For a smooth proper variety $X$ over discrete valuation ring $\mathcal{O}$ of mixed characteristic $(0,p)$, let $X_K$ be the generic fibre over a generic point $\textbf{Spec} K$ and let $X_k$ be the special fibre over a special point $\textbf{Spec} k$. We say that the variety $X_k$ has ordinary reduction iff the cohomology group $H^i(X_k,d\Omega^j)=0$ for any $i,j$.

My question is that p-adic etale cohomology $H^i_{et}(X,\mathbb{Z}_p)$ of $X_K$ knows $X_k$ has ordinary reduction or not.

  • 2
    $\begingroup$ What is the meaning of $d$ in that cohomology group $H^i(X_k, d\Omega^j)$? $\endgroup$
    – Will Sawin
    Apr 2, 2022 at 0:41
  • $\begingroup$ @WillSawin exact differentials of degree $j+1$, often denoted by $B^{j+1}$. $\endgroup$ Apr 2, 2022 at 12:22
  • $\begingroup$ @PiotrAchinger Is $B^{j+1}$ a Zariski sheaf? $\endgroup$
    – Z. M
    Apr 2, 2022 at 16:27
  • 1
    $\begingroup$ Good point, of course what is meant is the sheaf theoretic image of d, so locally exact differentials. $\endgroup$ Apr 2, 2022 at 16:36

1 Answer 1


I think you meant the etale cohomology of the geometric generic fiber, equipped with the action of the Galois group of $K$. Moreover, in $p$-adic Hodge theory it is usually necessary to assume that the residue field $k$ is perfect.

In this case, the answer is yes under the additional (customary) assumption that the crystalline and Hodge cohomology groups of $X$ are torsion free (this assumption is needed for the equivalence of two "standard" definitions of ordinarity: the one you gave and the one in terms of Hodge and Newton polygons). This essentially follows from the results of Bernadette Perrin-Riou "Representations p-adiques ordinaires" in Asterisque no. 223, 1994 "Periodes p-adiques" MR 1293973. See Proposition 6.7 in https://arxiv.org/abs/2005.02246 (I claim no originality here, but the statement gives what you want).

  • $\begingroup$ @Achinger Thank you very much!! I is exactly the thing I wanted. $\endgroup$
    – OOOOOO
    Apr 2, 2022 at 14:08

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.