2
$\begingroup$

I am currently reading the paper Homotopy Fixed Point Methods for Lie Groups and Finite Loop Spaces by Dwyer and Wilkerson (can be found here http://www.math.purdue.edu/~wilker/papers/analysis.pdf for example), and I'm having trouble proving a basic result in the appendix on homotopy fixed points, and any help would be much appreciated.

Assumption: All groups are discrete groups, and $EG$ and $BG$ are CW-complexes (for example coming from the usual nerve construction).

Lemma 10.4: Let $W \rightarrow BG$ be a (Serre) fibration with fibre $F$. Then there is a proxy action of $G$ on $F$ such that $F^{hG}$ is homotopy equivalent to the space of sections of $W \rightarrow BG$.

Here proxy action (as they defined it a couple of paragraphs before the statement of the lemma) just means that there is an ordinary homotopy equivalence $F \simeq X$ where $X$ is a $G$-space and $F^{hG} := X^{hG}$.

Following their proof, we're supposed to consider the pullback $\tilde{W}$ for the proxy action space $\require{AMScd}$ \begin{CD} \tilde{W} @>{h}>> W\\ @VVV @VV{p}V \\ EG @>{q}>> BG \end{CD} and since $EG$ is contractible, we obtain that $\tilde{W} \simeq F$. Now we need to show that $\tilde{W}$ has the correct homotopy fixed point as required. I've tried two approaches that I can't seem to make work.

Method 1: Recall now that $\tilde{W}^{hG} := Map^G(EG,\tilde{W})$ the space of $G$-equivariant maps, and write $\Gamma(W\xrightarrow{p} BG)$ for the space of sections. So a point in $\tilde{W}^{hG}$ is just a commuting square $\require{AMScd}$ \begin{CD} EG @>{f}>> W\\ @V{a}VV @VV{p}V \\ EG @>{q}>> BG \end{CD} where $a$ is a $G$-map and a point in $\Gamma(W\xrightarrow{p} BG)$ is just a commuting square $\require{AMScd}$ \begin{CD} EG @>{s}>> W\\ @V{id}VV @VV{p}V \\ EG @>{q}>> BG \end{CD} So we can clearly get a map $\Gamma(W\xrightarrow{p} BG) \rightarrow \tilde{W}^{hG} $.

To get a map the other way, recall that the $G$-Whitehead theorem (for instance 6.4.2 of Benson's Representations and Cohomology V2) gives that any $G$-self-map between $EG$ is a $G$-homotopy equivalence. So we obtain a diagram $\require{AMScd}$ \begin{CD} EG @>{\bar{a}}>> EG @>{f}>> W\\ @V{id}VV@V{a}VV @VV{p}V \\ EG @>{id}>> EG @>{q}>> BG \end{CD} where the left square commutes up to a $G$-homotopy, call it $\gamma$ say. Using this homotopy, that $p$ was a fibration, and looking at the diagram $\require{AMScd}$ \begin{CD} EG\times 0 @>{f}>> W\\ @VVV @VV{p}V \\ EG\times I @>>{\gamma}> BG \end{CD} we can get a point in $\Gamma(W\xrightarrow{p} BG)$. My problem with this is that I don't know how to show that the two maps so defined give the required homotopy equivalence. In fact I have some doubts that this will work since I've made so many choices in the construction of the second map.

Method 2: This uses the usual identification of $\tilde{W}^{hG}$ with the space of sections of the canonical fibration $\tilde{W}_{hG} \xrightarrow{\pi} BG$ where $\tilde{W}_{hG} := EG \times_G \tilde{W}$ the homotopy orbit space/Borel construction. One can show that the natural map $\tilde{W}_{hG} \xrightarrow{\bar{h}} W$ is a homotopy equivalence. Then I sort of want to say that because of this homotopy equivalence we obtain $\Gamma(\tilde{W}_{hG} \xrightarrow{\pi} BG) \simeq \Gamma(W \xrightarrow{p} BG)$, but the problem with this is that the two fibrations are using different copies of $EG$.

$\endgroup$
0
$\begingroup$

In case anyone's interested, here's a solution due to a couple of friends.

Write $\tilde{W}^{hG} = Map^G(EG,\tilde{W}) = Map^G(EG,EG)\times_{Map^G(EG,BG)}Map^G(EG,W)$ and $\Gamma(W\rightarrow BG) = *\times_{Map^G(EG,BG)}Map^G(EG,W)$.

Claim: $EG \xrightarrow{\simeq} *$ induces $Map^G(EG,EG) = (EG)^{hG} \xrightarrow{\simeq} (*)^{hG} = *$, that is, that $Map^G(EG,EG)$ is contractible.

Given the claim, we then look at the diagram $\require{AMScd}$ \begin{CD} * @>>> Map^G(EG,BG) @<<< Map^G(EG,W)\\ @V{\simeq}VV @V{id}VV @V{id}VV\\ Map^G(EG,EG) @>>> Map^G(EG,BG) @<<< Map^G(EG,W)\\ \end{CD} and use for example (the dual of) Hovey's cube lemma (Lemma 5.2.6 of https://web.math.rochester.edu/people/faculty/doug/otherpapers/hovey-model-cats.pdf), together with the fact that any space is fibrant and that the right horizontal maps are fibrations to get the required homotopy equivalence of pullbacks.

Proof of claim: The unique $G$-map $EG \xrightarrow{\simeq} *$ induces a $G$-map $$EG\times EG \xrightarrow{pr_1} EG $$ between free $G$-spaces, so by $G$-Whitehead, is a $G$-homotopy equivalence. So quotienting out by $G$ we obtain $$(EG)_{hG} \xrightarrow{\simeq} (*)_{hG} = BG$$ We know that in fact this gives a fibre homotopy equivalence $\require{AMScd}$ \begin{CD} (EG)_{hG} @>{\simeq}>> (*)_{hG} = BG\\ @VVV @VVV\\ BG @>{id}>> BG \\ \end{CD} for example from page 52 of May's Concise Algebraic Topology https://www.math.uchicago.edu/~may/CONCISE/ConciseRevised.pdf. But then the space of sections of the two vertical fibrations are homotopy equivalent, and we know that homotopy fixed point spaces can be identified with such spaces of sections. $\square$

Comment: The result is quite interesting as it provides sort of a converse to the well-known fact that for any $G$-space $X$ the homotopy fixed points $X^{hG}$ can be identified with the space of sections to the fibration $X_{hG} \rightarrow BG$.

$\endgroup$

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.