4
$\begingroup$

Hello,

Let $R$ be a commutative (not necessarily Noetherian) ring. Let $I$ be a small filtrant (not necessarily totally ordered) category. Let $(M_i)_{i\in I}$ be a projective system of $R$-modules with surjective transitions maps.

Assume that the natural morphism $(\lim_{i\in I} M_i)^{(X)}\rightarrow \lim_{i\in I}(M_i^{(X)})$ is an isomorphism for any set $X$ (where the superscript $(X)$ denotes the direct sum indexed by $X$).

Question: is it true that the projective system $(M_i)_{i\in I}$ is constant (i.e. isomorphic to its projective limit as a pro-object of $R$-modules)? In other words, does there exist $\overline i\in I$ such that for any $j\geq \overline i$ we have $M_j\cong \lim_{i\in I} M_i$?

Thanks for your hints,

Luisa Fiorot

$\endgroup$
3
  • $\begingroup$ By constant, do you mean that there should exist $i$ such that $\lim M_i = M_i$ ? $\endgroup$ Oct 3, 2011 at 14:31
  • $\begingroup$ I've inserted the interpretation you wrote as an answer. It would help if you registered an account, because that would prevent you from making new user-IDs. A new user-ID prevents you from editing your question. $\endgroup$
    – S. Carnahan
    Oct 4, 2011 at 6:03
  • $\begingroup$ Write $K_i$ for the kernel of the map $\lim M_i \to M_i$. The $K_i$ form a projective system with respect to inclusion. It seems you're asking the following : if $\lim^1 K_i^{(X)} = 0$ for any set $X$, then does $K_i$ satisfy the Mittag-Leffler condition ? $\endgroup$ Oct 4, 2011 at 11:44

1 Answer 1

4
$\begingroup$

No. There are nontrivial filtrant projective systems $(M_i)$ of $R$-modules with surjective transition maps such that $\varprojlim M_i=0$. Both $(\varprojlim M_i)^{(X)}$ and $\varprojlim(M_i^{(X)})$ may be identified with submodules of $(\varprojlim M_i)^X=0$, so the canonical map $(\varprojlim M_i)^{(X)}\to\varprojlim(M_i^{(X)})$ is an isomorphism.

For how to construct such examples, see G. Bergman, Some Empty Inverse Limits or G. Higman, A.H. Stone, On inverse systems with trivial limits. J. Lond. Math. Soc. 29, 233-236 (1954)

There is also Bourbaki, Topologie generale, III §7 Ex. 2, but the argument given there seems to be incomplete.

$\endgroup$

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.