An answer to a question on the convolution of functions (Q438609): Difference between revisions

Let \(G\) denote a locally compact group with a fixed left Haar measure \(\lambda\). Addressing some questions recently raised by \textit{S. Głąb} and \textit{F. Strobin} [Arch. Math. 95, 583--592 (2010; Zbl 1210.43001)], the authors in particular prove the following: (1) Let \(1<p \leq 2\). Then \(G\) is unimodular if and only if, for all \(f,g\in L^p(G)\), the convolution \(f*g\) exists (in the sense that \((f*g)(x)\) is finite for \(\lambda\)-almost all \(x\in G\)). (2) Let \(G\) be a non-unimodular locally compact group, \(1<p < \infty\) and \( 1\leq q <\infty\). Then the set \(\{(f,g)\in L^p(G) \times L^q(G): f*g \text{ is not }\lambda\)-a.e. finite on \(G\)