4
$\begingroup$

$\DeclareMathOperator\CM{CM}\DeclareMathOperator\Spec{Spec}$Let $k$ be an algebraically closed field and let $G$ be an algebraic group over $k$.

My question: is $G$ Cohen–Macaulay? If not, are there counterexamples?

I ask this because it seems to me that the answer is yes. For every finite type algebra $A$ over a field, the CM locus is open dense in $\Spec A$. In particular, $\CM(A)$ is nonempty because otherwise $\Spec A=\emptyset$. As a nonempty open, $\CM(G)$ contains a closed point $y$ because it is locally Noetherian. So if there is a point $x\in G\backslash{\CM(G)}$, then we can translate $x$ to $y$ induced by $G(k)\rightarrow G(k)$, $g\mapsto yx^{-1}g$. Then $x\in \CM(G)$.

$\endgroup$
6
  • 2
    $\begingroup$ In char. 0 they're smooth (hence CM). $\endgroup$
    – YCor
    Commented Oct 6, 2022 at 18:40
  • 15
    $\begingroup$ In arbitrary characteristic they are locally complete intersection, hence Cohen-Macaulay. This is in SGA3, Exposé VII$_B$, Cor. 5.5.1. $\endgroup$ Commented Oct 6, 2022 at 18:44
  • $\begingroup$ Thank you very much Matthieu Romagny ! $\endgroup$ Commented Oct 7, 2022 at 6:14
  • 1
    $\begingroup$ You're welcome. $\endgroup$ Commented Oct 7, 2022 at 7:03
  • 1
    $\begingroup$ @MatthieuRomagny Why don't you post your comment as an answer? $\endgroup$
    – Leo Alonso
    Commented Oct 7, 2022 at 11:27

2 Answers 2

10
$\begingroup$

In arbitrary characteristic they are locally complete intersection, hence Cohen-Macaulay: this is in SGA3, Exposé VII$_B$, Cor. 5.5.1 or Demazure-Gabriel, Groupes algébriques, chap. III, §3, n°6. This is due to the structure theorem which says that (after extension to an algebraic closure of the base field $k$) the completed local ring of the unit element is of the form $k[[t_1,\dots,t_n]](X_1,\dots,X_r)/(X_1^{p^{n_1},\dots,X_r^{p^{n_r})$, a local complete intersection.

$\endgroup$
8
$\begingroup$

If all you want is Cohen-Macaulay, this is straightforward, and your sketch proves the result. Every locally finite type scheme over a field has a maximal open subscheme that is Cohen-Macaulay (equivalent to local equidimensionality plus flatness of one / any quasifinite dominant morphism to a smooth scheme). By homogeneity, this open subscheme equals the entire group scheme.

$\endgroup$
2
  • 2
    $\begingroup$ The whole point is that this CM locus is nonempty (this point is mentioned by OP). $\endgroup$
    – YCor
    Commented Oct 7, 2022 at 15:28
  • 2
    $\begingroup$ Every Artinian ring is Cohen-Macaulay. Thus, each generic point of the group scheme is in the Cohen-Macaulay locus. $\endgroup$ Commented Oct 7, 2022 at 23:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .