Einstein distances in Hilbert spaces (Q1818222)

Let \(X\) be an arbitrary pre-Hilbert space, i.e. a real vector space equipped with an inner product \(\delta: X^\times X\to\mathbb{R}\), \(\delta (x,y)= :xy\) satisfying \(x^2>0\) for all \(x\neq 0\) in \(X\), of dimension at least 3. Let \(t\) be a fixed element of \(X\) such that \(t^2=1\) and \(\overline x:= x-(tx)t\) \((\perp t)\), \(x_0:=tx\) for all \(x\in X\). Define \(C(X): =\{x\in X\mid \overline x^2=1\}\) as the set of points of Einstein's Cylinder Universe over \(X\) and call \(e(x,y):= (\arccos\overline x\overline y)^2 -(x_0-y_0)^2\) with \(\arccos\overline x\overline y\in[0,\pi]\) the Einstein distance of \(x,y\in C(X) \), by observing \((\overline x\overline y)^2 \leq\overline x^2 \overline y^2=1\). An orthogonal mapping \(\omega: X\to X\) can be defined by \(\omega(0)=0\) and \(\|\omega(x)- \omega(y) \|=\|x-y\|: =\sqrt{(x-y)^2}\) for all \(x,y \in X\), implying that \(\omega\) is injective and linear with \(xy=\omega(x) \omega (y)\) for all \(x,y\in X\). The main results of this note can be described as follows. Every solution \(f:C(X)\to C(X)\) of the functional equation of distance preservance \(e(x,y)= e(f(x),f(y))\) is injective and, if \(\varepsilon,a\) are real numbers with \(\varepsilon^2=1\), of the form \(f(x)=\omega(\overline x)+( \varepsilon x_0+a)t\), \(x\in C(X)\) (Theorem 1); all bijective mappings \(f\) of this form establish the group \(M(X)\) of motions of \(C(X)\). All 2-point invariants \(d:C(X)\times C(X)\to W\neq \varnothing\), satisfying the condition \(d(x,y)= d(f(x),f(y))\) for all \(x,y\in C(X)\) and all \(f\in M(X)\) are determined in Theorem 2. Using the Lorentz-Minkowski distance \(l(x,y):=(\overline x-\overline y)^2-(x_0-y_0)^2\), a real 2-point invariant \(d(x,y)\) of \((C(X),M(X))\) is called locally Lorentz-Minkowskian if, and only if, \(d(x,p)/1 (x,p)\to 1\) when \(x\to p,x\) being a point on a line \(g\ni p\) of \(C(X)\) which is not a null-line. This leads to the following characterization of the Einstein distance function \(e(x,y)\) (Theorem 3): Suppose that \(d\) is an additive and real 2-point invariant of \((C(X),M(X))\), behaving locally Lorentz-Minkowskian, such that there exist \(i,j\in K:=\{x\in X\mid tx=0,x^2=1\}\) with \(ij=0\) and \(d(i,j+ \pi/2 \cdot t)=0\); then \(|d(x,y) |= |e(x,y) |\) for all \(x,y\in C(X)\). The assumptions of Theorem 3 do not imply \(d(x,y)=e(x,y)\), which is proved with an example; however, an additional condition (Corallary) will guarantee that \(d(x,y)= e(x,y)\) holds true for all \(x,y\in C(X)\). Final remark: In the case \(\dim X<\infty\) similar problems are studied in chapter 3 of the author's book [\textit{W. Benz}, `Geometrische Transformationen' (1992; Zbl 0754.51005)]. In this notes, however, further methods and ideas are needed and used in the general situation in comparison with the finite-dimensional case.