On \(L\)-factors attached to generic representations of unramified \(\mathrm {U}(2,1)\) (Q1668466)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On \(L\)-factors attached to generic representations of unramified \(\mathrm {U}(2,1)\) |
scientific article |
Statements
On \(L\)-factors attached to generic representations of unramified \(\mathrm {U}(2,1)\) (English)
0 references
28 August 2018
0 references
Let \(F\) be a non-archimedean local field of characteristic zero and odd residual characteristic, and let \(E/F\) be an unramified quadratic extension. Let \(G=U(2,1)\) be the unramified unitary group in three variables determined by the extension \(E/F\). The author has been developing, in a series of papers [J. Ramanujan Math. Soc. 28, No. 1, 91--111 (2013; Zbl 1288.22014); Trans. Am. Math. Soc. 365, No. 6, 3355--3372 (2013; Zbl 1276.22006); Manuscr. Math. 141, No. 1--2, 149--169 (2013; Zbl 1266.22020)], the theory of newforms for generic irreducible representations of \(G\). The present paper is the final step of this project. In the previous work, the author introduced a family \(K_n\) of open compact subgroups of \(G\), playing the role of Hecke congruence subgroups in the global theory. Given a generic irreducible representation \((\pi,V)\) of \(G\), let \(V(n)\) be the subspace of vectors fixed by \(K_n\). The conductor of \(\pi\), denoted by \(c(\pi)\), is the least integer such that \(V(c(\pi))\) is non-trivial. Then, the space \(V(c(\pi))\) is the space of newforms for \(\pi\). In the papers cited above, the author has proved that, for a generic irreducible representation \(\pi\) of \(G\), the conductor \(c(\pi)\) of \(\pi\) is well-defined and the space \(V(c(\pi))\) of newforms is one-dimensional. He also gave a formula for the dimension of spaces of oldforms \(V(n)\) with \(n>c(\pi)\), and showed that the Whittaker function in the Whittaker model for \(\pi\) corresponding to a non-trivial newform is non-zero at the identity. Furthermore, he considered the \(L\)-function and \(\varepsilon\)-factor attached to \(\pi\) by the theory of the Rankin-Selberg zeta integrals introduced by \textit{S. Gelbart} and \textit{I. I. Piatetski-Shapiro} [Lect. Notes Math. 1041, 141--184 (1984; Zbl 0531.10034)] and \textit{E. M. Baruch} [Isr. J. Math. 102, 317--345 (1997; Zbl 0901.11014)]. He showed that the \(\varepsilon\)-factor attached to \(\pi\) is a constant multiple of \(q_F^{-c(\pi)s}\), where \(q_F\) is the cardinality of the residue field of \(F\), subject to the conjecture that the Rankin-Selberg zeta integral of the appropriately normalized Whittaker function corresponding to a non-trivial newform for \(\pi\) attains the \(L\)-function of \(\pi\). The purpose of the present paper is to prove that conjecture. In the previous papers, the author has proved the conjecture only for generic supercuspidal irreducible representations \(\pi\). The technical heart of the paper under review is a lemma, which claims that the zeta integral attains the \(L\)-function up to a multiple of the Hecke-Tate \(L\)-function of the trivial character of \(E^\times\). Its proof combines a rough estimation of the \(L\)-function, based on the classification of generic irreducible representations of \(G\), with a technically demanding calculation of the zeta integral. Once the lemma is proved, the conjecture follows, because if it were not true, the functional equation would yield a contradiction with the fact that the \(\varepsilon\)-factor is a monomial. At the end of the paper, the author gives an example of the explicit computation of the \(L\)-function attached to a generic non-supercuspidal irreducible representation of \(G\), and determines the \(L\)-functions of the depth zero supercuspidal representations.
0 references
\(p\)-adic group
0 references
unramified unitary group in three variables
0 references
local newform
0 references
\(L\)-factor
0 references
Rankin-Selberg zeta integral
0 references