Let $R$ be a commutative ring and let $I \subset R$ be an ideal. For any $n \ge 1$, the ring homomorphism $R \rightarrow R/I^n$ is surjective, hence an epimorphism in the category of rings. What about the natural map $R \rightarrow \hat{R_I}:=\lim_n R/I^n$ to the $I$-adic completion of $R$? This map is no longer surjective, but is it nevertheless an epimorphism?

If it is not an epimorphism in general, then I would also be interested in hearing about classes of rings for which it is an epimorphism. For example, does it hold for the p-adic integers?


(1) If $\mathbb{Z} \to \mathbb{Z}_p$ was an epimorphism, this would imply that $\mathbb{Q} \to \mathbb{Z}_p \otimes_{\mathbb{Z}} \mathbb{Q} = \mathbb{Q}_p$ is an epimorphism. But $\mathbb{Q}$ is a field and $\mathbb{Q} \to \mathbb{Q}_p$ is not surjective, so this is a contradiction.

(2) I think that $R \to \widehat{R}_I$ is almost never an epimorphism.

(3) See here for a classification of epimorphisms $\mathbb{Z} \to A$.

(4) There was a seminar about epimorphisms of commutative rings. For example, Prop. 1.5 in D. Lazard's "Epimorphismes plats" tells us that $f : A \to B$ is an epimorphism of commutative rings if and only if the following conditions are satisfied:

  • $\mathrm{Spec}(B) \to \mathrm{Spec}(A), \mathfrak{p} \mapsto f^{-1}(\mathfrak{p})$ is injective
  • For every prime ideal $\mathfrak{p}$ of $B$, the natural map $Q(A/f^{-1}(\mathfrak{p})) \to Q(B/\mathfrak{p})$ is an isomorphism.
  • The kernel $J = \ker(B \otimes_A B \xrightarrow{*} B)$ is a finitely generated $B \otimes_A B$-module with $J = J^2$.

The second condition fails for $\mathbb{Z} \to \mathbb{Z}_p$ if we take $\mathfrak{p}=0$.


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.