Lorentz meets Ptolemy
Felix Rott, Zhe-Feng Xu, Matteo Zanardini · 2026
Read Lorentz meets Ptolemy by Felix Rott, Zhe-Feng Xu, Matteo Zanardini (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-21596.txt.
Introduction
Felix Rott, Zhe-Feng Xu, and Matteo Zanardini carry a classic geometry rule into the world of relativity. The Ptolemy rule is an old statement about distances among four points. The authors define its analogue for spacetime. In spacetime, distances run along time-like paths.
In spacetimes that behave well on the large scale, they prove the rule matches a curvature bound. The bound says a certain time-like curvature is no greater than zero. They also link the rule to hyperbolic inversion. And they draw out rigidity results. Such work gives relativists a distance-only way to detect curvature bounds. No direct measure of the curvature is needed.
Catalog record
| CREATOR | Felix Rott, Zhe-Feng Xu, Matteo Zanardini |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-21596
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Lorentz meets Ptolemy Felix Rott, Zhe-Feng Xu, Matteo Zanardini Abstract We consider a Lorentzian analogue of the Ptolemy inequality and we prove that in the setting of globally hyperbolic spacetimes it is equivalent to a global timelike sectional curvature bound from above by zero. We investigate the link between the Ptolemy inequality and the hyperbolic inversion and establish some applications and rigidity properties. Keywords: Lorentzian geometry, Lorentzian length spaces, Ptolemy inequality, curvature bounds MSC2020: 53C23, 53C50, 53B30 1 Introduction Famously, the Ptolemy inequality says that any quadruple of points in the Euclidean plane satisfies the following inequality: \macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{A}\cdot\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{C}+\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{B}\cdot\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{A}\geq\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{A}\cdot\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{B}\,. (1.1) Moreover, equality holds if and only if the four points lie on a circle or on a line. It is named after the famous Greek astronomer and mathematician Claudius Ptolemy, who mentioned it in his work Almagest roughly two thousand years ago. The inequality can be interpreted as a relation between the product of the diagonals of a quadrilateral and the sum of the products of opposite sides. The inequality is also valid in higher dimensions. In a more modern context and outside of Euclidean space, the Ptolemy inequality at least goes back to being mentioned in [ 31 ] . The inequality is more prominently featured as a characterisation of inner product spaces among merely normed ones, much in the spirit of the parallelogram identity [ 32 ] . Even more recently, there are a number of results about Ptolemaic metric spaces, meaning that ( 1.1 ) is required to hold for any quadruple of points in a metric space. The Ptolemy inequality has a strong connection to (sectional) curvature bounds from above. Among other things, on Riemannian manifolds, the Ptolemy inequality is equivalent to the space being CAT(0), and Finsler manifolds enjoying this property are necessarily Riemannian [ 13 ] . It is further known that on Ptolemaic metric spaces, the notions of being CAT(0) and Busemann convex are equivalent, where in general CAT(0) is strictly stronger than Busemann convexity [ 20 ] . To the best of our investigative abilities, we have found no mention of the Ptolemy inequality in a Lorentzian context. We propose a definition for (non-smooth) Ptolemaic Lorentzian spaces and collect various results. In particular, we show that Minkowski space is Ptolemaic. To this end, much like in positive signature, our approach makes use of inversion techniques. In Lorentzian signature, this idea was already mentioned in [ 16 ] in the particular case of Minkowski space. More related to physics, the class of symmetries generated by inversions plays an important role in conformal field theories; see e.g. [ 18 ] . We extend this concept of inversion to more general Lorentzian spaces and moreover, in the spirit of [ 14 ] , give a definition of an inversion time separation function. Our main theorem concerns the interplay between the Ptolemy inequality and a bound from above on timelike sectional curvature. Theorem 1.1 . A globally hyperbolic spacetime is Ptolemaic if and only if it is globally non-positively curved. Finally, we will use some rigidity properties of the hyperbolic inversion to single out spaces of vanishing timelike sectional curvature and give a characterization of Minkowski space among spacetimes. 2 Preliminaries In this section we recall some basic information regarding smooth and non-smooth Lorentzian geometry. Let us start with the smooth case. For further details we refer to [ 2 , 28 ] . A spacetime is a non-compact smooth manifold M endowed with a Lorentzian metric tensor g with signature (+,-,\dots-) and time-oriented by a timelike vector field X\in\mathfrak{X}(M) . Given a point p\in M , we say that a vector v\in T_{p}M is timelike provided g_{p}(v,v)>0 and we say that it is null provided g_{p}(v,v)=0 . Otherwise, we call v spacelike. We call v\in T_{p}M causal if it is null or timelike and, moreover, we say that v is future-directed (resp. past-directed) provided g(v,X(p))\geq 0 (resp. \leq 0 ). A curve is called future-directed timelike (resp. causal) if its tangent vector at each point is future-directed timelike (resp. causal). The class of future-directed causal vectors in T_{p}M is denoted by F_{p}M , and we define the hyperbolic norm out of g on the tangent space T_{p}M by \|v\|_{F_{g}}:=\begin{cases}\sqrt{g_{p}(v,v)}\,,&\text{if }v\in F_{p}M,\\ 0\,,&\text{otherwise}\,.\end{cases} We refer to [ 3 , Section A.3] and [ 22 ] for an introduction to hyperbolic (Banach) norms. Notice that, in our framework, we define \|v\|_{F_{g}}=0 (instead of -\infty ) if v is spacelike to have an immediate compatibility with the current literature on triangle comparison in Lorentzian pre-length spaces. We define a two-variable function that acts dually to the distance induced by a Riemannian metric: this is the time separation function \ell:M\times M\rightarrow[0,+\infty] and it is given by the formula \ell(x,y):=\sup\int_{0}^{1}\|\dot{\gamma}_{t}\|_{F_{g}}\,\mathrm{d}t\,, (2.1) where the supremum is taken over all piecewise C^{1} future-directed causal curves \gamma:[0,1]\rightarrow M such that \gamma_{0}=x and \gamma_{1}=y . If there are no such curves, by convention we set \ell(x,y)=0 . We define the causal (resp. timelike) relations \leq (resp. \ll ) on M by saying that two points are related if they can be connected by a future-directed causal (resp. timelike) curve. Moreover, we use the notation y\in J^{+}(x)\,\Leftrightarrow\,x\leq y\,,\qquad\qquad y\in I^{+}(x)\,\Leftrightarrow\,x\ll y\,, and similarly for the past versions of these sets. Definition 2.1 (Strong causality, global hyperbolicity and future one-connectedness) . Let (M,g) be a spacetime. (i) We say that M is strongly causal if the timelike diamonds I(x,y):=I^{+}(x)\cap I^{-}(y) form a subbase for the topology. (ii) We say that it is globally hyperbolic provided \leq is a partial order on M and the causal diamonds J(x,y):=J^{+}(x)\cap J^{-}(y) are compact in M . (iii) We say that it is future one-connected if there exists a timelike homotopy between any couple of future-directed timelike curves with the same endpoints. That is to say, for every couple of future-directed timelike smooth curves \gamma,\sigma:[0,1]\rightarrow M such that \gamma_{0}=\sigma_{0} and \gamma_{1}=\sigma_{1} there exists a smooth map \Gamma:[0,1]\times[0,1]\rightarrow M such that \Gamma(0,\cdot)=\gamma , \Gamma(1,\cdot)=\sigma and \Gamma(t,\cdot) is a timelike curve connecting \gamma_{0} to \gamma_{1} . We remark that on a globally hyperbolic spacetime, \ell is finite and continuous [ 28 , Theorem 4.124] , and between any pair of causally related points there exists a maximising geodesic connecting them [ 28 , Theorem 4.123] . On a spacetime (M,g) , the Riemann tensor is defined in the usual way via the Levi-Civita connection by the formula \displaystyle R(X,Y)Z:=\nabla_{X}\nabla_{Y}Z-\nabla_{Y}\nabla_{X}Z-\nabla_{[X,Y]}Z\,, (2.2) for X,Y,Z\in\mathfrak{X}(M) . Given a point p\in M and v,w\in T_{p}M such that they span a non-degenerate plane \Pi , the sectional curvature of \Pi is defined by K(\Pi):=\frac{g(R(w,v)v,w)}{g(v,v)g(w,w)-g(v,w)^{2}}\,. (2.3) It is easily seen that this definition is independent of the choice of generators of \Pi . Such a plane \Pi\subset T_{p}M is called timelike provided g_{p}|_{\Pi} has Lorentzian signature (+,-) . Definition 2.2 (Timelike sectional curvature bound) . A spacetime (M,g) is said to have a timelike sectional curvature bound from above (resp. below) by a constant K\in\mathbb{R} , denoted \TSec\leq K (resp. \geq ), if for all p\in M and all timelike planes \Pi\subseteq T_{p}M , we have K(\Pi)\leq K (resp. \geq ). Remark 2.3 (On the convention of