Talk:Elementary cellular automaton/Infinite length

Edge bits

(about ignoring cells beyond the edges) But you can't stick to a simple version, and infinite padding beyond the edges must be included somehow. Suppose the visible part of the cells ends with "...110" and is followed by repeating 0s to infinity, and rule is just 1 (000->1). Now after one iteration those three cells become "000", but followed by repeating 1s instead. You can add another constraint that padding/edge cells don't change, but that would make the "infinite" part kind of pointless because changes won't propagate beyond edge cells. --Ledrug (talk) 17:54, 21 March 2014 (UTC)

Is this the only reason you marked Python wrong? (You actually added it to the Perl section).

If so then it may be in error as the task states the meaning of infinite as being:

In other words, to describe the state of the automaton, you need a finite number of adjacent cells, along with their individual state, and you then consider that the individual state of each of all other cells is the negation of the closest individual cell among the previously defined finite number of cells.

I think the Python is correct under that definition of 'infinity'. --Paddy3118 (talk) 06:22, 23 March 2014 (UTC)

The description could use some work. Assuming "adjacent" really means "consecutive" here, the intention of the above statements seems to me to say that, you need to write down a minimal length of cells, and assume cells on both sides of this segment, extending to infinity, are all of the same value, which is the negation of the first and last of the cells you did write down. It is not stated unambiguously, and it's not capable of representing all possible transitions even if it can represent the initial state (you could end up with all cells having the same value, so negation isn't going to cut it.)

Regardless of the wording of the description, the intention is clear: try to represent an infinitely long row of cells, if only a finite portion is not of constant value. Given the simple case of rule 1 (000->1, everything else->0), and a starting state of every cell being 1 (what else can your "eca_infinite(1)" mean?), the cells should flip between all 0s and all 1s at each step. The python code does not reproduce that. I expect you to argue that your code fits one interpretation of the (imprecise) task description, and I expect myself to throw my hands up and say "suit yourself", but please do think about it before that. --Ledrug (talk) 07:01, 23 March 2014 (UTC)