2
$\begingroup$

Let $D$ be a PID and let $\mathrm{Frac}(D)$ be its field of fractions. I want to classify the intermediate rings $D\subseteq R\subseteq \mathrm{Frac}(D)$.

Theorem: Every such ring $R$ is a localization $D[S^{-1}]$ for some subsemigroup $S\subseteq (R,\times,1)$.

Proof: Given $D\subseteq R\subseteq\mathrm{Frac}(D)$ consider the set of denominators $$S:=\left\{ b\in D: \frac{a}{b}\in R \text{ and } \mathrm{gcd}(a,b)=1\right\}.$$ If $b\in S$ then there exists $\frac{a}{b}\in R$ with $\mathrm{gcd}(a,b)=1$ by definition. Since $D$ is a PID there exist $x,y\in D$ such that $1=ax+by$. Then dividing by $b$ gives $$\frac{1}{b}=\frac{a}{b} x + y.$$ Since $R$ is a ring this implies that $\frac{1}{b}\in R$. Now we can see that $S$ is a subsemigroup of $(D,\times,1)$. Indeed, we have $1\in S$, and if $b_1,b_2\in S$ then by the above remarks we have $\frac{1}{b_1},\frac{1}{b_2}\in R$, hence $\frac{1}{b_1b_2}\in R$. It follows that $b_1b_2\in S$. Thus the localization $D[S^{-1}]$ is well-defined.

I claim that $R=D[S^{-1}]$. Indeed, we have $R\subseteq D[S^{-1}]$ by definition. Conversely, consider any $\frac{a}{b}\in D[S^{-1}]$, i.e., with $b\in S$. By the above remarks this implies $\frac{1}{b}\in R$ and hence $\frac{a}{b}=a\frac{1}{b}\in R$. $\square$

Thus we obtain a surjective map from the subsemigroups of $(D,\times,1)$ to the rings between $D$ and $\mathrm{Frac}(D)$. This map is not injective because it is possible to have $D[S^{-1}]=D[T^{-1}]$ for different subsemigroups $S,T$. However, given $S$, there is a largest such subsemigroup $T$ defined by $$T=\left\{a\in D : ab\in S \text{ for some } b\in D\right\}.$$ Such sets $T$ are characterized by the condition that for all $a,b\in D$ we have $$ab\in T \Longleftrightarrow a\in T \text{ and } b\in T.$$ Thus, since $D$ is a UFD, these sets (and therefore the intermediate rings $D\subseteq R\subseteq \mathrm{Frac}(D)$) are in bijection with sets of prime elements.

Finally, my question is this:

Is there a name for a set $S\subseteq D$ such that $1\in S$ and for all $a,b\in D$ we have $$ab\in S \Longleftrightarrow a\in S \text{ and } b\in S\,\,?$$

I have been unable to find any references. Thanks.

$\endgroup$
5
  • 6
    $\begingroup$ Isn't such a set just a "saturated multiplicative set"?: e.g., see definition 3.15 of web.mit.edu/18.705/www/13Ed.pdf $\endgroup$ Feb 5, 2014 at 19:19
  • $\begingroup$ @SamHopkins: Thanks. That's good enough for me. $\endgroup$ Feb 5, 2014 at 19:54
  • 2
    $\begingroup$ In semigroup theory, a subset of a monoid is called unitary if it satisfies your property. I think the origin is that the units in a commutative ring have this property. $\endgroup$ Feb 5, 2014 at 20:24
  • 8
    $\begingroup$ Furthermore, saturated multiplicative sets are intersections of complements of prime ideals. In a PID, this amounts to giving yourself the set of prime elements which you did not invert. $\endgroup$
    – ACL
    Feb 5, 2014 at 22:49
  • $\begingroup$ The title doesn't really match the question. I guess all the information is in the comments, but it would be useful to clarify what is expected (since a cw answer could be posted). $\endgroup$
    – YCor
    Nov 11, 2019 at 14:25

2 Answers 2

2
$\begingroup$

The question is elementary and was already answered in the comments. I'm posting a cw answer so that the question can be ticked as answer (otherwise it remains regularly bumped).

Let $I_D$ be the set of irreducible elements, $I_D/D^\times$ its quotient modulo multiplication by invertible elements, and $\mathcal{P}(X)$ the power set of $X$.

Then the map $S\mapsto D[S^{-1}]$, from $\mathcal{P}(I_D)$ to the set $\mathcal{L}_D$ of subrings of $\mathrm{Frac}(D)$ containing $D$, is surjective, and induces a bijection from $\mathcal{P}(I_D/D^\times)$ to $\mathcal{L}_D$.

The surjectivity follows from the fact that if $R\in\mathcal{L}_D$ and $a/b\in R$ with $a,b$ coprime, and $s$ is irreducible and divides $b$, then $1/s\in R$. This is an easy consequence of Bézout.

The injectivity is easy: clearly the set of elements of $D$ that is invertible in $R$ is invariant by multiplication by $D^\times$, and if $S$ is a set of irreducible elements in $D$, then the set of elements in $D$ invertible in $D[S^{-1}]$ is exactly $SD^\times$.

(Such a description does not pass to the case of UFDs: in the polynomial algebra $K[x,y]$, $1/x\notin K[x,y,y/x]$; more generally it fails in an arbitrary non-principal UFD.)

Next, for $D$ an arbitrary domain, one can consider the set of localizations of $D$, that is, the subset $\mathcal{L}'_D$ of $\mathcal{D}$ consisting of those $D[S^{-1}]$ with $S\subset D^\times$.

Such an $S$ can always be chosen to be "saturated" in the sense that it is a submonoid of $D^\times$ satisfying $ab\in S$ implies $a,b\in S$. (About "unitary" I find it hard to find a worse choice of terminology, since "unitary" is widely used in the sense "contains 1".)

The above description of saturated submonoids of $D^\times$, namely as indexed by subsets of $I_D/D^\times$, holds for arbitrary UFDs. It's actually purely a property of the monoid $(D,\times)$, as is being a UFD ($D$ is a UFD iff $(D,\times)$ has an absorbing element $0$, the complement $D\smallsetminus\{0\}$ is a submonoid, and $D\smallsetminus\{0\}$ is direct product of a group with a free monoid.)

$\endgroup$
0
$\begingroup$

you can find such a classification in the following link: https://www.researchgate.net/publication/283199027_Subrings_of_Q

Ali Jaballah Subrings of Q, Journal of Science and Technology Vol 2(No 2):1-13, 1997.

Abstract: In this paper we determined all the subrings of Q, the field of rational numbers. We established a one to one correspondence between the set of subrings of Q containing Z, the ring of integers, and the set of subsets of prime numbers. We also studied the rings S between Z and Q that satisfy finiteness conditions on the set of prime ideals or on the set of rings between Z and or between S and Q, and we established equations relating the cardinality of these sets. Finally we have generalized some of these results to principal ideal domains.

$\endgroup$
1
  • 3
    $\begingroup$ Can you eaborate on the contents of your link? Answers should be self-contained. Thanks. $\endgroup$ Feb 14, 2019 at 11:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.