Approximate converse theorem (Q2339793)

The paper under review gives an answer to the following question of Peter Sarnak: Given a positive number \(X\), a set \(S\) of places and representations \(\pi_v\) of \(\mathrm{GL}(n,\mathbb{Q}_v)\) for \(v\in S\), give an algorithm to determine whether or not there is a global automorphic representation \(\sigma\) whose analytic conductor is at most \(X\) and whose local component at \(v\) is within \(\varepsilon\)-distance from \(\pi_v\) for each place \(v\in S\). The distances between sets of local representations are defined in terms of their Satake (or Langlands) parameters. Let \(\Pi_M\) be any given set of local representations. Following the joint work of \textit{A. R. Booker} et al. [Int. Math. Res. Not. 2006, No. 12, Article ID 71281, 34 p. (2006; Zbl 1154.11018)], the author constructs the quasi-Maass form \(F_{\Pi_M}(z)\) for \(\Pi_M\). This quasi-Maass form can be lifted to an automorphic function \(\widetilde{F}_{\Pi_M}(z)\) for \(\mathrm{SL}(n,\mathbb{Z})\), which is in general neither smooth nor cuspidal, but its image under some annihilating operator is a Hecke-Maass form. The annihilating operator is explicitly defined in the paper and one of its feature is that it annihilates the self-dual spectrum. Consequently, the author obtains the following result: \(L^2(\mathrm{SL}(n,\mathbb{Z})\setminus \mathbb{H}^n)\), \(n\geq 2\), has infinitely many non-self-dual Hecke-Maass forms. Next, the author proves the approximate converse theorem, a particular case of which provides an answer to Sarnak's question.