Two remarks on a theorem of Dipendra Prasad (Q953052)

Let \(F\) be a \(p\)-adic field and let \(L\) be a quadratic extension with associated quadratic character \(\epsilon_{L/F}\). Fix an embedding of \(L^\times\) into \(G=GL(2,F)\). This paper proves two results on the local theta correspondence and restrictions of irreducible admissible representations of \(G\) to \(G^+=\{g\in G\;|\;\epsilon_{L/F}(\det g)=1\}\). The results in the present paper are motivated, at least in part, by results appearing in [\textit{C. Moen}, Pac. J. Math. 158, No.~2, 365--386 (1993; Zbl 0789.22034); Pac. J. Math. 127, No.~1, 141--154 (1987; Zbl 0675.22008)], [\textit{J. Rogawski}, The multiplicity formula for \(A\)-packets. The zeta functions of Picard modular surfaces. CRM Workshop, Montreal/Can. 1988, 395-419 (1992; Zbl 0823.11027)], and [\textit{D. Prasad}, Compos. Math. 94, No.~1, 19--28 (1994; Zbl 0824.11035); ``Relating invariant linear form and local epsilon factors via global methods.'' (With an appendix by Hiroshi Saito), Duke Math. J. 138, No.~2, 233--261 (2007; Zbl 1129.22010)]. The first result is that the Moen-Rogawski theorem on the theta correspondence for \((U(1),U(1))\) is equivalent to a theorem of D. Prasad on the restriction to \(G^+\) of the principal series representation of \(G\) associated with the characters \(1,\epsilon_{L/F}\). (Note that unlike, for example, Moen's paper, this paper does not assume \(F\) has odd residual characteristic.) In Prasad's 2007 paper, he describes the restriction to \(G^+\) of the principal series representation \(\pi\) of \(G\) associated with characters of \(L^\times\) via the theta correspondence, \(\pi|_{G^+}=\pi^+\oplus \pi^-\), in terms of \(\epsilon\)-factors. (He also proves an analog of this for the multiplicative group of the division algebra.) The second result of the paper under review is a new proof of this theorem of Prasad.