ConwayLife.com

Why not limit such checks to [pattern] tag?

7363 is 37x199, so guntrue_7363 is a composite gun
Anyway, this is a smaller design
x = 123, y = 108, rule = B3/S23
...


x37 multiplier makes it even smaller:
x = 113, y = 106, rule = LifeHistory
16.2A$15.B2A2B7.2A$16.4B6.B2AB$11.2A4.6B4.2B$10.B2AB2.8B4.2B9.A$11.3B
.17B7.3A$9.2B.B2.18B9.A$8 ...

I'm not sure of the best way to upload these to catagolue, and it feels like it might crash it or something. Can someone with more knowledge advise me on this?

I prefer to upload guns one by one (with yet another script).
If Catagolue fails to process 300 rles, it can process 100, 100, 100 - at ...

New 1 mod 16 attachment for gun family I mentioned previously allowed for 6 optimizations:
gun_1873: 13734 -> 10332
gun_1889: 13860 -> 10414
gun_2017: 13298 -> 11880
gun_2081: 14430 -> 12788
gun_2113: 14916 -> 13254
gun_2129: 15162 -> 13490
x = 126, y = 82, rule = LifeHistory
56.2A11.A$55.B2AB9.A.A ...

NW31T198_NW35T275 - new timing
x = 57, y = 57, rule = LifeHistory
5.D3.D.D3.D2.3D4.D3.5D3.D4.3D3.3D$5.2D2.D.D3.D.D3.D2.2D5.D4.2D3.D3.D.
D3.D$5.D.D.D.D3.D5.D3.D5.D5.D3.D3.D.D3.D$5.D2.2D.D.D.D3.2D4.D5.D5.D4.
4D2.3D$5.D3.D.D.D.D5.D3.D5.D5.D7.D.D3.D$5.D3.D.2D.2D.D3.D3.D5.D5.D7.D
.D3.D$5.D3.D.D3.D2.3D3 ...

I think a better idea is to continue fill in missing periods from 1000-2000 (it goes to about 1250 before getting patchy)


There is slight problem with this idea: guns, which consist of base and lots of multipliers, become obsolete all at once when base gun is updated. The more such guns in the ...

It would be more economical to use one glider + sparky eater.

Here is G -> H -> MES implementation for gun_3634 (12312 -> 9956):
x = 131, y = 76, rule = LifeHistory
30.2A$30.A.A$32.A4.2A$28.4A.2A2.A2.A$28.A2.A.A.A.A.2A$30.BABABA.A$31.
B2ABA.A$32.2B.BA$31.3B$22.2A6.4B$23.A6.B2A3B$23.A.AB3.B2A3B ...

erictom333 wrote: October 7th, 2026, 1:55 am It would be more economical to use one glider + sparky eater.

If someone finds compatible glider to spark (crotchet) converter, please report it here.

Did anyone tried glider synthesis of spark for LCM guns before?

x = 67, y = 45, rule = LifeHistory
11.2A4.2A$11.A.A2.A.A$13.A2BA12.2A$11.B2A2B2AB2.A2.A4.A$11.3A2B3A2.4A
5.A$12.BA2BAB7.2A2.2A$6.2A6.2B4.A.2A.A.A3.2A$4.A2.A6.B5.2A.ABA.2A.A2.
A$4.2A.A.AB2.2B9.BA2B3.3A28.2A$6.A.2AB.4BA7.2AB4A31.A.A$6 ...

... but this is merely a proof that a lot of high-period guns can be reduced by this POG method.


Did you used code for gun regeneration?
I'm thinking that maybe some semiprime (two factor) guns worth generating as well.
For example, gun_7519 (73 * 103) was a good candidate for upload.
While gun ...

hash0 wrote: October 2nd, 2026, 11:50 am I am more than happy to adjust that switch if and when we discover further classes of patterns!

I think this pattern alone is enough reason for switch tuning:

Looks like "INCLUDE AGAR" switch in LONGEVITY_CHALLENGE needs rethinking.

Could it be viable to construct an "ultra-high-bandwidth telegraph"?

I doubt I will participate in this challenge, but I have a question:

Is it important to keep particular reaction/wire described in Telegraph article?
With different (thicker?) wires possibilities are endless.

Also what about ...

WhiteHawk wrote: September 9th, 2026, 11:57 am ... I have a secondary source that Gun and Guntrue submissions aren't updating for some reason. ...

I reported it week ago.
(Syntheses and guns are processed by the same code)

recursive runner LOVES stable memory

I added an example to my previous message, it does exactly that - interfacing with stable memory.

combine them without overlapping BBs

Good idea.

I had one more idea, but it may not matter for your algorithm:
I thought about having only one (or maybe two ...

... if we want something like that in CGoL, we may need an approach like Quest for Tetris , but with multiple layers of abstraction...

I think it is better to just directly implement Subleq computer in CGoL.
It is not that hard, all required components already exist, they just need to be ...

Your suggestion looks more like new compression algorithm than optimization.
I know it is fun to develop such things, but it is rarely makes practical sense besides training.
I'm almost sure RLE is used for cellular automata not because it is the best, but because it is better than nothing.
For ...

NNlk05 wrote: September 5th, 2026, 1:08 pm For some reason, when I use nbsearch2a and stop it all the cells are in this weird greyed-out state that counts as dead:

Looks like new layer was created. I suggest to try clicking Layer -> Delete Layer.

... I had assumed that bigger tori dimensions would yield other previously unknown Agar patterns. This does not appear to be the case (so far). ...
... I had naively assumed that increasing the torus gride size would massively (exponentially?) increase the longevity of patterns - this, too ...

I don't know if it is useful or new, but anyway, here is MES to dot:

WhiteHawk wrote: September 4th, 2026, 1:42 am ... and made a very dirty version of the 2026 stamp collection ...

P152 is broken here, please fix.

Syntheses verification is failing for a week now (since August 24, 2026). Please fix.

readsynth data (2 megabytes) looks ok. Local run with disabled upload of results looks ok as well (42259 original contributions).
I guess it is /commonnames request which is failing because of "large" payload.

Do you know gliders can be intercepted with eaters? Like so:
x = 51, y = 46, rule = B3/S23
44b2o$44b2o3$44b3o$44b3o2$25b2o$23bo2bo$10bobo9bo7b5o7b2o3b2o$10bo3bo
7bo6bo5bo7b5o$14bo7bo7b2o3bo8b3o$2o8bo4bo7bo2bo7bo10bo$2o12bo10b2o$10b
o3bo$10bobo8bobo$22b2o$22bo2$44bo$42b2o$43b2o5$44b2o3b2o$35bobo8b3o ...

One problem is that when Golly is zoomed out enough, all colours go back to the state one colour. If we’re making a screen, I don’t think we should require a script to view the screen. Then we might as well have the script-screen and not the real one.


I think we should take into account ...

I don’t really know much about hashlife/computing in general, but what if we simply make it hashlife-friendly? Each hash could store an operation, and then we would pretty much be back to normal speed? Correct me if I’m wrong.

I think there will be problems with HashLife memorizing huge variety ...