Converse theorems for GLn, II (Q4232525)

[For Part I, cf. \textit{J. W. Cogdell} and \textit{I. I. Piatetski-Shapiro}, Publ. Math., Inst. Hautes Etud. Sci. 79, 157--214 (1994; Zbl 0814.11033)]. A converse theorem for \(\text{GL}_n\) is a way of telling whether an irreducible admissible representation of \(\text{GL}_n (\mathbb A)\) is automorphic in terms of the analytic properties of its \(L\)-function under an appropriate family of twists. In this paper we prove two converse theorems for \(\text{GL}_n (\mathbb A)\) in which we consider twists of the \(L\)-function by cuspidal automorphic representations of \(\text{GL}_{n-2} (\mathbb A)\). More precisely, let \(\pi\) be an irreducible admissible generic representation of \(\text{GL}_n (\mathbb A)\) whose central character is invariant under the principal ideles and whose \(L\)-function \(L(\pi,s)\) is absolutely convergent in some right half plane. Suppose that for every cuspidal automorphic representation \(\tau\) of \(\text{GL}_{n-2} (\mathbb A)\) the twisted \(L\)-function \(L(\pi \times \tau,s)\) is nice in the sense that it has an analytic continuation to an entire function of \(s\), bounded in vertical strips, and satisfies an appropriate functional equation. Then \(\pi\) is a cuspidal automorphic representation. In the second theorem one can restrict the ramification of the twisting representations at a finite number of places. These complement our previous converse theorems for \(\text{GL}_n\) in which we twisted by cuspidal representations of \(\text{GL}_{n-1} (\mathbb A)\) but with restricted ramification at almost all places.