For a Lie group $G$ with compact Lie subgroup $K$, we say that $(G,K)$ is a pair of Gelfand type if the representation $L^2(G/K)$ of $G$ is multiplicity free, that is, if it is a direct integral of distinct irreducible representations.

Can there exist a pair of dual representations in $L^2(G/K)$ for a Gelfand pair $(G,K)$?

Or do there exist non-self dual irreducible representations in $L^2(G/K)$?

  • 3
    $\begingroup$ What about $(G,K) = (\mathbf R/\mathbf Z,0)$? $\endgroup$ Mar 23 '18 at 21:21

I gave in the comments the example of the Gelfand pair $(G,K) = (\mathbf R/\mathbf Z,0)$ for which every non-trivial irreducible representation arises together with its (distinct) dual representation. But actually this is completely general, at least when $G$ is compact: the representation $L^2(G/K)$ is self-dual. So by uniqueness of the decomposition into irreducible representations (here I use compactness), for every irreducible representation appearing in this decomposition, its dual representation also appears. Of course, the self-dual irreducible representations in $L^2(G/K)$ appear only once. This is for example the case for $(G,K)=(SO(3),SO(2))$, where all the irreducible representations of $SO(3)$ are self-dual.

  • $\begingroup$ If in addition we assume that $K$ is a maximal compact subgroup, then do we still get non-self dual irreps in $L^2(G/K)$? $\endgroup$ Mar 26 '18 at 11:19
  • $\begingroup$ @AlesandroLevi Yes, essentially same example with $(G,K) = (\mathbf R,0)$. Almost every character of $\mathbf R$ is not self dual. $\endgroup$ Mar 26 '18 at 13:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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