From LifeWiki
Revision as of 23:16, 7 September 2015 by Dvgrn (talk | contribs)
Jump to navigation Jump to search

Bounding box

For oscillators and spaceships, the bounding box can refer to the bounding box that contains the object in all phases of its period. This is what we usually use in the infobox. For example, weekender, which has a minimum bounding box of 16x10, has a "total" bounding box of 18x12. For the loafer, this is a 10x10 bounding box.
~Sokwe 05:09, 18 March 2013 (UTC)

Sorry,I didn't notice that. And I think it necessary to put up a notice on that. Zhuyifei1999 11:23, 18 March 2013 (UTC)
Shouldn't the bounding box of the glider be 4x4 then? Codeholic 08:14, 18 June 2013 (UTC)
The glider bounding box should actually be 3x4 (starting from generation 1). I'll go change it.
~Sokwe 04:37, 19 June 2013 (UTC)

To everyone trying to edit the bounding boxes of spaceships: we currently do not have a standard for what the bounding box is. For oscillators, the standard is to use the bounding box of the envelope of the pattern over its full period. Extending this to spaceships, it is reasonable to use the minimal bounding box of the envelope over the pattern's full period. Under this metric, Loafer has a 10x10 bounding box. Another reasonable definition of bounding box is to take the minimal single-phase bounding box of the spaceship. In this case, the bounding box for Loafer is 9x9.

One issue with both of these definitions is that they don't always match with the image in the infobox. For small patterns, the standard for the image (and hence the pattern file) is to take the phase with the smallest population. For example, the image for the MWSS shows neither the minimal bounding box over all phases nor the minimal single-phase bounding box (and the infobox for the MWSS gives the bounding box of the image, rather than either of the minima, showing just how inconsistent we are about this).

Personally, I have been using the all-phases definition, but I'll avoid changing anything on any current pages until a standard is agreed upon. At the risk of making the infobox too big, we could include both the minimal all-phase and minimal single-phase bounding boxes. Whatever we decide, I think the infobox should indicate that the bounding box given is not simply the bounding box of the image (e.g., label it "total bounding box" or "minimum bounding box" as is relevant).
~Sokwe 22:22, 2 September 2015 (UTC)

For oscillators, the bounding box across all phases is natural, but extending this to spaceships would give an infinite bounding box (or, if you only consider one period, the bounding box is dependent on your starting phase and therefore not uniquely defined). I think that the minimal bounding box is the more natural metric for spaceships. Calcyman (talk) 00:22, 5 September 2015 (UTC)

>if you only consider one period, the bounding box is dependent on your starting phase and therefore not uniquely defined
Under the all-phases definition, the minimal bounding box over all starting phases is chosen. To me, what seems like the most natural metric depends on the spaceship. For example, the glider fits in a 3x3 bounding box in all of its phases, and this feels like the natural "size" of the glider, but by the all-phases definition it has a 3x4 bounding box. On the other hand, I have a hard time thinking of the following as a width-10 ship since it has a width of 12 over its full period:
x = 38, y = 10, rule = B3/S23
If we do choose the single-phase definition, I think it should be labeled "minimum bounding box", which would hopefully avoid some confusion. The single-phase definition is certainly easier to describe.
~Sokwe 07:45, 6 September 2015 (UTC)
I'd certainly like to vote for the "minimal single-phase bounding box" option, and I agree that "minimum bounding box" or "smallest bounding box" would be a good label.
It was plenty enough trouble to find just the single-phase minimum for some recent high-period spaceships. With the Parallel HBK, you'd have to apply some slightly fiddly logic to be sure you had the right answer. The naive automated bounding-box-finding algorithm would have to run through a full cycle of the Parallel HBK for each of its 245,912 phases. Even for a simple glider, it's easy to get the "wrong" answer of 4x4 if you start with the wrong phase and don't check the other one. Dvgrn (talk) 23:15, 7 September 2015 (UTC)