Let $S^0_p$ be the $p$-adic sphere spectrum. Let $GL_1(S^0_p)$ be the set of unit componen of $\Omega^{\infty}S^0_p$. For any map $ X \to BGL_1(S_p^0)$ we get a Thom spectrum call it $Mf$. Now consider the identity map on $GL_1(S^0_p)$, the Thom spectrum associated to that map is what I call the universal Thom spectra and denote it by $MGL_1(S^0_p)$.

My question is what are the homotopy groups of this Thom spectrum?


Assume that $R$ is a connective $E_\infty$ ring spectrum. Typically $GL_1(R)$ denotes the set of components in $\Omega^\infty R$ which span $GL_1(\pi_0 R)=\pi_0 R^\times$. I would call the unit component $SL_1(R)$. I will use this notation below.

In the formalism of Ando-Blumberg-Gepner-Hopkins-Rezk, the Thom spectrum functor takes based spaces over $BGL_1(R)$ to $R$-module spectra. Since the identity map $BGL_1(R)\rightarrow BGL_1(R)$ is terminal in the domain category its Thom spectrum $\mathcal{M}=MGL_1(R)$ receives a canonical map from any other such Thom spectrum. Note $MGL_1(R)$ is not always a terminal object in $R$-module spectra. $MSL_1(R)$ should satisfy a similar property among such Thom spectra which are $H\pi_0 R$-orientable.

Using the ABGHR formalism we can identify $$\mathcal{M}=\Sigma^\infty_+ EGL_1(S_p) \wedge^\mathbb{L}_{\Sigma^\infty_+ GL_1(S_p)} S_p.$$ This smash product only involves connective spectra, so $$\pi_0 \mathcal{M} \cong \pi_0(\Sigma^\infty_+ EGL_1(S_p)) \otimes_{\pi_0(\Sigma^\infty_+ GL_1(S_p))} \pi_0S_p\cong \mathbb{Z}\otimes_{\mathbb{Z}[\mathbb{Z}_p^\times]}\mathbb{Z}_p.$$ You can see this by examining the $\mathrm{Tor}$ spectral sequence, but there probably is a more elementary argument. As Neil Strickland pointed out, we can identify this torsion product (I would use the beginning of a free resolution of $\mathbb{Z}$ by free $\mathbb{Z}[\mathbb{Z}_p^\times]$-modules to calculate this): The torsion product is $$\mathbb{Z}_p/(gx-x)_{x\in \mathbb{Z}_p^\times,x\in \mathbb{Z}_p}.$$ If $p$ is odd this is 0 and if $p$ is 2 then this is $\mathbb{Z}/2$. This implies $MGL_1(S_p)\simeq *$ for $p$ odd.

Similarly $$\pi_0 MSL_1(S_p)\cong \pi_0 (\Sigma_+^\infty ESL_1 S_p)\otimes_{\pi_0 (\Sigma_+^\infty SL_1(S_p))} \pi_0(S_p)\cong \mathbb{Z}\otimes_\mathbb{Z}\mathbb{Z}_p\cong \mathbb{Z}_p.$$ If you take the subspace $G$ of $\Omega^\infty S_p$ spanned by the components $1+p\mathbb{Z}_p\subset \mathbb{Z}_p^\times$ (for $p=2$ this is $GL_1(S_2)$), then you get a Thom spectrum with $\pi_0 MG\cong \mathbb{Z}/p.$ When $p$ is $2$ the completion is unnecessary, $\pi_0 MGL_1(S)\cong \mathbb{Z}/2$.

As Neil Strickland pointed out, any $E_2$-ring spectrum, in particular any $E_\infty$ ring spectrum, $R$ with $\pi_0 R\cong \mathbb{Z}/p$ is an $H\mathbb{Z}/p$-algebra and hence a wedge of $H\mathbb{Z}/p$-modules. This implies the mod-$p$ homology is a sum of suspensions of copies of the dual Steenrod algebra. The generators of these copies of the dual Steenrod algebra are in 1-1 correspondence with generators of $\pi_*R$ as a graded $\mathbb{Z}/p$-vector space.

I would like to thank Neil Strickland for his helpful comments.

  • $\begingroup$ In the original question $GL_1(S^0_p)$ was defined to be a connected component, so the tensor product should be over $\mathbb{Z}$ and the answer is $\mathbb{Z}_p$. If we instead take $GL_1(S^0_p)$ to be the union of invertible components, then we get what you wrote, but $\mathbb{Z}_p^\times$ is acting trivially on $\mathbb{Z}$ and by multiplication on $\mathbb{Z}_p$, so the tensor product is zero. $\endgroup$ – Neil Strickland Jun 7 '14 at 7:22
  • $\begingroup$ We could instead use an intermediate space $G$ with $\pi_0(G)=1+p\mathbb{Z}_p$ and then we would get an $E_\infty$ Thom spectrum $\mathcal{M}$ with $\pi_0(\mathcal{M})=\mathbb{Z}/p$. Any such $E_\infty$ ring spectrum is known to be a wedge of suspensions of $H\mathbb{Z}/p$, so if you could calculate the homology, you would have a collapsing Adams spectral sequence for the homotopy. $\endgroup$ – Neil Strickland Jun 7 '14 at 7:27
  • $\begingroup$ Thanks Neil! I was in a bit of a rush. I will complete my response and add your comments. $\endgroup$ – Justin Noel Jun 7 '14 at 10:35
  • 1
    $\begingroup$ @NeilStrickland: This is an unpublished result of Hopkins and Mahowald. It appears as Thm 4.12 here: nullplug.org/publications/p-torsion.pdf . $\endgroup$ – Justin Noel Jun 8 '14 at 11:08
  • 1
    $\begingroup$ @Prasit: I did not fully understand your comment, but I will reiterate how my answer relates to your question. The case in your question is $MSL_1(S_p)$ whose $\pi_0$ is torsion-free. Since I suspected you meant to ask about $MGL_1(S_p)$ I included that case. That spectrum is universal in the sense that it is contractible and hence terminal. I am claiming that the homotopy groups of $MG$ are $\bZ/p$-modules and calculating them is equivalent to calculating the homology as as a comodule over the dual Steenrod algebra. I did not claim that the latter problem was easy; it is just algebraic. $\endgroup$ – Justin Noel Jun 8 '14 at 15:13

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.