Potential inequality (Q690046)

The authors prove some general ``potential inequality for concave functions'' of the form \[ \Phi[s/M]\geq {\textstyle {1\over M}} N[f\varphi(s)] \] where \(s=Nf\), \(Nf\) is a ``potential type operator'', being in fact a more general operator, generated by a kernel \(N(x,dy)\), satisfying some generalized maximum principle with the constant \(M\), while \(\Phi(t)= \int_ 0^ t \varphi(u)du\) is a convex function, \(\varphi(u)= \eta([0,u])\), \(\eta\) being a Borel measure on \(\mathbb{R}_ +^ 1\). This yields some general estimates of the type \[ \| Nf\|_{q,\mu} \geq c\| f\|_{p,\nu} \] for \(1\leq p,q<\infty\) and \(\sigma\)-finite measures \(\mu\) and \(\nu\). The authors show that many known weighted inequalities are particular cases of their general inequality.