Let $(R/A)_\Delta$ be the prismatic site over $R$ relative to a prism $(A, I)$, then it is known that $(R/A)_\Delta$ admits finite non-empty coproduct, for instance, by Cor. 5.2 in Bhatt's lecture notes V on prismatic cohomology.

My question is, if we consider the absolute prismatic site $R_\Delta$, does $R_\Delta$ still always have finite non-empty coproduct?

If we restrict to the existence of finite non-empty self coproducts, it seems to me that the answer is yes, as it has been suggested (implicitly) in work:

Arthur-César Le Bras and Johannes Anschütz: Prismatic Dieudonné theory.

Or an upcoming work of Bhatt and Scholze, see the notes of Scholze on this topic in the RAMpAGe Seminar.

A more concrete question is the following:

Let $R=O_K$ be a complete discrete valuation ring of mixed characteristic with perfect residue field $k$, and let $(\mathfrak{S},(E))$ be a Breuil-Kisin prism in $R_\Delta$, i.e., $\mathfrak S=W(k)[\![u]\!]$ and $E \in W(k)[u]$ is the Eisenstein polynomial of a uniformizer of $O_K$. In the above two works, they claim the self product of $(\mathfrak{S},(E))$ in $R_\Delta$ exists and is equal to a prismatic envelop of $\mathfrak{S}\otimes_{W(k)}\mathfrak{S}$. Why is the tensor product over $W(k)$? I know for $(R/A)_\Delta$, the similar construction takes tensors over $A$, but it is clear that $(\mathfrak{S},(E))$ is not a prism over $(W(k),(p))$.


1 Answer 1


Yes, it does admit nonempty finite coproducts. If you have two prisms $(A_1,I_1)$ and $(A_2,I_2)$ with maps $R\to A_i/I_i$, you need to find the initial prism $(A,I)$ with maps from both $(A_i,I_i)$ such that the two induced maps $R\to A_i/I_i\to A/I$ agree. For this, start with $A_0=A_1\hat{\otimes}_{\mathbb Z_p} A_2$ (where the tensor product is $(p,I_1,I_2)$-adically completed, say) and then take a suitable prismatic envelope to ensure the required conditions.

You raise an interesting point, that if $R=\mathcal O_K$, then one can take the tensor product over $W(k)$ instead of $\mathbb Z_p$, where $k$ is the residue field of $\mathcal O_K$. The reason is that for any prism $(A,I)$ with a map $\mathcal O_K\to A/I$, you get in particular a map $W(k)\to A/I$, and this lifts uniquely to a map $W(k)\to A$: This is using that the $p$-completed cotangent complex of $W(k)/\mathbb Z_p$ vanishes, or some more concrete assertion involving Teichmüller lifts. The map $W(k)\to A$ is also necessarily a map of $\delta$-rings. So all those $A$'s live canonically over $W(k)$, so one can as well directly take the tensor product there.

  • $\begingroup$ Dear Peter, thank you very much for the answer! And for the first part, when you say take a suitable prismatic envelope, I guess you mean that we should first regard $A_0$ (together with a certain ideal) as a $\delta$-pair over $(A_1, I_1)$ (or $(A_2, I_2)$), and then take the prismatic envelope relative to it, right? Or do we know that the "prismatic envelope" also exists in this absolute setting? $\endgroup$
    – Mayday
    Commented Jan 28, 2021 at 1:18

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.