2
$\begingroup$

I am trying to prove the following.

Proposition. Let $\alpha:A\to B$ be a fiber bundle and let $b,b^\prime\in B$. If there's a compact connected subspace containing $b,b^\prime$, then their fibers are isomorphic.

My intuition comes from this picture from wikipedia's page on monodromy.

Let $(U_i)$ be a trivializing cover for $\alpha$. Restrict this open cover to a connected subspace containing $b,b^\prime$, and then restrict to a finite subcover $V_1,\dots ,V_n$. By connectedness, for any partition $I\amalg I^c\cong \left\{ 1,\dots ,n \right\}$ we have $\bigcup_IV_i\cap \bigcup_{I^c}V_j\neq\emptyset$, which means $V_i\cap V_j\neq\emptyset$ for some $i\in I,j\in I^c$. From here I want to reason there's a "chain" of members of the covering going from $b$ to $b^\prime$.

My reasoning is below, but it feels sloppy and vague.

  1. Take some member of the cover containing $b$ and relabel it $V_1$. Take the union of the rest of the members. These intersect. Pick some member of the union which intersects and relabel it $V_2$ and take the union $V_1\cup V_2$.
  2. Take some member of the cover intersecting $V_1\cup V_2$ and relabel it $V_3$.
  3. Repeat, repicking members if necessary, until you have a, after relabeling, a collection $V_1,\dots ,V_m$ such that $V_i\cap V_{i+1}\neq\emptyset$.
  4. Finally, apply the fact that if two trivializing opens intersect, the trivializations of the bundles over them have isomorphic fibers.

How can I better justify the existence of the "chain" $V_1,\dots ,V_m$ as in the linked picture?

$\endgroup$
4
$\begingroup$

Hint: The compactness assumption is superfluous, it is only connectedness that matters. Fix a point $b\in B$ and look at the subset $\tilde B\subset B$ consisting of $b'\in B$ for which the fiber is isomorphic to the one over $b$. Then observe directly that both $\tilde B$ and $B\setminus\tilde B$ are open in $B$.

$\endgroup$
  • $\begingroup$ Sorry, I don't understand how to "directly observe" what you say. I want to use the proposition in my post to prove that path-connected points have isomorphic fibers, so I can't use that fact (sorry if this is not relevant, it's what I understood you're doing). $\endgroup$ – Arrow Jan 4 '17 at 21:45
  • $\begingroup$ The point is that the sets $U_i$ are open in $B$ and since they are trivializing the fibers over any two points lying in the same set $B_i$ are isomorphic. Now think about what this says about $\tilde B$ and its complement and use the definition of being connected. $\endgroup$ – Andreas Cap Jan 5 '17 at 7:52

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.