When is the range of a multiplier on a Banach algebra closed? (Q2457745): Difference between revisions

In a deep paper, \textit{B. Host} and \textit{F. Parreau} [Ann. Inst. Fourier 28, No.~3, 143--164 (1978; Zbl 0368.43001)] solved the so-called Glicksberg problem (actually, the author of this paper recalls that this problem traces back to A. Beurling in 1938): if \(G\) is a locally compact abelian group and \(\mu\) is a measure on \(G\), then \(\mu \ast L^1(G)\) is closed if and only if \(\mu\) factors as a product of an invertible measure and an idempotent measure. Several generalizations of this result have been obtained: see [\textit{K. Laursen} and \textit{M. Neumann}, An introduction to local spectral theory, London Mathematical Society Monographs. New Series. 20. Oxford: Clarendon Press (2000; Zbl 0957.47004)]. The author of this paper extended it to all locally compact amenable groups [J. Funct. Anal. 205, No.~1, 90--106 (2003; Zbl 1047.43009)]. In the paper under review, the author shows that if \(A\) is a Banach algebra with a bounded approximate identity (\(BAI\)) and \(A\) satisfies the condition \((H)\): every proper closed ideal of \(A\) is contained in a proper closed ideal with a \(BAI\), then every multiplier \(T : A \to A\) (that is \(a T(b) = T(a)b\), \(\forall a,b \in A\)) with closed range factors as the product of an idempotent multiplier and an invertible one. Condition \((H)\) is satisfied by all the \({\mathbb C}^\ast\)-algebras, the commutative semisimple amenable algebras, and the Herz-Figà-Talamanca algebras \(A_p(G)\) of a locally compact amenable group \(G\). To obtain his result, the author proves two other results. If \(A\) is a Banach algebra with a \(BAI\) and \(T : A \to A\) is a multiplier with a closed range, he first proves that it is equivalent to say that a) \(T^2(A)\) is dense in \(T(A)\); b) \(T^2(A) = T(A)\); and c) \(T(A)\) has a \(BAI\); then, he proves that if \(A\) satisfies moreover the condition \((H)\), one always has \(T^2(A)\) dense in \(T(A)\). With these two results, one gets that \(A = T(A) \oplus \ker T\), and the result follows. It is noteworthy that the proofs only use the Banach algebra structure of the bidual \(A^{\ast\ast}\) and some standard results of functional analysis. It should be noted that the name of \textit{Y. Zaïem} in the reference [Bull. Sci. Math., II. Ser. 99, 65--74 (1975; Zbl 0317.43002)] is incorrectly written.