The Banach-Mazur distance between symmetric spaces (Q788238): Difference between revisions

In this paper it is shown that the Banach-Mazur distance between n- dimensional symmetric spaces is smaller than or equal to \(c\sqrt{n}\), where c is a numerical constant. For \(n=2^ K\) the proof is constructive. If E is an n-dimensional symmetric space then \(d(E,\ell^ n_ 2)\leq 2\sqrt{2}\max(M^{(2)}(E),M_{(2)}(E)),\) where \(M^{(2)}(E)\) (resp. \(M_{(2)}(E))\) denotes the 2-convexity (resp. 2-concavity) constant of E. This improves, for symmetric spaces, the well known estimate due to Kwapien. In the general case Kwapien's estimate is best possible (up to a constant): there exists an n-dimensional space F with 1-unconditional basis and enough symmetries for which \[ d(F,\ell^ n_ 2)\geq cT_ 2(F)C_ 2(F)\geq c'M^{(2)}(E)M_{(2)}(F), \] where \(T_ 2(F)\) (resp. \(C_ 2(F))\) is the type 2 (resp. cotype 2) constant of F and c and c' are numerical constants.