Breeders and such like

[Dougiefromthecosmos]

Hi

Ive recently got interested again in the game of life with the introduction of the Gemini pattern. I first came across cellular automota when Stephen Wolfram released "A New Kind Of Science". I have been messing with golly since coming across the gemini pattern. Im really looking forward to other peoples new patterns and discoveries, very like real life in that there is so much to be discovered within the boundaries of the game of life, its completely fascinating, you just never know what ur gonna get (Forrest gump comes to mind).

Well anyway my reason for posting on this forum is to ask a few questions.

Has anyone developed software that attempts to breed known still lifes, gliders, spaceships, etc.

Has anyone developed anything that attempts to take a realtime input ie from say a sound file, and perceive it somehow as a life pattern?

and also, are there any versions played on a 3d grid, Ive tried searching for them but cant seem to find any, is there some intrinsic reason it doesnt work the same on a 3d grid.

Thanks for listening.

Doug

[Axaj]

Hi, welcome to the forums. To answer your questions:

1: No, that would be difficult to program. Also, breeders create rakes and guns, while puffers and rakes/guns make still lifes/oscillators and spaceships respectively. Take a look at these: http://conwaylife.com/wiki/index.php?title=Gun http://conwaylife.com/wiki/index.php?title=Puffer http://conwaylife.com/wiki/index.php?title=Rake http://conwaylife.com/wiki/index.php?title=Breeder

2: No, but that would certainly be possible.

3: It is certainly possible, but programs for it are hard to find. A program called Breve has an example implementation, which you might want to look into.

[H. V. McIntosh]

..... and also, are there any versions played on a 3d grid, Ive tried searching for them but cant seem to find any, is there some intrinsic reason it doesnt work the same on a 3d grid. .....

Probably not, but there would be a huge number of rules to have to search to find out. Likely they would lie in Wolfram's Class IV, but one wonders what other restrictions might help to guide such a search?
Carter Bays has published various results, for instance in Wolfram's Complex Systems.''

[calcyman]

I'll answer your questions in reverse:

... and also, are there any versions played on a 3d grid ...

There are lots of options as you move to three dimensions. I guess that you would want one of the following three lattices:

• The normal cubic lattice (6, 18 or 26 neighbours);
• The face-centred cubic lattice (12 neighbours);
• The body-centred cubic lattice (8 or 14 neighbours).

Out of those six neighbourhoods, only the 3D Moore (26 neighbours) has really been explored much. There is a neat rule, B6/S4578, where Conway's Game of Life can be simulated on a bilayer sandwiched between two planes. It supports gliders, as well. Most of the research in this area is by Carter Bays.

The other lattices may yield more interesting results. FCC and BCC are common crystal structures, with Voronoi cells of rhombic dodecahedra and truncated octahedra, respectively. If I remember correctly, copper forms a FCC structure. BCC is the most symmetrical of the three lattices.

There is also an interesting 4-neighbour tessellation, with the same structure and connectivity as a diamond.

Has anyone developed anything that attempts to take a realtime input ie from say a sound file, and perceive it somehow as a life pattern?

The reverse has been accomplished. People have derived music from Life patterns; you can see how if you buy Professor Andy Adamatzky's new bestseller, Game of Life Cellular Automata. 3D Life is mentioned in there as well, if I remember correctly.

Has anyone developed software that attempts to breed known still lifes, gliders, spaceships, etc.

Do you mean:

A) Construct a breeder out of a glider synthesis for a pattern; or
B) Amalgamate existing patterns to make new ones?

The answer to (A) is "no", whereas the answer to (B) is "yes". Hersrch combines existing Herschel conduits to make new tracks and loops, i.e. oscillators.

[Dougiefromthecosmos]

Wow, thank you for taking the time to answer my questions guys. You have given me a lot more to think about

[Batmanifestdestiny]

You know, that ship breeder idea got me thinking.


People have known for quite a while that by combining ships and adding a couple of extra cells that a new, larger ship is made. What if you made a program that took existing ships (LSS, MSS, HSS, glider), positioned them within a certain boundary, and have their "fitness" be tested on whether or not it moves (could be done by calculating n cells one direction versus n cells the other, to prevent explosions from being counted as ships)? Then, after it finds a couple of ships that work, then it could put those into the candidate pool, and then take concepts from those in future tests, combined with a pseudorandom number method of slight mutation!


When I say "slight mutation" I mean like one ship part of by a couple of cells, an extra cell placed here and there, etc.


If you left it running long enough, then it could give you an rle file of the ships that it has found, and some of them could be really interesting!