Re: Oscillator Discussion Thread
Posted: August 20th, 2017, 5:05 pm
I don't even know what that is.dvgrn wrote: Did you try checking the jslife collection before posting your question?
Forums for Conway's Game of Life
https://conwaylife.com/forums/
I don't even know what that is.dvgrn wrote: Did you try checking the jslife collection before posting your question?
Anyone who's wanting to get serious with Life-related technologies probably should.wwei23 wrote:I don't even know what that is.dvgrn wrote: Did you try checking the jslife collection before posting your question?
Code: Select all
x = 45, y = 22, rule = B3/S23
4bo12bo9bo12bo$4b3o3b2o3b3o9b3o3b2o3b3o$7bo2b2o2bo15bo2b2o2bo$6b2o6b2o
13b2o6b2o$2o18b2ob2o18b2o$bo6b6o6bo3bo6b6o6bo$bobo3bo6bo3bobo3bobo3bo
6bo3bobo$2b2o2bo8bo2b2o5b2o2bo8bo2b2o$5bo10bo11bo10bo$5bo4b2o4bo11bo
10bo$b2o2bo3bo2bo3bo2b2o3b2o2bo4b2o4bo2b2o$b2o2bo3bo2bo3bo2b2o3b2o2bo
4b2o4bo2b2o$5bo4b2o4bo11bo10bo$5bo10bo11bo10bo$2b2o2bo8bo2b2o5b2o2bo8b
o2b2o$bobo3bo6bo3bobo3bobo3bo6bo3bobo$bo6b6o6bo3bo6b6o6bo$2o18b2ob2o
18b2o$6b2o6b2o13b2o6b2o$7bo2b2o2bo15bo2b2o2bo$4b3o3b2o3b3o9b3o3b2o3b3o
$4bo12bo9bo12bo!
Code: Select all
x = 24, y = 23, rule = B3/S23
4bo13bo$4b3o9b3o$7bo7bo$6b2o7b2o$2o19b2o$bo6b7o6bo$bobo3bo7bo3bobo$2b
2o2bo9bo2b2o$5bo11bo$5bo11bo$5bo5bo5bo2b2o$5bo4bobo4bo3bo$5bo5bo5bo2bo
$5bo11bo3b3o$5bo11bo5bo$2b2o2bo9bo2b2o$bobo3bo7bo3bobo$bo6b7o6bo$2o19b
2o$6b2o7b2o$7bo7bo$4b3o9b3o$4bo13bo!
Code: Select all
x = 28, y = 28, rule = B3/S23
12bo2bo$12b4o$6bo14bo$6b3o3b4o3b3o$9bo2bo2bo2bo$8b2o8b2o$2b2o20b2o$3bo
6b8o6bo$3bobo3bo8bo3bobo$4b2o2bo10bo2b2o$7bo12bo$7bo12bo$2ob2o2bo5b2o
5bo2b2ob2o$bobo3bo4bo2bo4bo3bobo$bobo3bo4bo2bo4bo3bobo$2ob2o2bo5b2o5bo
2b2ob2o$7bo12bo$7bo12bo$4b2o2bo10bo2b2o$3bobo3bo8bo3bobo$3bo6b8o6bo$2b
2o20b2o$8b2o8b2o$9bo2bo2bo2bo$6b3o3b4o3b3o$6bo14bo$12b4o$12bo2bo!
These octagon variants are already well-known. Here is an example pattern from the jslife pattern collection (in osc/o0004.lif):wwei23 wrote:Octagon 6... Can anyone fit a still life in here?... Octagon 8
Code: Select all
x = 29, y = 25, rule = B3/S23
5bo$b2obobo4b2obob2o$b2obobo4bob2ob2o3b2o$4bob2o13bo$b4o2bo$o4b18o$b3o
bo17bo$3b3o18bo$5bo19bo2bo$5bo3b2o2b2ob2o2b2o3bob2o$5bo3b2o2b2ob2o2b2o
3bo$b2o2bo19bo$b2o2bo3b2o9b2o3bo$5bo3b2o9b2o3bo$5bo19bo$5bo3b2o2b2ob2o
2b2o3bo$2b2obo3b2o2b2ob2o2b2o3bob2o$2bo2bo19bo2bo$6bo17bo$7bo15bo$8b
15o2$9bo11bo$8b2o3b2ob2o3b2o$13b2ob2o!
I did, but didn't see their exact forms, which is why I posted it.Sokwe wrote:These octagon variants are already well-known. Here is an example pattern from the jslife pattern collection (in osc/o0004.lif):wwei23 wrote:Octagon 6... Can anyone fit a still life in here?... Octagon 8Please check jslife before posting.Code: Select all
x = 29, y = 25, rule = B3/S23 5bo$b2obobo4b2obob2o$b2obobo4bob2ob2o3b2o$4bob2o13bo$b4o2bo$o4b18o$b3o bo17bo$3b3o18bo$5bo19bo2bo$5bo3b2o2b2ob2o2b2o3bob2o$5bo3b2o2b2ob2o2b2o 3bo$b2o2bo19bo$b2o2bo3b2o9b2o3bo$5bo3b2o9b2o3bo$5bo19bo$5bo3b2o2b2ob2o 2b2o3bo$2b2obo3b2o2b2ob2o2b2o3bob2o$2bo2bo19bo2bo$6bo17bo$7bo15bo$8b 15o2$9bo11bo$8b2o3b2ob2o3b2o$13b2ob2o!
Good question. Would you like to see my octagon collection so far?Gamedziner wrote:Could some of these octagons act as sort of OWSS-type emulators?
This is off-topic, as all these octagon components have been discovered. You shouldn't have posted your octagons if you saw the base components in a collection but not the specific oscillators themselves.wwei23 wrote:Good question. Would you like to see my octagon collection so far?
Code: Select all
x = 4, y = 10, rule = Life
bo$3bo$bobo$bo$bo$2bo$2bo$obo$o$2bo!
Here's a completion of that, found with WLS:wwei23 wrote:Almost P4:Code: Select all
x = 4, y = 10, rule = Life bo$3bo$bobo$bo$bo$2bo$2bo$obo$o$2bo!
Code: Select all
x = 18, y = 22, rule = B3/S23
13bo$12bobo$12bobo$11b2ob2o$17bo$9b4obob2o$9bo4bobo$11bo2bobo$11bobobo
$7b3o3$8b3o$2bobobo$bobo2bo$bobo4bo$2obob4o$o$2b2ob2o$3bobo$3bobo$4bo!
It's p3.Kazyan wrote:Here's a completion of that, found with WLS:wwei23 wrote:Almost P4:Code: Select all
x = 4, y = 10, rule = Life bo$3bo$bobo$bo$bo$2bo$2bo$obo$o$2bo!
Code: Select all
x = 18, y = 22, rule = B3/S23 13bo$12bobo$12bobo$11b2ob2o$17bo$9b4obob2o$9bo4bobo$11bo2bobo$11bobobo $7b3o3$8b3o$2bobobo$bobo2bo$bobo4bo$2obob4o$o$2b2ob2o$3bobo$3bobo$4bo!
I looked into this with Beluchenko's version of WLS to see if the weasel had any more secrets. In a 12x20 search space for up to 8 stator cells, no new p3s with at most 21 cells showed up. So if there are any more small p3s out there, they're either sprawling or have 9+ stator cells. I give it 2:1 odds that there aren't any.dvgrn wrote:Yes, well, maybe I can weasel out by saying that your oscillator's bounding box may be small, but it's almost all rotor.83bismuth38 wrote:Really? because i discovered a small one with no unique properties and was new...dvgrn wrote: In general, any question starting with "is this p3..." is going to be answered "no", unless it has a very large number of cells in its rotor.
With 33 cells to move around, there are a lot of possible arrangements, many of which may not have been seen before. New examples probably still won't be all that interesting, unless someone is trying for an exhaustive collection -- and people learn pretty quick not to try to do that when there are too many trivial variants.
Zacinfinity's billiard table has only four rotor cells. At that size, if you aren't seeing it in existing collections, the odds are very very high that it's just because you're not looking in the right place...!
The P3 space has been exuhaustively searched up to 20 bits, so you shouldn't find any less than 21. 21 is theoretically possible, as I have searched that space manually (mostly for 400+ boring extrapolations of smaller oscillators), but hit has never been subject to an exhaustive computer search yet.Kazyan wrote:I looked into this with Beluchenko's version of WLS to see if the weasel had any more secrets. In a 12x20 search space for up to 8 stator cells, no new p3s with at most 21 cells showed up. So if there are any more small p3s out there, they're either sprawling or have 9+ stator cells. I give it 2:1 odds that there aren't any.
How would an exhaustive search be done? WLS doesn't seem up to the task, since there isn't really a way for it to stop expanding the search space to infinity for diagonal sort orders.mniemiec wrote:The P3 space has been exuhaustively searched up to 20 bits, so you shouldn't find any less than 21. 21 is theoretically possible, as I have searched that space manually (mostly for 400+ boring extrapolations of smaller oscillators), but hit has never been subject to an exhaustive computer search yet.
I wrote:The P3 space has been exuhaustively searched up to 20 bits, so you shouldn't find any less than 21. 21 is theoretically possible, as I have searched that space manually (mostly for 400+ boring extrapolations of smaller oscillators), but hit has never been subject to an exhaustive computer search yet.
As I understand it, it was done by several separate WLS searches within constrained areas (one rectangular, and one diagonal), those areas having been previously proven sufficiently large to hold all oscillators. While I can't speak specifically to the P3 oscillator question, for n-bit still-lifes, (n-3)x(n-3) and (n-2)x(n-4) rectangles are sufficient, or one (n-2)x(n-3) rectangle could cover both of them. Constraints on oscillators are even more restrictive.Kazyan wrote:How would an exhaustive search be done? WLS doesn't seem up to the task, since there isn't really a way for it to stop expanding the search space to infinity for diagonal sort orders.
Code: Select all
x = 8, y = 10, rule = B3/S23
2o4bo$o2bobobo$2b3o2bo2$4b3o$3bo$3bobo2$4bobo$5b2o!
Interesting. That's 3 new ones (that, plus two variants w/snake or cis-carrier instead of trans-carrier). I'm not surprised, as I only manually searched for "obvious" extensions of smaller objects. There are bound to be a few new non-trivial ones that would likely only show up in exhaustive searches.Kazyan wrote:If anyone has some 15-year-old email from one of the Old Guards of this hobby that can elaborated on the constraints for p3 oscillators, now is the time.
Anyway, partway through a non-exhaustive search, this appeared. Your database doesn't recognize it and it's not in jslife, so now I wonder. ...
Huh, I was expecting 'this is 21.135 from an obscure pentadecathlon.com list' or something. Now I've gotta go synthesize it.mniemiec wrote:Interesting. That's 3 new ones (that, plus two variants w/snake or cis-carrier instead of trans-carrier). I'm not surprised, as I only manually searched for "obvious" extensions of smaller objects. There are bound to be a few new non-trivial ones that would likely only show up in exhaustive searches.
Noted. I'll start an exhaustive search once I'm done collecting the higher-hanging fruit from this partial search tree.mniemiec wrote:I asked Nicolay Beluchenko about this (because he was the one who did the actual search).
Here is what he wrote me on 2016-04-10:
Code: Select all
x = 14, y = 15, rule = B3/S23
2ob2o$bobobob2o$bo3bobo$2bo2bo2bo$3bob2ob3o$2b2o$2bo2b2o$2bob3o$2bo5b
3o$6bo$6bo4bo$7b2o2bo$9bob2o$3b2obobo2bobo$3bob2ob2o2b2o!
Reduced (L=population, R=bounding box):Saka wrote:Is this random p3 new? I couldn't find it in jslife, but I am bad at looking for rotors.Code: Select all
x = 14, y = 15, rule = B3/S23 2ob2o$bobobob2o$bo3bobo$2bo2bo2bo$3bob2ob3o$2b2o$2bo2b2o$2bob3o$2bo5b 3o$6bo$6bo4bo$7b2o2bo$9bob2o$3b2obobo2bobo$3bob2ob2o2b2o!
Code: Select all
x = 33, y = 15, rule = B3/S23
3b2ob2o15b2ob2o$2o2bobo13b2o2bobo$o3bo3bo11bo3bo3bo$2bobo3bo13bobo3bo$
ob3o3bo11bob3o3bo$4b2o18b2o$o19bo$b2obo3bo12b2obo3bo$6bob2o16bob2o$5bo
2b2o15bo2b2o$6b2obo16b2obo$8bobo12b3o2bobo$8bobobo10bo2bobobobo$5b2obo
2b2o13b2o3b2o$5bob2o!
I think there are waaaaaaay many p3s with that rotor population for us to count every one.AbhpzTa wrote:Reduced (L=population, R=bounding box):Saka wrote:Is this random p3 new? I couldn't find it in jslife, but I am bad at looking for rotors.Code: Select all
x = 14, y = 15, rule = B3/S23 2ob2o$bobobob2o$bo3bobo$2bo2bo2bo$3bob2ob3o$2b2o$2bo2b2o$2bob3o$2bo5b 3o$6bo$6bo4bo$7b2o2bo$9bob2o$3b2obobo2bobo$3bob2ob2o2b2o!
Code: Select all
x = 33, y = 15, rule = B3/S23 3b2ob2o15b2ob2o$2o2bobo13b2o2bobo$o3bo3bo11bo3bo3bo$2bobo3bo13bobo3bo$ ob3o3bo11bob3o3bo$4b2o18b2o$o19bo$b2obo3bo12b2obo3bo$6bob2o16bob2o$5bo 2b2o15bo2b2o$6b2obo16b2obo$8bobo12b3o2bobo$8bobobo10bo2bobobobo$5b2obo 2b2o13b2o3b2o$5bob2o!
This took quite a bit longer than my initial guess, but I've finished searching a 10x40 area for 21-bit p3s. Besides the recent p3, nothing new showed up--only stator variants and such. If there are any missed p3s, their 21-cell phases do not fit in this bounding box.Kazyan wrote:Noted. I'll start an exhaustive search once I'm done collecting the higher-hanging fruit from this partial search tree.
Great! Would it be possible for me to have a look at this (raw data in any format would be fine) so I can verify that these match my lists? I had thought that I had done all the trivial stator variants, but when the 20-bit ones were searched, a coupleo of stator variants were found that I had missed. Also, since that search also included pseudo-objects, I was able to verify that my list of P3 pseudo-oscillators up to 20 bits was also complete. My current hand-generated lists shows 422 21-bit P3 oscillators and 90 21-bit P3 pseudo-oscillators. As far as I know, nobody has yet tackled exhaustive searches of P4 or higher oscillators beyond 12 bits.Kazyan wrote:This took quite a bit longer than my initial guess, but I've finished searching a 10x40 area for 21-bit p3s. Besides the recent p3, nothing new showed up--only stator variants and such. If there are any missed p3s, their 21-cell phases do not fit in this bounding box.
I hope you're not shy about duplicates and pseudo-oscillators. Here is a pastebin link, in leau of me knowing how attachments work.mniemiec wrote:Great! Would it be possible for me to have a look at this (raw data in any format would be fine) so I can verify that these match my lists? I had thought that I had done all the trivial stator variants, but when the 20-bit ones were searched, a coupleo of stator variants were found that I had missed. Also, since that search also included pseudo-objects, I was able to verify that my list of P3 pseudo-oscillators up to 20 bits was also complete. My current hand-generated lists shows 422 21-bit P3 oscillators and 90 21-bit P3 pseudo-oscillators. As far as I know, nobody has yet tackled exhaustive searches of P4 or higher oscillators beyond 12 bits.
Oh no, that's great! Thanks! My own still-life/oscillator enumerator already emits duplicate and pseudo-objects, and I have filters that separate those out. All I need to do now is to write a filter that slices the huge RLE into separate objects and converts them into my own internal object format. (Presumably, all occurrences of two or more adjacent blank rows can be assumed to be pattern separators).Kazyan wrote:I hope you're not shy about duplicates and pseudo-oscillators. Here is a pastebin link, in leau of me knowing how attachments work.