A new Hilbert-type inequality (Q2381105)

Let \(a_n,b_n\geq 0\), \(p>1\), \(1/p+1/q=1\) and \(0<\sum_{n=1}^{\infty}a_n^p<\infty\), \(0<\sum_{n=1}^{\infty}b_n^q<\infty\). Then \[ \sum_{n=1}^{\infty}\sum_{m=1}^{\infty}\frac{a_mb_n}{m+n} <\frac{\pi}{\sin(\pi/p)}\left(\sum_{n=1}^{\infty}a_n^p\right)^{1/p}\left(\sum_{n=1}^{\infty}b_n^q\right)^{1/q} \] and the constant factor on the right is the best possible. This inequality was proved in 1925 by Hardy and for \(n=2\) it reduces to the Hilbert's inequality. Many generalizations of the estimate on the right-hand side are known, while the left-hand side remains unchanged or it has the form \(\sum_{n=1}^{\infty}\sum_{m=1}^{\infty} a_mb_n/(m+n)^{\lambda}\) for some \(\lambda>2-\min\{p,q\}\). The paper under review deals with related inequalities with left-hand sides of the form \(\sum_{n=n_0}^{\infty}\sum_{m=m_0}^{\infty} {a_mb_n}/\bigl(1+u(m)u(n)\bigr)^{\lambda}\) with \(\lambda\) as above and the differentiable strictly increasing function \(u:(n_0,\infty)\to{\mathbb R}\). For \(0<p<1\) some reversions of the obtained inequalities are also given.