4
$\begingroup$

For an cuspidal automorphic representation of general linear group, we can attach its global Vogan A-packet.

Though I thought that it is finite set, in some paper, it is written that there are infinitely many automorphic cuspidal representations in the associated packet.

How is it possible?

Improve this question
$\endgroup$
5
  • $\begingroup$ What is the precise context in which it is asserted to be infinite? $\endgroup$
    – Kimball
    Commented Nov 27, 2021 at 13:39
  • $\begingroup$ @Kimball, Page 40 line 2~3 of the following paper. math.purdue.edu/~liu2053/S6.pdf $\endgroup$
    – Andrew
    Commented Nov 27, 2021 at 14:50
  • $\begingroup$ I'm pretty sure this is not for a general linear group $GL_n$, but for some other reductive group -- apparently $Sp_{2(m-n)}$. (For $GL_n$ I believe all $A$-packets are finite, because they're singletons!) In general a global packet is a product of local packets, and all the local guys are finite, but since the set of places is infinite, the global packet can be infinite -- you can construct explicit examples using restriction to $SL_2$ of automorphic reps of $GL_2$ (the packet has more than one element at any unramified place where the Hecke eigenvalue is 0). $\endgroup$
    – David Loeffler
    Commented Nov 29, 2021 at 7:50
  • $\begingroup$ @Loeffler, Thank you for the comment. May I ask some stupid quesiton? If $\pi$ is a unitary irreducible cuspidal automorphic representation of $GL(2)$, I am wondering whether there are infinitely many places $v$ such that $\pi_v$ is discrete series representation. Because the local A-packet for $Mp(2)$ associated to $\pi_v$ is singleton iff $\pi$ is not discrete series, if such places $v$, which makes $\pi_v$ discrete series, are infinitely many, then there are infinitely many cuspidal representations in the global Vogan A-packet for $Mp(2)$ associated to $\pi$. $\endgroup$
    – Andrew
    Commented Nov 29, 2021 at 12:54
  • $\begingroup$ @Loeffler, I think $S_{\pi}$, the set of places $v$ such that $\pi_v$ is square integrable, is finite because unramified representation can never be square-integrable. Since the number of elements of the global Vogan A-packet of $\pi$ for $Mp(2)$ is $2^{|S_{\pi}|}$, the number of global cuspidal representations in the packet seems also finite. $\endgroup$
    – Andrew
    Commented Nov 29, 2021 at 21:42

0

Reset to default

You must log in to answer this question.