Abstract
Abstract
Firewalls in black holes are easiest to understand by imposing time reversal invariance, together with a unitary evolution law. The best approach seems to be to split up the time span of a black hole into short periods, during which no firewalls can be detected by any observer. Then, gluing together subsequent time periods, firewalls seem to appear, but they can always be transformed away. At all times we need a Hilbert space of a finite dimension, as long as particles far separated from the black hole are ignored. Our conclusion contradicts other findings, as these assume that information that entered into a black hole, cannot re-emerge. But re-emergence of that information is exactly what our version of firewalls is supposed to ensure. Indeed, the firewall transformation removes the problems caused by entanglement between very early and very late in- and out-particles, in a far from trivial way.