ScalingStacks

3.2 dd-Equivalences and dd-Topoi[0M3D]

[0M2X]

Definition 3.2.1. For d≥−2d\geq-2, a map of ∞\infty-operads f:𝒪→𝒰f\colon\mathcal{O}\to\mathcal{U} is called a dd-equivalence, if the induced map hd+1​(f):hd+1​𝒪→hd+1​𝒰h_{d+1}\left(f\right)\colon h_{d+1}\mathcal{O}\to h_{d+1}\mathcal{U} is an equivalence of ∞\infty-operads, ie if it is essentially surjective on the underlying categories and induces an equivalence on the dd-truncations of all the multi-mapping spaces.

An important special case is

[0M2Y]

Definition 3.2.2. For d≥−2d\geq-2, an ∞\infty-operad 𝒪\mathcal{O} is called dd-connected if the unique map from 𝒪\mathcal{O} to the terminal ∞\infty-operad 𝔼∞\mathbb{E}_{\infty} is a dd-equivalence, ie if all the multi-mapping spaces in 𝒪\mathcal{O} are dd-connected.

[05Y3]

Remark 3.2.3. Let 𝒫\mathcal{P} be a reduced ∞\infty-operad. It is dd-connected if and only if all the spaces 𝒫⁡(n)\mathcal{P}\left(n\right) in the underlying symmetric sequence of 𝒫\mathcal{P} are dd-connected. If 𝒫\mathcal{P} is not equivalent to 𝔼0\mathbb{E}_{0}, then for some n≥2n\geq 2 we have 𝒫⁡(n)≠∅\mathcal{P}\left(n\right)\neq\varnothing, and so there exists an nn-ary operation μ∈𝒫⁡(n)\mu\in\mathcal{P}\left(n\right) for n≥2n\geq 2. By composing μ\mu with itself, we can obtain an operation in 𝒫\mathcal{P} of arbitrarily high arity and by composition with the unique nullary operation, we can obtain an operation of arbitrary arity. It follows that 𝒫≄𝔼0\mathcal{P}\not\simeq\mathbb{E}_{0} if and only if 𝒫\mathcal{P} is (−1)\left(-1\right)-connected.

The main result of this section is a characterization of dd-equivalences of reduced ∞\infty-operads. But first, we need some preliminary observations about Cartesian symmetric monoidal structures.

[05Y4]

Lemma 3.2.4. Let fα:𝒟→𝒞αf_{\alpha}\colon\mathcal{D}\to\mathcal{C}_{\alpha} be a collection of jointly conservative, symmetric monoidal functors between symmetric monoidal ∞\infty-categories.

  1. (1)

    If 𝒞α\mathcal{C}_{\alpha} is Cartesian and fαf_{\alpha} preserves finite products for all α\alpha and 𝒟¯\underline{\mathcal{D}} has all finite products, then 𝒟\mathcal{D} is Cartesian.

  2. (2)

    If 𝒞α\mathcal{C}_{\alpha} is coCartesian and fαf_{\alpha} preserves finite coproducts for all α\alpha and 𝒟¯\underline{\mathcal{D}} has all finite coproducts, then 𝒟\mathcal{D} is coCartesian.

[05Y5]

Proof. By A.2.4.2.7, the opposite of a symmetric monoidal ∞\infty-category acquires a symmetric monoidal structure, which is Cartesian if and only if the original symmetric monoidal ∞\infty-category is coCartesian. Hence, it is enough to prove (2). The unit object 1∈𝒟1\in\mathcal{D} has a unique map from the initial object ∅→1\varnothing\to 1. Since fαf_{\alpha} is both symmetric monoidal and preserves finite coproducts, fα​(∅→1)f_{\alpha}\left(\varnothing\to 1\right) is the unique map from the initial object to the unit object of 𝒞α\mathcal{C}_{\alpha}, which is an equivalence by assumption. Since the collection of fαf_{\alpha} is jointly conservative, it follows that the unit of 𝒟\mathcal{D} is initial in 𝒟¯\underline{\mathcal{D}} as well. Namely, 𝒟\mathcal{D} is unital as an ∞\infty-operad. Using 2.2.3 we have a map of ∞\infty-operads G:𝒟→𝒟⊔G\colon\mathcal{D}\to\mathcal{D}_{\sqcup}, which is an equivalence on the underlying ∞\infty-categories. We need to show that this map is symmetric monoidal. Namely, that it maps coCartesian edges (over 𝐅𝐢𝐧∗\mathbf{Fin}_{*}) to coCartesian edges. Since we already know that it is a map of ∞\infty-operads and hence preserves inert morphisms, we only need to show that active coCartesian edges map to coCartesian edges. Using the Segal conditions, we are further reduced to considering only coCartesian lifts of the unique active morphism μ:⟨n⟩→⟨1⟩\mu\colon\left\langle n\right\rangle\to\left\langle 1\right\rangle. For every collection of objects X1,X2,…,Xn∈𝒟¯X_{1},X_{2},\dots,X_{n}\in\underline{\mathcal{D}}, let

μ⊗:X1⊕⋯⊕Xn→X1⊗X2⊗⋯⊗Xn\mu_{\otimes}\colon X_{1}\oplus\cdots\oplus X_{n}\to X_{1}\otimes X_{2}\otimes\dots\otimes X_{n}

be a coCartesian lift of μ\mu to 𝒟⊗\mathcal{D}^{\otimes}. Since GG is an equivalence on the underlying ∞\infty-categories, G⁡(μ⊗)G\left(\mu_{\otimes}\right) can be considered as a map

X1⊕⋯⊕Xn→X1⊗X2⊗⋯⊗XnX_{1}\oplus\cdots\oplus X_{n}\to X_{1}\otimes X_{2}\otimes\dots\otimes X_{n}

in 𝒟⊔\mathcal{D}_{\sqcup}. Now, let

μ⊔:X1⊕⋯⊕Xn→X1⊔X2⊔⋯⊔Xn\mu_{\sqcup}\colon X_{1}\oplus\cdots\oplus X_{n}\to X_{1}\sqcup X_{2}\sqcup\dots\sqcup X_{n}

be a coCartesian lift of μ\mu to 𝒟⊔\mathcal{D}^{\sqcup}. There exists a unique (up to homotopy) map

gX1,…,Xn:X1⊔X2⊔⋯⊔Xn→X1⊗X2⊗⋯⊗Xn,g_{X_{1},\dots,X_{n}}\colon X_{1}\sqcup X_{2}\sqcup\dots\sqcup X_{n}\to X_{1}\otimes X_{2}\otimes\dots\otimes X_{n},

such that G⁡(μ⊗)=gX1,…,Xn∘μ⊔G\left(\mu_{\otimes}\right)=g_{X_{1},\dots,X_{n}}\circ\mu_{\sqcup}. We need to show that gX1,…,Xng_{X_{1},\dots,X_{n}} is an equivalence in 𝒟¯\underline{\mathcal{D}} for all X1,…,Xn∈𝒟¯X_{1},\dots,X_{n}\in\underline{\mathcal{D}}. For every α\alpha, we have a homotopy commutative diagram

𝒟⊗\textstyle{\mathcal{D}^{\otimes}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}fα\scriptstyle{f_{\alpha}}𝒟⊔\textstyle{\mathcal{D}^{\sqcup}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}fα\scriptstyle{f_{\alpha}}𝒞α⊗\textstyle{\mathcal{C}_{\alpha}^{\otimes}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}𝒞α⊔\textstyle{\mathcal{C}_{\alpha}^{\sqcup}}

in which the vertical and bottom maps are symmetric monoidal. It follows that fα​(gX1,…,Xn)f_{\alpha}\left(g_{X_{1},\dots,X_{n}}\right) is an equivalence in 𝒞α\mathcal{C}_{\alpha} for all α\alpha. By joint conservativity, gX1,…,Xng_{X_{1},\dots,X_{n}} is an equivalence as well. ∎

[05Y6]

Lemma 3.2.5. Let 𝒞×\mathcal{C}_{\times} be a Cartesian symmetric monoidal ∞\infty-category. For every ∞\infty-operad 𝒟\mathcal{D}, the ∞\infty-operad Alg𝒟⁡(𝒞)\operatorname{Alg}_{\mathcal{D}}\left(\mathcal{C}\right) (see A.2.2.5.4) is also Cartesian.

[05Y7]

Proof. By A.2.2.5.4, since 𝒞×\mathcal{C}_{\times} is symmetric monoidal, so is Alg𝒟⁡(𝒞)\operatorname{Alg}_{\mathcal{D}}\left(\mathcal{C}\right) and, for every X∈𝒟X\in\mathcal{D}, the evaluation functor eX:Alg𝒟⁡(𝒞)→𝒞×e_{X}\colon\operatorname{Alg}_{\mathcal{D}}\left(\mathcal{C}\right)\to\mathcal{C}_{\times} is a symmetric monoidal functor. On the underlying ∞\infty-categories, eXe_{X} also preserves finite products since it preserves all limits. Finally, we show that the collection of evaluation functors is jointly conservative since they can be presented as the composition of the conservative restriction functor

Alg𝒟⁡(𝒞)→Fun⁡(𝒟,𝒞)\operatorname{Alg}_{\mathcal{D}}\left(\mathcal{C}\right)\to\operatorname{Fun}\left(\mathcal{D},\mathcal{C}\right)

and the collection of evaluation functors

eX:Fun⁡(𝒟,𝒞)→𝒞,e_{X}\colon\operatorname{Fun}\left(\mathcal{D},\mathcal{C}\right)\to\mathcal{C},

which are jointly conservative by T.3.1.2.1. Now, by 3.2.4(1), Alg𝒟⁡(𝒞)\operatorname{Alg}_{\mathcal{D}}\left(\mathcal{C}\right) is Cartesian. ∎

We are now ready for the main proposition.

[05Y8]

Proposition 3.2.6. Let d≥−1d\geq-1. Given a map of reduced ∞\infty-operads f:𝒫→𝒬f\colon\mathcal{P}\to\mathcal{Q}, the following are equivalent:

  1. (1)

    The map ff is a dd-equivalence.

  2. (2)

    For every (d+1)\left(d+1\right)-topos 𝒞\mathcal{C}, the induced map

    Map𝐎𝐩∞⁡(𝒬,𝒞×)→Map𝐎𝐩∞⁡(𝒫,𝒞×)\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{Q},\mathcal{C}_{\times}\right)\to\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{P},\mathcal{C}_{\times}\right)

    is a homotopy equivalence.

  3. (3)

    For every simplicial set KK, the induced map

    Map𝐎𝐩∞⁡(𝒬,𝒮≤dK)→Map𝐎𝐩∞⁡(𝒫,𝒮≤dK)\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{Q},\mathcal{S}_{\leq d}^{K}\right)\to\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{P},\mathcal{S}_{\leq d}^{K}\right)

    is a homotopy equivalence where 𝒮≤dK\mathcal{S}_{\leq d}^{K} is given the Cartesian symmetric monoidal structure.

  4. (4)

    The induced map

    Alg¯𝒬​(𝒮≤d)→Alg¯𝒫​(𝒮≤d)\underline{\operatorname{Alg}}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\to\underline{\operatorname{Alg}}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right)

    is an equivalence of ∞\infty-categories, where 𝒮≤d\mathcal{S}_{\leq d} is given the Cartesian symmetric monoidal structure.

[05Y9]

Proof. (1)⟹(2)\left(1\right)\implies\left(2\right) Consider the commutative diagram

Map𝐎𝐩∞⁡(𝒬,𝒞)\textstyle{\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{Q},\mathcal{C}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Map𝐎𝐩∞⁡(𝒫,𝒞)\textstyle{\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{P},\mathcal{C}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Map𝐎𝐩∞⁡(hd+1​𝒬,𝒞)\textstyle{\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(h_{d+1}\mathcal{Q},\mathcal{C}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Map𝐎𝐩∞⁡(hd+1​𝒫,𝒞).\textstyle{\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(h_{d+1}\mathcal{P},\mathcal{C}\right).}

Since hd+1​(𝒫)→hd+1​(𝒬)h_{d+1}\left(\mathcal{P}\right)\to h_{d+1}\left(\mathcal{Q}\right) is an equivalence of ∞\infty-operads, the bottom map is a homotopy equivalence. By 3.1.8, the vertical maps are equivalences as well, and so, by the 2-out-of-3 property, the top map is an equivalence.

(2)⟹(3)\left(2\right)\implies\left(3\right) Since 𝒮≤dK\mathcal{S}_{\leq d}^{K} is a (d+1)\left(d+1\right)-topos, this is just a special case.

(3)⟹(4)\left(3\right)\implies\left(4\right) By Yoneda’s lemma applied to 𝐂𝐚𝐭∞\mathbf{Cat}_{\infty}, the map

Alg¯𝒬​(𝒮≤d)→Alg¯𝒫​(𝒮≤d)\underline{\operatorname{Alg}}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\to\underline{\operatorname{Alg}}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right)

is an equivalence of ∞\infty-categories if for every ∞\infty-category ℰ\mathcal{E}, the map

Map𝐂𝐚𝐭∞⁡(ℰ,Alg¯𝒬​(𝒮≤d))→Map𝐂𝐚𝐭∞⁡(ℰ,Alg¯𝒫​(𝒮≤d))\operatorname{Map}_{\mathbf{Cat}_{\infty}}(\mathcal{E},\underline{\operatorname{Alg}}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right))\to\operatorname{Map}_{\mathbf{Cat}_{\infty}}(\mathcal{E},\underline{\operatorname{Alg}}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right))

is a homotopy equivalence. Using the fully faithful embedding 𝐂𝐚𝐭∞↪𝐎𝐩∞\mathbf{Cat}_{\infty}\hookrightarrow\mathbf{Op}_{\infty}, which is left adjoint to the underlying category functor 𝐎𝐩∞→𝐂𝐚𝐭∞\mathbf{Op}_{\infty}\to\mathbf{Cat}_{\infty} (see A.2.1.4.11), this map is equivalent to

Map𝐎𝐩∞⁡(ℰ,Alg𝒬⁡(𝒮≤d))→Map𝐎𝐩∞⁡(ℰ,Alg𝒫⁡(𝒮≤d)).\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{E},\operatorname{Alg}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\right)\to\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{E},\operatorname{Alg}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right)\right).

By adjointness with the Boardman–Vogt tensor product and the fact that it is symmetric, the map is equivalent to

Map𝐎𝐩∞⁡(𝒬,Algℰ⁡(𝒮≤d))→Map𝐎𝐩∞⁡(𝒫,Algℰ⁡(𝒮≤d)).\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{Q},\operatorname{Alg}_{\mathcal{E}}\left(\mathcal{S}_{\leq d}\right)\right)\to\operatorname{Map}_{\mathbf{Op}_{\infty}}\left(\mathcal{P},\operatorname{Alg}_{\mathcal{E}}\left(\mathcal{S}_{\leq d}\right)\right).

Since ℰ\mathcal{E} is an ∞\infty-category, by 3.2.5 the ∞\infty-operad Algℰ⁡(𝒮≤d)\operatorname{Alg}_{\mathcal{E}}\left(\mathcal{S}_{\leq d}\right) is just the ∞\infty-category of functors (𝒮≤d)ℰ\left(\mathcal{S}_{\leq d}\right)^{\mathcal{E}} endowed with the Cartesian symmetric monoidal structure. Since the functor category is invariant under Joyal equivalences, we can replace ℰ\mathcal{E} with any simplicial set KK.

(4)⟹(1)\left(4\right)\implies\left(1\right) Consider the commutative diagram

Alg¯𝒬​(𝒮≤d)\textstyle{\underline{\operatorname{Alg}}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}≀\scriptstyle{\wr}Alg¯𝒫​(𝒮≤d)\textstyle{\underline{\operatorname{Alg}}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}≀\scriptstyle{\wr}Alg¯hd​𝒬​(𝒮≤d)\textstyle{\underline{\operatorname{Alg}}_{h_{d}\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Alg¯hd​𝒫​(𝒮≤d).\textstyle{\underline{\operatorname{Alg}}_{h_{d}\mathcal{P}}\left(\mathcal{S}_{\leq d}\right).}

By 3.1.8, the vertical maps are equivalences; hence by 2-out-of-3, the top map is an equivalence if and only if the bottom map is. We can therefore assume without loss of generality that 𝒫\mathcal{P} and 𝒬\mathcal{Q} are themselves essentially dd-operads. This implies that 𝒫⁡(n)\mathcal{P}\left(n\right) and 𝒬⁡(n)\mathcal{Q}\left(n\right) are dd-truncated spaces for all n≥0n\geq 0. Now, consider the commutative diagram

Alg¯𝒬​(𝒮≤d)\textstyle{\underline{\operatorname{Alg}}_{\mathcal{Q}}\left(\mathcal{S}_{\leq d}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f∗\scriptstyle{f^{*}}U𝒬\scriptstyle{U_{\mathcal{Q}}}Alg¯𝒫​(𝒮≤d)\textstyle{\underline{\operatorname{Alg}}_{\mathcal{P}}\left(\mathcal{S}_{\leq d}\right)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}U𝒫\scriptstyle{U_{\mathcal{P}}}𝒮≤d,\textstyle{\mathcal{S}_{\leq d},}

where U𝒫U_{\mathcal{P}} and U𝒬U_{\mathcal{Q}} are the corresponding forgetful functors. By 2.4.4, the associated map

T𝒫=∐n(𝒫⁡(n)×Xn)h​Σn​⟶∼​∐n(𝒬⁡(n)×Xn)h​Σn=T𝒬T_{\mathcal{P}}=\coprod_{n}\left(\mathcal{P}\left(n\right)\times X^{n}\right)_{h\Sigma_{n}}\overset{\sim}{\longrightarrow}\coprod_{n}\left(\mathcal{Q}\left(n\right)\times X^{n}\right)_{h\Sigma_{n}}=T_{\mathcal{Q}}

of Construction 2.4.3 is a natural equivalence of functors. On the other hand, by 2.4.6, this map is induced from a map of symmetric sequences f𝐒𝐒𝐞𝐪:{𝒫⁡(n)}→{𝒬⁡(n)}f_{\mathbf{SSeq}}\colon\left\{\mathcal{P}\left(n\right)\right\}\to\left\{\mathcal{Q}\left(n\right)\right\}. We want to deduce that f𝐒𝐒𝐞𝐪f_{\mathbf{SSeq}} is an equivalence. For d=−1d=-1, there is nothing to prove and so we assume that d≥0d\geq 0. Taking X=[n]X=\left[n\right], there is a coproduct decomposition

(𝒫⁡(n)×Xn)h​Σn=𝒫⁡(n)⊔J,\left(\mathcal{P}\left(n\right)\times X^{n}\right)_{h\Sigma_{n}}=\mathcal{P}\left(n\right)\sqcup J,

where the summand 𝒫⁡(n)\mathcal{P}\left(n\right) corresponds to orbits of points whose XnX^{n} component is a permutation (note that when d=0d=0, the homotopy orbits in 𝒮≤0\mathcal{S}_{\leq 0} are just the orbits as a set). This characterization implies that f𝐒𝐒𝐞𝐪:𝒫⁡(n)→𝒬⁡(n)f_{\mathbf{SSeq}}\colon\mathcal{P}\left(n\right)\to\mathcal{Q}\left(n\right) is an equivalence. Finally, since (−)𝐒𝐒𝐞𝐪\left(-\right)_{\mathbf{SSeq}} is conservative, by 2.3.6, we deduce that ff is an equivalence. ∎

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

Tomer Schlank, Lior Yanovski

Original source: arXiv:1808.06006v3

Original source · 1808.06006v3