The singular cohomology of the loop space $\Omega S^3$ of the 3-sphere is a divided power algebra $\Gamma_{\mathbb Z}[s]$ on one generator $s$ of degree 2, so the rational cohomology is a the polynomial ring $\mathbb{Q}[s]$. Likewise, the singular cohomology of infinite complex projective space $\mathbb C \mathrm P^\infty = BS^1$ is a polynomial ring $\mathbb{Z}[t]$ on one generator $t$ of degree 2, so the rational cohomology is $\mathbb{Q}[t]$.

Since a symmetric algebra is a free CGA, any lifting of $s$ and $t$ to cocycles in CDGA models for the rational cohomology of these spaces induces a quasi-isomorphism between the cohomology ring, viewed as a CDGA with trivial differential, and the model, so these spaces are both formal. Moreover, both spaces are simply-connected. Thus any map between the two inducing an isomorphism in rational cohomology should be a rational homotopy equivalence.

Does there exist such a map?

P.S.: I guess the same question is equally (un)reasonable for $\Omega S^5$ and $\mathbb H \mathrm P^\infty$. What about this pair?


There is a map $S^3\to K(\mathbb{Z},3)$ inducing an isomorphism on $\pi_3$. Looping this map, you get a map $\Omega S^3\to K(\mathbb{Z},2)=\mathbb{C}P^\infty$ inducing an isomorphism on $\pi_2$. It follows easily that this map is a rational equivalence.

For $\Omega S^5$ and $\mathbb{H}P^\infty$, I don't know whether you can get a map $\Omega S^5\to \mathbb{H}P^\infty$ which is a rational equivalence (the above approach doesn't work because you can't deloop $\mathbb{H}P^\infty$). However, there is no rational equivalence $\mathbb{H}P^\infty\to\Omega S^5$. For if you had such a map, then it would be nonzero on $H^4(-,\mathbb{Z})$, and so looking at the induced map on the whole integral cohomology rings, you would get a ring-homomorphism $\Gamma_\mathbb{Z}(s)\to\mathbb{Z}[t]$ sending $s$ to $nt$ for some nonzero integer $n$. But if $p$ is a prime not dividing $n$, such a homomorphism would have nowhere to send $s^p/p!$. (Alternatively, you could use Steenrod powers: a rational equivalence $\mathbb{H}P^\infty\to\Omega S^5$ would yield a map $\Sigma\mathbb{H}P^\infty\to S^5$ whose induced map on $H^5(-,\mathbb{Z})$ is multiplication by some nonzero integer $n$; reducing mod a prime not dividing $n$ and considering the Steenrod power $P^2$ then yields a contradiction). Note, however, that they can easily be connected by a zigzag of rational equivalences: for instance, a map from either of them to $K(\mathbb{Z},4)$ classifying a generator of $H^4$ is a rational equivalence.

  • $\begingroup$ 1. Is there anything more geometric to be said about the map $\Omega S^3 \to \mathbb{C}\mathrm{P}^\infty$ in question? $\endgroup$ – jdc Dec 7 '15 at 2:51
  • $\begingroup$ 2. Where does the sufficient largeness of the prime come in? $\endgroup$ – jdc Dec 7 '15 at 3:04
  • $\begingroup$ 1. Here is one neat interpretation of it, which you may or may not consider "geometric". By a theorem of James, the space $\Omega S^3=\Omega\Sigma \mathbb{C}P^1$ can be identified (up to homotopy) with the free topological monoid on $\mathbb{C}P^1$ (as a pointed space, with the basepoint becoming the unit of the monoid), and $\mathbb{C}P^\infty$ can be identified with the free topological commutative monoid on $\mathbb{C}P^1$. This map $\Omega\Sigma \mathbb{C}P^1\to\mathbb{C}P^\infty$ is then just the canonical "abelianization" map. $\endgroup$ – Eric Wofsey Dec 7 '15 at 3:06
  • $\begingroup$ 2. You just need to choose a prime $p$ such that the induced map on $H^5(-,\mathbb{Z}/p)$ is nonzero. $\endgroup$ – Eric Wofsey Dec 7 '15 at 3:07
  • $\begingroup$ 3. I'm obviously missing something kind of major here, so I'd better ask. The cohomology of $\mathbb{H}P^\infty = BS^3$ should be a polynomial ring on one variable of degree four, and that of $S^4$ should be a truncated polynomial ring inhabited only in degrees zero and four. Both these spaces are simply-connected. How could it be that a map between these spaces could induce an isomorphism in rational cohomology? $\endgroup$ – jdc Dec 7 '15 at 3:09

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.