1
$\begingroup$

Hi,

Suppose $G$ is a reductive group over an algebraiclly closed field $k$ (suppose $k$ of char zero if you want at first). Let $X$ be its flag variety.

Question: What is the moduli problem that $X$ represents?

EDIT (to clarify): What is the functor of points of $X$?

Thanks!

$\endgroup$
1
  • 1
    $\begingroup$ Can you clarify the meaning of "the" moduli problem here? In any case, $X$ can be identified with the set of all Borel subgroups of $G$. $\endgroup$ Apr 22, 2012 at 17:23

1 Answer 1

5
$\begingroup$

Let $X$ be a space and $H$ be a group such that $X\rightarrow X/H$ is a principal bundle. Then $Hom(Y,X/H)$ is in bijection with $H$ torsors over $Y$ equipped with an equivariant map from their total space to $X$. So maps from $Y$ into the flag variety $G/P$ are in bijection with $P$ torsors on $Y$ equipped with a $P$-equivariant map to $G$. Or equally $P$ "subtorsors" of the trivial torsor $G\times Y$.

For example if we take $G=GL_{n}(\mathbb C)$ the data of a $P$ "subtorsor" of $G$ is equivalent to giving a flag (whose type is determined by $P$) of sub-bundles inside the trivial n-dimensional bundle on $Y$.

If for example $P$ consist of all matrices whose first column is zero everywhere except in the upper left corner, we have $G/P=\mathbb P^{n-1}$. Our description says maps into $\mathbb P^{n-1}$ are the same thing as linebundles inside of $Y\times \mathbb C^n$.

Taking the dual of such a linebundle and restricting the coordinate functions of $\mathbb C^n$ to it gives the usual universal property of $\mathbb P^{n-1}$.

All this should work over any field.

$\endgroup$
1
  • $\begingroup$ I think $Hom(Y,X)$ should be $Hom(Y,X/H)$ in your second sentence. $\endgroup$
    – S. Carnahan
    Apr 23, 2012 at 2:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.