9
$\begingroup$

When I thought of this question, I was sure it must have been asked before on this site, but I could't find anything. Maybe my search skills are lacking, or maybe the question is obvious and it's my math skills that are lacking. Anyway, here it goes.

For a $CW$-complex $X$ let $sw^*X$ be the subring of $H^*(X,\mathbb{F}_2)$ generated by all classes which are Stiefel-Whitney classes of some vector bundle over $X$. It is not hard to see that $sw$ is a proper subfunctor of mod 2 homology. For example (and this might be overkill) if you take the right dimensional sphere $S^n$, then by Bott periodicity, $KO(S^n)=0$, so $sw^*S^n=0$.

Now let $SW^*X$ be the subring generated by all classes which are either Stiefel-Whitney classes of some vector bundle over $X$, or suspensions or desuspensions of such classes.

$\textbf{Edit}$: Perhaps it wasn't clear from context, but I want $SW^*$ to be a functor, so I force it to be closed under pullbacks. For that reason I am puzzled by Nicholas Kuhn's suggested answer below. Also, we know in retrospect that $H\mathbb{F}_2^*X$ is a summand in $MO^*X$, and that thing is sort of tautologically built out of characteristic classes...

Is $SW^*X=H^*(X,\mathbb{F}_2)$?

I suppose the question is equivalent to something like: does the identity map of $K(\mathbb{F}_2,n)$ factor, stably, through some $BO(m)$?

$\endgroup$
9
$\begingroup$

A 1968 paper in Topology by Anderson and Hodgkin shows that $KO^*(K(\mathbb F_2, n)) = 0$ if $n \geq 2$. This implies that if $n \geq 2$, then no nonzero classes in $H^*(K(\mathbb F_2,n);\mathbb F_2)$ are SW classes. (And of course, $BO(1) = K(\mathbb F_2,1)$.)

| cite | improve this answer | |
$\endgroup$
  • 3
    $\begingroup$ Sorry, why does this show that no nonzero classes can be pulled back from some suspension of some $BO(m)$? $\endgroup$ – John Greenwood May 25 at 5:06
  • $\begingroup$ John, correct me if i'm wrong, but this follows from (i) F_2-cohomology of K(F_2,n) is polynomial algebra and (ii) multiplication in deg>0 cohomology of any suspension is trivial. But I don't see by now any trivial reason for the desuspension statement $\endgroup$ – Bad English May 25 at 15:50
  • $\begingroup$ @John, upd: for desuspension of the form \Sigma^N K(n)\to BO(\infty) it follows from Nicholas answer because after adjunction we obtain a map and as a class in KO^*(K(n)) he is equal to zero. Is there are a way to see why suspension on both sides is zero on cohomology? $\endgroup$ – Bad English May 25 at 22:43

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.