The Mazur-Ulam theorem (Q1931421)

The classical Mazur-Ulam theorem [\textit{S. Mazur} and \textit{S. Ulam}, C. R. Acad. Sci., Paris 194, 946--948 (1932; Zbl 0004.02103)] states that every bijective isometry \(\alpha\) between two real normed spaces \(X\) and \(Y\) is affine, i.\,e., \(\alpha(\lambda x+(1-\lambda)y)=\lambda\alpha(x)+(1-\lambda)\alpha(y)\) for all \(x,y\in X\) and every \(\lambda\in [0,1]\), or equivalently, \(\alpha\) is linear up to a constant. \textit{A. Vogt} [Stud. Math. 45, 43--48 (1973; Zbl 0222.46015)] proved some generalisations of the Mazur-Ulam theorem, where \(\alpha\) is just assumed to preserve the equality of the distance (\(\|\alpha(x)-\alpha(y)\|\) depends functionally on \(\|x-y\|\)), instead of being an isometry. Using ideas of Vogt, \textit{J. Väisälä} [Am. Math. Mon. 110, No. 7, 633--635 (2003; Zbl 1046.46017)] gave a new proof of the classical Mazur-Ulam theorem, simpler than the original one. All the above cited proofs employ reflections, i.\,e., maps of the form \(\rho_z(x)=2z-x\). In the paper under review, the author gives a further simplified proof of the Mazur-Ulam theorem, using a remarkably clever reflection technique.