6
$\begingroup$

In the book Loop spaces, Characteristic classes and geometric quantization by Brylinski I see following result when trying to motivate geometric description of $H^3(M,\mathbb{Z})$.

$H^2(M,\mathbb{Z})$ is the group of isomorphism classes of line bundles over $M$.

I guess they mean there is a natural isomorphism.

Can some one give a rough idea of what obvious second cohomology class we can think of given a line bundle over $M$ and what line bundle can we think of given an arbitrary second cohomology class.

Intuitive comments are also welcome.

I am familiar (not the proof details) with following result:

If $G$ is a group and $M$ is a $G$-module, then the $H^2(G, A)$ is in one-one correspondence with the set of equivalence classes of extensions $E$ of $M$ by $G$, in which the action of $G$ on $M$ induced by conjugation in $E$ is the same as the action defined by the $G$- module $M$.

I am expecting some intuitive explanation that looks similar to this.

$\endgroup$
6
  • 7
    $\begingroup$ H^2(M,Z) classifies complex line bundles with respect to tensor product, by the first Chern class. One can find the isomorphism using the exponential sequence, found in any introductory text on complex algebraic geometry. $\endgroup$ Commented Aug 14, 2018 at 11:38
  • $\begingroup$ @TobiasShin Yes, that is true. Given a complex line bundle, we can associate first(and only) chern class which gives an element of $H^2(M,\mathbb{Z})$.. Can you give some reference whose proof you like it better than others.. $\endgroup$ Commented Aug 14, 2018 at 11:47
  • $\begingroup$ @TobiasShin Can you make your comment as an answer.. I can upvote.. $\endgroup$ Commented Aug 14, 2018 at 14:14
  • $\begingroup$ @users who want to close. Reason some one mentioned for their close vote is "This question does not appear to be about research level mathematics within the scope defined in the help center." I can not argue that this is a research level question. I believe this is serious question. I am ok if you think otherwise but can you wait till Tobias responds for my comment and makes his comment as an answer.. Then you can close it.. $\endgroup$ Commented Aug 14, 2018 at 16:21
  • $\begingroup$ It's possible that this was all supposed to be obvious to me, and it's certainly all familiar in broad strokes, but I upvoted the question and all of the answers. $\endgroup$ Commented Aug 17, 2018 at 19:07

3 Answers 3

11
$\begingroup$

I think $H^2(M;\mathbb{Z})$ cannot mean the de Rham cohomology group. The coefficients are wrong.

Anyway: $\mathbb{CP}^\infty$ is an amazing space. It is both a model for $K(\mathbb{Z},2)$ and a model of $BU(1)$. Homotopy classes into $K(\mathbb Z,2)$ is in bijection with $H^2(M;\mathbb{Z})$ (By pulling back the fundamental class) and homotopy classes into $BU(1)$ classify (complex) line bundles (by pulling back the tautological bundle). This works well with natural group structures. This gives the required isomorphism.

The cohomology class in $H^2(M;\mathbb{Z})$ corresponding to the complex line bundle is the first Chern class.

I can recommend Milnor Stasheff "characteristic classes" for the classification of the line bundles. I can recommend Hatcher for the classification of cohomology in terms of homotopy classes into $K(\mathbb Z,n)$.

$\endgroup$
15
  • $\begingroup$ After identifying deRham cohomology with singular cohomology... $\endgroup$ Commented Aug 14, 2018 at 11:46
  • 3
    $\begingroup$ The point is that the coefficients are wrong. De Rham cohomology is isomorphic with the singular cohomology with $\mathbb{R}$ coefficients. This does not recover the cohomology with coefficients in $\mathbb{Z}$. $\endgroup$
    – Thomas Rot
    Commented Aug 14, 2018 at 11:47
  • 1
    $\begingroup$ You can also look at Chern-Weil theory if you want to stay in the de Rham world. $\endgroup$
    – Thomas Rot
    Commented Aug 14, 2018 at 11:49
  • 1
    $\begingroup$ :D :D I am very sorry.. I do not know what I was thinking when I said that... Yes, you are right.. isomorphism is from $H^2_{dR}(M)$ to $H^2(M,\mathbb{R})$.. $\endgroup$ Commented Aug 14, 2018 at 11:49
  • $\begingroup$ Can you explain your comment “you can look at Chern-Weil theory if you want to stay in deRham world”... $\endgroup$ Commented Aug 14, 2018 at 11:56
15
$\begingroup$

For what it is worth, here is another approach, in the algebraic geometry style. By taking out the zero section, complex line bundles correspond bijectively to $\mathbb{C}^*\!$-principal bundles on $M$. Such bundles are classified by the cohomology group $H^1(M, \mathcal{O}_M^*)$, where $\mathcal{O}_M$ is the sheaf of $\mathcal{C}^{\infty}$ complex-valued functions on $M$. Now there is an exact sequence of sheaves $$0\rightarrow \mathbb{Z}\longrightarrow \mathcal{O}_M\xrightarrow{\ \mathbf{e}\ }\mathcal{O}_M^*\rightarrow 1 $$where $\mathbf{e}(f):=\exp(2\pi if)$. This gives rise to a cohomology exact sequence $$H^1(M, \mathcal{O}_M)\longrightarrow H^1(M, \mathcal{O}_M^*)\xrightarrow{\ \partial \ } H^2(M,\mathbb{Z})\longrightarrow H^2(M, \mathcal{O}_M)\,.$$But since $\mathcal{O}_M$ is a fine sheaf, its higher cohomology vanishes, and $\partial $ is an isomorphism.

$\endgroup$
3
  • $\begingroup$ I saw this approach some time back... Thanks for your answer +1.. Do you have any specific references in your thought where this is done in a better way than others?? It is done in Brylinkski's Loop spaces, Characteristic classes and Geometric quantization (I did not read that seriously though)... $\endgroup$ Commented Aug 14, 2018 at 15:00
  • 1
    $\begingroup$ That exact sequence you gave has nothing to do with "holomorphic" functions right?? Given a $\mathbb{C}$ valued smooth function $f$ on $U$ you are considering the function $exp(2\pi if)$ which is non vanishing function on $U$... That is it right? $\endgroup$ Commented Sep 12, 2018 at 15:17
  • $\begingroup$ Yes, this is right. And as I say in my answer, $\mathcal{O}_M$ is the sheaf of $\mathcal{C}^{\infty}$ complex-valued functions on $M$ — sorry if my notation is confusing. . $\endgroup$
    – abx
    Commented Sep 12, 2018 at 17:38
6
$\begingroup$

Although Milnor and Stasheff is an excellent suggestion, I was also led to wonder where one would find this result in more recent textbooks. Most ingredients are in May's "Concise introduction to algebraic topology". Specifically, May proves on page 177 that $H^2(X;\mathbb{Z})\simeq [X,K(\mathbb{Z},2)]$. Here $K(\mathbb{Z},2)$ is officially defined as $B^2(\mathbb{Z})$, where $B$ is the simplicial classifying space functor. However, on page 121 May gives a sequence of exercises about Eilenberg-MacLane spaces. As a special case, we find that whenever $Z$ is a connected CW complex with $\pi_2(Z)=\mathbb{Z}$ and $\pi_i(Z)=0$ for $i\neq 2$, we have $Z\simeq K(\mathbb{Z},2)$. In particular, it is not hard to find fibrations whose long exact sequences prove that the spaces $Z=BU(1)$ and $Z=\mathbb{C}P^\infty$ have the required homotopy groups, so they are both homotopy equivalent to $K(\mathbb{Z},2)$. May also proves on page 197 that the set of isomorphism classes of line bundles on $X$ is naturally identified with $[X,BU(1)]$.

$\endgroup$
2
  • 2
    $\begingroup$ One reference would be Husemoller's "Fibre bundles". In the 3rd edition it is theorem 3.4 in Chapter 17, p.250. $\endgroup$ Commented Aug 17, 2018 at 12:36
  • $\begingroup$ Thanks, I never saw that book carefully.. This gives enough reason to see that with care :) :) Thanks again... +1 $\endgroup$ Commented Aug 17, 2018 at 13:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .