This application is very similar to Mcell, and the author does invite users to send their unusual patterns to him for sharing and future upgrades. Can we do the same here? As long as users have the application, they can swap the *.rle files to run on their machines. Zipping would help the time/space bottlenecking of internet traffic/cost. What think thee?
I think RLE swapping is a great idea. Somebody should set up a Life-swapping site for that. In the meantime, I could host them on my super-reliable webserver, which is online more than half of the time.
I recently got golly 1.3 and I absolutely love the flexibility and the ease of use. I'm still getting over a little of the learning curve, with the more advanced layering, hashing, and scripting functionality, but I've already come across several standard 2-3 life patterns that I'd like to share (and hope to get credit for :).
Anyways, Alex, you mentioned you would be willing to host .rle files; what's the status on that? If nothing's yet available, simply email me at wilsonethan(at)bellsouth(dot)net with an RLE request, and I'd love to send a copy to whoever's interested.
To note, I've run across:
- a pattern that takes two VLSS on a collision course and converts them into a single glider (one going W and the other N, creating a SE glider).
- patterns that result in dual-shot gliders, new(?) still-life's
- 4000+ generation methuselah patterns (based off of some lexicon patterns)
I'm still researching, so I'm adding to that list as you read this. Thanks!
There have been a few attempts to put together online databases of Life patterns, but they tend to run aground on technicalities. It's kind of hard to make them useful: if they're big enough to contain everything anyone might be interested in, they're so big that it's hard to find anything interesting in them.
I think Koenig's Life Objects Database --
might be the biggest and best organized collection; you can even search for your "new" still lifes in it, though you have to rewrite them in SOF format to look them up --
Mark Niemiec also has a huge collection of glider constructions of various objects:
And RLE is a very Internet-friendly format; you could just post your patterns here, or maybe on comp.theory.cell-automata which occasionally gets contributions of this kind, and see what people think.
For methuselahs, you might want to have a look at
I tried to dig up your glider-producing two-VLSS collision -- but what's a VLSS? Heavyweight spaceships don't seem to produce a glider matching your description... but maybe I missed something in my enumeration:
# Creates a large set of two-HWSS collisions.
# Python script based on pd_glider.py.
# Run in Golly via File > Run Clipboard .
from glife import *
HWSS1 = pattern("4bo$5bo$o4bo$b5o!")
HWSS2 = pattern("b3o$o2bo$3bo$3bo$3bo$2bo!")
def collision (i, j):
return HWSS1[i](h-19,0) + HWSS2(j,7)
all = pattern ()
for i in xrange (0,39):
for j in xrange (0, 2):
all += collision (i, j) (100 * i, 333 * j)
all.display ("two-HWSS collisions")
# end of script
Keep the cheer,
Concerning the "VLSS," that was my typo; I meant LWSS. The reaction wasn't immediate, either; it took about 116 generations, and a block is also a by-product (which can be 'eaten' by adding a variety of still lifes in gen 0 - I'm still researching the best pattern on this). I can send you the RLE if you'd like, but until then here's a manually-typed text pattern:
Thanks, and enjoy!
> Concerning the "VLSS," that was my typo; I meant LWSS.
Aha, that makes more sense -- I was guessing "Very Large SpaceShip", but none of the possible collisions came out as you described.
Actually the script I sent also had a typo (a leftover 'h'), and the ships I used were really MWSS, not HWSS. There's a clean glider producer among the HWSS collisions after all --
2b2o$o4bo$6bo$o5bo2b3o$b6o2bo2bo$9bo$9bo3bo$9bo3bo$9bo$10bobo! Oops... repaired script is appended.
Your 2-LWSS collision produces a couple of six-cell predecessors of an "R-loaf" (a loaf plus one extra cell at the corner). These show up fairly often in random soups, and have been seen occasionally in Herschel conduits and so on. They're very handy in slow glider constructions, because of the way the block is constructed way off to one side. (Sorry if I'm spouting gibberish here -- Stephen Silver's Life Lexicon in Golly Help can clear up any of these you may be interested in.)
> a block is also a by-product (which can be 'eaten' by adding a variety of still lifes
> in gen 0 - I'm still researching the best pattern on this).
A fishhook eater can manage this nicely:
-- but if you want to wait until the last possible moment, I think you need an eater2:
> I can send you the RLE if you'd like, but until then here's a manually-typed text pattern...
ASCII format is just fine for small patterns -- you can paste it straight into Golly with no problem -- but you should be able to get the RLE just by selecting and copying out of Golly; no need to save a file.
##### begin-script #####
# Revised two-spaceship-collision.py
# Change 'HWSS1' and 'HWSS2' to experiment
# Creates a large set of two-*WSS collisions.
# Based on pd_glider.py.
HWSS1 = pattern("5bo$6bo$o5bo$b6o!")
HWSS2 = pattern("b3o$o2bo$3bo$3bo$3bo$3bo$2bo!")
MWSS1 = pattern("4bo$5bo$o4bo$b5o!")
MWSS2 = pattern("b3o$o2bo$3bo$3bo$3bo$2bo!")
LWSS1 = pattern("3bo$4bo$o3bo$b4o!")
LWSS2 = pattern("b3o$o2bo$3bo$3bo$2bo!")
def collision (i, j):
return HWSS1[i](-19,0) + HWSS2(j,7)
##### end-script #####
Thanks for the feedback! This is great stuff!
I've also been looking at the various possibilities around 2-glider collisions and their results, and have come across some interesting stuff. I'm sure this isn't new territory I'm covering, but it's fun for me to discover this 1st-hand. I made an RLE of the 2592-some-odd different collision types of this and have been using it as a basis for more complex glider-collision activities (building, eating, etc.) I'll be moving onto other glider collision types/combinations when I'm done with this "simple" 2-glider study. I can see this kind of research becoming a LOT of work; something that I may resort to scripting to aid in speeding up the generation of the "possibility space."
Anyways, all fun stuff, and this is just with 2-3 life! I can just imagine all the fun with the other rulesets... ;2)
> I've also been looking at the various possibilities around 2-glider collisions and their results...
When you're done with your independent survey of the terrain, you might find Golly's Syntheses folder interesting, if you haven't looked in there already -- two-glider-collisions.rle is a good summary page. There's been an amazing amount of research done on glider collisions, a lot of which is catalogued on Mark Niemiec's website ( http://home.interserv.com/~mniemiec/lifepage.htm ) and/or on pentadecathlon.com . But the search space is so large that even simple three-glider collisions are nowhere near completely catalogued.
> Anyways, all fun stuff, and this is just with 2-3 life! I can just imagine all the fun
> with the other rulesets...
I don't dare think about it. More research has been done on B3/S23 Life than on any other rule, by several orders of magnitude (at least)... which means that working with another ruleset basically amounts to abandoning an entire machine-shop worth of developed technology and jumping right back to the Stone Age. Or worse: for most of the millions of rules out there, people haven't even made their own first-level tools yet -- they're still just picking up things that they find lying around and bashing them together.
There are certainly other rules out there that could compete with Conway's Life in terms of the diversity of stuff that could be found or constructed; the problem is getting thousands of people to pick a single rule and play around with it for decades from thousands of different angles, to collect an equivalent amount of miscellaneous knowledge.
This happened with B3/S23 because of Martin Gardner's 1970 article in Scientific American, but it's hard to imagine quite that level of focus and excitement getting going again with another rule. Might have to wait until SETI finds an extraterrestrial civilization that happens to have concentrated first on, say, B368/S245 ...
I ran across GoE patterns, and was wondering how one can prove that there's no "father" to any proposed GoE? Given the successive generations of GoEs, I came up with similar, yet much smaller patters that result in similar children, and seem to have the same on/off state as the larger GoE patterns. Here's a couple I've come up with:
x = 5, y = 5, rule = B3/S23
x = 7, y = 6, rule = B3/S23
Oops, I meant to add this one too a modification of the 12x11 posted a couple days back:
x = 12, y = 10, rule = B3/S23
> Oops, I meant to add this one too a modification of the 12x11
> posted a couple days back:
> x = 12, y = 10, rule = B3/S23
Unfortunately it's not easy to pare down that 12x11 pattern to 12x10 without picking up some possible parents. Here's one for your 12x10:
x = 14, y = 16, rule = B3/S23
And technically it's okay if the parent pattern adds extra cells around the outside -- unless the required blank cells around the edge are included in the GOE, which would have to count as a bigger bounding box. Here's a parent that also produces a scattering of 9 extra ON cells at the top and bottom:
x = 14, y = 12, rule = B3/S23
The posting on LifeNews shows all of the ON and OFF cells required for the 12x11 to be a GOE. I ran that pattern through a Java beta version of WinLifeSearch (see http://www.ics.uci.edu/~eppstein/ca/search.html -- the Windows executable for 'lifesrc' is in good working order, though it's a bit tricky to use).
What I found was that parent patterns started showing up as soon as I tried relaxing just one of the OFF-cell conditions around the edges, or even in the middle... though I only tried a few cells at random, so there might be an ON or OFF condition or two somewhere that could be taken out; I don't know whether Achim's search algorithm included a back-check for unnecessary cells.
Anyway, the fragility of GOE status at 12x11 suggests that orphan patterns much smaller than this one -- say, 8x8 or below -- aren't too likely to exist.
Very possibly there _is_ a 12x10 hiding out there somewhere, but finding it is quite a challenge -- a brute-force enumeration of the likely possibilities would take until the heat death of the universe to test with WinLifeSearch.
Wow, there's a lot to finding a GoE, I see. I got WinLifeSearch, and I agree that it's got a learning curve that takes a rock climber to figure out, and I'm still trying to figure out how to use it to search only for parent patterns. I did find a rather interesting-looking 2p oscillator from an R-pentomino pattern while fiddling with it, but it looks like learning to use this tool will take more time than I have at work. :2)
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.