# Homology $H_*(TOP, \mathbb{Z}_2)$ of the stable homeomorphism space

Let $TOP$ be the stable homeomorphism space, with $TOP(n) = Homeomorphisms(\mathbb{R}^n)$. What is known about its $\mathbb{Z}_2$-homology $H_*(TOP, \mathbb{Z}_2)$? In particular I am interested in the map $H_*(TOP, \mathbb{Z}_2) \rightarrow H_*(G, \mathbb{Z}_2)$, where $G$ is the stable space of homotopy automorphisms of spheres.

Can $H_*(TOP, \mathbb{Z}_2)$ be computed knowing $H_*(G, \mathbb{Z}_2)$, $H_*(G/TOP, \mathbb{Z}_2)$ or $H_*(BTOP, \mathbb{Z}_2)$?

My specific question: Is the map $H_{4k+2}(TOP, \mathbb{Z}_2) \rightarrow H_{4k+2}(G, \mathbb{Z}_2)$ injective on the image of $\pi_{4k+2}(TOP)$ under the Hurewicz homomorphism? So for example, is it injective on primitives? Does somebody have an idea or a reference I could look at?

Even more specific (and also sufficient): Let $\kappa \in \pi_{14}^s = \pi_{14}G$ be the element with Kervaire invariant 0. It lifts to $\pi_{14}(TOP)$. Is its image under $h : \pi_{14}(TOP) \rightarrow H_{14}(TOP, \mathbb Z_2)$ zero? (By Lemma 4.3 of arxiv.org/abs/1504.06752v2, $h(\kappa)=0$ in $H_{14}G$.)

## 2 Answers

I first comment on the specific case on the kernel of $H_{14}TOP\to H_{14}G$ when restricted to the image of Hurewicz homomorphism $\pi_{14}TOP\to H_{14}TOP$ (all homology groups are $\mathbb{Z}/2$-homology). I think it is a consequence of Sullivan's decomposition that at the prime $2$ the space $G/TOP$ decomposes as a product of Eilenberg-Moore spaces $K(\mathbb{Z}/2,4n-2)$ and $K(\mathbb{Z},4n)$ (for instance see Madsen's paper projecteuclid.org/download/pdf_1/euclid.pjm/1102868638).

Serre's exact sequence of homotopy groups for the fibration $$\Omega(G/TOP)\to TOP\to G\to G/TOP\to BTOP\to BG$$ yields $$\pi_{14}\Omega(G/TOP)\to\pi_{14}TOP\to\pi_{14}G\to\pi_{14}G/TOP$$ which reads as $$0\to\pi_{14}TOP\to\mathbb{Z}/2\{\sigma^2,\kappa\}\to\mathbb{Z}/2.$$ For the Hurewicz map $h:\pi_{14}G\to H_{14}G$ it is known that $h(\sigma^2)\neq 0$ whereas $h(\kappa)=0$.

It is known that there is a Kervaire invariant one element if its Hurewicz image maps nontrivially under $H_{2^i-2}G\to H_{2^i-2}G/TOP$. As $\sigma^2$ is a Kervaire invariant one element, we then deduce that $\sigma^2$ maps nontrivially under $\pi_{14}G\to\pi_{14}G/TOP$. Therefore, it is $\kappa$ which is detected by the map $TOP\to G$. So, $h(\kappa)=0$ only tells you that $$\mathbb{Z}/2\{\kappa\}\simeq\pi_{14}TOP\to H_{14}TOP\to H_{14}G$$ is trivial. Now, I suspect that $\kappa$ maps trivially under $$\pi_{14}TOP\to H_{14}TOP$$ which is what you asked for.

EDIT(added on 28th of March) I suggest the following route to prove that $\kappa$ maps trivially under $\pi_{14}TOP\to H_{14}TOP$. Note that here $\kappa\in\pi_{14}TOP$ is any element which maps to $\kappa\in\pi_{14}G$ where we have some abuse of notation here. Also, note that $\kappa$ really lives in $\pi_{14}SG$ where $SG=Q_1S^0$ which is homotopy equivalent to $Q_0S^0$. Furthermore, note that the Hopf invariant one element $\nu\in\pi_3^s\simeq\pi_3G$ pulls back to $\pi_3TOP$ and I think we can show it is a unique pull back (this uniqueness helps in showing that a triple Toda bracket in $\pi_*TOP$ is defined). Now, following arguments of Lemma 4.3 of https://arxiv.org/pdf/1504.06752v2.pdf we may construct $\kappa$ as an unstable map as a triple Toda bracket associated to $$S^{13}\stackrel{\beta}{\longrightarrow}\Gamma^6(\Sigma^4K)\stackrel{\Gamma^6\alpha}{\longrightarrow}\Gamma^6 S^3\stackrel{\nu}{\longrightarrow}Q_0S^0$$ where $\Gamma^6=\Omega^6\Sigma^6$. The epimorphism $\pi_3TOP\to\pi_3G\simeq\pi_3SG$ then allows to construct a triple Toda bracket for an element in $\pi_{14}TOP$ which maps to $\kappa$ with trivial indeterminacy, hence representing a generator of $\pi_{14}TOP$. A composition of the form $$S^{14}\stackrel{\beta^{\flat}}{\longrightarrow}C_{\Gamma^6\alpha}\stackrel{\nu_\sharp}{\longrightarrow}TOP$$ represents the element constructed as this triple Toda bracket. Now, by the same arguments in Lemma 4.3 https://arxiv.org/pdf/1504.06752v2.pdf we have $(\beta^\flat)_*=0$ which shows that the element $\kappa\in\pi_{14}TOP$ acts trivially in homology which is the desired result.

• Thanks. Yes this is exactly the question. I thought it would be easier stating it more generally, I'm sorry. The fact that $h(\kappa) = 0$ is shown in Lemma 4.3 of arxiv.org/abs/1504.06752v2, as H. Zare kindly pointed out to me (The more general Curtis conjecture is not known to be true). How do you think a factorization of $J$ will help? $\kappa$ is in $coker J$, so does not lift to $\pi_{14}O$... Do you think one might deduce something from Madsen's statement about $H_*(BSO)$ and $H_*(BSG)$? Thanks already! Mar 20 '17 at 13:15

You should certainly start by looking at the book "The classifying spaces for surgery and cobordism of manifolds" by Madsen and Milgram. There is a PDF copy at http://www.maths.ed.ac.uk/~aar/papers/madmil.pdf. That contains more discussion of $BTop$ and $G/Top$ than $Top$ itself, but perhaps that will be enough for you. The books "$E_\infty$ ring spaces and $E_\infty$ ring spectra" (http://www.math.uchicago.edu/~may/BOOKS/e_infty.pdf; May, Quinn, Ray and Tornehave) and "Homology of iterated loop spaces" (http://www.math.uchicago.edu/~may/BOOKS/homo_iter.pdf, around page 162; Cohen, Lada and May) may also be useful. If you have looked at some of these already, then you should add details to the question, clarifying how they fall short of what you need.

• Thanks a lot for your answer! I had a look at Madsen-Milgram already, and it seems to me that they compute the homology $H_*(Q_0S^0)$, i.e. $H_*(G)$, and $H_*(G/TOP)$ and $H_*(BTOP)$. However I don't see anything on $H_*(TOP)$, and I can't think of a way to derive it from the other. The other books go in similar directions, mostly focussing on $H_*(G)$ and $H_*(BTOP)$. Do you have an idea on how I could go about $H_*(TOP)$? (I will edit my question, too.) Mar 17 '17 at 8:43
• @user106191 $TOP=\Omega BTOP$ so you can probably try to study the Serre spectral sequence. See here for details and references. Mar 17 '17 at 15:48
• I comment on the specific question on the kernel of $H_{14}TOP\to H_{14}G$ when restricted to the image of the Hurewicz map $\pi_{14}TOP\to H_{14}TOP$. The space $G/TOP$ at the prime $2$ is a product of EM spaces $K(\mathbb{Z}/2,4n-2)$ and $K(\mathbb{Z}/2,4n)$. So, look at the Serre homotopy exact sequence for $\Omega(G/TOP)\to TOP\to G\to G/TOP\to BTOP\to BG$ which yields $$\pi_{15}G/TOP\to \pi_{14}TOP\to\pi_{14}G\to\pi_{14}G/TOP$$ which reads as $$0\to\pi_{14}TOP\to\mathbb{Z}/2\{\kappa,\sigma^2\}\to\mathbb{Z}/2.$$ Now, I don't now if it is $\kappa$ which lives in $\pi_{14}TOP$ or... Mar 19 '17 at 14:20
• or if it is $\sigma^2$. For $\sigma^2$ it maps nontrivially under the Hurewicz map $h:\pi_{14}G\to H_{14}(G;\mathbb{Z}/2)$ whereas $\kappa$ maps trivially under $h$. I thought this has to help in answering your question. Mar 19 '17 at 14:21
• This paper of Madsen also can be helpful projecteuclid.org/download/pdf_1/euclid.pjm/1102868638 Mar 19 '17 at 14:26