3
$\begingroup$

Let $R$ be a Noetherian domain of dimension $\ge 1$. Let $\mathfrak{p}_i$, $i = 1, 2, ...$ be prime ideals of height one. Let $T = R[[X]]$ with $X$ is a indeterminate. For each $i \ge 1$ we set $\mathfrak{q}_i = \mathfrak{p}_i[[X]] = \mathfrak{p}_iT$ the extension of $\mathfrak{p}_i$. It is clearly that the height of $\mathfrak{q}_i$ is one for all $i$. Set $S = T \setminus \cup_{i \ge 1}\mathfrak{q}_i$.

Question: Is it true that $\dim T_S = 1$?

$\endgroup$
4
$\begingroup$

If you can use the Countable Prime Avoidance Lemma (Edit: that is, if you are in situations in which the Countable Prime Avoidance Lemma applies), I think the answer is yes. Because any prime ideal of $T_S$ comes from a prime ideal of $T$ that is contained in the union of the $\mathfrak{q}_i$'s. But then the Countable Prime Avoidance Lemma would imply that this prime ideal must be contained in one of the $\mathfrak{q}_i$'s. Hence $\dim T_S$ cannot be bigger than $1$.

The Countable Prime Avoidance Lemma will work, if for instance $R$ contains an uncountable number of elements $\{\mu_\lambda\}_\lambda$ such that $\mu_\lambda-\mu_\gamma$ is a unit in $R$ for every $\lambda\neq\gamma$. This condition will hold, if for example $R$ is local with $R/\mathfrak{m}$ uncountable.

A reference for this version of the Countable Prime Avoidance Lemma is page 242 of the book Cohen-Macaulay Representations by Graham Leuschke and Roger Wiegand, but I am sure you can find it at other places, too.

$\endgroup$
  • $\begingroup$ Another reference for countable prime avoidance is Baire's category theorem and prime avoidance in complete local rings by Sharp and Vamos (Arch. Math. 44 (1985), 243-248), as mentioned in mathoverflow.net/questions/102546. $\endgroup$ – Fred Rohrer Feb 26 '14 at 20:23
  • $\begingroup$ I know Countable Prime Avoidance Lemma. But I do not know how to apply it for my question. Could you give a detail answer. $\endgroup$ – Pham Hung Quy Feb 27 '14 at 7:27
  • $\begingroup$ Recall that an element of $T$ (a formal power series) is a unit if and only if its constant term (degree zero term) is a unit. Now if $R$ has an uncountable number of elements $\{\mu_\lambda\}$ such that $\mu_\lambda-\mu_\gamma$ is a unit in $R$ for $\lambda\neq\gamma$, then you can produce an uncountable number of elements of $T$ with the same property, just consider formal power series whose constant term (degree zero term) equals one of the $\mu_\lambda$'s. Fred Roher has kindly provided more details below. $\endgroup$ – Mahdi Majidi-Zolbanin Feb 27 '14 at 13:05
  • $\begingroup$ I only known Countable Prime Avoidance Lemma for complete local ring before. Thanks you very much for the new form. But the ring in my question do not have this assumption. $\endgroup$ – Pham Hung Quy Feb 27 '14 at 15:55
3
$\begingroup$

I will prove that $T_S$ satisfy Countable Prime Advoidance Lemma in the form mentioned in Mandi's answer (See the book of Graham Leuschke and Roger Wiegand)

  1. First note that an element of $T = R[[X]]$ is a unit if its constant term is a unit in $R$. After localization such element is also a unit in $T_S$.

  2. Since $X \notin \mathfrak{q}_i = \mathfrak{p}_i[[X]]$ we have $X$ contained in the multiplicative closed set $S$. Therefore $X$ is a unit in $T_S$.

  3. Consider the following set $$B = \{u_{\mu} = b_0 + b_1X + \cdots + b_iX^i + \cdots | b_i = 0 \text{ or } 1, \text{ and } u_{\mu} \neq 0\}.$$ It is clearly that $B$ in a uncountable set of units in $T_S$.

  4. Moreover, if $u$ and $v$ in $B$ we have $u - v$ (edit: $u-v$ nay not in $B$ but it is a unit) is a unit. Thus the assumption of Countable Prime Advoidance Lemma is satisfying.

The following application is a generalization of my question.

Corollary. Let $\{p_i\}_{i \ge 1}$ is a countable set of prime ideals with not containment relation. Let $T = R[[X]]$ and $S = T \setminus \cup_{i\ge 1}\mathfrak{p}_iT$. Then $R \to T_S$ is a flat extension and $$\text{Max}(T_S) = \{\mathfrak{p}_iT_S\}_{i \ge 1}.$$

$\endgroup$
  • $\begingroup$ Can you explain why $u-v$ is a unit? What if both $u$ and $v$ start with $1$? Then $u-v$ will start with $0$ and is not a unit.(?) $\endgroup$ – Mahdi Majidi-Zolbanin Feb 28 '14 at 14:12
  • $\begingroup$ Example $u = 1 + X$ and $v = 1 + X^2$ we have $u - v = X - X^2 = X(1 -X)$ is unit in $T_S$ since both $X$ and $1-X$ are units in $T_S$. $\endgroup$ – Pham Hung Quy Feb 28 '14 at 14:40
  • $\begingroup$ I see! I forgot $X$ was a unit in $T_S$. Very interesting. $\endgroup$ – Mahdi Majidi-Zolbanin Mar 1 '14 at 2:03
0
$\begingroup$

Let me give a detailed account of Mahdi's argument.

First, we have to suppose (which is not totally clear in the question), that we consider only countably many primes $\mathfrak{p}_i$ of height $1$.

Second, we will suppose that $R/\mathfrak{m}$ is uncountable for every maximal ideal $\mathfrak{m}$ of $R$.

Then, $\dim(S^{-1}T)=1$.

Indeed, replacing $R$ by its localisation at a maximal ideal we can suppose that $R$ is in addition local. Then, in order to show that $\dim(S^{-1}T)=1$ we will show that every non-zero prime of $S^{-1}T$ has height $1$. So, let $\mathfrak{p}$ be a non-zero prime of $S^{-1}T$. Then, there exists a non-zero prime $\mathfrak{p}'$ of $T$ with $S\cap\mathfrak{p}'=\emptyset$ such that $\mathfrak{p}=S^{-1}\mathfrak{p}'$, and moreover $\mathfrak{p}$ and $\mathfrak{p}'$ have the same height. Thus it suffices to show that $\mathfrak{p}'$ has height $1$. The condition $S\cap\mathfrak{p}'=\emptyset$ is equivalent to $\mathfrak{p}'\subseteq\bigcup_{i\in\mathbb{N}}\mathfrak{q}_i$. Since $R$ is local with uncountable residue field, the same holds for $T$. Hence, countable prime avoidance holds in $T$, and so we see that there exists $i\in\mathbb{N}$ such that $\mathfrak{p}'\subseteq\mathfrak{q}_i$. As $\mathfrak{p}'$ is non-zero, $\mathfrak{q}_i$ has height $1$ and $T$ is a domain, it follows that $\mathfrak{p}'=\mathfrak{q}_i$ has height $1$ as desired.

(Using another variant of countable prime avoidance, also proven by Sharp and Vamos, gives the same conclusion under the hypothesis that $R$ is a complete local ring.)

$\endgroup$
  • $\begingroup$ Thanks you Fred. But in my question we do not assume that $R/\mathfrak{m}$ is uncountable. $\endgroup$ – Pham Hung Quy Feb 27 '14 at 15:24
  • $\begingroup$ Em Quý, have you tried to use Example 2.4 in the article by Sharp and Vámos that I mentioned above to construct a counterexample? $\endgroup$ – Fred Rohrer Feb 27 '14 at 18:33
  • $\begingroup$ Dear Fred, I do not have that paper. Could you please sent it for me. I think I can prove it. $\endgroup$ – Pham Hung Quy Feb 28 '14 at 3:52

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.