All conditions for Stein-Weiss inequalities are necessary (2110.14220v1)

Abstract: The famous Stein-Weiss inequality on $\mathbf R^n \times \mathbf R^n$, also known as the doubly weighted Hardy-Littlewood-Sobolev inequality, asserts that [ \Big| \iint_{\mathbf R^n \times \mathbf R^n} \frac{f(x) g(y)}{|x|^\alpha |x-y|^\lambda |y|^\beta} dx dy \Big| \lesssim | f | {L^p(\mathbf R^n)} | g| _{L^r(\mathbf R^n)} ] holds for any $f\in L^p(\mathbf R^n)$ and $g\in L^r(\mathbf R^n)$ under several conditions on the parameters $n$, $p$, $r$, $\alpha$, $\beta$, and $\lambda$. Extending the above inequality to either different domains rather than $\mathbf R^n \times \mathbf R^n$ or classes of more general kernels rather than the classical singular kernel $|x-y|^{-\lambda}$ has been the subject of intensive studies over the last three decades. For example, Stein-Weiss inequalities on the upper half space, on the Heisenberg group, on homogeneous Lie group are known. Served as the first step, this work belongs to a set in which the following inequality on the product $\mathbf R^{n-k} \times \mathbf R^n$ is studied [ \Big| \iint{\mathbf R^n \times \mathbf R^{n-k}} \frac{f(x) g(y)}{|x|^\alpha |x-y|^\lambda |y|^\beta} dx dy \Big| \lesssim | f | _{L^p(\mathbf R^{n-k})} | g| _{L^r(\mathbf R^n)}. ] Toward the validity of the above new inequality, in this work, by constructing suitable counter-examples, we establish all conditions for the parameters $n$, $p$, $r$, $\alpha$, $\beta$, and $\lambda$ necessarily for the validity of the above proposed inequality. Surprisingly, these necessary conditions applied to the case $k=1$ suggest that the existing Stein-Weiss inequalities on the upper half space are yet in the optimal range of the parameter $\lambda$. This could reflect limitations of the methods often used. Comments on the Stein-Weiss inequality on homogeneous Lie groups as well as the reverse form for Stein-Weiss inequalities are also made.