3
$\begingroup$

The group $\mathbb{Z}/2$ corepresents the functor $\mathrm{Inv}\colon\mathsf{Mon}\to\mathsf{Sets}$ sending a monoid $A$ to its set of involutory elements (those satisfying $a^2=1_A$).

A similar story is true for $\mathbb{Z}$ and invertible elements, but let's instead tell it in the $\infty$-setting: namely, the $\infty$-category of $\mathbb{E}_1$-monoidal functors $\mathbb{Z}_\mathsf{disc}\to\mathcal{C}$ is just $\mathsf{Pic}(\mathcal{C})$, and thus $\mathbb{Z}_\mathsf{disc}$ corepresents the functor $$\mathsf{Pic}\colon\mathsf{Mon}_{\mathbb{E}_1}(\mathsf{Cats}_{\infty})\to\mathcal{S}$$

However, replacing

  • $\mathsf{Mon}_{\mathbb{E}_1}(\mathsf{Cats}_\infty)$ by $\mathsf{Mon}_{\mathbb{E}_\infty}(\mathsf{Cats}_\infty)$, the $\infty$-category of symmetric monoidal $\infty$-categories;
  • $\mathcal{S}$ by $\mathsf{Grp}_{\mathbb{E}_\infty}(\mathcal{S})$;

changes the corepresenting object from $\mathbb{Z}_{\mathsf{disc}}$ to the sphere spectrum $\mathbb{S}$. Similarly, if we pass to $\mathbb{E}_k$ rather than $\mathbb{E}_{\infty}$, we get $\Omega^kS^k$ instead of $\mathbb{S}$.


Now, define an involutory object of a monoidal $\infty$-category $\mathcal{C}$ to be a strong monoidal functor $(\mathbb{Z}/2)_{\mathsf{disc}}\to\mathcal{C}$. By definition, $(\mathbb{Z}/2)_{\mathsf{disc}}$ corepresents the functor $$\mathsf{Inv}\colon\mathsf{Mon}_{\mathbb{E}_1}(\mathsf{Cats}_{\infty})\to\mathcal{S}$$ sending $\mathcal{C}$ to $\mathsf{Inv}(\mathcal{C})\overset{\mathrm{def}}{=}\mathsf{Fun}^\otimes((\mathbb{Z}/2)_{\mathsf{disc}},\mathcal{C})$.

Question. Is the functor $$\mathsf{Inv}\colon\mathsf{Mon}_{\mathbb{E}_{k}}(\mathsf{Cats}_\infty)\to\mathsf{Grp}_{\mathbb{E}_{k-1}}(\mathcal{S})$$ corepresentable by an $\mathbb{E}_{k}$-monoidal category for $2\leq k\leq\infty$?

$\endgroup$
3
  • 1
    $\begingroup$ I think if $C$ is $E_k$-monoidal and $X$ is $E_1$-monoidal, then $Fun^\otimes(X,C)$ is $E_{k-1}$-monoidal, not $E_1$-monoidal (think of $k=1$: what is a monoidal structure on $Fun^\otimes(C,D)$ for $C,D$ barely monoidal ? on $Alg(D)$ ? ); the case of $X=\mathbb Z$ is special, because $Fun^\otimes(\mathbb Z,C)\to C$ is the inclusion of a full sub-groupoid which is stable under tensor products. Your question still makes sense though, if you replace the second $E_k$ with an $E_{k-1}$ $\endgroup$ Sep 19 '21 at 10:52
  • 1
    $\begingroup$ Also, the lifting of $Inv$ to $Ab$ makes $\mathbb Z/2$ into a co-abelian group in $CMon$ - which is not a surprise, everyone is a co-(commutative monoid) in $CMon$, and the corresponding shear map is just the shear map $\endgroup$ Sep 19 '21 at 10:56
  • $\begingroup$ (Re the original question): Couldn't I take the pushout in E_k-spaces of pt<---Free(x)--->Free(y) where x goes to y^2? $\endgroup$ Sep 19 '21 at 11:26
4
$\begingroup$

$$Fun^{\otimes}(\mathbb Z/2, C) \simeq map_{E_1}(\mathbb Z/2, C^\simeq) \simeq map_{E_k}(\mathrm{Ind}_{E_1}^{E_k}\mathbb Z/2, C^\simeq)$$ where $\mathrm{Ind}_{E_1}^{E_k}$ denotes the left adjoint to the forgetful functor.

So $Inv$ is representable, and the natural $E_{k-1}$-structure (see my comments for why I wrote $E_{k-1}$ and not $E_k$ - it is possible that in this special case too we could get $E_k$, but I don't see a reason why, and what I wrote works for any $E_1$-space $X$) on this space gives $\mathrm{Ind}_{E_1}^{E_k}\mathbb Z/2$ a natural co-$E_{k-1}$-structure (in $E_k$-spaces - with the coproduct monoidal structure).

Now does the space $\mathrm{Ind}_{E_1}^{E_k}\mathbb Z/2$ have a reasonably concrete description ? I think it's something like a bar construction $Bar(E_1,E_k, \mathbb Z/2)$ so you can get an explicit description involving the space of little $k$-disks, but I'm not entirely sure you can get much better. I'd love to hear about a better description.

$\endgroup$
8
  • $\begingroup$ Thanks, Maxime! Would it be okay to ask a few questions? 1) Why do we have $Fun^\otimes(\mathbb{Z}/2,C)\cong map_{E_1}(\mathbb{Z}/2,C^\simeq)$? 2) Why did you choose the notation $\mathrm{Ind}^{E_k}_{E_1}$? 3) Is there a reason to expect $\mathrm{Ind}^{E_k}_{E_1}(\mathbb{Z}/2)$ to be an $E_k$-group? $\endgroup$
    – Théo
    Sep 19 '21 at 21:02
  • $\begingroup$ (Also, I updated the question to correct the E_k vs E_k-1 issue; thanks!) $\endgroup$
    – Théo
    Sep 19 '21 at 21:02
  • 1
    $\begingroup$ Théo : 1) it's because any monoidal morphism between strong monoidal functors out of a grouplike monoidal category is invertible. But in fact, even if that weren't so, because you want it to land in $S$ you would have to take $map$ rather than $Fun$ (and the $C^\simeq$ is because $\mathbb Z/2$ is a groupoid). 2) I'm viewing it as some form of extension of scalars, so it's a similar notation as for extension of scalars along a ring map $A\to B$: $\mathrm{Ind}_A^B$. 3) Yes: for any $E_1$-group $G$ this will be the case $\endgroup$ Sep 20 '21 at 7:19
  • 1
    $\begingroup$ Indeed, $map_{E_k}(\mathrm{Ind}(G), X) = map_{E_1}(G,X) = map_{E_1}(G,X^{inv}) = map_{E_k}(\mathrm{Ind}(G), X^{inv})$ where $X^{inv}$ is the full sub-$E_k$-groupoid on invertible elements $\endgroup$ Sep 20 '21 at 7:20
  • 1
    $\begingroup$ @skd : it's almost that, although I think there's a shit: it should be $\Omega^k\Sigma^{k-1}B\mathbb Z/2$, because $map_{E_k}(\Omega^k\Sigma^{k-1} BX, Y) = map_{E_{k-1}}(\Omega^{k-1}\Sigma^{k-1} BX, BY) = map_{S_*}(BX, BY) = map_{E_1}(X,Y)$. That being said, I agree with you that I don't see why there would be a more explicit description, except maybe for replacing $B\mathbb Z/2$ with $\mathbb RP^\infty$ $\endgroup$ Sep 20 '21 at 7:23

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.