Critical Hardy--Littlewood inequality for multilinear forms

Abstract: The Hardy--Littlewood inequalities for $m$-linear forms on $\ell_{p}$ spaces are known just for $p>m$. The critical case $p=m$ was overlooked for obvious technical reasons and, up to now, the only known estimate is the trivial one. In this paper we deal with this critical case of the Hardy--Littlewood inequality. More precisely, for all positive integers $m\geq2$ we have [ \sup_{j_{1}}\left( \sum_{j_{2}=1}^{n}\left(.....\left( \sum_{j_{m}=1} ^{n}\left\vert T\left( e_{j_{1}},\dots,e_{j_{m}}\right) \right\vert ^{s_{m} }\right) ^{\frac{1}{s_{m}}\cdot s_{m-1}}.....\right) ^{\frac{1}{s_{3}}s_{2} }\right) ^{\frac{1}{s_{2}}}\leq2^{\frac{m-2}{2}}\left\Vert T\right\Vert ] for all $m$--linear forms $T:\ell_{m}^{n}\times\cdots\times\ell_{m} ^{n}\rightarrow\mathbb{K}=\mathbb{R}$ or $\mathbb{C}$ with $s_{k} =\frac{2m(m-1)}{m+mk-2k}$ for all $k=2,....,m$ and for all positive integers $n$. As a corollary, for the classical case of bilinear forms investigated by Hardy and Littlewood in 1934 our result is sharp in a strong sense (both exponents and constants are optimal for real and complex scalars).