5 February 2014
Venue: Maths 103
Attempts to reconcile observed unidirectional evolution with time-symmetric laws generally assume that the universe must have begun in a state of exceptionally low entropy. My seminar (based on arXiv: 1310.5157) will question this, taking as an example the zero-energy Newtonian N-body problem, which has generic and zero-measure solutions. For all of them, one can define scale-invariant measures of complexity (clustering) and information of the instantaneous shape of the system. All the generic solutions divide at a unique point P, on either side of which the complexity and information grow between monotonically rising bounds. Any observer must be in one half and, taking the direction of increasing complexity to define the arrow time, will take the point P of greatest uniformity to be the beginning of time. For internal observers, each generic solution will therefore have one past and two futures. The zero-measure solutions are like half a generic solution and have one past and one future. General relativity shares the basic structure of time-symmetric Newtonian theory that leads necessarily to observed irreversibility, so similar behaviour may hold for it too.