ScalingStacks

13. Categorical equivalences[0MTV]

In this section we provide a generalization to Segal spaces of the category theoretic concepts of “natural isomorphism of functors” and “equivalence of categories”, and show that for the complete Segal spaces, these concepts correspond precisely to those of “homotopy between maps” and “(weak) homotopy equivalence”.

Note that, by the results of §9 through §12, statements (7.1) through (7.6) of §7 are now available to us.

13.1. Categorical homotopies[0MTW]

Let EE denote, as in §6, the discrete nerve of I⁡[1]I[1]. We define a categorical homotopy between maps f,g:U⇉Vf,g\colon U\rightrightarrows V of Segal spaces to be any one of the following equivalent data: a map H:U×E→VH\colon U\times E\rightarrow V, a map H′:U→VEH^{\prime}\colon U\rightarrow V^{E}, or a map H′′:E→VUH^{\prime\prime}\colon E\rightarrow V^{U}, making the appropriate diagram commute:

U\displaystyle{{U}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f\scriptstyle{f}U×i0\scriptstyle{U\times i_{0}}V\displaystyle{{V}}F⁡(0)\displaystyle{{F(0)}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}{f}\scriptstyle{\{f\}}i0\scriptstyle{i_{0}}U×E\displaystyle{{U\times E}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H\scriptstyle{H}V\displaystyle{{V}}U\displaystyle{{U}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f\scriptstyle{f}H′\scriptstyle{H^{\prime}}g\scriptstyle{g}VE\displaystyle{{V^{E}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Vi0\scriptstyle{V^{i_{0}}}Vi1\scriptstyle{V^{i_{1}}}E\displaystyle{{E}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H′′\scriptstyle{H^{\prime\prime}}VU\displaystyle{{V^{U}}}U\displaystyle{{U}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}g\scriptstyle{g}U×i1\scriptstyle{U\times i_{1}}V\displaystyle{{V}}F⁡(0)\displaystyle{{F(0)}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}{g}\scriptstyle{\{g\}}i1\scriptstyle{i_{1}}

If UU and VV are discrete nerves of categories CC and DD, then the categorical homotopies of maps between UU and VV correspond exactly to natural isomorphisms of functors between CC and DD.

[05W4]

Proposition 13.2. If UU is a Segal space and WW is a complete Segal space, then a pair of maps f,g:U⇉Wf,g\colon U\rightrightarrows W are categorically homotopic if and only if they are homotopic in the usual sense; i.e., if there exists a map K:U×Δ⁡[1]→WK\colon U\times\Delta[1]\rightarrow W which restricts to ff and gg on the endpoints of Δ⁡[1]\Delta[1].

[05W5]

Proof. The maps Wi0,Wi1:WE→WW^{i_{0}},W^{i_{1}}\colon W^{E}\rightarrow W are Reedy trivial fibrations if WW is a complete Segal space. This is because of parts (2) and (3) of (6.4), together with the observation that

(WE)n≈Maps​𝒮⁡(E,WF⁡(n))≈(WF⁡(n))0(W^{E})_{n}\approx\Map_{s{\operatorname{\mathcal{S}}}}(E,W^{F(n)})\approx(W^{F(n)})_{0}

since WF⁡(n)W^{F(n)} is a complete Segal space by (7.3). Thus, categorically homotopic maps coincide in the Reedy homotopy category, and hence are simplicially homotopic since WW is Reedy fibrant. ∎

13.3. Categorical equivalences[0MTX]

We say that a map g:U→Vg\colon U\rightarrow V of Segal spaces is a categorical equivalence if there exist maps f,h:V→Uf,h\colon V\rightarrow U and categorical homotopies g​f∼1Vgf\sim 1_{V} and h​g∼1Uhg\sim 1_{U}. Note that if UU and VV are discrete nerves of categories, then the categorical equivalences correspond exactly to equivalences of categories.

[05W6]

Proposition 13.4. A map g:U→Vg\colon U\rightarrow V between complete Segal spaces is a categorical equivalence if and only if it is a simplicial homotopy equivalence, if and only if it is a Reedy weak equivalence.

[05W7]

Proof. The first “if and only if” is immediate from (13.2), while the second follows from the fact that complete Segal spaces are cofibrant and fibrant in the Reedy simplicial model category. ∎

[05W8]

Proposition 13.5. Let AA, BB, and WW be Segal spaces. If f,g:A⇉Bf,g\colon A\rightrightarrows B are categorically homotopic maps, then the induced maps WB⇉WAW^{B}\rightrightarrows W^{A} are categorically homotopic. If f:A→Bf\colon A\rightarrow B is a categorical equivalence, then the induced map WB→WAW^{B}\rightarrow W^{A} is a categorical equivalence.

[05W9]

Proof. If a categorical homotopy between ff and gg is given by H:A×E→BH\colon A\times E\rightarrow B, then WH:WB→WA×E≈(WA)EW^{H}\colon W^{B}\rightarrow W^{A\times E}\approx(W^{A})^{E} is a categorical homotopy of WfW^{f} and WgW^{g}. The statement about categorical equivalences follows. ∎

[05WA]

Proposition 13.6. If f:U→Vf\colon U\rightarrow V is a categorical equivalence between Segal spaces, then it is a weak equivalence in the complete Segal space model category structure.

[05WB]

Proof. Recall from (7.2) that ff is a weak equivalence in the complete Segal space model category if and only if Maps​𝒮⁡(f,W)\Map_{s{\operatorname{\mathcal{S}}}}(f,W) is a weak equivalence of spaces for each complete Segal space WW. This is equivalent to supposing that Wf:WV→WUW^{f}\colon W^{V}\rightarrow W^{U} is a Reedy weak equivalence for each complete Segal space WW, since (Wf)n≈Maps​𝒮⁡(f,WF⁡(n))(W^{f})_{n}\approx\Map_{s{\operatorname{\mathcal{S}}}}(f,W^{F(n)}) and since WF⁡(n)W^{F(n)} is a complete Segal space by (7.3). The result now follows by noting that WfW^{f} is a categorical equivalence between complete Segal spaces by (13.5) and (7.3), and thus is a Reedy weak equivalence by (13.4). ∎

13.7. Categorical equivalences are Dwyer-Kan equivalences[0MTY]

In the remainder of this section, we prove the following result.

[05WC]

Proposition 13.8. If g:U→Vg\colon U\rightarrow V is a categorical equivalence of Segal spaces, then it is a Dwyer-Kan equivalence.

[05WD]

Proof. We first note that since Ho⁡(U×E)=Ho⁡U×Ho⁡E=Ho⁡U×I⁡[1]\ho(U\times E)=\ho U\times\ho E=\ho U\times I[1], we see that categorically homotopic maps of Segal spaces induce naturally isomorphic functors between their homotopy categories, and thus a categorical equivalence induces an equivalence between homotopy categories.

If f,h:V→Uf,h\colon V\rightarrow U are maps together with categorical homotopies H:g​f∼1VH\colon gf\sim 1_{V} and K:h​g∼1UK\colon hg\sim 1_{U}, then (13.9) applied to the diagrams

U\displaystyle{{U}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}g\scriptstyle{g}1\scriptstyle{1}K\scriptstyle{K}V\displaystyle{{V}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}h\scriptstyle{h}U\displaystyle{{U}}U\displaystyle{{U}}UE\displaystyle{{U^{E}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Vi0\scriptstyle{V^{i_{0}}}Vi1\scriptstyle{V^{i_{1}}}  and  V\displaystyle{{V}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f\scriptstyle{f}1\scriptstyle{1}H\scriptstyle{H}U\displaystyle{{U}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}g\scriptstyle{g}V\displaystyle{{V}}V\displaystyle{{V}}VE\displaystyle{{V^{E}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Vi0\scriptstyle{V^{i_{0}}}Vi1\scriptstyle{V^{i_{1}}}

will show that g​fgf and h​ghg are Dwyer-Kan equivalences, and hence gg is a Dwyer-Kan equivalence using (7.5). ∎

[05WE]

Lemma 13.9. If WW is a Segal space, the map W→WEW\rightarrow W^{E} and both maps WE→WW^{E}\rightarrow W are Dwyer-Kan equivalences.

[05WF]

Proof. By (7.5) it suffices to show that the map j:W→WEj\colon W\rightarrow W^{E} induced by E→F⁡(0)E\rightarrow F(0) is a Dwyer-Kan equivalence. We have already noted in the first part of the proof of (13.8) that categorically equivalent Segal spaces have equivalent homotopy categories, whence Ho⁡(WE)→Ho⁡W\ho(W^{E})\rightarrow\ho W is an equivalence of categories. Thus it suffices to show that the induced map mapW⁡(x,y)→mapWE⁡(j⁡(x),j⁡(y))\map_{W}(x,y)\rightarrow\map_{W^{E}}(j(x),j(y)) is a weak equivalence for each x,y∈ob⁡Wx,y\in{\operatorname{ob}}W.

We consider the diagram

W1\displaystyle{{W_{1}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}j∗\scriptstyle{j^{*}}(d1,d0)\scriptstyle{(d_{1},d_{0})}(WE)1\displaystyle{{(W^{E})_{1}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}(d1,d0)\scriptstyle{(d_{1},d_{0})}i∗\scriptstyle{i^{*}}(WF⁡(1))1\displaystyle{{(W^{F(1)})_{1}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}(d1,d0)\scriptstyle{(d_{1},d_{0})}W0×W0\displaystyle{{W_{0}\times W_{0}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}j∗\scriptstyle{j^{*}}(WE)0×(WE)0\displaystyle{{(W^{E})_{0}\times(W^{E})_{0}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}i∗\scriptstyle{i^{*}}(WF⁡(1))0×(WF⁡(1))0\displaystyle{{(W^{F(1)})_{0}\times(W^{F(1)})_{0}}}

where the horizontal arrows are induced by maps F⁡(1)→𝑖E→𝑗F⁡(0)F(1)\xrightarrow{i}E\xrightarrow{j}F(0); note that j​i=s0ji=s_{0}. We will show that the two horizontal maps marked i∗i^{*} are homotopy monomorphisms (12.2), and that the large rectangle is a homotopy pullback; this will imply that for each pair x,y∈ob⁡Wx,y\in{\operatorname{ob}}W the maps of fibers

mapW⁡(x,y)→j∗mapWE⁡(j​x,j​y)→i∗mapWF⁡(1)⁡(s0​x,s0​y)\map_{W}(x,y)\xrightarrow{j^{*}}\map_{W^{E}}(jx,jy)\xrightarrow{i^{*}}\map_{W^{F(1)}}(s_{0}x,s_{0}y)

are such that i∗​j∗i^{*}j^{*} is a weak equivalence and i∗i^{*} is a homotopy monomorphism, whence j∗j^{*} is a weak equivalence, as desired.

That the map Wi:WE→WF⁡(1)W^{i}\colon W^{E}\rightarrow W^{F(1)} is a homotopy monomorphism of spaces in each simplicial degree (and thus also the i∗i^{*}’s) follows since (WE)n=Maps​𝒮⁡(E,WF⁡(n))(W^{E})_{n}=\Map_{s{\operatorname{\mathcal{S}}}}(E,W^{F(n)}), (WF⁡(1))n=Maps​𝒮⁡(F⁡(1),WF⁡(n))=(WF⁡(n))1(W^{F(1)})_{n}=\Map_{s{\operatorname{\mathcal{S}}}}(F(1),W^{F(n)})=(W^{F(n)})_{1}, and since WF⁡(n)W^{F(n)} is a Segal space, using (6.2).

The large rectangle is isomorphic to the square

W1\displaystyle{{W_{1}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}(s0,s1)\scriptstyle{(s_{0},s_{1})}(d1,d0)\scriptstyle{(d_{1},d_{0})}W2×W1W2\displaystyle{{W_{2}\times_{W_{1}}W_{2}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}(d2​π1,d0​π2)\scriptstyle{(d_{2}\pi_{1},d_{0}\pi_{2})}W0×W0\displaystyle{{W_{0}\times W_{0}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}s0×s0\scriptstyle{s_{0}\times s_{0}}W1×W1\displaystyle{{W_{1}\times W_{1}}}

(where W2×W1W2W_{2}\times_{W_{1}}W_{2} denotes the limit of the diagram W2→d1W1←d1W2W_{2}\xrightarrow{d_{1}}W_{1}\xleftarrow{d_{1}}W_{2}) which is shown to be a homotopy pull-back by a straightforward computation using (12.4). ∎

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

Charles Rezk

Original source: arXiv:math/9811037v3

Original source · math/9811037v3