.


1 Answer 1


In my opinion some care is needed in the last step. If $q$ is a $p$-primary ideal, then $qA_p=p^eA_p$, but in general the contraction of $p^eA_p$ to $A$ is not $p^e$ although it is a $p$-primary ideal (namely, the smallest primary ideal containing $p^e$). However, in your case $p^e$ is $p$-primary (since $p$ is maximal) and this is enough to conclude $q=p^e$.

  • $\begingroup$ Thank you :) yes this comment is definitely useful. Other than this do you think that my outline is fairly accurate? Many thanks! $\endgroup$
    – user198313
    Dec 8, 2014 at 0:57
  • $\begingroup$ @user198313 One more thing: you proved that every ideal is a finite intersection of powers of primes, but you need to be a product. This follows recursively by using that two comaximal ideals $I$ and $J$ (comaximal meaning $I+J=A$) satisfy $I\cap J=IJ$. $\endgroup$
    – user26857
    Dec 8, 2014 at 1:04
  • $\begingroup$ I proved the product (which follows from the intersection as the primary ideals are pairwise comaximal - right?) $\endgroup$
    – user198313
    Dec 8, 2014 at 1:06
  • $\begingroup$ @user198313 I didn't notice in your question this detail. $\endgroup$
    – user26857
    Dec 8, 2014 at 1:07
  • $\begingroup$ No I suppose I state that I've proved it for products but I should probably explain how it follows from standard primary decomposition which, of course, concerns intersections. $\endgroup$
    – user198313
    Dec 8, 2014 at 1:08

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.