Three-surface twistors and conformal embedding (Q794002)

In ''Quasi-local mass and angular momentum in general relativity'', Proc. R. Soc. Lond., Ser. A 381, 53-63 (1982), \textit{R. Penrose} described a possible definition of mass and angular momentum associated with a space- like two-surface S in an arbitrary space-time. His idea was to associate first to S a 4-dimensional ''twistor space'' as the space of solutions of the twistor equation restricted to S. In the ambient space-time the twistor equation will generally have no solutions. In this paper the author considers the intermediate case of restricting the twistor equation to a space-like three-surface \(\Sigma\). It is particularly interesting to know whether the resulting ''three-surface twistor'' equations admit a 4-dimensional space of solutions for in this case further restricting to two-surfaces \(S\subset \Sigma\) allows one to relate the twistor spaces associated to different choices of \(S\subset \Sigma\). This turns out to be a useful tool in computing the Penrose mass in several natural examples [Proc. R. Soc. Lond., Ser A 388, 457-477 (1983)]. By comparing three-surface twistors with Penrose's ''hypersurface twistors'' the author first shows that \(\Sigma\) admits a 4-dimensional three-surface twistor space if and only if \(\Sigma\) can be embedded in a conformally flat space whilst preserving its first and second fundamental forms. By adapting the ''local twistor transport'' to \(\Sigma\) he then goes on to identify the integrability conditions for a 4-dimensional space in terms of the Weyl curvature of the ambient space near \(\Sigma\). Although there is a slight gap in the logic at this point in the paper, the conclusions are correct.