Vol. 25, No. 3, 1968

Download this article
Download this article. For screen
For printing
Recent Issues
Vol. 307: 1  2
Vol. 306: 1  2
Vol. 305: 1  2
Vol. 304: 1  2
Vol. 303: 1  2
Vol. 302: 1  2
Vol. 301: 1  2
Vol. 300: 1  2
Online Archive
The Journal
Editorial Board
Submission Guidelines
Submission Form
Policies for Authors
ISSN: 1945-5844 (e-only)
ISSN: 0030-8730 (print)
Special Issues
Author Index
To Appear
Other MSP Journals
Curvature in Hilbert geometries. II

Paul Joseph Kelly and Ernst Gabor Straus

Vol. 25 (1968), No. 3, 549–552

The interior of a closed convex curve C in the Euclidean plane can be given a Hilbert metric, which is preserved by projective mappings. Let p, q be points interior to C and let u, v be the points of intersection of the line pq with C. The Hilbert distance h(p,q) is defined by

h(p,q) = |log d(v,p)d(u,q)|,

where d(x,y) denotes Euclidean distance. If C contains at most one line segment then h(p,q) is a proper metric and the metric lines are the open chords of C carried by the Euclidean lines. Following Busemann [1, p. 237], we define the (qualitative) curvature at a point p as positive or negative if there exists a neighborhood U of p such that for every x,y U we have

   ----                        ----
2h(x,y) ≧ h(x,x) respectively  2h(x,y) ≦ h(x,y),

where x y are the Hilbert midpoints of p and x and of p and y respectively.

In an earlier paper [2] we proved that any point p at which the sign of the curvature is determined is a projective center of C; that is, there exists a projective transformation which maps p into an affine center of the image of C. We also stated the conjecture that a Hilbert geometry has no point of positive curvature. It is the purpose of this paper to prove that conjecture.

Mathematical Subject Classification
Primary: 52.50
Received: 17 September 1967
Published: 1 June 1968
Paul Joseph Kelly
Ernst Gabor Straus