Time evolutions in Quantum mechanics and (Lorentzian) geometry

Abstract

As R.Feynman has shown to F. Dyson -- who published it then in 1990 under the name of "Feynman's proof of Maxwell's equations" -- the only interactions compatible with the canonical uncertainty relation (for scalar particles on flat $\R^3$) are the Lorentz covariant electromagnetic interactions. We generalize Feynman's argument to arbitrary configuration spaces, thereby clarifying its hidden assumptions as well as its geometrical significance. In addition, our result establishes a correspondence between globally hyperbolic spacetimes and solutions to our (algebraically formulated) axioms for nonrelativistic quantum mechanics.