Ball covering property from commutative function spaces to non-commutative spaces of operators

TL;DR

This paper characterizes the ball-covering property (BCP) in both commutative and non-commutative function spaces, providing conditions for when these spaces have BCP and exploring its stability and limitations.

Contribution

It offers a topological characterization of BCP in function spaces and analyzes its stability, also demonstrating that BCP is not hereditary in certain subspaces.

Findings

$C_0(K)$ has BCP iff $K$ has a countable $$-basis

$C_0(K,X)$ has BCP iff $K$ has a countable $$-basis and $X$ has BCP

Certain operator spaces like $B(c_0)$, $B(_1)$ have BCP, while $B(L_1[0,1])$ does not

Abstract

A Banach space is said to have the ball-covering property (abbreviated BCP) if its unit sphere can be covered by countably many closed, or equivalently, open balls off the origin. Let $K$ be a locally compact Hausdorff space and $X$ be a Banach space. In this paper, we give a topological characterization of BCP, that is, the continuous function space $C_0(K)$ has the (uniform) BCP if and only if $K$ has a countable $\pi$-basis. Moreover, we give the stability theorem: the vector-valued continuous function space $C_0(K,X)$ has the (strong or uniform) BCP if and only if $K$ has a countable $\pi$-basis and $X$ has the (strong or uniform) BCP. We also explore more examples for BCP on non-commutative spaces of operators $B(X,Y)$. In particular, these results imply that $B(c_0)$, $B(\ell_1)$ and every subspaces containing finite rank operators in $B(\ell_p)$ for $1< p<\infty$ all have the BCP, and $B(L_1[0,1])$ fails the BCP. Using those characterizations and results, we show that BCP is not hereditary for 1-complemented subspaces (even for completely 1-complemented subspaces in operator space sense) by constructing two different counterexamples.