I'm making a new Methuselah

[Pigeonbee]

Probably been made before, but ah well. Made it from randomly placed blocks. I found a pattern that lasts 2,713 generations, I'm hoping to change it slightly each time to see if it lasts longer than 2,713 generations. Wish me luck.

5K1Glsz.png (2.15 KiB) Viewed 178 times

Detailed info about the Methuselah (So far):
>Lasts 2,713 generations
>Ends with a population of 224, (289 if you count the 13 gliders)
>I'm pretty sure it hasn't been found yet
>Makes:
13 Gliders
14 Blinkers
18 Blocks
1 Tub
9 Beehives
1 Boat
1 Long Boat
1 Ship
4 Loaves

[drc]

I don't want to sound mean, but:

Code: Select all

x = 7, y = 3, rule = B3/S23
o3b3o$3o2bo$bo!

[Pigeonbee]

I don't want to sound mean, but:

Code: Select all

x = 7, y = 3, rule = B3/S23
o3b3o$3o2bo$bo!

I'm kinda new to this forum, so I have no clue what that means.

[drc]

I don't want to sound mean, but:

Code: Select all

x = 7, y = 3, rule = B3/S23
o3b3o$3o2bo$bo!

I'm kinda new to this forum, so I have no clue what that means.

Click show in viewer, It displays a methuselah of 9 cells, and lasts 17331 generations.

[Pigeonbee]

Oh. Ah well.
Anyway, I put a block under the pattern, so it lasts for about 2,700 generations.
I guess that's progress.
I'll update the thread anyway.

[drc]

Oh. Ah well.
Anyway, I put a block under the pattern, so it lasts for about 2,700 generations.
I guess that's progress.
I'll update the thread anyway.

You should try other things, like beehive, boat, tub, blinker, ship, and loaf

[Pigeonbee]

Oh. Ah well.
Anyway, I put a block under the pattern, so it lasts for about 2,700 generations.
I guess that's progress.
I'll update the thread anyway.

You should try other things, like beehive, boat, tub, blinker, ship, and loaf

Oh okay.
I guess I will experiment more.
There's probably something about the pattern that I haven't noticed and is very important for development of the Methusalah.
Could you help me in making it last longer and such?

[muzik]

You should try and make the gliders hit blinkers in such a way that makes even more mesuthelahs.

Like this:

Code: Select all

x = 10, y = 8, rule = B3/S23
7bobo$7b2o$8bo5$3o!

[simeks]

I've been meaning to search for a methuselah at one time or another, and since the subject came up...

My starting patterns were random 7-by-7 soups.

About one pattern in 100000 converges on bunnies/rabbits.

The second most common pattern lasting more than 15000 generations seems to be this one, which is easily recognizable by the five blocks it lays down early on. It occured about once in a few million soups:

Code: Select all

x = 7, y = 7, rule = LifeHistory
A5.A$A4.A$3.4A$.2A2.2A$4A2.A$4A2.A$3.A.A!

After searching more than 10 million 7-by-7 soups, this is the first one I've seen that lasts longer than bunnies. It settles in generation 19065:

Code: Select all

x = 7, y = 7, rule = LifeHistory
3A.A$2A.A.2A$2.A.A.A$.A3.2A$A.A.A$6.A$4A.A!

Edit: This one is slightly better (settles in generation 19869):

Code: Select all

x = 7, y = 7, rule = LifeHistory
3.2A$3A3.A$3A.3A$2A3.A$.A.2A.A$2.2A$4A2.A!

Edit2: Generation 21581:

Code: Select all

x = 7, y = 7, rule = LifeHistory
.A3.A$2A.A.A$A$5A$A.A2.A$.3A.2A$4A!

Edit3: Generation 22000:

Code: Select all

x = 7, y = 7, rule = LifeHistory
2A$A.A$3A$3A$3A.3A$4.2A$4.A!

[TheoSwartz]

It's amazing that patterns in such a small space can create such varied and long lasting results. One of the many reasons Life is cool!

[simeks]

Edit: Just a test run of my new vectorizable evolving code, but possibly notable...
23874 generations from a 7-by-7 bounding box:

Code: Select all

x = 7, y = 7, rule = Life
3bobo$2b2o2bo$3b2obo$ob2o2bo$2b4o$b3o2bo$b5o!

[drc]

I've been meaning to search for a methuselah at one time or another, and since the subject came up...

Do you have the script?

[simeks]

Do you have the script?

The code is a bit too sketchy to publish, and my GoLGrid datatype isn't well suited to evaluate such a large grid that is needed -- a tile-based datatype would be much better.

That said, if anyone wants to put some time into running the 64-bit Windows executable, to search for a methuselah, I can post that. I will need to know if your CPU has support for AVX2 (256-bit integer vector operations) or not.

In the meantime, a different kind of methuselah -- 3192 generations without any escaping gliders:

Code: Select all

x = 6, y = 6, rule = LifeHistory
.3A.A$.A.A$A3.A$2.4A$.2A.A$.3A.A!

[drc]

That said, if anyone wants to put some time into running the 64-bit Windows executable, to search for a methuselah, I can post that. I will need to know if your CPU has support for AVX2 (256-bit integer vector operations) or not.

I don't know, but apgsearch runs in AVX1, so I guess not

[simeks]

I don't know, but apgsearch runs in AVX1, so I guess not

Ok I'll post a version, let's see how it works. Some things are hard-coded:
- The starting pattern is a random 7-by-7 square.
- The grid size is 768-by-768. It will be enough most of the time, but if it isn't the program will fail to find out the true life span of a pattern.
- The reported life span is not exact, and will need to be verified manually.
- "False positives" are possible, mostly because gliders will be converted to blocks at the grid edge, and then other gliders might hit it.

Run from the command line:
meth128 8000

where 8000 is the smallest life span that will be reported, 19000 might be a good value to find only interesting results.
The program will run until stopped with Ctrl-C. To leave running unsupervised, you could redirect the output:
meth128 19000 >out.txt

meth128.zip

(21.99 KiB) Downloaded 240 times

[Pigeonbee]

Apologies for the necromance, but is growth rate a thing, and how would I measure it? Would I need to somehow work out an average, by working out how many cells there are every 200 generations, or something like that? My methuselah appears to grow pretty quickly here. I'll try and figure something out, using precision big red square prodding.

[BlinkerSpawn]

Apologies for the necromance, but is growth rate a thing, and how would I measure it?

To start, try running pop-plot.lua on your methuselah and take a gander at the resulting graph.
For any methuselah that doesn't settle into switch engines (or something else!) plus a gun, the asymptotic population* growth rate will be O(1) (constant, no growth) but interesting dynamics may occur during regular evolution.
Try manually grabbing populations at set step sizes and plugging a population table into some program that can run regressions to get a basic idea.

*bounding box growth rate will certainly be O(x) (linear), unless there's no gliders.