home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!unipalm!uknet!gdt!masfeb
- From: masfeb@gdr.bath.ac.uk (F E Burstall)
- Newsgroups: sci.math
- Subject: Re: Spectum of a Lie group ...
- Message-ID: <1992Oct14.193148.26001@gdr.bath.ac.uk>
- Date: 14 Oct 92 19:31:48 GMT
- References: <1992Oct7.202037.14510@midway.uchicago.edu>
- Organization: School of Mathematics, University of Bath, UK
- Lines: 37
-
- In the referenced article, kwhyte@dent.uchicago.edu (Kevin Whyte) writes:
- >
- >
- > Let G be a compact semi-simple Lie group. The killing
- >form gives a canonical Riemannian metric, and hence a
- >laplacian acting on smooth functions. What are the eigen-values
- >of this laplacian?
- >
- > L (the laplacian) is a G invariant elliptic operator, on, say
- >L2(G). Hence the eigenspaces are finite dimensional and G invariant.
- >Peter-Weyl then tells us that each one of these is just some sum
- >of irreduciple reps. of G (each of which occurs as many times as
- >its weight). Thus, given an irreducible rep. of G, it occurs in at
- >most n such eigenspaces (if G is an n dimensional rep.), so it
- >should have n positive real numbers associated to it. What are they?
- >
- >-Kevin
- >kwhyte@math.uchicago.edu
-
-
- These questions are essentially algebraic: the action of the Laplacian on
- smooth functions is just that of the Casimir element in the universal
- envelopping algebra. Now on an irred subrep, the eigenvalue of the Casimir is
- determined entirely by the highest weight of the subrep via a simple formula
- which is something like
-
- evalue=(l,l+2r)
-
- where l is the highest wt and r is half the sum of the positive roots (I cannot
- remember the exact formula right now...). This answers yr questions:
-
- 1) The evals of the Laplacian are the numbers (l,l+2r) (or whatever) as l
- ranges over all dominant weights (all irred reps appear in the regular rep).
-
- 2) A given irred only appears in ONE eigenspace of the Laplacian.
-
- Hope this helps--Fran
-