7
$\begingroup$

I am looking for an example of an oriented rank 5 (or lower) real vector bundle $V$ over an oriented manifold such that the cup product $w_2(V) w_3(V)$ of Stiefel-Whitney classes does not vanish. It would be best if the manifold had dimension 7 or lower.

$\endgroup$
10
$\begingroup$

As far as I know the Wu manifold $X=SU(3)/SO(3)$ is orientable and has mod 2 cohomology ring $H^*(X;\mathbb{Z}_2)=\Lambda(\omega_2(X),\omega_3(X))$. Thus $\omega_2(X)\cdot\omega_3(X)\neq 0$, and in fact generates $H^5(X;\mathbb{Z}_2)$.

$\endgroup$
  • $\begingroup$ Great thanks! Do you have a handy reference where this is proven? $\endgroup$ – Samuel Monnier Sep 11 '16 at 17:17
  • 4
    $\begingroup$ For a closed oriented $5$-manifold $M$, the Stiefel-Whitney number $\langle w_2 w_3, [M] \rangle$ can be easily deduced from the isomorphism type of the torsion subgroup of $H_2(M;\mathbb{Z})$. See the second page of the paper [G. Lusztig, J. Milnor & F. Peterson, Semi-characteristics and cobordism. Topology 8 (1969) 357--359], where the example of Wu's manifold is mentioned. $\endgroup$ – Gwenael Massuyeau Sep 11 '16 at 19:50
  • $\begingroup$ Thanks, this helped. I also found the answer to this MO question useful: mathoverflow.net/questions/164714/… $\endgroup$ – Samuel Monnier Sep 13 '16 at 11:28
2
$\begingroup$

I think you can take the manifold to be $M = \mathbb{RP}^5$ and the vector bundle $V$ to be the direct sum of $3$ copies of the tautological line bundle $\gamma$ on $M$. Let $x = w_1(\gamma)\in H^1(M,\mathbb{Z}/2\mathbb{Z})$. Then by the Whitney product formula you get $w_2(V) = x^2$ and $w_3(V) = x^3$, so $w_2(V)w_3(V) = x^5$. But this is non-zero in the mod 2 cohomology of $\mathbb{RP}^5$, which is, as an algebra, isomorphic to $\mathbb{Z}/2\mathbb{Z}[x]/(x^6)$.

$\endgroup$
  • 1
    $\begingroup$ Here the vector bundle is not orientable. $\endgroup$ – Mark Grant Sep 11 '16 at 15:50
  • $\begingroup$ Thanks! But I was asking for an orientable bundle, indeed. $\endgroup$ – Samuel Monnier Sep 11 '16 at 17:18
  • $\begingroup$ Ah, missed that part. $\endgroup$ – Tony Sep 11 '16 at 18:20

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.