Let $\mathcal{C}$ be a site on the category of schemes for some Grothendieck topology, $X$ a scheme, and let $F,G$ be two sheaves of abelian groups on $\mathcal{C}$. Do we have cup-product as follows? $$H^i(X,F)\otimes H^j(X,G)\rightarrow H^{i+j}(X,F\otimes_{\mathbb{Z}}G)$$ or even the relative version $$R^if_*F\otimes R^jf_*G\rightarrow R^{i+j}f_*(F\otimes_{\mathbb{Z}}G)$$ along a map $f:X\rightarrow Y$ of schemes.

For etale cohomology, it seems that Godement resolution is needed in the construction, see [Etale cohomology theory, the paragraph before Prop. 7,4,10] by Fu Lei, see also Appendix B of http://math.stanford.edu/~conrad/Weil2seminar/Notes/L12-13.pdf.

But in a general site (e.g. fppf site), we may not have Godement resolution. I checked the chapter "cohomology of sites" of Stack-Project, unfortunately, cup -product is in the list of topics that "should be discussed in this chapter, and have not yet been written"

It would be very helpful, if someone knows a reference.


Yes. This is treated in detail in Section 8.4 of Jardine's book “Local homotopy theory”. See also the introduction to Chapter 8 there for a historical comment on cup products and Godement resolutions.

  • $\begingroup$ The construction there is for simplicial sheaf. I guess I can consider a sheaf a a constant simplicial sheaf, right? $\endgroup$ – Heer Jun 12 at 13:11
  • $\begingroup$ Thanks for the reference. The content of the book is far out of my area, I hope more accessible (for me) answer could appear. $\endgroup$ – Heer Jun 12 at 13:13
  • 1
    $\begingroup$ @Heer: A cohomology class c∈H^i(X,F) is simply a derived map in the category of presheaves of chain complexes Z[X]→F[i]. This translates to a derived map of simplicial presheaves X→K(F,i). See ncatlab.org/nlab/show/abelian+sheaf+cohomology for a detailed description of the correspondence between simplicial presheaves and the older language of sheaf resolutions. $\endgroup$ – Dmitri Pavlov Jun 12 at 13:45
  • 1
    $\begingroup$ As Jardine says in the beginning of chapter 8, one approach to defining cup-products is via Verdier's hypercovering theorem. The Stacks project has a presentation of the hypercovering theorem which you might find useful (stacks.math.columbia.edu/tag/01FX). $\endgroup$ – Jesse Silliman Jun 12 at 15:40

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.