This reminds me of something that’s always struck me as odd about large-scale physics simulations: using invariant spacial coordinates when doing a gravitational simulation is totally absurd. Wouldn’t precision problems be much less of an issue if you distorted space according to gravitational influence, as happens in real life?