2
$\begingroup$

Suppose that $X$ is a nice compact manifold such that its reduced real $K$-group $\tilde{K}\mathcal{O}(X)$ has a unique stably equivalent class of orientable bundles, i.e, $\ker(\omega_{1})$ is the identity element in $\tilde{K}\mathcal{O}(X)$ where $\omega_{1}:\tilde{K}\mathcal{O}(X)\to H^{1}(X,\mathbb{Z}_{2})$ is the first Stiefel-Whitney class. Here, we generalize the notion of orientability to stably equivalent classes of vector bundles stating that a stably equivalent class of bundles $[E]$ is orientable if $\omega_{1}([E])=0$. Under this hypothesis, can we deduce that $\omega_{1}$ classifies all stably equivalent vector bundles? In other words, is $\omega_{1}:\tilde{K}\mathcal{O}(X)\to H^{1}(X,\mathbb{Z}_{2})$ an isomorphism? Indeed, since $\omega_{1}$ restricted to line bundles is an isomorphism, it is onto and its injectivity follows from the hypothesis. Is this argumentation correct? If so, which is the role of higher Stiefel-Whitney classes in this setting?

$\endgroup$

1 Answer 1

2
$\begingroup$

Yes, under these conditions $w_1 : \widetilde{KO}(X) \to H^1(X; \mathbb{Z}_2)$ is an isomorphism. Here's another way to see it.

Suppose $[E], [F] \in \widetilde{KO}(X)$ satisfy $w_1([E]) = w_1([F]) = \alpha \neq 0$. Let $L$ denote the real line bundle with $w_1(L) = \alpha$. Then $w_1([E\oplus L]) = w_1([F\oplus L]) = 0$ so by assumption $[E\oplus L] = [F\oplus L]$. Therefore, there are non-negative integers $m, n$ with $E\oplus L\oplus\varepsilon^m \cong F\oplus L\oplus\varepsilon^n$. As $X$ is compact and Hausdorff, there is a bundle $L'$ with $L\oplus L' \cong \varepsilon^k$ for some $k \geq 2$. Therefore $E\oplus L\oplus\varepsilon^m\oplus L' \cong E\oplus\varepsilon^{n+k}$ and $F\oplus L\oplus\varepsilon^n\oplus L' \cong F\oplus\varepsilon^{m+k}$, so $[E] = [F]$.

By your assumption, every orientable bundle is stably trivial and hence has trivial Steifel-Whitney classes. So if $w_1([E]) = \alpha \neq 0$ and $L$ is as above, then $w_1([E\oplus L]) = 0$ and hence $w_i([E\oplus L]) = 0$. On the other hand $w_i([E\oplus L]) = w_i([E]) + w_{i-1}([E])w_1([L])$, so $w_i([E]) = w_{i-1}([E])w_1([L]) = w_{i-1}([E])\alpha$ and therefore $w_i([E]) = \alpha^i$. That is, the Stiefel-Whitney classes of $[E]$ are completely determined by $w_1([E])$.

$\endgroup$
1
  • $\begingroup$ Very nice answer @Michael Albanese! You have helped me a lot! $\endgroup$ Mar 20 at 13:39

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.