Bilinear Fourier restriction theorems (Q1934658)

For \(D\subseteq \mathbb{R\times }\mathbb{R}\), the bilinear Fourier multiplier operator \(T_{D}\) is defined by \[ T_{D}(f,g)(z)=\int_{\mathbb{R}^{2}}\widehat{f}(x)\widehat{g}(y)\chi _{D}(x,y)e^{2\pi iz(x+y)}dxdy. \] When \(D=\mathbb{R\times R}\), \(T_{D}\) is the pointwise product. The bilinear Hilbert transform is essentially the bilinear Fourier multiplier operator on the half plane. A difficult and fundamental problem is to determine the triples \((p_{1},p_{2},p_{3})\) satisfying \(1/p_{1}+1/p_{2}+1/p_{3}=1\), such that \(T_{D}\) is a bounded operator from \(L^{p_{1}}\times L^{p_{2}}\rightarrow L^{p_{3}^{\prime }},\) where \(p_{3}^{\prime }\) is the dual index to \(p_{3}\). The boundedness of the Fourier operator when \(D\) is the unit disk in \(\mathbb{R\times }\mathbb{R}\) was proved earlier by \textit{L. Grafakos} and \textit{X. Li} [Am. J. Math. 128, No. 1, 91--119 (2006; Zbl 1143.42015)] for triples satisfying \(2\leq p_{1},p_{2},p_{3}\leq \infty \). This is known as the local \(L^{2}\) range of exponents. In contrast, \textit{L. Grafakos} and \textit{X. Li} [Ann. Math. (2) 159, No. 3, 889--933 (2004; Zbl 1071.44004)] and \textit{X. Li} [Rev. Mat. Iberoam. 22, No. 3, 1069--1126 (2006; Zbl 1133.42022)] established the boundedness of the bilinear Hilbert transform outside the local \(L^{2}\) range. This paper provides some evidence to support the hypothesis that the bilinear disk multiplier should be bounded on the same exponents as the bilinear Hilbert transform, by establishing the boundedness of the operator \(T_{D}\), with \(D\) the `lacunary polygon', for certain exponents outside the local \(L^{2}\) range. This is achieved by taking advantage of orthogonality without the use of Rubio de Francia's Littlewood-Paley inequality that is limited to the local \(L^{2}\) setting.