I am reading Hatcher's book on Vector Bundles and K-Theory(freely available online). I am reading about clutching functions in the first chapter and there is a step in the proof of Proposition 1.11 (p.23) that I don't understand. In order to be concrete I introduce a capture:

I summarize the steps of the proof (as I understand them):

  1. In order to show there is a bijection we will construct and inverse $\psi$ to $\phi$. Remember that $p\colon E\to \mathbb{S}^k$ is our vector bundle.
  2. Given two trivializations (given by the two hemispheres) $h_+\colon E_+ \to D^k_+\times \mathbb{C}^n$ and $h_-\colon E_- \to D^k_-\times \mathbb{C}^n$, we define $$\psi(E)\in [\mathbb{S}^{k-1},GL_n(\mathbb{C})]$$ to be the homotopy class of the map $\mathbb{S}^{k-1}\to GL_n(\mathbb{C})$ defined by $${h_+}\circ {h_-}_{|\partial D^k_- \times \mathbb{C}}={h_+}\circ {h_-}_{|\mathbb{S}^{k-1} \times \mathbb{C}}$$
  3. We have to prove that the map $\psi$ is well-defined and that $\psi$ and $\phi$ are inverses to each other.

My problems arise when trying to prove that $\psi$ is well-defined. So let's begin and I will point out the step I don't understand:

  1. Given other trivializations over the hemispheres $h'_+\colon E_+ \to D^k_+\times \mathbb{C}^n$ and $h'_-\colon E_- \to D^k_-\times \mathbb{C}^n$, we define $$\psi(E)\in [\mathbb{S}^{k-1},GL_n(\mathbb{C})]$$ to be the homotopy class of the map $$\mathbb{S}^{k-1}\to GL_n(\mathbb{C})$$ defined by $${h'_+}\circ {h'_-}_{|\mathbb{S}^{k-1} \times \mathbb{C}}$$ We have to check that the two maps: $${h_+}\circ {h_-}_{|\mathbb{S}^{k-1} \times \mathbb{C}}$$ $${h'_+}\circ {h'_-}_{|\mathbb{S}^{k-1} \times \mathbb{C}}$$

are homotopic. And here is where I lost Hatcher's argument. I see that both $h_+\colon E_+ \to D^k_+\times \mathbb{C}^n$ and $h_-\colon E_- \to D^k_-\times \mathbb{C}^n$ are unique up to homotopy. But in order to prove that $${h_+}\circ {h_-}_{|\mathbb{S}^{k-1} \times \mathbb{C}}$$ is unique up to homotopy I think I need to show that the restrictions of the trivializations:

$${h_+}_{|\partial D^k_+ \times \mathbb{C}}$$ $${h_-}_{|\partial D^k_- \times \mathbb{C}}$$

are unique up to homotopy and here the argument provided does not work since the domains of the restriction are not contractible.

I am sure that there is something here I don't understand properly. So any clarification, explanation or help would be appreciated. Thanks!


It is explained here. The key step is to use the notion of equivalent cocycles.


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.