关于Hatcher《Vector bundles and K-theory》中‘可定向性是平凡性首个障碍’的疑问
Great question—this is a core intuition in vector bundle topology, and Hatcher’s phrasing gets at the hierarchical nature of characteristic classes as "obstacles" to a bundle being trivial. Let’s unpack this:
1. Trivial bundles are automatically orientable
First, recall that a trivial n-dimensional vector bundle E → X is isomorphic to X × ℝⁿ. This trivial bundle comes with a global, continuous choice of orientation: just use the standard basis {e₁, e₂, ..., eₙ} for every fiber. So if a bundle is trivial, it must be orientable. The contrapositive here is key: if a bundle is non-orientable, it cannot possibly be trivial. That’s the first "block" you hit when checking for triviality.
2. Why it’s the "first" obstacle
Orientability is the most basic necessary condition for triviality, but it’s not sufficient. Think of it as the lowest bar to clear:
- If you fail this bar (non-orientable), you can stop here—your bundle is definitely not trivial.
- If you pass it (orientable), there are still higher-order obstacles to overcome before you can conclude the bundle is trivial.
Characteristic classes formalize these obstacles, and the first one in the hierarchy is the Stiefel-Whitney class w₁:
w₁(E) = 0if and only ifEis orientable.- Higher classes (like
w₂, Euler classes, or Chern classes for complex bundles) detect more subtle topological obstructions. For example, the tangent bundle ofS²is orientable (w₁ = 0), but it’s not trivial—its Euler class is non-zero, which is the next obstacle.
3. A concrete example
Take the tautological line bundle over ℝPⁿ (for n ≥ 1): this bundle is non-orientable (you can’t find a continuous global section of unit vectors that doesn’t flip orientation when you loop around ℝPⁿ). Thanks to the first obstacle, we immediately know this bundle can’t be trivial—no need to check anything else.
In short, Hatcher is saying orientability is the most fundamental requirement a bundle must satisfy to even have a shot at being trivial. It’s the first filter you apply, and characteristic classes like w₁ give us a way to measure this obstacle mathematically.
内容的提问来源于stack exchange,提问作者user514517

