Weakly compact multipliers on group algebras (Q1883599): Difference between revisions

For any infinite locally compact Hausdorff group \(G\), the group algebra \(\text{L}^1(G)\) ist not Arens regular, that is, the two Arens products on its bidual \(\text{L}^1(G)^{**}\) (each one extending the product of \(\text{L}^1(G)\)) do not coincide. Due to this asymmetry, left resp. right multipliers on \(\text{L}^1(G)^{**}\) behave differently in an essential way, requiring rather different methods of proof for their treatment. The same is true for \(\text{M}(G)^{**}\), the bidual of the algebra of complex Radon measures on \(G\). As \textit{F. Ghahramani} and \textit{A. T. Lau} have shown [Math. Proc. Camb. Philos. Soc. 111, No. 1, 161--168 (1992; Zbl 0818.46050); J. Funct. Anal. 132, No. 1, 170--191 (1995; Zbl 0832.22007)], the existence of non-zero compact resp. weakly compact \textit{right} multipliers is equivalent to \(G\) being amenable. For the existence of non-zero compact resp. weakly compact \textit{left} multipliers, compactness of \(G\) is sufficient, yet as to the converse, only partial results have been obtained by these authors. The paper under review continues the study of the ``left'' case. In particular, it is shown in the first part that for non-compact \(G\), neither \(\text{L}^1(G)^{**}\) nor \(\text{M}(G)^{**}\) admit weakly compact (let alone compact) left multipliers different from zero. In the second part, a sufficient condition on \(G\) is given (namely, that \(\text{L}^\infty(G)\) has a unique right invariant mean) implying that any weakly compact left multiplier on \(\text{L}^1(G)^{**}\) is given by multiplying from the left by a suitable element of \(\text{L}^1(G)\). Both these results arise as corollaries of more general theorems; the first of those actually states that the existence of non-zero elements in \(\text{M}(G)^{**}\) acting as compact resp. weakly compact operators by left multiplication is equivalent to the compactness of \(G\). Altogether, these results answer several of the problems listed by \textit{F. Ghahramani} and \textit{A. T. Lau} [see above and also J. Funct. Anal. 150, No. 2, 478--497 (1997; Zbl 0891.22007)]. Due to the concise style of writing of the author, the proofs of the main results -- notwithstanding their considerable length and complexity -- look deceptively simple, yet rely heavily on deep results from measure theory and functional analysis.