0
$\begingroup$

Consider the diagram

enter image description here

Where the functor $G$ sends a topological space to the category having as objects its points, and arrows homotopy classes of paths, $\varrho$ "realizes" geometrically an element of the simplex category $\Delta$ and $y$ is the Yoneda embedding.

I am interested in characterizing the nerve-realization adjoint pair induced by the composition $G\circ\varrho$, i.e. the adjoint pair $$ {\rm Lan}_{y}(G\varrho)\colon X\mapsto \int^{n\in\Delta}\widehat{\Delta}\big(\Delta[n],X\big)\cdot G\varrho([n])\quad\dashv\quad N_{g\varrho}\colon \mathbf C\mapsto {\bf Cat}(G\varrho(-),\bf C) $$ where I can simplify again ${\rm Lan}_y(G\varrho)$ by noticing that it is isomorphic to $\int^n X_n\cdot G(\Delta^n)$, where $\Delta^n$ is the standard topological simplex; it is extremely easy to notice that if $G$ has a right adjoint, say $\sigma$, then $N_{G\varrho}\cong \sigma^*N_\varrho$, where $N_\varrho$ is the classical "singular complex" functor (the nerve associated to $\varrho$).

Google does not give any clue about this problem, except for a reference to the adjunction between the "fundamental groupoid" (which is a different functor from my $G\varrho$) and the (restriction of the) classical nerve to the subcategory of groupoids, denoted $N_{\bf Gpd}$ in the following commutative diagram of adjoint pairs:

enter image description here

(if $(-)^\times\colon\bf Cat\to Gpd$ is the "maximal subgroupoid" functor, then the left adjoint to $N_{\bf Gpd}$ is obtained as the composition $X\mapsto (RX)^\times$, simply composing adjoints).

Has the functor $G$ an adjoint, in such a way that it respects realization (which is a colimit)? If it has not, can you still give me an explicit description for the pair nerve-realization associated to $G\varrho$? Can you tell how does this adjunction "descend to homotopy categories", if I put on both categories ($\bf Cat$ and $\widehat\Delta$) suitable model structures [better to say, if I put it on $\Delta$: if I remember well $\bf Cat$ is quite forced to have the structure (categorical equivalences, Grothendieck fibrations)]?

$\endgroup$
4
  • 1
    $\begingroup$ $\mathbf{Cat}$ supports two distinct non-trivial model structures, neither of which have Grothendieck fibrations as fibrations, so I very much doubt it is "forced" to have any particular one! $\endgroup$
    – Zhen Lin
    Sep 2, 2013 at 16:49
  • $\begingroup$ Hence I didn't remember well. :) $\endgroup$
    – fosco
    Sep 2, 2013 at 16:52
  • $\begingroup$ $G$ does not preserve colimits so it cannot have a right adjoint. However, it does preserve all colimits involved in gluing together simplices (this is essentially van Kampen). This means that your "realization" functor is just the composition of $G$ with the usual realization functor. $\endgroup$ Sep 2, 2013 at 17:03
  • $\begingroup$ "$G$ does not preserve colimits so it cannot have a right adjoint. However, it does preserve all colimits involved in gluing together simplices (this is essentially van Kampen)" Thank you. Can you make this more precise (finding a colimit which is not preserved is maybe easy, but how can one characterize a "VK"-like colimit?) $\endgroup$
    – fosco
    Sep 2, 2013 at 17:22

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.