Methuselah notability thread (help for newcomers)

[dvgrn]

We collectively try our best to welcome newcomers to the conwaylife community -- and newcomers seem to be disproportionately interested in methuselahs, so we see a lot of non-record-breaking methuselahs posted with questions like "is this known?".

Before anyone travels too far down the methuselah-hunting path, it might be worth calling attention to a few key links:

1) Check out the list of long-lived methuselahs.

2) The forum rules actually mention methuselahs specifically, saying that

4. ... Some specific examples of discoveries that won't be of interest to this community include:

1. ...
2. A methuselah that lasts fewer than 20,000 generations or is larger than a 20-by-20 box.

This is kind of painful news to convey politely to enthusiastic newcomers, and we don't always manage the "politely" part successfully. It's a lot of fun making independent discoveries, and we don't want to discourage that.

Still, I think it's pretty much true that methuselah discoveries start to be interesting to the wider community, only at the point where they actually set new records on the LifeWiki's long-lived methuselah list. There are many millions of non-record-setting methuselahs, far too many to pay attention to individually, so the standard has to be set pretty high for including a methuselah on that LifeWiki list.

This would be a good thread for newcomers to post their methuselahs on, to get feedback from the community on whether they're notable or interesting.

[ehenkes]

Welcome!
...
I would recommend getting set up with Golly as well. It is used by almost all Life/CA researchers to manipulate and discover patterns, and will allow you to copy and paste RLEs with ease instead of having to transcribe them manually. It also supports an infinitely-large universe and two algorithms for simulating Life.

I have written a little search application, and I started with 10 elements. This is my best pattern by now:

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbbo$bbbbbbbb$bbbbbbbb$bbbobbbb$bobbbbob$oobbboob$bobbbbbb$bbbbbbob!

In Golly I found 2738 generations with an end population of 336.
Is this already known?

Best regards,
Erhard

[rabbit]

I have written a little search application, and I started with 10 elements. This is my best pattern by now:

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbbo$bbbbbbbb$bbbbbbbb$bbbobbbb$bobbbbob$oobbboob$bobbbbbb$bbbbbbob!

In Golly I found 2738 generations with an end population of 336.
Is this already known?

Best regards,
Erhard

Here's an 8x3 9-cell equivalent with the same lifespan:

Code: Select all

x = 8, y = 3, rule = B3/S23
2bo4bo$b3o2b2o$o5bo!

(The single cell saved came from the lone dot spark too far away to interact with the rest of the pattern. Otherwise unfortunately there's no general method to get the population down - you just sort of have to build up experience looking at small clusters)

I'm not sure whether this pattern has been explicitly examined, but systematic surveys of patterns around this size have been conducted. The longest lasting 9-cell methuselah is apparently Bunnies 9 at 17410 generations. It and its relatives have a disproportionately long lifespan compared to other 9-cell patterns, and I would highly suspect that there's no longer lasting 9-cell patterns.

One neat thing is that it does produce the aircraft carrier, which is a pretty uncommon object to appear randomly.

It could be advantageous to set your search program's cell count a bit higher, since there's more unexplored area in higher cell counts.

[ehenkes]

It could be advantageous to set your search program's cell count a bit higher, since there's more unexplored area in higher cell counts.

I have tried 11-cell-precursors. One case changes to a 9-cell-pattern that produces 4303 generations in Golly.

Code: Select all

x = 7, y = 7, rule = B3/S23
3bo2$o$2bo2bo$bo2b2o$bo$2bo3bo!

Is that already known?

[hotdogPi]

It could be advantageous to set your search program's cell count a bit higher, since there's more unexplored area in higher cell counts.

I have tried 11-cell-precursors. One case changes to a 9-cell-pattern that produces 4303 generations in Golly.

Code: Select all

x = 7, y = 7, rule = B3/S23
3bo2$o$2bo2bo$bo2b2o$bo$2bo3bo!

Is that already known?

I think you have the wrong RLE; that creates a pulsar.

[ehenkes]

Yes, you are right. The RLE is wrong. Sorry, I am newbee with Golly.

This is the 11-cell-precursor:

Code: Select all

x = 8, y = 7, rule = B3/S23
6bo$b2obo$4bo$o4bobo$7bo2$5b2o!

which leads in the first step to this 9-cell-precursor producing more than 4300 generations:

Code: Select all

x = 6, y = 5, rule = B3/S23
2bobo$ob3o$5bo$5bo$5bo!

I do not find it here: https://conwaylife.com/wiki/List%20of%2 ... ethuselahs

Interesting finding? I would call it "Hardy".

----------------

I have found another pattern with 10 elements producing 4616 generations in Golly with an end population of 1041:

Code: Select all

x = 6, y = 5, rule = B3/S23
bobo$2o2bo$bo3bo$2bobo$3bo!

Precursor in my search was:

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbbb$bbbbbbbo$bbbobobb$bbbbobbo$bbbboboo$bbbbbooo$bbbbbbbb$bbbbbbbb!

------------------
This 11-cells pattern produces 5206 gens (end pop: 633):

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbobb$bobbbbbb$bbbbbbbb$bobobbbb$bbobbbob$bbbbbobb$bbbobbbb$bboobobb!

After first step it comes down to an 8-cell pattern (5205 gens):

Code: Select all

x = 3, y = 6, rule = B3/S23
o$o$o2$2o$3o!

--------------------------------------------------
This 9-cell pattern produces more than 5200 gens:

Code: Select all

x = 6, y = 5, rule = B3/S23
o2$b2ob2o$o3b2o$5bo!

--------------------------------------------------
This 11-cell pattern ends near 3000 gens with an famous oscillator with period = 15.

Code: Select all

x = 8, y = 8, rule = B3/S23
bbobbbbb$oobbbbbb$bbbbbbbb$bbobbbbb$bobbbbbb$bbobooob$obbbbobb$bbbbbbbb!

Oscillator_15:

Code: Select all

x = 10, y = 5, rule = B3/S23
2b6o$bo6bo$o8bo$bo6bo$2b6o!

I have looked it up: https://conwaylife.com/wiki/Pentadecathlon (1970 by Conway)

This 11-cell pattern ends with a vertical Pentadecathlon:

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbbb$oobobbbb$obbbbbob$bbobbbbo$bbbbobbo$bbbbbboo$bbbbbbbb$bbbbbbbb!

This 9-cell pattern shows longest evolution (6929 gens) with end population of 1124:

Code: Select all

x = 6, y = 3, rule = B3/S23
o3b2o$b2ob2o$bo2bo!

Oh, I just see that this pattern is the famous "Jaydot". I am eleven years too late with my search app.

Here I found a 10-cell pattern running 6418 gens, and it produces a stable pulsar at gen 5928 until the end:

Code: Select all

x = 4, y = 7, rule = B3/S23
o2$obo$3o$o$2b2o$3bo!

[ehenkes]

This is my best finding by now:
11-cell pattern with 17408 generations (in LV) and 17411 in Golly, end population = 1744:

Code: Select all

x = 8, y = 6, rule = B3/S23
4bo$2bo2bo$b2o$2o$6b2o$6b2o!

Found: Nov 01, 2025 by Dr. Erhard Henkes
Is it a known "bunny 11"? I do not find it in https://conwaylife.com/wiki/List%20of%2 ... ethuselahs and https://conwaylife.com/wiki/Bunnies#Variants

[ehenkes]

Here is a 12-cell pattern, a "bunny 12"?
17407 gens, end pop 1744:

Code: Select all

x = 8, y = 5, rule = B3/S23
2o$2o4bo$5bobo$5bobo$4b3o!

Found: Nov 01, 2025 by Dr. Erhard Henkes

[ehenkes]

Here I found a 13-cell direct parent of the interesting oscillator "Figure Eight" https://conwaylife.com/wiki/Figure_eight

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbob$bbbooobo$bbbbbbbb$bbbbbbob$ooooobbb$bbbbbbbo$bbbbbbbb$bobbbbbb!

[LuveelVoom]

Here I found a 13-cell direct parent of the interesting oscillator "Figure Eight" https://conwaylife.com/wiki/Figure_eight

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbob$bbbooobo$bbbbbbbb$bbbbbbob$ooooobbb$bbbbbbbo$bbbbbbbb$bobbbbbb!

Two of those cells are unnecessary:

Code: Select all

x = 8, y = 5, rule = B3/S23
6bo$3b3obo2$6bo$5o!

[hotcrystal0]

This 11-cell pattern ends with a vertical Pentadecathlon:

Code: Select all

x = 8, y = 8, rule = B3/S23
bbbbbbbb$oobobbbb$obbbbbob$bbobbbbo$bbbbobbo$bbbbbboo$bbbbbbbb$bbbbbbbb!

Speaking of unnecessary cells, this one has three:

Code: Select all

x = 8, y = 5, rule = B3/S23
2o$o5bo$7bo$7bo$6b2o!

By the way it's known and forms from a certain collision between a mango and a glider, but I forgot where it's from.

[ehenkes]

11-cell with 10334 gens and end pop of 1335:

Code: Select all

x = 7, y = 5, rule = B3/S23
4b3o$4bo$bo2b2o$obo$obo!

[ehenkes]

11-cells with 11730 gens with end pop of 1158 and interesting intermediates (e.g. pulsar at 6240)

Code: Select all

x = 8, y = 7, rule = B3/S23
6b2o$b2o3bo$2o$b2obobo2$6bo$7bo!

[ehenkes]

Please have a look at my bunny 11/12 patterns (cf. above). Are they new? By now I do not understand the bunny mechanism fully.
Should I post them at "patterns"?

[dvgrn]

Please have a look at my bunny 11/12 patterns (cf. above). Are they new? By now I do not understand the bunny mechanism fully.
Should I post them at "patterns"?

This thread is probably a good place for now.

See the first post in this thread, which has a 2024 posting date due to Moderator Shenanigans. Without the date adjustment, it would have showed up far down in the thread instead of first, since I just wrote it a few minutes ago.

Some "Is it known?" and "Is it new?" questions don't really have good answers, which may be why people aren't always answering those questions. Given that there are known 9-cell patterns with lifespans of 17,330+, it's guaranteed that there will be a *lot* of 10-, 11-, and 12-cell patterns with similar lifespans. They're "known" in the sense that they're known to be out there, and we can generate thousands of them on demand -- but most of them don't break any records, so individual examples don't hold people's attention for long.

The other thing that maybe I should mention is that it might be better to not be in a big hurry to apply your own names to new discoveries. Other people are very likely to politely ignore those names, unless the discoveries are really record-breaking. If the community didn't follow the general rule of applying new names very sparingly, we'd get overrun with arbitrary new names very very quickly: enthusiastic new pattern namers show up on the forums quite regularly, often before they've really become familiar with the LifeWiki notability guidelines and the forum rules.

[rabbit]

By now I do not understand the bunny mechanism fully.

(Well, there might be a conflict of interest if I explain it )

The rabbits and bunnies sequences all converge to this single cluster of cells:

Code: Select all

x = 5, y = 6, rule = B3/S23
3b2o$3b2o$2bo$2o$o$b2o!

The fact that its cell count and bounding box are both rather small, and the fact that it's a well-connected cluster, all likely help give rise to the sheer volume of low-population relatives on the sequence. And, it just so happens to also last ~17000 more generations, so it pops out when looking at long-lived patterns.

[ehenkes]

Please have a look at my bunny 11/12 patterns (cf. above). Are they new? By now I do not understand the bunny mechanism fully.
Should I post them at "patterns"?

This thread is probably a good place for now.

See the first post in this thread, which has a 2024 posting date due to Moderator Shenanigans. Without the date adjustment, it would have showed up far down in the thread instead of first, since I just wrote it a few minutes ago.

Some "Is it known?" and "Is it new?" questions don't really have good answers, which may be why people aren't always answering those questions. Given that there are known 9-cell patterns with lifespans of 17,330+, it's guaranteed that there will be a *lot* of 10-, 11-, and 12-cell patterns with similar lifespans. They're "known" in the sense that they're known to be out there, and we can generate thousands of them on demand -- but most of them don't break any records, so individual examples don't hold people's attention for long.

The other thing that maybe I should mention is that it might be better to not be in a big hurry to apply your own names to new discoveries. Other people are very likely to politely ignore those names, unless the discoveries are really record-breaking. If the community didn't follow the general rule of applying new names very sparingly, we'd get overrun with arbitrary new names very very quickly: enthusiastic new pattern namers show up on the forums quite regularly, often before they've really become familiar with the LifeWiki notability guidelines and the forum rules.

Thank you for your answer. The problem is that there is no app (with AI) that give you those answers by entering/opening a RLE.
My conclusion: I think, it makes no sense to search for new patterns on a personal computer with an 8*8 pattern matrix.
My question: What can a newbee with programming abilities do? (do not say: rtfm)

[hotdogPi]

My question: What can a newbee with programming abilities do? (do not say: rtfm)

One thing that has been talked about but not implemented well yet is finding "naturalistic" predecessors of patterns. We have backtracking algorithms, but the predecessor is typically larger than the original, and after 15-20 steps backwards, you reach something that doesn't have any predecessors (this is called a Garden of Eden).

For example, here are generations 9 and 10 of the R-pentomino. The smallest possible parents of generation 10 do not include the three isolated cells in generation 9, but they're necessary if you want to backtrack further, and a good backtracking algorithm would include them.

Code: Select all

x = 18, y = 8, rule = B3/S23
3bo$bo2bo$o13b2o$o4bo7b2o$2o12b2o$15b3o$2b3o12bo$17bo!

Prior discussion: viewtopic.php?p=196293#p196293

[ehenkes]

My question: What can a newbee with programming abilities do? (do not say: rtfm)

One thing that has been talked about but not implemented well yet is finding "naturalistic" predecessors of patterns. We have backtracking algorithms, but the predecessor is typically larger than the original, and after 15-20 steps backwards, you reach something that doesn't have any predecessors (this is called a Garden of Eden).

For example, here are generations 9 and 10 of the R-pentomino. The smallest possible parents of generation 10 do not include the three isolated cells in generation 9, but they're necessary if you want to backtrack further, and a good backtracking algorithm would include them.

Code: Select all

x = 18, y = 8, rule = B3/S23
3bo$bo2bo$o13b2o$o4bo7b2o$2o12b2o$15b3o$2b3o12bo$17bo!

Prior discussion: viewtopic.php?p=196293#p196293

I have written my first C++ program, which works by finding direct predecessors. Challenges remain in terms of completeness and omitting unnecessary cells. I will continue to follow this interesting topic. If you are interested in the current status of this program, I would be happy to provide it to you. Input is in form of a file "input.rle". Outputs are "output1.rle", "output2.rle", etc., and they are shown in the terminal.