The Daugavet equation for polynomials on \(\mathrm{C}^\ast\)-algebras and \(\mathrm{JB}^\ast\)-triples (Q6201177)

Recall that a Banach space \(X\) has the Daugavet property (respectively the alternative Daugavet property) if \(\|\mathrm{Id}+T\|=1+\|T\|\) (respectively \(\max_{|w|=1}\|\mathrm{Id}+wT\|=1+\|T\|\)) holds for all rank one operators \(T\) on \(X\) or, equivalently, holds for all weakly compact linear operators on \(X\). The polynomial versions of these two properties read: \(X\) has, respectively, the polynomial Daugavet property or the alternative polynomial Daugavet property if the corresponding one of the two operator equations above holds for weakly compact polynomials (endowed with the operator norm) on \(X\) with values in \(X\). (A polynomial on \(X\) with values in \(Y\) is a linear combination of homogeneous polynomials \(P\) which by definition are of the form \(P(x)=S(x,\ldots,x)\) where \(S\) is a symmetric multilinear map on \(X^N\) with values in a Banach space \(Y\); scalar polynomials are those where \(Y=\mathbb{R}\) or \(=\mathbb{C}\). Since the paper deals with C\(^*\)-algebras and JB\(^*\)-triples the authors deal only with the complex case. A polynomial is said to be weakly compact if the image of the closed unit ball is relatively weakly compact.) The polynomial versions are clearly stronger than the corresponding classical ones. For the Daugavet property it is not known whether, in general, the polynomial version is strictly stronger, while the Banach space \(\ell_1\) is known to have the alternative Daugavet property but to fail the alternative polynomial Daugavet property. However, there are classes of Banach spaces (for example commutative C\(^*\)-algebras) for which the classical and the polynomial versions of the Daugavet property (respectively, of the alternative Daugavet property) are equivalent. In this complicated situation the main result of the paper under review comes in: it is proven that a JB\(^*\)-triple has the Daugavet property if and only if it satisfies the polynomial version and analoguously for the alternative Daugavet property. The main result is new even for C\(^*\)-algebras. Together with its proof it can also be interpreted as a criterion (and as a new proof of this criterion) for those C\(^*\)-algebras and JB\(^*\)-triples having the (alternative) polynomial Daugavet property via an already known criterion for the classical properties. The main tool for the proof is a nice result which is of interest in itself: every scalar polynomial on a JB\(^*\)-triple is sequentially strong\(^*\) continuous. The results support the idea that, with a good command of JB\(^*\)-techniques at hand, many Banach space theoretic questions for C\(^*\)-algebras can be treated naturally within the definitely wider class of JB\(^*\)-triples.