MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Birkhoff decomposition vanishing of the Chern numbers of the holomorphic line bundles of the Birkhoff-Grothendieck decomposition, is some statement I read off in One of Connes papers. Without going into detail with how it comes up in the paper, I am curious as to how this connection comes about, more or less how to consider formulating a proof about this. I must say I have a very tiny math background(struggling physics dude here) but I mean I can handle math lingo, but I need a gentle response. I am sorry if this is too trivial a question, and apologize in advance. .

share|cite|improve this question
Are you asking about the result of Birkhoff-Grothendieck (and, earlier, Del Pezzo - Bertini if one properly interprets the classification of varieties of minimal degree) that every holomorphic vector bundle over the Riemann sphere is isomorphic to a direct sum of holomorphic line bundles? What precisely would you like to know about this result? – Jason Starr Jan 2 '14 at 1:22
I would love to know an elementary proof of the result, one that I can sort of understand – user39066 Jan 2 '14 at 2:04
What is your background? This can be proved complex analytically, but it can also be proved algebraically. – Jason Starr Jan 2 '14 at 2:22
Well, to be honest complex analytically I would understand better, but then again an algebraic proof would work for the paper I am kind of writing since there is quite a bit of stuff with hopf algebra already in it. To really answer your question I am mostly self taught in both. Can you proof in one then sketch out the procedure in the other perhaps. I know I am asking for a lot here, . . . – user39066 Jan 2 '14 at 2:35
An elementary proof can be found in Lemma 7 of Atiyah, M. F. Vector bundles over an elliptic curve. Proc. London Math. Soc. (3) 7 1957 414–452 (this is a variant of the answer by abx). – Damian Rössler Jan 2 '14 at 16:01
up vote 2 down vote accepted

The proof is straightforward if you accept some standard results on cohomology over $\mathbb{P}^1$. Let me try. We want to prove that any vector bundle $E$ on $\mathbb{P}^1$ is a direct sum of line bundles $\mathcal{O}(k)$. The proof is by induction on the rank of $E$. Replacing $E$ by $E(k)$ for some integer $k$ we may assume $H^0(E)\neq 0$, $H^0(E(-1))=0$. Let $s\neq 0$ in $H^0(E)$. Then $s$ is nonzero at every point (otherwise $s$ would give a nonzero section of $E(-1)$), so there is an extension $$0\rightarrow \mathcal{O}\rightarrow E \rightarrow F\rightarrow 0\ .$$
By induction we have $F=\sum \mathcal{O}(k_i)$. Tensor product with $\mathcal{O}(-1)$ + cohomology exact sequence + $H^1(\mathbb{P}^1,\mathcal{O}(-1))=0$ gives $H^0(\mathbb{P}^1,F(-1))=0$, hence $k_i\leq 0$ for each $i$.

Now you have to accept that such extensions are classified by the vector space $H^1(\underline{\mathrm{Hom}}(F,\mathcal{O}))=H^1(F^*)$. Since $H^1(\mathcal{O}(k))=0$ for $k\geq 0$, this is zero, hence the extension is trivial and you are done.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.