Energy localization invariance of tidal work in general relativity

It is well known that when an external general relativistic (electric-type) tidal field ε_jk(t) interacts with the evolving quadrupole moment ι_jk(t) of an isolated body the tidal field does work on the body ("tidal work") - i.e., it transfers energy to the body - at a rate given by the same formula as in Newtonian theory: dW/dt= - 1/2ε_jkdι_jk/dt. Thorne has posed the following question: In view of the fact that the gravitational interaction energy E_int between the tidal field and the body is ambiguous by an amount ∼ει_jk, is the tidal work also ambiguous by this amount, and therefore is the formula dW/dt=-1/2ε_jkdι_jk/dt only valid unambiguously when integrated over time scales long compared to that for ι_jk to change substantially? This paper completes a demonstration that the answer is no; dW/dt is not ambiguous in this way. More specifically, this paper shows that dW/dt is unambiguously given by -1/2ε_jkdι_jk/dt independently of one's choice of how to localize gravitational energy in general relativity. This is proved by explicitly computing dW/dt using various gravitational stress-energy pseudotensors (Einstein, Landau-Lifshitz, Møller) as well as Bergmann's conserved quantities which generalize many of the pseudotensors to include an arbitrary function of position. A discussion is also given of the problem of formulating conservation laws in general relativity and the role played by the various pseudotensors.