3
$\begingroup$

Let $G$ be a connected reductive group over $\mathbb{C}$ of (reductive) rank $\ell$. Let $P$ be a parabolic of $G$ and let $P=LN$ denote the Levi decomposition. Let $\mathfrak{g}, \mathfrak{p}, \mathfrak{l}$, and $\mathfrak{n}$ denote the corresponding Lie algebras. Recall that $N$ is normal in $P$; therefore, $P$ and $L$ act on $\mathfrak{n}$.

There is a nice description of $\mathbb{C}[\mathfrak{g}]^G$: it is a polynomial algebra on $\ell$ homogenous generators whose degrees are canonically defined.

As Steven points out, $\mathbb{C}[\mathfrak{n}]^L$ is trivial, so this means that $\mathbb{C}[\mathfrak{n}]^P$ is also trivial, since we have an inclusion $\mathbb{C}[\mathfrak{n}]^P\subset \mathbb{C}[\mathfrak{n}]^L$.

Question: Is there a description of $\mathbb{C}[\mathfrak{g}]^P$?

For instance, We have an obvious injective map $\mathbb{C}[\mathfrak{g}]^G \rightarrow \mathbb{C}[\mathfrak{g}]^P$. What is known about this map?

$\endgroup$
1
  • $\begingroup$ I think you will have some trivial issues coming from tori. For example, you can pick a 1-dimensional torus $H$ in the maximal torus with respect to your data (and hence in $L$) which acts positively on all positive roots, so ${\bf C}[\mathfrak{n}]^H$ consists of scalars. $\endgroup$
    – Steven Sam
    Jun 21, 2016 at 4:25

2 Answers 2

11
$\begingroup$

The map $\mathbb C[\mathfrak g]^G\to\mathbb C[\mathfrak g]^P$ is an isomorphism for trivial reasons: In any quasi-affine $G$-variety, $P$ and $G$ have the same fixed points. Just look at the orbit map of a $P$-fixed point which factors through the complete variety $G/P$. Applied to representations, this means $V^G=V^P$ for any rational $G$-module. This holds for any field.

$\endgroup$
7
  • $\begingroup$ Can you please elaborate? $\endgroup$ Jun 21, 2016 at 10:24
  • 1
    $\begingroup$ This is a nice answer. @VítTuček the answer relays on some basic facts in algebraic groups theory. 1) For every affine variety $V$ on which $G$ acts, $\mathbb{C}[V]$ is a union of $G$-invariant finite dimensional subspaces. 2) The variety $G/P$ is projective. 3) Every map from a projective variety to an affine one is constant.1+2+3) Take a $P$-fixed point in $\mathbb{C}[V]$, put it in a finite-dimensional $G$-invariant subspace, which itself is an affine variety, and consider the orbit of this point as a map from $G/P$ to it. Deduce that it is a the constant map, thus this point is $G$-fiexd. $\endgroup$
    – Uri Bader
    Jun 22, 2016 at 14:41
  • $\begingroup$ Thank you @UriBader ! I wasn't aware of the third property. Unfortunately I don't have intuition for the 1 point. Do we need $G$ to be semisimple (or reductive) in order to be able to decompose $V$ into affine orbits or is it also completely general? $\endgroup$ Jun 22, 2016 at 15:09
  • 1
    $\begingroup$ We do not decompose $V$ into orbits, we just write the space of functions on it as a union of finite-dimensional representations. This is particularly easy if $V$ is a vector space (in your example, $V=\mathfrak{g}$): $\mathbb{C}[V]$ is the union over $d$ of the spaces of polynomials of degree no more than $d$, and each of these spaces is finite-dimensional. (Of course, you can also write it as the direct sum of the spaces of homogeneous polynomials of degree $d$.) $\endgroup$
    – t3suji
    Jun 22, 2016 at 16:52
  • 1
    $\begingroup$ @t3suji: That argument works only in characteristic $0$. $\endgroup$ Jun 22, 2016 at 19:41
2
$\begingroup$

I also suggest reading Chapter 3 (Invariants of maximal unipotent subgroups) of:

Frank D. Grosshans, Algebraic homogeneous spaces and invariant theory, Lecture Notes in Mathematics, vol. 1673, Springer-Verlag, Berlin, 1997.

It gives an explicit description of $\mathbb{C}[\mathfrak{g}]^P$ under the adjoint action and of $\mathbb{C}[\mathfrak{g}]^N$ under left translation in the case when $\mathfrak{g}=\mathfrak{gl}_n$.

$\endgroup$
2
  • $\begingroup$ Thanks Vit. Did you ever find out the answer to your question regarding $\mathbb{C}[\mathfrak{n}]^N$? $\endgroup$
    – Dr. Evil
    Jun 22, 2016 at 23:23
  • $\begingroup$ @Dr.Evil No. (I assume you refer to this question: mathoverflow.net/questions/208354/…) $\endgroup$ Jul 13, 2016 at 15:37

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.