On zeroes of the twisted tensor \(L\)-function (Q1318160): Difference between revisions

This paper concerns the twisted tensor \(L\)-function \(L(s,r(\pi)\otimes \omega)\) -- which has been introduced and studied in the reviewer's (*) ``Twisted tensors and Euler products'', Bull. Soc. Math. Fr. 116, 295-313 (1988; Zbl 0674.10026) -- associated with a cuspidal representation \(\pi\) of \(GL(n,\mathbb{A}_ E)\) and a character \(\omega\) of \(\mathbb{A}_ F^ \times/ F^ \times\). Here \(E/F\) is a quadratic extension of global fields. When \(\pi\) is distinguished, and has a supercuspidal component, the following is shown. Theorem. Let \(s_ 0\) be a complex number such that for every separable field extension \(L(T)\) of \(F\) of degree \(n\), the \(L\)-function \(L(s,\omega_ T)\) (associated to the character \(\omega_ T\) of the idele class group of \(L(T)\) obtained on restricting \(\omega(\text{det})\) to the multiplicative group of \(L(T)\), viewed as a torus in \(GL(n,F)\)) vanishes to the order \(m\) at \(s=s_ 0\). Then \(L(s,r(\pi)\otimes \omega)\) vanishes at \(s=s_ 0\) to the order \(m\). A cuspidal \(\pi\) is called distinguished if its space contains a form whose integral (``period'') over the ``cycle'' \(PGL(n,\mathbb{A}_ F)/PGL(n,F)\) is non-zero. They emerge in (*) as the only representations for which the twisted tensor \(L\)-function may have poles (at \(s=1,0\)). A precise conjecture parametrizing the distinguished \(\pi\) is made (and proven for \(GL(2)\)) in the reviewer's ``On distinguished representations'', J. Reine Angew. Math. 418, 139-172 (1991; Zbl 0725.11026). The proof in the reviewed paper is made relatively simple on using ideas from the proof of Deligne and Kazhdan of their simple trace formula. Similar ideas were used in the split case \(E=F\oplus F\) and the adjoint representation \(L\)-function \(L(s,\omega\otimes \pi\times \check\pi)/ L(s,\omega)\), in the reviewer's ``The adjoint representation \(L\)-function for \(GL(n)\)'', Pac. J. Math. 154, 231-244 (1992; Zbl 0766.11028). This in turn was motivated by the work of \textit{H. Jacquet} and \textit{D. Zagier}, ``Eisenstein series and the Selberg trace formula, II'', Trans. Am. Math. Soc. 300, 1-48 (1987; Zbl 0625.10024), which dealt with the split case and \(n=2\) for \(\pi\) which is not assumed to have a supercuspidal component. The Appendix -- ``On the local twisted tensor \(L\)-function'' -- to this paper, defines -- and studies the main properties of -- the local factor in the possibly ramified places, of the global \(L\)-function under consideration. This extends the work of (*) to show that the complete (rather than partial) \(L\)-function \(L(s,r(\pi)\otimes\omega)\) has analytic continuation, functional equation, and poles as specified in (*).