A new characterization of Radon curves via angular bisectors (Q1882444)

In a Minkowski plane, for non-collinear rays \([o, a\rangle\) and \([o, b\rangle\) with unit vectors \(a\) and \(b\), call ray \([o, a+b\rangle\) the Busemann angular bisector of the rays \([o, a\rangle\) and \([o, b\rangle\) (as it satisfies the metric condition by which angular bisectors are defined in [\textit{H. Busemann}, Math. Scand. 36, 5--11 (1975; Zbl 0299.50002)] and call the set of all points that lie in the convex cone determined by the rays \([o, a\rangle\) and \([o, b\rangle\) and that are equidistant from the rays \([o,a\rangle\) and \([o, b\rangle\) (the distance from a point \(p\) to a line \(\lambda\) being defined as the infimum of all distances from \(p\) to points on \(\lambda\)) the Glogovskij (1970) angular bisector of \([o, a\rangle\) and \([o, b\rangle\). The author shows that the two notions coincide only when the normality relation is symmetric, i.e.\ when the plane is a Radon plane. For dimensions \(\geq 3\) the two notions coincide precisely when the space is Euclidean.