Démonstration de la conjecture \(\tau\). (Proof of the \(\tau\)-conjecture.) (Q1401470): Difference between revisions

Let \(G\) be a (simply connected) semisimple group over a number field \(k\), denote by \(\mathbb A\) the adeles of \(k\) and let \(v\) be a place of \(k\). Then \(G(k_v)\) acts on the space of automorphic forms \(\mathcal A_G=L^2(G(k)\backslash G(\mathbb A))\) and denote by \(\mathcal A_G^0\) the subspace of forms with zero integral. If \(G(k_v)\) is not compact then, by strong approximation, the trivial representation does not appear in \(\mathcal A_G^0\). In this paper, the author proves the ``\(\tau\)-conjecture'' of \textit{A. Lubotzky} and \textit{R. J. Zimmer} [Isr. J. Math. 66, 289-299 (1989; Zbl 0706.22010)]: the support of \(\mathcal A_G^0\) in the unitary dual of \(G(k_v)\) is separated from the trivial representation. The idea of the proof is as follows. We may assume \(G(k_v)\) has rank \(1\), for otherwise Kazhdan's property \((T)\) implies the result. Now, if \(H\) is a semisimple subgroup of \(G\) such that \(H(k_v)\) is also of rank \(1\), then the \(\tau\)-conjecture for \(G\) is implied by that for \(H\) [\textit{M. Burger} and \textit{P. Sarnak}, Invent. Math. 106, 1-11 (1991; Zbl 0774.11021)] for \(v\) a real place [the author and \textit{F. Ullmo}, Equidistribution des points de Hecke, to appear in ``Contributions to Automorphic Forms, Geometry and Arithmetic''. To commemorate the sixtieth birthday of J. Shalika, H. Hida, D. Ramakrishnan, F. Shahidi (eds.) (Johns Hopkins University Press, Baltimore)] for \(v\) \(p\)-adic). The first step in the paper is then to determine the semisimple groups which have no proper semisimple subgroup: \(SL(2)\); \(SL(1,D)\), for \(D\) a division algebra of prime degree over an extension of \(k\); and \(SU(D,*)\), where \(D\) is a division algebra of prime degree over a quadratic extension \(E\) of a finite extension \(K\) of \(k\) and \((*)\) is an \(E/K\)-involution of the second kind. The case of \(SL(2)\) may be treated directly, and the restriction on the rank of \(G\) reduces the case \(SL(1,D)\) to a quaternionic algebra which is similar, using Jacquet-Langlands functoriality. The bulk of the paper is then concerned with the final case \(SU(D,*)\) where the author proceeds via a base change to \(D^\times\). This is achieved by direct comparison in a trace formula.