Let $(M,g)$ be a compact riemannian 3-manifold and $\Sigma \subset M$ an embedded compact surface homeomorphic to the projective plane. Consider the application $i_\#:\pi_1(\Sigma)\to \pi_1(M)$ given by $i_{\#}([\alpha])=[i(\alpha)]$, where $i$ denotes the inclusion application of $\Sigma$ into $M$. Why $i_{\#}$ not injective implies $TM\Big|_{\Sigma}$ orientable?

  • $\begingroup$ This follows from examining the first Stiefel-Whitney class. If you don't know about those, be more direct: if this bundle were not orientable, there would be some loop you could walk around on which it changes its orientation. But all loops in $\Sigma$ are null-homotopic in $M$, so... $\endgroup$ – user98602 Jul 15 '15 at 16:46
  • $\begingroup$ @MikeMiller can you point me some reference? $\endgroup$ – Lawen Jul 16 '15 at 16:22

Let $j$ be a smooth embedding of $M$ into $\Bbb R^N$ for some large enough $N$. Now the map $j$ induces a map from $ \hat{j}: M \to \operatorname{Gr}_3(\Bbb R^N)$ where $\operatorname{Gr}_3(\Bbb R^N)$ is the space of 3-planes in $\Bbb R^n$ (see Grassmanian) and $\hat j (x)=\operatorname{Image}(dj(x))$ (the image of the derivative of $j$ at $x$). The universal cover of $\operatorname{Gr}_3(\Bbb R^N)$ is the oriented Grassmanian $\widetilde{\operatorname{Gr}_3(\Bbb R^N)}$, the space of oriented 3-planes in $\Bbb R^N$.

Now consider $\hat j \circ i$. If this map lifts to $\widetilde{\operatorname{Gr}_3(\Bbb R^N)}$, then we can pullback the orientation on each oriented 3-plane in $\Bbb R^N$ to get an orientation on $TM|_{\Sigma}$. For the map on $\pi_1$, we have $(\hat j \circ i)_*=\hat j_* \circ i_*=0$. So the $\hat j \circ i$ does lift to $\widetilde{\operatorname{Gr}_3(\Bbb R^N)}$ and $TM|_{\Sigma}$ is oriented.


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.