5

In https://kerodon.net/tag/00PK, Lurie introduces the dg-nerve functor from differential graded categories to simplicial sets as a tool to translate statements/ constructions from the dg-context to the context of $(\infty,1)$-categories, as it actually even takes values in those. But as is suggested in https://ncatlab.org/nlab/show/nerve+and+realization, most constructions of nerves and realizations tend to follow a general scheme, i.e. nerves usually have left adjoint realization functors that are defined as Kan extensions of, in this case, a functor I will call $DG: \Delta \rightarrow \operatorname{dg-Cat}$, so that the dg-nerve should be given in components by $N^{dg}_n(\mathcal{C})= \operatorname{Fun}(DG([n]),\mathcal{C})$, should this principle also apply to the nerve of dg-categories, as the name and the general philosophy would suggest.

However, I was neither able to find literature on some kind of dg-realization, nor was I able to construct a representing object $DG([n])$ compatible with the construction used by Lurie myself (I however suspect that this dg-category should have exactly n objects). That is a shame, as it would make a lot of things much easier. Has anybody read something about such a construction or maybe has an idea on how to do it, or does at least know a simple argument why there can't be one? Greetings from Heidelberg,

Intergalakti

Markus Zetto
  • 1,046
  • 6
  • 17
  • Most of your questions seem to be answered in the first couple of sections of this paper. The answers are yes, at least if you allow $A_\infty$-categories rather than just dg, or else use the “big” DG-nerve. https://arxiv.org/pdf/1312.2127.pdf – Kevin Carlson Apr 13 '20 at 03:39
  • @KevinCarlson Yes, that is very helpful! I was aware of the existence of that paper, but I hadn't really read it yet. I'll try to figure out if something similar also works for the usual DG-nerve or without the $A_\infty$-categories using these constructions, but it seems like it doesn't, for the author would have probably included it... – Markus Zetto Apr 13 '20 at 08:52
  • 1
    There is a construction of a representing $DG([n])$ given in section 6 of Rivera and Zeinalian's "Cubical rigidification, the cobar construction, and the based loop space". They use a factorization of the rigidification functor through cubical categories. Here's the link to the arxiv https://arxiv.org/abs/1612.04801 – Aydin Ozbek May 14 '21 at 18:14
  • @AydinOzbek Thanks a lot for your comment, I probably would never have found that reference! The construction via $A_\infty$-categories had seemed very elegant to me so that I doubted that something like this would even exist, but apparently it does - further, the construction actually seems really interesting. Definitely something I will keep in mind! – Markus Zetto May 15 '21 at 12:06

2 Answers2

2

To close this question off, I'll just say a few words about the approach in arxiv.org/pdf/1312.2127.pdf, the paper Kevin Carlson pointed out in the comments, so anyone stumbling upon this question gets a rough overview.

Indeed, there is a way to define the dg-nerve via such a construction, in fact there even are two. If one regards the formula that Lurie gives, it seems pretty difficult to give a construction that accounts for all of it, as there is next to the sum term also a composition and a boundary term. The reason for those seems to be that they are part of the infinite amount of higher compositions induced by the $A_\infty$-operad, and if one extends the category of dg-categories to the bigger realm of $A_\infty$-categories, there is a fairly natural construction of a generalization of the dg-nerve, going along the lines I described above.

Additionally, there is some kind of bigger dg-nerve, obtained by truncating the morphism complexes, applying Dold-Kan and then the homotopy-coherent nerve (this is also described in Luries "Higher Algebra"). Although this looks a mit more elegant at first glance, since one does not need $A_\infty$-algebras, the result is much migger, although equivalent to the first construction; I am a bit surprised that a truncation occurs as this way one loses some information, I thought the functor was fully faithful?

Anyway, especially the first approach is very helpful and explains the ad hoc formula that Lurie gives, from this point of view the theory of $A_\infty$-algebras seems like a very natural generalization of the dg-theory.

Markus Zetto
  • 1,046
  • 6
  • 17
  • It seems to me that the dg-nerve is naturally enriched in spectra (https://ncatlab.org/nlab/show/dg-nerve). When you troncate, you should just forget the enrichment. – Adrien MORIN Apr 25 '20 at 08:48
1

In case it is useful to others, a construction of a left adjoint to the dg nerve is also given in Rivera and Zeinalian, The colimit of an ∞-local system as a twisted tensor product, Higher Structures 4(1):33–56, 2020.

Tim Porter
  • 1,081