ScalingStacks

[0MJT]

Proof. Condition (R.1) implies both that i∗i_{*} carries TT-local objects to SS-local objects and that we obtain an adjunction:

LT∘i∗:S−1​𝒫⁡(Υn)⇄T−1​𝒫⁡(ℛ):i∗.L^{T}\circ i^{*}\colon S^{-1}\pre(\Upsilon_{n})\rightleftarrows T^{-1}\pre(\mathcal{R})\colon i_{*}.

Similarly, condition (R.2) implies that i∗i^{*} carries SS-local objects to TT-local objects and that we obtain a second adjunction:

LS∘i!:T−1𝒫(ℛ)⇄S−1𝒫(Υn):i∗.L^{S}\circ i_{!}\colon T^{-1}\pre(\mathcal{R})\rightleftarrows S^{-1}\pre(\Upsilon_{n})\colon i^{*}.

Since i∗i^{*} sends SS-local objects to TT-local objects, i∗≃LT∘i∗i^{*}\simeq L^{T}\circ i^{*} when restricted to the SS-local objects of 𝒫⁡(Υn)\pre(\Upsilon_{n}). Thus i∗:𝒫⁡(Υn)→𝒫⁡(ℛ)i^{*}\colon\pre(\Upsilon_{n})\to\pre(\mathcal{R}) restricts to a functor

i∗:Cat(∞,n)=S−1​𝒫⁡(Υn)→T−1​𝒫⁡(ℛ),i^{*}\colon\cat_{(\infty,n)}=S^{-1}\pre(\Upsilon_{n})\to T^{-1}\pre(\mathcal{R}),

that admits a left adjoint LS∘i!L^{S}\circ i_{!} and a right adjoint i∗i_{*}.

Notice that i!(R)≅i(R)i_{!}(R)\cong i(R) in 𝒫⁡(Υn)\pre(\Upsilon_{n}), where we have identified ℛ\mathcal{R} and Υn\Upsilon_{n} with their images under the Yoneda embedding in, respectively, 𝒫⁡(ℛ)\pre(\mathcal{R}) and 𝒫⁡(Υn)\pre(\Upsilon_{n}). Thus by (R.3) the counit map applied to r∈ℛr\in\mathcal{R},

R→i∗∘LS∘i!(R)≅i∗∘LSi(R)≅i∗i(R)R\to i^{*}\circ L^{S}\circ i_{!}(R)\cong i^{*}\circ L^{S}i(R)\cong i^{*}i(R)

becomes an equivalence in T−1​𝒫⁡(ℛ)T^{-1}\pre(\mathcal{R}) (the last equality follows from Lemma 8.6, as the image of ii consists of SS-local objects). The endofunctor i∗∘LS∘i!:T−1𝒫(ℛ)→T−1𝒫(ℛ)i^{*}\circ L^{S}\circ i_{!}\colon T^{-1}\pre(\mathcal{R})\to T^{-1}\pre(\mathcal{R}) is a composite of left adjoints, hence commutes with colimits. Therefore, as ℛ\mathcal{R} is dense in T−1​𝒫⁡(ℛ)T^{-1}\pre(\mathcal{R}), the functor i∗∘LS∘i!i^{*}\circ L^{S}\circ i_{!} is determined by its restriction to ℛ\mathcal{R}. It is equivalent the left Kan extension of its restriction to ℛ\mathcal{R}. Consequently i∗∘LS∘i!i^{*}\circ L^{S}\circ i_{!} is equivalent to the identity functor.

On the other hand, for each X∈Cat(∞,n)X\in\cat_{(\infty,n)}, consider the other counit map X→i∗​i∗​XX\to i_{*}i^{*}X. For each kk, we have natural equivalences,

Map⁡(Ck,i∗​i∗​X)\displaystyle\map(C_{k},i_{*}i^{*}X) ≃Map⁡(i⁡(Rk),i∗​i∗​X)\displaystyle\simeq\map(i(R_{k}),i_{*}i^{*}X)
≃Map⁡(i∗​i​(Rk),i∗​X)\displaystyle\simeq\map(i^{*}i(R_{k}),i^{*}X)
≃Map⁡(Rk,i∗​X)\displaystyle\simeq\map(R_{k},i^{*}X)
≃Map⁡(i⁡(Rk),X)≃Map⁡(Ck,X),\displaystyle\simeq\map(i(R_{k}),X)\simeq\map(C_{k},X),

which follow from (R.3), (R.4), the identity i∗​(Rk)≅i⁡(Rk)i_{*}(R_{k})\cong i(R_{k}), and the fact that i∗​Xi^{*}X is TT-local. By Remark 7.1 this implies that the counit X→i∗​i∗​XX\to i_{*}i^{*}X is an equivalence. Thus i∗i^{*} is a functor with both a left and right inverse, hence is itself an equivalence of ∞\infty-categories. ∎

Original mathematics by the credited authors. Source collection and HTML conversion remain in progress.

Clark Barwick, Christopher Schommer-Pries

Original source: arXiv:1112.0040v6

    Original source · 1112.0040v6