A convolution estimate for two-dimensional hypersurfaces (Q986624)

Let \(\Sigma_1=yz\)-plane, \(\Sigma_2=xz\)-plane, \(\Sigma_3=xy\)-plane be coordinate hyperplanes in \(\mathbb{R}^3\). By the classical Loomis-Whitney inequality, \[ || f * g ||_{L^2 (\Sigma_3) }\leq C\, || f ||_{L^2 (\Sigma_1) } ||g ||_{L^2 (\Sigma_2) }. \] \textit{J. Bennet, A. Carbery} and \textit{J. Wright} [Math. Res. Lett. 12, No. 4, 443--457 (2005; Zbl 1106.26020)] extended this result to the case when \(\Sigma_1\), \(\Sigma_2\), and \(\Sigma_3\) are bounded subsets of transversal sufficiently smooth surfaces in \(\mathbb{R}^3\). In the present paper, the result by J. Bennet, A. Carbery, and J. Wright is generalized to a class of \(C^{1,\beta}\) hypersurfaces in \(\mathbb{R}^3\) under scaleable assumptions. Applications to non-linear dispersive equations are indicated.