A boundary stress tensor for higher-derivative gravity in AdS and Lifshitz backgrounds
Olaf Hohm, Erik Tonni
TL;DR
This work develops a universal method to define Brown-York boundary stress tensors for curvature-squared gravity by introducing auxiliary tensor fields to reduce derivative orders, enabling a well-posed variational principle and covariant boundary terms. It validates the construction in 3D new massive gravity by computing central charges and BTZ black hole masses, and extends to Lifshitz backgrounds with a covariant counterterm, highlighting both successes (correct central charges, chiral point behavior) and ambiguities (Lifshitz counterterms not uniquely fixed). It provides explicit formulas for the boundary stress tensor and demonstrates finite results after appropriate counterterms, with implications for AdS/CFT and non-relativistic holography. The results clarify the trade-offs between bulk unitarity and boundary energy in higher-derivative gravity and lay groundwork for extending the approach to higher dimensions and additional Lifshitz solutions.
Abstract
We investigate the Brown-York stress tensor for curvature-squared theories. This requires a generalized Gibbons-Hawking term in order to establish a well-posed variational principle, which is achieved in a universal way by reducing the number of derivatives through the introduction of an auxiliary tensor field. We examine the boundary stress tensor thus defined for the special case of `massive gravity' in three dimensions, which augments the Einstein-Hilbert term by a particular curvature-squared term. It is shown that one obtains finite results for physical parameters on AdS upon adding a `boundary cosmological constant' as a counterterm, which vanishes at the so-called chiral point. We derive known and new results, like the value of the central charges or the mass of black hole solutions, thereby confirming our prescription for the computation of the stress tensor. Finally, we inspect recently constructed Lifshitz vacua and a new black hole solution that is asymptotically Lifshitz, and we propose a novel and covariant counterterm for this case.
