A nonlinear d'Alembert comparison theorem and causal differential calculus on metric measure spacetimes
Tobias Beran, Mathias Braun, Matteo Calisti, Nicola Gigli, Robert J. McCann, Argam Ohanyan, Felix Rott, Clemens Sämann
TL;DR
This work extends Lorentzian geometry to nonsmooth metric measure spacetimes by developing a variational first-order Sobolev calculus driven by the maximal weak subslope |d f|, which serves as the Lorentzian modulus of df. A nonlinear Lagrangian–Hamiltonian duality and the notion of infinitesimal Minkowskianity provide a robust framework for cotangent/tangent duality, horizontal and vertical derivatives, and weak p-d'Alembertian notions. The authors establish a weak p-d'Alembert comparison under timelike curvature-dimension bounds TMCP, generalizing Eschenburg-style estimates beyond the cut locus, and develop a Lorentzian lifting theory that connects curves of measures to dynamical plans, enabling Brenier–McCann-type characterizations in this nonsmooth setting. They further integrate synthetic timelike curvature bounds with a detailed non-branching theory and a rich calculus of perturbations, chain and Leibniz rules, offering a foundation for nonsmooth Lorentzian analysis with potential implications for spacetime splitting results and quantum gravity formalisms.
Abstract
We introduce a variational first-order Sobolev calculus on metric measure spacetimes. The key object is the maximal weak subslope of an arbitrary causal function, which plays the role of the (Lorentzian) modulus of its differential. It is shown to satisfy certain chain and Leibniz rules, certify a locality property, and be compatible with its smooth analog. In this setup, we propose a quadraticity condition termed infinitesimal Minkowskianity, which singles out genuinely Lorentzian structures among Lorentz-Finsler spacetimes. Moreover, we establish a comparison theorem for a nonlinear yet elliptic $p$-d'Alembertian in a weak form under the timelike measure contraction property. As a particular case, this extends Eschenburg's classical estimate past the timelike cut locus.
