Thread for basic questions

[dvgrn]

Are there standard names for polyplets like there are for polyominoes up to a certain population?

I've never heard of any -- except for Life-specific polyplets like "tub" and "banana spark" of course. With 22 options even for N=4, one runs out of alphabet awfully early.

[confocaloid]

What are (useful or otherwise interesting) known ways to count predecessors of a pattern?
What are (useful/interesting) known ways to answer questions of the form "How many predecessors does this finite-population pattern have?"

Of course one possible way is "either zero or infinitely many, depending on whether it has any predecessors at all". (It is always possible to add a dying spark far away.)

I'm more interested in "low astonishment factor" ways to get a well-defined finite count, so that it would be possible to implement a software tool that computes the answer, and it would be reasonably easy for the users to understand why the program gives those specific numbers.
"Either zero or infinitely many" does have low astonishment factor ("easy to understand why"), but it doesn't give a nonzero finite count.

For "test cases", one could take generations of common evolutionary sequences and objects occurring in the b3s23/C1 census.

[dvgrn]

What are (useful or otherwise interesting) known ways to count predecessors of a pattern?

lifesrc and its variants (WLS, JLS) can produce clear answers to questions like

"How many predecessors of such-and-such MxN pattern fit inside such-and-such bounding box"

... including

- the same MxN bounding box
- the same bounding box but extended k cells to the east/west/north/south
- the same bounding box extended k cells in two/three/all directions
- etc. (could extend by different amounts in different directions, or look for predecessors in bounding boxes smaller than MxN or shifted slightly though usually you won't find any of those)

The counts are different for different starting patterns and for each different expansion of the bounding box.

The counts tend to increase radically as the bounding box increases, and once the numbers get big enough they become both much harder to calculate in practice and much less interesting.

The counts are calculable in a reasonable amount of time for small enough MxN.

The more interesting cases might be the ones where the count is zero at MxN and immediate extensions, but non-zero at various larger extensions (so it's not a Garden of Eden, but it's a near miss).

These days, if the count is zero, it's generally much faster to prove that using LLS (with a method like this) instead of a lifesrc variant.

[WhiteHawk]

Is there a collection for sparkers with periods greater than 16?

EDIT: In the sense of sparkers that aren't based on adjustable conduits or glider/spaceship streams? More like regular hasslers.

[confocaloid]

Is there a collection for sparkers with periods greater than 16?

It might be more interesting to collect unique ways of obtaining sparkers. Above some period, there will be ways to collide two glider streams to obtain a given spark, there will be stable converters converting some spaceship or active object (glider, Herschel, pi-heptomino, etc) into the spark of interest, and so on. Instead of collecting sparkers one would collect the "building blocks" (tools, components) that can be used to make sparkers.

[...] EDIT: In the sense of sparkers that aren't based on adjustable conduits or glider/spaceship streams? More like regular hasslers.

Ways to construct sparkers using conduits or colliding glider/spaceship streams also can be included in a collection. However, it would suffice to show every unique reaction or component just once.

For example, if you add 88P93 as a duoplet sparker, then there's no need to add any other duoplet sparkers using the same spark-emitting mechanism (BRx46B conduit), it suffices to give only one example of the use of a given spark-emitting component.
In those cases where all "completed" sparker oscillators are large contraptions, it would be better to show just the spark-emitting component itself, described sufficiently to understand how it should be used.

88P93 gives duoplet sparks and can separate closely spaced parallel glider streams

Code: Select all

x = 123, y = 92, rule = B3/S23
72bo$70b3o$69bo20bo$69b2o18bo2$66b2o5b2o$66bobo4b2o21b2o$67b2o20b3o4b
2o$91bo$79b2o$77b2obo$77b2ob2o$77b2ob2o$79b2obo$63b2o16b3o2bo$63b2o16b
obobobo7b2o$82bo5bo6b2o$83bo4bo$79bo3bo3bo$78bobo3bo2bo$5bo12bo60bo$4b
o2b2o6b2o2bo$9b2o2b2o$6b2ob2o2b2ob2o65bo$2o5bo2bo2bo2bo5b2o51bo2bo3bob
o$2o8bo2bo8b2o51bo3bo3bo$6bo3bo2bo3bo56bo4bo$6bo2bo4bo2bo48b2o6bo5bo$
6bobo6bobo48b2o7bobobobo16b2o$76bo2b3o16b2o$6bobo6bobo62bob2o$6bo2bo4b
o2bo63b2ob2o$6bo3bo2bo3bo63b2ob2o$2o8bo2bo8b2o58bob2o$2o5bo2bo2bo2bo5b
2o58b2o$6b2ob2o2b2ob2o53bo$9b2o2b2o50b2o4b3o20b2o$4bo2b2o6b2o2bo45b2o
21b2o4bobo$5bo12bo69b2o5b2o2$73bo18b2o$72bo20bo$90b3o$90bo19$44bo$43b
2o$43bobo$98bo$97b2o$97bobo16$105b3o$105bo$106bo$66b3o$66bo$48b3o16bo$
48bo71b3o$49bo70bo$121bo!
#C [[ THEME LifeHistory ]]

It's worth noting that the block in BRx46B can be placed in one of two different locations:

Code: Select all

x = 90, y = 30, rule = B3/S23
6bob2o65bob2o$6b3o66b3o$b2o4bo62b2o4bo$b2o67b2o7$6b2o61b2o4b2o$2o3bo2b
o60b2o3bo2bo$2o4b2o67b2o15$11b2o5b3o59b2o5b3o$10b2o6bo60b2o6bo$12bo6bo
61bo6bo!

The placement on the right has one extra lane of clearance.

[TYCF]

What new patterns have been found and what has happened in the forums in the last 6 months?

[b-engine]

What new patterns have been found and what has happened in the forums in the last 6 months?

Category:Patterns found in 2025

[dvgrn]

See also LifeWiki CurrentNews. It's been good times for new small spaceships for various definitions of "small", and new and reduced true-period glider guns -- along with some progress on small spaceship synthesis, new Life computer technology, and APGsembly tools, and quite a collection of new conduit discoveries including Spartan ones.

[confocaloid]

What new patterns have been found and what has happened in the forums in the last 6 months?

[...] Category:Patterns found in 2025

...which is a very incomplete answer to the first part of the question.
For a less incomplete answer, check also LifeWiki:News_archive#2025. There are discoveries mentioned in the news, but missing any page that could possibly be added into a category. It is still incomplete (many interesting discoveries were made, without being mentioned on the wiki in any way).

As for the "what has happened in the forums" part, there's so much I hardly know where to start. But to give one example, there is now a new board ("Help / Support"), which hopefully helps.
Related discussion: viewtopic.php?f=7&t=6809

[pifricted]

What is “John Von Neumann’s Cellular automata”?
What is “Von Neumann-type Cellular Automata”?

(I’m sorry if they have own pages on Wikipedia)

[PiSpaceships]

What is “John Von Neumann’s Cellular automata”?
What is “Von Neumann-type Cellular Automata”?

Cellular automata with von-Neumann neighborhood:

Code: Select all

010
1X1
010

[confocaloid]

I think it is more specific than just using von Neumann neighbourhood:
https://en.wikipedia.org/w/index.php?ti ... 1270072067
https://golly.sourceforge.io/Help/Algorithms/JvN.html

What is “John Von Neumann’s Cellular automata”?
What is “Von Neumann-type Cellular Automata”?

Cellular automata with von-Neumann neighborhood:

Code: Select all

010
1X1
010

[hotdogPi]

Regarding neighborhoods:

It might be easier to visualize this way. Von Neumann on the left, Moore on the right. From top to bottom: range 1 (what the vast majority of rules on this forum use), range 2, range 3.

Code: Select all

x = 17, y = 22, rule = LifeHistory
3.B8.3B$2.BCB7.BCB$3.B8.3B6$3.B7.5B$2.3B6.5B$.2BC2B5.2BC2B$2.3B6.5B$
3.B7.5B3$3.B6.7B$2.3B5.7B$.5B4.7B$3BC3B3.3BC3B$.5B4.7B$2.3B5.7B$3.B6.
7B!

Confocaloid has a good point that this might not be the question you asked, though.

[dl-rs]

How do I view .lif files from older sites? Golly does not seem to support them. By the way, I use Mac.

[dvgrn]

How do I view .lif files from older sites? Golly does not seem to support them. By the way, I use Mac.

Golly can usually open .lif files just fine. Those are usually the "Life 1.05/1.06" format files mentioned in Golly documentation.

There aren't too many of those .lif files floating around any more. Where exactly are you getting them? Can you post a sample of the contents? That might tell me what's going wrong with opening them.

[dl-rs]

How do I view .lif files from older sites? Golly does not seem to support them. By the way, I use Mac.

Golly can usually open .lif files just fine. Those are usually the "Life 1.05/1.06" format files mentioned in Golly documentation.

There aren't too many of those .lif files floating around any more. Where exactly are you getting them? Can you post a sample of the contents? That might tell me what's going wrong with opening them.

Thanks. I've opened it.
(In fact, I looked at a few files from Jason Summers's old sites and realised that the files I tried to open lacked the label of #Life 1.05.)

[dl-rs]

Is there a way to run apgsearch and store all of the pattern's RLE locally?

[confocaloid]

Is there a way to run apgsearch and store all of the pattern's RLE locally?

It is possible to use a separate program/script that outputs patterns (for example soups) as RLEs into the standard output, and "pipe" the output into apgsearch provided apgsearch is compiled for a census with name containing the substring "stdin".
It is possible to avoid uploading anything to Catagolue, by passing -t 1 into apgsearch parameters.
It is possible to save logs locally, by passing -L 1 into parameters.

Code: Select all

# ./apgluxe -t 1 -L 1 -n 10000 -i 0

If apgsearch is compiled to census patterns given as RLEs through stdin, then I think the saved log will contain actual RLEs.

[jeremydover]

I have written code that leverages Google's ORTools to minimize the stator size for oscillators and guns; basically it preserves the rotor, and then rebuilds a stator with the minimum number of possible live cells in the bounding box. Would it be of interest to create a table of minimal variants for oscillators noted in the Wiki? Without much difficulty I could determine minimum width and height as well. While most modern oscillators tend to be minimized, a number of historical ones are not: for example, Protein has a stator variant with 25% lower population.

[dvgrn]

Would it be of interest to create a table of minimal variants for oscillators noted in the Wiki? Without much difficulty I could determine minimum width and height as well.

Sounds interesting to me! We do have people with pretty good expertise with the optimization tools that we have, which is why modern oscillators are mostly already minimized.

I would think the community would also like to know more about the code you've written, and whether you have any thoughts about any other current open CA problems that OR-Tools could maybe be usefully applied to.

[jeremydover]

Excellent, thank you for the quick feedback...that it is positive is even better My code is moderately hacked together, so I'm hoping that going through the Wiki will help me smooth some of the rough edges, to the point that the code itself is worth sharing. But it is yielding results, even on more modern finds: for example, I can shave 2 cells off the p10 domino fountain, and 8 cells off the long/narrow variant.

Ultimately, ORTools provides a constraint programming interface with a SAT solver backend. The main differentiator between ORTools and LLS is probably the constraint programming primitives of Minimize and Maximize, but I don't have enough stick time on LLS to know for sure.

[dvgrn]

I'm hoping that going through the Wiki will help me smooth some of the rough edges, to the point that the code itself is worth sharing.

A good starting point might be a new OR-Tools thread here on the forums.

Once the rough edges get knocked off, some version of the information could get transferred to the LifeWiki (and then we'll have forum-post references here to point back to).

[jeremydover]

...
A good starting point might be a new OR-Tools thread here on the forums.
...

Per dvgrn's suggestion, I have made a post here.

[confocaloid]

Imagine

• two orthogonal spaceships (each of which must have a known glider synthesis), both moving in the same direction but with different speeds, the faster spaceship behind the slower spaceship,
• some time later the spaceships collide, and the collision cleanly produces a single "forward" glider whose direction is "clockwise" relative to the direction of the colliding spaceships,
• if the slower spaceship is advanced by several ticks relative to the faster spaceship (without any other modifications; for example each of spaceships must remain on the same lane as before), then the new collision must cleanly produce a single "backward" glider whose direction is "clockwise" relative to the direction of the colliding spaceships.

Does the above exist in CGoL?

The following collision between a loafer and a HWSS fails to satisfy the latter requirement (there is no way to cleanly get a single "backward clockwise" glider merely by advancing the slower spaceship):

[...]
If you're doing arbitrarily distant constructions, though, destroying the loafer is fine. All that is really needed is a *WSS collision that produces a forward glider, and a *WSS collision that produces a backward glider, and then your powers of arbitrarily distant construction are great. Fortunately, both exist.

Loafer + HWSS = forward glider:

Code: Select all

x = 29, y = 15, rule = B3/S23
$21b6o$21bo5bo$21bo$22bo4bo$b2o2bob2o15b2o$o2bo2b2o$bobo$2bo$8bo$6b3o$
5bo$6bo$7b2o!

For backward gliders, you have 7 choices:

Code: Select all

x = 108, y = 52, rule = B3/S23

[...]

[WhiteHawk]

Does anyone know anything about the details behind this variant of 66p13 (or really 70p13)?

Code: Select all

x = 23, y = 17, rule = B3/S23
5bo$3b5o$2bo5bo$2bo2b3o$b2obo$o3b2o$bobo2bo8bo$2ob4o8b2o$6bo8b3o$2ob4o
8b2o$bobo2bo8bo3b2o$o3b2o13bobo$b2obo16bo$2bo2b3o13b2o$2bo5bo$3b5o$5b
o!