Denseness of holomorphic functions attaining their numerical radii (Q2480595): Difference between revisions

For a complex Banach space \(X\), let \(A_\infty(B_X; X)\) denote the space of bounded continuous functions \(h\) from \(B_X= \{x\in X:\| x\|\leq 1\}\) to \(X\) which are holomorphic on \(\{x\in X:\| x\|< 1\}\). The numerical radius of \(h\) was first defined by \textit{L. A. Harris} in [Am. J. Math. 93, 1005--1019 (1971; Zbl 0237.58010)] as the quantity \(\sup\{|x^*(h(x))|: x\in X\), \(x^*\in X^*\), \(\| x^*\|=-\| x\|= x^*(x)= 1\}\). \(h\) is said to attain its numerical radius if this supremum is attained for some \(x\) and \(x^*\). The present paper discusses when the subset of the numerical radius attaining elements in \(A_\infty(B_X;X)\) is dense in \(A_\infty(B_X; X)\). On the positive side, it is shown that this is the case when \(X\) has the Radon--Nikodým property. For some classical spaces without this property, there are also obtained some positive results. For example, the denseness of the numerical radius attaining elements in \(A_\infty(B_{c_0}; c_0)\) follows from the previously known denseness of norm attaining elements. For \(K\) a compact Hausdorff topological space, the numerical radius attaining elements are shown to be dense in the space of bounded weakly uniformly continuous functions on \(B_{C(K)}\) which are holomorphic on its interior. In the negative direction, the authors give an example of a certain Banach space (the pre-dual \(d_*(\omega,1)\) of the Lorentz space \(d(\omega, 1)\) for some \(\omega\) in \(\ell^2\setminus\ell^1\))) for which the subset of the numerical radius attaining elements is not dense in the Banach space of all bounded holomorphic functions from the open subset \(\{z\in d_*(\omega, 1):\| z\|< R\}\) \((R> e)\) to \(d_*(\omega, 1)\).