I dont know how to say this well, because I'm just starting to think about it, so I apologize if is not clear.
I have a Event-driven particle simulator that schedule the events (collisions between particles) in a CBT. When an event is attended the tree change, but "just a little", because particles collide, change their momentum and there are new collisions to attend. As this is a extended system I can attend an event that if is far enough of the present one without changing the evolution of the system.
The serial algorith is like this.
1.- Find first events.
2.- Make a CBT.
3.- Attend the first one (evolve the time and change momentum).
3.1.- Predict next events for the particles.
3.2- Add those events to the CBT.
3.3.- Search the first event in the tree.
3.4.- Go to 3
My idea is to attend in parallel events that are far enough from each other, because the "information" take a while to spread over the system (something like a light-cone in relativity).
So, my question: is there a way to do something like this? is this scheme applicable to other problems?
I'll be glad to extend the discussion in this.