Forums for Conway's Game of Life
https://conwaylife.com/forums/
Thanks! What about oscillators or spaceships?mniemiec wrote: June 15th, 2021, 10:49 pmOnce all cells have been mapped without encountering such a problem, you have successfully unwrapped your still life.
Basically the same thing, except that for oscillators you're effectively dealing with an additional vertical dimension in a cylinder, and you do object-detection and connectivity tracing for 3-dimensional objects. Spaceships are similar, except the connection from the top and bottom are shifted by the spaceship velocity.
I was about to write up an explanation about using cell connectivity graphs with nodes embedded into 3D spacetime, but it looks like you got here before me! One thing I would say is that spaceships are very tricky on toric grids because the only way they can really be distinguished from wicks/agars/infinite objects just via connectivity graphs is the topology of their embedding, which I doubt is easy to find algorithmically. Probably what I would say is the most practical way (still a slow-ish way sometimes, but it shouldn't be awful) to tell if the wick or agar your program spat out is actually infinite or just a spaceship is to put every phase by themselves on an infinite grid and run them — if any of them are spaceships that match the evolution of the pattern on the torus, your pattern is a spaceship, otherwise it's probably not. Do note that this will miss spaceships that "wrap" more than once around the grid, like, say, a lot of the c/5ds here — there are probably clever tricks that could catch more of these cases, but alas since I don't know how to do any kind of computational topology (is that even a thing?) I don't know of an algorithm that would identify spaceships completely accurately.mniemiec wrote: June 15th, 2021, 11:44 pmBasically the same thing, except that for oscillators you're effectively dealing with an additional vertical dimension in a cylinder, and you do object-detection and connectivity tracing for 3-dimensional objects. Spaceships are similar, except the connection from the top and bottom are shifted by the spaceship velocity.
Code: Select all
x = 4, y = 4, rule = B3/S23
b3o$3bo$ob2o$2o!
I think that I was the one who picked that form for the Elementary Conduits Collection, and I'm not sure if I had any particularly good reason. It seemed less easy to confuse it with B-heptominoes, maybe, and also it was more likely that the various century predecessors would have settled down to exactly that form. (?)ihatecorderships wrote: June 17th, 2021, 12:18 pm Why is the canonical version of the century this:and not the bookend or the normal B-hept-minus-one cell predecessor?Code: Select all
x = 4, y = 4, rule = B3/S23 b3o$3bo$ob2o$2o!
I doubt that there's anything common. I've never seen anything converge to the century sequence at the century's great-grandchild.hotdogPi wrote: June 17th, 2021, 1:46 pm I use that form of the century when running my program to make sure everything gets counted. I can't find a common predecessor that doesn't reach its standard parent, but I'm sure there is one and I'm not finding it.
Maybe it's done in order to make the name century more accurate?dvgrn wrote: June 17th, 2021, 12:45 pmI think that I was the one who picked that form for the Elementary Conduits Collection, and I'm not sure if I had any particularly good reason. It seemed less easy to confuse it with B-heptominoes, maybe, and also it was more likely that the various century predecessors would have settled down to exactly that form. (?)
This sometimes comes up instead of a bookend:ihatecorderships wrote: June 17th, 2021, 12:18 pmWhy is the canonical version of the century this: ... and not the bookend or the normal B-hept-minus-one cell predecessor?
Code: Select all
x = 5, y = 4, rule = B3/S23
bboo$3boo$bobo$obo!
As does this, which I find to be more common:mniemiec wrote: June 17th, 2021, 4:38 pmThis sometimes comes up instead of a bookend:ihatecorderships wrote: June 17th, 2021, 12:18 pmWhy is the canonical version of the century this: ... and not the bookend or the normal B-hept-minus-one cell predecessor?Code: Select all
rle
Code: Select all
x = 4, y = 4, rule = B3/S23
b3o$o$o$bo!I've never seen that occur naturally. Can you provide an example?mniemiec wrote: June 17th, 2021, 4:38 pmThis sometimes comes up instead of a bookend:Code: Select all
x = 5, y = 4, rule = B3/S23 bboo$3boo$bobo$obo!
Not that one in particular, but I've seen the related predecessor (where the bookend itself has an open hinge) occur occasionally.MathAndCode wrote: June 17th, 2021, 6:54 pmI've never seen that occur naturally. Can you provide an example?
Looks like mniemiec's variant is just common enough that a natural result shows up in the octohash database -- just one, though!mniemiec wrote: June 17th, 2021, 7:18 pmNot that one in particular, but I've seen the related predecessor (where the bookend itself has an open hinge) occur occasionally.MathAndCode wrote: June 17th, 2021, 6:54 pmI've never seen that occur naturally. Can you provide an example?
Code: Select all
x = 7, y = 12, rule = B3/S23
b2o$obo$bo$4b2o$3bo2bo$3bo2bo$4b2o3$4bo$3b2o$3bobo!I think the more common "variant" was this one:dvgrn wrote: June 18th, 2021, 7:07 am Looks like mniemiec's variant is just common enough that a natural result shows up in the octohash database -- just one, though!
Code: Select all
x = 7, y = 12, rule = B3/S23 b2o$obo$bo$4b2o$3bo2bo$3bo2bo$4b2o3$4bo$3b2o$3bobo!
Code: Select all
x = 4, y = 5, rule = B3/S23
bo$3o$o2bo$3bo$2bo!I don't know of any conduits that produce much of anything besides 4bo$5b2o$obobo$obo! The only exception I can remember is the Fx158, which produces just 2bo$3b2o$obo$o! instead.ColorfulGabrielsp138 wrote: June 19th, 2021, 2:28 am Is there a Herschel conduit that outputs the xp0_16o4 Herschel parent?
Does xp0_16o4 have a known glider synthesis?
Code: Select all
x = 120, y = 48, rule = LifeHistory
86.5D9.D3.5D2.3D$86.D12.2D3.D5.D3.D$86.D5.D3.D3.D3.D5.D3.D$86.3D4.D.D
4.D4.3D3.3D$86.D7.D5.D7.D.D3.D$86.D6.D.D4.D3.D3.D.D3.D$86.D5.D3.D2.3D
3.3D3.3D3$16.5D9.D3.5D2.3D$16.D12.2D3.D5.D3.D$16.D5.D3.D3.D3.D9.D$16.
3D4.D.D4.D4.3D3.3D$16.D7.D5.D7.D5.D$16.D6.D.D4.D3.D3.D.D3.D$16.D5.D3.
D2.3D3.3D3.3D55.2A$99.2A12.A$111.3A$99.4A7.A$96.A.A4.A6.2A$94.3A.2A2.
2A4.4B$93.A4.B3.4B.3B$90.A2.4AB2AB3.6B$90.3A3.A.2AB.7B.B$93.A4.14B$
32.2A58.2AB4.12B3.2B$7.2A24.A58.5B.20B$8.A13.2A6.3A61.25B$8.A.AB9.B2A
2B4.A61.28B$9.2AB.3B6.4B.3B62.28B$11.7B2.11B7.7B.B44.28B$11.22B.B3.
13B.B38.28B$12.24BD21B34.25BADB$11.26BD20BAD31.25B2DCA$9.28B2D18B2DCA
28.20B2.4BADAD$7.29B2D2B2A12BABADAD27.19B5.4BAD$7.2BD26BD3B2A12BABAD
29.2BD16B5.5B$6.3BDBD4B.30B3.2B2.2B30.3BDBD4B.9B6.3B$7.2B3D4B2.16B2.
10B41.2B3D4B2.7B$6.5BD4B2.7B.4B7.6B43.5BD4B3.5B$5.10B4.6B2.2B9.3B44.
10B5.4B$4.4B11.6B13.B45.4B10.4B.B2A$4.3B13.3B14.2A45.3B11.2A3.BA.A$2.
4B12.7B12.A44.4B13.A6.A$2.2A14.2A.B.2A13.3A41.2A12.3A7.2A$3.A15.A3.A
16.A42.A12.A$3A13.3A5.3A53.3A$A15.A9.A53.A!Why did you prefer RLE instead of apgcode?dvgrn wrote: June 19th, 2021, 8:04 amI don't know of any conduits that produce much of anything besides 4bo$5b2o$obobo$obo! The only exception I can remember is the Fx158, which produces just 2bo$3b2o$obo$o! instead.ColorfulGabrielsp138 wrote: June 19th, 2021, 2:28 am Is there a Herschel conduit that outputs the xp0_16o4 Herschel parent?
Does xp0_16o4 have a known glider synthesis?
I also don't think there's a known glider synthesis. I bet someone could come up with one if they really had to, but it's hard to imagine why it would ever be an improvement on a much cheaper standard synthesis. There's nothing for that 2bo$3b2o$b2o$o! active pattern from synthesise_pattern or the octohash database, for example.Code: Select all
x = 120, y = 48, rule = LifeHistory 86.5D9.D3.5D2.3D$86.D12.2D3.D5.D3.D$86.D5.D3.D3.D3.D5.D3.D$86.3D4.D.D 4.D4.3D3.3D$86.D7.D5.D7.D.D3.D$86.D6.D.D4.D3.D3.D.D3.D$86.D5.D3.D2.3D 3.3D3.3D3$16.5D9.D3.5D2.3D$16.D12.2D3.D5.D3.D$16.D5.D3.D3.D3.D9.D$16. 3D4.D.D4.D4.3D3.3D$16.D7.D5.D7.D5.D$16.D6.D.D4.D3.D3.D.D3.D$16.D5.D3. D2.3D3.3D3.3D55.2A$99.2A12.A$111.3A$99.4A7.A$96.A.A4.A6.2A$94.3A.2A2. 2A4.4B$93.A4.B3.4B.3B$90.A2.4AB2AB3.6B$90.3A3.A.2AB.7B.B$93.A4.14B$ 32.2A58.2AB4.12B3.2B$7.2A24.A58.5B.20B$8.A13.2A6.3A61.25B$8.A.AB9.B2A 2B4.A61.28B$9.2AB.3B6.4B.3B62.28B$11.7B2.11B7.7B.B44.28B$11.22B.B3. 13B.B38.28B$12.24BD21B34.25BADB$11.26BD20BAD31.25B2DCA$9.28B2D18B2DCA 28.20B2.4BADAD$7.29B2D2B2A12BABADAD27.19B5.4BAD$7.2BD26BD3B2A12BABAD 29.2BD16B5.5B$6.3BDBD4B.30B3.2B2.2B30.3BDBD4B.9B6.3B$7.2B3D4B2.16B2. 10B41.2B3D4B2.7B$6.5BD4B2.7B.4B7.6B43.5BD4B3.5B$5.10B4.6B2.2B9.3B44. 10B5.4B$4.4B11.6B13.B45.4B10.4B.B2A$4.3B13.3B14.2A45.3B11.2A3.BA.A$2. 4B12.7B12.A44.4B13.A6.A$2.2A14.2A.B.2A13.3A41.2A12.3A7.2A$3.A15.A3.A 16.A42.A12.A$3A13.3A5.3A53.3A$A15.A9.A53.A!
I can copy RLE straight out of Golly, and someone can copy and paste RLE straight back into Golly and see what it is -- or get a picture of it by decoding it manually, for small enough patterns.
Why did you prefer apgcode instead of RLE?
One advantage RLE has over apgcode is that it encodes the rule with the pattern, making it easy to copy/paste a pattern into a progarm like Golly, instead of the rule having to be carried through a separate and independent channel. apgcode is also only suitable for encoding a small subset of patterns (i.e. still-lifes, oscillators, and spaceships); patterns that aren't one of these can't easily be encoded. In particular, linearly expanding patterns are encoded as yl_... which only encodes a histogram of the pattern's population, but not its actual bit image, making the mapping from apgcode to bit image not unique, and also impossible to reverse. xs0_... is a work-around for arbitrary patterns, but apgcode wasn't really designed for that. It also can't encode images in rules with more than two states.
I had an apgcode to RLE converter program but I can't attach it to Golly.dvgrn wrote: June 19th, 2021, 10:44 pmI can copy RLE straight out of Golly, and someone can copy and paste RLE straight back into Golly and see what it is -- or get a picture of it by decoding it manually, for small enough patterns.
I'm not as confident of my ability to get the picture exactly right for a small-to-middling apgcode -- just haven't practiced as much.
Code: Select all
x = 12, y = 12, rule = TripleB3S23
2F$.F$.2F$2.2F6$9.3F$9.F$10.F!