Basic question on almost complex structures and Chern classes of homogeneous spaces

590 Views Asked by At

Toward the end of "Characteristic Classes and Homogeneous Spaces, III," Borel and Hirzebruch prove that given a compact Lie group $G$ and toral subgroup $T$ (no restriction on rank), one has $w(G/T) = 1$ and $p(G/T) = 1$: the Stiefel–Whitney and Pontrjagin classes are trivial. In fact they show the tangent bundle of $G/T$ is stably trivial.

They say a similar statement is valid for Chern classes in the almost complex case. I guess that means that whenever $G/T$ is almost complex, $c(G/T) = 1$. Is that right? I suppose it's clear that if the underlying real bundle of a complex vector bundle is stably trivial, then the complex vector bundle is also stably trivial; just add a higher (even) number of trivial real line bundles.

I don't know anything about almost-complexity. I do know that $U(n)/U(1)^n$ is a complete flag manifold $Fl(\mathbb{C}^n)$, and from induction on fiberings $Fl(\mathbb{C}^{n-1}) \to Fl(\mathbb{C}^n) \to \mathbb{C}P^n$, that its Euler characteristic is $n!$, meaning its top Chern class is nonzero if it is defined. So, working backwards, I guess $Fl(\mathbb{C}^n)$ isn't almost complex, though I would have thought it should be. I don't know what happens to the other Chern classes even in this case.

I do know that if $T$ is not a maximal torus, then the Euler characteristic of $G/T$ is zero. But in this case, too, I don't know anything about Chern classes. Or maybe there's nothing to know. Suppose $G$ is a compact Lie group and $T$ is a nonmaximal torus. Can $G/T$ admit an almost complex structure?

1

There are 1 best solutions below

5
On BEST ANSWER

They aren't actually referring to an almost complex structure, but to a real-stable version of it.

On the top of pg. 494: A bundle $\xi$ is called complex if $\xi \oplus 1^n$ (where $1^n$ is the $n$-dim trivial real bundle) has a complex structure for some $n$.

For example, if $\xi$ is the tangent bundle of $S^2$, then $\xi+1 = 1^3$ is trivial and so $\xi + 1^2 = 1^4$ has a (trivial) complex structure. It follows that $c(\xi) = 0$.

On the other hand, for the complex manifold $S^2$, $c(TS^2) \neq 0$.

(Note that $S^2 = U(2)/U(1)^2$, so this is directly relevant to your question.)

Finally, to answer your last question, if $G = SU(2) \times T^2$ and $U = S^1\times e\times e$, then $G/U = S^2 \times T^2$ so has a complex structure. I don't know of any less trivial examples, however.