On restricted weak type (1,1); the discrete case (Q5951495)

An operator \(M\) on \(\ell^1(\mathbb{Z})\) is called weak type \((1,1)\) if there is a constant \(C>0\) with the property that \(|\{Mf>\alpha\}|\leq C\|f\|_1/\alpha\) for all non-negative \(f\) and all \(\alpha>0\), and is called here restricted weak type \((1,1)\) if that estimate holds only for indicator functions of finite subsets of \(\mathbb{Z}\). A result of \textit{K. H. Moon} [Proc. Am. Math. Soc. 42, 148-152 (1974; Zbl 0247.42023)] in the non-discrete setting shows that the maximal operator associated to a sequence of convolution operators on \(L^1(\mathbb{R}^d)\) is restricted weak type \((1,1)\) if and only if it is weak type \((1,1)\). Here the authors exhibit a family of convolution operators on \(\ell^1(\mathbb{Z})\) that are of restricted weak type \((1,1)\) but not weak type \((1,1)\).