14
$\begingroup$

A complex vector bundle of rank n can be viewed as a real vector bundle of rank 2n. From nLab, we have that the second Stiefel-Whitney class of the real vector bundle is given by the first Chern class of the complex vector bundle mod 2: $w_2=c_1$ mod 2. Do we have similar relations for other Stiefel-Whitney classes, such as $w_{2n}=c_n$ mod 2?

$\endgroup$
  • $\begingroup$ possible duplicate of Chern Classes and Stiefel-Whitney Classes $\endgroup$ – user122283 May 9 '14 at 2:34
  • $\begingroup$ @SanathDevalapurkar I know that question. But it does not mention if $w_{2n}=c_n$ mod 2. $\endgroup$ – Xiao-Gang Wen May 9 '14 at 2:42
  • 3
    $\begingroup$ I am confused with the close vote. I think this is an excellent question and should remain open. $\endgroup$ – Bombyx mori May 9 '14 at 2:46
  • $\begingroup$ @user99680: This is the same link others already provided. $\endgroup$ – Bombyx mori May 9 '14 at 2:51
  • $\begingroup$ @Bombyx mori: Sorry, just deleted it. $\endgroup$ – user99680 May 9 '14 at 2:53
10
$\begingroup$

The following is from Lecture notes of Professor Farrell. A hopefully working link is here(page 95): https://www.dropbox.com/s/80n4wd6xctpe6yr/Characteristic%20Classes%20%28Sparkie%20%E7%9A%84%E5%86%B2%E7%AA%81%E5%89%AF%E6%9C%AC%202014-02-19%29.pdf

We now reach proposition 7. Let $E$ be a complex vector bundle, then the mod 2 reduction of the total Chern class of $E$ is the total Stiefel-Whitney class of $E$. Since there is no Chern classes in odd dimensions, we have there is no Stiefel-Whitney classes in odd dimension as well.

$\textbf{Proof}$

Suppose $E$ is a line bundle. Then we know $w_{0}(L)=1, w_{1}(L)=0,w_{2}(L)=e_{\mathbb{Z}_{2}}(L)=\phi(e(L))$. On the other hand we know $c_{0}(L)=1, c_{1}(L)=e(L)$. So this verified it for line bundles.

For sum of line bundles we have $$ E=\oplus^{n}_{i=1}L_{i} $$ So the total Chern class is $$ c(E)=c(L_{1}) \cup \cdots \cup c(L_{n})\mapsto_{\phi}\omega(E)=\omega(L_{1}) \cup \cdots \cup \omega(L_{n}) $$ We are now going to use splitting principle. But there is a subtle point. We now discuss it.

By the splitting principle there exist $$ f:\mathcal{B}\rightarrow B $$ such that $$ f^{*}(E)=\oplus_{i=1}^{n}L_{i} $$ and $$ f^{*}:H^{*}(B,R)\rightarrow H^{*}(\mathcal{B},R) $$ is monic.

Therefore by naturality and the fact $f^{*}$ is monic with respect to $\mathbb{Z}_{2}$ coefficients. $$ \phi(f^{*}(c(E)))=f^{*}(\phi(c(E)))=f^{*}(\omega(E)) $$

$\endgroup$
  • $\begingroup$ I cannot access the file. Dropbox says it is a "disabled link." $\endgroup$ – Potato May 9 '14 at 3:19
  • $\begingroup$ I copied the latex file at here, but I need to fix my use of macros. Somehow dropbox blocked my file link. $\endgroup$ – Bombyx mori May 9 '14 at 3:22
  • $\begingroup$ @Potato: I fixed the macros and the link. $\endgroup$ – Bombyx mori May 9 '14 at 3:24
  • $\begingroup$ It works now. Thanks! $\endgroup$ – Potato May 9 '14 at 3:25
  • $\begingroup$ @Potato: Thanks for pointing it out. Now I found all lecture notes I put in my personal website are dead due to the same problem...:( $\endgroup$ – Bombyx mori May 9 '14 at 3:26

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.