Strict convexity of bounded domains and unicity of complex geodesics (Q5937148)

Let \(D\) be a bounded convex domain in a reflexive Banach space \(E\) and let \(a\in D\). Consider the following conditions: (i) \(D\) is strictly convex. (ii)\({}_a\) For any \(r>0\) the Carathéodory ball \(B_c(a,r)\) is strictly convex. (iii)\({}_a\) There exists \(r>0\) such that \(B_c(a,r)\) is strictly convex. (iv)\({}_a\) The Carathéodory indicatrix \(I_a=\{v\in E: E_D(a,v)<1\}\) is strictly convex. (v)\({}_a\) The complex geodesics through \(a\) are unique. The author proves that: (i) implies each of the conditions (ii)\({}_a\), (iii)\({}_a\), (iv)\({}_a\), (v)\({}_a\); (ii)\({}_a \Leftrightarrow\) (iii)\({}_a\); (ii)\({}_a\) implies (iv)\({}_a\) and (v)\({}_a\). Moreover, he shows that converse implications are in general false. The proof is based on the following result. Let \(\varphi:\Delta\rightarrow D\) be a complex geodesic for \(D\) with \(\varphi(0)=a\). Put \(\rho:=\tanh r\). Then the mapping \(\Delta\ni\zeta\rightarrow\varphi(\rho\zeta)\) is a complex geodesic for \(B_c(a,r)\) and \(c_{B_c(a,r)}(a,z)=\tanh^{-1}(\frac 1\rho\tanh c_D(a,z))\), \(z\in B_c(a,r)\). Moreover, \(B_c(a,r)\) is convex. The case \(E=\mathbb C^n\) has been previously investigated by the author in the paper [J. Math. Pures Appl., IX. Ser. 78, No. 9, 867-876 (1999; Zbl 0959.32025)]. In the second part of the paper the author presents a necessary condition for a complex geodesic \(\varphi\in\mathcal C(\overline\Delta)\) to be unique. The condition says that \(\bigcap_{\zeta\in\partial\Delta}\mathcal P(\varphi(\zeta))=\{0\}\), where \(\mathcal P(y)\) is the maximal balanced convex set such that \(y+\mathcal P(y)\subset \overline D\).