Decouplings for three-dimensional surfaces in \(\mathbb{R}^{6}\) (Q667620)

Given a compact smooth surface \(S\) (with boundary) in \(\mathbb{R}^n\) on the Fourier side, we can use the surface of lattice \(\delta\)-cubes to cut it into disjoint pieces \(\theta\) (called caps). For a function \(f\) defined on \(S\), we can define the Fourier extension of \(f\) to be the Fourier inverse of \(f \mathrm{d}\sigma\), where \(\mathrm{d}\sigma\) is the surface measure on \(S\). Let \(f_{\theta}\) be the restriction of \(f\) on \(\theta\). Let \(w\) be a weight function that is roughly equal to the characteristic function on a sufficiently large ball (whose size depends on \(\delta\)) and decays fast enough polynomially away from it. For an exponent \(p \geq 2\), the \(l^p L^p\) decoupling constant is the best constant \(C\) such that \(\|Ef\|_{L^p (w)} \leq C (\sum_{\theta} \|Ef_{\theta}\|_{L^p (w)}^p)^{\frac{1}{p}}\). In this paper, the sharp asymptotic behavior of the \(l^p L^p\) decoupling constant for all \(p\) when \(S\) is a non-degenerate three dimensional surface in \(\mathbb{R}^6\) is determined, up to a \(\delta^{-\varepsilon}\)-loss. By a standard bilinear reduction, the proof is reduced to dealing with the decoupling between caps near the zero set of a degree \(3\) polynomial. One novelty in the present article under review is: It was noticed that the smooth part is ``easier'' and hence we can reduce the problem to dealing with the decoupling near the singularity of this polynomial and hence make it near the zero set of a quadratic polynomial (notice the degree drop). Doing this once more we are near a hyperplane where the decoupling can be established directly.