I posted this question some days ago at math.stackexchange, but didn't receive an answer.

I have two questions:

  • I wonder whether anyone has taken the time to compute $KO_*(\mathbb{R}P^\infty)$?

The standard tools to compute these Groups in the complex case rest on the requirement for the cohomology theories $E$ to be complex orientable. Naturally, I looked up whether something like real orientable cohomology theories exist in the literature but found out that $KO$ is not real-oriented. Anyway, there is a way to "circumvent" Snaiths theorem for the spectrum $K$ if one is only interestd in the algebra of cooperations, in the sense that one can show that $$K_*(\mathbb{C}P^\infty) \xrightarrow{i_*} K_*K$$ is an injection of rings, where $i$ is induced from the inclusion $\mathbb{C}P^\infty \simeq BU(1) \hookrightarrow BU$. In fact, one only needs to invert the Bott element $\beta$ to turn it into an isomorphism, so it is a localization. This can be concluded from

Robert M. Switzer. Algebraic topology—homotopy and homology. Classics in Mathematics. Springer-Verlag, Berlin, 2002. Reprint of the 1975 original [Springer, New York; MR0385836 (52 #6695)].

17.33, which states

$K_*K$ is generated over $\mathbb{Z}[u,u^{-1},v^{-1}]$ by the polynomials $\{p_1,p_2,\ldots\}$.

By a process reminiscent of

J. F. Adams. Stable homotopy and generalised homology. University of Chicago Press, Chicago, Ill.-London, 1974. Chicago Lectures in Mathematics.

p. 44 we can describe the relations of the generators of $\beta_i$ of $K_*(\mathbb{C}P^\infty) = K_* \{\beta_0 , \beta_1 , \ldots \}$ such that

$$\beta_1\beta_n = n \beta_n +(n+1)\beta_{n+1}$$

and by setting

$$\binom{x}{i} = \frac{x(x-1)\cdots (x-(i-1))}{i!} \in \mathbb{Q}[x]$$

with $x:=\beta_1$ one can see that

$K_*(\mathbb{C}P^\infty)\otimes \mathbb{Q}$ is the polynomial algebra $K_* \otimes \mathbb{Q}[x]$ over $K_*\otimes \mathbb{Q} = \mathbb{Q} [t,t^{-1}]$ and $K_*(\mathbb{C}P^\infty)$ can be identified with the subalgebra of $K_* \otimes \mathbb{Q}[x]$ generated by $\binom{x}{i}$ for $i=0,1,2, \ldots$,

where we set $\binom{x}{0}=1$.

  • While snaiths theorem works on the spectrum level and the aforementioned result follows, I wonder whether a similar result holds in the real case, i.e. $$ KO_*(\mathbb{R}P^\infty)[\alpha^{-1}] \cong KO_*KO$$ for some element $\alpha \in KO_*(\mathbb{R}P^\infty)$?
  • 2
    $\begingroup$ Regarding Snaith's theorem: the direct analogue of Snaith's construction, using the generator you might expect from $\Bbb{RP}^\infty$ by analogy, will result in a ring spectrum $R$ that has an orientation for real vector bundles. This means that it can't quite be $KO$. For example, it would mean that it would have a map of ring spectra $MO \to R$. Because of known structure about $MO$, this would force $R$ to be a sum of shifts of the Eilenberg-Mac Lane spectrum $H\Bbb{F}_2$. $\endgroup$ Jan 28, 2020 at 19:21
  • 2
    $\begingroup$ The $p$–completion of $KO_* KO$ is known and easy to state: it is the algebra of continuous functions from the $p$–adic group $(\mathbb Z_p^\times / C_2)$ to $(KO^\wedge_p)_*$. Given the size of domain and codomain, I believe this to be too hopelessly large to be caught by inverting an element in (the $p$–completion of) Bob's answer. $\endgroup$ Jan 28, 2020 at 22:23

1 Answer 1


There are many ways to do this. One elementary approach is to use the Adams spectral sequence

$Ext_A(H^*ko \wedge RP^\infty, F_2) \cong Ext_{A(1)}(H^*RP^\infty,F_2) \Rightarrow ko_*(RP^\infty) $

and invert the Bott map. An $A(1)$ resolution giving the answer (since $E_2 = E_\infty$) can be seen at the bottom of the page


We see the groups, starting with $KO_0$, are $0, Z/2, Z/2, Z/2^\infty, 0, 0, 0, Z/2^\infty$ and repeat by Bott periodicity, with the evident action of the coefficients $KO_*$. (This action follows from the homological algebra of the $Ext$ calculation.)

  • 4
    $\begingroup$ A cute trick for computing $ko_*(\mathbb{R}P^{\infty})$ is explained in section 4 of Mahowald-Milgram's "Operations which detect..." (academic.oup.com/qjmath/article/27/4/415/1531880) $\endgroup$ Jan 28, 2020 at 14:25
  • 3
    $\begingroup$ Right: this uses the fact that the cofiber of the reduced transfer $RP^\infty \to S$ has cohomology which is 'made up of' $A(1)//A(0)$'s, so that smashing it with $ko$ gives a wedge of $HZ$'s in degrees 0,4,8, .... As a $ko$-module, however, the Bott map and hence also the 4 dim class in $ko_*$ act nontrivially, so the $Z$ summands in $ko_*$ map by an Adams filtration preserving map to this cofiber. The $\eta$'s in the kernel, and the finite quotients in degrees 4n shifted down one, then assemble to give the answer. $\endgroup$ Jan 28, 2020 at 15:10
  • 2
    $\begingroup$ There is one extension to determine, from filtration 1 to filtration 2, easily shown to be non-zero, and then Bott periodicity says all the others are non-zero as well. $\endgroup$ Jan 28, 2020 at 15:11

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.