2
$\begingroup$

This question seems like a pretty straightforward generalization of a result from vector bundles but its been on MSE for over a week with no answers so I'm reposting https://math.stackexchange.com/questions/4425410/fiber-bundle-orientability-vs-manifold-orientability here:

I read this question about vector bundles

https://math.stackexchange.com/questions/50809/bundle-orientability-vs-manifold-orientability

In the answer to this question the last sentence states the following (I think fairly well known) result about vector bundles

"Let E be a vector bundle over M. Consider the statements (i) M is orientable as a manifold (ii) E is orientable as a manifold (iii) E is orientable as a vector bundle. Any two of the statements being true will imply the third."

I am curious if this statement generalizes to all fiber bundles with orientable fiber.

In other words, is it true that: Let $ F \to E \to M $ be a fiber bundle over M, with $ F $ an orientable manifold. Consider the statements (i) M is orientable as a manifold (ii) E is orientable as a manifold (iii) E is orientable as a fiber bundle. Any two of the statements being true will imply the third.

If not, is a similar 2-out-of-3 theorem true for sphere bundles or some other more restricted class of fiber bundles?

Some stuff that I tried (focused on circle bundles in which case the bundle being orientable is equivalent to it being $ U_1 $ principal see Is every orientable circle bundle principal?)(Note that this is a complete list of all circle bundles over $ S^1,S^2,T^2,K^2,\mathbb{R}P^2 $):

Base $ S^1 $:

  • All three: $ S^1 \to T^2 \to S^1 $ (in general any trivial bundle $ S^1 \times M $ for any orientable manifold $ M $ has base, bundle and total space all orientable)

  • Only base orientable: $ S^1 \to K^2 \to S^1 $ where $ K^2 $ is the Klein bottle.

Base $ S^2 $:

  • All three: The lens spaces $ S^1 \to L_{n,1} \to S^2 $, where $ n $ is the euler class of the bundle, are all orientable. I believe they are also $ U_1 $ principal (certainly for $ n=0,1,2 $ they are principal since they are $ S^1 \times S^2, S^3\cong SU_2,\mathbb{R}P^3\cong SO_3(\mathbb{R}) $ respectively). $ E^1 \times S^2 $ geometry for $ n=0 $, $ S^3 $ geometry otherwise.

Base $ T^2 $:

  • All three: The circle bundles $ S^1 \to MT(\begin{bmatrix} 1 & r \\ 0 & 1 \end{bmatrix}) \to T^2 $ where $ MT(\begin{bmatrix} 1 & r \\ 0 & 1 \end{bmatrix}) $ denotes the mapping torus of $ T^2 $ corresponding to the mapping class $ \begin{bmatrix} 1 & r \\ 0 & 1 \end{bmatrix} $, which is the $ r $th power of the Dehn twist $ \begin{bmatrix} 1 & 1 \\ 0 & 1 \end{bmatrix} $. For $ r=0 $ this is $ T^3 $ and admits $ E^3 $ (flat) geometry while for $ r \neq 0 $ these are the nilmanifolds $ N_r $ described in https://math.stackexchange.com/questions/4367670/is-every-nil-manifold-a-nilmanifold and they admit Nil geometry. $ E^3 $ geometry for $ r=0 $ otherwise Nil geometry.

  • Only base orientable: $ S^1 \rtimes_b T^2 $ two of the four flat compact non orientable three manifolds. For $ b=0 $ this is $ S^1 \times K^2 $ with first homology $ \mathbb{Z}^2 \times C_2 $, for $ b=1 $ this is the mapping torus of the Dehn twist diffeomorphism of $ K^2 $ with first homology $ \mathbb{Z}^2 $. The total space is not orientable. These coincide with the two $ U_1 $ principal bundles over $ K^2 $. $ E^3 $ geometry.

Base $ \mathbb{R}P^2 $:

  • Only bundle orientable: $ S^1 \to S^1 \times \mathbb{R}P^2 \to \mathbb{R}P^2 $. (in general any trivial bundle $ S^1 \times M $ for any non orientable manifold $ M $ has only the bundle orientable). $ E^1 \times S^2 $ geometry.

  • Only bundle orientable: $ S^1 \to (S^2 \times S^1)/(-1,-1) \mathbb{R}P^2 $. This is the mapping torus of the antipodal map of $ S^2 $. It is the unique nontrivial $ U_1 $ principal bundle over $ \mathbb{R}P^2 $. $ E^1 \times S^2 $ geometry.

  • Only total space orientable: $ S^1 \to P_{4n,1} \to \mathbb{R}P^2 $ where $ P_{4n,1} $ is the standard prism manifold with $ 4n $ element dicyclic fundamental group. $ S^3 $ geometry.

  • Only total space orientable: $ S^1 \to UT(\mathbb{R}P^2) \cong L_{4,1} \to \mathbb{R}P^2 $, the unit tangent bundle of $ \mathbb{R}P^2 $. $ S^3 $ geometry.

  • Only total space orientable: $ S^1 \to \mathbb{R}P^3 \# \mathbb{R}P^3 \to \mathbb{R}P^2 $. $ E^1 \times S^2 $ geometry.

Base $ K^2 $:

  • Only the bundle is orientable: The two principal $ U_1 $ bundles over $ K^2 $ coincide with the two non principal $ S^1 $ bundles over $ T^2 $. These are two of the four non orientable compact flat three manifolds they can also be viewed as two of the four mapping tori of $ K^2 $. $ E^3 $ geometry.

  • none of 3: The other two of the four non orientable compact flat three manifolds, they can also be viewed as the other two of the four mapping tori of $ K^2 $. They are non principal bundles $ S^1 \to S^1 \rtimes_b K^2 \to K^2 $ both with non orientable total space. For $ b=0 $ this is the mapping torus of the Y homoeomorphism of $ K^2 $, it has first homology $ \mathbb{Z}^2 \times C_2 \times C_2 $. For $ b=1 $ this is the mapping torus of $ K^2 $ for the mapping class corresponding to the combination of a Dehn twist and a Y homoemorphism. It has first homology $ \mathbb{Z}^2 \times C_4 $. $ E^3 $ geometry.

  • Only total space orientable: The circle bundles $ S^1 \to MT(\begin{bmatrix} -1 & -r \\ 0 & -1 \end{bmatrix}) \to K^2 $ where $ MT(\begin{bmatrix} -1 & -r \\ 0 & -1 \end{bmatrix}) $ denotes the mapping torus of $ T^2 $ corresponding to the mapping class $ \begin{bmatrix} -1 & -r \\ 0 & -1 \end{bmatrix} $. These manifolds are double covered by $ MT(\begin{bmatrix} 1 & 2r \\ 0 & 1 \end{bmatrix}) $. For $ r=0 $ this is the unit tangent bundle of the Klein $ UT(K^2) $, which admits $ E^3 $(flat) geometry, while for $ r \neq 0 $ these admit Nil geometry. $ E^3 $ geometry for $ r=0 $, Nil geometry otherwise.

$\endgroup$
5
  • 4
    $\begingroup$ Generally a question not getting an answer at MSE isn't a reason to post it here. Most arguments that you can use for the tangent bundle work for arbitrary bundles. I would suggest the argument using lifts of loops in the base to paths in the frame bundle. $\endgroup$ Commented Apr 22, 2022 at 18:42
  • $\begingroup$ How is "orientable as a fibre bundle" defined as distinct from the total space's being an orientable manifold? $\endgroup$ Commented Apr 22, 2022 at 21:13
  • 1
    $\begingroup$ @DanielAsimov: he's talking about the clutching maps being orientation preserving. i.e. the bundle corresponds to a map to the classifying space of the orientation-preserving automorphisms of the fiber. $\endgroup$ Commented Apr 22, 2022 at 23:12
  • $\begingroup$ @RyanBudney I respect your opinion I'll delete the question if you want. But just to be clear you're saying that this two out of three theorem for fiber bundles is (from your perspective) obviously true? $\endgroup$ Commented Apr 22, 2022 at 23:28
  • 4
    $\begingroup$ Right, the structure of the argument is the same. After some linear algebra, it boils down to the statement $ab=c$, if two of the three symbols are positive, then the other one is. The key issue is when you think of the frame bundle of the total space, you can homotope the frame to be of the form "frame for the fiber, frame for the base". $\endgroup$ Commented Apr 22, 2022 at 23:45

0

You must log in to answer this question.