3
$\begingroup$

When reading Lectures on Analytic Geometry, I found that in the data of an analytic ring, the underlying ring seems unnatural and the module category should be the soul, but in fact one needs the underlying ring to state the condition on the module category. After taking a look at higher algebra, it seems that I have found a way to define analytic ring without the underlying ring; namely, Propositions 7.1.2.6 and 7.1.2.7 of Higher Algebra. So I conjecture that:

For $1\le k\le\infty$, $\mathcal{A}\mapsto\mathsf{D}(\mathcal{A})$ determines an equivalence from the $\infty$-category of analytic $\mathbb{E}_k$-rings to the $\infty$-category of $\mathbb{E}_{k-1}$-monoidal $\mathsf{CondAni}$-enriched stable $\infty$-categories which are tensored over $\mathsf{CondAni}$, whose tensor product preserves colimit and the $\mathsf{CondAni}$-tensoring structure, and whose unit object $\mathbf{1}$ is a compact projective generator in the enriched sense: the enriched homomorphism $\underline{\operatorname{Hom}}(\mathbf{1},-)$ preserves sifted colimits, and every object is a colimit of some $S\otimes\mathbf{1}$'s where $S\in\mathsf{CondAni}$. In the latter category, the morphisms are defined as the $\mathsf{CondAni}$-tensored left adjoints.

If nonconnective analytic rings are reasonable, then I also conjecture a nonconnective version, removing the projectivity in the right side.

I have not yet studied higher algebra carefully, so I cannot verify these reconstruction propositions. Are they correct?

$\endgroup$
2
  • 1
    $\begingroup$ Dustin has thought more about this, but I do expect that there is some characterization like this. You definitely need to allow shifts of the generators to generate everything under colimits. Also, regarding "projectivity", the external Hom from $S\otimes 1$'s ($S$ extremally disconnected) only commutes with sifted colimits on the connective part (this can be repaired, if desired, by taking spectra-valued Hom's instead of anima-valued Hom's, but then the "projectivity" aspect is lost). The connective part can be characterized as the part generated under colimits by $S\otimes 1$'s. $\endgroup$ Commented Aug 27, 2021 at 9:51
  • $\begingroup$ @PeterScholze What kind of modules would you expect to work for a version of "noncommutative" analytic geometry? Here it is about all modules, but should one consider a full subcategory such as that spanned by nuclear modules? I mean, an analogue of the standard setup of "smooth proper DG-categories". And also, are $S\otimes1$ considered "connective" even when the "underlying ring" is non-connective? This seems to the case when the underlying ring arises from "a coaffine stack" in Toën-Vezzosi, but I don't know the full generality. $\endgroup$
    – Z. M
    Commented Aug 27, 2021 at 16:24

0

You must log in to answer this question.