Hypercuspidality of automorphic cuspidal representations of the unitary group \(\mathrm{U}(2,2)\) (Q1096661): Difference between revisions

According to \textit{I. Piatetski-Shapiro} [Proc. Symp. Pure Math. 33, Part 1, 209--212 (1979; Zbl 0423.22017)], a cusp form is hypercuspidal if its Fourier coefficients with respect to the maximal unipotent subgroup vanish identically. In the present paper, the author introduces an apparently stronger notion of hypercuspidality for the quasi-split unitary group \(\mathrm{U}(2,2)\), and tries to construct examples of such forms by examining \(\theta\)-series liftings from \(\mathrm{U}(1,1)\) and \(\mathrm{U}(2,1)\). For \(\mathrm{U}(2,2)\), the author calls a cusp form \(U\)-cuspidal (resp. \(N\)-cuspidal) if all the non-trivial Fourier coefficients of \(F\) along the unipotent radical of the Borel subgroup (resp. the unipotent radical of the parabolic subgroup stabilizing an isotropic line) vanish; it is called hypercuspidal only if it is both \(U\)-cuspidal and \(N\)-cuspidal. In order to construct examples of \(U\)- (resp. \(N\)-) cusp forms, the author looks at \(\theta\)-series lifts \(\theta (\tau,\psi)\) on \(\mathrm{U}(2,2)\) coming from cuspidal representations \(\tau\) on \(\mathrm{U}(1,1)\) (resp. \(\mathrm{U}(2,1)\)); moreover, he shows that \(\theta (\tau,\psi)\) is actually hypercuspidal on \(\mathrm{U}(2,2)\) if \(\tau\) is hypercuspidal on \(\mathrm{U}(2,1)\). (Examples of hypercusp forms on \(\mathrm{U}(2,1)\) were constructed earlier by the reviewer and \textit{I. Piatetski-Shapiro} [Lect. Notes Math. 1041, 141--184 (1984; Zbl 0531.10034)]. As usual in the theory of \(\theta\)-liftings, it is already a delicate matter to establish that \(\theta (\tau,\psi)\) is simply cuspidal; cf. \textit{S. Rallis} for some general theorems [Compos. Math. 51, 333--399 (1984; Zbl 0624.22011)] and \textit{I. Piatetski- Shapiro} for the pair \((\mathrm{PGSp}_4,\mathrm{SL}_2)\) [Invent. Math. 71, 309--338 (1983; Zbl 0515.10024)]. In the special cases considered here, the author finds necessary and sufficient conditions on \(\tau\) for \(\theta (\tau,\psi)\) to be cuspidal; however, at least for \(\mathrm{U}(2,1)\), the author does not give examples of any \(\tau\) satisfying these conditions. Clearly these interesting results merit further development.