Maximizers for the Stein-Tomas inequality (Q343519): Difference between revisions

The paper continues research of former authors on the optimal constant in the celebrated Stein-Tomas inequality. This constant is defined as \[ R_N:=\sup\limits_{0\not\equiv f\in L^2(\mathbb S^{N-1})}\frac{\int_{\mathbb R^N}|\check{f}|^qdx}{\|f\|^q}, \] where \(\|f\|\) denotes the norm in \(L^2(\mathbb S^{N-1})\). The main concern of the authors is whether the indicated supremum is attained and, more generally, the description of maximizing sequences. The results known earlier were in dimensions \(N=2\) and \(N=3\). Here, the main result is the following Theorem 1.1. Let \(N\geq2\). If \[ R_N>\frac{2^{q/2}}{\sqrt\pi}\frac{\Gamma(\frac{q+1}2)}{\Gamma(\frac{q+2}2)}S_{N-1}, \] where \(S_{N-1}\) is the optimal constant of order \({N-1}\) in Strichartz's inequality, then maximizing sequences for \(R_N\), normalized in \(L^2(\mathbb S^{N-1})\), are precompact in \(L^2(\mathbb S^{N-1})\) up to modulations and, in particular, there is a maximizer for \(R_N\). It is valid in any dimension.