2
$\begingroup$
$ \newcommand{\Ext}{\mathop{\rm Ext}\nolimits} \newcommand{\depth}{\mathop{\rm depth}\nolimits} \newcommand{\dim}{\mathop{\rm dim}\nolimits} $

A module $M$ is a maximal Cohen-Macaulay module over the ring $R$ if $\depth M = \dim R$, where $\dim R$ denotes the Krull dimension of $R$. I've been told that if $R$ is a Gorenstein local ring, then $M$ is a maximal Cohen-Macaulay module if and only if $\Ext^i(M,R)=0$ for all $i \geq 1$. (This is taken as the definition here: https://mathoverflow.net/questions/214626/showing-that-the-stable-module-category-of-a-ring-r-restricted-to-maximal-cohe)

How can this be proved? And does anyone have a source where this has been proved? Is it true for Gorenstein local rings only, or also for Gorenstein (not necessarily local) rings too?

I primarily work with commutative rings and finitely generated modules.

$\endgroup$
1
$\begingroup$

Old question, but here is an answer anyway.

For the first, part, namely the case of a commutative local Cohen-Macaulay ring, one can use local duality: if $R$ is a Cohen-Macaulay local ring admitting a canonical module $\omega$ then for any finitely generated $R$-module there are isomorphisms $$\text{Ext}_{R}^{i}(M,\omega)^{\wedge}\simeq \text{Hom}_{R}(H_{\mathfrak{m}}^{d-i}(M),E(k)).$$ In particular $$ \begin{align} M \text{ is MCM} &\iff H_{\mathfrak{m}}^{d-i}(M)=0 \text{ for all }i>0 \\ &\iff \text{Hom}_{R}(H_{\mathfrak{m}}^{d-i}(M),E(k))=0 \text{ for all }i>0 \text{ (as $E(k)$ is an injective cogenerator}) \\ &\iff \text{Ext}_{R}^{i}(M,\omega)^{\wedge}=0 \text{ for all $i>0$ (by local duality)} \\ &\iff \text{Ext}_{R}^{i}(M,\omega)=0 \text{ for all $i>0$} \end{align} $$ where the last if and only if follows from the fact that $\text{Ext}_{R}^{i}(M,\omega)$ is a finitely generated $R$-module and the map $R\rightarrow \hat{R}$ is faithfully flat.

Since $R$ is Gorenstein if and only if $R\simeq \omega$, and the definition given on the Mathoverflow page is equivalent.

For the more general case, one does not have maximal Cohen-Macaulay modules, so a suitable replacement class is needed. Over any ring, one can define Gorenstein projective modules:

Definition Let $R$ be a ring. An $R$-module $M$ is Gorenstein projective if there is an exact complex $\mathbf{P}^{\bullet}$ of projective $R$-modules such that $M= \text{Ker}(\mathbf{P}^{0}\rightarrow \mathbf{P}^{1})$ and $\text{Hom}_{R}(\mathbf{P}^{\bullet},P)$ is exact for any projective $R$-module $P$

While this definition is not the most accessible, this class of modules has good properties over certain classes of ring:

Lemma Let $R$ be a Cohen-Macaulay local ring. Then every finitely generated Gorenstein projective module is MCM. If $R$ is a Gorenstein local ring, then every MCM is a finitely generated Gorenstein projective module.

Over a not-necessarily commutative Gorenstein ring much more can be said; here is the relevant part for finitely generated modules

Lemma Let $R$ be a Gorenstein ring. Then a finitely generated module $M$ is Gorenstein projective if and only if $M$ is reflexive with respect to $(-)^{*}=\text{Hom}_{R}(-,R)$ and $$\text{Ext}_{R}^{i}(M,R)=\text{Ext}_{R}^{i}(M^{*},R)=0$$ for all $i>0$.

Lemma 4.2.2 of Maximal Cohen-Macaulay Modules and Tate-Cohomology Over Gorenstein Rings, Buchweitz shows that over any Gorenstein ring, a finitely generated maximal Cohen-Macaulay module (using the definition given on mathoverflow) is Gorenstein projective, and clearly any Gorenstein projective module is maximal Cohen-Macaulay.

Therefore the definition given on mathoverflow (due to Buchweitz) is a good one and is equivalent to the usual definition over commutative local Gorenstein rings.

The lemmas I quoted are from section 10.2 of Relative Homological Algebra by Enochs and Jenda. It is a good introduction to Gorenstein modules and their homological algebra.

| cite | improve this answer | |
$\endgroup$

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.