A few weeks ago I realized that many exploding rules can be censused by putting patterns in a torus, where a repeat is bound to happen eventually. For example, in DryLife (23/37), most patterns in a 16 by 16 torus stabilize in a few hundred generations. Even in non-exploding rules, this could be used to find agars. Here is a pattern that creates an interesting agar (EDIT: It's actually a wick):
Code: Select all
x = 16, y = 16, rule = B37/S23:T16,16
2obo3b3o3bo$3o2bob2obob2obo$b4ob2obobobo$2bobo2b3obob3o$2b4o2bob2ob3o$
2bob4o2b2obo$b4obo3b2o$bob5obob2ob2o$b3ob2ob2ob2o2bo$bob3ob2ob2obobo$o
4b2o3b2o3bo$2bo3b3ob3o2bo$ob8ob2obo$ob3o3bob2o$2o3b2obob2o2b2o$4ob2o7b
o!
I was thinking a feature could be added to apgsearch to search such rules by adding the torus size as a type of "symmetry", because for different rules there would obviously be different optimal sizes.