1
$\begingroup$

Let $(R,\mathfrak{m})$ be a complete local ring, $a_{\lambda}$ be a decreasing net of ideals in $R$, indexed by a directed set. Consider the completion under $a_{\lambda}$-topology $A=\underleftarrow{\lim} R/\mathfrak{a}_{\lambda}$. Is $A$ still complete under the $\mathfrak{m}$-topology?

$\endgroup$
3
  • $\begingroup$ Since inverse limits commute, I believe this is true. $\endgroup$
    – the L
    Nov 30, 2011 at 13:53
  • $\begingroup$ @Liran: This is a bit too easy - you forgot about the quotient rings. $\endgroup$ Nov 30, 2011 at 20:11
  • $\begingroup$ Moreover (following Liran Shaul's comment), if we define $I=\cap_\lambda a_\lambda$, then isn't $A$ equal to $R/I$, so $\frak m$-complete? We can assume $I=0$, and then any sequence $(r_n)$ in $R$ that is Cauchy in the $a_\lambda$-topology is also Cauchy in the $\frak m$-topology, so has a limit in $R$. $\endgroup$
    – inkspot
    Nov 30, 2011 at 20:21

1 Answer 1

5
$\begingroup$

I assume that $(R, \frak{m} )$ is a complete Noetherian local ring. Set $\frak{a} = \bigcap_\lambda \frak{a}_\lambda$. By passing to $R/\frak{a}$ we may assume that the $\frak{a}_\lambda$-topology is separated. Now, we use a Theorem of Chevalley (1946) which says that in the complete Noetherian local ring the $\frak{m}$-adic topology is weaker than every separated topology. Then every Cauchy sequence in the $\frak{a}_\lambda$-topology is also a Cauchy sequence in the $\frak{m}$-adic topology. Hence $R = \underleftarrow{\lim} R/\frak{a}_\lambda$ is complete.

$\endgroup$
2
  • $\begingroup$ Can you please provide a reference for this Theorem of Chevalley? $\endgroup$
    – the L
    Jan 8, 2012 at 13:09
  • $\begingroup$ Lemma 7 - Chevalley: On theory of local ring, Annals of Math. 1946, 690-708. Theorem 2.1 - P. Schenzel, On use of local cohomology in algebra and geometry, in: six lecture of commutative algebre, 1998 (you can seacher it on internet) $\endgroup$ Jan 9, 2012 at 12:16

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.