A relative mass cocycle and the mass of asymptotically hyperbolic manifolds

TL;DR

This paper constructs a geometric cocycle that assigns a boundary invariant to asymptotically hyperbolic manifolds, generalizing mass concepts and linking to known hyperbolic mass integrals.

Contribution

It introduces a new tractor-valued cocycle that defines a boundary invariant for ALH manifolds, extending the concept of mass in asymptotically hyperbolic geometry.

Findings

Defines a boundary geometric quantity invariant under boundary-fixing diffeomorphisms.

Establishes an absolute invariant c(h) for ALH metrics related to hyperbolic mass.

Shows c(h) can be integrated over the boundary to recover known mass formulas.

Abstract

We construct a cocycle that, for a given $n$-manifold, maps pairs of asymptotically locally hyperbolic (ALH) metrics to a tractor-valued $(n-1)$-form field on the conformal infinity. This requires the metrics to be asymptotically related to a given order that depends on the dimension. It then provides a local geometric quantity on the boundary that is naturally associated to the pair and can be interpreted as a relative energy-momentum density. It is distinguished as a geometric object by its property of being invariant under suitable diffeomorphisms fixing the boundary, and that act on (either) one of the argument metrics. Specialising to the case of an ALH metric $h$ that is suitably asymptotically related to a locally hyperbolic conformally compact metric, we show that the cocycle determines an absolute invariant $c(h)$, which still is local in nature. This tractor-valued $(n-1)$-form field on the conformal infinity is canonically associated to $h$ (i.e. is not dependent on other choices) and is equivariant under the appropriate diffeomorphisms. Finally specialising further to the case that the boundary is a sphere and that a metric $h$ is asymptotically related to a hyperbolic metric on the interior, we show that the invariant $c(h)$ can be integrated over the boundary. The result pairs with solutions of the KID (Killing initial data) equation to recover the known description of hyperbolic mass integrals of Wang, and Chru\'{s}ciel--Herzlich.